From ad82bb2c48564a6f23e005e01c89e4c12318510c Mon Sep 17 00:00:00 2001 From: peng_windows <2686728826@qq.com> Date: Tue, 26 Mar 2024 12:26:17 +0800 Subject: [PATCH 1/6] llm-introduction-attention-transformer --- .../_config.yml | 1 + open-machine-learning-jupyter-book/_toc.yml | 9 + .../llm/basic/transformer-architecture.ipynb | 730 + .../llm/basic/attention.ipynb | 367 + .../llm/basic/basic.ipynb | 64 + .../llm/basic/transformer.ipynb | 20020 ++++++++++++++++ .../llm/image/attention_example.svg | 9628 ++++++++ .../llm/image/cifar100_example_anomaly.png | Bin 0 -> 155392 bytes .../llm/image/comparison_conv_rnn.svg | 1809 ++ .../llm/image/implicit-order.png | Bin 0 -> 34807 bytes .../llm/image/llm.png | Bin 0 -> 166613 bytes .../llm/image/multihead_attention.svg | 288 + .../llm/image/scaled_dot_product_attn.svg | 351 + .../llm/image/scaling-laws.png | Bin 0 -> 86847 bytes .../llm/image/transformer_architecture.svg | 118 + .../llm/image/warmup_loss_plot.svg | 1579 ++ .../llm/introduction.ipynb | 155 + 17 files changed, 35119 insertions(+) create mode 100644 open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb create mode 100644 open-machine-learning-jupyter-book/llm/basic/attention.ipynb create mode 100644 open-machine-learning-jupyter-book/llm/basic/basic.ipynb create mode 100644 open-machine-learning-jupyter-book/llm/basic/transformer.ipynb create mode 100644 open-machine-learning-jupyter-book/llm/image/attention_example.svg create mode 100644 open-machine-learning-jupyter-book/llm/image/cifar100_example_anomaly.png create mode 100644 open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg create mode 100644 open-machine-learning-jupyter-book/llm/image/implicit-order.png create mode 100644 open-machine-learning-jupyter-book/llm/image/llm.png create mode 100644 open-machine-learning-jupyter-book/llm/image/multihead_attention.svg create mode 100644 open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg create mode 100644 open-machine-learning-jupyter-book/llm/image/scaling-laws.png create mode 100644 open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg create mode 100644 open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg create mode 100644 open-machine-learning-jupyter-book/llm/introduction.ipynb diff --git a/open-machine-learning-jupyter-book/_config.yml b/open-machine-learning-jupyter-book/_config.yml index e7464e4cd..a0f9b925d 100644 --- a/open-machine-learning-jupyter-book/_config.yml +++ b/open-machine-learning-jupyter-book/_config.yml @@ -23,6 +23,7 @@ execute: - 'ml-advanced/unsupervised-learning-pca-and-clustering.ipynb' - 'ml-advanced/unsupervised-learning.ipynb' - 'data-science/data-science-in-the-cloud/the-azure-ml-sdk-way.ipynb' + - 'llm/basic/transformer.ipynb' parse: myst_enable_extensions: diff --git a/open-machine-learning-jupyter-book/_toc.yml b/open-machine-learning-jupyter-book/_toc.yml index a09d32cd1..5c9b8c367 100644 --- a/open-machine-learning-jupyter-book/_toc.yml +++ b/open-machine-learning-jupyter-book/_toc.yml @@ -122,6 +122,14 @@ parts: - file: machine-learning-productionization/data-engineering - file: machine-learning-productionization/model-training-and-evaluation - file: machine-learning-productionization/model-deployment +- caption: Large Language Models + numbered: True + chapters: + - file: llm/introduction + - file: llm/basic/basic + sections: + - file: llm/basic/attention + - file: llm/basic/transformer - caption: OTHERS numbered: True maxdepth: 1 @@ -237,6 +245,7 @@ parts: - file: assignments/deep-learning/nlp/getting-start-nlp-with-classification-task - file: assignments/deep-learning/nlp/beginner-guide-to-text-preprocessing - file: assignments/deep-learning/nlp/news-topic-classification-tasks + - file: assignments/llm/basic/transformer-architecture - file: slides/introduction sections: - file: slides/python-programming/python-programming-introduction diff --git a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb new file mode 100644 index 000000000..274f60583 --- /dev/null +++ b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb @@ -0,0 +1,730 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Complete the transformer architecture" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# set up the env\n", + "\n", + "import pytest\n", + "import ipytest\n", + "import unittest\n", + "\n", + "ipytest.autoconfig()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Transformer Model\n", + "\n", + "The encoder-decoder architecture based on the Transformer structure is illustrated in figure below. The left and right sides correspond to the encoder and decoder structures, respectively. They consist of several basic Transformer blocks (represented by the gray boxes in the figure), stacked N times. Each component comprises multiple Transformer blocks, which are stacked N times.\n", + "\n", + "Here's an overview of the key components and processes involved in the semantic abstraction process from input to output:\n", + "\n", + "Encoder:\n", + "\n", + "The encoder takes an input sequence {xi}ti=1, where each xi represents the representation of a word in the text sequence.\n", + "It consists of stacked Transformer blocks. Each block includes:\n", + "Attention Layer: Utilizes multi-head attention mechanisms to capture dependencies between words in the input sequence, facilitating the modeling of long-range dependencies without traditional recurrent structures.\n", + "Position-wise Feedforward Layer: Applies complex transformations to the representations of each word in the input sequence.\n", + "Residual Connections: Directly connect the input and output of the attention and feedforward layers, aiding in efficient information flow and model optimization.\n", + "Layer Normalization: Normalizes the output representations of the attention and feedforward layers, stabilizing optimization.\n", + "Decoder:\n", + "\n", + "The decoder generates an output sequence {yi}ti=1 based on the representations learned by the encoder.\n", + "Similar to the encoder, it consists of stacked Transformer blocks, each including the same components as described above.\n", + "In addition, the decoder includes an additional attention mechanism that focuses on the encoder's output to incorporate context information during sequence generation.\n", + "Overall, the encoder-decoder architecture based on the Transformer structure allows for effective semantic abstraction by leveraging attention mechanisms, position-wise feedforward layers, residual connections, and layer normalization. This architecture enables the model to capture complex dependencies between words in the input sequence and generate meaningful outputs for various sequence-to-sequence tasks.\n", + "\n", + ":::{figure} https://media.geeksforgeeks.org/wp-content/uploads/20230531140926/Transformer-python-(1).png\n", + "---\n", + "\n", + "width: 90%\n", + "---\n", + "Transformer-based encoder and decoder Architecture\n", + ":::\n", + "\n", + "Next, we'll discuss the specific functionalities and implementation methods of each module in detail." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Embedding Layer\n", + "\n", + "The Embedding Layer in the Transformer model is responsible for converting discrete token indices into continuous vector representations. Each token index is mapped to a high-dimensional vector, which is learned during the training process. These embeddings capture semantic and syntactic information about the tokens.\n", + "\n", + "Implementation in PyTorch:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "import torch.nn as nn\n", + "import math\n", + "\n", + "class PositionalEncoder(nn.Module):\n", + " def __init__(self, d_model, max_seq_len=80):\n", + " super().__init__()\n", + " self.d_model = d_model\n", + " ## Create a constant PE matrix based on pos and i\n", + " pe = torch.zeros(max_seq_len, d_model)\n", + " for pos in range(max_seq_len):\n", + " for i in range(0, d_model, 2):\n", + " pe[pos, i] = math.sin(pos / (10000 ** ((2 * i) / d_model)))\n", + " pe[pos, i + 1] = math.cos(pos / (10000 ** ((2 * (i + 1)) / d_model)))\n", + " pe = pe.unsqueeze(0)\n", + " self.register_buffer('pe', pe)\n", + "\n", + " def forward(self, x):\n", + " ## Scale word embedding representations\n", + " x = x * math.sqrt(self.d_model)\n", + " ## Add positional constants to word embedding representations\n", + " seq_len = x.size(1)\n", + " pe = torch.autograd.Variable(self.pe[:, :seq_len], requires_grad=False).cuda()\n", + " x = x + pe\n", + " return x" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "
Check result by executing below... 📝
" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "jupyter": { + "source_hidden": true + }, + "tags": [ + "hide-input" + ] + }, + "outputs": [], + "source": [ + "%%ipytest -qq\n", + "\n", + "class TestPositionalEncoder(unittest.TestCase):\n", + " def setUp(self):\n", + " self.d_model = 512\n", + " self.max_seq_len = 10 # Maximum sequence length for testing\n", + " self.positional_encoder = PositionalEncoder(self.d_model, self.max_seq_len)\n", + "\n", + " def test_forward(self):\n", + " # Create a sample input tensor representing word embeddings\n", + " batch_size = 2\n", + " seq_length = 5\n", + " word_embeddings = torch.randn(batch_size, seq_length, self.d_model)\n", + "\n", + " # Forward pass through the PositionalEncoder module\n", + " output = self.positional_encoder(word_embeddings)\n", + "\n", + " # Check if the output shape matches the input shape\n", + " assert output.shape == (batch_size, seq_length, self.d_model)\n", + "\n", + " # Check if positional encoding is correctly applied\n", + " # Example: Verify if the first element of the first embedding vector matches the expected value\n", + " expected_first_element = torch.sin(torch.tensor([0.0])) * math.sqrt(self.d_model)\n", + " assert math.isclose(output[0, 0, 0].item(), expected_first_element.item(), rel_tol=1e-6)\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In this code:\n", + "\n", + "We define a PositionalEncoder class that inherits from nn.Module.\n", + "The constructor initializes the positional encoding matrix (pe) based on the given d_model (dimension of the model) and max_seq_len (maximum sequence length).\n", + "The forward method scales the input embeddings (x) by the square root of the model dimension and adds the positional encoding matrix (pe) to the input embeddings.\n", + "Note that we're using PyTorch's Variable and autograd to ensure that the positional encoding is compatible with the autograd mechanism for backpropagation.\n", + "Finally, the PositionalEncoder class can be used within a larger PyTorch model to incorporate positional information into word embeddings." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Attention Layer\n", + "The Attention Layer in the Transformer model enables the model to focus on different parts of the input sequence when processing each token. It computes attention scores between each pair of tokens in the input sequence and generates a context vector for each token based on the importance of other tokens. This mechanism allows the model to capture long-range dependencies in the input sequence effectively.\n", + "\n", + "Implementation in PyTorch:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "import torch.nn as nn\n", + "import torch.nn.functional as F\n", + "import math\n", + "\n", + "class MultiHeadAttention(nn.Module):\n", + " def __init__(self, heads, d_model, dropout=0.1):\n", + " super().__init__()\n", + " self.d_model = d_model\n", + " self.d_k = d_model // heads\n", + " self.h = heads\n", + " self.q_linear = nn.Linear(d_model, d_model)\n", + " self.v_linear = nn.Linear(d_model, d_model)\n", + " self.k_linear = nn.Linear(d_model, d_model)\n", + " self.dropout = nn.Dropout(dropout)\n", + " self.out = nn.Linear(d_model, d_model)\n", + " \n", + " def attention(self, q, k, v, d_k, mask=None, dropout=None):\n", + " scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(d_k)\n", + " if mask is not None:\n", + " mask = mask.unsqueeze(1)\n", + " scores = scores.masked_fill(mask == 0, -1e9)\n", + " scores = F.softmax(scores, dim=-1)\n", + " if dropout is not None:\n", + " scores = dropout(scores)\n", + " output = torch.matmul(scores, v)\n", + " return output\n", + " \n", + " def forward(self, q, k, v, mask=None):\n", + " bs = q.size(0)\n", + " k = self.k_linear(k).view(bs, -1, self.h, self.d_k)\n", + " q = self.q_linear(q).view(bs, -1, self.h, self.d_k)\n", + " v = self.v_linear(v).view(bs, -1, self.h, self.d_k)\n", + " k = k.transpose(1, 2)\n", + " q = q.transpose(1, 2)\n", + " v = v.transpose(1, 2)\n", + " scores = self.attention(q, k, v, self.d_k, mask, self.dropout)\n", + " concat = scores.transpose(1, 2).contiguous().view(bs, -1, self.d_model)\n", + " output = self.out(concat)\n", + " return output" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "
Check result by executing below... 📝
" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "jupyter": { + "source_hidden": true + }, + "tags": [ + "hide-input" + ] + }, + "outputs": [], + "source": [ + "%%ipytest -qq\n", + "\n", + "class TestMultiHeadAttention(unittest.TestCase):\n", + " def test_forward(self):\n", + " # Instantiate MultiHeadAttention module\n", + " heads = 4\n", + " d_model = 64\n", + " dropout = 0.1\n", + " multihead_attn = MultiHeadAttention(heads, d_model, dropout)\n", + "\n", + " # Create sample input tensors\n", + " batch_size = 2\n", + " seq_length = 5\n", + " q = torch.randn(batch_size, seq_length, d_model)\n", + " k = torch.randn(batch_size, seq_length, d_model)\n", + " v = torch.randn(batch_size, seq_length, d_model)\n", + " mask = torch.randint(0, 2, (batch_size, 1, seq_length)) # Example mask tensor\n", + "\n", + " # Forward pass through the MultiHeadAttention module\n", + " output = multihead_attn(q, k, v, mask)\n", + "\n", + " # Check output shape\n", + " self.assertEqual(output.shape, (batch_size, seq_length, d_model))\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In this implementation:\n", + "\n", + "The MultiHeadAttention class defines a multi-head self-attention layer.\n", + "The forward method performs linear operations to divide inputs into multiple heads, computes attention scores, and aggregates the outputs of multiple heads." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Feedforward Layer\n", + "\n", + "The Position-wise Feedforward Layer in the Transformer model applies a simple feedforward neural network independently to each position in the sequence. It consists of two linear transformations with a non-linear activation function (commonly ReLU) applied in between. This layer helps capture complex interactions between different dimensions of the input embeddings.\n", + "\n", + "Implementation in PyTorch:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "import torch.nn as nn\n", + "import torch.nn.functional as F\n", + "\n", + "class FeedForward(nn.Module):\n", + " def __init__(self, d_model, d_ff=2048, dropout=0.1):\n", + " super().__init__()\n", + " ## Set d_ff default to 2048\n", + " self.linear_1 = nn.Linear(d_model, d_ff)\n", + " self.dropout = nn.Dropout(dropout)\n", + " self.linear_2 = nn.Linear(d_ff, d_model)\n", + "\n", + " def forward(self, x):\n", + " x = self.dropout(F.relu(self.linear_1(x)))\n", + " x = self.linear_2(x)\n", + " return x" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "
Check result by executing below... 📝
" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "jupyter": { + "source_hidden": true + }, + "tags": [ + "hide-input" + ] + }, + "outputs": [], + "source": [ + "%%ipytest -qq\n", + "\n", + "class TestFeedForward(unittest.TestCase):\n", + " def test_forward(self):\n", + " # Instantiate FeedForward module\n", + " d_model = 512\n", + " d_ff = 2048\n", + " dropout = 0.1\n", + " feed_forward = FeedForward(d_model, d_ff, dropout)\n", + "\n", + " # Create sample input tensor\n", + " batch_size = 2\n", + " seq_length = 5\n", + " input_tensor = torch.randn(batch_size, seq_length, d_model)\n", + "\n", + " # Forward pass through the FeedForward module\n", + " output = feed_forward(input_tensor)\n", + "\n", + " # Check output shape\n", + " self.assertEqual(output.shape, (batch_size, seq_length, d_model))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In this implementation:\n", + "\n", + "The FeedForward class defines a feedforward layer.\n", + "The forward method applies ReLU activation to the output of the first linear transformation, followed by dropout, and then performs the second linear transformation to produce the final output." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Residual Connection and Layer Normalization\n", + "\n", + "Residual Connection:\n", + "The Residual Connection, also known as skip connection, is a technique used in deep neural networks to mitigate the vanishing gradient problem and facilitate the flow of information through the network. In the context of the Transformer model, residual connections are added around each sub-layer (such as attention and feedforward layers) before applying layer normalization. This allows the model to learn residual representations and thus ease the optimization process.\n", + "\n", + "Layer Normalization:\n", + "Layer Normalization is a technique used to stabilize the training of deep neural networks by normalizing the activations of each layer. In the Transformer model, layer normalization is applied after each sub-layer (such as attention and feedforward layers) and before the residual connection. It normalizes the activations along the feature dimension, allowing the model to learn more robust representations and accelerate convergence during training.\n", + "\n", + "Implementation in PyTorch:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "import torch.nn as nn\n", + "\n", + "class NormLayer(nn.Module):\n", + " def __init__(self, d_model, eps=1e-6):\n", + " super().__init__()\n", + " self.size = d_model\n", + " ## Layer normalization includes two learnable parameters\n", + " self.alpha = nn.Parameter(torch.ones(self.size))\n", + " self.bias = nn.Parameter(torch.zeros(self.size))\n", + " self.eps = eps\n", + " \n", + " def forward(self, x):\n", + " norm = self.alpha * (x - x.mean(dim=-1, keepdim=True)) / (x.std(dim=-1, keepdim=True) + self.eps) + self.bias\n", + " return norm" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "
Check result by executing below... 📝
" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "jupyter": { + "source_hidden": true + }, + "tags": [ + "hide-input" + ] + }, + "outputs": [], + "source": [ + "%%ipytest -qq\n", + "\n", + "class TestNormLayer(unittest.TestCase):\n", + " def test_forward(self):\n", + " # Instantiate NormLayer module\n", + " d_model = 512\n", + " eps = 1e-6\n", + " norm_layer = NormLayer(d_model, eps)\n", + "\n", + " # Create sample input tensor\n", + " batch_size = 2\n", + " seq_length = 5\n", + " input_tensor = torch.randn(batch_size, seq_length, d_model)\n", + "\n", + " # Forward pass through the NormLayer module\n", + " output = norm_layer(input_tensor)\n", + "\n", + " # Check output shape\n", + " self.assertEqual(output.shape, (batch_size, seq_length, d_model))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In this implementation:\n", + "\n", + "The NormLayer class defines a layer normalization layer.\n", + "The forward method computes the layer normalization using the given input tensor x." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Encoder and Decoder Structure\n", + "Encoder Structure:\n", + "The Encoder in the Transformer model consists of multiple stacked Encoder layers. Each Encoder layer typically contains a Multi-Head Attention sub-layer followed by a FeedForward sub-layer, each with Residual Connection and Layer Normalization.\n", + "\n", + "Decoder Structure:\n", + "Similarly, the Decoder in the Transformer model also consists of multiple stacked Decoder layers. Each Decoder layer contains three sub-layers:\n", + "\n", + "Masked Multi-Head Attention sub-layer to attend to previous tokens in the output sequence.\n", + "Multi-Head Attention sub-layer that attends to the encoder's output.\n", + "FeedForward sub-layer. Again, each sub-layer is followed by Residual Connection and Layer Normalization.\n", + "\n", + "Below are the Python implementations for the Encoder and Decoder structures:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "class EncoderLayer(nn.Module):\n", + " def __init__(self, d_model, heads, dropout=0.1):\n", + " super().__init__()\n", + " self.norm_1 = Norm(d_model)\n", + " self.norm_2 = Norm(d_model)\n", + " self.attn = MultiHeadAttention(heads, d_model, dropout=dropout)\n", + " self.ff = FeedForward(d_model, dropout=dropout)\n", + " self.dropout_1 = nn.Dropout(dropout)\n", + " self.dropout_2 = nn.Dropout(dropout)\n", + "\n", + " def forward(self, x, mask):\n", + " x2 = self.norm_1(x)\n", + " x = x + self.dropout_1(self.attn(x2, x2, x2, mask))\n", + " x2 = self.norm_2(x)\n", + " x = x + self.dropout_2(self.ff(x2))\n", + " return x\n", + "\n", + "class Encoder(nn.Module):\n", + " def __init__(self, vocab_size, d_model, N, heads):\n", + " super().__init__()\n", + " self.N = N\n", + " self.embed = Embedder(vocab_size, d_model)\n", + " self.pe = PositionalEncoder(d_model)\n", + " self.layers = get_clones(EncoderLayer(d_model, heads), N)\n", + " self.norm = Norm(d_model)\n", + "\n", + " def forward(self, src, mask):\n", + " x = self.embed(src)\n", + " x = self.pe(x)\n", + " for i in range(self.N):\n", + " x = self.layers[i](x, mask)\n", + " return self.norm(x)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "class DecoderLayer(nn.Module):\n", + " def __init__(self, d_model, heads, dropout=0.1):\n", + " super().__init__()\n", + " self.norm_1 = Norm(d_model)\n", + " self.norm_2 = Norm(d_model)\n", + " self.norm_3 = Norm(d_model)\n", + " self.dropout_1 = nn.Dropout(dropout)\n", + " self.dropout_2 = nn.Dropout(dropout)\n", + " self.dropout_3 = nn.Dropout(dropout)\n", + " self.attn_1 = MultiHeadAttention(heads, d_model, dropout=dropout)\n", + " self.attn_2 = MultiHeadAttention(heads, d_model, dropout=dropout)\n", + " self.ff = FeedForward(d_model, dropout=dropout)\n", + "\n", + " def forward(self, x, e_outputs, src_mask, trg_mask):\n", + " x2 = self.norm_1(x)\n", + " x = x + self.dropout_1(self.attn_1(x2, x2, x2, trg_mask))\n", + " x2 = self.norm_2(x)\n", + " x = x + self.dropout_2(self.attn_2(x2, e_outputs, e_outputs, src_mask))\n", + " x2 = self.norm_3(x)\n", + " x = x + self.dropout_3(self.ff(x2))\n", + " return x\n", + "\n", + "class Decoder(nn.Module):\n", + " def __init__(self, vocab_size, d_model, N, heads, dropout):\n", + " super().__init__()\n", + " self.N = N\n", + " self.embed = Embedder(vocab_size, d_model)\n", + " self.pe = PositionalEncoder(d_model, dropout=dropout)\n", + " self.layers = get_clones(DecoderLayer(d_model, heads, dropout), N)\n", + " self.norm = Norm(d_model)\n", + "\n", + " def forward(self, trg, e_outputs, src_mask, trg_mask):\n", + " x = self.embed(trg)\n", + " x = self.pe(x)\n", + " for i in range(self.N):\n", + " x = self.layers[i](x, e_outputs, src_mask, trg_mask)\n", + " return self.norm(x)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In these implementations:\n", + "\n", + "The EncoderLayer and DecoderLayer classes define encoder and decoder layers, respectively.\n", + "The Encoder and Decoder classes define encoder and decoder modules, respectively, composed of multiple layers of encoder or decoder layers.\n", + "These classes follow the architecture described in the text, including the use of multi-head attention, feedforward layers, residual connections, and layer normalization." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The overall implementation of the Transformer encoder and decoder structure:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "import torch.nn as nn\n", + "import torch.nn.functional as F\n", + "import time\n", + "import numpy as np\n", + "\n", + "class Transformer(nn.Module):\n", + " def __init__(self, src_vocab, trg_vocab, d_model, N, heads, dropout):\n", + " super().__init__()\n", + " self.encoder = Encoder(src_vocab, d_model, N, heads, dropout)\n", + " self.decoder = Decoder(trg_vocab, d_model, N, heads, dropout)\n", + " self.out = nn.Linear(d_model, trg_vocab)\n", + "\n", + " def forward(self, src, trg, src_mask, trg_mask):\n", + " e_outputs = self.encoder(src, src_mask)\n", + " d_output = self.decoder(trg, e_outputs, src_mask, trg_mask)\n", + " output = self.out(d_output)\n", + " return output\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The training process for the Transformer model:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "## Model parameters\n", + "d_model = 512\n", + "heads = 8\n", + "N = 6\n", + "src_vocab = len(EN_TEXT.vocab)\n", + "trg_vocab = len(FR_TEXT.vocab)\n", + "\n", + "## Initialize the model\n", + "model = Transformer(src_vocab, trg_vocab, d_model, N, heads)\n", + "\n", + "## Initialize optimizer\n", + "optim = torch.optim.Adam(model.parameters(), lr=0.0001, betas=(0.9, 0.98), eps=1e-9)\n", + "\n", + "## Training function\n", + "def train_model(epochs, print_every=100):\n", + " model.train()\n", + " start = time.time()\n", + " temp = start\n", + " total_loss = 0\n", + "\n", + " for epoch in range(epochs):\n", + " for i, batch in enumerate(train_iter):\n", + " src = batch.English.transpose(0, 1)\n", + " trg = batch.French.transpose(0, 1)\n", + " trg_input = trg[:, :-1]\n", + " targets = trg[:, 1:].contiguous().view(-1)\n", + " src_mask, trg_mask = create_masks(src, trg_input)\n", + "\n", + " preds = model(src, trg_input, src_mask, trg_mask)\n", + " optim.zero_grad()\n", + " loss = F.cross_entropy(preds.view(-1, preds.size(-1)), targets, ignore_index=target_pad)\n", + " loss.backward()\n", + " optim.step()\n", + " total_loss += loss.data[0]\n", + "\n", + " if (i + 1) % print_every == 0:\n", + " loss_avg = total_loss / print_every\n", + " print(\"time = %dm, epoch %d, iter = %d, loss = %.3f, %ds per %d iters\" % (\n", + " (time.time() - start) // 60, epoch + 1, i + 1, loss_avg, time.time() - temp, print_every))\n", + " total_loss = 0\n", + " temp = time.time()\n", + "\n", + "## Train the model\n", + "train_model(epochs=10)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Test the trained model:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "def translate(model, src, max_len=80, custom_string=False):\n", + " model.eval()\n", + " if custom_string == True:\n", + " src = tokenize_en(src)\n", + " sentence = Variable(torch.LongTensor([[EN_TEXT.vocab.stoi[tok] for tok in sentence]])).cuda()\n", + " src_mask = (src != input_pad).unsqueeze(-2)\n", + " e_outputs = model.encoder(src, src_mask)\n", + " outputs = torch.zeros(max_len).type_as(src.data)\n", + " outputs[0] = torch.LongTensor([FR_TEXT.vocab.stoi['']])\n", + "\n", + " for i in range(1, max_len):\n", + " trg_mask = np.triu(np.ones((1, i, i), k=1).astype('uint8'))\n", + " trg_mask = Variable(torch.from_numpy(trg_mask) == 0).cuda()\n", + " out = model.out(model.decoder(outputs[:i].unsqueeze(0), e_outputs, src_mask, trg_mask))\n", + "\n", + " out = F.softmax(out, dim=-1)\n", + " val, ix = out[:, -1].data.topk(1)\n", + " outputs[i] = ix[0][0]\n", + "\n", + " if ix[0][0] == FR_TEXT.vocab.stoi['']:\n", + " break\n", + "\n", + " return ' '.join([FR_TEXT.vocab.itos[ix] for ix in outputs[:i]])\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Acknowledgments\n", + "\n", + "Thanks to the awesome open source project for Transformer learning, which inspire this chapter.\n", + "\n", + "- [chatgpt](https://openai.com/product/chatgpt)" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "open-machine-learning-jupyter-book", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.9.18" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} diff --git a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb new file mode 100644 index 000000000..2c4b9c098 --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb @@ -0,0 +1,367 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": { + "tags": [ + "remove-cell" + ] + }, + "source": [ + "---\n", + "license:\n", + " code: MIT\n", + " content: CC-BY-4.0\n", + "github: https://github.com/ocademy-ai/machine-learning\n", + "venue: By Ocademy\n", + "open_access: true\n", + "bibliography:\n", + " - https://raw.githubusercontent.com/ocademy-ai/machine-learning/main/open-machine-learning-jupyter-book/references.bib\n", + "---" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Attention\n", + "## What is Attention?\n", + "\n", + "The attention mechanism describes a recent new group of layers in neural networks that has attracted a lot of interest in the past few years, especially in sequence tasks. There are a lot of different possible definitions of \"attention\" in the literature, but the one we will use here is the following: _the attention mechanism describes a weighted average of (sequence) elements with the weights dynamically computed based on an input query and elements' keys_. So what does this exactly mean? The goal is to take an average over the features of multiple elements. However, instead of weighting each element equally, we want to weight them depending on their actual values. In other words, we want to dynamically decide on which inputs we want to \"attend\" more than others. In particular, an attention mechanism has usually four parts we need to specify:\n", + "\n", + "* **Query**: The query is a feature vector that describes what we are looking for in the sequence, i.e. what would we maybe want to pay attention to.\n", + "* **Keys**: For each input element, we have a key which is again a feature vector. This feature vector roughly describes what the element is \"offering\", or when it might be important. The keys should be designed such that we can identify the elements we want to pay attention to based on the query.\n", + "* **Values**: For each input element, we also have a value vector. This feature vector is the one we want to average over.\n", + "* **Score function**: To rate which elements we want to pay attention to, we need to specify a score function $f_{attn}$. The score function takes the query and a key as input, and output the score/attention weight of the query-key pair. It is usually implemented by simple similarity metrics like a dot product, or a small MLP.\n", + "\n", + "\n", + "The weights of the average are calculated by a softmax over all score function outputs. Hence, we assign those value vectors a higher weight whose corresponding key is most similar to the query. If we try to describe it with pseudo-math, we can write: \n", + "\n", + "$$\n", + "\\alpha_i = \\frac{\\exp\\left(f_{attn}\\left(\\text{key}_i, \\text{query}\\right)\\right)}{\\sum_j \\exp\\left(f_{attn}\\left(\\text{key}_j, \\text{query}\\right)\\right)}, \\hspace{5mm} \\text{out} = \\sum_i \\alpha_i \\cdot \\text{value}_i\n", + "$$\n", + "\n", + "Visually, we can show the attention over a sequence of words as follows:\n", + "\n", + ":::{figure} ../image/attention_example.svg\n", + ":::\n", + "\n", + "For every word, we have one key and one value vector. The query is compared to all keys with a score function (in this case the dot product) to determine the weights. The softmax is not visualized for simplicity. Finally, the value vectors of all words are averaged using the attention weights.\n", + "\n", + "Most attention mechanisms differ in terms of what queries they use, how the key and value vectors are defined, and what score function is used. The attention applied inside the Transformer architecture is called **self-attention**. In self-attention, each sequence element provides a key, value, and query. For each element, we perform an attention layer where based on its query, we check the similarity of the all sequence elements' keys, and returned a different, averaged value vector for each element. We will now go into a bit more detail by first looking at the specific implementation of the attention mechanism which is in the Transformer case the scaled dot product attention." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Scaled Dot Product Attention\n", + "\n", + "The core concept behind self-attention is the scaled dot product attention. Our goal is to have an attention mechanism with which any element in a sequence can attend to any other while still being efficient to compute. The dot product attention takes as input a set of queries $Q\\in\\mathbb{R}^{T\\times d_k}$, keys $K\\in\\mathbb{R}^{T\\times d_k}$ and values $V\\in\\mathbb{R}^{T\\times d_v}$ where $T$ is the sequence length, and $d_k$ and $d_v$ are the hidden dimensionality for queries/keys and values respectively. For simplicity, we neglect the batch dimension for now. The attention value from element $i$ to $j$ is based on its similarity of the query $Q_i$ and key $K_j$, using the dot product as the similarity metric. In math, we calculate the dot product attention as follows:\n", + "\n", + "$$\\text{Attention}(Q,K,V)=\\text{softmax}\\left(\\frac{QK^T}{\\sqrt{d_k}}\\right)V$$\n", + "\n", + "The matrix multiplication $QK^T$ performs the dot product for every possible pair of queries and keys, resulting in a matrix of the shape $T\\times T$. Each row represents the attention logits for a specific element $i$ to all other elements in the sequence. On these, we apply a softmax and multiply with the value vector to obtain a weighted mean (the weights being determined by the attention). Another perspective on this attention mechanism offers the computation graph which is visualized below (figure credit - [Vaswani et al., 2017](https://arxiv.org/abs/1706.03762)).\n", + "\n", + ":::{figure} ../image/scaled_dot_product_attn.svg\n", + ":::\n", + "\n", + "One aspect we haven't discussed yet is the scaling factor of $1/\\sqrt{d_k}$. This scaling factor is crucial to maintain an appropriate variance of attention values after initialization. Remember that we intialize our layers with the intention of having equal variance throughout the model, and hence, $Q$ and $K$ might also have a variance close to $1$. However, performing a dot product over two vectors with a variance $\\sigma^2$ results in a scalar having $d_k$-times higher variance: \n", + "\n", + "$$q_i \\sim \\mathcal{N}(0,\\sigma^2), k_i \\sim \\mathcal{N}(0,\\sigma^2) \\to \\text{Var}\\left(\\sum_{i=1}^{d_k} q_i\\cdot k_i\\right) = \\sigma^4\\cdot d_k$$\n", + "\n", + "\n", + "If we do not scale down the variance back to $\\sim\\sigma^2$, the softmax over the logits will already saturate to $1$ for one random element and $0$ for all others. The gradients through the softmax will be close to zero so that we can't learn the parameters appropriately. Note that the extra factor of $\\sigma^2$, i.e., having $\\sigma^4$ instead of $\\sigma^2$, is usually not an issue, since we keep the original variance $\\sigma^2$ close to $1$ anyways.\n", + "\n", + "The block `Mask (opt.)` in the diagram above represents the optional masking of specific entries in the attention matrix. This is for instance used if we stack multiple sequences with different lengths into a batch. To still benefit from parallelization in PyTorch, we pad the sentences to the same length and mask out the padding tokens during the calculation of the attention values. This is usually done by setting the respective attention logits to a very low value. \n", + "\n", + "After we have discussed the details of the scaled dot product attention block, we can write a function below which computes the output features given the triple of queries, keys, and values:" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Below, we import the standard libraries." + ] + }, + { + "cell_type": "code", + "execution_count": 16, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Device: cpu\n" + ] + } + ], + "source": [ + "## Standard libraries\n", + "import os\n", + "import numpy as np\n", + "import random\n", + "import math\n", + "import json\n", + "from functools import partial\n", + "\n", + "## Imports for plotting\n", + "import matplotlib.pyplot as plt\n", + "plt.set_cmap('cividis')\n", + "%matplotlib inline\n", + "from matplotlib.colors import to_rgb\n", + "import matplotlib\n", + "matplotlib.rcParams['lines.linewidth'] = 2.0\n", + "import seaborn as sns\n", + "sns.reset_orig()\n", + "\n", + "## tqdm for loading bars\n", + "from tqdm.notebook import tqdm\n", + "\n", + "## PyTorch\n", + "import torch\n", + "import torch.nn as nn\n", + "import torch.nn.functional as F\n", + "import torch.utils.data as data\n", + "import torch.optim as optim\n", + "\n", + "## Torchvision\n", + "import torchvision\n", + "from torchvision.datasets import CIFAR100\n", + "from torchvision import transforms\n", + "\n", + "# PyTorch Lightning\n", + "try:\n", + " import pytorch_lightning as pl\n", + "except ModuleNotFoundError: # Google Colab does not have PyTorch Lightning installed by default. Hence, we do it here if necessary\n", + " !pip install --quiet pytorch-lightning>=1.4\n", + " import pytorch_lightning as pl\n", + "from pytorch_lightning.callbacks import LearningRateMonitor, ModelCheckpoint\n", + "\n", + "# Ensure that all operations are deterministic on GPU (if used) for reproducibility\n", + "torch.backends.cudnn.deterministic = True\n", + "torch.backends.cudnn.benchmark = False\n", + "\n", + "device = torch.device(\"cuda:0\") if torch.cuda.is_available() else torch.device(\"cpu\")\n", + "print(\"Device:\", device)" + ] + }, + { + "cell_type": "code", + "execution_count": 17, + "metadata": {}, + "outputs": [], + "source": [ + "def scaled_dot_product(q, k, v, mask=None):\n", + " d_k = q.size()[-1]\n", + " attn_logits = torch.matmul(q, k.transpose(-2, -1))\n", + " attn_logits = attn_logits / math.sqrt(d_k)\n", + " if mask is not None:\n", + " attn_logits = attn_logits.masked_fill(mask == 0, -9e15)\n", + " attention = F.softmax(attn_logits, dim=-1)\n", + " values = torch.matmul(attention, v)\n", + " return values, attention" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Note that our code above supports any additional dimensionality in front of the sequence length so that we can also use it for batches. However, for a better understanding, let's generate a few random queries, keys, and value vectors, and calculate the attention outputs:" + ] + }, + { + "cell_type": "code", + "execution_count": 18, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Seed set to 42\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Q\n", + " tensor([[ 0.3367, 0.1288],\n", + " [ 0.2345, 0.2303],\n", + " [-1.1229, -0.1863]])\n", + "K\n", + " tensor([[ 2.2082, -0.6380],\n", + " [ 0.4617, 0.2674],\n", + " [ 0.5349, 0.8094]])\n", + "V\n", + " tensor([[ 1.1103, -1.6898],\n", + " [-0.9890, 0.9580],\n", + " [ 1.3221, 0.8172]])\n", + "Values\n", + " tensor([[ 0.5698, -0.1520],\n", + " [ 0.5379, -0.0265],\n", + " [ 0.2246, 0.5556]])\n", + "Attention\n", + " tensor([[0.4028, 0.2886, 0.3086],\n", + " [0.3538, 0.3069, 0.3393],\n", + " [0.1303, 0.4630, 0.4067]])\n" + ] + } + ], + "source": [ + "seq_len, d_k = 3, 2\n", + "pl.seed_everything(42)\n", + "q = torch.randn(seq_len, d_k)\n", + "k = torch.randn(seq_len, d_k)\n", + "v = torch.randn(seq_len, d_k)\n", + "values, attention = scaled_dot_product(q, k, v)\n", + "print(\"Q\\n\", q)\n", + "print(\"K\\n\", k)\n", + "print(\"V\\n\", v)\n", + "print(\"Values\\n\", values)\n", + "print(\"Attention\\n\", attention)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Before continuing, make sure you can follow the calculation of the specific values here, and also check it by hand. It is important to fully understand how the scaled dot product attention is calculated." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Multi-Head Attention\n", + "\n", + "The scaled dot product attention allows a network to attend over a sequence. However, often there are multiple different aspects a sequence element wants to attend to, and a single weighted average is not a good option for it. This is why we extend the attention mechanisms to multiple heads, i.e. multiple different query-key-value triplets on the same features. Specifically, given a query, key, and value matrix, we transform those into $h$ sub-queries, sub-keys, and sub-values, which we pass through the scaled dot product attention independently. Afterward, we concatenate the heads and combine them with a final weight matrix. Mathematically, we can express this operation as:\n", + "\n", + "$$\n", + "\\begin{split}\n", + " \\text{Multihead}(Q,K,V) & = \\text{Concat}(\\text{head}_1,...,\\text{head}_h)W^{O}\\\\\n", + " \\text{where } \\text{head}_i & = \\text{Attention}(QW_i^Q,KW_i^K, VW_i^V)\n", + "\\end{split}\n", + "$$\n", + "\n", + "We refer to this as Multi-Head Attention layer with the learnable parameters $W_{1...h}^{Q}\\in\\mathbb{R}^{D\\times d_k}$, $W_{1...h}^{K}\\in\\mathbb{R}^{D\\times d_k}$, $W_{1...h}^{V}\\in\\mathbb{R}^{D\\times d_v}$, and $W^{O}\\in\\mathbb{R}^{h\\cdot d_v\\times d_{out}}$ ($D$ being the input dimensionality). Expressed in a computational graph, we can visualize it as below (figure credit - [Vaswani et al., 2017](https://arxiv.org/abs/1706.03762)).\n", + "\n", + ":::{figure} ../image/multihead_attention.svg\n", + ":::\n", + "\n", + "How are we applying a Multi-Head Attention layer in a neural network, where we don't have an arbitrary query, key, and value vector as input? Looking at the computation graph above, a simple but effective implementation is to set the current feature map in a NN, $X\\in\\mathbb{R}^{B\\times T\\times d_{\\text{model}}}$, as $Q$, $K$ and $V$ ($B$ being the batch size, $T$ the sequence length, $d_{\\text{model}}$ the hidden dimensionality of $X$). The consecutive weight matrices $W^{Q}$, $W^{K}$, and $W^{V}$ can transform $X$ to the corresponding feature vectors that represent the queries, keys, and values of the input. Using this approach, we can implement the Multi-Head Attention module below." + ] + }, + { + "cell_type": "code", + "execution_count": 19, + "metadata": {}, + "outputs": [], + "source": [ + "# Helper function to support different mask shapes.\n", + "# Output shape supports (batch_size, number of heads, seq length, seq length)\n", + "# If 2D: broadcasted over batch size and number of heads\n", + "# If 3D: broadcasted over number of heads\n", + "# If 4D: leave as is\n", + "def expand_mask(mask):\n", + " assert mask.ndim >= 2, \"Mask must be at least 2-dimensional with seq_length x seq_length\"\n", + " if mask.ndim == 3:\n", + " mask = mask.unsqueeze(1)\n", + " while mask.ndim < 4:\n", + " mask = mask.unsqueeze(0)\n", + " return mask" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "metadata": {}, + "outputs": [], + "source": [ + "class MultiheadAttention(nn.Module):\n", + " \n", + " def __init__(self, input_dim, embed_dim, num_heads):\n", + " super().__init__()\n", + " assert embed_dim % num_heads == 0, \"Embedding dimension must be 0 modulo number of heads.\"\n", + " \n", + " self.embed_dim = embed_dim\n", + " self.num_heads = num_heads\n", + " self.head_dim = embed_dim // num_heads\n", + " \n", + " # Stack all weight matrices 1...h together for efficiency\n", + " # Note that in many implementations you see \"bias=False\" which is optional\n", + " self.qkv_proj = nn.Linear(input_dim, 3*embed_dim)\n", + " self.o_proj = nn.Linear(embed_dim, embed_dim)\n", + " \n", + " self._reset_parameters()\n", + "\n", + " def _reset_parameters(self):\n", + " # Original Transformer initialization, see PyTorch documentation\n", + " nn.init.xavier_uniform_(self.qkv_proj.weight)\n", + " self.qkv_proj.bias.data.fill_(0)\n", + " nn.init.xavier_uniform_(self.o_proj.weight)\n", + " self.o_proj.bias.data.fill_(0)\n", + "\n", + " def forward(self, x, mask=None, return_attention=False):\n", + " batch_size, seq_length, _ = x.size()\n", + " if mask is not None:\n", + " mask = expand_mask(mask)\n", + " qkv = self.qkv_proj(x)\n", + " \n", + " # Separate Q, K, V from linear output\n", + " qkv = qkv.reshape(batch_size, seq_length, self.num_heads, 3*self.head_dim)\n", + " qkv = qkv.permute(0, 2, 1, 3) # [Batch, Head, SeqLen, Dims]\n", + " q, k, v = qkv.chunk(3, dim=-1)\n", + " \n", + " # Determine value outputs\n", + " values, attention = scaled_dot_product(q, k, v, mask=mask)\n", + " values = values.permute(0, 2, 1, 3) # [Batch, SeqLen, Head, Dims]\n", + " values = values.reshape(batch_size, seq_length, self.embed_dim)\n", + " o = self.o_proj(values)\n", + " \n", + " if return_attention:\n", + " return o, attention\n", + " else:\n", + " return o" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "One crucial characteristic of the multi-head attention is that it is permutation-equivariant with respect to its inputs. This means that if we switch two input elements in the sequence, e.g. $X_1\\leftrightarrow X_2$ (neglecting the batch dimension for now), the output is exactly the same besides the elements 1 and 2 switched. Hence, the multi-head attention is actually looking at the input not as a sequence, but as a set of elements. This property makes the multi-head attention block and the Transformer architecture so powerful and widely applicable! But what if the order of the input is actually important for solving the task, like language modeling? The answer is to encode the position in the input features, which we will take a closer look at later (topic _Positional encodings_ below).\n", + "\n", + "Before moving on to creating the Transformer architecture, we can compare the self-attention operation with our other common layer competitors for sequence data: convolutions and recurrent neural networks. Below you can find a table by [Vaswani et al. (2017)](https://arxiv.org/abs/1706.03762) on the complexity per layer, the number of sequential operations, and maximum path length. The complexity is measured by the upper bound of the number of operations to perform, while the maximum path length represents the maximum number of steps a forward or backward signal has to traverse to reach any other position. The lower this length, the better gradient signals can backpropagate for long-range dependencies. Let's take a look at the table below:\n", + "\n", + ":::{figure} ../image/comparison_conv_rnn.svg\n", + ":::\n", + "\n", + "$n$ is the sequence length, $d$ is the representation dimension and $k$ is the kernel size of convolutions. In contrast to recurrent networks, the self-attention layer can parallelize all its operations making it much faster to execute for smaller sequence lengths. However, when the sequence length exceeds the hidden dimensionality, self-attention becomes more expensive than RNNs. One way of reducing the computational cost for long sequences is by restricting the self-attention to a neighborhood of inputs to attend over, denoted by $r$. Nevertheless, there has been recently a lot of work on more efficient Transformer architectures that still allow long dependencies, of which you can find an overview in the paper by [Tay et al. (2020)](https://arxiv.org/abs/2009.06732) if interested." + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3 (ipykernel)", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.10.4" + } + }, + "nbformat": 4, + "nbformat_minor": 4 +} diff --git a/open-machine-learning-jupyter-book/llm/basic/basic.ipynb b/open-machine-learning-jupyter-book/llm/basic/basic.ipynb new file mode 100644 index 000000000..ef8e960b7 --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/basic/basic.ipynb @@ -0,0 +1,64 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": { + "tags": [ + "remove-cell" + ] + }, + "source": [ + "---\n", + "license:\n", + " code: MIT\n", + " content: CC-BY-4.0\n", + "github: https://github.com/ocademy-ai/machine-learning\n", + "venue: By Ocademy\n", + "open_access: true\n", + "bibliography:\n", + " - https://raw.githubusercontent.com/ocademy-ai/machine-learning/main/open-machine-learning-jupyter-book/references.bib\n", + "---" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Large Language Models Basic\n", + "In these sections, we will explore the attention mechanism, which allows models to focus on specific parts of the input during processing. We will study the Transformer model architecture, which serves as the cornerstone for many state-of-the-art language models, and how it has fundamentally transformed the field of Natural Language Processing (NLP). Additionally, we will introduce generative pre-trained language models like GPT, delve into the network structures of large language models, optimization techniques for attention mechanisms, and practical applications stemming from these foundations.\n", + "\n", + ":::{figure} ../image/llm.png\n", + ":::" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + ":::{tableofcontents}\n", + ":::" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "open-machine-learning-jupyter-book", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.9.18" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} diff --git a/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb b/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb new file mode 100644 index 000000000..4143ec705 --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb @@ -0,0 +1,20020 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": { + "tags": [ + "remove-cell" + ] + }, + "source": [ + "---\n", + "license:\n", + " code: MIT\n", + " content: CC-BY-4.0\n", + "github: https://github.com/ocademy-ai/machine-learning\n", + "venue: By Ocademy\n", + "open_access: true\n", + "bibliography:\n", + " - https://raw.githubusercontent.com/ocademy-ai/machine-learning/main/open-machine-learning-jupyter-book/references.bib\n", + "---" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Transformer\n", + "In this tutorial, we will discuss one of the most impactful architectures of the last 2 years: the Transformer model. Since the paper [Attention Is All You Need](https://arxiv.org/abs/1706.03762) by Vaswani et al. had been published in 2017, the Transformer architecture has continued to beat benchmarks in many domains, most importantly in Natural Language Processing. Transformers with an incredible amount of parameters can generate long, convincing [essays](https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3), and opened up new application fields of AI. As the hype of the Transformer architecture seems not to come to an end in the next years, it is important to understand how it works, and have implemented it yourself, which we will do in this notebook. We focus here on what makes the Transformer and self-attention so powerful in general.\n", + "\n", + "Below, we import the standard libraries." + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Device: cuda:0\n" + ] + } + ], + "source": [ + "## Standard libraries\n", + "import os\n", + "import numpy as np \n", + "import random\n", + "import math\n", + "import json\n", + "from functools import partial\n", + "\n", + "## Imports for plotting\n", + "import matplotlib.pyplot as plt\n", + "plt.set_cmap('cividis')\n", + "%matplotlib inline \n", + "from matplotlib.colors import to_rgb\n", + "import matplotlib\n", + "matplotlib.rcParams['lines.linewidth'] = 2.0\n", + "import seaborn as sns\n", + "sns.reset_orig()\n", + "\n", + "## tqdm for loading bars\n", + "from tqdm.notebook import tqdm\n", + "\n", + "## PyTorch\n", + "import torch\n", + "import torch.nn as nn\n", + "import torch.nn.functional as F\n", + "import torch.utils.data as data\n", + "import torch.optim as optim\n", + "\n", + "## Torchvision\n", + "import torchvision\n", + "from torchvision.datasets import CIFAR100\n", + "from torchvision import transforms\n", + "\n", + "# PyTorch Lightning\n", + "try:\n", + " import pytorch_lightning as pl\n", + "except ModuleNotFoundError: # Google Colab does not have PyTorch Lightning installed by default. Hence, we do it here if necessary\n", + " !pip install --quiet pytorch-lightning>=1.4\n", + " import pytorch_lightning as pl\n", + "from pytorch_lightning.callbacks import LearningRateMonitor, ModelCheckpoint\n", + "\n", + "# Path to the folder where the datasets are/should be downloaded (e.g. CIFAR10)\n", + "DATASET_PATH = \"./data\"\n", + "# Path to the folder where the pretrained models are saved\n", + "CHECKPOINT_PATH = \"./saved_models\"\n", + "\n", + "# Setting the seed\n", + "pl.seed_everything(42)\n", + "\n", + "# Ensure that all operations are deterministic on GPU (if used) for reproducibility\n", + "torch.backends.cudnn.deterministic = True\n", + "torch.backends.cudnn.benchmark = False\n", + "\n", + "device = torch.device(\"cuda:0\") if torch.cuda.is_available() else torch.device(\"cpu\")\n", + "print(\"Device:\", device)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Two pre-trained models are downloaded below. Make sure to have adjusted your `CHECKPOINT_PATH` before running this code if not already done." + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [], + "source": [ + "import urllib.request\n", + "from urllib.error import HTTPError\n", + "# Github URL where saved models are stored for this tutorial\n", + "base_url = \"https://raw.githubusercontent.com/phlippe/saved_models/main/tutorial6/\"\n", + "# Files to download\n", + "pretrained_files = [\"ReverseTask.ckpt\", \"SetAnomalyTask.ckpt\"]\n", + "\n", + "# Create checkpoint path if it doesn't exist yet\n", + "os.makedirs(CHECKPOINT_PATH, exist_ok=True)\n", + "\n", + "# For each file, check whether it already exists. If not, try downloading it.\n", + "for file_name in pretrained_files:\n", + " file_path = os.path.join(CHECKPOINT_PATH, file_name)\n", + " if \"/\" in file_name:\n", + " os.makedirs(file_path.rsplit(\"/\",1)[0], exist_ok=True)\n", + " if not os.path.isfile(file_path):\n", + " file_url = base_url + file_name\n", + " print(f\"Downloading {file_url}...\")\n", + " try:\n", + " urllib.request.urlretrieve(file_url, file_path)\n", + " except HTTPError as e:\n", + " print(\"Something went wrong. Please try to download the file from the GDrive folder, or contact the author with the full output including the following error:\\n\", e)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## The Transformer architecture\n", + "\n", + "In the first part of this notebook, we will implement the Transformer architecture by hand. As the architecture is so popular, there already exists a Pytorch module `nn.Transformer` ([documentation](https://pytorch.org/docs/stable/generated/torch.nn.Transformer.html)) and a [tutorial](https://pytorch.org/tutorials/beginner/transformer_tutorial.html) on how to use it for next token prediction. However, we will implement it here ourselves, to get through to the smallest details." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Transformer Encoder\n", + "\n", + "Next, we will look at how to apply the multi-head attention block inside the Transformer architecture. Originally, the Transformer model was designed for machine translation. Hence, it got an encoder-decoder structure where the encoder takes as input the sentence in the original language and generates an attention-based representation. On the other hand, the decoder attends over the encoded information and generates the translated sentence in an autoregressive manner, as in a standard RNN. While this structure is extremely useful for Sequence-to-Sequence tasks with the necessity of autoregressive decoding, we will focus here on the encoder part. Many advances in NLP have been made using pure encoder-based Transformer models (if interested, models include the [BERT](https://arxiv.org/abs/1810.04805)-family, the [Vision Transformer](https://arxiv.org/abs/2010.11929), and more), and in our tutorial, we will also mainly focus on the encoder part. If you have understood the encoder architecture, the decoder is a very small step to implement as well. The full Transformer architecture looks as follows (figure credit - [Vaswani et al., 2017](https://arxiv.org/abs/1706.03762)).:\n", + "\n", + ":::{figure} ../image/transformer_architecture.svg\n", + ":::\n", + "\n", + "The encoder consists of $N$ identical blocks that are applied in sequence. Taking as input $x$, it is first passed through a Multi-Head Attention block as we have implemented above. The output is added to the original input using a residual connection, and we apply a consecutive Layer Normalization on the sum. Overall, it calculates $\\text{LayerNorm}(x+\\text{Multihead}(x,x,x))$ ($x$ being $Q$, $K$ and $V$ input to the attention layer). The residual connection is crucial in the Transformer architecture for two reasons: \n", + "\n", + "1. Similar to ResNets, Transformers are designed to be very deep. Some models contain more than 24 blocks in the encoder. Hence, the residual connections are crucial for enabling a smooth gradient flow through the model.\n", + "2. Without the residual connection, the information about the original sequence is lost. Remember that the Multi-Head Attention layer ignores the position of elements in a sequence, and can only learn it based on the input features. Removing the residual connections would mean that this information is lost after the first attention layer (after initialization), and with a randomly initialized query and key vector, the output vectors for position $i$ has no relation to its original input. All outputs of the attention are likely to represent similar/same information, and there is no chance for the model to distinguish which information came from which input element. An alternative option to residual connection would be to fix at least one head to focus on its original input, but this is very inefficient and does not have the benefit of the improved gradient flow.\n", + "\n", + "The Layer Normalization also plays an important role in the Transformer architecture as it enables faster training and provides small regularization. Additionally, it ensures that the features are in a similar magnitude among the elements in the sequence. We are not using Batch Normalization because it depends on the batch size which is often small with Transformers (they require a lot of GPU memory), and BatchNorm has shown to perform particularly bad in language as the features of words tend to have a much higher variance (there are many, very rare words which need to be considered for a good distribution estimate).\n", + "\n", + "Additionally to the Multi-Head Attention, a small fully connected feed-forward network is added to the model, which is applied to each position separately and identically. Specifically, the model uses a Linear$\\to$ReLU$\\to$Linear MLP. The full transformation including the residual connection can be expressed as: \n", + "\n", + "$$\n", + "\\begin{split}\n", + " \\text{FFN}(x) & = \\max(0, xW_1+b_1)W_2 + b_2\\\\\n", + " x & = \\text{LayerNorm}(x + \\text{FFN}(x))\n", + "\\end{split}\n", + "$$\n", + "\n", + "This MLP adds extra complexity to the model and allows transformations on each sequence element separately. You can imagine as this allows the model to \"post-process\" the new information added by the previous Multi-Head Attention, and prepare it for the next attention block. Usually, the inner dimensionality of the MLP is 2-8$\\times$ larger than $d_{\\text{model}}$, i.e. the dimensionality of the original input $x$. The general advantage of a wider layer instead of a narrow, multi-layer MLP is the faster, parallelizable execution.\n", + "\n", + "Finally, after looking at all parts of the encoder architecture, we can start implementing it below. We first start by implementing a single encoder block. Additionally to the layers described above, we will add dropout layers in the MLP and on the output of the MLP and Multi-Head Attention for regularization." + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "metadata": {}, + "outputs": [], + "source": [ + "class EncoderBlock(nn.Module):\n", + " \n", + " def __init__(self, input_dim, num_heads, dim_feedforward, dropout=0.0):\n", + " \"\"\"\n", + " Inputs:\n", + " input_dim - Dimensionality of the input\n", + " num_heads - Number of heads to use in the attention block\n", + " dim_feedforward - Dimensionality of the hidden layer in the MLP\n", + " dropout - Dropout probability to use in the dropout layers\n", + " \"\"\"\n", + " super().__init__()\n", + " \n", + " # Attention layer\n", + " self.self_attn = MultiheadAttention(input_dim, input_dim, num_heads)\n", + " \n", + " # Two-layer MLP\n", + " self.linear_net = nn.Sequential(\n", + " nn.Linear(input_dim, dim_feedforward),\n", + " nn.Dropout(dropout),\n", + " nn.ReLU(inplace=True),\n", + " nn.Linear(dim_feedforward, input_dim)\n", + " )\n", + " \n", + " # Layers to apply in between the main layers\n", + " self.norm1 = nn.LayerNorm(input_dim)\n", + " self.norm2 = nn.LayerNorm(input_dim)\n", + " self.dropout = nn.Dropout(dropout)\n", + "\n", + " def forward(self, x, mask=None):\n", + " # Attention part\n", + " attn_out = self.self_attn(x, mask=mask)\n", + " x = x + self.dropout(attn_out)\n", + " x = self.norm1(x)\n", + " \n", + " # MLP part\n", + " linear_out = self.linear_net(x)\n", + " x = x + self.dropout(linear_out)\n", + " x = self.norm2(x)\n", + " \n", + " return x" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Based on this block, we can implement a module for the full Transformer encoder. Additionally to a forward function that iterates through the sequence of encoder blocks, we also provide a function called `get_attention_maps`. The idea of this function is to return the attention probabilities for all Multi-Head Attention blocks in the encoder. This helps us in understanding, and in a sense, explaining the model. However, the attention probabilities should be interpreted with a grain of salt as it does not necessarily reflect the true interpretation of the model (there is a series of papers about this, including [Attention is not Explanation](https://arxiv.org/abs/1902.10186) and [Attention is not not Explanation](https://arxiv.org/abs/1908.04626))." + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "metadata": {}, + "outputs": [], + "source": [ + "class TransformerEncoder(nn.Module):\n", + " \n", + " def __init__(self, num_layers, **block_args):\n", + " super().__init__()\n", + " self.layers = nn.ModuleList([EncoderBlock(**block_args) for _ in range(num_layers)])\n", + "\n", + " def forward(self, x, mask=None):\n", + " for l in self.layers:\n", + " x = l(x, mask=mask)\n", + " return x\n", + "\n", + " def get_attention_maps(self, x, mask=None):\n", + " attention_maps = []\n", + " for l in self.layers:\n", + " _, attn_map = l.self_attn(x, mask=mask, return_attention=True)\n", + " attention_maps.append(attn_map)\n", + " x = l(x)\n", + " return attention_maps" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Positional encoding\n", + "\n", + "We have discussed before that the Multi-Head Attention block is permutation-equivariant, and cannot distinguish whether an input comes before another one in the sequence or not. In tasks like language understanding, however, the position is important for interpreting the input words. The position information can therefore be added via the input features. We could learn a embedding for every possible position, but this would not generalize to a dynamical input sequence length. Hence, the better option is to use feature patterns that the network can identify from the features and potentially generalize to larger sequences. The specific pattern chosen by Vaswani et al. are sine and cosine functions of different frequencies, as follows:\n", + "\n", + "$$\n", + "PE_{(pos,i)} = \\begin{cases}\n", + " \\sin\\left(\\frac{pos}{10000^{i/d_{\\text{model}}}}\\right) & \\text{if}\\hspace{3mm} i \\text{ mod } 2=0\\\\\n", + " \\cos\\left(\\frac{pos}{10000^{(i-1)/d_{\\text{model}}}}\\right) & \\text{otherwise}\\\\\n", + "\\end{cases}\n", + "$$\n", + "\n", + "$PE_{(pos,i)}$ represents the position encoding at position $pos$ in the sequence, and hidden dimensionality $i$. These values, concatenated for all hidden dimensions, are added to the original input features (in the Transformer visualization above, see \"Positional encoding\"), and constitute the position information. We distinguish between even ($i \\text{ mod } 2=0$) and uneven ($i \\text{ mod } 2=1$) hidden dimensionalities where we apply a sine/cosine respectively. The intuition behind this encoding is that you can represent $PE_{(pos+k,:)}$ as a linear function of $PE_{(pos,:)}$, which might allow the model to easily attend to relative positions. The wavelengths in different dimensions range from $2\\pi$ to $10000\\cdot 2\\pi$.\n", + "\n", + "The positional encoding is implemented below. The code is taken from the [PyTorch tutorial](https://pytorch.org/tutorials/beginner/transformer_tutorial.html#define-the-model) about Transformers on NLP and adjusted for our purposes." + ] + }, + { + "cell_type": "code", + "execution_count": 8, + "metadata": {}, + "outputs": [], + "source": [ + "class PositionalEncoding(nn.Module):\n", + "\n", + " def __init__(self, d_model, max_len=5000):\n", + " \"\"\"\n", + " Inputs\n", + " d_model - Hidden dimensionality of the input.\n", + " max_len - Maximum length of a sequence to expect.\n", + " \"\"\"\n", + " super().__init__()\n", + "\n", + " # Create matrix of [SeqLen, HiddenDim] representing the positional encoding for max_len inputs\n", + " pe = torch.zeros(max_len, d_model)\n", + " position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)\n", + " div_term = torch.exp(torch.arange(0, d_model, 2).float() * (-math.log(10000.0) / d_model))\n", + " pe[:, 0::2] = torch.sin(position * div_term)\n", + " pe[:, 1::2] = torch.cos(position * div_term)\n", + " pe = pe.unsqueeze(0)\n", + " \n", + " # register_buffer => Tensor which is not a parameter, but should be part of the modules state.\n", + " # Used for tensors that need to be on the same device as the module.\n", + " # persistent=False tells PyTorch to not add the buffer to the state dict (e.g. when we save the model) \n", + " self.register_buffer('pe', pe, persistent=False)\n", + "\n", + " def forward(self, x):\n", + " x = x + self.pe[:, :x.size(1)]\n", + " return x" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To understand the positional encoding, we can visualize it below. We will generate an image of the positional encoding over hidden dimensionality and position in a sequence. Each pixel, therefore, represents the change of the input feature we perform to encode the specific position. Let's do it below." + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:19.865866\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "encod_block = PositionalEncoding(d_model=48, max_len=96)\n", + "pe = encod_block.pe.squeeze().T.cpu().numpy()\n", + "\n", + "fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(8,3))\n", + "pos = ax.imshow(pe, cmap=\"RdGy\", extent=(1,pe.shape[1]+1,pe.shape[0]+1,1))\n", + "fig.colorbar(pos, ax=ax)\n", + "ax.set_xlabel(\"Position in sequence\")\n", + "ax.set_ylabel(\"Hidden dimension\")\n", + "ax.set_title(\"Positional encoding over hidden dimensions\")\n", + "ax.set_xticks([1]+[i*10 for i in range(1,1+pe.shape[1]//10)])\n", + "ax.set_yticks([1]+[i*10 for i in range(1,1+pe.shape[0]//10)])\n", + "plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "You can clearly see the sine and cosine waves with different wavelengths that encode the position in the hidden dimensions. Specifically, we can look at the sine/cosine wave for each hidden dimension separately, to get a better intuition of the pattern. Below we visualize the positional encoding for the hidden dimensions $1$, $2$, $3$ and $4$." + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:20.404975\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "sns.set_theme()\n", + "fig, ax = plt.subplots(2, 2, figsize=(12,4))\n", + "ax = [a for a_list in ax for a in a_list]\n", + "for i in range(len(ax)):\n", + " ax[i].plot(np.arange(1,17), pe[i,:16], color=f'C{i}', marker=\"o\", markersize=6, markeredgecolor=\"black\")\n", + " ax[i].set_title(f\"Encoding in hidden dimension {i+1}\")\n", + " ax[i].set_xlabel(\"Position in sequence\", fontsize=10)\n", + " ax[i].set_ylabel(\"Positional encoding\", fontsize=10)\n", + " ax[i].set_xticks(np.arange(1,17))\n", + " ax[i].tick_params(axis='both', which='major', labelsize=10)\n", + " ax[i].tick_params(axis='both', which='minor', labelsize=8)\n", + " ax[i].set_ylim(-1.2, 1.2)\n", + "fig.subplots_adjust(hspace=0.8)\n", + "sns.reset_orig()\n", + "plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As we can see, the patterns between the hidden dimension $1$ and $2$ only differ in the starting angle. The wavelength is $2\\pi$, hence the repetition after position $6$. The hidden dimensions $2$ and $3$ have about twice the wavelength. " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Learning rate warm-up\n", + "\n", + "One commonly used technique for training a Transformer is learning rate warm-up. This means that we gradually increase the learning rate from 0 on to our originally specified learning rate in the first few iterations. Thus, we slowly start learning instead of taking very large steps from the beginning. In fact, training a deep Transformer without learning rate warm-up can make the model diverge and achieve a much worse performance on training and testing. Take for instance the following plot by [Liu et al. (2019)](https://arxiv.org/pdf/1908.03265.pdf) comparing Adam-vanilla (i.e. Adam without warm-up) vs Adam with a warm-up:\n", + "\n", + ":::{figure} ../image/warmup_loss_plot.svg\n", + ":::\n", + "\n", + "Clearly, the warm-up is a crucial hyperparameter in the Transformer architecture. Why is it so important? There are currently two common explanations. Firstly, Adam uses the bias correction factors which however can lead to a higher variance in the adaptive learning rate during the first iterations. Improved optimizers like [RAdam](https://arxiv.org/abs/1908.03265) have been shown to overcome this issue, not requiring warm-up for training Transformers. Secondly, the iteratively applied Layer Normalization across layers can lead to very high gradients during the first iterations, which can be solved by using [Pre-Layer Normalization](https://proceedings.icml.cc/static/paper_files/icml/2020/328-Paper.pdf) (similar to Pre-Activation ResNet), or replacing Layer Normalization by other techniques ([Adaptive Normalization](https://proceedings.icml.cc/static/paper_files/icml/2020/328-Paper.pdf), [Power Normalization](https://arxiv.org/abs/2003.07845)). \n", + "\n", + "Nevertheless, many applications and papers still use the original Transformer architecture with Adam, because warm-up is a simple, yet effective way of solving the gradient problem in the first iterations. There are many different schedulers we could use. For instance, the original Transformer paper used an exponential decay scheduler with a warm-up. However, the currently most popular scheduler is the cosine warm-up scheduler, which combines warm-up with a cosine-shaped learning rate decay. We can implement it below, and visualize the learning rate factor over epochs. " + ] + }, + { + "cell_type": "code", + "execution_count": 11, + "metadata": {}, + "outputs": [], + "source": [ + "class CosineWarmupScheduler(optim.lr_scheduler._LRScheduler):\n", + " \n", + " def __init__(self, optimizer, warmup, max_iters):\n", + " self.warmup = warmup\n", + " self.max_num_iters = max_iters\n", + " super().__init__(optimizer)\n", + " \n", + " def get_lr(self):\n", + " lr_factor = self.get_lr_factor(epoch=self.last_epoch)\n", + " return [base_lr * lr_factor for base_lr in self.base_lrs]\n", + " \n", + " def get_lr_factor(self, epoch):\n", + " lr_factor = 0.5 * (1 + np.cos(np.pi * epoch / self.max_num_iters))\n", + " if epoch <= self.warmup:\n", + " lr_factor *= epoch * 1.0 / self.warmup\n", + " return lr_factor" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:20.987233\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# Needed for initializing the lr scheduler\n", + "p = nn.Parameter(torch.empty(4,4))\n", + "optimizer = optim.Adam([p], lr=1e-3)\n", + "lr_scheduler = CosineWarmupScheduler(optimizer=optimizer, warmup=100, max_iters=2000)\n", + "\n", + "# Plotting\n", + "epochs = list(range(2000))\n", + "sns.set()\n", + "plt.figure(figsize=(8,3))\n", + "plt.plot(epochs, [lr_scheduler.get_lr_factor(e) for e in epochs])\n", + "plt.ylabel(\"Learning rate factor\")\n", + "plt.xlabel(\"Iterations (in batches)\")\n", + "plt.title(\"Cosine Warm-up Learning Rate Scheduler\")\n", + "plt.show()\n", + "sns.reset_orig()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In the first 100 iterations, we increase the learning rate factor from 0 to 1, whereas for all later iterations, we decay it using the cosine wave. Pre-implementations of this scheduler can be found in the popular NLP Transformer library [huggingface](https://huggingface.co/transformers/main_classes/optimizer_schedules.html?highlight=cosine#transformers.get_cosine_schedule_with_warmup)." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### PyTorch Lightning Module\n", + "\n", + "Finally, we can embed the Transformer architecture into a PyTorch lightning module. From Tutorial 5, you know that PyTorch Lightning simplifies our training and test code, as well as structures the code nicely in separate functions. We will implement a template for a classifier based on the Transformer encoder. Thereby, we have a prediction output per sequence element. If we would need a classifier over the whole sequence, the common approach is to add an additional `[CLS]` token to the sequence, representing the classifier token. However, here we focus on tasks where we have an output per element. \n", + "\n", + "Additionally to the Transformer architecture, we add a small input network (maps input dimensions to model dimensions), the positional encoding, and an output network (transforms output encodings to predictions). We also add the learning rate scheduler, which takes a step each iteration instead of once per epoch. This is needed for the warmup and the smooth cosine decay. The training, validation, and test step is left empty for now and will be filled for our task-specific models." + ] + }, + { + "cell_type": "code", + "execution_count": 13, + "metadata": {}, + "outputs": [], + "source": [ + "class TransformerPredictor(pl.LightningModule):\n", + "\n", + " def __init__(self, input_dim, model_dim, num_classes, num_heads, num_layers, lr, warmup, max_iters, dropout=0.0, input_dropout=0.0):\n", + " \"\"\"\n", + " Inputs:\n", + " input_dim - Hidden dimensionality of the input\n", + " model_dim - Hidden dimensionality to use inside the Transformer\n", + " num_classes - Number of classes to predict per sequence element\n", + " num_heads - Number of heads to use in the Multi-Head Attention blocks\n", + " num_layers - Number of encoder blocks to use.\n", + " lr - Learning rate in the optimizer\n", + " warmup - Number of warmup steps. Usually between 50 and 500\n", + " max_iters - Number of maximum iterations the model is trained for. This is needed for the CosineWarmup scheduler\n", + " dropout - Dropout to apply inside the model\n", + " input_dropout - Dropout to apply on the input features\n", + " \"\"\"\n", + " super().__init__()\n", + " self.save_hyperparameters()\n", + " self._create_model()\n", + "\n", + " def _create_model(self):\n", + " # Input dim -> Model dim\n", + " self.input_net = nn.Sequential(\n", + " nn.Dropout(self.hparams.input_dropout),\n", + " nn.Linear(self.hparams.input_dim, self.hparams.model_dim)\n", + " )\n", + " # Positional encoding for sequences\n", + " self.positional_encoding = PositionalEncoding(d_model=self.hparams.model_dim)\n", + " # Transformer\n", + " self.transformer = TransformerEncoder(num_layers=self.hparams.num_layers,\n", + " input_dim=self.hparams.model_dim,\n", + " dim_feedforward=2*self.hparams.model_dim,\n", + " num_heads=self.hparams.num_heads,\n", + " dropout=self.hparams.dropout)\n", + " # Output classifier per sequence lement\n", + " self.output_net = nn.Sequential(\n", + " nn.Linear(self.hparams.model_dim, self.hparams.model_dim),\n", + " nn.LayerNorm(self.hparams.model_dim),\n", + " nn.ReLU(inplace=True),\n", + " nn.Dropout(self.hparams.dropout),\n", + " nn.Linear(self.hparams.model_dim, self.hparams.num_classes)\n", + " ) \n", + "\n", + " def forward(self, x, mask=None, add_positional_encoding=True):\n", + " \"\"\"\n", + " Inputs:\n", + " x - Input features of shape [Batch, SeqLen, input_dim]\n", + " mask - Mask to apply on the attention outputs (optional)\n", + " add_positional_encoding - If True, we add the positional encoding to the input.\n", + " Might not be desired for some tasks.\n", + " \"\"\"\n", + " x = self.input_net(x)\n", + " if add_positional_encoding:\n", + " x = self.positional_encoding(x)\n", + " x = self.transformer(x, mask=mask)\n", + " x = self.output_net(x)\n", + " return x\n", + "\n", + " @torch.no_grad()\n", + " def get_attention_maps(self, x, mask=None, add_positional_encoding=True):\n", + " \"\"\"\n", + " Function for extracting the attention matrices of the whole Transformer for a single batch.\n", + " Input arguments same as the forward pass.\n", + " \"\"\"\n", + " x = self.input_net(x)\n", + " if add_positional_encoding:\n", + " x = self.positional_encoding(x)\n", + " attention_maps = self.transformer.get_attention_maps(x, mask=mask)\n", + " return attention_maps\n", + "\n", + " def configure_optimizers(self):\n", + " optimizer = optim.Adam(self.parameters(), lr=self.hparams.lr)\n", + " \n", + " # Apply lr scheduler per step\n", + " lr_scheduler = CosineWarmupScheduler(optimizer, \n", + " warmup=self.hparams.warmup, \n", + " max_iters=self.hparams.max_iters)\n", + " return [optimizer], [{'scheduler': lr_scheduler, 'interval': 'step'}]\n", + "\n", + " def training_step(self, batch, batch_idx):\n", + " raise NotImplementedError\n", + "\n", + " def validation_step(self, batch, batch_idx):\n", + " raise NotImplementedError \n", + "\n", + " def test_step(self, batch, batch_idx):\n", + " raise NotImplementedError " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Experiments\n", + "\n", + "After having finished the implementation of the Transformer architecture, we can start experimenting and apply it to various tasks. In this notebook, we will focus on two tasks: parallel Sequence-to-Sequence, and set anomaly detection. The two tasks focus on different properties of the Transformer architecture, and we go through them below.\n", + "\n", + "### Sequence to Sequence\n", + "\n", + "A Sequence-to-Sequence task represents a task where the input _and_ the output is a sequence, not necessarily of the same length. Popular tasks in this domain include machine translation and summarization. For this, we usually have a Transformer encoder for interpreting the input sequence, and a decoder for generating the output in an autoregressive manner. Here, however, we will go back to a much simpler example task and use only the encoder. Given a sequence of $N$ numbers between $0$ and $M$, the task is to reverse the input sequence. In Numpy notation, if our input is $x$, the output should be $x$[::-1]. Although this task sounds very simple, RNNs can have issues with such because the task requires long-term dependencies. Transformers are built to support such, and hence, we expect it to perform very well. \n", + "\n", + "First, let's create a dataset class below." + ] + }, + { + "cell_type": "code", + "execution_count": 14, + "metadata": {}, + "outputs": [], + "source": [ + "class ReverseDataset(data.Dataset):\n", + "\n", + " def __init__(self, num_categories, seq_len, size):\n", + " super().__init__()\n", + " self.num_categories = num_categories\n", + " self.seq_len = seq_len\n", + " self.size = size\n", + " \n", + " self.data = torch.randint(self.num_categories, size=(self.size, self.seq_len))\n", + " \n", + " def __len__(self):\n", + " return self.size\n", + "\n", + " def __getitem__(self, idx):\n", + " inp_data = self.data[idx]\n", + " labels = torch.flip(inp_data, dims=(0,))\n", + " return inp_data, labels" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We create an arbitrary number of random sequences of numbers between 0 and `num_categories-1`. The label is simply the tensor flipped over the sequence dimension. We can create the corresponding data loaders below. " + ] + }, + { + "cell_type": "code", + "execution_count": 15, + "metadata": {}, + "outputs": [], + "source": [ + "dataset = partial(ReverseDataset, 10, 16)\n", + "train_loader = data.DataLoader(dataset(50000), batch_size=128, shuffle=True, drop_last=True, pin_memory=True)\n", + "val_loader = data.DataLoader(dataset(1000), batch_size=128)\n", + "test_loader = data.DataLoader(dataset(10000), batch_size=128)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Let's look at an arbitrary sample of the dataset:" + ] + }, + { + "cell_type": "code", + "execution_count": 16, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Input data: tensor([9, 6, 2, 0, 6, 2, 7, 9, 7, 3, 3, 4, 3, 7, 0, 9])\n", + "Labels: tensor([9, 0, 7, 3, 4, 3, 3, 7, 9, 7, 2, 6, 0, 2, 6, 9])\n" + ] + } + ], + "source": [ + "inp_data, labels = train_loader.dataset[0]\n", + "print(\"Input data:\", inp_data)\n", + "print(\"Labels: \", labels)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "During training, we pass the input sequence through the Transformer encoder and predict the output for each input token. We use the standard Cross-Entropy loss to perform this. Every number is represented as a one-hot vector. Remember that representing the categories as single scalars decreases the expressiveness of the model extremely as $0$ and $1$ are not closer related than $0$ and $9$ in our example. An alternative to a one-hot vector is using a learned embedding vector as it is provided by the PyTorch module `nn.Embedding`. However, using a one-hot vector with an additional linear layer as in our case has the same effect as an embedding layer (`self.input_net` maps one-hot vector to a dense vector, where each row of the weight matrix represents the embedding for a specific category).\n", + "\n", + "To implement the training dynamic, we create a new class inheriting from `TransformerPredictor` and overwriting the training, validation and test step functions." + ] + }, + { + "cell_type": "code", + "execution_count": 17, + "metadata": {}, + "outputs": [], + "source": [ + "class ReversePredictor(TransformerPredictor):\n", + " \n", + " def _calculate_loss(self, batch, mode=\"train\"):\n", + " # Fetch data and transform categories to one-hot vectors\n", + " inp_data, labels = batch\n", + " inp_data = F.one_hot(inp_data, num_classes=self.hparams.num_classes).float()\n", + " \n", + " # Perform prediction and calculate loss and accuracy\n", + " preds = self.forward(inp_data, add_positional_encoding=True)\n", + " loss = F.cross_entropy(preds.view(-1,preds.size(-1)), labels.view(-1))\n", + " acc = (preds.argmax(dim=-1) == labels).float().mean()\n", + " \n", + " # Logging\n", + " self.log(f\"{mode}_loss\", loss)\n", + " self.log(f\"{mode}_acc\", acc)\n", + " return loss, acc\n", + " \n", + " def training_step(self, batch, batch_idx):\n", + " loss, _ = self._calculate_loss(batch, mode=\"train\")\n", + " return loss\n", + " \n", + " def validation_step(self, batch, batch_idx):\n", + " _ = self._calculate_loss(batch, mode=\"val\")\n", + " \n", + " def test_step(self, batch, batch_idx):\n", + " _ = self._calculate_loss(batch, mode=\"test\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Finally, we can create a training function similar to the one we have seen in Tutorial 5 for PyTorch Lightning. We create a `pl.Trainer` object, running for $N$ epochs, logging in TensorBoard, and saving our best model based on the validation. Afterward, we test our models on the test set. An additional parameter we pass to the trainer here is `gradient_clip_val`. This clips the norm of the gradients for all parameters before taking an optimizer step and prevents the model from diverging if we obtain very high gradients at, for instance, sharp loss surfaces (see many good blog posts on gradient clipping, like [DeepAI glossary](https://deepai.org/machine-learning-glossary-and-terms/gradient-clipping)). For Transformers, gradient clipping can help to further stabilize the training during the first few iterations, and also afterward. In plain PyTorch, you can apply gradient clipping via `torch.nn.utils.clip_grad_norm_(...)` (see [documentation](https://pytorch.org/docs/stable/generated/torch.nn.utils.clip_grad_norm_.html#torch.nn.utils.clip_grad_norm_)). The clip value is usually between 0.5 and 10, depending on how harsh you want to clip large gradients. After having explained this, let's implement the training function:" + ] + }, + { + "cell_type": "code", + "execution_count": 18, + "metadata": {}, + "outputs": [], + "source": [ + "def train_reverse(**kwargs):\n", + " # Create a PyTorch Lightning trainer with the generation callback\n", + " root_dir = os.path.join(CHECKPOINT_PATH, \"ReverseTask\")\n", + " os.makedirs(root_dir, exist_ok=True)\n", + " trainer = pl.Trainer(default_root_dir=root_dir, \n", + " callbacks=[ModelCheckpoint(save_weights_only=True, mode=\"max\", monitor=\"val_acc\")],\n", + " accelerator=\"gpu\" if str(device).startswith(\"cuda\") else \"cpu\",\n", + " devices=1,\n", + " max_epochs=10,\n", + " gradient_clip_val=5)\n", + " trainer.logger._default_hp_metric = None # Optional logging argument that we don't need\n", + " \n", + " # Check whether pretrained model exists. If yes, load it and skip training\n", + " pretrained_filename = os.path.join(CHECKPOINT_PATH, \"ReverseTask.ckpt\")\n", + " if os.path.isfile(pretrained_filename):\n", + " print(\"Found pretrained model, loading...\")\n", + " model = ReversePredictor.load_from_checkpoint(pretrained_filename)\n", + " else:\n", + " model = ReversePredictor(max_iters=trainer.max_epochs*len(train_loader), **kwargs)\n", + " trainer.fit(model, train_loader, val_loader)\n", + " \n", + " # Test best model on validation and test set\n", + " val_result = trainer.test(model, val_loader, verbose=False)\n", + " test_result = trainer.test(model, test_loader, verbose=False)\n", + " result = {\"test_acc\": test_result[0][\"test_acc\"], \"val_acc\": val_result[0][\"test_acc\"]}\n", + " \n", + " model = model.to(device)\n", + " return model, result" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Finally, we can train the model. In this setup, we will use a single encoder block and a single head in the Multi-Head Attention. This is chosen because of the simplicity of the task, and in this case, the attention can actually be interpreted as an \"explanation\" of the predictions (compared to the other papers above dealing with deep Transformers). " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "def scaled_dot_product(q, k, v, mask=None):\n", + " d_k = q.size()[-1]\n", + " attn_logits = torch.matmul(q, k.transpose(-2, -1))\n", + " attn_logits = attn_logits / math.sqrt(d_k)\n", + " if mask is not None:\n", + " attn_logits = attn_logits.masked_fill(mask == 0, -9e15)\n", + " attention = F.softmax(attn_logits, dim=-1)\n", + " values = torch.matmul(attention, v)\n", + " return values, attention\n", + "\n", + "class MultiheadAttention(nn.Module):\n", + " \n", + " def __init__(self, input_dim, embed_dim, num_heads):\n", + " super().__init__()\n", + " assert embed_dim % num_heads == 0, \"Embedding dimension must be 0 modulo number of heads.\"\n", + " \n", + " self.embed_dim = embed_dim\n", + " self.num_heads = num_heads\n", + " self.head_dim = embed_dim // num_heads\n", + " \n", + " # Stack all weight matrices 1...h together for efficiency\n", + " # Note that in many implementations you see \"bias=False\" which is optional\n", + " self.qkv_proj = nn.Linear(input_dim, 3*embed_dim)\n", + " self.o_proj = nn.Linear(embed_dim, embed_dim)\n", + " \n", + " self._reset_parameters()\n", + "\n", + " def _reset_parameters(self):\n", + " # Original Transformer initialization, see PyTorch documentation\n", + " nn.init.xavier_uniform_(self.qkv_proj.weight)\n", + " self.qkv_proj.bias.data.fill_(0)\n", + " nn.init.xavier_uniform_(self.o_proj.weight)\n", + " self.o_proj.bias.data.fill_(0)\n", + "\n", + " def forward(self, x, mask=None, return_attention=False):\n", + " batch_size, seq_length, _ = x.size()\n", + " if mask is not None:\n", + " mask = expand_mask(mask)\n", + " qkv = self.qkv_proj(x)\n", + " \n", + " # Separate Q, K, V from linear output\n", + " qkv = qkv.reshape(batch_size, seq_length, self.num_heads, 3*self.head_dim)\n", + " qkv = qkv.permute(0, 2, 1, 3) # [Batch, Head, SeqLen, Dims]\n", + " q, k, v = qkv.chunk(3, dim=-1)\n", + " \n", + " # Determine value outputs\n", + " values, attention = scaled_dot_product(q, k, v, mask=mask)\n", + " values = values.permute(0, 2, 1, 3) # [Batch, SeqLen, Head, Dims]\n", + " values = values.reshape(batch_size, seq_length, self.embed_dim)\n", + " o = self.o_proj(values)\n", + " \n", + " if return_attention:\n", + " return o, attention\n", + " else:\n", + " return o" + ] + }, + { + "cell_type": "code", + "execution_count": 19, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "GPU available: True, used: True\n", + "TPU available: False, using: 0 TPU cores\n", + "LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Found pretrained model, loading...\n" + ] + } + ], + "source": [ + "reverse_model, reverse_result = train_reverse(input_dim=train_loader.dataset.num_categories,\n", + " model_dim=32,\n", + " num_heads=1,\n", + " num_classes=train_loader.dataset.num_categories,\n", + " num_layers=1,\n", + " dropout=0.0,\n", + " lr=5e-4,\n", + " warmup=50)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The warning of PyTorch Lightning regarding the number of workers can be ignored for now. As the data set is so simple and the `__getitem__` finishes a neglectable time, we don't need subprocesses to provide us the data (in fact, more workers can slow down the training as we have communication overhead among processes/threads). First, let's print the results:" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Val accuracy: 100.00%\n", + "Test accuracy: 100.00%\n" + ] + } + ], + "source": [ + "print(f\"Val accuracy: {(100.0 * reverse_result['val_acc']):4.2f}%\")\n", + "print(f\"Test accuracy: {(100.0 * reverse_result['test_acc']):4.2f}%\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As we would have expected, the Transformer can correctly solve the task. However, how does the attention in the Multi-Head Attention block looks like for an arbitrary input? Let's try to visualize it below." + ] + }, + { + "cell_type": "code", + "execution_count": 21, + "metadata": {}, + "outputs": [], + "source": [ + "data_input, labels = next(iter(val_loader))\n", + "inp_data = F.one_hot(data_input, num_classes=reverse_model.hparams.num_classes).float()\n", + "inp_data = inp_data.to(device)\n", + "attention_maps = reverse_model.get_attention_maps(inp_data)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The object `attention_maps` is a list of length $N$ where $N$ is the number of layers. Each element is a tensor of shape [Batch, Heads, SeqLen, SeqLen], which we can verify below." + ] + }, + { + "cell_type": "code", + "execution_count": 22, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "torch.Size([128, 1, 16, 16])" + ] + }, + "execution_count": 22, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "attention_maps[0].shape" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Next, we will write a plotting function that takes as input the sequences, attention maps, and an index indicating for which batch element we want to visualize the attention map. We will create a plot where over rows, we have different layers, while over columns, we show the different heads. Remember that the softmax has been applied for each row separately." + ] + }, + { + "cell_type": "code", + "execution_count": 23, + "metadata": {}, + "outputs": [], + "source": [ + "def plot_attention_maps(input_data, attn_maps, idx=0):\n", + " if input_data is not None:\n", + " input_data = input_data[idx].detach().cpu().numpy()\n", + " else:\n", + " input_data = np.arange(attn_maps[0][idx].shape[-1])\n", + " attn_maps = [m[idx].detach().cpu().numpy() for m in attn_maps]\n", + " \n", + " num_heads = attn_maps[0].shape[0]\n", + " num_layers = len(attn_maps)\n", + " seq_len = input_data.shape[0]\n", + " fig_size = 4 if num_heads == 1 else 3\n", + " fig, ax = plt.subplots(num_layers, num_heads, figsize=(num_heads*fig_size, num_layers*fig_size))\n", + " if num_layers == 1:\n", + " ax = [ax]\n", + " if num_heads == 1:\n", + " ax = [[a] for a in ax]\n", + " for row in range(num_layers):\n", + " for column in range(num_heads):\n", + " ax[row][column].imshow(attn_maps[row][column], origin='lower', vmin=0)\n", + " ax[row][column].set_xticks(list(range(seq_len)))\n", + " ax[row][column].set_xticklabels(input_data.tolist())\n", + " ax[row][column].set_yticks(list(range(seq_len)))\n", + " ax[row][column].set_yticklabels(input_data.tolist())\n", + " ax[row][column].set_title(f\"Layer {row+1}, Head {column+1}\")\n", + " fig.subplots_adjust(hspace=0.5)\n", + " plt.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Finally, we can plot the attention map of our trained Transformer on the reverse task:" + ] + }, + { + "cell_type": "code", + "execution_count": 24, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:26.716937\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "plot_attention_maps(data_input, attention_maps, idx=0)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The model has learned to attend to the token that is on the flipped index of itself. Hence, it actually does what we intended it to do. We see that it however also pays some attention to values close to the flipped index. This is because the model doesn't need the perfect, hard attention to solve this problem, but is fine with this approximate, noisy attention map. The close-by indices are caused by the similarity of the positional encoding, which we also intended with the positional encoding." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Set Anomaly Detection\n", + "\n", + "Besides sequences, sets are another data structure that is relevant for many applications. In contrast to sequences, elements are unordered in a set. RNNs can only be applied on sets by assuming an order in the data, which however biases the model towards a non-existing order in the data. [Vinyals et al. (2015)](https://arxiv.org/abs/1511.06391) and other papers have shown that the assumed order can have a significant impact on the model's performance, and hence, we should try to not use RNNs on sets. Ideally, our model should be permutation-equivariant/invariant such that the output is the same no matter how we sort the elements in a set. \n", + "\n", + "Transformers offer the perfect architecture for this as the Multi-Head Attention is permutation-equivariant, and thus, outputs the same values no matter in what order we enter the inputs (inputs and outputs are permuted equally). The task we are looking at for sets is _Set Anomaly Detection_ which means that we try to find the element(s) in a set that does not fit the others. In the research community, the common application of anomaly detection is performed on a set of images, where $N-1$ images belong to the same category/have the same high-level features while one belongs to another category. Note that category does not necessarily have to relate to a class in a standard classification problem, but could be the combination of multiple features. For instance, on a face dataset, this could be people with glasses, male, beard, etc. An example of distinguishing different animals can be seen below. The first four images show foxes, while the last represents a different animal. We want to recognize that the last image shows a different animal, but it is not relevant which class of animal it is.\n", + "\n", + "
\n", + "\n", + ":::{figure} ../image/warmup_loss_plot.svg\n", + ":::\n", + "\n", + "In this tutorial, we will use the CIFAR100 dataset. CIFAR100 has 600 images for 100 classes each with a resolution of 32x32, similar to CIFAR10. The larger amount of classes requires the model to attend to specific features in the images instead of coarse features as in CIFAR10, therefore making the task harder. We will show the model a set of 9 images of one class, and 1 image from another class. The task is to find the image that is from a different class than the other images.\n", + "Using the raw images directly as input to the Transformer is not a good idea, because it is not translation invariant as a CNN, and would need to learn to detect image features from high-dimensional input first of all. Instead, we will use a pre-trained ResNet34 model from the torchvision package to obtain high-level, low-dimensional features of the images. The ResNet model has been pre-trained on the [ImageNet](http://image-net.org/) dataset which contains 1 million images of 1k classes and varying resolutions. However, during training and testing, the images are usually scaled to a resolution of 224x224, and hence we rescale our CIFAR images to this resolution as well. Below, we will load the dataset, and prepare the data for being processed by the ResNet model." + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Files already downloaded and verified\n", + "Files already downloaded and verified\n" + ] + } + ], + "source": [ + "# ImageNet statistics\n", + "DATA_MEANS = np.array([0.485, 0.456, 0.406])\n", + "DATA_STD = np.array([0.229, 0.224, 0.225])\n", + "# As torch tensors for later preprocessing\n", + "TORCH_DATA_MEANS = torch.from_numpy(DATA_MEANS).view(1,3,1,1)\n", + "TORCH_DATA_STD = torch.from_numpy(DATA_STD).view(1,3,1,1)\n", + "\n", + "# Resize to 224x224, and normalize to ImageNet statistic\n", + "transform = transforms.Compose([transforms.Resize((224,224)),\n", + " transforms.ToTensor(),\n", + " transforms.Normalize(DATA_MEANS, DATA_STD)\n", + " ])\n", + "# Loading the training dataset. \n", + "train_set = CIFAR100(root=DATASET_PATH, train=True, transform=transform, download=True)\n", + "\n", + "# Loading the test set\n", + "test_set = CIFAR100(root=DATASET_PATH, train=False, transform=transform, download=True)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Next, we want to run the pre-trained ResNet model on the images, and extract the features before the classification layer. These are the most high-level features, and should sufficiently describe the images. CIFAR100 has some similarity to ImageNet, and thus we are not retraining the ResNet model in any form. However, if you would want to get the best performance and have a very large dataset, it would be better to add the ResNet to the computation graph during training and finetune its parameters as well. As we don't have a large enough dataset and want to train our model efficiently, we will extract the features beforehand. Let's load and prepare the model below." + ] + }, + { + "cell_type": "code", + "execution_count": 26, + "metadata": {}, + "outputs": [], + "source": [ + "import os\n", + "os.environ[\"TORCH_HOME\"] = CHECKPOINT_PATH\n", + "pretrained_model = torchvision.models.resnet34(weights='IMAGENET1K_V1')\n", + "# Remove classification layer\n", + "# In some models, it is called \"fc\", others have \"classifier\"\n", + "# Setting both to an empty sequential represents an identity map of the final features.\n", + "pretrained_model.fc = nn.Sequential()\n", + "pretrained_model.classifier = nn.Sequential()\n", + "# To GPU\n", + "pretrained_model = pretrained_model.to(device)\n", + "\n", + "# Only eval, no gradient required\n", + "pretrained_model.eval()\n", + "for p in pretrained_model.parameters():\n", + " p.requires_grad = False" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We will now write a extraction function for the features below. This cell requires access to a GPU, as the model is rather deep and the images relatively large. The GPUs on GoogleColab are sufficient, but running this cell can take 2-3 minutes. Once it is run, the features are exported on disk so they don't have to be recalculated every time you run the notebook. However, this requires >150MB free disk space. So it is recommended to run this only on a local computer if you have enough free disk and a GPU (GoogleColab is fine for this). If you do not have a GPU, you can download the features from the [GoogleDrive folder](https://drive.google.com/drive/folders/1DF7POc6j03pRiWQPWSl5QJX5iY-xK0sV?usp=sharing)." + ] + }, + { + "cell_type": "code", + "execution_count": 27, + "metadata": {}, + "outputs": [], + "source": [ + "@torch.no_grad()\n", + "def extract_features(dataset, save_file):\n", + " if not os.path.isfile(save_file):\n", + " data_loader = data.DataLoader(dataset, batch_size=128, shuffle=False, drop_last=False, num_workers=4)\n", + " extracted_features = []\n", + " for imgs, _ in tqdm(data_loader):\n", + " imgs = imgs.to(device)\n", + " feats = pretrained_model(imgs)\n", + " extracted_features.append(feats)\n", + " extracted_features = torch.cat(extracted_features, dim=0)\n", + " extracted_features = extracted_features.detach().cpu()\n", + " torch.save(extracted_features, save_file)\n", + " else:\n", + " extracted_features = torch.load(save_file)\n", + " return extracted_features\n", + "\n", + "train_feat_file = os.path.join(CHECKPOINT_PATH, \"train_set_features.tar\")\n", + "train_set_feats = extract_features(train_set, train_feat_file)\n", + "\n", + "test_feat_file = os.path.join(CHECKPOINT_PATH, \"test_set_features.tar\")\n", + "test_feats = extract_features(test_set, test_feat_file)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Let's verify the feature shapes below. The training should have 50k elements, and the test 10k images. The feature dimension is 512 for the ResNet34. If you experiment with other models, you likely see a different feature dimension." + ] + }, + { + "cell_type": "code", + "execution_count": 28, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Train: torch.Size([50000, 512])\n", + "Test: torch.Size([10000, 512])\n" + ] + } + ], + "source": [ + "print(\"Train:\", train_set_feats.shape)\n", + "print(\"Test: \", test_feats.shape)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As usual, we want to create a validation set to detect when we should stop training. In this case, we will split the training set into 90% training, 10% validation. However, the difficulty is here that we need to ensure that the validation set has the same number of images for all 100 labels. Otherwise, we have a class imbalance which is not good for creating the image sets. Hence, we take 10% of the images for each class, and move them into the validation set. The code below does exactly this." + ] + }, + { + "cell_type": "code", + "execution_count": 29, + "metadata": {}, + "outputs": [], + "source": [ + "## Split train into train+val\n", + "# Get labels from train set\n", + "labels = train_set.targets\n", + "\n", + "# Get indices of images per class\n", + "labels = torch.LongTensor(labels)\n", + "num_labels = labels.max()+1\n", + "sorted_indices = torch.argsort(labels).reshape(num_labels, -1) # [classes, num_imgs per class]\n", + "\n", + "# Determine number of validation images per class\n", + "num_val_exmps = sorted_indices.shape[1] // 10\n", + "\n", + "# Get image indices for validation and training\n", + "val_indices = sorted_indices[:,:num_val_exmps].reshape(-1)\n", + "train_indices = sorted_indices[:,num_val_exmps:].reshape(-1)\n", + "\n", + "# Group corresponding image features and labels\n", + "train_feats, train_labels = train_set_feats[train_indices], labels[train_indices]\n", + "val_feats, val_labels = train_set_feats[val_indices], labels[val_indices]" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Now we can prepare a dataset class for the set anomaly task. We define an epoch to be the sequence in which each image has been exactly once as an \"anomaly\". Hence, the length of the dataset is the number of images in it. For the training set, each time we access an item with `__getitem__`, we sample a random, different class than the image at the corresponding index `idx` has. In a second step, we sample $N-1$ images of this sampled class. The set of 10 images is finally returned. The randomness in the `__getitem__` allows us to see a slightly different set during each iteration. However, we can't use the same strategy for the test set as we want the test dataset to be the same every time we iterate over it. Hence, we sample the sets in the `__init__` method, and return those in `__getitem__`. The code below implements exactly this dynamic." + ] + }, + { + "cell_type": "code", + "execution_count": 30, + "metadata": {}, + "outputs": [], + "source": [ + "class SetAnomalyDataset(data.Dataset):\n", + " \n", + " def __init__(self, img_feats, labels, set_size=10, train=True):\n", + " \"\"\"\n", + " Inputs:\n", + " img_feats - Tensor of shape [num_imgs, img_dim]. Represents the high-level features.\n", + " labels - Tensor of shape [num_imgs], containing the class labels for the images\n", + " set_size - Number of elements in a set. N-1 are sampled from one class, and one from another one.\n", + " train - If True, a new set will be sampled every time __getitem__ is called.\n", + " \"\"\"\n", + " super().__init__()\n", + " self.img_feats = img_feats\n", + " self.labels = labels\n", + " self.set_size = set_size-1 # The set size is here the size of correct images\n", + " self.train = train\n", + " \n", + " # Tensors with indices of the images per class\n", + " self.num_labels = labels.max()+1\n", + " self.img_idx_by_label = torch.argsort(self.labels).reshape(self.num_labels, -1)\n", + " \n", + " if not train:\n", + " self.test_sets = self._create_test_sets()\n", + " \n", + " \n", + " def _create_test_sets(self):\n", + " # Pre-generates the sets for each image for the test set\n", + " test_sets = []\n", + " num_imgs = self.img_feats.shape[0]\n", + " np.random.seed(42)\n", + " test_sets = [self.sample_img_set(self.labels[idx]) for idx in range(num_imgs)]\n", + " test_sets = torch.stack(test_sets, dim=0)\n", + " return test_sets\n", + " \n", + " \n", + " def sample_img_set(self, anomaly_label):\n", + " \"\"\"\n", + " Samples a new set of images, given the label of the anomaly. \n", + " The sampled images come from a different class than anomaly_label\n", + " \"\"\"\n", + " # Sample class from 0,...,num_classes-1 while skipping anomaly_label as class\n", + " set_label = np.random.randint(self.num_labels-1)\n", + " if set_label >= anomaly_label:\n", + " set_label += 1\n", + " \n", + " # Sample images from the class determined above\n", + " img_indices = np.random.choice(self.img_idx_by_label.shape[1], size=self.set_size, replace=False)\n", + " img_indices = self.img_idx_by_label[set_label, img_indices]\n", + " return img_indices\n", + " \n", + " \n", + " def __len__(self):\n", + " return self.img_feats.shape[0]\n", + " \n", + " \n", + " def __getitem__(self, idx):\n", + " anomaly = self.img_feats[idx]\n", + " if self.train: # If train => sample\n", + " img_indices = self.sample_img_set(self.labels[idx])\n", + " else: # If test => use pre-generated ones\n", + " img_indices = self.test_sets[idx]\n", + " \n", + " # Concatenate images. The anomaly is always the last image for simplicity\n", + " img_set = torch.cat([self.img_feats[img_indices], anomaly[None]], dim=0)\n", + " indices = torch.cat([img_indices, torch.LongTensor([idx])], dim=0)\n", + " label = img_set.shape[0]-1\n", + " \n", + " # We return the indices of the images for visualization purpose. \"Label\" is the index of the anomaly\n", + " return img_set, indices, label" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Next, we can setup our datasets and data loaders below. Here, we will use a set size of 10, i.e. 9 images from one category + 1 anomaly. Feel free to change it if you want to experiment with the sizes. " + ] + }, + { + "cell_type": "code", + "execution_count": 31, + "metadata": {}, + "outputs": [], + "source": [ + "SET_SIZE = 10\n", + "test_labels = torch.LongTensor(test_set.targets)\n", + "\n", + "train_anom_dataset = SetAnomalyDataset(train_feats, train_labels, set_size=SET_SIZE, train=True)\n", + "val_anom_dataset = SetAnomalyDataset(val_feats, val_labels, set_size=SET_SIZE, train=False)\n", + "test_anom_dataset = SetAnomalyDataset(test_feats, test_labels, set_size=SET_SIZE, train=False)\n", + "\n", + "train_anom_loader = data.DataLoader(train_anom_dataset, batch_size=64, shuffle=True, drop_last=True, num_workers=4, pin_memory=True)\n", + "val_anom_loader = data.DataLoader(val_anom_dataset, batch_size=64, shuffle=False, drop_last=False, num_workers=4)\n", + "test_anom_loader = data.DataLoader(test_anom_dataset, batch_size=64, shuffle=False, drop_last=False, num_workers=4)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To understand the dataset a little better, we can plot below a few sets from the test dataset. Each row shows a different input set, where the first 9 are from the same class." + ] + }, + { + "cell_type": "code", + "execution_count": 32, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "JVBERi0xLjQKJazcIKu6CjEgMCBvYmoKPDwgL1BhZ2VzIDIgMCBSIC9UeXBlIC9DYXRhbG9nID4+CmVuZG9iago4IDAgb2JqCjw8IC9FeHRHU3RhdGUgNCAwIFIgL0ZvbnQgMyAwIFIgL1BhdHRlcm4gNSAwIFIKL1Byb2NTZXQgWyAvUERGIC9UZXh0IC9JbWFnZUIgL0ltYWdlQyAvSW1hZ2VJIF0gL1NoYWRpbmcgNiAwIFIKL1hPYmplY3QgNyAwIFIgPj4KZW5kb2JqCjEwIDAgb2JqCjw8IC9Bbm5vdHMgWyBdIC9Db250ZW50cyA5IDAgUgovR3JvdXAgPDwgL0NTIC9EZXZpY2VSR0IgL1MgL1RyYW5zcGFyZW5jeSAvVHlwZSAvR3JvdXAgPj4KL01lZGlhQm94IFsgMCAwIDY4NCAzMDAuMDI1NjYyMjUxNyBdIC9QYXJlbnQgMiAwIFIgL1Jlc291cmNlcyA4IDAgUgovVHlwZSAvUGFnZSA+PgplbmRvYmoKOSAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDExIDAgUiA+PgpzdHJlYW0KeJxVjjtvwzAMhHf+ihvbRSJpS0pGp2mMjA4EdA5cJa3hR1MDffz7ygH6GojDHXj8KOjIVoLzDEaX5x2CGnab3p7bdKg3aGfinA/kV2XW/qoFs2F13msO+L99IhrpgmD0Ot6vjYcGNo45b6iTgNeEB4ywlS5kyWTJZEadiz4sPM4V+TnSDrB7wXZCQw0u30XG+W958bSJsDuBKOKJtCzN2hWiDroqTPj9ID7STTVOw7H/RPo4Di99mjGNuNvvqoMw3yJ2uI/U0BcZLEFQCmVuZHN0cmVhbQplbmRvYmoKMTEgMCBvYmoKMjA0CmVuZG9iagoxNyAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDg4ID4+CnN0cmVhbQp4nDWMuw3AMAhEe6a4Efg4gPeJUpH92xBbLrh70hPnOcDIPg9H6MQtZEPhpnhJOaE+UTRabzq2SHO/vGQzFxX9M9x9he3mgGQ0SeQh0eVy5Vkpej6X2ht+CmVuZHN0cmVhbQplbmRvYmoKMTggMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAyMzIgPj4Kc3RyZWFtCnicNVE7cgUxCOt9Cl0gM+Zvn2czr0ru30awk2ZhAQkJ5z3YiMSXGNId5YpvWZ1mGX4ni7z4WSmcvBdRgVRFWCHt4FnOaobBcyNT4HImPsvMJ9NixwKqiTjOjpxmMAgxjetoOR1mmgc9IdcHI27sNMtVDGm9W6rX91r+U0X5yLqb5dYpm1qpW/SMPYnLzuupLe0Lo47ipiDS4WOH9yBfxJzFRSfSzX4z5bCSNASnBfAjMZTq2eE1wsTPjARP2dPpfZSG1z5our53L+jIzYRM5RbKSMWTlcaYMVS/Ec0k9f0/0LM+f5owVEcKZW5kc3RyZWFtCmVuZG9iagoxOSAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDc0ID4+CnN0cmVhbQp4nDM1N1UwULC0ABKmhuYK5kaWCimGXEA+iJXLBRPLAbPMTMyALENLZJaJsSGQZWJhhsQyNrGAyiJYBkAabE0OzPQcrjQAA3EYkwplbmRzdHJlYW0KZW5kb2JqCjIwIDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggNDkgPj4Kc3RyZWFtCnicM7I0VTBQsLQAEoaW5grmRpYKKYZcQD6IlcsFE8sBswyANFhpDkxFDlcaAKVEDOQKZW5kc3RyZWFtCmVuZG9iagoyMSAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDIyNyA+PgpzdHJlYW0KeJxFkEuOAyEQQ/ecwkeg/nCejrLq3H87LjrRbLAlKNczuQMTe/HITJRuvGS4O8wVn+EZMHP4SphsxEzoTlwjlK4U4VSfCI7L3rzpoIl7RM6jngVZ1c4NagFnkuaC7YIu54wVN87JrUblzfSj1xC+aXcf13mH9kjj3sNUvs451c67ighpC1nVtL6QbBTJDms/Kk3bzssQseBsGlboHN4Iu1d3J0sYfr/yMCUTPw/d+lF8XTej6xRnJ1cma8956EnpX/XKow/FcSnoF7HtzCT3X6dTkqlTe2fvaf2nuMf7D5BuVjkKZW5kc3RyZWFtCmVuZG9iagoyMiAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDMwNCA+PgpzdHJlYW0KeJw9kjuSwzAMQ3udghfIjPiT5PNkJ5X3/u0+MslWgEmJACgvdZmypjwgaSYJ/9Hh4WI75XfYns3MwLVELxPLKc+hK8TcRfmymY26sjrFqsMwnVv0qJyLhk2TmucqSxm3C57DtYnnln3EDzc0qAd1jUvCDd3VaFkKzXB1/zu9R9l3NTwXm1Tq1BePF1EV5vkhT6KH6UrifDwoIVx7MEYWEuRT0UCOs1yt8l5C9g63GrLCQWpJ57MnPNh1ek8ubhfNEA9kuVT4TlHs7dAzvuxKCT0StuFY7n07mrHpGps47H7vRtbKjK5oIX7IVyfrJWDcUyZFEmROtlhui9We7qEopnOGcxkg6tmKhlLmYlerfww7bywv2SzIlMwLMkanTZ44eMh+jZr0eZXneP0BbPNzOwplbmRzdHJlYW0KZW5kb2JqCjIzIDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggMjQ1ID4+CnN0cmVhbQp4nEVQu41DMQzrPQUXCGD9LHued0iV2789SkZwhSFaP5JaEpiIwEsMsZRv4kdGQT0LvxeF4jPEzxeFQc6EpECc9RkQmXiG2kZu6HZwzrzDM4w5AhfFWnCm05n2XNjknAcnEM5tlPGMQrpJVBVxVJ9xTPGqss+N14GltWyz05HsIY2ES0klJpd+Uyr/tClbKujaRROwSOSBk0004Sw/Q5JizKCUUfcwtY70cbKRR3XQydmcOS2Z2e6n7Ux8D1gmmVHlKZ3nMj4nqfNcTn3usx3R5KKlVfuc/d6RlvIitduh1elXJVGZjdWnkLg8/4yf8f4DjqBZPgplbmRzdHJlYW0KZW5kb2JqCjI0IDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggNDUgPj4Kc3RyZWFtCnicMzK3UDBQsDQBEoYWJgrmZgYKKYZclhBWLhdMLAfMAtGWcAoingYAn30MtQplbmRzdHJlYW0KZW5kb2JqCjI1IDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggMjU1ID4+CnN0cmVhbQp4nEWRS5IDIAhE956CI4D85DyZmlVy/+00mEw2dpeo/YRKI6YSLOcUeTD9yPLNZLbptRyrnY0CiiIUzOQq9FiB1Z0p4sy1RLX1sTJy3Okdg+IN566cVLK4UcY6qjoVOKbnyvqq7vy4LMq+I4cyBWzWOQ42cOW2YYwTo81Wd4f7RJCnk6mj4naQbPiDk8a+ytUVuE42++olGAeCfqEJTPJNoHWGQOPmKXpyCfbxcbvzQLC3vAmkbAjkyBCMDkG7Tq5/cev83v86w53n2gxXjnfxO0xru+MvMcmKuYBF7hTU8z0XresMHe/JmWNy031D51ywy91Bps/8H+v3D1CKZogKZW5kc3RyZWFtCmVuZG9iagoyNiAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDE2MSA+PgpzdHJlYW0KeJxFkEsSwyAMQ/ecQkfwRwZ8nnS6Su+/rSFNs4CnsUAGdycEqbUFE9EFL21Lugs+WwnOxnjoNm41EuQEdYBWpONolFJ9ucVplXTxaDZzKwutEx1mDnqUoxmgEDoV3u2i5HKm7s75R3D1X/VHse6czcTAZOUOhGb1Ke58mx1RXd1kf9JjbtZrfxX2qrC0rKXlhNvOXTOgBO6pHO39BalzOoQKZW5kc3RyZWFtCmVuZG9iagoyNyAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDIxNCA+PgpzdHJlYW0KeJw9ULsRQzEI6z0FC+TOfO03z8uly/5tJJykQjZCEpSaTMmUhzrKkqwpTx0+S2KHvIflbmQ2JSpFL5OwJffQCvF9ieYU993VlrNDNJdoOX4LMyqqGx3TSzaacCoTuqDcwzP6DW10A1aHHrFbINCkYNe2IHLHDxgMwZkTiyIMSk0G/61y91Lc7z0cb6KIlHTwrvnl9MvPLbxOPY5Eur35imtxpjoKRHBGavKKdGHFsshDpNUENT0Da7UArt56+TdoR3QZgOwTieM0pRxD/9a4x+sDh4pS9AplbmRzdHJlYW0KZW5kb2JqCjI4IDAgb2JqCjw8IC9GaWx0ZXIgL0ZsYXRlRGVjb2RlIC9MZW5ndGggODAgPj4Kc3RyZWFtCnicRYy7DcAwCER7pmAEfiZmnyiVs38bIErccE+6e7g6EjJT3mGGhwSeDCyGU/EGmaNgNbhGUo2d7KOwbl91geZ6U6v19wcqT3Z2cT3Nyxn0CmVuZHN0cmVhbQplbmRvYmoKMjkgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAyMzYgPj4Kc3RyZWFtCnicTVBLbkQhDNtzilzgSSQhAc5D1VXn/tuxw1TtKoYYf0gP6bJVHutTYnWJ7PKlTZfKMnkVqOVP2/9RDAJu/9DIQbS3jJ1i5hLWxcIkPOU0Ixsn1ywfjztPG2aFxsSN450uGWCfFgE1W5XNgTltOjdAupAat6qz3mRQDCLqQs0Hky6cp9GXiDmeqGBKdya1kBtcPtWhA3FavQq5Y4uTb8QcWaHAYdBMcdZfAdaoybJZyCBJhiHOfaN7lAqNqMp5KxXCD5OhEfWG1aAGlbmFoqnlkvwd2gIwBbaMdekMSoGqAMHfKqd9vwEkjV1TCmVuZHN0cmVhbQplbmRvYmoKMzAgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAzMzIgPj4Kc3RyZWFtCnicLVI5jiQxDMv9Cn5gAOvy8Z4eTNT7/3RJVQUFqmzLPORyw0QlfiyQ21Fr4tdGZqDC8K+rzIXvSNvIOohryEVcyZbCZ0Qs5DHEPMSC79v4GR75rMzJswfGL9n3GVbsqQnLQsaLM7TDKo7DKsixYOsiqnt4U6TDqSTY44v/PsVzF4IWviNowC/556sjeL6kRdo9Ztu0Ww+WaUeVFJaD7WnOy+RL6yxXx+P5INneFTtCaleAojB3xnkujjJtZURrYWeDpMbF9ubYj6UEXejGZaQ4AvmZKsIDSprMbKIg/sjpIacyEKau6Uont1EVd+rJXLO5vJ1JMlv3RYrNFM7rwpn1d5gyq807eZYTpU5F+Bl7tgQNnePq2WuZhUa3OcErJXw2dnpy8r2aWQ/JqUhIFdO6Ck6jyBRL2Jb4moqa0tTL8N+X9xl//wEz4nwBCmVuZHN0cmVhbQplbmRvYmoKMzEgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAxNyA+PgpzdHJlYW0KeJwzNrRQMIDDFEMuABqUAuwKZW5kc3RyZWFtCmVuZG9iagozMiAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDg3ID4+CnN0cmVhbQp4nDVNuRHAMAjrmYIRzKPY7JNL5ezfBuy4QTp9IJQba+QBguGdbyH4pi8ZhHUITyq7JTpsoYazCpKJ4Vc2eFWuiva1konsbKYx2KBl+tHOt0nPB6XeG5gKZW5kc3RyZWFtCmVuZG9iagozMyAwIG9iago8PCAvRmlsdGVyIC9GbGF0ZURlY29kZSAvTGVuZ3RoIDEzOCA+PgpzdHJlYW0KeJw9j0EOAzEIA+95hT8QKXZCWN6zVU/b/19Lmt1e0AiMMRZCQ2+oag6bgg3Hi6VLqNbwKYqJSg7ImWAOpaTSHWeRemI4GNwetBvO4rHp+hG7klZ90OZGuiVogkfsU2nclnETxAM1Beop6lyjvBC5n6lX2DSS3bSykms4pt+956nr/9NV3l9f3y6MCmVuZHN0cmVhbQplbmRvYmoKMzQgMCBvYmoKPDwgL0ZpbHRlciAvRmxhdGVEZWNvZGUgL0xlbmd0aCAyMTAgPj4Kc3RyZWFtCnicNVDLDUMxCLtnChaoFAKBZJ5WvXX/a23QO2ER/0JYyJQIeanJzinpSz46TA+2Lr+xIgutdSXsypognivvoZmysdHY4mBwGiZegBY3YOhpjRo1dOGCpi6VQoHFJfCZfHV76L5PGXhqGXJ2BBFDyWAJaroWTVi0PJ+QTgHi/37D7i3koZLzyp4b+Ruc7fA7s27hJ2p2ItFyFTLUszTHGAgTRR48eUWmcOKz1nfVNBLUZgtOlgGuTj+MDgBgIl5ZgOyuRDlL0o6ln2+8x/cPQABTtAplbmRzdHJlYW0KZW5kb2JqCjE1IDAgb2JqCjw8IC9CYXNlRm9udCAvRGVqYVZ1U2FucyAvQ2hhclByb2NzIDE2IDAgUgovRW5jb2RpbmcgPDwKL0RpZmZlcmVuY2VzIFsgMzIgL3NwYWNlIDQ4IC96ZXJvIC9vbmUgNjUgL0EgNjcgL0MgNzAgL0YgNzMgL0kgODIgL1IgOTcgL2EgMTAxIC9lIDEwOAovbCAvbSAvbiAvbyAvcCAxMTUgL3MgMTIwIC94IC95IF0KL1R5cGUgL0VuY29kaW5nID4+Ci9GaXJzdENoYXIgMCAvRm9udEJCb3ggWyAtMTAyMSAtNDYzIDE3OTQgMTIzMyBdIC9Gb250RGVzY3JpcHRvciAxNCAwIFIKL0ZvbnRNYXRyaXggWyAwLjAwMSAwIDAgMC4wMDEgMCAwIF0gL0xhc3RDaGFyIDI1NSAvTmFtZSAvRGVqYVZ1U2FucwovU3VidHlwZSAvVHlwZTMgL1R5cGUgL0ZvbnQgL1dpZHRocyAxMyAwIFIgPj4KZW5kb2JqCjE0IDAgb2JqCjw8IC9Bc2NlbnQgOTI5IC9DYXBIZWlnaHQgMCAvRGVzY2VudCAtMjM2IC9GbGFncyAzMgovRm9udEJCb3ggWyAtMTAyMSAtNDYzIDE3OTQgMTIzMyBdIC9Gb250TmFtZSAvRGVqYVZ1U2FucyAvSXRhbGljQW5nbGUgMAovTWF4V2lkdGggMTM0MiAvU3RlbVYgMCAvVHlwZSAvRm9udERlc2NyaXB0b3IgL1hIZWlnaHQgMCA+PgplbmRvYmoKMTMgMCBvYmoKWyA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMAo2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDYwMCA2MDAgNjAwIDMxOCA0MDEgNDYwIDgzOCA2MzYKOTUwIDc4MCAyNzUgMzkwIDM5MCA1MDAgODM4IDMxOCAzNjEgMzE4IDMzNyA2MzYgNjM2IDYzNiA2MzYgNjM2IDYzNiA2MzYgNjM2CjYzNiA2MzYgMzM3IDMzNyA4MzggODM4IDgzOCA1MzEgMTAwMCA2ODQgNjg2IDY5OCA3NzAgNjMyIDU3NSA3NzUgNzUyIDI5NQoyOTUgNjU2IDU1NyA4NjMgNzQ4IDc4NyA2MDMgNzg3IDY5NSA2MzUgNjExIDczMiA2ODQgOTg5IDY4NSA2MTEgNjg1IDM5MCAzMzcKMzkwIDgzOCA1MDAgNTAwIDYxMyA2MzUgNTUwIDYzNSA2MTUgMzUyIDYzNSA2MzQgMjc4IDI3OCA1NzkgMjc4IDk3NCA2MzQgNjEyCjYzNSA2MzUgNDExIDUyMSAzOTIgNjM0IDU5MiA4MTggNTkyIDU5MiA1MjUgNjM2IDMzNyA2MzYgODM4IDYwMCA2MzYgNjAwIDMxOAozNTIgNTE4IDEwMDAgNTAwIDUwMCA1MDAgMTM0MiA2MzUgNDAwIDEwNzAgNjAwIDY4NSA2MDAgNjAwIDMxOCAzMTggNTE4IDUxOAo1OTAgNTAwIDEwMDAgNTAwIDEwMDAgNTIxIDQwMCAxMDIzIDYwMCA1MjUgNjExIDMxOCA0MDEgNjM2IDYzNiA2MzYgNjM2IDMzNwo1MDAgNTAwIDEwMDAgNDcxIDYxMiA4MzggMzYxIDEwMDAgNTAwIDUwMCA4MzggNDAxIDQwMSA1MDAgNjM2IDYzNiAzMTggNTAwCjQwMSA0NzEgNjEyIDk2OSA5NjkgOTY5IDUzMSA2ODQgNjg0IDY4NCA2ODQgNjg0IDY4NCA5NzQgNjk4IDYzMiA2MzIgNjMyIDYzMgoyOTUgMjk1IDI5NSAyOTUgNzc1IDc0OCA3ODcgNzg3IDc4NyA3ODcgNzg3IDgzOCA3ODcgNzMyIDczMiA3MzIgNzMyIDYxMSA2MDUKNjMwIDYxMyA2MTMgNjEzIDYxMyA2MTMgNjEzIDk4MiA1NTAgNjE1IDYxNSA2MTUgNjE1IDI3OCAyNzggMjc4IDI3OCA2MTIgNjM0CjYxMiA2MTIgNjEyIDYxMiA2MTIgODM4IDYxMiA2MzQgNjM0IDYzNCA2MzQgNTkyIDYzNSA1OTIgXQplbmRvYmoKMTYgMCBvYmoKPDwgL0EgMTcgMCBSIC9DIDE4IDAgUiAvRiAxOSAwIFIgL0kgMjAgMCBSIC9SIDIxIDAgUiAvYSAyMiAwIFIgL2UgMjMgMCBSCi9sIDI0IDAgUiAvbSAyNSAwIFIgL24gMjYgMCBSIC9vIDI3IDAgUiAvb25lIDI4IDAgUiAvcCAyOSAwIFIgL3MgMzAgMCBSCi9zcGFjZSAzMSAwIFIgL3ggMzIgMCBSIC95IDMzIDAgUiAvemVybyAzNCAwIFIgPj4KZW5kb2JqCjMgMCBvYmoKPDwgL0YxIDE1IDAgUiA+PgplbmRvYmoKNCAwIG9iago8PCAvQTEgPDwgL0NBIDAgL1R5cGUgL0V4dEdTdGF0ZSAvY2EgMSA+PgovQTIgPDwgL0NBIDEgL1R5cGUgL0V4dEdTdGF0ZSAvY2EgMSA+PiA+PgplbmRvYmoKNSAwIG9iago8PCA+PgplbmRvYmoKNiAwIG9iago8PCA+PgplbmRvYmoKNyAwIG9iago8PCAvSTEgMTIgMCBSID4+CmVuZG9iagoxMiAwIG9iago8PCAvQml0c1BlckNvbXBvbmVudCA4IC9Db2xvclNwYWNlIC9EZXZpY2VSR0IKL0RlY29kZVBhcm1zIDw8IC9Db2xvcnMgMyAvQ29sdW1ucyA2NzAgL1ByZWRpY3RvciAxMCA+PgovRmlsdGVyIC9GbGF0ZURlY29kZSAvSGVpZ2h0IDI3MSAvTGVuZ3RoIDM1IDAgUiAvU3VidHlwZSAvSW1hZ2UKL1R5cGUgL1hPYmplY3QgL1dpZHRoIDY3MCA+PgpzdHJlYW0KeJzs/WmPNEuyHgba4u4RuVS9y9m6+172FSWRkERgRiMO5gt/HH8evwigAFEj3YW3T3ef9V1qycyI8MWW+eARWVXve7rBAQYYgDiOs1RlVmVFhJvb8thjZvjv//2/h1/Xr+vX9ev6df26fl3/tSz6//cF/Lp+Xb+uX9ev69f16/r/5frVtP+6fl2/rl/Xr+vX9V/V+tW0/7p+Xb+uX9ev69f1X9UKz7+Zvv4flte/RwBwR0AAcHAAp/4VkgM6OgEwAoGRm7bS8qwqiERx4GGPHA0JAMABABBg+5z+wtPy65vg4IDgBEjoCJrNP1QsCgGNwBHBAQERXdEV3R3JkR0ZEAHxi9Ofvrn7x/6xKaV/9+/+XUrps5v1Ty7B/fl75ia1LbUtbiKtPDx8vL+/u7u7u79/OJ3O4PDl129/+9vf/ot/8fvXr94SkZmrtlprrQURYozjcDMOr0OISASA7gSAAIiIT8/i6WKua31DVf/Df/gP0zT1bx9+87cffv8vEfpWID77VQQABLw+Qvz003/pT714bX3VHcARHMAADMHQDd37ZSuyITuxAxE4OdD6We7Pf3+9IgRAcHeHw937b/7p/+x/hoj+1b/+VylGVTdzBHQAcO+ihf3GrleIBP1hISICMSCimZqqiZoZAJrpZZpKKYE5xhgCx5iGYQghAoCBmQoRpSEGDojcb/y614iwPtF+/euy/q+ZmamZObi7f/uH7+Z56T/69quv/9v/8d9st3o9H95vaH2iCNvdgXv/I6a65OVjnu/my+M8T0uuAD5EjoEJsVaZ5gwAHJmZkSjnsiyFmVII/WE09So+L7UKHI5vXr3+4u1X39y+ekMhAbEBORIA9idr7tovYL0yrJfT/T/8p+0RYN7/t0Z7Xw/ds8fy7Km8EJRfkN7tBe8PAt1xO18O4Js8/OLyz9+4ylJXBb695k9/BQBgqN8FPfVfOZ8v/+//65+QiJgBQc1VtWlTETclohCYORKxiahUEzE34ITpkG7epuOb4fiahz0gOgKCgRuYSC11mbRmawVMCcxVTZtrA2tuzVXM3QEAA3KMwy6O+/H4Ku0OQGTuLc9tmer0KMvFpIGpuwMyUiBiRNqN4//j3/wrojWs+te/+bu3x9fgDu4EiL4J5bovXXrd3N3BfXsJAQmI0N1Vu8haF0tdl6kbIBJxijGGAA7uJqpNpNRWW1URdyfmGOMwDMwBAMzdVNXU1AwMHBABCYmYmQEBHPo5uZsfTuXS74KH/e1/979sl/9CXf3Cbvt1u6+q7a/8/PrDaE7mDubo3q/JcdUn/ROx68SrUtw0yefC9lzzazv90//q0vq3X//tv3771e+7TvhcSeMv/Q9WdbXdxtNX/T8OL178hTvtm/zXHgA+++SnV9bvCIEQf/r+j3/6w99f339h2pc3f/f4+/8XuKM7OgKAowMAu5GDIjuSIxB6QmMXlNKm0/TwQVoBDml/O9x+iWlvHAEIHLdHDQ5uf8G0u69vokNEZDTCemr+p5nOzQesAWzbf0SrJBkAnILzoDwCBSdCgKtpjzH+23/7b/f7/WeP55m6g65h+glCADBvIsvp8vEy3WldlvkUwsOyzKrvz+cffv75PQDcvPq7/f7V739//N3vvmFiVas1L8s8zRMRjON4c/z69e3fDsMxxEAYHQMCA1C3U7Dtn18v5pllBoDW2n/8j//xatrPX3z9/f/0PyPi9aBcRbY/D0RH9Gfist5bF++rVKFff/vqS2174o6uAAag6ALeyBppAwcAEkoaBg0DUCRABuTrU3Mw8O4RrKYdEYDc3dS++vYfn5l2/Lu/+/1+t2/VVB2BNgNkjt6lBAG6FgJkQEJaNRczIIG22kpptUoTRBSRDx/gDBZTGochpbTf729uXg3jCIhm0loJgY43x2EYmRMCbyZ3VVKIQIjQtWnXUqb9X1EVaaLibmb604/vr6b99s3b/+F//rebD7p+gIMBOnY95Q6bInZ3AwYHB2nl7vH+Pz/e54cPD3C/VJ/AbdynIQYGAsxzvgBiHBKHAERaL3M7Jwghjoho7tV8qnp3nqdsbwnTm9fx9c2rv/mbtD9y2jt3SevW3dS9uWm/MkBAnN/99PAP/+l6+trwty2+9X4ir87hcw24WtpnLuQzdXRVKKs8OV1Nu4Otzl3fZYAXqsj7Nlw97E//oF09g6t/1J2/desg6j1spn2alv/9//h7DhxSMkA1K7WUWlorqhJjTMMQ48DEUhbJs9Ss6h73dHi7//rvDl8PN199kW4PTtwVALq41LrofLI2F80TamVXl6o1W8sus7XFNJu5AQEnCuNweLW7fXvzxbh/ffTAqloubXl4nN5/nx/ea55dKjgiB4474oSIr29v/u//43+/WXb4/Ze/+5df/a2bgXnottHMzNwdEBHRHczNzK0/oNW0OzGGQGpWWxOVpl1PgjSprdXWmioRhxj3u904DGhgqqXVpZTLPE/LUjGbWQhhN+6Ox5uYEiCpqrTWpImImgEYEhFTjDHGCIjubqImUqReTTvF4ea/+b8BB3NzeKlpPl1Pu7tK3pOF7G9/agS7U4NqrOquRuZIQAGBcAs8HQEIe7iAALhaH3xp2fHpA7dvtC7nP/xvV9P+9qvf/92//n+6ma/+/XaN/Yw8OQubGl6lE5/FJV05d+fCr6L/iQeDz5Q64jXEenr7esn9qTz7zBcfgICMwIgI8BdN+3aG+904AFr/fUcAIFyfIIEBiLZLO31Y7n48v/uu1QVD2r/5mqnF4xeUDkDJgRGIsD96QHD7fIevOw2raUB0AmfWgV20jTZHq2QK3W0tky0PjgjpALvXtvvCaOdE9Nd9vud3+PlydACRNi/nDx///PO7by/nh/Pjw4f37z5+fP94elTP4w6YKcRa5P79h2/VLoioKrWVWmtrAuAhhJvj3TTNN8c3h8Nx3N2kdERKCIwQnm3Vc+3mzy7s0+QIgvcoGbcQ8aWB71qvG1bsPstTEAYG7gi+QgerJOOTRK8xuoALWI9Iiuviy6MtJ2oF3eJwCLtb37+18RXwDikiEDr6Jtbum6+0CrSjG6IiPNtqB1AARTQix26Brs8BcdPduDkHgOBg4CYqImatLFOeL8s01Zw5kJl+/PDhdH5EQA4cQ9jvD6/ffLnfHwFJpC3LFFP84u3b483tbjwETojcFzGiE61/rMfWBm7gBG4ABsQQCIBqK20Lz9b7cLeuYhwcjBDdzcGuB/i6n096Ax1MVEspl5InNfFN3N281ZZFShFEikMcxp2YLaUspTVRIqpqptqa5OrzopfLcp5aqT9clpprfTg9fvmb391+8fV4eB3SHhDB0Ry0R3jrJVyf7wvBwqvm3WKLFz+FVy9l8xM/+fVnNwt9r3sg0KUUr0CM+ioJiMAbPmP9xU0SNg/hxZ9yd0RctcLVtD9f5lZqZQuG5ACqIq1qq+DIYQjjPgw7ABcRbaKtmoi7mzSvpeWlLrO0ElQACYAc2AGBGYYQXw20e+NSSIVcQJq3Ym2xNptk04JIQBEwOidOuzAebP+mphtg9mCkmJrVaZZczdmpEnNMY9odOQRVjfvd8wcaKCROhuroAZAcDNVwjdIBwMCxSxl1HdB9IANHUReVKq1KayL9N0SkSWuiakYcAEE1qgYwN1FprT8ob9VaM1VRLe6MqG0gDmbWWmutiUh3J4iJAgMAEQGim6vIhqI9k4vP9+mX1fKnr+IWfPhqkTdV99z363GEC7iCGRIjILijmyMiISI+wUv9QTy5EE8G+KpsrpfyzLMAAHBTV3l2hq7Bfz84V3/hycXdXttEGQE3SPX5M9lC82f+wXZB+CyA29wGhOffbgbgOeh4/YuMyIiBXzzVT0y7A6165+qVbOpPcX1GiK5mi0wfpo9/nH7+9vzjH1qZKY3eTsPIiEL2ltKBwogYiVYDZJv38UxZPu31qgAIEIHQo+mATW0Zyn1sU7CGKqYi00O7vHcKuH+N/jsYRoNgHPi/nDPg1x/16/9MbVnmh8f3P/70h2//9H883H04PTyeT5d5nltrZi0mj8mB8rJ8/Ol9OE/v3N1MRZt1XxoBEc/n++kyvXr15atXb16//vr29usU90xjN1rPXNlfNO2fSzzQup2I/rTxuNp7R3Bcw+Zu2nvGBLpF7+g6uZEr9agK0ZFWyNMUrJFV1IJa3IpJ1nLRx5/19I7LhU1of8M3X8KbvwX8rQ1vnI4AyZG75u/xdr9uh+5GKLiACXl7cR+G5EQACIQYnp2mFULrYu39LrqDYE3ESs2lTMv0OF8ep9MpL3NMAcEf7j48nh5UxMGZabc7XC7nw+GWKLTWLtM5paEu+fXrcjzWcdzHkIZh3O0YiQiQgK7CbU7oRl1xoAMaILuj16ZiL1M2oG79PBCgO7qbdbgat1vCq0rBFRewZi3XMuU81VZF1czRwczNpOSs6sSBQwwhliVfpjzNeSnNDMBJRHKuuei86OWSL5fy8Lh8uHt8vJzvHu9+n+e/cXtL4RgSEAOArZjtJlMrnPqZXG2+4HPhwyf5vL68/QK8EE/EVQTAHdF9lUNcZRYA+6N1RRAAB+xagBw63KP9A7t1f9I1/fE9XQY+PxgIL86IAzQHc0QgdzNVEwEzComH/XB8lXZHLUvTizn0INhXr8pd1a3/Yz27Z4COARBgGMNwiG7ohqZkgiIgTVs2WUyLmzAHDtGBDdgpQIg47jSNSAxgDAQGQ84qZjya1BDTsNvtjrdE3GqJQ3h+I0whcDQkBw/YTTsaqEFPFNkKPwMSIgDYCte7mTtAE6mtltaatO56ioiIqLk5BEQ1EpFGFcxVtHW7LgIqoOKqptLMsoM2CTG6Q2tVRUXF3A0BjdidiEIIiGhmqqqiPT5/Lld9C5826mVm50XogVtw8MK0u+MWTz7JAeKKPKp7dRN3Bw5E2B8GIBNGoI5hrqbE4RpyA66e4hUuxWcS9anaJTAC9S1VebWy3XTTug0vLC9sqCw+iaxvFvop2oYns71i0LihCtQ9cES6gjJXf+Lq3q6x0KdvIUAHh+NfNe2IBGiItP0WAljDdrE6qTQwR2K31vIpP/48vft2ef9dffjRtdIwykDz+9jKRLv78fbr/euvmQ9d4/T9WZFbv4LST48WARyBEBkhoieoqUzl8YPcf4flgVFcSi1Fl5POjxAHOr5lgDAeMUanlPATGfp8PccAr/91AHPX2vKHjz/+6bu//8Mf//G7779dpqmV5k4p7kVm1aqqXuvl8hiim+k4jo6O6MxAhIgUAocQLostS75/fHc43Hzz9b/47W/Lq9uvD7s3iATAL607PDPwz6/w+W4AMeIWBOGKT/a9dQInMAIhNwB07AlX6sgoghMogJBVkoW8ESggAoWOh2vNWheoF2pT8MbWasllnur9h3b6CGVCExz3fHOKVaMCvw5Ig3E07AKC6OgA7ABg6ApWwYpL9papPj67DaDAHAIgujFCeI5PbA7K9RF4Dz41t7qcHh4/PD58LHluec7zLK2O48ABXSu6mBZVUQST0mp5HPYxjgZQcw4hmNTL+bzf3RyPr29vX799+/Z43McYgByRsEMODmCGSE4O1uUBCZDZmQLhi9Ph4ObekQpfc8t+tT/dpXq+seQEZiamTbRqze30OF/OS6kSCDQiurkbIiKRNDk9To+X+eP9+XSe5iUnDmOKiKDml7lcpparq0Kppc15yctlOjdTA4hpP47HmCIRd/wGEbvntbp6L7UXIRohOj2HJZ+ZTYTnOhmf7um5ZHZr3XM67gZga+gECN4BPjNo5hXREYkQARgA7Mn98M2pIwdEMAdCcN+029PlPWUNnl0Dx3T7FccYUzKpmi+AFEII+5t083Z89VXa35b5nB8/du4MABMg727SzdvDF1/v3n457A4colMAYlrBJ0fsHisRIDijRVd3NdSjmYIroDEBM8IGmAERcEQmR3AnTCMfXoEZD/thmdw0pCENQxz37kB5juHFnXS3sMsPdZIOMhKgq7rTGlAxETiSuoOtKF3PJom22oqodiQJiNjJnTggEBESEogUk2pmrqYqpkLoMZAbK3ZehrVaVDVI6sbb3ZmIEQwRGJEIAFR1owg44MsUKwIRAncQ1NcA+gq6P/+5F85cD0K6lHW7/hk+g/0z0UBFFpNmDhQTE5i7qlFMHAOuph23c+jYCSf+XJy73PkmVhv159lKkY4Dqbta59H0/aH10mkzxy/y7NDPA10d5O5I4HoUX7oSW7i+fcwVyaT1sD7HGFb/4OpbPH9rSwwAIzDAJ/HtC+VFCNzdD7vCCwZWdb5rlw8tTyZCHE1bnu7mh3fLx+/b43uYHggNUezCMwFNZ9q9RdPD4UApIUZHega0bc8W4AU8DAgIDBAQErpZS/lEDz+1n/4Z5veBxVouedEyQ8s07LDOlHbD7VvaHXB/SPgJ2P9X1vOMBrpBzsvj6cMPP377n//57//83bfv3v/sqgw8DjeBI7iYLGboYMuSicHU0xAdjAOOY4yRiDClhDSWWvP0ABBiHEuZkdzNCGlIyMyI8aWP+EmQ9Onq2/Hk2rlvPqQjOLuyFdZM3hAAiJAC9IPXM+imqBVlxnpBXcgrEmGIAAhuUmZZzracMJ8TtODic6nn2R/Perm4VHfDZeEmFMaQ9mF4RcNRKRlFAzTATm4ENHRFq6CTl7OVs5UzLncvboSZQgAkcAYI/bEDrAdgw6kcwMm7sJhbLfl8un/34d0PrRYTbbW6KrjEyKaC3uHFCuCtlXlekNM47olYVZm5tXK5nIe0f/N6Bofj8UCMHGmLC3vKAhypm2vHTpwhWLE9RuIXiTqHq2nvDAdf6aa44iZXce6+tQOYmTST5mpSbbqUy6WYOyQyc0JHAgREotZsXsrlNF/Oy+Wc53kJRDXGYUwUKJd2mRY1FqVaJZdyOj+eLydADHF48/abV6++jGGgEK8ChfCUvvsE0iImZnK3zYZudLvPJNABVk/mxbvXD16dYwJzULoyIB3Q3VFRG0JFMAKmDvYgIri5rTkd7yhaVzbkaNfnCJtph2d41YtgNw77L/+GmDlEyRcAJGY0GW5ej6++HN98E45veDwBRTFTc2sFAYfj692br2++/Gb35kvcHSBGQ96AXCdURCNUWrlZ6ETG7JbciYA6u4VJmYSwoQuAArh7dRfofMZARjtiDoebsRZ35zhwTBSSmUGeEzbEJ1irx+awGb2n4MMBNyeSiHruqv8sOUBPyvfoXRXcCBGJCNcsHAXmwN6pc7WoqGlHKQDcCSEweWBCNzNTF2mqZmZEjIhdSJAZiJy6d0YdL3gyV59YbAIgxCemHzxXttdXVuvrm2kHo9XL7r/p6ztr3gE6S8kRDFR1kVbdOYAbgbo1kciAuCNGADSkTV5WUHPjbTwjcOBzE/CpJzEEPIwsamJunbP3lEPvpho/EcjrA6FnQTbCmiK4onoAm3le397C9y3m20jX209uj3BzDrpTsEV3T6E78NWxeLZemHYEIFyfPSGiO5lYPeeH76Z3f8jnOykLYHAzqZPki81nrAs6MHIA9LKUux/hcqLxcZcG/+JLHEYMEdboBzdnBrqd2vzxazQKASCCj+CuLSwXPN37/Ts5/7RANi21NTdhMHJljnG+G6YPfHiFchtM4K+tzzBGAAAws1rrTz999+0f/6///If/9Ic//uP9/YfLpYF5IHQFRqoLSYsxHkMUQlPhWl1V1RoHMLUQgMjbUM1MBedZpQEAm7moLEuupb15YzfHToh+kodnl/Qcw3q+HRi61BCuVOwtT48mqBPWE+YP2E7kGghjjNw9N1dXcWkuFeriZfK6uGRgpBhDCExoUrUsmi+WLyQNpNml2KVysWQQYqSYPEZMiV24XkJ5COXgFIxIMCkG7bvmDlZBzjbf6fmDTXe+PPhjfnknCEhIDM49YWhXWusV07+KCLi7q9VWp7Kc8vxYSzG11sTURBtHbq02sda8NYPuyYCStgIz9XjKGdHNtJaWUqryxkCQAIm6YVlttK+JX+v0+E6K7a+sVu2FDG0vbHQ5eIk5XCUMAdzNBayqZtfC6IzkSqprUQEiMjGEAECApGpEGJnHlDKXDBiIUoyHwy6OKTc7XcqSW87WZEUCFp0//PRuv7/95pu/e/366914jONewdVW/bAlfvAT084BKCEYrULWA3DvCnBTIrBaDr1ys1fD4+CKoGbqZh2f7D/SGQzuAo4Aht7QG1lDMCZlJCLtjxoBDINjWNWe05rcR3LYanN88ztW9bDK0XWFYffF3/535mjmy/lO1CBE9oZpb5wUI+HA+7cjD85D2L1yKYwwHG/Gm9e7t2/jcW8hOndrrYRKKIyNoaI3cDVVAzZIDoPx6BDNA6ITWuAaqQbM5FllsVZMHQyJEtKgvFdKNAZMgce9OwInoKgU3J3DPkABeHdVSm7aQWZ074x0U+28TrtmfPujRwczNMOVJEaOMMQoQ1I3R3R3N3N3U+GefQJXVfXaWjVTcGfkFd8mA3ryo6hTuTomgESEIYQQIwaG7lg8EyEiwsBXkv/2IgJ9GrV/GoKv+u5q+Ay3m+uXh8wd/jF3UHV3pK4FHcHMqltFSK5NqzeVqpUC90wjANDqHeGWrHxyMp4lB3AN6bfrfq57E8M+uhCIgRiou+NVT/WP9mcxe3+rp/vhass/+eeKUTwH2DeR9m17+/t+DeWfLvca3W10gs2lWC+B1r/+4lm/jNoBeL0JJwR0wzbr5a58/NP80z/mx3c1zw4MBq7VzQicDNAJnVxMZW7anM88LvnmdTn9joZ9SAMSANKG4K0ifc1FbMcXETCAR/dgyq3SMuN0gvOjPt6JXNRFwJmJmJErlpmXU1g+xvwFti+C1U8l6LkwPXMhn/bXXUWWefrxxz///T/87999/5/fv/95mUur6GYN1Fpj4lqRYNyNNIzuUIgdIblBrxtpwVVVtaQazdCUlkVqNRVQM9Gq4q4EMAzpiOlKCrg6vZ+DnS8ue2V8ETigIRgggJEpeaF2ouVnuHyP+T1pZbKUYmRCdHe1VrVWLcVK9pKtZq2FCDEFHyKlSK4sTWvWkl2a1QZTxUWCxYBDSimMCULwRGYZ8z3Me0qBmJwQAyISABk6uoGqW4W2QD7DdOfnDzA5wPD8VtbbWNOF0EGWrs6vEr86fu5urtpKXmqZa55qLSLWmqp508Y1ALgaqLroFqT2IEYVAZARHM1aa9ZA5rzLZSplqa1wjADkjubo1tOWYD0tDFc92r11MPDPd8afhblX2YUN8lvjATd0c2sgs9SztMm0rZ6CE6AjMnOIAQL3gxkQHYDMXM2liZmmEI673eF4CGM6TZkDi9RlqQ5r7hXN2rLMj4+X+7v58UG/+ob8cOVe9ZDA+ue+DK9WFUybDV9P/BqeX9HGNY3ei6oc0JGwlx1ZJ12aNmAi4H735IRA1gsn0AEEvaBXdCUgZiTQra6SHRnoqoVX1XZVmU92HTaw6uVJBgDmsL/9QgxEXRyaKsgcoFEcMI1OwZB4TDwOFGLaH9Eak6fdkHZj2EeMClgcFIEYnakRVIZMnsmqSW1N3BlxB3zgcHQciRKAEwp7ZlsYZ/YJ6sXLhM3AiHgk3nu8sXDEsAtxgBgdk2ISSIYMDhSUMQO+fzLtbmbSXSuD7k/ZZtdhRTo6lmGO4OTeAU5EdCQLwVJqpuouIk1VW2utAhqQOpqZtrbUVqDjN9hJVwT9zjd4/JoEsU6LIiBGDsQpArEj2pW3D449DfnClmz44vUobwfkE+PugCsVFwChR9nuZq7NTAnXglU0c2nu4KxIjETYc+1ewVFNteVm0lzibofotAGA19B3C4w/vwRcS7/6DvCLdxl9YGdwRmB08RVb3mDs/rVvxhquNrnHzU+2nNZQHq90nGu4/uKxbVe8oaDP4vHNbj15Bi8O8mo/1uj/E8v+adSODOgr4u9gpZzfz+//OP387fz+Tzo/Qqvg1AsfkQjTQBidSFRLbWbNXDCqQ5wePnz48U+vOd0OY2SgkPrVsRirdahFAitzt7e0ORZkZiXrPMF8obxArZprK4uBw5CII3IECK4KdcHlkesjtxNbgf+ytZlTN7BcpofHDz/99N13f/72fDkHSjGQtVbalFteZmMqwzAcjjevX4/jHkRnQBmGARBazUg2DFTqfLmcmcUsIlApakrg3Gq7XE7v6Uc3juFw2L/GG8aREHhFfZ+8ON8C90+u1ZEVkXqqHoEYAF2DZW4TLh/w/AM8fgfTO5AFXDySMyK6m2ittZQ2ZylVq5ioqiIhMaYhDkPsYDOu/HAD0BB8HEC1uVkEjy4M0T2WOsllEs8gcwIhMmZ0QkPe0GgG24XxGA6vrS4yT4z1s6d+JQJ2K9F91Sdt4JvcOrgZ1qo5l1qaqkrTKq2bdgOL6ESE6EBduzkCBOYQQoyBmYmhMyDcwVxLWR4e7o43N8ebm9cG+/1N95U6Ct9Ld3v61wHXqN3dwOwTmlBfTzXHV1FCfzLw1nUzaIW2aDm1+W6Z7pb5VOriYBwYmdIQh5SGhAQJiYGiGaj6uFviENMYbm73icNuSBSoucbIIQRAVHN3Z8LDOO526fWb11+9uRnZoU2WT1qSAhoyciTkfqI7s+DlHWzwp18rhDbU8hqaw+aHIyLyqqzcOqpnbiBF60yRELgDIOszNTV3ZAIE9AreEJyQEc0xujNARGTCYMiA7EjXqp8VzX2C+rvsXK/+xW6oyHw/OSeIQxiPh5QIGqGCGyKGyCE0Co5EgUbbBe6hOQtiNVss92xyTHEIAckErJhMptm0qpRSmjobjpxuw/iK4wHDzsHBBHQSvZhNpBets9Zs6mAkODgm4QcNO97d8HAbhltkqpgcyCGAdxPwAmI0VzNZxcgcbSXIIWyMrf40fMtYrUDe+ogCYCBW1dpayTnnvCxLqQXZMQIyAHgr1URTjMQspggU4kDMyIGMoAF0nWyrZw0AZCSmaIamAOBA5ptpdyciok4MeDriHUjYYuWnHcRrBnY93x1+6C/aemuu2opoCxDJGXtNqjRzAEvAgZnRKnoDq2YqYq02Q8fIaEoddUDqVGLYLCb+onMBL7wAYHz+A+iGKv2DuOs3dEBfK5if7Oimu7YvnkftVzN8DbWfXxO8lG94FsNfL/ZFdH59wi/PwbNH/AvoyKdRO61tUBxBXXI9vZ/e//Hy/rv540/YMqquBTZmFGNggsCIrKrLvLhLCBgCEFLL8+P7H3i3H49HRAtwgBAAmZukLK5uSL4fLSIAkIPbar4AVFqRZfL57PPZSpZS21KNkDlZIHM2Jze3ljU/6nyH8wev06d39tdWr2Zu83K6v//p4f79+XwSEcKEDt4rlFxFqxPe3OwOx3G/H4cdNFGAkFJEwsAAqMTuBVvxhpaCAUKv3gbvuG8zRWlwe/PVV1/+bhh2cRgIEwCgkz/brF+w6wCITuwEhri6umBG2kKdeHmw0zu7+9Huf4LLO2sLeCM2YCQGd621laXky9KKaDPzFQQjgpaijJEDEyNRR8oRCZg9JVcxVw3o0YQtuLA4aIsqF7QMiSgSBAY24MEgOJC7qCsAEBISO32K1F396A7FwUpl3Q7CFrNt8Kw5oJlLT3b1iioHc1MzNCQjIuwxBwX0HvYGSiHEGDkwojshAqiDGzSpp/PDx4/7/f6ASESBw+COvobvq2vjz0zHNcf83JzgSiz/5MaeWr/0eN2taVukzFbOkh/a/H6+3E3TKedFTftxJyImTiGEAEjsyOao5k011XKE3bgbEoVI3KxJ0e25da0CIfCr2/Htm+OXX7z98svbQ1QrD/PpHWEzChh2cXfDcccU+jN+EV05rBWRq4UH8LUG8SlyXn9uXVuNmhEYuYA39AI6u1ysF9W4gouZgImZOgCEhMToCm5EgBBXZwKHngEgZEQywl4GtwU23Qwg4vOqw2c78OxGVOrl/h0N+3C45f1x3H3BgQjdNYNkQmMqhBWJISEmZAqM4F5MZ2kX1UpOjDH4GIzBRWVp5WItg4mpqJgBO46uFaCiLeSjA7iJt8nkYjKTziqLtWYKZmi+mAehCGFIOrNXIkdiokAUEaIToXunEz5tiKm5gK9FF1ck6ImWuCIXTj3x3m3M6gQbmIGbqtZal7xM02VZllwW8arQkAEJTAwd9rtdSknVAGgAjXGkXtNPVx53L+xc7ZkDmLuoY3d9e85qCxSJXppEAER6ops9p5U+fQXY2Xc9c6Ddp1NyMamtZpECHlgZXM1Umjgg8OAhALO2WWVpMptCK1JyQQ4J92jSG21tSX7cqsnhhYp9OrPPLw3sk3C3F0f0WrGVLgyAwHTFvbvXBStwQRtl/jmN7vrn8Um+n3kTW1CwfflZGP+XXBJ48fLzniafrU9pdAFBARwdVUzmen63fPx+eng/nx57sSUaAYChswMNLXEgAgBptQBRGvfj8fX+9dtGoUz384cfzkNq+ZKOr2i35xjjpeKlkCKGSPEN7wMAohF10jW5o6m1Vud2+VjPH+v02MpiTZ3JmglJ97kCxyayTKf2+HPgkPLyy/f32fJVSEVkOZ/fv//4J7F8e/t6uszLnPNSl2VKg+/3I0AIPNy+SrsDik2yNERBdhBhJuIeEZbWJMQBMYBHU29N8lyX3MaRjjeDNnQNp/PHy3R3vLk92MFpw9c+b83wcjFiWvkRZq4OZlq8Tj59lMef2/uf2vt37f6jXx6j5ghS2ULAkBgQRb0sPp28ZlNRRCRmJCQkV3YlIgdU8wao4xjGXURyQqeAyL3VYLVaOmGVMYBWRqcxESOhsS4hHZ1HB25l0dOjTvdyuYP53nUBN4D4JJEERNecqa2Ez03utyQWAqD7avhiTOM4xjSEEM0dOTiQQaMegqIjYgjUm+QgYmAKgUIgIvTeTiEwO6q6gy15uX+4izE5gCOM45E4MScO6ZpQ92cHHrrD//LIIyARufkGg8FW69F5trh6hJLz5WE+37f5XpZ7yfd5OV2maV5qa1KaWGuMJvuAzoHYAKpIqbqU9ni6PJ7PiJRCMvOmUkVqbjVLqwqAIUUmP+zi11/d/va3b7758s3tqz2FXKefPvy43N8fncfh8Pb2i7853GAYD8Ro8DJqRyCMiMnArjnspzLiNQXZNbgjGIIxKoKSG3hDq+jZfWY/k17AXAXAs2s2ya7NDQEDxT2FgciZEAMjuiqDRyRCYiJydEcFBwDyzliEK3rTAXn/RfN+XaZtubzDtg9QdpHi4ZbjPnAAW1zP6GfwyU286aqbCZwMPZPnYDm4IBB5xVqkgkitdcnzZK2hewxhGEbk6ESODepJLWtLnbWGXtAKQUVQBgQkBVOzJk1NkY3cPZMCVgMU19E9IgZCTKtp0Gc34q5m5N5DTwIw3Aq4wFauPoI59M5AV2vvZqqttpprXcoyl2Uq07lc5jzN82Uup6VeANyR3IAxHG924zi4GSKNw2FI+yHtA0ZTByAkprD+LQCgEClEIBYDMO/HFtE6WZLIAXQrNNsWIVBP43sv1EHAp8Kep1jTsTe8EzEzckWrLrnms7QCHl1Qtao0FXVkDiNzEPJWTvN018pkKtJUm4e0J0uozWs1HIAZeK29fzLdny38xJS+dNefAerdU0AnBELerDuuygGIoJfS9ppt2H78F8UVn+R73VqA6/l79rd/6Wqf1meBxV/I5QJ8ZtqdEQDdzECKLqd6/lBO72W5SG2gDgqoCgDO4ExBRKMgMyIic4hpPBx3h+O420NreT7Xh5+nQJqndvuWj8c4juFSwqUGiDgcuO0M9o7hyhLouDyigzUrZ5nva760klEFMLiokQpA4ACAplryrNOjxkH+ywny0I9nazJfpg8PDz+YlZvjTc16qnNr1b0OY7y5TYFTCGkcibjVtpiWEJEdmrg5D0MidEDjwIf9DUJgTqoaFTNoLYpgKZF7Bjg/Pn68u/9xf9jv9sMAHriXidAvi951O0yCZnBBbwHMQb0WzZNe3snDz/X+Y314bA+TTVmsCIpFjxFFGIjNqBaqlUtRbYaAHBAJiVDFpRnhatodGppHphCRGTvGvbXEEjV1dEICFUKAx4gI6ALlRMON884h6JL98b6d7uXyAHnCurQlAbx+JlerpXzeWmxDrVYT/8TBQGCmmOIwDikGJgpM0GkWglt47IhIjAyd6d3PZ2fCgbkiEvnKBhK1Wsv5cupekiPs9zcpHXb7m93uSByu0XD3n18m2F9sEMGWOlt5ZddCmrUSyaVqnvL0MJ0+lMtdm++lnUqezkudllqqiPZQxa8IhZjm3M5zOV2Wh9P5fJ5ijPv9IRGzY6ltWVopImpEvNvxbuA3r8bf/Pb17377+u2rfRpwWS7z+Xy+3BnvIB6Ob3IYbsbxSOM+IKp/khMFRAZisrXg7/l+AMKmrzuLwciFoJILuZgW18Vscp20XqzO7mpg7ovpIm0xEcKBw46dVpsSHIHBRBWAiUKgHnBiA2PC4JiQEtIWDuF6RAHwKfXx7NKff40xYCRkRGYKA/M+8ABEQMXETGaX2VvteU/oHgUKeGNv7p0o0sxAVJu0WnKds4qgIY1IkUIgB1erqsWEgMLq+7gACJCtlYZEsLZXdAdnhICOUmxB7e2bMQAlCgk4IDD6J6bdzDY3q8fjAOD21JJqtTKw+Z9dUq0jg6paW53Lcp7PD5fHh/PDNJ2n+Twvj7mczXukz0xxauNuFwGdiVLej/G4H49D2DOlyGMKFEJEYkAw8167D0hmuNV4G6KuZHETNbdPyMvkSFsfq2dXv8Xva3SK7h3cklatNdBK3lxyLZO0DB6UvNVZWjFzQI5pTxwApJTzcrmv9eLWXM2NGA1ksJa1zsgJMSARdfD3ysH/bH1q2n/hJxABrrG449rTlwk3QB4QAQm43zI+/9BPQa/tE1+4GrjV/7tffw/x09/6zG7jJ29e9eEv3MTLXDs6s6G7grR8qef7enmUPAWEcRikiGpnxjoAmpg21aCEzmm4eZPSsDve3MRhQC3YapTs00MG12VplwsfDrwbtZo23w03wxDQKotYZ2B2b4eQmeOQxsQIzWSROquUHp2BG5iCI4EzIrip1KAtuvJftZEvn06nW9VSL/PyMC0f3XVICQBqK0S229PhGG5u4zgOzKFJFWkOFcmYCQlaFTOKkWIKh7g/7Nk0ugVwMlPZ18AD+ADQwMzEa60P9x++//4PPVvx+rUdDoGp54R/eVfWJQvM77WerU2BIRCDAeXaTvf18UGXBZqBETi7kyMRIzFY14lIQICciFGFTM0aICiCIjUm5ADMQGzMjEbeECkwkpm7utlm4gzMmkFjRYDZ8E5UuJxhPDgPRtEgtixyyfU0LY8nnWfLOdEbGF8/l8getW8QmW+m3AA21gGgrRQwI/aYaBxDTIRk6k20mTckhZVQRE8Rp7m7i4mrqxkSGBgicmCkgMju5I615tPJzSznvNvf7Ha3b7/4+ssveRz3IUZHNLerG4xP6+WO9DfhmkvplAECB1WRmls5lfmxLudWLqWcS7lIXZacL3OdltaaIOFuHA6H3TgmImpS51xOU7l/nD7eT6fzNC1LCGE3l5txtxvGXOo019LUndLA+3388u3hm69v/uY3r96+2RFaKZfLNF8WqcqKI8Qbg/GrLxZWiW79eNPL4AB7OpIcVyfl6T79ui/o4ArQCDppvII1b1OrZ2mP2k5lvtQ8oxuBOVa1Wpuo0jjumfYUR+YAsJgUbeLITgVDo1ApJKQAwOAMPFI4Yjxir99ZdSRci6x8G2axydHT13Hcf/kv/41xxDTw8JrTDfGeMJBncFBtVidZLlpzZObAjKED9uDgRqbQCSgioqrmimaBCDE4OChYU4VuiAxAV7+YGJF7iYIx0MoEJEBHghAZMYxpjBxEvGmr86RKgfchHnlQIgMneqmyHcDA3RwQxYFXQhX5U2W2IWIncXQeY4fuO8KFRIa+1Hx3uv/p7t2Hu3eX+bzkyS27t63xSiDi6tNUKARgZqYUebeLN/vh9rh/fTzwLh7SMDAHB1A1RKIQ3JEQ3NYUDnXwRl1alVa1PRXCYAcvenrN/Ur5vh6Vjaa3pp/NVKTkeZIyMzTypm0yKUXBrZXlLC27O3OU8UDEZlLKeZ7uVWYAQXdwFsRWU8unWm4wDDEMDGH1g3DFna7n9rkF6Ij6dj2fhs6weVPUu9TTijtSz7jDNZtuvqYTN+TxEzj++SY/e31zYHGr+NtCnK3y86+hVS9y9FfA8ReixJemnQDJ2R1Vm2TLkywXLYURU4yg4M0NtF8c9qJBdVPnGMdxTOPIaXSwWrK1FlyhFZvODuxAps3rUAwQCJhdRqoFayVKENfCd0QkphgoBQgkBM2tmTZEciNQISZ2ZnDqD8aMTYI3dob/L5aVtpzP96fzx/PlThsj7kPwYaAQIwcbRuLowwghglyK6CIqiOCeEGjbDAqcYmKEYMKteimCiLvdABDQx5zn2iZTN9Xz5eGnn/8YoiFpU3Pgww45xc3f++WFLigzlJMv9w7WgW2oQtMHXD5yOyNkZLNIqBiIYsSQwHojOncOmFLorCoTddNNjRi4giMhBQpMCIo1Kxgau4ip9PJXIkYAFjMHVAFWVZm4KucZYwQiIzZgE8DidKl4WexcZMoyJBif3cgWoa+2c/OmtwhtTcOZSe8sa1JqWUwbgCH1/lRKZESrwnDYynzWVKObm6GraXdsOqBGHIgiIDtwE6211taWXHa7825/cYcYk7nt8UhMz6k/vwSMPfPLVzzeN7WO4G4qrS51uZT5seRTrVNtS9VqYOKwVF2KuOk4psNhd3Pcj+NAoLlprnXJeZqW02k6X+Y5FyIqWexguvdc5DKXJbcmGkO8Oaavvrz57Tev377dH/ahlEWLmIlqW3LNrYrXEG+h1QFgZGIiNAj46Z0g9bIjh7Xq/RnNoDteuPINCAStghaXojJLW6TmVnIrs5QLWkMURzWw1sAgOTJQMDfRxe1iOqlmd0LKyDOEASkSEWJETCHdxpGII8IAvRncKjDrn3+iB/kmSNviNB6+/tKInQPggLgDCL2MG1RdDRRc1GozFnd2UkcGdzd0A1dyM1NXERHp0kSAQGRqqppzDkK9VNNcERGZOASk0HsgO4FuiUQAIiLGwBxTjIxsIiZWtaoV3ElQYwByMBN8Gex29pmDqQMYGOHKK+x17Wvkux5a6Nn2/pMADmSIhpBbuT8/vr97/9PHn+fl0lqJwWPssT30pjPq3tRDQmZECIzDwnMeiwFwins/IO04MiChGPT0U4eYemmDNpNs0NxaK7mWLPLiRggBCMzXmS14TVNvpn1l13fkC8xNWst5uZAX9uqWQatpk5bL/NjqAuAc0iiZOJi0Ui7L/Gi6ECkCoDM6MIU6Hmu5CcM+2o48kPMKsuMvBLX4/L9XFP3T47FZWlq7yfSc+4tuthtp7oWC+C+JMDvr1qDkUkqV1twtpTQMKcTAYWW/Xi/vJYT47LsN+Xe/Npl4sT7tRgdM6EYCDEbavDVvjQCIg7N7sE4SCxwic3RCBRXDADEmB5pLMa0mhREiBSImol2I+2FwDmYOBA1tslLrlPI8lJzSLtCKDDIhOzBoQB2DDdECe8P1jxJhCMQQuI8KMCLE6Ba00dN+/dWFPSLQeT6///DTh4/vHh/uwTnGutv7N7+5acJqbL6UsgwjUojmVaQsSwUAZko4MKUQhhQPKQxEYOoicpmWx4dzGvjV68Nutx/j7cM9333MAEaApUx3d9kgL+W8lCZKv/k6pnRY2TB/6WKJQ4iIkTVAvlhdCBpqHsoHgo/OF4uLjWKkIMAE4w5ixLWBiCoDMoDF4Bpd3a23mTFVUWnM3AvcAUCkzCUvAIgoDZqAKDhyHIYQAxL3RAk1pgpcJWagWIEcCJABkEbgECCNnhsuNSR+4Wn1sHzLvnWsq8PZ6+iL7i21VpZ5XpZLns/LdDef70vJhBACOQS71pd1detm2huGghlArwkGBzADXRPgSIARKSIFQCbCJiqipdZSKjMhoWhzsGHcxRi3rPTae/+JaLatl7v1jBgPrlJanWu51HKu+VLK1LQAeYhjANSHUtXIMYT46ubw6vYwDKxS3NEMzUAVVFwbasMGJlLdsBYt4lNul6nkXNIN70a+OcSbY0qp97pFinF/cwSWovky18fzaT+eWe0Qwk1MGCOJxqtTtcoVwdqBewsf1jvqmnd1sx1o7fmhzVuWumhrbuCeAA+IFXF2EPcFAByROTLFEAixlXwRnVzPZrNrBUDCs8OgEAADUghhF9PNuIcQ9hj35AqgW+cveNGlBjYCwCfqiwKNr4GCcQAkAHZT1SKSXSs6Ix4IhdDcmoo2VFV0BXckSISMGIh6rKmtaU9CA7iTNW2lLkwUAoODmSIiBQoxcIi9WVKvoOAQOAQOMYZAFJgCAoiaGKihNDJg1MieIkQ0aLWYLs/1LhMFZnFUM3VA7QxP6FWKvvYshY4e9O/XCnQAAVcABc+tnKbTw+Xh4fxQWwY0CiFicNRevIjdypvp2h+WEEqTqqZIyJHH3RhSpBiYkHhrZrgWa/Y7zppPncsmNbdWlUfgJxee+mynVdGuRv2F7dvINd0jInIEdWu1zi4zQQWrLc81T2V5bHVx1xAH08JhcNNSpuXy6FYCA/YDr0bEtRykHK0eYNgDB6Su+Vb/HD9t6/DCQBLhZwNItgoE3KrZrmVsePUGrvXrvWBh658Iv7CuSaV+srpdb6Xdfbz/+OHj+XxWaW9ev3779vXt61f7w56YP3U2rh/19DxXxePrf/0zMP9TQB4IHc3MxFvWMkmepWRWQbPVNSV2N0ICQxcz6MUUMbSmZk1FtYK2yCGk3nXQGTx0KqSDuFZw0Sw1YJ1jXVB2pKHTGxkDuYJk1CVQS8FSgILeTNCdAoMJqlitFck9RmJZphaDxgPw4fNn8fkjcjc1mebLx48fHh/u53khBAcl5nEPVLEU1AYqLqLMaKYATtQTekPgnSMxJpNYHMxbb95Yay41I0eAIUSkmPgUVBCRmSOhmbVpekDUkA4x3ezG2/3uNsYj03X47Gci6I1sQS0u2fLFl0ezmWzm+oh+dsweqiczMCXr/o87OLoamIIpghIBrXxk6NWhpIqt4VqH03WZo6r1KKEJiWJVUqcGxBY4Bgor95OUuKKqUzViZTYKygE75DnsYjIOhjuM8OnNwLWr6JUh35VmrdVMHUxak1ZqXubLaTo/TufHkpd1BAUiITFD34hebr3qj5VwTgC9iaaamaluRQGOffYHKCIGUVVrIq01B2/S1AUJXuPbGG7w2cXC1aI8exHX7Prm7a9gds8sKFg1zbXO83y6XB6adN0US9MiKgaHGHfDsBtjigQAZiCGoqAKZojOxDFEFBVVWxapxcUgV6tV3Ww3xFc3u+M+pUDSWm26lNrEAgUKkaiBoyugQULecdjHgDE4YgovRGvtH09Xvx83697jju7A0JomwQZAaqZa1ap5AzSk3qwPxJppAQTHAMC0PgSv5VTKo8jFdAZXBCYsAEk9IkYKow+OmFSbu4ALggCEKyUJn3scV+P2iUAhYRiRAnLsouSmhmBCzinAgcE9KUhzU7XqYijuCu7MiETO5A7S84pNzNTW1lBgatpaI0S36AYqioQcqKfT1Uxkbe/GwUL0lBAdKZCbVAM1EGPHGNMIw00M+wARxVyKLRezBY5XhLYfia7wfW1n1Acor8KNtrZK32ouN/kTt6o6Lcv948PHh7u7x7vT5TSXxVyYwYB9ZQAo9PFH/RtVBAckBHHXojBljnMc511MMQ6JeVULhLam9K1qm1s+tele2mRtVqmmzcY3V9OOvUAGaQWUkaAnfK4sms3I9UTXhhWZmbS6SL0wNtRSlktZLi2fpGV3U6kAEOKAADXPeZ7cagrYYzRErDHWci75MaZ94LSZvAQcV/76L8Tlz6Xos6wbYi8ZXP2SF8a0k/Bh+wo2bjD60w9+0pShP4+eQFmdUxVd5vL+5w9/+uOf7j9+rDV/+cXbr7/56re//e3bL98O4xhSJGYi7N1F14/YnvN2ZJ8w+V+8vRemnQHQ3VWsLjI9lvNdvTzW+UIqtIqdUx/2oa4m2sxJjKv26qoQeikL9r5BrQUHQ6pSQssBESGoteotkBAitInqTCUCiqk68rA/sjUtk5Yze06sKSIzZG2IliChidacRctS0i4MEpu0vEy7V9/A618w7dftuoqXmbZapsv5/v7ucl6kOZFQM0Qw82WpJVcA4BCl9fYvFMM4xCGEcRx3TFEFRWC6WEdSKdh+HwFgHHquCtxbk6W1KuIhxhh2+wOPO3AUczmf7959+O54fHM83N4cgYfbXhjw/Aysq11g/gHzBesjtgdrD9rOKjNaRWuEBozO3nN/TWSemzRV7Z0uaG2igMgIIVBMSMwcGAkAuFartbseiIxpjJ0MmZRFOSpXJXEWCAqRMYQQOBAwOFh1YdUIyKhsRirghTnFgRPHMfFeBvisPSB2hAt73LFOE2mtXS6XVrO5hh73ETKA1Dpdpjzn1qSJNlVzQAwxphCiqosYoSqZqiHiOCZEKGWuGb2YufrKCXFHc12njfWB8WbWmtRaL9PZ3ULgPhm211Xj2qLyc0f4Gkpee7b1Wh9H8kgmbBlEJJ/ODx/u3qtWcAMKpdo0FXA6HA43hwMjaGvqUGtr4t20A1AIaTeGmLC0mkuV1koRNRRzdBhCeH1z/Ortq/1uUNVezzEtxdz3ux1iUPUQ0s1xf3t8NcQUmVMIHIMRDjE8OwEALuiCKzZ6zdIhwJXODAjep0mhK3r1tjiYWlabAYWCcCx9irm06o6AhuBAptVdueVzzadcziKZkIgCoREpIoaYmCLzyGFEjobuqI6CqIDxqQ/9U3S1wpPgnxKKiQwIelM2RAdkoERw48wsuygDgzAstSxSVZu4Gxp2jjtgj2NFpLWmrfb+b70DK4IT9cbyHY7vXY0AkJ3MVby1Pv/dtTUpqkklthgZiZq4OlE68jAeD1/G/Re8u6HAVmapVeZzhALHZ6Jl6i4bRoRPXA7qrP5Oad1Mu680eUAQ9aWW93cf//D9t3/64U/vPr67LJOB9iJmAxNTNzVXcDVwJCMFprVdUKeGAeaip0sO6ZJSirv9LnIgCBvxwcC1tanMpzrdt/neZXLNAEJu6DfPT0dwpD5R3VdvuoMLG+RyNbFISLDBCK4q0lrNChU01zpLW8xqr/43lTzPHCSG0GotS3atntaum8y1tVzLOc87wmgKO9HBjIcjpx1hYnpWQHj1FT+35c+WO5qvjsfKfMeuRroRh63PO7htOYgtP+dXMs4nKXxfYfh+HbW08+ny84/v/vjPf/rw/qe8TO9ujx/efTVfLnn521dvXh9vjnEcYooUAhI9AYibOb8Sf7dPh8/NxwvTPqAjyJJP9f7d9OH7y/vvy/lO8uSi6MAckJiIwU2bmujaqCpQNAH0MCSMofcxNvMmbhqDS04RcwhSkUK2XKxEijgsJXBAs3KiNDQzSuOAv0F2bwvIQl4i65h4HGLJFYEiI4Fbq+oK2AiHlEAW15rb8FzINkX1/Mn2XXGbl+nu488//vD993/+4f7+JM0dtUntcaQpBR5FRJpNFw0RiIgphDDEMBBGpoFjMpWH0+P5csl52u14SCnENIwcI5h50dyy5JJFXARV4zjcvnmzq20qdSolPzy8+/jxz4fdCGCEGMKuN7r6xJTI+WH57oHKjPmCdYI2mS6mpef3+hh4IAB2J1bXXKBmM3XwlZIEbojA7CIuCmYMEDhwjP3gOWEn0xHR0GEnNVYLg4fm3JTEmUKgGGIMHLh3rhTpE9MA3FXAxAAVo8Zk5MbUI42/uLatMQcUqfN0mueLtLrfja9ubnr9tIrUkmsrItKaNNUeCQAQEQMAoisyohEZM+12O0IANxOVJhulYK1JM7VetG1kpkaEjlpLFpXz+XG6nGvJW8JiPcv0S84+bnb9mTvv4GraXKu2XMo8zZfzdL5cJkJz96Z5znKZJIUhEKfABNBqy7nUJqVpqVarqTgAEjIRArA5tQa1rPApAo1DOOyG/Ti4+TQtD+fp8TxNS3aA2xuJcSgFmMfXr7/44ouvDvs+/gDcBEyxNzvbFvV0LxEgO+FGq+sj1q0TmNEFTNHVtZgUteJeAQviApbNitvFbBbNVQo6IQYmAEQVVwVtk0iWWkSFOYJD76ZIAZlTGg5xOIRhRyE4uHkjLwgBMSJ2X/R53gP6bjh2dt/2MgIhA7FT2OjH2HtuAgduSFghJZZEmkCSqlvrPYS1B++OCtBtJhAHd3Np6gaGPZJxhD5dbQUSqFOnXM1U1VbWigmoiVkUi4GIqqoCs8cYZAgUIgFUKdpq05ItL8YKvr9alNaWki/ioI69xUSXLkcC5M61UtBtWM4WGwJUlbnku8f7H3764f3Hd5fLqWnBjfPV5dKtIxUdkFjtDzoQOnF3tpr5UuQ0zcN+3OV8kzgkHtzJXU1FtdblXKaHtjxaOYFl8opogPac6I/gZApgLg2QkRIQb1yiT7LdBERMzBy6TkF3UxHJrrNJcWiI5tg5vKYiKGjBaqklN1MxtbD25KiYF44X5sGNVcCdgMJARDEihJ5YWhPnfyFj+8mr5qi61ex5H2sAWydYh4023xFIX+etdFv7vLD+ufbwNSx2V6kly937hx+/f/fu5w+P96eaqzXN83y6v/942Aeikpd5vt0fD7vDcX/ch5TMwbR33e0Xg7Tt8nYLho5mf9m078kGrMv5w/nHf3748z88fP9P5fRBy1KLgGMaxpSIQ0DAutSyFGliYBjITUKAyPsh7hCsNxISbRqC6gDshkoYQbHo0rQMIUDaeZnq+Y7SCCEphfHm9SENuB/RKlklL4FkN8a230kzcBxjcvcqzUwBE4JH7n5Gc3k+RfRZFnR9tFfilT7cf/znf/6Hf/j7v/+nf/yD6jIMqGpNCoCHEG5v3+zG4+l0Pp9OVSuy3dzsdvsozV2qK9GQdvt9a+V0ynd3F7NKtCfcDXEIpI6llKXVkhebl9ZUIft8gfDbN1+8+e2cPz6e3lVp0/n+/ftvAyuCBcT9/usUA2w0k+tl548f7i8/sy1Bl8jKZKsXT9EwrNWbyEbkRIaoYAaIxARESG6+dqI2a1V8rmWA3Z73h2G/H2LC/b6bSmQOzAS9esvA3IERKTj2hqBEhBycyRBRzUvxJmBK0nApAEoIbAE9qatJM4b987Yc6235Rs5Cwz7+y1xaXubT6eFuWSa5ud1FMqlm4tZUm6qoSpNamxggUWBugQNSVwnE7KrGzCkmQmhhaEGIFQmQbWOJ9aPdbScyUuQQYjAHQHK1WqrUZqrAkZBWhwiv7v6TCrhG7avjToDuqqJtqXla5vPp9PDw+HCZptpkjIEAypIvl+VybrtBpVY3BaNa6/39Y64KGHJpOUutampNrIjl0nJupVqrLqLuthv4MA5jDOA+TWXOy93j9HCe5yW7W8ltHEeHtD8cv/jiy6+/+c3x9hXHVFtzqVlaKctzuSIX8goQ12mB1CdqOLkSKHpzLW7ZpKgUKXMtF20X84UwMy+iF6tnyVNdprrMpRQiDuxMjGhqrQ87djNEZqLIkTiCE3HgGOO4G/c3aXdDcU8hOJh5dc/uATxS90p9tRbga77Ttyl2z10UogQ4ACcA7y1BDRwpICP6BK33ZQ1IO2LHFt2KNtHWVJq7IhkxxhgCR0pR1QqglCq1gjtvsBcQciRm5kDE1Cu3zLzPD+jJH1B3MVfnwGZqrmJnRYr7AzKKs4hJKd4amntigP31Npb5cgnsiEAcYuQQHHs/JXYIDtxbhHnv4ghERD2Or1qmMj+eHz/efTifH1srgNb7UFFP+HQiiymvubfOk+5dVJ0RmB1Q3Zv7UsrDPA3z5Tgg8XDjDtqKtCItl/lcpkdvE8rMKETai9rxeV27O9piojXPSCEORwwjcuwkWLg6yp3nCBxDGoexpqGklIndvNWishA0QncmcBIFNRRFAFPxmlutfZ6shUAxoGFTWoACYjAlU0ROnIYwjAmOa5Hp6qw/uUSfMeJffK/mVRx7G6eniS++NqLb2jPD1m+LcCPNr3jkhuXh9a47hQUM7HKZ7z4+/vGfv/vTH75/+Hjvjq9fvUmRmCxFrkv+8PO7y3TZ3x+Pr169/uLtV/z1jljNpUFr1sEAYgy9B/DKmLHeK1tUn9/IC9OerO7kEuY7O70rD+/y6aPk2VRaa2ZEbCFiQAIEEWmtalMDQwOOaC26BBB2BFMVqWaNPCJDq5Ohg5I1Ey3mjWLgOGrN+fxAcaThEA63KSZvxTWQKYKiC6GFQCnF3Tgg0G43SGvSGduohMBIxEDu/Evtz665KV+HHJtqu7//+Ic//NN3f/7z/d1DjBhoLNWXLIiQEt8cQuQxBWM2LS61MoOpmmRCPB6MOe0RYgwxckrRAFPaI+5apWkqasUxq4n0UMxcluo2z5OqRvDBPda8NJ05ALERsDR/80qORxnGw4uhBQAg6p7dZvcZglNAMDIgMzDTlXzm4OZSoRaUgi7MFAMxIhqYmqk5dmWkCoghkqkDOFFvyWcq4MauwcG9z8gCJDQkZQxIjIRERIAMyMhGDqxuXhQFWGFnJqiixepUTUSacLqBV89P/UZHcwcwMAXqYJzVOi/T4+V0N13O3paA4q7LPOUy977Fah0ybWqIpP3YxJCI1oy7u5thq40IESjwkBIARlkbcfeGsWuyEt1MmxshcJ9DU/J8OT1cLo/LPBFxSls7rWcw9XOpWiGgFeNzAm1SW5nLcl6m0/n8eDqfci4qKzHdxEwUwQJ5H/tp5qW002leckMeWtNlaa2JqkqTUqUUqU1b0ya93ZAPw3g8HkJgkVZqW3LLRUv1UlxVzOY4VYdYJd3cLLmU0mQpuUoWrUtdzqeXg/iImIMTO5Ez9ba7vSkYegHLoJO1s5RJ6lTL0spsltEXsBns0vKlLpdlvszzlEtprTEHSMAc2EnNOsCqqu4IjqrkDn2ILFDvNKiqCqSAQlhNwKqDGbIZDU6BMFAXtz66Y+Uv4fPtQECiBByBIvZ+/6v1BzQGBAMDZORdSIQwoGe07LqoL52NQYxETBRjijEmU0PTbK00dbU+toeQmYlCCIE5EAC4d6b0NkZgq4pDZFN0VVFRV6VmDm23ZzRHQjWsFd0DxRCej1eAeXp8tNLbJA7jEGPqzQopRNIAGMxRFAAwYAjEhMEBDFSs5ZrnPE3TpZTcG/9tnP2ngTrrvBFAcAJDMHcxIEd2BCNSBwSvTS7TlB7jPjiGIzBAzUsrc6uz5EnLBa0wVEKjldL3YqirmSynH2srl/MjhmF3/GI8vkn714hxG2GwQvxuCIi9RR93q11zmae8XEymFD3wOi7D0bs/b4airbZWW5PWGkIMZImNwBlCjLEuIexiqiq18wBcGrAgKZivDXE3W/tJmE4vz7iZae/7C5vlhjV70xkqsMUKiIC0zYXbnM+trezGsYe1uZ8TqNrpdPnxh5+//eN33377XcuFHPAwDENIEYdIDKClnGqZpnnJxQD3NzcUUhcA7aEbAClawOBu3plSBiuL4i9H7SRzyHehnkK9UJ29FjdzRzVXNTU0ww7UqGkvNOp0Bjc1NasiVN29rYG1EQcCd6nNOnFeABTRTZOolrwYPqTxsDu+GXfHMSYC9HW+BHjnlaATYxpSDOF4GEvGXGZzQzJCR8QYwxBSSp/xtqB7aJtJ8T4mod7ff/z22z+8e/dOmgWOqtgqlLyOFq3FWoMYjsd9akJtPi+Tz5dc5guz+dc27kazEhO//fI4jCSuQxwJ9+fT/NMPD00vaRSOyMQi7q7LMl8u9cOHu68+vnFcapFlLkuZEA3B2uKnh+XLL5evvsxffPmbYdg/Py1x3B1uvsAWqHkii+jaXKq1ZalLa01FwSk6BFeQKm0xNIgDEfA6gNHUrcfhBivUGd1YG5ioWqul1SLoATEQ4Rrqg1Pv7sbEgZmZiRAhEMUQkBgMXbEIN0+YdgimbS7T1JaztmIq6WZ5btphLVQ2cAdURO3EaxWpdZrnh/lyP59PLZ/y/IgAIq2Ui1ozE7WmKk2aKLj3BrSS0sgcYPUYkBClVCZijjEk5JjMm/aq5SYiKq1PKgNvIo2bcXAHMPN5erwjfPXq1etXr3vVQO9iu8JoL1ytPkvKAbZ8J7iZSF3ycl6W07w8TtPjNE2tSk8I9NMdmY8jHXYpDUwEal6qTHOZpgIoqt5qr622plJKrc1U++gxMbQQeDzsj6+OSFRKETNAIg4hxMCqTedLE12a+OmkqrtxvP3m698RA0KrbV7y5Xz/+OLMDzsa9t4nYCKvaUMz8OI6uVysPUq5r/mxLI9SFtXmVsGbySJtkZLLUqYpT9MsUs01BAXofb67099VsLqTGXlTBONAjsJWReZ5fojNYrI09OKNppadM/JiODhGDrsQ9xwGwOFq1z8BUQCRKAJGoNAH9DrainyhmSu4E6WQjsA7CxIoB1zAT64dukRmDwFjjCmFGMkN0IMbteJq6m6IkZhCjCEGDiEEAnBTUzUW055VJmaiGCIhi0irpbXae1SDm5wHA40pIWEkRQ5hGHZjeH4j0/T4sDwCAQfe73fjOBITc2fHJIfoTmpIFIYwQOidsshAxaRKqa22Vt1s7enO6KCIvc37Wn2Ga6PYNSuMndZFhuwYgIjcROuyLA/3xiw+Akekskw1X2q+uBS0GlCIdSsXXwn7TxZR6/37f5zny/39R06Hm7d/88Z/H8cdBzbvBV2ru9yr9wgNoZnklqfl8nh5vC/LI9jiO4IBQ+ghsiNiClEV8lKkM2RaATe1ABgxOCmoibk4OfWeMgbW1FrrPODe3AJ6FgI3SvuaJrg2nHpuNFS94apYyDd2/Gq2bTXeG3GIgByfEvDX2o4NtYd1JAM4qPjj4+X7H979+fufv/vhHZjHwEspOQ9fvt7vh/0uDUTwcLlc5qWp8jC8vnwZh31HcTrPWt1VwYzMILjDOjEPsDOL/5Jpf/z48wSyPN6zVXYj7wx3dAVtJkUaNnQwaSZCCBwDIChYb8ApIlB6OgqQmEPkEAl5nerTxFURgBjdTHqg4gCikROqBKRuRMxd1MRd3Jt7cxc3RoOAYQjjfojixCkNAdyZaByHFH/JtK871tWxqpacp8vl9Pj4kJeMgG7YqougKTMjWGgNSlZwcg1SqVXCxIgsLZs2UzcT0YwU0wBAESm5hLyU0+P0eJoBGwcOgXrjsJTALJ8fy/n88OHjuxClyXKZlnmaAATAarZlbtNUT6fzeX487F81eRqswkMYbg9eGuQKJqpugGquUlqjmmtr2kf19CmfWhu6QzFLHCICGZMSOzEgMiARoZmXLGaKYGZSlpqX5trAAjEig4MBeUghJKVAzMRM1DPARCkEZlYKAgkgcNyH4ZWxlUJqNS8uRcxaTS+goSvBxNykFdUils3EVS7n++l8P10e5vmMgHm+IAGAq1axaq79CBExqnbWEyIisMeuKdDd1dEAjAImTsM4DCNyULNlWU7nR7OqquCKcJ2GqW6tH8LWlssZ7j7+vN/vRaW2utvtQwjuW1vzZ9K0tuPY1EFnAUlbluW0LKe8nEteaimt1FYaGjCBmw8x7nfhsN8xo5iSYBMzBWkmUlRd1US1qZkic4rujo7IgcURQuQ0JmLu43BiCsC0VI1FYghCodWWsy5LnZd78yHF/fF48/D4fhwQoIosZS4v2rpxwjiAO3lPtK+tvLv37ZJdFpfZ68XKyaWQqWpVKbXOJc9SW6tSSmvNWlNzcUcikVYQVLo31UQVCAdCduztplyl1TKZOVXRpGhIbuiijRydQ+Kwo/4PAsQIlFaS8Nbx67ltR+wlE1uTWjdHcnR3c+sMFISOXJu4Se+RQATEGIcYHThACMChp3pb94MJYUjRAxNzjCmlFFIKKXYP181EpOddhUzFvPeb2yZEl6W0Vkxb739R4iUg0X4XYgBXQEMc8CUTJefzxaqjEVGruyUNHJlDpJiIE2ICjOAcQoQ0IoyMvecPMkGKMYbA3IfFOPb5uBvsvBZoIyEogJn3gb+O6GYOBoi9dAYM1cFElyU/LnGoyx6IW54kz9oWtEYghP68kvsTHFukvP/h/5zn6f7xIQyvmnocb27e/o69T7LBTnfC3jHXrReLtnJp+SxlkjLX5WK6EMbAIXAkXttYcAQgp2ocPQ6MGAF698yAyH3IrToDRgwDcyJidPdWnRjQjRiIgQNyRwqQnhH3EbbedU/aysCln3LzdepMp83hVmb2xJIjB1snVvR2GptqWMP6/qcQwB1ULC/1fJ7P5/k8LegQmWspNS/sktDHGGJiEWsirTapHbhTCpGIEMjIUdc2hKDenQtfuw+/vItPTPtPP/z58vN3Q0hkSkTEAYFBEQSgmUKrClqKg7oqEw1jQiJ168QXETNQCswhpCGlIfYJAh0VXSlQREwMgLIiASZUWs0qzd0iUwxczJtYM2+OxSCbZalKOljkQLvjjiGEkBzQ3JhwHIb40rT7VhPQu5s5GKCKlGk+z/PUWnVwwmCGtZgJEkYmIkoqviwV1EuxaWql+H532I3RKgAsHKKD1ZYBq1rmgLv9mCd7+Hg6nR5breOe9vv9sCMmHMYUY0J8rKXlevr48YdhRECZLtPlsjgogGuDWvXu4eGHn394f/fz4fi2lqcRdhQo7IJiUh9bURdwj4os3DxUD9lbliItNymqtVqr4K2SpYH2x5h2FBOGAePAzAhErUhe6qUaXnpbCS+L5Fm0ofW5cIRATgHGPaWdU3Tibg4VHYkosodgENxTpHGI++OweyWsugjS5B5USQU+gYZg6/loZqXkaX6c5weVQuAP9+/Pp/t5eszLDA6VA3bPGzttWIkxpuBOAAKivZChEwiYAwKauZsRIriZKSIdDjdp3BkgPtw/PD6UUrUVRAvBQ6AYAnNv9YmEKKqq+vHuZ0CY5unN5fLVl1/f3N6a6nZgnmz7VgG7Tu4zcACtbVmW07yccrlIK9akLHWZc801RiKiYUjH42E3jgBQWgMENWCOhFVbbWJu1sRKU+Bh3I3RMKmZiZsiIwWOMTRRBB5SCkMQ8yW3mSAF1hhrFfdWm0mea3sn6k3qb3/75RdfHA57Duyg+GzGLgJFoIRuG+XX0N1d3RpqdWkgQqKkSqboq+GqtdRlyXlqTVRMTQHJnESA2bthU63epxyruTMxMkew4O5qTaXJVCgsKRVyNwqGLlYVsJlQSHHYpd3tQAiw6+3KKRAidd36OZZKjEDYs3FkjL0dgZp7BGcHEqutXbRlaRVETZt5piAhUuCBIyCZShNpVdVFVQTcx2EIIaSUQowxJYqBQ+jFSCoitRERM7Wg0lSbaVOV1qouS85LNmlugkQGsoSClAnZzc2FzDEUDeE5s6y2eWkXcwWAkhOHyDFwiiEmDol5ZE5IMcUBfSTYEe5CGBDTEMLt4XBzOI673TmftRXupqgHo46wlhKCw8ayR0fsMTysNBvizhlARgcxLyKXWu4Rg9UCVskbotI6KqVPpdkSyc9sibRy990/55ovU467L4EON6//VpvAsHZ222BsAHA3aXVe5oe8PNZyNs3oTVuWtkhUTQOMgdYoz5GF3Dl4Ghh8dI3YSy+JvI/K02iWAEcKO4oDc0RwkAaEDj33QBgjxwHiSL0nvz/5GvYJka6Pg1xLZ82dcKXrXCnu2zSrdYh974EBa4vdJ/vTG3B6/3U3EzFp1ueHuYOqm7SytHwxkIzWhiHd0N4BkKhjpQTAiDEycYDesoP65BzrWDnSdZgs4V8x7Y8Pd+/zdNgdAnHJi6pJU2nWi2VdVHsSp6cT+nELlCj0Lnx9/4hjGocQAyKqqqn4dssEFJCop3nVpImaImJtS6uzSibQyCiEiAQYgAMGxKAYyBGy1EhM4EQQeokz8RCYO633k7XSHVYnisDVas7TNF8ul2meMnigTtnpHGQAdzWdl7kiJFNUlRjj7c2r42FnTUp2E5wudV7u1ay2MuxSGggQzDORDCMeb4Y3b/cholqNEWL0/YFvX0WidpkfHdMwcK+kzMs6K9rclkXUPp7naRxfSXt7HQzl3rWNSyNr4A3Ro1PCNDBqpME9tXyWWspctVUCZXIk5+AheowQIjITA6KhG0ixZWqtiPbG8I61WF3c1MGAmTp2xaEzq4wVKPT+sz16YkQCY3ICAzCxuhR7aFlbuZhVJFi5m5/V6F9Jo+7WynJ6+LjMZ9d2OT0s81mkqKmbqymuQKK6Sx8TBQC9SgHW0hpvrZl5GijEQEC9k6qDrUM+lsUczLHkIrWPsDZm8546C+jgTcT7DEM1MzjRvZvVUkrO6OauIeCVrXEVKO4xka0ZOHc3a63NSz7XOteWc87znJc5L3ORyKZxGBP1VkwOORdp6KpahQCHmHDHfU5mLookFPdxPBqiqJqJmawwoIu2xjiMQ6LAIAJmJn1AWW9BpKJWm4rNdP+RyHN+mKbbV6/G/RgSjwf45vlWENHK2+3sTRewnmXPYNWlSi0qBayBC4K5V7OiWkSqaSe7EFEkdjJ0R1MwMkJ31y1t4eaGZkyOjgZ9KkFFVXB2C62RQyXZOyejgAjYMyIxUQjIPeFMSORrnfHLmgXEFYBlR1PCBq7uarKAXKA9QntEOZOeVQtodVUwjcEiMzMSI3F3aEya1CJaq4oy4pBS4JhSrz5i4N5JXs1AWl+1tWrmSB4CbCVBEAKEiE4EHigETqm3dcUYOaUAjJGYnfAFpqVaRGY1dQDVxhJIiaRT6iJiJApMscXB2mCydznEuOewB+cU4OZm9+UXbxeZy8MC0LbCTFg7MIMhKKIh9jPROd6dNABuaLZ2rUfqLYoEIJtcDNlNwJVw7XjTfwUArgfj+ekw0/P0vrZWihHfuAiZBbd4rVjcOOemYq20PNX5InUBaCHAMLK02GoKkZHI+iTE3tSiVRFT6QNjGrjHEJmQENXQBE3YNaqyKthaQi6quZVmDcXRkeK4JwTqc716l0WET13F7U5AN2gbCRDc1kr1NauOTyYe3ADRVa+ddVekHgigk3GBGZGgNVnmMs85L6Wzwte+Q+oNZFry6RLO8xzHgeNwjHxzvDkeDmNaS0WRUa2Dec6IyNTzhVtF5JrCfn4fL0z7Mk/39x+meGKk6XSutdUqrSoYUAeY3LBLMrF1KB4hjgMiqlpH4VNK424H4NJqz+USABMTd0cMGUlVXNSkiQigcVtKuUi5kJWIYyOMHDhEDikmHRSgF5C2ZtCCI7IHhHEX9oeRA6M2eNm78aWd752yTaXlvFzOl8eH0+l0YRwIeWWIrQFa6+0dIqcQEjLtj8c3b1/fHA55nlqd81KXPC+5FCmq9fb1YbeLTDEMtjsChfDm7fDFV3tAmedqXtVl2Pnrt7taW23LzpFD2O2Sqk7zuZzyMESOOOdpXvTxPBEdfvPF/xJ5t575Jm0uZW51VmiAnfeNSClx7PEQlvNsprUVkzIkjmMYd7jb0+7AMREigqEsYOaiusxtubSSRZq7MRhLda2AgIwAifvsbUPXCo3M3Bk4MhIGokAckBkC9UHtLrnWUtuHVlSWplIpGEdyY/wFVmO3KMhM7jpfTo9378p8yXkuZTE3JOxAE3axNzXXjm10HD0QY2QRFdXWClKjwCEGjowAYOqq0ppM2tSYB3Ocl1xLNXMzQO7IPmKvzOxFNOpm1gtTVKSWWkvt5Tj7wz4Nsc+u3kwJ9nl/SN4HOaoU09LqXMqlSVaVJZfLNE9zKbmZASGHCKp91G9prQQCNHVzBtgP6biPhOwAc9bzpBB2cTwYopqKVpWqKqpNaoMmiXifBnHLYlqlldqytFJrLq01tT7MXXNZHu4/9gFoeTncHHfH3e3h+PU1TOTOIVwzgeBg4M09gy8AGayo5FpnqYtpBa+Kap7Ni0MFUEAnZEMi6unngKBbwg970xJCV0cVcawQ1mFvjk7mxBqCIiytqWrGeMPjbUxv4m6fdse0O6bxENLIFHvoAsSAa2nvcxgFEYjIGZAMvaJPrtm1QJu8nDz/DOUd6jn44t7Am5h2qxA4YofwwdQQyRxEmtcs0iQQEYQYHddBVaAqYtpp8VJFaqu1tFaJIDCHECNxCH3MUAoRzBgBQkqURghj2o3psB/3A7MjOaAFtpcmRQyqey/QdEA16T3ggiiZQS+yTzG2Okjdy3AchmOMRww7wng8pm9+89W5nB8uDwYNQAFW+MIBABTRCIzQDQ3QkQnZHNe8JwhSoI0NCAGduJlNBgFce8lfr0zorFXapMY3qsuqZN1rXZqoGTPhLoZdoBF9AEPXNZXloKYqxcosy1SXybQQeRxod0zER2mMBEgo6mK99aGKq4jWXPKyLMtESDg6cgRyV3Yha2wtaIVWRaWZNzUgl96NQAyAAiFgDBgTejMjAyLsA7o243i9kXWQLeI64marJ0dAJ+oDsK3bFV/Zc2DWCzM72N/boAMiAhHGyByo5Ho5TZfzNM+LNOnCS+AOjG5mttQ65XpQPRwOx5ub29e3r1+92g9jYg6EBqDWSc4OgIHZ3FS9Z1Yceoj6ItHzwrRrrWW6VGQErKVqfyghmSGQcuAYw7hLHLgPFnaEEGNKo5mrFlh1Zy+8MBFptUqpgThEAjQHdSIHdTXfnoabmTQts84nyyfPDDWDVFaL7iNhSGFPgymZCTmQYyCMwcYBDntmIgePn0WJfZtEaqlzKVPO08ePH3744Ye7u/ucmzQHAgUHkI6tdJEtLavU3X48HvbH4+3tq3EYGVBV27KUdlrEioE6mCO0astSxxHTgAhxGHDckXkBUGIlAGJnjkMazud5msW8mTEHGMfYJLZmIlJKMVMHmZfJrH79Rq9lY6gV8snzYnmxBiCEPqNFdHZDq60ti7QJqY47R+DdLu5GSgOG4O5WFlUBaaANVd3Mm2iroA2kuYma9Cykh477gBM5B+SIqxsGeKWO9WoQQO8lRoTIqI4u3tAaeHMQQENWDsD8EsnuxSJd4JiHFPe7YY48r0Oam6qoWQeakIgd1US1AtqmSQiRCYl5nVhnJq1lJOAQmYhA3dV63JYBMLtRFQEwZgLkEDAm5kBEpL06s1lrYmbd5geW9v/h6996LEmS7ExUbqpmtrdfIjKzu6qbQxKYOf//3wxwgAHJJru7qjIzItx9X8xMVW7nQc0jPepM0xEvBRQiY7ttU1ERWetbXdu+v72+MslTf5yXxe2jcxeGqQUOj2u6q7Zt32/7fjPrACnCpVTilqlmtm1hZqaGiKWwqYKAm1Qp59NCyCJ1yEBEhtb7VJbz3vttu7W+t7YxURWeeZloOs1zEQEPRh6qbCJmer9I5TF3zeFKCtPe+i69iol/dBYTIR9x7RlDWEVILFgqQoXogGnem66uV8gO4GbNY09QRDu2joNzCzBWmwjvGt0cnOOh6VWAKMVFWAoyD3YyMRMQhHuiE0RhLtNUamUuzEVKlTpJnbnMKAWQAeg9GeDDlwqSoadDuqdeo72m3sPW0DX6PdsrjH/80FhAIHhGJGQAU9KQBLm7dffm1lxbmA6nVY41CjBgoA3JaUT4iEJ0QizCRMBMIihMUqjUUufJzUeaChIDcWCRietMdWYWBEhz+77XOR4HAzMgDtIMECegEw45O+bIZE/UZAj2fm3yWsq5lAeq5+R52+6RCgQ4RkpjTfQ9N48CIegA3o0xc4wvcmSYeSIIABB6JEAEgiOrMX8v+INScWDtv/emH2wv3w9bj/GXFM6HCR7FzrnOXjHK+MgB6RHkq+kquqKuqZv7DmjEKRUBeLhd3xWsOPpajxi2g3BHTIjkgkUKioRIqVzIGY3QMtR11zBPNDXXQJZSZ3fL9HQNBA8IoORCKUAE+XfCoBwf45CmQXpEjL02MREjEAQgIBMyAY90M4rxj83EyEgHjyREFsQsCKJd29ba1rSNwQMzohASMNO0TPT8dP75H37+xz/96fnTp8fHh9NpOZ/PtRSIdHXP7Gbd3CxG/EfkmCR5pgMAIbXWP36QH0p7mOveRpczwn64LuVEAWu48cTTMi3nU63lHdczoE3Sm+3ZIOBIwVJ3V1cbf7gMqUZmeAAeREhPylFQENyjbb5e/PbqBX27QdvZdE4ogshCyJA13MbBhZjMeT7Rw4mEBRJQ/j4eZowp9r6/vHx5ffvy9vbt999+/8tffvvy5ZtpIgiiRKRZuGuEsyAhrNvW+44Up3M5ncvz84Kk276v6+16vd2ud484P87TUkgCsfTuRXyaqNbJVZD0vl6GWqcUqZWZamYJMLU9U7umcJ0XSTi1zpnQWifCWkvrYf4DWgS9SX/jvkHbskcohGIoRIdQcLUwC2ul2DJTrXJaSq3ElG623fu+2r5G30A7ZgAiECcX4uQYd9UY8xggRGaUkjJFmVgm5sLjAoeAEADuSGNuF5kUQIhYhHkc8O6h3UkTHAm4IssPLz2+q8kP2eM0Pz8/9+2yXl4B0t3UbMQ3mwUhOZO5mnXEJDqWYsxCLMQohJ4Y6l13Cyu1FjkUtUABASOPJBIjkgUACYNr5alWEUIEd4X0iByzekQopQKMUK+83W4eqdbPD+e/C8AYNhY4NHTp2tu+tu3e93tmZ8b5NJ8fH+6rDrZGM922tfe9ToI4e7gkIWKtpZbRmJKbt96QqcxFloeynF7e9O2+ruvr/XY7L+fl8ek8nR5mOS8TI8XgEEgRKaWQWYrI2LqMEaEQLLPM0yA8Qb6vy//4FDSkrBDv6l1EAZqRIxggHZgtrenWt0vmThgjgSZTAQ1wfA3AR4YQERwOb4DMoQqMZA9QtUg167UKcRURRiERHMp2JMDCPElZpnoqMjGMuFkpdZL5hDIn1SFehJHf93EenyG+W3S3e+wvvn6N/pb9FtbSO0SHNMTRwgLkgBiFqlkCMSNSArqHtq67eo/QdE1LdzNimk9TEiBBQNrwUXpgAiPVysQ1IQCSeYh/BUkQ+P07Eh5unpFABaeaUpIKxyHK/uGwIsZSaexlmMeQOJBC+Jh9x4DHR/aWPWANZJpYzlhOUU4vm15u99b3d+X2O7osEyDG+pIAhh//qMnDyBVgCu88ZjQ3BmTKntiJWbIgMQoNd1i+D33x3Sz5o8o0c5inEiGrxOPsT2V/iMukCTiYr+iQnk62u9+qr2xb9NX6LXwH9LEyMNXedeySGQdswzMBiRCPsRAiCcs8TcKFSUiQpBdujB18tyZBnAHb3rTbNJ/OzBGeMW6oFpGJyDElV2KB+IGJkhEengkeoBEDQuwx5Pc0xPYYhIBCVATniUohKTR8rRHhDmZuFoQohQmQEL2bDllcOBNOtRThqUihmAo8Pi6//PLpv/7X//LP/+mfn5+fT6fTEC+P8yE81GNX62r9QL+OKchoiAwAAXnf28cP8kNp19a32xqRkSR1Ya51mhirdo0wFqYRYAxx7MEIAcYZcdxqMAkDciQvxSGMToDI8MOkS+PfjEhj9TPML9n22O52e82JavanmenTY58CgJlQJPlwauLIWUaMqZZlnpiZgO5tuWw/nMGQEWn3+/Uvf/233377y7eXr19/f/nt129fvry0XSPGjSGJ0gzMzB0yY9+6WtcumT7NdV6m2+369nJ5eX1d1117AlI4AvA81fNpOi1nEdSu6QEJpqbrnShFaF4KgJcCTMAcUhLSzZ0Fa2GWZbbJtEd6ncuMnGBtR/xwo0d3Ui0ZM0MnN3QiAIYQSEhgQACi0euACNZChBBmYQlJ4eSavfm+OQBIodNczk+FZZCk0zVcMzwZUYSmB57OXKpw4bGHNhvEwYAgwmTKMVJHRhGUMva1GI7hhMkIkAYZxPXvblqj8GImqGlrbd/21pq7hY++tqu5eQ6UHiC4m4czI8sRTRYRCZYJnhk+UuwgM91teI0FkYlQCAA9Ic1xZNsilpQR7CHCRAhJI1vC3QEyczDwS8aAv95611pLKfLDQH7Efn9PWUqIiHFLoKOdp2mup4flcVciKgSE6WYsdDpVKQiKpcpympdlYSrusbfN3SICiYW5FpdiUkzYiAzA3HrbtwJFqaiSFbIMc+9mzayp7apq6uGYOSzNQliYGDE9tfmGrWCHn3+oJcRDT4WIkMQAFQFhREO4YVmQpwQ2T1cD6JmWaWbmrmbgjgkiwglj1TUW/jFMsJmATIDMzIwpgkOFbuqJHMmMpdZTnRaZnsr8eTn/cjr/QlQTCKgEUB7BBwRMmHjU9fw4j4fMiH7XdlvXF9++5v4N7Yq2hjY3xfd4bUTQ7tqG3ljDDCIID1BaJI62B4GYJDmQuFSu00QD+zIoBomUlADDVypCIhzpEU5ESJwBbpbpAAQ8KEAlB6U/ufcMDM73FAcuH4viUY9HbkgaAkAaRgTmsLONMyFztGnu3TV6YlO471C+rfbbdbvebh6OnERHwz56KaIkfA/zg4TBrhn7JAD3DAjAAAQ19wCApALiFDWWUgqTYIzIKXzXJsPYOP//7aodcKzuJs4Z9tmuc/s2pwIUIAJEh7AIb622u7QLtlu0q+43zxau4RZmrqa77XsP9xGiS+//sXBnIiZ6/zhYBIsQUgQ2bW/rW0RrPm+lzkSie2uqACaUbRKhQDoonEhcy0OpC5caqh/nDyMJxwPMoanv3Zq6eYxfXAJAEgZiIiNMhc9nOZ3qCQszhqeZ926qpurClFGGACUiSpHz6fTp+Skia61zLVMVTK2cnz89/MMvn3/+6dOn58fTeam1jpmLeYS5BVhEN+/dWze3iIBj0ny4PQEg44fpw4+lvXfdb1skIhcRlvlUUBzbipeMHFSo3puHlbkWLsCSAcM4DN+TMjzSxlZ+7LBxiFB61761zCSkaZpqrQDv2LgesPdcb359zUUWsofH+ZP8DH4i5sJYJEvBIowokZSQCQ4wyCMsyH/9Un4o7ZAJ4a7X29u//+Vf/+1f/+fLy8u3r5dvX67fvrxtmwpVKEQEgmGGmanq7tqaeoSaeUQphZnf3t7++pdfX18ubesAlZkjEQCmuTw8np6eHt3tdtm0GyGp6dZ2xCiF1DSyzlOW4pmdecApE1C45CwTZLnf76p9nmcuNcFrQfqwpU7P7CGAKAXMgJwKciGYAANlWBHK0IX4EMuEerMR/XzQtGDsQRFQYDrTp1/muhBxRGZYmoapQwITLA9lORcWASTt2Xbbd48+MAPBCEIAFEiAVJhJmAAyAqNQhiAgEY8SI1OFv/9BBIyM1tr1dn19fblerr13MzXVrqpqHpCB4YcAHiCgMCEBD/1dhpn5AOxlIg0ZZEZ6OgQi0/B6I7FFRA4WKOChNxVCZhI58Ng5GjFECM9BJInIbW8R2VWfnp/ix4EjIgpxQkTEOy4hIIKJapGhHytVpqU+Pp3muT6eprlKuI0VmYcq8zyX88NpmRfT2Hu73C8RXoqQFERM5ARAbCJeC9bC4bbd76kMWqYCpaBm7r3vva+tbXvftra1pmaQMDy8I/oKE9yi7ao9GZYPiuxRn96NgAkADEQAAsmJgNpJTswL4hQhahmuCXoY28xU0x1FSGrJQxBvHt3VwjwDEahQYeFSiBhYCHEkrngkcUECqeU0zz8t55/q8tP0+Mty/jmRzR1lCuBIDABGIDrURsOw9ENtD3fd2nq5vX7x7Svqq/gquVtv1hWlUp0HBr4177t6a64aqsfjG1cFIqKCgEyCBUao+3ya5qWKVMTvQxocoTjMKEylsDB7kDuNOWXv2ofgIRGkUK1SJuZKSBC874QOJalOUovI3zl1R17rsVEaX/QYqSA8UJCEhGPmC4NL59a76k3XtwZf1v77rW1qnl4lafTYx180Suofo3NCJBr+bkLEDAiDIVY1C7BUj+REw3DIpKV4ElEARtI72G00z+/4h++vB+SQVgpXiup77de6vUyuiDURE8EyMEJ7k23D7RX2N98uul2TItKjW3SL7rbrftu6KguVSaapEGF4ZsIA02ZG+JB3MFMkJURv97bdbn2++enT6fxY5zm0u2rPTqCFI2076FUIItOytHl5KmUKt48CNA9XM3XolluzrfnerJkPffy4NWGMPz5Xbjp5LMRQC0dE77ptrTc18yJySOXcImGq9en54ZdffiLi+7Itc52rWLtT6tPj8vx4Op+mqQohuI/FOnikjSyiSA/s6jrUhBYjB3Vgl2AkBf6YcfdDaRekyoW51nl5+PTztDwO2RioRVdvqOlmhkyznzJnjnCPfd37ptpcSDKiOLPRGBC+GxlTzcc/17pmhDDTPCPhOwoxrO3tdrm/fGmLfP50ejg94zkgGzExJbMJATMicKQkAKBDBmQwMqNMtwT4eGnxCHPvbV+vl9fX19fXl7dvX19/++318nZVtSBAuDOPziVLkXEdBkAmRmDr+fpyT6j7FgA1XTKtDGJmdg8zh8xKHIDIzM21afdwyHcPWZBpKNmAc0sRFiLMUpklAMws7/dt21oklAl6AzP6OJBXzVvzcHPr2pppm2rhkQlUmLkQsSeqhZtGGEOmZ++oSmaZACQwLQTExDgtvDxQmaPOSHLYGdIhRmOSOSirg8jRu2v38CTiUrkUYk5ICI8IA+seCRFIGO+UbWIpMyNWomnmhx/OrswIN/N9/Gzb7Xa73W7rurbW1NTdzC0CM3Bk7r1Tl/FQpvMfd3cAH1O6HoaUI7MgCSkxOJ0Swc1ctffu5oBURDgZhooHzM3VXROMOEUwjxGSh7t5AmKEICITwQ+n1x/u0SGSJSRmrrXO89y7d7V9X9f1pm7EuCxyWmpv6R5ADImxyFSrVPaMbt3CuLAglSJDPWXWokXvu2rPBEYJOB7QsNZE5IEwcO/uu+puau4ZOeZ+hbjgSMELt+jNE7Ty/vFTjGbokCuOGe3oLHykbHMmm5M7ZxbIAskA/l0SfaDfIpMsMhIc0YgSBYHG6ISlFGJKcA/vrbunOwAWloKUXbWbegQSlWmaprlOE5UlUZALSpE6C/N43KNh+G5s/+NZEMv8UMIna0qWzWkM4D2DEpHHbCYhEbhIZcAgViDr3c0jBnQRkoeVkY51jxALA1Lku4c4xnQHkICQQCASLDIDMtA93L1tve1dzdwzRajaci5yqtP8yLIoCkjlZZK5yMTC+PGwGn1k5nsETUBGeo4nEkxjyjISI5gomUfGbHrErr53bWqeMdbtR9eOI0cuMzIx4d3FjodJK8f8f3zbScZJTYEZCc2Tu4l4ieDMEYD0bg2Bd/HL32uVAYGnsZVPDqgBU8QUXl0H4CkQMtzdYN99ventpV2+tutLW6/JkACh5t1sV91Vm0VEnWqt07RUIjwwmaNNNGvQeKzWWodD/EGAXBINwNEi5zRN0/Tise6xxTZlYiIASZlOAsiZLjXMPw7nuvp9b+rQNJvG3n3v1tUjMg7XAcIwHGQ0JQPTMA0dVVnV9q0NVauIW+beEId8GBCR5nma52qmU+EipLu3vq8rX2/z/bbO8yrViQWRA9Aj1KJ39QAAVvO2d1Uzc2IqRUSIhIgSkPPHVuSH0j7X+fnheZrm0/nx8ad/lLp8/fW3W2/Re/Q+8nhaa8gH346tmPn9tvWtu4ZIyXT34doQKSMwAjPCISATEd1Nu57PZxFG5kQ0M8wI7e1+W99e+vN5+uXp09MDJgE2REB0REVwhMykTMkcPdJINxPCwtIA/mjbMw+0V+/bvq/rer/f17fX67cvL9u2pXtAuDozSMFSZJrFXBsc6n9I6Q1+//Xtfg/hWuRMtCGaCHLJRI0E9VSvAToQaIi0tx3Ap6lIIRYiHAPbkXIGo1scWiXitN63LW/X++3WzKLOpsYRJZ7y+yauK1y3tNZt37xvkIYnknPBiZElSBxYe/bmvXm4MQIFhpF3GDQbLrAIzafKgmXi5YTMhhh47ESQJqaRhRHhFtYPu2NXN4sEHHj2UjnBzMwGJw2Du3k3KUyEGZmBSFJk5nIu5TTlI3yQdES4ad823db7tt629X6/3+73+yjt486XB7BuUPSGMz4BYmgzhYY2FWUU6LBId43MIGYZ3Q2EsQNAYqpZa713Mx+x0OLMbg4JhmaDumKW4MPWh5BuZp4ewCLM4wssH0eO7x6eDx4XYpGp1nmaZrVNVe/r7XZ7c8upVuKTSPSuiTrVSUoZioFE6KZ73z1jPs3ESISq1nez3lJ1W9e2t3QQniITM4twqcJMgOCRQ3ZoMazi5hkwGjKgQsxImBAWo4s2z7n8WNoxD1jhWGogBGZkuKuZjugT7W4KkCOypYxx8OHszRjrkTRNGEaGJE4WZmLhwizDnKMW2vq6t75HhIjgfE6CMG9dd7UW6VKoVi6Vy7zw9IBcAAmHMJ4AwOk9DO7vBsDIUh4+z7VmlX4XvTO2SloCZOjmk3gw6USEhdNriDbgFhjWBskcgN0hMYccEXCsKoYKeMiO081VFXCMH4ScIkZI+mA8mzbtrfemZuaRQcoR8/zAXOblU10+K81RZl5mmaVMWMDg9tc/hsAH0ekPnXYmZmBEIoZTUiAHMOMRm8ASEhyBBAHuMDg1Q1eJSIBx5Noct/aBnsvh5MJ3LF1EoAeMt594XAnAPTShRezhM2ZFcBxDi+NK6we2ZewIfvCPlIlDM1qkJTsWxxJQMhIsADDB3VB7tLutl357aZdv+/Wt7WuMLJwAV7em1jXUgbCUOs/LvFRmzGP4lb11bS3cIbO3fuxdiEqRWktSSwZnsywDipVc3Mre35QKESMLlgmjOYtlJrCbwweOW1O7ra15dstu2TVUXdXNcswSM0eIUmJGN9TUZn3rbZ6kCucAPFhmAEt0i/GLGqDi1hVgXK7cA1Vt39ZtvblpJkzTGVBKnUqdyjQTS0R29b31CCAW0+it967mzjxyepkCmYUI8n8Dmv3nf/qnx6cnImYuWOdm9hJN2836qm2DYGIcxKz9vrk71+KR2jpklsIizDJKublFju3UeCURE4kQtbObjogFEWGRMjFCFEKpVAqNTQDKAyUTqci4AMSwcLinau7btm73ImWe58RiwAbwobQnpLrpen/b12tY194ul8vb2+V6u7lpEchMs84JiVin8zxPmVNG9q5mpj2ul70r3u62zGcE7M0jEinmmebzaT7jtNA04RDYSwHizDBAQyRmKlKEgQmYmEm6qnaTIgLfu09ghuU0R7J73K6bGkX+4LYCwEx2Q93TO2Jgg8QMbn5EnyJlkBtHYLi4W7qHeqi5GSNMi4zhIREABpXMSDc48ASYATE6XjzywAIThTDZIEwt1NxtR8JMDwiATALklAp8vFTDOydMFWXOLK2DfviOZea+3V376+vldnlb7y9vr7+t90tre+seQcwVUJDMLQOzVGESM1fVUfVVA8CZgYmIuZYyVYzIvTVVA0QetR/Q1cMdEHx4sY4VvZk2xIGpyZHjaaZjcQlAhELoMHB3SaWW03mZl3laJvogfRh5NnnYhQFghJidynQimRPYYxROSAxMdetmBJRMY47ImWAWve1jZ02EQjySZNVya961q8ftvrXNl+nx9OkpPdNtqXCacZl4hNKHWnqmZ7yrE3Ck5X6/9QAOMoma7a3v7WNpT7e790uaQwAMYgmE67bf3/r25vulrV8y7gAtomc6AIajavburQ04fAzA0cEvzUREFuCCpVIpBENXDEGakBABphFhUnUMYOCY85RlrlNFQcXcMwQwkMtooseCHY+pyndp9vu7QczzQxEJRkAH6FSAlYkYUY77IiAmMCWPGAWEgTvKd1YCIkUmIBAjH2FiCZAWin5EfaQnDppeWLpb7+7hFqMWD66Am0UEIoiQTFJOdZmwSBI6YCKz1Gk6P07nZT6VKTve/va9kI8ajIhIDPDOPMs8ckAPz5glYCQRJFIyoxTiklSYhJFsSCcAA0ZHiZEJQykJCJyQmOP3PpZdQ2eaechHiACJEhIMIsMJnNxwoIsYAw4r/LsIcIj5f3geQ5VrmZb73i9v9+tp3R87S6EDEZEZ5qYxAjHN0Iy6wabdY5wYEdnUPHwQbtxC1aQjFiFEJvJwQiCm8OhtDy5FpkCASJZhFokAberNsal29yqliBAgI8/zXKeZULGHr2C6EZWR9vz9c9zX/uVl7Z7dwSI9MiyHwyjiOEHGSBEhKcCSNFIjp+ZTIcz0IW4LQHRCHes/wIzw2+3y+vb68u3b9XIZKpK23137ts1dw5Pfrts0zcvp9PT0vJzOpZQcY7IYQouATCJkIGJKBDMLC8ROSF3/YxndP/3pT/+lVndXtXvbv76+hN+1XVU37c0MmIlZOKGvu6lxLYN4wSIDvibCOHSy4R72ztQZchAIpr6LKWWGuzEUKuPLCYxQ5zLNVWp1KOqVESumlDHbQyIgyN7VdN12ffmm00SAU1LRoF0/6rYyfDfVt9cv1+uLWzdtt+v1ertu+0YQRTghPA0CKIl4qRMjTgAIN3SL3r01X3cvq55PISStGWQAZp3488/n86MgWa0DNemRmmCRijDkhEhD/gEAyZlkGvumNUCIpcC45pTKT0+liL1d7uu2tZ6R+pEDjMN8GRxOqZROGplqKIBCJAOhKAgEyZni3rw36xYWkM6V6lSXU52nmhlm6mFuNqZ1h8g7HQCKFGF2y3RiZhJmNmaNtfXedW1mMSwXzMiFqCQAhWQMUAMdLBiC0g22pjMpzH88j227b5lff//19eXLfn+531+3+0W1mQUil1IEQ9zGDmmq8zwvqrbv+7ZtrTX3yHSiFOY6lbnWaZqJkO/3fd8jAgGEeWjW8khRBkJiYhrUeQfQAKg5vnmudlDbgzBZkCiQhsUFSi3zMk9zLbV8VDUOCkuMnmdMALnWepIyI5dIMgdCriKBKQQZ5m5EBERJbIGurt1at4xggioCGYOQ0bqvq65b35u2pubw86fHf/7zf8pw6/skUdkhVa2FmjYNjSE4PpxWiIkZmerObhYACZZjetH6j64YbxfbSmqMWTUQZZq2a7v8tt9fdL/07RXyTrgntExDgHDoLdZVt61FGmIwE/Hw7wIhMGfKQBkmlwQIcCcfUTRDujgyQi1coIxUz1Kk1MKFPGMNg4QkmQkmYiDiREzMQcM6UmL+biA/naAICGTsoTdCF8kh6+1tV1XCJGSCwPA8dJPuEXCQ2AZa2ZGAC0lBZhwzpnBzd0JkKkPH5pFuri1iyIa6ffeYQR5r+FplWsryMM+nuUwkbBG72RpUCp1q5dNpfnhcavRjNH6cVpmRLEeI2JEglIfRDAgS/8iaEwRBQEYGlMqlJhciJvQYsbd4BLcEjNd82M+/x5bQ8CdlRibwwMsd+bGj/nJAZDA4uZEaiAIDYCJw4h+co7Ef+EHVmNYjNVFz3/T17Xp5uG8/97rMZYBf0iP8AKADFKQJqSaKge+6qwWO3F4YmsYANNW2AYLnVGop49kAADNnRFcFpMoEgB6RmMSAAsnRvPfd9qZqNtUySYHwQsjwWNmRDMDMG8ptKgsBf5TR3TcL2DXSMg9s31BmJAwJSAwnPAQCUIAlWIIFdskmxMczHZkVkTkubdC17ft6uby+vr28vny7Xt4G2xghhWlv3jRbz28vt2maHh8ffv65ff78+fzwUEoZN4kICz/ik4/sXwIz79ozAwF6/49L+972pr2IIIFp29d72+69bz6EBokRQJQxsMDjrRUGxFrrsiylVhGOjLE9DY/3HZYwc0YOqx8maO/r/W4QU3oUzspYJBEdcuv9t5dvW+vnqT6d5lKmOghPiYSp++X16/3Xv3z529/+rdTp0+dVkTeNffsgcYx4+fbrvt3/5//6l3//93+9vH1t++ZhkIEQI8kx0xhsXsr5PM8LEw2+9LjU+gAPeVh2EJYsBcARwdzVPCLdM8zVhzfa79f9fm3r2uepIhQ3WlXhnQAV6drV3ZdlzkCkwlyIRiJFxiQPDyckuVz33uMHzWkkWKADBUdQGFhEuBEjSZQJQMBBPcKtZxohCDGVkgQZSQSRoD0ydBQltVB1wCSmMTwd02giZOER3k2MLFC5LlDKInUt67q3phEjECKHOsK7Neh9V/dBZhEuwRUtRDt5tY+l3d0yvLdN22bWwjR8BAgBs0zTjAhmymziUaSIYAKWoEiBsc0PjAgDkLAIiTBAJEqRUbfH4QLf36Uh50GiobnINAQRZgB05AjEwyAEgDzE9Qg0jPMZ2bXf13u9FP+Qk4gIhDSQBgerPIfDztX0vq7X283Uain1NM+1zPPEIpapHm3V1rVv4zZDwjQJO0RLBUCL3Na+3rd11949k6pMDw9Pz58+97av4MzOhK21bWvX23q9rm3vPjhbSH4Y1DMyI3zvSveVj86mj0HUR197X78mGgYDcA6UJIS3e/Y76Aq2U3ShkCE/TsrAhFDrZhphR/ITwftW57gkmkXumgGmQTwoQAEIdRJEqDURuc61TrXUqRZhAre23l5b3wKYp0dZoi5ZmTALghN/UHwdu/YfhOVViJMJxAjDzfa17RffLrbdtTU3LZWpCB5qj8EhBmYaKyQiJGEhkcrTPEDdkOluGkbhfsjmI81G1JANTaT1cH3fXeMwT/I0l2mu01ynpXKVpDTfdH9FM7IdqeWkKJvgqYB/3FQTIY2ERQIEjENSEDC03O/5C+NNigxPQCCiUhinQpOgUMQxMjREH8BTHNF1nEwHy/xIdY3BNk+EgHcz1zvkYHSjEZAaunfOwI4xY0xUK1IZxHUa644fBfIJsRuYsQMOfXkE5gCgBGRCGmMUplOtcD6352dVtSA1bD12VxBEYa4lAQ8IU0bbd/feGk+lElF+h+UAATKXUucamakdOJIiBaBgKBhEM21Njw47zAGIINJqqVKmgmWuiyyPQhV+2LVnUhy0rOOCheNimXDsI4bH74/o1sT09Az1sb+Ao23y8HDtfW/tvt5u67W1fW/bvelu6RYRQYAFsSIHknrsexstzb636/X69Px0Pj/M88ws8B4XzzQeY4ZH+MEEgcyPEI6/L+2323XX/vT4OE9T6/u+3VvbVN9TOhCGOx3hWDOWaSpFiKiWOtWpTlWKeDoaakcFRXxPtR10vjgk8aa672PD514lvCDOntHdrvv93v08r//w9IlRTgtVkgAIQiHYbvrl19e//tvf/v1f/52LvHy6tIBbt2X+9PTwy/fS/vtv/369vP7Lf/9/fv311+v9rbUNMpAOOkSZAAAj4OGxPj8vhOyucaSbe2aIEDGnD8jMnplSgIDCve22b4aU6i3CgVrb++2y9t3cCVKWjpDRWod0pFBvA1nPRG4AgMxYpJSJmAlQAeN0mkWqeQL+6LA0j7VDD7QRcosGI5ydZViEMlRVtal2xJiXqRRCkSRwT4Rwg+bW0sYj6D323SIDEcsk04ylcKl0YMTkOAJIsi4kImXhOgtXLNtQTB5aSAAIj33rbtFVkbDUWuYsEyTWjHJsSP44huH92x5HSsWRrAXEQw+IAEDEWYIQgRIpRlIoUVEN0xwTVvcxVEoKTOhIjmQZkSkABEMQD8TMIgUJzS0h3J0oS2EkGeBjN6cEhGRhkXJ0lkhI6Kbbut4uV0L80deORAcn2sMDTGPvtnXdWltvt9vlcg2LwvxwPp9PCxF5Quu6N71e1/t92+4bAi7LaZmqQGK4dUsAD9i2tm+tN/cAkTJPy3J6OD08RHreIRI8Ye92uW9v1/v1dt/3Hn4IERxsnMuDquKqfgsiwgQ3t8PV9Edx1+0lQYVnxGoHyypc92x30B28YRpDCIMIQZAnAHiEAjrRu8MZgeggGQ+ctJlrd+0p4qUwIXhEBJTCpVBEErKUIlKklCKE4L2v1ysS16RST3jipdQZIQmTIIkQhd9T0w+Bw4+lHT2JgHdCiOi963r39eb7zVtLN8KlyEg3hzFTJSIRHjsCZGShOtdprnVhKYgQ4W6KYZQR49Lr5gkAqscu5ghhYURERiQkhvk0Lad5mqc6VZkFhdxDdVftgHepV7Bbz5vZU7Rz8g9tIjIyEvORJHKUkXdX09Elj6EuHFAyREEURiyEgsCQBA6ghHYo2Q4UOIqACOQosXyMd3DI4jBJkjgBRi5DxvCoYCaGBW7aVXMDO0k+FMQhphjtIgHT33kRM5pRBgNVoipSiJkQMTEzwCEcIQsTz1MR9vCIbJr33d5u2xV2IqJSyjIBETDD3lrrZqbmtGOTLiw8kiiJAGjQjcpUI80BgTMokhMEMygQNWLv4/1NTmcE2LeBD69SCwlMfUagMgPU798tCwh71zYCfOfkHuMTiKGjAwAaYZhEZQw/AjLj4MoPK2NE1369XV/fXi/Xy227D7miJSUXgLEexxTmaSnzSWoBhLZv275v23a9Xm+35+fn50+fPi3LQiyD7fHubjhs7RkDlxz/Oxrdru223adlqnMdTNAPF7N8z18aAxMuU52WRYpAhIevt3tvrc51jNdJmDPdXVW1dwSgoSt1F+Yy1TpPLDL2ox2SmVtr1/stiYmbuZ+m+tjn3lr2vt/vkFGEX759+fb7l+vLpd23wGyt75F39Z9+4j9Ku/tf//KX6+X1999+e3t97dZG9gxREsPpVH/6+cwcGXo6zafTvG26btt91e1ukTAtUguxIBtFRqlWJz7PCyHd77Fv9vqy3lfysYb2CE8M1k69uW79fnkbLq2Hh/r0eZkq11rGbjEjbteGQJC0nJOZLm/3ddWpnoCKCC8zfVzuttv29rev6TBEY+kBGEGIHAGu0BS1a1Ntph0RwDSKMA8BXwxIZCB8xy1p97Z5V3OPeanxSPw4gNqY4BkZkESQgag52A5AUWciKYhl0DDCQnu25vt26OqEpVSeKpdp7F9xlh+E5efTAwHsz2vft+320vam1iMdCSN12+8IOLBxIjyIKIAB6EgjfZUQgZyOUUSYqgOGWXfXSB9N9KDhCgkzIdMRPp+BmCzEBYkT0ZODJeoMFYQOXxwDQKZ5ZBiodg9/OJ/mef4IexmOj3EmarSt3d9u317ffv328uvt9nK/X7b7PR1cyjxVIlL1+95er/fbfW/d+q5t71UK81wom5seya+YyOnJSJXBCUut0zwjUTPfmt63xqgIdrncXi/3y7pv3XQwx8cJdNx4xgDw0EQSJiGmfw9J/uMQBnCimCZBILvfe9vSuvZ1v1+03d23jD1zT28IPsysgFErEpd5rqquapneu4+/b7CGIiADTL1hMNugOyPhNFXhIbYJxD4W2BHSG4/34uFheXj4eX74aXr4XJeHUk8iM8tEVJFKHjX9h9ixUe2wDBrhPD08np5+JvRWeJdqKO7XsI2UkSmPJFaMZGJiKsw25vIkUCapSy2zIIGZWphaQgAhj6ywROLMCWcuMkzwvVlviowsKAWkUp1qKZPIVOq8LAsK3W7Xtt23tbklc9nrS3/93Z6e8POnh/n0XE/fx9lD/JZH5vaYWI9TGg+XSA4kyYEHIsAMNPPWc99MW093ggCKcQ/KdzgqYjJDKRCBQ2AQkYBJBDiG8QNF7QkB4Zljw4Yjqw3Co6dhJIxvU0mGMnwLQ5Twnln/vbibMD1M80/Pn/78y58+f/6p1hmRDxOXR/jw5SIKSy1lKkAUGUBERYIxIPfWEiGSWKRkjglrvv8MQ30mMIsUqbUOADoVRMEk8AwNDwRkygRT3Uy992UqYx5gkWmRqUlZcG97SYvMD6SBcfIM+8i7ZzAP4E+8J5iQEBXCyjhXKUOECJCjxyeMDHNv+/r69vLt5eXl7bWreWKZZikVuHJdRJiZM0GYPj+ePz0s5yqYfrtetm1Nj97a/XZFiAzbT6d5Wuo012nmkaw9Rk8JjIRDh0Q/0ER+9LVrX9v25JoIpZZlmaZaWRiJRgDMWD0iIRcutU7LzMzee+u631diXmyuyyRzRURiGlMsU3U1TpDEiGDmeZ7m0wmJEsFdXcPM9t5v95sBsLTMfDrNW5u3rWyq33773bWXKte3t29fvtzeLm3dNSzv6x6xeZyWT98/RWRe3q63273vGgGM40EAUZYCp3P99PmhFMjUWliE99bU9ta3plZknutcK7NAzXE25Tznp08To6i2+03fXnbgHLZpsxAp53lJz7btbW9ut8hABPzz86fP0zKzlEBK97hd1+3ewsB6PmrUSpfLfV3bMmedFiGSuXwccvWt3b680UEVRRqjsPHmunffPWLkQ4cbIoBZlPcGNAORggeic2yIQLvZnm333i0MmWVZKiEPfJZHegQRRoCHHvQnQEQolUVQhBgxHPbVMLs2JwRmKsLzXOdTqZMgEyROHwyWiHg6nauINW3b+vL1r4PaCRlEFNm3vSEgE9WpFqKhNxmMlBHEyTJG7GMWmR7mGZmu1tx1qBIjhmKRmYWFEtLcht4bEEQGf8xzmLXIWJKZiwz5HR7rdxuSPcqI3rqp/vDOj48DgJjmbd3eXl7+9tuXf7vefrvfL31ftfcwCMt9a5lxve+vl+tvX1+v9z2TwsG6n+bldPLKCSOxJwMAEwcBE5kIE2sptVZAat221u77DtEi+vV6f7tv96bdw94V1YPUPLQtR583NEsQQ3Wcf6R7/PEhiGmaJgLe7m+xX/t+bet1Wy+qa0QDMCJ1bxk9snt0RC+VCjAit+aw0r431Z4BUdHUx+8NksIhxo4XAjBEmJCx4nsfc7i4TSHDM5Co0iM9nZ9Pj59leeRpoVJJCo0hNfDI1fpuO/zhSRQgEkSa/PnBrMxTeXyi5QHqCfCb5iUR1CnTIzJDABIZGJOYOdzcSIALcWEqJTMtrFn27phRWQbaZWx3KnOd6/h9tl2ldWKUitNEpQ43oSBMzHOtj0x0801322+bts5EJhLXwut18S6PT8+/nP74MIgJB8svHNzSHQ5g/kFiHBeyMSZgwrQE9dyb3te2t+5uicE40CP0rvUEhGRGFgSDeM/zJsqhSEU6QDnhEQMmCAmQA085kAeRme6QikmSOCFVZkBEGsaTv/txZp6X6eHx/Pj8XOale/quAaluZn18QUUICVtYz9hdN1PDTCbHNLfeNRFKWYiYRZAwgSB8XABERhwalMLzPEvlwU1FIRyysox0IyAazuaIrr1BEj7UWka82cGNBVf33hsEADx/OK+C6QjUORzCERFgZhDGBII0c62ChXASnCtWYUTwDPWhKYUwU9vv6+vX19++vb5erzfkqc5nKVOZFq5ZM5dlmabZIxjx0+Py6WF6qIKhg+m2b2u6a+/rHTLctNtJl3CALFlJyvDCju8HIyEB03/sa1e3XZuBc+Wn8kiQv3/+7evXR1XNYwOEAMmlTPM8z3MtBQcL9FgvxGjTh+QKAQhAiBIpAcK8eyAiFSGRMi4NgOocEIQcHm3fIwGwgflblYdCZ0lb11//+r/atpap3G/3r1+/vLy+XC5XT0fGntAiTf8YZTPzP/+n/0vb/nj+h/vt2rX95de/3O/bOt/cyzTLEcyTaO5umql1xgcs04nHJk4EapF5Xmohi85Cnz4XzLLeat/71gadQllwOU/zXKUWN3eD3kKbQ2YtDC4YVUhqzWFTSmfdab3216/9p5/t8XlK58KzaWT2+TSX6Ue3FSITM75rEflIH0DmSOi9taajrg/5aHhamqYNwREhEVpCZBpCIqF7hgMEQmJ4evfokZZJEAcr5nA7Ex3g9wgIj4SUgiLIo94bFJKH06kWU1eZZF7KNEspnImmkRkfs9+YeKrz8/Nna/t2f4nY81vf9jUA0gMxxlQTUHWEFYYPmVseIRRD/eZmY4yfQ1ceDuFj10MYgJlBg+MVnqamnpYZw12GmAdapXUzcw9m6sOkCIdOySyHOHdsrTN+mGOPfSUkQmRr++X2+vr29fXta++3DK+lnpfzvvZxwXcL7brvfd/7tjVIDgfrgcG361apLA+LELpb29u6byMp1Sw8gcosMqxuOdhYpt2t9YBkSbIAw/d79nvErY/T/B1zjwAI33OhfqzrGiJZkacqcl6m3Dju+9bebH/pfY3sgIbkHn3sesw0Mgb00zzM3MxNvffIJPfozVQdkQhlYPqHNAcAIcgUDqIbQgg6A7Mz78ieSXWuZi+t/y4dsQZ6J57Tp4CKOBFOY7X7vnD/+DkSS4Ig0FTpmaVM+umk+/nnf2rX1/7ta//2tW8X3W/hzW03gugZaZDBQgWJx8T9wOlEZHoPb2E90s3BrMhU4+hnvru5EaZZpqmwkByjoPDBdwxNp8YbEVprqUYRE9P5dDrN81TK08P56Xw6zfPH59Et2OJIXnWIwPe0XQQARAIcuTL5/VsICIHQPO9d167dHCmIx1sMEceA/Z0ijJkQBh6JAFxQaAgGR7TEd07d8QEHpQIJBzY0EDOyuzXTxjQXHPpUHnmQfzwOwMogZJhv2/o/f/v15XVdyhmQPNPcLAwIkHGaRAptbXu7Xv96ef22rVftN9XdmqYdoUMYidhVM5w4i1CdShUREWJCwJFKgKNMUx4JqIiDi1ikEHIppZa6b1tv+zBqzVOttTKPwAxGZuTBa/pwWGFWCuGxkGIiMcvefdWupiI8SZkZCiNBUBoEhBsSRbi7tq5bb9fb/fVyeb1cLtdbNy/TXMpS6pmpZhAisRDLQjLnYK+DeHAMebSUWmu4pZswCSMjZLj2HSDNeqmTlAmRIw8x13gzBk3w/720m1uzrmFA8DCfJpZPnz49Pj621mMQKzI9QkRqqaWUd6bhYSUZ/4+RHs/CMqarxEABxAZj8zbiFJiERWQMhTx9PCXt6h7ukF0fKp8Lnij0evv9179s91upsrX+9na93G63bcsMLuSDpfRB7sQk/+n/+D8x8x9+vqz327bfWcrvv/96X6+ZJgJmg4OdHgahRPBwnk4PJcZEKgiRq9TPT+dlKXu7J8bpgcNxmlmEbIV99242n2Sep9N5IuTePEZKkSYBApF31D3dEJEAPRMyKEzul95aH9fz02kqpXZVSyeCWgl/qIhUSmE6ZuY4wlkHEtvdrPfeBjqHRAApA4beBxL5MOgPbpMiAsvIWqUDYZMYOs6yI8zCDNyOKCHEAcCC8BwZKlIGGwuYcKTAVa5EQkFcqAgLExOZgnVzcviBuEUi5eH8lD/pvr5qv237RW1XNwCHgSUhjEzvmt9LVcQhKIqMQDNVtWEOlCGRcwwfdxDEwKFNygRVM9duLcAH2XNU68GA3PdDHo+ASP1Q9x/X82EHooNU8tGucFR3AICMbHu7Xi+vl5fL9RVyhwzhUsusFG7+7vrNQ73l6e6u6T0odb31h8npqdYiCn31frtubRgk3CNhOj28q1N8/FHT3rtnDtVbvh+pMXAq8R0MMEyN7wnZY7j6fgn4/hM4Jy7IkwjNVbTkLbboF+sX6/cAS3AktxicQNMR1wEwblc2UtAyI8AsM6JraHcEYAkWlFJGSs0YK5tmuBMCEYcDcxI5cZCoCKnWfZ/uNwHWwL362W1GmVAm8QeZzkAVuQAWRMYP1xRE4AIglFK4VJxOkRlprrttt/b0pZ1+vX79y/31b9oRyDM1g4fqUsaVOQkAwr23zoERYM29R/RwNXV30ZxcyrgBHJNxRCyl1DqNiU+EmmvouP+kQ2AkIui+g7sglKl+enp8PD9MIufTshwd0YdT17LruAe9a9pH0hq+8xOO8dlR2gEpID2je967bl3VgiU5IDwDv/f9OfyQGRCe36+tWEbICWamerrhUdvwUIrBwWwgFgLgYIwOpt7NOovlOwVY8EfqA5JwEmr4t/ttfeslXxlqBFmkhTsECUql+VTqRL3vt3X9/eX6Zb1fW7v1ft9XcztNM4GM/cC+t0ivheRUq0idhHCAfUcYt/nIimRi5EwIyzhMjlxFmGXIy1vX+7YjUSQkogAQccBYEdHf5bUzZKWoDFPBaWJhaT3W8J7ursJYEARBhiFyRCs7IIK67m2/revb7fbt5e3ry9u6N/Ms0zKfTrWeRU6RGD4iDApAyZTxWD1ILc0TaUQolVprOhXGWrgIM2WE9Z5mqr1z6UQFSQYPbahMw/9jGZ267b2t27bt2+N8mqbpfDo/PT623pEJAcx93zYggkzr6mZ55M8kiSAi0rjbByPyeBNzYBxrIc5pAmGsksIaHgYIYGYBUVEkiBJAva177P2tykxUzP1+v75d23Yn4W6+qxkClMKYpdDImqgf2I1I+PzpuYo8Pj70trd9a7b+7bf/te037a23/mZvdZJ5lsJYuNa51LkAJWAyC6JYxww6T3MRQsBufd9s39u2beZGgvMyzVROD/XpaamTmAXQUJ6jiKSHmV0uV/pbOJ2wLiLAI9WWqZRiDtvqAHs4nx+EuEolZvq7GRcxlcojbighAzzcM45FwHAYy+CrkACk6XiDkpAODbcPKvOAijAziiBSqSTMgEnesa3pPQHT1KwfO1sWIYKEQUWOiBh5lBEK468SkVpQICnqXEQCMg29t9xWnSaD08eKGJBAJLUuD49PD48P81xvd/S9q1oE0HAPj/XsqKl5HKOAGB5mY0o/KFr5d/dTGAQAwTpNzNI7hGpvFmFSCINpVMkEd4AoEARxxMOO6CVmYuLB9ghMH783dZQf4sYQEJIgSLvd79vttt5u98IhmPva79et7UqIwrxMky627/00zftmm45g9QzKcDCHMV0JR+25rnpfN1Ubj+mh9db2tq+lCqZVoT1Dh8siHBGZqEfXtvfeeu+mY9ENcKjJj8sKAlAygtRaPja859On5fRJmDN21dve3tZ22fpdbY/QEXQ5gs4yQHvc7+Yj0w0AMJF4PhUkIhk3EEIGKeyeCFAqzQsLMUD0pqpmZg5USqGhpR/REUTEHKD7fr1eCdH3dp0vv5c6sVSpC08Pp8d/WB7+kaczlQVlAvmB+kCIM1MwpRAkATNABlJYGNe4YRNL6OZN+9b7FtYBggrxMW0e0yHv3TKRuQPgYAagwYh7673bvo9YIhLiwTEuwozCCWmm3nt/59UEoRCq9z0TzBQzaynLNJ2X5bQshRmR1vum3afp+UNZ5AQZfotDo5gwZuFwjNAQBIfOPQEi0SO6RVPfxxUkD3/cWMqEpjsM+HxGmKUdaRHATEPYnJFmqS3MiTlxmBEAD9PDIBYyIVEkOYBFOkCPEa6SiMCHuffDeZWUkaoe3rfdUlsoqaV6eEZQSiWZedpICrR9v923t+t2vbdNo6mb5VgXoJOam/p6XyM95oIJgqSqg64zAK7hMRyuIylcWEQECQnJMSw8PZBwmid3A8S9G4ki8wRIiFsjSoSgyhkfr4wJFCCABUkSyAPN0JwBhYSIIdFGxMww92eEu2q7329vb69v18vlel933bsnMslMvBCfEKcRC5RDwungBkQAiESYMfhmwHD0yaUIClWhKlwKszCJACIkZUZvO6IiFWYhLt+HdP9xaTdb9+16v12u18/np2ma57meTqfz8JYhmqr7AY3ue4uRfjfUq0gIh4kCwpEYB9vDgzJxCFOJkmmolyzCIyBy9NDMCCBj0apb0+zXeZqQcddY17e3q7ZdihiARgYx1kI0ujdMQPmQ/IYAhWAqJFirQOF4PC+P59NSJ0rUXS0svVZmIi7MyzSfzjNyIsM8z1OdTMVNCnK6h8O++/W63W7r7b6pGzHNtdRZzg/ldC7EEOjIhuzEgCKeamq3291gL+d++gwnPq7KLFiqeGRvqtqZJ+KoM3GBgIi/4wkBJAHiUEqNibS6mR0s78hELlykElF6ZGpEHIahIQIZ3juHCACLUoGZCo9oHyDMdNQ9gwEgtMWIrPZIEWAe3/Mx3wPAIwggwpGJheusXJkLZFCRgSKN3qKvofjj53i3h2aOad67HMDNVd0Bkdx8FIrvkpl3qxOahVrGuyIs3+fMRGOcfsjEkJCLFCnmDkBh6R6Dw8WAQB6BHhSB6ZRObuA+bNM59M4jujbHQj4hMj+SZo9tb+J4t/e9b9u+rvskUAju9/12WyFgnooQ1SJTkbmWudYq0sDAA2KIq4WgwPD6EAxVZNvtvm2AUKe6t7bvW9u3qTKECQNCuFk36+6RcfzuTE3VRnpd5Ht/PpKx3r1PiExcyg/6mofT8+n8qUiLrua7evOMJKRShAAgPdQswsEVtMW+uaplBAlKoWnmOhViJDl2umWqEaDdwlMKMOdUmQgZg3B0+EkwZLiJIziQRhfgve/ripDe99te5qEtKtNDXZ7RvVDBzGMgPs6X748DYWIKJmcCIMzjYmYY0XvEXfult0vfrm279r4iOI8YVh758hEW2q1tPTwJfaQJE0JFSU4biVba3QGCCAoyMWEpUgoRj/5JtWnfrTd1dWYnGpovGCAhYZZB+gbMxNb0ujcR+fnnf/7Q8nKGhEf68f1+10occxhmgKTBUQaEAFSNrjFAaeYxdlaR6RaBET4aKkQYRCM3TTeApO9Wg3FlN03zBMBx3YE/xjs0FnrEiDlskuBxIFzyMILCx9KOgBUEMmIkmTTUrbfNu0a3cExgkJlrSHVkznXd7rf1fu/r5hYwtPlMBIaRaaq99bb2hGAgYduwE6OHM1OdpoxUVWIsVQbNXkSKHBJ6cAgJM0fAeZoRsXfNTPNU87EsziPSmqrgPP+h9h8peSOB3NM8XZt5N0wYp2Zkqhn48AxCpPfe7vfb28vL16+/Xy63+9Yimcpc6sQ8Ey9IE1ABZDiI0WPRmWE5BEyMB/M/E4il1MoMDDkVrodRmUh4bEjNBpQ+AA8JCb0nB/zHpb31+/X2It9mKb88Pi0iRCCFkOlgRjBSIU/vvcHBEoFDXjdwf4P7SjTKkaq13g72ShH+LgaDI9s23T0cACIYAJgZgCDAzfvmV9z01trt+vr7C2Y8PJ6xiAMBCQ6AAo9Ra3y8sIT75d/+x15YVdft/np9+9f/+S+///W312+X22UzjVLLzNPMUyWkAN3b3ZULlUkmKeVUfvr0y2n5hIH363q97q8v+5eX1+vtbt0gAGuZJlzOtc4YqRGO5CRO4sQRhAcGyqOp7b20fi/TUksB1MROJdjT99Ce6xrEmvcuc/xMS9LpcfpjPqSm27oOhw2MnLFR1MceOgCRQyPgWEowEpfC4+YNODQWnFCAhnZamItIqSLChICURJxHHCeOcHftaJoNjTBIxtWBSpVxWQeIiPFSY9+TzEmAMH1iBklASqqFivzwvRr3jL7v1+vl999///rl2/2+mQYBE7qH2TiNRsNJxMSZ6QdVJvUYMDIgESGzlCKlCDMBgA0JWSTLmI4mYorINM3aMcPC0inHlF1Vuw7/XiJlraVWLtNQ55BpjJiV42WSH6QP76akg/g9ZuHazfYObrfLre9tmad5KkjgpqNjRUgZhAliZJrqdFpOjw+Pz0/Py1zB3T2vt3Xb2nVdzSPJ9tb3fW/7pnMJV8wYs7HW2tZ1PAMY3rBhLCVCSEgaY1QkPEDkAN9ZJB8+Azw8Pp4fn9IuXYV4rvPz0084LZ/Nmvatt3Vbr25v3qNvXfdMGyhaBMcBSynIJElQ1DwiRAqTqHrv6mbbZph1nsrptJxOp9b6SBAcY61REzLBA8gJhDPILI2C0WBQOQaR1+4YK0QFF8rCWH4cyGMRjsJcOMcxi57pqrf9+tv1y7+9/u1fbr/9bX390tseoaUwTUJYCdEiDlSvI6OMsHnMZKa5lNNcRqKNWet9SwgW5CIyTTJVmUqCR/QMizBGnouAYocG5onORYDHfDUGSgTfruvaCMnVtm2f6vT/+b/++CDHRbNnGL7rDAej/+jakwE8wMj5aBP37m0z6waRBCM8JzLD0xGCQIhlHMHhkJHpiIM4RYRA4WAKquhOmRxJg3o0pHMZGI4eMCxrkT7gGjwSqJDGizpicb//ENJzObtpqEWAIGVaM011iDyuCixSRCZiBu4KyJCMmRUoiTwsNLr1nt3sQPQUqRNPnOIdFdI9pLAgR4Q2QIp0Ixq5zRkcmR0SSimlFBYkonla5mkZ4HWAHOqVHmlp1iJmngtN0x/mi5Ftqea99xgxLZ7HShsw0tKOE97d1LT1drvf3t5eL5fL7Xrt6gHCUkkWLDNwBS5AgizAMppuGHxFQiIsgnPFU6WloCAK0XwuiEtYQ/DKVBiFvs9hISNZWFzc021YXW0IiCP+44H8iG/Z1vV+u7W2R9rAUpMQMGZmYpIwjhDObm6RMDQuBDTcmZTCXCTIDd3MVA0HEWXc3DMxgXOsciMIKQkRRFiYi5SAJGAPCM1962vX7fL2drkVJl6WwgP9A5YQPqZX47b6IW4hfPvtL0bQrb9cXv/65be//Ntfv/z6++Xltl53IlzqVLAWEiFkDGt9X7XOBXLyaYLIh9P800+fwgADE2jb9H5r29pEsEzMU04LTCeUkpmOGKXCNGOdyVp2HYGoR1QikAM5S5QKdYJpGTIswBVth70FoFrei9ryBPX0A+I/x0gjPRMxj+p+oHf8QFBAwNDBEYEQD5LQUMANJPZgTo+4tFLLNNVahWX4jgIJwjNjVF8CeM+wjwSwMhEQjrsdERAgWdL4VUfGgW+AMEYv4HIcuAVEfti0Z4RpX9f76+vL16/fXl8vrSkkMguTGxikxwFC5+OfP5jmfdRHMEekITFhEai1TPPEhJmBlqaBkQDedZMoCFwKwTwRQGsRntYTKQLxWOMPB0vhWqVOpdZCPKaRQzM0QN3MIpl/NIpjWeLp6r1bM2sju7Hvm/fWWyfCZann8yRCHvF9Sz3AZ8LgSSL1fH54fnp6fvp0Pk2YERnX6/VyveDrkduhatu2b+s6FYqw1trAn+173/YmPDpzKMK1lBEt7Qet4P1bMWpgBCRk/pB+AYCnuZ6Wqa9sxCLzvHxiOdvZzPu23m5v31oDt63vuK+pHTIpg9+B5G5szowEhSU9FGDMBoAhyJuadgVPSqqPU60iQr331vqgCsLoPRMggZAgK2AFqAk8kBlHi0KAoa53IEnkWirD9KP5DaRAFvSCR2wQYkSg77q+tdu37fJ7u3+z7ZrWIYNkZpSjLT3CxQGRRCQx04MAK/Npnj4/PZzmWgu799Y2wGBBEqFSQBgYuu6t5xC8SZ0pyy3vd8uuW4ZhEUTKzBGZ6qFmd6adWVx93bbTbD+2IuAdvIPrMZN/Vz4firkkAMvgIMbRnLUe1j09GZEHQIYScSTCADEW5hEJA0kZPNBGwixMg01ijmYjNE0SOOHgyI7SE45uYBZIaRmenEiJlO9am4NW9MEaykg/zY993/d+98gxOGppFu4BCZjAIjwt07QwC5jFtPu+JmXgGLo4uEYzHQQLRCpShAojY3Iaunu3DIuCGZHaEikykhmIMSgdwtRVTYRLKfNS52Wa5kmKEImx9d5cFTwMIA1k2CXwhxSoobvpqr23seoiOm7MhAA57oTWtW/7vu3ruq7X2+3t8rquW1dDKnVauMwyLTItUmepw0VfiBhwjGQQ8iCOC8Nc6DTLqRJmCOOyTKVSRMf0giAEgpCQnhZxzMkiUM0VB4zbLQmD/AcIx4+lfah8T8tpniYSTAxgIGEqhEym6ukkRMK+d1W1bhl5DAEJfUzdRdAjEmpiQBILjWUVEQZAJFAKU5WK41sEgYTn87Iscy3VwEupECYkENha35o2cyByQPBsQ7arPcIQnAVFsH/4VJg57VthQO+4Xbdvv12/fblfb9va+m6lFHBKw+jJtcyVb6tuayNGhNnd+76v61stlE7rtka0Wun50/nhaTotIhUNDQhEkiggoU7l9DCVotY4ddO9JXpi1Eqnh/L0vHz6fHr+dF6WaZnrNJfrZQfu0y69p6pHukxBjAcV5MNFWEpZlrN2V7WxR2Vmpveu0QMA3jEONMphwlgbDWUJYaEcgD33zJimejqdRrN7TAHMuhkfqad1WtIDuvmgjI0O0N0Bc6SdZ0Umd48j41JQKte6sFSi4hlH2h9+mABn7m01699ev768/r7eb2bBPNfiONIc3UbOFhIR8bhHZCYiZaJZqKZFEuZo5gEGUwgMISHGUDo8IMF6K1KneREpwhMSeGhrvbc+crK5lIf5NAB8xMhC47MjMQFxpSqwd0MiJhJhs48z4HSMzba3/dtl/Xpbv7V2HTxca32q5fz88PnT+eFxScC9m0XqONgQ61Q9XFWR8PHx/Onzp0+fns6nmSkj+tvrcvpaqlACSy0JsO397XI16+He2v52u1+u29p6Nws+wOE8l8qy1+l237ddXXOM6BiRWDJzoJQT4u/0NQQq0A2dCWupsJxLgd5tay0xNW5bo9satzXanhYYWRxA3cINN2t739c2L1OdShh6B932zJaJbtE26929NWtpFvNSRODIdO4D3ioslZidgYgBJ5HztJwLE1Eup/nxfJrmuZQZENbbGzSTZkXqMp9+HMgnsyULcSISJBJKRiksQlyKTPMUy8zW3Cgh6jRJLQCQkcyFqaYgBIzVC2YSJEOeT6fPn54/PT2eTxNBmqsIz3NF5kBo2jfde1tbP7Wlafe5PjDWL/L1d8/b3feRqxRxAFUz3dPdixAxoRCLk9SPtcS652ZmeTgaj4170rvhzzETAQmSBwAs3RGTC8tSalK0DKQkHk0+inARwYAMhiQIJ6YkEiEuMCLLAyjH+BMZj2TaPOrOwLdpEgdyBkYCIzOiEJdDcMAjL/fDYUX06fxph9tl64GKnCRpAtnDLUIogUqt5/PDdGLiROAM3rdcb324o7Ufy6WMTBgLwYQI7ZaJtU6UEK311rNnQroZCUByFuSghLDIfdu3dUeiIjyfpnmZ56VNcy2lAIDv3ppmjMBCWJZCT8s8P32E74Sb9d16M+0IWAuLFGIaY2ZzN9N1Wy+368vry/V23fa99dZVIwFZSp3rvNTpVOelzqc6n6Z5LtOExJCDUB1DrgsAmUFIpdD5PD/MMtJYRMo0VZGFMY/dIQRk+AGmyfA0T6JxdlnrvXfdd/vfgWYZSFAKCzOb26arhsYIEkIIiEDgIuwxfrOmmkeIAiKhMAekJ/QxDASkEVfE4/sgQjRapkXqaVmQEzAsA4UeHx+WeWbklq0Ig2cthQDWBM8cfKIE8sgRMNstzD1CxbEm24cldWbeXr4VCkW7XV6ul5f79bqtrTczSx4L5u7bvYugyMjkc9cIC2268nq9vGBYBN6v3WyTAo+PCwueH4pU6Nk8LBLcMgxGXDcs9PCU1iAMRchVlgd5/nn+/PPD86fzw+NUaxFCJgBKy9w3arvvu2v3x5lqpWEx+CjaEpapLhDdbWyXh8l2DIlzZNvScEEei+CIgDHJ4kFjEQEcG4ujtE9zJcL3iIgYPWKQADCNm6UoC8EQ7wkSRaa757iry2AhfdcMFqlzmaaZpRALZMLxn/2ja0+A2/2CEF+//e3l9cu2rxEpPEHxDHfvIkOum0jEJIAEOZQ8SeREjGgIw7AdgJDp7tb72MW4u72PzDKjZ/FSeCw5a/LeGBVNHTILcRVaTnMpR17rsAjiAZrlQZhkGuAJ/DsfdWaa2327fX397eXt9+vtpbV1UMk9cpmnnz49PT8v0yJ789j7mGoM5G0mqyJgJ4bzaX54WKa5znMpgufzdD7Pp9M0TUKl1OVcymQet9vats1MW++3db9tzdwCkjITswiVUulUVIORMbc9ultgJDNOUjOzRUZYWsIPEo7s6+squd3f9v3q/Waq5tQ19mZHQHWPXVODgSeWSUTIDPve91X31TZrm5nC6UzhqRqtqaoTcia2FqphaNbTPVqXeWYAaM3aHr0ZQkglKcSFpQwD+Wman0pBRj+dH56en+c6E0v3bO4JmtwxjCE+DuTTrd9fQeYsjbgyCTISmIQLpBBJlToLWk0nhJRSiflIc2XBEW80sgDGcMY9rePBJESCo0aeT6fHp8daKzLtfd/2e9dd+76u67Y1phpOt+lWRIgFlM0x4ohcH/vyyECMkkgjPOpHtEhYgMXxmMbdJYam45AdwuCWY4bhe0YaEHLleqrm6pEGTCyIObgIdZKCiekJIeO/D0kswJLAEZlEQIyAMnj0gO/L/YBhpkcCs0R4t9SQMDELibAIMQNR4ocbIxE9zw/FXLl4OBPiRLZIRgQkCedSlqXOp0kmBHSptVRHYotovfXWrIfb4Dgf86dwtzGJAcJClJAGqqr7Md+QSgClBIOABbjFdmvrfQcAEu679ZO5zhjJCzGztdjvFkcELp9qeTh9/vT8jx/ly+Fm2cx6uBYpwlSEiNljSIP6uq1vl7eX15cvX7/dbrc+RE9EXKrUpc7naTlP06lMp2k5TfOpTJVLyWOkH5gEBAPFmhlEUAqfTtN5mTZANyMam1URGcl/TuDphq6BQQBBSRT+PZ0B0v2IOP8PS3tvfbut01R6b6+XV/Xt9X5Z+7711k0TgJhJWGLg1gdMIg6zbwIyDSFnRAwrPJfCKAJSuDyezqd5aXsz9eeHh+dPj0AeobsZEn1+fDyfTxl5Hf0Z5jJPTNT2VacKcaoDRA8ASDJNvMxmqtYQIwnyw9tiZv/3f/8XQjeKS1t/X29X1zZmlomRqJZ+2283XXc+rwIYiOSR29Y0XM0JQPeeidum23bL1FprqTKgZg/TgpQesK96e9vbbkg7JE1T+eUfps/PpE1NdTnz00/zp38o50dGDtXdR8rquTrQvu/rGntXt5Bynpe51lERP46GRjB7CqcfRoR0GPJUKrWUIkiUAdY1hucqMCwAMTCJMjIR8TDtAAHiSJkaWZyRY06fAIEWiE7sQCn1uIkhRoK7w4jEQAzmQkRlrIqE6yTTPNVpGrKo40pfyiTLx4r49va7e/v9y79fXl+1OUAKC2QxHfcUAuAxecwkHKopZGaoBRCQxcz9WMKPHDvCAUM6grQBMfCAyHua6jCFA7pULMEOnIDENCJEBz4qISCzlKmWJYPdYG+9NTVLkiHLt4/IGo/Y9u3b69d//+u//fr7397eXlrviFxkLpin5WFZTqUIZGrXvitEVBZcFuFyX7WRMaYw1ImYQW3fmntwYM5LPZ3n83kBns7PnwFIzXrbb23rrbXexuBqHKyUwJET0Xmu59NDJAkyJoDH7o0yBfFhXjIz1foBK/hQEQF++8v/oN/lcvm6b1eEnulNqWnuGtu+r+tltz2F6sO5yJPUqUxzjNTly9vt7dv98rbf757aOyJCRLbm2n3Q+szAnQIhPDJb177vTETppB37HuGG216mnBYuFdQQsE7zw7JILflwOp9Oj4UnQhGhCYmnUz09PCxL+cjGBNB9u/+3/ye4Bi91Pi+n8zSXIgDrjXtD7xk9YTScRYhIBJGR0z3Dh04zkVFIKHNoWLTv4S2jv3z9HcMLyzTVn3/6+U//+Keffvr89Px0flrw+RNCuOvXr99+//3Lt5e3b6+Xy9tr6y0CAEsE+Oiy4eC+DDWrqSGAm2X5cYiCyZhE7wiFBODhLIr3N2gokwAjDokdIrNMJWd3Bfew5OCKgiRIVaRwoSQM/CP9DYA4BxP0XeWOEZQ5suYy/aCjAwIRRaIHUhDSUE9NTFwIhVMoaHyuD4FpCDhJ0SJcOFFYaBJ6KIhLkeZeC5yX+XkuU3h6b22763rf13Vft7Zue289LCGQIgmACMcSUcRECnMFGLpAGM9pgA4jBDLSSwj7EEXuaj0yATuAKwI9nOYqUpkTUHffb5YOTFLn0+P5p//05//8j3/+p7/++799T4sYmcYDizvqZYYHhFnsrb1dLi9vb99eXi/Xy31dzYNYhIWlcJ1KnaflvCyPdTqVeip1IpYMMLVMjEMnl5FjbIIBAJAiPE11Oc3p3hpEZO8GkHXINpiF2AF03919tHK1VvdQ1XGpIuZSit/sprf/99KuXbd1q5NcJibx21bXrW1t79rNbdghREpK4B/6yIT3w48Qh5Xd4vihsUAlKqU8PT799PTperlu6/r8+PjLT58ATa1dbqtnnqbp8XQandGyVEacl4KAxCjCp2VhGuFBGQB1npeHB3Pd26Z9N+vwwQ9u4f/tr3/N7CbZIDa3u9nWrVskYCS6p6aZ7cDMU50mKVIQUNW7+97U1fdbIxL1aK252bjJQw8UnkSkUgRaA9NNh6ScC2OZztP0vAzkSp1xeZT5lIjufihu6lRIkCT2DbbN9gaxBRIRVh5Zuj80ikg4mGmQ3keTm5kH4KuWWmsCmNrQzxMQxMjkQJaITIIx5pahlyeiyKA86BTDZzP2sXEES5IUJuFapiICEBFqhiM6EABFRIoQDy8Q1alM88RlCNzGTupAsn8s7a9vX3q/f3v52/16G+xLYXYbvjx3C3PIIUBFHsMxwoCRD0XIwpZ2eLqQhtjdPXpXt4EdIRoxYYmJ6eamCpiRiZSDODYuOQAR6QAw9Nk5VDPEQAUBi0PWlMpSuFR5h3gcP+6+7dvb28tvv//t27ev99t931vv7o4MxCTCnJHmprt6Vx6IjAQm7bszQmFkyght7R6xdy3LXLu2JJChHqjLw/mhq3e1fe+3y23YqrqHZxJVQqyMc+HTVB+X+XxaPOl+b1MpTMdShhEnYUaEefJSmOjTwyN8eB7//r/+e7P29va1tVuRIEJ1siBHjswIJbHlLMS11mmalzqfADDUbw/naZ4RixpGeDcgBMxD/BHDRQNUGImRGJkDETMAiKVMhBSuCmO2hJmUQaYwAJWn8/nxoTxM86nMggVTgAWllvlUlhOXgmk/xHis61/+v/+3Y3Ga5vPjw9PzPE+1Utve7m9f+uU1e0NIERn+URyWYUKm7KBuEekIwiID4ZAUmurdMzuY6b4z0jzN67r1rto1I3/6/Pzp6WGeJyLk5La2L79/e3t5vd9uvXePTODhsBgguxFy/25DVTyAfD/4R0rNshzinHdqTL4frQdrYRTowyEyLsPCgXFKMSruJTmpUhWeWApxIWZgAhp/zbBCIsW7NN6rhBp4UgT6YXX0Ae0bmV50wB4EsCAUAhaiKljfSztCfHw/EFEKSRWeSkqSoAQtFWku1SNKgWXCCQHVtOne9ntfL3tbW286pCQDm8iZDHC4+TwzcpgsiQRgLOlyCOIYKSLcAtEh0C2se9iInkZAHGhIBp6kTmXKRMEqaFzKPM2Pj4+//PIPf/7zn//0T3/69a9/+V7aMyPe5SkJ6WO37rG1dr2v315fX94ub5frtu/mjkgsRcZ5Mc11Pk3zQ53PtS6lzCOWc8D5x5bjPVkqM2AkCxGBCJXKpYpUUTPPdPWIzJqEPEK5w909VQ0da621CHMiIrO7B4vUWlddP36vfpTRueu+X16967b38+lh9oDWzNQyciQtClO8Bz4dM0uEES9CTKWWAR4Z/3MAukVkmupPnz790z/+qRC/ZD4/Pf7y009ItrV1b33dtnQnhGmZEeGnXz5t6z6dpLeu0YJsOlXIbL13sx55Xp7//E9/Dojb7fL6+nK5vH58VyLh233T6MruQsl097zvrasCIBAFABKVqZ4f508/nWWkJFMm5t5aV9svthRfzmditk69h64rEJRZTlapzjOWCOot1uvgqGCIEyJWm4qVStPEQK66+103VSlQC9d5muvJIqWaarp7b31A/ltzd8HRvP5xCsOwcU6VwrJ3G1B0rmWep1qrFO56MNcCEtzDw3SgkYYIHKVKLRWPPEYPMAAYgxwGQj4kd0BEcqRsIlKtUylDrWLu+l7aqdSp1jqKHwtJkVJLAhyh1dqHDmgqM0x/fIzXt6+93a7Xl/V+x+AhxQnX3tfWWtvVLDNwmlCkEtE4ABGTGJGYC5bkMS50T3M1D9d082GADwDOw9bzfnp5YESGDlj48Tpl7ztSRpTvYP/eezgKz0Xmp6eTlCdipEJlmqSWfcfvw2x327bb9XZ5fXtZ13tXvd/3+21Py4llP8291TCP0L639JhLrQXNIywxg9JrIQK/XF6IgplOy/T4eDbTte2eLkUQyd1a123b7/f1fl/jsLUNUAcJ8Wmans/T03k+LRMg9e5dXf1QQw6ZFHqf5/rw09OyLA8Pj0/PP328af23//4/3q6v9/ub2TZXKlWIi5RJplpnWWYRIaQZCRGZBaQ4c0Fczmd5fFimeeI69X1z7ZiK0YXJOaxBeMpEZYI6lzoJF2QmZBSutT6E833VvZmaI3ApE0txA9N0x3l6/Meff16EORIDKInrzHWhaUKpHm77PfwPSc1+v/3Lf/sfmmhQ5vPD+fFpbPxCN9c1rUHqhIhlfocHJWCICApBenNV7RBMsxSRTOAKMiEYJAWQJ4W66tq67m/X1y/ffvvT13/8r//5P/N//S91qsu0nM4P54dHYu7au6ragfrNwT+AhIRBAM5MIIyBjEZk/gEhtJzzNMVYN72b3kZk2rsV7hjHQwQNk1siDDtZIIYwWHEGKjSXstRSWQoNsNGIPELiocg+rrYWMV7oTPJAj1CzPkjLw92GOHizo9OnRAGYCBaBSVJwwOrzxz4EkJArl/PkBgEBCaUAzzknGJJxGjRTh27Yu9329rbF1tlTEh0pkDIR04deO961kUXqVOd5WsycWZhYhImx1jIIdADj4k4iBQIwjuNMCk+l1FKnUpd5IZT2BJXPD+fz48PDw9P5n/+Pf/rHPz9/+jz/gNX7XtLyQDj31u7r+vL2+nq5XO73dd9txGuwEAtzQR4x25XLXKal1FnKRFwGCST+4ERiYgaOtuq7gxdZhpIvkQAI3QbkKjJTGGsBPJwv5B7WDQGXIScSMbOunZxKkVJ+EC//UNpHmPX91rd9NW97W5iLB/TWvCtKQQDNtN4H2GT8Q4e2mEVKLVKKuyd0wqF/ZkKqpZyW5XRaToNKX8qyzOfzybOpNcg0HRbRTgIOxpXY0aBrtiTjCtNSwqJ7D4uAzPSj94KDh/rjVwxhWrTF29Y6BBW5b2rmhDTPtZZSamEWlnJ+WM4PJwRP92muUuW27fe1iU8FlypnKlyK1mKlUBIkpluu967qEHy/9vtVzRxBvDhigENh5KVyYY/obU/rpD7NzMgQTFgKQ1Y6P2jv5X6bzCAc9rXvK0+T5EN+R7QO1yQh4TtSbRAeT+fl8eFUp4KMJMOdgNos2rAYATJKlTrXutRpnqdpJqSMNO/miJTC9P9j7892ZFnWNDHsH2xw94jIYQ17OENV10R1dzUgEGgRECheiBfklZ6JzyToAQRBoMgGdCGoIVItqLrUVbXnNWRmRLi7mf2DLsw9M9c+VS1BdyKO72ntHMPDzeyfvkHdgDCIBbEdrknIFEIIIeTUPf0ANuyIqSg4cogxxth72oG7qEUTsaK11FIaAjGHMKXn0O7u1+u5rOdlua7rFRTBEADdVLS2Km33d+2NJg7kbqrYB2+b8DUiOpiRunRGct8AXWyC3M32LePeKhoYqHdkTS9FOrdeDaW1Tr/qyak0F9KUDBGGYRpH5kCb0Sv5F/ga19rWZb1er+dlXVovrM+LNkvMd6dBZGBwBAgUciSkoAC+tg2RDD4NKSZe1/nh0YhxrWPzZmbzulRRQGoi7XK5zuv5cl3LYqbExDHYPi7fnS+YKJjBXMrTtZzntTRxwJhiZjodhrf3h7d3N3c3t/f3b+7v33Iaz/JyIz/++MMvH34p69WspkgphZRyynk4DMdTjjSEGHNm7LAUUmKJYQgBhhjG4UQEIYayFKkFdEUturY2t+Vca5U82XDAPKU0xJA62ymmNA351jxel7as0pq6IQEjRY55TDnHfBiOd6c3AUyXK4CReyAjUmtLq0tpa21V2/J8F1Lrw0/fNQMBzuOhnm6I0E1NVpOSIuUUUs4xJd9kRdTBGZwDIyihgoubmhZFFSlmDRmYeYwBlGNAqU1FHbS05TrHh6f09npfSjUzJE45D+MYQjSH2mQtRYEMyBw3egLslmq+41+QO/rp9XEVsw/JutzcC599p79tuLaNCre1dMVB3EiNAqIyaxZ0IMwxjDGkwJGYAHFDyCB3/bnuYGJmfRrn0LuYYiaipTbpPqPbjunOE9jNbyPjlOiQaGCMCNzVc15XVOiGagwwBBQzU3Bkw4gAiOrQ3KoZmnS7iqDGTYNZBBBEQ5SNU4Rb1Nt/LnOIMYUQu2UOABJzSjyOAyJ28zNRQUcE7/ae2J1PEdCh62HkOEzTkWmU5jenw/E45SnevskQlmrkryYLat7MRFVE+mT5er08nc+fHx+eLuelVlFF6r4kvVkeiCJxpJBCHDhmCqnTOja1gU3koEtwIaF718forRRExC6e1PozMXORTQakRspRNTIghRCZQ621NVHVEHhDKUFgdncI4Z/WkO/4rFaqLtqnaDEld1gvay2VuVRmImq1rcvaRHR/szlwHHIcEqegpVts4ibl75BSOh6PyLiUa5Xq5MSERGVu58t1WdZSy7wsdEa4emn1aT6XWldZVIUGyDnFzK6kmDyAF7len/7D3/57cxMRNQEzehXdmeg3v/ndp4ePn/9+Pl8Wda+rsMNxGsdhCiEQQwgQEx4OIQVyMwB+//bt26/fl9aWRajmYANnbt6AfJjim7dvY04fPz8+Xh4u589NKlqcz3J9KqaAUPvbKlXcFHxkGkWllgZBA7gSCHoBRW0cAxAGxmGgu7vJWvj0sVwvJSYzlW/vPe7PxN3VjLd7MyQYcx7HfH9/e7o59ETPHVStg5bn84LL6gQhhcPNYTpNeUwppxTTHtpDU+7poapRUHM0J5Ou94GmgJFCCBwpJIqBu5omAOzeoJvkIHVfSPPW2rqUy/k6z+u6VBNzQ3oX4d3LuiqlllK1qbZOYjffjK42dbk+dTgchtPNkQhUpdVaya2IiKipu3dyOHqXbtonfd1E0dHVN/q3gYKgIDBs/UgHN0DosBPs8Ps+zXJ3RGauqk2tIhSzubOq03QIaXAf4EUNv7cNWq1Lq6U1ac1q1fkyk9tX9yPYTU4pp3HKUJuvovNaWtWyttaEkKbDIY+DuZZakEndxcS9kz5EzS/X5bp8Xta1lMoI4xjzMIQU57XMaxExMF1WSox9VvrwND9e5stcSzOOcZrG05i+fXf7V3/y9Z/85ptvv/763duv7u7fXBb5b/8vf/t8Xl7Ol6fHR2nVVVYEQmReOllouc2yDnqf8TbGiF0iGiyaZtOReRzi8O7NeHMczNANSZvLWp4u14fz54+P18tlOFo+OGcOiWNKQ56Ox9vDdDcOdwDDvGqprgqmDmLu4Minu7u3t6ebaRrjYHUppaI3ZGh19fq0rMu8zLXOTdYyv0ypESFGAAMCy9FSVEI007XVWua62so0juMwjl3TvnsFWQscCMDRLLCDWy3zuvhaiokEwmEc39zdRKRWVm3iajHEnMdpnI7T6XA8IpMaqBkgxZRiyiHE1uRynYECcuy+6LsKq/cmvCOiGyIHxsBfTt2ga4+4YwfKbtU6wHOo74u4FzFdTMWbeVRLGbKHybmqqXkgjGyRMBIgEqJ3D2hm72k6uAOCE/kufGcABmSGOqB7BIAtc1YVU3N0JCKKxMcUTpnHgAE2SfvXt2HuRWpzUQYDckAyRHUCZER2YCMGZ/cQiCPVbHXUttYa0A3BulZ0z3Do5b2BXYPJu4KemBkixRjHcQCC2pqupZSCAIxobk7qpmaiQF71cg3TZfjqq3j/5u2bt0QEQ2aOoNiEnn789Df8GPRVN6g2qa2T3+o8z/P1erle53leW2migNRRxtgtPYC7KFSIOcQhpIE4dUQDmHUbBdocUxk644h6PwccFHFTs6utlGJNRHebrj61WhkiQWIKDEMe+uljZsuyuNswJESIkfvk4j9mD5OGNJ1GJKilMTECuYH3Teybc1An5cUYYBw6nrK7SE3jmHOOITp4rL1bwH201Zu3pS4PVpc6G+hay+Pl/PT08OnxcalF3YqssJiC1FYXKerqwBRxuslMIcXkBmEIQ5VataxSShFVd88ppjyN04vjAiJO4+HpfF5nOT9c1RwNE/JhHE6nEyA0KRQ8BoyBUuysSDpO0810QA7mFHREjUtdnpanaYpxPHz77btxOnDM5j6vl/U6l6WuF62LErI3Ik4xMphenxYTlSZdlpUzciCpuIq2ebmQpjGGTKLVTQidGU21FZWatPHrRNjBDISQARDZKWBMnIc0TnkYc2+Bw67T1RU/gZACcuCQQkohJe7/Zuo+SNSMzLXHwBCdiIlDXVqr0quGrszGiIEwRU45xhiIUJp0LLiqqvSq2c29iS7LMl+vy1LaKmaOhvrKrQfcW5VWRKpqlS6op+aA2L2HQ6TIIcWYcoyxOxF0Un5viXWfGNgYlrhLmRG4d7VHxK74iFuBZIqbUtTGOSDspN4QAgcmdHvGDzqiKFH3miNvJkFUkfloNp3QPT2HduxgYXLazY27BkBr0rn+KYXDOIw5lgYwy7Vc53k9n6/n81yKIGEehpRzE9XWGIIBNNM+81irNLF1LefzWVQQYTqMN8cppQhE6rqUVUxV5TKDqS5rAYDLdZ3XagZDCodpujsdv3p786e/ef/P/+z3v//tt1+//+ru7u3pePvzpzPA3z4/kBjCkAceJuwTDnNCSilMQz7kPIQ0cBpjTgmRuhgUEgC5DgGHcRjG25xPAMmN0BRaWZ4enz5+/GH87vOnD3HQMLgFoMB5GG5Od1+9+83d7fsp3xGNVUkV3UGbSq0qqm7D4XBzf//29mZMg7hqzgTEpGuZr9fz0+On89OnWmeRUu0rgEO/CyLMOZCauKcEQwIicoNWXKRqkwJdlFljjLQ9Ym0m0jYEdsczaXU1U2kIEEMYh3ycDjkGq7k3YIY8HqbDOE7TcLi9uR3HA4fQGWq2R0dRk6YYO3R84wwhAAI7ovQ+FQC6bcojry9HcOyjpS26Q29395+9zd2xz+8B3VEd2Ezds6MANaCmJl17ET2gMTmi90SByIkIeyYODhtAAn0DQoOjA0NnhdBWx6MqiqIBQIe7EI0Bp4iJgMHJkfwLR0F3L1qriaAr9f4AbOgYYvLuW9sZtYZmhwPUGy9aBWpoGItcrrquql9YxfbGeAcn4K5kyd6xY4Te1csiBWdCCNyF6by/n9KsG0CXag6cch7HmDJyMIVyWde5zg8fqxuY/a6rwALAWst6XUspy7L2sN5tuBzBN5IxdkEEpMghhTzENKY8pPEQ48DcMx8w3+154MULYJuLODg5AhC5ufVSGV1c0fYlZepiVouvhDkGyiHGNAzWmkirrQkRBkYOTATM1HW2/snQfro9fPXtu/m8lLXmIccczVWagHsK3Fv+IQQAn6asXWSWMYSQUhrG3vU1CghoSBwyd1ddYqhaHi4rmJg6An0+f57bfLk8Xq5nA6eI4uLaWxHqbBw555Rz3LQDkTfEqCMYrXO5XuZam6pN03g8He+mu1eLDNayXq/z0+Pl/HBFoDGmcUw3Q747jLXVxypaGjqQ0RSnKechpuiwPJ6Pdzc3p+PIRxP8/ufPy/oJuU6ZpmM4HPJ9uSmlrXVervXh/Hl+qqAxppjDdHs43dwdr/PTLz//cH66PuQwDmGa4kgJUxT10mSd11ZsOKbxGDmpWbteZL5IXSsoD2E6jndfEGPIIIiRIxJEYCMIaOjqLqKllFJKq1KrtCqttFo7CMOh65BKY+ldGuJAKQZAVKemIirgDEBDjkNuF5pnE3AncHJHc3JngECUmXNKIbKlDZNeSllaLbWqe++GlbKWtqgLJ8wcU0zT6dWkHUCaStVWtdVeDqiIUgghcIo5cOgyecwo0txNpIl2/dQu6OW4xdSNtBOAEQnRFNwdyCESoaGgOrx0FQxcVYkhhZBiTikBvoj6PRtwGBECCEBxbQXKWgyROMc0YnypTZAwBMo5jdO4rrlVQcDOx0+Jp2k4HafjYcyROzrh6fHyyy+fP30+ny9zkzaMQ4iBmZe1OiLHZI61SteyL2srpbXWTDWlcDhO797evLk/mtqyluuCSG5d+L7JPEPqRsWOiTlP6TBNb+5Ov/nq3Z/9yW/+7E9+989+9/t3b95M4yHlKcUhcHkNCXz/9t1hnI7TMXKspaoIEwxTurmZpkOKkaZDPJyGGBHARGprxdyQ6DDe3t29++qr371991uzUJujGmgrl8fzx5+nFL/PaChO1tyR+TCd3t1/9ae/+6uv3v4mx5sYJoyJiBFBai3LXGup0uKQx8Mx5zGG5AHHHAjFbS0/z09Pnz5//O7x4adWF9UaD0dKz6GdxmFgaVUkBcyJY4zguC4zOHTTHITV3YYh55yYCZFFmlRxN4cNLsYMhJRjCsxTztMwxhBTCMxxSPl4OB4Px+PxdDreHA+nnHKMMYaISCLa21GtbtrmxCGGzqvnfqATgpsJo5t1hSURU6lfRHYLpkGlSwgTkTMiEBAads83cARDBNoM4ahjyH3jg5Iimbuauym4oRuAEXov3IGs68xtAHjbxJL7nMLAN6VR7NhoBAIgMO4azZuuPCElhEwWoAsyP2cvz9vcVykVRMAErPMy1Dww+YbpQ7LILKYaUPOJDkCVFUc8ms1r05/OpS2467l3c3Z4nmkAMIdxnBxMdAXw2iqyG3qawnCTQhcriNxZKir6+Pl6vRTgYB6bQJWWuvwSSpX5XD4/zY9Lu5rBe/iW9tA+L8vnz/O8LNd5XtfSWkMkipEDI5JullGEFIhTyOMwHYbxkIcpxIQh4gZf6tkOPKMecZdJdu/vX3eCcRVd1/VK6hIDR8BuzODobmateiUpucVAKcRhGBBgWeZlWaRJZUpuGDjErs/5RTT/4n9u39z8Nn67XksrLcTIgVoXSq7VRIGQiUMMiBs6ABCpT2djCDG6WqutLGtZDgDIoZtnh2ka8xTcxBQjBeaAERpWjx4PMQQOiTkSBTQIBja4E/fQnmKIzITPGE9gBq5FjvNUaxO1YczTYco+wt6rM7eHh4/nx4e6LNA0Mh9DfDMd7g/TachXsBXdGYchvru7/e03Xx1iCoClllYKNgluOZohiF6W5aFpwZBqmZmDiTDSEMccRtcHMMwh3d/effPtt+/ev7m5PX3/w3c/fvfj+VKQyu3NkEJwZbDkarLC/FSulzIUrU3zAMjQimsz6GaLEAPk17iBmOhwin3kSyFyBAIkdiAFVEQjMARlNAwQkIdEvcJKOR6O4zClNMQ8hJS5S2QCgaOJBlHp2DkVq2tLAYYEYBiIh5RyTnmInQQ/DmkYc4y8DwC91Xq9pnVde/WtYKWG0w27e+SYUsopT/H+9boycWkmzbSpiqmYiLKRsAU05C0HN7NaS6eti4hoM1fYxFIRdxF33CblQIQQwA0YIBGRAxMbGIYAgQxRTM21q551BFPHKttmLwg7QAUDcuKQAzNhb7bmEFMIivg8hWPmaRyPh+nmdFzmy3Jdc4qHaTjkcMjh5mZKiUMgRBTRZV0v1/npMp+vy7xWRIQuzelQW1VzZEYKYlqb1NrWeV2WpbYG6Mfj+PU3b7/+6vbN3eFyme2TcABE687sfZzBMQw5nE7Hm9Px9vbmzd3N+zd3v/n6/Z/9yW9/8+23799+NY6HDTIAwf0LIvWQTwD5dLpJIa7zqtqYYZzS6eYwjkOM8TCNNzfHccwxkrR1WS7zcl7KjB4JYwzTmE9rMWuFIQYMAqW3eFJIVVgFyIEpZDpO6fZ0uD8cbt0ihjCMQ0opMKm0kkMtpUrrwu7zPM/XBwAPAWMEBL/Mer7Wp+v6dF21NTM7Dp72u0DEEEOPU+OQT4dpnA4hpO5q2Goz98AUI6cUYmRCMNN19VKkOxz2QyUQh5iYOQZOMaQ+D3MAxBjjNE43p9v7u/ubm9vT4YaIzJ36AQh9paqZpRhPx2NMQ0gZiBxxmwOZuyq5PbPUVcVEXo+pA2GE7iOOmzELICH1wr3X7tiNQbqnBLr3gTJgt3l2pG60tw13Dd0Nu8YbvuBFes//ucnVX9/m+N0dCDYVYNx9bLtKMSFi56cGgAC+mSR/KeLmvSGP1roLtBlu+ABwB0Zg7xj94IwOwBNkgiOPOEJT5bk+nGd6cpUNL7RFwV5/i4oIdFO0EMzJXFor5IiRAseYYoohp3g8DofjmFIA9w8/P3z8cEEbYgzIYCBipUoDk6XN5/nz4+WxyOJO78aXDH5el4enp2Vd11K6s3sIgUKXne5d1S7mFYhjzGMcppjHkAbu1OWdq9UHhO4dbEjk1mchW7TflEOgY6i6pDx0FRhkAkcHFTG12mRdWwo85g6N77RkdFPq1lnuYbP9/mLQ80Vov393e/w2SVMTQyJA2Oie0kyNaNMA7hwqgA1AR0zdi1REaimtO8Jt+rO0mWoxwdbS3BTOAcBsdDfkTRpsGx73fUvEHPqP3vND3zgNTmlK0+2kpmLaj21eCa57IFH95afvnj5/Rm1TjMc8vD+dfnt/dzoeYoqopSSOQ759c/rzP/n9X/3lnwT1epkfHh8u1wuWKvO1IbhbWZ+W5amoVK0ffvmQ0jLPvsyNgIY4DmnEHI7D8bfffPvXf/2v3n/9PuWk4v93/H/W9eJmU0LXgBbREjmgBxeVKmVBYgcLIRAqBoZummXqUr4w75mm9NVXRwcU9bLyupI0IbAYLWUPRDnGbu1CexeQqfMUQp5SzLELg3CHvsfQ9Rz7kLsnjiLaSjsdua6ZgCJxiimn2KHyKYaU4nNPnplSDOjeai9XalNRcHVVN2LqBskpxeUxfXhpAIN1J4ZmKm6td0ZRTNXWjtMFMMIIoGrYyfZqoipd0mEzHQDcdRY3b7guKucOATBFCo7CZs4Yo3ftwtYUBH1r7as0M1dR0y2dQsBAnGIah3Ech9M0jjmKiAEe79+Mx9Nj4WcJ4xjC7el0d3N3d3N3fnx4Yjwccgo85nAzpfv7CdCaqorNy3K5zstaapOmaoAxBA4BAEVkWdaqqg4cEyCV2uZ5mS/zcplVhAjv705/8ruvv35/ezqmnxmeLmdCBzN0ZTcmzBxupunt7e1vvvn622+++vqr9+/fvX13f/fu7f1X79+ejjcpDIhB1c0QoEsyv1yiqTaYV6yk61JVJEY0ZKcmlqcxTYfblN/d3t3e3Eym5Xp9+OXD99efv1tW5fOSx7PTw8On86ePD2Mcx5jmh4+PH378+OHhei7rCqoUUwhT5AOToovO6/XpvBrS8ebmME1DSmDW1tqaNDVZa7Xrzz9/+v77nx1pnA6HwzDkeH0sl5nnOix6NHQgMnzldI7Q4RNAOE2H29v7+/s3h+Pp9vbN3d1b2MCGPa6Zm5Qyr+u1a63UUkWk8y1zzOM45iEFZgRjJget6uSQWmoiiDTkcUxTirl3fTqWq+fc5gbux2k6HY85j5xSES094Ktaa6ram91MZK4u5iqvt3mOdkDf3GE2iCkCbKKyiA5oCNahhX0ahUB75dyh1dg9A9w7Qd7dX5kVw4a070EFadtN4Ng9kPzlBHHcBW4J96O3/6JtKAOBKBARETzPpXokA19VBE0QfNNndkDUroYJSD20gSuAIEKiwDzGTBOsa2ugcWCO2NbO2iXETsMndWhiSynUR+kAZu6g4No9LjpIDwkC8eEwvf/q7vb2kHO4uzvdnB4uT4oeh4whiOi5LVcnXepyuT4s61XBCb8AlpdSLteLqDpASJm5C/QwAAJx5EAcO/cXKYYYKSRHNkcw71OePuzZzykjJGQwQ0foaVtPXQgpBEw5DjnkRDFyijGEiBhNoSBVhFZra7KskCKdjkOIOcU4DHnIudYirdS6llKZaMjpS1jjr2ftMR2hH6C4uX71JdrdO3qkZkTcgEkbxIH68lFVaYNqt99A2gxjyLdUp88aaLtv3wnxaJti4bYKN25A9+DZFteOAekEKGSiwJvccc89X4orQISRXAK9nQaLeDMc3xyOd0POzO5GqqwWETvxMAINgYZh0CXrWsDcamu8KhiAh5CckiNdL22Gc2soCuiUQh7TON0df/v1b/7qL//qn//z/+T+zVsH+PmnT8fx7pHnanVIpzd37+/f3Y6H6fHhfH74LA1BgzVqhRjYI0t1sJBT9kgErO0Lyuubu9PtN7/riM1S2lqqNkGAm+PhMI4dAEHPxIi+agh3RgYTU4eS4TadZsB9ura/Va5bh9xFCTBgD88xdrImc8dhMnMnbqfIhOgqrbUiVU27CqaDExMHDszM/PF7+/C3L8gUFZOmZW11bX0AGUMwh666UAp2xZkQqTPfOj53q8uBe2Wh1pV0AQCpFw690PDO5KbgQNA7+JtqHRGmGImwd366v7KbIjgTdynZGFPK+XCYbk6n+5vTzXFyd0BK04nzcP2wtp3w2vOAIaXTNE3DkFPAKeGUbg/DzSEx2bzMoEbAa2nSXVCpizn1Rpy32tSt1FJEHTCoEYXaWil1WdbLdUaHcUiRwzTkIQVGdFGpzdUiU5qmeAzHYbqZDm/v7r5++/b3v/3222++ef/+3f3d3fF4OB6mw3SIIUJHRILtNOkv9nxpMK9muhJ6bQVco4Ut5kirpYicS4Xzeb29nQhV2vVyLlodArS1PX16aAv88uOnX374JXhIyHU+l+tDq4+61naWVoBzwJM4o4xP508fmtTzWoxYvNY65ZjJUZtq98MDUoDOZAREBxLlVsltDHQbQuWI2n3iX51XRJiHYM6AOB3yOKVhSuOUHCZkZ+IQNpEbdxEp1yvRWVRnMeyqxuAYmFOk7jwTGQKHcRyOhwMTmehhmA7HaTqO4zSkHJm5MzbcXVVrk7XWJmIOfSENQyYOoupm3cbJa/PWrDVyJydwMxW3L9QBY/CBTDoEBtx7KO/xdhvlb3U2odHLGB52SZyOxOvFFoBvWetegvcVgADQKfL7x8AByNAcfZ8CYxe9AujYN3oBA/ZjFXA3CkGkX1Hf3KGKGpp2sSjbBFm6bGynUm29V0LikCJEDHnk1tK6NqK4nE1XeoJluTQ3BCQKIeWUcqBgoquZ1VJbKyoKsEEPvKu6r44CAVhu1NVDwMMhwvtDCnB9aqb09h3f3KiBqIuiK+I4RPHBABDT6xxFzcQMuo9ETBwi7InNJtoREiIjsQN33l1XwoQNu4O4o4S6OfU21kAgdEdCoq4suh0qnacXQk5hGFKMiTCo9mmJ1OpNlZpXiU3NASj0Eh05UFlBpElbC9ewhtb+aQ15c3OTfhJsK4C8m7YBPos8mm3Y/eexgvXELjLHxL2s6uAJ7vldB9+9ms24ec92fVM7tef2hTmY9g/ZC/Zgg2XYZiPtG05K3RzBCQZLz506Rvr9/f3MYbJgFQ/5MKWcHaA1cZC1tFJCQqv1+vD06Ycf355ujnkYYy5pUCI3qkUFPMXj3S0pUBO/XudSWtd0oo0QMNwe7/76X/6Lf/7P//k/+4s/mw7HUuTt2/d3t+8fH5Z5me/u3v/+d3/21TfvxsP4//h3f/Mfzj+VVREDGGqjCqyNm4gpjcMQOAVOXQHp+fr63f03f3rXzKpqlSYqAM6MY85j3kSpI4feiOnMmc7oxE1RwqXXpzuPZvdE7WDbjnHdBmz9IZFDDCGGFGNiTjsK5lngFrdY2iPk7mJhW+exg2ocHC5DA9jXmYOpt6LrUspShxRiSiFFcy+lmmkppbvsAKQYQy/Rt4oLsav0dJujbljATCGE3cEUwJ0BBiIyQOPaVFSbqwIg4TDkEKjT+DYLGvCeoaaUck4hhBjj6Xi8v7/76s39/e0tMyMRhGwcfnz8aam7loVZq4XAx5TGIY05JPLIcHvKhzGen87nx6UMkkIWUSZMiVNiYgA0BxORdVnFtFSpog41mAeOTVRFamnLXBggMmtTqbJeiyx+ebwuTwuIH9JwO53e3N598+79N+/ef/3+/dfv3n/91fu3b98ejsdhGJhDL8jUN8nSzgVw77JkLyurFpmvZUEjVEAlQjUwC6Za1/WM9eefPpnXmGiawpB4SJgTDgOFIUHV6+fPTz9/+uW7X375h5/Xx0WXGshT9HHggA6fL3ZeMTNds+k8o3wInt/eeYyUhit4WdeAEYFxg3ZQzEMep7dv3uZhQAopjujs4qsEGkiFxQa9XNay2Q9v25zpeDM4OCKNhxiima+1oaHE3HWRgLfoFFTNgJvgUpyrR0cOTMiBYoxAIFpbwDSOx7f3N99883UIoSxrjvnmdHN7cxqmzJEdTUS7K1mVdl7mp+t1qVXcaHNqEDNb13Wel7U1aYKiqOqtEbh1z2qV7v7wcgQjRHJC2Egee4TeEsItAAACbpKNXfgECMC38h2ouwVvjc3uDdpVZwF7tdjP0q1Dvv94wk7V60ZSvUK33vanTaVrTwS2bKHXELQ12l8FdwcQM0NzA3Qg3ax4HcBUvTc6ugt8F75KzIEcxExb9WOWaIeRb3/mpwe+1ibuHjLnMY7HxAwic21tXlZpBgrcTzlAU1AXV9fVrHrOcRhCTpZiGwaYfpP1fUbn27twPJgjmw/VcCkDc5rW2tTNmBZ63h5EzByJe8udibhH5s6X5u7yAh3cIBvcjJjZ9qeFfbYoIj3qbXuQ2AkIFZWAnAxd0dS7TDUzD2OexiHn5E6tmaittRi6uLKBqLWmTSybMRMFjp7cfV2LGqxrdbdS/mkNedgIEb6j+nbyAdE2PHL3ra0C+wC85wE7EIOIAdx3h1ck2u4N9u/e4rlvcPvn7Ka7u4Nv4ds6sQ6RAEDRn/WYu7Z/L0nc3Kk7Tn4JOgUk5pxzA1XTyzpfVjeEBlBUMMSYpzgeMQ6KCfOYTjeHMPh4wpw5DxBYCU61fSvqyKW2D798uFzOzNCaPDxel2vruP3pMB1Px+PN8Xg6tea397fH21MahnldAQNSdg+1wPVaHx+uy7y6AyfiFkJwpoC09UKI4zgcxuFErwgMp+Ph9789NJNmUlXEFNGJMDLG0BvtHLrewfbcXjU/ALvNVsegqGsf0jr4c7r9wizpJ6A5uPXamzgRRXwG2Gw9ued+S3fTMEIAog7wNbPdX9cBv8gfVayJ1iK1CiMxW9hYdAjqIorotfTX48T4LLq5NwadunaZe19THHgb8aCDGwOGvvDIFTu7B5AxRM7dsBX6xM4AwNyp24fkNAyZA8cYp8NwOh1ubm9u725jCEDcAJsBvnocpSwffvnxen50bYwWA0bmFChnJsSytvPTUlcbs4bA4zQcWp3WJUUqxR2sN543tRA1aNLt4cGhk2TBQM3b0s4Plw8/ftJlGlOwCjeHm7vj25ynd/f379+8+/rdu6/evru/vb27vb05nQ7TIaTIHDrlpvsA7azZrnbya/0zkbW12V0JISZiDoTYU0FVA3fRVuuCZMuVhxymHA9TAhlA0IqX9fHydPn0w8dPP3y4fjq36zod0ulmjHDkENpcytMl59gc6pBpqv7pEo14zBgq0GLGpmAKpsYh5Tze3t7dv0F3C6ZkQFbaIuVa1/O8XmasMqw4PzV/vPhhgCnvRzCOY3J3QGSC1so8e22rmokqc5AQmJgJCd1c1AqgcqA8xBgYnIhCoJhDisjskGLIITKgiYp7ldp5ViJ2uV4JQjcrqk3NvUr75eOHH3756XG+VGvaBMsSFnaEeW1zrbWpqbJtWG0EV0MkUDD9UusbgBACghEQ7io12yEMsKPQO21qC70vm7wr1vU/bnL4QAgOSOjbMBwRYZdE2ZKG3mlDB7I+AO4ULdgEcvdu+yY2COCA/qxR5hvlfW+i7RdTQFcwY/Ng2DUJu/9I5/P38jfGmFLsmluAamAqVAY/8e1tnk/x4dPxaV4XxZYmzBPliQF9WfDyqAZQF3QlwhAoRIZI1p3OMZC7r2t9erqMRxiPcHebD8fIxIG5B3tAUGMTD0bDmCFENTANuuLr0B5C6mKKnXDbkQ8bOhJxL0fBfYMkwOazDB2/4Y57qxue0+seHQFxbz+jujeAZfHMcJoiIPaxqYibu6g0kSpi6jGRAYm5bFQh5MDgZhY5BEAqtaxlnZf1nwzt+8PbCJQvuKX+T1f52FNKhO4f2Q98cDPb6/I+GXfr2ojQ6/juvb1V5ntDvv+fPWerGyTSbG/Yb3b3pntTsVsz+8Y8Me/tg9ebxcEfSqlNKuMKXtZz62oMDk6cpvFwc5revD++/erw9n58e5fe3KfbG1LP6tPpbjrdYYrAbLAxrsq6fv/9d08PnzjA4+Pjv/sf/+bzhydwaCLXZbmWRUAph3Hk6XYcT5kzCdhS6/myGjya2U8/ffr8cJ6vs5rFIaUhMdfAcZhSzqE2Z4bDdLq/ffdaojXneHt33CBoKmICmyS1IXhkCrTlV/22Cbdl1JUlA1FEdAR3U1N1Md8JGfBcxTkAdC4ZbF7RXY9cdSc94R7a+5mg4Oqm5rqrCLuDmvbUqx8cTb7Y82rWo3trwt1VjTchPCRwdRNttW98D07Avam1HW9ExAEQwdg7xqePILwbtG7NQ2YCRmdEZkfGMISYOWUKiGioYhW9VyNEFGOIqcOnOKWQhzROwzANwzSEEAHJREX0dc/xOl/+4R/+psp5LWfXymjEmDIDYG22Lno515VVDvbm7d1xGqq26zIPiWYGABfTIiLqamCKhm5oTs5MnFPNOYfU1trW9vDh6XsO9Xpzf3uaxunNt++/ev/1V199/fbNm/u7+9vTzXE6DCl21At1uU0XRDYHNQO3QNs+dDf1P4gltphe3R2ZGIcUKCeKAQLVLlYbWCMTIoUYIjM6tepXa3VRQjw/Pn7++OHpw+Pl4+PydJFanQ5piuIYgUuT61LYKSSAGnzldgaGRhczWJtAqW1e27qWWso4jm/evvnmm28iIajO5ydQYOfr03z+fC6XImtLRMENf/5oHz/4V0e4P+yHFcUQ1LzriqgqEjlupwd3RSWmTX+3Q/2lIVHKA6bu0xIjpyHkgWPGQO7mulyXn+RnBWutAlDg2L2yVbxPGt3BAMT1fLk8PD09XB6K1Xmd67r2cqWr1tgzEx27lquKQQ+5Bl88DvNgntCctnLHHARAcRtiOuynMgH1yAL9r23vb4j1bfiCAED9d+Mz82r7b0dqbxlBn31At4Dbfkff80hABIzdNwfcu1H7zsTr/p3SNar3i5BSGFRb00ZmCShjyN2tgSmkEGI0N3VNIeaUxpRzjv03qJNM9Hbwd4d2Nx0/vns8L0+KSz55OkAcTc3nS/78IXKC65NqJfQYKAX2yBoDcmA1EzdDm9f1Ood5SYebgDHmEVNEdBE3RFSArq+BhCkF4OgWLh9fCHfEgUPagIzIXVydmHvR1VVjNxktgu5r0Utz1a1w773G3Q32WYCbQwyw+fD0CKguhtrY5OYQRUZAIEJVLaXOyzovZSmNEEZkoKCO5hsFENyNmViJAlIo9TrPl+U/EtqJiCM/x9WXWAneMxHYC/Q+hKGX6Xivp59j8Evgfv6TvUTs5x++l4Fb+reH9i3e97GAAXTM5VZlIu06vOauvVy09OrwQuI3v/lNXcv5soxF3NxERUUNHOl4c/vm/fuvvv326998e3t3PN2Mx8M4jQMAOfAw3QzTDYaAO4PRVM9PD0Q/mvsQU4zJzFtrPa7//OGXm+/+IR3Tw+Uzh/jx4WcIMt3kOzkdbw4hxdLq5XKpreUxA4E5hBxCDH2q0afgptqkLeuc4+Vgb78gJ25mMOhIaAzgDtSHJtuW3sxU6Hm79o5bx19u83U3JEV79ifYB3B7EwW2QLD3AHvi5Dv99TUQAkDdpUs82ZaM7q0YAOhTchN7eRzec7UtrduaC+bO3qEXHNxhAxsDNQDgDcHrvafAfZbQb7PPBgBIxVsTFUFwIxqGTRqCQkhOFrqONoK7OZKjOwMYE0JkItqoLGruQCStXyKiXW9jg+C/HsK1Wj4//OKwil1VK6L1Lb2W2lY5n5frpTCju+cx98IrxTAOaSxNDAxQTKuYNFNVdGdEdMshDjlh03peJIYUwtfv3v32m6+++fr91+/fvrm7u7+7f/fu/Zs3b4/H4zSOMaYYAm+ziFebtVP+O52f+vntnd38+i4AQNqllceuYGttkZJbijnFnAIhqHaELIXAhCBmWts8m6n2d3O+Xi5Pj8t5XmutoEYWyQf0lQmZr4SP6ALaQA7WRilpuQYUJjSHWnVe6vmyLGsRaW/fvvn6q7fTEIfE16f584ePTw9P8/l6ebxcHy9SFdWHGCPi58+fHh8f7srvntXwa60/fP+Lbq5JG+CjtwYNnIk40N6QB0TvQiCqYu5dzo9DiBQjxYQhI4O5SDXQjmJW1x47Vb0LuJgZdt2jwMispg2EJzzwwAPUwibaMdUd192XMri7RrdN/w0R0zG+gpaDARkwEfAGizB3dCDq5VMP0g74LOyKe2jfkOS4Ydz3XnsHMe1l2naybp1YRHTfhRq2VvxOAnXoSkgA1HuyToD9x+14KXDbMhsQh9etUkI8Dkdw82wRKBOPFHMIkblLtyFj1Vql9i0YCeKmKQvuho4h4niAW8s4Hg9CxgOPyoNgbKIaD4CZaeD16taYIAWKgSGwxMghsJg1MSAPiY73YzyOmJIyC0HXrHM1gGCGxV3IPfbITf6F4Bn0M6mL0mx6cyEgkYN7V3J377165kghdDbB9sZt8HB7LkxpE1vnEAKH0DMAVzVVcgUEQWiitcq61rJWBFhXWdeyrHVZ61qEiAaFpt7ERU1UgxMzMbhqby0EVVxXre0L26EvTV0Dc4wbn3hvmPu+Y8x6OwCICDsqfltOXQKkZyrWkaHmG+IEtgi9c/2eI/SOvtxm7ntQ8r3tbnsbY0sUAbuQCRj6VjyCidcitbSURtjNxkIIf/4v/uW6Lr98+OwON8djZO6QazU43dy9++qbr3/7u69+89s8Rg69Y+TQZVpoIE57O8JNpbbydD7/+OOPv/z8w/39zbKsl3lelrXUWpt8/8P3Rnquj6ebAwX+8OGz+HL7ZhwP6d2bt8fbw+V8vi4XDH7/7sYBOu/A0WsRUY2JiN3ARevj08dW1je/+22ADTYgqqXWbqXR383+LrpyH0WY4ZaNbx1V2LPGjp1j7OM5NOq7u2NLXwJ7P0QcDK1P7baBfIeG9A7eNqbvY7pe9HcpOXtpvuyNHsBuVKP2xSLrp0anMoReSiETEaIzkDObiWnbaGmMtunCdrVt3ITreirAgYnNTZuWpdRaAUAjD4yckjMxBaDozM7uplWFABjIbCMXxUBbRuVeqwKKqA7X+Xy5TtNxGEoQJ44eOiz2VUTUNq9PgNV8Vi3gaoa1eV1lPq/np2W5Fg7k7hSotOoATDwOw1FsbVqaNfGOrjJVB0bjCHDI6fbmlIFsLgx0czr9sz/9k7/6yz///e9+8+23X7+9e3N7uh2GMaVMRIBbnmQ7r8j7cAypYwuhKxRuqa8ZIBDjr11EZ61PIuaG60y8cQSG42FCxC7s1ZVnwbwhikgpZVkXaU1Eu1aSmLQIhgGA6hTWREsiSHzJ4SnTmn2J7UTLZDRWSRgiE5jXYtdrPZ/nWhsi5PTuq/d3X3/15s39abmePz9+/vf//m+/+7t/OD+d61IQO38hENGyLktZfyvrs8jhMpd/9z/+v7QDc7ZDdUNmQwd1doUX2sPbloI6IHLofwUmRgN2Ck7Yp6dgTo4MyBtISJpKEzdD8JhCGtIwDnkc0jCkKWYcAUYT1a4k1mUbdBNR3LZdl27fZFz9MEyvcy1HMOqsp20u607uumfu+9/7IbpH7lfRyPcAvvXaX9qnz/sP9hb/jtDrHdityeo9rm8tCdh/nHW/pT1jcHRTQAMQw+Zor4btRHwz3cVAmeMQ0hjiQCEhx8AhsIKItbnOS51Vm7mYd5CCG3oRLWriVAHpYNPEAx+dBw/NuCiuIC0nuB3SdHc0YYIYKAYKTE5kIRAHVgORjecaU0gpcMaGYNrYmu9zKgNsQBp2VSwC/wLU2NsWnYMWQgjMsWMLrFtXqCBgjBxjohCB2DeQkwN0Bu9Lw/mlXt/RyGrd2kJNKiAwdXU6ak2XpV6vi4qsxZalrmtb1rasjZnXoiVbbVqbVpFoFGLYUOzExBEwqGGnwzxfv27IE21Qqw6Xem7bPl+wtd/BwHdI1R7Rn/96jszPtffW030NzOp7EV79ih0C/7pj30Ei1O2PvJmomFZtVcrS1rmuc1mXAu/5/e+/2R8N3b17W9a1GgDg/d1tCsFUWtMmOk6n4TBxDOa2llXm0uraWjFDN1YjU9wMlghU5Xp+/PnH7/+H/+H/9uGXn+7uTiry/fc/fH54OF/OiMiRnLXo5XAaQwq1KbIcbzNBOB0zRTUshpWipnHbL2oqKsgaCGLmnAIhBwpTzDnGX3ETYa+GcYvVm5i6d+ED35vl22bfYQ/P0xTftynsj6D/uH2ksp0MvdQxQ1ewnpgTIAADwj5t7uk1gIPTRpnpkNB934P55lDTKRIvi+rm9pbQTVpdh5xTTinluDkMIAC4SBMpxJASh0jMmxAmISOyqsvm1sYhBGbujRPkyKW4WwoYxoFiJGdzBggG5GQAihACcuDsDqrdQKuTNrFTgDv4oIid5zWfL8hpGKaYcxgyhfh606vrKoJYAbqsp7VmIl6WtlyLmYUQzG1ZW/10jtc558xMOQ8nJ1hW9aqmTJACcIjTONwcD7c3p7f3t2/u79pb+c2bd+MwvLl/87vf/fZPfv/7d+/e3N/dHsbjmEfmgLhZg2wmndgXRd9H2DuH3Uj0GaAKXbALkWN6dR/w22++Ok2DO5hBt/ckoBhiShkcaqvuEDfuQAIAVU0hpEhVatPqHhzUzNTVTRBszOk45cNdzDHefXOEQTl6GsJ04mGCnDUGYARQcPDRzZFaC0R0fze+uT/e3x1PN9N4SDxQg/pYzp8un+fLggCdOYuEoiIqVV9UDkXk8fNZXc03CdbeQO2d1BACBgYC8j458p78MG8Qledm/cYG3waNREQUMUSOKfaCzMRMFMAJPcaQc0jjkIechhxT4q5Q1o+6XuBIF2ZQ7dMCd4DOb9d+LCYe8VVs35VnUGEnv2/N89eIt62Lvp/Irxcm7ufk/lHcQW/bz4c9C8dnmAwCEO7n8jaqx70O2wZir35phyujO6uDOFbFavA6fw8cvn73fkhpykMOMXPgHiSZMGCR5VovNmuFUqWJmbk6mBMYemVt6LrnDUTIzMDBMCgyGANJCJAz+YHBmYB7T5K6cy0REbtjJ2ZtcD3E7WNGBtzlztCDQ1eycETeCX0dUrBdRMwhdQAdc0DubvZ9uI4b9Y8DEO2oM8Tnp7BXptAtuF7147c3Gnp8UzPlvalkACJeq9aqTCziKtaattaZ7VCqrKWuEdeBSwkxYke02SYAM4zjNE2leLnK+eWJ/CqQbE4afU73qsB7XmEdBNcdZ3cY3BaOzZ7t8DbMxR4OYA/sW0OLiDpJccPEbUfRXmo8NzR2UjsgmYK61dLWuazXsl7L+eH69Pk8n5dlLpMf/+r3z7EEh2kipvF4AIdxHEMgbeQkiqhgS5kfHj9XFdW6rpd5vizL3JpJ81K0VBMTAyMmU316ePjl5x//9t//zefPnw7HAQHPT9enp+vlciEkCgjUql2max7GHFLikMZpSGmITAqL4oqxURTU1psbpZXaKgKGEGIK05GmYZyGcUqHFEYOLwUWATLSNv9GQKDAgTl4x8ZZ93B+jtBbv25jWvSHs61YQ+yiSNsp8wyW7OtNzVzVTKk3Nw0MzYm5M+YQCQiwu6SCbUM+N+hQXTR3g65Mt1m3vO7UIeKbr95Oh0wEZVlSjDmnlGJMkWOXXkJptdSVyENADkidEdETGEcz7OUnEsUYObCbN5FhWtZSVITI8xBjjOCkRi7oBoBG6MSQYspxdEBRbdKa1D76cfD+ZiB6c7iuhZ+u5uFwsPEAA4WI4fVgytyqNsIG0As0W0pbFqlLlSLMNE15Wetc6nKZDeHm5nA8TuOQA6emVpsqmwWkxENMb25v3t6/eff2zbs392/u7mIIYHB7e/vVu3dv3767v7/v6H1CBnBV2bBV+37boY+bFkRn42yGNt5tEQDRcRvyDc/YCkT8iz//c1clInAU20SEVN0VdS8u+zHGzFvf39W8NVurrgbNUMzVQMGNwBLTEOLxEGOInG9v3iWgxtFyDimHyMyE3eknCQxTODbuGKO3b6ebm+F4zMdjPtyM0/2QbhKO1NiuWlTtOUntUae9HvSY17o5amyDUWTeOEucc0opdY8UVTVXZg4ppBxTjpuzEW5pbQ/pHVjDkVLmPORhzCF0EN7GJyOEGDBGDilw7Lkp7xhj7HOmbZrYe4SqYtu8wHwbVroDW0T5dWjXvUnZB3W7dPIeyjsgskdZ+EL9YgvmsH95/4l7nu8bVa+rRfShRcfEO6HTrgb1HJ4ciMhf5D87xH0DzbE5iHoVKAJVsQV8DiCB47df/X4ahsM4xi7T7y5mzmjB1/WpXmTW67nBCt5ss58BAkfUwLp70mx30sG9TgTMGIiMAiESWJ8UwIuez6acQ7v6Be1aOr4p8AHvuQ8hRMIuzbp9vTvYFx0QIA5dgRt585juFYA5dskc5oAcHNC0W1O8kASfO857vb7lGM917CbL4RuaeWfvgpiJuiiYbk960/FXB7dS2hppCb6uXGpMlYgcHdSMmYZxOB6ParboDK9sXb8kv60ET+G5cgbYmzHg4E598e1xeofPbe6DCI7P5fbzifjyp+cmPJHiztdy6jN138r352/3nolbl10gQHQDNQ2ggWtOMlod7HCgmzKVWtv9K9tKM/vhl4tKe7oaOKgXIjTVbg9OBGHxdGkxnVWl1qWWtdQiYiremjVRdXM3IjKzZW6iYTy8ccwpR0RMWY43tZSC1P0nUh44DTGlGGIMPWqF1KkBOePtTRxyaa31tKVKFRFCYg7DmIchDynllFPIgeNrSPbTY/u7v1k6xL1XaMxMxHtzw+C5rwH7AB1fbfV9SfVWWx+T7Vg333P/fThkCrb5QtsW9o3YiLgnab1L/ny47IS6bZrXBzC9SlH3+fPLSwFADjmNcLq38VB7a2pTc2fuwlechHJDdOaNX7vthP634d7174MrcndUwzCGQcwUwVNkZoZuSiEb0gjJewHHnGDDMwlrl5jt978lrMzsHKqF82oN6iJzLBpifE0VVYH1SogBwFs5QHfTJMGkiYwGBIU0Sm5trNXcx2mYxmEYEgAAHceh1i5wK5pCPE2H0/E4TUfio1gCC4RUhB9nETg/zUIdUPkyQ94eWH+AuKe9e2gnZEbEHkaed2DnE3x+fHwdC8bxgLCB/7dum5p2uJ09Hz6bNEVP4xzcUdWaerNtzNqnrobgjBiJchqYIoTjcGiOQmQcKPSye6sC3CN6Bu/m7k4UpsfLSj9/ejivHz8/uYfbu7d/8qd/dnPz9nKeX/d++qK+vb19dRfj/+wv/rLfaY9QzLzNQJliDBtDEny3F+oAqd3botfaewGzdRUJiTFE3lLPrpi1Y00Ioet1kxMpoyNKX7/95MZ9vYK7kbu5hU3uzR3AaTvqCPl1aH/SyWDXkdnntXtA7z8XnhtmDntg/yK6P2/+vhj85cO9yW4bABD6+K2P8PfcHvcpaf8sedfCsK1i79QZ6LBlUANxEIRGUF8pCInaDx/OOa05zkzUW4vi1iX0ljZf1+tlafMKzVgsOoSuj9PvzXboD27DAoYNSGvuGcCti6g49kakbWI+vncmX1CDXfMRABB69RHBDTasQne3pe39dgQAU3hN45tGfnMHu2BP974EczJj960wfQ7kgIakhI5k24HiZu6IxuxMSiS0DYcIkZTVQDM0CxJYIysRBSqmZVmWR7wsS1T1Ulppi/vKJIjoWssazh7cHksdhyHEyAjg1iVDfF3XZVnXV76IAID/zX/z38Afrz9ef7z+eP3x+uP1x+t/Khf9f/6SP15/vP54/fH64/XH64/X//9cfwztf7z+eP3x+uP1x+uP1/+kri9m7d3o/j/y1X3s1kHY1GXJATrM5tdT9uehUZ9+vKAMtjEI7mOuV+P4/9+vzh18/l/qJjx9lvwlUeT/+2t7YV/CBWCH/PVh16tP+v4b4fmz2/ds0JjXfJf9dT2PxvdvtVc68qqq+gWLDHf+V9dJeAYn/lOv3189FN+5DM83gq+ekX/x+LZ78WdtDHh5I59/45ff8sVFm/nvdnUw1h9cv/5Yh2huHAx8xZh8/uwz1hc2Pj0+v6YNIfjlctrh41++eHjmYjy/fvzykcGOHd0ktLYPEzNtoKidtvCyxvcben1XOxvBXy395xf8+rb+8Xf49XN6WXLP3/fydF4tx51487IB/dfvtKhs6o348mv/kVWELzf0sob25/OPX/iH78EXb8av8V9/sAR+/VVffn0I4ZVcIwJuvPHXhw28fnc7Inj7zBdolF/dxuvf88XX/yN36u7/6CeeP74/tn98Y7o7gL9gOL68qX/0O3aIcV/9m6/ol2C6VwsKvvgwfvFT/mO/Zv+yP7y3VycgPB9zHcv5jMN3F23PvxH/4GftSxp29ADis9adv2z917/weeUSviCq+8/y56XfF/zGCgB/dZf4B09wh429YMH6Z0N47RDzvLH+kQf9fOpuz/eL5frq61/9CP8yGD7/86ujew8rrz+FL7HVNvDd9ho2KAC8CqHeY/LzD/witP+bf/Nv/u2//bf9z30Z0S4Cz0w5JWYSEWI6Hg/ffPPNX/zFXxymcb6cf/nw4bvvf/j0+eF8mc2dYyBmYu639ebt/bt3b8C8ruXh08P1cr25uTkcDufz5XK59ACmJg4eiBFQNxL3dscbj9f8eccA7FARBw4cA//Lv/5X//pf/2f9lXMa/vQ//99wygCORBQicSAOPRi82gQAG46ma68A7I666OYqUta6ruU6S60dGxnHERHqfK3L3NZFahURkapSOs4D3EwV2gztYojOQ5puh+N9Pp7SOLl3HLmDY8xDzAMPmVPCEJEZzU3k7//b/229PvUX92//7f/1v/vv/08dugsIgcPp9Oarb3//n/4v/rN/+df/4u3t7Zubm3dv747HCb/IHHriQOfL5Yeffn78/LBcrus8r8v8d3//d3/z7//GwY6HaUgxxRDzwCFdl/XpcjmfH66Xx1KKSCMkcz9fzssyqxkAMnOKOQ/j6Xh68+Ye3D59+vD4+DRf51qlrz3VLfD+5V/+5X/1X/3X2+Mg/Nf/4mYaaENmYUBkghfyAwCagYpdL8v56VKui9R1zHiYaMphSASmKro0fby2f/jx8adPF+BukwiBwt1xPI1hZKut/vQwPyxanByA0YYcDsfpcJimw5hzjDGeTsfT6TTP68ePDx8+fv744aGerz4vo+noxqiAUGNegJ+u87LWr//6Xw+nu34jX3/9zV//y//59z/8+Pd/9w/L9Woi0zBM44BIQJxyTikG0oAemRBRzZtIqWVe58fz43ydy1rBPHGcxuF4GNx1mefWmpkDE4eQUhhTOhwOx+NNHsacx2Gach52YwVV06a11rrOy3pdr8u6lCqm2sTXWq7X88NnA5/uT3wYPTA4YjNExCE9nyD/h3/zf/zw+QOAI2FgHqbh5uaUx4zQ3S4cGSlgijHHzMTgeJ3Xx/Pl6Xy9XOdWRKuauG8SkdjdHXuyj7Q9206mwk5FclBpruImaErY3cSfE5leM1hnV1iHc6uruKltXC0HcPwv/9f/5Z//2Z9vN8K3fvhPVVSkEAJz1wehlIacBu7eriCEGghjYI6ZQ3aMBlzUm3ZS24YhVOsgUCCm3B24QuwZ9zMqsYOiRXRd5nWdXbXrqoIbEgJoXS7SVo4pDYfDzZs8HomYELm7vIKrSJ1XKU/l838HvmXt/8V//r/60z/901/lAXsiBYigarXWTx8//N3f/d0P3//Dzz/9tMzzziXqPirEzEhs1g9TcbPIFDkk5kAEm+K5qtlzvmfPvOXnvAE2Nb8urNY9ytxdrWs8cwiBujy0iqref/3tzdtNaOAyP/3v//v/HZGPOeYccooiVlZp1UR9yMN0GCkhRY8gI9nXQ36XEldvi/zyVD88refLvJYaCB2gmhXzAo7EhxC/Ph3+/P3t13eH03GiHCuHxfRa2zqXMq8IyInxwDrxo5THUt0sEN4cpsOQ17WVooESYFhXuV7r+XFZ1oroqlJKY+T/4n/5X6e0CRgXPK/+NC/zPM/rWmqtImKuIXrOfDxOp+Px5nRznI45HgiyVBIBda2ylHJtsqiU1qTV9vh4+fzp7OgxhphjTJG2gE0xxrfv3rx5e5eHhATnp/PlfC1LAcDD4Xhzc3v/9k1Kab6un3759N1/+O6H73/45ccP67Ii4f27+29/95vT/TGPSbQtyzrPyzKXr+9+86df/cU/HtrXdX18fNxPAHeH7vPWY/wwpBBYRUMMiHCdr/MyE/qyzNfr5fHx8fPnh8t1NnAOgUJgDg6OuGlpBA6uVlvtzM/a2rIu8zz3zKZJUxUEAHPRXZ6uvw7d6HDPq7+XVVvkYI6B1+UFHIhI+XjDaUAEJKaQKIQvyq+XYsGhc1a6Fc1W2qq3qq1SB3ECc9JOcuAUwQy5IDGFSIAcoreOfO1sIgMUhIAQDdA4cveLzmMYD5uAjCMAxZS7jzrGbkXA4GDSXsuklFIeHj5vPigAIUSDcLi51CZAFGMax/F4PN7cHL9M+jv3BUWNiOHZXRfJzUpZHSynwOiEziEQBzMT0WVZzudzq81MEVHNrpfLPM/m3uHHLYqIMfE0ju56uVyenh6vl2utYvYS2gHger2+vByEMfOUabOIQO44VdqtpQBY1aWJFFzRFQ3AEuLAOEWfEriBoANAjZDIeNMSBRNwVtQQHBMaoJA1UHFj8+4NjeQOJtqKoBEYwTgkVuEYkdER1KRomU0V3AAVqHtQclvnsqz+qmsSmKdpGtIQQgwhAXFKOaXcWcghxBhiJE7kMRAidvauu5sqOgQOmImJh5xPh8Pt7cFNLue0rmtrgoTdDyB01WkRVGMAcgDvQrxirl3tQlW0NalVRdy6RR45ETp4U3f1pt7UHBA3d4vXpUfTWmVFBHJ0IBZvGkkM3K0rnJMjY+RQQyRgd7zO68P5el3KWqpU1WrS5ToNoCsfdr125s4sh06MdAcA3te2a3Nr5Mro3GXKYddk2EKKiKg22fh44qbu3QDKwQ2avPDakQKlW2AFqoi+MZ2ZOA0UB2YMDIzKqMwYmLuAikF0o86pdDUAdVQg60IKZECMIYYYOIYA0PViQfey0tyM1Agc3am5aXcZxC6fSOqIjslpcj5AOHbN1o7IdnSAioGw1dfbdRjH0+n0hyX+3jWEUoq0pqK1lFqqtO6OvFFnNoqKA+LGdugMKoTQbYGw0526pom9Movq6fge2nszwcE6mcgdiMgAzFzUxIw5WIiBkQh1U1566T2Y2eX8xAQuyS2BRxFb11aLNnE3IfYAxA6IomA9YZNqpci8rNd5nde11haYEGBVXVVXM3dU5EllGbEkH1hBwoq4qjVRLy1UYcJAwYVRiKWiFABHZMQGyOq1aTMw8FBaWcp6nq/LtSC7mZa1MUV/JWYq1qquRZa1LWtdSvf/BQvgyKEpqyfHBqQUjMBQvXN6nFAdDN3BUNWgNSlr7ZpQiBwowFaFm4KCoVEgTowIQKAuTauZJ03qzcmAAdkdTbzVWpbler3OSJCnWOo8SGRzsdZsrbKsbWn6H7OHeWng0EYUZURsTfbaFrpTFnNQ0afHx1rWVtbLdb5el1LrJmVijmaG1jsW1+uspqfj8TBONze3tze3OY9EtCxLzTUPiTmsdS3rWsvaasNndaVtxfl/pK3ubqLwJZMaiKn71RMFjBE3xQD8sjmJu9rKSxPLVEFanc91vpopAHBKYRjQFFS0rVJWuZy1FndnhhAjoWst4BhjBFcjZ04xuRo2SCHEQMiEzDHmkWOmrsnCAYlhk6BBAEBC4D8QrHHwTvHo6kAqrdW1zMsyt3Z6lgbaZGt9b+27q+uyLA8PD4+Pj5v8gT0fUNu1JUxdYJhZxJalEmLg2Nt+jCFx2gIxMVMgYjfsEWbTZd27Rbssxh/0f/fjG3YFtX0Yg741kdCgZwZ7B8fMvOtaIxG5GxIyUYwhD8M0St18sPd7AmCi0FnGvXvXzWwpMiVt8LRemTDnlONwd+Pgm/pNDGFF1F67dS97AFETQFGTL1bVcxMSQ4jjNDFhjjHFuKmLdsbw7qALCOZWar1crvM8A+A0HfMwjtPheDzc3d68eXNrJo+fP5/Pl3mewT0QSi11WcpaylLr1FrVsBRgmudlWRb37hLrqm29rHUtYgDEPESKbBYwBSRGcWjuqxp7CCFyQqbXweR4yM0OYYvBHmOYMgdGUTVp2qp0a1xDdgRDE1hru5YKFIc4woAWYV3W1UsvLPvqCCmGGLsHhm0KLgoOSNyfPxCCAQPHAJEphN5R3DpppiYitdSine2IPWXZ5I51Iz0/X0QUx6kHKAQgxk3UiSJwQCYKFIMH7grpoEBqVNWaWOsWtn3vOBAQB+ptPUYPBAGVtz4oqEEzki63KaYi0oq7uDUw8W4uihEAkRKSISXHoAZNjMh5UxTe6oadWPhy2T53w/16bhi7u4ldL5cfvv/+u7//+x+++/7zp0+ttl1padN07D6hbgqqaBYQA8cUQmRGgK7NBOhdzkd0c5jrGjT9CxD82SxC3BWkc7kQANxMW2viwRjBsMsndPbjF81uLwgBMDF5II8BLTIogrqiEzmxh4icPCRw0kElV7GztCf1K4CkiNE5BkRIrVhrrbRW5LrqA8rna815qWAKfl6rmTPxwOHAIRNHJFVcKzQh0eCBQoxjHEJIgCIuZu6uRdra1mW9zusaAgKAqCJ+mfi2ttSl1NpaqyJNxLsOqBN4dxLvYgjmKJt1LXGMMXgMIdfqtSICihgRE9KQh7vbu+k45pxVrRYppZkDU+IwxDAAAOHsxiKg0v2727pUBFa1HoW750Wo7K4OZiZm1R0BBFG3v7+cHfxKje5lBPssfgsAm3DuPusFQBGZ5+XT588xhFrK54fHeS1NdEs0uzIoGhAiQCvVVFNIh+lwujkep8PxeAohnk6ny+WccybG6zw/Pj18+uWDNCHaZclhi7r7UOY54m3/fZ55/Crl7XVqVynCzb2n32E/on0fGu2xBvsMxrw1KUu5nNfLEyBQiPkQYwiubuouzcpiZZZ1ERPkkKZDiJzH0c0I0M03VqgzIDmGlGIecx6nNB7iMHHKHCIy7yzS7eVAV5bbX9jrB9Jvj/Yg11qZ58vl8lTu7szUO8l8PyPREQjcXEVLLfM8L8tCDqC65fgvU6adHu/eWltLMXcOAdzNQZu01tAhhIhh9xjuCg9m0kS01dpE2iZMsDctn1/zr9YVIVo3s8HdVGofFu2itS8j8Oc/bT9qE4AAZsw5D6NCQ+sC5t34xgyBNuooAnmvJJExBIzuJtUUHYG0mfcqa7/crHV5LAIkBwcRKWa1SdNnU6L98HInpBhDl1pmpu5pq26kogKODEzmoG6rtKriRGkcc87DOI6H43g4TMfp7u723bt7kxaGIR6ehsuMZhFhPj89iMxLLfPaitaqlJIzzdd5Wdau04sM7rbOc12qA2IIwIgxIGFI8XA8osPhdAzj4DHkYTgOo5j9w9MvL3s+UkocAjGhg8dAMWBKmCGKYqleire2yc15N7MxRYCY4jBNACTNtEmx1XfDK+iyquZICF2Nai8GmBgREJzAiDAy50QpcorcPRp2hWrYczkmcmZEcO8KyY6blNbrdYWIIW4Va2fp99+B7J1/DmiI3mUO3NVAbFPrbKKivh1nhBxCSjHQpnmJ3tykqZq79LjuJNqFMM1FzbSnLm5CgND3LTJSRDagCMDqKGbo6O7k4IyIsOUSX47DfbfW2g+3l/Fpa22+zL/8/Mv3//Ddjz/8+PD587qs3jneXbFo82ndkgYCCESRKDBF5tBtx8Cxq8+gmwNZT5J6cuCbY8wuI98H1tCrensB6biKIbrxrka2w1xejlw6hQnJkwdWJiUCwkA8xBxgGqeb03E6TtM4TMwH5rtxOKShKE7V8tzerM3dCTAwgVsry7qu81Lm67qe1ynYdIo8RE+xtVakatMAxhGGvNXMDgiCwW1yIObIMWFAIDR3UWWCrfaBXlwEDiFQChRDplfrqklb16WUUmtttbbWwJG533uP672TIa1VAG3ihCFiP9QJgN25VdpPbg4x5WmYpikPQ6sCUEUB1IlTCEMIA4AzJ6IAgGpWWpuXdZjn7jaDjGlIwzSM49hqLdW0tWVeYmYDdbDaaqtNWjP9pzXkX53OuInpEMLuTMfM3VusdiMNMwUD9/lynZe11NrP5Z4Ggps7wq68b+6m5giH0+Hbb7759pvf3N3ez/M8z7O5tVafLk8//vRjK6X3ct1ddUcW/GNIluc/MVOM8TWG7vnzXZ1gGy1td+abIeFuqLDpOIBDd8Qsa7tc1st5uTwhYsw5xQQUulmuq7maa9P1fD0/QYhH/CofTsebWxEpy2xm4Kaqrg0pUMI45vHmZri9y8c32F0HEGC3b+qbArby/ddQGkSgHT7SIQUAXtt6OT89PH5e377rI7GXOcWGhyIAFdHueaKq1NvSz/2P3qbYjgRXkfPl8unzJwM/3t0s1+tyvZZlldoIu/sAO2CtzV0BsEkrtUirpZTeztnBIv6HOdbztu8DV8CuGsG0dVHA1EXapgK+X/CSefS1iEzEBEyQYhyGbGjNFAHUTERU1O35aIRegQTiyJGZEdiiIkKKKXDoHo2bZK+7qNZ+b4EZyQFrk7laqbXJF6G9535EGGPsg+Rm2sy6WripAXhiUgdRrdoupVSww91tn5vkYaQQOIaQUzoe8mFSldjKAIAhBfMMgOaX89XtUtdSV1mWRjlTSq2JiGcKTIkCuknAqt6aqKspEZoz4JCH49fjkIbT7c14OqbjdDgejtPx8fL03f/5w/O9qErTCsgOe++cNQ/peDggQi1lWdd1WaWJGZiTAeWmUTTmaRyP2qxc1uuTSlmsD896WtmdvPf95uB9/NIlO0LAGCgEDBxS5CHHnIO51WJNtJQqzVRdxQCJOWBXIbVNSAeoj29eH1Yg6vsO7sLXPcg7MaqDAZqDMPQuEDi4uUjrja9u/BpjSsOQI6UYuHfgtUnVWrXWom7du8W7P7cruIErgjH6Bg7YdFQ2hwPiiBSc2ADVd6gQWJ/3u8Fmj/rFNid6UXN+Va+bztfrjz/++N13333/3XefPn1stfZ2FFJg2goekdZac3dCCByHEAIRbzaujgiOZAi7bI1hl5iFDYuKDj3cvzh7mbqrqfavAMTusoFuXYedEJy7F9PLjUTi39y9VWkGhgJcMcTAOeIYmOLp5ub+7u7N/Zs3d3c3h8Nxmo6HwzBNELJSEAFV6Pqm4GatluulXOdlXq/n+enxScs5wxyDcqR5LZ6e5uu1zWsxx6pVlMmsuCAoYGIaOAxGZKCiXBuUghExUOaoaZqysqfDYRiGHDmlONCrMai0HtprLaWV0qogEATuj7znM/3QMHVwkuaBIwcIkQAVyQC8y3u6ASJjbyZ2laTN6JwckSkxZ+aE4DF2dU725rXV63zlyGIyDBkjjsfheHu8Xi61llpbWevjp0eRNiwDEKhZKaWsrd3806H9+QjbEWyvPUC2plo3Y+hS3iHE7kDRI/B+MG8jItrwiNA9OFWt1KZuIfDhcHj79t27d2imtdV5mT89fGxS/34ciNCNEO3LkP66ePLXSGIiDjHQr0J7v7ZE7bmTt80tQQUQkULXR8fntLSJtwomTBg4dMSSlELIu7aruit5IysoM2Jm9JxTnk4iau5mTaq7KIpy4ADIvSLIOeYESAa7NNzWH0FABGIAckQ0+nJggN0RBLHrKhOgq7Z5uV4uT6UWM927Gq+hwKCq67Isy9rbLd2hb1PwtZfsrcuSN5EmrUrDAEioaFVb1abWAjOAu7q5i7Rt/GbSWhNpz8KZv+rA/yF8uj8n6KJpO9oKoUOnNtAkPAf1TQzNnpmZz4jRrY0eAjUCAO0YAemFO4HZ5m/lmxUWYm/LY+RABJE3eTHsw1EEcxeRVpsgKIISOGATa71k+3JEDXtTwcyaiHYVKzcVc3MMLoqlVjUqUpZWFqkUw3Gabu7vT6dTyhkAgYCZAqI1aaXUeS3zsl5XqG2Wdn58vF6vy1LqWgGZnCJHDujOCIgYmRMxAoU8KDkOZsg8HKc8jDnGnPKY8jhO4/EwHKY0DcMwDCnhhy+yRmaKm+cpEnpPm2KgIQciRBcRaGTKBgwOLmAVpFgTB21uxWqtra3gkiLHEES0iYg0bX2FbT03JEImQ0ZGY9SAymQteItaYyvB3GqttbXWxDuvIgUwMDUVUTF3RbQ+s/pVSHQA7f287qDtBmYOvqFZzZo4iApqt0RGM1A1qVLWMs+1NiQOMcs0yTjVPHCIwKzSyjKvy7ysa2/LcczIEQBwU9+z3sKm7kSOm53DbvOwe513HVQHAOgTeHJ0B/Vf25zjnpT3zdNTfpE2z9ePHz/+8MN3P/3w/ePD57ouaB6pj5yACLc3SUSlMWHkkAMNsSs67ul7L8N122y269g/NzF7+1K7UrQbALhtAqMduUxMAcEJiSCAETohWC+cXl2B+fdv79XUAChQGFKehuEwpTzmNN3e3Nzd393f3t3e3B4Oh3Ea0zTGcaI4UMibSNzWKlBtrS1zW5Za2jov5+u5XJ90eTApgLaU5f7pcj2f56eL1gqioMW11LqupUiXke4ldqvCSNeSxWPEwME5HOkw2FSbDWNKOUdOgdNrhsI2GdzeWDVV7Erd9gxHd1NTMTABJ1VEUJEKQIDWNU+baGvajbHdXVVaqwjQ510q6s6b3bICESARR+bIGFBN17LCGdXU8UiAFDDmmIccU0REqTJfFgdvZhTYwUVU2q/Jbf9IaO/b0txAe6rnsBfuZq6qHMI0jnf39+/ev0OEwzR8+vT546fPakpA0Jd6byADgDsThsRmNl+X8/nydDlf5rmK3N7ejDmX1tLlUqTkYeAQkHjzkdiv5571S/HtL71qZgoxML88G4S9OdujB9i+jh1cXYq1AhQoZIRIzNuKr1VrAdMQ+HA8phSWy7kPPdycQ+wxB03YJZN5Jh7SYcrjNOXjSUTd3aSulydthuaRgQEdCIHQAUyRujB0x44RIPlWwtO2xX7djiciekmbCPvgq5Z1XZbWqu0x8RlR26Nja3K+XK7Xq6r2rp0ibK6Wbv2N61rbDihmyBSHVHStsq5WGooHRwd1VRHbfP2MkEMIZmraOoKGCDf3ud3ata9+ezWkxh0E3f+zj0k28VEi72f3cxe/bxxzeqkZ9sSlt087el9EpImJ2AZypj6b6OPD/rDAFMAIN/1pos3g9KX5aaZdWB5Rup4zoah3r3riF27P882oai11WeYmjZkDk6mBAzup27WsMrfzfJlbccLDzZFyzIcDD5lDRAByDw641rnU6+Xy+cOHx4fH89NlPZ/L5VJaqdJkLba2NEwhpJzHOB5aU20aQmSK3Tk6IeN0DIGHcTjd3R5Pp8M4jcOQ8hBzpBgpEBCCmYn4F0YeMKYkw8ghECGAhoCRicBdRZqt87zM1+t6KSqCUMEXt8taznMBI/KIDbG6VgkJ72+ON6dTqXWel3me12U1cdduJAHuBEoGagK6NaKdCUPoVljUuShASIGH6TBMB6YADq22uhb3quLQHRQ2UtPrhvzWc+jIVFdzVwRH6HMStSZiC8qqdfa6eGtQq0uRdV6fztd5bQ5AMQ1THMaYhs5VUZWyLMu6zmsZTzdvvv52PN7GDMCE4L3HD1svm7xXFM81DYKDIznynsR2OXgApC0pEHN1+PWqem5uE6K7qtV1/fjhw/fffff9d//w6ePH1gojEgEj8obk8Gat1CJSwWzgeIghR4qM3NuCuFVp7o5gbqqmsoX2HZCy9/69NxnMDPvNEaE5GIEzIDEGICRkcu7i7c/3vF8x0O+/vmOikPNwPE6nm8PN6XC6GafTOBxuTjen0ynnFFLgGCgxxogxEWbEhLQLbPclYuoqZuq264OvSz1fpCyma22lrMt8nS9Pl/lyna/X9Xperk+fP3+4fPxwmUtZrnNZD/PcvXeB6SbGAcMQc0hHOA3lhquiBwQkN+puOK9OXdwHuFtHcjOM6+Ag31yltjOia/JCN2gXB6ulrWtb11bW1sQcQEXLWsCt8CrNWjVtThQ7roaIQgBzRyaKzJGtahPRy1VVkCGGINqgY+4CE1ETraUiE3LkBEjohgj0KzjaH4Z23+vzZ96Z9fvtWYGqIVofoXWrj8hkZufzpdYGtK2m7pvTS5yeZ5pZHwmoKSBwCCmlNGRHiCnmnKfDdPfm9u7z6fw4L/P660O1c/i25fjMAARmSjEE/iKF3Dh7r9Ze92YCE5dqdQGKu0GpmzQpa7lcynx17+ZD4L0gdjdpuuPAXBu6poBpiCMNPA7pMObDFA8HMyBmMLVWNBJZ4RDDMOacU0yhu5Piy07fzSbJdwAc/HrDPyc31ven9yk+UQyhO12+NDNeHQ7urqqllFKK9n5p7010Al8eYuTp9nY6HHIejLCoRoSAsixay6pTIBpBEoiZiIl6E2/y/B6YSi1rNxHpo/fnMbt/kZB9eXZtCUDvq9IGj3InghDA1FX3cZ8/Ayzg1VG+nVU9rdetCW/ban0eN3b3374H99ybKRDBZvewv009F+l+XNpUmUxdUc2wUz7RO0Lm5ZmYWam1j+CkNTXrYvhG7mYcGQiv83qdr0/XS5EahzwcDtib0T0zVpXWSq22rm1ZLk9Pnx4erpfrupT5fJmfHp2AckQD5hBiTDkP45gPB1c3g5zTMOQ4xJiYEZgghphzGqdpGIYUYo/V1lqra0fR13Wt8/zh4eH104hEmTnERIEdlNgxsAAsTWqrj/P1slzmNjcXQWjg1VxciASAgkFKIcWQj3ni9P7t2zf3d2q2lnI9n5frVWux1vqpaNAh1iBq0qS1Vjr63bbpt7qLWQ/1IY+Hu/vj4TjmQUtbrvPDx88Pnz5LbU4WOwMhvfCPcZu4IULPcB2suRTXxYVVq7cibUaZrc5QVzQlU5AK62KXR72szcGAW8wUIiJjCBSCmrVay1pLa7bcZjIr1zQeqDNZulh4D4auvA8ge4sIAbA7IW++oNQ9m/oxZOjqoOaq+utjt58CuFHd1uX68PnTLz/9+MvPP56fHmpZ0J0IGIgQCVBNpTWp1VUCYIrxkOOUOnwCEAF903hAgM3Fqffnt8Ngaxhug38HdEMARurghk0uwI0QmaifPIAbv5yIeB+gfXFa1SWkeAz5mPk4pimFkXFEHUCGttDVdUXthQyDEwExQERIQAG6q+U2f+uuMY5MRBAZg3uK0YgMBrWmx7oe19PxplRtYmVdluX6+eMvtx9+/vzx56fPH1ErgbTWpClHioi4NocFGnOkQxjGFBp4VS1V1SSPLxudiEOIIWgMYsFA4dl4y92ktVpWAne1GDRyYkqAoKJIAL6xE0yhe270M0tFKjoiaTNpZuKEdr1e0uODSImJS1tFFZFCiG7o1lotKg3AArM2Xa5Lra1PYN2sVeu93KiJY+h2d7+qQ75cY3uFteV0r8qb58tsA7Ku67qWMgzp9vamlvpj/Jmogy47pYZCZDdTdSCw/pcrEHDgPAzDNHIM7i4qqsKBj6fDN9++W9ez6s9lrVvJ/fLK9uE47KSQjr5lzDGELxryyBw2H70df7fZLqp6q1ZXJwUgd+AeBufL+dPHy+MnB9umg0gm2m2WHZs3AASTRmBjCjEMNACNBzyOPI1xOjhwSikwxoBejsFmc3RK0/EwjlOIEZEBAyDD1gmH50RlG+FsmLqXq7/V0Ft/ambOgVJIp8Px9niTU9o2McJLytMXoG89pZ2wa+6AFEIeptMpT+PdN19Np9s4HoSQpT1NiaLYU5XrSgNnP7i6q0ttWiqWRqXZWqEpAKq2smxRszd1dENMPb+AX51b+37dT4aOqO5DgsCAwELWB/l9XrDh4fa93vHAnYqrIq01qU2bgHsvYmibOGwo/b4O1URaE6kcusFi50mg+8sK6sleZ/t2Vysjx96rNlfQ109EVK7zvK6rNAGHGDinPAxDT4A5sLrOZf349Divq7odY9xQWR0s7trKWi7X5fFxfni4PDxcns6X+eoGTLEtRYqEMeacOSGJD9M0HqbpMI2nQ+AYOEzjOB6m8TDkKcXAm6/UduRIW8tlXsrlUq7Xssx1nuuyrpfLernMpj69+LqiKbkHohCjcQB2C76ir61e1+XzfJnr3LAaaG/pIsKUwimlhDFjnGI+5vE0DnfT+O7t/f3dXUgRENbrucxnWRZrxQ1txxCraBMppS1LuVzXy7VeZ1mrF8Gl2FJEiTml4XBzevvmN998+/Xb99D0+nj5D3/zt1LaSjOA39yeTrencRpfr6qOy3QkdER1k9WWR/CGLlZXLSvIAm0BKeiSYscNmnllrWw1IAu4N63FpTZ3xxDcQXtGq9ZkfrRlfbyJ4zEOY8w5cGSKm+keIfZSGpDQvL8WgB4OmV5P0MEAwMHM1bRTb15dvp9xjoAq7fHh4ecff/zph+8/f/igtYQ+WEIgJOhIjlbWeTHViDikOOU8RM5h84HdzgN135G63e02IPUp6fNBoX2Q0bspTAQAhOSOuo2fEAk5gFt3cDUk7lanAOT+uo+tIg8//ngcc2glWSvS/ClUxiuHyAGRHXoDskMbN08/8+AQHRmoF4MUiAJiJAhMnJhToJRCGkM4UB4oZ2d3EzJNzIc3N8PpDTKL6ny9nB8/f/zphw8//nB9/HQ5f/r8+dPT05MhNAFt63JZieYQh2k6xpzMvFY7z9qMxtu3vN8Lc0gxgwEaoCF5laZ9wOKuta7urdaaUhnikPM0DQGJvMM8kImA0Jki0eZ91ZHCqobooiaiKg5uT0+PCj4vYx6jg5o3AAwxEQQwqOu6rmsrBQC1aV3bellbk16QSJMuAJFU05hjiiHya2sx+FVoZ+aYAr4a/zCRBzC3GOLd/W0I4Xy+dPDUuqwPnz8T+P3d7fMK7nkidfydbww6JOrllslGhqEQQuxqCrgBPsDtecbq+whoG77C3pTF1//eIPCw9YJf7fm9an8Oeq7g6ibQB3JqZuJQwRAoSC1S1lbXVlcAR3p2PII+tyM3BEU3qIvraljdBV3RFE3IGlkDciIfUoDDiMmCkRoIhGEYUooUInQ7baTdPG9ngm3IgT8I7Nvus14UePd0gh4YeUfGbUZuuxLCdhFSCJsTmu0zdkUDRkws7E9aqtbogxiuKpdWrnVda6ndki4wxYBAmARTw9Q0FYzRa/Wqqgbg3R2vo9v92VfqBQb0xbUBrMAdCAjh5dX25AzcQdC2Axv2qL61j7wPMgidwHuB31FjAZgQOiDOEAyxIwtijEhMEIccU4oh8NaEd5Qm87w2k7IWbQLmvEXyfmQREUUKyaGJk+vrNLiJnC+XWhsi9hlQDMy0WT45uKqt0pZai4gDiFoTWddlXWZyJfAyz9fz49OnD+dPn+enp3VZW20hpByZ8kAO+Tge392NIWWKw+EwnG6G0ykfDiHEQCGFkGKIgRjBRa2UVrWtdV3m9XqZz+flfC5Pl3K+1Pna5qUupS5zna9yGOBf/SfP0555XS/zdUBIZFv/BLVPStKItzTdYOIIzMhEDBicGCgABSB2TCGkEHIMQ0xDQtBiTYhwynaIwcekFdZVatM+VzZHM9IxypHubmMpslZrguKhNlorzc1mtXgYiI0CpDFSTCB2e3vz9u0bt9s8pPdfvb1/d3873nyxrjYspvfKqK2X9vQheUEUr8VrQW1oDbQRQsicEruCyQan3wdigGLgFdSwU5ncAJTASIot56KtzeeQckxDiDlw6oGbA3EgCgFjtFCdspo5IHrAvQQH2CGm4LihidVUf73V0RHA1Gpdnx4ffvnxx19++uH88NDKSu6JaR+Supg2qaIN0HMMY4hDSmMKgTtyxnde0cscExERnRGdEK17pflm3uyOsAHhiXedATNzQjTY+g9kDr1Hh+YUIW5AvC+IMMR8++Ytm8zXua318dNTZAzciWIm6k29HxMpUA59cICrQFUQoJ40ROYp8CHwFDAHDBFCZMo5pDHkY8gHGicM5KCiKuLj3dtRNR1OOaVhGo7x3THnN7dvrpen6+Xx8+Pnx6dHaa21WuZzuV7XtVQRKrNoE4PavC3eILwOH4TIzCGwxeASoXcT3UPgPvZVNUQlUmELuzIlIgUKIabAQC5ScY1SQquhbgRF3IEN0PvhWmvB+eIgTXvyg4SUYsZIgYOJQBceNCAMgTEmD1EorIBoZi4Opc9VmbYW0Rer6ovQHmMYc96h0waAMQZEEtFxHH73u9+O4/jzTz8/Pj2pWq3l04ePHcGxrItIZ7AYMzEHRDTrGNTgAJ0JrfhMo9twykTMHIlYVZdlefh8efx8LUWemZ/bEMv99ULqW5qJAjMCmtivIAQItCeojt3TUKub4e4H7KrmDR2RzaSZCIBzCAAG4FKq1NYJThgDbGMZxXq1tlQQtUZWWDEOZwgjERNHV2GpmYwiBY/qIBhi7O6lBEybffrWcd64+5tJ4U5b+1VP/rl1BnvDzx1UtNXWWnsGcz2vmf4ehcA555QSInln67g07Y11mdf108clliWV1RxbLQ8ffn748MuyXForxBxizClyiBh6AA0eGWO0EnRevCr13lMh32vl57f/H43tzzC6XmP5zqTt37U33re76yiNDY7Z5WAQkEEMCBzN0YwRI7GSUedXIhqhGPYWx5CHlHNkzDmO4wBoIrUDLcranp4uzeS6zLU2Akj/b77+q0m2JLnWBJUY2Xs7iYhDklRWARdA3x7pnhYZke75//9hnodcFIDKPCyIk03MTMk8mMchhW64pJRkVmVlhof7NjVVXetbKU7jODClnGKMGBgBXLQ1UPnBKtqanK/XKo0DB+TARIhm6mpm6nZzjqiZATiCmJZa5vl6vST2XUCs67LOl9P5ZZmvooJEOY85D/vdoX+Np7vd3U8Px/3xMO6H/T7vpjAOISUEAgNv4q3qtm7LvM7X9XJdTttyul5PT9fz83I5letVl2LrZlvR2qSqSlVd6f3D+L/9z1/fyHmeny+n0doE40gpBQpoMVAeck5DTodxSPspjykOHBNyckT1bveR2kRFVEuTrdbTY/lUvGlDtPu7cNhxcNNSn56ul+umvdPk26MaAh/24eE+U6AQUkgT4qg2PJ7L70+nl21b5DrPT5fLwBpkaymln3/+eb+f3ry9/+VP79++f/P4t8t8Lq9Pxm2s526moq3W+bKdvgDWGB1FuAmj000CRsMwpGnUWkUNQnCmngYeArMyuIB6iBGAbsJMUYqBGaxtbZ0FuXKMMYWQOoQrRAqRQoohJuMBeADOlEZPEV4tYv7djqoPwM3/U9fer+4AKvX08vzpw4cPv//t8dOnuq7sTkg9f7ypinQRYCH3nOM+5kMaEjETGKiCfj0uvx4FN7nsVxkqmX6le5rj7efsbKPuWrztBV89TrebTxFXM0CkaMn6D+z43cOecv7L//K/Xr58/vCv//r59NxEI2POqFK3sm1NNzFTR/DjMNxNw5RjJDot5by1zVwQGWhkfjOkN0N8yLgPzijMToGJE3KiOFAamZnIRa0p7N/85H/6x/HhbTre0TCFPN6Pw3F3NHQ1XW96yDLP1+cvn5++fHz88ul6fmmuVZoqNSGHwBS/1zghAnVFCJOF4NG6UDOlGBPBbZPASIwcuiXEEYlCSsOQJ3eKLNqwLK2kWmrhgHRrdsEZmMH11smLtFrAQUPkkEJIOaUUY9JhADcOrGIdGqnNtlzcaJ3XQgWh9RRtFVUxFlOy7xVO8HelPTCnlF6Hq46IMUYiUtH7u+M//9N/++WXn06n88vzy9Pz6XQ+Xy9XV1+WdV1WEXm1MNFtEu63BWgvxW6utW3Ldr0s87Ks2zrm1H9ZvQR346t3MNTrffPrGvfr44zYNWX0Sp8KHCLRtzeCrxeC1/+fulYts7sTsbRat0UUHBtkxZhdBcBTyoiurWqrTgzsxBgYM8PAmgnBmlmpbS3STIrWQktLDdN1idNjiPH1PRuTArkSCQbgCBSAq3GikDkkovC91wJuzXvf38kPFAh81ap8d+cztW1dz6fzy8vL6fQyzw/lsE85fV1J3B5mYiRyAFERqes2n67Pn1++vJwfr1I2hlTWwWqkCE3r+UVfLtE0cSRjbGS1FivSfV192i3qTbWpigCS37zIfjsk/tOd5Iez6/Vt9G8CAJhZn1Hibfx+u/N8E0zeug5HN0QnMDDvwBAVtVdAoTmqQ1PYxItoE0OM4zRO08TkOYYhRWt1beIAHAnd3V1URSVmfnhznAK2+/2IPhIxRyBeTefWpmVay/b9cte9c+Ckw9KKK93s8UyICl57Ze9PD4J17JQpgefAKQQSKTEF4hhjTjmHPOZpt9vvj8cQAiDl3bA7TlMepzgQkzNZk1ZarVqLyrq2ed5enreXp/n0tJ5O9VLKZd2W07Zet23W2lCdRFGaVClFFMRZg+9++DiIujY+Mdzvh4eHcbcP4xBTDH0o0EXe1ICqBcYUoqmVzc6n9fnlvGx1q22rbauyrbYVU5UY4c9/PjLtE1Hd4OPn7dOXk4A7de8iM3EMIUeexjDtwt394WFCZqxVxwRv9jlEv0oj3U6Pn6yhVpzy9Jd3//DTT+9+/und8X632+X5sX0t7V+/JV1BxASMxqCJfIjkjqreR6JqXb/OxMGDdWNAF5ljoJCiq1olQIuhq1nAFSEwxxCG3Jq4NHQIgMGQ1MEQENzQjExZa1CMzgMPRyRG73cJ5Jtgw2+jNQCjb938dy9XlVrKfD5//OOPTx/+OD93Pbx+BZP275QCJCSOkYnGEIeQEoc+WYTbNuCbzuXbFRmgdw+At4O9GwEJuuwWuy/VzQxAb/DPGxsQyfsWg29zQ0NXMO3zzO/njBzC219+VXf98Ol6WWbRiLSDtDY5Xdu11KWqihL4uyPWkI+REvKnTb6crsVcgSLxLkVkzmPaD9EiWFNxJXU0A6kkhepCiAhq5uJoLuKST5/TtEvTPk37MOxCnijlmDIzj7vdcdzV493D8f79+59Pp+fL5VTLui3r5bJerkXn1uzHz8OtR7RD3/ffBEIYQkgpQjcrhZBSzkNOKYXYKW4ppXHIkxuitxxrjDHEEGOgiLcGD4nZGAS9ISoxMQNg37yioimZBUTiSJzHqTPNGJkpStX5OtciIaaOlnkV9L1OZxz+rjH8u4E8pRi6f4UJmW/HljPf3x3/+Z/+8X/9X/5vSPhyOv/rv/7bX//tP/72H7/P81xKXZa1eyv56/Sh1yFzd7ltvt1VZbnOz0/PLy8vl/m8n4Ycg7uBGQFHGsa8m4bpGq590P71+e3D3ts9FLA3/IxISDHlcbdPOf9YS76Opzs4bW3rBQAo5Loty/Vs6kgJTTuRhwCGcRyGYVsuxQwTQXAKEAlGagNpZjB1Qa3aWtm2bVvWzX1O5znlzyHnGFMIHPoUInHIwYiUAm4bzUvzKBDG3XHaH9O4i3mE1wbX4dU76G7Wvn9aEKFvs74Jxd1F5TrPT8/P+8PhsN+/f/twfzww09fS/r2IzW/mrvX55fOHT397PD++LKerlBU167Zn2+UpG+Iy83nJeRjiBMSqej5f58ulbJuapjGGFADARHTbTNSRQK0X/l5iv/7I8H/2wtfrVrepd3dTP5eJCOnVNPD6WfurEhXN0I1A1VSb1dq9+q1JaypNBYyq+aau1WrTqhYC7cZhvx8dNBGOzFKtbZsjRY6RMEYuBgY2TsNwf5d/eRtdgwirurM6bSKL1GXbSqvzMHzrsBABQU1LXddlrmVjpMhxGIaUBgVoralot1B3ZR8BxMDjkA67acy5hghN1vM1c0x5OO7v3hzf3t3f7e4OYcgQmAIyAYlj0bKt23le5nW5rtdrna9lm+dyPi+Pn9bHT9vpc7uesBhWNa/mTVyJKKQUI4XohbRIc3IYCaYfrKG7KZuNeQh3U/7t7f1vv75592a/GxN3JtNSL5ft5TQv81rKlnM87Cdpdrlu//G3L//675/O17LV7mIwU3JDJNzt4rQf7x7YQ1oKfHzUv/77UlwF3AHB0JUIMLLvJ354k3/7TQBzYDmfizrFkN/s8x3l81xenj4ui6nFf/mn/+kv//Tf/vEvf/75pzdgTaXwd9vEm1APENGZEQNbDJjCLoXDFBptRftCF1q3SfYYiu6egVtnBkyR2QAEwcGIHMmxfy8dY85xHAM3UGWknFIXNxi4gRM6OWADa2DOThsic54Qup+Qu464j6v6JdWYmMg7Yvf1oTE3kXo5v3z59OmP3//2+OlT21ZyI7r5z7vEiR0y0BAT5SEQB755TMS7Jc8B4avkpQ85mfk2mr9Jjcyg36m/AfwBAZzUTFTVrJmpg3UPmDQmAyYGSOQGbm4Mhq7o9J2aGQAAicbDMR6vfnesW12pWsxpN61wfpn1BNsMrXklN+JhHPdxt9MQTtf26KsYIHLEgHEo+4O9PYb3+5BQ5ovV4l9BOn0+5wqqnU00y3J+/EBPnyPRkIdhHFPKMeU47uPuGPf3cX837A77Yf92f7Rf/9JMSyvLPD8/P//xx4ffP37Wx+dla99vFhzc+w2n/7rQHR2JOIaYch80xxhSTkNOKeZIMYWU4pjjlNPgCibelTDMHCJzpK4ADRxErJH0X1t3uwWmLifSpg2VWGN0DhzSGGJOaYgxBo51q0xhPi8cIt6cw7du7+afvn1dvr1+KO058W4KpZqoEb7qdAhDDMOQuj2GkYdId4fdw93d+XRZS7leLtfr0hfw/XTuqvhXfz7llFJKCF6Kqei2llqqiOor8hAQmeJ+d/iHv/xliHG/mz58/PT0+DTPS3cNd/+a928vABk4uDMQ8U/v3//3//4vb9+9/7Ga3L7PbmKyaVnqcjFz4rxdL8v5pGJAyc2YyQ1M5HZYqBJiHMfAkVADtBF8QAuMTQhC0jBKDGoJIZNbCBwiB0YCQzMEJrS21XXVpba5tqLUlJYiTfGnP/3513/457e//Ba7xflVU/Dtlv3jZ/O9mcGs7+m653O5nM+n0+lyuWzbJiL+ah5w99cG+fY1VVfRBuDjkB/wMEzhZbm8rLO1JvPF3JGTS9VWjQIIUEB39CayduOfYaQYIyIKYFXv0j5X0x9hbf1P/1NHAg5QS0nRe04PwqtCx/0rpwfRX5Vz4H0pYjfxhYkh9q9NvV7qspRSWxUREzFxp02MBYzAjAKnFBNTh3QZISZ0Mou1OWLImczMrNR2vS45DzFmzCkyBmsoAsbomCyCpbwfxawu8XvEkwOUWi6Xy7ZcpdWcMme6XT17F/NNMEIdABZCyHkYxmkaxxwzASNQqzXGOKXxMOxzTsFNl7lKa2Vr6yrzptd1m6/rct3WUrbWjMVYaq3rsjw/refnul5A1xyAgrupASpEYdLAECFFDhsDYDWXjDHg/rtP5P6wT1yRYAgxA2HR7fEsCCZaVlnm+vS8fP5yucxLqeXt2+Nv/8BIuNV6WZbn0/m6tCoInTzhDMCEbAjAHDhwCETigGLYFKpDBwZKA1AgsK0EILx7EHEUac/XcylGlHbH/d3DMRxC5PQI82UWImV291q2eVvOy/W8bcu3p6Nzmbt0g9lDEGYlCgFiYLs5Rd3dmog6iLpafx6si6SsiXV+lao1dVWlBkR9wdkHUdbh027MFLo0yG+Cc2JmAjcxUTdzJjBFd0YMTP2P2yCv3+OtMQEx/F3Mm5uL2LaVdVlrKSpCgLe9bn8L4MAY+mLjJiXvRc7E+8zshlj4ug/7bnWJr7ohgkCI7Ibd9vQacGC3oBnVzqB3RCIKCJGJCMMrH0Ld1Z161/6dg7S/VOTz3/44v7yIOoaIETBFjmHYDXdyDOMwNW1S0fX+br87TvvjbkjpZ8K835ujG5p5irx7M+X7XbzfQ/C1ldIU3akH/BASArsxJEA0oPMij0+nVhUVcoxDTh3+n8fdOO3Hw2HaH6b9ftzdGnqMOVFgtzDl+NPbKYXDbrgs24zf5IUhhgS5J1g5QP9F9YsSxdBLOwbu/kZiijGmmFMcYkohBANlBIROB3Lqu2ECDhgCI6KrE9OrRqQrbM0dALlbkiUGgGjmREwhhphjiNAvmimGEJj4+6JOcPtZ/iuFfE643xGir9trgXAnxJxjCHS9XD7+/geiqYrWliLnYUDieVmu89yN3Z0vawgISHxDxe2mcZrGfjsJfQbhN1GUd8AFIod4ONy9uTv+6Zef7x/ux2lnBmou0kxFtadB3ZimqmCmCBQy/enXX/+f/8f/7kBPL5cfC0qXxIvVrW1zXS4ihpDW+bKcz00UKBJhSqmvOUVEVUwVCYdxGqYdWmFZBq3JAdAKgnCukQURoiVrEW2MGAMyAbqBSUAIhOu2Xi6Xz8+nz0+nl8t8viyn07k2+e//9/8HII77w93b93hDqiN0T0l/GMm+L+evo+l+a3dVBSBpUkrdtrW/Wm03ATziV6bbrc0ncHQDVdOc07t3b9/AUax9eX6Mnz5dVOq8GLJnUGlVG9SKocRu+TN1V3BD9ICcOCD1rSXeRuF6GwbdROz+dccHf9+7O2zLGtnyjplvPj/s4rgu6uzDwj4gvZ1Zt8GgiEoDA5Um87ydzut1qVtpzcRMFdQBqjqLAyNjyIlSSgAm2hybI0cIbFZaU0QWAWmt1Pm6PD1d8igGLJIlcXDhmyuajIEjJ87AxNVB/evH4eil1svlXLeVwHJMMUYOAZkJkcx7OJiZ91aNkGNMKQ9xGPIwQbQU8jiOKkIOZMbNdF0uL9tyOV9eni/Pp8vzSzld5Xyt69zKKmqOnPf3eX+HTODiVpwtTJF3426gEGBurYmLcQU0UM027ZBWd211gytjQnz3+qkgwtu7w26w2lpk9KbXp9PL9VyXddvatumy6vPL9unz9boWc/tv/+Lvf36Xp0DRKRqxxOgh3mh+5uwQKPCwi+MYhyFkopYo5ZCGqEKgJigORuqqLu5NoSoqMMVQtV3qejqtrcLPZg93hzfH45uHGMOT+zNBLevLy6Nt1y+X09P5/FLmESB9rVqB6dXNS8Y3RWD/H/sz010itVV1bCpV1Ku0IqCOBtqkgbTSwM1LA7cG4Ihmjt3nXSqsGxIggjN1Y5u/wmlCCMwk1bvH+fVRwK7IZCZm6tJ2B3RTE+lH+t9hJ7sLR0RVlJBSiIDYTVRu2v3UjMRAAYkNbkr719AGc1czMKVXVO/rNrLb1fqYs7NugzHaa9aduok2k2529+5kRdObQYgJYyDEPh1AgE6wRjewm/Pqe6JTK/Wv//H/W2ptW0HtOH8L2I4j7cKhmlf1Js2t7ad0v88Px7yfdvf3x2YISCJ+ua7S2vHA0zHHfTbXq+OlmZnHEKeUYuQuKhwjE5E5npaXv748X69FKwSmGCgQBMJxSLshH4a4H9N+jLtxmHa7YTrE6ZjGwzDs8jD9PA33w5s3+3Rey//rb+f2+pjHlMa041YpMAB2srADYCAM3Eu+YSePG4CH7uKOKXa8imlfa7gp9CMFDdEInQiMXltuN7PuszZCdENAM0EVahIcXNSJ0L6K9Ii5D/7jrdP3b4rjGwHs77qqH0r7EOyQFcRcoKqJOiMGoBgDuH/5/Hm9XtZ1MbdxnKp4a7V737s0rktaOj3fXEGdmDjQ3d3x/ft31+v1er0eH9798uuff/v5lze7/RACvH47zT3E+O7++PbhwR1UvYnd3d2HQCptmed5vl6u17JttTa/YUUdwIeUH+7utyrfl/buCLtJtl21tbpuqkYMKq23m51RrdLQUUXdnRAocAghj+MwTqBg67ZclsfT5+fz+VK08EHCHuIEzA41oFfEeBOGCVmLIKnPGW73Ilvm5cuXx9PLi4rO14u06nYLv7iNw27vpF+6f3BbwaubCwDM1IxCwJzS/f39u3fv9/t9iMHB7Wv3/PrZ0tc1G4GBi6mouOlWtnm5vjydLs+XVUSY7LwtxJfHl8vLhXG9npcQGQhrKWKtKxGLVFudCEW1mhi9Iv5uByzdYGGIX7uEH/Onfdu2HCHlAcLr2vCmpeiXm44H96/bCXUrza6LkIsLM0hr9Xwtl3mb57ouzUiQbIyBKPZJFxOSG7ujdUYxMWMMnCILdwKAI9i2rJe1/P7x8d9//xhSPhyvQwo5YiLLDNM4TOOQp5yG5AYA/L2GABERA1Fgjjl5ZJzGaRjGmDKFpO4oPdzW3YGIxpz30zQNORLJVpeitpa6rNu6SNmsVl1Xn9e6rGXZ1vmyXs/zdVnmzUvB1lSLaK3qAqTBLTYegiFsuK6kjhhjGu9Zg13P7Ul1dt7MHeTe7BhCyIQ5qOBioPb9x4E/vf9F7XBdZ9OaIrls63y9vFwuc9sKiNJ1acsmpZijuQEHTpHEYBrwsA85OVAgdHAvikXR0ZENyXslYEYKgAHQEIECMQe05KYAjtMQ94c8TTENSMbTITfFugEgXecNMMQ8Zqb73Ri8Xp8/6XaKgZflsi5zwF8IX0s73Lp2R+jhZWre1Eq1lazWJtJzZ40I1eByXZai1MRLac1uF+qbJvwGlripdx1Ae1SauxkyMRGZuoghOvirJ/u2t/ZuDe/Nbv+b6eby6JMoB+wEhtth/OMhTBRiTLtxXw/HMs9WNtk2EwMDIAxIiMTUdwUA4DdX201U9HpMqJpKD3HqAQ1d5kZw27f3GUsn99rNu3pbqoA5AoSuXbIbwdHMOsWSuoTeDRFCoJsczH/wuwKAqPzx+LmIzs0upV1L1RwGHHPArn410bKVVgu6Jw5T1iE7ckgpxTyownWRJgU8B85Mg0mbKz7N1kRDwD2kSOpSdzm8PUwpRSOqiqe1vcy1tf4Lvg1UUuScwpTDLvGUaEo8TuM0TmOexmHajbtx3OVp4pQUPPy4peYYE4/EjEgmINUaaadY2deFtrsaqr7284Sd7ot0y8lTa2YCtzpw202LiqjLNyKIQnMITPzKdk0xpRB7/o7epAzdSwndA5hj3/0SU0eRvuoy1ZX8v5DRZdZ9bJ5IG1axKp5DYKQYArh/+fL591I/f/ni7r/89FMax7mItHbTyhC+YsodXrGm7IyEh+P+1z/9si7Lsiw//fIPf/mHf/rzL7+8OxxCCD1Eor9tJtrv90OKW62lqRqVst0d99Lql8+fPn78+MeHDy8vL6ruJp0ojt4tm4Hoh3f1VVnQ5wKqKrWqekjxNv8PHFNiIr85s5WZAgciDDHlnFNO3qSs+Hy+/P7X//j//vXfXjY5/PLPu3d/znd7jlEV0IwFb0+yh2CU3UcQAgo5j7tp2io/vpTaapWuWek4odv2/NUdA9AnueJSvjcwfBOUQT+8gJnHafrp/fvffvttGMeUUt+mv6Javt0LsIuL0NW01rqtay3r6eX58+Pn55eXbvck5quZNN3WrawFDNAJGTBQTIECE6MBrGWby9oHQOoK4EyEDIjWle9wWy0Q8y22kH7YiUItpSUyM+zGiz5d6Ma226bFv1b3PqUqpUkRqbStRKAq7eWyXZcyL2VbW0ieEg0pxDhASBxCQrjdhFVMG0BIKQwppZAgBmQEACTctu3zef7j3z/97d8/Ooc8PTMhoQ0R92N4/+74/v39Ax0poouoodkA8HVRTYgxxDyMO0w5Bx7HMaUcUkIOpuqI5tCtJSHwfhqPu3EMgUS255d1rdvzeTtf1vmyzde6XNvlIudLW9a2NW1FpTT15hQD5UScAYBL1SJqMIu1CMGJZqwXEEdIgccpcNSXi3yq5VJxUyAswP6LIxFgCM5U1eBHodBPP/+JuL1cnrf1lHyReTGXIu281lKJQ3YOFBMbIELKKeeYM6jBfqKHYxJ1pIDgZn7dHIoXMwPxG5+ciIkCYkBXRIDIzEzd20PO0xjf3I3HuzyOZEhv3h5TtroFF79elrpJSitxuJ8GNL2+fFkvSIyioiZTlm8KVAR8rWJ+M4ZZUUOtqG59sk2OBBy4iZ0v19LO2TG5gyoQM3MnFJooqDoYvsJaCG7MZwdgBwYgda/SuYnAgIHczOkrKIl6vx7CV5oxdEk6IRlgJ0lDL8Y/5jsG5hzzYX9wadvlUufrJk20j/OAe+I7EphbF2nCLXus34v7+W4iJi0yI8W+puhPGb3+2zrEyfoeVLTVpqL9ltHlPDEEQHTn/s9vbtavOqZdD023SAkGIvPXZOWvpd3s0/VSmpRm16Vc1k2mPLJ5ikw4Fzkv5Xydl3W7O+zcKHJmzhgIE+8SO+F1q5fzdpxG9MSezWCp9LLYtgkHKC6M1rb5YUpTShwSEInhWnWurTZrYr1Du20lmXIOYwqJLDGOOY4p7gLvmKeYdimP47Cbxv1xx3n0+NvXx7xL5DqGX6vV0IiqaTcLdt4luiMBKbmpu/mr4K5PM5toFW1q4l1QiUSA5mYi0qz1IUlfhdwcEBhDyDmnIcWcOLIBUjMEwy6kQGdC7tV9iCkFDtip+7fZZiNC/K/iYYCIYxhHNqCl1q35lPNxv9vlMXHs6OwqJqovlyvMy+myXud1WTdzf01ao1f/BQK4KUjTJqqqMcTDbv9wd3d/vA8h1aZOCAQdLtaqCGKplRHMbZiG3/7yGxPeHad1vhJYKdt1nretbFvJTPe7BG6IFFGtdffa96X9xuc1adKqdvW+AwAQU4yMgJE8MqRADmBGIcQQopsSQgBLqMguoMv1+uXp6enlvAjuzGLkMcc0ZFcmsH6LJUICCx6DxeQpghBoSMmRn54uQ/6yhmDuOaVpGFKMfVnieMtMdHBXsVa0rt9fhIc8PNw/9Oliz1hj5hBCyimm0Fq9Xi/zPHfn4U1WZNZU1m3bSimlbNt2OZ8/ffpwPZ22dbnO1/Plcr5ervOMiCFGcAKnHKbhuGNk+ppkwOCu27ZKKT259XbvBycEQwS7CU1eE7noqxz37+1vnYBE2IN1uzb+tjGgG3WE3bX3F695OEW0lfV8kb7pMNMmvjUXBXBkh+SQzYNI081VeQjoVlt1DwieUzzs9rucg1NJUXIEorib4toY1oiYmKpqWTY1M9P9lAJzSMPxeNzvp5zCurZWxO2bPLNvgojikHfaNjNtolirAiBrVSl1E23uniPvYtgTDrWWzx+/PH7RedPLKi9zW5fWSpOiUto2t+XaSm3NiDGM0ATmpikDTAgMAvgs+lRqoG0MNMXEGJ+qnjd1tJFxBzGTFajVW/PoSMjOwQN5MAjgDEZo3198EfHt25+HDHmIyxWh2AZz2Q+8tHqGxSAIb+rVQRFjQE4hpZCTqcKY8TBRHxX2rS2SK7qW+tq5hpCGCdPxzbvj1dvLXGrJGcaBxjHklGJIuyHfH8affjocD3sFqBqGgVRymct6uazX5eXLyzTtDsdjHoaQQgcLhZgiJcYfyFTYFZgEGKKniMRNvG3rpltCTwzMHAK1qqK6rHXdxEMgpoDIgYYUPQUKwVV1Ca5KHAHcREEU1V69Ng4OFAJzdEBz7+TWfsr3pkrtlp5GiIROHcnSM0GICUABrd/l/xM17NZ7vR5W0qqrEADz19hKwA69QXR0Q1DzZlpFqrTWqtSCrgyAN6dtxy66u/c8vFtvf0M4m4MxIjJRh80TIfR0dpev1+4OanQDIuir4hhCSurQxKtoFRv0m+XY3E+1tdpa07WWuhUJpE0FqQFclu3pvJyu87ZVojiNMm81zMtSLlVgd9wTh8fHJ6nN/L5fj8TIiZRIEHuQgbkXw+pocDs+DKCqF7Wq1g+pPk7Q3i8yO2M1JvVTaQw1IWTCSJSYhxSOu/Hdw3Ha7/VPv8KrDJmQgEOHIFWukQIDqnlHh37dH7pBr26l1sAbAGqMTFhKWWsprVNmkTnEwCGEXvn7ltl72k5XdLwGUfYNTp+J3OLU1Mq2uIu04uZl20QLM8TEaYhmamLetU8V3V3lv0h+IwZOQ2YEfL40BtiNw8PhkNNAzOISYx6maSt1bbKu5dPnp3UrzISI/efq7Zp/HQ04qnjZ6rpsOcYh5SHlGKKozVsdCEIgUzPRWpuZX64XbXHdFmL66ef3u9142MXT8/Pp5fnpebfbTedLDiHsRvr5zWQipUpAaWWV72zt7m7aTNhMpRapRZu8NoZOBDESmTFoJI8RAckMQ8zMsdXNVdha8kooBbSWbVlWU4ucdzndTXm/S8OYGJjRA90SwrsbkiyT5YAWEfIwAIWHzy+H/X5bliYy5DQNOacUmG/tnRugoSpo1bZpXV9nggAA4zi+e/uemADger0syxJChy27SFuWpZTldH6/LIuIdjGXiK7rer5ez5fL+XK5XC/PL09//PHH6emprGunfG6lrmUj4oQ0xCHH3TRN0zimnFJKPbSqSlnXq37+uF03lJ6/DQCd3uOGr8qgXvF7pGY/EG5xQt/VEoDQJ07EBGi3qOnXoWXfPHHfuyO9btyb2GUuy7ys62o3GeCY4gDIgSChJPcgQuKqDUMkHgkBanE3Qkgx7ff7KSWqainqkEKIab8fsIx53Y/5bjdc1rqpSW1NbUiJOR8P9+/fvx8HQpeyVm3Vv/s4zKA1ReAYs7Za6iYqVVqURiFU1bUUkYbumWjPNJmE6/n68ullXrbHi5wWWgqoWkQIAMHVa9G5ghbCvEvhfpBi87XWZD4hBqhmX07t921NIPvg98OYw3ASfykKqBJhMyVwRyGyGCAgpsRT9kyQwIJrJBuih/DDbevh/v1+H0OwSyiyzMHTWqawmDzBhhosFNWi7ogxckicckhJVXHKuB9J1IiRkN3ZEKt7kR4TRkgh5pET371t9wtcGwrCuIe7Y7y7G/a7cczTfhyP03h/Nx0OozgKjNMuAozLaT6ZzU/np4+PcpQpjml3tz8cBbSZhEAhUFuTtW/fq9vdkDv3KjOzqK+XxdfLYUx3h4lC5MTQtta0Vq3NpoCcYmJKmXE38pjjkNBc5tXVmAO4a23WxEXBABFVRFsDYgzRHVS7ztz7GoqJgNgcHAyxdytO6ITer7NdSIOO2g2++JpG+e17paatB0Ns61K31VUYIQaO3VVyw8vc5lrqva63tZSt1NaKtpoIOAYm7FMDQuoCu9sgrC8nXyPqCZACgb/akF5nrdIVTQDIpAZdfIT9ys8h5pCGVJqurW4ia5HDd7XEHJbWh/zadQMmBg6qXk2va32e18tcWpVxlNJsKRXJPj+eX87LNI0xpnmpQ0ru2sGRZEiBMbJL/23emHnO3G9MfWZj2KlxiJ3RpeZmZG4IgQOHjERqtm1rLbWjRRwcEWKg47T8Uurdsbz/xb5OgwiRiLu6IXGMFNiRHBggwCvG1xEATbySrtvq7iIaYwjMtdSlbFvtCVPIHEMMKQYFENXbYhAQsMca+qt5E78uV1z7m1BVU2u1rj1ru9VW64bsaQjDlN2tbu02tf2KLv2/LO1Gpjf1HgIQYoo8DnEcEweuQkBWbBdyBkB1ysMAAIEREXpMUL/iulEfzA3DuN/vpnEkohhDzsm1LdeTwx55UmPyG4hFRUopL88vawrXeXbE7ld3bEASU0gpIFHXGyPaUrXPN2qpslztO187uLf1alJVVbalLnMrq0jr7aK2VbY5gseQSDNqaeKtaQsZQ2q1oOs0BFQikIR2HMdf7t/tMWPIv7y7v78f012gocslOrrBrEM1zQHM0R1RESGmNO3u3r39+bdfRNv5dA7c13o3bA4CEDq6iou5MJrhDxI05jCO036/H8fhdD49v7w0kdbq88sTMc3LnFP6+f278+XX1po7mvm2lU9fHj9++vjHh4+/f/jj44ePLy8v27qYa0ppnAKHSC+0bFdKcbo7vnv355/e/eXtw/3d8TCkzCFszdYia7meXj5La2XdMAXEAGiu4nUTqc3FrN/YbxQeVQdwtG/5699/ryJyRI5IAQmIwIC7OQRvUU+gTmKkymb98TP10vy6yuVaWq3g9uYupUADE0SIKKxitaqhOlMe0DP1BaFZj94VscZuDhBDPuxiTMNu3MQCYyCIAZjMm5o1V0PwHrUyTSN5a1XBgDH8oGp0N9Va67Iul8tpvp4CYcopDUPM2RFFRV3dFaTaoqVcLy5aZlkWvRRcJRlhYOHkQ4CBivtS+bK1edUp4MOQTq18KnMyqIGDoQIsum66kTsQh4g50zSCCBLhYef7zPsB6mEI4EVH4JiH4SHoRIpbJfch4dsp8V3+vpiM025/GLQdsE5LoYIGgSFn3nFQ5eYkQKkh2rBPw243jMM4KuK03+/L1kQUCBHJjRrCanptDGYUqOeemWFPpMEY0jjuDuHuPj88DMfjtBum/bjbD+NhN0xTVkcjV0sAOVNkbeV8qafrfhj3eff27s3DL78oeXNlAiJ8/CBLey0nr/CGPisEVyJW62NOACJmCsQBsCvAYwpxGH95/+bnt3eHKY+7Ie7HOOaYEwJYqa6OzK4mPdS9St1qWcv1cr2ezmrgxGoG0kNdoScGu6qbgd1oiV2H3hPTA2EgZTRCUChmq9iWvKL/cAabiNR669dNCYFCCOixs2rMAb336+reVNda11K2UloTEUX3HEKOPKaYYww3ivzNXIoAgE7mBBCIIUQz6yGufVpw63PB3VHd+pieCBVBkRTd3UWtNSm+YWm12Vp1rVpE5e+yxm46QnToOlz7+l/6q7jmNUQCHVxUt1Iu17nUGkKozXUcL/M8L4vcDfgKLumCzVd/Pry2zcBM0zC8ebhDjrXpzXRjAO59uRLzEIeBmd38+fnZRHuD4gCGUBGvCo9LqcRvO1q/V0Ri5gjoaiQxlRADswgFDJECMmEPyjMQV7ei0mppJdcUYwhBqi5LKaW1pu7OTIFCiskJxQxBwBoIukAP+O1XMQA09VoF9ZakquqqZtK3pN2dpmoSUtgd9ky8DMt8XWoR7bC4av9V126CrSC616aq1j8JjpQyxxhY0RkaUmxOFGLaVH1dZrDmKirytVnzruJP6XjcPTw8HPa7GEKMHCPVspxevjh5yCEqI7mZgHfA9np68RB5ayXm5GAOUtrWdAsRQwz9cU4pmbXzKtoUHcpW23K19A0u7e51OVNIpiplq8si26at9WxvqbNul95AumQtuZW2rAU4QkjSGqHdjwwTkUtWeTuM9PDW8i6mdHxzTMcoO2+xViliougqpipeDZvdIu+AEMnUlGg47N/+8v5yvdS6EaGr9OCZ7tRkNAJzaO7i6PyjLYaIYkrHu+P9/UPMGYhOp1Op5XR6blLXZRmH8fn58XK91No6FW7b6ucvj//613/7H3/968dPH1/Oz+fzqUkNgYeYx2k3DFPT9vn5UxyH/du3v/7Tv/xP//y//frTT+8e7nJMCHi61tN1mZfTl8+Hl6fP55dnpiHEkQK7Sr2ey3pRWVUcTEH7rLCzMBTsJpL7u9LOgAGIDfl1RRiA2ADdVJrWTcVN0UsBqSCtB861aqXaVnzbBN3udhCQUwAmJ3GXplVEwTkihA71dsSO3agia6ngGAGAadhPKaVhTHHdmPG2Y2VnskAGDCnSkOOQU05Rtqq1oUMMAeU7Hp2bmpSynK8vLy+Pl/NzDDwMeWi7wSaOUfupYqK1taJL27QuUhevlZonC8gZIsvIuo8whc3hWtYXwJemB5YY5OzlcbsO6hjDAIyIBpVYYuIx826I+zHbEWLwwLqf4H4X9hnxSFOEYgOGOE44WRuuc9WGBjlyuh/omL+/Msach2mUbdIlC9+cDR5imHJU59LUIIoyw3QYp8N+GMdhMuYqTVRBpFOMwAwb6wZt1lZE05BCjICobk1ETCjwEMbDcbi/Hx8e8vE47sfdbtjt8zjmnHM0YAhsHh1iIgpW5Hxnl2XM+7tp9/b+3c8//2YpCBqAE/j88sdyucLrIc9dYYZEgEDcZ9DdlhBjTDFGog5bRvTdbtgd7v/hv/32D3/++f5udzhMeT/GId9SddTcwZFcTZtIaXUtl9P15fH0+PkLEpbWDKi1BrX13R4FIkTx2ttTJCNwRghEgTgyB0aGRq7sim2jcqYyh7a6rvKdWtZuCUUCbkwYYwgAAZ3c4FtdBwdQsKJtKdu6bmXb3ByRUuAhhiGFnLpmn255nX0p34shAAFGYgJ0VHM1RFM1BHMzUHc3QHGtLmrETAJgxOaubiZgqrqJmjWDZtDUxeGH0o7QmfrWM4r6Jr6LLzr4G4mJIHAPsUFEc2+qpUlTZSI1QuTrvFznVVQ5fAdIgb7A/ArK6+I12k3j+4e7nIbaRNXcDAEJyBG9O8dSijH2xE40ATAANEQBVHdA3wyw6vdDLUaM/S0wVI4x3HiikUPkREwAiG7iqqLiUoqVUIdWU0qRozbb1la2JlUR0QkQKHLEQBGAXMAImrvcxDmRQ+BAgKouriB6U0g6dC9iX3m6exdPpJxjiLv9btwtMV/my7Jc11ZaE1H54dT9obRfN/t0aq3UUupc3TAutZ3mTQ3GYYjDkIacveYp3t29baLjdHj+8ul6+rLNrePJvi5QO/c7x7CbxmlMKRKYlW1dlw0vVwVHplprCOyylnXWuoDWsmETUpA8hCF6ZC9VVCyGyMQiIiKICMg3RimBqM3rin1e8noG123loO6u0rowl9BFm8jmuhE4gKtKqVVpaaXJVrrFHkQyASeLWFkaXK8Pl/NQZiorQcC2Xwp9mF+esF7KXLQJoFWzVWi2uBhJ5wjcFAcOMNdy9U2CUWQE9baBFnTpN+vY5b2uag1UXdsPAnkAAy8ifXmj5hzCgBg4oIOptlZrKXUrrVVVcQc12bb15fTyxx+/f/j0YV6XWhZVDUgdt7KVrUpzgJiG3eF4uH9zeHh3fHh3vD/GEMDAuVLIw4Aul/1ujInNAIiG/f2NpxuoFbeiXg0BQgiM3YsCPZKtJ8l+/zasiCxlo6umoKroKNhJXbWuy3K9qJgBX+dSzvM6l8tSLy/z9bxI1ciBhzEg7HIeQhhRg5mDG2CjwEAx5jCNYUxAIe4mJXSC2urletVcp8DkEgMMERJ7jjgMIeUYU9jjNO5YxMzg4f747uF4txumyKWAujsRM3xPGnA30W0tl8vlaV5fSrsaBBQLHgx6iwXmTbRsrQSryRtDw8EpM1VwoWpaoZzVVCRoMhJnCCnkwYcUUoCcaZrSEH0ceRpDYAZM4zjtBro/xLd3+900pgi7nSPDmHC3CylAqlsxrcoYYsiRW2h1Kxs1Ccr0mjP97SWqpsbEOaUYIzEboBqYOSJwDhnHQ6TAdNzlabdLeRgGzMmJQoiTaicPoRkM15b3ddjtq+jP79/c3+1iiL7WMcF+ILPAhA+H/LDPD/t82KVpCEPEFCCwEwBxGENWiGaEGn3K05imHDOHCDykcb9/wDErd9aqp/QM8F1pd0MwAnLVtm1Si5umnPIu7qe8m3ImAJVAOA7x8Pbtu19/+cd//O2Xn98NQ8pDGqYhpvjVMYeEIYRuvqxrXS8rArWqtTY1qE3UYN3WZVlaE7P+JbceCQ1AjMgIAYGJYuAUA7NbWVq5bNtc5uvldN6WpbVmVn7wjSEQ0TgOfjzU+YBSQRqogL221dBr2s197qqEMKTE2OsNxZvRGsgdTLtq+qtf5Sbe6wPG18wHMTVTB+9bFANXsabWEMQMWzP3zi83hNcsAEAMzACMwAb6gxHmVZdwq8X9TiDg3BXANws2ghN34DmC4y2EBYEc2JEA+fYnQO5oZt3ojPSV99+tWLd/Wwp8mDICtMZ6U6WFzoihEB0JEPs0G0EP08AIhKhI4tBE1IS9xRC+P69aa7JVcnRxEXVz7PS5OAxpACJzdxNDx5567qJgwo7gxqrVaxEpTauCOYjVwDWFNCaOHANrCpKCNO1ZoEMehmEAIkMQvzkTezgIAvRtiPeVCgEF5LH//mhby26/f3k+vTyeruerXBXwh+f8h9I+b/L5pczzspXKHPJI89bivImCejiOkWPmxjFP7376te+XXaUt54o9FvCWjR24Z750lz+nGBKzaSulrFtVJ44ppEzLSojBV5NNSzWxqkhCwIoeI3tkKAZgFDkE7mp2dXMCCoGBAyM48SYS1OD1a+buWmv3A3abHDOGgFLV2koukRERmqnVKsCgRi7aitcSpY3gQ7QkC5SK1/nw/GV3eeZ1thaXZfyyyL+15a92fd7OizRB8s39rOlk48m43lwiPWOAY2gRK+kKDSIi6i2KSjdEJjd2I9Bm4j3tT+r35jcOzCk108s8z+tSakHElFIKkYnAXFuP4FuXZZ7nKyKt67qs8/l8+vzl06dPH2urCBYJgdFMt20TW5d1MfeQ0jgdxukwTPs4TJwyIjl6zohIOYpsu92UU+R1U3UPwy4Nk4ELSaBVfTMVQowcckgxJnOordVSS6l/JxRqay0m0IRjDz50EEBtKGWbL5fTqVUx4LXqZZPzIi9LPV+2eS5INMQYckqBduMwBB5QkpsjaOBIQUOCPPAwpd3oGCYkQ8hDILLWSkMzSpE9Eo4Rc/RpoMM+Hw7jYd7vgGPMXQd82E9v7w/HIQxgYCqm6Bh+jFJSF63zup3n9XmrV4ENPYqjelYQ6ueqFZF122bSMiUfB4xTYIq4uW9eN529nFVUZFANhDHQLkc03g84RWyZj/ucA+x2vBvDkOM08UPlxDBlvj+M4zhwsljdiVLgNEUGgMDCW1FHxwGTGGihbaUmoTFI9dDse9JsKWUroc/4uR+CQKbu6gjY+00aUgq0H8I4DimmcYzMMQ+7YVf7UK4nhB3mepzLm+u1ifzpp4e3D0c3R/Q3x1zKeJgCE75/GN/eD/fHvNul3C/p5IE9EGIIMQ4KLApkAYY45jjEEIACYA7DbnfHu51HJgQGiPF/fHdlNJCCbsCuZSvL3LbNTdOQjod8GPM+BWpVTFMI+xx/++39b//0l59/fnc47t0dmIAZkToJCxg5hjykQIEcGbgtDZEQOeRhPEBUMzMK0QGxlNaa9/gANVNlRkJgdCYPjIE5BSaConOdv6wvT9eXl9Pzeds2Vyey/Rv/9oj0PJJpYvR6PXhZZVusmjkqwiuesS/NDd0JPQcOKUWOmSMTEHhPwOrTe1cDMyS6uePcb1IYAO+LfdHO2QEEDBRi6OnvDUERFR2+IjKI3LHTIzshm4mckFVZlX/I2/y+tGMv7d3cfbud3FwEfczeZ2y32HvkwBzAiDkyR6KAyOA3m95Nn/i6Jv7uXwcp0D4HcmuRRMzMmVOMOQ5jyFnMxYyZADBH1rtjImQkAxTzUmuTArIhIX+HGii1qs4EBApbqSIGgJFjjnmIoyOqGdziHBV6cqurG6qCm2kDbapNvZqrtaaVqcSAiJ1l+BVnZBgQKcaUUnYEdVdp5iBqouKqxJg4MdMr/wQDx5RSTimn1Frb7aeQIgCoaSkF/8558f1flFKu81WadomXmm0iRe2Qcpwm4NgM1lqL2MvLExGWcgWXmMK0G4chdGwQIxKhODUnEZ3neb+baL9HdSkCVR1ca1uXdamXJrLPlMiXpbSmMagzKsgWoZU6DsOQJktefc1IxzGfAs+X65jzw/FIhI4+3R3T/gAxQ/sGTOnBM52iRJAQBL2iNbLaw77UtIrkoAPTMI4pHLCuvF6H5bwv8936OG7Pda02z3R+pOuJ6maa63Y8F//g5//wy1mXzdSRXQGq5eK7Alxu92zsWFzGmrFGc6tDImYgFSuzLC/MgRCd0cBaWeq6qqpJ+15Gd//w8E//8s/LMl/Ol2Vdr/PMiDnFWquKLPNCRNfr9enx6V//9X9cr1dEPF0uf/z+t+fHx8vpNF8vIi0GjkM2061IqbJs5brODsghhZQ5RABqTedl61vDKY+HKW+U55ymMQ85butmtboYOGIIGELfBt1uMIQhhDxkJEqaF1pFfuja3f1yOkuwFClFCgFBpc0FpQYXbVWXdVvrdW2npb3M7VL92nwTN8OYU4oUA+VIIQCwEUEg5mFARkiDx2wpUc552HNI944ceJj65YdS4CmFRB5AYuAYI0eGQHlKb97dqYI7rUtd15JC2DHQusiz2PWCyxyJAzFB/uqKqbUs27yWK5LGBBgI0RRb1YUKqkVwaHVWWUk3t6pEGgMPiITYvLcxRJQT+RCGRHmgFLkVHYLkBDlgjpZTYHR3CpzHYYrDEDVqq82leg6e1ibXqkaUgSfIjr40PC96Kubk6mVfW95UxUy9NliWFrM8+Ldz8eXlCm7kizdpHCAPHLbIOrIZQZd63woVeACIxGOedvugbk20BzMRMgJtTbfStrqpyn6MUw5mWsq03w3/8Nu7UhuC78e4G8I0hpz4NeqUkSKGgeIO016Bm+qGQmUYcsoxkBEjDTkf9sd0vPMUCYHcYvwWTavSltMX4kghtrKtl1MrM6MNQ9od9nf73TGn7XSqWx13u/vD7i9/+e3X334xt+fTadsKAB4Ph2maUuQQA0fmgIRMjlLk/Hz52799+Pzp6eXlMi+1NO1VyZxz3iNnbqVuy1Zr61ozIuhJ5AhI2HsJIq8uVtd1Pl9OLy/Pp7JVQowJ9/Atm4ADc4rgyi2EwDEwIgoA9r72NRWKACKSx8gAbq9q/FtGS+e+4m3ZTXCLnSD8bkkKhi5gRVvVHgnqdjNom7kX0WbuzDFQ7KYf8KK6tKqtq2ecCAMxBWKmaBx/LO398O22v05otD4neNXVduE+03ehzcx4ywFJ5CGmIaQcQiIKAII3Qgl+XSt8nQ10jEZgHHMg8CZUSitVCIzQGZ27NI26OZkCZbfYS7sDimmOoAJkiIT63RV+K3W1a6fStK21rapoYI4ccxiI2cAjxUChIjZGM3SwG6fQrOOxg98CK5mQFWQthcxQHfFmo8ObzVfEWm1OqG6qrUsQRVqfTzuAWWBmvjnfY8wp5hxSl/f5vrVSSi21lZZS+v6j+KG0t1rLthISM3fUSW1SRSEwpiSATWTbitnCDIy4zGeRjQPlIYMHRuiiBwBsBihgqtd5uS8VAdzcq4CoE7TWlmX5croua/npfn8YUykiTcjNFarVLdK6lt3OhzR6tItBJLzb756G9MU0MN3fHTiwuE+HQ9jtzQna19yIr1sBBAcIgSAxDuwteXMjRChNoGlOaTcOh+NxPw25XIcr7mgedaOtWmm61LqscH2G9QxSjXetLYuEE16fYV6kNHfAAAqupgqgwPJqCevKFISq3rJPaIcxhcjoanVpy4v1WNAQ3K1tayuru92Q26+vw+GQpvGPP/54fnretm1Zl0gMZn3nNF9nRDy9vHz6+NFEP/z+ByCs2/bl8fHxy+fr5VyWRUwwJxySu2+lXufldJ2bCgZkZubQIQutyQqurYLpLudxSARpHNI0jsOQ0RaRTWv10TAypdh17fQK0bupU0OMEaUpdUT219IOfj6fC7QYPEccB4bWyvmKrWZ2MFfR9bo9Pl8fT+XpWlejhglS4pRSoiFzYIgBKDgGp8Qpck4UU+Rxh8OggSHEEHIMOYWUUxzHGCIx9x+WAzq59p84DilN+e5+/6df39aqZZOnx/Pj01mrQC3lJNeCul21bpQihgh2//WN1Lo9Xp5bWRyFI9zI21ZLQ0eTxgDY6myygVZEwRAhA2QCJGRHBDMLjNMQaUppjOMYxkQ1CnsJ7CkQsxGze18aphAnop1bqjpvdU3KIHze5PmqAjoMsBtgYLyucrqUx0UUsG3aTO9LJVV3F/F1c9nku8oOLy8XlRp5CVScmIc8DMNx9LozBlmb1B4O4OTsLqJihCHmKQVKbkSRMQIyAA7WO6vmImDSA3yIKOf0/o2JirsHgkAUCb/iNQ3AMQDeTiujEExJMuQhpxQDgyAhpph24y7tDphz9/iG8O28Umnz0ycMkUKoZVuv57KcwTQwjXnc390fx1G2JnCNOR3v7t+8ebi7O3x+fHx8fLleFwCsRe+OOo455xRzRKTMogbbpXz5+PJv//rHx0+Pl+vW02lD4BxDjCmGGBMDkbT2vYnzxsUjAALkHonbY41EW6tl27ZSS00xBCP4rrRTCByjKwAhE/Xa02EX/R/mNxKLM1EOIRJ16VtHw5pbR9m8yqw7J8e+TuDVeiFxcWuu1bSaqEFXpdgre7KqOgBxSCFOMQYid8C6bdaseTMjV0RmdiJKiA6dpPvDk/7KAsBbXe+W2ddXt/7RLe4ebzAeDhxSyiNz3u/347ALMTOzmb0ihV6LOcBriN3NYh2YhsSEkITQTUWAnHokPDhhXyEgMwVGBAiIjOgAUTGgerCAjkQnxK8dVWttKZuru7jUZk1NNTIzcuKYYgKkSjESbwiNUZzMxV6D79GBelRoH1ozuYOW6qSKAkyG1PEEqmYKXCqAA6GBi4mqqknHBQIA1r4rAnKGr8ISIiDCCMHTMI27w76spRXJww9BKj8q5AHcoX+/VE2a1FqXbTtdLkSc8+gOHf55PT2DaSlbLVtT26quy8oI0xBC4A6/HackaiIios2sqFysNlN3m9drcbue5lrF9yFwSJHAuXWnNGNDmDeZVg2RDam6UeC7u/2bh7uXy3Uad3HaExGKAEQR8u/bRHDVhoSAhK6gjcByDMM0QgDEHskXkGLMQx6ncTeNOe5m2PslXyv7ubV1LdXFzQWgAqqROwEwRuRJ0rHltLioIUctLrV6E6vN5BaF5u79gpqIx5jucnwfh/1uh0xm0upq2oRQOYB7a6UrY/9u5FRrvc7X6/myzLPUBuatVSlluVzNbNtWDuHTh4/k8PmPDzlnRBSVeVmenp+361VqVRcLBGAOnQclTZqaMoWOAFRRqQ3cc4wUqGeUMfiQ035/2B0OedwhPrW6VZkTjmEXh7irZed1awqmrVVx3cwg5Rx7bOCPmCpwmJe16BpJhojeYgB3EwRXAxGbV3m6lk/n8nyt100pxXEYwjTGMU/7IQ9RW1FtHHPapd0uH3dxGvIw5jhOlAbl0BxqUW3WSkPRBD5gHFNMmULoavzQbUTOTdFDxP0+tqLbXMrl/FiX09Npva7rGC4js1UECTlQDLL701diTanly9Nnt+ba3JtD16a6qqqIcECHtm7eKoPnEKYh7aYcxhyRY3OrdZsFgCgOYdgNU0qZA1kpXsUdMGMU060SOkQKzVLVUAWWqsum0ixldWzP5+Xj43XZSs40+Lu7MVxOp5fHp8+nrRnWw+QBx7JlVXII2KGn37ord59PJ688jnUYlpR0P6Vff7rbD4f3d/Dp8/Xffv/0tFy3VYSJd/n5fP3weNIIe9kw9jxzRqCmVkVu0cKl3f6o4qp9C9bVJkTY3RChdzBfT2tiCjmN87jfxmnKQwbAGFMIAYkUXFzMFdAR1Dubzbuk6PYyadeXT4DkRLVs63KZzy9lvoyZpUlMw/7h7el0rfjYmvK8PT6fKKfT5bqt1dV7VvWyFTFdSwlMMfIlRRPfLu33v3359w9Pzy/X2rSnjDNBJQiBQwjEDEQcwjTtAkEdoplTCMDkiEiO5D0FJhCnmPfTJMc7U25Nb1pD+AbD7wXOrNevV5k5wFdW3te//Mru7JtX945IAuuwoFtJhVssYhdad0kdmLk1kWqi4N7HmUghhr48riKybeqG6Bww5kgAtbbS2mXdrqWVZgAeGwzmg1uMoRPRf6gd8DXJqycM9HrySiWzHneM3cT3SswkJI5p3O0f7h/evbm/v7/LwzAwB3Dp7B0CROxKXOTggW/hmcRMbBRDQAoxdMGhA1GIITIH6lEA3S/elcuBeryEG5obAEK8wa6/vfosEtAcblCY2xVFlURTpEAxRU7IgaloaF5Fm0hXLwCSAjqSMWMCTAGVrKCpWi1qRAYoDaRaK03V3FVUKBASGlj3xjEhhAgI/ck1cxHpJkW9MYYNHETUDDmkYdwdjjaM4/dv5MfS3u9dhMTcWdim2mq9nK/gsN9rCokIHaGuS6vbVrYm6gbrJs/nJRCADylFYh9DyjnZVsta122bl3Ut27kVbUZIsC7RrJXiZuCCoIQOCKW5osccKWXHoEaiIAYCSCkdjof7+7v7l3PMY8wDGLi4CWxF6cf7o2k1AkJ2N9BGbEMKKY1hDMzMMedxN4x7DMGJMHAg32nYJQ+4uV5KW1qVYqGAGaiDqXtzqEhoPJa4n5lfXMQogTm0Aq2J1M2q3OKlsHs3OFIYOLyd9u93x/1+R8xqUuvan4rGgZBMDMCp+xu/ey3X+fPHjy+Pj/P5KqWieW1VatUuBwVPOZ1eXlw14i3gFhDUbdm2tm2u0hukntLDgTkwMTrcDllyc2lSi7VGnnPkyBjQXSqxxxjzMOVxRCK1UuXabEp5F0LM46jLYFWsapOmUnpkOyK/SnN/uC+u20ZtYa8tOlrKgUmRHMV8LXaa6+dr+3Rt19Wq4i7HnIdhGob9sD9O45S3lUrB3WG6e3O8P+7u9uM05JwTxQQUGpAVqeW6XBbdtoggu4EepintY07U32yMQGTeJ+geA0Umi5TAhgRo9XI+ff74dInhZYyZLUaImTlyS/K1tDdp58sZQAEU0RBMVU1M1UxdSQhQWwOzgJhDGHIahxxyisjDwJpJaTWkEHIadtM0EoPIVtu2bJKHOEJUsKYlIDtl9VSElubnVUpRd5vEkPQ6b09Pl9PlEgO8nwLfDcv1Ml8up6e5KpDKbggCms3JnajPmn8YnG7LAoYAAlgchQnGMQWkzFiX8jep7Xpdrw2ZwOTj8zl9/PJsZTcPEKEXBgXYal3LttVaS6urtKXVpbWt28F7RccensQcetBUoN7LdnoBM4dhtzsc7+/v7x7u70Ymaq2aNwAFqy7dFKomrmig3/xU/Rk3aevZEBW8rMsyX5frtW1rLaM25ZDy/sjjzkIW2XxeP378sra2bq0U6bE9RGtprR97t2k6oitaxY8fnv726eU6rwbI6IxOoOxKhEwcUoo5BaaYMzOmHGutCojM0CnbroDWbV6BQk55GqdaQdXGHJh9g/WrqgZ/UIJ/mzz/3cHs/hr9/oqIwm6xFld3tM4I6xI5M7NuwlE19VfxnLsiKIIieod35UwcRETslUXv2heK6l5V1lbnUpaq1RBcC8LN8orEzH8nz+zokM7qoVuk42vVv0XQ3Eo7ImI/hG4w/nGY7t6++/X9uzf7KDEBcgC5iea6Ag/7ZJ29I4DQDE0JPBBwQARiDwHdHYGYAxK7Opj3vdJrybwN8t3QiAzAEzn8MGQEphA5GagaKjYEeDVoG5kF9UwUAImjowEjOIMGgCYg0AckaIQEZAlhYGjoiq7m0kyRxMkUVKw1kdb6R8PGHLhTkwGQXoNNbxAbBHNrzURNmqiKmSOgqok6IMc0THv/u/jT/7OunSiEQBgAaMg5MbdSFsQpD2kcd9NkqvP1pWzr+brW2mIItbXWBAO6u6o29ZgVwKW1y/WKn1BEiLCb0dwhxLzbH/gugoManM5Lq17Ei2Ic8t1w9/7h/W+//robd6WUWoVCHqY9k+/3+91+HygOSFVr29Z1ned5zkP+LgPdQTcKkMKA5k0kEe2nITOwWQgh5XG3P077Y3NfW93atq4zblctM8mqVmfTi/nZ4Cq4Vl9WnWsT0GH1Obs8V/l8uXz43ETyYRfHlALFKlpW3aqbxRiGMXEgZAAV2sphf3e/3++nKaVgrltZRKuZ5jikODDHyAHhJuj5+jo9P/+P//f/Z1mWZVlExVStSdegu0MMYYhpSDkSy7aJamAOIXKMESkg9v8cQxxzDikN48iBnVxFmOgwpl3ECNXreT4L1NNhN+zHwYkKoqGc5xcxCSmFHIChtnlen9vQEN3RKQQmFiQA6pb+2wbC8etu7Ou51NSgirWtsrlrjhEtuIGIXpfydFo/n8rn2asQEWdHNXETItvvx3fv37i9AfP7t3f3b+7GHHPowzedL8t1Lee5vJyW58fTfJ512xLC3X747dc3jL+i7zkzaWZH4OiAW7O5KBNYZDInomkcD3dHCk+XIudr/XTCHCFnipljCtN7+wY2ReLAtUlphRACYT9E+yW9j5iJOEaM4ExEyADUGbxMjBwZIxFjGGLeDbtD0zpf1+dLfTyXO6B7iiHxMFGO6XgYQ7SiOheZNxRxInQKyNGR1bA2N/WqJu6KAIwUmJFCyCHGAI2hSRNwg55I8d0LY7bIK5R1q+Wy1XqWWtumdcFPf7z88fnx6TSXBhBoQ1s/+xPV/WUa7wYnby7iJm7FpGpTFWkqRWUV2aRtolVdte9rQmD6ulmFHhnSh9Z9Mwwppd1+93B39/bh/n4c9yFsz+eLNsQIZJtLkUoiRGRgZvp9LQmBjnejA4j5xVtZ+l4X+tEOSOoQh3E6HufZt1b+4/fP/uFpWUqtOsQw5Jx3IxBe17WUevso3cc83h/uz5flw+PpsqzmkJimSClAIKc+GK8lbCmmkGII5EwhZXLiENjdai21bKaKyCpWSpvX7bLM13kR0VpDiJCm76v3rSf/ukT27xyX4LfpWqeodi8/M3OM5lZbq6VIB/120kf3U7t3wZr1XDcAIO7+YbV5a8XNGZmByXHe2nVer+smrkGjOpoDGpRa1ybSIVDE4IzArygPV/0+zRkQIPpttJmHkKeYhhhCZAqOAoBmruaGr3Bwih0wgTdvbHKKQBEInNwB1UFEVFrnXd5ol65uYHX1pWHD5L5/pbsBgmd0dwBFaoS31Sg6oOLtd4Dc94huKlrBhcHtx5zzlNIuTCKtYXFpJg2RYqBMlAC5icsmbg20QmugQqhG5uRGam5645MQEoCBKZC/psih93g3QmQXUkTpmwuCm1IQuyAdgUKIMaQcmdHMVLVJ01ZbcW0Ghohsaq2KNAXEmCKHH6r5j6BZv8HAQ9fwUYgpEkDZCpjbvYRA025yt1ZnB1hLraXC4GYdH/26A7kpP25QAju5qo05ReZlWZvqtD8czXLKxMEB1mq1SG3WnCgNQxwP0/G428cYLpdrqdo1X1pWIg4hROLEJKaty8OXGRkGzq/Pgm3LiUDGEBC9WSWMOfGQIjnEEFMa8jCkFFsppW7X+bxcT9fzUz6d7DK3uWxFFuGrhyvAWeh58y+rOegvm+Ni8lL14zz/22OV5m8bPezyYQiirmJSTGWMwyENFNiBN1VZNZlPecgxMwZtrcJVtPRvZLd8dnKl+Q8Oy+v5/OE//sP6IAvREVTk1Y+BzJxTHodhSGleFtk2DpGIx5SIaZ6Da+BA+5yHEGNOzsSRYmZXDczv3z28v9s97HiXhGwu6xJxCD4aMxMJtFIuRD7uhsP9XqzxLoRo7JWQ4phim7A0q6rlpqYFwFtK0Xdm1P5qota9wqRIUATAQQRrk8vcnuf2vNlVUI0SsSESeow4DOGwH9883EeOTLS720/7sVvha2nLdT6dTs8v58eX+fnlenq+bNfFW8sE8y5HrD//NI0jBUgEaB6AwIG3Ymt1Qm/iwR2bckzT4ch5XI3O89aq5IGHoUsM7Z/Epm+lHZBI3LZaGdGYuxPJgYDYERwo9wpr+jqIdDXj256RAQNicAiACcPgBpvgXG0pNigYcohpHON+t7u/24tcrsvLvMplQQcbEqqhObqTI3MYYgQMCULglNI0HnVQD4fjYYoY24LmDtLxQfZ9bgTicLiLQ6w+X7b68bE9v5y39VLXopufHq8fnk7LtSIGMzJo57N8pjKUYVgGQ61am2kDFVBFczcwt2ZaVYtqEak34ATezrBb9BnewsNeszLM3JSZhhyP+93D8fB2f3gz7qMBiQ3MTLC5VZVk+gpJ/WHRw0S7XTZAUW9lu/b7i3doJ9XW5nVxhGGampRS6+Vlvq71ci3SdD/k3TTkXW1unx9fLpdF5La5frg7/MOfoFQ5L+UyFzXLgS2HHDGxg4mbMjOHmnPOKeVEKXDHLbpDa41LKVsxUWB2BzUQ86ZapLWqDhAB4w/l5AZ0ec17ec2BfUU+9pO997xwuyIRcs8/JiMUcDEz0dZaba2X3EAxICq6OUCX9aRkZqter1WYeKDAeWRi2ERgLQLNjED0lasqIkVEoevfGJ06HPt2XfjRVIkAGRwIOBJO0XTajeOYMiOx4xDiEGP3+aUQUkhjzDHm47grO5h2+904DTFF5sgeegCmO4MnBGJMkXaJmSAYZGhUDMQBNBEmusVVMRFzcEB3IGU0VHHT266/w947t7fDXFWbgyG5cMDh2zuJIRCNrTEBmDRXQfMUU+KQEFG1x982UCETMvUbp8z9lu/siIRMLOSKZIzwSgsmAgZkIoTo2tRV4Wus9e1md+MMEWIIlFKMkc2stWamYlJLdQHGgEh9ad5v8BSI+P9aIe/QQ0yIOcSYCIOIllJqaw7QuhHh24fphEiMgYlS1GnMAVOOeRgoDTHGromMKYcQ4OYZsFJrKfX56QkI9oe73f64O96HmPVyBV/RiQjddJuvHz98oBCWVVpTJramT4/PXz4/nk/n4/4QY+BI4tLaVrd5GCK8ighM9cMffz0c7wJiCFzKHDg3HaMzUlCHrZS1VPvy+Hw5P52el7K0suLl2U9flg+n5XFGZ+Bo41BjeKbl0eaPRYAgFZuqWhNo1comrbSStEawGBmHKRlKKzoN4bAfkFMVXC7zvJayCShhCw641bWwpByGccjDEHMotrUm2Ddj340cXc1qBXByNwR1b9KaCDgwMyIFDjmmHOMGoKqElInv7+6b+7LMkWA35DwODOiiBn447H7+5e2Q85jy23c//fTzz4fj3TBNalCr1q1sy5oPh900ORHFcHef3/98F8M/lvWXlAcgWlp1t/yWynX5zPTFQDZrTfp51EN3ifjH5DcQUVUTIANYjZqgG5pCUxRgDCkkSmq1iZnEkI9307v39w8/vXl4uBuHDEYmfnk+vzw/reu6Lss2r+u8rMuybltpWqtG9DDliDkHHDOlgdWaqiaKYmFepLmZk5qLBDCdVaEWK5sqhLyjcedpvNpyKXWgOAZKSMHtL99BIMy8tSZNpDu7FUSaqLJBM0jEiRiZGbG0tqpvlXNjDu4AXAmbKVBzlNps3WgZFIxCTsM+TUIhqxJCGHI+7u/ePNydz/pyejpf1i/PkgcmDPNStNVta8zp3c/H43G8fzcNE0xV3sX92z8fOYyJaCpLevzgWxXHKrZuVUv9/gj+9bc/D4fpy+Xp9FE/X//93z6u5/OpbAuKl0u5LEWKBDR3diYEYnQFF1VDE3NxF3CjVxUymLNjAI6AA2IjFriNZvtUsR9dr4tjt77BVBU3txU32WSW5eV6/jTs9mk8xPE4pECwkVd3RQhI7j+mkfTq3oFJt81uv0wgMTjp0/lxbQuKo0OmbHFnkkRaTSWwTbtxvxvTOGxNEIq5IRKSIRrxjnlIQcY0aUMRYQJ3qE1qFZOmKjGGlECBFLAKBoIQiEPA4kEQwn7YmykAckzj7nBH6IixaaxVhmEMARB//zaQd6e+QX9dpaub2GsqG4IjOqEZqlvffJmIbUtfu7bWMfJNmzgAhtAP6ZByGlJt1URCzhRSNb9s28fn87wsb968vT/ev/3lT+Mw7s7n9PioHz+crpdqos0clACsJ5zc4s66y17VsDnHEDD9cIMngAMQE+QUWgr3Y85pOE4TAUprwS0zbU3c6d1x/3Ya3h36ADb/6V0bp/vpcH847PdTnCLuowxgCPowRr8byHHIw90xMkEbcqjb1OY4r1i2GHDIiQgcravyXu9Hnbbf2yLvwTpm/lVfDei3XzC6c4a9wTfQLCEzeiQHVI8YwHwMMaVMFFA7MQduYj4CAHUzAESMgdgDuAImZdSElrkzid3URB2B/Aagc8iA6Aq3hD3vwTN9SuMu4lghhEBEgRkjgaMrSFV3l9YAoINeTLTf9/zH5+M/lXa4IRRCiES8lbpum5uFGJrIVirhDG5NmrsTQc89Q6AUQowYAqchp2lPiKaNqAesMSJ2tk6/V87zzJGJwjDu8rQbd0cxr2boyDGY2TzPpW4cIvIQOBBQKe3L48vj08t1XlJMRaSIVGm1Famryjfjrpl9/PDHYT4T4JCH1jb33X4dCJA5gqOLrds2z8vTy9Pjy5OYmAsss54vp5d6PSMzxUw8JOd4onyi/AKKHlaFeHvg1LWZNlO5ZR8TEjMw3/TiKQNG1x5Oum1FVEGri+iG2xYK0YTjwISAPrfrXGYCAgPz72GBjvZ64zIT6027IvQAZkC6ZSvfLvaIRHQ43Dnj5fxUAu6GTIGbQ62tNR+mdH93uLu73+8Obx7evnnzZtrlmMkMasUXndeyBR7HkRRlM+MgKeOb93eMtM87MHw5v4i0aZfXKct1Wc/rhRdpBl3jY4bQ90P0/ZdKzURdDQ0IBalD6xRUTAyQKeU4IuAKUiRG2u+Gw37cTwMBbMtmAlJ0rfOyXZdlWea1bKWVaq2ZGTAFojymwDRETgFT9HEXeyYKUXQIW7NNTMzVwQzrqmVZ2nzWskzTnuPIecQ8VORr0xZImwVXVpDvnhZVq1vrmF3seEgODmjmpTYnd3IOzghYS2JvEkVCF1KERqjQw7dFHZrW2ihSHqfD0YvSmPpIDlLgKaf9MKwzSZN1nq+ngj5ansparLi0FmN48/bNu/f3d3eYuE77FjJOdz+ntENRfH6E81MBUkcxlda4/QCXvru/H4+HizbF9LL4h5f6fFrLtrC7LbXVBk0DGKJBX34QOJiJdAOvot9MGt/GyIhMEJwMSMCV3OA1m5rQHW5RLl1BDh21BRa1Vanb1ra1rMu2ndftOO7XyYzzaLKabNomlWCiKqLi3+HP3Kyumzmqg6oCYggMHohRrJ3n03m97NNux2OKGTOpNfVmntX9eJgOuzEOA5c65CVHN2QzN9eUp5gGwjbmwcQ1iLs4iJibautRV8gQnNRBrLqhW2AKUSlCgpT30itzj/zkEGMa81BSGhwkpIHZXPEbwcL9FhiCSDFyztQGdkeNZopmlAybsmpTVdP+mW5d4Ocuqs20qYi07lPp8nrOOQ5Dc3czTjnkYdvK2uy8lnnZ9neOMcVxGg8HCLGYP12uuBXZVMwQNHQGUxey9ce1J5o49jDRv8u4QYcoGh0mRQg05RACja4E4GQx8wS5SjSHY+Z78nuyY4D9cai7IQ1jGkJILbkOIoO3YA3K9R3JfqRINGY4DEYIgiDa7HrF+ULbkhLvfeTQoRJ9NU+vTrkesKIiJqrYsQTUA0/65EjNTRw0DPjD/AERkTkgEEHIaR+QxpSOh8OUshdpTdAEwRAVAN20G92IIlB0ImOEYAwaAqRE5gYiUrW5Ngd3YiBCSLE5SLOmYPgaQ4KASH2vLdigtkDcHQnEHLhrs8ykVTO7AcI6xo/4+6fj70t7l2v0qV+XDoi01mqMkYlqqc+PL1/0kRCGiO7WXRpuoGK1NQQyD9AjZpkYO6OPAbGHALqoOnSNTeAwJNoNdNyP0+HQ6qaqACGGoABLqVB0yMPd/RQC16LXrT6er+d5NYNlLb9/+rRt63VZ8jg2+0E6q6Yfvnz+/PL8+Py8H6f9MMq7d2PMdSy3gYn5uqyX6+V0vZznKxAEhmCVAozHg3MsVVYFbOamFmKcdlkI0xhSRO5crB60yMhkDpuou5elyap1w5IINwL0dWuXrc6lbbXVJhts4FKjqFtd64pLAKbIz9enl+0au+/9uzClLtkBN3OV25Ld3LzfR+0WUC2lwVbbWlszm9ziOA7j8P7tu/kc1ETAxX2t9bItYo0ZRal51lBXXPIypyid/5XYaUrDkJBw3bbn8+OHzx9+//SJiKZxmqZxHIadDaacM1ulFLq7GbknjLmhGwJwDBx+0G31waKou5i6IRn2UAypKs1MIofDlAJY8TZETsS21fPj8/PjixqAIpirVYXijggwDYmGoR89jq/YSUImiOwpwLgfY04cA3FADEjuCApYmqxrfX58efz0Zb08W13/9Kdf3v6UIHBKKYQAgCZatypMRGzfZ1uZ1a2qCKlNeTzuj8Qs7pd5vi5rH79XreRK2iwBmIA3N3J067ZQAg6cUwoxRuaU0n6cxjxO4w50S7St0tgVdLbGVq/eVq+L142FoiWq7qgRbJfC3S7d7/OUjF0yMRIE02CN0bxrANRuLg139h+e+bauyLRer8syl7I2bQboRF3BbB1qqErsAbh3QCpVpUKfO7I7O4oD9aQRePVqod/ysTp8DLsRyQXAAMl7lUAmDhxzSDG0bb2c1MWtN4Vuy7p61eB8nA6dIrBuGUBVVLpb6PVVaz1/+UgUgEMRpUDDNKBFDrSWwuAhJw9MKSEGRE6OAzgGBuK7/bjbjSElinx3t1M3URc1cxrGkDIJcciUjPvhJtLVYQGIyCyEyDEBBXUEgx41JqA53IzUTGDaal3m+TKfntfL+Xo6LdfrVtuyrSH4cffN06OmZmqAGFM+HB0pH45ai4t0Vbmqit3iWKu01tq6bXFdWmuiuq1zU1W05s0BbyFuSCGlEJNvq5gSUxpyMogpxhiRqLT2cj4PXz6vpYB7aVX6mWLmZgIQIqcUEE2lgaubqb9GO8KrGv17XK7ZclkiKM6ABEbgzBhjChgJJof8Gl6Xm+TrBjbrPAakSCGUgZfMFAhdrBRoIVr0+vM2k7cBKVvI24oOTWVZ5+t81m0JJplojBjiKwD1dg/B/jaEVLTn12mn3oYA3QXX8SNq4OZkf/dGupWZkXIacwrTcX+42x/e7aeRaL1crperbluptccImwCARQoA0YHlpmIHcISQwzS6qdWSUAQVRcXMm6oIADJHJ0BQYuyqfyQ0px6GaqbSWkVyA0Sy26ALVEyldGWidoobdinDf1Ha4fWramaqpt6aiEiMwcG2datFtrIFxjd3O3QPDN2z0S+PTN6nNq7axxW3LHOkwGwG2tvMEGKMMUZCN611XZizqYRAOeaUcooAJnUrkWhMnHJ2d0Csas3M3NeyyZO2Vlrdamui8v2szh3WUnVZ5vPluNv9/PD2MI7LvLTaSq1q5ubruszzddnKJi1Epr5VGBKHSMNezvMyb/9/vv60O5IcW9PF9gTAzNxJxpBDVZ2jHu6V/v9/0QdpLd3b6lafoaoyMyJIuruZAdiDPsCcwSjpNFdWVq7ISAbdHYYN7P2+z2vdQh2QeJqzMuUiORMRvJH2IBEnIFIL6wpr65v2Fl4ibY4YrfbaVd0cAYh6OJgquwV29bp34R2Ertf9VveBIXjfVAnAQPYAc1MLVT+EKwFxzL94yIia+aaKEQ8QVMrp/PDh4cnb/vX1ZbVuxGtt67qx4LbuZak0dU9tw5q3PfP6cUlP00SRikzM5BC37fbt+dvzy7eXl2+cJMC30zkxMTlHcCh6R1d0G08xxRFzOHAkhAT/sLDGinRXd2RH8pEeDeAikJhBMmOQtyLMgFa197523briUKixEXvKOeWcJYmkMZjScMdgESAIVw0D9+7+FqtJKCxAgQDQzW5b+/J8/evfv15evljbyvn8+PkzEeWUMg/QSIB5eDj9MNwND+3mauh+ytMvHz8t5zMw/9tf/9b2v4MNf7BZKIQRoFAkBmQXcCJHAkBDCIZgD+hKSUpKUgqFWEPvrUHnsOhr39D7zmETxTnBWfDEVNAIYhF0oVOGiT25kXl2DHese4zA5rZH7yNUSpgWgZx++Di+/fFFrtfL5Xm/3XqrbjZ4Z2+cETcPDegeGqiOaqauvceQASUEgRG4BYNlcegFR34AggMqoAMFkCMoguEBFqXAFEw8p+nhfNpJ9suqoSMgFT1U29XaQtPtfLveLq+XbwSw5atpmFlrb+wKMLPrtbJkKQVYyjIVT2AKiNteEyHk7EjBzAkp0ilcGCyAWB5O0zxlzmnKZO1hyVSbdVV3PS3lPNEOUAoOMUxrEOZEACjGYQ7MY4J8uLePUZRFHoep0NDatosb1+3S99XqZq2C1tDu0d+i2MbXbd2eXy/D4+MiNM9EDJJCNUwpnNzQjNzDXcxUlXImkTEcAozWdthBQ8GRQgiQmIB47A/mEQgjoGsqZZ6mddt6by+vr0h8W7eU0r7ve21DfA3hjgGZkzCCNCJVxxEkM9K/APweW/b+6bhtla0bGtMIPkFLCIk40chRHXk3bCgNY3/VnLJISsKSmRMhYYTrbqCeQSTOaDPFBJSCuCFEqAfta9suXZWPnCdMQnAn4IzS7gHmABiBjGqgBDw0riJCI8PKOMAiMJjye02juysYkjBlSo95/nh6+unDx48fHpeJgPOz8cv+ekXYojfTTQ2YIkthZ1NANAB1RGSgNFE+uym5sPfk6tHdm6paKAiyiGMEADEQHaWdAhHD75N7M0c0QA+LiCBEgxgdrKEhMHcIRHT9X+S14xAhRLjZtu3q0XsfzYFeG8RKLABAIkPJWbIwEVFyi7e4brPW6w2jiCQIIMCc0jJP7t5aAyI1m6ZJUtpqa7/98eW5pnzKWU7n03Jenh6nkkVb/bqvDPG4lNPDuZT87ePTx6eH2+u317q31rqqaw8169aa2rtXRUhPDx+0teht5jTnnEVGwMPL9dpbc7Naa92rxkiYyyWliVKBSR0iq3Tw6vtWTTUnBEk8CeUZJQPyyB5AESZIKQsnAFL1vrdeuzvWZuu6J0Z/s9TPUzmfCKfeQtmcXFic0tbBu4emgudZ5syy4ReHo33qEc1GnL31IYuPOxICMU/TfFryNEGAAu4egF4DgoVTzimbxh/PL9feeJosHJCnMj89PM1lxoCByNTesa+7WtoopcdcSHsPosvl8vLyYl2FCdzbvr5++2K3K7lxuBHsl0tdb9oauGE4BIZbqEIgu9N73dYQ+XpYdzOnFBhx0LOIcqLTnJCzh2BE6MQsbuEKwMScMgsP1gQ5C7AIsUTg3nrv2npvakFUlhkZW9201dC2bQ9Pjw+PHz0gSDAVEUQCd4w2ZkJqtWlrtfZm1hljEpqSTCkxemIcYyn6se2IgKAOaucy/dMvf/r1T38qyyIg12/Xdd28d2YsxBP4LHgq8rAkLIhAYujk1bpb+G0NQdAatdhUAqQbhhkYoQmHWe3r5ebNljT9/Bgz6WlZHuYpsyNaTqHIyarfXp0Dw7g5VdP9VQGZAK83rw3MCGJJfH7Icprx3WfxP/7P/w4iTVzXPbqCWnR11TEzC49wCAVs4at6Vko9THXdHAITYSGaEEZrlgGEMDMlGogQMEeN6IGKDJgBBRIGmXl3q94iMUHkJZ3LmRQYBAxDkQYprLtX61L31/Xy9fnrb79tr1dEVjVTv91u3x9zprIUTjmVSUrhnLS2dtu23tumzkYlWtcmjQOK+HIWCsZAQiqZc0Iq5BN/Lp/3z4/71npvbkoESehF/ZZALNyd1BzVEYnIAnTcVn0gu5kgYkDfwhGdolu7bq8UseYi7H3JcHqcJrTQmgWkZBIA/I6U/ttvv1v4PE2lJIJw7dv11rYteg83BHfT2hpApJSYGRCZIidOiRCRoPf9ut3AtLs5ALG4ILrbkc0O0Lv21hLRecqPp7mua1O9vr6q6uv1mlLu3V6ut6027UYYwccbRfd5so8IdARA5JFlovZenukAV3M03zHYncGLhWqgIndEIsG7Y48gEL2R1hFSziQSLIhMOJj/zkqUgsiA3Hj0IobD26NW190diHMgHZm3oQGIMFRpdEShEyIwpSQjc56YUkLhQcvjQLQAdcEMP2JRHAKCIjLG1HzebdpsUj7DnMRzhlPCK9Ma15v1qzox+ZyXDKjRk7dq6OyIKCIBZMAK7OhAfvD13dyMCDEE0RBoLIbD3wgBiCKCiDzYW0hDXcdEKcnYTdVhRFyaHrf33v/jUFcYJ1Bz7dpdm0bv6hHaFQJUnSWlnBETsWRBgpMHEpXaNJWcGB5OSQ5DHhILkgJgzvl0OgNgU5Xauuq0zJILAHSP9XVl6p8+P2WmqXDJiGFt3y+vl77v1+fnjMhBM9N5nqaSX8NNewC6KZhu2/b8/PLw+OH7/ov4eDp32X3DSVgoCB3Bh86w1mbae+va1RFBiIiEhSkIGGC4GFgBtlpb3ZcpUco+juOAMfIFCUlEgPM8SSkK7oFdwwwQ0c37XkEQwQVhypxzwpSq4W7e1V2jBzQL06YWhsRpTniaJO/47e2FWEQzV7N7J/69XRwiwtz23sOjBnTACKgWTbV1VbXa+uv19lp3aUqJhXFETQ6l0ZzS08MpE7DX2bcMlgWnKU9T4TLN0/JwOv/6uS+lmDsTLXkqzBLObuB9B6CAkRTpMsaqNI6Goyf7/qt1b+Ns4mPMGu4GGEyQcz4/nBF5b77vHQkdoJujGgR2hI5uAQ5BBOxAhkDRu7eutY3S7kAyG5FQrb3t1faNkV9v7VPvpzAg4MTk4dV6r7Vtqg3ChiiVCIgiCy4lnefyeD5NGeciI4ShpO8PiLAs87m6mSkDCVJhOXGekBMQOURARp6ECliiwIjD7uPRW9huWndz9kuycN+zbn2bLbhoJACiKL2Lm+07uEbXBPAwTxOj55STJCZH8CQOAdZgiw5EElCr9I7NNMIQgfYgYwv2UPaYPORHfc1vf/stENLj1HWL1rFrdI2u4A5qYe/E/U2j9dgZrMPWASJ0zN4RwAIMGFEwLIUzMAESmEdzqI59xFaiUCYkMnAz6B49nBSqsxE5oRM4xiCFE4GGd+jV69aul/Xb15eLrF291tZa9xXeLlhEVOaUcsnzlKcp5VyZUM0ATBWRwaH3XmnPQpL4w1xOJXMMbjoyE6WELEBsGnVv2vrogO77bhVOCUHJ3DFRJHIgYjGHI7sDYrB4D037EWHiYa2vl827tss05Yc5zUlSyWK6XlgwLaeZBL5tz28fyZc/fm/19nBezsuUGN1sfb3s6+Zdww0xwq31RoyllLHj+ziUEwkLYiShQayI8HB1I6MRmD4mLFhbo3XNOTPAoDa2W6+tq4VUTbmo+7a31s0thOCdFp8QcQRGW4APJA65ug/eyvuKuAcEYCVJCAKhodo7mBFDEBTEICQipyAaRoaRF02oTDy2JsTwRJGZM0EGFXCKAMJAQQB0A68QCsGBGMiB7EBHhO1BuDlOmQEYBITOEgQMRMgCONh9CIFIgegIP2Q3u4MFArGBhAkZ7R1qh+pcMJuAZ4rCURMkDsZAZ4qpnCZ0tU0M2GIMWIjwGIaPi8w7pd/gdBMCgRk4hPsdTjDGW8zMxJwS35MO3A2AADiC3dndVcNGdVAbc5v/sLQPX7ua1daaRlMfZ4nupurMJskAwWOSPJ3PM1MIC6cpgMyUIDKHWWt98xGHtVsQcirz+UlSCaS99do6gwnjdH5KZdHWBfGnzw8fPzxMuexb/fr15fff/vjb3/6WEPpt/ctPP318eNDrlSGEOY6AdDQzV/32/HzbttP5/J//y3+9l3Y4ZTFMBioYvd9au4U3ISkpoYcd8Z/Y3rQ5iOrqvW2tX2tb93Wv+3W97dsaliXlbYdkob27kB9xSZyE8/mBp2K9BrWgxIIyJm3aPIgZElMWIoqt1XXtL89tdzfCpXhiXW+tq53O5cPT8jAXofmHRRahYeOv+LGym9v1djXw67oh8trVKAF4a/r8fCH168vz5XbtXeteb7XnKZ9Pp970el1F8lTmP304/df//Je5NLbH2F+h1Tx/Lg+/np9+StPy+Dj9+dePL8/P63oLB0IuOQsitNr3db29JpAvv68vS9tXYOqHmT0IPAL8fdxCBFxuVbdOflAnHNBUAYMEhdMyndxj27fe+1ZrYdpNFdyar9437d4V1PBAmAkQ7623fpzQHZAYK3RJ4o6mYpq2hpe1Xrf9wXpCQwr3tu+XdX3d1hezPQksU1ry/DCXU0nLlJalPH08h9CnD6dPT+d5zjkLnt+8bzBN8y+//PnyLd0At63927/82/pynVL6+7//db/eXI1JpiynBNl6WL9eNyPDQuAQz47PBhcHZwCMZjbNNrPWU2dRTITEOI0D/Whd+1BNqimYO/fOwxurw9/dgRBvQBRoRubhoQ6ORGSbYASGWrO1olePMv9fvz/mvVZzD/Fum+97tAa9Q9Nwj24wMuAgHMIQCYxQkSNN7BBOEexH5z3GToTQPWDskA4a0SJW9eoxTmUJE2cgBHPobuoNYMvrbb7ubbduYeAGFqh0BJmqc1Pcqr+uPaJvW71cb+u6/jR/XNKB3CLEKacy5XmZUs5MRIljSpJlQQomYHRtu9u8FF5Ojx/Pn5/OjIQw7OnDSTwC5sF1CjVwu91uX7/2csOZKUTcUaZIhB4QwKPIDRnowL2NBzaAx6G1bztot7qnXWCZznAmXJhBAhIgMT9NMyV83r/rmW8vf9iKsc42FyF01W3d6t5Mww9y6ZBpYbfdI8ayGIxxZsbwAM2ZlimbI+DIIHNV66oe6IDbXlvTqZTBM+PB/D0O4gfO9h4MdzgYIgaXjRBGpCGEx5jXakQ37+Y/lnYMROcEZYJhiar7vl5VW6AphzImAqZxJ4hFkIkBMTElJiZKTIKMYJljEpoTLCKJhj4fkQQg3EHkINUiUqAYZCQJhCCKtw8UyJEcMIAwIgU6YACgwTizhgeMN+4+IPn+QhzNKQiBIMKIeu9779teZ2ba1n7dfW3QnYCSpCmlnjNM+TSB9RTYnMQMPADUre9XC4wABB+ht2NeycRCQkQWNrQxx8Epjpo/QnNKHnxGDHczVHRwFCZPYmbvSoSNTKD/sLSP6q7mANrV1eLosUcghIUHAluyey5QhLPQclpSKUREEeS6b6/Xa1XHHuwAQzknwrkU5IzJKTWta7hRmsvpQ5k0YZSpSEop59b6uvWvr+uXlyu0Pfa9Pb/Cn37dtNcjEnHw2d3d1bzpdlvXdf0ObkQAHlZSb2qqvi+nSW2f5oenp0dXsyGsr33r+251WqYyFTaEMOw9XDGcKYhhUAgZnGIosQ4aCxKSCOac5xOWhG6ADEiAzEwI4ao2NhAKImytPr+8vLz41y+1A4aktWRGul53NfvkjznlfbFC+r6TPU7g5mbjtnt8Pocafa+7um+1s2QAppTDtNX++29/tPXm680ASply6/u2I5LPoWr7XnvvDP44pb98ejjPwZH6tmhrUp7y6cN8fsrTMhU4T/SY8nq7ATCxlGliBKvbfnt9fZHrZZ/npZRbKQ4ggHGPao9wpx/wZ3FZq61tSpKJ0X3MihBBWNzBLNS81V5r29u+Ca6tMKG53bStvXlT6I489Fc5iLbam5qkxJLH82ID84EMnEgCSJraXlurFVN1JG/N2wZ95+iTAC3lnEDQP5zmzJiEppIeHpc0l3/6009/+dPnx/MyTfm/f6Xb3ThWyvTzz78uJd/mid3q3r7WL+xRt23OGSUAsCTOYmzo5utaGzjNDAbx2vni006ZkDXAwJRqzReYq5yUijBnITqeqoFnCYdw9kDvhIojLSIcw8FHVGYPQsADIo7mEGqBllI0gjUCuWmvHfP+40Nurn1fda3XervpttlWvXf0gGaoDg4Iw4vuju7kxJgke7iBKajdJZMj5dvH/dwAkMAimnvzqBYWEBCdkhgxWbipKriHr5fba07dujV1i5FGbQ6ITImBuGm8XG5/+/sfZrGu+/V629bt9JflrbQjYcoyT/nhNOeUmGhJcpqyAQ695N77kCKFp0FmRUZgOKaOEDA2kkEcZSBCDE6W0pTzVEopHuhuLJSSmEM4+DB5HikRQwI/gLBj6mRh5uHdDIx6eEvSkCOl1tTNjzL9AyIFtN5q6ztWtowRprZutVZ1RwdERhZKWZgw3Ltq3ffWuvbuEQhQhIsQIeaUzMCB1MHMaqtO1M0CoNbutveujGzmY+gMSKlMKWdiQbVEZMxvyDZCFhYAYB45bH2QqUYulPug1v2wqob62zhhmlwkHA2quHa0BkAIDshDXznc9gFv1e57IByQkCfGzFgYE407DSLxkO8hjZ1WgFJwNp6AZHgtj5ZqYAB54H2AiUOyOKB8oB52sI2OQPQfh242slxRA7vxBs574rXIepvC4Xprl2u93Oq219YbmGVOc5alLDk6bLsheoyoBHC36H0cOHBIfo9cgO94ArKDxzDaDhiHRJ9GPG1KKQkihptigHOIAzgguFlvrKg4VrLZ/0ohP5bpOJeNYnKk9gwotFNEuLuptb2+Wt/X52VZ/vSXxCyQCIcQutb98qKYQqC1tm217bvXzZnvO5KrdtNO205Sc6IguKwbYDw9Pp4flnLa03RlmbZ1/f3L1xT2NJdd9esff1xeXq3rodbzMB+DwXi/yiJ8r5dtW19vN2stIeS5/NJvD58+fnr4KQVDs65We7/ut7WuJCI5k/WoheQ1oltPrhPGYz1NMyMhhRCVU06ZjiVITIKSc54hMeEKEa5BPu47x3toDijhSq+vl45/fflaX77uXOa0LM4ZAbd9D4g+Y53l5Zl029js+1gU3FyPedlAVL3D1ZlZRHMkQc5l4sSx17b3f/2Xf708LJ8ep+l8/oWFpDT7ww8IZbxxPxgig0/ETEvh4h4gE6dZiNED1aOarbVfN+CcCuFEIgyunLPkzDlTSpwyF+NgAHvz5UI45PfQcriuu21DABXNe0CoBxOL+7a3L19f3P261a3utbe14nUTIY7w5mqmCCFCKSXJhXMOJmBiM0lZUmHOIkmSIFOEgRMmnmdhIu9Wb5uHdNDYNfd+JsCSzoGWJwHIgo/nmdzQnYVOy3QS/ud/+vW//ue/fHg8z3P52//9b7e7dKuU/PN0+unjk7c/19ulXi/9ctW2P54fzh8/PW913Xchw2gQaBq1eWdlJnSIark7OU85z6cFz49Nnhp9bP5p03NIgUQ5A1EggB+c6IHCGGf+iCNjAMAdDwPaQTYZ/4LIu3ndm1lkz8mEnFDDzaL98MznLO79cru+vH65Pb/sl1vfdzMjR+yOCugHYQYRg9AIOIswu3toB/Owflw6h/nBLdSJg5DBALqHjtsPtDDtOykhQQBYGFAgIKyHl6rVPsISBw095ZwpC0m3/vvvv2/X3dVbs97V1PpPf363WQETl5IflmmZppyEiIHoGEVdbq+vt83UHMhRu7+8XKzvwAhEjO/iyWAgFI9YlQjA+ZQfoOxmtLkpm4jbQWqFI7zcjqgIOLZhxIjovVlvg9gVDtr1tm7uwZJ717UqRuTrSukHCzIijC7iYLE21bXb3sORkUToqO0sQkyOHZRDqyL11nrvXTyKgAcPx2mQq3dVjY26BYI51NZbbdpDWHp3JC5lmohP57NI0hH6VRIjdg8IF4wkqZSJte9Smfq4uw+t5VHT/8H9BsesDdwAgkVcxAkZIBMQoREGoRNlgkJQgHKAHOKhN2gPjhKH4MdongD9sGQHoAdbiEYyEpLiMqlMRuxmrsP8NKYlA5DpB2oDxjkUAvyuFx2H6JFCD+9Vjaba9/34JTSvdolgxClP2v3ldX1+ub1ebvte3S0xnaZ8zrKUwgo1INS093AnQgrncaiAsPHThXtYhAEoBlAQhmEcM16kQYIA5gORLywiCSIcgNCZBRLQUH142DimdHN4f/37/1faBwIpAv0eQjCe3EEuPorCKDHuventcoMAU4PB38HA0JFMwwjBiABu0Wpbb7e9m+NqQBZhbXP3uq3EGbIAU4AT4FY1A+uRRD8o+7K7f729rnt9eX7dbtsAHIaPT/FAVbxfZgHRtrVu675vocqEZnvHnWf48OlxhgR7b91387KXqU3ARCIZgqyXBOhtHGXmSUx7ih7mpRDm9LAwJSnnUp6WonuUzEtyBsgAGXBCUIJMIwNg3BIoAWRuYX693i63/bIVb4lUUTAIrYsw6+51Xa9gNT3Rd7LpUDQODjzc2ZxvYWsiwilRypJySoUphVOoXi4XQv/08XQ+L6fzI1K61d6s5ZymUs6nZVnmaUrMENa7hpPD0JigwJHM5a5uTXtVbUbJIcO7BiYDC8lQvmbJbsExzgxH/pEh/7iu8OjeHlNaJBZKklJKjnTdau99q21vzY7UmiMrQ5CcOCFmpJSSJCYhJ4LMMsJFGYicwSAiFNwNwzkCDX1v7WW9EkuuHXLX8Npp29K6YfcwSkQ5MNb91natLYmcFpKcnx7PTw/nh/O5lEzv8MUppeX8tJQyl7S+vrz+8fvr739sL6+Pnz9NT09/f3n98vxS+j61G3iL6GOjI0QaGMnAoULg+6EcIFtM3Se3AorUXQgIwz3UwQI9aBh1YxzmgRAxwGC4EkbXEI/eEhFE1+ibQiPPGJxGJ8MAfoAcQpnEgm1t27rWvfatWxvgCxrK9u/LDe8QcCSSBGZkjk7Dz+b3x21cjYAwCMAgNI7LwfGH+5HZPVBdcAxmzQ0RLSwOeTxEBDOVnDjITF9eni/fXqyF6/HytfW3V6Fq63oxDybqrU85pyTE5GpaW79c6stta25Ap5TH0vXRyuUDGB/HfjcQz6MhgggEkiilIfEKcEIWP5g546JsdtAhj82HkIgiwu+Eh7FDmkFrFRDZondrqhRQe+cfCfFMmJg5Jc4FHJg8QYkMgIlYknBJMnqazKxmZW6ttdb7vu37tiW0zOEYYT003OlIhWkd1FNJfM8bVdMIMDjMoiIyz1NKue67q3WRCMBDIK+EJECIlJASjiUWDngEZAD9gweGEOfMLbDjcVoZkHghTIxEpChAxEiFbWKcCQsh4bHW7rX9OOHFd3s6Dur88XsQA8e9lwJw3DMBws3inoLj4RT+VtodwCHudob4fh4eTYBA/TGZFsPRFaMdaiIUV+sKrTuS3tbterlcXl7rthMETSLpJJHQqveqvfZetTeIQBFGyjJEpWGjeR46MlMwFMLDEHxgmQAAGZmImO/heG8I/vHsEzMFCIwTirCUnGOYWTzC4h98Sf9ofoN7AAHcL8LjIIXH2w73KCdhPOIdmVPiJCkTOrCV0/RojxqiOJWSRbj1/u3ltfulqSONbV1ySr2uCGHAiSVNhTn98XxlXl9fnrft2nuTUj5+fJoK/rFfX19eX9e1NX0rIQPgNB4teP+qAvS2e2tsIUQPheeCkRqWmE959hSq4KRAucyRGBPLlB5zntG/iKFtSJCEIwy9UX3xtm2GUOLhETWnVzxfs24nUER4Qg+FHtRQlMEIEyMwAEEizCSZkwhHgR4MJtQz1ORDJIpT4lJooS666+7e5XEhuL+UN2AvHLMAHDTfsQM+nE7zciIpQNmoADDO4r217YaE07w8PH0SEZB8a33bb5Lpw9PDn3796cPHp4eHMyfetena3RsiMadSYgoWNkIyta7eHTSwsBDLETx1t+QFCbKwJC7OQBGGoyHkFq74riIiwrKUjiolUWJg4CQpTyXnSbL3tl1vt73d9r2ZMXNOaU65iBBSd+thKbDg0NgMexsCOFOYd6tH7ywIA9xMKbwQVSst2U01Xjck0ZBwCI+67/u2mgE4K9FOYOKdQdNS0klSSE4CqLWvtNWm7zEDwvJwOv/y0+dff/q8r9fnP37+4+mv12/ffv3Lnz/8/PP//P2Pf/v773C7wOtzR+9XU66QIxeRQMwurORhrfdtD95qmlR6sEeMoAhsm4+4sMEHcaBAPm4ux8kKEQgcwEgwBCFnTIlyoZSYCLGq7Ki0IjIc1/x7aNgPpZ0tGAndwy3CAtRBHfDeu6RB/wAMwOFkUxgaN+yEDbGiv2Ff74jpGJutxeg6hOOd2DXgX/fc7iHgcTe1e7onHkBai2HcQkdXb9vWb82ahYJwEsnvjTDbXv/Hf//7ssy/PX45LdM0FcIAdwEg9+vrfrlsWxCW+bQskqeHD08fPy6cmBhj7Pf3u4t7HMb57ta9N3c1VzXres+kOS6OcUQb+ggcGQ1qP84Iburex/t+2Ied3RWM/MgFg3F8eH+/EuGUME9LOT0AigMtjhGEJEzIGIkxCaUkIgmIAoZA0/d9X683a1v07fby5WLdwEIhMIZ0GiMYOItYYjBCxDfgmapBYBCQDAUbHV9DQKBA5tiVzJNDAkxEPcKAGCARCB1t8fcHlI8P06rx6gyEfoBfUIBSIqfcsRgSIzxxX6QviDMhMdqRijWGeeOKe/BHAo+SPN7871siArqDNqurjuI/xCGHisLHoo+A+3cYTk2wAA0cd92BWBFKCmPWev84KCYOgsBwx0DhUpaUz8CTBfVurdW+b7ZvAA4gWNw3219b9FpvL22/9t7oLeaauZp6b+oa1sE7ghKqoB581lGZkYhkhAkxMxGOxpCpjds8DOkDMQUYupsjYEoZHNGZgAk4pe9Jwf9Y2gmJmI+T+jBQ3++ICCNtiIZVHQCQuEyL5MnMa20WiOgQ3ZwwzWDoBoiYkwDE1va96lY7ACThp8enZZoSIbru+1YBz/QEcXJzCEC3THCeC9P88dNHivbty7fXba29qXaAYMQRa23uRzvnh/NJoIcEzIiJcRZOCKAGZoQozCGCXW34eYfMtKTldH5KrOvz9lJab+DKiOwGqA77KQSmVCarMz+mc10Yn3IPl2lSbcI6Z9JlBg9EDgO3SFnKNHi4hI18BzxbDioll5IhAINKlpwFoPf91SoH0s/TJ/heF8cKPy4HY/rCRAiQc/786ePT0wek0p0vuzeLlIqxmFYiGknYKed5mR8fz7lwyrQs85g6EUGAN9NmWuuOiJLSA2CSDGCEIkIp52megXCapjJNKTESEBIgB7A5dQXVYZ9GQEYEdiRnD+J3t3ZE/Pnnj1YnIkIZOZc5lymnXJD327buTZE6ADDlLEUSO2Az9z6eVQFKCKyAGIZICAzAEWpmFmoHj9PD3TtFBHNtfXdfb9VZIkh9zPDQtUWt4RghBuAYO+kuQY/ADzMxsUO9bd++PHPKwNLeXRNLyb/89PmXn3/+0y8/9/3Dw3w6lfn26fnT58+PH59wXqaHx/bysn/54xr15ttmZlAjLAIRHNy9g/ZWX2+uXEvqywPMHWnYT0JNx7bl47ILZDAMu/xuCSB4oAcHZIbTSYhpkpKnhMQanUWJM4AgjCgTGI3v918vL1+vl9fn56/X68XMJKUl0TFJdmALdMQAKIgL48JYEiFDh2jhe0AD7jw69iNIfARcxIDWDQW5AcQhmIK3AwAMiN09o8xG0xUO+ue4P/SA5AfYdHQy3dyCJRH/YEXs3X7/8pxfby8vlzKlnNKQxZxSXjjta1+33jgJFQeUVObz+enTY8pMhOMeF8cTFu6uptZNm+63etMNhr3oYGHF274SAWFhar0f3rKB9CGmAd4rSZhw3OGJSYR40DpHUufwtfgPZ62uRmGpKWeTJJLSJElYCMfc2JgiJUpCLEfjHgCBZErTnGm9wvV1N8AW0Bz6aIOMB4diopgEKFMKMWAHUg8a/i6zaC2AqGsyz+ZgLoFgTh5TV1l3CJ+7dQdFEoQKQERJOBMfvv7vpZ1+flxea9ctmhv2itYYnTCCyCWbLDrc2WRCIOTEBCNcfex1ECM+9bhZwvdfjXfG82FMcHQGY+vYtzvx8GiYBEYAOaAFju1bIUZRbwbNorl3CzeAgMQO73lnAIkZcxLMQiVoxnwqDw/T6QE5qftAQBAAgbMr9m57r7ReLEXXfb1Z7xBAzCySc5GcodWuHSPGHI0Bxvk8wNy6uzk6oATRMBwKCyKYuluY2gjZw+NUQwBHMDwgiiSEAWUZgUH/cfIbC+eUx3FzVHeCY6YyJh80+gWIakaSptNZUlr32v2FmY8kiDAMVvW9NnfPWYaTxryrNlXtxI8Pj9M0zfMMiF9v11bbfJ4T45JzFulTjtOcCYX54/m07aGBzd1CIzp6SMpcJqJuwwFwDzd6qyVFWIInCsEohGwQN7Or2aaeM7AoWK1919ZQk4REJpaU55LLnHkSaOwJTbABbAHXiQRYiBsWfFzO8XQ69bOGIaK2XpfZnlp0G/yUbddtq3NKj8uUENHBumuxnhZ7NBKWEdKFmAjD/eV2vdxeu5oD2OdHuKPc7tNrgACkkQeTcxIiXOb5119++umnXwHzuof+cbGt5ZxCWNuKzBau1kEhwHNJqfCyFGZ8/vYC4CVz12rRVdttuwFAkpQlnZcF0VmiTOkUJyToveWSRCTC1cyZCMkNe/Nt7evW9r2bBzGM5QVMHvQ+gwgR/9P/5c+htZsbQuREKXFKgswOQZS2nXtHV8E4iWRAaFr32vYWBMjEJIZHHqMj+rEBhKr2N3zsQK1YJwgg3nfbm5ecgggcu4YQY2IAF1MEslCN0PAd200AYUq05JxQ7cXs9np1Tk6yxwJ3uvQyz//0lz9/eHg8n86Wpyxpmea6/ZQImelPP//09Onz9eXl5Xz6vb7a9nW/vqp2aOGOXDs0lwaq6qa2+b5QjwcsO0thcAfrrYIPGS9pYDNsRgNU7jDmcXiIghzIrAh+6BMxLw+CMhEyMhM3oozI98FZfL9eH4sq/uf/+O9f/vjt62173XcNn8/TfJqIqWt3tdE5ZWRKBBNiYShYW99u1W5qm6KChAADMDrGnWbih9zCwt3R72XffYyRhqP5AGqOfrUb2BBTBVKQBEUEmkZnYkJMKctDqtx6szIvy3zid/cSd99q22u/btuwYGehOafP5wdZkgOjYMpznk8pTSJSpjyd5pIHIng0QkZc5hGh4+rW7MLXdm3kd6kceBx03GFyA1NvVYcZT7vC4JEKpcQPyzzPM2OMTHAizDkzCwYTgwnb8cf8sO/etn31Wg32ZvO8nJalnJYpAUGEq3sjCIHEIOhNu67b5gEysPWEbtv1dnm5rZfdejOzMDMET4xTwlOCWWIq1Ch15x5kFhWAkFQNrjfnRha5W9Se1SKAPCRirjr5ChGsShYZeaXYA8dLBcJweH9mZMZfnpZ03Wvbb23XvlOvyTthqINzgukUHmbNo0aoEzgFENOheTjm3njc2g9l/FsI3rhXEwRjCAYQJoJMnsAQ3e8aOgf08QQB96BmUM2reQ9sDqvCpr67dw9wIIBEStrfS/0Tp0xzkSWnE6YHnp7yxyc+nc2h9j0AWSTnxJZIjbz19Xbt1legiFCDkCQllVSmqUxzKsUiCIkC0IEDAYkoBYa5dYtwswEXZ4LwMcVFQovu7qaO6GMSN96bwwU9ol6ImSSlLCIiUv4Xoa4iqZTSWrURVYlvYTSDKi/MjBHa+/V248oYAYj8chkogvHHj/lfBJi7qgLA6PLciQXHlMstBmzP3ZBAhEQwXLW7uQZEGn+cD9kGQ9xpB2EY/iYqosEleDf6QcTzPLugawNXcve11i/X5/z3f4vzw+mppKKOiMDCTEHMgEgsKZVSplKKMI0rDw14B3RCNNSb7S96/YL5SqBiEU5AxDGlxMEiCI5udON2pbaIfJgmAUSLTtrZPEPoYe5MRAlBTNu+fn2t1+vLVnd3d//fAd5/QsfAaaD9hCknEeZlns6n0+P57JACNPHOaMwS4OOM5WFqCgRdW607MxFO7rbXHdCZcZoKEQNGbxURk6QwDIW9tmmazbR37a2aaTeDiNZarW2v9XK5ffnj+a///sfzt8v1sm3NHZwZmUgoHVrUH09ap/NELhbQARqCExGxkCQU05gfWoNwYcE4Cc+1023vTdu2ASILK5mSjMtdEMEYTHiwAlvEPbNoTAoBIjiC3MTdDdAhEM2JnQGIQJA0UOHoxHp41d4ulzDMJZckKYkkASkg2U7l7aSVU/r49HiaTyUX58QkJRfdl/r63NZbmaaUCz88QKvrw8NtWSaboO+UCPWogs3de5CpO3dqtij4gY7HMBwD6hj3X3ZFszGlIwsciGwMIMdB2zICkZYylzkhEXOqTXt396B4G9MdFIz3X7XurTdCLGUiCkhUTgUJooV2AIkAQmZKBAUhISQChdFSDnMIIDiePzt6Zu4+LEoj1ICBxsweCSnlJCmlku+x8XjgwMy0q7YWMfqFwpIAUJs6OiNJSpKTGQQoM9N7ZhiMrAY6VL4WZsBIgFKm8+OHzxCkBlDmdDqd5jm63p5fv1JPiUV44Czu89yhujouSdfn2+Xl+Xq51G1vtXZwDwcPMzCF1rS2Wmuvrdfaeu8ig5qTxjZbyozgAD7GzUkS0xiOSExorkDg+AM17Pm67ftrua3T6+W0zB8eH8I/Mp6KIKMj9JEgLBQI1nXX9dUCsBQqE4uo7pfr5bbutcfQsxMxJi6Ec6I5yywUlE2gO3dHtWjEybzuDZra0EWqU1MwhwiCkPBA6HUoeQMhEsAUI6YUgQZx7Yd2KQagaYn+AduE3cwAlMgyODq4qrUWEWRKZMOt5hgcRHepxvc+/DFVPy6U6IchDBABGCVIlByJGZCOU36AAihgB+iB1XE3vzVbu+9qzQEkBUl1bMEVXCmQEQPYg8Knd0uLRYSmOS/zdOZ85vmczgvOZd3bDoGAhDTuMGgAYW7NsCtCZi4lZZlzOadyTtMDc45Rld1gDIGGVn0MseC+B92V4Mc3J8aj1kE4uIXjeJvwaGQ5jEeG76MUYWHmlP/jhnwSKaWYG2hHJCKgQwE6GgzCzO62b9u+72rWalOzu75hjOMPd1hKUkoqUymliNCY5uaETImIAHDfd1VDRALIU5mmwkz7djOz67beWlV1URLw1hoFJyqJZHiDwG0Q1SGCmBOz/HhNfDwv3rntqNWt9X5d1/jyt5s9/+3l4ePnj59+eXj6dDo/zVmYUiQC5GEIyaXkaSZKQw/mHhBBgUTcAP5o679e/V/ZLugIIEgZ06nL48anzsWFgQKFCCnzRLJgToGE0UQaW2SG4CBEwkKUI6Deorr1/bZebuvVVP3dNHEsbAjw0dADYESh49KbxkHLAoZ/ARGHpQcH+tfNFS32fXt5fRaRXIQYeu+vpnVv29q+fbtM05RSEuEkcrvU56/XaZpzLnCfKrq7u3Xt27qv63a77a8v1y9fvv39719/++3b9bI3BwdHsuEaGNEwvff368rDiWBKhSN63aw2RYI05SmXaTo9PUKStCwSPgPw9QZVx3UK3CnCCDpF8Gh1AiHx0Q0mRu4E956bKwKAExKCIDAjJ0ICEBryDiEhYORBh3AOM+q9N395fb1db5JzzmVZ5mle8gypEL+bwjHzMk05JSZGYMgkzI7Q//i9/fEFpwlPJ0m5TCXPU5nnc5yKKSaI5r2ApmhozSwZBrsauBOEoAs6UdBQdyEABaKDMfgRd8EE6DFyvPGQXyBCxL7py8tK5K21Uop6bPvuvXMYguO9tPuPI6v5dH4MP3NqAde+79aMrHlvrurd1AgxkTDQGBuBYw911CAPPuDnSENADwR4yJAJkEiEiXh01F2DmZfzefwlKbm/HcQgIuq2XV+vY7WMCRKoa+tgigAzoEhi5uAAtW77+6eDmR5Oc4xebSAgTqWcT+effv7TX/78T8IpgCAVymlGt9v+2/94+fovnURQmFCQGPlNFBbDUO/u1+v+7cvr88vtetm23pqrmmm31r1W22vb99pV1ay1purzPJ3OC6eMlIgzcga3CCdmEiZORDwcW5QmC7Mw9Qb28vZCvlyvX5+/EGESOk3584eHiD8Tff7wMM+JiE2YEnsigrAeHbQSBAdxMAJo32/rdatNg0rK81QoHMMyQmEojImBkSBYndRIzbvoRGmj9fZy3bfdavdueHe4eISBd4h6mMDdEYPYj7mJubMP1eP7Z9z9+voKvX6MBqwI6uTOoeZq0bZNqyFSwsilJzl65wTOHkRj1BGB6HehG8Jgr7Jj4JuAFxkTQ4qwUBIHQsfmsHvsDjWiAbWga/fXal9v9XlrzSKIp9OpzJlYQkgFjAEJwD1aI4ryvuHPnITLNM3znMosc5E5R5belfFNyIxx/P/RT00Z52l6mJdlfpyXJ84nlKU137c9XMM1QgEMyI9JUxwrd2y0I42BD6bYAA7ymAXdRWWBh8T+qO7AxCQpJ2FhIkSUH8XL/2h+u1fo+68QMhIz82gjj3HX0D8GAQxKkR0uxWMeAABghx0hADA8iQiTzFMeo+1lXkQSISFCTjnnzMzuXltTVfUIJA/vrrdNtTcIlJTLtKDHMTjRgTcLIZ5KTj+m0I8dOYmgC1gP7e3yut/Wr398yb/9/bfHvz19/Pz44TMuU0xZ5mk+nz/I9CTJQDCfUQqMlpMbukG4ozQqr4C/W/v3/vrsjRATSqH0yeas55OLAAtJEBmCYSTA0UZ1QNdQNVU3V2RmYSSHcFJzwDQtp8cPnHOYvXeEj0MVOCGOGI6RPHEk85r2um9dqe7qvYJ2Q3NrYRrOMIKqwltvvfcIV22naT6dH/e9Xi7r6+v6r//695yPN5+ZSko555yyJMHjBxizQe29b1vdtr3Wtq775XK93eq22gDLBoTbMKiP7ADo7XuKaES8XFYGn0qo+cvtuvUWQJK2qezCgkgPj49Pnz6SKW61e7Rtw1K491CHCAfsAR5OYYPV6EH3ACM60BsBjnfjS3gN21y3QEEWGkHJiAk5MyW20dNQV0AzCmZzbzHiszjUwgzcIIze5Wrj0PuEm2rAaAEQGNjrbf/rb0oIyyy//CyJc5JpmZAWc+OEXm298fUlXmHv3jMShOpIOPeqvrtzRAOww8oGgAQMlDAIQd4mkMMFCfchTbi79Wa1Um7MwgEAbuCK4Thg7m8/+rtF9cs//fPnsLI8GNIfr99+//bH37/87bauvaubR4CFW9/HpoOKIOTdLczQQw5l/LilHx6OgCGDY+FpWco0iSQE0m5ER2mfTydmHl0FACQEBMzzjqkMmjURMXG/rfvWdG8jUCuIwAEQrKtrd/1e2nPKf/71T3D0JxARUy7Lcvrw8fPp8SlLRmQQAUKqm6/r5fpl315bhAIiJhhgWwRGEIQMQBEBftvq15fr663edm1m5tbNWtd169et77XV1kYvX9XcfFdt5oGS8nLuuISIZGESERYRTkzDOXBw1z282/Z6/dvbJrs3u26NMIRRtSehdd+adoCJhRExMbEwE2PglPPD6RQBnDOmNMotE+eUkuTTcno4nwkPA5tAZAQBYCREEkN1UHNRo9JCUtVAj0ACPsyrFndmJADiwFGbwzBFgw3NSBiyEI+srO/rqkwnxCCrNFSugd2oBaDpFJ0jBHliOjHw3X4IR/uHcHBBCAiFE9Do4rIgMQ2FPCIcoQSsyJt7b6OziqvFVWM12N2dOVhuZleLZ6fXkCAkSSyFJYkwMLWwqqpgEcZkSX6AZyIAYozOP1MIQeYAiTlhz1wzayITBEZwZOIkSRJzgjTl6bScHs6n0wPnU2CB67beukdHVGJnODxCYYcP1O+Qlre+7JBSAQAiIfrRmx4C1KFFHR+JOQGOej+6Gzyq6X9U2sdNbfTvRlVGALrb65iYiZAwiUzTBAi5lNZ7Nzd3d8eDAgwREWYjAxGhhkEUKmW0u6eSS8pJRO5WxhBJANi7IoB7EImkYha9qfZqvXt4ymk+PxDhHm6t6bi1QyTmJaX8npES0fcNw5gARSCJtl3rvnfbmrf4q2Ga5tNyfkoPT+nxw/nTp08//bw4zUARYXxyLkGE5uQdvIe7Ua48b1wu2L729YteAYmBC6UI/+RzEDElETYOccgEGKGgHhwBe2+3277t2nqklHKWTJgwxCycl4dPvy5Ppg3CU8rvnxYmJoqIkGGmx7uMMbxu24VeVXHdrNfduoKBW7NeIZehDurqY98M8Ag9naZffv389cvLy7fb33/749u3Z/eRcwwIyHzsRkSHlHKU9gg309q6qnkcUYmEUspJZGImPLzWrqZDc/z+1h4Rv3+9hvYp7Wr+fHnde3dCJGKm8+n86dPHn3/++eeff4par1+/XVtrt5VOKoBee3T1cZ47Rp0dnEaOWAQ6QkdQHPJTUzADgwhymz3YIzCXodMVqBIpoxQx9Q5eHSpGZw6YGD0RBgtI8iQmZAz+D/qzCFfr3qwHcUYWUre99a+v+7/+7VpXLfkpPP/yOQvP88y4RHgu7Nnxxre5v6LfvBVJCN1sDb1Cfw6OCDF3Dxs4kMP2RuM87kE4aCFgEDZkw+PwHgEaQABOCClRADAHUIwrOx7uHvyxHw//9L//3+bz8uHjz0D873//6//53/6Pr9++1Vs302Gq9NC9Nm3d2EZlwCA0BAQQiAGsGiJzhzEhHTsTpzSfzw8fPszLSVLqTQNwmudcpmNh1+ZmMA6pTFwmysXM/OhVxE1hj9to1DugE2bJTGxuXvv7Qc9Uyn/55z9BAIYP3j9L4pwfHh5REokIjXZvYLj1vj1fnr/8/m1rN/VAASIkIAyKKAgLUSZE9Futv79eX/a+9VHTQM2q6sutPl+2rXU1zSXlnCHC3an1be8OSdLpdLZFOU3H4UZEhIWIDn9dHAwVsRWv3z+UcHCD0Y8akm43jMCxDRIFEyInQKKgqYjw7AGO0AF2V5Y0zzOhE5bHx6enD09IFOEYw7Jwn84EjEhPckc1a01yFgQpCfZm3cLD3MJsoMmAEIjMRmqOR4Sa9mFPAS+Jp7nwOwwzsXz8/LNepfUNQIkTODiheLhBwhDxIjQnPI9s0gOGx0EMKAe3RlASpoIpEydGZiAayWyH8GjwwiFe1F+qKUkkuig8t7hp1KA0USpJgSuTFmF2kZRTmrKURMxk4Lr3275d+x7k53NeUno/6RlRe2bdtLE28EbRhWwpHJpsT76zNyKlQBZIOWFOzuIkiUqWueTTLKlEpH1b3RtEJ3IRBD74t6HhjaKHudsQIg9rGxEixR3T8kZDAo9ABxgtFDM3UyNAMzMzQgy3cIcfH/QfSruHuykAEBJQIKKMdj7TPZd9uHCOFhgzSXAgouOIgKMhA3Xv4dY9QgGIJCVEkiQ5c8okCYkBBgUBmYlEam0eUXIZ6Fz2EBE31TbmeHFsOrnYNDtR9BYAZvqDK/JeS2qtBJoEMZwJgDAQKBSt9V1v1V5AUP6O5czzw/L04cPnz9sfX1/++Z+WZUKy19vW1NE6aUNzpITlkZaPIcf1brcGyIKBAR1694EmJJEkhcideIRZgTXrre97vd3W26q1R0qcMzMqg3MIgDTkBikcEfwdk2aUdgpmRBCidIwemIki/Hq9bFtVhb1G3XXELIMbmhIUBgyDWtu21m3bk1Ftu0cXgWlOp/N0Xufaam8aEWbmFuHdVHtnxGPyNay8Qwo8KnHOiZkAApEIeZh5xhiAjqiigRd9VxABbmtr+wa+tqaXbevumEY7NAxwOZ8lycePj9EV3ay12tqmXvfeQ10jIWbmcZr2wQK7GzEDySgMQsHN3dAMPNAbwApeKETIslBiYsaxmJGCQFm6YA1SFExesvCUKWdOeezJOaUBCXl7Idr75fUVAgF4OT+eTg9hrntrr5f999/Xl289cfrwQEtJLPN83i6/b9faE0f326Vdt35RvWIURiDvvvX1d/9CkReENDQk410N4BiXdi6UEklOkogSJgJH70NXi+YQ4SKjVxopEyKWTE2QRgTGwG/5j61TxE+//vnh08enp09q/rLuOZ8YMkcaadbC3LVi3918MDdhCHvuut7juzngIHE4DAQYhDPDmJ1NeUllYlHzIBJ36t3Dw/oQGXtgAAM4MAiLEHNJKYucJC0s6+tr3bdUcllOy7JkKfvrbbus8m6ayMTTvIQHuBOxMCMzsoT7uq6dRcbcJpzqHnvt6ubkgebQXS3uG4tbQVDmWbAIaett7/tW1zponFS7rrV+u27fLmvt3cJSlZzHVYfLxBOnVKY8LZxnlAJUgrIja7Aboo+ngO4tGbSQH4+MMKTRTJwk5zQRlwipHW6bI44s1Xb4si1GiI9FtLBN+9fnda2GKFMp0zxP80x0INzAI46GT4THkDyiO6EJUHY8PQLnyUa3ZhBzxuz3AEejuo/+qJupqlq3iEDKJaec338cAbBC0pCGiSgy0+jfCzAhC0ERnlKaUiqMQsgIhCDDcoWMzKnkac7LQ1oKztbEurt2NR93VXMHUMDLZl+2/q9X/evVOmeZSYkbpC6swEEJUFAoCRQ2UB8m8QBQO6ZOt71uralHEpnn5Xx6eM9FUWvmlSACYoYA4dxPHNOplEyFY8nYJupbMq2ODoWRSANBwTftqVVp+4RCRIPkGAyYiIkP0F+Ad1RwbdTiOx3u/jbeOTJv9soht7rriQ44Pg4ZhJoNCu13Z8F/UNrdzBUGm35we4fueRwoju+J4dGbwsjc9mP5jOnDnV+BEWAWgUEcg3eBIiDiiD3CzShsjPBBGAj3fW+tygdJJaGOkJ4UbtDYEdUDPBCCmfM8oRB39n1v2+Di6/vjfEDUXsk7aAghEyIzJHE3V+0Ym7V1X9fmzf5QYCnzfH7441/+9d//+Z8///zp8Wkx26NX1x10FweRIssHOX8WUGpteOjvAJehhNAAQ4jEInnOGEl0HDE3Xa2vdd/Wdb2ttjeQhKkCwAbRIU4Bs0EyQHdF8P/tp3j7TI5OC/PwPCbmJJJEBjf6crn0PtLvBrbsOHsRuNxHNXXvt9t2va25421L637d600EPn1+LEU+f37q3Uy9jQA1s7cBkLmbmqp27eMKfjotp/NyOi2lCILX2l5f1223bh4eTAjIQDSSX96rGiGgm9+2ut62dW97t2BOcyFBoihdmykynpYZw7XX2nXtZrd6/fqyd7XWTymjCCNguIKPhBEDh5BhyggKj/ADFO2AYRQNYWPKiSMnSnI4rJDIR39PjFkRzD0Q54dTejqX05LnH1Sm4fxWFltrz1+/mkYE/oz0cHpwd22tXa/127f6+2+dYP/TL/L5Q8ppmh72W/zx71eB8G6X1/3r83pV2wgtiTNsvq/X37bbtyBhToiMIMxJOI0eKnFhKanMucyynEo6CWdEVjHl6IpkGAEsgqRIyhzMnDNFImKKwStUJdX3MecIcPrw+eHTT2U5+W1Tg94cPSWahjaICEGDu1Ajd0L0YfKFw77txzN+4E2GQRwDEZkM3PaICtEJhNEgzGpX8B4B8AYIMydEYxkXX0kyLelhOT8+nuPpqX36eHt93W5XljTN8/nxaZrm56/PL99eymmCHz6ewx8KAEhjPujbvt/WTfDeegpP7qTdUGQ+TVJc+1Z1b90d1CwsHIDDE8mMIuRMjAGqPgA0W9PXtV7W/brtXTXQW297xZzLPC/nXB6eHp8+fnz69PF0fkhlBk7qaBGAFqFjknM3EQMimf8goyMARiREYSllzmUhKd3k9WrbvoMf/iJVHw+kmpm7A3Szqn3dtnXdT/MynwRFkGRczGDEMWLEMai6BwOMkPLBmDzxPJ0GZg/Hme2ocTjsh+6hI3pS1UzNTcPtbtfI+XuLUQN+vzWrHi4Z8AQpMzKDUALOmWASLikVSYmYmcZLJkJkCiRIkk7z8nT6+Pl0Lggv3+Ly3Ner1WambhqmGtECv6zx7xf7fz/rf3tWS748yHKSZZlY8kQy5NwjLZa6Qu2qqmZhVsGa99372lUBSi4P5+Xnp4+Pjw/vYS/NutnNrHfrGgoMuS6zzctcZJ5m8XOOU471Evvq2hyGUdJgd/Xt5oiBrI5TQQMPQUiEmWmk5gEBoBEBeN9pB+gwdMB3hZwPe+MxX7+X02NMPnSfiEAjSCfMAkb09NGa/49K+/g66jjgMcZhYh7jXRwTtWG9DR8Q1O+jgrs/jlGIJeeyEHFKaZrnPM+cSxA7YoSrOiFM88wpdbPaO7hnSoeadxBaiIbgDt78ih6EMGx/TNC04xg5AvzDKbi1jt6DMTEKD22VoIQUmAKfkJLUaW9rta31tvdr3f+9t+vLt8cPD8v5xIkSxxOvH6SeCy4PZ+fHkMeC7Sn2n8sDRh+qzURUOnpvFfeKJVlGKAZgGEMys5GtyS6kL1ova9tvxjyicdvoswRYQAqgCGeC9zmJb9X9sB8QjXwVIvRwU2u16UgmH4d+kGExYvoOegiH3q2rfvv6OpVUUno4nx8fll9+/jyVgkjjN2gfbTc3i971dluv1/V2W2/r7XZbAeHzT59++unjx09PpXDbt+dvz+AKvomB+bgsgzta0Eh9+f4SEB4fC0LuumLXMAcmzinPOWc5Pz3M51MumRjDMZiMqTLthBvAVa3V2ixaQCZkPAjla+vVbPAYM1EiTAiCBCyA7BQiDMQKtDu6ATFChPXuAzUZhMDD/FoBOuN5ms+PH+bHU1nKeI7GROLlm2mP759GQN23bavTtDycH1Gtu9uobtrD+n55xdfX5eNTcLEodZfbutVtv9y259u+qikRZwkRc+2t3faLOtABjJAkOafiHtodkIlSylMu87Kcp/mU0kSc3MGO/G4CIBYBLIgnpNeUct3V7Opew7r3hq3LGFMM5R0AAOQypWkmThbrttXb9bZd13rdJI0HP1rddW3eDvQDENw1fHcJHMbbQ3mYkMeZ37TnbcVL1JCydrtXIvOw0WsOQpRDvcOjMXR6PJXT6em0/PLTx8TkXW+vl8vLa+8dEB6fTqfzeT7lDz89LLLA+4WFMhJOHEiDImCQlGH41YgBAIMUCQkb5536FnZt+nK9XG7rtvXeNdwnpg9T2nK6CNXefr9cv97Wl611CyCu3W611a6IVEqRxKXkeS7n88Pj49PHj58+ffz08eOnDx8+LPMiKRFxIMEdGwlxQHsBjg/hvdUKAOIYs0D32Lu93uq///bt9bILIgGCg5mPR3ikBbk7EnKSIVXso+iG5NLzVll2HoPxGMrrwwZyXMqPi/n9Hw6+KMKh3r7/SHjgQg5oBSEJAQkG0cFTGujT78Wk9v7//Je/Yt+T1YXgJFAIEyFjMHAJLEAFOAcLc/KhGUIOZCAizC4NskMmF1Xn7tHMate9N/Nm0Q2rxc38X6/2f7zYf7vo/9wsWj/F7dH9KeC00DKlY5qIoOF7b5d9r72ZW2JIAsCeinx6XKZpejidPz0+/vzxcZ6W/0+jt04jMwFxYHRvW19xY75MIjLlks+Jl1RoztCu1C60r+teu2vXFhoO3QwopdykaMZhjCAQAuaxRwMSAxEAeQcRZ3KniGPTH2UbiRjA3D2A7uJkRIQIU7AwdkrASMBCI6ASGMiQfiQE/ljah3eF4S5tQGLkwbIdo97hSY0wVzO4D+bf6vp9LC+JhJGH3W403OQ4VsNoDighLMxS8vX1UmudSyHmALT7knuz8Y6jA0aEOwYIESETJOFD4Rp0QLTeHpXeFNxcRD2SORECSyQi5EXSPNljq73ut71etva82mXbX7/tL8+/swilLGWZpvLLOX555J9/evp4WmZbIubC/Dmdt/LxBKCmFuEIk7Jp22O9YsLG2tkRau/drLmu2K7FXqR/s3pdt+25ERLJYAAgRAXogRlICEGE3huVEADv0VLHPyEiExDCwFLcnbfIyAhCyMxIKETDzjyaYhGx781MIyIM/vmf6T/9p4//6T/95Z//6c/zNBPJQRCDiLC699tt/e23L7//9uXLt+evX5+/pZdA+NOff/rLP/3666+fUqLL8zeGfvnyVROYoDp0DXXsBhDoIf+gr/nll/NpUaQdWXHTkDw/LsvDw/k0f3g6P354zHMxcHNr4Xv4ar5F7IQ38FX77r67T8yZyQCa2cu63VoNJGE6pXTO6SGnkiSREN+7FswWvCsYAWNYeOu9da2qgEKSkASZFRFATix5Xso8T3MhIaZxnKXb9VnvugEmyjlfXy+3y+UyP18fHpm4B3gSmidOgtrqvtn1SudToACfAk+3rV6e++vWLnvb1aNwToI5oWFoN62ttQAYYIooBaO02rZtMwsAEMmSSpmWMi2SJk4FSQBlGCYIiFn2Pe37tK6nUibGjK1Bv5huvVVoPYWF+Q/QDxImBgjVvq2368vr9eXl9vIqiUd6pLbWt920uzscRRzu4UiHwGoUgHv1Opg60XuFFRrU1xVFunXtamreLWxcICGltMxzzlmZPcLMCmL6/Olc8ofH07LMBHQ7n5d5eXl+3tabEEwTnR4fkB/xq8D27vmgBDCOrxiO5mruRMgjWGukYQJCoKFV6lffX6p+vdz++PL12/PLdd1bUwRYclofllMWcd9rfV7X173eau8egGwBHhhI0zRPUzmd5qcPjx8+fPj06fPHT58eHx7P53MpJadEg3V3P4rfgSD3fuYdq/K+/QsAFqHu4ehqtta12pdvl8HmGzY602hdtau6ejhg5JLmZZKUmAkoCAEgpbQiihsenrA4RtRHDEjAG8PzTVF1/9v9667cHtsuxD2sw+9CgfENj8/7B1Nl6/r/+B//IuAPRU7MM/XMlBiTcGbKzIkhc2S2xJA4EvmY8zIHExW1x+TXrK+0nUin65ZuO2yt174aroqr8lXppen/vMD/68X+bY0/GiD11ftuWtU+BUoSpoREGtpae9luX2+3qhoQy5zOkpY5n8/Txw9Pnz9+/Pnp4+eHx6f5RCj/9j+b3Wt7yqnwPKpQt7buN3z9ysiPy5nmuUialixeyLN36Q3W1qu1bsMwTpT65LYABBMKIxMMFcVAThMDIgdTEhQO4TD87mC5m9GAgMHRYQSWHJftAMDgOEiFRMCJ5O4OcwYOen/H/VFYTpJSHjfE4wp+184dznWAoYqnI46WfLSmR/lnIZYxLCeWwwl1xKnQ2BbGKKfklFLOKeGgM5oyLzkXgBFwO5r8cKSA02iAwwBrvX0x8ZSzyF0F8K6264hgwmFxAgFkciaUksMl1IRQDqFRVIWtuXVVc+sYteqtIqfbc3w7p99v8HGbP7Tf503wMT+m/J/9069YOrQN+4W6JvfiFzXz29eto11607btTbWH7WR72O3ldluvdWu9KpEIpJQyiyAikUhZOE2EKPzDySsO6OKo4AiINmJ9DyXVPVTzrqWnMSxFco9tXVm1mwGCpES9tWavLxvFtyTltJx++elnJlnmuZTJLXq3bd8ut/r12/Pvv3/567//9ve/f3l9vbxejls7ErbWr5drKbKvt+cvl+u19ebElJiS8MCVdoWmUN69CiL8L//5J+vLr5+Xry/ry2oNMk+PaT6VuZzm6eE0cc51aEIAbTAaSp4eTl0VmdHch7PFQ91r14vq1s0xxD0CCHBKeSYREWIyDsqSp4klIZERmpMG9KDdYdcwUHCQhDxiFjQutxt/+7a1Nc9ZjtNpEhZTffdCaBBIPGBd1y9fv+Y8QTc4P0x/+vO+rf78NUoBws0UiNPjw+mXn7du3lptdcW4gbmGrjfsSTG66QjYch8MCYhws27e3frQM0V0j2q+1XbhVFgKSSKSgXg45GvMt+v4ibNQJnPY17h9876jO/8jQv4o1R5jftp6r23f67ZbH4UJXO2+9+PY0r83+5DwrbYfxf4o72PFRjfoSjlzhJlDV1t3rd3V0QMReZ4oTZKQAG0EvnXdrrevv/0BrimlcNjXerverpfLvt2ul9fb7fL48Xx6XJIveOcwB2CADGrrKGAODAcqnCwQBvEcCQLU4dr9Ze/frtu363Zr1gGDEyQCAGNePfrerPfW2ta9ArkkRiJO51xKmXIuOZfTspzOp8fHx8fHx/P54XQ6lTLlnIf4lJDhe11/V93He37/FPzHxqmq1dqhG5IitsGJBQsYWECgMDDzg/MHTgSp9r31lEQSJ+GUZJP9Iis49qppOL6HORG/f3b38h4HjOA7ZW9sMnfc+vEv3sQ2R2v20LEd/wkERPoxaowQiJhTUcTX7tE6gDFzFhEiIRIcXHkcDUYhHCHtQpRYptd1nvMy0Vn8g+1z26G2VvVV4bXDpcelx6Xbl+q/Vbg6BgtSGPqugXuTsudNurbE6BHqrmFloiUtueQPD8uHx+XDaXo8Lw/n+fF0eppPD2VZUvEgxO+Onsw5ZxpNDQgk5DDTuuu+2r6mUhhcvIt3tGa91r3urZoHIgsLS05lytOcp0nNZUyz7/1uIHQkZRQaHMYYrMKjZg6ltCDgXaKB7hAGQzAT6qreA5wE+Cith5IACcAQ3o16fiztLEkKHqV81HUeqno6as2Yo4cZCrqxxLjoi0guLIlYDgwtIN1pBHTMOQExLAIAlnleTidGtN7RnRFyzqmUQFQba+zYt4gGgvroGzncu0kRTLRMEyGMdsX7ryNgRyOYDcQhCtiQRwVGdwiSIGFJOXvOkcTfxHqGvrV97dtLxG/f5PTsH57xp0v++aKff/308Lg80BlhNqsvuP8Nb8/U9tJ30Ofa+25Ve7vt7WXt2jVcE5qgvxq89qgGGiQFg5NgKYlFJOfl/KHMp0H04Xet7KFYGsQMJqRw9XAPvMMdII4x3vEO01Hi3f16u3FrzgQApZSufd/runZrr8ySc/744eMvP/8slMDRLPat/f7t619/+/1f/uXf/uV//ttf//33L3982/bWu5oZIn75+vxv//q3x6fTNGUC7LXfXvZwn+bIHLkQInpEb1ArTO/8Coj4X//5c+an1z99eL22bzdfVTotxgmGhTQxJdnV3NxHmyjJtMwPHz9IzvPDQ9933avWrrXX2veICtiQA8ED0VzMTgEG7JSQyNGJc1oeOImZq7uaWaChKHlH7xHhkQBGDre5vV4vPTSVJIVTGojeLCIUT3h/RogoDXtDSq3359eXVDSh0Olh+ss/83oDAnx8gGmqERGePzw+mN20i+7Wt9pvu0brbb0aigBTILIIErv7gD2N9zBiPDzhYIAeoF216U49s2Qaxnocmpo30ynEmIsFYyC7oXa0ShBjbbx/OkY8aTiYa4SFq1vX3sCIhTkJwrB703cIZ4wWEuBbmsE92AOOowICBDqOaVpizjkhRHTtHtE0moYHMUHKAijERILkgaFqL8+vtW6//+238FB17dabtbabtrLIw5fp17/89FP78ISS4U26hW+1fBgBAREpAMIB1GGwzQDDHXr3remt9uvet2pAuSws83FwJnCEqNqra0fxLIKUiERSzvnx/PDh6cNp4A5Op9PptCzLNM0sMjLk7vvb96IO/1jdh+foXgJ/KIig3Vrt44p+5MGPuh5IQYxyuKlHbxwJIky97c1VXRlLFsLe+rquYN73NkLimN52zyNqBP9/79rHbnnvAEIMx+T70u7fr/lvnePjOFDeTQ+J8MN5EaLTsqjaxfat99ba4PYc97OIN9I70egyohAlJKFBD6ec6SHTL5kewKH3vem35s/NXprfeuweLagDBnJKxARMYQirWq4tb2vVoTMJRMxTPj+eHh4enp4ePn94+Onp/Ol8fpgLMwphRs6IGaL/qD/LlBZJfoSeIgQSUKhZ3W27BRhiQN+i71a3vu/7tu+tBmBKxDnlMs/LaZ6XMhXtLTENo9wxEBviQQQ+6HkDwUXyrrQjj051ONqx1ZsOVcqYySCiDCoUYuAIQzge0Pdf/1DaOeXyltkwdLGHXmykzsFR2hnZKTzGzXrY3vNo/I/FPA4HR/bD+KkhIBxTYubzw8M0z23f3WsZ3uqU8W0DwgNgMFwIZqZuZvdcxXvziJAyC+K9vn0vJsDCEOpm3SMMOlNnZDWqoGqtNdPuquFhigaMwilCGJgwEM2tq1eNm/Xut73Dde1fv3z9+G8fHh/PpymXLMRwTf0Pub1g26z3ZrFab9q72lb91lzNHVAYRWgPWM27BwCiE7ow5sSppDxNyzLlebo3Ruj9CwHCxDkdDllOiQ89zPhORDQA7ix04LVgOBS09wAcntgyTWoegdp7V3t+vhD+lYnr3j59/HA+Lb3b7bb9/vXr3//48scfX79++fb6ervddhsDnwhEGNb2y/WWUmJKGOQGTKioBdAAU0JhLIIJYc4/7l6tcvZzFnlIKfOt881TRw4+tNcGsKsjAElaTudPn6KUcjqf9n2ve617bXvtW21b27Za9pq3tg83nhmEVsRXVd+2uTVicoEM0edG7r1p0960AyKLOKClNMSJZZ6nZQaEgftGANPuodZVpffaROQ0n994+APfkafy6dMnkixlagZrVWKB8yP9/MuURf70M//02UWGsJPrnh7P+faY65asMTrU2s3AFIEPfjq/IfdjuBUinIgAghBIkBjN3dwBDdAw6JhuH7qBt103wh3srZ8TGDAW1UY/PPcOoTCKcEjmMuUjDjzc9O438bEA8bhujl3kcMZAQLwz/L/17MfC07bVjVhVHQbOE4nQB7g9orb2cnldW6XEQRgA1Pa17ZJF0j2ze/SxmZCzM7XAtdnr1pfsP8YFj3cC/FCEw9gJjtPHQbUlQBemZV7io5eSPn/6NOJY49hnBvFLTbWrRsBxGyISkSRpWebTcir3TmNOWVIWSeO2Am89Mz5a8fC+F49vs4u38oH07o4IAAwikIY/6aACRyBhIk4kSVIiGTKbkcp1sHCEmEek9ziMSkopEZKbRyio0T03a3z6eNAM7zKSN0HJW2kHx/C3KLa4N+R/XGMAcETSxvdLPAAI8a8fHhGpTPNa26ZGYYQhLDmXGJgN7WqjugNa0LjoHzDRQT9CEZiYvjDOCGHW1G7mN/VVvQUoIGKMMAoiQEJHAgpiCIqgoExTSae5PMzz08PD08P5fDqdluVxKed5OpWpiHhohDPQELX9Q0kMg+hwjMaHZA1IECWCQzk6hIO10Gqtamva1bsjESfKkobRgxHQFLWTNtZG2jwcnIAlkMIUvIMbgg/cjAgRjfLdQ90hWt/UlOjOgvMYLoXeDRHZDm374UNACIAFU373Un5E1hBzSnQsyPiuk0EiICEipKM1g+h8l++NiQnSQMiP9cwjyoYPjh0zjb5yTqWUaVkWSWy1QsRcJkop5XLk1WMMF5e7mh3nFDNVP0xP96HPCCKFgH9U0SEgM7viwE1oRJAcrGLrtfW97hFO4IMl0RyBmQEYsDAgQrdo5urQe+x7rbW9vrz8/te/TvO0nJbz6bycl+k0tSm+pe1GvaFps1gtunt3UqfuoIBGiSExhYI36AZ2P6whOtGIIWdJlOQwNh6NmPtLQaYyTad5GdCfGNesOHw0hAQMCPexyfAwxL0KkUuAsExlGuq8HbHv9Xrb9m1b1/23v395fDjPU2mtX2/r1+eXb6+vtTbtYyRy7yJFAISZ967rFkRMmCWVlOeUUFsomDnMgVw4C6VC03uZecR6uWIBzqXIFDihiPVAIJAhBQoLaOZMzCmfzpQkn07Ltu97q7W1Vntrra6trnXf9n2tw9G37/te996ra7+Zt7alQCIMoYJRS0KWXvveWu2NRaZ5kpxERCTlnObzaTmfmAlhHCJ1BJaBmYWDu6vOxd+ICWNFljLln+ZUZkr5j+fr8/oSgZQn+PzT6XEpv3xOnx57wN6aRfC+58eHpe4P1neKnhhu132vZgaHPAlpsPcjfDAW3QKCeHDpiAVJENQCfJD+Bk/j7kHFu9rpqFMeGm7j4n88hQD1XpfHl4b3MAp39FTSdJrKMqWSejuCGu49sHslx0MnFwDfLzh4r+rfm80BAdp03LzTniQnQERiEoFu49HdW629oTBlYRFOgsK4E+eUcs7TVKaSS5YkKTGP3N5EjdKlwWd5I/of5el4+UflRDhcEONWcVfhBkai9HB+OC8QP0EMPDUFEoxjgZtpdzeHkUwpTHx3Gg/aw2B/vv2x+A+N98OUfMBQx296q+tH9Yj7rxPmd5UeiuQlzc21mymNQBpkoiKp5DTlXFI63DFJskhKMs76B+GEDgbrIYQ4hhPHWg46VPI+1DrDSff2Ux4D+ePDi+9X9nt1h7ci/q4hf1c7v9cDCtPPHz9AAEoC5puqMyZNQwiqqnWvNryxjvFdEj5u2A6ADgSA2IAxvkZwQAQahIH3OKIEgVAIMwNREAMRAnLKOBdcTnk+5Yfz8uHx9MuHD79+/Pjz09Pnh4eSSxLJBILIJBDYHMztyPeFf0iwA1fX8Ld1RExCnJETRkJjsAgH17Du2l0VzNEHEIuKpMwsAGjd6xb1Rm3FtlLf3N2JwsSITdW0QxgB3IXqo2Gn3YY90u+lHXHIzCLenAqjIBPScZEekBsMytP53cb7j6Gu4W5jLcaBtRwijWC8y/gw3p74YyR/GBmO4OLvQ3rmt6iy8Igg5DLN8+kU7vu6WVdEHOzocVIeq4ZoxLa+DYHimOMN8qn5fTAUd3fAD3cIj7jt2qr2Znu3vVdDAcmESAHaW20NIZggCQqhIsfAd8UwUgUcygf73oQawoxWt3W9vF7nZT49nnCWXpwlCoZ0003dHBwScabsCBaRQhKIQXQEg7Bx/lLd183NRu6yBaqFpCzMMafvBhQAAswpLfPMRAih6qZOTICIdHhm7zY8pJHc/faFI1mdshTNI1AdMEA7qvXXy622/uXLc05ipr31dd+3Wn38vkPF6KOVd9AMCA+yBCELpoI5SxKS46LFyJlLKkXS/Ph9SQHsSsQgPDDzRYjZlRwcApGExoSSYuhCIVBASpoYZc7TMNR2bVtre6t7rXure6vjPr/vrdVea7QWrfsI5MJoGLfWAE27dbVuQWBeewpIHhNzIoLj6WAaVGjnCIswQCCCQyVC39cV0oi9QQCWnJGTQey9t64YcH54POXH/OGBz7O4c86DQCmSpuV0enr6+PLzt9eX18vler3u+957710HBQgQzfRuEAscpiBCiKBjHoYUFgGuAeQj+vS4dx21Jd75Fsemiz7OyRHu+l4h37XXVsN173tw5FN++Hh+eD3fLrdBdIc3KxRgeKAfIWl4Lwbjgz32gLfPOY7Hz9QiqpklC0lCxJKyqfnxMMV9ngTIRDmhcDCGsAlFJpxFTtN8mspUUkljM6KUTDjefRxENJXp8Ng7HIeZt6KORwfsuJSNkTMCD9k50fcNHe9qlgggImQGJriXwHHtP5rqfujHjr7A/X9vnffjfgzwdnMHePulOJSI5CZ3cQIAwNPTk1t0632ApkdrHIKR0jCM5ZSzlJRLziWlnFLiAyDGg8YI8NZt/66JHx31ATqyI8bUw93fhHXf7913GdfxLsH30n5XOb3d1N///b3ml2g5Pai6OpTCnz+l83g0ABCp1l3VWJXIwm20HkejCeBtcY0PIBwjRrzneA6REiIP5gcCj7hswZJoKnmZpw8P808flk9P508Pp6fz8nQ6fTo/fjo9PE7LksuYdDMMsABEACNFBBKOSPmwox91fzoa9P5GV08iKSUKQ+9gHSMhwqj4QqlwWlIRZGKeJCVA7Npv1601Q6iXl/byh11fdF17eCcyEiXpEarKAFPKB6+G0MNVO6Ghj0GYQljEmDCNj2+Q6MI9Qu394htP5Qwd/qPSHu4+fLsxdHhARAE0bgR+H6i9PdZ4TJjuveSjth+b4jHmZxr3HUSmJJJLKtN6ed2uV3cdTLIBFDx+xmOpHU/TsX7u6syBUBlz8beBUAC8t4xFwOut1dq06xiw9WAQZZFEFKbajTCY0ZASkgMakgaYYbcg8GZ4dxW+/S0iwJq11tZ129at9T6dJpkkp/8ve3+yJEmSpgli/8bMIqpqi7vHkllVXTONoR4QjiAChnDCEXMBEQh3vEu/Ep4ADwDCHHABMFM91FtVZkRGhLubmaqKCPO/4MAsaupRU4TzgFLSKdIXc3MVEWb+t29hYTKlbQV3AKAy5blMilFde3rTwk2joygCwM3XZdm2NW1bqc0C1SPnKecU8fi70C7EiRmHFXa/3d5sJMBuhTkO/9FG7AYD+0SEkZNE9o7CHRgGqFBbW7bKgEzYtcIjnCOI3klOowro6E5EZhLJIkIiknOeUilTTpmQQAMpBU2YMk2F8uPdssIKhQLISBhzFiJhBRoqOyiZkXis4O70EQ6CSVKCFN36Wk2PrW1Nm7baWmujlN+2utW6bm3d6rrWZd3q1rQ5oBFCDNtrIYRuI2GGROYeu31c7G0nhNSBKD2p6Y/wVqn105qF3TGCANER1G1r9dIqmB0fDtPjnE6FioiFSE4ic5kOh9PT84dPl+/fLufX89vr2+vr6+vb+Xy9XC6XZVmWrbZmWusGEO5G7ggwNAjd9yAFAOEGAG4QREwCvbLv3eA+yiQHowTmYKPDBRFDVua2OwC2tsG6mNZlWxwtHdLDp4en5QkktivF8JYCV7dmrt61sHb0VcAubgljXezO2ANVHeCgrmhDcqNMhVLW1tzccSBaSERSSlNOxwl6zBaixDQJzklOOT8e5tOhTFOfo0e3AfoGnslzmfsahcC+K25H0XB2G2CfrrgzMtP+1G7Q9fGl0JuvA9vUqX5+g5eDR/faDQi//fUddYA7KXAE8xH4b8dlF5qFiJ4gE911HgCen55Kys1a09ZVRNXUXcMdEUQ4JyklTblMpZSUi+yhHXto32EPuzhOH82MUi9czZtaU2vgbdAD3cytDwD2IxTfkxAc6nn7rbyHc7iduQHwDYsPkdJ0tK3ZVjnx4+HQ6R+taWttSF4i0k6cDjffQ3uMp0Y4lkcAwTAy5N5AGXJp0Usywinzcc7PD4ePjw9/8/3z3/7w8bvn54+PDw/T4VSmU54PaRJgjg6IG9YmHhHhmdOoVYaV+zfTQ23N9IrDVSSh5yBEV7AG1iAciYlEOJdUDmVSd3UjkZRyYgbVerkEnDdrennVl896vdi6VoBK1FiUxBDNQYgOeTJw7zKBXUJJnYDGirlJh92ysECAngR4hO854wCXt6T3N/K70G6urd/zexkOA8fWF/hdnkr4HtfvSsVbtGcGBHMLNXBPU8nTDAHr5bItS2uViFhGXL/N12EMDceU3bskjaqpvvMx97W137nfLTzwiC/ndds2c29qWzN19+aSEuRCRJxyuKlrNFcH91CN2rxV7U+4Nds0mu3Vxej17Y/ZTU1NLTTQMIkUEmMHDgWPAEFOLOCuaB3R16W5iUhISISIdqggBYCa16YOzeObHhcEgPl2WV7NabRlLCCUtAt0uwUPaz+S2+zllmcFhAUyiKQEkLuPM4yJQGuETYlQ9jrl1prrhUxE9BfQ80TCAMaU0uFwOBwPZZok51KmaZpV43pe3WNrEFdo6nm6uwsEnE4hvlrlZiCKko7HOba6na9qhgCSC6cUBGDqjkjvzxwRCHvvQWIqpmaqTa211mqtW2u1am3eurxObbVVbe6BQBFg1iWP1XdDBiToZI4AV1Ni4ujEDiYUQu/Ra8yw7qAPEaFmtVprnj0kT7W1ra3VWrhXj+aBZmIIQMJMOAlLSnmapsPx9FifP27rsi7L9XK9Xi+X5XK5Xi7Xy3K9rMvlcj6/vV4vl+v1atrc3FR9VEcD7eUD+oBISjaAMMyUSxbKIgwArZGqG41dQhCMUkq+z1G2ZXFB1bosF3WVwo+fHtW/n465rluYdwHMbdmur9f1urVFPfbGGO5Lc0SBfWh73zsbbDg3U1JyG7ZUSOhmgIBMnJNMKc1TOkzAFASxGyREeG0NlyUQ1KPzOQcxO8WtIU+IORfcy/IYdfR71N4/7P4ed47PXRtq//qBYAZAGNCd3YqsNwpvom63FY0AfSC2Q5/vns19xb7nyHBrawQGfINqnA+TJO7aUObd96rL0liEAwYCqvoa1Sw0uWUrknPKyEi8+33iHp0DCSJ4RGwOZzYmF/Ykrhbmt2PV1K3/ZD9VR7Hv++kKe2iP/QHvHfpeaN2njKjBGlo9tm3Ti26tbrX1qyfhvU/V/z2IXQAdgHsZMzIzIA4WFO4/QAhyoj6lySJTkUPJT8f5w+n48fH06enhu8fHj48Pp3k+lDKlnCUVTonSsFNCpLtzDWJIs8A+u7gbVQEAmNZYFyJmThThRAgZAd28Mzn7MZtynufZ3KXkIMxlIhEA6seEt+rb4ttKptKXNiAzBTMyMaAzJkZ31rAe3Ueu0fkDCIlEO3nMhwZkl9gndAoEbL4v+4heD7nbNznK76t2cMMd8LmHCYK+1uOWldJ9aH8v3m+dqdHsxADvFlcI3d9lNm3rdWnb6m7EjCLI3CNcN5fe2+632P7+053Hfevojf7BDqZ5P4Jfrtu2bb53p8x7u9OZpOTESVxrq2YeaO4R5rE1r80HXM/cPBxGkjmy1753ItzG9wUPCmJk4UTomiJAzQKBMQgxejo/JPsC+ryMUiLmrvrbR6kBXW7PyQLuQjtGgPl2XXRZO7Sxw06o43ccCQhljD5Gyb4D5QOwf0jkYZapbv0NEgEzMquxMYH0lyD9xMHbOWXm27Zt22ZVw80Q2IGISykPp4f5MBNzKdM0H7dNt7W1tdXW1s0ZdZ63u2WFVA6Bum0bWpNiE8OhTIb4drnUqtUDkCR1S0CO8AjqvIDbcuutCELsTnRqpmqtNu1xXa3bIvU8sGnr9xojbzPtRgfWV0JvawURxmh+9mY7M3VCvgOMSZ7pu458hLfW1rVe1zYDFWS15qEWFtEFNgwaBAVTlxjopkoplzIfrGvv95xk27Z1Wa/X9Xy9vl4ur+e3l9evL1+/vL28vL2+LsuyrWutw6K17wfcuXEQAAauw0U8ugNJksSC2LVgDQMdAwGYQJhK+UbEra2rEaht27qaK2c+fTgB2+Eha229x6bN3l7OAWHurWr3aNkjYWdijii2l3rv8+SAjk3v7RY1VSKUnCSLQwAhJeYsXFKaSpoKMDkO12cQAsTW1C6LqtdNU06SpPer/GTvs3ZEEelQsf134Bbix8n9flyNBvuttYxwF/1vcKr3kgECwMK1A9ZH39j3DRJ70odj9H43mbgP9DdG+xgpBiCAfUtG7K4abZ/Q7OpvPf52MVFraqpem7Wm2rJmNwtPnlMS6iSm9z4CvMcuiHBGZnShMI7R0Y0e1731mknv/j2/8d1ioBRxb87jfji89+PvQnvEaraqrq2er5fz5bxuW78l67WZderuOJVx70YhICNKZ1oxIQEnZMGUSJgyU2E6zeXpdHw8zqe5PB7mp+Px48Pp+8fHjw8PH06nQ5nmlBn5VmJyJ4TB3svZGw+9yXKr07peH/zu8kA3BCRyCu/wBCIa3A2zMOlrb57mQBDNlGQ+HIGoNd3WbWuLNvVtw1opjBCZRZCcBYWZKW6gCI/majFMm28WMYAYEM10bRuEeniv4oPA3WvvZMY7qsrDv+3NAfwzXjsKUW9Q4nuOi/vAqCNXh9ErDtOervW5b+wxoe8NxC58FCIplSKlIKG2ui1LhAF2txraV8vezYtRuu+qSRYRhAhEHWNI3Qx57Jfoyuolf3MjzaPZDvnoDVbo7MQrxIQ4IQpLmDU3C0AAJgZJaB1JyihA73Or0Sfow4Aut7uzRTzCwjQ8xgkQAaq2bHWQ0fv2JmZByhPnmVNCpmHT2xc0MRPNZZpKoXvj+X0EjWPsCT1FCHPEXv73arO34XHH0owd2BEyXeCAiZlEUhBiThI+mZo2izAEJwYRymVOedozg9DW3l7f4u2stqhCRKjGsmzEVyKptVuJXFO5qNr5cq21uQU4AuCH53z/OizQI9aqwsjCuSQSnHJ6PB3j7XK+LlXVAnIpXeMIAVTVQPvb4x3QNPZYELMIWxKxlKy0nj72w79L5Lr73r8icx825j32j3CjAdBJ4R2eepv59eRIBInI7T20u3vHRtSqmBokZeHjaXbEVpUTA3WMfUXomjfCvIMfkEJ6AZg861xKmw6nU3us7cO2Xtf1cj1fzm/nt7fz29vlfL5ertfrdblet7pt27Ysy7Isdau11v0Ajr2F5G1r6ODVgMh00Bqgi4X1PfatTFW4hWsnOnVNKylSDploNhXoKPe1bTXJxJSGCuB72ELqLdv3hbq3BKL39+7KoOgTRMZcSppLmicqAkKYhLvPbhJA9HBAgB2J1g8Dq+26tX4S9c3e0vdwt7LcHYfUG97yC3yvQWBvit+AAO9Hwojv/au6nOaem9xm152BEoQwEPjxnvsi3vqlY3y4H1033E/sze79VfVgidrqfW9OpHs1EzOramtNiRSVnMyMunuDm0dggDtstZlFrdplp/uDTMKpt687y/gGhe+OaUDQKWGA2B8bd9UQYCIhNhfrxiPv1c1omr93Su/Kp/353/dK7fXyZV2X67Isy7KuaxvS1bFvKgCinjcD+o7sJsSuGUo8kmuQRFNJc0mHUk6lPB6mD6fDp6eH5+PxcT6cpumYp4epPJTpWMokSVhowCqgyxDGYIbBvnL2gcsADuz51t6XuL+ph+NxPkoPs0DEkiRnlBSSjVJ1JA13UMBgDqRmBhHIDQB7yr5er7peY1u5NTG3vpK6hjSJEwPAsHEDR0QL6mc9MbMQiSBh1xlEQMYWY5QDPV2uhAldPXpNqGDoEGPqc7eu7n+BiLx310fIvh8ovYcN3JOhW+bV23GBe/8p+kw8AhAll3I4sYi5tVZb3Yixawnv4gpwF9dvV4/sDhGDGRBBgE5IgB5desenwoeSSnm/kQCwAO3+VAi3IY+bNTMEJJYkTCzh7mHjxlkQkQnc3k+qnmGM9W2mhmDjQETonBkzs1bVw60PjwDV3LfWzUsG2JCRmSVPUmZKgswj5w2/DUtzSlMp943T3hkkBNqpqYON0js3TCIs3OMHALzH9XhP2h33ANYZVmOWPioqM61mDRFYaJpPh+OpCzV6eK3bVpWXitgiLLqQyVo9wC1yuvajT5IERPNRDHczgbp9vFtWUZtBtHWtJUsSmXJypCnj4+lYa/v69UW35oAecTgcmAiZwz18tHpvwCiijhYjoiCi7usUJqMU971N048l73g0dg9V7aurZ2BNtTXtC/S2lPvK6WAs3GELcPc63L21Wps1NWwKrRHz6XRoPUsjdAh3wzAPRSSJ5JFupy0hIjMABXOW5DkO7s29mbXuibOu1+V6vVwu5/P5fD6/vZ3P58vlPH7jcl6uy7purdbWhsrbGFc2r9a0Wo+R48RCDEKCZMz+jbRI9NA+gPT9S4kosUAixXADAnOnxJSYhFAQDW+RYj8ld9DN/fkBETio7TFO/NCmKJRxkikfnh/kOEHmLlZCwl0XL/aN0M8Wba1utW61rXUAyyIioH1q8PB+H+PE6De7H1J3J/UN+Yz/LCrh3Q/qUrqwv+yulT/0oHbeya23voe48N7PvUWLfl70HOS9tTj4uvdjRLPtvst4m6SNpbKvRjRERHei8HDZC2pXD7NaoRESCyfhLKlkySmlJMLCN3bA+2nSazAP2NmBMIoBIjcmdxkYZdsb9WbaidShsR88e7v0/Q3cr6q1va5trbqaN0TtvWePfTjUcygEQCdyYuQe7ZFyppx4v7AUPh2mh8P8NJ8+HI+fHk+fHh8+PR6fD6eHcpglF5ZMXMYcF3YPQtgroL4YqI9Xei/mPbR33d+d+wUAOwVuXMfD4UM5dnjZzQuHkkDKLqkGoUUAGFCwONLWzL15oDus67ou67Ysui3RVvGawwEguhJNl8MOhjEqCCTv7FAYMYoZmVCwb2SEjAkJAnzfHBHuGSFhqIcBNbUaGOZORvd78Z8h5DH2QnHkp3t9/h7F96HW/offBn6iAHAfivfEiVOXC5+0bXW5tLoCRmfE9XZ0bwbcplh3Gyx2uYT3o2NfWUEQiCDMh5JPh6nceRBBQG/k9887GupI/bm0tsUlvJRcCpCQgLuFeV9lgRLSdeoA3G6b0t3NWdXUwtyJmBgdrOpm4cQVkKLDcJBHSoQYHenWvXSJiWmPzjcjbeq36trqujBEPD/c83sQb68BEZGBGACRmDmJyO0wgPdc61ZVIAyb447bEZIOE+j9KCamRESAhm5qDrVWgGtEJydrrdvlfN221cxupY+51bq5G3dDGJGcM8uAkhOiJE6cpvJeW0XA+e1r+LYsV+GZhXJJnU+JhNcplyy6tuv1ah5ullNKTASQmMfR7eFgt2ixl0oBONgvzHSrjEbK4nvp2r0wOY3Q7m7unKiU1M/K8SEHI2Mc3OHQSUj3IcHN67apggFVs6gVKFJiAF/begUtlD9M5ThNnUWH3S6hqXuMPnQ3+cSuEYFEROFJxCK8ZJsPD8fjdnpcn9e11+mdpbhcR3C/dEX/6/VyXZY9zOtIa/yWC/sYlSJirwKX6zVuisUAzSq0aFq3uq5b3aq2Zq25VnNtfaRRtZcFfXiTkK2/SryVqHDreMN7F/t9C+6/72Bg0Np1WSIxljwJ5yRdwjqQFGEAc4QTS0R4s7rVtmzr5bpdVjcHG/WVtTugECIhj+A1VvweGW/T8L2+fi/Q8X1bwQ4/g/cK+9Y77A2vjrR5h9Pt4fn9v3uEwD3X8P2vwz8P7WMh+f24Cu72LnbR0P7zLk5iaujoHbjZPVn3lWoBptbM1tZ4QxFKIklSlq5TJ4l5qNXsSKm4idJgFyC64XADe7+6vwxnY0/hKsm6M8yYDvQ+6uhk3t8CMT5/SnWzukRryVqPp9TnqbepCBGwBEuIEBG6IQAd5+kwFUYS4pL4OJUOdH8+PDwfjk+H48NUjjlNnAtL98DpoXKPzUNK6XZcdoID7cnmPqTZm9BjM2B/juzflu1EmLh/IwJAopSFk1BKIUnHDmAQJABszQO2rfYjp25127Z1XVvdrFV2zWhJhEUUqJk3UwsbjCYm7N6VHmb9vGmwr1ofA3Rz8E7lh51LzokSi7o38w0U1J3Ib1bc+/X70H7LZAH3euk2Rn8vDEchfxNguj3Tbp8KER0cQJ2NmXJKuS3X7XpRV6CbXbMQ8d7QupFT97u7bbW75b8fKDuiVqikPJeSJN3fRuwl/u19E6BTRHiHqCCR5MIkTBR1g1BEICAQARqe82gabsAUe0ggJKJxatOgQzdzRyVioZSJiLg3GGGMMJCwJ6gkQ5FpQKEQ+trp8rum2raKHn662/Cj6/A+t+gBnphY0gCP9vyq56l3I9C9fBnfAZCZzckGZqJD6gFoGEO5aqvNLVQ7Pm1crelolu4PVlVVGyIgkqTk4QVKzpkQgzClNE9zKeX+ZSyXc/i6rethzkSUErsDIgRSyZKSwFrrVjurMaZCpYgQEUdEL0zD4r623ssAwN466Fs3IiLMDZFwKByNBinjaMGhIxgw91T9lg70eiXGNLBnkx703siFvtlaa+YcQGoe2lJGZgrwqnUJnQVYDofDrGbNzPqYVGtrFuZMjNNEmFBoVJkBGBSICSAgQQYrpU035FGnAKzrtl4ul/OlV/IdYv/y9nY+X67ruta11la19yHCBuTuroHaWqvrBndtcjUNjdrqNnBO2tS1eWvd3KuPevsyJSAmlmCMEet2hPxtUv27gnjEzL3G7T9Tr1vFZeWl8JSTT7cywSN6P4NTkpxCXR3CvXUKx3UNdfCxi+MOKITDR+OWsQy43C2471XrvxTaR/12sz/tiUjsM8Eunx9jJOo3PF14jP78qGRvjJIYgnLfhvY9uo9/ARE9Wvzuoe0nHBH16D62jodT9MYCxIDS4q623fuEZuZhAEEYaWeuTLlMGQDwJnex30tHZI3pAmAnN3t/q4SMEIQUFBzuEcJuzsP0rQuNuJK5ooXZfSghxqdnaTXXiayF6yAeuLmqUh86EhFjypAydC8irQBBj6fT6XCgoIx0LPlhnj48nD4cT8+Hx8f5eCpzYZYAHjJDI94QEgyv0e5etM8h4obx7G2/2BclDs22XvZQz84cf6cOeGtO01DI4y5zIAzM1tN0Iuzq6bI6QNPWtHVNnlpb96poTTnUMJyhIHtg82imzYeDryASkQMZeDNX1Y5w7O+2LyPaY0iXfCPB7m4BKak5NXUNJRLEREjfhnb8t//2395+8fb6er1e8X0P3BWA+3O6tXn2bXM7MvD2P4CRBuNuGEPMrk11jA1oZ8nBtxKYd3VAx3Gamcaw5473cg32NkpXIRZOeUrTcIUyt59+/snNb2FunDF7jwwAO/Ksv+LYE+o9D3hPIH5Xl4wB7j5xGze/VwkDqkPvmrx3T+6utwE3MOt4bLddTUSPU+a9sfL2+vr185f7p4zvj3nkVfvriP19fPNtATrEkaMD0CJi4C330D+aCB4jZ8dR+d6Osluh8n6Ex97HhTtmxHiPRMRE0zQ/Pb/z34RaR3ukJKfTMaXcH7AH1NbWtZsrxC6IQJ2KemsV7HeCY+verZNv7nX/bPtH/v1wMPZF0P/KN/fzjvTYv2dfBf7erEspHeaDA3gH2faMF2KttbbGCIl4ypITj305jvVeZvWWFu8w7fERbv/a7ZNExJ2NRy/D7XbpuDo4aQc+3b4+/HYv77eHME3T3/7N374/rikF9Xhl48S2Pp6w94mYh6lpZ963W5v/m2Xwza++vX7/eno9KMQpSeookT7vG896MNaIep2rna2l5mb34KDvnz8cprn/nChN+cPdw8Pb/9+tB7xbHPjN/+0fM/4nfjZ+eXuIA1AWo1cR33zdP3sc+2/eds37T8bCc/fX29fd9k7cnTb7Zo0Iv3+Zt5Qp9k+01z8jtnVgxy5Udnck3H2+3323uPu9+PbO7xYyfDMuhYiAqZSbr2uErfXLwCTfxhm3Au3WTrxpCozDFyBQhJm5lyfdwT2JJB6RQ2gYn+0NFniPNfdnNHyzq+8i034zeOuvfPPSLOKf3lbbv00WTHshtR+1va2UkOj2rXAIruu2rmbaH8G+Z8d/MIIw+t+O7mkct/AxGhl+t+nf0Rp3C2uU1DBAbrcDJPpQ1d12QESSUvK7NeI3of2v11+vv15/vf56/fX66/U/94v+f3/JX6+/Xn+9/nr99frr9dfrfz7XX0P7X6+/Xn+9/nr99frr9f9X1zcwOrPqpreRCtxm6wA7xLFzTEFN3R0COtcbCUVE+lgEYEfx7EiefX7z+wuHGsn4H+JQut//4m3weINN3Y3OB/q0jyc45ZQGdMvdPr/+ZGZq3WihA3QSAppHrW2r1d2hiwTxmEu9Yxf2aYyqQYSIdMsnZjJXhwACNd+21jbVam5dGRT7aIgYu+TvjlQBGFCOd1JBh3YPmEEnxFsQECI9ffcj7ziazc+rvXSufKfh/X4G2J/E+3h9DPL7HDyncpgPJc8pZcIb6v42AjYz61Ct2mprzU13VHjs324MrJi4lDxP8/F4zCUDdVoR7jNqd7dOy0RkRGqbXV62/SXH499dSW6f/v5zgMXQDfb3P73hWG5DaURADiYjXbVtWpupGQCA95kx3BxRBmyRSYRyllyEmTAwgeQQrXVdr2urm2qXqvEOo1Po8l9Dl2sXgHp+/vG2rjCMvN18FW88JdrVyGOglLBLDdxQEIBAYzJnhJRywhv0YR+ody5f93vo83W6LcsYCjF9or5vE9wBwrcd877GbtsrItxNHVZ/p11Urx4WQx/pJj5+ezc7FORutIr7970BXu5xeffIh3f85t1Q8za6HW933+13f/t9QD7WHd7/dPyhULoZ1OaE333o9hWgaltrVW1rujRbq3ZP1JQKMZ/fLtfrAuYsfHo8TFNR01rbumzhMc8lJQkANatd7HUn8fdVFBBmHULLbqrWUuIkouqtmlt3MrvNRHeiKgIxpZLLXKbDhADr9boua10rAn3/9MPtNX1ZXlbdIHYM8z5K7c+FELvDGzMT8g271BWKOhzlBlYf74UYiPvh3EybNTAPV9XmbkQknHKec5klzyRlh/da267X61f0NjEVkSSpAxX7mBcAOsXshh86r3Wpg7Ng0d62fwQM4IBAMATFMOKgBN19Fd/1/RkwBTC4Q1i3Nr2Zj1MgaHiQJtmQIzA15WWJtoG27p4TnZrtEWaOGMSQ5ygzIKI7bQu2DSSBZGAiCNw2V7NcPItLBBpUxWpgAAF0TH+D+wl5qbG2d3zPjvj5ZkR/22V9tdO+Jb/ZM9Ftlnba7o7t/AaxMN5d7N/ypiq9n/S3Mfwt/sE7pOA9TkcA4pTwcGf99q3Sy/q2LV8Hfk0tIoYcHVLn7osISzL38/m81QoQXVM9pfT42HGOM0Boq6bNVPsHuwmyQ2cE9dgPQN2EK6WUE6eUkgzvMx9IjO4jbbVLMHaSfD8R0COaWRtYGz89fff86W/GXVj9f/4P/7e3y/l8qQB8mA4fHh+/+/ABUZat/fr55edfPtdtDbApy2Euh2meSnazrmsWHkwYEZfzxdSfH54+fvjw/fffTYdyqectKmQ6L+tPf/r8208vr79cdWkcMc/p9DAdT2U6JFVbluoKGAQR2P2JEUREkqAgJUgTS2aICLV6brZ5ipy4/G/+D/+nw8MAoH2t//EfL/8PU9AKrVqrFnbDTt4CcHd5hs4QBaBwRGCi9N3HH/713/+bh0//6tPTp5JmiK684Z1Kva7Xy/V6/vLrXy5/+fW3Xz5/+W1dLrWtXWg+qOtcIAEw0HE+/PHHH//+7//+7/7m33z/4xPmBExOA42tbat1ScKHw0OSmah8+fP1H/67n/pdkMR//d/+4/S8tY4nG2hi7DLAi9ar1s1aDTcARyRgQun62MN3CEhCjnrMS379p7fPP71ev5xfL2sYBESttm2+VDcgyZImyYmPx/L0ND1/9/DDH57mktn5g59+8MfXX17/8z/++3/6/Mt5e62IwWmz2FZdz7aebVnqurZt01bdFTHof/ff/J8/fvhjvxH2WrbfXl9f315ehblMUxfY6dI0u5CDE9Fxnqdp6jljRBBiYjazpV1F5OPjxyS8VuvqY63WprVIPs7HaZKU0rKuy7rmnKdpUjVT68I3rao2ZZHBNiS8BUz3MFOMSCKd4dFhVKa+rPXS4D9v822bv7avW1vbUr05IgbgILlHxFBjh5FhmJmaqjJRSglj4FtNbT9m3o8YANz/WYadMjuQWDBkyACdGVNK3YcUAGOQFAchZ2R0uHNuCXk44SAAPJUPRYay3tMD/u//tzIxT8zn8/XXr+tfXs4/vVz+029vf1ku5waK5fnTDyWd/vuf/+F//B/+va3t4Tj/r/7X/+aPH74/Xy6/vn75x3//J9v0v/gv//j04dEQztfrL79+fXu9rNfqhkTpeCinp2LW1mVNPE3luCyvl+uX56fDh+eHt6/bl1+v6yV0g6FjS0AMWVgSEkM5lec/fPzxX/3h7//mv2CCn/7Hf/rzb//0yz/+Kp7+2//m/yi7WfD/+5d/949f/4weFCBjzRMEugIGJMHDlJ5Oh+N8KOlwLMcHeZim0+M8F8mEpG7VtEO/O6SKyhHz0R1qq5+vL6/LAnXR5fz29ttaLyVNp9PHHw//6vjx+8cf/vXh6UfOM0Bovfz287/7/A//d66fv5umHx+evns65jQBJpLMnCA8vCEYoAMykPz3//Tbf/jLS7+Lam//n9/+ryGVJ0fDWMXfcrzlo5VHSsfMhzJS7VQoHxAPBidvK+nCrFOJ6SQyl5RSdoZXXz29Pjz+Jc3W5MOX1+nlP8TnV7t8NVUndkpOyWrTZW3Ammf/4V/5jz8AUrI1fT7T59/w9NGfH/GQJrD0y9vydlk+fr98d6ynZrjC2yqfr/TmalD+l9//XwQHAO1PL/afv1oPVV3keIhk0EikEFDkhlNERmbhlBIRdsJCa63j23AwP1LKSUQYqVtgmQ6fSXNVa97jLEvHDtIwLIyuXjnUAs06U3eAI3tRZGGmNupI/LsP/F/98M4U+ya0E5OIEGGnNYS7iDDxrYRyt1DoHD6mIfNiatRVOYaAxV7/RnQOJBOJJEZEQOvsswg1A/M2pA7NipnJ4GFCpyxTvxNm7iT13UMW3QPsHbB5zxkFAAJ44Exp4oncgdFtW84vwJwDU5H0/Ph4udCyXk1tXVZBzIwEkHhYF3Sz2JxRAQC6hn11Z2RgQMVui+3AABzAgNCNKxkRXcOq2WauAU6EQRQolISnOU/zBBLAgQlAwM3VQwVag9qUnX7HFsUuWcPDtzGgV/C3wxR2hYb+3KjX0AARju5mrm46dPIcAd6x1OM8xd1UkPjdA7PHi/4VxEnSPM8Pj48PT09lPnDKwByEN2dngO4eFG7m5De5zm9vA/rf2HUrEQBt5w99e894A/reEn0GRiAQBNmFjwFgd+XoInQagB6CIBNND3z6mA9PxNLQgzZypWvjl7fzb+fza22bUwMw97Xaummt3rrrloVpmFpY9wh5/1y3lRYR5q6qhBg39SEAYkokSSTnLonEAOg2NPUQYVcFsS405O5quiOUo0PhqfseBgxek7u7IyF6IBIxR4SpBQfGbg/DTES3xk7X6ukftVeg3+jKQGzLel0vujSrhoAO0OLGjQeE6JQfM2uttqammlKacokIVW2t1tr2Gh9hhwbj2AeMJL0FNerPeJcXDTARLiVPUylTIUJ321N2vFXkg0NLGL0LtHcL5o/HW2i/rtv/6z/+/OFw+PHhQYAe5wdwYchgTMGfr/WsMIOCrmgNw9ysNttq25o6ISYGpu7xY2EdVN7Pnv7w+ypGIHDQphhmaZwz0ZU2octSdvscxN7DcjQ3VAwAbbpe1+v5ul6XUlI/5YnpWym//UEGuIOBB3aWOUYgITqgB5hDODIwIUWEuq5aA4JJIm4mb7GzMBiRzFVNHQJJKBUyTUmaAqB5bE0X1QXARLBkQULFmlJC4lFn7hnW/m664M1gEuI7q3C/CQc7M02YJkiz86Re0AuWSuIkjJwpTZgnnCbJM/kh7GA5wzTJXOfZjhNzFqbMKm6ITRCSrIbnC5zfDCEen+DDs5AQJ6TkmLRqW1ZqjcwbS7x9dUJzBd+QPUgBGjN68pgBFYkrtTc6bxBXOL/RuqIaBzF8d8ejILy34wCEodrZn0cEIvZt1l2mIBpW7IEbAbuIb0QgRE/BHUHdvSkiYmcQmEU4dfpmCLj5aKNRIGXm3rglCSJlpizS/zkcoR0BQM3UtDXqwpoAIPzNeP1boVlmyCmcjZQQzDznlCT17WldBlvVzMONiaZSRMRVAUaKTth3he8Bxs0sSck5JRFCqrVusXnz1lTNA1Ck5pxUc8ppSKATMLOI9IcrwtF5IHsDRGG3Hh16cd9EBwR8lnkqMHPeOuNwu7wuZ0nzND9NaUrPRYhUW10XqzUjWtd3EHCCCCAJADAjwkAytdp0U2MSIMIAdbAgBw5KyIqMmDKnnBjJmuvmtpppQBgzUIJENE/y8HA4PhydTEmNTEG1hXpUgcpQNwXvivXvAbE3KZjDGKgLhDvsZEAYDM6uSgWC3cUomgcgkbuqVtXu3qUYnaBvOzVj9G+ZhyRq74cNli+MczpJmufDw+Pjx++++/Dpu/l04lKiU2shIKxPG+4ikyHc5GL2iwAIdmlD6NbRPe/TboJxmwHEfpDQu+VTJ5ASMihiQpTO4O9pxTjXzNwCOICEylFOH/Pzj9PxwAQNtyYr2RKvi/3y9eufX15e6rYGNY9qtq5tXZttYA1Uww3c3FqXgPzmLnpgM3MAMLNWKxF19di+d0RkKrnkxETC3QEWqmkP/IgY7gagaimFiET4VqFnqxHQuW09KkO3h9c9tMMwdYAuwWtG0fXukYX7PnE369qr+9Qqdr2Ue0hNACyX5XI+67VaUwi0gNYLCRsNfyaWJKrazWfdrJRi88HNtm1b1nVd115td+WPnqDtGZkwyTDDvhmijx/mbpL4cJiPx8PpdGAh8yEHiUMrAiKGWrUjeLdU7dvc/ePpe9i7D2/L+t/9w3/4u48f4o/4h6fnD6fTQz4+5eUo07GUP319/eV8daxrM7aNws18a3Zdt2utJMBFQNgJmpu6JRlJJEK3HoBubEbIEdiqQmhOFgFEBMOlK/pQxN2J0GMIz5r1AAFQ2+XtUr6+Xt4uhIckqZQiIr+TBO1TtN7a0ACM6EIUvXqLAAsyo3ASFEGGiKqKuKpb4tR7uAZh4YzDJgoRV9eqm0GQpEyMBOuWW0NAdV9qO7d2AWhJaJ6SMDdqU040Crn+JHZN8T3AA96U9/HbyA5gqF8lP0p6tuMhpoNFA31EvmLeKAElwTzjdOB5SmWW+mDtUIWwmDyth9P2yE5IGBm0NMq0CVeYrlf98hWWqyex52f5+F2eTpmzYHKQ1rQuK1/etrcXeHmJl89OYBzui4sHKeGWM8SM+IgcKLxKXb1dvZ7j9UKXFc2JJIHDne0QsfT5MiNR+JjT7YG/1+7guz6UdhsnBCISlq6vgdS19liYDMB0jC2691LviuYkqXNAAbfWNDQQUbq6rVCX4SJCd4bdEAShz+wCQFVrq7Vxq62LOWa5mwH/vmonRGGI7gwPbCbC2PcbBhKSIxJE4GAahkM4jTrMIwzCu6wpIgRTQ0SE3SdDmKiLxiCiuauOrTIa9n5jIxIEugeA78TXPjfsMwWg/ewno9vI6v0ukB7yY6acoi26Lutl26yZWtvWuEgBTlPO6XQ8XSG269XU2loT5lwYMbw71CNywuhDUmrVFtZAAkN3NMBGbJJCErBzIcklST+gm3lXeQkAhJz5cOKHx/nh8XA4HXIpS9u21pq7hjeLZt7cWphTjKrhPZZA3FjoY16zkx99nAmEmCUfDg85TQDYdFvW16YVAdy01q3WVbW5WK/Ze5djl7N+5+OP7lKfdCFLTmXKh3k+HY4Ph+PHDx9/+PEPT8/PnIv33sCY+OAgcZoBgLmTOYL5N8qm4GEB3m1teqVjERqhEQYQo/065q+M0pWUiYiJpfvaoRAxKoEg7DaBHn1A4Q5GHHNOH7//8PGHh49/mE4fqEzuXuumuaVDHK3Fy9vlt7fL53V7WetqoB7q0ZppNavhtRsxOAEwdzuvb47g1trl9XXbqqr23CqZxR7aJaWU0jRNU07hXUIEELD36vsq7fuqV+79nSZJhFQ8IwAG9giGiEkEELvDoJtTQpKEiF2B0Htg6Xo+Gjw6BxH3HPC4Ff0R92ViAHX/T4/QfbFZgAFowLDQAHLsPzIlSVNOOXHSQEUVZEFu3r1qbjOB6C3LIHfabawwYCBvOqTB3A0DlM2zgyPFLkffK6KB9Bhz/9g1bRBiOKHdrytmOz21+bRJaSSGLIKnKQJCEjAYeb2o+VZzaMa4hqvqsm3rth7yIU8lT7nVqmG1VWLstrH9vNlJ1NSHzWZBaG5BRDlnERmJ96jgAyC6cFRXawCPMADVWLflslzfrnPJ0zSdHk7TXLwnue/bA8ICDMIgfFjjAvW64t5fBhlJkHt7zd26ZkxiERLYH9dQDXTftC51VXRiLnlKSbZlqitZKISGbxAtMU05HUoSouqUGWkcw7hbgg39oXh3fx3KRbgjqcaycmwX4SSweSI9Phij26nFFfGCKUhIeGY+uswok/icIisiuOlaa+jqKwYgHVBypIPIPJOGcwNgdyjFp5nmowRSVawV6xWrUmvibmXKR4dA9FV90YwmBDNMB3s4xocDHRpVh+t65Vqvsba2mq0UG6gC8s0Zbl/KPWwDdNfIrW61tt4aTSmxyJieM4eqq3ZvUSQUMSZGQo5eJoGaA0DcVM4QgRlvh2eAsKSUcinmYREQ6KraFRoRMIIRM+/M/ps+CoAJ58StSUs1IgCxlAbwLnT4ew35Pu8iQoiwodY3+pCE708hhrSbutlo+HW4FwQRJZEABghi0oY3QVSiYcE78gL33oojGvCrrqNHRDjGenFfOsUACgEgMLO5M7uZI/o9ygERD+VBUhZq3BiwASiCa7O6LeaYAhPz48MDgocqRmhVT4x5PHQACAxKXccynFr1BZsxYWA4BWDr+uUlAwAXTkUEEcK6xHN0OxBimE7p4eP04dPD04dTStmBrm/rtrTqpuFNo1mYe4CxYNrRKvv9ovvAVNyhtbzPq7vogbDM5fTh8Yfj4TEilvWM4Nf1rWuR1LpudVNr5sajd9r36V1bfjTebkBGYqHj8fDw+PDx48ePz88Ph9PT09N3P/x4enomkZtWM44zxM3N3GHYsqlCc7f7reLeZdW76QJEgJnvcwKAYY3Zlz8TSpfAJSIR2a2pmYjBYK/aoScq1o2nMVLG09P0d3/3/R//7rsPP2RM22X5rV1XvDQAPkg5q395Pf/6dv26tq/LtlQzh25TPwx+mrMFAggTAfYQj3e6zN0vx8aZirfMvf9pP/RLySWJNnDT3sQSpi55jIjSQ3uEmfVoK0lyb/aoaa2904uEklIv4ltrrp5k2IcQUUopIrptfYS7QWutV+0DZIeAu07z0PK7LxMRGCWRGLbR0A/0ru6N0W8kURISYgwxkXSYZ2YGgIoVHbqb9gZbjQ5HhS7s2R1WsCdht4Z89EkE7HIj0SdUhCTEItJdbIDwfbIWN9hez/l6LPFemtzug3I5/Pi3MpWa53Ngqm2KSBSnOeV8Ml21XX871zXqhFo4GKO5rdu6buvh+ZjnUg5lWzcLq62KkHd3zQHbhDENw1E6u4eZS+KUppTpXXoowsMxEBx2cXKIQI8Iddvael2v5+Xx8fhwmh4eT/NhbqD351VYgO6Jqo27pG5xwrCbhI3jmYkSCePQ9nE0JBYkxsAIDdDR4LS11bWulkVSnqdDiXJ9m67MoQrhGEboWWRKeU5JCKhhpsBh34KA3d74HvbVldr9dpP3sSMCdBHN4muQw1QizwFudq06Q1IRSHQgOBIegCYWyUDeYtvaVu0ci9c3dKMJ5HSU74pMpxM2yXN7+gAp+XQwQFhXen2Nr1/17cUuZ63NzCNPVA5SJpBM20vUqpkDhSY8HOPTAb474rHRssXbcvH1YqDhBqCIDl7jTtd7RBl391obYrirWbejdQgiDsSJuyMICyISVlNwdQ/wQHNAFOBeLXlA03ESMjNCV0YeoEQ3U48sPJWcUkaEZd22Wr1ZUxtW3V2gHDExU3el9D4lAhHKwJpYmwQAIU50/RdDu5m5tj1psXf/17tmeEdkdnxQL7hSSh33m1PuIHnYDQ+ZSbtnFxFA2BC49t5JFmFETimn9H6UiwgJQYRau9UQvu/ykQkARHTMKjGRDQnz99MLSxHHIHYy9RkIJIsqaGNAQQYSypwITokgWqVwJLKI7mrlGAFh4QaOEeigVqW7tfaaE5TQhSEJhyFCmGl1p0ACTElEAgVloqePxw/fPzw9Hw6n2czrWptrszZkxgETC2UI5kycpdDdvCQChhA+7P2vIS/V1xgy85Tn4+Hx49MPjw8fA+Jy/Wq2mLe1rha61mXdrrWtU5o7gGCXVe0AAjVTM42B00QAEsKS0/Pp+OnD88ePH56eng/z4eHh6XQ6zYfZkTzAhiRsnw4MTLkZNG3uCGG13ml9AxiYhWEgRiBRRGgXdUUEYkSi3p1AGpbddBt2CXHv7hIgYQKZpMyZhaG3rN01jASPp8MPPz7//d9//OHHpzz5urW3BfyCuWaGFErnt+3Pv77++fPbr2/rVZsBDmXTDuer2mGDiRAYIUCNHL6VbkQk5o5Zu/Um+5mHiDmleZp68E7CwX2R9sbLGHykNEAu2qMIjj8ABEJiljF9d0SE3ZfoZszhXQ27w6+q1m6KGBB9tudm2E0MCG+HcdfSYqD3dlDAPu9zdUeg0Si3vSYOiLBwRIzcPcWEoP8r4EiQEgPkvQpHAMg59YTDLFTN3RC6/ur+iCB2GfYhAI07Yq5rhGLP5Xus7KiakWcSBQDYt83f/mXUOH2p2tYvP2udrR4JHoQ/PhweD/PjafqjP6W0WtDzsZwKv3KoNtuWtq2MIDnPh3lbNkDfHTFjV3HrFRcSjzYJ4EhhC8vhmEpGJiDufjy9b4NORD4EpHsjIjyi2bbW5XytS5Xn0+F4OJwOFdq3rWwI7TiEAIc+KXOOsAiCHXo9liABJWIm6R4oCMBIidix5xNgQA7Y5/EaLafpeDw9zA/Z7KsUAI6gCAIU4pwk545eButBrVdoeAPd3KDhtxFg3DEW7qeHAWxIleEN4w1iCcwqxQmcUGPbmhIKsJBmtxlJsiBFxAbbFS+Lba0likI0B2d1ViUEElYSE1FCu170Lz/rX37Sn3/S62trS11XXVajDOUIh0c4PACHEDlLSanM/HGmD2GHq8tlxcsSr2d4O4dE99yCTGDSm7R368p9+O/15hwiEqdSAICYc04sgqOPjsRcSmHhvUrCXSiUEMkdIox3c04YQ/qAMPQgNwZozNljZskp9ZfbVPtYbUyIOqmAuzD4vuLGfnIM4F1gmeFfnrWbaa1bmPfeV+wtl9EVhwHUgi4JieDuBMTCLFJKLqWIJBbqfSREEGYVAQ8E6Jp4OpDzToiUhDlNJeecYGC6WJIws0Pfb9Hni2jej6h+wHrELpo+Kn2kb+qSEAYXgUicpjQhhWQxC7c+6A5mYGE5zlPiui26rUigAYQIzIHWvbHUDAzcIgmlggK4t5MdMBgxEVuXkne3gMQySaJEmJBnLg/56bvH5++fD4dJErfzZW1109pc1QAGMY96RjelVFLh+xsZlg370P02nOjAPcIkaZrmh8PT8+Onj0/fB0JJcl2+LPVcvVrYVpe1XmtdNTdGQaA9bTNVbVpVq1nr+KkIRCAhnCV9OMzfPxwfj8fT4VCmwzTPZZ5SmTxAPTxa2F5J7SgpBNe+G7xu7T6jDw+3cPCGwEzs0bFLEUQIfR7je9EuvewbMLqOvBhcIsSEec7TcUpZEMHC1cwhpkk+fX/62797/tu/efzwNG3bUqvHBfCakgtF2Ry+fl3/9OvrT19eP6+LMaYp5ZKn4+xqbdm8AdSQgIzEwADYDMzxPmUkopQztDb8tnGkuX0t5pznaSJwiEiSEEFba9HchwA6Evb4BwDWPTp7W2/MmIhZPJo2vQlj3uoid2+mAc5AzEyI4abWesY9mgcRO/wJY0fEESGRsH1TmJh5M2tqTRWBPaC5uRmEEUAge4B5EFHKzIwRZu5NW4/uxCjAycVc3A2JjofDNE+qVmtbltW2GkHDUGuYrPTVYn1S1WsaRIwuYQ4RCB7gMGL60KknIuReDt+6EbfLI67q5/X689tLXF9pPT8KfTqU/yp+LId8OE45ZylXA/7z05enQ/qaoKqG1miVIUqSaZpKKQDr+GDd1Zo7gjDg5nTRu4jubipcDocpJ4Qw5gowmg19Qh8IEXTzz4kAV29ru75d1+tKRNM8HU4H8fYtAC1COwwVhqBvB68QOCEyDOXzvXIXFCHpLZZAIERGJgQfxrQESN2E0MNzLg/Hx8f5gVsTKR7szogMkImKSE59Turmw6/Bhyv3OG4GiXf/xe2I/f2FgBlQGtOF4zX0xb0ETCE5EEyhVUcAIOBEKsmTCDO5N7S68vqGZjRliMgTJKqNcBWWhGIQm7aw6q+/xV/+sf3Tf9p++qe1XVf25XLWLy/RnDDz0w/puz+kxwd6mEqhMvFxoucUD7XJ0uzzpf72tn1+a5eLZfJZ4lCwZATEMV2+rSv31tqyLNu2EaKkNB+PeZp6iJU0HKA9hrJvzrnQMOPukCMfguS9cxbMwH1AHmDuauZmYMYQgkiIQnSYD8w8lSLEPauIuMGuyANtkED7xKe/KAMfXMQbTOR+RPJNaFf1VnVP0BDg9n5jb5cZxJDNv/HQcfcM7VCsPhrtf2xkhGTadaq9C+e31iCg++GklFOWJBwQOPoN1ocJxAwQQgIREGA7IijGCGhHyMMYBLxvlQjX1duG2lg3UVOL6qaqphEK4BzUnBsiCDgicpLxRrCbUbsHqMa6qq7BYAycsiRHCjJHU7AGVkEr9KhOAKk/DcI0cTrJ9FSm5/n0/JhPRwvcru3tbX19u65bDbf+9JKgMOGeBNq34LN+PgxwInarjpG6AQ6btalMx/lwzIc5zw6+pSwdgkHhoM222patLk23xBkRPPawrk2HOWhTNw+IIHQUj8nt0Nph2+DtXIFTKt2haO8ZBBFDdAvvYdawI/R7UvrPNn7vDQNguINroLl7jzzYm45jxoC3U7+f5fvoof9TwJgP+fB4mI5FCvu6Odp0LB8/PP7N33z/xx+fDymwrnGttMBkBzWlCusGdb3++ZfXP//29fPlWsFSyrmUPBUpKYQIghw9ODsUAA5CB9ZQg/uEPiLUrKnVph2mr9ap6ogBDDDsogKp+9FEt3sNAHCzYZrwfkRGRLh2AH0wcUe/+4CgIksSSgC0+XbdlrVdUqKSkrsyJ9UaroEEQO4DxwrIvbBSt4DoRvRM3L51aqiqW21bq602BA9AA49wDB/tNupwrA6zsaatmanqgIUHBAQJTIdENInI4XgopZj5tlWgcHDTPqwkwLE7HTwoiFAy5ynlIpSIhIMiupsM3sJIP8U6/ZnQYTe15PuGvJq+vL641jBlkXw6ahE/TnY6+TTnMh2QLE2LwY+fHn/57uFy3YDXp4McEpOZ10YRzGgWarZtaxhGOBIwowgOK2noBXlX/3ZizLnkBO6tdyI7dgUxRmjHCAcnAu89stDNlvP1er60WiVxKYVmvt8iYRA6nGli+MugR2hzAAtA431whdjt0QQ5OtovvDfyezKERIyCJOTAiEiY83SYH0uewxeEFCFqDEge0p0IE7MweeBA177P6HZEY5/j7enmLeb/DmRKCKeMifHgyFdpv9mWHYulEpLD1Xxrq1K7kiZuSadScmLAShz56PNH87kmkPS4hshSwwPnw4xG1+sF/DLJVZo+H9H/AFMh0JhQX1/az7/A11c8X217tZ833Z7FnhOi5JDF9LJdrwpvW/v88vrl6+vLy3Vd6iTqM54O/HCQRyTi6X6b9/MRAJIIM0vOvRXn4eAB2mHyOwa5c9aIRQQR3b1FczUIACQh6gIDrbbeGOsVs5uFO4EzQKtVWwPEptp5qw4BhG7Q3wiYVWvchEVwR5whAHcH9n6SAJg7wzdH7+8b8q0Z9Txmx9lix3J57+T2kE8U0TkSg34zCFQymg88VDsUFQLDq4WZeqvaamtNoeMDc845syRi6p83wM1taFowEaKQ9ABHYDtmarcj3/HRsDO13iNJvca64lqxNbIGXs17WqHoIC6BYli7F7VQEOIGYBZA2C28I8A06hbrRSloKj43cEcMdENT1IpawbbQGhGGRJiQBDlhPsjhQzl8Oh4+HMs0I6f1vF1elrfX6+V8Va0QwQREkThEEJEgsFVX09+R3/bNFETdVgGJe98EiJCF51KO02FKU+ZsoULCxMyMjOamsVVdar22tno6IEKXLNBbdNeq1qxXkQ7oWNyPZoday+VaW6yG08MTIMVwx+ozVQ7sfVPa7QFHy3yUF78L7g5g7uDuSOEapI4+ju9eZvYStdsvjhyiA4c72J4oPIIw0pTwEQ+P8/RQZF0M+fH59MOPn/72b3/4/uNjCvPLEhenjYrNqGrXup3r8rb99Pn119e3S9uwkEjKpUhOJIyMjMBBgZI0sgU7ogWG8ftKA+iwM7PWtDUjCkTqYAEEFMIb6KjPTzqcarBjIMyNAAZLDSAAKNzMBq8GAgUR6ObVAwCCxCy1aVNbtrPqdZr4OBVTzTI1rREGKB3uAADU0R4It346CXcKPAMCvAPQmupa27q1tlUiwWGbFdzppRREIImY+lzMqlbtgL49tUYEFspccsmlTNM0pZzDPddsYRa2LtWbAvgY4XWYHQAz5ylNh5ynTIlQCBh7gbOfUfsSG8R2HrPADga5P6xUX1+/IgEzzcc5TQ9pSjgXPRwXzoc8TylNBg/L+unjwx+/fzpfFuQ4HcspM6labBjOfSIQUVsNA48gAhbsEF7E8FDAkMROnbzeh4Zg5h2HtKNXIGhv0ESfVu6dimbLebm8XpZlPfKUcuKZv9kf1ifksFM5O40OXEPBAd3STv7vnU0kJorRmnFzV3cg8h18Q8jR9biIkuSSJkJRhwj2EHcy4wBBFCZmISFs2Mewdqva95hxz8Xp1V7/GIFk99EdEQ4ZEmEJwisvv4ol8xmPzzHPQOJIujU0hSbW5Fq9zN7DJkxHJEJvxrFySoq4bWaNSj6hw/VLAz2nw9epxI/fH58/8HUTBp45Xr7Gx5/gp5/85z/bL7/o5z83vU5YU54hJ7dYF7Ovm329bi8v59e36/WyaGurBJMQp4fTdCwsadro/U6s+z6LdLykpLSP4cwiBs4mAnrFxULDz7qzBQP7ICZg7zqSqg4XTdUd0trTN0fwRtzZa1V1yjnJMAj1cHVTbZ22TsQs0vFWhCRMWbjIGPv1i8nvcQPfhPbeOOvn9fiN3pSP6EB3IhgFVERAMHOnovbOlQjn3NF8QEgExBhOrjccde/cIiKRiOSccym9xdFMzX0fwOGuSjS2ANFuyL1bJ49l3i3k0O9nogiQzd00aqudqeOtgUU4RaAhuqNrR6MGAwqikPTpiHZYTCAAkyQBK4yBlBmEAtEDwRmMooG3gHBCBwqW4ALpINPjfPxwOH13KKeSctbW1tf1/HK9fL3UurgpIyL3vY2m7tYAyB1aC4gxs7ndCEHXSECAwHHaQURAICEnzvN0nOeDJOlPgJCSlJzmKc9NKwGa6rpet+k65aNwMjOzplqb1taqNtXqql1rwBLaU+LvUjoFcfUI88k6z9k9VG/guDucU//1/olxGLG//x4AsgopA5qja6BB1+fo49XeZyTCG061g5b6lJoBwQPJI8gROJjkyI8/nn4I5aO06t9//PTH73744cP3j2m216strRipVruezy/Xy5fLcq7r0t5qpcwlZSokmbGjNA2RkIWpYDgzOoaTB4WnGGPJu8NrTH+IvU8imXmapsM0HeaSRDou9PYsduU46odhR8B1PFpEmGuYhTUCSpKYxmS+l18BuGz1fF1fX1++vnxdtot5fXwo9ThdeS1SckmSEgb32AawW0LFgML1Xtpw/f0202pNx1FTlRmIGWkIJzBRkpRSEhJE6JiKwVoE6JsQabDjckq5lDIVSdL1xjLh6fGIjCTXdVk7eW+vJVBESsmHw2E+zqkkJOpziIgwcOimsQMnO/BrO+4bbhzru8MqwjbizCVxyZBli3i5bHD9cvnt+v3p9OEwa12vy7Vk+v7TaVmeciacRcK4NXNOgFNOjjOEgrtu1qBFGKB3ziWQm1URfHo+uRFhSklUW5+a9SnL3n3H6ONPxF7Gj7m0hYVt1+16vl7OZ8mYkkAQbN825O12oyN6444k6sJEBqARLTpUsrfJQBEtQsE1rOduEQ5hPQdKhELobVsvrwiiy0VVERiGZSYTCRADUgeTdlUUgKAB2aVveqEjsvcT2pGNJZDe0bIA4ADVIGq4Q1tZnGdk85AfgjjSDBaqFbba7Eoaq3k6lXlOZSo5p9TqpnVd1C7b4svhkOLRTrPgwyHY28Phmo+Ap1QjLxu6ErrY5D/MJo8xPbkhvL7Sttjlq71uFQtsWs+Vvi76stS3y7pct7o1N7MExyYs0+n48OkkpUz/SO930jExwtzl1ESk18Sd/BHQAeMDmAABG6A7mAMRDvY5S6/wXT1Ca2utNjPzkRUjQgx2I0ISJpEWca1bNZPuIQvobj0n0NbU+vyuSzixiGRJU5I5de2qPpIzSnr8l0J7P1bvMBLhneO5T7VhFBbQ4TA02HvQwfnMnFMiIncf+jKA3Kmig3HVM3KmzgLOHUKXgNDWTgfdYQNEMYgXY7lj/77DenicszQ0PJi+GVGDdNipuzXdrutmTbsgHCI4hEaYoWmHmFIiyolTopQMQGGM+RInLwzggMGTQEYfrG8EIzCkQCbkhMQkmcss01M5fjwcP56Oz0cSCvft7Xr+ej5/vVzPVyRg6bxE2jMt9wB38kBzvAtvYzPdicgQQhAGYSfdIoMkziXPUzl0aFWXQCj5cJweDLS2FcIRotZl3a7HeQUIs1BT1d7DqFrVmmszM6XQifw55e9LngKhekCAYXfNDUDz6ICm+2v/gH3ytL/C+2ZdAFbBJsHNKDTAAK2zq/qC27GB45yDiHAECCDYiTgBDoRE4cQyyenTAROWU/IGP3z44cen77+bPpWWrpVhbUS8tPN2/vzy+e3zr6/XpVWLa1g+FkDBDJQEEMLMFSkxMYEwZESDkcg6SO9R3G0NIiopgUPnviNCSmma5vkwzVMZnDRz3wlsMRhMY712USNmgc4H6wqlpsxSEiGhQoCGBaiFe6zbermcv3z57fPXz1vdAr3pwbUmvBQpz89Px/TQbd/pNsGgkQzvWDACfF9At0vVWtNWzZoh0K5ai0wkLFlS4szEvSaMvbvC/axHYB7hP5dcSkmpIGMQAKKQHI4HEu7Q8lqrtt4tRiLMOR8Oh8PhMB0mEQHqID/yAciNzgGJO12OznyDztqB91wfxhhfCYWYArFpWKvXZfm66p8bfH86fvd4nDgSas7yh++fMPw4p3NrG7iYIbgAJGaUAsiuBl5pa4hBFNAFttDDW8pSphOEgDNh1FoJWfi+6YQDXzagFwExJCjAwdzrWtfLcjlfpmNOKQklqHfvxAJ0L8o7boqHtzEKISMQOkQLr241TMHS+HydN+fW5agAIxpaZ9EhgVG4LucL/KIgXmtrFbAL5uAglvYZxwCNdGpiYNeXgpupO+yaR4N2hRTESknvQ3sArAodDFiX2Ayp0mQETIeTlRnyFGbetKnH1gBoI8yzJMqHnA+AeMVWdbtuy3bhqJsbbfX5KPnxGJlhmk1m92MD54qkK2oVzz59iOfsPPuXF/jzn8BXXy921habLQ1eV3+56uvSllW3qp2uqY5rA8dcpuPzUz5M05/e3oehTJxTlqEil4XZw82NmTzeOScRDh4OYeaB6kOwRgMpMAeGm2pT1aZN1QadbezLAYJBpi5shx6xtIatdX5Zp7O6WW2t1q3W2tVkoJNnUi4plSRzksTMxBDhZvlgMP0Lob2fD24eZu/IVgjuFMvOnIcdFOPmPqQhA3wvNnrJpdA7O+GBhmwkQRxAgTzWLjIhETAGQvjQ5k6YUe46vADgMBQxOlj43pY4gomC+8zjLl1BqEyWSKfULLuXaBSxs60iwg0s0ILAEQI9SJGYKRCFUyITAzTBDl9QIJcpMJuTQrhZoHsWhINAZibgRGWScpoOj4fj06lMBRDW83p9u1xfL8vbVVvrIl8RuFW7KQcCdPmoRJJTKTnP9zkKInSJzRv3hQiEQ90jkFGEcuLMnPYQiimVh9MHEjy201aXVhcEigDVVrVCV7ay1rSN6K5Nq3pr2NoM/pzoQ5YHSVXtqpvJIadpKsepHFhKBzBGvLepe99xB3EP/M3vJwoA64U8i86TJzdkQ3bgGBhzGN/vHpY3MswgJqB+jI8I5e4eJgKHOZMd0PjD6fg4Hed8yjjBcd6i1q1t6/Xl8/XLb2+vr5dqYZKw5CkTsxnbKAAJALxDtPvKDfNQU3U360v6/l5ySuXx8bpsjEufxnXVh4676aqNXV2RuQOrkYiCcNTiAapdwApNW6urtSqERaRkQRYMuFq7rvWyrMtSL5e38+V1q0u1Vj3cYd3sIltGgKzmR0IUls5339/Fe9oFAGYWBk5ojvczkn2m2wPugEkwYcm55Jxz7pJnPtC0burRpTxGNTPkpPrsjYmAKKjTZUBSmhARsZRcawcE9YYTSkol55Tzu8TlLoC44+don+MSEXo4hu8EPbpD9I77IFCta7V4s4Bm0lSawubU4iWffz2WP346/vHTw3cfT4dSPj4//fb55U+/fv5y2Tylq2PU2tYVxFGGUIMkLhMie6uu6kSRspQpMTO4mFKrW1vWcCpZ3H3HeuKe197a5oC9CerhEabYmq7bpmrTPGfJ97cxeO0dYIRIRMJEibAzghNxQhJ0jM3atW1vbXWIxKzRx2juYeDgCO4tbKvhYb4ul7ausbyZ/DLlIwVrXQfXKZCIhZmIhDAzeqe9xc79u+3IfUrQH/gYDrKRNOJ2H9rN48vVOqy1KWzquEZeiIWOp/jwIxyfaAIMgsbgEjljyXKYPpzmPzDlZlurr5dzvbz6diEICM0//+VzPB2/+whyPFh6PG/ty6/45UW/fMZWAaIEkBMChYXKIT78Mdpnx6ub2uKxVN9Wa1uz2tHbAxuOjsumL0s9b8pyPBxmPL8fW/tuAjNDVQRgppQzy5ACjBsl0MwiWpD1NMg1wM1B0Vuztq1am2lHxaIwdX1lkaEOthMNwj2aNXPtRaoDFuSUhmd9Tqnm2lRV3buUMbICetNaKyPJngef0jdqIr+v2qGzkN1gTAkDESAoaBwct65pp/XuHcfYWVUGBOot0BCCaACOU2AzII3B0GbqXGWgjry61fRjZLF7awAgOO5kntGK3/8Ed2gN0jecMYCaWEEUoqEHOVRBU1etrTUzDY1w9EiMiREh0I3MxY0xsSQSCiJBRyRmArY0BxZ3UjdUdTMnhDJJEk6JU0nlmKfTVI4lzwUC29YuL5fXX16Wy7VtGzFxEkA0j2Vty7p1Sb/OCEhFCrOUkqfpfp5IiIJoXRwuEACYEBnCMBx7zocoGDtMlzClfDo9l2lqttZ2XdZz21ZV83AzbUjgeDdiVzNzU1RNpieGjzk9plSILq2eVeUEcyp5OkzTAYQdIUz36eJd36G/mv2D/x5EF7Bc0ApDQhM0QAuKPqeE6NMpGHMfj7AYs9lOuh4qhwOeFhFh4cGBQphzFpdHTjNyDkk0xTR5rcvyuix+eV2Xt6VtakTAzJkPedrIKjQIR3dERwwIRx9NyV5Md8aZo8eunDy2CvN8PBAiuNdKXUE5Ilpr4JaEhElNW9P+SIgSDvZzl9sLVQUECDRtbdsoPDPnnFISJw6ztdWvb29fX85v5+vl8rosZ04oWRxZAzaFrSqymwCEE1IPs7fXcTubsCvAqxoAIarS/U4fczfA3lRjYuJISUopJWdmgQhV7R2IbiiBhEmklNwFKIm5f4sOnQ0ip5GpMYyB9DRNPYOF9zZbR/cTwA6Spx0QDwAD4N3bD4CEYLduEPlevN+tqzDT2nTTra5qS0vq2cBX180Y/FgI/Ifnh0M5HH747uOHp4cPT0di4F++bpBg7UpUV09BmVgSMUtGAAfy8Ka6ATALpsQi4kYQsJou1yVcAMLC8eZzALdlijfvgDGJ93DrohqqnRlP6ZtMq696HJqBhMiMJAiMKEAJOSEJILmGrbZdmgR4dgFEA6foZCKwQGur1Rdrq9btstR1VXDbCO3wLOmgdR2BCff2KBAhJgZnIPBOar+lgX1dwY76HBA6DEJHNCC9B3B4wMuitakGqEZTh0vwGfPEpycoMz08c54dUgi4E+bEcymH6flQfvTw1l7bBuvFtrPVlShJhfp6XmdJHz8JlJMRnM/155/9pz/Zr7+YNurqMSgos1NRKvb8A6wOrUZorOrbpq2aq/U+2ygOPQJwrfr1sr4uNUhyLt+cWQidYgDuYIZIKUnOacolJcHRYsQ+fFf3zaCZa2uh5hHVow51eFVrfe8wsyTOJZWcc0pdQ2QsDbMwNfemHuHOwR4OQCRJBJNHKWZNVau6BRiQeqiatrq1Sj1MEDORfpvBfxPa3V11FLc3jKSHG0BPXjoekJn3nt+Opez6GNpq24JdYQN0lpimMpWcNaU1AluzDTHCO8dXhugdISFP0xQevSpAgOiIlPvG7wCX3Hd6R1r5u1FiILYpN/ZGoBQgKKqk5tu2XaO5rs0dDTkKcwhx7yBb29ZICAkdA1GAgZOIzEwlYtpcmkdos+Wi29l0c0lyOEyHh6kcSp6TTIwMzVq91PXrurys9dxCkSihYBA1863qstZta0HAzFxynkous+RiDJvrvUIrIQp3rNZNypOEEZgMkIDDoG26rjVxZZIub1rSnCS7HzUdpzRvadm2VSRBtwCwMGudrAcDdQ+FoBB+V8oPcy4sm9lbay8KR4ATC3FmzsDUyb23oG5mNqTHO7B2HMzwrUJgAKyb2xZsDIGGYKMvA9QV5UfmZgCOAwstg8hJwoDogBaszgbiJkaiRC1iNTFP7Upr8ZwVTFder9v56+ft9VWaP0gpR2nELeeaeE0AqAEYYBSOoQDGABhGgRwRGM5gAMbQFDT8Xvc3wl0bRiRmSNDvfF3XxARJIiYiSpK6ns6+NfqMBcnDIsy08weiZ3UoWURyCZZrrb++vv7jzz//pz//+Xxetk3dDZFyLmkqCt5ZDBbBArmQJO6aoO6h2rrwHCKaWReH6Nu5qwQ24/gmtA8tOwAnxlwkl1SmVKbCxH2qMMifrurqYYlTmvJ8mOfD3Lvl/aXzSNAHVWZH/ndUAor8zm9w1AW4T9hw6PTvgnS3Ld8dEzp/bt/+HvejdjCP86Lq0QzBUCgTuDa7vCznL+fQWgRTwCTpdDidjqc58+PT4cfvThjtvALYesTtAlWBiFKZZwcObE2bR+0MczWt25YTlSxhZk3rsiyXFaIQszkAIwh0UmxnbPbtggRgLdC9q+10Y4ItwpBzSik3fD/JOjd0QAvQYXD9Ayl6XxyJhSJxFAlCq7YG6Oap8/LYydwC0IDUTVXXy+V6fjtf63U1RshJEFKeQpuGK0IAkAVUg+bhHkLgHISjgBgg+J6w79XUmOsEAbAHgUrH8XX5OwDwiLV6a+6d/2LuEWrx9Vf8+T/J8SF//JTzg05zFWwGPqVyyKfEBcC39rpsv3mr4pgIOPN8fDhN3z/nPzwdT4hNm1P6NGX8+MRgW8mvBluaG8tKzGuN8zWE25QdJ+KCi8FWo2qYByFkosauBs06aRda8+tlO182VQT8JtMiYk6jUETmjksm6h1yl11/rXcb2V3QvQO4KBzh2ncVpSIHbcm09YN7KjLNKecsKXXevA2S+YDoEHOnmhEzAKk7mSemlES6fYO7mlf12toWUZ00CDwIQRiTdLGP9/Pq96HdrLPlvB++/t54cFKU3tvfhbHuun89gwYgj6QqG7NDITrSdCrgkldQl7VSA/eGKXHKXWdsqKAx88AZjkmkw94vHBqkjrgjgOFbUPzvdBsBoSWuwDXIgIOBjMlDBCgadOAheKgDBRAkAgE0c2/VCBRdgiW4k/MOhynNtDFuEVpb3XRd6rZ6GEgp+TjPT6fpWKQQcLhp62avmyXI5TADAwg4h6JdrmuLBaoGd2PFNB3naZ4lZyRW8+rt/t30dEzNekkWgdJHDwIEiECusa11XdYsRSQl7owpAUCIcM5ZUuIifO3NzGHaZW3QeEqaPLnl3ESUP03l0zyD22XbXpu+OqB7gwjobC4CQCaOYbM7itsB6d+7kZ0zQfezdoA2MLYAiDZw5HuRA8M/NUY/MKjvJ2TuLjSGaEAGqUaukR2SRVIXhViC1aZt5fQWCRW31kQvW3v7guv1gTkdDka4EV+RrgTBbhAtKDAYAYIgnMLJgjsrRDAytUA3VLRNw+8Wlru3tkFEEkKEZuhu66KepOs9d52drjyDCLTvkF79u1nEMG8hRJZuIZOCuFq8nK9//vW3P/3lLz/98pdt0whKIjkXlMJ5ouZALdABXRJNU0rSlSrAo9MZW3eFucX4iNDWuqaMfdOogx2o6ATBwtNU5uOUSxJhj9CqFubg/QcgkHAqeT7O8/EwzxMidmR2RD8l0HeIFWHvMAUBAhDKOB+6rk7/bB1D04UpEckBHEdWgNCF7cZJS+M3xnf/HQ3GA5Y6Ut7CaU6Th27n6/Wtfv3lNbSVxD/ncijz48PD4TD94dPpYZIPj0cy/fq26BbPE6wrbsKQ8nw4GYrFWusCXfEzQlWXZc2Z5zmHh7atbtu6bMSUSnEAYERBFCAgJJGUJIkkQrTY1IayG3hE928MQ04iRRrY/cvoqI4h/rnL342zLIDAe1yfJBKZe93AtrDEqUBS7wbH5Ige4BbrWt/erufLtlYXZp8i5Q2ZOkmvE5Q7BM48AoAJmDox39/JbdDBrHty24t5jAgKF7eAwDAFaPttgBqYjW7FGJ8pXF/wtz/Rx0/p8ofCGfNkQEqAU5oP+YkpNd+W+nLdPrtWARYyKvJ8enx++O6pfJp4Ur3UFWc5HKeJPkmSS85p81csC6BDNH0toQ5OQmbMJuQBVb3qTRYMM5Oxu6PtDedtq8u6bWr1W1ISCXPuhoSBTMCDo+sehs634ejAzwQCJIYc6AoVQRTdMSUJTNpaa7WrHOYsOXU66qDOD4J7xyd1Jht16TlyCFXDQEYi5pKkV/nNrNa2IQpEIlDB6LrU3HFrBlD/p0N7r777yHQ4tQT6AOAN6coY2luSRBCpd0g9Qq3qtm0e4BW45kTpkOnE8pgTJm9cLa9rhlANTElKTsSCw8tgr8r7B0CEnU7QR/v983CfCsSwXfpn8KD3rbLoVnVrVjswBDrjVGQ6HoBFcl4uy3pZ1Txag5w6u5MdWmhbmoTkSFCmlBMwAhOCgIqtra2uzQGhHKbT0+n4+DAdD6kQUEQ4Rkooh6en+cPhYXp8ODxyZmffbL1u1y9fvnz+/OXz5y+v5zdA4CTzPKeSzUHN4Rb/9ouRMnIDhz4hijGRZEYGDAvXti2XM/e3kXojD3ZjKQwiTJknKGBmEG6urg3BS5b5UFDotC3LeVIJIvtYpkMpL9fr101fNN4Qsa3H9by0pakmLkTCHAAQGvsshpjZnMm6KIFhJ+nTNy8mPYgcNcQCLYZP3WDEj9bL6HwjAKAjOo2ZkgEbiIEYzcpzw+JcnJJjMkRwAqcKWBdA0ziDEi969IXmOH46VJ0q0NngRR20bqrkiqFBFhJIQYgUSAQMJEhEBFk6XJjAvx21g7vVujGllAQpAtq1brVWiDKVDAMnwyCDjcaMxENFteepAOg3GggAIRrQurVlu/zpl1//459//vzy6hBlzolLBHpQU8DN3aOL+DDRNMnhMKUkvYy6IekAonfdtE+M3LdamWiaJiHG+m2/C4b42mGeTqfTdMgkZK6uFuiBXa8MGTlRyjl3Q5dcShKJDqbdxYpvQoa4Q/R7z2/MzfbY3tuAt99ioa7Np2YEQYM5jdSf1o7q6WkRBgzGxD3IFEny1BOFh+nhcXq4+NtFL9u11vN2yOWUj6Dp62/bv/uHP22X63/9r//w93/8UKQ8Pz0jSmtxXhVSedWwPE+Hk1PCyKBkq7d1qx5quoQmoZJTOEa4Wau1Zk393OfEnIgSYpfUkpRyTpkAtLV1oN4RIsgtujoUEnEShHsNi0AMEU6FOdOwpI9wi5HzuDNYZi/iwqCBjuR9vDksligRCwkBA0iE2JBJJuTuGhmALswTECNLnqeUs6QilHq9Hrtc9BAbQxxi9qMzGnHboRTA3S0kfLt7HZiIkKkzGSPYKTyAHLY3/PKT/fnfrxr6TAYFUYqU58KfPKi2t6V93fSt9x4imDifjg+PDycy29q1btf5IKenY5md5bItL7a+fP38+nJdzpd1XVZgk5TA2DdrG11XXKtvGlVd3TuoowhBcPR8w4EEiSOone3ta80e390slIglZTbVrsZqvZIJ6FRvD9DbsM69kB8TPAo/JLDNXmLtI/eE7CxVmKjn9IEIqtqaWkBPcyPiVhF10MaYq3XiLBJ6VHd0R/fh6haQhZPQYcrqpqbQ6WnETDjz9V8M7aO/Pdw0u8IG3ppvo5+GQF3mWxiRPQJMXZt6M6/mG0lNpCknnB0PKSYBypSYJqOCtHD31UkiJGn05GGMpaILfQ1IMeIutTHkJhHBrVtvwYDt7+je+9kVxOZb9a1ZDXP0Lr0IwJg5s0jKGYG0eV3W7rXF3U9QCJqramigYiruHM7ghOCCIWCM3vevHE8Pj89Px8eH6Vi6IxwCCKZM5enw/PHxu08fvvv44SMlMtTLen59+/rrL7/8/PPPh+P0629ZXQEpT4UlbbVF0+65ft9/IMDh3zyez3sjEwHNI6yt61UolWnOZeKUiYUJ4l3ZgIgkcUZophXCu42jpJLmIiXXLU8cup5xTbMwEyzqXzZ9c78QYrueltfreql1o5SEE2E3O7EBiPBx4+Gh1lqrIpxSasrvVkoIckQ6goo7dluboO4ufxPdHhU8QgAFYxAFiXM2zErZsTgdVQ7GxTA7ZSABJApEC2+mZr6gX72BaxD5fACb5s1xceLN2nU9XyOshtaIBuJBgQxd1w8DiYiFKeGgEDVLqk2/4bX3XhERinCAoYK7NdVsMgYUPgSRO98tItysA2Y6m5ODqSeuHU3t0NTertffvr78019++fMvv1XdkvA8TVOZtxrramZgGsJUhCeJqeA85XmaEwtBwE24fEegmGp/9e5W1y3ljIiM36jRBUTHwSSW+TAfj3Mq7OBa3cAcuoEMMjELl5Ln+XA4zPNhujH3PLDDkdTsFqSw61FEb9vC7T7vkZWjizFkd9kdzKwL2Pdn0nUS+qfEnQ42sJqjJNx3B9NhPhKiID8dHh/Lo77UtjVdFaqXkh/yAztfXtt/3n49f31hh4z8xx+eT/PDoeHjKT6ssYbYuS4oEmRB0o00o9s6gpprxHrdrnkVlt47cFBHR0FhDsI8iWQOJQTsMiIsPBCg+P7EzaNurVWFHVNw/0IQu07OUFUO3JOmgG4zRxCMwRg7+qqnz2QQLQxCCVKC6ATGMd8gYgaS3nFzRGNmQCEASTmn1INEGsLErsMR/KbNsHPaBzLjRp0iAI7AIVh7d2VhGn1e2L0cAwBijctv/ut/tnTw8hgZOFGhmABy1W1pr0t9q+1qQQZoAN37jAhNa2u+rpUFkDRnI3id8pdCX6FeL5/bX35efv11LQd6+pinLBLQqi+bLS2q+qZu7kIoDEw4JR5y1hG5YDkETfVKry+BAf+L93VFSMzmFg4RrhZNWcgIcXg+dihsBIMXhjnjU4oP2atpja04lWDKEiXtkd0jwjqZrWlrhthpYAgAfUJHTEw0hOhhTwjdLaABEIA5EQITSWImIUL3aK5dXKofN+kursM/Q8hD//u9tXhLuzuXMiCkM2UHhICERlsVEBy1+tLiXNjmA6YjyCwu58W3pZFt+LrAtYK5IOZep3cH4w57rh3COFrx0JF2xIzdJdYdAsmZnJEM1RSMGQCgz3m/EZqFAGwO1X1zM7DAoJtfBSc5pOwenTG4LVV1q2qHwzzPE5VcpowElBmzYGbICAnJJEWZpmBOD4+cSzkcH+bjcTrMZSq5yGGeH44Pzw8fPjx+fH748PTwNE1zzslBq27X5fBwmuYkhZnMwXxZ12rGJICcExILIBDKNwj5Ht0JsoBpQDgR8EhzwCHctbZ12dK0XqfpkMuczHd0buy+aJ0W4A6GBJIllVTmiUSCoE+luCm5tW17ifqyXF/rtoIq4tLe3pbP5+uX6/YmJZOI71VaHyF6mLu2Vi+X6/X6tm0LMaWcaT0BfH+7ES8O2YM8wAmQIhAGFnecHAhjrgqELmSSQqaQo8sJZAaZgCegAiwAAiBAXeIFxQJaYHO1KbSiVvItewWsRBCkTnjWttbVrst2Xr05mXtg75J0cBdiCIdz51aiY2Ka1oKN+E7HTVKaTw/Q/V3cWmtImEshEQtf63a+nPuYipkQ0MwAIecsScbsvUuRenQr9tZs2ezXL1//6eef/vLl6/m6JuGSS19S7lYJGSWzPJ3SwyxMlBgPUx+yDCq8umutTZ2YYyfsMVH3Bh0yf7/rbkUg4TTlY5mPhznnZN7aMKyP6CT5CGIRkamUw+EwzSWV0hsDXS7PRysMvJvKDIhbvJfmcJvK9A8wTE9h78aLcIejYq/acRysQozQm9IR4UPJQq21doMKAoCIfPfhueR8mOaZZzH5LX7eLguoHVI5pjJRQkevelnWdr3+x/xLxsxQ6IfJ4UASRC1sratetu1y1RZ0Xdr1cl1eL95qQhLJAR4Nzq/L8XiY5znPOc2UZkozlzJDUKttvVbdui9Cqw2YE3MQcZIU2tS8zyuX5Xq9XmqranrfDxrbaXRFsWdCg/O3n8EBqE6bEyFZEBAxdmT0HgTCPDTCEJxZk7Sm7oDCLuwD+QGMIA7h4V3fVPo8JdB8eOvBrgo4oJkDABoQNgoPfF9E92uKEKZEinwL7XBDQ0HYEtevsbxCXbGcUk7ZYb1sf1nbeq2Xbbu0Zh7WABqAaf315ataPvCDQM5pypkh1mgL6S9P86v8/fL0gJ++n0+Txva2VlxemI8lFfHYqtWmVt2au2oYuQVmRmbMQMhgFPMjPnwP5XttD1+vRQPv8IAdf6pmah5OABRA+93Gjj1AiBThKdgAa2v1sry9Xb68LJUanegBy/GUJDF6t0puqo0UoCI0ROxaFNAtDSOESVjwXTf+XcOBEB0dhr4JBoKBNXW37uCO1kFx7kkalH8ptO8Yinv8WgSy9NgZzMTE/QNRQHdrg+iNITWsRiukSHPOM6YJQFqNptW2S5wXWbeMRgLDU6H38XoXsW5bawNxgIgoN3H4rr1rPX9Ep35CUHTNJ4a9Y//N2QUtoppt2tQ1GFgkMSUSFknCySJq63psBu7ATCXlw5RL5sSBAQzlUKZDSXOWxOEDJTNFSEqlzNPhOM2HUub5MD+cjs+Pzx+fP3734fvvPn7/cDxN0wQQ5q22lTZzE8vpUMqpTHMqhdIWFZpaGFj0Llh36f3djSCAEGbBDoHtwwGKjjRwj3CtW13WbVm3ZarHlBIS93FOhEeYeR+uKyJQZmIpc5kOB4eotfrW/HLhWlOAatuav9V6da2kig6qy/pyWV6W5W0+nMRztyyJcAAfPiNutW6vb69fv3x+O78AQsn5gN9/uIV2BJdAGTuCYvSeul9fn7Fi9GYuITCDSOQp0jHSE+RHTDPw5CSOrIi9qd/b+R4RNND1vfTBEHLjUAJgcACzqJsusF1tuepSw6KzvQdQGIABmQAInDvBmhwJMeeE+RsNeWGeDwezHmNCTYm5ZElMSNiaLstCnfadEhOp6qjOu03y3pHpm7KZXtf68rb+8vm3n3/7y9fztTZIsstMSUrCKWFinpJ8fJh/+DB5UASWRDh0IbgfykOwovM/AxBHQWkqu9bTt6yYxKVkSXQocy4JCa1p02ZuXcofCJg6fzZPpaQpc04kNFQ6eij2TjPwYedwg1/tBwncYZMibm8aejdehJl5hPmdYwOju9gbwhHeO/sDuGlq9zdCiCXLXNJxKsnZN4umqDZLSid5Pj08HY4Ovllty3Jd6k9/+iqYisymzILL2l7f6vltu7yul2VTXJrF1qyuVZcVIKTjPUg8oq3Wkk0zpMzTMZWDpInmYxHO61LXS12htg5Q8GaBBMzCKSXTcFcA6ojLZbludVU9fMMj+abtHTsrg7r0EzIiYQBbcHMhZO+tBUDuoFXoY5AOcVUIJVThJhwWXbLBkAJwd1SOjpIegToCzUENzM17B6V/JhiTshvfedhU97cYd5lbf+WIRZhx/0v7OuhrYUpYAAtEBshEiVjtWvWybNtaq6mFA2BggkjeWv16fglP/FBSngQTE2m1DVbYLgTL02MrUzo8MDTYXupffsPzKrYlIwzrnCobAE0I9QgIRmLBxCiJaIqHT/jxj/jwo+Hjpc4ed/dyB/ccdxzdK80c0UyHy6UQdik4joh2XZYvr1+/fP3t5bXiwpeZuXx4LlIypn5iVU2bNCasfbLGo5Gmqu6RRFi4V589DXvv+eAOFEdwDA0L89bzcIuOQ+skiWn+l0P7yNR87FKi3VSGunv68J8ICLNuOmwQYeDNW/XNuFHGNHGeci5TzhMLQrj6plvVitocm4OahjYkJDaL2uq21W3bzKyfWJ01i53+vrtH7CJ0jtwV1YdcQ8S3wFkAgDBrplVb3bamm/5/+fqzJklyZUsX0wmAmQ8RkUPVHg67L4V8IN/4/38IHyhCitzb95w+e1dmZET4YGYAdOADLLIid0u3S0pJVUlWlns4DAqorvWtJHn4/HOZOCVEOlAAe57k/HggpCRyOMzH4zzNOZc0NixOwolHeMAQoO0HMGIiCSJO+enT599++/23r799eng6HU+H6TDl4hD35WbatG9bva/rdbnd7pfrj28v3/75/eWP72/Pr2+321IbCYOQDd0RI4vYF/v4nYzSXhB9zK8IGEcbJWCHSAdqq3Vbt3Wa1pQEKY1IDx+zV1PThgwpsxSRklLJklJvLbrq7d5frrnahHmNtpo1AGVw0oBuTq2v63Zd1supP2Uvw9+4n+UhAjzctm19efnx7dsf18ubRyTJnw/509OH74MiCAlkSPLfeQUAEBhM42MBIjKhJCoTT2fPD5HPkc6ekjur++atuqm77ZIr8HG2UA91b+bdoSm5MTamDXUJuDpcY7v6evdt9WaIsqvd9ov0SJFFYPcRTeBhgAyUSIp89CIiUeK0w5QIAiClVOYDIzJGAHTVn0kTQy0/jleqPRFC0Dihx3jTFve1fn95/fH6el9uap1FgqIZVCN2IYnDTIngkPnpNP/26VENW3f3ZhED7RZOgZ7VAJCE3XYwf0pD17vDDT9edgHgfD4hBipk5HDvrXbt6mbvfj8RKWUqpUyl0DglhI2LjaMDBe6urn0evm9J75radzf0XsvHyx2IIIKYKY2BnBACJBaCfX7002iFEPQ+7XWP2L+bX+Zupnr58WNhvnJiY9iiLcvD8UiRZ8UvD09P58fuet9WQb/e4n6t/+d/+9Zq//f/+Oc0Zwv//vL24+16XVtzB5YAQnfuVgaVvRsDcUoO0MNDY1tWwDg/HqZjToXLLCWXx6eTVr3A/R6beQw7HDEmSLCPN9l6AIJq2+q2res2bQzzx/1ql9LFHp2DRCSIgsiAHEjj3igYiSIREgNyBEMMpZUgI4B6d+tmGqBMfUTFjCnJMBx6oO8t9HBwjegezaAbqg+O97t9Kj683pvy8K/2w38p7ZDHxvO+IxOjCKXEpcjDQ/r0Nf/1d/rtDCFbs6XpWn3dau/dU+SCmSe0YsF1Xdu23XI98CcoU/LG6rTcZbsfbH2ACE53Ys/p/vvXxf5v7XDA//xO20rLG3jVgziVEIiEuBFWA3foDughhKXQ6XN8/Rv97d/4699gelDk9vGYMuKGSVCIx0lUiIUZkNSid1UzRARhmiQLMnar9fb2+u37939+e71UMy4smD4/TudPWIqi9KCqUZNsIi31d9waIo4UzaHAxd0Gbg4RvJ/fGRA19shUM2/aVLU3NbPw/ZY/Tr+afoED/ouvff82B43h5wzvo88N38VPDmEQEK5gGr17h+x55nJIZZ5LmXOaiCDcB9TDGlgL6B5qo9ugFkh11PWBthCRsQ2JjMnWT1vM+5J6fxO71AP+hws7QAS0rfeqbrArEpAkpTKV6XAg5oAAgsAQ4flQCFiYS8llSmNmPWZJgBiD8eJmoQ4OggSMOHandJiPX7789re//ttffvvLw+mcU0aAcNvqWtd7q2ur93W53m9v9+v19nZ5+f76/duPl5cf9+t1Xdat9YGXM4ygQEZJOT7swrF36EIC0xhR7iereA9UcQ9Q67Wt67pM5S7CgEUSDRayuap1054HN30uMhcZxEBVNMet4bJxROa0gHUgmEqZJIAEGjBIkghXa2rVrHnAuKm72xBQmmut6/V6uVzerterWxBJ8S/w9PG5h10pCczBCBGoI9GLgRgyAdO4sqNMeDjS8RHKo5dDSAnyXut9rbdel16b9+EZiNjZ9KhG3bk7dxAHCRBS9B6wGtwh7lEXb5urQhBRYhFhIgAMIJcx1wtCJBhFxCPQTCLyL772d7UaSErM/D4wTiP6YAwoCGg0EhCRhChiRDQSMZJ31cEp6ubr1l4ut+8/Xq+3u6PnwiiJWQxYgxylFE4HyOiHTE+nw+PprI612dZWdwWWMbNKQNkSECGKgv7k0o3z8TAy/DLaRTgej4hhm0b3rr1r9FCFoWdDJJKcpsM8TVPJOYZnEmFko73X6lGF450q//5E7p0/fIcO4c8W7g52AWLhlER4lDAcf/c+4Bk4P8fRi9MY2b/hNvbSj8+6m233BSLQkQypoameT4fjxI+UPx1P58Nx663cCaIjeK3bstT//Mfz6+U6zQUJb8u61qYeiDgEKinQiUzSSE4fQDjHoZENbZ0mOEw5T0kEiIME52N5fHqw7r2bugEiJ2ZiBxABKESoHTXCVLXVbdu2WrdDTB9b2x+LKY5IAmYSRIFAH6UdgAkHAJgIgcEp9v5mogFDVVNV7e46aCC0745Dc88BHMCAhAQOoBHVvWpUhdajq/nAqf7Ze4nRSXufpMCftyj81yEPApbETgADV5y4ZJlnnqd8OpTHx/L5c3n6xMfkt3hpbVPoPTYzJ+eZyykdykSOiuKBdetNfdNYu1fzpFYwMjraZgHEQpKWlKqwPZ5jrbZafVZaLhjdMwUmYIBEkJiSRrddiC2Mc6GnR/j6O335TZ6+IJRxHv7z0zBhYoaRxr4v0nHaAVM1c1MDpBEAZWatrv12eXt5+fb88v3lrXZLJcNyTuuPeWLhs2GuKIkpI2fExhwwkGyIiG4+2uGjVrqrth4eiTlJKiUh4shpbKrN3SxcRxdh0ET2g/b7qOB/UtpxF8bvFBr3QNwv8cOEOiRTe8cOCYksvIc2V8WeM8znfHiY58OhlDnxhBAOSqHkHQxdIbqHGoab9oB1kMlth8yivb8QsZSy+6w+nh5jT0pAImJilz189sMzHxG3SzXzxHnKOZFPpczzcT4cp8MBENWsq7/H50B49G6qy7pBWjEVlAwkEIDuoBqq3lUDIuWc01TKnOfpdDx9/vT5r7/95eunL6f5xCSqZta11229L/frtly35Xa7vN4uL9e3t+vb6+vr29vr9b5uQSqZMkrt2ro6eBBI4qBfbldm3qoObxE7pxjxpzamZYEeNCbp2lrdtvt9KUgY0JMyDsOzqnmPMMmzCEmSlBInIaRAKoEN0AAFIJAM2SQfz4d8LM03teoY8+FwPJ5FJMK6NhiIbFP3bqZqfaSIbdvaWx/f0Whe/bLICIVwR13R+8AniIIYUqJp3NoJWTAdcH7A0yMcHqFIYKjfb+vr9+X+ut4v233rS+2bWndjAuTwpFE0TkEn4MJSKAmAhzdTh+bYQnsMEzJlSkcusmv4zNEId1444niHoNC7947uHP4/dB1HUc8pJ5GxYsM0XBMzJ8Z3z7eZ5ZwBoLWmZkM5tbVuAUhyX7fvP96+ff/x7cerg53PD1IYmM2KaiYSIjmf5qdjyeSTxMPxMOUpgEqGoqlZR/Cqmjkzc8mFxAdbPKUU7qbqROMMFxHWANafWzCUks3y2qybtjYuxe4MTkCZc0p5mufjcSp5oGlsYFF2keDP3rn97NciBv7k4BCN8zZ9tKnjSHfbyRgpCRGFGwCIsBOamfbQHdNoQ2ON6qAjsgSEgVLiD7kqiMhAex6XAwCKcDrw1+n0t+P5KEkQZLEIRjpMMy/bWlsLDECo3ZEo5Smlef+hwP7Z3GOHERJpQA9XAGZy8kBnAmESJkLovUYgSzo9HuvWamu9qwckYUIyswBMKQslwqbaPFrvvdat1jp/7Mjvo9IRS2LhhMjCSExBQ0W99yyYJUuifZz03rxCYkoBFg7drDfVES0ZhugR7M7mmaK45wghsgH60oiqvnRj4qVFbaZmP4fpCEEcLMEJfvVLj7r+r5ARRMiZEDAlPh7yw3F6Ok2P5+l8nM6HMk9pmgSz29qi33vCSMTCE6fE09fp6fP0cMjJsCN6GHATNH1bvtXeyR4LfHI5JZqYEkBRS7Vdul7W63J/FY12furLQq8/0B3BghHnhIXhmMEc1aE7jIJcDng40+lBDqcyz2hMZvKxfDBSZmEZF4BxAwi36Ko7oG1PaI3W2rWu0J/t7Z+v355/vL5dllWYHlM5YcvbC98A+4o8EebEhagwcUpiu4kAPHygxQVxt9A6tX1wnktKU8mIqNob4ubBFCwogAysPA7WuDe93Ik7fDBV/mtpZ/7TrY7vg5W9e7AL63YBFBA6gEGoq4ECh0x0eJgO53k6zKXMiQtEuHWOji6hbj2ie5hBoBPYjsyHn+2JcU8dsXrmo6W8U/j3Zt27tgMBGQkYdKgY+cNl1+H+1gKNDjmLcIKUCrEgMdKeYNa7rWura7OmI2LP3QAtFywzlgOlCSPIDXuP3q1pD4AyzccD5jxPZf786fPvv/3ly6ev5+ODcHLz1mtr21aX9X653y732+v98na/vN4ur/fL5Xa73pb7qqtTcCEhVMSwkbvmwMDyax4UgJnXqgNLhUAE7AGBNkzJu7UGIsJVW63b/X4FiIiWsxCPM2UPMEIAKCKUk4z4IAyIprZsWLtEjLadS0KR46fP6eunET7k4JLTp6fPpUwAYdoBKSDMh8hEzXrrtdattaqqPw9i70FV+2vkxTjATz08AXLA+yRxt+knkAL5BNMDzOeYZsu9+rJsz8/bf/7n5e35dn1brku/bu3WezPjRJwRi9EpxCizlDkVTIWBMAI7AARyMGDClLn0NHGaRYQIwQ1id/vEHi8wPFgepqGG7vTr9xH70BjfXX8xBLOq4UYIw6jyzmiOwX5wd9U+mkfNTA00+uv1/u355dvL5e22TrM8lGk6ZCJunRuyEAjFoaRPD8dMkclKzu8qWgQCNFJtg/1ARMRBAKYOEcIS5BFhZsN+Mkrrr085AqK5t961m4ej0ACHC6dS5nk+TNOUkozrJBK5qnlA4J49xQQB4WiA4c60X2LGN7tD7ffY552jbyb7DiOcOAV47zqYCQgIBrb7rwbJw0IN1Eh3vkXOiXP6wMAB7fp6uXgYuFMgOyeXGUqa+HCeZkIy88BASWWaDjJtsrXWzS3C3AEhpzTUxcMIxLRT5EZWkgJu2pdWm5sBjNVCFuQkAIIYps23KUmZ83wqhzrV2sycicHR3NW8pJ1w0jvWru7eemu9fhCc703JscIQBmhWdj4FuQGMvZeYRXgUAID3QI7hmkIcAzpXba1rD1N2H2eG5DG5z+6Tu7zToHdkngc0jQV9qVqbqg6bVsAQLQtIMkmGIaFj6QSSDzMivEME94rI9OlxTolOh/J0nj8/HB+O0/mQpywp8Ris1Luvt6gHgBMzCVJhklkOn+bz1+PjQzo49GprV6WOdeurvnWDiWZCaE04H6b5KDI75G2TtVntS9UJOA6TlsmQWneOTiMwhAEKg2QkQgNUREiQH+D8CeZzpJlYRgpS+vh4DIEh0yCeYwC8M8M8YGSkIQAyhIc2tXu1tunb1m+1q/Yp5dPECVq9fI+2UT64TEYT5gPlg8iMVBRoKPf20RUSwY6BCwAmDiCm4ZdgHpNxDkqQiVWimdWkTb374J3EiCxk+p835JlpjPdxZIYO+8t7sYd3wQe8k92RXN26WbCnieeTHB/nw8Nc5imnSXACc0OkSGAcCt7DNcL8fQBEaSfrprEbqL63Wne7y/BK7f6ZPfHSHNwxgJCAQD1660L285NExP1tC3KBTDOnJB6wtQYbOUcE9t4v19vz99f7ZVmXDWwI+pwojmc5PWZDmVCIBOL9E3Z1ACb1HInTw/nh3/72b3//298fHx9TSmZaW6t1Wbf7ul3v97fb9fXy+nx5/bFer9t9aW3t3ihjkSKO0sHuHcz3DxaBgYQ0Ul7+3LzUt2aFBguJAjAogiJCIxz8PQgOwsN7r8uCbqa6TSVJJsTwMCLIRZgxJckixBIevdb17Xr99qzXG6hGEmdCTCLp8Pjp4fe/PT48HecDIAQOXRkgknpHpAgw19GTV+3btm7b9jP4PRzGF/VLTRyD2ZENDEgwQD5A4EDdEQOEPDHKjOkE6SF4UoRqy6V+/3H99//+9t/+jx8vz5e3t+W69Wvt196ruRRJE+UDTJWPmI6MkQXFgwAxgpwYBYALH46TGQhSCp6HKd1DwQncR7EkG+hbDEcfZT3+R5PSqJlqNhz8Az0B4bR7ynu4KY5J3N6x38Nzew/EYKnd3m7L8+v1x+V2XbaqwY7mFMYjlhUpCluhPpHPwllQUBxoUxV0HM54CEFG4RGd4BG992Wt7lFEmBMEmHu93YZ4pYfAu8AmIFpr27bVOmInhoWWWEg4TWU+Hs/z4SCDueqGgCNCFNwQiXGH4xtbYtXe3UwIBHEXxo7DBvNobxAzMcG7gdYjRrPZzHq4uyHGgN/t5gvEIfW30DA1VWZKlKc5T6eDpD8fkOW+/X/+3//fAEcOBhKSh3L6fHw8J37MDDkfhPKEJCl14BaUUjYOYAdU83BgxrFSmDAnKSnnnCGwt15737Tf68YEW6+192bdVYfYh0pJyBrRTV2UKeVJDg8TLdCaYrA272qtdWaZcir5kFVg7Yhopr3/YlIa2lGm0ceWnHOeSp4SkgUaoyfhlFkyp0TCe3ecSQZLkuhPKqr13tamLVzTOJkiHBCOGIfwOZzepXaSmSfJQhKOa/PbpmvrvWv8TIwnl+wpq+QeJmYyfKIkTqxuERofU6BE+L/8X76c5/T10/npdDgfZiYMj7W250u9LtttqU27cZ/+XqezmE/RcSrplOdzyac8PZUHALj1a+0b2HqhXjcPOKSUE86oSXg+ng7H41lyWVbIr41oi6jm6LElCRbvhsuKbkEYiWMSeJrxXGiaWCbyo+Ojz79revSQrsiMmXn62ENhQiZw0007IHiE7mw9QMCfrCgMl2BhkARqPS5vlO5TWo8Zj5O4tW/f/4n0LGkiLiQllWOaTjCdI5+cUpCAJKQEQEHsTmqAjuGhjuEYht7BXYVA0IU4TSMtIrrp1tvSdG3W1cxsQA/5V5rIrzK6vc8C8KGZ9vMV8EFIGIM6hR7uocCRJppP+Xg6HOZDSbNwYcwxZHxDsuxBMDyz+/9GRFLK0zyVXMb/feya4b5D84bp76fNcldk27g7jffrqr01FP1YSOraA22VjYhIiMC6OfSI6gDQu7W29d5qq+u6hRlCCEMpw5BgZtA1hAmBTF37ECwQASdOx/n0+PD0+cvXx8enlLJH1N63uq3bsqzX2/31evnx9vZ8eftxvbz2rZp1B3NBCMag0DALBdAYsysaJjcWkfeoj720m2+1owAn3ilVHugOERCOgBQ4LKsQbtZ9z+xp2nOZhIWQImVmJkmSRqoHsoaGelvX9XKBbcluGBwQwEwll+Ph/PD0+fPvj+cnZDLXtd1bX98zYCLgHUUKoWbrum3rOsKIB5d1OK4+rqv3/JUxm1WP96ksDR2Ws2cOKo5Hl6PR1B1rbXd4e73989vrP/54+8e3++uP++VtuWztUvWmunlw5jzxvNFR2SAQUBDBrFbAAoiRUVgoT/n8iEy8ElOP5IDq4YbuGO6ETgHy3lAwpwgBZEKIXy2V79NgM4sAIhrtHgSAERIT4GhMKMxANMZM720MHzdeVX273l7fLvf70ntHAkDsCmrIIkVYEh4yn+b0MKc5iYz42YDa1QB5J4gF0PD1x3sz18EdAViYid30/f7r7t4j4oN2tvW+1bqu67at4UgslJNIOkyH0+F4OhynnAloeBtHpU6UOLOwlJQHa3MsAtXuZgzDqeIQewdup1XS+4RvkFci3G2o4rsqonTtZhqhA7CKRENdYwoBYeEAlphzkWkuP43149WbfvvPF+CgBNOUD/N8PAQW7hj3VufEp4GPwVi2hbZIE2tAUHJgVXcLxj1ZUYimnKYyzdMBgfr4+bS61HWp2225Xe+3ZeuoYT2CkDqKU9CY1ziQp1mOMI+LrCv27mraWhduw0DInLsVFhryl/hw42UhSTT6/CKcsqScJSUkBURHG0EhSTgxC1GYwhgMESML4ODNqJnaSNlSCOd9aMmTyAHpgFAQDKNHGARklimlzIwI3ax27b27j+APGIY4oAhpITkgD8s8kCMrpQYI/qtIPif++//2l9OUn06HnBIAbrVfl/X1cv9xud/u67JuDkbFf/vNinDQ5CET0kwpETPjEC0f83yeyhqtgZo5WjAyBptFa7atnUhLmHsAAYlInqKadjRtrtq63yuqRkRkjpqAEXOKU8KHR6LPjp+dPpmczKV2fwcF/4pFiQjfVdPh4d3dPIhIiIfQkRAIiJyRMgDgdJbDuSxvqV1niUyhvV5viwcwJ+GcOOc8pXLA6YTlBOWA5UBlpjQBiocAR0TaAZABDujhpqYWiSAzZua8P1SDi7CblGhwu4IiQvjPIvivpV3Nm+rHgAPzPQd+p70iECDs2lgIpIBwUhIrMx0O+TQf53RIww8ebB6qpB2tMwUmRmQcayIQU8rzYT4ejyUXdXVzT/vgHBEIyX+GjI0Jrpqamo9cURozxd57743tQy8iwNW7t/sdkUEKYk4MCObROgC4BYnOR3YrAGFdIfw4y+kk85HLjE5mbswAgaqhGkSSU5ny8Xh8fHr6/Onpy+FwYpam3cxrbVtbt7Yt2/16v/x4+/78/M/7/dLqigBUhILBvLfear9vuix9XXvtboaIQujMnKRILr+WdltbIwQRLAKJJAxD0R08hsERGqAiGoCFuXl3g9HM9DxgGsySS8k5c0oj/5n2rrpqW8Uao6Ob9x4ZkXHEdSKzIxIKMhF3cg3XiD2rdDBmEdDN1vt6v29tM9Nw2FU7v3aAAQEJaNjA3N0DlCgQCR0wKICdDgpnp4fgqTu0tS2x3Pq3H2///s8ff/y4XG5tXb3WWBa7b301qADYvTWw7tijdJhqbJv5FeykeMZ0ynA4zFMqxyQlplwXZrvXWDuohzuGgbsCwiB0snuAm5NDxt2gTh+gHBGge9jJyB1mGPHZiAFobhYeBISccp5LYRHYfyuPYxggqtr9ttxv994qYZTEzNjURTFlmebp0yE/HPL5kM+H0zzlYYE3BwP3CPawdyIkMUHi0TwgxGkqACjEEWHmAJhSGs7S/lH5EKBda633+329LcxSpnlmPMzT0+P5/HA6TBMRmQ5+vCEDIaZU8qGUXEougzYNAOOoEyMJ9Odxk3in9OBPXfWu4hn/yTgfmlmf+1bXrS7btvr7pH6049+tcoBEUsp0Oua5SE6/EJ0CojNCUKbD6fz19y+/ff765eFTcW5GISUfz+c5JQG+EjIcCB3ZgNRxBPsxkRAKsjAJ85Sn0+FEJKo2Ao9bq2vbXt9evj/DG7ncfXPXztyQOqbClMYBy1MRYjJTUx1YZTWvvY/pxJRLTpRzzoLgbvrLFpwzl0kYhneARUbkVKIxZCfgRCzETJkoIbRwIh7xu8Kpu669jpxe3eMtA4Bw/01F8sRcEATAdpUOQmKZUpoSJYambmbuimhJQBAQ9rEUBQA2xCkQgALJiBvJGkDoAh+0QTml/+f/9b8Qggdel/r99fbt9fL95e16WZbbhmCJXRLmCbPAnDiI3EvGoEBzb94b9IwhCaeSSnBxNkWoFD20G/TW7q+X1y6y5nIluTpe26am3Nu03mm94XaPWqN2az3MYwFIDLXDFihnnwucP2n+vcNR4RCG0RzQBf0X1YDZUFYBAHqYmpvbMFg7eO/NTZl4qBkDuAeDlOlw4uNBN8pg0KtrdFWzIDDlblz7thBeQDLleXp4mh8+SZwxDkTJOZEYeukR5hjEEGTg6gQBHVGde1D3IHKEiJGzjrvmI2QIAjD/Lxry5t5UmRgAcaD/R/HhEGF651K8T+/IBuAdgwuUA09TKqkUmhgyAgNQuKtCb9BbhKMQo6AjjuTylFMpI24qo6LhQPfH+66w7wvjGPGOuX2vFSO0y819Fwt+rCRDpDy4zJT2wRVidO3DpNibmg5EQwhTEjk/lqfHLNlRTN0dgPbd3Hs3YYJM03R4fHj6+uW3T58+lzIFQOtjwL5t27Jut8vt8vL64/nlx/OPH7WubrrjRB3cfKt929qy6rZp26xvps28aQAgEgQh/XJ+9PDu2p0UqLCwjKEcgg4vETogB1bANvIvI9RtZKISDaIbEqJIYkk/q7Kp9da01t42UqXd7uDILFPJJaeURmnHcTYf/CnQYUTqut2Xq2oFwMvl7e1yvd/W1s0H54CBOJDjw7cBjCMP0SAc3qMsAoEQBZAdZuNzT2dLR6ek4apbb291/b68/HH/8da2Do7COWdamutIVUDv4R1ESbu7gnWom+nV+4PyZ56cJ0IonHKeMiVk8agRTQc2OtABA2jQFUbQtntYoCIpUiAFfiDW7BrmYclkdhHvjmAWsPPWMVCEJad5muepDLr7uMOOm7cHmHnvaqaMnoTKnECSI5mDOgrL6TCfD/k055ISwf5IvOtewHd3BPgIjVILiKo9IHLOTLwbIgMAYNytW+st/OOD3ntvtdZWVfuoI9NcjodpeD/zXoNBgJhTTnmeD1OZS56naZ7KNH4UPwWDboYjb2Z4W/g9Fy526QXgyNzgoc71CDcbTaBSF7mnCGzdAHS/twOJcOTkBOheDnOZZ05p9H/xz3VFQlMqcnicvv7+5a//9vtxPiDJcuv9bo/T2TFTmnKG2TpQkAiIaIA6hAMEChHjHt6GgFnyNBUmcYfZTa2bT10Pc4aEfWIsJJe1Lz2oga+KLJJkwMNIkEWmXsw8vCP1YbEcgobW2vDU5Lzb9n/ZggvnWUB3hzr+tJPgns3GYzL0wYzAhGWUdkoQXiHCbQT5AMYQMhIPmi9JGt0QQHAYsPX3P30c9NEDw4W9lCgnnAKRwyO6dXAM7xRGCAR73x9DPXg4Qf58zBFTzq31t+v6z+fLf3x/+/by9uP1Wu+bVz1MWE58OKTzp3w8RSlOnAiYKIQCgrrrZmtgAkJJOVtJKZJ433Rbb9RfE1i4a78SLcUWzo2TBgZniRqtQ9163bC3UIdu0S0gonboDp7sM+CXgn7sfNKYIxI4WY9O0dHbxx7j8NG+A9sIEYawBBEi9nQBDAjmCHQcP+p0PJ1AH8BO2eqE7uAc4WYBhhhECGEeLfri21W9qldod5qOTskkR5shFwMCSpInpKw+GohoSNVFDdq+HwWiE+2yprEq3h2jv+ASfi3t5l0dhBAD1A2GYA1yDgCUHXkKI0AekDqQAgliLjTNXIoIMoNwJAwORPBuPXq1tpl3IkAa0D8aHpjEMvLbR7J1eDi+a/mRcOgDfG+R7C07GjJZfJe07nrRjy/EJCyR5mk6HObDaTpk4lCrtS29qzZdbv3+VpdrrYseDmU6TqfTfHzIgZt6F0IEYiQNcLXeVMlznuZ5/vzpy++//eXTp88pJXVvrS3rcr/flu2+breX15c/vj//+PH6+nq33ve5XdUxiO59RGWiOVp3rdqWprUjoM8xnwLs19EuhIdZmIcHASZiQVQkFXJAQA8cKctOZmGO4Y7m0C2kAwuIEgQICxO7g5uBQdvqelu2ZdGtJtVBIAFEKRmPxzzNKWVmGnOm/ReEh6t3j37fLs8v/7gvN3e8X5bnHz/uy2pmSEECOExBv4BNgYkZycDAA8OByHkk1CI7zUbnPj/1w1lLMQaAjn7H9hb3F7u82ttGgTMfuUxZ7m27LC7uagHu2AM7kQoFh/q2NLi53yMpZgHPCDNS5pSFAtAUeretjvEcBqLR+/FjKBbCNbAHKIV7BML0ky0N74pSJuKUYGSOdYMhZXEnQpA0zfM8Hw7zlNtoEfk4NESE6whIBUxMCanMcjyXTuWu4khjnifEjAyOpg7QRz9fKInIsL2OFGRT1a51bbXX2isJPT08SGHzvcTCUEWY19ZayL+U9t57hLHQNJfT+XA+H4/HqSRidHBDQAFMKQmnw/F0Pp+TTMyplHkq0zjLMrJIUtcIQGDEsFD3AGZEGQ+sDVTqSLNiYWZmidjDbZkp58Isqn6/LW6h3YdMgUqRJOFGAPNhTnMBInUjjJ/9ICKe8+n8ePz6989//a+///a3L+uyvHx/bs8L3+Kczn95iuMEJXHOkwhxTpST7qA0xNgjj8KGkxgQKMJ8JDkjMTEgAkrmh8x2KPlQDvlleX5ba7N2W5knyYwSY+CERNOcIdD6ytLLNGkHMzX3bVsJ/XDIOZex6UH7c9PiLDIlqxZ7DNh4P2PCEkg0hiNq1kyZkWgwF1LiREhGJGP2C+bkKMAjNI6RBURUpCGN5N9O1AVdBByxB6ijOKCDYEw5+AQ8k/TgFhbe1KyrZU1uOcYqBrNwQFd0xY/pSWr+j+f75Xr773/8+OP58v3ldr1v21pJPUekwnOWz1+mr//1OH9pqbSpTEVmNAT3wNpN1353EEUgmUQ0MSBsvdfl9o2tPxzXPAVPF0nrlBsxBUoESaatkQU2xdZI7c9y5w4BEepJYyXfMtSsKTmzIO0B9eoVfPvl1h7R3AVG8LgIYwwMwFDYjNvVT/d1uEcI8/F8muVzKStvV+8NvC2EROARk+A8p+GL1tqtV7/1rd14OlE5KCXnhGWiXEBKmub5/CTloI7dyVHUubt1UEAJ2FMWiZzBGXbFuQFGEOV+yP+T0v5e83fJZgQM2fMwmmEMk6NjAAEFendX6pBDMk+HVKZhWBXBBEAOEQambj28BzowEYIgCsl7aX9PkCMiEEBH/+ltRXwHo4wYnvjlPe4AsjHO+1eI2yDUTtNcpkMph5ISkgNAxwaupq6t99pAPRNOiecinDBoj7rKLILiSroDUjDncjqdv3z++vXrb49Pn+b5AEC9a61t27Z1W+7L7X6/vr5efjy//ni+XN7uvXa3qGvflqZ9NAkiIogTMoOBq2tTbcpEYEBA/OutfTcs0EBfxHDCDAMZKRIQOjIqkZEboRFCjIQiD1XQBpoZIpKknDITQ4CZ1XW9Xy7b7W61hRkIe4QDYkr5MOdpkpyZZcxH322H4WHmrfu61re3+7e3t7dW/X6r17dlrVuQM+Ngu+/y949fB+6oowEfgn2gQuJSNB1bPtS5bCUUF22NbMH2pssPvN/LpkdDBQGUDMRRTjBX8hrYwBTIYYbIEdRdQXXrvjkoZnROlrkv0hh5dAsAAhk5iYk5jVEx7DSUsd4tooc3gGbYAc1CHPLHj4IDgQzvVuxB+RiSEEBgFkmJWGB3L4yOLDBC91hrv6+tdlMfUcYiKQGljMKcDlnmzFmYiEZ/BexDdupwFRPvo8DArdnr5X5dbk1bmfI0HyXtzsNhN3cYAzX4RcoMUEo5Ho8CiBHn88P54Xx6OM5zESaCwHBEFuacy+Fwenh4enz6JJwROKWcUkrbRkBDYaukSnvEU+99WP6ExdzcnN1gRGklEUm8E+hguFuZqeUCiMuyiRSA1cx5ZF0JZ8wBjgiSMwkPWXv6cGTMJf/9f/v7+dPx818+Haa5Lfb2fP32ny/6tuXOb4/rZa1TESEoOU/zVOYiJTuiBZiaq4Z1V9v1uurhoKjvGn8QCmESoSmnOJ8ZU+GD4I3j8tbut77B2lyGNh0DCHhwDgB4A8Y0lxnJzcHM3Zu15OSQAZHwFy8MMpEIOAYFMiEC4eB80rjjE/C493TtMpJNiDMlIR4Og2HmdLJgBwGOYHfkICKAzQMoDEAIV6AGQQim2mqra60gvat6KJNG0pScyEHHwIq8YzRACBmXPDZHQCAfQoMPT3lX/d///Z/Xy/2P7y8vb/frvdambkYxPg6IwHTC4xcsD5gmzDmSeDTxTt3M2wbQOyaUwnyac6ot3cDNtuqviTrOlg/MvAk3ZgXgcI6IoDCwdTceIjElCfPQHVAM5mABiqHiyqYYEU4xomwY4l9SKset/T2tgJhHmNUAnQxaIjOz7JoSIEaaEU4MTwd7PBnc0vL2Gmq1akYKgOMxn84zEbqZ1tbX1rr2dlfrvt57kCGjJM455ZkPJ++bT0d3QkqSZ+ZCmHpwdwqgIBrRNEjABO7WzNTCAicy+J+V9kGw5JEDTAPkGe5m5mYGgOFo4YYQEgaxqqp0KC5ZprmUqUgWEZaQAIowcPAe/n7tFGJGQU4szInl3RlCH9zAQ3OkqrC3n96Je0NO9CfTGABil/ELfQSCDrs6spQ0T/mQqPCgh5ArT0oKUcGDIOYsZZbDscwTRfR73VBMGHLKE5a1e/QOBsLp8fHTX/7y17/97e+//fb7YT4QsZr12mutQ2lct/V2u11eL68/rq/Pt+vrfblv29La2lvV8cYRAZlSNkmZiQYQhyAGiqKkNKX88YzCTDmLCAORQehwX+wnRgIfNkAP8JHoOEp7GIC5qSuBVgKPKZepTDnlCOjat/t6fXnbrnevHdCB2Bw0AiXJNKdpSjmP9MGRArLTBNzUetX70t7u28vb/cf92pab1rubBSdEIUqBFIHxaxwJwHuhw4gdWhaQnCZNxzaf66Gs2Ve+tr7qcsV65e0e2x3WejT5jYHNscfiHpYe/URCG5Xq1ol6TAaTE4DX5our9UCUTO5cEUSQTPt8TswAbgaBSUDcyHt4H7wSwCAIwDCM7raFr26bRffPD3+W9o+ndTdrtfXWXI2Yhch+YiF2X1l0NR2BCMMd2/S2bC+39bK1rVsSJKNszMQlpeOUn47z47GUxPwzXeb99TPECHeCvXfzpfU/Xq8/3l4t7Hw+Pj55yWg2tH6j2xLmTiwC6cM3AY9PT2XOoCqI8zyXqXCWd1PPiCohIjkcjk9Pnz5/+vL581fh5A7jIJ1TLnmCiEAYU62h2Gytm+k4rw+2x/tRiIedi3eS5t44RIStipnN06Hkmek2UkcBQJJIkkGsIiJHNPN39OT+OQ6n+f/x//q/p5ymnK/X67//t38+//H99fsLO57KfF23H/ebcGDMn59Ox3yc5kOZJ2B293VZ1n7btkW3FdSiq3VzBx9xQ27CMGWeci4lI1FK5XwuUz4nngpJfgO892XtzZTMCcgzYsKRYusRhsFznksRJte2Xa5d29Zh6uwxwa/3EABCZs6EAExMQ0/HMk45gEDIPHLAVBmIpRCQkDCy7jR4ZAJgBw7kIPcEfVAwzLr3SrIwEWJnCvSM0bQty3K73u4dJvfo1txrRDNoMZ7TYNISXWIjZnCKIAMxlAAkUI7gj/eQ3vT/97//e922Zdm2qqAmEUCQANPotFJgVp5rPniag2izaB5ZgzZbui0NfMIy819Ezkc891R++Oq2BN+x9PyE83kWYoxkSgNEaaYd+9Lr9b7VrixQyh5PO863YYBjGSEGkAcMISwD5ZQIsmBGzB8/iLn3rsEUHonHSS8iHDwIMQ2MIjMjMlFmKomPPB0pPRH+BTAu+BrNW+9VuyAxPTweHj+dWSgitGrb+tvl/nZZtt6a1q1ZMyeklBJMBzocl/uF8qwgVObDw6dyPM9lbsirhzp5DL6w88j5jHCPbt4t+v+CRrfTtXBEI49ZD70nqUaMu10AQHQzC+/ajVUg4B01OdqBMcZGIztidz3tm+PIlBsdup+37XhHx//8N2M788GlZB6kev+zPbwLtd0t/oz//vipCAEzScKUKAsKojsqoxAwBBJBZkxMs0gpJBwOqmESgciMIiBkHRUSpzSXz58+f/369eHhIeesasuymkVrbV3WdV3Xdbldbq/Pry/fX1+fL5cf99vbuty2urTe1PsYfiIRkgSjAAHAfg9HxJJkLrkkycIfPwczlcxEDCPnR8c63eVe5m4eHdRQHRXQEYFHHQkicMJIIlMuh+kw51k4ta69921d19tNt8oeTISAFlHdBUKIA4cLHfdi9j74c/eubW3L1u7dN/XarZu7QyCjZKSEKBAA9k6r/PC0KLhGOOCwejAHZ5VSEy3UblYv93bpt1ovur3B+oZrZ1WyzWxrBhHIwROgxFwYnkqp3qpbDdic74ZraPNW49astUCH7KERiJQKEyFCcAZEBwsQoiI8Ozb3qhEBioAITqARDbyGbtbX7tV/QbSOM4l7H3HL2l01zDwskMydCE27mgEgMqMavAPZCMgs3q7Ly+V6W1c1DeTkWBUPSQ4lfzodfn86Ps7ToSQCjAAWYeZBcBpPxG5UiYiI2vp12X5c7z9uaynpiAk4kxREMmjqbbSIADHnAph/dhwR8Hw6zYciEUKYkhCT006fIUKWAewoZZ7m+XA4ng6Ho3Dy3dkLTJST/JkGGSAsSKRdzXxQA0aM2B7z+t5bA3gnjMP+vIoMS2YZOWS0SxsQEZERdg44GuxpNL8845k/fXloW7tdb8/fnr/99z8uPy7brSbhxvJyff33fybrT2CPJMwpGXH1AKRwa9tW161tdRdndqOBI5E0lAwCkAJBW7ceLMCJKJc5P1hQ7Nh1WG993Ubzmw4okCQT8aBH8MATlTznPIOp1qF7jeE4+PhBRiTWDgpDHKEGQzi3K9URiYYnaweM74NV2Pm+wkQSaEriklA8KDygm5spubMEg6DAUC+A69bWt+X6zPPnFgSAbbsuy+JbTahixBYMjMioApQC2HH07TqEMjkR7pmP7y81++e3V1V1tbFOBIAIpkzHmR+/ytNf5PyZ5jPmCTlRRFPV7lvTqL4p1A5uaMBrjmLA2r2tsS1eN+XElCHPUmgyNfXNTN17rbbc9PW5//jW71dXA0QQgZJxpNJ18IgwxfsN314hPTIV4BKcDEIxlCOTwy8T3Rib80+0IgzJFidCQhmgqvcpujsBBJsJ9i15FMllOh1P1h1AuhkSnM7zw7FIEkRQjVZNWIjlet9i2VTVuiIEopJjtLjUpQV3YCrzebmcHp/OT0+lHJmSs4xcPGKxgOoGAHvu8b5h/09KOzEP1KuPsGjYKe7Dmfp+byGAIV0bE6p4D12MXXEGPuTPIxvyfWt87+7u63FfyD+vJv9qtIuorZm7JJEQkQQDePlTOxBgqr03tb7TrH/uwAACRIHsxE4SLCDjJI1OI0AyMVHhBFQYhR0RYFdVISBTMAWDGgZPUy6n85cvXz99+pxzUbXr9Ua8AWBXXZd1WZblvlxers9/PD//8fL2/Xp7vd+vW1u7VgML2gHYgYDkxMB7vlvs5unDlE6HPGVO9MuXM3yuo7iaR+82dgW3d7BPeAPvZEY6QtUQg5jHQbUkOR3m8/F0mk9TngIxrLVa67rVdYNuGUekI5lHUzV3Du/vINnBSscdOoIW0Xpft3Vrq4NSwlSSqbsrAqSCJBBEtjOGPppJoGkTbz5gR4FinC3llvIq+mbP35e3b7fn728v23qx7dXWN98gg5QBd6NsPJmUkvKJD5wKSVPrzfUOevX+XPuLtlsszW7mW7fompW6KQnNx1JKylnCIEhHsBjPUgC72taatwgLAAJD6B4VfQtrrtW0WfxLZSdS1d57rdXGtGYcQT08gplqldYKnijnYrar2QEgAlvX1+vt5e2ybisiiGS16AoAfCzly8Pxr18e5yQMOO6no/s9hmm7Kfz9MfGAtbbX2/J6X+9Vp/N5Pj5M82majxjW6tpUPUbTWyQnCIE/k7VhmmfALIAycsPADWzQGYmJk6SUUy6pFE6JRFCG4Gw0cTxnBmDfhfrvjUnYYw6G6svMInxcOnco8a7Ctnf7OiIOfAoPOgdLSpIDDRD2kTPiCMbAcRD+tbojIhW4vV7+49//+8s/X64vd90UAZGwu35/+1HbYn1lsKDortN95DwFuBEEuoZtvdXtfgW1wmUq85REhAGdKbJA731dVwWKlFOBXKQcChEahVP0Z19fa7t3N8yQSoKc2YROZbqn+lKvW9MppzJPp/NJS6q1vsM+fyntAwedEgFBoDuFMwZzEAK6BQREQmTmSXLhxECwp4b7yNf2EYFgllKkRKgI6uE9oKuxgUQowU4HCsMey3Z/pfwd8mO1QOK2vd5eb75uBWJCKAyFKbCgJ7IhnwlVC2qIyuiEgQL4QZJtHm+3Gm4MyIhCIAhCeDzSp6/y27+l3/5L+vT3fDwnmsK5q9UW26Jt690CgoI0HCHgTbq6y/XabtftfvH7giwJ4pDkVHhuXs1rV7Wuy81fv9mPf/j3//T15rHzjSDnd3QqQO/Rur++YP5PkikxUz6pzOruphHC4uVj+SBE3k9SQ1dDCMGIKSciUrNu2lXNFDwIKTPfod3iQmV7OPUnwHI4PQZLmnvXAMslFeGcJSVRw5aDU8mHY3678eVGvDBtAcZCZWZDf329vCx1c0LJx9fvXz59/rv+5enz18PpkcuBBXKWXMqt2rdbDQdA3pOC+KMT5l8b8rhzp80A35+/9xtxvHdqBkI+RlDEThQj4hFoILD/omFzRWRkYRFQQeDxwxotp6HKMzdQYGFGxvetAvaGpJntklH4kEU9JvBqvdbaW1PV9BG9DmBdFWK53RkJA1IRQG+6bvXW+hrqTCRzTohCTBkgeZB7mBs5hCEEEItMBznMh9OXL3/921++/vbb8XgI8Mv1ravGOzhvWZbL2+uP5+dv/3x++f56v651U1MAH+CfHblJiMTEwhk5DSGBx3gyDyVNWSjMWv1VdAq0AypilEwmIsCBTzc0RVd0H4cSBAwQxITCOTGl08P5y29fHj89lqkgkqpuy3Z/u9XbAlXRnIZOHCFYsEwyHfJ8EMmDu/pTeonjR2rae2ut9t48bAxwiYEzEoIUJB6BQei+f4M/v4/Q8BrexyTbvIM36husS19ft8u36/Pz5fvL9W2r196uWu9WSSgVFmFhzsATyfGY52PJE8tMQuTO/Y41m1XVRetia9hmXg3QEQh49fWmy0UPJ7cT0kiQJ0TBkhJOCQgCYbu2ulhYuILX8OpWTZvpiDj91wwMNLMRe9B7M7VhKI+hIQjQ3murXcddPtT3RdKaXm73233pvWaJnOU455QTDmxPhBBkIWEMjZ0NTZRzHpOpcfwdD6G5t1bfLpfnlx+X27VaiKTD8Xg8HE+nM2PULXVVpC0ChqLFg7D++UlqrRaNPHiYmAczYfjP6adqMsxsOOC3bXNJ+J5r+XNT2JtwvCvq6f2CPnYSh73B4O5m2nsfpmsfJGkiYVG3cTIiJBFJJQMoM+yPh+Pgq9KYE/zqqOzavj//8fzy8np7WfsK7DIzBRNDkN/70mwtGVNCQ1PsJz+Wkr13NC1CZch9C5myVu+h5C2ZZMbCnBMlhg4Aps2jQ4T1rlU45eP0gBCE6h4Wt9aqOi8dcAkPKvnI9FDKM+Laelu3QjznkiQBEiUMAPs1rQd25wXQkKPgsJGMfzeswjhyQBiFUABJwzdtgZGRAw3IkYM5QAgTeqM9yid0n2IqKoIwAlEgRdTWbrS80v0lKEnJ1lfrzZqzkzKlTHuQU4xfqDsyI5gsizIbUif+s7SHx33rEJ6IMhMOopgDSZRznH7Hp3+T0+dUSg4OQ1LQgO7QLNQCR/yQQ7hXgNgavL2217dtXc1tOK7HwdGJjUiRHBmIMCFlpjkTzRAIHmiB1h0hfIzZwC2irbS+prjmtIKkhWhtZg3r5syW5AM9kyDYHQAc1CDADWGYCoCIupmadlMzH1HjHXTzbe1LystZu1GbvJtGC1Igd9etN9Wjz0c6ALIkPqc8H4+5lJxTYs5C5goElNg8LLT2eu9hsG21unUKvS+LHM5cJuY8z4fz6bw63TZfXDYoDUid/pfxMPi+0twdAwaadVgkdi//yIbag/8QAkaDXUboQxYpzIUpe4BDjFwiliS5gBOrvA/ehNPwT8J+/CfkPaKd90kAIdrQiDuiRoSpAcSuK3PvXbdae+vWrXzwtUfEVptqU9W2bduySGIEV6u9b8AuiSSzzIkZiMeMyj3MRgo8m2JYQpmKSD59+vzlL3/9+7/9/evX34hoXbfr7fV6vXbtiCAi27q9/Hj59u37tz++v71c69rUApGJAZ2QggBk9NeYmJj309UAW3HOPE+pJHZtbfV/yd/8YP/zgGiBsEu2w8h9bzu9d32DGCVjmtJcpvnpy5ff//b705dPUpKGr1u9XW5vP97qZaXu5AHhGtGDI2c5PcwPj6fz4zTNwvwzhibet+eBENDWrPfBchjSb07BjJKJGB0QCCKC+RdGK1ZCRt9CF7WrwUpczVfoi17ebs8vL6/X5e3e7psuzdbWm3ZiEuYxRBTClGit5ezllKfjoeSSEwshWUOYsEts4JtbD/AgQvJAVahbLDfd7m4VU0ZkJkISylOacsqTpMxvspgvfTHrblV908HvsAGh/ZenI2AQ6FprrTV3HwtypM0goZnWrW6tcksjIMDMam3X2/Lj9W3ZVgA7zXw+5PNxDirVCNxdu/Vuqhph6gPZN4br8OvQfVzi13V9eX35/vx8uV45F0lynOfT8Xg+nZLQVnJTJRZzQ0QRUvvYRIm362Vr9+hK4TlJzimVJDmxCKADKfQewEwb0U0kJU4lZxlCnD1G7Oezticy/nRmwTAJvb9nc/dwNa21du2DqtJ7GwcXj2i9q3ZA4MSlZA8iikDflSSGyD+1G78MerZa/4///G/bsjWrPGEq0wgZMtOuXZuq6o/llV7CxSJ7FIKMbhWtCzFkliI0Z0yHbdnWZTNbsYFAHGguQEW4CCXhzWwz21xbvcN8KtNxZmFJGJQwvVyul/t9W7u2C/TGx+mQ50+lfJe0eNTrnR0Pnw+lzBHAyc1j2KY/ioOGP4qYhsnK3Jt1BibEwN0LM7qogBSAzc365mAIGUWdDCmECVmAIZBH24wwKAwjXEEDe0JgQmRA7X2F9SrLm6Qp8VGgCzoAgjGEoDAE/dQuO7iGoYUFJQMAZXHmYBnnZQAAD9i6oXtIIAARgCNgOCPOmp/K8XeZ5iSYHRHAaDecKPNgmgQAm1Pv3vp2vbfXl/p6qbV5yZwEwLvbhhhElbmxBKLMM+DZvnzi7XeqGxhQ7bhW3BZ185wQASGgG6AybnluT2cV1hdtXrdlbV3vK3b+a4o/Gyk+IBIagLYDYoAQa62I6BC7AfW91jQ16H2pldflWNdKbfIWZqoD/6Hem3v//PQQyCknEZkPcy7zNOUh8xSKrr27KwCYMTOLeO9tBGGZt9rk2w+TFChIfD6ePj99xunYZG5y7HzsmDToMf0vaHQREU4IKMLMI8zupxNTbRByHMExwhEcHQc9BWGvy8RMQpQgxoGTBzS65ILBxEk4s2QpSZKMJp5Z9z3Lzolpb7shEjOHjwU6Imtx4JMBB1s+fDwgAxD663DXrPW+k1DrmoQHQl4YEkvKiTKGREf1UHALUHMdSTjD9YqJj+fT8fz09fe//vaXv/3++2+n03mE7ZSU7gi9rq1XQLzflufvP378eH67vC3rZhoQGMMHysBIiWgkbCIiEAxtKREwsSQuU05zQeHu6r1/FJ3Gh18AYQEIFrD7f2K/Ro2c8uF5JCJhTtNhfnx8+vTl0+npIc/Fwtf79vL88v2f33784w+73HitbAroLWLTMJ+Q6JhyKpMk2Wn+f66K8eWoWw/TMNsbJxhIPo5HyB675xJFkD5KgAPaN8Xc6731e7e7+wrRm3avrd/W5bXXq/cFbAlbtJs5AnIAh1PAmIQ5oG7YbrAxQceFDbzqCvXm66VpH/1fDASLMDNvGASyarm18rbmI54wzYxcBAU4U5qZZWJhQAyIG2/qW9RxwnPXEVz+L+AdGBllY/7t5oQgwiXlnNIQ23QzN1uWNQKHGCIs1KPrQEJquA3b1WAEEucpcaagYVMHdAdAZuIA0FHJxybx59DqJ4Nxa60mYibOueScU84ilCEO53MQ1W0bHB33P0sJApDgQC5TEHNiEh5KrCH4RXYAVd1iA8cs5TgdMjPlHbH585wRA1xN5B5m78SJ951kl8u4BYa51dq6dlNtvY9dMpccAN161S3IWVAymzugw7t7aY+doRjNEfjgvCDEVIR5nucyMJGj1TTEL1oF1Iilkv5YL/biFfVJHw5CBybjaGCm1UNrq0vflr6RYXPz0HC1mIJKntJxOkwQzfy61re1Wm+NhIES08PhwE9QOE2SLsvtWu/9drdW8wkKymPOa56u67bafUkHBJSUy0w5DULFh3VFSEKUiBMDAhIjQwzvPexxDE77MDMQHRAi1BqSC0Vg7167ubu87xYUu4FoHAdiYPNCxlzWPbT3VePK+a2UeTqLJKNj1ABfkXCM9Qdpz5wCSJ2IgHDvYnlgD1TA6WdpBwgf9ltHCCeEUvgwy3RAllEZEpMIMiIZcB8XRdgMam+t1lBzjQDyZn5b6u3etXtCeTpPj+fMbOabmno4BkVEMwuEPMfpMT599QDiLLcb/XhB7X7TgEARzIVS0Pk0fT2ffj8+fEmy2XZZ1rrocu/9gqgQ//bn52i13q4N3mvQANPAu/godhHz0IURBLgaa03WPNpNt24rtE219wE5JbBeXdtqUQEfjvP5OJVpYvIkmBMRhYOpeVNvDk2DMRWJLKhmEVDVYq3YzJiBhFi2bretUTnhdKKj0SEpgQaZ0ccT46+l3T3cBlgyS84pya4+RzOLWpv13nt4MBJQmDjBiLGxUWPwfSSJ/qcSRlimUgiEJREnksQpcZIAcHMHhzALx5EdObJPEFh4+AaHgZCQUBgAw6M7uGkAMSVnDEf8VZM99t9xG9raKiK5pMNxmqZDOZRURElb1O6qXgMVcZ+Hy1CkEkspT18+//773/7+9//y9be/HA5Hkdy7htrj+dS29X5725Zl27a3y+X5+fXt7WXdlrYzaoa3y4OcdmnweFTCRjkkEEZJkuZSpgmnpERVw/6E8vwsi7vDZDyrGqa4x/OMv9Ju1mHcsbNCKefD/Pjp09Onp/k4I1Pr7fL29sd//uPbf/zjxz/+kK0dIQQVKVbTpbXQOUE8EjMnJtozOX0/OuzeoGFhNB0sRBx3gjFYIwj02GM/h8fiY2mP7T+q47be176ob+gdukMLq9Cqa2f0nFzBq1kEMWYSERCGfQFFEASZ++K1a3vGulmr5tW9hxtpJwAiJiBzdO2GHobEG6ZbpdfwtDWavsyHdEAgRgFKlDLnKeMIyESrfcPqvniAjZQz2rvOP0v7/s9DPwYAwnzI+Xw8HOcpANX8tixr7+t97d3fWSEQAT7kp+6hZm2ASHHKfCjlkHlOyBimo5lDQkIpOUBtTXXntEfEILzuhPYhl1SzrgC4q66IApBEDsdTALpF8821u0L8ac/H+TDLxGBOgZlEiJgIBXGQTiUhoJmHNTBoZTXtiJBzIhwcDutdVbupDcu6qQ/x4M9VO2qSajc3JHSw3vogO7Tatm0DhG41ELr3pjVQx5cShoM1P+oauPtOHAB3CPpTxSFJvnz9jIiMlFJKkjxCtV8ut8vbdYiJ0RzVrrotr9vq7d7Xv35+LNOxiwdarWtt61q3rbbalYMreIu+tm31ufPp03Se5/koKQCYbq31Vfum90SpIM8plweeUp7zlJnA9XV93ZY7B3E+PkjeyrRcluV+f4PkDk+/Pc7TNE80Sfrl0i5IiSgxZSZCInkHiqIDqDsEOvn7KR93yDtYt9jYvfcG2+auJmiA5u7oPvI5x1FwtEGGZ3loK7S1FfpV0qsdDzMdj7NLwEKwIqAjyZCUh6MauVMgoUAOyDsDItxDI37pAA91H0EwRqI4HfDz53R4oJQQXbwnzJwSCrIDbmOTxqyRarXlYstiVQMLGNtSW+tOLqeSvz6dnh5z4nDfmlIogyXrsW4rO0yZppM9fLbpgMdHfv4uDvD6o9UWjJAy5kxJ5PfPh//6+8Pfn85fZ/6jXW3J/dXaG+mV0BL8DX+W9nXdXt9uY7QkKTHLz9YlAODQdCMik0giRHZN3g6sWXut9b4st9t9rVu1noUORUKb9XpXXyx++9QCLRVhpt61a6t9W+p6X21r0J26Y0TKQpOQWW/mAdEHW5aZUiIp1eD6dkfpMukc5ZAfnUV9JCn+WQf/BVmjvXce0AmIiOjaR9Pd3bv21lurDQKnPDNxkMPIY+yt1rZtTaCCLy4YAd3UfA1oCEbvghhEJhQc7OyfczvCgUracdgRsFuGOHCXkPouyfx5eGCmEIEhM/8X0ek+uxIaFrs85TKXMhWeshK01nrU5hXJmPFYjvOcM7EQZ05Tnj4/fvry+cvvv/3t629/+fL5y2k+ArA1bbW1rXr3sPDu21JfXl+vt1vvnYSmQwFC3ZFQvuOwKWCwpmnAhUbQWKBIJILERrhsrXfftrpfTX6p67uDAOD9xDysQwgB77mZYwxHZIFAjMy5lIeHh4fzQylTa35/uz3/849//Pt/vP3xvV6vbpYAUNzFN/CVkBNLERCCfbI/XuYOFtZ764PetzPE8D0CxIcZf1jkhsYRYk/+/fBdwPJ989h0SB47qkEza6gdzXDkxWB4UGDhlBmnxMxOaBjj4BfoyIpoqAtYi/vNtqWHKiLkuTBLEtYCqaFa2D6dQXPsHZalxeuGpcnBKLtMWCxp2JB9TA/5DAd1Vbew6JtCBdiDy/81HgbfZxSAyMzzVB6Oh4fjfCyTR9Sm2trIJVW1IZh2i66+1VZbG5h3dapK9xZUYE50mOQ00ZQEAszDRlSDexiw+wC32QgOACAikVTK9PBwfnp6/HHbmvntdn+7XO/rdjodRQgAWVIuUykdIgypg+MHEcdhnp0KATKgDCB1OCCg8BDSQqCjM7BIQkK1Xus2JnMR0VpvrfaubrYHiI23GR/VSDsxfuDuHVxN1cZKatUqQBiKY3Tval2jBxkKIOKeSzlyXYfOJNw8zMDZfx5RROTx8YFonEZEkphbbVVRuzfvjg6JOBGTBwVkYU+8ut28s2QqiaZgQ25ETUg1DFvwRW3ttppt1debrd4epsOplEz0NM/FsDl7d+gqLFxGQgMFGUikK9y3NQBba9y6qLG51357uURAmdLxIQnlnMqHRQVICDzgkRyEQEySmLMDjYUEAMGJ9gM9wtjcYM9/at7WQagIBA3o0Xt0dWFAoQhyp64UiJyRg/Za70192dbrtlx6PVkO8A2xcTYMJFEMDkUHA6gITkiAxDBTFBgsSEL/M5MLRrsOY8guIXGcDvj1q5y+pPmznObCxuRIIzsu9i2CpUg5AkFvcXvu62ry6HTwQE8Zj4/lqZw+nx9PUwpu6K1V9Q5t47ZBrZoIsgiVyIdIk5N0QDfD1mzdXBiQKGc5HtLXT/NfPx+eDlIIBIStTO4erAH/kvyGJCwFcY8CBMARDDomJkiCyKPQqxqjSbSZ69ccqUVb+22tz/eteUdxEGSOCDSjq5nfF0Wv2t+Wdpxuqn5f2h/P12+X9br42qIreCAQaUTtNgKNiSlNZT5M02HmMqOU2mFZ1UEoTSllJnEP6/q/Lu3WepMQABihUr11N2NkAOjee29ba4xymCVlQXbDbm6tt22r67pRJDdISQHDXNU2jxqgER4jZ8sjAMPBLIaR52cQzTuYfxcVI9GAtRiGD3b33iJFRGKWlEY3HiN2Lc+HdQbElKZUDmU6ztNhmg4TMXn4VrdlWdSqQzuUdDrNX0+ffvv8+ZBLEWHinNKXz5+/fPn65etvj49fSj4gcGu6be2+rLf7stzXurS26XpvLy+XrW6UeDpMKCzLer+v1g189yRAhI3gZMDACAmAMAQWcCFDahbLbV3vW90aBPmvSecfbkI/e6o/rQr47ioiZAnEIU1H4pTK8Xg+Hc8kebvd3p5fvv33f/7xf/7Hdr1KGIVv4YPktgk1SWXKPBfK9FP5BIHgYB7d+s/S7mMiFkjIjMzsCMGJA3YXmFvACAT8iKwJWF/WgA0RIyic3MMtHBz4/W5hAc3EIXEqRQ6TIGl4c1UP84AwAMMI1BXaEuvFt8UhTDJm4ZQTEjtAb2C61xlmQRB3bFX9tvHU5GAyRTlInlIqgoQ5icx85NkgLMBatHurCw596BihfPgc+319rE4Wmaf54Xx6mMuUknZDhy1JU6mBvqvHxVRb1/uyLssa7oQcmFow9MjqRDAXPh3ylBMCDcBAmAeZh8vYvYfh4F3VKiLzNH3+9OmybP/8ca1v99fXyx/fvv/t99/OD6cSSYSReKSuE6Jy69Dhg0R+nmZKODoAGANSagGxx5oSY6CTJ84lT5ykm97XpfU+HtLee+ttBNDQyMqAPZjIR9bnPh57d2nRoN+6ulZrzVqPCiM/ELxbN1cPMxjGMhin2BhyeN+nhGau9osShZnO5xMxkjAnIiG1HtVzTaVmbUaBp/l4Oh6nVBJyWzetdXXDbU2nlA8lTTOjYV9Jm3Tr3ax5X7rd69313uyi2+vt8vvx4W8PT5OUp8PxBKkF3S/3ZbszUkoILCGzi/NEx0Ne7rfr2l6XSrViraJO3Zb1puZlzqfHTHHIvzbk9x7vmKMjBjFxZs7hEGFDRwljnxwneoDRwmf2gK7Wt9YUIoCjg7dozXv3Uf49SI1aJwvijJwxJUQKwO6+1u22LNdluYnAVle1iuzMQKmikTs5dIht5wvgLCgMCVQMxEPsAwsfAVIiAshCOWGSOMz46RN/+i0fvpT5WJIjesAOpISADmAsKeGJcwT4doH7axwQcnKeMU/8eJw+TceHw3mS1GHVjs21V2ur12qtGyRQpuCgDIa6VV8Wvi+4rH3bXBiYoEx4mNOXh+m3p+lYSNAEuXA5Jc+FrQe6fLTxSS7TgeknQ8LMoCOiFBIRoozBbu6u5h2gC66npL+dAe/wT9et92tXTHA65VQYGUEBOJqHaa8Xuyxr4VsiMYPW/fW+vN6362ZLddMIBxFEGgMPMMCSZZqnx6fz4+NDmg/BpSrda3QjgMTTmVOy5q7qhvCBWfM/mN9S4p0r7CMBqbc+HOgDl8mEU84P53OaZTWv3Ix3RwcykzAyI73f2gbsPULVQY3UAtSgIhMyDU3OcIUNNyH4O9kLkSICCXjozXz0MhGDiAJwWPI4wp1F4qMkGxHKIXHAdJyn0zQdplQSC5upttZ7desEnkTO8+HL+envn//699/+cp4OU0oj++h0OJ7KNHHigGF02Gq7L9vltry9XZ5fX5+ff/zx/O3Hy+v9Xi08i3CSSQQYgaDXbt20mcYwPxsO7dEIaBhCckrOYBrmttzq/bJqU0L+NekcAWgMtsePJ37ul0j7hZ1+imvA3Z2QU0rTnKcZWWrrr6+Xf/7j2/d/fru+XbxumKBHeHgDBOCOZMBFcpqKJGEGRPdwGmhs8Pdpy8gZ2vGXEeMKEQhB5DGU0jg48RRDrfBz54poW49oxByI3bWr19411NkMo0NYVWguhglk5nzICZHcomnUZn2Nvlp1QHPbtK+0rd6bAXsIqpu4iciU0AqFMUVyD2IRZgLCIAqy7tut3i9SygB589hgmIkTzafJjbyCVW+b1q1R4K+VHdyjm3kAsSDbnqRNo4ktXd0iWFIu0Zs6vGfAB7Sul8tyvS0AMc9lPp84FydCdAwliCQiIjBgLjsMjhJzIgy3AAggYPgpgUCiwzw/PpwfTsfbst2v12/fvr++vT49nv0wTSWnlERknmdGqohi8LG07ymC47gylCoJCXb02ZglirBQ4sQOsda19Ub7/T7UdHQm3H08qh9MrE77UsD33lI475Apc+uhHbqhBgS6OZiG2iDwuqprwLDeIMIQiO7UUKKgX8UPxDQfJ8ABQB056K21Royn01Ew5VSmMpVcKMhaXy/X69tFWy0JO3rjOMshz+yc0YldrZvW3sgaUOthMJgomFs9bBUnOU5TTmWmxB6DYLDV+2a2uQfB4TjPSR4Pp4e1ztd71+frbS0IEoFqfduur2/Xl9zuR5PyUUaHu2+DEwoCIBAAOQyEl6spjmhR861rBDJTonTIGaA3NxsEaMTRcxvpmGouHrxniJEO/rSCGDADAjA7m7q3Wutt7UC0bRbuIi5ojEpsnAKQIBJaIA2wlhAIeAonMw6rAHt5R8JDSQRWEhUC5nCPXrurJEyMHu5dI1qQoBN0aAFtnFfTlKeHaTp7b56TFwYsnicqRMyh0dctmmuzXrVpA+2xaeumrgHdddPagQ0IYdu8rtCad3V3kI7nwJTkkOVU0pSZcxwT10KpYJ/IC4POH2DSwMyS9hXcuw5SMomUSVISioTBYBDeISCBnpk+HenhLJtx8+4c50/zfOTHRyEwbU07MPOYKnvg3fG6mbbeu9dmt63dt167q4UQiSDulpP3G1aEmWlX0869hQZjPuaCXIgnSHMQo7txpF+ut7+WdklSpvJ+SRj5v722hnsnkIhRksyH6fxwThP7uhmuKMIinERykpxltAgG3wZ1WC77oN2bWTR0AyLiAakW2INYfs6UdyWoD+ECcQQgObiZOWIEAOJoRwExkgyi4sfSjvO5OMrh4TgdpzIXHHlLvWtbQ5ugs3CZ5PF0/vr4+S9Pv/396a+Ph9OUcmtb102ExcF7r9um3lqPZW3X+/J2uX3/8fqff/zx7fv355eXZV00lBJ5ILNkYRIiprrVtlaA5uq9q+roOiNqAEFAEFNwQELr4d3r0rd7c3Whj33THUawy+jeo7EgEIe3ZEefERIGkEVYgABJLrlMkosH3u7r84+Xf/7jjx8/Xre1kqsRAoZGmKEFOgM5O4lIeiechoe/QxAc0fdAVgDYef4wVDkAPr6lfQhNRATu6L8SKAHAurkrihvAZt669d49HCgUo4PrptiNXDJRQZ44IeK+t1bcrr5eLdRAuzXyTqOPDxJI1FxTaMLMQpYQCgmIG458UQYQxERCpn3ty2UTkQG+HO3eUlJKVKaJaY5O1mO5Lev9jg3pFx0zuHvt3QKQBUljwLQBAimINEI9SCQT1VjdnNARzN1b0+v1fr+vJcM058enI6Xc2qBkOyIwM9FuSiUEYUwiJXEickX9iYAa9FkAiCg5nY/Hp4fT2/V2W5eXH88vL6+fnx6HT4eJsqQ0TYwYrtJ+iRob2QA+qHD4XlvG+RwwAohQOI1AF3WtbVdXRMROOXc3Nx+DD9z/xMGPYmEZeGTEAHcA573T5ANXjOY0RLDq4RZqYBqqrkMqP8Zzo7qPUwNEEAfHL2pZRMolWZi5Wutbq63X3juTzKfTcT4fpiMSh0evfVvq9bZ8e35dlzsTdIjG8HumU5owUYhgIAkCuYP0UK0BiqAIAZPavG1CecpRmFJO4CV8frtcttvtUrfNfToejvNxmg4C1Lqfjtdl6y9vt1k4E2bE1vX+er2c03p5bL825AlQgCbiiSUgHHiMZoYjws2c0CO6+9orAMyUE8khzRbc2ubhBopBRDye2t1F9g7vdECP0YED6SCMwsAMwt6tq/b75ga43YMiTnMEBYYJG2WP4IAZHFCJsBAnChnmeFcKt5+lnRCORQgpCwkGo2uP66UdHvBUWRqhkG0eZpQQBKpVjWYu7pwmOj6V828KaHnylJFKpAkYw0E3XaPT1tpWt61WNQ9o6tajA8aGas21RkIkgHWBdfXewjzcofUAoMxcEk9CJTFPcUykgHyKfgScS9Tp+iF5fvi4x2Y31rlkSUVKSSkJu1AwOlAQh03UzomeDjwdZVtw9UYZfns6Pz6mTw9Ut+XtrXYKSzSc4ercO9603vu2LX3d+lK1dSPELHKYyyEnYnKPtQ2sBrJIBFo3bR0DPCrneZ5SmVKZD8GlBws4O8+/NIP+hUYnkkoeGVVEBDIATalum3b1MAkSTgg4mLAWCugkwAlJAthBHBiAGB3RMaCbsxm5IziMsgCjI+845BmjoA+O5M8w0A+lGgHQfM+Hf583DbE5BqIEI8BHSTYSnj8dHe1wPkzzlEtCBOstoSfQ1kCtc+IxgOecmvl1WRkEJnBzD1TzqE3xjltvClvVy3V5vdx+vFyef7z+8fx8ud2rdmTOnEiGl3f8rPJMOFi9Irlka0ur3Hrt2tTVAYAYCZm5pDSHRbeuFrbTtn6x9+wffx9/w7s6Ya/r46+w76EQAO4BSLlMaZqCaGn12+vLj7fXZdvUHYkR3MakjlGRlSQws1Nd2v3tsj49tPVBSJgEAjDQ9/CuP4GCuwX3nT7g+z0B3CMcISgCw39aoN8/Q0cwdLUe0dza4JyDB+xmcjWPADfvvauQZkoJOQRUbKO+UL2725i/OqBHwiCMglbQxQOdORjAUoQCGSiEeSA4OZIHjwFfi3rrV1jAOYzD9mhA4ZTSlFP2T6jdlutda+tv6tsv98Q9xatrH15sAAM0wO6RBnpNRIgAorijdsJw67Vu27Z13emvA+zEKFNJ53l6PB0OU0GisbRHCKEkkTEbcR1xebs/lHbvgrsz4XEqv3/5tLXufzyr9pfX1+/Pp+EAj4goXlIexTh+aQVBSokzkSBgeNigFCDCOCCP0k7D/4jhGL5rJO3d/zpWhRkM9qub92E5c/fhbOUdyxSBEDxoXj+J0SMLGgAAPcZpYm8hAPxc7W4Q6qD7DF8HGj/Fz2Gih211iZHZhzjlkkQsm0QSTBi43tdl3WrtiRM4QAhTDqut2+XS07RKTg4+n1OaOLGMcQQC2n6Bd222bWFk1qIrNPeHvs1TCQsknScOmHp0rT0jTCLHcpxkCoCUph+X2+v1fq99Uwfst9Zba9v1fnt9O0j6LH8eGxPTnOlxpuNEm0ENVHA1HTYMi8AIdVe35ppCEGHAQdSih6obhCMhj2P42BiGvn2oWz3c0R1Gz0TTgBkESwQZknWzaLBWSxAzC2Qhh5RMpEOQa/bY4ZbhGIgBFMF7+Nuf+xQmYUacEg0shpnfb/7jW49Yjk8xncPFlRRSQAoTNbEO3RCBiJM/fEHJbBqBgYbQ0VM4K5CF2Va3y9t6udauBkwWrqEWuwklPASIjL9/h5cfsW3+c2aZhKaMJOpSIyNMAdkCu0vDBFNGqul2+/M2QuPqFOAYTBhBiBiO7gHuGMYwolkq+yXBNfniZveN7/2urPlInz9Pj2c5FienKug99iRyQkpSSjJyZweBSICZDkFPD+evT49/+frl6fzAksxhrX1r1nSAqjxnPs4JwutWI5A4DinOB5apAE9VfatTzr9w9X69tYvkUoZhZtjepph666/urfcwAyBGMbdtW9mhW42iJEAJQDxYHbujAhRABOBwMhvXOBpXbdizZUerZwjg0SN8hGK9A3IQkIl3ixUMGhLQntm0G2uR8L31/+eZYPyG4+MhyKZpKiXnLBBuKAKWMLcUTQOTpKlwEUW4rEsotNrr4bSrjG3MmDYHrN2Xtb2+3n68Xr4/v/14vbxdr908TSWXIjkhhYXuphNmFkZkRC4lQKHmusp2e7v3atYjAiQzY8p5ymXe1tq1auxBN0C/tlTeK/2YXL7/rt3zjzhgiCO5YB8CE1HJU8q5hy3r9u3H84+3t2oKzFImMAbSYAQmx2SUgBidtqVenl8eHk4Pjw+SsqQSEehj0DlG7LZX97Ej79EVQ/gwtnn3IaUdIIdfb+1sEko93Nw19uNFIDqCRvSwQVoPj9aaECQBKomZopFtqCvpxr5LIwM5gBwEYOIoBCmQQzgEQtmNDTjAEcIhkAI4gAPIERTaXXsz62hKCCDCiVPJyCXlPPvZu7ZP90dv/Q73Cu3jEM7da+u1a21d3ZEoEA2gu1c1B0RJnIQQpjBsgQhb7+u2rtvm7oO6ZN5aXSXJYS4Px/np4XgsBZFGoIswpbxD5GMQO91h+NkdRlD6uMwRwlzy718/W8C6tftaL29v375NOYvw7hUjIDczU/dflMwp5zQxCwG6WlWLcMOBGEPAeGfCQjj44CH7eCcQMHJZYU9u8jBz69par601VWVmkdEE4t2aSoiMkoTeGfJIiDHkKEAO/j5kGmpK9zA162atW+t/ni8hLNtPHH64b3Udo7mUUs55fEekFB2W+3a53J5f3u739enx0+lwAkehQlB7b8tVmReicNfPcWIsaZYkkpkkIJouDSr2RfW61C16Tb6prdqeaj6XNKdUJJdZchH17qFzkjnJnMucj8REyF8/Pb1eb7faNw3A6rD25d7u2/31ekslPsfPlZUTHUKeZj4f6NoBOqnGfmEPG+kTFqFhCupg9C5MsrBuam4QQQBEhIwfZp0UowEXEQFuYYrWwAq4ACGyAHiQuIGGedMeEZYSaEJDRs+5YSSDZCYehC4xvnjcFb6/PuVDI02JmXGsDLjd3ayvN3v8BOdPZOwdurF6Mp6B5jAOl5DCnOj4BOXAtzdd13BFA3QKkODk1ry27XrdfvyotTsKGEQ36+rdHDCYQJyh+fdnfH2F1lwEiDFnnieaZ8Bcuyw9IZToqXVqKh0SSo5UARcA+/kpgPH/z96fNEm2bNfB2O7c/ZyIyMyqus17DyBIiCJFIw0wTUTjiBMa/xf/mAxGgeKEZhzR7BuRkkEAXndvNZkZEee4+2408BORkXXrvgbA91GUPa97qzKjPY27727ttUZ70Gbaw8HVvYOFJ1LC4OhsZ9IPhE8Aayd7Dj61E6SY5/zmbbmbKEUzxpKkrubWegQQTjmVHVFKkiNnyAW1R5byJz/72Z/+oz/5P/2Tf/LtN99Kmi1oWXWpuqy61trqiqAlYVtPT58+nU+nvrY96rsSuwOlKTuQeZz6+tyWHzHtKU00jeT4sB/gQEilTKrdVAHC3dZl+RgfKIOnhXKfJEgcqFuszQggHA2cTKP76lCBnBMzJPY8wu2RYKTNQhHB6M5AU/WIrQUcAQAcAgCZeWPBGHn7gaoeKxwGQu9mmiGQoDlY79WsnUezR1Vtqs3AAmEqiVNSiOf1dPb6CZ5P63pa1yzMiBamoT16N23d1tpPp/X5uDw+n06nqtoBEbd0uYPD4JPRFhsxIXHOBQEpKEvOkty81TrymElkKiXnwiSmS63NurkHbWmJ2zEQVHiB0CDjxoCIhLaxhYxSY4QHeSTkkhIhHM9PHx8fv//066fzp46d5jSJoBuCIwEyMWwSQghgXU/P58dPT/uPn/K0m+f9MOsOoytbYyMM33rwBg8P86Ym7YNcZBC62UZmdnM3MMkE0VQrQBdGZIA0Imd169yMABlpwKW06+nJLOXM0p81lkAFAhxlPhQPMaAAAZoiF5gKT1k4EVhYqFl3hVAEJxROQkkw8SB/B+2qZ1eFboAEIpRESpkkdeBmqFRg93by/sCOC698U78y97q2ptrUWFIumVIygKYKg74HGYmIaZqyyKjNa+91qYtFJ46UiQXNVbtCCSGaUsqJR+8gQpgZ9hZGdFEBgpuge4B6eu9uJsIl89uHuwA6r/27j4+11o+fPr15c7+bJ0IU4pJKmJleMudXHwUiEAaJUgp076Y9fGPEjQGKcwOCIeHnowJDm2cZQOCIDmCjPSUwmEJoEGpvAhPR9NI+yMhMACEhPCz9WKSBI1Xj6q7u3a1ZGzXztdW11aW2WlV1IyQj+Kr87H7azoKI5t18bQBva9OurTZbzVY7HZfn5/Pj8bSsTZsf82l5PJ0eT8uxare+al3q8vzp+Fja+c3X396/fXfY7UtKnDmV3WFFX8Qf7fjxqM9rb+3ZwDCD9d6J73A6UOwYS85v7/e7kiXPkuYIWJY1ImqrJaV3b+6PtfdA4pMFLMsCatbUar+9HTmXkGm/l8OMXrEBrXZJbsLGGY8Eo4UtyB1dw6v2wXUcgEyJSZg4JYkpUZ9UM4ID9CEixxAGER7mNKBNPrgEOTGPm+KIiuEYDC5hERoghjAQsYRAMBIAdMng4asUYwCs6k6RmYJGy2x4j2X18ykQYZpJDpQzrmrtrLEAEChESKS7KHdYDpBKcAHs2A0iYE+ckLz5euqfPtan59bNNXw0Lq/NBl29MCQBs/A1VIES7O9RmEqWu0P69ls6vLXYn4+5kSyJcKWzkTkiYDRYDU6vSCfD48LkbapmBkDkuG3zG17oHPYk/jH4DGIRqC6U/P7N7rCb9rsyCVCPOe/ikMyW2s/WTBW0qtBKCBNHcJCAlHy3O/yjb+7/5CcP/+hn7779yU9yOQCXrtgVukJrvdYFvCeO8/On737581///OffnX6J9VlqSQWLjEyzra9W+WemnUXSdPWgIYZWVkzT5G69NTcN99rqWiuVSHc+AwzJryC1qN0Awh0VnVxBvXs0JBBhBkmRL2DfTe1y46UEJGZyr2amFhIX9C5GBCDQ4K8Z1BiDqmbU/jZii9fV3RjC891qC3Xr2nuvvWp0BSOhPGUJRKTW9dSrdUfD81JP5zUTM2A37dqqrk1rd1XV2rw2q6t2dSJCZt5Spja2XHcPG1G7DDqhRMLEziJEdV2Xc3Z361AmmeaSkiBi71qXqqow2iZfa6HCS46IBraQkWTjk8DwS5krYnBSZOKJpUgK04+Px+/ef//h8bvn9eRIMkveZQbCGGjcYSJHQO7d2rKsz4/Pjx8/3b9566EI6MPXdx0h+5YWiMBRgCegoEG6ZxojiPChBRmfeygkmTyHGaCzIGegHUUKZYoWCj0AEol7aAdd+nlRRSuU26JenQKSICTCgi7gEoaBHKlAKTgXnhIjg4UbmIWFAWxS18FCSVA4AtB6qFpVberNnAVz4pxzKRNyMgy1FmTTIbMfUD0Js/L1Lpj5WuvAc3MqKRckVvM13BCIhJgJkIhyySmhqp3XqtZbr+6KHLmQJBo5KgwQpCIszGpbn4ibNTdmSoNYnTfp2IFkGL2hZuZuCSgL5ymT5KWbIX3/3Xen43FZzrXuR629tQamvbcNZX1j2h0BCZMMHVc2JVXtujHPDEX4jfUUwEb1lmFoggAiBZhBUAw3XTDFhQ/eTEeCHiAQiQnRESKYKIBABvwTRz7aHKK795FuNqtWz/V4WpbTspzX5byuy1pbVetAQIz/7KcrvLlMKubdtDNz7dbWui51OS3LaWnn1k5tWep5qUtrTa2tSo7nx+PyvGgNDMprqgudntuyCEMXtEmwEE6QJ0n7qfRE64Socn6up35stSajyRJY146Go0eJEstung7zjmQKkOPJzsu5ra33xoAP+8PXb7p6oGOr7YnJPEAt9DXXd8qBU55wmrEDLkbHihGwhVdENOw7DiY6D3D1Xg26NQ9HIMbMyIScJPNMyWazKTbxPQonlaH0i8Pzxhjc1MychTMTASmxEQQCg0t09x7GgBi41UAQATy2lFzcZn7HpIo4NisUQphowC29WnBE4djtQzvMWdIbiqWuT6EraMNu4AxuToRlQp6AGIHRDDBQiNmpn+35Y/vwfX06dWRwjG6+dltWNwNExIyJIYIgUAru30BJsNvR3S49HPK7e76799i15+RKxwTo4IaOyA5eZUGLwJc7csWRmNqgmwYgdIhQcINkAav1T2SPDE+AHYEAJTDlLG/e7g7zfNhPGSACMScEXyrnU7Te3FSrtbCSeThAIrCf0ru76etDfrtPbw7lzf087R8kHxwzhACKqbe2hncmf/7wXUFvz4/v/7baedUTO6tH9Qjr6jxDuvuyaR89jxtZBxECOhEC7PYzM5qO3lRtva99VVRiD2HOfNFiHkyeBEQB6LgxaLuCK9IIFmLEuQ4EEAJ0oWQZGTcf5NWbDBMRjjCCKCDQCS/5Xr/qRkDc5Ewvlu756XmtS6h7N+u2kYYSQKJMCSNH53r21nRZqqmDwYrtIz5yxGCvdjSUQAFkiLEHBaacUhlxxyhyR7i5m7oOoBHJOEmQ0dsdHmCBLhPv7icU6E3znLiQhVr1dV1qq6YGERuM6WaM4sRWpGAaK30w7xMiiWQSZAKEcGWiOc8Pd/eEcTw+fff0/rsP74/PT613QgFG5GBiBhkYZonI5qamqlC9e2+9n5e1trVro6ELDhGho5Z1YR9DYvKtH8kCbbiAQ4J6iIMz0y1lTQCcrTarK/ROGuSckCekGRNznKmt4QA0nJtAN9JK4O7QrVuEs0ARxESQ0AVd0IiAPWcuwgRhphpgDpAwTwmDvFNzDMEYmQq4hBEarg4VO/fz0/KJaRx3bW3Xd0ijHcAow3Sfiai9J28vm1frNoRFRwBdW7hBIsyMOVEiB1MHT+SEKMxJRhNIIBmBDdFDoFRyvpum/VSyJGGJCBbJqbTe13UdukFMYkhubu4smUXMw0yJR9UOgIAZpynd3e3fLOtyfISIMNXWLKVW1yOE9b4cn0/NA16gWwNUpeZCxkTCkphMDKt2dBvMQ1s47BEQ4QCOEYEUOEQaYYTxEBhAHLyV0FTcbMi+xdASoNG2PRwfpoERGcj3APCRKHbtps16U23m3cGRUXIqA73hHl1b62Y3raFufj6t2q3XXtdW19rW1qp1DRsblwh7iKP2Xtd+Pp/WZXFDYSlJyiHN+3J4m9NhZ8LPvcH53N33BeaShGUq9PbtDvGr5e3U1oVMlzjBir1SS7ZmOy1tJ8skeUrlcJCSU5nIWizL8XQ8QykcsCf6ilEEXOKTxJkd0R1emfZRLO5BPYgYMxMixpB0C8kiCWMwP6AjgRMaYYtoACthFQKPQsDgzMSp8AwTyh7U3aMb1A7pyR+PuDToCtq9dwQOZsxJcsmSJEZcB8hD+L2TVmJg4iLIgMgD83MZlzT/i303j8dzyxThnpkQQCO6h0SEw7nVcz/uUtq/5cMb5HelV9LGtauG5x2VPZYU4UrWxxaPAdzRenz4Vf31L5bH7/XcnCbA5Cgx7UAKmYEbpoxzwUlypqxq6r2UOMzxMPNdmViJIgJCVZ/7QgKZEnNiSoQWthi129sR7tZ1WHcEECYCY+wCC1l1bxYr2jPhOZWWS8hEkgPZASCz3O/37x7ecUTF07Oe13quY4N3wpAw8x6UU5FMhR37LkkC7+en46fvzs9v69s3ZXdHjNu6QXFywTwYMXJOUyklCUO09fz4wXtdpv0OEN297d79qGmHq5IdwdCa91H6niMJh25E4ud1qbAambH7wKgLDy7Bwam6acMMiIu6K5g6uW8TIizAYbAWxdbPE75xsV3aaAhGe/rojUOgDSAyZhhcIsMBIcNbTSiPeH46LevJmrlGWAAyi3BiFiHIBJN3WlXPy3o8rWGADsd2xmrRO5gBA2ecD7nsc5qERCCImEUycxok7rY18l5k2OKi3hHAiEEEYI5uphYqhfcPM2VsreecOZGFarPWq2qHACZkJOJXoq4IMKg9BxsfjI0RkQETSy57yTOKAKFbF6HDbjeVbN6fn59/9etfDQCdeyQJJxralBtzMG3oRe3aGlls0uJde+u998Y8Joa7q/t2rqOWz0xmsN2lwRMdSMCDpYAIROizxt2jnqude6iRIVpORDOmmYGgd+AI1/Dm1sA6WEdr5BqjFQogOCMzkSAyh4CLGBlIZMHEhOHduroFEGUuQIziDVGjkykqRmzMR0N/1SAwvHs9tucAdPfWe1c1TTNJQopAjLQXSaKPL6Y9PFQ1EAFpS4w7KoIyWmIgATbrIQ7ESIxMW+840WgRjE0igaXkcreb9lNJkoiYybPkeZ4AYF2W8IjRTRLhZgGRmFLOaBrgFIRbV4wTAgsf9rs39/X8tO+tgmmvtQuvEb3W3tp6Pq2OIK9Q2QEx2tKDCYmSJCZzJ0Alc3O3C304bsv4SpcUcCXm23wMEgEkIpFIbuZim7r6UDN6aedgvjTGbQrYWyPecBEGnTQys+RCwnmewtROOR9PfFpPa1tvKYTcfDkuvWtbe6u91W5qZhDIkJAB01a/6q5m2s16hJFQmmn/dnrz1d3dm+nwMO0eEs58DtN1Wbu1HhFpysyCd3fTbqbWpnU9n5+el0/PtYYrtQ5L1SdaJ17mXA7zrid+EEpMOXn0tZ3OCBhMU8RbhkNGKPirCYGQxqnfNF9YeA+rhs3IYwgGRQAyISFDiIBTALpDEIQDdgANUIBKtDIk8gLA4EiCOeGc87TfU1h4dLVzdQdrZt2wX0pmQIPgllkIgBCYSTiAkCDIlLQlwexSQogIx6Qb2Tu/tEp85qA8n7tgmEURYho8tVEosoBi7WQxZXlTJFOJYp5MZa2qaiwsjBymdWVf2V3QBQgb1Cf7/m/rd79sx8eoDmSe9pGnmGZKmdSgVRDBacK7XTrME3AL0lJ8V+Ch0B5l/RDrk2tA777WGmw73s+ShTOgN1oNX5/IgIiHI8DW7OGN4izwzH4Cr+6VY5GsOUeZUApyQkYgQA45TLs3d2/BHTo8Ptfz2tbaVYckEEWEeyCK5MKIIZQRCXo9Pz5/wqfHdw/Hr3Z378pQKMOAMAIQRkdyB2bKOeWUhPnU2se6nNfztO6Yh3zNDC+W/XOiWbPoSAhBvuGxgxlTEkJQHCJUgYwWrt5AB99EsgvZNYEAOKG7galbd+8BSmBoFhCKAxhHsFVuwoYxH4BQG4pHYw55bFvCBZYLVzcxLn8i3PzKcnM5jejNevfNHucknCQlZEKBJEIg2sJ6XdbeVkMgAo6uNnL2rpIRgYMGfz1ecGO8deFjICAnGV5sN6pKXTc8S6iOjjElQ0RzDfRUWPIkE7euQkzA2syrIroIkpAgU3BKctveI8wD5DzE6LbiJmIq+X53//XXP3t4+IqkAKO5RnT3vizHjx/fv//44eOnx/OyqhsSBGgEWqAHBvCoJwzC7QiWcDFx8jzlPBUkUOujkSnCzHVgFNwUYmRTiJE5mIhGA5TpyGqM2QIpo8jrqB3WFc46xKfJkBNNGZO3tfel+wLtOdqx9QX6AlbDe1AEgZMAMVIiFkokiYUzQyIlU1KkAAgbNhcJSXaSMos4WYCi97CO3SPQEENHdWnwapEzKNhiCyzQLQLAY/9mmg/JN5VwBMHb2xER5kOCPbQ3xMg5UxIL6OpMfUDcwRENgDGVQhd1SCEi4sGHCsRZMAtmISYm5JFT0K7hzjyaPbDWrtqJKKUUMCRoiJzIycK7mXpkSpRknnfv3oLVupyOjFDPJ+stpSTCg48mPkvObUIPDMjqhDboWwmQkojItt5Ge7SGk9uGnoRrF+SGAhnyGYiIvL2LzNlpcDMN1oUNNjBygdurBwbPR8ogFe1dc+ultdEsOnh6yCEszsv5eD4+Hp+O51OZyu3tGDECjn6l0UU7BOcoEjOnHJNNTRkATSnc57K/3z+8e/j6J1/dv7tLBaUA50CxgL709fj0/ETPp7k+3D88PNylxJJEZJoK73Ju815P3c7W1/bUekuoJB1t1fXx/a/mj++/mnZ7yFx4PuwWgK6dot8l3L/ZFei/XE/eKwLg64R8eHdv1r02awrPFVejwJTSRIARAqbjXiQBDTv3BdFnbIEVqAHMhDmMIoDBSvKHHb3Z7TIHAJ2O/ePH/nzyRDrnmDLs91h20sI9eG1qWAueicI6SjDwxhUbzmEpiMFHM9LovNj0dgfq83bXjYDWvYN7QGJKPARBcbfH/T08/CQe/ljLO4bsmIbrmwBytsktAgA8oCo2mHjxjFMEqNlzf/6Ovv9OP30ArSmQOnYqLgnu38jD2wSEtQUSiMCceUqx9QBnxIQNDXRVCpUg8KzYVmpua/TAtqPCJOAF/DPwMiBizmVrVUG19dHbU8ZHjlNoQzdBE4IkTIlQCBkEIRMXKrsyJU7qqh5dtfbWtfuFeCooIoWJG3dAAzaAADZMAlTX9fl0fHxoa4QNMRN3R0BhQGGPrPO02+/3h8PucHh8enp+Ws7dZ0dJzEjpoLf++6s1r6pudXjOEREhRAiAzAjAo1V3QN8izELB1RwGSlZNTc3RECxQTdGaWY8BaKKgsDBTgCs/6uhYdR8qLq2Z2dbWhpu5J4chOud+AXLFZWZdMF2jqei21h4Am1o0SUplKnNORWQUmp0IGaFpW5ZaW1f1kSf0oOE+AIYIUiIUhCGoMFrOBqlyOEYIsciG4WfbtOdbD1MdShmGEMxI4OBIwcIsLCbZDAPBEKN1wZSpzClxEhJwFE633VZMPEnq7t1tUGhHhCCVaffu7bt/9Ed/8s03f0ypBJF5W9bjx8dfP58ePzx9/O7D++NpbapIMDJpPug6iQAy4giBCQJHH23BLCCHu93+bmYhtY6A4W4+kMq9926mET5o/jiEg6kRAJpdJWMCCVg+N+0AsXJbozk4sFOKKMwzQsS6tPXY+xHqU5w+WV2iNwBHChQOSQAClJEzpyQzpolSzgkzVegtSEE1bPDUGTmLc6Eiwgo9nMDD3cgsHA0oDLeyzEBuMhlBh2rd1o6BFIQRjEAJhtIOMQPebl7hbgMd4qodQpgxJ4BhBY2NRlRlHgqUYdMwZmJhDojEnISRKQtmAWFE4mEb3azW6u7CPDaXddV1rSknTmnDUuHQXUVX6KoBSDkmkl2ZkxRSe0rpfD62utZ1YZFSssiIO17VrEZOAZED2BwDBuECQoxAbrTYgapCtLBRJYMYLS03adktht64kyBi6DoBjRQa4qgR85aB3yiWcGvuiIggCc5ZbBT6ehlquBdmCw5Ai7Wup+V8eNo/nY7zNN2eCDgMFIJwQmBmNwuLSJsjguhuTUeb9ZQFEb/+ybuvf/r1Nz/7av8we3SHZtAt6sDjP52P2GhNXqsH4H43lUQ5y5Tzrux87/VYl8f1+em5hlNiLmyMC/R2PFPr/fDwzXRPJWeQ81L72gV0TvGT/TwL/M1y7ufTiUen460lsYBuFq3bsflzjVXZIpiyELVOgxPeAwLRIFbrpJ1xdWzdNUAIDEayCT0x7Pb08G6aiiM4CS8VcpaSOYmR4HzHlCXWWBp1tWZrj7Mwonqi0Yc4IMIMIRG8wZZHzwTGJepyGH1wt3cjwC3ULbEnpsw8CaZM92/pzbfx8NOY7hEYgGAIMBIlhAROzbo2jYYENnEiEQPvPdYnf/reP3304xEoBBJBALOVKe7u0ruvC2VUi0ECJkgM0XqEIiZ2oarRrQcF5khGZFRaCscQsgRQmKkAIqHUG86aESayyDRNJRFBNK2uz4WeORYljYjElAaqShhoaHPjJLJP0y4XBu7eza27dlMLHbXG0eqFCZxNqQMqYjccqXp00HU9H49Py3Ka20KECAoOTDRaSREp5zzvdrvD3e7uXvLHqs9hqlhTEmHed/tR015rXZfjcK0lpdFSkkTgSo5zXTOEjCiJJRHRyA56xCAaNYhuHXpz74HOggk4maKOWNCBcOhbEFzI/FprXXVU/ghxQOEAYFPc9RE/DgBtc4/BQg8AIyF+u1oQEEGEI5dpKnOZJiYenWJIQ69jFPS7uToAEyATZQpnYMVAnklmhBQ21OQ9Eg8MsxMyIzkixKUMGcGETBgO1k21Q/bEiIQ8aLcRSIgEZUNGIToIciJhwHluSRJzcgdGuWXMZaTMyaO14dNEDDKTN/cP33z7k7fvvrl7eBfI6r626F0/fHz81ffvPz49n+tqrpf06WgDgxgC4tA2PQokCBJkEt6lnUxp/+ZweDjkOTv44B913yI3tSFdpjE0TJklBDfRIujDwAKIkAgOSeibnQvpgDzK9AIy4XyXeZZ27KcnPX6w9gnqJ1weYa3R3YkgJeRMvCNOQAlTSbsy3dG0x8yEDh5d1XyzRYTBSAk4sRRKTuIA4BSObiPqsHD3oBhyk1sLISEjIHhY8/W5PtJzoDvodMhlTiF44T19Me7gzkKShuNrCE4EjMQEOUkWSUNrFlyYRJIYCMsIzSOMITJBKjxPkhIRwqBTJSJVba3RZRkDQErJ3XPOOWUA7N0gYvTXmnnvRsRJpt3uLuedTUEOCKDaVbtFIDghJKGcs2KGG/TshpxwcgwkNA9zh3BwJQoxFmFiZpYUEAYKm7zni8DD5bqMf/ySWIvXVt8hRnJo0KgiwjD7G0wGB8XlkEiEEX37BtVEBiQPtCAhzoKJy36+jdq3HllEIgIeZAuZWSRnTmmwiXnXvtYp85RTWyshvPvmzduv7uZ9KgUBOTAFoEd0dTaN3W51W2v79XcfHx+P94f9m4e7t28e3jzcCTOLp1kCBIqU+x0MTXLEiED2qPwp1Otpz8QlhyGqB3RC3c1TKYd/Zt/QcfpVXxvLbd1NBIXB0FfzRW3pXjuY28oolMwNAHATBRpIeDCnx9U1YtC5CRsDJtx04ZrhqVHzALfjWevqmfHtveRJUkESqUZLBeukDmAGtDojmWliv2Ccr2OAqgkwtpNFC0AOYr/V5ELALEPPDdzRETBBKXw4lDcP5X4HO/EEgA7gDqjoCqAI7EHael1rNA811lz0oFb15MtHPn5yrWQeQVFmOnxTDl/j3R2mBE0jcXCixJJY3Ny6Isig9sssAAxAxMaTpwXSiiWyMXUkpGlPb6a0YybndMIX5fmuutR1UCmSe8FzjmfiZUrOSI3ZLRJhKiSJeIBekLLIbip3u900JQB3U3AncCYnChwcTR5MPCVOBBHq3sN18AFzt9R6Xpbp+Pzh43dGiXASLllySql2FmEiNlVJadod5sOb3d2n/WlZ66pqg2X4M5LyV6a91X48nofrzSwpp2maSi5jr4lNWzIAgAiZOCdMicdmdlFxMQANB+2g1a0TGlMIUQJ0Bd16nmPr4B5BagBYhJttkwgRYCPeGtBhH0pD5q31ZVnMgkYXJxFcaLJenZUIMc3zrpQppQQIbkYEkgjCrY/WWfWrQJIgBUYgCSKizMQTAYW5wSgLwEASDY7XsccNLhEbzfbwojFt4LRtWIjAI31POKARY+EEZoaSEjPt9p1EkMQDEF6ZdkKUC25ulCmIMLMcDoc3D+/2h7s8zRZorfben0+nX3///tffv38+nWvXiOFTjyYCuvgG6tE9FMAQBQmFhUTmw7y730+HueynwSF7wcj4KHqMLttBMEhEQ8GXkCDQ1E3DbejlEQsiObwGCqV7CUxEKJnyTNMsmEkd6tnrc+gT9WfsJzQHZyAJmkEOkA7EDMRYSt6VeU/zATNGqPbaFuwOECBBTJgYC0pmycSBOQjCJYANhLAHuA1lF+RBkksIjDCQx04RXld1XILD0O9iH0QZiOUVDBg3qjjKSYbXgxDgDjxyGZiE8wAzhAshi7D66PIeoVommATnKe2mnBMTYbjDRrcMvauIpETMjIijUTvnnFICgCET4tfqUyACp5SnsstlF45gbr2fjs/aq7kxsxAKUxZB5FvTPljh1DamI0SwrXNREULYk0fOQ51AkGzogI6qe7y0yL1Y10FN46OQuJn82PrUYRSJYUjSjtByG3QhPAIARGIGRLqYdgwgdIRInCgxCKW5JMu3M2uYHsKtmTalXMo07aYyT4jo4Vp7T0IQJYu1ToQP7/b7h13KCGTMwUxDgVo1UgApH6MftZ/P9dPH5+en87qauwjvpkKEACQyw1wk+byx7Y5rkiASn86t11WpTMhKNjrKmKxkzCn/E3wLc+Lnj8/+KqklCSVIcUDKNxCim3dlYB9Irswj1Eo55Uzkxkv31aI5iNAEQeIkBAhqcKoRJyew0L489WXVlPHdu7K/ozTh2sCPABGmqAEY1nsNR3I1fgmRtis7Gh+ACMmRcBBaRxA6iQH5zb2AJBRBttGIoghPO7k7zG/uD3cFJ3fSFto3qsHRIw8Q7tp6q4u3gAa0cJxnbVw/6ekDnJ/VlJDMxdOe335T7r+W6UBAup4Xi5gIWThzCbA2tl4XwSSYUASEEBpZo3Bx3vku8dRSCi4F7zLuUsoOiZCu68PDdWipkJdoyMeM55x6TgMrgxZBGJKQZWN5wICx3HMWJtDetDd3RXBhEA5CBzSIYKREyBibNLJFJ6AIMmPUqeqyro/PnwyFacpp2k07kQQ4FMCzoBDSNO/u7t/ePTw9H08eUddlFNDCXhvB219612Vpw4ARduK2nrukdQDkCC51sggikkSSMGUug1uWMcK1Kw496Ya9YlTCzmRMToyeBAhe5F4H8YmIAICbKW+pyySCiJeoeOOUH5zTrfXzeW2tR4Cw5DxSjgyvokTcTZOBT1NhYd8Eq2OUy7VbbWur1bqFb+oWiB5kkIwLCotkBiY1AA0ZFLsgBAIwGJ4hwlXj0gIZDuHmRF4y5zSlnHJOLDKUsiIAg5lSIEUghGM4EYggcTFLo5nMA5ES3tDqDSQBBWYiM49wweGiJma28KYtAGo7f3r88N13v/7uu+8/fnqqXceSgcH/FzAWGwGMBLt6F1finJJMu920n+f9PB1mmRJl2doUcIhIEsVWISUkIh781gidLhoeQzAGEYlRBJlhfO/tmt+9mUpSFpaEqQBThJmHMRI7ag3sniCoINzJ9EC7B8ozpjxa6HmSXNIkIRTEMOTtiBxczTSG0qlklkmIgyISEnHeJdcGatW7myMgAlMMtRKBkKGCC4gITkCkgael909HhVD3w/1+ml7NKyaaS04ppaFH7B7hdTkzoQsXBshClJIIuI1VAAAppVxySgSOu0L3c9rtd/f73W6ahcUHTSMiM0/TdLWazJxSQsQxvQPQPczNzcNRWGhikczE7mbuCJRE5qncHfZMvlEYuUOEWtfXTSRbVcuaGooIIqqam4YpRBCiJJk0Rv5ADdWhm1tYXBgJ3Uc9LIgusHfEq/TbxV8fZXnYuBQvQf0lgU8bV/YlFeh+G/b7IEclDyCC0ddFSCu9dhoHqHGAvwARPKxrjzXMVHs37a4mCd+8u8+cUpI0MaZQr7X2nIhxIEs5iDLmjPtd9v1sz0/n58dn7fr+/afefDn3/W6ai5SJS6atGx9cNwb7cIIQNCHTiLZm61h76nXHsSMUtML+1T43nKudS8fjzQphZgqppmYIwYWxiypBIi8CmUvhNLFMJU8lzylPlFQZRt+4uQDvRPazzIXCYG3RPJ4Wd++mzc+Nuj7c5fv7XGb2gNOHutRee1gEoAHqEL8Fs/DNKd9oAy/tOXhJoI+uRYuIofL8Gkk3ZD1G0UUEpzkfDvPD/d27/ZsDc6qK69llIUZEQZqAijl2N/Ou2kNDF6i/xvU9rAufj/T4KWqNAKIMtPPyQLs3uL+XPMna4vQx+Gxugd4ZZBKa57yQLeGACWGa8pxYmj1r7cbgQoXnh/QmzXvIWQ10gUA1fUUoknPa7ZwhKBrDmqju2MvA9UCIECMgGHGEqykQhkEoa239tCyNMHRttbZWEaOUlGon7kwQPOgfNZQQkTb6aLTARYPBgxJL6V3XZZkyGVANWMJrqxCQU9lNh/18mKfp22+/PR2Pnz6+b3WFMDD/YUPiZ8pvod23gA0CQFfufLG4wiLMKbODEyIzMbswSWIR3iy66+j56hX7StiEOw0MGI4ettEdd3EJYWM+kShFRAgHyJ4AwMK2fQFhy+95qFprbV2qmqeUt1AKX/WDI0LOycIGUYaZxQVdDhCu2mrtTd0CRosJAIwyMLkkzJkJGYLMPAwlCUMiEAK+gFrxBkjiAUNwDXOWDQgskkZ9FKCrdlMORkRkxtHnG0YQBJFYJEI9dHhTeCHjBIABT1cjgAnZ0WOotjiYWu+9tcppCYzaTqfz09Px8Xh8XpY1KDa6r0EhcN3Nw9VNvbFldU0InKTspsP9vuzmPGdKAkwejuEO5gCE7OGgA0jIA8oVyLFJn8BWlhsV1s1Z2YjMbu/H/s0cUzATcRCpa2va3LswCmJT5fCcgu+5fCu7b2T3jjF5gEFD7FIwZ8hk6OajcTyJJBKwNgi7CIATc6IYnAo5JaaJU2u+rqphpoNoDTeEUCIUsNF5EUBEG3hAo5970GIBiAxOdOMIi/Ddfkc8ekAgArr2buoWBu6WIHwLSHC75gAhSUpJORE67grf7fJhP+/nXUmFiN0cwmgIdONQBLEBqx8u70Z9Okojau5BhMwpCbKIh7W2OhBTgnAR3s0zU0R4115rjQhXc3zV105MSGTuGD4q6a3V3rqpgjsiJkldI6XETF21NlVXDxvtl+MgVdXcRUREaLBQXFJnOJABYyoPYiW7qMfEhTOZrv3aMMA94/kxp9zczUAd3GmIThEJIbXX3aGwiS1tkxzcTKN6V1TrpkoAwryb5/28m0phpqZ1acvpvKq13ZQxSuEkzMicMDLTVGLexTTvcs7Hp+P5uJyWFeBxWdbdnB7uZ+bJw5r2ZqpuG227eZiRO6mta88N9g5TxE54L5TQE/rDnDvm51UA8XjrayE5UHM0w0xUEgWABRTBWWjKuUjJzDlJyTKJTMgdUpfULHXrk/Ah093M00TnBZbVW3UN7abWLanN5G8zlUNCll5jae20hHoExSiRBLg7YtjgEr7a9S2m2hoheGTnHcLdHAxjg31flzkxJRyAT2SmaSq7/XS33z/s3+wTF7cwNgNyZBTAHJDMzNrA6arXaCf8+D0+/RyXM9YF1+5VMQIoA83Os3MOkkBCbbA8BmYDMkJMxBPlnMUQqocBGgJmEuZuYBFB4BmZ0yTTjmY0eVrOq9bFfTV2/6PreZSUdjOArhydYUm4ZNbM0XW0oQ6XdKPqD4vAcEJVW5Zq+oxWtaGbuVc1ZUZJlIQsjSanQaEMEFtjiAOFg3VtBCQllTkCTQ0k0ENb79rOyzE8NE3kOEnJKb17++7x3adf3R16ryJk2sPs0tZ02a9eLRUiIgn3Ac6LrSndiJCIB8F1RAa2uCiPwyjFxCixBIKHmTboC7YluDoquIK5mZmHAQIPxNKoz4GPOSRJWBhjm1TDuweALVNwYT4ajezbZAQYfVwDevdyGiOkMu1QAckDJElOmQhcR7LEXAODCK7ZZ2MBZM6JRMgUrYMbozNBQkgQ5D7m+TU68Y2fi4SJRZJIImIEGjKrHtG71bU9PZ8l9TJ7mUueCgsxUUS3jYgrBjmx2UjwvtgSdWvadywzy4j3u9lpWT4+fjrsv5ecJaG61bogWkokiUQocBPGA/QIR/RBaRXgDqpB7F1dHQKJSAgzBYeCCTAj0WgwpA0KDWQ2ckmIg1oWg90JHS+Jl1GTdmZgAaShJHK75GF3t4MpAN2ttl5bX1pr5sYMwoGonGxfcP+zcv+n+903qbyhHrXW6meCRYqJKGtXa5qJBKmU6RC49F6rhlqYYfjwwFtQI2IULpiRs3Hv1NEGqXQIQkJIFIyOoRE8+h+ycBEiCIxa3X1BwF7tXl8kqaep7L75umvvXQfexCGbm/UOboQQ7q7aEdF9FNKRaGTpp4ko+DDLYcqHaTflmSlHoJmZBW0t4OTug6vVL2OEuwPpEBGIKJJTklElWc7HtbVUapIsEOE9Z4ZItVXYyINiI7O73hOEMhWZ02iCQMTW2lLr+XTqtUeEiOSUuvsow6lq6w0ZJVFOOeWkaoNOf601pVSGU840yCJHyM4XZ2XIWmu30ctHxADhNjBhW0Q4GDGu555yighV1dZtMNemJDmxyC3IFDbC2rikKAA8DA00gECE57kc9vu7w34373LK2vqynD89PX/49PH5+GymD3d7e4CERSAxCiIMqkQpkKdyuNuvS11PtTfzbuFWawsvmdNS9fx8fjodz3WhJJISAoKGnzuclRffO99N+7uc37G8Yc8EgoEZd04HxpUCbyKsZt7MBscPCyfGkpEQZ0lFhIWYAMNEhkaBeYBBB/IksEe6m+B+tt2EUrh1AqSutnRVC/CBZ4q12+OxNm3Hk3/82E+rA2LKqE4OHCAXpmrHi5TFlk/Zps7GaT34bQdbdAy+rptlnjIjcM55NOxOU56mMs+7w/5wt09S3HJ0NiRFAHdVs3a2emz15PUcvuLyDI9P/vEp6gKtgVrUcBtBIHt3PZ1bAIj05dTr2UANJJggMaYIaLEstpxVyTB1j1aVez+Z1kKCmbrZqS596dr94/n4oZ4/6LK4sP1frqYwJdwVAgrULrCAH90XwxZhiMF4BesAhRMAExKQKZx6e+qf6hLrWQkhFyCMgV2epkTE2oPImZ0QENBta+NCQHFCzmU6zPMdUhLMiUqiDB5ICcuMiDmVKQmYEspuyg8Pd199/TUxLeezaXNT3u9+1LQPFE+QoweA+iXztilCGIa7CBE6Ezqia1g1ra4tAoIIAsBb9MX7mfrCqblYmDk6uccW3REFog+H3kf0PgTXNxdhdH6OwOg2VTKgJKN26R4Xjhz6oWXX1po2VCUSFBlseGNxtrVrU1MHh/FNsWlvgCSRxEwUBgZDgyGJTMJpsN/B6L+NgRceZp3GEeWUcy5EhBeVFzUf/PmtmnYYiCtBARQkModurqpmo0kZ3YIuSMUxzL2Z7oAScAHqQM36WtvxdHp+fnp486C9qLtZE6FpyqWklPjSDxQbE+SgFxmlUnQL1ejqamBAgYwkPPToImCriCI4+riqo4Yz3HNExKBwBMVQQicMRghEQwRiYMEAAP0cAbzb3cWMaue6Lq2tbV1dgwjzzH1vvHMqUe754Y/k3Z+U6evEB1y701ktEYiklXGBZtrr2pEKpyxpnmhKpWpzC2vGFsOfw4DqVDiIUhLKWVpXZhvNJ8CIQsAEdCFIGNT9iSmlQY2kfV3XTrSo2o7sWhfNOb/dvVnXZVmWGLEvEkCsy9JbZWIAULUIGCmaCCDEktNhP1k/EPBhv9vPu920S2nCYPOA8FH5Zt5y3dcxLni4W0Rrw59AERkUgO6hrq03b7V3TSllJhkgGH7VLxpBY0pcRyopzXnQMkBE166ua6+1tohIJqNrFRHM3FzdPBWZucTm/QcSOURXHemZ1vutzSWilISZEcnN19p762aGxNNUEFCHo+8+Ok9EJKU0qAI2BpVwNavaeutElNwmwtGoc7vULzH/BVMyTBEBAaUs+8Pu/v7u7nCXU4qIZV2O5/PHT8/v3z8djyd394ZokmAClSIiiTkjCRJDmvLh7uAa2nw91+W09rWC9t283097Nyeg0NBqHDg49kBDu1tTaVbCOfuENBNN7ELGjJxoDrrLvGjcnoWZDx5AAkTilKkwZaJCsrEyQ7gHszMFgHW3Ht1AhSML7EtMyWS47wgB5B5uw3WXzJ4zesBx6Y9P9vhkxyO0BiTAAm4UwYAEboOw9xqoj600tj7HrbiCW6uNDm7pWwVqQpxzEsbdNI3dL6dEIpJyzrtSSirRRM0r2AK9m2urdv5kz5/68dSXJazB8ojPT/F08t6iW7hBB1cGDwB0jVhW1R5g2KrVGugA7MIqBLBC42irL6sZA6SmurYG7orhSTgQ1t5hOeHibWkfz6f39fy9nleQnz74Vb4jEc4Sph3ijH70eDZaDfswEkTwQs7gG/iGSRC5dT+f+uNjffp0SoJ3d7lkESEizFkIQ8kjFNC3jVXduiOSsORS5v3Dbn8/7w7gJFSKTEUmhNFAURCBicGx1xWBEWmeyrfffjPv57quvTfrbUn784+adiYpDMERYYqmtJmzjVN14yogxCJZ0fqyLo9+nhsG5gMju7n2hfqJ/SRxBnI3cBq6bSiDwwOJHQKGylqEB8K18LDl8gbBCFpYqA8QAQCklMo0zb0jInNLKQ/A0RXdc5mPsZyWta/InMs0M9HwEltbT8u6rL31IXq/UccxMQgSMyciRmRmpEKck1AuqQjzUFmMMAfzwLFoJG3ZSCBE5uAtFYExYpIQgVJkf5jdAwnDTLsSYyBoRDOvzbR3dKTABCz06nY4RHfr3nt3J0AaggwXXJuZWyBSkrKb4XA4zHOZJlHvMKQ+aPQfAeLgdR+ukrureXNvDh0xmIhJCBkcHAw3pxQCAdxi6+DoQ4IMI9xCW2gDVwm3cIsYhJhIRBHh+CpsR8C5vPFM57qq9vW8WrVZ9rxL/jaiQm0NwOc3ef+znL9mOkBkQwpBZmRgIQzrfY16WhfSKCgPh7sp5f08O9rRam+OFYhxzNAOXYyycALKknLxYoEdzGwAtQd76lCwItgkSzaEIEJ4MoPardvy7cGvUmPMNO8KC4jQplVDfFkgwSkhyeBho5QJeUDA73YzwrvDhOD1bpJpPpSyJy6mAOBIA1RiFzDZVlwfDisz9961Ne29t1Eao3Evt3J1aJhpLKEdhH20grwkv8MjwMzJbxWpB3hiBGIRAASSOOVk7mYeAM10rJANNIuAjhauprUhACCTpJRLdvfaa611HN7YyZi5lJJTCgBVO53X1joATGUi3gmLQ6hZ693dh2lX1RG1E5HnLXHvEd0N3XTLB/pkk1zuBwIMhxuJaLjniIgoOZWS94f9fr8j4mVZPz0+rst6Oh6fn49PT+e6uqu4+emorqe+wvOh3R12u32ZDoNZwlk4Sco5ySHHA5g6qLPHYcqHknKeiNJ+d7/UOtQDh1RLpaXjitRKc/SwtSqDFfSZQYQyF0yHXVkioN7cjs1ihgE4IJKUjEUQHByDUBGBN34P0HDV3qy30My6z5EZzGyt6OpLhe5OZFMGRkky3c9yKN1Bl9Yfj/37D1orulOZSABg9EB6uBvHUOfc4oGBqtpSoJf06iDXgNAA21oybkz73TxlpsM0mce5NQJUjW6owECTsDRtVk9hFVpflvb8rO9/rR+/19PS1tXcaD3j8clrC48wiEHW0RywAzkMh68tcfoI4SEFhUibr88BTRtDInVD1QhBTOA9bApGFGRNUMH9VM/P3Z6tnfu5txqOTIn5FZsIWoLueuzLB4CPxMdELjSQBChb6nJU3oCZc07CCVHcg6Sb1+O5CgNJBMaEiYmFEewCyxhX1NF7gEMuZb+/e3P/5idff/vmzbvdfABDhjyleZ52U8ksCKhqvdd+fj4/PT5rt9F/9ZNvf/JT/hkgtLYu6/KL5/WvHl8m1mcJeSRGhJGRBeaLrY1wc40OAOEOIYkLkasmPdnxfTfFsggl9Ahr7AvBitQoEIehiwt57UZjtcW2g+YqcNN/ughkb4cQ5t5bjYgB4yNiES6lRGwcHht031+ZdoiotdVeSZiEEAuAmdXW1nVdaq2qG9zbMYAccCBEBIEhBEA2dAGmhEkkMTGgO1i4WhCGE1MuWdKWJNy69AgDgAaILAIhCKMU3vs0REtHZm3sWTYYaAPcAT0IkJmS8K1R9MHLbtY9jDEkRgcKMw1ikwggTokQkOZpt9vt5rlUjYiR68bNqQm1UVkx940fqHt0RCdGIRHOyDywTgABAwa74eKHVNuW3TV1bd5Wbw3cKWxoNsNWFdk8K3glCoWwK3ea9Lygtt4XRcNSprTLjmZqzTNwzPe5fCV0j5HdwUKCxvohCguv2rmtUL3pOtiriErOQPt2jm7qDUIGW0you7MBxahYpxLFAwgGkzrCUGMdbVdBQ4lcw9Q37EUgAmnvAe67FxVRRBj0TcM/AoDB/GPa3YwlAbH2Du6Dsj4CmGlXShK62+XwLhhJMksBoAEQJWIcql0MF8b4AAAzG9Z9tNfThlQD3JieHHDjwMDw0O5oFgKGhqiqw1KOuzEybrcL/dr7sIFdhKd58vCUk5pDxKjFXYvfSJhKkpwGiH1g75FQkvSuvbfeNaCNox0qCiLCwqOwLsyRgIjKPE3zPLL3Y18xMxofONh2mUdHAPOgL0cgiggilJRIBNvrWg/B4I8eFLhwxWGkzMwRUGtrrZ/Pp9PptJyXZam1WjgSpAiv1aw3U2wtumJz2GNkR8qRzD0xJOFBeFtEkBJgIUqI8wwWNE2HrubgHoPfx9tc+67GaUnndVqNLUbZGXOCkjFJQtjP5eyB7SYjP+aKe8AIIBBFKJGqRQSjEYJABkTzcIvRKk1kJcdhQoYhhQW9Q+uh4SxRcPDp4v2B9zOtFc5N1+rHs/WOiMCCgw8KAcwVQwFs7FpbDvSqKxkRo78CgdApDMkI1FEHBPlq2vc5Z6ZdkqZWEd29Nl26rx27pwkm6/O6FD2hQn96qh8/tO+/00/vba3eewCiGvaGSKMZMtTAPcwBOpCCWah5XeLxPUDgfEdlxIE99OQVu6DF0AonQMGyszJFYk5EkNyg47PHM/an3hfrEZYg7VJO5VUo4p30HO3Z1ieik6fVncwZKOiyaMIvSSLEYTIi+FL8GjnpsEC1aGqJgAcNHEYgYOBFOw8ZaCrz/d3DV199/fXX39zfPeynvfegSElyTmWed5LIvPoSrZ5Pp/Px+bnVjoiH+/s39/e7u0OZp97bsp6Xn3/3V4+/+rJpH376kA1l3phjASDMDTSYrlqDBbMwAUDr6+n7dn5a0lAmSUlIEkxsmSBxCARs8uKIQWEYFjZ+xUCMgJcZ5HjpqRwi4aq6rKv2joAiUkqJACIUFk+jJfilm/Zm5xpd9g5Dc5zMomnzpa21r2pDAnq0/QQKSAJJiDiwAjj6R4USBAeQBxEQCwkzUBoeFzGllIllww0NMTRCxqCAMIvQoZacGHZz9kuWfPBrICEHJxZMJSFjhABNKRXJ9JpdJAjchqR2GAYkkMTTNM27naRMyMyZkQMopXmeD7v9HawYriVxSpISm+u6npvXUd8Lcyc3UwBnwSScUs6pEItdGBbjKo/QzdUxUJAdQd37auuiy1mrmtIA+w/OFd56qOKFlfRi2XFfcpMUzftioCQkueTE0tFK8L1MQSAz0549RaCbu6M5GWZCsOgYDeNgcbbWW9d4WhHR7ve7w26ubq7NDL0HjnZ9JIdBvYejK6XMgIyto6m7OXT3HqOHFwTQR3rbOxFcJNfGh9zWR9y9926X5rOR0B9xdi5la8O2iE0vyiOAkNJwDqHAIHuEgatQ1VGYJpGU0miOs9tBl/6ReZqYsIoMJQR3V9OIcLdwuNDBAcAGMu29q47ux+2w/TMqzTBzHnqfiJhzurs7zPNkg5lmK1PFqJ27mfvGyC5XpYgB1fOhQdCX81LrKjza/JgGMAeH1kC8eRNDDGok3gnRdjtV7b3H2G2G+lNERIwP8eHWa+9mA6ctzETUfrVa1+vEogvNNBBA0FYy8ui927Mdj6eh3dV7b72bBxIxU0qD/8nC0CG6xdINzrVjrN5LpVTGyVqilvgsnLPkxFKYJ5HM7BbAWUrifNVZHkkUBDV9PsKn4/z+cbf2/ZR3uyQT4IxQiCjKlCcLeHxZ5hRAAYzhEKp9VW6eeFSkwx1MAAhhUH24hVkkgang/Uz3E5niaUUz0WD1iHBJMWVIjFl42kWZ2JHSQnTl/4aw8ACXYYasAzQMe6kI3ZSGPJwiApUIkIGpi6lxJ644WKW2uwGZSEYD1dg9ImKB47k9r/pQYUpSaz4+5acTPh31w4f68UN9frTzKdwRkDghMZWEmQHce/cw7R3DwhtYRa3QZ+ud6kLWYHTSOYRxkIcESLh1VB16HlEKTAUSQSLL0hNGLOQL9JNpC0PMO343pZlmvck/RF99fYr1GfqJkjLhtrSbGToTUfDghYhAZfTow//sLVpzJr6/uyOCMiVAq7UpqIy09GjIQg6AABR0ET7s9vf39w939/eHu8O82+ViCODExADoAa3bsi6PHz99/913p+eT9h4ehNRb77XGPDGilJKzzB+Ot8v8NYc8AAxiZ9ym7CCndYBwRBo+lGnH3kSABffuXNfQY1/ZyiS7XZJpkjwLJgKm4A0Ix2MRDqMaBEyvqYzGhBrNx1eocNfeWmu1QkDOeTDyAIx8yMZ1M2z5a2sCwICBJIDsDtrNete1tqE3MvpvCIESSoGUiQUAwjQ4MVNJKSdO4QhBhELMLMgCyD7CbiQSSQA0MsDb9oJIYBiOAe4e5mEeCu5gAQYYjkjDByAiLMiTEEgCD4RgxM/k2pGQmMAoEDV8aISXkne7eb/flzKnVFhyoATSPO3fvf3WIU7LMbyXJCWJJOl9fX5+WtZzrbVha9ARWVhk8BGlnDgnyczJzBS6uZo5jrp7EAELpaCRIDNtoBVcGYxHfi625sStvOtwsXw39xa9e63r81qPispcGNGB1bnj5Jk5CDARZhyC7xbm4A5Gg6ephO+RHjwpOqCTd6tVGXEuKd/vDqx9EW/ohkCIgGiABuAD7SYgw1wTKmqPbm7R3A1psDVoeHPn0beAQ+hyyJq9SgaN1GRce7ojIhx8cME6IHjQgDmNpsDL6whxgFcV9CIUuWW6caC86aWbISJGY/NNECwAgIhDIGW4CNeU+/CLxzvNrK61a8cXxeRrtfRlrEvtoOEOgElkJCpEUr5A1i/ICwCAYYPdHWCYxku9YHDCpHAzoTTlaXzpgKsM8SyIAMCc5ULSDENqIee85d62yu4l5XNJqsfWfO+2EVsgAhDg03s1uKL9B9QBYPBbbmx44O7RdfQUjKrV2IYdEIhZECycLRxGndAiqln0ZmsoenNOnYQMURkqAcrmJEtmzsJJRqvNlrmmoZCOxMRJRBC5FN57rpa5yTzhLF7cUgA7RuRpmizwBkdndtWzHLNDu4cE2YCChMHgY4jQ0TIANom/2cXdjnaZasfq0Br4hTcQADbGfmEgC+xmXXsMquiIAcQBIJKcCUuYum678IDAX3ni3W30XQYqCJAwmgMqcANpwHRrQRDCzRtEbb23rhgGcDrVx+N6v2uJ7fkUn57i19/7d9/rx/f96VOrNawDC+eEwpJzEiEEsK5r9JUQIdxAG1AFbaDqvUNvoBUDgItLDQfADinAMUyhq3dFN+wMXTBhCHliJXDvZA1bdVMMwoPzuzvOc7Ib1klri54fvZ5AK4rjaDt3NA0PNzFBDA930AhQa9oQ1B1NwXpAxDyV4euZmXUPBObNgx2cQu4QHomZJc1lmktJSQgRzMMcL3kTj+hmZno8LsfTsizNA3KZNzoVD229nldh4iRDnOB2mb9OyI/A8pKF3/ZlGMmZzYcb6Hk1yyXPu0lgnwOjV609vLDsc9pPsCOggE2VcMCuRzPooIe9YHcHECOuWR2PrWRmaq230ZAzEL7jxURbtlB7dx/91ER0S342sJoSzJSJJNSq92hVWxuVxEFJh0SUJywzUkZENIuIIJZc5rmUJFfTzojEHMg2uhXGjj1w4OHgQYO0FQcMxRwM0Tk8rMf5XI/n2sw0HBiZMbFMKe+naVfmKRcRcreufa1r7/329ozsogAh4MBhi6RpKvvD7u5wd9gfpnkHmBwIiQ+Ht3/0M3p489Val3BNhMJEjOty+jR/PJ6eTufTeTmv6wqAInm/u9vt9iVPTIkx8YUrTxX7WKeDDj0Ig0Oxm1sjV0YviccN1R5124SDYosHRr/ijRcccfr03dG+f/ru+fzYsjABNjt1iOpq4LChTgEA6ALa9BjOkUNYiPhM6SvYlzTt2B8NHgOrOxgjfnW4fyB8jnqEumJV6ABhiH0w7OIAoBAD5gEDNAgAVY8e7mEUQWDoil7BjEBKkpJQkJnjBiiEgMyCI+DbSmYb7FFEHAadBzibCCOhQ5ibuwO4dg0I29BjNibhMMddzbxutc1tiV3awLbu+cGVIcx+vaTjg/BC7rRZRI8R0JdSmHnzIF43tUPA08cnf44wI6IyTaO9HggkCfJAVLpHMBGLqOq6rqrqZjnnqUwDkTAs7jgY7aYaqh0A5hlFZMTx7j7YMQd7rrl1VU7pcDiUUlg2UQa4bEoIVyHZ4RLR5coDbVoON8t8i9u3QhhcW1HDvfs2hwIi8AL0HlvbKGmhGxCjQzi6hoIjGKByVFKL0asMHuFDwWaj9SAOog1eBh546SAb/QjMlJPcl/md5J8w4X7XmJeRqiMvgEKR57m83oKrwlIDIIicRT26eqiheYQ7ggUikZO7RzD4xHZX7O0+drNwSi6QwoQMawtCc+oarUNgAZ5qh/D+9Lw+Ptq6uBvQ4IVg4lTS9IZpF0raA+wMcbXrW4djgBM4xRp4RkbMCbqYY4duUW1Q+m7TKlpv7h6Aa9fz2hxBPE6n5ePjcVd24Px4PL9/rL/8vv3iF/34pPXs4CFIOWHJfLcvu90kTOGxLqt2u+YqFBxXSDW0h1qohgNgCUjhHtHQj4QCqQSSUwIw0k6woDt2DKEYi009ukJTUEckdLJ1sV1xOLy0kFhb2umT1zOZIlAAq5NreA+AcAswIOQAGNQQ6moGYQEG4MjIImIRrambocOQMMksjGKOzca2EFnSNM2jwNdbPZ+OT5yxR+KSZAYSMLUKrbfjaVGFw90bZp5K6bWenp8RwNVOz8/n0/MoPZ+en2/n1WemfdNPRxzijZtdpy2XjI4jHxihFg5CGQnDGC2hAXSBzqCMzkijv5fGmrh2ssG2PvES2m3dMohbTX6YfHOzIbiCOKIEZkG8kMurde0DWz7S537bYQnDxR/6Ztib9WZ16drddftC5lHdJGGOCO8OQMQygHpTLjkJBF9NO4ACmKNubYmIOKrrAILClIRQKMBW8xY+ZmT07r1pa72NQq4BMA4rkIH2Ke+nOaWk1pe6tHVpvd/GiUyUR21WWHtoxDSXw37ezfM0T2Wap7JzEAc0C+Kc8nTX3/Xe3JTARsFj3S2lHA7n43k5nZfzsp4jglnu7x/uDg/ztE+ShRORIEYADwIqJmdKBIxB3gNdoVMokedELCwYrkboo61u8/m2uzpaNq9rPuLDd989t/eP3x/X1fKDIIF6DbA+UnE8uGsRcGO7wYFxg43hDMmxIDJw4kgckrp1chyIvcO8k1KKL9nOjxarhVMAkOFGahnbx+JlIjOBoROYuQaiB4GCK1oH6+QKwREkzOS3GOABs9tySwg4ij4Rg4RvOOME5GwjsxQx9Gh8A6vDaOCmQbdwyei7e/RuwzaPGXvTqQYvpv5iv+Fi7+GSPx1PjTzqxdelYf7HvRZ8tcyX89JDVTsizVMTEY9AQkkMNDgbHAJSklwKDHnA2lqrbW2t9PG9eAW9Ari5mvfeAQCRS8ZlqefzeSTVSko8+vpMm2qZp1GJL1KI8LYnJK7/Xuq9G3nCSA29TswNX2v7ccP7Boyyd2xyHFtIvwlBBGEAB7A7ATNceL03HJt6qDliuMdIvLlamA+WhxguLA6RptGAC3ghzgN3dEeEnNLbu3s93E3zYSopIs7mce4cugsrYJl8aa9CkdZ8XWOgbimNr6Zu5KOM72YYxD0BYGCimMV2yeYULOTohhgcQRaoCImQB7O0QzbYdc/huqxPy4K9Q0Qgg/AoMU6SDkw7lebcwCoAqGnXTpe+GnRAN08n9CMGRpTqZempQVeyfqNzExG1965mAbVrbT0QPOB4XD58ekqcWrXn8+n7T8cPn9ZPT309u7VICJKGdJscdmW/mxCwd12RLEA9ultX7x5Yo6+gPTwAJSRH3jkStDPAQngiK2EcWIIEUB0JzSEqeoBuUSR1t+bYDCxQBKfV19Xa6J2/DNdmbQHviBCA7hgaSDFmqjXv4CIEQOFuGmtTVR94KQGiREnI1WtVD0uEQMQiiZNQqhqw5ZMAi6RcALG1/mRHrSbO4nJ3kJyJmJFQTVtXs5CUp2kqpZSU1+XsHmCehNZ1OZ2O6hoYj/WmGfGLUfu2hDYRlxjpdAnyxDHANRER0Xs/n85IOBqzB6jXTFV7134hsiBkGoyRAMjIL9m3CLgh9L48ENvO5T4WWBKhhCkN7DpqN3NXVe0DKxQsIuFjE7lMMugNNJATmaJ3a4uvz6bNY5CjE6bCBTEre0ePUIc8yVR28zzPU55SzpJok/URHMn6UAczDBu6iBHgyIiJS047JkToCmYdPNzcunkzc0RJwnmoHgMBJJZdLvt53u/2u3lOSbp2B8cjbqy1l8GIRXgnZc8JK7Dx7rB7czjspinllHJJeQJMHqRm4p5gnsvQ0Oum1b0DuqQ8TfsH67332tfWlsGem3OZpl3JW+BONKRQGAIJeOAVKjAoNK1RITqRp8wUFIAdvC4Q6E60Je/MB2c4BzDcCM9HxC//9vvn+v7T+zUw7h8Gba856PABRoCIjISjLZLQiZzAB/4cARmEGrlSYAgo4hFiiQgMg8J5P+8zJlGyVaOHjs056FKbjrFDmpmrgwE6MpCHx5AkigiGQAaEoOHEGTYnJH91Q0YK2CMAAwEJwwNgFDcswD2EOMRdB7x8q2oMGAcRTdM0MORqWtd1iM2M2Hp4RgMVPy7bxdKPLK32AX/aiNw3M2+XPMDoiUfEUsp48EpKX6aJqNzaxd61Wq1rdfe6NhGGgeZn9PCROiLCUqbdbpdSIiI3b7WdbYmnp1GAvcmiDzwfDLKdnDIitdZr7dq7m60IEK4j8HffHQ7TPOeSU04IDPG5yb6Bl123BID43LAPJ/ICSIKgGCpRA1eHo1rxAgXzrXqCQ8FiGGKkGFolG6OeG9hWvt9st0eYjiKK+kjvx8iUD1nUrcp+5buuZoEtsOUpnGF252Vtn454Wufed2j7KSqliPurA9ebr4sRRmQsExNIuJiKu6pB7yoYnIlZM2EWnBMkAndvXSvhueHSorVwRQYpUkRM2JEFeHK8t0D1j2rFYQVyJhgwYaSMlIkLSiHJoBLha6sCFNnNFJVAIsSKnL2flUnAjos/nay6KXq9wfl7wNJ679osurqpISKCnk7Ld+8/qfrT8bys9fH5+Xg6D3G/sYSEsGSeJ9lPeVdyN1+7ruanbotq7dbVOjquIBV6BcCY7kJS7O6ineH0AXnBydABekHJgOJpwggwp1DQjqBoDj2iO3aLahARE2BXa7XXpZdX3rRjGBIMImR1hx6EniEQoFUNg1IIGMExAs0GZ2gwc040FS45NWu1WYRjJkAa1RomGcrmGtADJiJgVvPT+dxWy5wTlMN0/8Ayz3Oa50BeakWEXLJIur+/zzmPUtncOgGUlFq3x+fT8/GpteWUZpjur+fxyrQzSxq6zjH4RS9KTRFODsAIjKTG5gOhRII45ECACFNm4gTAHhSXLT5wgNFGQXZk+nHDZ1wEOwBHGhAuj0MEAzJyYhRhSikh0mghQ3TixGIBhO7EQpzopkxNiIfpXqOXiYHAyJIpl2Y0FnsEhUwyzTLPeS5i5uZWpv1ud3c33e3KvkhKnJiGdWcICNJwdQwH102SZtObSVyKzEwYweTq3CiQQELUJSFoEkcerQeBAEI8pTJPd1M5bJJ0mJJGTueeCW9yjkwpyyHnKaVcUELblA6Z9uCpV1gXBzBAjKDRcwWIAeiGbqgdfetCY0JJlCk505R4NlOPICKM5Moa4ejDuQIIs7CNVAC9ozfyRtEYNbETggBFICdomXZKBpw9bIuuNuMnCW/IEwLWZ2o1YwNJJL5jg2gJWSlwAL2ZRKIQCgWFOZqCG4aPT0WmIIpwjCAUFKKsXoJsAt4h70hmoVQQi1oLoXBzjEB0xAjCCAhDY3Imcwkr6mpO3cUQAgiC0RkYLZEHD17ysbpfFDAcoBmao/oWUXrwEFAnYAOwIXyJgwACbORyiSDcgihIgIkSMrmDAlk4IRmEAnkEALpTInJAH3VcIHA0jd6964bRMA/3rZSuMUSRAYPAEYmCIILUrDs2RyFOlIMT3PDRoSMagkFYOJj5xh8HBObWWx+szAyqrBxMQugIht58eBJb1D5g+gAIBIDugULW3MXBAv3S43HREDQ3D/DuVlWrmigMmuYrBdpr6x6XKAO3wja8qo8EUpeBSBq1+TGFcdNw3MrOIz2yYQJHrQcuSGFkg9icGiByYRcxHlJ14eRmNm6GB1qYwpCSGD2zG1M2bsCD4eagk3WuDU8NHjOsGrhGPVk86Vx1R77rYcK3jsoQgGT0RFBQMk7smSxBOJsNxCsww4VTAILMoFZ3oxVw7dgqacPojJ4YMoViGLi4sqFEkNsEsBPpJWMSkJRYdkITAVOQDOIMmQjMgrsDG/hovogIcO8AFRVQFM4LnBdojgqi+nLbEDHlXZCGOUqEOyIKE0tRx6U7DsJDR+Scpx1TBrdEMGUq+5LnnZQ95QLdQBCSYnFRLJZ20BU1zTZln8QQMd+hJJj2vhp0QhGZhXOJlE2yc3ETECIDMiRoCArqgA7oiA5oAYBTwjIRigTPt5ZdUirzTJiQFGVYyCCMRJAQyAOJgQqxMIWAJmtEJgBZqKRUSsm5ZEspQ4SnwlIS50SpECYKZ1d2CXbOe0xzEHtEVVP1c7VT03O33C1xd7DzurbazTQTVXdTdfO1a/UYNPyL+bHp81rXZe07gRtlRPwP/+E/vKyozwFQr8e1Dnx9CX7+kmt2HW5fgl989Ze+6NaF33KWl3a422O4PHUt0yPeGsXoqjcl/JG6/OGhbgcc17rA5f+Xo74e/Q9O/dVHXYg0Xi7gzbWKm6+7+fbrd20vH/isAYy8vNUhfPvwgBcCr+F0bYw0nx9T3P51C1V8ubav7t/14G9ednMTLonf7QAulzS2w3t9XfH6LxG8GMUAvQKzNi7dH8ycm4/48Rl4PfYNvocIMDr84npUN5He52+9feZSD7p+7RffwvQis/tS6/0shYw/OJlXn7zdvJd5jNd78XIwcfmKVxfg9kq9oPe+cD1uXvfq5l0P79Z4XqngL2/E6/tvb/7tWr4uoc/xqp8PvIAGrm/6wuuJ6GWd/ciK+sL4ASIw6McPJr744+W3+OHj4+h/8N2fTacvnv7rCXAJYvBSWRnVmK0hdLzWXqmIusdWwRqAgsv9iOthXuoT8PKy12f3cjNvD+SytuNGlu+6rW3YxJfzum62r3cEBHgFnsGXw7q8Y3zCBdQcr9563ehG6Q5uWUgup3aBSg6wxbaNxrbrxGVt4IVUOG4AJOGXU4XNB3yZUyP/8vpC3a413Hx4xBtCkdEXcnN417V7eezm+rwyLS8LfOM1gwtJ+vWu3q6h7YSvlx+ANuzrCwnbzYVCfDnnS4MXXDSZLo+8atCHm/EyG7848Dc896PjRzfa3/phn9uc8SBeJt9vemeS9Jue/zuMm/39N77qxbj9vh//QwOBI59x+eU6CS67zW/eZH+XETd//5YDhFdXfUwU+h1PE0FerY3f8Tt/4ye+zPPXH3bdF7/4rr/PiC9s7Js9/E1fFj94WfzgNbfPxGe//y7f9SPP/HBKbR/0YrO/8NSXPPDrRf3iu354KL/l9VcX7O83Rp3vt7/sN/z22fjSRfzd3/3qc15I3DZzHD/yXsLPOl5fvuyH8+O1If/th3ETWfzG3f31c68+PuBqzn/TQGT5Yb/VZ6/5Ldfv1tn8XRcr/faX/F6DiH73D/1Nl/PHH//hk5fSFlhcvJXfa3zp1v5DX5g/jD+MP4w/jD+MP4w/jP+p4w+m/Q/jD+MP4w/jD+MP4/+vxqsUyvnjz+vxPV4oLkcbzYVtCWDjih2/bgRzl6cuhTn6LDNwU5MY/8YPn/xB4ehawoibsk5snCijsnD5Z1QZIk13ZfdmvLmU8u///b8vpfwDXJ7/Y4eq/sVf/MXxuJEK/fN//s///M///B/8W8yumF8dxGHjD/5OydbfafzN3/zNf/kv/2X8LCL/7t/9u8Ph8A/02f/HjYj4y7/8y/fv349f//iP//hf/+t//YIM/19nvH///i//8i+vKXCrDQOILxwWF+zGheVmKxPfjvHGK73MlkJ+hay5rPQfuzZxqR7GS8l1+zVuah0vxci4tPZdSsQIVftVSPTrr7/+t//2337pe16Sx3ibA75WO/Dmq17KRPHqodf5Z3yhmMGXD371pZ//AK9ffTuOx+Nf/MVfXJt6/s2/+Td/9Ed/9AJ9+J1n1+/1lguY5wYFdJNkv72TP8RL3X5RbERDCAD/7b/9t//+3//7eCrNuz/+v/7fiAkiBFGQClMRzsyZZbD60IVkyCGq2WK2tH5c66fHp8fn07qs2jUzTzLQ8qVMeZ7yLpcpD92u60TlTRb+5jivhb5B5OQAAwEZMTpqX8Zo6h5zLgBabf/l//F/bxe4/7HWc++DUSqJEKL21mvtbdVewzVCB5YSRxtZoKn32geQWXLKU0YhhzBzM221t9rNLGAwm2LOafzHzL1ra9sLJIkID/LyDak91BZa69pVNRww0puHb37203/yRz/70z/62T+eyh5xNHHL99//v3/x8//tepqvTHtfn5dPvxjs7MOOD/10vP46qBpp62rbEPR02RzoQqh2BRTApuf9Q9jKS/H1C67AdWW/LHnf1CpiI+K68KGPrhZkKfBmOyWRf/Wv/tVu90rh7n+J0Xv/z//5P19N+zfffPMPa9qHS7TWuizrutbW2pA3GB2TIoNZ+B/AdDHz1bQj4r/4F//i3bt3v+XYAAZ5GAwyBLqw2PzPGxHxX//rf72a9jdv3vzZn/3ZpjT4d/q865+bXz4fr3fXf5gr8Nd//df/6T/9p5cN2gMDmJAvtpyJLjINdGvgL7AeujX1Y3MgfCGuGeMFRPEDTOLYAgaO6NoD8/mAF5TRdbwI3SAEgrpfTft+v//zP//zq7GBl++9Me1xuaI3wKrLlv7yjXBblY64gEVftqchs3TZs/BSC/6sOI0/vJ8vjsDNrXz//v1//I//8Wra//E//sf/8l/+y1uT+TuOH1rcz/5+CYFuLuYPr/rtnb3qEr2aABf37jPT/stf/vJq2jnlr/70/8zCDJGJJpJdkl2SOckkSYgSEQEyQgAYwFn7c++Py/r+6WS//m5JH/W0YLddkvtperufH/bzvJ8P++lhmncllU16AIhYiJl4TMzrIonLfx5hCDaUIgA8wAP82m0xpgluk8ADlvPpv/4//yNcFHtW1ae1DtWPnJIgaat9Xdp6bm1xr+GKYIguoz3OUasu52rmSJTmXPYTCll4V621red1PVczDQjiwRSep7nM05RE1lqXc12WRbumnHJJKSditov6Q+t9UI1p726IkX/yje323/70p2W3//bu8I4pp1RyLrWefgE/YtrxKn4V0ZqqWQQEIA7BrC2OZ2IeyjNE24PXBY/ISJfZMj7yNdjyFsKDiDeP3LzrB/N7TCm/nZ3j7USAAf77LYnfcfwd/Oj/Hx9DKuP9+w+//u67p6en8/kskuZ5enh4eHh4uDsc9of9VErOCS/wvbgocP9e4/fFSHmEmh+Pp6enZ2bazdNunqZput24f/NhfPFlv+97P3vxZ49fd8Mf+7QvnfVm6bbw8zJ9N+2VGH9evnfLi/EQhbnRLH5ZRF/Iev2+o5TCSEOFZcvLDQ6Ky87++d9fMu23Rv269cNN1P75gY6g+WrTrwH7lwz8Z9bI3IdpX7U17T88o1c3fUCxt8cvVvflBR5gsPFz+AU57nGTRYALIHlsRjQ2PSKirZX95vM+82J+sHHBb8eO3V49+I1z9XaCbe7UxdCO01AdTMbb6L0PTYExxoN2w1J8Zfq6UhrnnAcN8Gc/pJReNu0vHScCCEImKki7JPuUZpFJKF/0NAmBIPDSZjMxBXCtmCEy0ZzLPM0l5a/289t5ejuVfcm5SE5cmAdT8YurhmjhEYOYjIfBuWaCro0UAXFpXYSIKwN2XFLAMJ6qZq+gqwbabK211goBjJhZElFKs0g2bxHGFIkxM6ODNmvSCVd3x0SUGROZu1pvzVtTU0cIYSZGEiYmRHZFbQHmurpV82be9UJXaynnlBMID7LkoQoPnBwQjE319Pz04cOvf/2rny+nJaXdfne3399pf7U0Xou6Ig7SVvNora+1djV3uGh2Mg4ho6uE5zWld/mZaJDSDAjztv4vq+syH15M+rZr4WWqXDMAr5yA8boL/2X45lYPZwMCAOhHIYm//xir5erejm9/6Vy9TuX/Fez9dS8Yy/h4fP7+++//+m/+9q//P3/94ePH0+mUUtrvd2/fvnv37t27d2/fvHlzf3fY73elTIOr5FVU9Pc+6xu7+CqyGmzJT6fTL379fcny5v6OiEopP7TTt3b3Mxt8axduj3acwheP/IsuyG+In24//Dd81CUs30RVw8y0a+/W+2XztbHLug3jMtrKNzvKzCycU0opp5RSksHcfl0atyf3xYP5zSPnnFhSSoM/h0Ze7nXITvT5Iy92/caW3z64ffqXTPu4QC9pi9d36mrFf2johwLd4DcBwE/HZ/jBeGUXPz+IuPwyvszCe3g362Z6sXGb8PhtNgHxpVax3Q9mJiESpEFSMWw8XWORG5fic5v3u4zfMEu/OF65PhdSo1rruq6ttdbauq7j1/FDrfVq4DcZp2tCBBERh+xeKWWapvky9vv93d3d/f39w8PDmC0/vOaXRyAz70T2LIeUDjlNwomQr1vnYO6D0boWhRCRV+azyKEUu4M87+72+2/v56/m6SGlWZgYCC8cpY7u1/g7DMLcMZB9yFJvpGdxsd/XIP4SnYdv3bqxsfwhRIQFVLXbdIv23pZ1OZ+W9axNCfAw7wdtKLMgcqALQxYuIuhA0MK6ZQkMLoSJgKCbBYgpKHowcCIi5MTIjINQHlkwC7ITGYGCuRv0rfEukDhnQhpUsACAQExAQeHkXU/H508fvv/17ufreZ2mO1cjpN7b7R35QdSO6AERrqq19dbVA1iEWCjoJRsVAe5EsUXaaFtkj3r1Qjf7C7fZrleZrkv2/WWXHxE/QFz4Z18MPsDoD42AYOax7eWch/OKP9I+8ncb16VyVen4bKcj+l8Dfnhdga210+n085///H/8j//X3/zN3/z8F788nU6995TSNJUPHz78/Oe7oTrz9s3DV1999dOf/vSbb77Z7XY55x9+2u/+vT/++CWeujB/mftS2+PxWJLkJPv97seM6A+/4rJ1xHWbAwC+jN/ryG8/8Ivv+h1ObZvSHmGmvS7L+XR8ejyfjnVZeq3atba2rrWrmQ+6060zavyNhEPyfJqmu8vYH/Yll5TTFewC8FsoAn5sXDICWyS65QheG/Ufs+5XK44/GJcLcfNNtxXsa3/Pj+SNEfHFn77cTR/k1A6BGF9qEP3CjdjuwPCYxkd5uJt11aa6qq69V+19UPmN74cX3yOu1u6SRcGLVm1iySKJOREL0VDKYSSCl+TKNQf/e9+Y3+p9Xl8w5nlrrdZ6Pp9Pp9PxeHx+fn5+fj4ej0O8dtjyWmtr7RrHX8359RJ95gpfY/eUUinl/v7+66+//tM//dN/+k//6Zs3b0opRPTFhYmIu5z2pTyktBeeZXN/YNy9z+4/BAImwH3OfndASffdMKec06GkKQ0nAhEulMMX8sPY+H7BInqEm4V7YimbeBgADPt1sTRjYTlcnbYXlnwA8+jqtfVbP7Sej88fv+vaTJt3i4DmirWyJGSOoSiAQQhJCANdXZv1pszICYvkabdDYndc1vV8OmsbmrURBMSMIowinA7zfsrFWlvOxw8fvjs+fYxogBoe3tyShaO6WR+sTkPAGcNBVet6Ph2fnh4/opNplDTN8/6WkhW+RDRLI0Vo7qpDmhJpOBmchtTrRZ0br/mmawARl1pG3Nj1V0t5m64vpcYte7IdwSVXclM/u9k2IsIBPaU0RRBR2ti+AvEfzNZGhGqvtfXezGywBG5/Rt7qsjX+vgbjf8oYW+T5fP7uu+/++q//+n/8j//xi1/88sPHj4OddKTsHh+fxt6akrx5uP/mm2/+2T/7Z2b2k5/8RERebdz/Ox4nqOq61jCvrZvabzLsr9444CobtWptrdeKRNM2ylBd++L4u5n83+Vlo3qkrR6fnj5++P79r3/19Pixr2tvVVuvaz2ta+9mgQ7kI3WxKRQ7IBBSymmepvv7hzdv34ykyv39/WbgUyZm4NfVrt95XA32lgm42O4vWvQfZulvL8WXrftlxEa/enPpLhH09oLLGHZ9RK5X2zMqq0OKMi6CFr/TGcZI8Pm2+5iattbW1pbe165Lb1V7v63uwa2zufG6XE5pBLUkzClJljQUwJNIYhFmIRFixi1tzIObD+B380R+h1OJSx5xjG2e13o6nZ6fn5+enp6enj59+vTx48enp6fn5+cRqevNuPWZxrhe6mt+foTyW8bikp+/v7//5ptvlmWZpomI3r1792NLgBDnnHc575LMTJngEo3hNg8AYyTOAwCAIhhgJ4l2zHnau3cCJMwjYx1D2PMizQ2IAYMRfGRx1L37EFG0nAKYBYEvEoqbst6rI716MRGDVNDBzFvX9lqUq7elnh4BA8HR1dVrbwpnQEESSolklKPiIn9Iw3UQ4ix5P+3u796mPAXwstRTOWnrZroREzETCwQx8n7azWWiXehuzVie0qx67lqrVoegyOAEDhQhGMA2/JQgFE7u3lpdlmOWiSn33kyHluPL+II8DG2pQUIiIUaWMu1ymUgS88YMT1uMDpfg/PLDuIovT2zYkqsvfjtZb5xHgI3/efOxb2TYr/6Ae3hYN1MkG4DD37Ct/N2Gu5tpbW1tq5lGBAWNvw0dEBCH5HMqMPIzf7fvvfFl/nce7t5a+/Dhw1/91V/91V/91a9+9avj6YiIu91unmcicvfn5+fHx6dWVzP7RZK//du/fXp6Oh6Pf/Znf1ZKGTJi8Hetu//4wJvr8FJhGXXcm4LpZVzS3Le0KuHRWlvW9Xg8PR9Pp9N5WZfeesrpzcObt2/fvCWeJuaXKsDft6Dwm3MJt6801eV8fv/9d7/827/5xd/+9fn5MTEThKm11mut3cKBHCiAPDZq8kvVyQHwkfDDh/flF9PDw8O7t2+//clPvv3227fv3t7dPZRSeABN/i5R+5dCc9z+3Fr9z4z67Vr74c/wA9dnu1YvbvqWXLm+7NZuXY3NNXx3AAoPINxq37+baxUR4eHqboPqtrd1YJlaPUdoRHczDCMEulAmXouB2wfAZbJs0Z6Hm0UPr6qjcDw8/MQiJMI5SyopZUmZMBEJwJX55O815cbVGPp75/P5eDw+PT093ozT6bQsy/l8XgY4tta+JSTs1qJvWYhL0iUiRtw/svS3r79+LxF9+vTpw4cPADDyo7vdbrfbfTFniYg5pSzMhIjhEBvmakObwCX2i4vhBQzgUZ4nUwAPcw9HMuBA1MBVrbsZBESgb3fGzLupuumApBMGE6omAGGgi4jky27xasFu7iEGuKl21d776xI1ojMZMQFwNfXQ2lbvbgqIkuddmeacEzG5ORCykCQRTofd7s394eHh4f7+gSVbYOJKkHzelOIggoQRaTmvy7IsZ+3r+W437+f73R/v9ds/XurzaTk+nZ6X/v/l7c/a7Mhua1EUwGyiWysbksWiLFu+lq/P8/72//8N5/0c771lydWRzGblaiJiNgDuAyIik02pqmT5xqePykpmLkY3J4CBMQYSECmCEw6hRp9FBNEt1Td654L3ARQWn2Yb1vvpjvl5aHdEAIq0ANFI3oXYtF3Tts4F2og38Gk/nRa24padwRbdXzBPATbI65O4rks/BdZltKTuIus4dPuOlAoqWp9Pl5Yy4u8SclSVueaSS0mlJlUBNOtDsz8EZlYFBAglS9sE7xEQ8JOdD16kNT/373zZiPzvi/GllNPp9PHjx7/8519+/PHH4/Eoql3X7/f7/X5Xa7UJXeM4Xs7naR651qaJtVZy7s2bb16/fg2Abfvc9v7vqOARgKzYeXbQ3f5qDQOgz+iqKAvXWnPKl8vlfL48PZ2O58s4TSllrrVtW2Yg55u2RUJP1uHaRqv9tx+qatvx09Px/v7u44cPaTzvhz56L8yllFpyZRVwFtpZpTIbVmmm6yJiHVFAfHx4eLi/H8exlKwqznkbyI3r+/mbzm2L6Ev8XnZ93GL8htW/jAfbf8KnxTp82r/4FI9f/sG1kAJ48f5YjN8Cz8u4Dtt2IaAoAKQEL+G7v3KIcK2l5CnnmVABuOa55CnNYy3J5goQyJJL4vMpGUYAiCsVcCksdBlWpapVBUWWGZWEDtFZve6a6GMbmzZwG3zjKCJtRTzCr8MUt1uxXogws/XLrUB/fHx8fHy8v79/eHh4fHx8enqyGt0i9Fagb/W3rp2Fz3Ca7cNftuet//hZaHfOHQ4H733XddfX17/73e/C2gb9/KUCCN7ZKpOVkL44wy5f2u6P5liNNooK0CMEwqiaRapIUZlJHAKKjjlPtWYWBA1otxtFpdaamTMzErrgtTJD9syeXPDeFGv46ev36Zla5i3LlJpPN2PhWksK2DjynrySALKqEACCOAGnENB5561x5rwPIQQf+65vmz74iEAqKmJsWUAM3rdENuCSEIEzzsClZgbpGgRqhrbxnuY8dfMYx8uUZ1ZeVd4iXFUVwLjkUJm5VOeChca1lP78+FrVrkCICAQ21YT8InhDJamWWi33yhJ9Z6mLR+8JvSIB2uCP57W4QF7ri/NctquqgFjVvv7o9l7Ki/pehIU9FUQU79F772iR4VlA+KWF81cPBVVh4VzSnKbMSZDRwTKkmdB6J4VrzplLcgApRUekokTW+I8xRhe8IxsC+3Pno18L7f+NwWaa5w8fPvzwww8//vDj0/EIiLthuLm53e93TdMeDo/zPKU0M9fKJeec5jml+fFweHh4uL+/u79/sCwueO9/0UjybzyWe4JbDvXcr3y+N7b7M0spdU7z5TI+PR0Oh6enw+F8HnNhVrD1g+QEaM71fJma47nWGjy2TTRe3n9HavLlBamCiNTKVhSJAqAjH1zwiggsisgqlYsCKZKhrKqCuDJMiTwRqzLzOF5KKUikAM55HxtyzoXokcj98ul8cTxv92tavgR1wueI/isP+CK0PxdNuFXtqPjJFvoyhLz8GhanTxARQhQj54us/b1ffHZaap7H89Ph/nQ6OIIY0NkUFOIYkZAQrMgRXWpJtPANAmKcLOfWWW6gAOpIBBYtkp27VUrKCiyCKlg1lTrn3ITQBN/60Abf+BC9j0TBLus3PSED287n88ePH+/u7j5+/GgR/XA4mLbFGurGmPuSAP+yWF+7iMuT/SzkM/M0TZfLxTB8+76h8XbUWu/u7v70pz/9wz/8wx//+EfjzH/1pSIkQKqgCCigbqv8lmgKYEi4ooISgCB4UFINCB6JkIRlzHkWTZhIdS51zmVKlRCv2mZoY3SeHEnUqWROSRAEibnOuYACEXZt03Vd9N67bbT0s+WvRXlTUDMzKMQQoG1f7gnTZT48nLpeum4Irmm6DhshAE/BUQAM5LxzLjaxbVsfwvbZqni+TNOcyD0pAItOc5nm4lzbxCGGJvjoyCFBLow+ND4iYUF6mvJcJIYQYtvvhnb/RlBFTQePBv6IcC5lTmma58vlfLmciRwspvmoX7Mx/qLXbnQVQBuIBOiQPCIh2HxtVqkovNAdCYUIicB7lEjaQASiiOSASDcu3fb5L7pZsHU/ZK3a159eq/qXyL2KMFcAqCreO3xJBIK/w36tluznnHJJogyo6NaKhhCQRDVzndKc55GkluxUJM0JAEJo2q7tu87G/jjvHQXLkjcJgW1MW5z6WuH+dz4Mxzs+HX/86acff3r/8PBYcuna9ubm5s2b17Zox3E8HA6n02meJ4PyAMHSUEdumubD4eC8V9W2aWKM3vsXiqy/ZwVvTVj7ers1G44jorzAkmmcpuPpdDg8PTzcHx4Pp/Mpp0IuhNC0bReCs5ZbKXWa5tPpknNypLuhv74mXGrd/z8U7sqiVaSKsgKQoxBciBQ8AZEoMZMiKL98w61bZuWVdZoNoC+1Vp4PT08+hN3+ari67vqh7QdyDtX91qvZAHn3HMW3733eaP8yln+CzKO91p+G9gXKXod46Iq4bMt/XQCfZ7iILyO9AKDIch8+30t+5qar1pzG89Pj/fuH+w/BY9eG3a7rusY5JAcEiBaaBWw0HBGqqnCtudSSHGJsovdGhicEMqheEQVwmT2HoDbaV2EhQrKwVubCJRU/B9/W0MbYSmx9aJyLSM4A1b9+/tubWWs9nU4//fTT//7f//u777778OHDw8ODwe/jOFo433Ru9vVLSdtGfbdHacvW7k/O2cbvhhC6rrMcIqU0TRMze+9DCABAtMgjnXMppbu7uw8fPnz8+PHm5ubq6uormDwCEApiVQBVByi25dltWmOPKrCqADhUBSBQv+T0WnId55RLRubiXbDfVqjMqJC88+ycOE9oj40csQozV5ZaWFXJMBnnGDSCd+isZbV2A5Y3hEWqsAAgofce5NOBI0oEPlDbhqGNTXABRAgguuDIl6oiNu4XPQVPoTKzWEpVmdmWOyACUq2aszpXk+cQYnCRiMghekfBuxCcc7XUUnOuJZa6I9/H2DQBHbFUC+3BUfReQXMpl/GC+FRymXBCoMUqAgHQaHafXMeXunYU1RVxI0BnslaUwnmEOqMwgbpV2woKUrFWqhR90wGobxz5iOSB3DNjdHn6+BKQV3ieebQl7uv91+2n7P+Q1SaR2q7j1o3p7wHIq6iUWlKaS80iDAjk1jIGCRAFtHCd5nkcRy4TSuGs83S5v7+bU3LOtW272+2sF9W2Xdt0sWma2IYQg2+MUruOe8EXTd+/f4zZtstc6ul0+vDx43ffff/xw8c55aZpXr1+/ebNm5vbm6fD4ceffvz+++9++PH78/k8T3PKWURurm/evv32j//yr//0hz80bXu5jID30zR1polprfne2Gax3fbPsMTffsovnz8uxFZVXlRKtRa7+dPh6enx8GTlyziNpRRHzofQNl3TtME/0wJEJOV8vlzOo9aSbq+vfIhErjWJ7Nd0Sn+3A8GKVFEUQCVPoQEkDC0GT1QdUkCnroAzjhN4JAWQdfi6cw6IFIAcOiQlEoXKepnS8TIeT5frm3lXavDR/faqfVk+SGbpBfhpm/3TuP4ZLL99/XmzHZZnh0ZqtriOa1937ac8B3jLWdb/NP4PEemLTvzzv7BU7b/8gqlKTtPl9Hh8eH/4+EMTney6xr/qGyIXyGKyiLKxpNUUvaIspUznp8vhXrl2Xdu2fdMOPjTkIjqPRErW71UE0VUSC86hIoHh+yBSRVhqrjRVF2uIoelj04VmCKElCrj04L965p/kOTnnjx8//q//9b/+7//7//7LX/5iVfWmUDcU/aWkzeK3HZbTfwmuiEgp5Xw+55yJyPw6ce2AWLA3Co6tcUvlt1r/eDx++PDh22+//fbbb+EFT2K9AKOvg2HHorC6lW2chWXjY1BRcKCK4MG64VprPZ3O96dLYXaE2jVXXbvvu0HBh5BKTii5TCdODpEUyBE6YpZpzqJKQOQceldVz3MqLE2QxofovOGBNhJORKpw4lJZ0BGZKr5+8lJdX938/t0f9lfXw7BHBS6cLlNOaZaiknIupVYFdcFfLhf0rtRahVlYQATYMI/YtE3bOx9aR8pYOZUygyo6dJ663dC2A1qPQVmUubBI9TORSklOQVNOzBUIQvBD34cQgFAql5RLLsyC+MwA/TKuw1cAebTSXbZF7IjIhlBzAk7RYVg4oQ4IFJQFimjVIuyYixMmRPIOKXyNDrU1z+Fl5H7RF4FVhLLaSSyXICS0+NzSS48NE8z97Xu0iJgoNJfEtarq6sNDSAToVKmy5FLmNKc8kxSUnEo6PT3c3/10Hs+qGmLou34Yht0w9H3fdn3X9m3XtU0fYxdjF33rfPTOkxn8GRqANmR9vfK/02H3bZqmu7v7H3/86aef3j8dj4g4DMOr16+H3aDCj4fHv/zlL99///3Hjx/SPFdm53zf9+/evfuXf/mXf/ynf3r16jUinS8XXQGAXEpOi/LVJLBbgP+vpVYIAJuumUVyrdOcT5exiiSrUHIep+l8vhyenh4fDw8PD8fTSZiRsGu7QI6c20QM9jYxcyl1TlmkppzI+eF08T6E4JHcZwDBf0cdbxFLkZQcuMAsSbTmWkoupTJLVWBEBthGoG//E0Qwg0zbEskjgACWKnMqU0rJPkP1bxhxaCSaFZPajCY/Kcq/Wrg/f01byf7Jo0eApTWm+NzLtkj+XKava5+Wu/QJSKe6pPBryHn+6F9Zt0vlkrjMUiZBD9WhVAK1cxVeBBUqiojogExSxTVPl+PDXZou3ruuG4bdTdMOzjdWpbg2+C56T45Aaq3MSJ58QApEjgBUgUhXi51SOAsHkSycTZwfYu8AkP7aE9vK7vP5/OOPP/6f//N//v3f//27777bdGtWqVuRbYC8lexWgltcgU/JEKpqNf3WuZ/nmYhEpG1b86IxBD6EMAzDMAx93zdNY84HiGj/+jiOd3d3x+OxlGLf/+S+A7AqiqAKWx9d4TNcxh4fqwqo2dIFIrdGd65cSi2qAq6IVgUFCI7a6FnrVOZSmRAjUiTXuaYJAQELCYsYTQzJZeaSy5xKDH5o2q5pnCMkFAAWyVwXfAPAqlMB5U+J5W3T7nb7ru28dyWVlNI0zXlKhuOXnFlYQSl4UUBHmSurMAigqNPFsdtRiE3wDYFn1lpq5cLMSIKOgApSASAARayklbkyQwGYSyFAEcklVa5KEKLXmpu29SGUnGrOXIoKq3pbVovn7Urz2o5Pu6eISA6QFzie0DkKDp0KAROIj7Qfuq5vQ4wW2kWgiqTCY+KiJMoMrKhI5IM9frTuqW7LemUvyoJFPufyz6140A1bQrLQrsJ1LdtpU6B9lun/9kNrrTnPKc+lJGPdExKClaQOkEQh1zrnlHMSKdEjVp3ymOZTLZeczjlnADg5attuGPq2tcZu0zRd0/RtO3Ttrm13XTu0bdfEromdKWfQ6KRre/VvvYRPDjQHmFKPx+P333///fff39/fl1K6rru5vX316lUp6af3P/3pT//nf/2vf7+/uztfLrVWInr9+vXbt2//+K//+n/92/+12+2891YWNE0DgMZFT/NMhCGEtm23BH/L7v/mcxZR5oU7lku9jDPS0zinlNI4TdZH3GDDcZwqsyPyq+exsORUUAmAMDzXKGYOA4TkQ2U5HE8hhq5rV9UK/DdV7brGdUEUJEEqgFPldLrUkqdxtBuOWyVRWUSUzb8GiJQU0ep+ERFFcuQ9kgfnBKCyVFZ+QXr6TQetXrHPNtGfhvOX9PivfrEkcwuV9hNTI12RF1j7ZBaUdcPmwdq9aE4kVlShqqgAgIXzl1YqW3yCX6EmQzVLgDj0XdkNTXRD3zYxrJ07UgVm4VJBdYlPVluK1lLH8fJ4/2Gephib6+vXbTs4H0WRAfevbl+9+3a3730MOefL+YzOh6aLbR/azk4NiYBgm3OhUkoSQ54ic6vYNOTx6zCLpTJGbTNJy3ffffeXv/zFoqmuhPaX/fVN1QYA3vutjH75mGwPtZ+0RWSpwIbbw+oAYVhR0zRmVrN13+xzDCp4fHw8nU45Z1PNfLrramUBYBBB1dVETF/iEZbuGWDiAIUweCQiIPLB912zZ8mKigiEqfLxfImLNL6WlApL8M417a5td23Xt21ttItlzGUqxazl5lTHaVZVR3S9G/Y7advovWfhypxKqczoPHkHiIbnp1peWtawcq4pn5I8yTRO02XmxMAQ0Xsgy0jJe+dD9AG9I+8ZRAnACXj1MTRN13W7obvyFAlMDAmqRTQLVAUDDAqJgjgq4itjKVBUKxdwy5pXw/tQ1SVkqcmHJpdqIrd1ssKqKkME+jyAfBbaCYxHt6bmzqziQVCZCNrgh6Eb9n2IkbxDRFEtLC5VcUWLZsYqnEoGFylET94RAeILJ6rn5f/yG2tS/5zB0xrb1Za9Kq9buRVpGzdEt67ebzzs80vNKadaskjdfHMBEdABEgCxSq41l2LND+9AKuf5kueLSpGapstpnKacc4xxGAYzSW3btmm6prXQPnTtrmuHruv7dte2fWy62DQ+Rh8i+YDOO3K/fU7v1w+u9XK5PDw8/PDDDx8/fpymyXt/fX099AMAPD09/fnPf/nLX/7y008/Xc7nWqtzLrbt7e3t73//+3/8/T9+++23OWezwgAAW+rBeyJUlVoNSlm2oQ27e+lu9utPVUQqyzTNp9P5chlTKqLKoimXU4jzPE/TZE6jqsKVU8pcGRWdMx6lSW201oqw0IWM7qcKLFKZHXrnYxU4ni9NE4a+I+xj9Gj55n9b3315n3FpQM6llpznabpczlyrt43Tke3atlZthyHnaM16KzMzk/MhBEBqZVnXi0D0bzqxhftKq5fkAlB9HZD/ucIdVqnYZ6EdV7ablQe4jNAwUHQJPghqhdqWyQuoSZkRn/8U+WJB/HJsB+9907Z9P5TdPnhomugXegWu3XGtLCBCzm/cgMUnKEYXAk9TqXVOs6giUGHJLBBo//pWpQPEktPpeBCg0LRXNxBig/ScDImg4NbvZubMCQWIKDoK9LWqXVUNVzcXuXme7+7ufvrpp/fv3z88PByPRwCwYn0L6hae7deJyBbjZ0F9RT2F18N689tf4QvGXErJwr8dW+S2+l5EpmkyXk5Kqe9792krSBUKVwFQ5uWhv8RolqeHuBjDKSEyoQNAbzWka0JoQ0mppMKMwAXEowY/tE2Lcag1VfNDI0DHCnOptldYBVa5plLnOc0piyoRkid1mFVC8KvzBQvAwh3TVSZQP7GsYeFcU6mllJzmXEs1+mt0MaJHRUJETz7GZujQu6IiIEqgXimojyHGrm12XdM78KZ+d4SKHsArGvVShAEVSBUBFT2TCrJXywyZ1MjoBI6MDAIsSgIA3nmjkK951c+m918y5AlQXrBmiXCB05zDEFzThraNLgZ0DhBEAVgrUlSqyDVDqTXxWNWBC84FMs8T2EL4ykrFlV/zHNcXKJlt7gupqBjfH1WMi76d1JZmLi8x4N+wzVnzKeeccxKpuOSLK9IIqEoCUEVyLYWr8xQoOiyVS5rOeR4JFETG8+Xj3d3h8Oic2+/3V/v9br8bhqHr+rbtmqaLsYmhibFpYts2Xdf2bTd0/a4brtphH7shNn0IDf+tm/VnR87l8fHh/fv3P/300+FwUNW+71+9ehWCf3h4+POf//L//r///v33P1wul8pMzg273evXr/7xH//xn//5n99886ZpmsPhYDmBJfJN08QQ2rb1vkVcUiLbg8yfMsbY9/0wDIbj/ZqTtHw+lzrN6XB4urt/uHt4vIxjrUKXydIFezoLD4g8BVdzQUBHDtwLJfKSItRSKznvvId1E8fCSC42nmu9XKbgXde2jjD4nQ13+O8j1C0bqyqrskgq9TLN42UcL5Nw9Y688yE4RFATeW76Y2FEEqMa5FJqRXIhhL1q2zb2w6tsFH7rFeCqa8G1ubXo3twnhfv2xZfFuv03fBrgYfMbWJ1lUNVyZDMvkTWyo6neAGCl2QgYMPfJYZyD5Zw/nUry164O0YfYdrtu2Od5JKjOOSS3mG/YKYiaCZ1XEAUCS3Zc7Ibbb76NbbsfzyASfQBVLkUTswqRGmKIADmn0/Ep5YIukPPDsKOAuHTZXs5WUQUUxSq1llTy5H0TYvsZTG3x3xzlpmkypP3u7s4o8SZe36D4z9jv24cw8zzPFuDtX7dYvt3Mlzd2+/UtD7ZSvpTinLMvrOlm+af90+M4nk6njUv/GU9eQUupLIvyABcipS6Z2/p0ENFGehFAITTHgo48OgqOSOV8fDpcpujd0Ebsu75ZGO99P1zmdJkSg55rPZXClVc3Oue9z3M6XaZUWRQsIs611nGMtXjnTEdtAYRVEch0Xgr6mTtWrZxyrrWI1NiEoe/bpm9D17jGq4NqLw366JuhI++y1qospOAUPTjvnWuij548qXktKYgSAdpkG4rCoqQogEIOPRBVKOKEjLEtimAhGMGTOieOwHnw0YkieVHNpSwC/mUf+FzCB59X7SbBxHUKAjlAEkBhEWVyxkv0LgRawWRQRVRiJa9QQEBy5VRmxRCaPkYBXMuDpfG+vpRLXrnW7ggKix0X8oJI0PIuWmgXqh6X6L7g8UsBAQC/ObSriJRiAFcqtSAKPVvRW4JPolCEUynMjAgxBqzKaUxpntOcc6q1pHl6Ojw+3H98eHgAwPPpeBz6YRi6vrd6t23bGJsYYtu0bdMN/TD0Q9fvut31vuZBpQWs4KKaReJ/Kbgv28R4+fjxzlL+eZ4NS2jbdp6nH3/68bvvvvv48cM4jkalaZrm1avbt2+//d3vfvfu3bsmRqPN39/f11q3tT3P0/XV9TD0XbfYy9tOYXuKmVkys/XqvlJvLae3foFg8ODlMh5P58fHp4fHw+U8cVXjm4poLZWFjc2LCI7iUqYDETrFlxaEtmOvU1cAFgaDoqwVmZCI6jSnx8NTDL5r2/jFxIu/67EI1M0WzTLXlTPPXJmZq2NmQlQwB+Xn7ZpWeFZM04SIXEMTQy0D1yJLt1jgb3pb1si9WNYD4ZYz48uU/ou+O6zZPyIqKCtLUevimgh/geQWChVufD1zdFw6AEt8XGnzqrTt/WsUt0/aovtzMPvFPAbR+RDbvun37TRKnREZgIzZtWQdSEgOVDeNrv1a7PodvW76fp9nZSZUqYWzObXy/nrftO1Sqooq1/F8Kiz7/VW+fUVIFNxzOwKRiOwfVDUMutaSa0nC9cXlAgCUUqZpMlM5eWEPZ3mtdcc/k7Rt92r70/725ZP6DLH/7Bdt5RrMrmszvpQyjuPWlTccznh51pi3df1i130+VLXUukzxUmNMWyG3jmVZT9V+hADcyvBwRAExeNcG7wkQhVwgHwzOtALWO+fIiepUcp4mFhYRD9AQNV3XxSCiOVdyTgDBO3VUladaklRCBAVCjMEH5x0p4cLuWwCrFxfiyEffePIIGmNsm67t+iZ2HoMT0iLKoqDOkY8BHSKSB2EUIMN5vUNP4jbsyhxigVmZVQg8eN00gYtiUJ1HVVokVbBQQ4goeHGOkYQckAdkEt3E1RuHeU2iPnkin/fa1248KjklFXSqWKsQMwRtBQUJyK1IPaEoMAugKIpCFU2Fx1nAlbbnuupLLB4jLTfR3skl3/ik167CAiQovLUTUNQ0QGZz++XWs+zmv2WXM+pcSvM0jaVmUbFbqfbSIgKQKBWBMZVUkqgETx6JNU9pvlwuOeec6zRPx9Px/uHu8fHhcjmpyjyfD0/e+xBijE0z9G0/dF3Ttk23G/ZX+6ub6yvl68qpaqUQMXbqW8FYZIkFv+EyvnZdc0rHp+P79+8/fPhwOp1UdbfbdV2nqo+Pj3/603/89NOPxqMZhsGq7Tdv3hjx9dXtbc7l7u7+48ePj4+PiOi9n1N6eHzou26/2796dfvmzes3b95cXV0Nw4CIG5BogKHpZz5zM4b1+W5/iuic8tPT8fFweHg4nM7jOM65svMBkIQZVGot1kUS4VLUO+dX13PEupqB2mevqJC9UbDxiKwZhKDgyMXYMNfD4dBEv9sNRLRF97/vsZSHIipVpYIKmvlT8CEE570wqwrXoqygAsrP92fhjeCSrdhEB1EG5Zq5Zi6ZS+JapFYV/ht6CrhaTJGV3bRSU79GoPuyfIclM+NcSsrzPJvnSTLGgIiAmgEReSLvfQyhiU3bdU3bxBh9CGgE+yUjE/ga8mD/3BaWlsj0SxeqikjBx65pd20/5gm0ziIgqqS6kImc1wiqSs4DkgIqErkQms4F3/YDKKMKKYMUqUWERcF3+2Y3eO9V2DlqYpRazsfz5fg0X87eReejiirKlp3YC2o7NagIFy5FatVPV8c0TU9PT4+Pj5fLxZZVKeXp6WkzbP9qXP/ateuGGWyUhS20f3ZYIL9cLpY4bklAztlgeUPsDK63pOHm5sYUQJs95af/OhSuqCpsl46KoOtI1bUjvFDm7Sfcysb06NA77/yua2/2AwQf265t2i5E9G7K1caZjylNeT6N4/EyAkIb475t+jZeDd1+GPqu6/tuKpyZ2RETnqZpnicuBVQdkl/sqpCAnLdZZkRIzvQX69E13c3wyprRIYQQm9g2PkRUVAbxrLzs1FwrCGIgt4xLBWBEJgAEVEEmZ9A2gGqtzJwZQBxF74JzuCT/VQSrAINpYVxwpIi6bBot+aBIAKRIwqnWuVQ2SzGwGlTEoi/+tdAOsDafTAVPik4Ai6CyEmpmYCC1cQjmzaKioCJYquaqpWph03qupb8hC+TQOUfPAyJh7awD6KLPA1BRcYLMLG5BzURsgauIswrS/Nw3cZpt419y///qGqi1ppxynnNJNleKFjrbZo+LhTVVseG2zpF36EBYdZ7naZ4La1U1p8LKVUHIYa2Sc5I8KyA5F4K/jLG7xOibJrbX+5uSUuPpauhVK9g8KyJBrKpQWeS5L/U3HLYyn56ePt7dffjw4XA4WC+86zoAeHx8fP/+/fv3Px2fjiJi/Pa+73e73TfffPPu3bubm5sQwuPj04cPH+7v7w+Hg7U8LEsxZu00j5fL+Xw+m7y1bVvbSqyyn6bJUPSvhvbtiZda55SOx9PD4+Ph6Xg6nac5F2ZEitEtFbiyyOKytNCDmZmIiGKMZou5vK8IAGDvhA/WwiYiUkB70KJcayIiIuCap5qbJg7DQIjhC6Lv3+9QUFEWFTZdrI0TMVsAJgJR+xFQWYKc7Q5bTffM4bee5WpwxZW5ci3MppvTX65lPz1wi9hLJU34wtLkq1W73SXDeOeU5nk+T+N5vFzGyzSO0zzP85xTKqUIi6oSokNyRNGHtm37rhuGXT/0/TD0w7AbhrZtQwyOnLUSXwLydpIbLAS/rWoHJEc+xrbvhitQLpMgOpsHggREDjwiOQXj4ZOhDOScx6jgAJRAEQSBUapqBVBARN9RaBBUuRJi8CRc5/E8Xo7TeG67XdP1ywP7pD4GAnBkL3VlNo/UT1bH5XJ5eno6Ho8559vb277vjbtqsvKXl//VY/sr/MKpFz6lysOLQt+WtuUN1ty0DQHX7vs2Xcl+OIRwc3Pz7t27V69ebav+sxPJlZHAdIUAuKrhzKJmteUH02wtDXL7TQeo6gOhC6HruqTAgClXLjIpBKkBIXoCwl3TgKqyIGLfdbuu3fVdbFtwHlGcqBMF5lrKXPk8jqdpLLmoahtj2zSOPAA654IPfstZq3v5XkXfDO3OkCwDLZwLhA4AhKyAtQkzqqrkKVLE4FbkDVUQgQicR++AnHX0a+FauYoCIKEGwLjkFRQ8oK9WwCIhuuC8I7SkjJFESZSqaOU853ma55yLiJAj0A35+grc+xV/saXzjQQISh4QBRyrSyJpi9nLfSFUVNAqkIqkIqmqKKHzPsQQGucjOfc8B9bm2S08taUbA/jMjV/gd2ZiFuee6UUAouxKIOfRuoKbbZYRU+XXVu2qlm/lNE85J+aqBECkRLK0LxZ/n1JlzrUwq4J3LjrVKlJrTinlYjIAUUSitm2HoScHOaWUsi7TAKpW1jlXJlR0GErKHt3b169iCE3TNl3fdkPsBhcadB6QrHr72w5b1dM8393d//TTT3d395fLxVZsjDGl9NNPP33//Q+Hx0PO2TnfNLFt22EYbALEu3fvhmFXKx+PT1ayGzzovd/tdt45RGKWh/uHx4eHH374Yb/fv1qPq6ur3W5Xa7VpcoYEfnmGuD7fNM+Ph6fHx8eHx8M0pVoXMSiS8z44R46QebEoUtVlehUzAHhybdtuYh5V3ShgPoa2aWLTWL9AREqpKaVScq01xtD3vQrXkp78UwghBr8bBiJaYdS/42GTEVVEhFWNP4vocZsJas6pgIIAtJFHzd1jyYBWMso6OGuJr6K60aK2DupvSlDweSazjWmmrZ1Onx4vo4I9iMvlcv/wcHd/d/dwf394PF8u4zQZcHwZx3mamAVUaUXjow8mB+37ftjvrq6uX71+9e3bb1+9enVzc911XfAB1ygu8lzybt+xR7nGpF++OCRHEEI7dFylZq15kcfaX9outO6JsCA9gEjo0EzSQFVBAEgdIXgw5b8LQJbbqznIg9SSpvlyni7n/VVaz29hHrw4IXCI5rmkXEWqvdjbcTqdmqY5n88AYJmriGzxFX6eKKWriGD7jhXcsGYDn0EvX2Zva+hyAGAUGXxBrLN1ZD/Wtu0//dM//fGPf/z2228ttH92VqqQKwMBr4RQQ64Wb9nlTuCqjxAEtV6UqoBIldA3EYjIB8R8Pk/jnJWVWKLUXRPeXO9ub3b7/Z5VjucJkGLT+BjIe1E9V8ml2EDFcZ5O43Qcp9OULnPOuQDA1dXOoZcIZFNDY7Ap8qCqn9J+vfMxNKSme3YkDgoI65I4Va6laK0GBAYMne98bIBIFHg1tHHkvHfr9Mda5iwiqIDgUJ2iU3IuxqZpmr4PTaPoGDAXFUXnPCGySs51nOdcWEByKeM0pjSXOuWSYZkDvQnIP0mLlwv5YmmA9coAre3vyQdVlUABZh+8ebQZoAcmGCQAp+ABowYAiBTV98NV2w2xaX1oiNZfsKC+0vRFF2mEndSafwAibLi7kaMBwDEbjW7NErZBisvJfvXt//IQ4VKyNctrzaJscyMXDwo0rzCowqUy10oI5Fz0zjsQFe99bLq2G4TFCKf7q9u3b2vX99M0zvM0TtM8pzknlqpYCUWVzY3eETZNbNo2tl3TDU03xG6ITUchovMOCfVvl8DZzmt28R8/3p1Op1rZcv9a6+Fw+OGHH+7u7uZ5RoQQfNs0fdftdrubm5tvvvnm22+/jSHmtPBvEcA5t3X+AMBGPRZQ5mrNP6Ov21bSNM2mscHVR+grN1+1lHIZp8PhcDg8nc+XnKvFNCCbshFiCN67bXaFfT4CqAgDeOeD923bAoCd2wIbBu9D8MEjUc4p5zxO4zRO8zSVkpmlaeLQ9845JChcq0jXttf7q77rvF9k7n/XCt52saXhRICrwbXz5IQIwDxQWMVGvcH6LuuatSydOg+ooI6cc97cJUXhWfzy20/MgvgCyT9L1b8e1O0/l6B+f//x48cPHz9+vPv44f7u7vFhnudSq6VHl8t5miauZlVpNFRsQrja77u2QyIfQtM21zc33759++7du3fv3r1+/frm+qbr2uDChsC/jO7wouP+Kx+QAi6FWWxD03JuCYUIjKBj8UXWTcfCjK5dDUQHQKtXtwI6MAseQkBvHh26tDbNvjKVPJc0MdeNQPh88rh2W9GIg2LDZUU+B+QPh8P79+9NoWoLbdOUvrxk/FSM8PL7Wwz+zGdiK9BfSoq+jPfwIlGwoG65u9lw2YDXf/u3f/vjH//46tWrL0XtYDQ6FlVgYVnmqupyj3F1JVAEQBE2fjQCFM/Z1ezK5MNU2SHOhUvlaZ7HaUaFgBSc8zE0XdN13dC3pgtLpdZS58oVcC5lSqmUzFy9W/RTnpwHsiiqAMDIVUvlnMvsnCNwPph0wn265/oQ2o6AEQUdLoIscIAIouLUeQrgmQCjDyE2MTbkPCCKaEVWVPIYgm+agASqnLPL0S+PXQAEvXNEgVwkF8kFF6MLEciFIsyA5BQQRapmSblKWTbDhdz3VfLsL4b257guiGZ6HWPTNTEAR2LXNuBs5B4BoWnACZ2jGFzXNoFIAqgjdH0/9Lt901hrDRbm+WrMbFFcREzLrKuvjqrCwrRxzjlVx7S0mQw9tIoDiBCdsWAWKP/XbcoKUrmYbUPOsyrDIs2FlfNLio5FS+Vaiwov9qCBHIKqNE13fXOLAI48IZHzzse+30/TZZou43S5XC6ny/l8PqcyljqrZJACAsHFm5v9m29e7a+vmm5o2qHpdrHpQ2zBRSIfyMnCJv5bDsPDnw6Hjx8/Pjzcz/OEiLYsx3G8v79///794XAQkRiNBtC01j6/vX37zTdvv/kmzWkaJ2ML9kOPRPM8i4jz3hC5vu+x70QW04zT6bQ5Vloib+xcY998/f6Lppwu4+V4PJ1Ol5RKLZVF0LkQoiMKwcUmxhAM1TdhPTMvpEtZmBtt25pRruEK3vsQIzpSkTnP59Pp8fB4d/fxdDzO0wQAwYcYQoyx7bp+6G1C1m4YXt3eENHgur8/LK/LEBFbboTknY9BSgjBOXEeEZlJeOEIIDnvF7lHrZVXUprtyLAQWA21WggEz8MRf+OBayx4Nqr5IqJ/FhvmeX7//r0ND7x7uH96erp7fHg4PjEzOTKLj2UJ0epRZYpqFiP0jtM0pznX0jTN7e3tP/zuH/7wz3/45z/88x/+6Z/evH59tbtyzi3t2DW0v6TRwcq9/VX3Hqx6d96H0DSo7FDNv8xcAnHVDcNC8bGmIC6kYzDik2nvYanaEWkl2Nu+tbRFauaadZWiocEqqy5fNxf1ZWzvSkd4UdbbG/4f//EfJnJDxD/84Q9bV5teuG3+lbhuU1tsLdijfBnIt++/NJN/+ZR1lRra+u267ne/+92//uu//v73v3/79q1h9a9fv/7222/3+/2XjXbYqnanxfySZU2CltcB1xUBdu+scHeEHp0jCs7FkDyRCM8ll5pVqveha5vrXX+z3+2vdm3fknOiFUDGeX44Xi6FE8uY0zjPzMWhfvP61ds3r0NoYmy9uzicUqhVJPigojnlM9jDq9h2FKPHz/XgPkYUgErIROAdkQtEzhwWVRoBFWT1RG1ozOxWVVmVVLwKIXiPXeeHXeuDAwQzP8y55lRyKiUzGbvUeUFXFFCUCH101vcXJVaFCoKogKxSSq41A6hpe0Ud1OeItRTvX6DWn4rfYEm2jSlPit67EKPHQOChSBvYe7doXox8A84FF8D3zgWNSq2Ni2libGN03q+jJ5UWsOrlXDdeHvNi/bhAY8558KalWQK6iKJzi94Gl6p9uS7dyp1fWvAqLDXnNM9jLomFDRRCK6AUcRmviamUORUWJsLgMXrnCQlUnW+abr+/9S40Tdf3u2k8TpfzOF2m8TKO53Eax2ncj5fLeM5l4pqYZ6lZmR35b96+ffX69TDsfWxDO8RuH5re+ajgEZxbfIz+lhijqjnn4/F4//Dw+Ph4Pp9rZZOlMfPT09PDw8PpdMo5OXIhhLZpu7br2na/291c31xfXe92O1X13l1fXwvL6XQ+nc+X8znlbFqyWmsuObwQjgOAmc8i4ul0MvjuJZD45SEqtdRaWESJXAyR0BVm533TLhI7vwka17hFW9tVlYWZyXvnfWvtKEQE1ZRzSvPpfDqeng6Hw+Hx4eHxIacZAbquC7srU+uVWuec+pxZ9eHh8af3HxxRE8NXC5H/4rH4oiqAAhHFEGIIhDhP88urW+KK2vhSMMTSwHwA9N78xdA5q8mWwv2/dGK4uc+tra2XIX77CyJEZOE5pY93H//8l7/85bv/fP/h/TTPRutiZlEhRVMWXV9dvX71imhZ6CrKtUplUK21KiqLmP2Dc67tuq7vQgiW4scQu65zFsjX8zQonhBl6xn/0oUbGohaVQpzVlBDkHCN26JauaaUTN69Xb3RF00urGufUFQFl9QLYPXbQRRDHEVBBURAqnKRWolI0S0Pb2mtkIJlO8ZhMuv6T3ZhG734008/3d/f397evnv37p/+6Z/6vr+9vb29ve26zogsG8Zuv7V9iMlTbSW+7JFvxfr2xeZEa+kCEVnqsOnpmXmr2pum2e12b9++/eMf/7jb7ZxzXddZqfDVlaKglpAWYRUFMZ9+1c1UUUFYhY0VyCICKisVCx2S98478g4JtGuboW1jCE2MsYnqfQY85zrnzLWmlEotIjXP89M0X+Y0lxSc67vGGgk+4OLT7N3lMo9z5lonqaJBQRbqpIKwdLER+SQqKiI45yh4MOdVZ+m0oLBZo4ogs0OiEECl5lKlImrwbr9ruiZ0fTMM3bDrwyIRJxFMucxzuozzNKZahAUACMlhcOqIjVOnIqyVNbOkwnNKuUy1zLXOypVQySOSBwyluqWrDWuM/+KJfF61bwT5jTzrvW8DeXJQSnR5HY6JRARIDr3D0LgGsAPXUxzs6dhchfUltLxNEcRMp0REuIoutCBl5sq6gJCkwWS+QM6vZYogLajlyyGUG0T3a4A6g+Jznuc0lVoAZAHgURFsbgSqUhWYUpnn2XuK0QePgUxlhUg+NuRd6Np+t7tO02W8HC/n0/l0PF9O7Xju5mmXppRm861TKcyJ61xLRoE3r9/e3rxu+x2FNrS7ptv72JOLLA6AbFv5GwB5S5XMK+ru7u749DSZH5Nzzrl5nh8fHw+HwzwnEQ2BYghd27Zd23f9bre7vr7e7XZt26aU27Z9+83b66ubh8fHx8eHp6a5jGNlAcRpTqLSBG9FeYyx67qbm5u3b986506nU4zx+vp6M6752VNlAVXvQ9t2ClCrlFp88E3Xee8tsGwqXqsheCP6inDlLOp9F2M0PktlnubpcHi8u797/9OPdw93T4eny+U0zZN3tN/vh77vu845l0pJOV2miUVj0z4enmL4sWua66srIufd1nRfWoO/9UF8cqxd1634ppX3MM8zIlpoeaFUhmXrBqtsQMRKPSPyWs4UXozn+QV21V85sbWh9ly4E23smef2rFXJKeWnw+G777//X//nf9/d3V0ulxDjvm+T1LlmK6wdIYC+fv36mzdvmrYNZhnEXHI5n04f3r8/HY9t2wJirhmRQoyIOM3zx7u7lBIi3tzcxBgXctaLiyIiI9oQbmSDnzmWXxKACpqF51omlrIojJYyQIU5p3w+naZ5AkDvXdO0IRgCtxBCF9WWGs9hhRRRdZmchQokVoQgOlASUS5cEiKSUxFdYgACgTEKUYFko5t8Kn4zOfvHjx9NxT7PMwAMw/D27dt3795dXV0ZV8ZYpLh2uzbLmo0MawEYX/DgXkL0VtnbG2i5gvf+6urKUL3T6XQ4HOxZbJ9vxjXX19c3Nzdb0gArdP/l/WcWUa28eCrZSy02D0bMFE+qeapWEREVsZRnZWag99S0btc3b/ZX+671RAKQmadaZZydiJaEKg5BVbsYpnmWPHNOINJ07fV+33U9kVdQH9yw70ITAMBmppVaFVvyQBUBkYuUwgqOPlXqskoFdCF41weK3jlHgKi8DlvmyiBVVVBUazlPo0gJ3vVt/+rV7vXN7vpqGHZ9N3QhNi4EQC+KKZcppdPpcjpfxkuep1IFBVAdKqEASC1SKmeeS025zqXMKc/TmPOstYCKcZFcDIA+FWcaeFzbRV/uV1/q2tcvYLOyMJGAQ2fm5zbjxCE6II/gCbyjECiCb1yMZCZ+8NzNQgQEmzuHQOgUg3cgodZSS7FxOSjGJVYLtoKLLBIX2wtE2nIUfLEpLVvhLx4KWmuZ05TSXGpRYWsTrPdm6YzmUufKKZcqHMlF74IzKF6FxdyGHVGIrQ+xiU3btn23G4b91XSZpnGap5SmlOaUEtcsUlWySuZaQGTX7/rhKjY7H/rQDLHdudAAeetvmFf/3xBQrB49nU4PDw+PDw/n87mUDACqOs/z+Xw+HA6Xy2jBPvgVjW9aM4s1nxnvQ9d2V9fXAMgs5q+z21+Nl8v5Ms7zzFxzyjVnKx9tfrN5UNdaz+ezSezatiX6+T3YrJgUnPMhAjnPLL4WcosOTVRs3VeuYCj0CgCoiBJJrSZ8B8xzSuM4nc+Xp+PTw+Pd4+P94fBwOh3HyzinVGsx9l8MsWm6EELlpYXtXOi7QRUPx9P94enm8AQ3sOv7Ja78Xar3DbkVVhUiijHe3NyIiPfudDqbXaiVj0t0X7cYm+UlLABqVQgRee+6ru+63vuA+MwD+xuC+0ZlXZxnnme9mMwXVbXkXLmWUg7H48ePH//z++9/+OnH4+kkIh0BeGIQIIgxNiFyZVWppZScDbFBMD82Zq6lFAUYdrvYtnNOIuq8V4Bcio0cvL66Op/P+92ubVtE5FWvj1ZBL0j8z1btCpuphQkMCte5lKmUSbTAcnGWDAEz55JtjKmCOudyzsEH6ytxE70PiwTHEQEtDhuq60AABOfJefJhxcAdgRmgJwCDVBfJ19I3YZPOg4m5v3xc5/PZrGC2fpaVyL/73e/++Mc//su//Mv5fL67u2Pmvu8tMJteznJfi+v7/b7vewvAGwPOOLDmM6GqlpH3fU9El8tFVW9ubiwDMFv4baiMGifmxaRX+8C/9r4DVGZRMgkkLFN4DJg1236p1sGowsxGLwVVuzu2mTuHTSJgvOlQxaVScymXnEqtygK1aC1DG1/fXu+7dr/zLnhB7KaUKjdtbGIsuRwORwFWUOtQtF0zFBbQKSVVyLmgJwUsKrWCqCP+xCos5TxOIq3HJgogqxAoggpAUUmVa61SslNpvFPO4zypMlJ0Doe+uboabq53fd/FJnproiMJUIwuRhcD9V2YdmVOXBmqQlVl1cI1lzJzzZqnMo7zNOWU81xrYi1Axaipql6KL6UIJ3JxbUV/fXF8Dshb/ERbV9vvrGR2E3zDy1HuYP+zSL8U6gL6IkptH6QISkiEPngihDSnBBMyIzMiCACDgjKYplRFQZAWC0ckVVTYEJyVlGdg/F8vdtUMamqZ5ynlJFxh9dHaLlLRLIjzOeXKBRGcpxC8J0Sw0QUl5QIATdvGEI241cSm74b91bUZnM9pmqdpmqZ5mktJtRaESsAiRYUdeu9jbIbY9LEdQts7HxEdEhCQ+bD8DWHFrKweHx/v7+8fHx/HcaylOu8t4pquJqUZ0cgdFtmbDcTrui6E4Ii6vvMhABKLILkQm1rKPM8PD/eHw+F8Pk/juKF2InJ7extCsMRiHMe+7/f7/WCc859/EIblApHz3vlAXtEtF/1y4LSI4otZ5Itqg1kARCXXMpd8f/94d3/34ePd/f398Xg/jWeWUmutpRqazSyXy4hIgK6JDZHzIbZtO/TDzfWNqExzejg87e/ugw990zpP67/2Xz9U1dpNyzC3tm1fvXoVY9ztduN4medkfAL7AQvzwuvghKWdKwAAZvziXIghxibGxjwrXkT3Z/nTrzmztZe2+NUsgW+FxAix1DpN02W8nC6X+4f7n96///6nHz7c3aWcnHdMmoWnNBWubdd2fZfnVHO5nE6o0rVdjHHND8plvMzzBAi73Y5Vj+dTyhmJ7CGWWkD08fFwOp1ev3plWmPRbRqstatpIQD93Cpf7wOCGk22lqmWqdTZfFG2fcIeSMk5pXmeJ6s+0zwbHhJCKG0Xm7iskODR5r2YFZKa5w2hc36RO3dt2wXvEUBqrSmpADoGdIqoYHgg6XNHX1fTik8u43K5HI/HlBIAWLZt5fXvfve7f/u3f/vP//xPc49h5jdv3gzD4Jy7u7uzdee93+/319fX1pg3v2dzh7R4/O23375+/dpWFgCEEGwk6/v3780vFhFTStuEN1u89kIej0cbI3t9fW1j4r5er6+vOwubywmLKitXrtWUmgbCSeXFb0nqohxRVQRFk2MhIOE0oTLdDiVgtr30lOZpntI8ccmk8u2bV9c3123f3+y6Yeibrj2O6TxnVlGVy2XMuYgKOby6vhqGXYxxtyMg5/w0pSnnTMEBkrJm0FQEapEXhfs0z0/HWcQpuoDZgXOiKCAIDDrXkmrhnBxIF4JKndLkUGNDSBCb0LZN2zYhBHPCQRMCojrSJpJ3bd/GeqWVtQhU1lw1V55SuswTQx2rljnNcpryOeURoCoWJAbDgRglU05aKgRc59kt5enny/9n6E7LFrDFdXsft3DvrDnifAAIXkJVj0pLT8UGcm6sSARcjOhNwquINksRo/fUNF0MWmtJU05zSqnUylJVF8tcoYwIDv0ykXbjteCSfyx9sL92GBRfUk4pp1KLrnOptv8JIitkrlPOKSUL6o5IVUphZSmlpDlfxhEAdiLSsGnqHaEj37YhNm3b9b1VLmmb31BAq2oVrsJVRQGoaft+d9V2QwgNUFB0ROARPSH9dhN5y6+Px6Mp0U+nszWiANFcrpZgX6v3z7MfLK5bMLbiGwCc80gOkAozOBJVJIpNc3t723ftPKfL5byiAsX2lKZpzBczhGCQoFXtP3+2IKoWOVLKAFBqzTmrCAKICrMAovmqeh8QP2m42ohlEKq1zik/Hh7ff/jw/sOHx8eHaTyVMtukTsscYvTBB5vIF2OMTfQ+dl2/2+33+6sYGwEBwjnlj3f3fdfthr5rG+9/+5DUn7vYtTtrrCa74X3fxxjmeci5bELBUkrJJeesNgtzpZ3AaldiH4hLg8xkO19H4X7NQWtPnazJtahOVpIqOZE0Xi6PT4en49M4TT6E4IOC5lKgllwLuXFKU8opptCE4IiCzUwIMQYfVz4gqLYx9r1VjVBqZZHKdZwnVplyJkQCOJ3P5/M5pWQNcUIUWmA+FDBL0BVb+9nrRQPutXLNpaTKWVUW1QuicRmYqy1O03FsFaowCxEzz2kGhI23SOZhJWj0aJAF5fAhdn2/21+NV9fOx1JryaXmAkroAYDNSwyIVBwKmlJeQZ3XtZP4fOScp2kyyqqhOwZWDcPw+9///n/+z/95dXX1r//6rzlni+u11u+++86GunZdZ4PabFu3otzAM/OWcOsgOOv+mL8kEZnJnf1DNqwohFBK2bo8pmU9nU6n02me591u94vvVWURhMK220ktXEvlKpV5+9PK98WQzIj0qqRbiAEAcQqnmylEWkb4TfM8z7Vk57Bt+9B2rG7OPM1ZFWKI0VcCXez1C1cW551zQQRT4pzrOKVpmqd5rpyRwCE5dIllLuUiSUp6WbXnnC7jxfsmxga9AjquLKWycJXFMYprAZBSCVGZinc4I446jXVMta/cBN4gWJvoCIjoiYL3SE4ABYgVWaEKFJZS65zzzbQ/Xa6Op/3hafh4/9PTUxnHKdcZgBEVHQqiKLFDIVh8aoxzD4Jf2NF9PbTDql1Ze1Trd3BZK7B04QNRrOwde5SlS6kW120ZKixbmyNUAQUQVlFhtjShibFrokOYx/N4cQgqwlwqK6MXBWA0IhXauECEzVcG15IdF0zna4eqFXucc0o55ZxY6gpFrj+DpEiZZSoy51xqjrFrYyRCrjXlwrnUWud5Pp3PoOAMFhYhxBhj28S2bTx58jEufGd7g7kycy3MmdcMVlSbGJu2b7vBhcjgFMhZivCV4T0/e2yFmuXXh8Ph7u7ucDiM4yVlU4pTSmnr3qlqCLFbxrW1MTZd11mj/ebmpmkakyYQkiBy5ZTLOM8IGpy7ur6Kb14BwOV8/qFt7+7uxnHc7Xa3t7cxRvOjHYbh5ubGgj3+jETHzltUcs6n4/F4POVsFPjZGvD2Oz7Gpu2G3d56h7DS6CzCEZGCFq5zms7n09PxcDw+nc8nromZVSqA8YBi1/VDP5jDrpViMcbdsF999Cj6EEKopdw/PO53u5urK3LkvHFh/yb85JPjGbOyNnmM0aK7Kexr5Zyz2b+kea6h9l1vMdcaWdYkssC/zfZQUeN00urk9eVb8Yu1+7Kqae11beSVVYwuIuM0Ho9Pj4eDc+7qar+/2gcfVDXlzBMbR0ZBZzdHF/bDcDXs9v0w9F3TRB+ckQFrkSY2zoWUchWZ5inXaj4D4zyZpLrxYRwvl/Mlpwzmb49IRuUFm4W+DKz6uUxmW8fLsKeaS0kijDaaw5h5i5HJYiwNABsd7KVcc6tu1zu02uGJKpKikeoghND1w/7qerpclOuUSjunNiVQdPaTgIAOHJGIDaNWMwdf/il6eSWmN7GqHRGN0WnMjG+++eZ//I//8Yc//OH+/t7YdtZ6M3ZLKcVWHBFZi8E517btfr/vus7snw1XtwJgmiZTnTjnLJeyet2vEphNdYKIm2n86XSapmnLL3/uHbOqnUFKkVq4ZKnZQjuLxXUjTJsiyqaEiJHmYYU1QFWZC6FeprEdPDlVjwqCqjH4rm+vr6+HYVcETpfEuUTvTBdTS07TPM3JpCYxNrHpQGmeyjhOl3E6T2MqiZy2bfTkHDoWHlNJOXOa3AsP0FLKPE9dO5XSBUeKWsUg2FRrrlKrVpYqIFmAHLro1LsR+cThae73U+y74BDQC6mQetMIoHcYvA/BhYjOATlBUiABWoiZIpnLnObzdH54+PDn/3Q//MjvP4z1wpULoCKSOieIIqAV1FkzQwBfRvfn42uWNSt7dxXlbr9gIXUhPSA65wNR8Ogdeqy0Mdxw86SxF8DKgsWei7mUokKqCOqIPAE4sqYMSxVlMzLiKrUyFAqRHSJwRV7UkGtUh1UJ/3MYqqqqaM05TfOY0lyFYe03wMqIUQVWTYXnXBQkeNdEH4PjWuZ5vhyPJWWDrJsY7Z/OKY3TCABt15XSpJxVNZcSYzD31hC9qtrQQLs0XbhgEBz5EFxskbyA7afg6BdIQp8d266UUjqfz0aUu1wuKaVaamUWkXGcrFsGthl1bdt1bdc1Tds00WKzOcoBQC4ZiRRpLvV8mR6fno6nU9c2Xdvc3lzvdgOoXoZeRYZhUFWr/m3KaozxzZs3r1692rQ6P3coqLCcz+fvvvvuw4cPaR6LzShUVTUyJTdNt9tfv/vd77qu30S6G4UbFBSVEL1zbdf0fRs8IVmNFbxrYohd1/fDbr/bD8POQqnzbh0wH7wLqppzbl3XtC0olJyO5/OH+3vyFGNwbvWU/ltU48+HnWTw3uhvi165bQJ7FmFmNzsWMdhDAWKIXdubdaAF8lKKqGopsvrziShtw1y/KGJ/LSC/6p+Wqh2fLWsWhIHIkbO71DRNDDH4QIQqUksttVSpSGg2hcNu+ObNmze3r4au65omRuftkQDWKjnV83k6ny9Pl9OYknNOQUupWAsVF0NABUMvLN03REHE2DnLJICFAPFFKvPiUJNfcM3MWbgCiHvmfKiolJKneZqmpan0koRoV280MetVWWt5u5+6aNmXbZHIxabtd1e7q8vxcDidR8RHEdpdXXXdAOQRCUhASYSVnTgH5kpAzjnvPh3x+NI9AvGTR2FNHOO1HI/H0+l0PB4R0YjrpoO35J6Zx3HUdc6brIMZLZCHEEIIOWdVta+trw8ApqOxL7brRUT72HEcj8fjFtqXvsbX14WWUqpiLpITVxuNXkQKizXaPw3qVrLb/bRbDKb655JnndKUuBnaZtd0bXAl5VqKIrDI+TKmKeW2kb4dhq4Nwce2bWtlFHUiqoA1i/Bs/IhahZzr+r7RFkmdI2GYap6mnFKpzCL6MtXSxdOslDJnj+J91qlQ0sjQCCEG9E5JcDGsAIcFNHO6n4/xEc3UlPf1ptsRgF8ILQBsVWhRBGVSRCAH5Jz3iA4QlTC40Drqgu8cOtAuuOjww0f3eHic0sRabSY8Vy/q4Vk28pL8+3x8EdpXQord85W5C4tB7Rq6DTUl55UCoSMlIifPNler8gcAdMUOAQGAWXJOnItyJcDgCZVjcHme5jznkpmrtfaXEKUozI0PKozMKAbIfyJy/fkSEVTZxrFP0yXn2dLmTRmsQAAkgJUXMyNH2DShCd4RplIu5/PDw0OZ083NzbDb9d0igE45Xc5n221tlqItg91uhyZ2DyZOAL9OQdMVy7CbIRjYwFBAR+DNJ+Dr1/HVS1NbfvM8H4/Hw+Hw9PQ0TdM2wUlFDeUz6fkGxbdtazKz/X7/5s2b/X7vvRfmVAqQE8DjZXx8enp4fDyfTzH42MTrm+vbm+taS4wBFF6/ftP3vapaI3+e577vv/nmm1evXv2cMOblExHmy/n8w/ff/ed3/zlNFwD1PiChLv3m3Lb99c3rfui/efsWMdKqbDZ0dBtKEGPY73c3N9eHw+NlOkslR9S17bDbXe2v91fXV/urvh+apnGeEDXG2LYdV57nbK47IUZH3ofAUi/j9OH+vuvb/W6IGDz+tQTl1xwreG4+Os9HCEGMoOxcrRURWHhKsyPXNG3btVfX1wBQcr6MY7ZB9SVXGydTqqiSI+Mfb+vgt8LyS8lOhEuavkaSBbhBmyFL5LiyBPFk5peLSMyMKIhc8GHod69ub7/99tvfffNNG5sYfNP4GJ13AZFK0TTX83k6PD3JPV3SHNvGpSmXIixEQgDsvLWzYaVTESIgKayKaLJOzNJV/NphWKGIFK6Ja1apSOCWKTRqkHvKaZpGS6S2jjK8mG5uJe86zym+SFJ1pbUvxGAgciG2/dDvr46n8zg/MR9KEQUgci5Ect78alVISFSBAHxwjoL3wfvPQ7sdtErYt0VkfDrDe7z3VnnHGG9vbwHACG42WuZyudiGsA1LNA9HewmtDZ9zRkRrxhmp1jjwxl40Pt32htjn24e/rNp/7lBL0RYMnGsSriLF2uqyFDj6jMNvaPzzPV6wqVKDWlfR+XY/dM3VnksdL9P5Mp3HacwjKtRhCI5803hAcr5pOxYCcDnXUmrNlSWbT71zoWmaNngkZ6lpmtM053EuuTK86Dgv1w6KKsx5LhME8OiSTsUVH8kH85zd8GlhlcqSS0o15ynXMnMtoAqVvaBT8GBtHQVRZQFmKTYtBHB5HYILwRhkHpwDCiHG3XVE6rx3qlh1PI3ndJ7KXEUBA0BjM+CXV2XJt/CzOPhJaF/g9MXxV9eieK2LF6P2xbhBVCqLgjCjqFtbdtYFVNxq9kXNA4vDDSICqYKNTC4Fai2OlGuuJaVSKgsglcrjNFdmwzItzbM/7cN/3aEsNec5pSnnVLlajwFgGawj6BCpsiaTWqkEF5oYADTP83g+jeczAtjwdUdk/a1NEdvEaMHMhpillMxBfRzHcZrNft0W1Yt0eMt8nRpogOAR3FKR/rqrWu19Si1GgH96ejqdTuM4ppSscWson+0XZinTtq2dTNd1V9fXNzc3Nu+BiGqtXCtrzSyHw+Hu4XGcJkS0yn4pfM3y2AVhcc4dj8fHx8fD4ZGIhmG4vr62RuAvnjkzV64G57pAAOACERGCJ08ueu8i2BDfF3XVFn6WtwCVHLVtc3W1/+ab194TV/be7Ybdbtj1/a5tO2OcxRiY65wmVfU+OPKW4ddS5mlyznnv2rZj5dP5fHg67vr++mrvuxb+y2y6TbG2dU8RUVhyztM8T9OYci61Ou+G3YBAzntFqMIqkkoe5+l0Pp8v52mabGqQtcHMF1YRRbdJgfqbovsGhJDJZzZZu4liFswN1VxLFWKMw9DfXF2bbK/TDhDJudg0r1+9ev3q9dVu1zRNG2PXNTfX+92ujzGC0uUyn0+zQy+iNtlmLrMhWDkX4w00MbZN23d9E6PNCyEkwE+nJJlT7F+p2FVUqnCpNYlUMPybFnuEWkpO83g5j9NoHke0jpWCtaduuY1BOxsbfKU0rhF9LdxNoxuiFe5jSqXkfB7P/hzRu64fYtshORQSBECH6ENou3Zou977SPQVG+at4N5UbUvWtkZ6U6PYWxRjvLq62jinhqWbfaQJ2LY+Tghh+2TbEwyKz+thEd0+xKp5uy26Dm4+Ho/2K1+dCvPy/GuuhTHPnItwFqkiVS20r4eufgAKzy/v0vIwyF5YhEEYhLVWYRYI4GxfLuxTBg+eHDk35yrH09PlXGstpdrsSVREpCYG8g4dgTPbd8ws2bbvOc1zyrkWFjEES/jlFGoTX6Wc+CIFa8RQXGXi5IGIPDgStFGOqGCpSJrnaRyh1hFwvkz5Mo9P5+nV+GZ/fTPsQwjkHJm8DAER7BLJOxdCMFc354g8oAcrO5U958HT691wHPofkXhKp6dDrkK+DWEXYwAHi2p7IbbTZ5nv13rtujoHLS+y2TLKsox0EdezcOXC6gqDqAfjjSxiO8VtHpUZY+PKqCNHziNVBWRR4ZJSBakAbJwhVgXVwjKXwqyt8xsKZKzKtVj55Dq+tgurgtRa5nlaPGWFyRvWulDtzY6/cMm5CDMhRO+a4IVrmqfL+TxPlxhi13Zt2wKocVlLKTHGq+urpm3cOmLBSCv29TiOKRdyvmlbAGyaFp/bzwtaIIveVh2qp8/NDn/xsAdTSjmdTo+Pj8fT6TKO8zyntLCzbO1uxJy2W0K7OVG8evXq9vZ2v98bFmey08oyzenp6enw+Mhch77dX+1vrq+bJiKi994737U9s9jWcHf38XQ6XV1d7ff73apcstP7udrd7o9awhEDeViwdJMaWSK5qiIWTcb6aYQoy5sgYM3IGPf7HfM3w9BzZe/9fth33RBjQ0TMC1Wkcp2msWRvJolN0yLkkss8z4Cwv7rq236eL1NKh6enrmlM90+/6Xl87TBs21y6rbMtIkbaOB6Pj4dHEWnaJsR45a1yBWae5qUher5cTpfT5XJJ86wA5jXknV/mPi5kO1n9VX7DsayfJYy+tLXaED61SFBLAYA2NjdX12/fvlWArusVwXnPqkTu1e3rV7evurZziMG7vm1fv7p9/fpVE6MIHA4nRydlrcwMUpRTSUasSiE755oY9/3uan91td93besW5MA6ZS8ubEUovnKhughglKuNxVPh5UrMDx6k1DxN42W8zNMESJtf22dtY1splofhC0c842YCPLcq7fd8jF2/u7p9xayHx/tpHC/z6KaIwVOMbimnFJG8C01s+27Xdb3/QkW2If9bAP6sRMYXvu522obVW/zeRq9aFW5UOMPeLX2xj7V4v+njYe3x21UbLN80jfXvllurmlKyRoCVNH9F16qqtdRSMScuRbmKVtWqVrKrrBNVFp+fLU9afldElvnrNlJGgKuWXFMuHsGBM2ptE5voNXjvkErlaZ5ymStXWSTyFEPTNk3bxdg26L0gplzHuUzTfDpP58s4TamUKqy0uTgoty/ih43qSyVNda6e29CKRwlagUXAiSMBZUERBwhVai5pmqdxqjmduE7ny3y6TKfLfBqPV7dvdtdd25pKMjibVyGsLKjknV9sMsNiNOm8hVAbMkcl7QJdtbF1CCWPh9OUqw+l65wfeo0KssjOzJblsyfyaWi3Gw642QOawb31QRRkfR4LFJ2rZ6TCUNUpRlAFFCAToxpp1Vw3QHXB2pzzIUZZJrKDqAl/YQFCwAECKyj50PYRqe/7tm3AOeHCq8fBX0d91/MWq6XnecxpFqn2Eln4gIWiQyKYWXJhBAjemb+DgqqwSlUR6zDUWmplWzbOOSQUlvFyybnIag5jzHNVZZZaqlNkS0XU6iMAy5qABG00O3pCR+AQHP7y/K7PWlyikks+nU6Ph8fL+TxPc0rZINxSy0YbdN6R67xz3ruub29urr99+/b3v//969evbRCFLebYNJIzqpY052ns++56t2tjREAVVYGlB4vEnJZ84ngS4WHYGZfHfWUY1OeHBRUFrVzTnHKZRYUQnX/2YwGwFjDanJjlqS2m1ICwnAmgE65d08LVTdf2oLC0iLz33lv+AQCqERRiaFRhmmYi530AROfd0h0sRYW9c6Bhnub7+4eh64aui02k/xqVTu1ls4lHuJQuBvDYEHAACBq7GJumSSmfT+dpnqZ5EtHKNee0NDgJQZcyZ9kGF5OFanUQOIBf3WgHq1RWRH4VibywmCXkypfL5Xy+zPNMgLvdLoQgqv/4j/84zlOptdR6eDpdxnHXD02IoCDMMYbr66vb29vXr14FH0U0+Db4BhSqcJayS93Vbse1dKEhomG3a5umC/H/889/+ObNNzYjGGBR88BiM6nPcf3rV2iCt8ol1ZKZC6AsAWiNG7UUS0ZLKc57RKhVwHyEzIDLkfe+7VpTQm+hd63aZe1LrsUOgAKSC6Fph92VKqCjdjz7EF0M6ElJhYAcKpLzLjZt2w1t18fYOvOxf3GY1I2ITC5hoOB2Ana8rM4NdV8nBC0lvqXH9mPbJxhcb7idJQfGcrUvVPV0OgHAMAzmL7m1KrZ/3YoHM9Wx1+CvvGbCwFVrkVpEKyiLsgrbYBzVxYhvje76fHlL/lFZhAGhipZSUyqpBJ8y1xq9b1zTtq0jx7VqFfs056l1DVFjnnvLMFIgBbiklM9jyjzNeZrzOOVpmU1YhUUVEKw5i6TLGNrlKkiZpFZWljql4gDRi0LmkoVVAFiRxQO0zjtArOIFG3DkvJAq6aTp4+WRSz4cHn4K/RCaLsShabsmBu+cd77xFJypxR3ZiDnnvCfvwZGFXyBgEKklarlu403f3QWfUuVSiyvcvJAPLhjI5xPSPgfknznuz0W7ERl5zbLMZpGFq2KugIWxgldiBBEWVAQQRKStC4gAiKpkto/eB4nrBGIQcwNQYQXTeKOoOsIutD6Evu+Cd2qmw6qqixAXXvTav/6eLWNgpnmeSkmismgnF40AKhALVtFcpTJHT9Ebg8rMc5gQvCfvPSLM8zzPaeOheO+ZOU3peDwpwDAMwzCYJWRKySSwzoZWK4ppYda9SZAYUJY9DD2Cszb8z1/LZ0sdEUQk5TRexqfj09PT03m8zGm22S7MtVQbop5ySREjIhj3u+/7V69u33777e/evTODWKPRtk1jahlHQCoeZN93r66vWx/UIDJRZwJhoFKK2d5dLpe2bYZh2O12v9aoFRfXzVrqNE3n86nWgi+8LWNsQgiu9YRg3d0131y6RMZ3CsFbe15FHbq+EyIvotPi4WrjqK1wAYPcSykpZedSjI1zLsZQK3OtpeRSSvDkY6wlH4/H8/lmvk6L1+R/7diwUABga5hbNbzuyKrqne/7AQCtqVlKAQBAZK6IYAZBvIwBVV0nL5j8YsGxNiH4rztwEbaj6do3h4c1slPlejlfLudzTgkJd7vh9evX+6t9rqVyPV8uh6fjd9//+P7Dh6ZpiZyFE+/80Pf73X6/v/LOi0DwERGncRzHy5ybvm13fQ+qV8N+v99/++23fdd5pHffvH37zZu+7wlp7biDWHTfMt4NYPjiFoOwcCl55ppUGJxue4Oa4KaUnFKtRZSdOlWtla38sArJimDraju3Of09w8hLl32tCtab6HyI3TDY4prmXnXZ4IAI3LOdZ1g9okKM+AXP1MplM8/5rGq3k1jSwZztmwb8pJTsFdogdIMiLK7LamJv+Lwp4iwwq6opV+2jVNUcamutG2KxnZsR7M/n8+Vyubm5+ay6+PRBoCpZWOCiwKq8Mj+tWn/+c3VnehnXl0GGggQsMqU0TlOTPDlNqk2Irg9NiN57ziXNszAgaiTnPYboQwzkPCDlUudcLuN8vszn03y5pGm2oVfCNrZG1j65Lv7FBJ9M62GU4piRQZVz5Qs6VGWqUhOXWliZSaVB8jE68k7R2ewjAkZA0krylC/j5XIvNGjYu7j3zVXb79o+NCE0vh26ZmgBBUFI1SME50NwFDx4sjhNnoCQETDPV0246fu+ac6XPFe2RszGPgabWKx/NbSvcX277bAwGkUFTEWnW9dJrD6FKuJUi2hFrag2u0VtjNmqWLEkWlSWBUfkQgiI6gjZe6lBpaqY5IiUDCpHchSCR7AsDZeF9UlQt+z+y4JXmIvtKTnPlSsioNLS10EEQFHMledca6mGcXnnhOtcueZUa+m7tm9b70Otcjyenp6O0zQiYghXpn4mxBAcIHrvWPh8ufDxKZfStf3V9Y15WrDq8XyOMTYxOu+RyMQKtmURgkMk1F/TZ9+CgfHSD8fDxw8f7+/vTqfTPM2rTtqmRprcQIwmoyrOkbHiX92+ur6+2uL64+Pj09PT1X5/c3MDiN65169uu7a5vbm9ublpmqi6+IEROlDMJR2Px48fP97f38/zbI3JX3SqenkgoAKw6Dynw+EpzTPAEsBibNqmubq+jrFdnFRo5Yao4mq5SgTOMgkxMFYrc/BmNrcgK4jYtu26NWOMjaqa8+E8z0aVQkgp5ZJTmr3rh6ZpEJchQKUkkeavSEN//bExlg1Q2eK0915VhaXUYlwHWQJ/WQ2/nYi3t5WZRNQ5WnoadRETyTpX6TcdW6vdvlgk4+t/2znnkoyl5Z2LIV5fXe13O0VglQ8fP6ZUgveEDgS4soJk5WkaL+dzmrOwAhGC1lpzmnOeAaTv26uyO5/PNVf1+ubV6//vH//11c1NcH7XD9f7fRPjQoOBBYJAayZaX3N1z/vsUBWVUutcysSSgWTps9mdqnWe53mackoOqW87W4POsaggkYJWqQ68iz400XlrdgCARUfWpZKxPdG8qFcqkYV350Nsu17IeRZRQB8CuWjDe5U8kAOD5siATP1sC7ak1upmo+xs3T0Tl0/mgjXPG2f2/v7eeHMbt86qjm2L2L7YnrgFflgdareWPDOfz2fzntocankZl4yWRlh0n+fZEoKfealgQdkWWtQy6E437twS3O0L+bxk5/UmI9Raj4cnxBIiIvaIkAvXLI0P0UeHSOQCeVp4IcACZZRcx1TqlPI0p2k2AnzNaVlbZmu7oMiwdmXhK9TMCiVDQofOLCay6CQoFJxXwAxZQB1CcGhmbQEQgJxSESqKAsJSF7NhJlEOzg2eKuiU56fzMZfsGh/aGBvfNqELoY+hi42GAL4qgQIDiKIISBUeS8ZSA4ATVZZShHxdkCNQo4MbHfav99r1ZZEEuri8i4VytGi/oNoqIiCirFJFq3AGIhQE55BWTTwsQxERUQReJoTOO8TokES8eF7pUWgmd0s9AWBom/X+bZ4dPmcMsAB4+rmuXaRK0XG6TPNYalYV59yL8GmDs6DUOqcEwp7QEznCUnKe55wmYR76romNKNQ6jeN4PD5Zl92WSuWqICF4Igrei8g4p8s0TuPl9vbV1c2ND97HkFOZ59S1rQ67BtB5VFpOd7HU+y1dduvCPh0Pj48PHz/effz48f7hfpzGnPPq5FZf7PnPTDTrF1ztdze3N1Zkm1GG1d+lFBtjDAA3V1evbm/3u13Xdaq6mpURETHz5TI+Pj5aaJ+mab/fb1vGrzvQ2uTOBsAXtmnKhFio1sK1ctcP3jlyzt6Vtfu7yLCN80TOWKqoCjlVZpNhWLbJOWfbLkVkmiYRRURrVtp3DJ8M3tWCXMs0zT40se1CoOjJWEWqv0AJtuNnIituf6vPzN+yGO3pMoGCFyPWakjsEktUAMgRoneq3pZkrcTMFvnEbOpW149ffedfnNwaxmk1W140LwseBiJSS2VhImsI+r7rQozeOwEYx8kMqBY5a2UFKYrn8/j4+PTq1dNu2HUdq+rT8Xh4OpxOx3menI8xBBtXD4j7YfcP37773bt3bWzcYu+xkqoQDZ5Z/pOspbW8Bi/v/UreyFxSrUmhotf1UgBYjeKU5rnk3IQYQkSH1h8z2NCgdnAYYghNIL8mBXabN5HWsnUsY0lxwUwtZDrvVGOH6KuIKJAzp063DKgkMtMSXTnKn4X2jQCvqtuYlq0it1huvjEGmB0Oh4eHh8PhgIi2SE2h/lmT/rPoruvMVnsbjTxvkMA4jmZjZ8fLaTS24RjH6Jd48hic824Z7Ql2c+WZiPj85zI9Z9mgtq/t+SIol3I8TFKn3a4JwaF3CDxyiuTb2LQxtjEG7z2QRaCca0rlMs2XcR7nNKVUCtfKwktNCwCrUcXS89GNB4aI+skGLChMNQTvnKdKqABJUTU2npxzCOLQkQSHwTlP6BUQwAkRIDBUXcmCiOKUFagNzdUQMGCRy9PT/cN9looeh92w3w9XfXfVdfuu72KDziEhkgIwS6mcM+csDCImYjdmqzPyAjxj4vBZwQsAX+21b9H9GZNCFRAlAVnK9gWdElZFZRTGqqgE6GxYlS0vWBVzLzC/TY6rCgrkHBGp86t8noy7vozCsnska8drHdSw6WDWk/mkdlHVaR4BZJ7HstBJYCPt4xrXeW1YeoLonXeIIJzTNJ4vlwvXoiLcSSnlcplqzSH4tm1iE31wuaTz6Vhrdt41TWt+AuSoaRrvabcfYuMVeJqncZzGy1jrDohYNcYIzoFbKZPrzYDnZ/S1RYNoy+x4PN7d3/35z3/+4Yfvj6fjNE6Vq7GnF2BkUZqwLZWNfWPS2GtzlYmNbSLmUmfEHHObZ+abm5vrYdiajiaKteCdc354eHj//v39/b1x8m1e3Bp9f/mwSqdp2tdv3vzjOPb9UEpecGCbGEZ0c3Pz+s2bvu/sV4jIpn3aB9D6ZgGA916kCSHVtenY9z05msbJ+IO4qoStDDJpr9lzxhhD8N3uak75PCXGC4O7vhraYd8Ou9h25MLPXsYXT+eT/7ZVw+tIwzXB2g5QFdFa2WKnvoi1651cXni7ZS+7Tytwvnxnq7N/5f1fn8K6Frc+ET5/1wobW26bobQDJFG0zlkpeU5c6rp9AxCpwPk8q94j0NPh2LUdABxPp6enp8fDU0qJfJjmfDxfpjkhUkol56Ki3nu35FL6WaK05O0C25V/sURYpXBJLBmAkXSzn0awVp/UWkXY0B0xtFvVBk0pADkXmmh+pascHww4thC0bi8CWyIgum2UsBbEznkFRLYPdyZhACQB3MLcMmAGPw/tJky1xDqlZKvSTsPcqH788cePHz9eLhdexxwb3gYAZjKRV+8nANhKc3sl7FemaWrWwwB8IwwCgCnibGQzvHDy2Upqw+TXeZL151pvhNA0oTB4TzUD28z75UbKBv6ucV1erorttC1br7WM55PkaXxz0w+9axEAOfMk5XzJjshodASgCsxSq3KVUrlWLpVLVWZg2ZyYl5xqSQzN4szwHgNP8ZN8JQY39DFgDBJ8IZdJGaQggAaPwQcM3gdwRqhQAWGo6lVRFAVZSczQzfmIrsXw6vb2d2/fDaHFIgXk7vx0PJ4v56kr066M3bkZmrjvhn3X923fN23XRUdUCpdaqy4pPCg4RzabDdaqDVd8/SXcvh1f6tpXso6uORezwPauL0C9iKAgEIKwVKhFU1UldRFRlXSjihhVfrXOQthC+9I3RaB1V4fVl8fWuMhKAuYVsrE0DD8Rwrxof23f0XG8qMo4jTknInRu3bQAAFAVBcBMZBDEE8VADlW55JzmaZqnqdYaYwOI4ziO45hLJodd3zZNQwhzTuN4Zi790AFEBRMRaAw+DM1u1zeNV+VpOo/jPM0zOR9zi+QA0YE6cgTgN8G/tRV+/lBVS64/fvz457/8+f/5f/+f//zPv1hQ7LrBKtp1IBityZCqdRm8jzGaW9zN9Y1RZpjZ7KtExJg1FunNBAMALOT3fW+D2C06TtN0d3f34cOHx8fH4/E4jqPtC7+GQLcciETUdd3rN2+Y5fr6RkR8MDESca211r7vrq+v2razR+2c2fQtXpAv/x1ra4YQS6m5ZAFtuxYJp3F6yRy27WmZ/8EsIraB7q+uu76fC8/pwiKgsutjCCHExoWG3Od0p587lh6imFHIAqvPc3p6eno6ni6X0fZcWZ1mYMWKWQTCEpgX2RkthEt71Y2EufpR6BqBAQBFl5EbzOK80qdk7196Clu68CxRXVYnLldkzbzVkwPJBC3AqsKp5DnVUmz6B1dB7wBxmktKpcz5/u6+bRtEHKd5HKdxnnOpqpBKOY9TLuycn1M2F30EJaJnp0+jDL4spKzeXcGb5++rqrJwqWVWLgBCzzq+lcULigA2y8URAaJUqcLkCJ0DBHLUdW039LGJ3ntc68rlj02C/bKE+HTDwS26AwASqfnWIQIKLlU+LAkTro/yk6NpGpv7gogWhi20A0Ct9XQ6/fTTT3/605+Ox+OidmlbVR3H0ZI/C95lbfdYwN4O+yj7gW4ZrkhGkje3eV3HD75M0Legax2uLbSXUmy7+Mo7Rdg0obDG4KonY2itE9rXLWmL7luS+9lrqwqinPN0OnOhdJlqquiDEmbWmquUrKqmDwNRGzyztUhs91MBEVSllXm6gi1Lr8dQ+SVfI0SFT0J714Sbfe8wegguEc1QRi5ZOAMIBBe8cyEgOpGapRYQJlGv6IE8EIMTEIc++Ni40Pvm5ur6zes3+7ZHkeN4CvdtPslhvoxYL1jD7KL3w+Wyb/vrfn897K+5j4FKnisnAWHlLDKlvNVwtkK3FfPMR/z0iXxWtb8AjSyeM3OtIIrE3lX2JAtIjySIgMpQS81zmVJSx0HIKxCEjceyYGvPS051mX68yNxpcZa0TpRZ51nlwyoMatBEYbHsWxCcwQG6pWSfhndVOR6PqnI6nUVqjBHRW3Q3PMBK9iqKADG46J13xLXMOVlQt8VmExFqrdM0z/MkoiF46/7az6Dr+r5zLnDhNM/TODWN78K+IYpEudQ0Jq3SxCY20QUPjngdyeEQPS6Gky9gla8cqmpA3Pv37//0pz/9x5//4/37n8ZxdEQ+BGYGQDJTmn7pwXAthsyraghhv9/f3t6+efPm6vqqaRoFTXOyMY7e+2EY4EXyPs/z3d0dAMQY3717Z6EdAKZpsibf3d2d2eMYZdfC/6/E5IkwhtD3/fX1dUrZduK+72OIzlFKeRzHEHzTtsZeJHLeebbp2Gsdux5W67um6US01KpcCSn4hW9sb/vWjDSw0TYmEZnnObRtaFtQJp6hnIVPtYMyhDzEMvQs3v8KAZyqllLnlM6Xy/l8GadpnuaU0ziOT4enw+Pjw+NDKWW/2/XDIMIAumbfsNTvy9Rw84yxzVDWxbi+2yv3ExBFoFQZp3Q8j8P57Jtmp2I8LPp10X1FJr8QMlkDTUGYcy7W5Vm7YBagdJnHXAuXyrUWgIQkwcRXBACXcZqnMUTviCpDFVFFFr2M4zTPqVRAQnIiulBDdBn+tWY9sPIxVlbvmvx85sNsflTChWsWrtvuviwoUVgHngbna+kBQEVyKYXrUoEjBB9i0xh9bFH72iSDBY3XFU6Al2jmJ+HZbg0BrSOm4PlxIZENk2m8j0Rf95K2UG2YvIjYbdd1EEsp5eHh4d///d8fHh76vn/37t2rV6+urq6aprGuvEnVTem6BW9c1ba2CvDFpFddee/GxUurWONwOOSczb92+92XVbvNsDE/n6+8VIht1xSBGF0thEqsWhVQtQq8eIlfvtNASNaJsbtqyEopuaTk0Gnl5YYut9iBs8HgS1feostG+AYTKq4ZmW79ZViDGiIICCrjmq+tefN2IfuuJy9AHtW5TBh1gjRLzixSwVXn2QfwjlSCU3TC5J1GRCVfEIrkuurDCNE7753z5IJzRNR3zdWu74cmJI+t40jqsKBM5XIo08fLqTvEoYlNcAiMKIBQmMeUH07n+6fTlLLoC43qcza+dMlfPpEviUL6XLSLMgtVVmTAGoIIuwWhFwABB6giXCVnTZOqA8WgSM7wrgUDUavLrUZfIC0VS74IER0RkhkvG8FN0WI7gzIocy3mWyhcaXHah3WxfSWPVoCciwjP00InFInWFrDXSGy0qAgiRO+jJ0eYa52nyRpaFq4QcZngUbIhuvaiW5JLzsXYxNgDQMq5pFTT3PgukgvkSEBKLVNCH5q2aWLjvEfvwC2mngvI+Sv67MZk+fjx43/8x3/86U9/+v7773LJpvIywYwBXM77pmmEhbnOM9W6zDtvmub6+tqE7GuXvRgUYQ7StkfYiJcYY631cDjYNvHq1StbtDZE7uHh4eHhwZh3tshtOsWvD+2I5ENo2mbo+2Hoc56J3NV+37ad834cR6vXbPwXVzbgwYrUVca9hiMzekeMMYrKOE9VFu6PPTvbHw2iEDPEEDVvuJTSnHItlZmdcoeMPPqpwhjrpa/zvpRrkbioyn7+UABRnXN6Op7ef7z78PHu8HQ8nU7TPI/j5Xw8Xs7nabx47xXg5vbG+BkraAxmTW5A1MZr2zq927u99qARERWQqxZOoqdSlUXnXG6vr6/2u77vYghuw6PXW/7Vx/CiOfXFRakyiwkH2CIcbDs9GDfTbt1G3jaQKEIAwlyycqbZBkF7RKdIxpqc5pkVyAePKKpG/LSNWtf7CcsdeL7D9h1Yw/anJ1qEM9esUpAEl379steDanAe27aJzdpZ5FJqqbWw2Z5qiKFpmvhC06UvBG8vaqE1tq//91l031zvl4EW9ieS9yHGtmm6GFtHHuBFMr8elm6aI83LHWYL0jnnDx8+/PjjjzZZ0ZJ164UZBGUWe7IOGHzZo1l3cjEmznaZL9n4srrSmj5wg5e2O2DqdjOuWcUaX3mnmi5mltguob0aOrp8hrWfVXW5R4C4iqNxE3GJSs25pCSlQEAH6NZxQRswYu0NWN4aQYTlP2VdN0vQBovnyyuuiguMAIIoCKI2Sxvdp2jQELo2kJJTQIqkTigDJAEpVcBVctVRcUseroJanapHAgJwXkFEFl0HoTWcEUARxCN00V/v2v3QdFOUxkMkcSiqeUo1ZazqFKJz0VEMLnhH5Erl8zifp/k0TbmIgF/a2V++SZ8eX7jRrfmOsRyFuZYCWgRrAOCo5i2EAsAAgMygzDaqUzSVPAstek5YYTGL64uXzbp0QAUXeYgDIniO6ytsZcAKCNdca2EuKqxLpbFd3RrdX6w0Qtzvr3LOT0+HcZyeDoemjVdXV8Ow73oHYW3eMzuHwZMjQLULfX65AWCe52kcTet5dXW1mUJM05RL8c5XViXv0AlLE0K7vxr2w253RS7MqaQ5c+UYopnWkXNk1GdygVxA82v5hcejqjnnp6enH3744c9//rM12Iy7Pqd5HEdbb0TOUjgzM3WOrGq3GTCvX79+/fr11dVVjJGZ52k6nU68jn820vg4jtfX113Xmf6qbdvdbme1oG0f9/f379+/f3x8PJ1OT09PAPD27dvXr1+bZ/6vhIKtrnKIHiRqbbSCCNQiNQBRrmWaZwTgpTxHb3U2OR+8KNSXDlYKulIcYox93wHq1kcEgJWYpk3T3NxQrWx6obZtY9MYrU5FOk+7oXE+eS69JpcvmmcpxcSvf61NAgAAInoZ5w/3j//xn9//5bvv7+4fj6dTyqWUXNJc88y1dDH0fWdsRxtbgs+dSFwUQSssD2iivlUdtKjZl46VCE5zGqc55TsA2O+H29ubd2/fvnv39h/effvq5rptmhD8p+X75/H7Wei2kFCe8Xi7YoO5S61LPiUqIhVhWeEpzSmJig9eFVhES2WWnLND9E6J0DYEVATUypxyNqMbFTWDelGdU5pz4soanlsVX7wwawb3xQumqsxVuIhklUKgwvTC8nLp4Bo9bcMEWotzxmc0B/UFIsKl7PsS2lxA40Xxo5/+xfKVpRPLWFkDgtG50LTDMFwN/VXbDI4iLt5hn1yLtbe3athCu/lDhxBsgJPNdrtcLtZxb5rmzZs3qvr09LQND7TwbAn9ZhtnGNXlcgGAWqvNdL+6urJ2Fa0HAIjI8Xj88OHDhsnR2t4w2p3V9xt5/vMnRRjbGFliG3JKwqCMoLS+bQq4MLJBkJwasVBEq01zV0FSrnWaLvN4RpS29U0bYhPQI1iWiChEIAACulCllkEdwmKUOlhb7Lp5s9hTEmVVUJumgkIABM5T24ToHZ6fH2ZQF7FZWJAKNdQcyAf0jFqBFCVrloKzoKsqSdLE5oBj09gUzHbdu0DkgFBUmIuwQ8LW6S76qxiuYkiexCETFGaRnMvEpaowqjqk4Hz0wfuoijnXwkroHKKwQDXW+bML/xoAP3kon1ftKzPeUHdFrgCgkgErO2TG59COAMgqoMomdhfJpU6SkRHQuQ1nlxcu8ta8N3rRMiGayNz1LK7r86YkCEoGL5fMtYiw+ZThivavKNknixGR9rurUvL5dJrneZ7mec7ez0gBXCAlcGidTEI083YQc5ZZRkTYwsgpjeOYbXRHjGbwYi+3MItzIppLdcBaSnSu69ph2DXDTpGmlFNlAHBEzjuyYt3cCch5spL9Oa7/XGC0lPl0Ot3f33/8+DHn3HX9N998sxuGu4d7M048nc7em9mRw6XzDbUyALRte319/ebNG/OLdc6VUi/j5Xg8qmrXdRtL1pptRo5lZvtF08hZyX53d2eh3Qyr+76/vb395ptvbLP4daF92SMRgVScSgPKKlJLKRkBUsopJ+MdhOCrcKkllxybxnlPLMgoImujCVdAzjnv2q7j9enYwBsL5GvXYLChtLZhOe99iIBUS2mIdn3nXfUVPBHYIIuVovyLFySq05yejqcPdw/f//j+/cf7w/EkzIAQHXoCdA6drwYF12JFDyE8w4jrIsVVoWT9ww0sXJcOApAqpFyPp8vj4TDNkw/+ar+7u388nc8Gkt3cXBG1uBmCrnnwy2PrssOLH4J1c1NVFs61Vqk25IRFSq2gkGs+jZfD8XgeL5XZe2+hsHKVqqBCiG3jo3fMVVVdUEVMKeVSmBkJvXPOBfJOVKY0z/Ocaw3MshApvjIz8OdfLRshVYSzSgUABALbYp/RShsVCOuLt5joLlwU++ZCyXquUuElbvD52Sx3aoMVNrzB2BC64KUO0aYPXg3DddsOwbf4vOI/D+0Gnpkp7BbaDXO6vr5+/fr17e1t27bmHmMD1F+/fm24uqwedst04FI+E62oqk2MtYbU1dWVOXCYmYTxUWyZhxDO5/M4jjYrUtdO+GKePY7Pvgtf3hjEGEKMHBsfGlcLkqACukXw5wREgERWBnNlZa2MrFiNn41ay3w5PZXp3LXUX/Xtrold0OAIEYCEUFCUQS37VxQBGyDLS7W4EDVWdFoRxHIIVFWTMRq84tA12Ea/33dtgHqHW8OdBJ0SKgKBVIHKqEwk3i31pFSpmQUqYAaYsU6ibL1D9A6RHKAjH1xw5FS01pLzxAHRu0AwRL+LYQgBSTIsA9AdCGpVzf+/9t60SY4ryQ5197vEnpVVhZUEyZ5uzsxTj/Qb9fMkmclM356mzaQmh80NKNSambHcxf198IiorALIBqY5mmcyXMAKiaqsyFjuvb4dPyfmlFNGAEveu1xkJDJZAJGccZxS4gSZQa8W+GiqPm6gfhS1C4NkkAQStRqtLTecgXS3mUl+kRGWfjREISPGSs6JeeKEiEDGEhpaFOIXGogFDccszDjDXuauphlpIouvhWAQDaHkmELkmGVFRQDKjG6ExazfXxYiVlXTtp119mS7ubq66oc+xTiFmPYHW7AphIwhg8sOyyDsrNUANOesnr5OoJgSM2u8Ppu9qmqatigLJBNzDuM47ffeemttIsquDCwHjoJkCw/WKCAUmUnAAjlEh7ia9l8fatqVDYqZm6Y5PT39/PPP2rZhkN1upwsvJXY2e+/n+i0Lc7bWbrfbJ0+erCG7LEj4u92uLIoVMbfyZqtSnAb0p6enqiu13++1xP727dvLy0ul2q2q6uXLl8+fP1eP4a9dx/qENK2MzEKA3vvEkAWnkHJMU1AtTgKAoihPNhsi6vsDElbOaU+7ApwAQGSW0WZGJNQH1/cHzcM75zabjYYazFKWRqk0VfXEWouEwxT6w4ClK+vG1B54k62bqIno0syk+CEPCDRaUs6NGGMIkQjLonhyerJpa2vQElpCY4xC7KyyoNHcQi1zQXfRdTf23u6AEIEIKbp0dmhZcs4hxn4YYYApBMXvWWsQQekZsEB7/1Ae1oYBVtanI0Nzj1thlphTiDFl9s4IQsxpmMYY4s3dzZu3b//yw/dvr69A0LmCAJlzmqaw4JsARClimHNRVWhoGEetBFvn6qZ1vkQyLNwP/aE/TNPonEVYIHDvWwL6QjO4D36glHwpigQRRDJAVpBFL28GTM3YXVkYxzSyN7gkCHHBti3J6+N09HJXjsrEcPxX7++cnVQqb0JLxvuiKau2bbZl2RlTIJoju/7YtBORrkE4IkJYp/HTp09fvHjx448/qn29uLgoiuLs7CzG+PbtW6Wpmemlp0mzjEt/6X2VARGbptHdQOWd1svUT9SEpRy5mOvJ9H1PRDc3N1rUf6/fg4CKXS8KP3iPNkoSMTLTcQoZMCIZEoc0xf4wHPbTMAIVaAtyBRgTUxgPd/vrS5TpybMX2xdn5Wlja4dkBdASpMRRUlSG85DTlGPMKQpnmDsY+EGhVtmcEBbWeqU8FwESZ01T19vT+snTk9LTz9/SfZkhCYeUZ+nFFGOcxiAxWTIGKWWIkXNIIQ4pHxCmwjAaYAJkJDFkjCHj0Fm0kDnmaRwOQ18EC0ylMeCdLQtfORs4hJwMWTCmsh5dGgUMQAIEQWtsYX3hvCHLxBk5QWaSyBnXkvnDkPbXona9GVlk4edn4kwZSESQWUATH5lFPTHkubaFJEgMnDIHUT10ZkNGo/Z5Z5qz4Ilz0q4SnLP0y0LEBUa3QAGtIW+NcJp1f7OAhfv9dkbQPY7aAcD7oiyLsiqapmma9u7u7m53N8WUtayijFfqIOS5g4eIVNRbDxdidM4V3vPCJpZSmkIAEWWPKstSQNIYhINIBLLoDFs3oh1z2sfsEYvCGe/QKKEAWUCHaOFD7bo+Ea21q7Osi/PZs+dNU9/t9lX5xhgrLDFNKUTODAjTNKUUAbCqqidPnjx99uz09LSqa1pEncdhzCmJ9wCzWdIqu9YjNIPdtq0W5rWz9u3bt9pKe3Nzs2pPvXjx4unTpxr6f2jUPjdRI6Ih633ZQOYMJjOE2YWyhkjRT9uTkximoe+VcoIIrTPz4zgKrTKzQbLWFIVXA68ZCC06anWQOSOi8tAJCCJa6yhmAGCy2VaZKkIBJLClmGImsf/r1wOy0K3PNEE555yJXOGL07OzZ+dnhSOQPA29Vv1TSs4YMmQMESLPXMsMAFYRV3lpeJl9GFgTU5qtmKtJzCnlxFnvGxpsmrqp65Nu0zSVc86YX4x3ER+IMc6lbAAGSTGN43i32x36Q0yxrEpAnGK4vru9u7n9+c3rH3/+6YfXr+/2u6KonC+VrBYQV8xziBHmMDEnRGPNFGPKGUAcofPeeZcZUs7DOPbDMI6T996QwcXdOSamOV7Y71oU4Sw5M0fmSIKLv6/c1lrBwJXGThCUAXvVTNV7oQdeMHSPOxXnn91Dspbc5zE+AEBAM8/WoiVTWl9VVad/na+InErcvHfRa+isRO6wFJLUtBtjmqY5Ozt7/vz5kydPLi4uxnF88+bNycnJs2fPttvtZrO5vLzkRfBN++I0HD+O3Vdrvdbv1+z96hOM43h7e6vUNAuCktfVNI6j7gAaYygc79GFGETF9lqnuCKWzEIiCGiQGAkEcgz727urt3fX18Ph4MrO1xvftGhd3+8Pu5txd1vVpt222+dn5aa2pSchyYCkCWLJIYUphTGnkHNaaEp5fXTqgC21d+GZNX0mZ8kI7D0Vrtx2/vysOTttvcXXxyuFASPmMceFBYxTBGZEATQ6zZBFIocxgEzkxVgjlhCNEWfRO2MtGQKKaZzi2FvaF+6kcLmwiGCdKZyrCzeMccwsRgySBXJoxWj4xwBoyTnjC+sM2iw5URYQA2RwqW4vtu+X5tXjWvvaiqj/su6pAApzTQyZkQWZCcko2lzmRZEzS+YskhEzKYMjIi0sqrN3mFLOSZ0pXNKzeJ+NB2bIC2zYWYPegbDCdpaFNq/zOViX++V2PIgIwFZVY41v25PzaRxDnGIMCrXXC8w55kyat4H5PPRfZ+2ma8vCx5R0CQzDGGIMITBrJTETicSpsFCftFVz0pyeQ9nsAx+GqR+GrvTWV74sXFE6p0rh1iHN8rzvPopfGLxAwxUj03Vd13V1XddVXRZVWZTWWu0jGqdJhHWVOu+2p6cvP/vs+fPnTdsaoimE/eFwd3eXmTebDRGt9TON6XWD6Lpus9m0baurdxzHy8vLN2/eKGfFMAwA8OTJk1evXr148WK73f5VIdejMT9wMs6Vlas36KPow0gxB/QenDGkwlZde3q6HfsDcBbJYRqtKwrvOd9TZC8TgXPWzcso2l/rkQqJr6pKZe0058nMMURCcs41VVmV3jmHzkWAJOCsdUXpi6IsnFVWnL8ydBJl5aLPOWdmADGGvPdN3WxONqWzkuMOxBhS0w5FoWEVIq7rTIV3tQ4mWS9SYyZNt+bMTEBrkzAAAhGIZMEp5kM/3t7tbu52u8NhO55UZSUOj1zgB6uDZlV0IsLFxZ7X/qHvX79+/f0PP7y9vEwpnZ+fkzWHYej3+3/5l3/58ccfX799M4SQAci4KQajnSfGuKKQRSZ8hq6KQAiGDRAWVeGd80VhvQPAzDnEOI6akR/LWDjrFKFugOR9nuICDTi+9cJZxcWysGb0CYAZ8B5Ht8B5YCHfY0FanBukheJ5hvcu7tPsSB21aC0hO79/t9HY3xhbWFcVZVuUbVG0zlfWFoQO36mvH481K65taaueGyyKNV3XqRv97bffHg6HN2/enJ2dffnll1VVPX/+/Pr6+ocfflihcMoRyYvk4/Ht2u/3P/300/X1tZLahsV0KdmD7huajV+LAnlhjPDev379+u3bt69evXpv1A4AwnPsrrgipHQPmWZhgRzTdDhcvf755798u7u9DuPUnDxpT2PJEYy9vb7q93cOc123m7PTzdmZr2uyziRgZo45DKHfj8MQwhi1CMMzJH4hwVnQj8wzya2oUigoRC9zDIakKuqusU/PN+fnm6JywA+E+AxYA8jEgqpSDigsMYc0ZiC0pSUP3oPkOJkYJXIyDjyV1pRV2TnvDJKgZE5TSMMw7IF3zg5NldtaQIio9LYpin2abA4xpCSYxswRSKxHAw4RkJSiUwwKMAMk5pSBWVmNyZhZBnOZou9ith5H7etbly5EZhZEEKDMHBOEJC4CAxMjImWAlCEkiVlSBjW/CJkJiYRBTTvgDKLIOSfmvGYBEQAJYK7HKFpSS42auZVIiKLwzpmBB5bEmICsaf73zTVEIkfeu6Kqa5Ukn0IYQ1RSVoXmcQogeSkBzLUlQDCGqrJUEYWU8jiNxlhmNkQigIg5pSQpp9EZKqq6qGuqmgHtdR+mIUhi9Q+s98Z7Y52xVtFzZs4U/tJKf/cy5j1tdb21/7nwRV3XTd1UZTX0w7Ssw2kaWaRpmvPz8+fPnp2dnZVlKQDjOO53u9u7W2ds0zbCciwDta7hruuePn3atq3afgXGK/ccAKgm7JdffvnVV189e/ZsVYz4EOuO8yWogKEnXwJZpEwpGwDLTFr4RPDOlWXR1JU3iJL7UXVunLPeWmtMzJptW2YCIituV3MPd3d3utNpa980BeWWVwSyLClHa82saorAQmSMLYu6aeq6rApnzV9vX1icWs6Kv8i8NrNZa4vCV1VVF55ziGECmbUttbTpnDMx5kW1yDlXVaV3bkXJA0wh55zT7DawkEEW4JmSU3NlwiIx8TDFfT/s9odDP4xTmDnqcakMPzRFuDD5oKqjIbBAzGkYxzcXF99888133313fXNdV3VZVcbYw3B4/eb1n7/5848//Xx9ew3WFlXFgimxVhZkLQLlnGPKKepypZzIkvPWobNgFbDJDCHmnFLl3TRN+l6Nhu4d9vcCtR59R+ZNam54EhRmgIz33VQiiMIw0wUog9ES5yAREq+sl3PW/b4T4z4C1ALrEiLJ44YcTfCTI+N92RRlW1ZdUTbWVsZ4RIN/LUOnOTONs7WZbQ2adSK1bfvixYuXL19ut9u+729ubt68efP27dvNZnNycqJ+ucIUjtWl10esT0dhKHd3d/qd1a6v1v0YjrfG67on6KRdk3Y8U7Y+uqj54R03esx3jhcxt5xTiP1+d3tzNex3ObMpDqYvsmRB3N9epzDVJ5t2s23ajS9rEJMC50nimPp+OvTjbj+MQ4hT4qS1czXcs8XSlkWF/Cu+ToRnPmrOklOKo3fkXNV25clJ23VNBgkpHF9GSpwCxyAxSk45xRynGKZpjFHQFE3prDPWCDOREYEYk0UEIEvO2cIaBwCcU8oSpjT04QByKIZxDCmxiBCid7YufRXsIWDOAkkgASR1P0XmpQDMnDkDYNbu75xFhHBm257v/zIV8Z1l/g6MbslwzyyzLMyChFkgMoYEw8RC0QRBw4jIgJlhSjwGiYwMqMQzKJl5aX4DTYXJQv4AS21PBAR42coAAOn+VGHJVSprHD+gv182gffU2o/W3Dz5EMkYKsk466tSSYs0f6ViQFPO0+z+K7JvpuCcUwrGWGOt90Vd19M0aWvQMI7jkMYhJkNkWTJMkW5Ten03WsmnRVH4Qpm2gAjIEBlDZOndbuJfG4sRsmsMrQB19crbtj05Obm7u9NapoazROSd05Td+fl517bW2hRjfzjs7na7u7umbrquY2QAUBosXfOKo1HJV6WmOhwO2vCmaJ26rl++fLnZbL7++us//OEP6jR8BF+NmiOcW/sZ5ogdEAyRNSZrEApzX5YhqrquKvybi8urm5ucEhZgjSm8nyTEuDa5C4AobHCdG2uhoW07a8dpCgCgcYzC/heFP1ZL6Zwvnd121ZPzzXbTlIW35oNkXef4jnlN5S5PbebTtc4CgfeOM+o5VHUtzIe+n2LMPJM/V1V5crKdpsF565y1xvSEnPM0cpimmDILGAEBUhyrfhQLZAFgiSkPU+zHqR/HcQoaNN8LPT68kiUhTUTEis6QfOj7129e//nP3/zP//n/fveXv+wPh7brmq4lZ3b7/dvLy9cXF9e3N4dhQGOmFM2hJ1Re1vuC7rzElZUEGAkUNTqr+pEBMMyQsnRNUzkb4kwzvPLIqrP+S1H78ZUIwEJhRzLv7xkYNPl+X3JFFGEUQCbEvICsYM7JL8zbMkN4SO63QJm7rO4NvKztCvcbjiCAMbayRVc3m6runK+sKxDtfX39V2eSusva1x5COBwO2pi6tplVVfX5559/9dVXGqNfX1//9NNP/+t//a9Xr16dnZ21bdt13d3dnWLlFPKmtlxTcWt6/9jSqyu/vtDX6zjucEPENSd/OBxWWZr3rAWcKQqAhWOSlGcwOgtn0NQ+gKrqFLgxQmhsmVKcbq9zSikFX/jN2enJk2feNxxp3MeRY+rT1IdDPw3DNIxTCDGnzIklM8CM1wIAYc5KosRJmEUyiQCw5ChZ9WVTTsFiURa265qqrsi6YRqHKR5f0X6Yxpshp5BCSGGKYZzGIYZJEG1R17bwZc1CMWUhk2ePUnNBwJJTQmFJ6g+McRrThDSVOUwcI6MIonhLlXe1c5WxOXMSITAEmhfPMaWck2QmJG1n5byq/cJcy11gp3M32VIyOh7vi9qXBIrmA5WQEREhw5SQhjRFUSJsMgatRbKJIWRhANH+TmGcy1uCCDxTvMtRbnB26+Z9al66hGZt8J0XD2cWzinPbJEPTnU5iDDPXaWPx9HOhmDQGDIiVpzk7KPzk3WT9SG6GGzMU05ReYdFkRdqiRDRkDdGrYLmqaZpMtYREijjm2+Y/JThZohX+6H1dFaX1jlnnTEWNUxVmDzgB1mM4ydkbVEUWorTipfC35xz5+fnyl2l3r1qSFhrvPcnm01T1YiYUorTlFJWFRxF0S8gkajkVoqes9aWZVnXtSLjxnHU+OD29jaEoGy1RVE8ffr0D3/4w2effdZ13cdk40FTFasuyYyYFCEUQVBuA9YWSH0nYFWVTV31w7Db70FU7Ya8LzRjrZA6nG15EmZAVLC9NcYaUxRF2zbGmH4YZtLl2aThPBVRrEFDpq7LrmnPTzenJ5u6Kp2z9D609qOhcy9lXY15zpQvn2KssdZ458CQtW7W7clsjDXeVFUdU0YK1nrNz1trYzSyLJHMklIOMY1TiCmzCCUWpCmEmNJ9kZEFEFLK0xT6Yej7YRzHqFQz9B4buZ7ebOARU4yHvr+4uPjm22//95//93d/+cvV1ZX6jiLS9/3d7d3rizfX19f7wz6kKESYgrJOPNhO9MMEEWAB4wgSoEUyswIAKO93lnS63bZNihG1LWUJRR6l3vHRU3h4NXOrxNpzrkoyuvHcI3IEkER0H9OWarX4gEz3MLo54KR1a4K5y2r+exytL1sYAhAZi1TYYuPqbVl3RdkY64jMu01uvzScc2VZal+7ms8Vzapv8N5rCeyrr75SP3u323333Xe66plZ94fVbGubu9I0HRvs+aYdpa5hgdGtr2cLwHw8STRsUOqnldDm0WCRMcSQmTOnmMIwTocxjcyZMpucRfOkzEzGFHVdGmu8TzGHKcZxSCH4wrVd127PynbLbIddBIwp5tjHcQj9EMZpZocXzrBC5hjh/qyzSNambRBGzsKJcyTgtq2qqmb2ZVlUdUXGhpxhDLt+HA7j8RWFlA7TxKqMPQ5hHDhHQiqbpm63TXfqiibEhDQwcOQkKRDJEEYbCjHWkBWWpKROkUWQGVOSGLO2U4EIITpDhTWFtVPKBsQisdFHwCr2klWPDmc5sXWx831qCe6pHgTgnYfy3lr7HIiIhszAiv8UJImQJeY07A+HxOx84auqrFuyjoFkYXsjZERBzLqaF4d8WXNLvoYXyovMbK1VZb6FzmPOt/NSb2RmxNU9EBFhkIUYcvYhfn0cPz+lZV1qsSHGaQrDNA3jNIVpyjP5CeDCpjtDv4zR9LX3RVHVXXcSpigAxrke7D7xYZoO0+iMZ0JtsbJkCY0jcnPfnnzggtehmduqqpQdTyWetDL37Nmzk5MT770a46Zpbm5uDodDTJHI1HWdc756e0lI8XTrnLfGbLrOWbuKQqqMiiqkAcCKitd83X6/f/Pmzc8//6z4WM0QnJ+fP3/+/NWrV+fn50r69jGmXQDXoudsV5iQBJQ6EhGQSJlH9f3W2LL0m67r+6Gf4jj0VdOVVZFyCjGkFFNKAsI5xxBSjCknzkyERVFtNhsFJRhjQ0zTOGooT7PGDFlLztu6LNqm3m4225NN27Z1XRlr5BdcxYcXA8ycMocQxnGaNIkqrFhvY0h9LF94yckYE7STe5oysy/Kum2BjJ+CCJAx0xRubm73u93l5WXfH4Zh6A+Hw74/DOMwhRhTygwUWWCv2LO4sMYIAEjiPE3T0A+Hw2HohxgC5yxEs5T7O5NqDdwBMaXh7duLb7/99p//+Z//8pe/KGGRUjHe3t7u97v9fv/69eu7w26cxsRZO1RjyHHmvoTj7OxCXAswg5IZ1kwcICoaigWFn5+f55y1CcsYuxqS4wP+mnclC+oNju26INCxddeAbmncxwUzp7VAWQy5dkXhDJ9EFND6fV5Mu6w5+zkQFwQkNMa6yhWtr099tXW+JOPxr9NVPBgaM6jfrD3lGrjnRf9N60ovX7784x//eDgcbm9vFf6iP1Whl+MGVKWVnZ/v0T1crfV6/x57Tsub8eGQhZZOk/bvN+0st7tDFghjDOM07A93l7fjbiQqnKuzYEwpTUOIExqqmrZsG181h/2wv93FEEBgsz09ffqs2mzBFn2fUt7nlHPkqPFzTFMIIYbMCSQ7C84aEE45x1mPgFUiXDtwgFMOU4qTSKrq4tmzF19+8UoABNEWVR8yX97h3TCEEKeRj9GaCEwAllAoMUfmoijatnv2/LPN6TOkJiRIu9ssEnOa4phCHyOKSEi5SWzJo6bCgAHQu8JaB0ApcZiipcQpCTMCGDLeOmfQUc4GLKJ1rjQkAGoaJTMiZM7TFDKA5Jwlx5wxxsyzKvGsTDCf+oMn8rj5bQnb5yTjXIHOmodCFkmZwxRubvsQo/WhbKUD7wqF5ih9MjPmlTRecEFPaYPbrOm0oBgXpV5EpSBbCwYIc8KNc0opJ+CsTJawhOrzlNKo/UN43Zax7iBKCsFsU3LOWeecs+NofYxzres+O6c4OwBdG0alHD3nikUECTHBYeLG00njNpVrSlc4a9BYNAaNQ+MIldt/2W8+dCiZqxqq3W6nERUinp6ePnv2TLN5RVE0TXN5ebnb7ZQ20lpblGWYptubm5xSXVXK3rrpuimEYRi0sTXnfDgcdCNQ/teVoVoV3q6vr5m5bdunT58+e/bs+fPnT58+1TTgR4bs97f/GGhAC3Jxhi+uHROAsBQXurYdpzFf3+4Po3BGUmYeM448hSmnpLkfnUmIWBZV3dR10xRlaYw1JjtnY6QYI4AYY6w1hfdl5euq7Npm07Wbru2axqu25i9d0zo3BQRB18IUQj+Oh34YNbhWmAggETljvbPOOUZAIgbhxMMUdochC45TDDGHmFNKuR9vZQcA+93+5uZ6HMcphHHW8pzGaVJQk/K4D1MYpinEmDjzWgrOWaX8DvvD4XCYxinF5I0BhXY81LZac99KgpdS0hzv999///r1a+XbF5FhGH7++eeU0m53p2RkKSdAIDJIhpljygvZDty3ps2wEJhL1LDktGebYTTJHcPEOYkILQCSNWT/iKkkDMIoR8E7AiADHEnJ6UnN/XDz0pu3GT3JZWiD3LxJCQMoG6bMDcQznb/6KSikO3Phy7aoOl92rqjJOG3d/KgVrmk5lVRHRH2Ox1G7tsadnZ19/fXX+/1eeavevHlzeXmpN+44ktZ4SdPmx8mPJV9FD10oOH5x7AQ8esG85oTfzy+UM19d7xFJWHJkFIjTdHv1VjI6WzFAyjmOfRh2/WGfOVvnm65D8gIWgHKKm9PzbvvEly2DHYY4jolTVkBUTCmmFFMIMRqHdeU2m2rT1ZLzeOhvb3fxelAZ4jlez0lyzGHkHG1hnCvOzreffflZzDCGNEx5CDzc9hkx5cwplHK0QhTxpUGlkriQcUVRt5um26ZcpCGwQOKccow5hhgiAIth9ORSYQyymgogNGgLJMwsMaYwBaHEWTGKrG2BIMDCMecpA1lyxpZl6a2dWfqYY1LA6TD1g5kCxWy9XxQuliUjj9c4PDbtS7Cu4BSVg1GCOEHgDCxIBDFjTDIFDhzZBFtERmesQQMkjKqvuai9LRBUJFTEKvBS8xcWAFQmNW18Z2bI+vO5cIcy3woEIUI5Womiq5eZhemv5U6Pp+mjoeG7MeSLoqrqtT1U1bwzq43nORpZYh0EsITeaUzPZBCsrTydd6729qwqWm8NkAFyRH6O2u97aD9w8eMiynR6enp2dqYAV+V/1vLb2dlZ0zR1XW+327dv397d3g7jyDkjETPHEMZh6A99VZVd13WbTdd11jnnnDrgyg5NRLqzdF2n9TkVZb+6uur7Xu36F1988fLly6dPnyrlxUdptB9dzfwQFqQxLIX1WTx24STUzsTZl2qaKvN2nMIwjJoaJIPOOWGexmEcRxEui9J57wGdc0VZllVVFCWSSTmHe5LB7KwpSt81Tde1J5t20zVNU9dV6Zxzc1/Q/em+M6UWEJXMhd4pxsMw7Pf9bn/o+3GaQkyJs1YZwBpy1lhjImdBYEEQGKd4cXXt7vYzgfcYxmkaRw1I0tAPfT+EGPScp7n/JiXO6jVk4ZRyzDktgqN6Wpw5xTiN42G/3+93fT+EEArvzaJl+Z5HsXxJOe3udldXVzfX14fDQfPwGvldXV3FGMZx0BtoyDivcG4ngsxwjzs9ArYsX0ApQOeOG5ixf8rz7JwxWhzCB20y7503R2v+wbeRM+ran/OTmn6ft7rlES7JyjlwpzV5uYakOpYG6KxhBWp6AQR1D9aiPgAAClokZ4ta7XpRttbVxtr5Xnykv6uMC5p4W2nk11o7Lnx8bdt+9dVXioFHxL7vD4fD69evFYij3jkv9FPrfTu24o9u6SPDfxzEH/+UFkZ6EVkb7uUdPETOfHm5M9Z5Y0VMVbdEb/e72/5uxyFpPjmFKU5jjAEJ625DZOp2Y3xXVJuccrfpyrojU6QsKUWYaQuyzu0Qp5hC4lQ17bMX559//uzzl+eS0+31zQ/f/yQSry6noZ/SNHGYOI4SJ5BsHdbtyWbbdqcn1ck2DSHmccI4xhjHSSEpyKk4ytHNti7FFCZOWUQpm3I/BTdMjBSScjUGpd3JIolFQvYRRTxisU4YMGSMEchTTOMUpmkSYs5xmtIwpSnmlDimNIWw64dDzJSmStqiKqumrovKGuLMMcVxmsZxDP0wDtMwRbGFKUoyFlSUGVau1wfj3Vr7ghBlbXFPolh4PCZAAAASwJw4hhRDNDYhGgIUAUBhJEJkRCLgebaLINLKGoFLug7JGEuqxogEIsJZlt5EXWBzNAZiDK2YS2Zc4fHM8ojN4iOHBuJkjLXGeZed9c4Vzk0hTDGHmGLKUSu7ayFKvREEIBBE9oZag87ZTqw3pnGmMERIFo0n44ksIt3rOX7o4kdELYErqZzKQszE3cyqwl7X9RrZ7/f7aRhTTogYQzjs+91+t9/vU8qH/UGPqVZZKdtW6RSt9q07i4bsGtxvt1sF6L548WKlsvp4u/5oiGIl1njCKNAaHzufCphvm7prm2GaYpIwjdYYVeBk5hgDAJjGVnVtrfPOF4U31gJiTJEzhzAhQFk4Kos1TO+6pmubtqlmup4P347VZLCklHd3uzdv3r65uLi8vNrtduM4pZiEMyzkrSIww+dZBIDIZJb9vgeAQbPzYxjGYRiGcZjGMYQpTiHMPOdJVadzzHNFKqsbO/vdMCs2CwCAkr2OUzj0/f6gAgGhrln7SN7xe+egVRbH2xgzE+kgIswIal15yv6lmIzCF2VZAGJmyQwx3ktRrYkXuPclBAQFSGYDPxcHgRlInR7CY9K0h6vir/rqamtmhgxARJpxtg9m0IyomA+vBhrmJpfVjN3X+QDm+GP24GZ4HgLIXEs0SI5cQa72VVuUnStr4yuiNQ//rxnzEjAGj2Buj2Jo7/3JyckXX3wxjuMwDJeXl999952SVimdAy4g+fyQ5n3Nqz8qnz0Cuh9/3Grm5/116cRTNff3+grMvN+PxmZvnSVwvi6btmyaaRjCOGqLkxbbwRjrvS1KV1TG1YUYV7YpxLIoyHoWlKjM5Sw5MyfOOYUphDFzBoNl6c/OTs+fnJ6cblCy98SSWdg6B0D7mzyMQ0qJhLfb7uzJdvv09OlnL3zTDln6yH3kmCExhsgxJgEhTscuI4vknFPMMSadtTGmQz9c394m8L7klCFwTJIZhax1ZSWMBAWagqgwtrIGEUSQ0QpaJg6ZwzDl27vBQuQ07vvdXX/oQzjE2E9xnMIU4hhiTnlKbMlqR0dVlIhonauM9a7gog51GKc4MUbjDNE6ZY8yivfjsfLbwpWhyJEMHBkBWBDNLOIi2htmLFHMDDlzjBIjGquE+VpCVeii4qOWyEx4mRRLdgwAjbGEArO3q9HQ7BcqjEV3mEQEAkb3CFYmsntELgv9LaZ9ntIwu6hUlsa5oiqrmGKI0xTHaRqnMClRKDMDIJLKDAiBEDMjGpKSTIlokbyCh5AsUWGMoxks+wh39CGnpR3PKymV6i8pCywd6alry7t2EwkzwkwPeX1zfXFxcXNzs9vtrq+u9rtd3baqM1EURdd1AFAUvm07hcrnnPf7/cXFxcXFRc55s9k8f/78888/V4qMoig+mFb2PWPGiM0xk+6UQEQIRsTOPZfKM7EQJoCAssSfnJxkgbeXV7v9garaEDqVUCMCIF/WTXvii8I7Z43JnKeZeSOCcOFNVdRt22y6brNpm7quysLP5uVjLgbVUAozxxAvL6//5V+++/4v379+/fputx+HMccw+7Gaw0xpChiDyguhMRaRtHJ+OAzjOE7jNLd3TyFMc00xKmto5sSSsmQGFmRZdSppTXksXLSirWhTiId+3B/6wzAM09TlzCK0Ejw+fAqyFJ2VzfTJkydnZ2dTWDqvMguIMaae4dtlUZZVVXrv+364ubkNIRsT12B5tur3kbfcI9QAlSJ3DZ0RxVlrjKGFj09UJ+cBlhwfnvOSdT8aK/obVYpC7u2TLJudzBG8fv58DwFwuYHHB5xn53Jz59PAZRqKlllc6arOV50rO1vUZDySA1wxRR89ZEEdzbpTi2jho7dpQX273f793/993/dv375dNd80ntbYfZXA0COoOX83Dy8Px5pmf/TOVV5WAQGK+HkvwkZEQoyYJGAuC18UZbt9+vzLUFTN3eWV4u9yzsJChnxRdKdPqnZjiwbQGj+GYVKdlxyzwAxx5xw5x5xiDNM0DEBUtnVdFZtN55zvx8ESuMI/ff60aru6O3FF9bOhFCNiLlz5h//n63/4xz9UJ52ra/TlxfXdvo/DlAwYwCXXI/JQTxI4S5zV31kABWic0pQPQS76kE/O2NgicmAQIOOKsqprBCPZlkVrXOl9U9sCETIkIRbDkgYJ42Hii6s95iGGw93+bnfYBeYIMGYedI0zDuOw3w9TP+7udoezs67rCusUr1MWhStqrnic4iGmXWRCApaV5V0gP1KnfUfUdaYdMNYYZzWRxYgJQEC154FQxBi0lgQYgYGT5MCRUPJC0kxr0XRBTQEsZTiYy6pr1Xmx96urPMNRWZdlzjGlaAwtGYWs4sIAgItXSfLw+Xz0wCVs0CQDEclMsuycS967ooiT8jzMOogpAIB1xhqwOJNNEQghGqXIRXTGqPk4cnQ/evFr4K45+c8++0xXr8boGj3rV63YpZRk6XvOOYVpartWc32Xb9/u9vtpGvvDQcVw6qYGgbquy7JQR2GapmHor66ur66uhmGoqurp06daX99sNh/b6vbOmPd2jfCEOcWgXYUilhA5pTFF5gxsmTMs0wIRyZi6rjLz0A9qQXVTt9Y45wDJWqsqS4AoCpHlbBGcd97bqizbpu7atmnqpq4K75yzj7Y5+OWSzfueipaZ5h5RZXUWZmAmMgbREs2PfbY3aIzyhHsATMDCq+D6HJUv9Cuw/uVZQEyn5VLmEsY197zqWjFk5hjTMI6HQ78/HIZxTDkpOuw9bu+SyQcB6+zZ6dkXr74YhuFkszn0vTqIgGCtqeu669qmaRSuQYbevH4DgDHlGFO+tyKyhOyLaT8KJtbGGI1PEaAofFmW3jsknCWaZzgL3h/g1yeTSIj5sO9vry4JpSgKVxTWFaTF8hkTnGEGy833aUHQPf4AnP+ve84885aTJkAksmicLWpXNL5qfdmRr8iVAAYWbr9/3VB05zAMfd/HGJUCcu0vf5RRL8vy7Ozsiy+++Prrry8vL3/88cdVEklB8sfSbXA0t9exGuY1jr/HDy0bnrUWlrK9ktLXdf3FF198/vnnJycnv0AsrcWdnGJkBjLO+Ko7fYJona/HcYwhLFxkZK2tuxNXNr6qyTjNk8RpijEqLTfzLPzDKaQU4jSOw+i8t1g5gwgyDMNuP1Sl67rWVVXnyvPEITNzRhCOY13aL77+/Vf/+PfofGDYj/EwxMMQQ8yFpQV2gXMh5mgyIBkih5gALRIYS8AihCGmw9C74eBLZmaYM9lKj0qzMIjuVVb1uo0YBiMZhVPej4dDOHDsU+gP/X4YBiAEa5iMVVYEwkFiDOEu7aYphBDb3b7wviqrtm66qu6K2gA5sg7BYNYs/IJpE7xvdZ3Ho4Q8gIhB45wrS0EyqmKkPf+aXhQgELYE2aAIGhKCDDlyBGCj0lWL/b7PuOK6mpaKNdwv4mV7mvea+0U418kk55wAnIibLXnm5YCzw2k+oNb+1wY+XO1KcYHkjbO+LCplD+mH4ZZvhj7sbw9JclmXReEKZ9RaAWnmCgXQGOOsWRKPq/dy/HEfdlqIiqPZbDavXr0qyzKl5L0/OztbC964oPeddfermjnXdVlXbdeebE9OT7dvLi4uLi52u93l5dvd7q5pmq7r2m7jfeGcFeFh6C8uLn7++bXCpLfb7cuXLzVe/5vtul4LLJVVRX6OJZal90RkEMdxiNOUcmLn10Ij6KxBKAoPgMOmD+N4t9vtD31OSVMagAYQU0oIwCkRAgJbAl8UdV11bdN1bVVVZVE4a4wxy078kXNmTT/hHEK1bXN+dnZ2un3btTHGOEVGQgRLxilLkbGIBEBExlkoy8r5QlhE0NpgKUQV+EVa4FczIGZ2NNTHZF1Hq10nRCbQ1TJnCBABBDhzDGEcx/7QD8MQU8rCCO8UOdZrAWAQ79yTJ+dkTdM0V9fXh8N+GMYQAiI6Z5umOTnpVLrXGMs5/7n6ZhqnGGJSBllRBW2Zz37Jxc9pAW0hg0VcE2dF46apu7auqlKl62fs2/xQ7jeLNbSSd2wUC4yJL2/uvvuX7xGk67rt6fn2rPLkDJFIlpmKCmaE/LL3vPfZak6BlyLRunsJAKJBNMaVtqx91ZZVa4vauBLIAVpZsHn/6oWhJbDD4aAJOQDQ1jUlmVnt6FrhVprn3/3ud99///2f/vQnbWdfS/WKwjtWfF9/V8OAVbhdqXKUNlsh+lrdU1guLNW6vu+NMZvN5j/+x//49ddfP3ny5L2mHRGctWlK4zDlJMZYg1BULZGr6i7GmHJSLmbVVHTOG1e4ojDOJU4xhxA4xiA5c06cYs4T5ymlKcUpTFMYRoLWAAPnceiHab/b35xsN8aXvjCMWDT1kxdPEKAqC+BU1/7F7748ef58DGk8TBkkc04ZUhJLbBZEyFwaPhrGOl9UnCRFRjAiUlgDhhKzMIcwAqKwGLIGrWSc4kRAziDblDlmDlkcgRGF2VsAMZmwn6b+5i7HPqcph5AzO+uccd4U3jhMSSSEyCnLFEII+3GcrtyNs64uy5O2O+9OZHNa+lKAtNeblgwLmZkY6ddMOwgIAyE56wDIGl4IepEXS4uo6WdGEgpCxhTOOEfGLDRES7/yKhepR76vCyzLd7Hoy8HvO0eXzNj8XgYAmtUc5lKUVtoBYKYLeS8f3UePx8tz9kmMMWDFcDYOwaSQxyEQ7PI0DTLklHLhrGMkQ9aQMcYZJLLGeGeteZS++tfsABq4V1WlLWe64FWZkY5ooh/bXTEGZF3DVVWVZVWUxeXbt1dXVyFM+/0+a0okTCFURCal9ObNm7dvL0Vku90+e/bs5cuX2hF3TFr5tw2Va0NrSAME9RgMkbPGF44SWuVgfgBMQCL03p10neSsDx/HCXAmPDaEqm1kEAtvC1cUzlZlUTeVNgd6BW7hQvJwj8n6qCeyxHEI1prNpnv58nnfDznlsixf24ub27sYE4jEEA/7w83NLSJMIdzd7kQE0RSJWcXL+74fxnGY4XSTpoRSjimnlOJKJM689KPOHJqgvCBLO7i1xlnjnakK1zXV+enm/PSkbarCO5qjyV9xx7QNwVRVdU5UFsWzp09nTF9KiGCsqUpNw3rnHADO/KMpbTab65ubnLKWJ0RLY/MxF4Yq7QuHhckE7nEVT86ffPnlF+fnZ0XpjxLA96t4ffVuano+dSJX1OQqMD7nHJmEvPWNLwtjiDkxx8yRObJEkazB2YNg4oFvd38FSOr2aFxkyDprC1s2rmp92biyNsajsQBGYFl9HzOHHg11E9XKisgwDBcXFz/++KNqLym8Bo5K4Iio4BsFvmih/fz8/LPPPnvz5o2i5bUGrxAcHSoauxpyfaHmvCxLNepq11d5WRWwVsBNXddffvnlq1evuq57L84GAa21OMYwjmEYJExlWRTeG7LgSzLOLtQ4syCIQEy5ALGOytoz+2HHKQ05RE4xp5DTlPOUY0hxitM4jZM1yHGM43DY78TgOIViSsOYpiAhJSKsm9p99vTspOWcjMGq7cYs+ykdhjgFyVnxbSALnZPWPh6BJHLKIcz8j6oZbZ0DQs0UphAJDKHxxpW+TJMfxpE5MpicphiHyRpLSNaIQWbmyCkMcTzsx34/9pwjCijLCTprnPeuJOuQMoCZUh5zmoJ28coUokGKUzCCjSm4FvKGjEvIxGmJMgBg6dX91ahdQICQiNAYJx54VnbRFhZN72HO2Vk0BgiFjKmqwheF2nYAXGv7uPD3yuLE3xt3Of5AAIAs6zfm7JnM1p1V090aUu4rmWvdOL+fZw6Gv2FxfciYA+OiwLblENN+t48hhHHiFCSV1meyxjhnvfPWG7LWOmetNR+B0PqVQQubtLahq/P+V8reCAhorCFTWmvLqqrrujvpTjabpqkVAH9zc317e9M0TdPUgBRjur6+7vvh9PT05cuXn3/++fPnz5Wa5iNb2H/hhBY8kyEqvEdA5xwRapuld67r2pyy9rwBwNHmiwBAhF3XeO8U1XV9e3e3711iVuIgAkdYFXbTNl1Tt21dlYX33jhryCyBujw87McOAVDMJlpDJ5vOOmeNretKxW0J6e7uDli0GhpTFJAQwtD3iDRNwfuCWaYpHPb7YRi1DSOGEFIMMSVt9MzK0cRze+iyehZgFxKhIeu8KctSaw1tU3VNvd20p9vu6fnZ82dPTzZd4Z2Z2Q/feXBzSUy7VtAaU9d1VZYyf44wK68bGEPWGCQEQBZOMcGrz5u6+mr3ZX/ombMsijUL1eSc59R82xK4r59Iyl/fte2T8/Pz8/OyKpXQ5v7hLMg23UXgF6y7MbY+2Z6M49luH0Pw3lfdWdWdlYU3hjLHnEOMY0qjZGIOs1KILFxyR4dcenE5i6hl1zwKoUXy1ldF3fiqc2VrXGmMA5pp5tbt9G9ZGOq1b7fb09PTH3744ebm5vvvv++6brvdruqr73asrclzzZ8/f/78P/yH/9B1nYb7Gmo3TaMNqyrr3DRNVVX69ThGV9OuwboeUONyTSco9t45p0k+Tfu/5zIQnbMEkKYh9MMI0jTtyXZLZJmXTjJlVQspxZSFjY1V4w0VdeVI/K3JKfZpHFIMOcWcAueY45RjDNMYxzEYnPr9cNjvdne+Lo31AGYY0hSGQ3/ouurJ+clpd15aE0KYppCRLq93+8PU95GZctZ+L3Xh5koEACA+sB2qtZGmkFNyxlpjgEgQWMlqQ2LM1hlvbFPWHPsUDiln5MQ5hNgPBgHZZMMEUeKUQ45TnoZ+PAwpEoJzzprCKw2asc56YywZATJDzn2K1saYMygoUvvDYhIAa11ZVMb6nKLJc/noXvXgHeDpo+Y34CxEgoYICYhU+lAQAHBmdUDMOROJSM4pGWurqijLyhgDREdTXX1kWVfOPcZlte8LQpXl/ivAvMT1dBWwSsZYa73zRkmg1wuRFcz/m0Ttf2WoeS/LctNt8rNYN1UIIwOTMQKQRYAQBQnJWuus8sz+jUjy+XP1q5rzR4CXv/q7iIjOGWMMkXNWhRcVEn93d9f3fYzh7i5p7wERaR7+1atXWmJXZta/2a7r2QAhGABD4CwRFeoBaWgKIKrtZhCbunSOFh5TXBK6aK0jIgGwzldVvR3GkLMSp2jcX3jf1lVdVWVZeG9pPvNjuw4fXtD9tUvRfDXWz54+sdZaY+uq2m5OLi7eTjEAyM3NzX6/SzMJfCSku9s76xwIaudxjGkR4WCWLMwEYg0YMoWb6yyLCb4nj9Okqve+rMq21S6+rlNvpqm7pu66pmvbtm0K52ab/PjcBR7uBlo2RDM3o8G9CbkPDQBARJyxhKdVWT0NU0xxWaMLd4yIKK/04perpLTeclgOjoiF91VdVWV1DHpQVMA9iG05TVhrxkfPi8hU9eb0CQDYlBIRdd2m3Zw4ZwmBOWWOLk0xTilNKU85Z5n1LDPke/L5xUQjCQiDIC74V2tt6XxZlE1Zt7ZojKvQWECzNsj/FktiNu0vXrz48ssv1bRfXl5+8803Xdch4jRNz549q+v6mCj+9evXP/zww8XFxeFwUH6h8/Pz3//+95vNBgC0i1Wr8l999ZXKPajie7EMfzTUnCvf5dr7rvf9WKhmVXb/pQsREe0pH3d34/5urBuepqKsyThZOuNzzlHp6nNCgqqxTeuatrJUeo8iIcZDHAbOmXNMKeYQYghxHMI4ksjh7rY52TTj1tVVUVaCZrcbQ4wxZmnAGmra6qSpwhT3/Xh7mPr92PdxGCKo3ELOuMyqe9MOD7Zo51xVV8HYnKLCrhhBuzcZOOVokqYU2RB5V1ZlE41DJDQmCUdORhJy5sxTmsY4CifgjM76ppmjP2M8EQEwQCIrZASAjPUiNUBGY30JMCPIC+t8WRlfiLMBhWMYYoqZjcy7qW6oAObRfHxg2rXdAAQJROHwoKgXRKTZlyWinDNzijEaImtNVRR1XVnnUH9hQdAB3PehyKN/l/8tbcKaJbnvP1nQu0sCYKk7ziVJXpa/gDDD+0kU/k2GOrCbTVfXZYwhpqB45ilM/TTGnFjAkvPOO+vMh2mDfvg4XnUfdWRERGN8URhrrHN1U2tYoCrsu91OeaeNsScnJ2dn569evfrss880Ff+b2XUABVgaQkvorHFonHdFUYCItSZ7z5xB2BrctHXpvTFLjWf+dY0jbVPXZVFtT05yzjFn5jxfIpE1xs4F9TXKWU7+PVfxcfPm2M7pca2hbtOWZVGV1fnp2bOnT3/66ec3FxeXl5d3+7t+6IMCg0UQjbOWzJw/OJ6xMyDHGVyY5+duNGssrcVRrZBa5VYqq7Jp6pOTzfZkc3Ky6bSNz3s/U9BbY5aI/Z1U3cPqtay1sdV4rq4kECzwONQEpp5MXVdaMlOjeFxrAyHlmxDRrjOeKyDLs5hX7j28VqGG+PAE504KBc6v53z8vJBMUbbe15uTc5mBiioNsJTtZDYPKYU4h4qBU+QUcoqSIuQEnDQNgyCC2r2LSEjGGlf6oimrrqgaXzZkPJBTmdj16f8mQ9Pdr169ur6+/uabb3788UflkdXC+X6/V/JHLaXnnIdh+P777//5n//5m2++ubq60jaWJ0+eKMm8akBon+rLly+//vrr3//+9+fn513XrZb7aDu9f/Hg7gOo8VvL6us7f/k6hHMWZhJJw+Hm9U+9L/I41t1JUTVIRpZGUPV0Q4xAUjWm6/x2W1dlWRSGMOU4hHEnmRXqm0II4xTGIQyD5Ly/rtrtNsVIxhRVnRLc7XoUcarZos59WZCxUQj7HKPEKDGwZBbJIHkmv5xzmirsRsc4q7ptQeppGqdxSrP3nVISRgTCDDlyoCyIAITOFzWc5Jw4syAImkyUCQEhcZ5SCiEigbOu8L5qyRnrrNPOkJxTTpnn+0oA4o1pnTdFHVNUm5hTUml5V9XZmgOnYQgh5iBYCupWOqsVoIGHPsqjhDxz5pkNVkXbUNttQBN3DzYJjR1hoYBGQlJaS1pVnAHgPRAegCOfd13EvHSezNmyZS0/ctaPHIL5jOeGvV+ec7/lwGUojShz1tpjiLEOIaSUOfuiKH1lrf1ovvgPPod/3S8SIYAty1KzrGVZNk2z2WyUT5RZjLEn2+35+RPN4P22dh2WRGJdV+dnp23bCJB11i/K8cophyCGsK6qun5vgV8z84YInLNrQQaWsHAdH3I+H+sQyqPZjIgg1hhT4Ol2472r6vL09OTZ2/Orq6ubu9vDfh9iSJmXQrPKkeNs11QqZ05RIxGqjruxxhqNkexs442x87/GWmud9YUvy6Kpa02vlkXhFo+G6LE5f3QvZlyL1u11sR2DQe4RZ4D37Wgruk09fDMTAyPiIwS+0Fprn6uAcE8IsnbrAdyn8kRWBPvqdiznqd97EMDfnyeRQWONdfBgUehkEBJmcmSyccnqMuXIOeYccow5BUkhxymHMUxDGPuUInNGY6wrqtaVbVU2XVF2zldkPJJdaeF/21Wt0cL5+fnvfve7f/iHf7i5ufn222/3+/33338/TdPl5eW3336rik1KOzEMw5s3b7777ru//OUv4zhut1t1xJ8+fUpE2iX7j//4j3Vdq41/9uyZdrq+a7/XEzi++etP8d1Zgb9YH9EHTkRK6SiZ97c3YRzqtqu7jXUe0ehqTTmlmEIMgFAUUjfu7Om2qOui9kXp7lKY9jvJnFWGLsYYQo5RUs4hTf04HoY4Bo4MginncRpL7+vae2eY8ziFu/0QYj4cQj+EMCVOAgwpJs7RGDCAgiyIgAYRBQjwgUV89vxz/2p72B92+7v9bnfo92EaQ4zMGVQ4hdAaQwi2KIpc5ZyEMzNnANFeHecExOZkyrJIDRFaa4xOVjKGNPMxayDP+DVCADA5FSm1Og/ncJcNoXeuLEsoqwzIjGjFo7VFhcYBmiV2f9ym8Q4bXRYUnlNnauHn9KUIiOYuZDZomqXA4z4gAARtvkUEbVdfEStHW8YRVhXv/8F5A9F+BIHjv/f1Rli+zlOReZG5/D839K4YIiKjCfKqFBZOmSMnQlJ5kr+lJeZ4/IZXh6j8+YU21HVdd35+vt/v+/7AIkS2bWc9eLXr/7oTeO/79ZvW2rZpCu9TziJAhpy1CMhwz6+mJ7kS4/zKp2tgsXJ7/ttPA1yUkY+3eAHEsvTe27YpnzzZvvr8+eFwOPT9OAxz1A6z+OHieBCiiqY8NO1Ea9RurDE0u83mSM6XZjlsNfPzXVpp+e/PVJabAo9tkVZesjBxnq0urcjwh+9cCmcIiArHh5kpcibrfpTH0DtzXyBf7fqSFjg+FXz4Mcs7ZPE3jjx7zd3Ju8/33hFYj7qmapDICBlnWHtrhCGzZGZFaQXJIYyH/vZm2h2uru6Gw06ErXNl0xnf+qKq6s75Bo0DMPDO1vkbDmPMycnJq1ev/umf/knFXf785z9fXV3d3t7++OOPTdO0bat1Ma1/7/d7ZZsXkfPz8z/84Q+vXr1SX/zp06dKWrcC4I+BMuu9eu9Kea9PLO8T4nt3KGqarC2qqqxq54vD3e3t1UVRlu3JxheVcauATc4pxxgB0Vmuaj+Oz9ttWzZl3ZbCadofZmngpGIwDAIEhEIx5DCGMIQ4xhQ4xZxyNM43XVkUVjjvD0M/pTDloY/7/RjHIFkIkHOOMWqvyiypRAYIl1rR/Xjx2Rcvnv/jzc311fXbt2/fXF9fDsMhhGmZvgjzVBCY2aBklWsRJCHERbNUFiA50VraojX/L0d/NGhm0RqWzPQZqHxoWo9U6Ip4L4JEYI0tyXkiS2hoETE8vpB3TXsGAMlIADJ74DJLuInSdYDifHJalACzcGaZe0YQhUS13Jf7Nkfnj0z50ccu1ynrKy3Ty+Ler9l5WQp48x5wxIz7VyffbznWFO/RlmjAGCOWLSwiKL/Zp/1mh5rXqaaANAz0RVHXdQiTCCCR9+UqNHfsqv8mQw+1atQuFS+N/O6nwGw4P/hz1/fhB8i1/Q3jPfmA5b+yWGVyzpVF0XWtcr8sGi735gvXaPc+fn841IivIf3Re+D4t4h+bY798k+mMPVjn5KzmmZb8G3HVzhf1xq8P0jYze7XsVt+P2bX5z740xw8zov7/ad1/9QWU74i8hQmpHdgZVB55/4/fn18F5AQBNAAAhEYEWuMFeeFE6GNUwLc9UO8ue1TCkVZbU2d2RhTGluSedDh9m80NHDvuu7v/u7v9Bq7rnv9+rVKLe/3+3EcVyobHcaYJ0+ebLfbP/7xj//pP/2nV69eKSAOAFTh6b1YnNXAv3uvfmm5fegyFMjMgkjeF03bbE/7w353d512tzGMviicn7luFra9BIB3mHxpDncvt0+6qi43m84SpRBzjJwFwFhj0WqWyBjrjSlSkN3Nvmx3rmiZkJCKwrdt3TbeW4yJhzEMQxz6OAzjNMZjCT5Z6M5kxndr9eUBQt4XddueZsbEEpMImnI8xBiWfIb69LL4sIJK37gUqGbDpYZpLuHhmsSew+CjctzxfrWoXs7rBedIXBCAgZMWPEQAiMDOxYeyaeq2LMp3mb8f1tpFMvOSRJvB7WraUQhZiIANaOchqyrHwrlBKvuCMFO0ofJtPNQ2fn9koFf46Ovc2zsX4u9Td6IdNceEuEeE5P/OQ1FO/95n8SsDj6Xn0BhHZKwty0IWIlC1UO/13z/qY37lm/cJwCPp62P01oqc+qhzeLBI3pdv/PAj/ML5w9KA/v5t8d5nsqYsi/vU8nsmJ8LjGHA+wjsn/eGb8IeOYRy9c0G5lGTxyYlg4TNZHavFvD7+1NnNgJV26tEF3u+k64/eidaPvZ17p/7Itms2XxHNyhDyHtP+cbcDEcHMpU1xCIxCoZjIllOA3X4ax6FqsOqEwaIpiLyik/Df2LTrqakuu4ba2+32T3/60/fff39zc6N96jpggd21bfvixYvf/e53//RP//THP/7x5cuXyhMHAI+2+HfX0W/orN9/igqPIqC1rmnas7Pd3TVemPHQj/3BOeeLwlpDZABnXTAQYI6uMPu76zg9rQq/2bTOWs45JwYhY7213lprjEMyaKyQSRnubvauui2rjatL60xZFHVTdV1VOLrb9bs+TCFNIU5TDCEY4wmN0rsICAsTEKziJoQkD24XgiHjnSvLommawAJlWacU1HirEgTiYtdnvfM5ZT1/a+4Zmdf/7KGDGvaFxhiWuvP8iABUAnneA+TItOu3OIsG0Twjy8ka8oWvyqIuy9o598j0PDDt9cn5GjeskcL9OtZogTBn9nVoQ0wp6nP03iOh5vUWp1+OS+1Hu8L7p8VysbBcMx79DI5Wuyz/kzV9JyK+bNejxRj/+3//779dH/b/ucHM4ziu//3xxx//y3/5L/82H/VwL37gGf1NRl3H69ev19fM/D/+x/+o6/oDTubB536saf+3GHd3d+vry8ur//bf/tu/+yn9K8bNzc1xJvbt9dX+cDBHQf/iVszLHQDuY4vZrsODRfyOsX/vWD/z6JffTdc9WPbrNiALAmD12xHhMPTrL9/d3f3X//pfP/geLMedT2IuFuQUp6G/vbm+uOn3AWO2eYQ3l4cIP9zsYlHWSAbgyB3+LYZCVtf//ulPf7q5uQEAEdFW8qurKxFRHNzJyYly12ioC4ucjLa2l2V5OBy+/fbbi4uLR9X0/wPjp59+Wl8Tp2r/s7DklEo3tKe04dMXze/iOAgzGVpb+AABYGYiMc5VbbuFQNdvEKST8PtXTzb0D5x4Dj2MITJERnHcAmp2q2YjHd0ZGTCj38Xxhx5LZ4mGcYrDBCHZkMuQbMyGDCK5lDInFUA1vJBEoUFCFMaj3unLq+9YOIRpmoZhHMI0ppyYMxzPVVntEshs2mepwSVSlePVo7qrKCAP6aPuZzvA7GscrwacFwHOmWsWWb4nKEwsJoS9sGe5Hsdyv784fjr4n//zf/7NHvWn8Wl8Gp/Gp/FpfBr/3uM36Lr+ND6NT+PT+DQ+jU/j/z/jk2n/ND6NT+PT+DQ+jf+rxifT/ml8Gp/Gp/FpfBr/V43/DyuWvUIKZW5kc3RyZWFtCmVuZG9iagozNSAwIG9iagoyNDczMDEKZW5kb2JqCjIgMCBvYmoKPDwgL0NvdW50IDEgL0tpZHMgWyAxMCAwIFIgXSAvVHlwZSAvUGFnZXMgPj4KZW5kb2JqCjM2IDAgb2JqCjw8IC9DcmVhdGlvbkRhdGUgKEQ6MjAyMDExMDkxMDQzMzArMDInMDAnKQovQ3JlYXRvciAoTWF0cGxvdGxpYiB2My4zLjIsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcpCi9Qcm9kdWNlciAoTWF0cGxvdGxpYiBwZGYgYmFja2VuZCB2My4zLjIpID4+CmVuZG9iagp4cmVmCjAgMzcKMDAwMDAwMDAwMCA2NTUzNSBmIAowMDAwMDAwMDE2IDAwMDAwIG4gCjAwMDAyNTQ2MTQgMDAwMDAgbiAKMDAwMDAwNjg1NiAwMDAwMCBuIAowMDAwMDA2ODg4IDAwMDAwIG4gCjAwMDAwMDY5ODcgMDAwMDAgbiAKMDAwMDAwNzAwOCAwMDAwMCBuIAowMDAwMDA3MDI5IDAwMDAwIG4gCjAwMDAwMDAwNjUgMDAwMDAgbiAKMDAwMDAwMDM5NiAwMDAwMCBuIAowMDAwMDAwMjA4IDAwMDAwIG4gCjAwMDAwMDA2NzUgMDAwMDAgbiAKMDAwMDAwNzA2MSAwMDAwMCBuIAowMDAwMDA1NTkyIDAwMDAwIG4gCjAwMDAwMDUzOTIgMDAwMDAgbiAKMDAwMDAwNDk5NiAwMDAwMCBuIAowMDAwMDA2NjQ1IDAwMDAwIG4gCjAwMDAwMDA2OTUgMDAwMDAgbiAKMDAwMDAwMDg1NSAwMDAwMCBuIAowMDAwMDAxMTYwIDAwMDAwIG4gCjAwMDAwMDEzMDYgMDAwMDAgbiAKMDAwMDAwMTQyNyAwMDAwMCBuIAowMDAwMDAxNzI3IDAwMDAwIG4gCjAwMDAwMDIxMDQgMDAwMDAgbiAKMDAwMDAwMjQyMiAwMDAwMCBuIAowMDAwMDAyNTM5IDAwMDAwIG4gCjAwMDAwMDI4NjcgMDAwMDAgbiAKMDAwMDAwMzEwMSAwMDAwMCBuIAowMDAwMDAzMzg4IDAwMDAwIG4gCjAwMDAwMDM1NDAgMDAwMDAgbiAKMDAwMDAwMzg0OSAwMDAwMCBuIAowMDAwMDA0MjU0IDAwMDAwIG4gCjAwMDAwMDQzNDMgMDAwMDAgbiAKMDAwMDAwNDUwMiAwMDAwMCBuIAowMDAwMDA0NzEzIDAwMDAwIG4gCjAwMDAyNTQ1OTEgMDAwMDAgbiAKMDAwMDI1NDY3NCAwMDAwMCBuIAp0cmFpbGVyCjw8IC9JbmZvIDM2IDAgUiAvUm9vdCAxIDAgUiAvU2l6ZSAzNyA+PgpzdGFydHhyZWYKMjU0ODMxCiUlRU9GCg==\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:30.487860\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "def visualize_exmp(indices, orig_dataset):\n", + " images = [orig_dataset[idx][0] for idx in indices.reshape(-1)]\n", + " images = torch.stack(images, dim=0)\n", + " images = images * TORCH_DATA_STD + TORCH_DATA_MEANS\n", + " \n", + " img_grid = torchvision.utils.make_grid(images, nrow=SET_SIZE, normalize=True, pad_value=0.5, padding=16)\n", + " img_grid = img_grid.permute(1, 2, 0)\n", + "\n", + " plt.figure(figsize=(12,8))\n", + " plt.title(\"Anomaly examples on CIFAR100\")\n", + " plt.imshow(img_grid)\n", + " plt.axis('off')\n", + " plt.show()\n", + " plt.close()\n", + "\n", + "_, indices, _ = next(iter(test_anom_loader))\n", + "visualize_exmp(indices[:4], test_set)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can already see that for some sets the task might be easier than for others. Difficulties can especially arise if the anomaly is in a different, but yet visually similar class (e.g. train vs bus, flour vs worm, etc.).\n", + "\n", + "After having prepared the data, we can look closer at the model. Here, we have a classification of the whole set. For the prediction to be permutation-equivariant, we will output one logit for each image. Over these logits, we apply a softmax and train the anomaly image to have the highest score/probability. This is a bit different than a standard classification layer as the softmax is applied over images, not over output classes in the classical sense. However, if we swap two images in their position, we effectively swap their position in the output softmax. Hence, the prediction is equivariant with respect to the input. We implement this idea below in the subclass of the Transformer Lightning module." + ] + }, + { + "cell_type": "code", + "execution_count": 33, + "metadata": {}, + "outputs": [], + "source": [ + "class AnomalyPredictor(TransformerPredictor):\n", + " \n", + " def _calculate_loss(self, batch, mode=\"train\"):\n", + " img_sets, _, labels = batch\n", + " preds = self.forward(img_sets, add_positional_encoding=False) # No positional encodings as it is a set, not a sequence!\n", + " preds = preds.squeeze(dim=-1) # Shape: [Batch_size, set_size]\n", + " loss = F.cross_entropy(preds, labels) # Softmax/CE over set dimension\n", + " acc = (preds.argmax(dim=-1) == labels).float().mean()\n", + " self.log(f\"{mode}_loss\", loss)\n", + " self.log(f\"{mode}_acc\", acc, on_step=False, on_epoch=True)\n", + " return loss, acc\n", + " \n", + " def training_step(self, batch, batch_idx):\n", + " loss, _ = self._calculate_loss(batch, mode=\"train\")\n", + " return loss\n", + " \n", + " def validation_step(self, batch, batch_idx):\n", + " _ = self._calculate_loss(batch, mode=\"val\")\n", + " \n", + " def test_step(self, batch, batch_idx):\n", + " _ = self._calculate_loss(batch, mode=\"test\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Finally, we write our train function below. It has the exact same structure as the reverse task one, hence not much of an explanation is needed here." + ] + }, + { + "cell_type": "code", + "execution_count": 34, + "metadata": {}, + "outputs": [], + "source": [ + "def train_anomaly(**kwargs):\n", + " # Create a PyTorch Lightning trainer with the generation callback\n", + " root_dir = os.path.join(CHECKPOINT_PATH, \"SetAnomalyTask\")\n", + " os.makedirs(root_dir, exist_ok=True)\n", + " trainer = pl.Trainer(default_root_dir=root_dir, \n", + " callbacks=[ModelCheckpoint(save_weights_only=True, mode=\"max\", monitor=\"val_acc\")],\n", + " accelerator=\"gpu\" if str(device).startswith(\"cuda\") else \"cpu\",\n", + " devices=1,\n", + " max_epochs=100,\n", + " gradient_clip_val=2)\n", + " trainer.logger._default_hp_metric = None # Optional logging argument that we don't need\n", + " \n", + " # Check whether pretrained model exists. If yes, load it and skip training\n", + " pretrained_filename = os.path.join(CHECKPOINT_PATH, \"SetAnomalyTask.ckpt\")\n", + " if os.path.isfile(pretrained_filename):\n", + " print(\"Found pretrained model, loading...\")\n", + " model = AnomalyPredictor.load_from_checkpoint(pretrained_filename)\n", + " else:\n", + " model = AnomalyPredictor(max_iters=trainer.max_epochs*len(train_anom_loader), **kwargs)\n", + " trainer.fit(model, train_anom_loader, val_anom_loader)\n", + " model = AnomalyPredictor.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)\n", + " \n", + " # Test best model on validation and test set\n", + " train_result = trainer.test(model, train_anom_loader, verbose=False)\n", + " val_result = trainer.test(model, val_anom_loader, verbose=False)\n", + " test_result = trainer.test(model, test_anom_loader, verbose=False)\n", + " result = {\"test_acc\": test_result[0][\"test_acc\"], \"val_acc\": val_result[0][\"test_acc\"], \"train_acc\": train_result[0][\"test_acc\"]}\n", + " \n", + " model = model.to(device)\n", + " return model, result" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Let's finally train our model. We will use 4 layers with 4 attention heads each. The hidden dimensionality of the model is 256, and we use a dropout of 0.1 throughout the model for good regularization. Note that we also apply the dropout on the input features, as this makes the model more robust against image noise and generalizes better. Again, we use warmup to slowly start our model training. " + ] + }, + { + "cell_type": "code", + "execution_count": 35, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "GPU available: True, used: True\n", + "WARNING: Logging before flag parsing goes to stderr.\n", + "I1109 10:43:31.036801 139648634296128 distributed.py:49] GPU available: True, used: True\n", + "TPU available: False, using: 0 TPU cores\n", + "I1109 10:43:31.038146 139648634296128 distributed.py:49] TPU available: False, using: 0 TPU cores\n", + "LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]\n", + "I1109 10:43:31.039162 139648634296128 accelerator_connector.py:385] LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]\n" + ] + } + ], + "source": [ + "anomaly_model, anomaly_result = train_anomaly(input_dim=train_anom_dataset.img_feats.shape[-1],\n", + " model_dim=256,\n", + " num_heads=4,\n", + " num_classes=1,\n", + " num_layers=4,\n", + " dropout=0.1,\n", + " input_dropout=0.1,\n", + " lr=5e-4,\n", + " warmup=100)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "We can print the achieved accuracy below." + ] + }, + { + "cell_type": "code", + "execution_count": 36, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Train accuracy: 97.77%\n", + "Val accuracy: 94.38%\n", + "Test accuracy: 94.30%\n" + ] + } + ], + "source": [ + "print(f\"Train accuracy: {(100.0*anomaly_result['train_acc']):4.2f}%\")\n", + "print(f\"Val accuracy: {(100.0*anomaly_result['val_acc']):4.2f}%\")\n", + "print(f\"Test accuracy: {(100.0*anomaly_result['test_acc']):4.2f}%\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "With ~94% validation and test accuracy, the model generalizes quite well. It should be noted that you might see slightly different scores depending on what computer/device you are running this notebook. This is because despite setting the seed before generating the test dataset, it is not the same across platforms and numpy versions. Nevertheless, we can conclude that the model performs quite well and can solve the task for most sets. Before trying to interpret the model, let's verify that our model is permutation-equivariant, and assigns the same predictions for different permutations of the input set. For this, we sample a batch from the test set and run it through the model to obtain the probabilities. " + ] + }, + { + "cell_type": "code", + "execution_count": 37, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Preds\n", + " [5.4543594e-05 1.4208173e-04 6.6922468e-05 7.6413504e-05 7.7112330e-05\n", + " 8.7848457e-05 6.6820685e-05 9.9929154e-01 7.3219831e-05 6.3545609e-05]\n", + "Permuted preds\n", + " [5.4543532e-05 1.4208158e-04 6.6922395e-05 7.6413417e-05 7.7112243e-05\n", + " 8.7848362e-05 6.6820678e-05 9.9929142e-01 7.3219751e-05 6.3545544e-05]\n" + ] + } + ], + "source": [ + "inp_data, indices, labels = next(iter(test_anom_loader))\n", + "inp_data = inp_data.to(device)\n", + "\n", + "anomaly_model.eval()\n", + "\n", + "with torch.no_grad():\n", + " preds = anomaly_model.forward(inp_data, add_positional_encoding=False)\n", + " preds = F.softmax(preds.squeeze(dim=-1), dim=-1)\n", + "\n", + " # Permut input data\n", + " permut = np.random.permutation(inp_data.shape[1])\n", + " perm_inp_data = inp_data[:,permut]\n", + " perm_preds = anomaly_model.forward(perm_inp_data, add_positional_encoding=False)\n", + " perm_preds = F.softmax(perm_preds.squeeze(dim=-1), dim=-1)\n", + "\n", + "assert (preds[:,permut] - perm_preds).abs().max() < 1e-5, \"Predictions are not permutation equivariant\"\n", + "\n", + "print(\"Preds\\n\", preds[0,permut].cpu().numpy())\n", + "print(\"Permuted preds\\n\", perm_preds[0].cpu().numpy())" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "You can see that the predictions are almost exactly the same, and only differ because of slight numerical differences inside the network operation.\n", + "\n", + "To interpret the model a little more, we can plot the attention maps inside the model. This will give us an idea of what information the model is sharing/communicating between images, and what each head might represent. First, we need to extract the attention maps for the test batch above, and determine the discrete predictions for simplicity." + ] + }, + { + "cell_type": "code", + "execution_count": 38, + "metadata": {}, + "outputs": [], + "source": [ + "attention_maps = anomaly_model.get_attention_maps(inp_data, add_positional_encoding=False)\n", + "predictions = preds.argmax(dim=-1)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Below we write a plot function which plots the images in the input set, the prediction of the model, and the attention maps of the different heads on layers of the transformer. Feel free to explore the attention maps for different input examples as well." + ] + }, + { + "cell_type": "code", + "execution_count": 39, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:35.755092\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Prediction: 9\n" + ] + }, + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:36.649271\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "def visualize_prediction(idx):\n", + " visualize_exmp(indices[idx:idx+1], test_set)\n", + " print(\"Prediction:\", predictions[idx].item())\n", + " plot_attention_maps(input_data=None, attn_maps=attention_maps, idx=idx)\n", + "\n", + "visualize_prediction(0)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Depending on the random seed, you might see a slightly different input set. For the version on the website, we compare 9 tree images with a volcano. We see that multiple heads, for instance, Layer 2 Head 1, Layer 2 Head 3, and Layer 3 Head 1 focus on the last image. Additionally, the heads in Layer 4 all seem to ignore the last image and assign a very low attention probability to it. This shows that the model has indeed recognized that the image doesn't fit the setting, and hence predicted it to be the anomaly. Layer 3 Head 2-4 seems to take a slightly weighted average of all images. That might indicate that the model extracts the \"average\" information of all images, to compare it to the image features itself. \n", + "\n", + "Let's try to find where the model actually makes a mistake. We can do this by identifying the sets where the model predicts something else than 9, as in the dataset, we ensured that the anomaly is always at the last position in the set." + ] + }, + { + "cell_type": "code", + "execution_count": 40, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Indices with mistake: [36 49]\n" + ] + } + ], + "source": [ + "mistakes = torch.where(predictions != 9)[0].cpu().numpy()\n", + "print(\"Indices with mistake:\", mistakes)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As our model achieves ~94% accuracy, we only have very little number of mistakes in a batch of 64 sets. Still, let's visualize one of them, for example the last one:" + ] + }, + { + "cell_type": "code", + "execution_count": 41, + "metadata": {}, + "outputs": [ + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:37.728184\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Prediction: 2\n" + ] + }, + { + "data": { + "application/pdf": "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\n", + "image/svg+xml": [ + "\n", + "\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " 2020-11-09T10:43:38.742709\n", + " image/svg+xml\n", + " \n", + " \n", + " Matplotlib v3.3.2, https://matplotlib.org/\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "\n" + ], + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Probabilities:\n", + "Image 0: 0.06%\n", + "Image 1: 1.63%\n", + "Image 2: 89.63%\n", + "Image 3: 0.01%\n", + "Image 4: 0.01%\n", + "Image 5: 0.01%\n", + "Image 6: 0.01%\n", + "Image 7: 0.01%\n", + "Image 8: 0.01%\n", + "Image 9: 8.63%\n" + ] + } + ], + "source": [ + "visualize_prediction(mistakes[-1])\n", + "print(\"Probabilities:\")\n", + "for i, p in enumerate(preds[mistakes[-1]].cpu().numpy()):\n", + " print(f\"Image {i}: {100.0*p:4.2f}%\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "In this example, the model confuses a palm tree with a building, giving a probability of ~90% to image 2, and 8% to the actual anomaly. However, the difficulty here is that the picture of the building has been taken at a similar angle as the palms. Meanwhile, image 2 shows a rather unusual palm with a different color palette, which is why the model fails here. Nevertheless, in general, the model performs quite well." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Conclusion\n", + "\n", + "In this tutorial, we took a closer look at the Multi-Head Attention layer which uses a scaled dot product between queries and keys to find correlations and similarities between input elements. The Transformer architecture is based on the Multi-Head Attention layer and applies multiple of them in a ResNet-like block. The Transformer is a very important, recent architecture that can be applied to many tasks and datasets. Although it is best known for its success in NLP, there is so much more to it. We have seen its application on sequence-to-sequence tasks and set anomaly detection. Its property of being permutation-equivariant if we do not provide any positional encodings, allows it to generalize to many settings. Hence, it is important to know the architecture, but also its possible issues such as the gradient problem during the first iterations solved by learning rate warm-up." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Your turn! 🚀\n", + "You can practice your cnn skills by following the assignment [complete the transformer architecture](../../assignments/llm/basic/transformer-architecture.ipynb)." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Self study\n", + "\n", + "You can refer to those YouTube videos for further study:\n", + "\n", + "* [Transformer: A Novel Neural Network Architecture for Language Understanding (Jakob Uszkoreit, 2017)](https://ai.googleblog.com/2017/08/transformer-novel-neural-network.html) - The original Google blog post about the Transformer paper, focusing on the application in machine translation.\n", + "* [The Illustrated Transformer (Jay Alammar, 2018)](http://jalammar.github.io/illustrated-transformer/) - A very popular and great blog post intuitively explaining the Transformer architecture with many nice visualizations. The focus is on NLP.\n", + "* [Attention? Attention! (Lilian Weng, 2018)](https://lilianweng.github.io/lil-log/2018/06/24/attention-attention.html) - A nice blog post summarizing attention mechanisms in many domains including vision.\n", + "* [Illustrated: Self-Attention (Raimi Karim, 2019)](https://towardsdatascience.com/illustrated-self-attention-2d627e33b20a) - A nice visualization of the steps of self-attention. Recommended going through if the explanation below is too abstract for you.\n", + "* [The Transformer family (Lilian Weng, 2020)](https://lilianweng.github.io/lil-log/2020/04/07/the-transformer-family.html) - A very detailed blog post reviewing more variants of Transformers besides the original one." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Research trend\n", + "\n", + "Attention is all you need; Attentional Neural Network Models | Łukasz Kaiser | Masterclass:\n", + "\n", + "
\n", + " \n", + "
\n", + "\n", + "The Narrated Transformer Language Model:\n", + "
\n", + "\n", + "
" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Acknowledgments\n", + "\n", + "Thanks to [Phillip Lippe](https://github.com/phlippe) for creating the open-source course [UvA DL Notebooks](https://github.com/phlippe/uvadlc_notebooks). It inspires the majority of the content in this chapter.\n" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3 (ipykernel)", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.10.4" + } + }, + "nbformat": 4, + "nbformat_minor": 4 +} diff --git a/open-machine-learning-jupyter-book/llm/image/attention_example.svg b/open-machine-learning-jupyter-book/llm/image/attention_example.svg new file mode 100644 index 000000000..45fd2897a --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/image/attention_example.svg @@ -0,0 +1,9628 @@ + + + + + + image/svg+xml + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/open-machine-learning-jupyter-book/llm/image/cifar100_example_anomaly.png b/open-machine-learning-jupyter-book/llm/image/cifar100_example_anomaly.png new file mode 100644 index 0000000000000000000000000000000000000000..7e06e5a5d24157731d1976feca48d9bcb492e50c GIT binary patch literal 155392 zcmcF~MNnPg((QqRySqd1;O_3ho!}na-QC?C4(<-Y-66O;!QGudckoZ{o4l%5uXcA2 zy1v?b@~!Ib)oVv6%1a``;lTj_07PjiF=YS%_|E~&Fi`)ZTHJI#0D$J|sjBI$Z0JT} z?`UUcX>Cg4>|t+8V(M;b1^~FPRcBZ_l5;19e2vh&!mJP=5kOb2Zsf;pZ^N@RvQ+*u z`#q)2B8`C?4cN%*`SzUc_c`P9*{3m9%s z_i^!JzV`GZqSSfxsBUNfjuR=jRdDNiAJ5_Q^L@U%7WBwJ`xHV3oO=1t_eCOGiDrHj~4^toB62lHj5=xAnf=HV(TZ}=(vy;R#b&d>e2 zl~L^NZLjCs=R?1sMzXJ;;iCOlL(bJqSI4H8KnH%Y^6aa)s=KeMOw{3dUH^gC5D~g~w*tMx2&x_5)ix%!Ed2?^k(~_P->3Fe{)eW9SJit$ zyW5Ss=@`8l=iF=$o{yih-nHx7XVoihx3L77WK>a|fBimuKWnkSlCF-niqVF!>mRBy z&tyaxCKQkEvx+RFI=8!5b#{oiw;#jw{foyK2ufHuIk@`YvyHUhfjTc2PKR9kSLU>P z`)^+0GK0-ALrmb`Lh%}h@PbMYn3YE)f}rRW+CkFLg<5O$EX*YA{JA)|HB(-E-6NE; z>?f9O5fV7_2lpjM!i_S_*Zai*1%|7U8EQ^eaW44m%gWVDr)m_Od|nQBrCxVb>w4&rto-hft4p!2DD#;1ULrehG5A`pJsbvc^XmE*Gs9`d^ao~Q zxl&G+Q^{)ktx9KFu2l)LhNx?DJTs_TYt8rm(Q&z(DzkKd;z-&iRbctJal_+}RI0OQ zN7Qx;i>%aHwz9b?m`zShn}D^6Ex=)^I0Bn`>NjLw^7y3cU;9s}Z z@4h(A7w4`5AvDJwl^K7&R?NgR#ZGx3@ zlqUFUnZ{(7^?54X-}XS&4ZR>kEr+U9#wB*K*zF1=^=>AJkRKBnp29WbhS;AC%R z%A_Y6=rUeR{00^Ib@C6D#2jz0?a24L1^FW`i3!)SR-V!lK9~vaf}nSnlIfv-_%|QP z%*>1Sl12ASLre?tHq&X9-g#iv2>gkV;k?2(waGM)b!A}$w=2DAIh}h zsHhED9RI4evHN`q&lRAen7An0d+1~oZ*l)M`neiK&#WzTm^$F7m;G672he$hPFaT4 ze!o}i&0s=dwO+hn7X5{m*qUejUfr^LvXX{}svTzB6i4j`L0;tw27B?)M~`nj{}>T- zq)f-k9BLy2Pzc2h-|>nXEjX5+bVawaF!Sh*b1;-h`E_N z4mXFwX1B-HTsmA74iR=GyiydA0&J+uU4QXj8AGaJ^_r>Mb2JP^;(Vy%MCBS_yJ<50 z@oOEAKpUJxb~iE`_n**m?4z3I4RfkkM>gT$a5S+>>mD$%7||jn5)1S>$w~7_zI^ zid_a=*rty5F5{K%>i@}1K&=ST);1u=)bfwyjdoIAF*US)^vDO`vK$(v-N%|w@-Of% zIOiW-eFH;xh2}9B)hiJm><8zuHI7IaZ;URKALIMG{L77IAcmqJoV{^Rc+0&6aV>ox zfPQ1rh6)*By-H%FNFTmGD@W&7&Nwk?z-kS0M#@tHc!)Y!�)u10{q&+5K^ouC_28!XT1-mR^-CVAP;2 z35;+{m`)5iJEP@6io%6K@zsDibDKMR!8xCbpfilF7l-EbM3fN8JVV&9zGc4Uvylw- zgoe@on#BC=b!Z2rtr>k}c0;cCkcz!DLR!ac@KaufRjFWNEq~PqxD177^CuPx+(dkj z<>Qj`P} z$+{h%^q^tSpp2jPNuBcaC_nsizzT67S$}eg?`Gnn4@vP+DL9o7ex<@<;&0L1j#e3c z=WdXfsgY~~q+!!@{xm0BRR6o71Deyi;AAB6+`X?d22YH(=wZy(j@13Pv& zg%sCMN1Jw3G(C)iO$-~XMN;3Enx`5qHAJ2^J&u3pQX;thX>AgV^hQ({)3Fjwg&iiQ zgs#l91QVFi zJO3kNdO^)5)l+FTw0C2+Xv~IG&4Y{?OlvI`HrRjs9c9rBXhFpG> zK$amf6P8hZCF*W2%qt^m3;eell5SHfJhgQ+%;SJ#|3m>~dR+i8%*Xi%5wK6_E^4j@ zGrL>mzu!A4M$L$iX=^V7>ivx0p*4V^J6e_{y@=kvjK?VFjRWhwasLCDC&FQNPraY4 z8i-}GRKK$4jH{~d8$HMwvbZ`EA>aoWL0QwWy+>k-zvB@mWHF^4Q;%Mg@+E{1@$oq@ zwIK7&y9Q43wigm5!jk4r2|uBbOBJ!1Ft1HuC|fVXN}F>|#t_!uAgq>X?s6*|4ny+X zcKIkuGKf>k*>l6(ho5DkR>GnuaYS}ds5=MnuEPfeTg_G?b?pI74&p8)a7PQ=i=YRE zn=)5D`A)+@aZ~Ip@DJv7I9SnG!V2F*ryI<44=1MZBSUG2dclbc!QCYBk-(~RFPL|jZ&BK_SwN@ut5wA!N_q-$SNg#mzZn$>{DV$ zjqcl|kkT4d-^QF9W=hB3cxl`*Pn>M>?<*K>JO0rt={!2|40Zk$T}anLNal;^9go3_ z(Je(&dytSZ>-%1h09Q&)C{VIr_tG5MX++{+a!NPNhlP5_&t zKMRF&?OGf6z%@0PtBh4uY+c}?%IBwB!V40 z9*AokNLK{OUP*-Y%y5Cv-Z9H?55bQLvu6NTeZ%xE#-`dMnxY2O<&IQ`Eum~Np&xkf zwP>*-bs>6Y^eH!N&MhR3DLYd!ZyhcbxaqDVcc{v~3QU<{JPAJi7;;$o(6Ck;FlFB8? zq#MtDB`M1ghIGA)Cs}1ihi3-8lBtCEv{E4=jCRLR3Ktz|fA5v+P0pEPC5fISeH@V zQB;V{c>aXZ(ugY&9tJkgMu=q_20}*-u%8G_uhh{>G~Yy%%j1xv(CiQ25;YvoR>yra z>?J^}93Nf4CAS7+=EG$Qv_H@5g5O@c`fg~cA=9QDJgcijKH_}iQr z1Qf4jOsD9&!>Iid=~07zgsp$(Nx5-|fv#9^)iU+|Gkzd* zz;|a3K+J%-A++dPov|D$I8K48vRInVPkUUI7sk;oS7;C zC8O7(L2^pTNk;7l_*0C#eR-Q1${T{v`*XFno>`YVSd$%>O9gzqs6{aPl^}{cDyrvI z!g@_<%)%>$F|7>VN>gms;nRHg_ZB3zvmH~DVsJ1js7hvfjqz?&6WLXJ2F{^Ur6&h) zGk+SRw4A>xhd{@K#m70aHtE2yIf5cPd&^+p-`u~mTirp{vBQLHeBW&-&rv29=D|Dk zHn7kk7|6c|eBC(Idi!mC+x9h&1ZsO@Z4AkbZC7^Z&Fh^{4~g6fal(wby;r)ObxCol z88q>DAg>0>!b+EPjNEbPw#Hj{A4hf9h*d2gk+0ZB>yhfHc(DL}mf*)oX{5-aT~C-2 znhZ$pgC0b(>~T7J=>r|;XAISL6lbUZR>9dWUY8^ert*0%R6wvO ziRVkTxEV>MiT8*>VIDr?b6^WOM)5S)W@dfHvNoFSLYN!q=M?Qoc&`6E)___-No$5f zGDs?h33wuptvlhqS#lqAfvt8i@o^6xJrNM7InV|6?8y6MZ|C`eT_fRKVHa~$0svqk zmZGAH(xRgO%Rm7D)CoQb{89r#1Vj2N#o}~?=Olaa1(D&9N~n_6BB->%+68LRNwR{$ zl!($%QAIo2_SeSsr#pL9l|hvxFuwj=+&nbDr6Rs$;l3O3ab1pdc*S+N`71IqPo3fb z*}9BM&>WbPbWwq|#}X)p(FU4ksmyfJ((bXfUeC{-*!x;rN1p`zzow1+6)18IL$yft zsB+h;aI&8w$g8L;2ag7eD0SnX6V4(h?6TpJ;Q3gMN>rQ%QH|>V49nm*7M7zbph>Fz zDD@`YZy5K8YE-XiIOKnU7v=Wxh&W?|j_(D>A{7n}q~%Z)fn?TpQKko32wv`9xxJ&c zV>sIQCm0i%TEr6yJ`=xRE{!WhcSsj-{dqB3Ck_%az_wVhetTPK!Kd>7!fl(=`E(>q z|D@s9Lv$Y&2vTq08^W!$q@FwP!#Lob)%7hkZj;_{R)F8OchD0lxn8L0>4KA2T9^hMoG!?N978mkbt4Fx!qT@sijCuv}ItKUD^_vsB5S_ zb-6+1bFP@EnI`tZAiMFp#u5T^_qqRX>@zXwG|KON#Bbr~#69-tzYJ$Y@MB(Z_U7u@ zL;L@Ho3Axrz=9Ls$bT8S6#svxoc=!?9Ql9x&U**Wi=>~h+D*^9!WZjSw=X0B#veR{ znnY3RO*-(G-=^Nl?nn2XQv~%VtQa6%=yOd!U?49p4-p{q;Dkx?=B@SsR4St#{XDsr zGY~`)`RIH`Z1N!hBdHm&iF;lXgo6`m>IqdAB*yD`d;Wf^kg&$cx<12%p6;Iwb2JgX zJQ0{S5uG*>IByA+3`Q|9lUa8aIHSLP{r3#NyR&QmE{zig;*7^w&z!y#c7jJVLNT&1 z+<-BXEYb%3yyLmiXdBXLJl&~3lWKBRFX>07>xJ5f4@e*?CurC6!R}4}tMv=hbTr@)#s97YwAc1{mHL7(;3%1lG`Y*LliRD4lzW;(4Y zOlU#RioZ2XTIEYmEi40gag-h|oZTz;nO&EwPG5>ydKdZ7wC60af9p6rF#*X~i{06o z?I4iz`f>QQ^Ke1v_j#9h!}xW}_?+hVg8lxI_Z3~M-?Qe1wB2=R7mh22ke_>YJ`yjS z911D2o`G8S`f{^;;~T!6xAtREUg`_B$L?G@%&$kE#rZjh=kxRGi}fA(Yv1on_wLs9 zE%NK*>scSUQtwMWPF?Kwi+QoRrkfIFo%O9>sJm`~9o0qPrN3y<{|gDAsLO=pru(yJ zxM$&-E!WDY3F_4fPkp!Bw!!>*ZM*!@MZNih_pN#ez`}aSXE?sPl$y-^H`=nH+3Z$Q zH*Y_;J}N(av!{dhQQ027PaUpB*`5wEt5$r7xC17VR_uZY_Vvsc5?Lb}$^4G1nJs(d z_fgs7$B6#AXMyW>*~&K1^~)lU(?K>`mmJ0U!N<*|67Cd&Wa3lLd|@7ivdhM2WgH-Wg` zZM+Wm*k!N||5z<#cotHP3o90#43q6Y2{HZ1-C__fN%^Z$jtYgv9~Oj5vJD&~u5-!d z4xk}bYG5B-zIgltzDO8J*o$rdGz>cV)rDP01gIbeI!N$mK{GAooz1X zA^f;Z%PV&NCR4x6K@~Nn_1$&rtQsX3j3lK;GSTn3#^WWg<}3Q^?(FNZBV@`YG*G$^ zLnHTfyKN+4L`4~*0e;+erh%?0oZl5JvdO3ElWHvwe(=gCPDQn#P+$s!M8pV+*<+(O zJvG`ghq;>da6Z^FqnA4XEdVX(Lu+t*AWs|+{4wV%lmH`zKmnfaJ-%cxx9qJJH?(Mc z!_hN(iS75}+C%u<2?pgm%6IC|D(dd;vf6z(%Hd@512#h$bjQr^ww7Myn{F=pMr0_B z7fKJWiW_5Z3@gjZq1B%p*bzABKhl*Y4x=Nem`05xXgI=TZ>-2sdbA7YVxO~#H7Y-o04%T;3=)sJ^E7ND7S0^f89o8*4YJKd>}5@>R*38AnFVH1(XR*1d_3; zVgq=Eg1d0SD`wUyca`d{BEOp^(}>Vs^khl0l7v@<&I6D$d{BW+u8Y zI`wNWcv6)WAB|>^azV*ZpnsmLEc2LMid0DOlgY^|?op5IVy#JB7vjy;1;xdMqs~zkojqg_B~~aBNvO%(s~cWQ3DJnx+HPh|bfbH^A5F6g zi>W0gzsnRn!e1?3H9mh7!h}o?Vp1+0jkk_CK$WcT5*ZEtT?{@L4$hSc4hCEV0nTUZ z^&v{=lxzLems2bz?%I4b|HKfe_;5TDkX(%mm%uXl@S2a3g6tn@0Ym5KWNQ?5d>Hm! z5rg{T5g{NRI+42Nundd_9vqeytE`?naZO8ACU^DDL#yu#0HT+oum;!wwgK%TCs2J* z!ostJ9p@_`cY~BQR;555Me?j`UpoBrYJ+j*O2|KBHT97fewcFQ)xucF&kxmOurNZd zMCaZZ)UMDft*)1mw$z^Dv-qxE!fNr{Umrmth`Jk(%Xb9jA2x4|qIGJCkFn>7B)j4A2ibkJq4BeT%T^j4f zX~m5TqEF`L*TdEK3wCxCQ;R}MPM8STOq}sOK@D&VTBwRqea>UAXNqQnEwkp}hxyCg z9v^P|z=##aOq&da5z&$Y}0m)$)+)sG%#+2cBF!B3ml40xCKZ?Ca;lAag^@55*!Za1*AGxR8u zOYd>%S@*}AJs!G#Z<}A~Ax*VzYc#JoDj%`%U;W}^(B0P+hScmP#Pj16Mo{?vwo=jP36*VOyvluV7Fa1& zz(GC@D|u;f_IpPA7}KvMMb+PIz}zkvL%H{9jE2NuX51|-b_G7RawBuc?@?!d(-{e@ z#+hg(lj%pn3bk~;FV6jpdFt=Cp8bCt0OPsU4G0St>b+wPkmB%LcAV$M{^%OfP&7uP zldR-peQE9-4axBzZqhO-IheDQQzxAA@qN!f#6ZFgx0;2I{Kg$N?r*n4tlB&>Z(`Xo zj+S{VXF2G~Yd_{%&68f*H)N_Fn);C#4rNVBoCM)g#OB~AL`%Kj`L4Ma*9 z)Q4z6^Fav@-4Se9J03%+@E>V>D>CCU(YAn)jR|oqha6fyo>R7^H%87Sw#i=zG&zwo zj8`c0CDB@_a}BuEr0qwc3kgvyP)$MZt$|aAjPjSU+=jK*QluTyYW01mT&G*fvDy4N z?grn`owdELv_+L$Or%*({)K+NqR=N!X;^QYVA(t*cnSBm5 zw7FUrGb@Wct{r%%sMRoG=Y&&Xw0*2TI^6g5kf;k6M`$er+cFB4|Zk2{>`Atw`T z=3$?Zkr2VG0P8fZY34P_?G>-1IDRSI0(n{7tG`aUpdpAyNkC}oT_9A!&?<*>)+_ZR zE~+AzWYVd>!K7*;*khWj8vQ>gs_n_WF4doYewfE)WQ133w{fnBemX;nw=vUMy?*#0 zTp0xg)cAcT`)(=3M3N%($GVAIrPPmoR2jiOeA&VIu+p5GE}4C z>r;PphnqI>pVZk&u&I_rL?pP*^yt_O>BCw%?N`Ule#{BN#evhM&P{6Np`q zk;2%D zT6^&CDL{1J6gM7ZUXmlsa841l7R18D*!-YIP^Ecf$h+}Jr$woqKVd7#adOQ!Y-l1Z zv_8Xd%TEFUjf3%MW)V9ssnsg>drPL2pt$NMwpNybJk$tfMwHNDU~&OMC`_>7yd?E& zOAIp{?GDgtaJQb!mOB#CuWRUcIBJ^(>=Lj5nI8g}UzG@TBGr_X8-j$eCm$PlkMgqq zXb=8*WcjK9BViikL{2twxF%vc5lHE{l);)(O%cgvXOwSxvaA~$Y=9x&xpQdnr3%mq zX-ktQBOH4;9>r)yz|%bRy+{1)uPROkN2;?V?N?M;A#Hh4=)aMBb&Mg4iNe)H5xP)P zNF@trFaH!1o}~ySk1uS2Gz7gF+dA})(DaMN^A{@xi0253pi~@=V)RUBJwCHbJs34T z58|NkIUqL#VkFqxJD~8rdlpa@{_0&D9I|LSwwr{!GyUb0j{UHu)Fny!MHH|w5~2M{ zNv>uE6i9A+SIl+-m6=L?L!m*_LA5d)PR1$q=(o&q5SZesW|f~XF|b^>U*|l33$BPV4QU7IF?QQv)gAZCD2um589} zkOZ^@>KDVf(n<{~%8SNmZBmb%t<;8`N($|)@4I&Fou$oOJR$hsdj^#WhJNV+WvhwC z!Bi{obSwW_!rQuEa+6%(E3AHxIx?VTsOdrq6WTi>NzY|7CHcLe*aQW2JX8Y%mf;E| z>Zjr?J>cE=`**NqZhaN&()>Sc!T-b7_3H$ko?zGB!#q{CJl&GZ&`Hx?Q%(9&yXByj z>kLk{wpFU1oiX)AH>zrh@CV<2`u$8zp&xub_;oGa|7eJfexTfV*cE2`5L8VyJPH~LC*q`?Co z>B*ZayCZ6x(ys;`BnkAXX!_r=G%*JB)D&zK(xX;Mr=F9uI|YF>^6{_ysyr}HnNQj( zY)t$Q(>KNiaC7PFuYD$Cu?5P?yloNR5L~)g42t<})Y0GLp;X#31YB6@Vaxa%4rrLv z3ZI1@i63d*zuV=Ais2w?E}-$|4kieBt15=KnQ=5v&%RSN^oJe%?fF4cJjJ)0Ki_6v zxE6SG6Em!oxEJR&UNa5PBh`T{f_^KlP!y!$^rCEO%Na2Sa-=cTMVl@i%f6nbn?AxV zK3}ngU5=^WglWaM@1ohN9b#zHEE(Ofub}9kt$ze0gz`Me->5_P!-ym<3o+9uH;8na z%KxiwdJOOkX_$EeIoED}zjp-RPawZ`6&Rm81aodq=5Z^6$skZgl=_!j@om>bTc7Cu zae&5pl-9*fdoC`bKq%60S=SLUL-C*$^6-izO^E)~P16%Jfk>@e(O$^~!Udz9+kxnX z7Dx%WOHcMKgZNJbMyzi}iZXg!c_VTgp*1>TEJHz7QUSse;8mfOKu3-`?b(-oYgL!9X_s&1`dB6TeUo!1$ zpifW3J!>!+$JGX>BMnSub##3i821%p!W7UK6_USH6*oX`z!77~tmTV;d;p(|gS>LH zf*@jbkHc^yo*?rR&;mUpTIw681kOUtD^~eIDD&52^r}7o(K|C+1_hQStXm{tQUoBtJM)>WtLRpkfTif{8?d>mxcfJ(6L_h#>mahW-FKx!e;b`#Rb=_-6aaUJ!>k;m zwlG^)l4iV%@=u8`nOm0D*s4lz*0%bk*7U>e8Br@SyLj#b0$3rr9pCC*_-! z>Jzk31td6_+-`y*wSQS0rT~8ozghR$!BsvjGbz<@Hm!GB&VV84&_csVe+^pz@@hvWI9F<-nM)pC0K^z7;Pbnoi- zbnZH5)?W;nI_Q=_*R8Igky43FM=N7+J5*-i^(EWH{lt-iORN}j@qmJj#Uda1@n{~n z^x0Bf#tM@r`n#2NM(w|$U)iD-V$~5TIHIG z3WTa+tAhbu1V0#uwP~WyJ>R~WNle4uu^94KbLq?5I(-Hzt0Vi> zS7#SFRi8{C#U5eps4*6(43otxd9lE^gCJ!&BQL=iR6xQ;iELc58uS+}i5+L0e`9IY zuaaf@Q@5p8%4X>2PNa61TELom!nl*au%` zl#W)>ksRY<_2pJ*`5ZwOJ1cu@<`OM!c-O|!rH=xNCHmx!dl;MAQde*B;(*djQ^Cs) ztOd-43^L5JwsmagARV*cD96iIrD}KjX?Zzo<9^$zs~ZPEn5sQLhYsw#N@$t^K4i)z&!rGV$E z;u-=8xDl+hH%pT&(u$$B@Tbhw+EtZ?*hHJtqeqHTk8J`5@q)2K=0!DSs~6C&>`T-E znO1hq?MnfQ1*m`}t0Ifo;1P7JZ>e}kS;LjH@U2oEfIB;6P+?XiRC(eYN+-KFH^%so zPhK}HXfKa01=e)C?t*t~ zyVl*SXYb-8^yv)q>MCnd4)YdT5iWD9^MOgxotYkwNqUkp&G98>tORDYPjwm!q^G|O zm9M9-;r;47d}i9mLd67Ty6)iBAri6KkO(G2sXrO*lHtI^=*L(r_X3HYJDmzi9 zSsQuY1;JrmHCp~$FRjDGCiNbUquhwDHXn@?yU31(3ByD^ddP&8<{P)wa}Le(J{!0U z#I#k0OS^MmQQYYzF->|?gI{duSPgzkVexWGOB_US>f$cEXIT16_(i}G3R5VtL?%=| z&@X;ZZCOx2HkO8P8THDiz?Gw(J)@3pim*VZr>F1G-nrx5ZJ;}o?r3{A69^#U6GaHS zwXpENawx&8)KS}a6)x~jAlm4_AOQlDu%F9H!Hnyw%vRsQI_D0A9*S;4wnirVl8XcD^ z#KNdcZ4m+|H#1(lQ>mOZ`|qM$Ju zCyzgW$ezx(R->eh9!(1<{itZH&GWv>^Cv-7BSi3SYIyI~Qw1_&-Qr-AxO~SB;SsM+ zoDtA@LG~AZc>*gNaHjNL={Y%87Y&c0plc384p^vnQ-x5m>Eq!5aOjR|ssy_0TeL}g zb%(BMrg>!AYnx@KL8i?|HUi7;%iMx8BXGXK3K{qwm1OFUWy)O2zVDr@@ox zC49ng@lKG+kTf1QoQu!>bI@YoW;Tu)E$K@{Vv`2n!r{_&;RX${Zk`ys*A58UfRwYU z6dVRM8L0BsLjV2*I-r-!r2-`@buu4x|0B8r8?II+Vtt^UThF@?qAq-Hj(rB^ite2! zk!DE^N;*uTw(UXuZKWZAq6V8gK0{sX&YukKANP^^U-xtZAGc9p<=Ctln#(H2ED~;< zwhh6*x#ole;s1mwlk;4P6f6(eZzd04p*7ZHXdSKyo&s0U{n0n8O(Sc*y}?|+cfJf7 zD~vPay11DH6@ZB-#2R-%AV!qG8BLfOU&ScVg^yeBSUS4WvIgbwQ*s!Gi4AL;$+sXC zA2d_XMXS;W>>3OOEavmseJIX4fl6T_ zkD|g}5K2F=s&XzqmEN2lcY3CNNK^_MRrtCUVD-z(z#E-rN;XFzX2%#I@q_*lieB~u zfuwkZCs0Qtdrmq60OgIP>r{-tskLsHlC^%#gPL}Vn`nCro9~_;LdV2^4I;g3OOKC_ zMpR?+qwDYnE5XEO3)JC|`U^Mp^oyHj>A~b)#7os1MGgI&Fy&k_C`4{vYctRBe%bsY zlu<=dN~??*ZJQcRe;3zb+4m6*B4}fjd>5so3~`KBuk>i$o*iWpGbTlTWxsbgC#m3| zh!vt89fd`4MDE8(A!bYG@@<_*OtIJgc-9aAq# z7b0AGkeLwK*iYj1_`6+w0LuTt4iXegLMtO**mUK;h8#*6mJ}sw@W34J>C9IV)Noyc zjT|ffB}&JHFo}WQem?YXYAc02a_RHPaL_v%mie)&6^#J0W_x+N2z`3O->SWG@4%yT z>A>vNhg{a_yU@1pP?K400GXy8Rw--Yl93z*o9fj~q91xVL}23kj|=6(sBnb$Xr6 z^Qr-7X(9Q*WLFcn_W8QHMJOkc7999SY+)ks)Pto{nSw?x=$Cb508KSR#8bhQfgv8d zF`|Ly5{tQ0U1KImCE_@@@ittG&4BBAooQuEQB3ASnIbNh3&Ck(Ea5LHOyU>xQ_=m^0ER~Pv_ncZu$HP0rmCefpJ!k zhP6}ceC$|)qzisUR|^SDMo_jGw9a~c-_*X-iOQNV2t8uRoykCK2)hTG4dz8TE4L zhcsCT_L=8%i#rr`0>$o*eZq?_>kjK2Kke{qI(J=-H%#1QW%_*4kGorG<0vP>9}G9% zHfbuQG1RI*L~gayL8If;lSOntA(SgbcU1Yr?1UnTAlffCN{Snl=T7S=7L3Z*7dQ#y zS-tp;WxjKDR1(3mEAyn!@ljV39F`DiuH+fZAxepdPU8Ox#ZpA9;uIP~p_N+0kC-F5 zGi1EIl0c{quz=aZKN~H$0JD0o`L5QYSkQV z8;lk5kk4uoK4)Y1F8ve#gNHR_jv|O%saVNZZHysux`RBv0x3{z4ySJkMZ@bmwugEsd_L3~=cYv`4me;qi)D9E`Quu)dzgJ!MM+Zwmoec~K90jaL$DV|rW#u} zYZDiDCdW?ANX}L}X$25j$2jrZ05@;CjBtpM!E@#*nxN%|vY>1|CWJBe(C87rq+3;hCSnMVXX)uWf% z6>*l%S(f~Pd&allL?6^-mqV3(2s@F;JcT=-$vlH=X|!PrmAh4Q6_exXw|(>2vU~A+ z|H9{A4MVH9*{;6|^r4GMh;%(t`L{Wss+tIN(XD5IKm50rGt7ihfhx(cZ=Y=YY1Y|Y z^)F`2qaU;s{j^)D#cw%7+JrArWAYeGb%aV_wwqwdXtHY*D!?!<-14yc97stxofF`~ zPpL*mZ?rNm|K0=wCk?O&Y%XfpscJ7-xLhviRZ-)&E7lviBNvj%1TIP%cmWpI2wTH( zX{ZipjQ`28&Z6f%`x_Y2k8sy*MlL+>&JUL;qRpIUkZ-B(9qOkCIu~bW4};D(+(zUI z?D-cK>#{;X*>Y>G3~fKB-F}+{#`ib%fXq7Yv;bK_@zD+s4 zN1FatG-6~sMH}_X{g(m@g&VMQKB;`9j6ZLGPm4lMyg)v$Dgl+By)*o(-1By@Bx+F9MK)uOmOCG~4Q(RN5haQjJ=KS-Or;hDLRVvN6J$iQT zNBAr?qXzb*2GC+r4|(g5o^+Q#OSa0l8VV*6Di{hPz6)UrtrGlGl>5dEQQnm77x4>J zpyzemu$v1Yvr&K++8`-f82(ZCza}%5dsKa;`SuCD`~ups=?Jdwf`)wcBb}*#LK8Kr zAcvez%Y9U-2>J|4)h|DCG_->ATZT%bwP7UyK%gg{A{h^93-Bqn1>isFc7C5w>$Z7L zHO+4OY1l%LZ1ioundGIls*TldiAW0UT!Bj<(&N5%Kq8h~LDO=xdobPKDNzQS<{uV2l2zBHrR%X>0wvBSkeNDWdvFoYhux{(# zvvvL0y?eHfsK0&w;<=f0i8JVUo(1{aut)S#5Dk+EM5GKr-R)M0Yn0c>feqV^IL2(LG^ z#cWY$wVHUod!PSD`h2h)Ec!PJ(=)0jcxO6lDe7O%s)wMH03?h=g7i`(P}TPU!bYZ` zRX`^Sr%JAGl3S(hS;~QO=^8QZ+VGpF$sxBM^61#KT}lJRjI~$H!R49c%?zm+D{#J^ z#>MJ!nwkfpX#eL>L4ku1&-C0$Q0VNU%bEE;pk`BBl)l#1S;OjEvU43iWw@*(=*Le> z9zi!p^gs@d3Tzd1Np`pglR-17g*Z`Yl11dI+i+B=;-cWA0>AQJV;uNc{9B@m~~WkHdgZ!#m3m3 znSzoBy(Wn&TI=CL4>}GaVeKtQMASPGbT+KI4y$&1xhjcd^gc}N4Y5IV@Lv15SxOJ~ zPra8ndHqklG+rm^^5Mon1&N=@{1j63fipMtq{86Phr1qmj~?IN3(#9k)b+)iUXP5_ zpwI3lFqAxb;IBFPo3dY0b|BQDD;+qWc(CM+s8f@F+FYq~vbLnDwM7F1I996l*V-Ec zH(_6^wYRmgdPyFoeRP+O4vhEE`XP8Ngt)-!_INaW@z|A=0nK2kujzp^<6j9#X6t;=)OZ-!lIs~tQL&~!jva)V&o}fo_DK9IqkSyH{7nBi7LfvZ=^&E zc3U`e&RCZf>tezEvP@WJHc8*am}Uc1>nUShCoF5mG&9oVQWF?Mv+pKJOWA6DYQGXg z=@j}N^vpwVKwHeesMbJKIbyY;4Gf1ufW6j2jTx?5+$iDJsgi)f-Hpb+LqAd@b+^DO zqKJE3bbS}+WOZoEvS2-(a5|lFIvaqhua|j2S$d6B*$ZyB4PU>0#n&&N@%8JBX`XPq zy`$`#)rsso%Dz=&+Jn?oA23&s3JCPWf1%jOP9IBJHCQjF%CNf4`LeEfdAWF_U3;8P zE0$%!G|gtX!L35Pv3;fNDEmG*A=W^;X8XoWx_)PW8y#w&@E|-SI)WdvI<01btb_T+ zhj?x{y$||)eV9J{9big6X44E0bKI5Cl4O1%y|?IKIo)bwG&Z`X@l8U*&1T?Hn>*Lq zCIA}&U_lPTC5095?rOH-MUB;|4uq|ZM77RkMnbmiHz&(MassD}XbXBSI4;G+n&qExOUzSiVENShdwwo zk}Tb8!1pIT*kCDwgi}ALKl+{|s0PkgTw3S$-rDOHTQIK&6MGNBC?--VB*kj2iXSuk z&sV7*QJ3zTDx-p?(bBl}bkZR0yIVi0fuI^(4Q3J_rL9!Q`bmm)rP}MKq7^zUTi-3f zV~?7+5c}~{MW=^Mdj*mVQ0i{3#Bx^_(+cMUz^M|%psX8uHKIDtB{JtZre^#B);`}t z?sSyG=g+STZj^u_L%qb|h<%f3iI{-CE?Z&I#EZm#?OrT*;vPW-l ze{{n2g;sGk$ae5zSXHx16)D8Cze$58KWa44vw!68W>TNWP|y9b51$6qHMd-Dv1!Wgq_ZNcPFL*Npx zIZs$->z$LmmgjaWD7Ota7Hr#sZ7;zxSCKM-Gg)t*Gyx3k`;PN%$L3CW+ve_XyXkYw zCe2+&wKir{3x4Mr^PDj+R-Z6ABb8^y^DeO;#&(zFcs*ycnQ58zZ%zPgs{V=+7<4+qMm#KYzyOe_gSD{)~B^ z@cuU8dVR<3dd2k}P}!j+aYZm&wd2|uNj@@Nga{`%x=`fy+>He z$n%V8DLC!s==Zk1UO>1?du!9zQ5zE5~zq< zzSnHGn=Pc|K+OujyFI!yiPWvHp&xjhaIDSX@Fe%1A|E6lz6(f^^lPSKPF}vJwal^} zg8${Au%^5M4}@h z9X$D~0Y(4L;!K=0R;nH!1pwC)%g3h4$JQQ^y>tt9u9oGFzbVx3Hv4RE$AQ`q+k0O1 zenj0#6`&5F5-f>OO?x|O*PB99?~51{-$%QXsOr!f%tKJXNMbTd0V8?4Uv4($-h*tGprF@|VO3FQ;8uuuS%p7$PWf)2t*dP1KMLlHNY%~iko9egWTWl5EpAG&H&Mtu(+6!}4w+#(GEonHt{K|I z^nCc8xNz1W&;nC#S;|_7pOd+_&C`r|UaZl2xmiZCY(UxEBtm;JH#6pw0jQiZ<~3uQ zQ)6>P%+0D4lZZ$KQH^@`B4+PZ#7vl{j5H@qQ?j7H0Vp;wH4*Y`(mD$WJUf`2DUcnz z1F30V{Btb}SnVP2pWTduWs2 z%!wPA`Vna#-PINyGmZ$_d~a}^67oD_Syr4*CtNNUoX#f)Q|HFskT<=n8sD<-Sl0#f zJYkv>I03mzkQ`WS8wLCSkumx+1!tuNMPR=L^m!bLgAq1)L@%2U1;fC#`wszV9RZR~wnO!rpPSiC~KEYACx4QYY6! zrTbkI???4#-3qvwLO!w{rbEzqXy%B>aECzaj{sB}vxYIz;~PWG9&#%^qv70{lILKT z(Hdz|D^9P{i_S3PDCBRwku3PJAc53sfD(5%C1nB&2d6t@r3O3TA)t2|o~OK%&w<5j z6-p^smc^1&^E*~Iv9RwhDBFg25XiI1Wq0SlvV)jVX6t39MWA!G>~2mTr#BZvbB9OhtAntI@fu|Ixm=~1?w^+dE0}?O1;0o<6p_L?YiwK*UkFRh}D)Bw^}Dv zBxF#kJ%tE|&cpUBy>*MlL*xN_$d-!qIuv)KsuU%W-ibVa2uN|y{8Q}xq~1iALy!#} zoO!!jFNXkbG*qdE1*#TYiFbxl^m^dhjLC?b-c%EnfvSHFCL&$E(V1C|yb$RSjP>PU zRJgRnLbYwCLtB)V{dd4q!IU#&M4{H!Wrh@8YZ{QOa^;ZFt?$jDkt$oe3s>)o2?%L`?fA9Ex(!La?CpxS8Gf<*~U zvOWlWj~U%ZEYx?@G7O7AT*NSX_A!NKl_@pL@p+!H%nL}(we5B**xuia15Rb&zLMB*1TxwV<0** zGgT5qi9soWb24|OFrWVL5eo5B%3dS3;(WGdw5QV73;cM4OO06)~1w#WnC~$ z8Pk+P(_~x3x7!t0w!UF^VCvfyP#*?F)a$6-&bkNJ8rckKJ^PPtw)f2H*PJwFCGf{l5LYQAwUC}#Qs z6LIGR_&@>l2oyam>cw|BhF zE8dqkye*5l00P)=8*cAcYv)-Ma$?MuN#fl|2sG zd{-P;{j0fhu8TJaqo^6Nh|7Q3 zqpT#|z@oZ3v}V37=n>p?DUNa=?lmX1SW2T=!#7P*erqlCT8OITAgfg)l9uJQ8YZ-Q z&|r2w0@q6GRLx9KH%Kk{As2VI#;VnYu1m11#0*-Xx;vN}<-(7peFz9ayDX0mu7aQg zGu0cPc9t&4`bgTdVzgn`)*|&zYY-7Y2P_9>s(luB;AYIOI56@SA-?7(N+6n`#1PsI zfRYZicu^&c=9EpKOr4F2$$E_dsj`4xNiN-Xqq-fppL2&@4(1XG450*}?8z7nvNA#X zoorNhxx}5sw6y)8y7U!~tBNN912grYUt6tlZncME_o|TAvd4ju>NU{^>vA2{PD8Il z?fuxK*m@zif^sufinAKt1-a?ZXiCyu9)speJE&^4{vb$`advYTV~*I`T2R776DFE4 zrx_DZnDdOpj7@J~B_w5VVHDb14@ooPNul1>lw1`^vDYyUvOJ}pw5)7Yb*lFL`?sWh z9L^&Jm6O#OcSfq4wp3#pZkp+;DAFC!qFqM)w*XSdX-Yo$y|f*rWQCjafWoszusGonovsuLCK61{5iKJz=(*!aOD9nUR-dOw^pZm8;hjl~TZ| z_Q+85?(E&@p=Y;353jiKsUM3JfNC$6&Pvrz+D1LwO(~)%qFtgF^yz~O)mT30fm{{X zT3tnx_x*@v>UaKVqqyCvuhvLebJ`2m(+P_MsAX9(ud6eYRtq&xGeo2pLRwc!uiF)< z6ztm#x9hv#+@-Ox_892hZ$|sw!Hv9GwCm3kH=_VVt+(_BrS-E^)!8VwJHMQpr0;UE z+NgD1F)s^NokDBvXCnY6PKMhDpllIj>y-O&*cTJ(#v>D%jJ-_)G-{p5PHU<8H zRMPi-G11ZTI(%wvU}!y(9CR_8q_0UhLOlY7785e4QG9cum}Cybyf*Uy6@_qV+4L$A zYPPU!n@{4(NbT+Z2X)KX%GAP}S#z>S;?#8awQm^#R)>_#@MNAcGBczU+^#q5w;SHS zE-2gPh2kbPY%<|epoM_Kkllb$Cuz*w0N~!Yh&K%S- zeD%VS8sH{payiJht)9i@RG7e+!J4hE#DPt5wmVAuDDJ6b+P5*+686ENV^C z*Bhw|Gn8!}S^vNZ6Hl0U#>^9DUN9#EREZOkSbtAeOLB|sZf3!)F0wIF+`$w`iHaPF*AOJ~3K~%OxP#RlDNGA%k zP;997Hvm$xlW~|hTEsRqQrl>?+E>#ztX#i8_IaG6{>Wxp9~O3xz-uhP?jWM~36YR} zs5aRafP3RbNMn7j*DJ2qg!!6qeP_%IW17s%zk(=c@ocIuVB!)Av8Jvh0@4Du*3yQh zgAqo~#!_W>?aNC~F-hw{LEy>2*0g5K=LuymSWXk>b;7hvnC1-f8iQKESK7i;>s+K| z2dic1F4km|l3QImBLxO>@g~EGkvR6Sh>N1OVs`@_3>0hNNYyeO3|0K$g9b}zm+EnE z!aIBasHxQmrPf!W7Wy(LaL!-=%X-53d}~y)F;X?jEP(8IzrK6(-hxuLwy?JzBe?fR zQH+ja%vCySE{jvn)+9AqfT}gaR=c=Y&=We*pVtMK^NN@A2``s3USADF8GxGI7C02a z`E_2TJ|qUDSqUB|LcrHfw>-&Dvwwv{bN4&;4=D%7^4P+yCgzAyDzFk zzdfaHnt&#g@L(^Y*R-g4j*%O^RNx zYM`ltDCX?X+!{5rJGSL1VV1eiYbH0cu;dSC!s)!W+NrmpUGAryI|j)jlJ#0 zgn`Y?a9K?%4T7l4`;vp%q9`|$a?`fEd^vfth=UY9)cp;;$vT zWORRPn44i~Ol4`n&iRGfnA8m<5*xQrc?b$_V#Fp#CwH=|0eO{<+HrDk%2^%7T4V3N z7ui0Md`}K03^*pD^kCp1=(gyBd*hx8o+20gC;0e zCnLpd{<#OIEjh#DPJkK>(L$%{dI~Fu+Fg3tec8m%J6FFpa(0sN18SlXO!be}FyXbk zV1~_86-^amvC45TLQq1*8ov>g>2JiP^bwJOuGSn9XmZ$7sLQME5&Cn9$I)IlE6!q( zn&jD93owubn!LZAs{LJb_H#KuIMpE1l1MybrWsROG4qU>7po0TGp00Q=D8(mNvQSL z6x!{|HjD9|pkjhEQtYz6-7Lq|8PnR}xF*o61E4!ks>(pDj2w$MJ4at1A*J_wtNWy? zyYLk^m0)W&tPGN5fJ#i`sI|$Kd{l1nSCsZ?Fn_2;cxLy}n|cvfKN-67$A5@_?zKAy zCSLEFY7TW1hZ?9mn_wR_rZT2w0z5-o*&h)reyq$w-_T$Tm54J^yJeZ`4eEMX5k)S7lI+1;Mzv9JDYHhN z)QXO6k?`icqE;Xa{1BkJQ-lr)K^-^F)||FB;hiuq3+8pjG|$M>1Ww6y?Nu&hflx{r z#DwLvLc{{VY9rsAG*tj8nBFpOx2wxS2^-zqGWxz2S%*!bOEpuX*8gd4wH+1^CeP@% zj_sv%U=_gB>9pc}I^l9LyYtuA7rb1~IGs+I=NUPtUdIs|3m!n!Adr==3AcV)rPWH^ zJNU))#Yb85gMDYC-RZzgK4P!+upa0`GIKb`gRD;;Bwjw+T7M#gB?@}LfYDujXHUMH zY?v*n!re-nf>eWi-6aXmaH^)UM>h+Q2Bf%&l8C7F@=*2`qYk2k6(Zbfe)eK!IXP{; z#=_T*#zLjsdi7Km%4A9Yb0FktG9Z==ROKmKK$+l%IT@GB8J|CY!Iv*z@cFZ~oY?m* zDhfTxWYRtZQ4Dsl!D{OXCvQ4Vj5+1j*nC|V2T=Z4jfF~7Szfkmr+6L1W&mp6Hk8{2 z6a^Dw&J)hdYRUrZF^p{&>|_$iQWT}^wrAIUBWIH^Co|5d>kw6UwQ7yth)46aVkWOO zDQ)kfz*I+*1pE+?6702*esI|2CW&jwCzl$bvNi#b7>UhhnvlHTR;eZ~#?sq+MX60s z-XNTZn7~xEpEYSnz|#+_^Ex%I#|t9KQX9xQrObJpw?R+Fy<4pZH>8Jg`O8KfD|1QoU|MhjF5r1iRT z5XG_0i?LQD#qWNyuZ7;|y2`^^%~su$s^Lj&H0zD33{1HVw-SKV$|bZWXov@^|L+Z2 zdutS5J~@~`1owIkAgbvJ&NGy=+|qE zl&g~b=TKEjr0(KeJdscx&#ehLF++0(iy|eDPhAlJt`3^6R^cAA3X(BTykJf%W?rzQ z)m=(wdrlJ))ph}dqzOqf3hjf?PT|RYVkAb3QEeAVAT0o85Q^21I@A_bx8rY`cLy{y zN^od|sBO=yQfKM%VIi@5(jA&AP}2&cgaeRr0^1bKO=|tO15&YwzXnRwQyD^|ZG;Bq+Nr*F2Y-tIQ!X|aqpf;;Artldaft7Xhw@vhr3;QVPNvcvYB5F(w&|ga}Lm{v7(C(1QuuVs6eSz0L7$e2l%Wm#X zdBav1RIJ~UdJ@E#qTJTSR+P+c9z)fHg^(zN7{(Au!89cc!V|SQeLkIhj%Q3cAv18j z-P{$jH0d#FNxK?5Wk8C+Wa|g1g?5=4OUgJ+vrBl5iMpK6IIkz1PiHLi94v~xAi7i_ zRjgj?x?$gTs3;N$Q<`y_orNljZCA``$HqHKvCQf2EdH(vo)U5v@RX4!n`4~VuTc%= zorS{KFRYeqOUR&ii|?Ng1a>LdQMyF7?niK=3r`HiM!j^j)hFsLZ9EAMHBioMcF#oK z$hJ1eM6$o5ba1EZeXA_tWHQ(0w%xj~QoARrK}~B;TxE*t^0`{;RgGAPyh&#Uk=Z#P zffNy;ICvr@W22PX+N>TF-u0|ICh)50rPZ*=kPxX7DIj;_D8T%$xZKvF6l-=lJx6gHU65uAw?wzH` zj^`&BsoHO-iS}9@SeK3}Xh=y(ONQHb-a{NCI<`Qm<_9#yj|a9%4=T=Dk`fdJAIV9yKP48;BxJ#^Ba-EFIv=@5fQUPuh3v-kU(7HpqjF2l75{u0-rEb(Vco2=)SCQ-BNRE93o*vCQOs zo`6JxluXu_S&)3ON!6sHx|&D<8jzAfJDog3kaI$IAe9%wv{;U*)a+leAha@4^C@FK z&me^Xsme&r6NJpz0e6g4koE-zse&kWo6eM6?v@yw8M$PXRIC<(ckEHt7dMlpIaZ+f z=AB&D2)pTrIrfLj`)HK*rHoYjg)LTa&MCJ|wHEb*YVDRu`|>oIHFQo`mlMK(d&`C#?G#{$@u0uTOG$~vF51jikuU+ZF3_Lv%0rn zQ@bs{2+k*$_Pw0(`f|qW>nkoV7pwnTOv*?UYItf5__5=<9<1zJQ^P9lD=r7q#z=jC zFR8ztv~Mg7{uXG`KjveHM$t6bS0A&rEkET!gLtnjZJprJeMhP}Kyg-TD?9SOIU8?s*z62_%9dO(AmywSCkIi~CDB!`q^VU& zN}0X3bi(<3!sTVf%a=1wrwQwlF=s|*#oO06Y`d9HY}vRk^7C9{Y6>VU_a zr?2yj^Sa`4GA8zNz8H`?pRMk3FJ?#V^$PGlFXgsjzi!yx&0q*cnzS!@EnC?Yw_R{c z8?J1}r@Nx;8?M`qTiI=$nFYB>ll0`Az>{CUgeH#C(BDa;E8}2u->s0T)d&699xJM? z#sj-|DTlYU|?t@hu(Rhjn1KU zUlI)scbk)JXrkLXTgUdX+E`xwj+pUSW8#TiqwRuB{0I-WFxQ^t-R zn@%XS_d46=Sg39`5u3eBV=T$%o@fE-j-q?Jo`TwhTe?{W_es}#?Nv33Cz=3Hz#t?H zH5jHjJ1m|YY;06+4Ga&UKyP&x5>m(xXAVZJ&Bn`HQ4+_41vxxlBauvM& z$Cig<51&U}wwo=mYG+^E_B*@Qvjz^(Y?8g&-$kpH<>s|88D+m=-`I7>0#rtdtA#wd zg{f^x>OiVKsAhA$p1{DAB4sn7ND1IvHS;EkQ_o_hiIJC#`7~j@neKHtEtt;}rX?dS zwop@|-Wfee!KmGg+s$y2xWJ(v6ncvG~5pDSh$#XrBF0ah-mv zR*Hg4V>WqTHdd~qh59Ia|I^xk$7FtH@th|s z{;w9K5+trw7SVL(!}bySbx3uZEJO3sk;(e&4tnHOa`2bAlZZ5F+&v4XA#>TX^*;!v z26r{;W$y8QPzr&VAQ%&pG?M|8w`Q>#^iEw%`It zMx?ZR-v2SDPq$rCMv%by0e}QkD!Wu&p6+S;3#Y2QFH&X%$%Sx)RTOG|xtBwu#EAVcNYGA@n=l4p+`u3|rn)>=Kr?tnIPbK3 zFIOa6lEkx5U!9@fN%3a6dkBNO9gt_oSXpW$$Q<1WajIeX86&k-WxPnf*S_fhR?>c? z%bFpY?Hgam^7Omodive5erEk?17t}Um}@2?ffakWg-TMUB)3*DQ4#@GUSJDWmqFY8 zqg2+v7h~8MY_aCtmS=Ci{p8ZV1FalrY**Zeb%vVo+%Mu`vU8wr%~b@Q{{2Fj`LHVx zXD0q&Z@qHQldjO5NG??>FJ>HPH>tQ+DzA+zwu}PxyP}751t*a>xwE>(xAv9)Rv`7y zP7M=zi!3sg_8T_nQ2laK{~18H_cr_eTlDVd^@+m^d5NuTw^a*~ zK=MEFEH|{x-wp%aZhqesm{Ph-ss~=iSZ&x8|u=f$K$|nzdi7H9C&{_(R<%}E`}^d)a7^pP;2!PW=l9bj%NOj(?HPU z@jxkr=dcFpG56HMc=qM!kI4eHA z?E3Sw;`yx06Oyhd+O5@?=Gq$G9|wLtPP^V7cz?5mzVkTHs`aNBBVmWXDA&F38=ucl z{Px>#_}hQ}4S)aJ-*MmF%E;ZjAgU(c8&;Bg9tXHq`yL{|C1h~q`pWKL>>iM#Lc3ig z1KK^{hLO6b{43$wPJqu%9Li!UhwJ+WJEl3eh=^K3o9_64T{WdOwCXj^weqsD8Oaf8zTc@PN&1AAX z_ruo)j^T%<_fvGr1dR<%gA;%MM`#68X)-HK{}WdNYe7LpqXUgj9Q;5lhmGgORdxc9 zkcCa~VfP!#VsbR>N*+=oioJt7QW=+&HnT#@^Thi;o1~*r{LSkVm_@x`8p&xD%?I7B zR5p+Eik11Ag)n99{`0YrE6e;iTGru``F|jgT75RVP9^u;(2G-nGF$W-u}btJ#rZwo zjBefr+8&~P?(nK9pOL{~r4;j;-2KLNU3fk(P%9?UV?r$zh1f+heIBrw)r;v|C#oF~ z(`!(+Sw-aNH0ryILvrb0M8<}J*O;&prL``l$c6W!b`X+EC}+Vj4wLyYVLSw_zM;hd zx_JUv!JKqxDoy2JYFfh(xKyY3Lh@BdoD-aKAD8-=Y`&)U@T{Fy51$b6gev?QB=vG{ zk!~fm+pkJKsW8aTj;LT*{4gb|en4v|^~PQA`7;u;d#4z!9XQ|K@ZK+o+#ra@jyDG| z4Ihsa{k|bl%shsLMdZ>z*GEU7#JbEw@;I$L^gPd$nB!!E42(f9#lsGy9%i&Z&jZKN zEMTJ0><$}!V5p+I=c9*2Vswj-P3eUpKM?E;l79Ezur-O-B#Sg102e#;1&2)_^|RI2ZpGaFlO}kCRDQYcdrFhw*|YOH zf!F=s@KYl^TW;I+piTx?tP`*Hds^{(5&P-vD=$MnzYY08|&ap3A1LJ(iRq#d^?) z?SQ#~Cs|(I{Y{hbWy|bJ+>chNisrHP-JgOB2q)gt*TbZ!93L=i8HVEL^`4oauLaE=j1){43XND)UJ`?8jy2>@W6YS?yrQw1A(lzK zv&uX(k+e6bals2mAl8Si3WXo(Ime;XE5ktG=W~*ob`(!7SZl z$r^R~o!C2VOs73W(7UXk!QfvG2RzA%vPI0+-YZGWVE4o+kqs7yC26Vn(Mn?|z{@N@6J@=; z6ZZZP1WXp#vpS=6UPOyQicUfcKgldq0dcTY0i?bbW(A0nED3}$Sdu338<2{FDRwXw z0UL#yc~ZEd@`1)D3Y}>DK&1m+EEP}8cxTc3m*BpD=jXuZv*Y=^fO3P10V%4sk8=t4 zngeY1i05QkU=rIO$oy<7G`%~T=wf~GpUnio8DeDNQxQ0@z?4-s%t-xhBbFb5r#wWb ztWwk^jg#TZ5q}-w+0~UXh$MLa)-mhB8^0Tfw0T?C8lK8#=AnNX!Z8`C;c@1aR0iZ~ zt?6rH^7!t|1icneM~DhTt3`Q}uqErEqLTGOooxQ6)%7wMNHOArhzJSMtc&9*Ga$EtsqMr3$DWMuihJ|Ik)?fC zzj56cp3f(0u`)?2=39B3)fp6p^lmJ`MB0{C0_i4Lmgq?0H7yG7!4RMi50}XD4L8O4 z7ffJOltMr&kjDWX0HJ|!0B8Ue&=^zNr<@q6P1<*u2f#a( zXzK*S7nX-o0o?EsPJu$k(g<6>NJh|#KYC#v< z_`4`Xt!#O8!951fJqp#x5VI)e<@_MV{9IXbA4RUbshxj%o)A@x4RDz(!RR zc>07(Grz21peaqujM0yF>tu8%D#kd@=flA*TZyM7dlH~e3zcz+ud1ZSFLgAzTnXXR zbD6h-ml0g-UL=q6whl1sLgVJIo|5MQ{ZLnZSg3+{*P0{=DNjJfGa)bmssL5ISgF~J zn^MA*R-g+<%=ZLQG<}b=fYd!13fkp;{wJ6p$`8Dyp|Z)aDm_s7fkr2Y1cHvP9T)`Z z9sOopm*Vp|@bPiu<9UJ04Xzzj2Y4uo0HuZzF{bdr-q0Lw4Pv|3gr*&P5~x8?*sC|FY8R0fwk{~jQQE||eQmV%i+T2A-8_p1HR^hnJ(=NFVYk-WNG zf!5dfvcvvic1+V_OV~f~uT;hQcZ$@5H36SJpk2)Oaa|WaTg7oOq?x>?RNJfpe~`=4 zlpYb;Tl|n7-%~sg!vWH8aM2aogGrOkM90G0MyJ~iBuf&POn4LF7t^DIEiPS!0BwLa zi&I4bSMP!;E0Hc=g~Kx>u05x%&t}OYUpvBRdZlDalys-+{lKa(o~18Ru76Qb!K)U2 zrY8C3x(uOXGA(0t^8)pb-fy%T6B17=gIZXHT3<(H%xKVB8LbkoqvGCx)Ma0fF*|z# zNVV2*oQr)rjsv9{6V^N+Lq+VLYr$QM#YvML^lNnclrnu)=g}<0q7>I1(~L-$K_=51 zCMOwV;C6jCdr$yz@nJyO!Z;>~ngEKnV2@WuXA|Di?|Il(a$-a=&vY_c9);Ium;(r< z)W`^j$iNCO5&Iln7_>UPCXk{XNS){Tqx-FZl5sbuqfwgG>PtwLj~ZxIIIJIEXUI(W zFV}(?gBxAV@v**WIvD6WaM3JIWn;iWUUciR;;_^&(>`}-_v`MEdPCS+?zI%C*EoQ8 zc~dZJulK;j0KC?3TVpNMhfqOiP;DS;01hw>lnT_Fa92XDgu55i+R%>E3U&|cg?oED z@pv;PHr8j3^_j`bP**5H$YcXhs~ZSpt_9-KB9rUYtIc7E`Nb543C)uN<;@sfcECP{ zcT#naX&=r$+|lcm!qZ%al1W6OxN z?Y!_@At0^>gM=|qAj3SmItUWwn9iQmLZHUpNMjPm2u$I`J{gEIiJ>9@RfsYrCIT_+Mpr&l43}pU+vhE5+0%eR9Bi|d}x&mDj%q9tU;p#l@1g%kXDd* z+Fd7%TXA0l_jBO+d2iR})2s$u2a1Y^KDjrHy*5OBAVLu@k?UV%mXbztB#4|{rIc{D z4yrUkkgVyyzRaPvv}Brc@zLwU%P7O8vz5q0ANzUX9r-w^4YB>DPy~0*dQdm&#wvzywr&3i(1SFXBwbvzQ-XPFFv#z+j{ZZQ!aN*SIhK zyxGx4nHj`nt&JjywL^R_j6!hv9}1-Ygi>(B2tP5)ejx*w(G_1zg5O!y=_>*4#`}== zwd;xVDtJ5X1LI=-O&q_~8W%xmcsvT;9+nby)QZa9v!G^~8Xo;XdJh*bFwe80%SW|H3vs8cNVP{yzc zVZSkYo0Bi1V1Nja8!8RZV|q)76qN}_VcbW0$#{|NOQrItmy$l61%p$3(+IeAtpo z>p^J35HVoI6=5u<5-H2UCr_*2pH&Evu-2l_a zh0otU@!Q7}A0HP!KA!k|UikdH@cg{+yeK z7p2^7o*gmC52MqnfatoKpw-n8B#VNlEa{V<-Xhc%g}S2QUbd=G%^eoReK z>Dw7VcTd!7h{kK+AZ3>n?b60|K3eRyx=cDGp`4Ti2$|nYcj1z8mpN!Bg@f{n#s-|u za^m!R$^s`0q;uX!wv^00{V-HQL^KEA5ri=%`l$UclH3bG^ZOcp2NEImM@BbmXxosKxZf zyhH;u7&Outl6XZz61iYHCT(nGeh#Lf)qqqnAQd2rxq-NY%UI=_?$1!xEgCiMbpsT% zEeuoROevi?4299h6^80Kv_>m;`s9CnB*?nzO=JB^@25#m>XjnqovKe=Vv*!GoRzXf zKR*<^reCF@iplIqty8KnFbdpYYjftRS;6p{#@7kNB!wBMr|iYVeX(#Z-%vXogg1zb2ak9e?SAetn|lK@EVd6>pmOt%WP%Pl7Q2Eo^zwP!YAA&8DdnEmNQLp9TQ?&vQ8<7f#h))K1&dC}eV@@1Go#LR)`sN4sPaaVy@X_? zfU^88GWI0C6sYx@2)~7~o_cvnO1L0#8 zkcae^Ng%|jcQy@3ofGEyS7w;&BSb2Ge>1@1;Y{$g!Arpd;3Zc3uNv0>>9C27j--EH zoV(yt7|>`->-ZW-PXAPnus(b@jU{6^Z=n)2J6Gbyl}L^ICV|c(ZHpsv& zH+exJ@Apw#*ile;ppf9GWLG1!N~lbW!cE3c_~sGo{r+D1^>Tg-2OaoCmDpVU(8mKa;R*&_r#!8h2E*;F>h2Emg-vEICzq=z|;ucxdW-;MUQjZOtg7s;=Y#zqO8EzOy4n0KhDDr za_4E2^;(UQa%zYDP^eRH9;Kum{T?3O%0M+AOD)ag_)X#`s`J-}v@aG)o^reZ53lX< z`XY%AFdYFP?rCrj+v+Ey`MN-b%VE>KKY>&j%7baj)PDw29)n-KbhgGMoN3Db;_v(^ zBUQP)tan-5wgyDx9t@S z#l1T`578#`#eG8qKoMgU`^>agXssr3i_aa)ccaYyeN*c3lF3Hdf;QY6YSTq!MJZN7 zZvK~wQWb5Gv!)f@(;{nasK<%(d~+amqSnK%R#7UWaCpvS72Jxwhb8q{9Y@9K$(g4! zQulq&;Q?-V=aeVB-WNVUKdspEJbjNaAZ0@V03zNmcwyX6A7*&^)QGIJb(;(YIDEFn zQ~W}{me$+6M*L{WOwzr1kfu~e!AahenBR%==5UwXe^Eiy{IF+W!un~dB`Z?Lt=*-J z9pe&=0gPZ3hMmKifZ;%js`;8K3K}<0S7k_q8aR*=9T<9p#LGYBb_a!fLA;Mg#41bF zJs1%`4T5zmI#VCf?kbU3*H=HDqz*O*Q5uYtRxp{Ne;#^Eb0(@y<^Ujc88ex0+yl?g z%K+5J2R?rLz~BG=X>8Qh@%i*}=i7R$h%Eq$E2D-t%g;K6;q@a6L|OZR?DE911~eaPI8-Sk-H0zQmvMkkKmPhk{H<0?PWqn@>s~<6AInsQ}aB`tr z0#fl|P%eVkT_6~VG<^`Ro4~BN?h`}C#z?ipBs+}+wFoL($UtL7D=?V|t#Y_vpsciX zX_R$DgL95vR&Z=CvPWrhgfh6i&(f%?dkj`-oIB@1TmV#vG=QtOJXhbCze@Y|t9XI1 z^;PK9awH9~>et9?Q1Zs41&T1eFC2$uFRP6eYVX&ff6aaOCH8bh-6DaxIXRj#0tZvW zP4-c)%GNW5TIUHvRDeR>QeQ!GfV4v$K}?P8?HT%eb)=`~6= zZOZyh0LT_gzxR6?Q)E|sy_fO@G0PjnxVp_&WIg3{F2CO>G9OJhrBP!PrRp(Sin_n#u{}6 zLoo)=GU(y!YHbHvGf7`CQgbaUbpe{VZ**yYvb0ZX!BLBcyV$d2KE>+86laB0A!2Jn zFjA#fa4{e4aW=fYz2P_ys0#Xh;raZ)^Ya7OXSMI&8G|ty-IX9~&qa69| z^QEwe+`Fy`NNuJD+PObl!^3Y^?Aa|3m`_WOx9VGD8I{UKG>FX)OveJR*9im&PR{Ul=}NF|W+J{Bi^69q3&# zY={TCExAH4FaVUHpr+trT6!*Oaj1S@QbdwOFNe>kK^0udlj<63JzNHMpw!y{&dNpq zQZD|xA`x1CKW4T~CrScIUrC6UAu7F;sQxAL)o(!5GImp@x_DhdG4{SSJ~Zo0Th1X`e{^{Q<1JCufN7Pd??DF-u}BLLszTfTMXB zLRN1cmm(OlC3QAi!WnAz4zkFB))~&h%BH9KNoQ|6sC5NI>WNVnocu#9ErA zsg>qwYIa|m-|H}-Qx}jb{^zevUbhJ$Xa!Yp;T6r#D_M+|4#WXeK}$f&fY*}B1duw~ z@iS>(>YZ8}&WofkDJ_(~(d1`a7`@hIEg7vVW}Qi!`7m3#u0_sk(vEOS8gQdMhwO)@ zI!33Ic~2d*K|;Z3&6wID9w5Q1gC(TUgsoxcCcliv=Ta zF+kQT<2VY=qvCN|g6I3&f$Q_cxF6<490PaAdR1pQ0xLjXRzzae8kWB5nxEpQBhi&g zr9boZo4lsXubsbHp(wepY$|pl$F6{>q7u2J$PMpgp$Y~Z+Ps(rPlBy44xDWb8IZ1C zqae76QOiKT26`#z)w;xl6cqxGur-8n~VoCh_t4 z#7D^bK0feyb=*U6-_ex$d13wFUK6(tAzGey_d z7fn`$={cKwioPguUIr)n0-*2$#5DQPl7tbF(CqgrDK^k!UMbnZx?~BSvCR|th$axM zn~!dOk3O6-f4?VvQj+KM>IPHD|0W=%UnT*$c+eYcwunkin()Ge#_SQ{EmskaKPf7KKXz{n|zu!DK)*OxjBd}0sd$y$VRXQy3DSG9zG5jT++eoB`xEn3viXJdbohvmaEFp z$_^j^77OJOw)bupmWRoGf~B!gpdTQhJd?^quu7$Shx7rxO(w>0FjX0kRsoU8_8GXZ zCjEO@MElV!p|e#&t(Fy03(y+jI0_zT!~6RKqu;oD$9;E{(b4aYAv!~NR+5|prjSuk z4ji)Rbc*}thyP(%`qvs{t?sCm?zrTS_Lwa%R3(9MnMq-0fVYWwqn3cw0nk8WL8Sqv z?#Z7~@@i{p9|NTm`pvj2qZS{Iy4<&yj>-%vY$dclFtMx!@5NzH1yf_Kz9gh+dcLOg zc=h?LV%6l4_j3bNfpUeqq)^xvhlgsu#Z+v`S z_;_CCi+Vos@%hBv2fAPeDr|D#J@?}!YGZ*Y3*+e1u0QuIn8c9?HVuA>=~{y+XS9nY z0JAcxnmmd&&tZ~&y~@_0KUy<6MZNYEsq*E|nm{W#Zs~>L3H6UGYsCigbZ6x`4|`t% zCXfIFy(n-vGy9p1B7AvoH@}f!_Nz`1=RqaLJQ)jhaKq6KJkEEV#~Y6JKr1H-HxP>T zrgs79?H!)pvE@}en&o&|s<%nBmZc$NefkAR$=1U)hpXSDCM=KJ0#oKgjDd`I&9S7w z`WU_rS1=HO?weEUZVL~*CoIQ+3?5o!F7kK=;+`t={r$dS?7tC6=|A~r`sVff$nqr9 zA(|f?Q?`@ulob(7H(ksmTnk&0jUQ-@nzuK}9i8TR20!F>GIjL67{HaLFIxs0HJwK5R39bTsqvwc|iL zt1(jt<7gJiUL8c00<_9_92M8wfpMRZapSr=u5zPv#uMy5Tr5vn0+;r=xtC{%8K{ab zAoVBuP%zJcijBs!Aqh0}nJB4k{{WMwi>? zU_-4NofP+N?_Y5_Zed1e#wZ1&3}`h6AQxUePUI$bHXj{JTkDgj?pa!53SD48M;PLh z*Nk}2LX4Z0%kqDOG?&@@QGPKzE!A1}=KEdSe7z8o9KnRiVpp*U+J*N>&oIS=r^oLhCzK?o@}C3`>& z=;A{OrQ>AcbGM;`D2erzo$+O3qAC~ET5;3^M{PLkfwP(i^k@$QQuRck1f+C82Bq_g zq~Mg;7hmEEUoDmhELX<>03ZNKL_t)V&w29l&#z$V6>?(hgraoTrAX3A@;<1oK_-;& z^A5+LolKsmJy`;1FTqatw0XNM>4Ip$vDHjx)ko>|0(h3x_g$s^)UT`-%sCN zMALvf1wwHsV$6rWR+Hq|Wiid!h&UhUTxjD@3cmy*SO*Bt^rjMrt5(u$u^8QuTzTQ0 zj<|*$pu6C@2A-FJspsWDYHmoj5S9Y8W>*9^6bDhkMvd-2KN~Y;hX|nFjCzPqf(L+V zS^HGLJj@rRx=Kai8%@EKQPpXs z;gQ-W$=*^QU|nc>6Q*>J_Ggahg`$e*g^>~uzk;bbxYDH0+uk+zXt6)*tWdTtADiIH zN_QNx4kl{Pcx?ux`rW-xDLp%rp)BjOqE~CXeqpC3BjqWMN@}4CWQQRVkLwk&HB&I9 zn+uTo<~p@KW>A&^NcS#3_oCWbV4f?H0qr+P1_}hN5RPWar-dy+uvNz02bc?Li-A;` zopHfPO(4}6=UMS~ACP`Rb>LZVW6>`AzPQs$AoBx{=E=6CQ<>~w>N{`89|BS__M(v4 z@GP|r0o{uFj^g(ce*j2flX9{!g)`VGGEY-=Uy$-Z<$+2Ag>Mjbmz1u`k={F)fX0rIko!L_{rNMWZ)^poo-ui9G_Yaz} z=t&mmEGb}tAV8UM=e0Bony8kbCeDHk-xuyX&ITQy&yJ4?raboi^NEkoPfI&=*`@kG zLSLGjL|01OOlTvQHDxg6N}2F{i8FHy-lFv5pL1}PZ!a#{&iv5Qm6B#ZO|TyF?fLMZ zt(gl*<(U6(4vpJjZZSL9dXpU64c!3MChgM{ytB!VmjZMPTX@AHDg767Xf^4sxq)O$ zgXRwr*%L)GCAZav^EmNnCmzR%vp#UNH`D~A7}eKg!cc+Az4qu4gIq~*Z~LA?c#(B# zB=+eHo(PZRI!j|kzGF!-HHCx_?dL>GZB3txhf=Asof8v-IN5g{OF04Ypp#W9KhH+N zL`j^Fy0cQEYPCbq;a`WhL}zq3W;z^@ z4wR(sfxwi76lb+o8k)=ds>%9lD<~Kq7824v1%{X`#4ouAo|oYHwAasT;J$Th+g4am zz$u>C<%jIyQvERkNMZJ5NF+UxqIEWQ1%d`UXcS`829^dY{vgou?wGU>;9`=T<_Dn~ z-Z17ZK3r2$*tgQYe7NH^8)8-v%beETX=Rl`=-E&oGHy zmM~4Ola-3;wob{`79Sk4($xD!Iv~Cy5&L~sc*Y$sMqDw(4pvj@l@^Lv=tyNH$`QQ7 zh3)ff2BM}^Eh5Epx@8op`qhS-C!s`augf)G(qUGYhlfmf5$R;Q{6K5J&Pz9)#h}R~ z+Xh)(+IO5M9&Zo4zrW*rd$a5Dj`Mure4GZj;#mx7AMMh<$v(#rB};qL#=p9%c!H@` zZTdHXR4q%gl35=B_w)M0K+1ezJpXPEk9BeNVY$2y z>mH)bT$?A^eh-LR2$=xF)=sBGb_O;vQtHW#?L5&OL`mus%hEn(o)PC?AM8BpO@|M^ zw^c0hFNz*DEQk@5;ovV0d=p5;VL(krsvbElT`etbD5RVGdZgfrA;C!X<&&z`om1TV zbhqO509MFdngv>v!e}AyD~pwCbxHeloPdhUQ~i-2x{W@;=td!Dq_~9W)B(Deo)d60 zqudUne2utEL2q@c1=yz>q>O?(`)C+yzE3CyZro0JDF%k7ULwu_74y$BhRbDwL=M$M5ln0^%pQnw9RgO#$rsCa$i^ZE3s^9u6LF(h_7t6C&6 zQb}Ww`nCpS0x6jRQL{QlO`%|C1c7TOU6va@Jo`L?p~_yE&99`C_KBb0ms#4US(;7X zcc3xKK-v7W*y?jy zH96mTKJa$F;j9m|`amfsT796F6HqeCh_7K*CE%)&F(#gi&`V-6&2@9Kmm=He zIm>pPedFaK{CbAV(ms5Z_L2H9Q~azL%$(EUI)mc6|5;^vTcr|YOxk|s3(y`KpnJpk zk{S`<@){OG5r!zEM7bJe5c_*nZe}pruTS!$FJ354 za*>%rl%jaOv~?C->k6C5<0{gZ@fFd_DMeEWcVw&k;i^PgYOzC}UzD!dMK?n}BqeGn z_6`+I-6|IX)B*-pFp_OVGam@w@Oct%FPC~jks4{jsuKVtgenH(Q z)ry7n9IZgK&Y_aM2LYn&XFkr;82+=G%#RhLSeolm38R_^XcS{!YI=sJAJu&mtk|T~ zR?Xrd$_s^AWtC6;=wLwGkuldmkPlhd56_G&AGIRch_bI-XE>34E8@~nUENrz115%f z5pbcAN;(7`9fb%NmI@%AA*Ka{+HvAgLUmtM3F(g4GuVKs&MrCYbYeeTxsP`;xz_bG z6CSBfZ>2>v`@W_rxFYs%&97elz#i*NF;IdrRuhYnc@Vi;jQ~}cT8G|CfVuwTl(zgtaD;65pIg|yXuc+q z_eZYF{pgVt7ojiwX;oQ{Zum#LN! zu`@^j=hvjy)jP&>&!A+3SOvTe%;tLS_7sZJU7a5J9O9Hg+Lv>kCKK?Mp5$CSN#q}6 zs29G&QjmiQo+nbdoo()3p`mGT=oXj6Fy}~o{R?o+e;G*W57ukKtF&v;s+99)wusdg zeMnC4ZKyh;z9;UmLp&8rJwl*nNkyPRNjgq3;AHZ;;&I0Yq`XUp%be`*Obu36{Sat3 zHcD>m;L)q2Ujv`dfzM~h=hGOe>-GZMv=Fz(eh3X#%0N|2&Vw>jRg3}L1Ln6HmfojK z;HD@I5V6B|ap(AwDI^eu6~%6G*EjmGOPms2hh6fb=oN*Usr!&mGUenw-;^D|MzLqf zU~5Xol5Ap@`2BS^WWMiWvA6fw<$bd^siWWbTy&z;ca`j|pEn}H4S;oA*KLP^=!kMK zg)#h)Gr%4;Axqk8hY^ZMR9&G4N|0g)t?I8*N(4(mV4fO8nonU=FXWBbaGr-V>k%T_ zlf9IZgo123T%47ub)XhTFE(T#7jvsRO&ke!D<9R&OuTBnj(dB1~M%92)#&`=N{(q z=kYaVqvCE(NTDz^|E@Ktc7Rkpn!Z(kZQxqd6@bcpD@@xEXGvRhGeIxVkUIc*+YgX? z=`f6poLEYb8_Ucm!Shv>JUKP16`T$6vhDls?v=a%=^zph7b^2-nbGPZY~I4PXGV;De5O{gOn+*WxgBxl+?jwu=IJV>0vKBaa>xUjt?w#oKPHe$ud@s^`C}%d8xr-+`JKiTX84;v7;pt~ z1-Jp_29;)CN63r6mk3$i)_Rb^)RL;@P9<^`&Sa}J=cncpXH5*Y_x-S^T-w-gS`k$D z$a~e~Oq!CoNRMYc8oH*KvR!`J=qI$3dqk7!F%N6}z z$*O_1zKQ#oW%&a)2{p~Lf~kjizu7{Xj~9l@yXW$H9CiWBXU3j0P6fl;*+1<+N=f=Y zYl(&jlWI5~=9zYd!l-$uJX``^9v`1$q@WtWD{Bh*7yjo@0jXD}R)6Q~e~$R^^1IkJ z_|VU*q*%&aJV~=_38_D7$^D#>aoL^7N*W{DH)1BBcsfui$>8eJrs~twi*C?Fo2h z#AVX0!)JEaXaT=6JShMmrTg~(&-KK<$Y2&$ zngm6<%*3thdExndx*qw0j2k*UM%m4eG4zsU;+DY-PK=Vm?E9!!x-p8A!3B$}P>VqE$gJ!z8S=ppN14Kc5nf?hAHiExjPO zR`X_^1IHLHUu4`DO|z@Ll{rj7-S^aR9tR$e2i||ZXh?-UDHs^F-i)_q|`uA;qAM!LeN>VbN2aVA$OYFRF zJg?g&z9w5`3{-#rYqL}~1Ke$bC$8&foxZ~Oj8xHCakHW^(4C#?MNvsnNaq%BEL0Rw zuEPw?3GM<=Y&upS()@|Tfu>HN>6=VlZA245MVAfK%YogaKr!oemBtu8IC20*B;c`z zQwBzs%qDKyIr#AVGKPH8OQuxjVbOpTPcJYP_xrE~fRs|5MJO1hSpm8nP?z$POZh3f z;;2qQ#Z;C=77>(^0%LSd)WxNVn*5n@AI`yBq%JZ5E%rQe1xmFp$9e!&K&1l2z7wpy zDl@f|K-=;dF90>%heg>VE{nZdlPuH5nx7Thwbtl#R!P8dmN0KY4>X@eyKjxPCDClg z`FYp`>nu`dGAZhSJu?CYt1(llV#sf87jcm!&s5Hm`)Va2-NPp+z>g2!C%firBs+kL z()cpv;F5q;gx}B_#M8plSpsLisD@T8Xz4uOa0VMyPBf1?w`aBh!;`49UwejotYl)_ z_xb|Dl!%Iu;TMy3GRwDTDoG$pR9vCG(hEQWb_G-^e%5xCS3dAlwu^++FAkXTAq-F2 zb01&y?b&0**+~|(0A{KC95rLIPz>GEdvjW${x?ed^qYVB`oF)!4(#)=q_ z(QdND7U5MCh?o}gKFOK@u%q`IyV7Fqqq2>hq+O#S^Sfc^Y_fC`V4%k-F*zI!G?r35X}#Ypx0wj#ynZHvP& zUxiAavOWtH7+`a@+nkm$lfzVa9tb(^aSx!7VjyZrK^5^Cuh*CQ0k}nZs9)uUiHJZI zPTlpv%prhcDF8bYHRHb-qx0+uv4u*P{4)-}ey>I+S!)C5u_Rm?4BqINCcw;g`tx<1 z2hPWdx3@REzrW+{*E`<-?HAtP-|={VU&m@n>%~i~u?-p=NO>Pve8yu4ga>HQWWfx? z)%mq0cGlEW^$Mcw{<%x}(Onrq55capFZ~q5eXK}&CQB6J08SAeQM}H;zvjYIex7lM zY+81#19?27ktXGb^Vz|XWqsQM8;_$8tsFF$xH2R44J$Q)I=hZuBt_-!=Iy!9=ecl5 zsz}%TRu{`Q-+*Z&)IIEl)DLBr_NQPoV5f<#^|##^_l5hqa6O;q#qswtBG5}^>u5U< zoTtb12Y~W>h`D%2i_XKx7=q4<=FHTc7^&(#guzOwKdS+zHXnx{V520NA;vO#IwAc~ z*|%_7N*$Z`SNNb(A!v>xP?wBr3@v@gARaY5hE+L0FcKrRJQO^Q>T@tT1Va^IAMz1l z1oIUKWz{7QmRY(T;3>&bgON(frCE)! zsxFW`nXVLxd(0VAa74-bCRfyS_@%j@c>vs?RG`Gu>#1ZNlbNO6JyKa>zq3-Jt8Yq# zW)Ik6cHRFJz_I|PP zyu#W+)C^Sk-2p4TCGhB2YEy=g;4S@FWr2EOw`5}Ra~}ky`eiz3oh2X68$TF z7y&6nEr{a271vWys^UH^)lUXPq;g~YZpzH93Jq2_-NZOHz+O;SX#!*(C;ce-(3_w1rEVvxOq zE5+5r4@NHSyKZNsR6x=Jy-laWyQp6C}+Q%j*tK#JBv z---dLqZJ&rnmJ!(J1K2gRTq$2G2jIhQmxvpN}v>s%Bx{~PPzJGw5I1R{80yvq1(nQ zBA&*@5YjD`jbf@-D-LI*-rnEv_Wp)n@9%j3^~>3*_mnZGEA&EES+dTl_Lx4lAt?=v z2L`VIgOFbZkP7K;$oe7`aF!3Qyjzr?0x8um=$?J?Uul}oN%~q=3W2IWKP6Z{2U6h! z-GG$NDRxZpB2)#@`Ql%b=nO6+)pT&+Wvq`v%p^V`&<2zut!`e~XcJVHJvYtA8&m|V^$YxW*MlMF^?x$bZ<)yIW?KTVpON*oI@ zS}S;aG&~*$-rgQ~yuEEeid|AW*Hpjvx`<2q(iKcpx;XLi6ktmAi;83V-5f-jbn(Z6 z`}kWnIp_ozHDpvv-ZZE4QOJv8J`I=q&i55ScP5G<7&I_k9isEYD=UpI1;nTfvpM*7 z(9@Jb`LQOe=IygEN)ll8&Zc-z~;56$x`6FjZp7EVA)Cj+ID zpS9)lx_U>ajy{Yz%XI~Ff=ihOGKrtML~%*`5{SwvZ^^iyeNju|7yq1-WC>f;b?i56 z(qIdpGHIVW)uxhT)|0$ecljc1a?KRBAest(|9C!sR3^m1x;UUMehF+ZG4ZBAvu0^! zf;bDon@@+lH*ZPxG^O;VITF1DP!O$-x7bIbRwOQ$-%mfLl0Ba+UP z(*FXG$~QyC_L?TInpup01fY`knmwIoF+vNptb=-D6yhKS1rIB)<(raxh`#MV&Wx)O>Gun=fOfvWkLIJ!Hq3U>xG%`{gxq#%zkqRD6svX%q6se%d>x_a1M&mPly=hU%W`HVA6R~LYi4OR z7DE(8yo&)EtI&Q5&veES$Mjo=F-epYQiwo<&q-iUo9uj1 zNVVaZHvf5w*B+Er(o2;5fDu5Fq@QN{;i(E(>#^^-e$`f&Y%IJws!18L9!~N5tgt1d z5;4x>#MvI()%ZZ=hC+2Rzj~%35m;J2q4wKyT z0GAWYCqM@v3>9lumjP$bGLzi0+BoT@s0v19S0*Ro<0Vw2scK&m~#5jo1C_okaE1^yq z)Act103ZNKL_t&_74lY(KW+z($BA~v>xtuRwrm5*>}I412Faw0Ip;Tg~OuFH4R=A<_|lL>Iwxn=Lai> zG=U#Cv|rHs#<*_CsMDek%MP4e-eC{ zM%qE1CT87|rWzT8Fgh8Cx}NCQh5LHqzMkm!g)wd~Pfei^b128o7_9d-r{DswZDQdR ziGwWrN3-^Ku-|jAy$LD(FPB!^kT^9j#J0_=kIx>)*z7}u9_!vu<FLosh%z>+Qu7k)veF@Q$#4QsB4htLc@tmX+YL*&L=|F85M_-c zQk1<9T4ln$9~)zwfF012^VZ1Hkm;(AH@&jgtQ+N) zrndowO?0aVGSu48nwli0G}J0+O~72HT=dyKQA#z6J|#biELI6=6AK_wJr$|O2KAz` z4k=Ac%ZOx`VY?xpxmf%@?+jE-Y4b#Po0QL?k8CZ5ME?b$0Ii89aR4KDunZ>Sz%_tW zo`kwbsSvo*T&Du+Dp|CDAL1Zt0E0H44b%+UjTwy@2QdydCaNM`y1W6r;FJZ5JPsU> zM*>o(RYbIE)7t6N`@ht^>z3Tea;3S=+#Lv3&-D90Xseh3xSP!nySO2kM@^R0>-0z~ zl&n&b$&3KtcG>&eoBE_d)YHvlrkaYWrQ!7Jv#2Te9-GX2VUj1OL?NaU(}G7MdslY_ z9o}Lgv7dy)d~DN#)(L$yEEyJKAj>DOE)^7UnRmw+LFvNwVW>A$Lbi!oH`eurTo>m1 zjdks<)VeR|{k939+*WjBWi!=$8)et?rm1oF>!8;rYHzXsG&>@S-<0Uq>6kaJQp0!r zXH@5gs!zz5hjNFc_JQg3n|dvF6^8m!a2xq!%MzOw#91sRk#2LLPq8LyDegsvMAiny zurrR2Ef?f1N+KWr#5)sPs2-6Ku1wkg{j%no{4;ggBLq~T-R%Pfnm)tWCHUx zY1JgiBwIu-!n|(G`;GgyChC5F`BrLr1c^T6WhmRI#}==^KG)alX)xAHHfHM$O`p9)XzREYZwRSxffUB@aU>6QHIyOXM$zJx$#k+K0Cx zrY(xc059>VMasb?oQ^&oHu!Uy%5x)Cw+P!j`)Hzoy#`+k%Vid_ULF+NaU3UJKYoCy zZ9MNZ$wvIXdEdDB?L~%#Fpjg#e=g(!G%C_I_k@;SvA>#J5u7 zQ+Wav`dr_Q6zu{=2@ot9rW%AsHz`h4jkINb3Xy({dMQ(;k`_yxE6l<9G z3`DrzZ(P?K{pUz>8;LySx}RI?GGH;RtPNLc^FmHgR8cpw+Z2q}Zr-hN;`BL)QaSXW zjr?;n^=y{tp>g=98!3Hm0jhs6HchQGP(Jyr8!2*`pLwLh@LbX1?zdaZM()dt1V(A1 zQrU*i4Wl&-Za7-QAU3(5I_Fc1Nw*kw`#5mC;&o!Y>PE@}YjY!&SqYT-eW)by@mmgV zRYU3qMx+ESA~RGIsm^Yi-w>cUJ^A6UNeN0wC)waaAVM3AzC5gS?fY9}(;_S7=N52H zk8HLL62O-A&C436%U*I{ko$t(mu;-pGQ7HO${hzbJhoD>VDm(hXI|nyJGtST46xrU zSIZ_No2fv2C!JNY3Fd61*yNDWNUfD{7wB$`XjzfHq*=C+7F2ZvW7o3* zbv{7i=$ToPmTQH&?Chzsl`3h9_A_&(A8shtcFS{mUb1QbC!132fcM7DHc(=+y)OD} z6x?$hGG1-Q^oT}^AlW=+K<|nnr{;zSs`>s#^c$~xK$dpsvucSc`30&#kIG%$?{B5>(?GMg z(tCiQQhuyxst`?7-AExL74rGb$eIHv*;G`&Up*Ewu=PgD@)G*S^V2W(VzW~1m1U=Kt4cZ9!3QSF^Rci%`k!Rt5%$CgoHmSf$w7fXBmQq`76Z}Fs$^)om zekBeRk6^k-EA=Oh6z}U26|t#hfA&b*cFD?Gdn0wF7f+VlpU(G8Hd&zrKFrc-wZb zHNy)Fixv5WE|jMK;+IN!pEqn=4q_t8jTU7iMb=dL-!p=jxiMJS(XQ;Fo4N3CbjZ`#`Zo6iHmREqDSwJQG1qw9y z7PEDuKDM5XRI_0%T1c*R7ijE0p>Yn3quXnAlP1UCkq{?xA;QH})8YUido#6VTu<7b zliKY%nXr>9o<#I(|i{gjX>HV(-TbapQuu~AL61~~iuZE*+IC=gJx%-my( zk7=)}OZ#qXoN^Q8P1L-$N6LXT;g4#h=8JO4V1$^54829O-FR;%)BOHlSdrFEMy124 zWN~R>+{?BN>RUZCA)1;&ma7^(`aq(IV$f<(>E8VK9*Q}`e?z*()Ud!ggpFp(8d-a; z*d~e|O_a&{#xuP7v5^WgB4zVbbG+!GV6a0he5i=n0A^4v;))Mu|9fjPJyzGHayEF= zrQS#Z^8)T7>ASBl+*jF1t%x6?LhTIzRV!eOVo~5?Svm z)LNF2*h&(HXWsE{a>&70TI7$X`aXtu-&l5^^%C^sGpxR3&BjBEY9O|PV2fg4 zF%S4sI{=YDZof}LkgX}jQ@`fQE~#MWH308nIo8IZQpM$xTAmF`_-9|>wa}!Y%feEFHT2F&*X!N# z2Abc1Oa*V+nolD)xyK&nVu!H08Bum>cpR*AE?n1*_xl2A_n0P=Juc<`yK-l2!;TfBrP z$a5p%)Ivtmm`D#>c<6O`pxnV`kryvy{05vJsOmu)0M|vJI;Y>l!7v8JCiRRZB915Y-K-wUV z+*M|%7l`1B$}eee-bjVy?l{eZlzMtU;N}__!{lCl3^YG*L>44Mj$;5)$dUJbVcyf^ zRdc3P0g-tXL4W@H?b@TqCTUCe4Ab1&o_tY#j(v1dn|effPtZ3P;-5Zzm7W_n^~M;P zHDqt>6U+P&ikbkET_ElT7``RgTZ4ut{gh}+a~jMS6|YV%b<*H_`jbPq#Dn7ScaqYa zH|XX7fdbfs^MT%PEPH1U^oWjuoQaUyB)o6%zE} zCEh5zxV$$9jo8^m#oYTmWB;(hmj#vS({lF_uk8!0Udp{a7Wc2MMA1LIPZVoVlZ$D{ zEwMAvxHFnG4=lFWgzhh03ro!xt=w!ozyYD^-u#c;!kce~6EFRcX?Zikv?-{)mB%D+ zAVrleB``}13`uszB^Oy=HOo^uGcv2L56nd%w!qM}#0RohHc{8xUUB|d6J?n~tqrt( zpdbDojRRvee_5^5yZv0P{9IM8Y(of5xpMVL>BC1HR2JQm-gpMV0*kx*bs7z$z48E) zhYOcxiAW#CdXu)N?7y`s_;hpA4xRED z)K5yHpM9>aGmfLpXkj#m)?dC5xg0Yy;(=8m7`gWtI4+N9Y4m)kmv2d z;DPSIrah4I|EsMuIa>a?f!hwxjqJ<%0Y`mTyVe+V>n|!ZD)i%epZuC9`zYFIgDW&^ z#;0;kKb0EEFLXEx1dNWxg+>UmVM39(2I5>x*gPt--G_~$l=r&mj;;DPHd5qduwNUg zAhi+X@!=5>e(8ZL)#sy$+PNdXnhh;(n`qK7R4_F{x-^@o{c1@bs)2#6Z6csO=&0PJ zZmt)_V&B#zV%g+M^ImvgH-3IyHdP%AjFyPS&=Ff&ycqJ+P0S67K@EbAxd`_J+ASdV zq;)iiGL{hTMYtznPBMSkWUOWWBujagRE!2jb9-lrk5G_l_+8>xb|(w&hd zE}@JB>o_>!2Rt%(x8N~#hroRzjsqRtytr*>jz!f+;3j>yo%-MW&o~Bt{N;_5lj3G# zh-Vpb*u-2#^C2R#Fqda(%^NH7pHLbpAJ2(+fyWG83?akLs~Dk(9j_a(6k$KPv3*H# z=my2CefYwxJ0FgovXIz5HC4~Y@EW!Jcd_O84#O>CpMMloRb#Dxm_MPsi(K8I zJPDnsp}UV}Dk)-d9hSpFkt*=IQDhag^dM_4Lnw`T)kGDI(90`4>-HZZD=6 zPD`H|D35^KEf;B3?ismW#{Do~om!?_=91LSST<3T1zF)wJDyx@!{jJuYoHjbDTY&2 zV{k$enPS+5h@<4W*;fRZCfdS}n1|oKE|)dM@Xt%UdDKOZ^F4-Q^3kMhKbxeq)Slxe zTQ@*;^G3?IDh^TU>I0S|o(~6lKdfCr1*^{QX9%#xDd(uXzS-`mU)wZNuM>QU^vqG5UcqUKstbTFD2Blo?6i=&}NS z;gdDyjx#n={lsYANRfJwvie5Hn}mpTAbo6))Ze7X5mlQgeZrpfshG}%MZ#2q-dQx# zs&#Wkzcx{&5`9v=E4b zavlGx=BfYEMk*h)v9a)N6kO&N(;$=w{Z?O=06=Vm!lR*T4FpAZ01P(*FLt!3V(11a zMgx@%RcqTg=@SQbKS|KWY?4oC9Z6ejK})-Bx;u42opZjH;$EH!)o#!7=vI9Als?R} z8mt_{{eW6SkCmoQ+*8o{LTd}mw|i-9TaZ+r6@(vaqUN&0(>6uXD5*79b-GZ45CZXw zYA7ry?`rn13KiXfeoHj{IFT3)_c-kkNApeTbo}z#0dVScwUADjLAaYynE+*ZtaHEN zbpBo=H3t6r>n{KXD7o$nbl(|Ly1|&mMdeUK0DPNaz?sQ107z=N`Oy?*=N-?7f3$Ow=nd zVp-uI*XXI!Xa7M$7SRHGPL4fy&j>%7 zs1I~|dX^Ss-U*Rdvi4-S?VOsf-LIvVLBi2I!AWK!W%KBv*fSbx-%Q^3og3%Y-4EDp zLqSzuM3i#CwG73bazPCmsT8?yw2iyiHu*BAv^Ack8rw^ak1q0+q9}2X#gc3icQd6$ znquFvc_SqZEcSv=?S8bwj$KY>en|8NDuWS?ZB^Vu|B&NrP3Q7wr1KEj&mww8jYt+j zHI!4*Gm(>aQZKA;FZQ}MQq-*ZaDSzuqmNad5%v;)??k3;qKK+Pr-@>DYF z_ElxlrW=IVzjpRUiaQz&G#(%{D0P?ZHHUpv8uFfSvHT&mFFP<(4u#*THhwkx>?stB z0jd2yVNFj|!3#e^ymJt^Y8;2;rk@*O_%_O34PzV@mfI{icl4JH5#oFK)(a1RXUSnL zp;v^4!~IdYk)m!s#OBXl(uD0`8l-wWIqxOKnM**8*bbfO8{N-f_?NpsbQx<-8W{P<-XRg3`Tt34<}N+pj_w+F&M znl9f+l?dHTY3FP*0wNHoW0MZfxg3LD5S*mwWh`X_g;u<4`n;z9pknP~q_l6XnFVUd z{9;-X`OHF}(mwY{Ic2H(lggUTjh5wKq~SZ-~n{=iJ-K>$-7W7q0h(`@W%o$+_HVQ1Pz<0j3ks`uk zYg$I*O&^1L=7w(-#5Ue&-Gf{0@2PBoq;>H1Pvp=VpGWvOO_S-Hr5FMf-$}fey6nBw znc?45yoB~Xc7y5A^ugK6kWY>c#}E&4j!@?VZ91TdJA`j1pf)Sw`@rY}V;ni4Y3UiF z5<2c)JcUiTi4qhLzP39~%S;G(GHV$ZfaZk~iUv-DUtrTL95JJ-PS1 zaNlJk_4UU4>&v%M*o{G66@Dab1n`Y**)i{jKyWkFR zlA;2bkin=G3Kb#6jT<@-YhZa85!g45R`Oai#pUea3MQNFPr9fFoWd=P$f;8(`&ho9iVe#x#CS`|%#*Y%appA0#BntHIu4WK^%u_ZmwB3W zH&Uv*k)pDlqhNMto|B#Iy`R>Y^#h{~8}=9PrjnF=zLw-V8)h_r&e=ddo(c4jt?x`f zBK9V6OS82o$G)q>^gmH#5US!wpRb3yNWZC5Yv!5P^x-SUmZT~qWt%CPS9$nwNwvrk zt@#v}Rg1G#B;#Y;bH@LUMyim0qP@!pO?gXxW%_+Q#Pt&e5h zc${hNdi6IlJ(;*U) zTKO%vbfdI{oWNc!33I?ypsQa;`h2`R%)*iDP2|Y&mLZfor8Fdd9n`u7RjNr`Q38GQ z@Kh%!YL(X#GJqufQbE*fgf75P#XEO8drPvXm%(=G~snRJ`xY;sWNq zWnC%HTq`r6RG;#t&*r91hnYFOZXp8;EUXO*)}s)Oze{Z1GZa&)9AMbfyx*otCilE} zo{*Bg_;=x(X}oD%ZTDN{=-E`!Br{_0QJp%w)}=IOd_3%I^6yx%vj_l4`a z03ysWtua50G8!^%19P}{D4+I_WPm(>^y+A=|X`256$`lWMxDW8`P6HBuI-fYzYgd>f^Y>1*{Y{PuuVmif7& z=w$ZufUOw~e*!%cU+lfq3DF7FmBhkL(SkY9T;3;`F8%UBlW)zt`%I4GQQ#PJ001BW zNklvVszK+P+o)P= z8StStQlSphR0RRdV%ehUk1A0#_z3U?ej)8XB#IyG?&^YiR&x3wKBf;Nu{Y2fbw3_q z$>M;D_`ecB7TwDWEqzIm^ElLT+gRwo#U1ruXr#nZr@)p>iEg;Sceurxz)*iO5tEq_ z5hhJsI{_531g!-T&uCS7V+L~530BPmO?NXy+sqvwpa>SoR-znuMi=7|WBTMP?!=g4 zL4e^zG$UQLqRWoeH#r7YHqIo8R1T5wk;;7w-q*rC1-AIGK!N*GeEpmtyd6hwQ@^#W z0GM-PoQ@an>{BQQX*dpe7x(t_2U^pnE)#3sxThB?*BjS$rJTkxq=MQU?V6cA;KrQ~Ia8^Y362ua|in?%Rt>WWEXW`UrqkjN`;{ zoH#;v-@DW0c5@Xd+LqoRhAs=_w4&Fw_91&(F;%~yVuj&--_{^30j3}9ZZx&@n~tUp zQ*8p(WFvtY%~fEWFp(^b{hFnZYdua7O|P2l&xd?#m0mu-Tp5ODH!XxurYZ9-;9E5& z@n*^f2R*XmNE6z^%$Q@~@FwawPn@rpQK{}X?)%!3#EQDE@QGys$RYWnvbnY<2Pq+4 zYlB7VvxmxLlLO4+jfMETPan;? zZ9+%4+`o}Ut{;LPevIi7Lv1SRS z4e#l0do!Tcw$_~Z?45o0F&xO?(~V?P>_~+iN_}hE{>_55>23^Jpfn-Ix`5Iy$dwlG zjh8^b5e-nO(w!z7UkH10Th!*w)VgK>4nyqe{djgHQi^Q3Y^ z<{N7c%&X&mGcH~jY*Q#=fu^~7Z%OUU&X6$&1c}3VoRr!XaG#7gy^eg7W7pSm7}0X+ zSMz5l+)rY4N+-7GXd{R+e*86YE#MYQn$pZuWFJznKUbb3v~isN3>_W@ZK!|SNZWY> zxXcl+d4_0&_)1*o zE$=BMTZ#~t-!t;du$g0{_Q1%Bk=DqOs^Nsgq zejS_EPQ2eYe*BLE=WF2jF>wAEMk%DWn?+p^y?vv)8>tA7HR+V;vfuA$q_!VLR@F{> z5S)q^Pre0Hs$isT6{OvKwWeu&Ut&!Zd&uCg*W)bkB^xOwpf|!XOd7{-T#j%ap8J@L1p?>J#xsbQEg-sy5=Fr zh9|dXip82KJIbutimhqDJxtEvQ(155_m;TiY6NaQi?KZo*2o%(?SlfO9V;5C6+R61 zLcvz(tc<2ciCR9R)qoMFB>UY9-u1w!5rFrZ@R)S5slOX1FIX1l5hq2T&e3nYNQCj5dh$l04I3jr^t`puWiQ%V?_Ga^a}y4cN>T`Pn4Q0VaHhsvh@QxXyq zKcxK2y?DeH#v6@PWB>VFYPiBTt2>lDxK89%%6&& zOfOQ?zbMX~Hse|p#Ia4TP4d3p%Mgp6-eTHTyJr^LA2d>|@{ph|KPzc@P_L70AE@Et z&#K*@qTV(kj1q`HDQsKfNNEuCXizyEr`>Uk7v4w5_0{nn4OeGe@8)ocZp1w-V7(!e zv95;sCR`n8m}c}F9#+j!G9f+J_N?4kZz7w@16%$RA9j%4BCig*G4SvW5l zC)u&XNQ(_-UiwRHdoR?0c9`dl4Ga{6%Q@F$OJV2QksE!3_cYtOK`cIFT@ZbP=>qt& z>?+-jlqh@lSJ}Gfvvvam>>RL$q&cC_8mVp@DB7bR((WD{F4SfE1#IwWokV~8(hPKX z{F)`|NA`so6$K8gX+{|)()QqPTGDeH1>KZjlfzlVsor!>nV528$&D%YZ0W;mz!4kL z6V1I$jXGFcW*Eu8YNTS27qW8MVYfm){N(NJ+_!xUQ6y=9Et|n;q!=n@DX>_OvQgD4 z!J`dp3i+G+{UoWiVar&*G*S?KNH8f9cYzOZiT<*bdLVI%`f0NhwuLqGvs`X1Bp{uZ zry7uR-SsDHAy(6lS3ss9vV*QE6=Q^dQzIp7Jqii=D0Mw$OL%5#9R_arkkFZWRjWN8!02B6@p*cUYL<{U-*`F zxiez4oZ!S>XghpaWPu1+Os*N$lwM|rPpnZsq za;cCtQ@tN1zi`cc8gt{oBt>UncnC>uNZAf#@~QWlF=-!AiT8Wrz6JO6>u-lsI;$MgxpxQ zh~ji_6qmgN<tadgHpLBckouH<#G85@EvX@ElRQ zCL8Ba0f8(=6#h=y)E!U>wJn*j?}t3-T170QPux%IlS;Hye+|He$lgZT7W_lL z9^viLMA0YViqb&6y;uQb+5t&g35-Sh@=s(c#P(cpt)gE)9&SxLYJe4w5bG!QgdQgP zI*C`c;!t3L!x35=z?kkidyP7JJ1`Hi>!|OXuGWy$K8}slx}cLWZ)>FP1~e?t(ZJL# z&Y`zb>^Z{T8>#S!V3AW9u^U`}G>cdBnzB~Scmb6K=-d)l34o6}bHxT+;(!J;^-Yvf zd$kWJH^+FBHQs7xRpAlkP|#o#fr>XUkV)1jL~eUO_yT(5lie<3BHL29L=ho7Q%(EA z)`L3Y-&GJy9Syc^Y1DnJ*RcZ<+5_@a57+PpJ!(MFKl1dGmLzT8&}-tnISF?AGxnbW zlprh=ITw8fygtHa&5R+wnOiEj<;J}jQwY2~CvsT$Zq$H{4+n2_-$w=5YfD3=+{YC6 zkq<9diA;zyTb=ePA};aV03SQ|xX9as_(us8(Z*2iX0k`rz2D*GpXF9;q^g&z2L9$S zL1bq>5!jVm@k8TNXxb*k3PY@r!g2M-@cksA_j+? z8-4SI?6szOID?QQP?f_ zI9=C``+eiSrr)G38z~Xsj;3glZL36aEn6WknNXSd`FF>>Ca$+{q`q#f+b*Or2LAei z%*u*ov8B!V11IEtU_%zm#GDuI`@;Ke=GMBAnyWTbv0dsm&8mfEA^kk+R+pi0Yk9(( ze3GomSOkG28S5$SqeOf^9=Ct+y>7)-uEV6-2V-*P|2V$c6*cEXkPK?3$AxbFrg@O|PyQ zvRj}jm1Ot86Z=DXSvVx98Tm4Ndfx0>lM%R9_hDE7hhIOmr|M3@TVOO|bh2DDux*!F z3^c*qu;#>Z44mhI^Yy~(x^R8DS^B3pQpCG})Igz2W(lQF>Yjd2@*{Y8c@St8u>nS* zVn{Y|3!Xx&cwV)H1%Nb%JKHOz5C7CZ+ei@x%0|i?C?n@lZlw;F^_jFUUO?BK??fr6 z(T7(xiGF_nh$T2fh_|ea$Pv`V6YAGdPdsvIq==sKi@ND^@PS4n)rVyQIYF(*c11Q~ zT-U;SzcGJ);r{z?y#M|i@1H;M^>yLzpI`X;y72So8`pJXy77C7VlGANBMDe7Ou|F3 zG7!`}M%&=7JEBlOD03=lfEFJE8d0!=mIA$7|HW1UaZWv{WWW?)j692tgd9nzHFh%_ z)m-M&1pSS8%WrGcL6P#sly(0z?Kp&gsEVfM=cNBV#Rr-(owvNJhRk=tGx+yA*g84L zN62Zf-pD4Xp{WBAOz|PyFgigFf=uA(9q(cArqO_RL&1;RMoQ&^Ov1XoiDG-ku?}=T zu((@u6Afy_o%C)zofQ=dhD|{O~=on~pctP5{k>Zq1KDnlTJw$$+l-LvfLuuctO;xcIgb&v5 z{Un}dxbs$RBy>?8>9vml<~S7uJ1lys;+hw(#kiI=QUXa*WN%KQjYf((8oCX6f(%z; zVYd)_i?T%}H*e9LFiU!!n(uSBlE6VpN&Dl+mJtttQv7@?$-J5K_R&OVHvuKlKDRMX<2&wcOQA?xsl~9X*Vi1JhsMU$BkA(xdNmlWP|)9 zvPS)c-Bd*ch__vV@A`Lmq(0G!l=|uK8mW(XujhNI0uh_!sP$CfP>1?Nv317mjJ0}p z?6F=fP>^k8Ti&GQr0|&!k*mr0Go0R%%fHyfprw&?^p4@{-0%W4=8WD|CxNUfSkvf! z_hqPb3)yvKxiUn^65!_!kDQUAGH=XzW6cXPT@&sX5sBg$4d>bM*Iy^*wIEZ>H0ZN4 zBA0HHnUkY>0k}X2Lh_ulPD=*mSN&FsNT0!F7?1J?PEZ@T^sQpMFVKl z(%2qO@O44f!abu=6-UWeM7Lpek3FY95H3^UnSG5I=FCpBi;qZq#l5 z*-2`IGjaeM5@Jo18`{}qUs^mKn#KI~+BB3<))XH57S}g^=W*gbUwFSxeEn_E37ast z0&>|hsc3%ft2I;1j5X(`hUT=)fCrHp4|-zI^9ZAAGPcczh=Wteo;hLKPw0s3e^%q3=8*GP1(DXVXmG9Ml0{O3x&*yWWEo`QePO-7Fu&fIUtgF%zcAky zR=lnYGM9Tc>^^M)W_Seh7H??%d*e|d><^#>Hs`%kq(>wh+t2RXt!8ja`I9~*IeS2- zk0=QnHu|;c5=-j?$v>a6F$~mrJZynS=2tYnLYq30fz%OnTsBhl2OU)|x_~T{WJR$J z%LUQ`|0eo{L|J8hQ4etW=g7Z_0D0Sv8(9uaCQa`RIy$iITJKHKNYIV^9*pYTDie8R ziF-E5$!bI{hAv*+lt8=WpZJN;ex}^e11=Ll1AF?el=Ybp+2BITuKEy{@Mkcu?!hiG zj?wKpAd99@g~{pTYyvQ^>6v1ZjSl6ufl|exuq~Eq3|}u`Ezq(MXi@{{)i7fC^ugZ>C$gX`&f5P5DiiInR06uSyw(# zWRGQuf47gMUFRj8@zG3iay3=i&H^$@ULepFYTLm8LbQQS0}ZYW+DzRo8Hd(=)B`pV zlD~~Z{~*VfP*1HEl}}uM;sKu-ip;is6sOuoHr*$APwNVVe)U-D<5?G*Gq=VNXXa&Y zBSX0)0Q+X1g0!QQ(2UPO-WqGRr4>dKrbgi|EoJ8p2+!|}Pn5fEdN%f%vfC(t{#}jK zf4*k=m?}X31x2WsYr=z5{`>gEW6QIGz}MEMdW5fdYN3tzR!{}=TA0_wycX7N@*j4Y zPfX@}W4y+J*Ky!=47^XB%Agc-3C10{>;kb;r=Ayyixkf)(&X;m~+%m1{I zVzw0d_!Y`E8E!!y2+!lH`Y^&yUHkmW!K&{)hNLX`)42MA2hREmK zYgs0UEZlb#RWQ9eTYVqgbJwwV&{O*_HZ5|h&;;HW7?Cmw#2MT?XgzEN`Hzqw^DN(5KoIE{X6z|Ho@TbR*NIc`?Y=NcGMbOk38}0%o%FS5wxP61H0K`9`-iPT_G2Fhs6t zBBazee64*pB^zpjPf$<^<4(DnPY+22SktaE$DCt2$?m>!|NM#d^@hCP(CY@~1dC!2 zL&pGtutcHMJVU5KsiBX7<7L;)`Fi0v5A#{lQ}&Y5EbKtl945KE54v-P;goN`7(c)Kl3;``fxuh1*vVh z|8+UJP&Y(LUFN&W=1HeDPIB9?^#=6zt&1x!mTg7jSsFQG{_cXrMjJ;1FbMkUkof|u zHGap@F}+R+YO$<{Hi!+HdYuryTuJyBdE;(=qUOVh4TVS`RFwR*Av=4kV`rreDjY1K z9V#C9UUuRi;~e*z4)`{O!YI>uaSTRu0nt9o})FI4PHB3j_*afZJ9P9@FUVQX`pZ^T4}vZ9TG~N0*(?D zyQyV2Hp|sR%LC2p(qK#X*yzL#ZD=v~ESYE>`}!2-V{PeniBe*0Tzz897aVU(6BtK`O%iCbPlyH(W2qdjL95v~!^A zVIm=BFfFGR>0{dV_!JdSjeDJrWI(7ra8=1C4`p(ZDYKFZ+4>Lyp%k^8$Vujq4VECo z6hiqrbGjsu9E0tj8-UiUZ}cn zgnZS#)-+;ZZvXrm$@jK{TQ@n>Fx@*dfjcLGZ~K$BS5^!tWrwfR#G2FDxZGXYMEOCV z+H0?t5?Nd}5Fm!O21qng?zW)@(1b2_@WpbCvEl34uZ^^h_%kxHw8rQ-&J$}byk9TN zLOJ{(Y9ljyVyd6W^I0}i%czSqXP!(|`>ph3mer}8U8j;{BqU1S~cMMN}cdH z80tn!{$L};O!$9)oH?`&nL=|(ardTaFrl~IKv~U6*+hk{uI6e~6Ec-kJMzR(Hb>c9 zF_(OplH2!?$WTsnfp0dRWkdO-WC_UM8=a0izvJtT_4S4I^~Sm`Z=@DjfzAx-CL>c~ zP#>T^!2Y#4{jqz|JdR<(xepg}$w?_23{7fZX{Hh^BEdVff}urLlW+4xTzZ2Xj79Mh z4SuDd1;{4PgBU9*;Ku=OB`hUO5T<(!&0?d!?X=yaeWYES<#)~B<$ykdLVjtX*Lf6qyus~D!t)#p19}A=E>xtf;n#=EL{#VTDT|j zI?qN*tQijFO9n6Lvx$N?dA3y+Z~8n7GB>>r_Hv1oYV(JB4BvJqP4T_r`($>EW*{{l zZo}WaHa{%{Wh)i67scV|Ua0UPvF^x-9V0=G$kx573m=~nRvGY%H^8A9X-=%|+*95+ z+Av>~aQ;0aDn?6SFab}6M1%;3AiD(JrtP5&Mk3xq${F>b)vi^VLZnoay*6Rqa=*xV z4P}zW!eCo0%2KgXZ;72sTw(zpUMOtIpNQ3Sv7L_K04}j4Hmu-BWWzV4bgS<(J3=9} z90FoZ)bvIwsSM#|^xwnu^!#%5YbpJyd*r`)BlU|gTEG2|=yL%@RoYkD&PPM_iB?sp zHzV_2XuQ0m1+dgG;a#@;CRd(YCKL(xHpyx}0k-YZdGyCN>NO7A4g?vHV{_FoxZ_A< zu!ld2Zlm$E92}X;R{%mszL zZ1Eq+)g(2#JSTqb&K2KrW+ZliQ7)CZ_Ol3M*1vf&%~nmYCERgiu_8h zeOKLBKaE*wQ7q7xbFI zy$r?9Y%)h!6V=WW?KsfR6MUTL$Kipk1O2c7*Fa51!xCm~@y)Bq1_KC8@{~T=x~0wR1^`_6{YHxVGq1D@p|7QHEL-E**_x-mpK zv7#~AdCVPy24oQCL6}Fwx)03j%x4$m_6?Dwbhx#aE7p_g001BWNkl=pzLj~}@_m@Gffa-ocVC$6TUUzAik%W5}LNNd307?_-PgoyJd#OeUMGl==CG;u_*L1(R%3BTb>xG6Y%+ z$m*u+Na#j20ugN!0yQ|p(YkM-YzuWpBXtaGq!`7^0}O8HY?&+FiC4BbGC7jCA%5N7 zL`{KCYXW61%*((N@5}r$p~tri8g5dRlmak3&}~lhj$W1xvrfkSlW-k`_aKaxBXg50 zHbNWC4uQ4ON46iJ&zFA4{1kg?K+KlOA^ZQkli;g9*QfpT;Y%nFk$k3P$l8_?nSK*a zs!_rJ1&6Geo%&ScV%-b6&pEj#uDpoX1%_J+T?5KQ~p!r7X7z29R=lS~aN{Lo? zu*b8J61$HwFnVrjUu;`K&Zm7h0Ik`wjdbU1OX#9{P=KJ`Xw%lo!eDWvA&S#+|rs3N94 z&pjJ1nra1Q!x4Hz$?dXjqm~e+ONgc+V6Y-+lAj0UW8>(XC@N!iZa#Qv@wwUW`&<9S zBjxgHFRnt?r{x|fI**k0>AskEQSDRa6oW?-_1H#zY@}GxyG2TzZQ$%DbPdQFko&OB z{=97?BokP-H#SlRU?xZIqkCZOiE(sms7A-=#(ap4ls$;8nL^6U`d zKn~@l0=TR%fDPgfqULf^*6G3MiXNXA6OKM`^nvqK{FuPCfEo2fnEaRE@1OsHpMU=c z8Y_qv=KM++S4jKjT42w`v_{GRSJ~X8Kcx6YSyOS)sDpTb`2f=>COCKcI6*DqGIBDj z?BAmS+D+aU8!7imW_`zh&`2pB(!Oj2BGaw*X1a0drY>+ChH6GkG$NykHpiTkgJE3F z5mAdVIN*@>p-B6f9W-IvsL==5qigm+-K(e~&!Y^74==UL(?#~rNT1Ta{Z@&qVnW># z^7pm2^gRY(=DTeUT4L1fX{g%9)N6;3&_~l2pbUzkcPfa21HfCY@xeRD4VAbniq$3OgHSo`KLtQ9!#7Za65Rjwk#2s{OgRyJ5zYQ}hpDZR<3R^{)d*_x;AjaojH@un!gBFI9^!3~~?;-sWK zfre)mx#^aDa!mqO0f`>~^M>35GACqp*L2H+P($BHGPfT3(=+vy|f6_^{*c8HKIIVT`zOaVpd5Pv~I~sUuZF^~$ z?a%18Ng)cL!7~3s2zA(D1*}=D5;xll!LC1AMG8h+XG}zBeK_DEXn}`^`tBk-0)7c~ zb5u0Qp7%l`P7qWNEi6da^1qRjgd8wP`Sr#{+uyUYOGin=!kt{i|&H7&Dtr zJOhIn&3#e|wahLLKkj|_`%IbjWPZY5At>T{ds+e}3)7}0X}l+VLMAx>-0g?g2la+( zCJz-F)F+ngn_1BxHTVR)&i$z6aj$EMB;5`P9-;cE@>d|#03jr;4$P1L+6*5yN|X}^~p$G~V zKnwJ{Z>8e*{Efw_rl()#RQkQsCKv8*Y$^xc}5wHv9}sJ4bSZ2LXVVU5&rxbINW*`Q41I!&qs$H>SL%BCSG z!Vw|iIVGhD-!JjZ9;^ky^lge@tp?%=o^0EmF5qtTSndj-Tl_5(bXY^UWap19jqJk- zT{q_a#=LF2#^pt=fmg_Jo0Ij(GkBb`r!|XgQ{?OxcbXYTPG{RhIDJxjNhztoR2wO2 zAT0^r48ZUerA(G^@j$6W4uPuHF?g&y@}I znhVT~(QPnE0H{C7$Kmm!$vKdV+6KB&;1Q3awGb!lp&ICk2i50LF#sOq&6<&0-ODYm z`Iqo4X$kCLd?I2Mu^IRKO9YDqQx}#lOf|ywa_V6QZ^Li{(}{QL+TWF7NTKn#x+vjJ;(Asx<$$T6x^YirZ9uR=; zTMnySA+tydrxd!_o}6P@$YGk@noB6u4Ei@S<_Z9X!6pE8c^8~e1-y=i*V*vnw6Evc za2$r;?k!}8iYjm27SLyoc81ltX}NjRkkE*lg^F~)5PH}H6(cDDYrQK}8`}v^z zhBcb>9e%P$Be#Q@yg8$sBF53Up%-O9(42Rqgr%#<7_7OK?%P#@`&--`ew;YQY2lx~ zm9l}!miC$cd(+Tr8hmXe@_r6qVZ2q5=f$TWfNSu+<-EsOL|xs_w-?>nMyNYo?vHO1KDSfW?4pLGFQ%k zU6;!>w(PpYAU5OZ=0C)=ZLeoYDoO*DpS@(vgZg$nNNrLS465alOKF*}GqCJEkfN-~ z&0hoeH(DEDS0aErBr>n41JvEau?dvS=Ev?{ix`~9wr}PQvuEkTi9YOJYK_egL=2*y zGQ9_tFhDi(y?^HkQ;V93H#JZFNv1{G=Mf(1x%tq%hz^MS?jtj>`~DX5lLAdHnGRoP zLL~~NF9^gza~1IgYLIw!4{@VGuuNr(od8=iYMU+FSdV_7x5Io=ykioim3Rf%#l8ti5QYVk0%@Lhs#bisRhh7xa+r6YT}TV-OB| zhTD7WO%9_diD)Y!isvinnV*f~uVsh&_2Ri^vcYl&ASty?o#@ohF_sqOGJl;|uIpB>HItxuaj{e`*jS)@c(}+=5wZJfFsd2 zc!?@`@*66F5BatiDMu~Xswt4F5xO|y3SLKwXRV@gGj@{puMNf|={X}+{F8La-aDEH z@>U$NQII4N^%0U6+oD*z?+3+hU?d1n@B)nN*XmxT*2UCt-F4JTiQ``DNnwtEIFFG{ z)Zt;k9GNSd(dPwU&?$G~y*S2IRIzfSVg#l|0uGTdSBEa1>+{&i#X~01u-Mpp;v^7> z?t(%TSP(Q7izyJ3DUfGL32FM&V#}wl770*5Pm_4h@K00|Yz^6!EyW5Nq1H&yhyMI2 zSBpXbxIjn0PJK(ZJ%ZmAn>FH=9-AvF5>@=+!nN{mBWioIxG!T+w??swFHU1&rQS|DU<6^vvWlu zdA~*Wi+U5qw4bN6*Ox^H_@`?89c8j)lMHLlRqCeT5QAJ^a(QNYAc8ki>9O(VKQ>XM z3yM2)4s)lF-$u|q#CC84v1R%-KdTxKhckq~uNBOB&0AtWPoqf_{~oeG|GJgZO$kms zbB#<9WTPGA)JLE##KS_Jwww5Jr#cd$^#iS$zpwQJ!~JH>x4s5Et?ZKg6Rqz`liYc^`JB zCznVC0GPA?b|V#@#Y9xZ42gCv63x>l^X=O%w;Lkq<4L-(YH4~lbDMY95o|edQDcDF za-$+gifxFZjll@nFqh}|^O;{aHc|#RT0l*JWs8zuUuy}j>y4j3zwq_-g}=YPaEJ7L zMKdKBJv{NqfOTemoLfOmenwktEWX{3-avx(?`JBWV)KPl38EtUScmu-{TK>{lBUQr zts$2W5D67T|Felwvcc90DY;rhW%*a$z&lQ)ViW6i!$D323jX~6GLKXZ^ZMm4_aQBL zB$VjCU|7WTlQqLtd6~?D;^Ow-FugOsr%QuGe6wIOm&SxeMkb0L!E%8NH#~a7adf<1 z2VSoO$6@k-^B5S%z`PbNy6|2XJ-e^l z^t$(k^JM&}8>yptBW2B2(!B5_lhCJsq7+_Syd#n2l&HY6It;Nu36 z8qiGoDJ0tEsy9-)6d!KGKf1*G7&w2t@?w&Y>@ZP$JiN3>821A~#B}-MSo$f;q$e;5 zis}Tqk(QtXOx6SwHHfe$&J$*HBy{nbFfq>4;0xzz;k31pO5g~VQSxMNSxw0_TCt%j z7|%FdJOtyDh_ta(2+%hR{JEAY}T$z<8h)RWe><$TX;J3LZqS#>1etiK6(Kv3!NJj=Wf zmyrm`pEptjj6tcNqHx4UBb%S4fA&-+bo7nO@QxU9rX9Hy$#T)nAny2es8Yw~XR@osfV)7uS zfyEZ6s@=9y=qISzMhatqMhB0>8Y;2_maFt?PAK2jO!0)$4V4a(2_lzOJGda=6V_ab zzi&R!fo~|tuX}HzM9IRcy}TlJ2@~(A3~*_t!T{$JXxm6RRaUKuV)s4pny#iWb%jl= zSAsNzG&fX*8I&HZe^L*K2$o-KL8gqrV&cl0j;)P*Y=XiS0}D9q4W6-w66UN}k%z``3^ zpgYen(Tfj3fT_ zJ1KmH)}$PK|4h+Dks>{Nl#(SX?^W4oiRNZl$QvlQv`?o8fA4%t+MM%!t*80}C`af2 z|D=8YVuAIGVypuFC?{rqh%CYK9aN2yBfw0+ZUnxg2x|;!4E2sixP;o_We1fh*FKmk z&FUleuuhlKvf~eG{RV~kn;WUuj~|bf@-u@XpbfjRN)2%|`Kp-rg$c!_Grs<_fMsLDG3ix zU`~7`8>-;YJ#M?ysGEy6q*rN9E`BJ|k7=N4N|KfWO9?Ct?&R}4$tj{h3kgyU-CtzP zL5tFCrj`eZmqzNgM(TQBxUV;^_l{%u{A&Seu{wNaP-()8+y?p$@Iem`=g*82aLd$;vQjwD&v+#@nUQkwUF z&ONFs637Vmec^}j0IBX~tyxPm{b5K+RXhNRNI#hUHgUZSaI=NLi${$xbIN47ZkehS zPoMG;OU<=yw1SFA^mO0dpsru@=MqpBT!FA*#w1Acez2M-+GKs1%SV1T!bq$`ho`jZ z=g@F0bpcg6alM!RuafpXy#e<1@fT-1RnP?e+iF?!NnCIDS&IyP!c6-|WW8c!7MmPt zO8T1nqXL6|JCC)IIy#Qt?eOvy^}6tu!tc4yKZFHrm~m z*d01zT7;D_6_$fTpXiKr>P5PrB>yLuG`oy!H$EaIaq&B}ItikEgO?k5otSb#hE4HR zEkIdiK;!`^;kXV@rXN1p7MB;;CZ;!^YBiUq_(btM51ft!KOEajQC&Mx3Z82b&wm!% z`4WladKU&gB!x<30EqAgx<75hdDq;V@?|TK1fC2!lCmF&18?|YW{)m*2qCpDjgDYmKL^P@^&OA{8wW&Az{ zGa27F6SZu^Jf}*5~fDB$XjRf+r{u%Sxu6u$#pMsX)Z0>88VAqgQFo zcS_gN*Iw3tdu{TQOedP2D4slXEW9?v?w?$ttV>S5Ah9r@xWQiq2dwD~u+(@>29jyUwiQ_Oys$cKZ=(1B3 zG2XRT$>K~NkB2?(7aorXRMZDY7kWQ&^nrd%9G9YB7y4tM9}l$S1joRjpCInI-ygW{ z9~k3dZ~z3rh*m>|1KeTn(pg^@iu2(6hUsh6gy88rTree?vbHj`a8`Kef7CkuU$Bpq*NTT9Cb>&mpf@<1k!Te z`2Jg=*4(0m%BQfHp0-sb0)Rf}BmS@3NPX4qDeY80O3~8apZj&~ZZd@Z*H$$=+j2Iq~*lTY}bGl1^kUW1pmVV`gd@Gl9C{MjOEtlJ1(M%0pD!^Ka+K1q3EV{RKN&Np=t7UaJ}+gD!RY z$KL%0FcT-Nj+~dEJ4V6ec_Eqaf^y?myRk=*CVQd40Z&XLwfSPGD0iql0vEQ@sUxTuu_2%+)&L)q&(OXTFQQE8!;*_^V7LiQZ{8EN#wf9Ech{R36`joDY zIt}#qeM$tM8Zj%X*SQfg24)K;W6H9gf1LZH(;&yI6zqn5%($Tod zjBvsf_C_k2T|3XlAk0CSlQ1SQC*d&_kD<7G82Mx9B3Ai(jfF4|!=M0w`Si~tkTU*EdZc1vEv3qi|5oj5ezveTeo{Wq zKNs7_GKA6%PQ}mwJHTQ$(vTHC$=no#U(<<$N#yjw%5R*RhDdI^M^<^y*77Yw<_l3N zW=gi;a+MXY!xz}gkWiX@N`iIm@QemN8@M;{;TxyEHc#Am`C&NDCgfI}Qyv|;Q^Qf^ zvpn&!c?zj3`Jm_NB^Qsf2%qf~%%QoRwSS%%N*L-N1sI~Z-yit6KXAWaxZf`$mL3yF zKXIOlvkQ6`wEIAN?`Zu!y-!>p?ZC%1alZyWu3`QdZ*Zu@Fq9T) zdv0GuEMg%VCy`BN8TUHLds%$_fiW-eZF}>= z_#XnG_y5VJNlCY@RQ;G+jM9NGdCq5p)T?Z^EdT%@07*naRCBWw7iVmv^6$6i8>yAQ zkmgX@3#zz@vwNm`CmfDF-%`=fV>Jq%ZIf9x84&7Hn3xhbc501yGi93(j;=$OYf_bQ zC_$7lRnv@00e@fyHJ2RrXnH&|B$_F2plX0ymSdZ)xz2QATS9I~40LRm5V4xun8*d~ ztGrNJ=szJ7q-;<^CMMgPZzHc!Arraoc{9oBZ(fIm7PPIKYL3=Z(Mg5+ljK4%sZt*g z%&TE|s=36(dJM=E9Os4WdSHwb^BSoYXP7w}FkEvqP=@=q4b*uw$DemESe;fVY}>r0i8I8=PJ(1k3HUm{YV+hXwpH9!(|j?{Baq_FZ`3{oA~lHONae$3kJ@`A8ekny5iWBE5hyfX6^KQjbgV7&U{-8il5S4kOlK2zc0QOx{4TKN<}qrj~F=N&NJ5 zD&abta|=};-eg7dG>1=bhc{9wbH%4z6RJ&AGSJM}EVM*Ut8XffJK67hS_!8y-H-Q6 z6#70;q2`8c^+o`MNV*IOCbZ&zzk6ny91Dmo#%k>CySeWLeW9^wpEPfA7=X+4o^I;8 zD)2SzGcuyGf_k9b{l0cEsPk%~GAnD5XS0cl#UK0mr?Uj+`a{>?a}Q50O_K+vO8|BX z`FzUC2Mp4#Ty7_bjj!0_F$C}T3m+fWOugSt+IKqz=c%|2#qCh^Ho-yKvp%0bNb2>a3-2l- zb{`_GI!NN(;YBSwaO;~a6edFNeFKwe$#||H zC_nFByDd=1rperl2j2_ons4H_y z<}eu4bONu0Jl$(xhf+);5DP>o@2e|I3d_$Gq{Dx+f6Bo4a^EV{jRK;Od_5DalnrWu zmt~E}zU^$*Oi@F};pYvPArE`B2<6V4HxOIYHjDE2yEW(kb!p%Bv1oOdn(rxD>gffl zGPQgWYZzpbl@u(z5a|+Oym*o>t+Oj`g+50bSJ&-E7?SyCzd(FG|Bbx z(FKc}@wKd-sU(ZD-)E!vsm{-Mj>v*e4{M@Y+qY5Vja2jeiSS7PPew|J7l^Cr;~)S< zpkdX`CI;Q#m{WX)vBnMRA%64(uF#2u0uVjDwz7V%FsT|H#^*~Kw%B0g7!LrzFKUkmG@Ibqs%gCTLv;*9FGJ@p#z8=y73Q4|@*$`!ouYR7XF~ zMAi0g8vov0qS7rF#HP+& z_$oe0_F^Mdb6?``%hT6e##)zMsR0-u0*U7oz#Q#c=i=H(MWGrJ6xj;cnAqp@N(G1# zTPUJ!8x=PN_4lW!VyNPpwt;%T5B&VN@ayBkulEZ-KQ7!aLF)s(PqgDi>xzCT&Kuzv z4d*#<$^jx?g{JuRG8xn5P9~-1wUHXY7?h3FeZU&2Q5&fxw{1d@YkZ-2R=$zmwHqts&TSgs^9_T$hm}pQIotnEGmRPwX<;y@a1~ zXt)e=s1DNTa{;0u|HJZ~urBC~gl950RBM|;Doz5#g2sGvy2`#`u3(|HA%!(r)NPnU zhtsiKo*JcGDeP?M5HSN<1~$pITWE3j7AdJi8_73O)rHUG30`iBms8Q0+?Oi(nzz=_4k!B^ma`|+F}YzWvO__kodmpQ$;N(3D(TJ|+hR&Fea$v8cpc2&&3p^bD~l z-YKO2=|-w|E~NmMplg1joyPT5l{l^rq2k9YF72we0L)MqoNfjnNg!n-Wm~B==!rRHKAX`< z^$zkJzNi(Y%JKYnMbJIVK=&(eQO(DqlGPXl*`-R;O^va!c>R58eOACJY&FHyyn*Tv zI-nR1fO2W3H%AM6Ly-oO&D078E{)XuU)D%%%@onsKB)v1=<1FBg)E^WBa2H8GbY6> zJ~b*U*67$$49&^CXNLQuY~yntwv~Fh9eBB&=)K{(hN09wgJ_mqo0xi%0{|@Rg3+4e zG94meYsH<8lZQ<-FdpU#rZs7ZX(Q9xEO+GQi9LjI1@c9MP+}9mn8;H~`^bygwT((C ze<|oB(8fV(;u=&FMW;7r+}sbt`|JKps!?wM0t?1P+V|W={RYKkg8{g?G@%=LBlbg+ zIi-$kv!>QWxrc|jW%iKxA$MH^q{DQ}*OdiX>%+Cwip~JMVd$rsjF0o?%~anE#JzbM zJWoIQle*l^*V`=sqff$6SOu|TV_suAMPlZjff{@@&(AX{7dzRF(3P(;BthP;%&j9V9|EE4KhaK-Ju5Uu}vNc=utFp!dgxAMbbk z`1Qb#Uk`jdYzuW9g4PB70FDDVXTwbn3^6PKvyY&>iAucxj2m~z`PRoV_l?xYgvtEo zYIH+0-vpo`Xx=cjDVeRrQ_zN>T?Ta+QILpllAOF24O2{^w~z7q&gzY_K;U@qUZs&* zO%f%UE*q%)@4g9OpLTlEz$WcO6akd-n%t0YgT)~3gz_dtDJR}xO%dG8~Z4$!o zx=Uj)vSugs+C%~bpa%$tmkd_0N8x1znk8QZ1*`$(=UTqlkR(fNuh{>7C(yy5qk%ea zO;lpe>mVU1v@E8*xkAcTE z@G)%T`|IPv4{xM?eq0bGJci`m(*_-oW1weE6vY;bx_yprqCt{1;4gB% z1eQRyknn750#`nkhbbr5yzun%X^V9Q!M0?)E%?b&V{8fzF<4WZc2<#)L7`~oMk-=S zq^nC9>NZH$epz>WO)NI3s(@Ux^joR>p&Txu zp~+-xrj!cGQ_^1&A%b?3=o#2384BsT`3{%zIY#~H;KQa>{dS<=PPFr|kjie`sg**E zC(n~?jaBt7?M16((P1m)QWVh6)28WfK31QcjNIZqPid-GEhnH-tNfh=_}%{#$1vBO zrj|R1h3lF8&ZuPOYg4G-08pM6pK41y!iy2Ur8Tj%{4y^BA$ff=FQat#5XIwkYLn9< zJG03Ei{~Or8Bzt>lJSxH`&}giaqV)abrAVY&bx7dZE|J_^An5#Cf-D4tC+I`C9dl* z@_vRpJZUhg7G9J$LN)fPU5fr|^nTzxPuy-VczJ!r>)RiIDjsd0is!b9_Mqq01eM0T z==o)O&IJe)(+UUCoYKu{VZWI(rezaF`|6(%fyz?0fA(iv<8NJ?8PvEvU%H`uQ&y?G z$X#5*19@d-VH;{JQc)JK)!SBlkv+0@1ZLdtJwL`Hjs5W&c=1aMa^;EhT(+9JKLXe7 z@e^oP7_X#tzkjw4bR(IVm#--b7u!IoZ>OFbsd|?A zAZ)m4dyEU?x-7>DC72sj?mN}6%-%$KVUGwlZ(yGJc?=)+u+tsJR5L2<-%H%ZwzpQg zC~~{w@BQ##d6V`#@TnG3rZ1X&gz^9#gwq~$WFw`1ri9;*9G;?SaG>-gQTWM%OOz2< zHM6P|UWt%zmbb)#scoww_xo06?Lna90gMmmd_cw>Ixo*STUn3SHSpoyrH}iC$9*{D zhVs2*&1<;_B*~A{`>GtoK48kkHMPiSZ6o#L849%V@RWFb%~nyBnXl0 zMt{yc9#8V5g*Uf;l|2Roph;lret-VPMv5vpiq@_98H$9RJeWWVVlOp3e7p{n6Ybr6 zXkyn^ZliqQni5=Z$TYlkC0B1#M)-3+y&f22dT{Pcd0uQb(<9}Ot~x}VUTauZ>D-2V z+#0$p8Zn1=w+_5*A|T+o*p z*sIT!j%72sTg$-zma>oV9?T1WUM!k54?fvLqoF`q_ZBEHH#dh_tpj+a}n|;45BTpdB`AhZ7+ZGs2jIg9-b_ox>9eX;=pENOrC3E z>_%^WYNqzwFx{peYokC}G<Jid5aQ-UbWu-qJsV8t;SI}OI<13JyDULQ80 z)7G{f2g&IF1v??ZA0EGCz;FA?}~ym$dDQl3DE}e+?C3p;nvY& z1JNkuUX~MLlhtcnU|x;X$n8=@qt0dWPMP0K0L4z{ljV=f$IOkCkV#po%7cq4DTZQ+ z*e6EgUMQ#?tx3V~H{YW0g7|ICRLPOZOg=3zTT0Iq$!e$}^wgvg^f#tYoSbkh1iuz1 ztP}efCzx(vIYsHHQ z+rcmt+B7|QQ<@j=|0%Vv@4lHWqe-8DYx?Ot4H=**k+kf#7G;yMXnh;Cq(Id*sfYUI zL(4E)Rm$elGfjtm9!-*X6oqcVFqCSO3P#$RD7T{|0j~@fml6os^k^YBXG9254gu}Y z&rQa+)J(EH_xP0Ih+XYe>9Ww>)}c?4I1>2abuTByCelo^a1i`t)w1f_MCRsl1CUBjxR)mLSTK)pNww=8Tw{EfG3b_P}MKjeL zP2Si&QohnAU-g|w$IFY;_+C%^`Ns?X>(4j*`Nu2%{PPum{PBjjw;Nt=2acnm(R$I= zHcHksYlcbOF3uK#yU|1i{#=1_T=~sNTt?xNy;vpt%~nH)<+&6=drBE!x6GFoUYqlV z?t!N5JQm%kH~*n*%9}YrNq3QLdQy$pX~GV3WZg)qrras~FGL0ji0Qp(7$ggjLw55G zWg)g?AbSY#zK;5atlFm4wUMH|k&0`vNw=(-)*LhJ&8OP7FdNAcjnqt1lIntORO!bM zMVofVqozBeHo^&AwqdE`$R^6memv&8}4dsmMNCp87r7zl%#l3 z90d@w2Q=D^C;E|Su2t(6b9o)0`z%Q_O`M#z6h9&3I1b#-8(v;)8})X-yY`t&#*3o$ zZpQ0!gV5rU6kUab#t@+VS=wlkia9TgYZcCRqu6t$(l5upnf5KY`ZcZ|==gnUUoqcS zNkj!nWJAnrQL%+ZQ#N=N0D_velBtS5H0X-1Kv&?D10!TiAgA-)(2komQtgI*-f)~Z z9Je#6lFU6N)1u8L_bycwWz5L3f^4{tka|*E7GZGT_kdu1?z7jwjx=riBgZ6u%XX7OR3_rzUeHbf~OA3o5r+U z?Vj7HTb>iYfceFmB_6Apg30J4?wc-mB5}ou(2^fEtPLt7rIQQ&KlhyjF~CO-5W)*+ zzRsC2335M_%m?JU=SFHg9tM5*xa0jJKDXxP*L~pU`-S(rP2`Ve>XqKuVo^Mo=j;{& z%$?KTYd;DD-uzD*7{a!d8iM;{;^RK>^M2vy`#m9)ms$U8BUKtH_{ZMJ;!eiIIR^Uh z?Xdy}7)?DSwoRZ-lY_GuT!Lal>oFZFS%n}RBU-|9E54qipe$Hw5{Pe;?Zx%$ZwB#I zi>^UjQYfjHRGyGA&0hy#%F8z!DW^#W^h!`N_i9{4b$l$v40#g;TKC61V}mM4twa|z zSMzuj!6pIKEs%LNJDP~|l=OMy6dS2AMv>jEj3|Ag%Suufqc-z66W{(cZZQ7F4uTNX zEqFPbA@@|(p{@;0e6TztGwW#s6;>G~&j*lOcK%Y0OEWDo}6@zQgZWM}EQXKRYF$JowSkZrIBUSS-|L)wbH~;9C z1B$@WNa@-}#f3z3CgOZTQ`cBix+~xynkkbNc(|7rDaX-qJ5Rj6-0;WS3;yd*YoPx8 z^9_IgVGY&W+beE215Ct5${-7xP1LwnBNbUxvHeQeM26t42C4<^o_(_CD9|W~?0IoL zuIuNOG?Sde&%=R5Kdg~zr#<@ZM8Dk}6U+O9E2>5}9J)Mjr5xuj1*9QU*2g%EVkOIz zIRI27DvR>>dBeC}Fm)sK)fn|REmCs~T-T%EO;HGgl-G-jnR?Z4D5jxqpk~)wZlUT% zYEC~m*VoLv9+&rZ5pb;COYTp2*&?8@H_l0CG~#w(d3dzO6j3+h_abA3q^+-pOP`{T zbCRSv?G=pNcoNGtyV-p>T0D+e+bDTj_oos^H!U>j^6SuEc^D=0JaKz@!Ryu=a6Se1CrdZ;+Z=_aJ+42T4hKKyRoNJv?YdEuc!F*F3lu3gwh({P8 zGp4@9KwnC({W%XZk#E5EMrui!?Ly&VQ0AwMl@i?R8_(VehhiD(y+H2+;BN2}Bln#? zpjlzb-8WJ*3uL?20n0aHkg&%M%hTUql3GV>ue*Xm}+5*JR5iz z*=)LJGd^xjR9xdh<&%-0XY*8tMt(hv7?O$XH?)-VlmbBHNoirlSoin!(Y*oef-W7c zEn?k*FHquxi(WnZpb8;cc~vMrfl_3Y>tF#(z6qOPUIfZ08sWZ)*hq0hYbQG2(Ata3 zga?Fe8zQ(YS8Pl#iZn6WsZ3?hm<+2QoEk{g4@e8DGzKv5fN!E? zSOX>Vf?Rhz9(Vlu^?_f%KJeqGeg5VBq^Tg}z0B$Fk z2?v@#*G`9R?ziFz0b?rRnpnVzhkK>oFWXW`$olN@sgY7|qy`)C#XlYjbPy&QIuUxv z?=ZcQ_P{ljCA-vFJ;16AZpfSs(X_qvN8w8J)0?RMBiv9WKY?eCz` zrkmX+4pOpLvN(B`^)1kbgmF5nWG?ir70>%jQOmUK3TgS&Ea?+ygnql364^IWM0-Kc z%uq4;P29A}9}>aJN>Fd2W;9ZcN8Y4u_$=)k(`0?Ecy4yMhxt<)HvT#tK<&2J z&qjq)rehwdTQ*YHxH6M%%*&tinMUi{NSTMBOtb<8ENd%84uwbpo;JJ5J;}MMf2eML zs=TMQZa28_1ic~_swaMxE>SW zW*e#2PqgMTzr(y4@e$mtS#{Jo`*S*{lSH{ARHZa;-r%X^MoMIC`G5OHYHn05$>}YXvkmLIkF1p_#ITz1qa8Ss z7`um=o9EUVPf$2AsV3;<0Bw8CbLvb5gU{ZemV6kB2`p`L(eHSqGO{2{-j-uw#cMQD ze4w`zy}dwrLa9TWZKcG)@3xhq#4)$jDKDuu>B{y?r3u!sC;UjO5DFL*X9J%18c{o| zuLtyc_eScDIUbO4LB<`Ak2~IfzT?M_cl`C^9sm8~7k>P@@Ynl=zdk1Z`ho z#hAX4n&=%kC*znC*JPZ+IEir(L#Y`#w7KlC338M|YbymTX=@|SLfi`0koA>EX;6K} zo!d|IQ;9dkelHo>`}CS&!L}H6M_<#MnYHo9v5;%7krE-94Pa zQPOsSISm_h35F{85cHd3424WzKRd#f8>PgL`;lb2Fwt7TZeUy|rc0g7ggdi&jCFNh z{MC5Pi*~Z|dKNE2v;;dQkVT6U7<=0KC*Bl24f4O^krJT5CU#$mA09pl4fRYu!n5of zsc5367x(L7g%`mmPpO4g^my)ElEUeMYsGk8LTSaGo(1y0?9JZBu z^AOerFtV`f*Vk9PzP;gmdBJg>)>QG+hN^gk#0{$y`BHAkXP?+zS0N|OS+cQo+j6fB zl~2T!Qer@~wCvyZq8z{JeDixtJQRRKK0ez!B4TXP0ja1MK`0cG> z2*AyTFq~isLhI^}32v7KExSQ?LT%&PE@)rZ2*W6qvlGT-%+WAschq&qtkcPmOvbg% z77uXr6RjOM4%>_!$G^u!`TzhR07*naRLwu0IQk95{G?teKp00yt6Kis7#FVjF#6_r zSOZN1G~E%PmJqF~Xr5S2an>fO>N7?^fjP9^XGD~tk@Z1JTW(4Es*1mgax=5)H{Eg| zzSvB`Js$;s5z<=SJgEQa&3rF_mhvEi>;0b91l$GOOiHG#wmoNpsBJk^4aDH0jg@gF zCJ*+tlx=lg?q^$zfhhw_14I`@F3fpBbYQAA#U@KL0$GHC>vcgM56Ja^KJJj~!y2gR zk89v@39g|gty6}=0z8#NEdYxb2?@48xP0Dn<9I;HHcz622|^o!Ngab6LCy!Xoe*?P zWlXSdsRqt|;OK|R1T(jgJ<6gLetyfztVW~TY-5WiAf9&plA+@X)8C<&$t;7a=oG^> zz_l_d>Aw^bDbto%oxtW33{XW)mex4D^b=u#v}{uw(lfc=-R1Xe_yzFB9v7q7v!S40 zNm|O#q72KPcM$sJ)GObRxhhV7bWyp`(|xE)0c!)MPjuX^y2leWSL-%y@x7SCAE#z# z@rPZfYtyt7YD>0i0sZVlx8>s&ZRo0{455vDt4{;fR2_-h_A%(0WuOMuB{>JLfhckn zTAUbR7oZJtCFT4jay4ti8g7&}3T42!DYQw=Pnp?R(u*e3f-tZc8*$TpFo3de0&x-dw#g{(O7M#~*Ju{PB9j%XwJH zEmM+160|^Ac|k+W$U6NI$@bIrzYFt{_h)GvS6)-U}I-tL9wL4R5Ao zbVJw|KkSiCZ!OR z%u<6-jNze%8WA81gc6MuVg%O{2`t;@qfDK)45tO;N$F>~DX7afj^1&*y(Dd~9^9>Y zIQHucZnqco<1`d~jSQ$#zan#pLz+|}EWY_x-$unILQ;;4f_Eq41OwD?!R;FTjV6kS zz~@oWzWxf*dj6{rfpP;}Xs1-rzh!NxNx58J6Jl%98d26RM&~kU&Dqqvq})NRJ2{j* zgHjDd!Nb4L%;K6tMArt;hx-e~h^37nlQ4xLvzg4L*cO5c83C7p8n(dd6TO`tT5TJ$ z<1opgeLUSqs%22)5^VQEsn`^LTz6dl7^CBQG>mqE`2iaK9EqDwViT-O7-{hUyS|@d zSd?3pTAa~hZFp82ZPeOzzgqV%2P7qVQI_w4D*2^u=u6=hQ6BmIxX63IE7K2YBr+R{ zSo7;KTN6cng4(c$7f0IcT9btD*pD!qUw5)RlqLyHcE7_j%mXH3;Dkw+fs4eNsUa7p zTzezs!J>dduYq}8kh?Wgay^XD=ThuB;y!83o>B+e3F>U>fTQCW1E&sjCY%T3I2w+F z%`erW@mUXL3j$SQ2)XQ)4+wWqKSBKj_tQ62=6#l_xL;RZ2lkOujUh42Q_qNuP~x_- z2u-pNa92>5)lg9sgwthx1B#Jb+Sph!Yewg!4I!nK)=q^aol`y)`J3KIgfWll3shUY zE=!uDbzA*y(W#crWG%#<$me^|$QfD|hU&8}Bw5|Jy=$XVhY*F|t@I@GE&kIQ=vso) z75Fqv%m82AP)z?CJ`lhRUA+L8KT#JH!VQfjUq>o_t4*yD3jq3LL;nJ1p+zNObwZGs zg(x#rV_WFXyp?^})cyU1LBu6Wh5ej`mL_Knk|^y9lwPq^0ueGALQ|HM^BPjS(i{?@Zh~b)v$QmfQEIVv0UriX($26~z zc`S;lbZRyolv~1<8{CiteKmNavPH`sgHNv|{C|Y9F|pU4hz)QN4Vg<~7g}`bt|cS^ zF&k7_8`RiLi9q|bHE8S5)}Yi@e!OF46Q^Gf!=xLzM5x}tX!+&1G5UGpMKqt2Xp(yG zxZO-*cAh6*+_Te;BhQk6LYNF}O#sOQwC&6^vwK_@_PWiaD4~&KSlvb7GQa9cN?*%Y z4b?YweSACCR25FV%0^bxYerGA*T}qLB(8E*&Ac|14I){iq0{4HR-G!kzvfItlTJu3+=E5j=RZ}+X?nY zirdYas*oIZgJBRC6mkr4JS=pI`x#~6dVJvV_`u`wj>qG`qcvQQyAQRjNQ9e=$$fOx zJoFm2vCqv=gkbAe8}p}TM@#maM#0z6GMw8AA8K{`>4j?K>XnSd?dz(htS<<7q`YnX zcYz9hr(7+Y0!L|=EmdryS{vYYf&1lF8_Ixar9Hc+Pr?v5QO}E>X@wE76Wzp#engnF zf#}L%#*l|S=9Lc{7OqWz>co6pn2!hK@xZ*_A=ll)Op84Iy3|N{+Msj*4cSC>891dw zXo3m29fUKoB|CfbDLHBW-ag0t(OsnYjn%R)9ds0% z(o8S&@a;{5AL@o!i%{7}g^$8z<2e|J$OEpy&C#g;=a$liOV=fD3o3>p2M`2V-HI3Y z>wEfrQ&$??wzcp8^sVrSKdU9JO4mR*LbMlfgL7k7skWlaRw2)K40`gls(#X^#Bsnb zMq`(@HZ@7ERCN{;TKK`lYB(GIRuG0MG3w|x$~&vk%l(Q8BRt+rV9!yVg~XRs!P4m1 z;Dx}Q0zzzmUl(zbiITc#()oQ}Bqcs!^OcF$+`=0u1qFY<>>>JzjH-V9&5#;b((ymi zNX14<DK5=f5c|w`62zU;sXMo` zn=k6^W?QM(7yR*h;&yhN-BzK#Ix%^{bX0w8!~7uD`Lv~(Vp2cRjL(EiLbQ;Rs&|TF zBgMTZ>v&K`Yd<*PQ@xJkl_VD z1eRKKU?#zv$#EuX+_uQNrXqk#`Yeo4Hj<&s!^SWe%8VrO*htBOQIw4oyAFGk3_O!< zP%+qXT~tCF7jH(;tk@+Ow1PBaldg6XRRmSYn(@fNnl9rrtnbQn30WVh%UYZHWYnQ3 zTIFquPcV^erV5qO^!w=q_J-+z-h8+ahb6QJP3rnQ|zdB za67G;>IYgo{GU%yJI%M%4sgF21fy9)MV_^Xf_^tYr^~>GeSF+c^m|9|E-9yVyG3M; ziOay5p)IyjvTC*&WWo|Apm<7+4>bDTL{;20mkoYP40;R$$%lzCvc(rdU)?{?tyI}a z#YXKnRDE&xwq_1V2wl`Om-Vp~?zWLK6n#W&FxzlNCZUVT_N3bpJ9$Zdr;7k+wgpaM zC%@7#4<0clQEs88Z=}K_7d{_*?+WI1VLYzYP+fP->F1p*EKJr%5M?NJyC%D!$;2TY zr%nh48Ube~oSkrVLht0~kDo#Qu&vV#)GZ)asDb+l?l<(?3(ngI+Htxp-_NBna2(Bw zS}`R_+$+RonxY<^(a7^4EfOH4x8y1wZoZM?3Ge{u0;-%3Qnrkg+hi+n3+20j2Xf2f zQ6^HhvnwNNY0Q+Ds-$jZBQ!O7Ud^6Gs{`a4&D7V@ccz7MkKw{9D~jAaVq3o4NYN6= zl;PD5)RYt0;uQF^tglM@;-##ftx(*2GBE$YY@`~kMyh1ZEIt1+PIEDEH(qcE zZk*dFEdb})?L~=BTgxPZ!CA;hE|0C?J8nnA?d-UnhAVx&9Y*SVeZilvH~jH-!`sV= z-q`(7rURCl)Nz|idWO`zd_&{%KE$S`G*_`%(P*hz^Dv>JJ&jP!lyd~-j@BD&MJN^7 zj#{GnbqVg*#N(RaW8n5NaXkcM*v~H##a3*k4mWhWcHO^&xJPREM(TP561w=KW<{23 zF#)2W=Ud4;T3c@&=1Y-Wck9>WvSOx;Fr%+!eSm&T+UJD9IB>IEjk9ptv@Enn4iE51 zgUl$1ME7W-!ZS6K{AMQPKxtcUguPaxaRcHdU&?F>+QZ?hq^P(a=xHnRuO2A4=J4h% z>^Tw@9yt2xNL6vj29tY=W45a}$-24lh9SQD~s`CLb&NV7909z_)VI@CI1G0Omj$zf88ZQQ-o$ z*|F%)ewihj1>d;bM6m5LG1?O z1H?CLthzN-++O_8^aPf`<+28XxiH)hHOGZ}KP<0}?TiokB$1k_-(`SQph?!iHa8%% zZ>N1Yp>@J(=LVN|+#cY}WitAeKZ>@3CGH90nMxD`in7j6z(l68Ia<6Nz*MBu^g9wN zk}W72Tny?gKxv{(4qApp&EEa3DQ0ty4V$-0{Jo#_F*WI^`J)&l3L%EDBkO6H7fvor zd3Y0LuWtgGG@>WeJid>C`M5A150m-DHeBqwoGv#tPo9DgMTg+Pv}`z@pf=G+aW>%A z$Ug#Sc=V3gN*&+65vaL@T6~|s<&i4FW!rMMJlE}KptXf6Q7ein`@`Nup*A;~o~y0$ z##fO7^yFXK-*Hks!W*Yy_jRp6Tl%tDjIW(}j4rJVuX}UaX85Ftbn{4Qp)%`SpOgEk z75juRfKyQZ6tr6vKOvt_HyLlMJ z1aY(Wr%v2^6q7!iz-xwjMuTRwGn=yZZtM8l(QrFDUeAu(dEoWsz{`2yj4YbiCczGP zL+p|joa}|?v~=8y@dBTpxGzgE2cpMuW`eMmll>wjDHcnxyZ#rhmN$ z?)S@0_>S&Ogd6+~C5yNgA;?=Zh~>vATY(H+i<@Uj;)X2JPbR`@RSy6DMrxb0Z1`y$h)$n8*{j@+6HILLaa9tOmit8A-#>5fLRN&D=7wu)TxJ+Tr3Fcu993m@Cm!IWR zK`$$4+NnF6yEeir317^(${7+M1`Q9CzeaOFF5ALhx+DSUhYhZ}ZA`SS4JrM*vEs8i zKLQOs)WDt}O&EnyXPxGE^Y^-Y9-nEvgfAq}MME#5F2O6rNLgg1Pn)Cy6|ldnApcm+ zv?-7D8%f+SzjI|$<092wsOP%Y-8ZzWLyDW|c8~CzY&Bf7?h?xg&D}D-ia=hme4QDj z7{RcSZKTWFmZ>p4$XsMpM3I~f2bl>_Cysgp6<&{ipmp0uaX)M`#U}Mb$dS3P28{jh zQD0d?yf*N}0@iL@G*ADHrRpN{0$Q71YoRX^fRH6`~f+PulZ zA#!1i4_x#PP`8s(-Jc|;u9%{xB2o6Ox!+@(U%LWm*8r8GO~pAC=McOMvBqW!8Wf!r zM*~h)bOPR9PrSao;JjI8BliP*9N^;wA1CNIfp+4Ez;TXDNZ@|Iv6|*e>Wh@ z{lc9G2$yA^Pl>y9K3lyL+QIA58Lb0+*d(62f_prtx(%v`)$5s+cWE9+_fAoBNjoph zcRuOZU{LhTdf1u+LXFv=LTrsFhf1RSt^f0Agl&hj#ipgSzT5HtC20a$eyTnT&2(}fLq6)_smo}}=ug|3W9wk9q9obIj&Xs9<5XjX==^=GRR zynrkPAg65IHXpTWDmaEWNAWR+kV`c~n-?@|6n~y$pc_H%QwS_3xv&21g_18@MXKJ~ zy{D}|rwCG5-e0olD$dOXef1r_SrIA_*=&n{JNuWX)XyX@_k`YDhFhW|? zubBa^>!t!M;~OWr#r%w&2}g5H^U)1odOPuQJMs2%;5@ter(g|}j%idj_fptY%m_r5 z;nUEjy|AM+QYM3Q4E9`vJ0n;Tgh~VD^6w}*#KU>RV}kI>_{Fydp5Ad!SaEy7(%A2o zyng?9;NyOw9gN%CiN~)K*M~ihlhHb1o((#BHd46?_C{)4g6kT1JT83PANaUG@VLx7 zA~B`ll>}}|8lNg;=E9mNC@SK7hg9S$*_#m5M(Qc#vI3<3PcP{jZn_yJE8rvb$@J7< zmB~gW^@KdkWqpr_HB$W;I8TQdOwS$*1ai?+~x$11LV^l_V8A685ptb8ilW#sQjp#=~g~b zRF_7anGLl~HjVA)>DxLfT6if9M&0BUVw65@#nK}dSm%OXe!?inq>b6KqRcZuTLV=a zDHZuvhff906R$&{h4sn(yL?%?B(jnGUOLRUo>{T!?d9o7S1CgZT{No*-ApBZW;)d2 z-U>wzNJ(CI((-c&w5=BGF7IY}=Ds&j-YkK7Hd63_%m^%J-+Um8p?Gh&yvC(-$_;4h zp}5vat!7oX4HPqQjgB@hjL|TzXr=_r1I)G=;bHG-VIQuEU1-fsuUr}_Bh&6A#Tb@Y zvBzHIaY`#Qho$65#b0L6U=o-y8u4Braa#&6*_3-3^mk5AnoyoaK&YD*fMjA|q%Uo% zY#ZUoZj^b)oOjH5SuRp~b;RXhzICD6G;s;Rc5g7dd<+v)9=PTmj9*YXAm~QjLxU>2 zq{vCD)Zhjpe{L+WQ(=8Y7vKKJjOWQw1gKP~G{fpm` zpN7Bqk}9YRN0a$$%(P_@QtgC+?cNn?3|^)ve@<4O)fzfV#)QOUkBQLoyP2#|*lGsz z#ZkE?jrMB|6dCM#>XrprH`ZmLxSemi)KN>r7Z3?QYoJdAzLs8UWEAY=YQg#75 zaceHkxLq7Oep#kkns8o3Wm@Hv&?zQWQ$cFX*H@6r_6fH$bd( zn=jCL70c>Dc0(~P#kgO1eBAMPzcQJ<9fbQI2k!4D?!RuhK5iJdhITMybja)mTp%pp z0z7Sd3_QG%vW?UokH^3@XHu17vH&h}62I9|X{gfBkF|OY#AdwNq)7LuG*RjW9zEeW z{~gojKWU;^pBGTHrYF^36nQrOSU`=6nqDdiDI2NB{Q`92I8I!bZKY;-q*4>Tnl zMyf^d9s7`@>6Rsm67Z*+%ug0QC_Pgi#D`iG+jZ5*wi5k3{W>xUmU;F3bI+Nv>uRPK zD#0aBH@ek8Nlu?OIen1eA6^qQ+D?Jts>$pHoF%7}PY8Q;j|-IUx=%S-9HO zNY$puD5B9o`BusssWfmHn#{IR5#PY>=Sn?1FCStwH9n1O{qRlHL>ZTq&8{_4efUP| zYPdRsF9u(;e7^zakULv5MN?c!;2S94N_7u!&cY_ALb??*0r{#@AH;BkvBgxxbr8+e zlZM&Ee!K>aIhFmc4~4Cf+MpT*!%Ps%I-2kR?6JcLe3L11LDhm~#mRAYZ;xh5QrSQr z2hyr!H|782h0(;E4-9;Ow7Cc9@DuH@W@^Ncz#%wvc#w8nKeok;Gufo60ddjWbEV_)@$_R>RtPPR)LPlUC0n-E|>x@kN5j zh7ihr7PJIO=A9(TbF^hcW*B@~W|B{}ayz?}zJJ?L{JTbqsKS_#kOIsSAAqq_{Fh%~ z+koz=@p$rXt(?1nx8&bY-D!74g|oSxOEX2N^~V#0)4jstQdpx=@+T?IqO$syM#+%& zv;uZBGGl|(#LqH4pnPIZwrkxz^X$0Z3=8(Gyx2mGVd|S`q}F{^N{wXzwH9Yup)^KB z1&Lw`gC<$&!F7mf$hlOs7mA-gHBz+E5j7QD5I@IMcMGelGSWv4d8!S3|IYtpBh}YM zj^TIv2G?8f(mo3^TkIfU8wXUvpC>eF8wY8B-Y2WR@+wF$8ms;9nT5p-U(BS^C=4mVl#h4a~8pU!2X>I{vp2?5}dt|9B=_0#c zYIuPtLogqL>+!(#@xbHbj*nj-_;`PypX}1U7hE5AT=zT1{XoBUZ}Mi@NRi{^4OlV8 z+DP4f({)`&o|7o%Lc-;*kDa%i1xDmg$X2|5u`-Q>^Ax*cf2UNr-;=bDzP43;ZzEL@ z{5x)by({;%F+BzQ?ZlpFhT?EhnK-Z5Mnxl~C6qI+3)cXPVrHTaz$9KITQfzySwZ4) zA)qxl2nm2@P8qglr3QO`!4q>DT$tv?3V4y+7K6yHZDO!kpztKi6VcY(M7~gv3Y(D=RRU=|jPN5f@u&Jp@==N;_XK{vmgtjgQ*w%H zrLvF}R{`qyw2@Nz-gQ|vQfA8Z9igBQFnvnrKUloaIHta;`?__*z#Fwm$!=C$Xi!;j z>DnTPE044Sa(tTCyy$3@=TuVnD#jnx;4Hb7j%<Q4>fB5s+a{m~pt8{Z;rDSwFSBBdhITQoM&J?tyICR=$llu|5K_#D z9|&n*P&rF3RE9efwt+%O`@CO5faGH!RX0+RUA7fvHEhPW*D>S*FVpsOwE4HCeFeE3 zmCS_s9)!e!4_67=tYOHOt!LE9nOo#&qHw~)3%{U2S+$~Nc-I9*AB;>H-q5> zF_4Hb?(euBZLiQ;ZKP}C~2JFX@l zfU$3f-S-yTCapy%oZ_4EU@zXto2e9_J~h|+6n9ej4Dsx0*{;UWn7(~kJ~dL58F&d| zN>D(YXicE+Nh{^_^AAItk4nK$;zruvhp7m`7Er%Q_^49qzIqPW1wiCEky3tq zErUW71x9Z(w;2EcAOJ~3K~xOncX{869a>wKw2FkZ&x6qufGM^W{%Uje+)TxAyw+DX zWOI{BDz9Q_qUp;g#Zsk-LMHE?YPf8=HdD{izUp5na(ttI`bMit`$*RY&s7OuZ4OQF z@()S->bjCVa8{L<`Aca}S%#JZggh~w^2VK}#BHL$1K9oRabukRyNgHT*+R!81lFKi<^+3C z8A!L#jLz-i56WNJN^MlW$X5zB zxU045MyeiwzrT^9)O)W`RVv#i{v_qoP3Bd=85=agNrz)%__ZK&xL;~X`w}?eRHo_8 zmlZl{nBBpcq)yATkBtN(4cZ%k9wsn%f@@{NrYxvz8|?>-r|XrF2> z9IiK0RNsG8Ox;CB9Ycd?z*Q4Ykwtjj+|jd)4~evF9HAiOLGY zb~pq)H;S5X3;c7X?)1R6nxlZUG!O7~DO?6(DkZs)QXXS6!jSY;(8S8qv7>$ryVlxZ z@YNhoJ`9j?HJ9uOW}BG9-p7+;BNY;7laZybktzMb#I(zCqY%mhv&3NdbDrLkYDSc9sBO@@CyhdfeKIj)1>LprqC{_g_OMbY70s zzkoSB6t@B0GZ{mE%8PbM5gici2Q&Jd=*NlkIB`2qT(=XC#|`iIiK7d88+O*I`3RMo z1@AIj^VG!xvAZ}5eL7%M9sFV0QB)+)?5{z3Vc7@LJ{-vC3i6FqMu2eSp%zgYKM_j_ zPm#}s1))!k`LbCDO?Ic${>QFT-?%O)3ksk`SysrZmUFGr{3EM@D3ZY&$APrt5B@Kq!X7fromgE1oaIt$H&x zu3_XGH_)*XhEhtS$PH4YamcG5wH~6{oYX|4yO@2iXG&U3L?O>b-dsdb| zM`BYj_}{UxT`TObIdDzJH3{z@1AqPZPyG1n7k>Qp3qODS!q1=Y_;~;D&D69ez>ClN z2-LTdXE6DjVAkTPNcR$#?k=F*TGb7eu1P>_Rp<+8%6CxH>^UuIR&6~4F>R_{bp55# zkBzCg6veWmF3;Jzw)J+AMk7?OP8eb_`I?180o)r}$*g-v`;z-H!- z&n2=;B7|AOc+Pp_5XI9_#0yk4d1(IV+#K!@P0ngyXp_}NKyZ!-TcWAVpIg#hxcl&mHzDObEu~(qAL!~q)1;a2#us}-)o`^5?iX4g z2A}Y=RHBg@j^B;1@5+nF;tAkH_@;WZXpV^eXx(VD+*dfgrU%{|qJ*U^_z1bLDza5{ zTcv9bFYy~?!4v~2Wf2`gmp%?&e&;1q68H57(vtQ^ji(@LX|3OkxJ>)s-wz9 zWBTqJy#njsuk>(h%23^ufQD5!rSjluW0mjv zRNj0Q_wk)Glm3MyFq&fXMlLE_MWENtC5^NjmWrN(c*_~A#g9{EZMnG--`vcSH(45w z`p~}t6V{(oeTtGfPtU(Y5?=KQkAX!5Eb{MqK8LCzXp7{wj|4u2UrTpFpQ@sJ5N~>N zxwPG~tUGqknWkjn31yhFqb=4)ZjuU>Z-*_?2Aim+_vh_rrB3aeyV~UImhBby9!_~W z%)0PTG*XwZ{GyT4Cvbpj(HNDYMu4Q`vEPVUi-vVSLY-!2p|;IEQLPbP&l9gVZ=zmr z_~Z44+i~DL8ak6r?dB+`Sf8TNn)0cVgCYQ$9?nW+a~=_l$4E?gG*S^>J02!ELS#M( zImw8UN{6XV0JH*)5@G9&R1~x=H12>-)0o#riXrF>oCgEia61@BCp3cLs%5MuY8YWq zu7T@Pd^{BQ%O1aeUHI$&`wM^l_YeH_*H8TQ*H8TX@e99x?a+zDW7kGXm;6H&gEI#n@VR<6Q7~UK#x(-89u^yj)Zlw6BX0`Xk7$e_*;~Ef= zHC>GjMLBfYQMJnU-)AJ|IyYZsC%FkXNNqgV0frZih zVH)^ny+X1U4Q1t5jI$nJ$PocE%krA(Sik&{zlPP{g0 zDI7dhhJ3uy-+q*S-K*qbS;4XiGbW}JkXJ5w8~?RpQHisYswn<-Dd zUw4Zq`BrJBT0e3-64zJj9c?tU-Z30J(fdR{299gsI)*n-K42Q&xND(RrSCEtsrB!> z`-rWHT9F?S-NDT^eUYn65iX5Fx{QsKh=s1JXXp6fFXZmeu#)c>{O25)JXT>;0`+1b zmXfNJgLzX^XqcA#;qei$L81hdgk%qKCEd2Oy+Gr zDcQybG|~J%Z(Csxu4j`illzj8zmr&_;>V+5;MbI&*v=pwmxqBXl&+FdXT!F~{R8hm ze&Oej5B&Q1!2A2auluz7@gQ8nnA#y54v34F$A`WSick;T6&;v5KstQeYR@bAP<2Hr zz>(7n%4fD8&97^X8Dr|@S)J@?YX6V+LS`-vc^ zMa`ps<8h#;d+;Z?5mn)k7@S4Llvu=8dpb%ANyavUZe}dTb9fw?(MZ)ZTcz>Mw@&_^ zQU_2-&?K8ncA>>@r~5pm8a#ZhY4^Cb4W$AFqe?%0$||dwUi*+{q1f4<5`;~EnzoYM(?y)#~%%=dab@pe1$`f|hRxl%pT zo#!${XD*AbX_F?bGY1tCrF1q@vTvlkcpc+{j0s|3c*?vTWI>t`pjn|q0nUj35kq`$ zAb!J0`L{Pwwq?3XBc+f=z|kymy%D-={%sp&(=eUlvcQ4sA$WWU-tQCd9|J#rUik6X zPyG1nCw~6*6F+|Z!q1-{`1mMZsS+wX3pUz(a6vz_^(;z9pY%Rezh{OxjZ=cPD8H-&nbqGQLq`A*-7T7~UfLD(VHlkpFU<6ZrBNe}=xGYz7F$77h(Q@e! zRZ?VR?SMhr9RDDgw35uj6#Jank%eu+vZm~bunl}fqF6Id%YStw;=D{5b_v@j}IPechNgGF1(v%ur_!NB6*Gw8&~& zu?J>I=ap+x?$SyjMJb!8O*T<9?q!B5#m=pG+D2u7qptjnijL4KHPWYMDt@;>qcK?aM zs>&n@AmR-I&z3PBS8r(Jz~}@08W??|A3oXl0Ox3ysA^7YP_v?LbkOdFQ){%c6XR!d zgxPZGykbdXZorhtPtr0i;tiQkzs;T!v5q>5uN$t^$Sgz@=_sQp609+!Q`1#|BkM|4j9~1AV;Nw!9(+Vlk z?)yI%$P*Pgy{5NJ$wGq#7f3IgBIO#dIc;JuXh}`QBU;%BYu$+{-_+^EXj@_JK&fS# z$ZK@@-zSCC7U2EKbpNG((&fEG0e#?QK2`sDQB_5~ zfbzl$8x+IO!~Q}VGqklVJfj%os_FqYML@{%CeiYh`u$9F^XiH&jJ`xTa`8V#K(9AB zRx!Wd57i@_-qhj?t`R3x)KU@fZQF4FX9}q?Ca!A~Yu+N05P19WvtudtF#B|IbvLl` zbE>Z22pa~K=p2VkJ7|7;JQ64o6RA!GT&ZGbG(Wtl)-fLyDG;TxyP}(APF&}7PfVFi z#rW2X6*NkOItybovN3aX*ru4*)eW2(%D7m6YZIO12Upi>&*9Q6MPuO6ZBu?6jKigU z3_FbK9Ka|v!E0hX2cADK{Cpqy@$2X+gi&H9c|N zWPDNzg#dxrVUs~~mc@W>qEuhVy%08&@*3&RnzHR9;iZ4aOQ70UrF~II(PsX*Z&ti_ zfMoF&Wm=(ZS4^a=sIqY0xu$CQJE==4?YU^@3I2G3xTE)Lc~)Xdq&Bqy!r&~~BzSZy zqPlmk4XbRQJrTK2r}(c_+yg~Pcry@-Jq@e#?L}2U3+D8xClY-MRipQf10ohk{87gC^Zxnx>pLDT?NBSQMGp}nEahCVw+Orz$7F)#Bv9@Zp0 zhLf&pAvMH_vZauAQf3(YUqa}c2TM0zO_6r<1C_wYJh%znve+n1>AI;u^rhSXC28Lr zm~?H0)DoP!kJyp#N(+oKfkYd7quU`meweR*lJq;2%IV%ImX!muLa=MTE=;}hz6-yDc^2Il ztD%%Blbqc4Vrn;CI1}`NmXp8Oewky!P2TVEd(i)#HKP3Eqt*1Fy95|tpGe)a0`E69 zIemi58Cu|H3#?{mOe9R6UMRN_8IqF_zImsM4@`^4M|Z#M@TpXy!b>3)QgGEzFfhe) zSW=_Pv%MyMbdye2{rU!$`b65NWx_ROT60y&KuSgG$VJx3bf3!UChbcKVaA{F6Y zg^I??nTxQ@z`uZ=q^5dhMp3ECgPf%#Xlxz&M{h>vdvttzJn-#t;O%kP_s3B(av})U zizpP@)y6+TQrc3U=-1&nIl6&?akRaO84h#bkBFE1=BY ziI)UxFvO%@Udgiu1h&SIZhz&sw~oid3d`nlEsEFe1s|RlKHg9K{PDz(A1D6${={E@ zec-P@zT^8}Kk)PWJKldj@$vD*H3VZQt|1HIU8Yen(iQhOSYYWgekQ)1s=jBx!aud- zeX=AO(XA<6hZjK3-t9M|V!8f^_2( zDLPM3`!Et+Z|PNH^+H2J{^pkmdLtaYVMHO-+2ui244KrAy(6Y6!Gvnw2Q7UT1VTq6 zhc){|q4{KL6?CgWM*1;4dnYy^6n|93UEel5TE`o^r&Y2M5W^QFf|OlW;6;GHFPR;M zf{tZ^3MZkCZkX_mBo!V_-eql>7M^~+((x}CTwx;?s`>=slf$a&*9-htq#`G7eE)vG zw2pmZ^DiC%5?y&44js|Ewu7)uWf5BV^gXoZrw#)N(IZ_;(NdNHC)rBeSpze4OZclxW=I2-%m!9j@Bk-pJ-#E z&nTc=+ULboM%C0Xaa#LVe7}(1J5&L@u){%@ir-?vVob4NTiGf`$%(VfloHdB;cOE> z`pRBzhn~&zn8<+z9uYO4ObyD6p-rgrq!>CF#kW{mTahbRbSFN!@cm&bgpIB!De=P= za33gd0bzPJ&6qClqv;j%SgkAW29#24unH~LRK`%k5Lh8K*B>X($I)p8+chT6dE%PD zbqUVv!g-pP_B>C!p6`Br;Cx;<&w-Ci@f?aV?dOG0o3|-3_UH?CS-nr%1QU!|o=4sQ zF(JB$2pV=63o&iMru!#Q9cX(_;L2@^7zJ$JC>tWCV7jOj^j0K%)g$!_rD~B9M>uul zsF89inuuk~M`BPm3o{E*Z<8FNO&Yecj)K0aNwupPOIQ7R4NQ;bSkhP7d|y<3oJEC` z&RDw1ChuIqvbNVL`2EUar1(ii1?)0ERx5JIiKtsJc*YShw0BcNY)~)S&QQyq8c`K< z5qPVNPxj{9Ru?L56y@lFYo0gjKDuPovARh3ZmCSvU9KF2&QOIh8k!pFztKN5k=n`< z-6>K;wB8VX)q)1Tkx;d)QZ@mX1~}sNXpBeicpTk}C@Y|j-f?gVM6AS928%0njcQtGk46>ovF>$D;OJkKRQ@|@vxx4 zH&1!yZd_aSlxPLx+kbw;fBxq;eEW}Y`1Z}h0$cA;pA_o*2@JL3>CX?FME&)FKmL5j z_wVm`|LYyk_b28x5jGZ?;pHk6zJf5n(LwzNcfUoB99Q^mMNdl}MqDpEbq`3b2pw9! ziQfUKFgLXR*!9i2%-eKt;S`92&SA*im`1sjb56SgBX0mR|GfQcEU>Yd(serFsbLH& znAwX8%T;j9t}f*6(HdrNkfWPNX*7(+-qq$jh)tDzFQ?dU-<#*9vH7g1n~X?XB^dgI z%Y9hdJ7(mo$(l%ow}Vp#8JQ#NstfftZMsoC*=1&lf8`QXToT(&vu64nnG>L~$X^bP z(F$z{WOD0iSfy zD(KfT%~FX&3ul~yR!?0PX>qvcMTP_Klco*ZcY9rRLi|Y~m0Am>WtQj2q$&9GfCo+kkHiS=4fH&y_#!=RiH%8qGZ{rELDX z4Havy8spQtcpXw{K}U${bWdNs881Pf_v4lRVt4+V(mp@b?}ZdnYRZ++CF?Cr?-YBE zGMwD!(23msqv{GTqU^FYbdO?)Jr;uI2Kd$pD=uLLF~`~w*-&#ThI>`R|1%|K!7pD7b z$1w8lm=o7jTvKq(PyuMYs?1~F7JR7-n0x=zB&+4P) z7tTCLzJ2PqQ*Iu^rF`ALFX52u%#|5+zrSLB_8-)&Ad_bye;0)XKWCB0E)``5A+WwS z-)OBD8ERe{l{3$K(_s>}$6wcl^Ew^=WY&n&zn6LjiXX<55_`nw5EDtR6{9!U%YL(z z9AQUr^$j;`5QI{s2K2`Cn`sNkjY%1SFfjl$U-OD;C*n59qK;}|x5*-J>rP?)IO&x# z#|PxNz`;NQgn;BaS8WbWAiZv z__|Uq6`h@&JjRqp@0lfn*&622pyPn_4(UBJyX+upFI);OXo>}CK6=OO4M*?Lql5IY zDW*6Pk{PUO)7@?`2|b!^xUN;sLaKM0e))k{oIz(W#&9Y~@LKQ^bLn`m4JZqkI&CUp)f-TSzYA>Ah<$$~E=YQ|GDuXfU zlR_$E7Re_Z1Z~Bmz|Ko)3wgZ?6Is&D@JmgT@kr^CGn@=64b7+04FYsce`sq`8&J-a4dV@R8J;PeBWY=HWLE8agY{Cxkw z&-W*Oeth8P`v=bR#PfOLe4h9?pSUiAKg}S#x*UA?p>UQ2gpfWjv-Y&|k0gtzc>lR^ zA!``Wz-?p)zdrgwAVxGq&^htG#*-abrXOI_9tdQmj3R2=Z zm=>r$C4faC5W&zUG=$HgteBz>c2>bZ2AimzI4u=lV2k6C}f-4B}I$ z!*b0I^Dk8maIBRwCx#4P+y{t&c^r_(1Nu0ebV=y5uLN>0q(UMnjyN}qkZYyAi`KeL zq%s6`!(NwS3U#c$o|oDmjgDv>C%MRSIFFb$=fR|vn7j(97b;OSA*@(hk^wVhVFmyI zAOJ~3K~(iF^KByKg;ebf!bycuL@m^7)zI7bTUcoAaVdRq$)pXuKsWe8rJ}{=x5cZ` z4tCM>k;LY=x?52w#eHgFz_$W04qwU%&YmzrD z#RH681E`a-3oR)dRy0|Ys0{6;1%_Jk+_t1b-k34!xnhV~6LLr%q_cTR3>`7~s-%?n|M3&wfBeAn`Sk0=^Z7J+cX5)^*3#ugykvYL z-I{VU$OkDTV`U8WoeiIH1>Yc0Dx5v`J;lLPps`pW`PNJ4!Or`^L1{Jh_aHE$v4>k9`$E#S3BHQ6w6~&yI$KR`+EOCy&B! zk@~LiMFpwP@updnkd1;vt_0U!&pAQY0H~RH**AG23wvA>D6uJ&I3nEte{yulaX7~Q zfFymeC?{1?Y8V%jLf*bP?eT%;o++7!PvK%pW1cH?3;rePcK?Q~FEGYM+Lr)=oJieu z*kx+u^17TFr6*=de#7$t-{dMp!BFheq43xCGnS+yd?q%L;?{HG2&ZP+!rxw`eeCke z84J40g0hGb*<@5k0$57iFA%=gF-|>MR%A(lB3ofW*vN8Ql&d=fn3h1 zo82n9?fl79tp2-OfOVdn_hPCH%={TH8!|5mA8|~i+M;NpPNk}@pI(Zn9dbb%DNNJv zhXOG|ar-Q3oo%nvi+owxIkNEb2EmDgXUovt@-P#6BT5v3LT2m6WN;`@3(_J}S~a4Pu`idWXyHt%y>vupNfM&^|3^N<8eY>1* zp9IU^ye>Yudr++YJ9q0%;;N-7&Q4_v${@)0wZLZwm{4Q7aWA3@4N$P4SaBuV++}@H zJU!2+7gJV9oi6d)Kt$m=YK*3g)>fICivg&|r6ku{&4 zELj4dcTkT(R3r|>;%jn97KMrYVGifD_dsQ$tAb2JYdEfBN~=<$J< zu*1yLX@og#L@rcrOwX{fZgxq7uhgpY_b}Pd8UtuuF(%NQju$egFoSYyXosRpIoIx~a`5+?uh(1|K! zxQ}aE!AE2!@XmH!t>GBm4{fE8qJ?|beprW`$+|=7MFQyP;NF3D0Ii!*+jE9cH1#j= zxff~}r}S6vZ5G1HZkMQk z?(=3EZd><~)vb_E(CdMC^IqM3S}(o`y-k~RYlz&*t09}xPwNiKhTo#*S{|GD>9+Yn zbWNXI2Jddacrm`lp;8@KA|4bg@v zdBD(fK-&dvCxlOzZTj!z_szUK+TFe zsgDngDd_!y)&*T3anP{L%wmIJUJ}#zKBRdZ*!2XaZ>JIv=SFw5%eZJjbs< z^UtKiD;|Tv>HSLh%&p0_I+puBV(*L^7Tsx)T0tREU{yOuWY-nw7Ny*<+=}dStX9Y% zvvkJAJ`ZC5Tu`NthF{OWhb>MB-JV%4^+NpI8WqE2(F9X>bkT|+@`5d?ef-Z9QZcMZ zvW^)3(xwzIpvOf^i!n1661I1;oTzSIp`#x-I^#GF9L)%Q|MATL6NZK+N1|Ji6DkzO zGPlJznX?s#Kwd}RC91EV} zZEiuE0f+ELKy;^bQ}QWa6jwBDGNp}^a%b6TH^Pt8l`a?^Ql-e%T!$YyXV-CIM7lbb z4x72#g7mP($Yhsxhzx@)9p6C58|d){9B-iG0csB`s;DJ)Sj)Yi#Yi+2(EL8X**)`F z^GR@?jVo!0ZiSPk1asl%=WA9-nOdokNtS4?LO+5L3smAW8x!4$RfWh_Q{8`mGF5+9 z&O9m8p*P!STFH#iRZ#84Af?}`>H+AEr_HW4HWA*v{Hip=I%Qc-Pu*lEDD*%r-*3OE z7P(;1lQLB)euA+0Vg!CkbK zQb#GI+I>nEg?bW8RW{Pq4nD^%q_Vu$^4edqPrvuS)(8i;au{8mZ_AL<^uNf6pxDKb zP%IBeG)ETNF9?mRCmiT|h+QlI4TA&-8CacYf@xy52^s_31{l){pm+|c`i0^iZ#kv1 zhzDTGg^3Rg0nQ;f&SA1;mjmZ#H8myN0Z7UK@9@%y*aX4@+e5Js>k-8s1V7Il-Jn>< zf!4Z5QFwEsu@_8YUYBOR>fYU_+JSx$j&2QBwCTTRblUkP&LEl=x&@|6mJ3PG<$cnr zw2Gj;sTDHRxf+Cy{WgN-PIz5(TBkB04!iBka%dt32RJI$Y*!8-$ieKHU>X$5I<;D7 zjgb9iV7u7ldKR>O6j0)^B6EmYCA>g7&w=N8;(R{w@qFUrdE(>aiSx49gfUGKH0DyS zFjfD~+S^j8N_DLek34YMQsV!2FCSqa^WFC{t?>c3z6e!vaYcvcI_Ms5;%zd=;`dtx zXxy4;M1G$}W?55!#R^~zUBL4Gl4l~QL)3KIxtxlOHCY2m#dZi3D1jD6NX3en=!Dh@ zqSnNo@tmi3Z>RnJ-}>KgPNZ}@5Z6#P#ci{p@lRks0^XE)GMO4)#u!OW)m!05d!|uoC#GK`%A2jc@QE}cV^z$qb0`Qk#K>#rszA^rGdYN9!LTG@0E$&lwoh21AR8x^nq z9^M>U`l4t5snb1*sepgXdnztE|uoqTXw9wn!slrwC=8@W_RCFtVUz2M7W+BDs z*^%|)M;E4^$$Xt5fySIsBN>>9>KzraC%o?S46JA1mLmrBtfL9&PzX&|Xc#qdvB+OvTU-TvO1miT1o4oV2pXD(x|__FIOl-y?HM0tJQ> z+@?3I!XFY{^4=InZ#a&Q$Kx;np+H{0`F?twc%jOhXWlh&DHYHg&^vH+LO&SPX72Zi zcdwB83Xx&js&mdM(hTf<}B?~*xfF)mb++x@pFvEe1QVrs6YOA`CYX^So8YM)O$ z|HrN|wk%i+@0D59YB%U|Vx*CSOz&GwF&0mHL+Dnkb<#FECAfzTK9&6LuuagVfZ9E>p; zZFnFm*N`){RCg_1QW1!@LV@9807Mp6#6UY0$51>5aEWa!LJeqaCex;=U9FvE?gTPj zV~*p7D4y5E`^SmzKR)p1pFi>a`v<=Nc;fy2#K*_L^Az;%@-b*?OruL-(@Qsgc(}>i`S;!yX;Qdoi_evys6NQX&aLU=#sUJPlj0c-_9Qt`PIZFQME3w{-D zw;edP|J=3z;jJhWARSOy+uZfUm_sYkh-oiktzKs3JSzH5+CLFQb%1t~^7}P>V3{GLae{XA+dY>pGLp`0+fwkb2@8!|e;}GPJo( zQ?xXSGB$>(18JWX1)4Z>;hUzjd}s^9$tBaq3!~m#&Q~z!-`?=; z@rJkFEK902JR0LDvcBdOKS$+H=?pMr=2$ZvrK8D-qW*k!s9IMYGLg zbzH}k(#2UtNImF<^rd>+wRi>SQ@_X87T=(>K?uTPU4HiPIc7H@`m$4Kg6ZvwNz@#h zUy6t=BP{M`Zh0@StL49M5oRYV$Pm$qc4?n_hZiy+a^$_i zTjnn@=3o$C77Xaf-4#sReDU_FT@dTkd=Iu+0@~m5lH(2NkL8teX&)RKAz}EglBA60 zO>EO`g0rvz?5&qYELy?I(-#?$B@>Mu)8#cG-HFD4t_gL`Rm{m#1seEsN#cW67|3EtpKS^?n}Ejzb(25$W{3Hd zO>vn>@uG~13dV#%0^FlZJOMgwD&_~m>TfgkN<(u$a%(2V3z=X{_Q$lq^(&=)HkA_b zf+D>d))lvjl+lKp1Etc};ba)%xenDnk!sDG%!lGQ)QTt~Oy&|!N_$pZ5YN0AJK1(euI#L!v_kSU?m92lGpjNF;m)S=a=hp8sB;4H0pUi@o}Db|9E1| zQPda?E7hPf7NAm%x8Op&-29b5rB>!DNmbehZ?Kne_S=vno+Y}$OQ~eN?hUq;k-G9x zZ5jod%gW&Ly?W*$T6;FR98X;-rP1v7{iC@zTGMOY7J{GbKu^uyKyo__WiHuj*d#12 zfN|bz?yccy4ZSmX6117Zh%rDIc^_`guYXqBha8*cW{`+US_MFu0!cHuol3x8ya7ja zZw||hI=cP4M`s+Ji=5A7LF}Q2cS>(eRTzPrRJ2au?h90pMJoQ;E96>8iIZH0U|tiQ z88fB>Nh(_hjp_cV%!c5VqHIJq@;nQ|*7%-eN;FHx-`FBiO0j1m35FC{hYdLI$pa-~o_RF0?B zHQU^RV7G@wbtge=9=Eqw1ns=b<)vfr6!b!?RX-1Jd%U!>uU<5JZYkSz=(+65RGXEylw>c+1p6@u% zrxg-&tk2U8_pv7T-pnKASMP^mtK%T;6DfO7#EZd+{w8?5DIO2S<4_!jwNR!{E2dBR zcEkexl0l*+!K_L_SMy|{%hzOTq7vVhSDMN*5uXda6!NN+d(N-mKY!zhSzXfzG(!3S zLdu7wE?M&KTh5iA2ko`-0$_QZ&APmy8l=9QNR7;k3b4d=4VU#9i8H-)YG+H3owxW! z0=IIIYV$|jlbWE10)5ab=d1o2?S@z?{;b#Es5B$7Pii9DOhrh)wOS+YkK zVT<>u+gWjW#d$0&E2R;F7$i{0r&pe<`G|GTn+8Fb${fY*ChRFkD31v&D+ zTS&bi6*aR?L@gyoAZBRy1(>yl584}K@;(a^eLM_t{^-jKWwJfZuMSNvM&DcJMOoaJ ztdD$5ZiQ3x?|TE<;V4wmj6%?980C3Q%+@g4#AHJEin=9kadpuwS=|EnBAgroNVCH1 z65u%$y#bFC_!twTGv?7Tn+2qPn}VZ9ZWcRiU@uYxfERWPfU0s-)WzP|QfcgQn!>3Rk0Nax%>kp^w(Wj8Mp#)2UUufA&dbT?7=^-rnR8df| zs1;KYW-NY17MDc8vAz~1>getC=;#h4C?<&2<--_x6JUtfyQR5@l0Y+M+m8ZLDkJvEn^q zDnyW=6NlW$wc$}fML`t--;$Yc(|sy6DE{vjQ47=(68|Mn`}dlw7C<~x+NYvGb5yjl zA2MPODKMfiIg*wL6KmK30-6L!1HsqW)C-x%gasaM;l0jBi{%MT0z8CK(Vd4 zy>}e#z~}=823DbbIfzdkJ>|p9%M7i2RY|VRr|B2`G z1K0WVX_Txzj{~A*XBq=dctO>Vlq6E)3{<5i4xE^!kP;kmJ*Jg5Dj3SXN8cJhK3ReS z79fFAM9!Rk=G2qwA(}m;Rlqv!9#&d8eiTIrL7IT4SRqUHlVUQG2i1r+C{qW}I)8k@ zhHt6_;1_w+RH(yRJhesAt75)yo?HH-?eFmXF)S%xhVJ^6unG4^U8l?Zh8NN?!Kun( zZ!{BN^}NaaG&JNz-iere>jcg9coAh*?*>yX9sF&|z3%M|H?h_Zye_eNKgVyRzRJ^$ z%r|>|3w2J}EJC3qM#hg;xp&1wDs4HGHM1(4Gd0A#G>*8fiB!!TjkP46^WjziBNq#5 z&LIMfr?Rbq|BOefnv?xQu(j2AUFro-1kWzA60Ol?lKTipZ+JU8j)xafZB1oj!j!>U zph6$H=SG&I%X26aF3}17h@#4ise^4IC64zHlZKN3qcw~^(U~xa9c)o2VJ>sMvC4PZ?i!j4f^QN&d^80rHY3GH6DjuL}rptqo)Y^Bus6XxG-^{ z3*&rR5p`w}Ww{u0+B8J5kfc?rp{g62)Fgjl#9hs=8m-15+H%wPbNf7?e?v-dqZt$dOoMe(72QgcoqH(TA}B03ZNKL_t)d zadTY1Z|p}$KN@;(F6RJvG|&B5Nbef@Cv)PwMowCzkb1iwm~S)n-@SE%NO*W1HxHz4 zE2GB+uT~sJ!h1QKhy0^4Wq6?)V8@HC1K}hCCKd(G?y@g%-(m?m?qM32{uc4tZNv6iSxTE;lUhBW!xq zEx$>MDPI#KU0u+b_*5h>me~@ZfnRu`^p>fG5(IpkL?NWd1q34G$Y(DAmG;?J;O(`- z5vp4&>MyAPh4r~f<_a$4@5)0A4^+TkEL>V-=Cyo#yEvcLJNw!^BJxQq<+^D~P5orD z`1L_4k(SU@+rNwrP3cE~@|p-}Ns*ORplSpV#t>5h5(cy_@~pBGxG3{}YJ6aP&VeTm z>CToHL^1X6n60CE<8>Sd=9oEUSJljcvw_rfPj8)W4=<+712Sg^w}$Brwi&RBVXBU6 z*hK32{J=G?B>UZcl;-)ieIN>|espVy6hH=dfZIgtg4QRv3;JWCKPHa1iQ_TR4r^|b zar8aln3nV_Z%=Vp3>o&>pI=ceQ>Owcp#aasU0tV9<=M#fN71O6Oy0BAm}KSF)y4)^ zG4-iO^KY(~BXSV+2I?uk&Q$mnSVDxm!{0aJy=iL_KC@v=m-DUtY>|5-H|p|y$?IH2 z)Ta%^JRfwB2&OXFrIlKsMZ{#lu!vvriCAQvQTT-wTN*_5G^xHyhT~1TTh8leS-s1v zh$_>rQn+N18e@Yk7Gz=NL$oP$hy3VH5z__Oh^#4Vbn_~tN;Cx0eo`Pwbl(M{K^&f4 z1K5b4mvtEra&NLGk)*sE+oR)X zrI-pCAE{$M7Y1P+*7ap7Ma9q1dgzS_$AcYeu?nhHNV%k70yGt~vw2Y(V>le6PNS&E zw5ABIDLAL5e@@&pHf98{+2M~#@piIJq*}vZ#_SE}RD65ucsn}gn_5?uG`%5X4osOi z1~8=IkcKyH=#%mEiPZTqah`(fRE(jRlZq5$g>Ak}q_EJ_HSkWAHbW=*Q6R!36}pFC zZpD-=jc}P_-Q|3Nu-qazb$^z>P)BXCHwiV8g?P6~WiuZ&1LP#Sr1oghNl(eCOLz2Z zykG1Ej8P|TqwL)#QLP^->zD+L-Y_rTCsO`%h8Jq-=b2)eCg+Lk+c&50c_-ze9uAD0 zNP*RHy2~f9CQ?)u`87@tA6IsrWK9f|dKOb)n@WX-oAN-id`n|aT#myH12hUGr3U4# zMNkS*BrDSl{nex$sm;Q2D&@(b+xN7Lqw4x`pVY+oU++d3-6v8Da6-H6ERD&rX-!nD z@RwvN#YbfH#S&sEHGbiKLvz42i`+Io3t3rIX_&U(U5LQG$rGm2KiQkYl%A)J&{T;?HUGR4x3o=A{vZA^fSMm{Lzb zrJ!C14Q92&0@cGrIk*w6P5u{!)b#aw6;h*|26ayw*H4a_IfcT_V~AakNZ5&v^z}lD zS#oM>(@yi460@)K%NVXOba*y#m0SxyeF#9=SY~k#1a%9i?XKCi+}V`@yJf0{Yhp>3 zp#e{3m-Y25qSYKz(YsGILe{^YCfZb&17#jOpaI zC$8(XwOFKhzsY5NUPK)|_ksR6>^+Dgio013j){H#EVpNPqg-uAS+r?SK;#LwXZtVl_L zHC9S%E=wsvs7Tgi8s(I!NoOwlhxCkbnPAH1gUxoA4^v``;yTgF^Mxk8QN1vhjd)sA z8{&;ZS@**~DD8_Er&wFEW`_d=m-$htnAaN|Sn|MxK8}t@?>HPl5gAeJ=)~m$sge%e z3#zmhXyQd{F?DoCKfI{w!0~Vb-L%XJOqYU8!R3&Pi_N3eh)R%GB&*ni8=Bg42!u7I z3c|qPlCmsTOnn@Tc{FIZi4=~8q3)l0gKMBoFFP^}&Im!*BusQn>QFv#Xvg_6Y$|n5 zTu;Hc1keB~RDl|?_X&;KrFTUTiNoZ?; z(KbnWp$Y8QmI-^Cl5WyId|le7`~6HI5v@>!pi_NPaM))H&os*Mj&}y>7NG6j05a_U zA{=2&W(hCSKK;~~c$-EQ?Cut6AmzDTTwZ96pVB=C3OCOb@L9IUzn1+GRr+BBcv=U` zdJWN#N>KWmno0THZ(cJaw^IIIO2uCn(4f*G-wUaksBoEeA|(Z#zW+x0w>m_61NXe} zOMdHrOoqEgzF);=meV7XLR&$lOG2B)W|79bN0WTriZz5<&&dWw)(n^5hxQyallGNJ z6Aft}FMm|`f~rSB)e{)d)xE*tnVM8&um0K0%PzUr#Igk4j;Xk=iSrVSVG$!EKoeJZ zq*55V_Jf#6r9#1`Qp8?RHv7JJpg+W~D5x6RVGUoiBG@4xwDP!O*{itgXk3JEN&WY= zinoAP=suCaTHACQgh2tOzWZvmLAYlD#Ys54O?CrCUy@m)Ak!~+r4&orM|Vp=G;Jyv zMzrKr`ovMkr%#wgbolud3eQfX+f``h7R@qT78RetxlU#2b&QPY32 zkh*9Rhj|@W%-p3_Ti# z0OJx|=V67=WpueYubfE99O$!Q>cEf-Q=T|3!u!t?&-V-0b7BrppO=+zZn<5$48GMY zT8bRvdN9%#!}X%VET2a;v{x%R6Rjli(mXAVnH5^+XwLi$e^d9rLojMSOwF*#dIH+a ze^NvJ5}ptSqi0{5(C&Iq*mX=<#Kln=q>hi4JgJWeCb(g&E)}ci_JUbY4c5ImLK{WB79Ta&c|a`@lVp-%3z$&(ZIR2^B~63Aqe5UJ7(|7_Xg4J?>jq5(t%_Gwi7 zS)}26p`kfqKHLvQWPYW_4b3N8A*Y_rsCY6aWIUiL?ek_&3kmXyrg#M=gul)GRJx4# z*YE;rSab3kMb0IXIX=-;BWkWvXi0FbR6%Q~MTmO`H=s4Kt20^`^llSErkMquO(Qt* z>`XB!T*1ys&eSE5rF+#LMt)wDwYAp6!90?Hn4Cz#B$Z6!pT&f*K$_mxgP0^z!A;}0 zASiqDiIP@Vl>SO1`;>l$3)a;cqF{4*a=dV;-ok zu;P7@vIDZJqKW&bqo|R{A*^8;^e^rJg4P2_oHcmw4&{1Xn)H=WtG7y#w7RHxKj-Ue z{@m3f|I3Bcw$tYBvjL}Q$MFU}e06PjT~Rc3#=%}lbuU)f9*YJqrhG!R_HmzqiK`Fd zrn-w#ax-W5E^!+4RF6<08S*xFYFqxeB=leNQKpM-V| z@bdyXLmQi^^)dqAI1L)nm|V&wiF|HTFvSX~Ar1W^{QU96^W7#=GQ>m9Vp%h97T62J z?CAJIk_JB;b;BiQAvda9clCK}LCm@WQc-Y~wQ#EIZ7r;cum$o$ccRwU3aPcB1k^39 z-MVOU(Zpa4WEE$ZeGVVj7pl*;K$Zzh5c`hatax&1U-v6KQrF{wJ_lOAz}@musEcVA znn0U%u4V>&noNTyECXds&^7E(QuW?UR=m6K366iS0F9}d0m^k&^E( z>P-sT48;i{AjlSVM%g@y+#1dMi&hU-Z^5>=rXbyat)P{DR^_PP6KIDAqK(mUih60BfKf`-OMOKpB{2ot1{liT*lg)q*C zA|WC@#$~eoYnWFmNpBkH?3E?hngfiufm@ZemAwZm_lnlgCYwd+Z_P%Sy>`uDV*_}E z&Px5Wh}^T!^6vEZ7&9Yt4-0GKNQJbI`yJdxyul&(l5P2o^mAowQYDyXyXBn+XcXCIpO@fF&3D4F(>08>w zU4l-&|76LWa40A9>Sra?KK|F5|K}aIY*^I{?^{nG)nP9-d|F|@P&)yKwGdJdum4`x z{{IeaVg>Rp0L1Dn?GOm>-1`*DJS8@wHvR|0dh{I% zdV51ZOU%M$>!7OmAyQW5trO5G<-oQ^h|Q-_N3&d?gBuP~9Mo`;pqJ2Hg}7OAjYDID z9Yl6GJ_oef0&k1fy2U(AuXb$-`^Uyc2X^I1F7^u@EPPF+v| z39UF5lG{zF~oJ6$gSQ6Kz^EV@~UMDyhp1 zO9xNIWRIam_5-^VqBLqDH5Ygy=zAoR84#+p=tD%CsEhRGjYD*a)Zr3hYP2{EDMQ=p znA`<|Ycf?}gbJI&C8)tOsB|s#dOMwyPa5qo7Za+ro>?KqzgI}4WAM#S*dDj&oZ54Yi)-q!!OO=&mznU~PaqEx&mT_Cj&d zuJZe`FIM1=2I^PK-&QlEO5{TQ{$CYVUs*Lt4zwtwa6q&}VV9t-iIl{}i#A%DC~WHy zhkV%EpF+Hf6#;1I%8+uRe)!WYgq3cYd^U;dQAAPO@aMgEXrGXyL+4@m>OL@sHE3m; z@%%dpq(ly@FT`pkrQ{ITY6vhfZh)VQVqJGnJ{6Poyljk9SF4DFC8);aU?g z92hiL^Dmpeyy%5!*$h=7%XD{ADd)6_{TY`vr6zTcFsJQ-8?U`LCmhoX>17@SX-n{r z6;uZ-+He6A>fV^MKc(L;py=i6`n%Nwp8+wYhX4LxhDysY?H6cO@i^L0f^xqdQ9OI` zba_r#_|&7AO8R37LC>?nOJ);333V?j;{pg08R-#BIL|hPqF+DfT7D)+nwxr$&IzQl zpbWK|kQ&Gv@hP&TZ#I=DY!PuG(RVqwO`|N^G9V|MV4gk1p0tu?%k(jPlfE%)4gIdt zz*<{y+z(ZWrln=2ARp>(qc{I9|$^wy;-Umwr=Z+!4|pcAFgk-uA^7<| z@bP}(ItB9zDv5`c(Pj-;2EdRK(Lxqe^^*ALlkRm;K_73@K9PN-U9#Bp6UqOb+P5!s z_$~9a`hvI$sTi!$E>CiKgN>z{KXhv|m02?;WM$=O?I^E;rhD+>&!tp*1E_>KN#pUgAb4^TX6Cq?!{g1$Ca8pEVYVX@N}i6EwMaOoETc z*!8`I8kU8BbQh^&XcwAW^Dr(6jVOGCjbIn)?J+rLuCnCyn&(#wDZz%W-7|7trF~S* zzqf*Fg&n^JvX&BTS*GIC4T&G!h%d!&WL;@m#SvAq9epy1+T|wLxE|%XDW7i^j9xHh z(Ls_OWGYiD1@>^s__m_X;qo(5_eint6Pl>xEIJjR1laD)(vWSfQy{(2ma-Jnh5SwF ztIErP1sl1Nc5l<)@Hy3@DM(YF7E)i4_RRy3H-8qog{k6zqFWK*#S>{;GSNwpVG4>d zE{IGtHqUNj0<{LA<_}i{@p?fvu!=IRM)#VHSBfa^=zRDzsxSPxPo$(Dz~PFB2!$UV zM<2MF(@Fi@A#qV;LN1tvh+(h;Bw&}*`lfxIQ*Q?O6$%{rQLVi8iUHRWN}T_F1haBI zFU4zO2cB3SQD$l_bX3VZ_E|# zk^k&51~Iw8O`UR|)zbz=z@j~4c>rgT_NltGFSA1_i=Zl)QX4-IX6?WSFd?3E9!k@IYBp7}=zKy3s*6i>c;5sfNxIojhkMa2x)24+uuwqyZT9 z4FzLDpYVgO;yNbADY%ZA*U_8%aBPCd%xGhx4@Mgo`laA5cz;gIkvgXrfwYR5O)u@z zDcriIc_UU&$EM+1^u3KPRx@@a@Nanl7W(|QS!d(bk-K+%e=EeVO8d0rwQ5anp&Q*V zJW|-gc&p4~3z{s&;DSv=j?`XAwSGVW^y7g(23q%NRCj6L5DQmklYUM$L^6frdqyE8 z5E>9D4s8$s2d(6E3H`X~4K2Jir)-vo)n-HEiIInjX;M9(YDhVtuOx9>t#-)G0yc!d zAQ0$~Wjdd->m%CIfVRw{n=C5)6H(Baz7No*o18%lo>3`=q*0_uuFuQM7s065 zG2lgy5r?e3klM2)`K!{tPsTJ_Bz~KJ>OOg(Uw^;p^tWesF#zHjweEOXt0M#%MU(z1 zDUJ$};WJ~;4x~u$nckaQ$Ra7HPdPn`W``7_MJXyjD-<-QQ5I~f?!u4%OK+s!{#tw{ zyW0zqq>35K5CJ?GtJ~;qDM> zCM8qrGK&{fP=-!GF2%eAqMU*nssk4>`I z8bhXerO%E%>`8`jI+)ScEDNT#W7kIP`x5($B^hi+@QjJoYt_sL- zU-*p2#nCy>ukh>|-3RP?_w!@( z2}$(KL)w=RR7x_Eh$}Adqe8xk_f5-0UCV?g z08}Bb(`b}?bO9HyEJ=;@khSh^9;rD0epX1)T~gph>`n`eb0WQT&|f{ATD@>})4CHz z?#gIdK73|M1izxt9Mjn1CmNF5Dldh+`cF-yNMB7YWzvD-yNPf^L!Gk|&OEHniD%tq zb3w!-pFm}?;}SAoF&42)R`8le_8XfAgh4LRi}Co%Mp@k#a($@ZK-2PV@$k$ZMVj#=lAaP^=G~(XT z)He;!p<+dwgg${b6>TWmA^122=jpNpS%kLeGBeTMv8I(;@?sJ%sNZ_{+XyjId{Nm*GqT%E2e&NTzR|>k!r}jc@%W#*hf)NDO)=}vyPUR!ML{| zG*$HZ!(bDK!zQkYJ_h<@pbtUwi`(=dMh{eUse*W3k`QAOVSvHPKAAO24!ie<{R*>a zr2*#UUA5-u)yQZAtsuV0y4Z>yUZKX=I-6P}i8gYRZj&JK*pn!VLcZtXdYS3Bhy^{5 zrYWlpiJ{RJMcF-5ny?+;lJ_Gqp|!r#wt9alq~czPPVV2X zq2_$0Fl?}in}4d1D;K@!m#U7&x&+3()ol3j=P?+#*pVU?~VKcy!BmI@8 zNpKiT?vpRFEt*)+SC{t9py>OD`y-adLTrj5HLS|57{q%S;Z^X?LJhTT$Oz2~t?;?O=f z+Ew6~p|XP4n}wYT5kz?DcUR4|HhgbUI(<4HsB;LOW8&i)ct0 zAB{1k;q6%R1RgeLmuV^Ebd$%$&y54`5Cso$`JfA}{DqD-R{<5ploGlEU5yLSMDV;6 zr!W4Ja=r2}@-3+v_fEEcVrAMTeAbDv?zu>TOOz=SVehQhb&RRzw5-lxW4;k=Zu8Es z6jEx5*6Stc)_somfQwcji`;EmA;m|dTb)=Ztuc{`B>P3|>u5(uKMp7ey$>(02KpiB zeW3M;)-Dhm3YXlk2NvIeCA*0Nw)@%nkQ0oiCixJTZi&;XrrMRpS(z9jXmesT%N80z zq2rzlRP}M-M%fIMwX%7oxJY*Cmg%tGQ$Hwp$JKfSQDpf+c3Q_EhsLU9vDHn!pv+KK zGzTr%qNwE{v~rI`ZF3LE#w-~j0c`;o)YLp9okz6*03ZNKL_t&&4(YIUsiKRIy=bH* z6&a1cbPSLw)hc`&nHI@KnO>2<4@$rG#8&%{e#%6;xA$n%qv)N$$3=1{nmFAGcff6O zy({-L0WIX~gwix~|71st*NIYTym>)&cX68}BZz#GolyJPlGDd1q%t2Ui>QG0gvVj# znwI-ZzI*(dC$=?9YoSxt3$0bojx{ zBO<`w%tU5G@g(SEh%jh2uyn9a7Z6W$I?(xnqrKs+e>38q6sUyA3bK#1234Qnbf0*v zyVZu!RyaLs5yh6t$fQ|JH755On;)uWrc*R=ZMa_*5VgR)`}w2*2-YFnzP4tE(C!UT zqhI;~x1yYr9Jd~I5~mdUe>*2YbAskDzbp+jE%9JV9hREtao0LH-#iO;Gbg z0SlC;H^?psE(iuhuB|b}DJ3TlOEv(L3cz@P=>YQsj0Xq@8sHwQdjC&*)`mT65OM0F zO%It`?%|X8cdMwi*4+UwdXxMWP|Ze>rd2Xde-8uWEwhk%N8KwV0Yt6&c@k7Y)O=m? zCZUi`a+;rOb1(obnXv?sxb<6h$}?Dg#vyJFnuf{`t{$)CLc^Fho5Cq77)c8gBLqt_c;yK9uO#H}WinxhN#iLU41LpTZH*(S@o-)bqm6=ZW`^6VG#ETy`*p zcg&8)e)BhD>5z3Sa7|3IC5&`05*AI%J6~#=QW65m^ez=)PN1`Cr{kT}ppgyv%C2hciQy33)8DRSTNeX-dD1kF zkp&okWg=y|{d$ z$7A9eg2M}`c1*NWz-^#;7m}KtE@Vcoi%nU@(ZLgJ9w&I1VB}H3^zrNxDDo-P(1|XB zArpNLj7De^7(O9pLLQb7;mN4}wlzA3Q?HbfUEPh!$+YhC7#@>?FpV_mCS?Kpb38IL zOJ&xT6M^B`rd{NwQ3qHM!@@Sbi@fT9Rj3(#X?PCESR}^i05?B0QzeP60_r#(IF7!e z7IZJ9T5DfgGi&Y?6{rMlp@;kK6+zZprpG6?oqCqYX|$I46jiC)g3DCTRn3SZmlF}| zKCaT@XWZ&ON-3Ynl52qqwSw@^CgY$-+%<&Wu{HYoTowS@TN~rCag2Ntkm@DF|!*d`h)yzv}sLAo(()ZSSFfuYg(Y8(;FV`fw%UCNB@T2Ixr__ z78QvTcu{;Kh{F_fY08eNhMxmkg6`LuRmVuI!g;uG+T z=(PQTt%YXAls!wV6RZP_0bx@%P!pq@GP`v9!qf|*Cj=MBp)7XT>?y+v-v~_yx8My- zZ&pO{0m7q90!kxN24oCqv#*eu(msy0+yC%I{PpUoG=C}27L1+vSt1RT%UL}&%sI_k zUVdHiATf0u;g5=puuKH@aNXwg(?v%W_tLM46g;}e8oL4=hZmNiY&izWeuzx zK~2u>k`>3FnW&_?3_54Mff=HyuBmWX#h`4^99cnS#axFMA+&@{I~oMO4~VRNE%va` zBq^pBpFWXl1j(}NK&SAL`uXEsep4m*Dr}>YDHxiA3-ez;qSxmQm5-tEQ6*O@( zsoZQ5n7Z7diVF@FP`|m_utus`GsUdv4c17BqKi%Dy9nCw45~r6#2(sOy&lY^k*e!( zTb0Gupjan9NfDMjnhd5bh;*Iv@VpMv=R!-yO%|@DiCVP&cT@FOHd0cgeRXZU4y7_x zpgtZ%tGrL<7Va(y1TDts*huwl^Q}}z+h&%O&^6QQx7{~V1Fc_ZZ3DFv%-so+;u|T4 zBCtIZLXKhvRo_SjAfw3n$nS+^BQ>mb?5lQ-*9NLwM>osCo!4GzE{h6Y5S03oQXYVXg+=?|?ijhoSZ@m|Gyb0hTWeCFUL zC;oDi^aTo4qmJeqDTBmlQu;6y9Q4r3GU#XkJj6FrY6Be7+eYePjZ~!E!`^c${s!`A zTH2R+X4U0JimVyhTWq9&&UOwmVk?E*mN$n>^vJ68<|+MAHH)&G2ieHRx$FahXr``^ z2VXvfqy}PZ5IM~9C97~s`wB%=yzul!3N(x`OT+R6uM16mBgH3(Piv(31mkiGfHzX? zL{ooRQ+2ug*~~HO{wvYBks5j$HbR@-3r!|{Pz@+zo2iCI8yYsSZeZC!yxDe%8LfvG zjBF^zC*(SzeA@F|SKmoIuJ|j`K3%2Fb;G2kG0{&ODInT)U_GxXexAaYG$A7);WWsE zN&BJ>h=F47uAUk5c_dDHPCu`hcPy6!WPd6Zn2^=2wA8S^^5ufCJ_6SEJ2F`|pEI{6zOKsr0U%4q7~WCElUWbkC;e?r#V#W6I{mx@yq$B0I1;N!4H zYFtvLymr%U&9Ugr7+tXqg%II{;w1x5VUQmnw$=19#UH8zFoY~?sy0%* zHd5Wg(Auy`zX)R}VQ3TsxP_}yPSz&@Cc-CV1w+(>Jbh}i$oxKs@aob;E#bS-Ncmbe zrF}G|eOWv&VRgt*z(3nag|sgyZb`QAEmE4z%fll=>*}gqOX!~7^oDaA6R-&{z-xw- zZ={46eH;FYjnswKxAX^4N}AXRec+_cxoytKpbBrK$UkdSatu{)HW)X5l#JCY#U}5w zBF5~P%=fV2^}f|ogvc${fGJk(NPV}GH{HY#l6jyh0Wu($d6mXxA=8sG)hvU;|4hIS zfH0S`hgGUk2bpOl^Ofs!jMzw>NfoSgA`3Bw%QkqC8%HzMd-vvv7jrHE#|D30+E+lx zw`?Mv6W5Zjqxgv}NvkKSc%W#F7A}zT{L`pzq*f&%N{ZqFSKzX34*BGrjj;Nod@4y| z&G#77MXPhrIkU#fpUe75$C%LtN$Lt?@w$-`r(cOT>*9@6Y@(+BaX2kXmSNFsr1)kE zr|GG~jm)Q|Sb%PW-1!bcsatvntOtJ_D`+Wj&FqOqKpcdL9LiY_ARJY$%3(In5-{z22+AM>3mcD;=U+UZ>v%lF#Poq?rDD9;sNx zEO8mQH&b!HH6Lbh%5{m@x9V2dYiX885+6ejeobi~1I^Ei0OOG7^2HWP#G+S}kk=qH zVYbtDb%#I7q@$9J%R02Z!+f|5sd))vsl$`-x2+yc%E*`4B+d-135qN4W%8&evTl1WCT{AM#1+bP`b1UGC$D)}gJ=QH4o^1YnO z25UY~{W}^d`d1hF8q`{kj%5;+9tY=VxczAaNHesht})Re1#&Qb+I3UPPz;I?Tx%Nm z1bSzjjnKP*FZTK;MQ$Gn3+V~rh1uO;t%gA_C4rP@}Wve z>WRYF0_Z96_E2CWcD=rUMA2jOnetA*2ntmyQR%I@Ppg4i_kV6>({uw++;cG^(2OWRua zS1YtFN_yQ255Sb+0r&|&`@f#9ij-p!D~hNbK9SRI@ri8KvK<4k^WX+^A`8_EbS#Pog zTZU97d{X)EV;dzuX{6laOAfT4Y^pq7fHVOYaWhh(_od*aLXP9!%It&>>(QAUOeP>A$kUW=MX$f%0rh*&0N; zPsisL@*EfX5HMfl5U`-VC_$)KeGot$QWDcNZZudk>n5>FFLL{SDL_Cf~*ntu|8gC#xK%2iZwo;e(_CMZLIR9XqLJ|KyoqY0RbIG#IMlqQ1keng-Es2aCan#I`}(m|ynm34r|Ji+-k zaPmJt03mc!LB!&SU7qKkrG!*4f2riZp#<`yNZ2tOw6dklMr!;`jnq${+i^jtu7_?J z&BbIO@)9O?+8#tqie-_YQ+TghFjRF^yg z;{2R#l5SJ~Qxu0Ea6w^)lFPa&jp~tPelc*8;(*voO*1I|(O)u~o<&?1)bgadtRO9k z#!BplV4^Zc6iQiUsA4tH;dXN`Q}nqV6>kJ0b0Zqg zrMZX=OGAb=1}Zli3lV`_1LHbVN-*53raHsoRAK0qS;F_65` zG&TQm>-qD!9V_4GLbfV7M)}_*Q$>r}clF$e=tRBxp(Q2sSST7O+ssD6AM!n2WYi@L zIkeV!&)6nn#Cv*Po^zufkXgcvUCI~tE;nzaP$UjAK9SZcq~uEzRhg&1g)OQluVcy2 zQXqpnNOgv5mg_jSyLlVqCxgs8*1bvtZ)=iD_FLp^C_!*b)Z0(!a_Jkosp-9mLNP-} z$Ro0uLTL&oj6o(foe&>cZmtvscXe zZv^b5*f31~CMPg>_Om+4n$4wr-LoiVG>90|J<}Xj(8q-~Mj52Uuxt1~`(P@d66x`v zo82Wgh%2vFm$21n!)Crt$X>09x)K}#TrhKh(e{8N?<+BmL{aP1W__$Mlo+q-fT%Y& zkOe4^8Hog_+BS;mfz}>?cJokq79ntfwWp_~wH>WLT%vVBX}7P#R0c8*$Y}QWaIX=4 zX(M&>Cla!*`G-qwOK+~e`M)BxG#*MwTD7u$73R4^qBRR| z){j*>jyb@kG7xW8^QKyai&vV;hAU;yKD=l5V=EQp`7LKJeX_GDqofOK@&#+NR08UlBh81a@d}9+U z+vf0$-H2`6%-nji;01q)>7j?iB`V3Ke%MkU3_~36Y&Tv@&VRwYOX!+?LQyU>@*6%j zf027J-9bFH;g|&w>7<3}EsmUg+hvCUPoGsh*l*~{H1hezqA|**W`;A{Mv7OaScOh} zVbCgK`492XrXUi5UDju^xo9Bm?RJV+=4Q;lOKwq0U~j+*WCS986lu@Qmtt>FnqUSr za_Hy1fgR#olfxP*mE20%L1~iB?#cG5>{D3sY&Fhe$QlCQUPh2&Y^1Vbm>VbdEfiX} zhN}0nk>Z_%^CaT_hNX6Z_6+z{uu@?fBx}nAeXWD z1;VH+{(IUss`_={#ay|Ly||4Dq)Aocp*V9%W?s~SjVm%iKUnf}qMKxJ{xkUoh@!FK z3HhkLo+{r_DQfpd`ky3ou$ThfFDWAg)==I_^^@H83nHtoG*pps z79^~qTu2FZJ{0ArxO3G%f2?P-=u>}T1vN%!BDxdumHwux+G5iD0<%lzEq8jG6G*9@a>8+ep1Nm-OjClM^J(d@3D-;%u95q+DLjJsT))3c8}18mP&+>axF-_KB_}V653`BATBSr2<9JNI-3v z^tE5^fc34E%lbgCN#s<@f?4m-loHpmY^-t}TAN-EzUWf!dDV->tI~@lBM0gvND1u8x71SuOED;YFdJ81xG{P7B!TX9K0;->q3W1L0$e0B9nA_a59V9Cei7p@mwSylnk{eNF~;ie^w0kO`#MF{i+3~ zpKF4b9MI{<vOh20m{m12rKYNXmq)4u zWn~DMmcgag$s}$~)HI8sES$#h(%iEdQoupFO6t6bsK!urS;(d3lm*aELf;7ICFqyf zDt}n9+N9W8X-+egW!*?cb0(6jO%AycD_Upld&6Tl$iPX8rw%;Fz;kpw$AzK5HWXJA z@QI(!R4ReY3NP|B?AewI*zlr4Y4l}XVJDL+SDyV1Hv@GEgRA)#GNAyULduAi&oTUP zF%7I4J5-GF^?;nO2s>q8ne&`1-V$%VLAg$o5-h=Zo`qnqmYcDlQid6q0B{quE@-y{ zUHxA31V(GBT*4O(RBf(eLuEN6T!?RT7llRQH@9z@iv6@DnV zM3IRdn~p*QD*LeF(F%&3mW|W|O&GOLVV^;35u)sno7QiKLTI?-0FVh1RJs8dzA(qq zxskH-Yl0Ryq>kgvzbFLdHO>EXw z9c0E_k-n}nn<;l&ruaOk=5(JT`zFo?9xkWCn8S=Ba~?V;>j|4sEe+JnC|kD738$z< zXMg~T20u5g%`w;XNv@7+_`U~}qaxAe^QP(6K*g(Wn5u+_;`6HrX#I;oiqc5MMha@1 z6HRzfkqi$4vslhl^UA?r)X46yB(i#@iz5+c(Nh{Jig4ZV%2EdCKDHfiz2mJlyfNWH zNU%ak`;TkI_f!UruQa>kHKyrcf0_*SwY5*?nQyjk0v0_iuX?wA0NWb!Irbfv@47`6d z^#q3;s1pzU`?4G#d;T*$(A%3qsMI!52I;xbV^dBIt!?Q24W9aLDi2X8pCEJ$^(}~u z93Fi|U9v)_X6Wh?4Xbp2!^;-_^RPb#d$ARYja+P{!XIVwzRT4&c4nty!8cJv@S!Kb zLziYQwE;bh3B~N934Y|a|q7MrED}->PUg-7fNSB5j1qRVQ8c zz*#iYPrjEZRsu3|T_?`-#CaY#4|^TQ;ajK;=-x>6Zm%GQas#y~GlCZ!LL-L`c8yMK z>{shn857E4k1N-5m!djS?jt*j9q~;L$8gajq;ZmoW|LG<=_!%;Je|E0EmyzPZ2}tq z%2$FZ$BU_M21-gHPThbDD*idj1igqq)5Or(PHLXssHF4w_Bayi{XQ7cNY!nQmOx3A zAnrmu)KYXs0i{m4RMniEqDohDAJAMLO{IyfB*%D$jD;jvH&QcqOFq;3{zYz7eiSaJ zo@mF${uCB8dmbry zZdF64o*9xoU)BIUwjF!#cyElI36GT4U5@CEzoT_@9yjA>vIXY$#*l79D3tI~_?fot zn+2u!3$35%m*8r!VN0{zMs5OblAbAwb?IU)kO`d7;;+rKMpKGMBn!P3L%`|Z8}36J zP)v!R;`HZmn{h4fD_5yHF83e~8cv8cAG!@MTWijGz3`lc}eQToP zFr8(gE`!PWO!-%@di~koIRhXIL~r>WYk(7J9tT>&$rl1(syIrBdO(BN6;hdrO2sDP z&r{U}a+G_+q_HQ4HjQ{ed z|CTiSkX4YOw=F9NsfP@tYDiz!kt&MOQp;*`N=}H>jeV0`b`8Qc7)=;m9RE4YJXym( z*$u1Rr2*U4ux%aUxfaCJA@M1o%#CjGdu*NY=nZEV%NA&MSX|)^qL96D(!;pqEQOb& z`>VBPI`l`!s3R~_*t+m001BWNkl8e@n=-bI99iE5Y2|CddrYv zqLq{_RGfVh@?>Dh=1T5-s!f@Led0 z6|)eCH$=lX8B)>ux@P(e-*lYk=|%brM}D=J^5#v+4&oE%Ib6PX&1sjv*9MrFsPgbi zjL@#@fNFzqgJ^>chBV68)d&KUr}ZJ&6i}C_Fn4g<(E5(PJFs?pzNBLTVL94Xz?sqIe zk9-fR)UN!0Wu1tJ2SVBI#Q^bCZhQb}BEKxCfQ!UkSGY(1<6u!-f`}y1i-xMS(cC z=4m6KK@96XqVb4xIqdjR|vSqj>kT5UBG#%M}~BO4woLe zoYaT3z1yZDknMB$xdt;=uiGp?NBQ-H&)G4w43}r1ta)Qe*!_Xs=y_=FRq6?c09^ow z&+4*lDY|SBRADF9=>WPzr9-G;5Myi%AJ-We_k;FFe6Fqy>LI z{ryuSJray0EQw(cN;Sd1uiPw$WN>@}<^2sFDJ%)fR9h%7Hkj6Cs#3<%xbGkQpsHKc zY7@&&Q%XN2g}$RDyF&%)k3b&O$+weW1BIuk^#N!W&X z7_-y)`UdX1Vc7Qv`u;FaQ1MDBHK#^ea5ZG0duSdU1#S1+D4P43cQmgpw!+>$AG4V% z4=JSuH=IpW6sI(&R-rRbLLJPw9*t~-Cm!d4j~^fS{^L77KJ4rBdEh)u=YAX)j?<9y z`@Xw3htPx8*LWU0TIFXU6fijrI5KYt5=ulzYpYD3pzq&Jr4B3#1{_qu5B1Z`Ecn zI7>(rhaapmyra`o`OGdGqbU5P7BGITL?OyTwh^uyDO8|9 z4cC-q`-L0kKA(eH;VG^(@GYcNqT?-wp-Sg<@g@I-vs_vZ2WHM8O}MKJ2;1L{X!w?D zE&OfJJ)3FCn1GNuBS))^I~sLt+|jvV^G0eTvPLSumPSed8U-;AE7SlBp#>ZWq2xXz z0kfd_v$7XXrfsCGff`D<)O{4Rf!fQS-E=r5(w=AMd7-t-&eK}JE}O`T-XBW$cF9+q=b{0I7%B~d2GR|R9ikoJ4H4?g7zP1xe--P*s0#v!G*bFZ%24v(3mrhi zr53U-Owb;|-8NM2t|}1-dh7OC>%h1Uob=(b7MO4<5y&{bKG0j?pkFx3Db*+;;R|KV zIy|3px?E_KiK-UWLex06d9nfM^Z9}2^8>Hv6R+dId0si>RBbs32uYjx6O3Bf4u_8? zp$PsAYwl<705xwo+UygufnuF!OO=IwIzjl(_Ssr>&T-B(1-ds8MEtT)ECnPl$* zz%sz@oluzpWh3_{? zMA|oRp8WOpz~iyoMvEJ`d1^VgoD_1Ut<2}0NrYrh|EJWY4o_*xFR3?d%CgiOsv%_) zbi4H{GWj-e3A)j^#6wz#`DXO5Y@}A{A}Xpok{Co&f&;b5+VF5#amn5?QH$cGWE0so zs^fSa-c0#xe*!HwQX8lpI8ROU>UDV|b-I+GG@;Ru#&&GDw33-2V}O}4nq{I~o{w@h zy9W2ijyKoq3$dA|+Xn6y3cSTe%D0*(?c*$>hnuIx2LhIrGVI_Nc>I7A!ePloVmAAg zPK4GNHQkwRUt-E0kty==bh&HTU&qTDsP8}UcEfxg0`$#TGtFW}}*Yb&91}3g6hYx8gI(iC>sLF|#+S zDbMUN(&{7%J3n)ZN!dqw`Q2V zi4ySKx+TfU>8R}fz6i zrR@2>>5>JbQUc0pRaR2rBc`G}Qrn61I;5a|EuOTMjJPz8U!VEgZz6SpdDv8;a(#D>zaVQ1-r^{(Wo zKX-N7pWi-R(Rm(|uFoMqlNjxj#B_QmO$l;drzuHZFT7q)+ep1$CI@$!{SZ^FggKuY zgF8PtNq;>)+;)(z~j#QsUX|pLb!OZ>8eE z5=o*@eSg&f3zHBRn$rogfPB`LPs)*RvqUij=HnTJG2rq(m#+oFn!~g@rG34zOZ%+3 z+4jveox{QhYZzHQha8}@zo zY>u^gdV4(Zc-!skWA~zvQ@7gO{2$hwkYcRmpDu-n$o%m>H`; z!g&D4(Qq6ulLs769Iq#i#{+1)O*eT5_0t=v!#7gbP1+}moNp9J!tkKlAOJEm=yHls zW1Qab3lxuQ;2eShn$o`3!M>I1o5}lH=2l*#=J#K@DCYD)=JZNX^44x7C6%vA`{u9b3UH`0F{kHbh~=`W zb7~(@nkWx1_UyZy!Sa$iUSxWKI>roBjF8>>{oILziafkl(&zN#g&BH*N0d`si<+P~ z9&9ct;Td(XgmT6%q}(ukWKu%fwdPw*3hF|FpgWAAHh;1r%TWk~34NfPj+4w{ah(cb zIgiSic#%?jZ#e-v3AIoUmkx@o;f(1PKAO#*%F-Nu`q;i8^CN8@qfo+!eWGbObpCk| zLuiyn3VHuiDNxk?vxW<4Nzf&Dj$Z3%)XLS=H2WOCuFS8{bHU zBnfN)CMX)KdD`Qd=pGfLP8yL;(V)};zM#}?L$48RPnjtcd2{!S!|8o1llqW$hOgTY zdtNxx>Nk#t0agBPo0C5PeJ{Z*B$^q9=$CupVGJz;r@zVR$~3$mt8T z259rI_V<;IOr;|pTFVam_vcG<$OLjDbwYd5sPZfL_7q_(zS|TB zgh$!8TXy`}KvY?QrVgH&4V1DsR6+a7jbUs|>^g@qFgPW^LCa8PhpAB#IqPUloMsz{ zbHM~%1gI(`>4YgwrTx>DzNk6H+hD;64{D1 zu;gkJo*jU4Gg+1`-7(N`fzeJu+ZuT5Xj@0yn{A)oc08hKdfP1I@cm&8m2aOOkB9wu zFB_>=vErP`?Zk7!kecpWBZU%hTt^?74^(_kHDe|kDVv~bd62FDs}cp6O?dgpaZ=m)7^#kXnOUPf9^v%ucZEY^IkrFXlR!j(`nQoq$ z(8RWn*ANVGG7m?j6c~+H*9F5jti_9nnm6ArIRbRhuypk}t zNYpC80+Z63fFTQ6<{A>2Q5QO~=29A-X}c}OQ&a<-#Z~N|qNu95QZrv_Zk1#V|Ni`p z6Po7aPwO9DzRoMj8({M3YYBr~Ep4mAIHCfG@)N#!6tqnH^~+nRPmNTzb}_c*eCFHz zW1tfQo{XS?&-10+6Cra?VkE-9Ym&a+Hfy5VhTR*f4q$`MMk>O8UE)jiigizv6@8LW zDnlfc5(GT%%oxo?Vw z`NMobqaInZRJo-wML=z|Nc`9&eu{1|2i<-6+rY$zE@xoK0q_CR!yBs8evU4HPTMYe z{bF(|oBk^^bP!_65R$Jb^5L&|4yz1P;n*+ohR35lKt{aCSeCi5itFU7L(%H9ilNDx z%ob$WC9N%+fH+sjuo~kEkMrxm%g^rDap1fThdzx-;$@TZac|(w8+BzHKBQ7?^+{35 zLNb38{e+LUvp zU4270{7kw`I*Xyc;hOb=Nr1<^Pos{wk?N)=Fyc7E8^y_4uuZOQYg_aNS|#K%}Zb-`)=PtyA-LMEFW(@cAwCysOAI4_*%zddy-83xVSQks>9W;k&aboV}VGksXR_{<=cb ziY=PV!3By>F`B7~M8ClRidUDa_fLxux`a{Bee4ACM$E(nM(FB+K#Bh-J}D#^LzYHL z|2vIT9X3Tipwc4i&cRv?K77(F=@A+jdO=_>rO5K@U}6JHV8_lIHgBRfwuY(^&{UjB zDUQCcBt(A(C;Vnq&a%niX!4vjrm|&&n~hbl{ut&jZ4BA$vl-XGxp!RG<{5MY`?lfe z{(WXlht!;xg$8z8QM$LWN;0*M2yWfDP9Q|3(M7paH*WzcsGd`pjMAFJ4Mb|*bB+ax zKgZZ?3FbjDl={{U7R00`cfU@1N;EFg;>fIHA27y1EE=f>W=(l2B{{a$uH9vse|7*> zJb{?kz4$8jJh3h;`jwmx2mmM&{E{E^kEEoZy z;Ndk_qXbDX$7iFN-$|rCXvBngR3#o>TOBY6JP2*5zpUovO~GA3T|iyz$2^A*A-A*TW;EwT z?&v16fRHsdd1(t!S+VZK=PiETsXs;1e~MBEpuj8^s1N;*B=Y{HM(V9MNbgf%C|+EP zeY>3)of{g>7lW)|h>d`hErNPzEO9nb23_zaZ9&JH%#W7P)ENvp@+uIYJfGaq{9
  • Yp93GSfm2}3)4rkY9h=kl9*=$cpo+||yH6{j6WfGQ1bu9wDA7!q#$C_-shOfu z6wyk%qayjpe^=JLshh1cAml?JaJ|(EvPgtvPPxY=kZWX971ziAb^H!|JpY{DznRyp zl5FZWEoGP@ka1z07KVEs2d?9V(_hzdU|d%Q9wG(+TEazT68w(i!0Y)$-*!;n&7ZVA zz{kTO36>8+(4uUMrnlhfxxU}{ghHOCh{pr&u3+6HQyHB)TOl#tPxRDjoU;d~u< zJ?-!LdL1~<3$NqEahy1>fnR@p$G`pdD`fm?eic%?e;{Zy+|Tke$j1s2iTtNpDp25Q znP-q7QvybmD2a)3o4${6;m403`2PKSe*OOa8@_$B*Y|JV@W;15a2%&&&1VSlwrzf3 z9vGJ&nuDjY7R8`UUWoffjXiWT!YpT~=NC3oYS}fFVHn$s*-*`brTY9R#qTA5&l?@8 zC{6cOPzv*Xj_<%2fn&oRuq;K+XIS8}k)SE6*Lx$SKiOI(78kdvX`TDg0vhou!=*!9 znkX#3EmYrJ&5p|AisqKhKRYFw(m*Bi{uR1kg!o1?rAZ$&il*J$pBt&%DCuhMrh*8r z-X~IBwr?ii3Gy3HBzqw`R+{}y`?(O}bRpVe!YRJGpPCiAeX?TF_ z&=Uh^HdFBoFnCkPMmsj%t%-_es&(^1$*_DVy}WVqIwA4NBWsKa65t{g#@xz=J}OCI zlv1x2_cswn2*cXDf5|B!nXVe%N6qHH^3Oapj-bg_JxuDl!APO}fH+c&h z^;dW@H2mmDwflrzrtT2zt3W{r0?z?L(xqxxm&8|sxo(c2$?;ryq;PUIPEBXcSsc5I7 zoj}_Z?Lpv8(bRbx&7&1)vDD4efe9q+YJ`~$7yUpGuG>#o;V~u=T<*bjWlnDG3dAB+ zRlGt_A-yWd0+0S#g%>Xf8m*SU8a9o1dB@iWRq`ZOZZxbY8Pzl>eS>x2vy1^&f78vyK zc|;ylWHc4?Vp(Ob0cqi;sd}J^TFwiwlnZlOBqe;g2Gr0UAmef?T_)ujY#6?^ z%eH{=%{~6l7l1$6NHL-Nb@@q`?IhYT2FtiG#^GCv@=!f&LpAcjDvPAlwo)`Rpu8D7 zjwiO`fp)yW{eiw6xc0-0%h`~5S;AB_?z5q)8>yMeRHUs$=`E{^^K3HoXzh!z|0?@LE6~^BSNM zh!B|+NneWpk1}SgvmwDl2QMzIO4Pa7G%=)n-@kps?|=M(Z{PmNhU&-nANcm|8?Ng@ z@0$-AHuOGl?z@K``=;@_%!E#}j?~F&UL+2Zp|LjW5-*y+)JVay=%!j(r^l)6AHzcm zxW|!b83v@>s$i0H)=r$wRNW>-KKbwg#&BT8P|YorEQ1uSA8c;3CM6Xm%ZO5SvhWh9 zV9hB|y>%;9e#}!!Rs*o_`u=F7q(ZiJ1)bMTkomD398zVpiSnOYsTqtr%5~?`-V&iu zG1k>D8l>9Dotjtn^cYx;KO1AdWge5py)xSu)Q!d+Kb!*={e(aN8VQ}#i&8Iv>ZOaK zn8^QbEQx`J4oGaHG&WKr&GbQ}?Yv=YJN9Uxd?VGNU^#8!B0-kZnVluxNEOMO<$ZZ< zLnuUuGGbxg7qOL^G+}i=TK5pz81TsOP1No(=WAfwwtSXEc<}IPcP&jbI2<$Vnb4Mj z<(jXipS_uJHrL|z7IGbs3t}mtE-KUZ1}@KKc8Zyh-78FX_EqFG;BXJ8qLG;7T|7H2 z7~Ym;L_X7|>8+%ez!Aoj1Ul);vpGxC{uL|!g8RQUm(Wik;5qcD?S?9Js4HFkygm+G z*+`wZu4AFn#-Ovc89`&&gN){0W{!;Ze3x0Y@ckqhbyw|Df% z4&EMU+XHR4?Zl-pYM?4kF=P1t*0DWyY>y55+uTTP-bn5Hj=uQ=yx*Iq8;y@D3C4BJ zF&U#=e%>Zxvp7%bNyU?cmO#7;u0*|@RX%eZl|Embce#kJ zw6^2e-_Xw&fbPGK6cc3m3_YJB{9UwUj(iU4WNumXrmTXR`LeQ!3TfX4;SQmWOAb7b zfp6cR`2C-M;M*U+VF001BWNklqfrED4|JcO2QpLG#!RvgSTpPptTmknD?MX^fzMr?*j zm*LJ5z!~p>>XP%5o2ifpl`OZ&fx1R*q^beF1T+_&|1?JPGCiWQc_ihsDG0#CI`sLu zv`!;5oxo|u3fbc1_QC#9c<-UCm zSv(g?sy9-^-<8|Oj*NB3WE&}QpQ1fInu8R2RhRm=YyMqOc-8#>M|~5!p;Ucv)HKK{&=y{MlV2|Dc@!{0cJcJD86KbRW6#+OC{RKrN`o)8 zz6mfYVThj(LerArTRCDy+thXjr=oR5f3w^+MUbS~-j|>(a`+kj+GUvMTM4X2#ckTa z==RUCk%Dm<3S%1s*F3Yd_?vZ#sqZ!Ilmz`n2+IFtBlX)a@7Vmp9hYL9itDtmGLQkX z7C_5I=jegAfh6>QjtjA6lQ93jx}GI-gAp$sPCRq;=rEs67^D|!iF-;~YXA?lHf-)9 z7H~;w>qBwbg!+oN^>`SR6(KYBU*E9*`V02oe!>3RJ0Ab`j{bPV_IN{oyjgb7zMH(u zum553>)0OVd)mro$_l`q{--dwQZub65iirPY=M{-*~QdtM<3C*TPyqhYo%YBAlt&X=o2gyv*ORe<4STYa@lv88f-X3`}iql+Kj-g_lN0;Ig$) zs-^^tBF|=+x9jxZC!rF){V@8Zq;8yfW-q0zuY}T0&3Ob0t{bVFh;9ZNrwZOec7r@M z#9%AODfoDu`2O+2AK!l9_up-cV}XSBI?ofmG2R|Ke*5hWuh%=S>$K?*2#o*ex@}R*bd;ocslL!hecDJRyQv5(hMW@F5WhA*gjP0D z^o3`p&S8}|P&v@fxEWaqqB)$K8*4jHMn;#|`F>^0h>Wmp%=eE8Q`YBAQ*5T<=huxD zd&s-xKDK6I&@F1;Xa+Jfls$$-orn_66rjT3VG`h~{E1Bzbk=tUve5+EZKZI+%1>l|L;bM0Q|52 z{0FY%0aeB8HE=u!Ue621bKp3{vLfOFi(ICS(06-SZ98Gx!sO^OzCHu#%o8D=d!RNY zEXZ=(7)73p-5~ZV&;l8_q~Q|NJRd`B8tc=;!3^qxL5dEz7sM{0iwM{{+S>yU5TtLW z2kwTBwi~M9m*3v-{_7il`?q)e^4lAJ`RyJ3@j!n(u0WQ>ozXTzqbVs;r{UBs(Bu*Eq}0_-Qpp&=vi_x;&w)Ov>w}UQVQl4ex;0XIlL?j) z;iu+nEuwVUR$*o~+`)~1wvj5ZiJ0nT@g-t3ESG?eF(0beb=ulLg7_TYn@Ie8_{~?Y z=Vu3CMVd=O6Q8d0LCB5JbY4H%D5p=V6$ux6BWDv$F3lCK*esVCT9JHRg7Xp_=fLZ@ z&|eo?Kk<5=`0;Vz_FH#6g;?X4} z?Jk`hs<>)HrO}|!`W`1PU9yCgn=i@2YF&!xO;vN*m)YGJ1MxhvqB??sNj0fuTs&eN zDBs&Nfz|=@@{&08OChy1QiP@8M`^OP1WZn(cx({&+$b=l=>}@aNn7HS0x7XnPvAv`Dk1RPRzUf5<+2 zVw<=>_zjQthDUo~?+@&4N9S&NMC^HXqEI#*H^jRHZC>>>*NeuFTMmpu#+om>`|u{S zk+OmC;etO`n%^7dBf8LNU}M3fD`W$XN5lEnasKjvjCZJeV$bXHyrI(r$Q?YegcMJ8 zh~+z)SAqwz3z*H6IAK*F5VBfo;?XsG@5}Re4YI{h>40=VEThd<6-5uJC0Yj#NK0C$ zD_k-U)jtDiF6Z^3j?5^JumGx)Ll?qd4D7?AAb&{<{XevkBH%y&^B<7w9jc0t9|wMX zKk?(+3m@%7;}h(`dqyZBv<~!*Jm8kFJ=hwm7LAlOQ*rRxDsg}c9y^*S5y?Q#QNlT| zGbRo;#g-vBMQ{##?U!Zc2-y=&d4Rh>SkOtFGN|YtYD&GMgRr*-JRVl~Zf0EX+lIDn zc>nc*_g^3Q<<|$^etqEm*EjTiN8ca$YTks+<-c7q7Q*X=`9?0`GeavkLTi&r0^HX8 zH+{9iI@q&hW+He<8$LBI^^UB*f}c@yZvUdX)2e@`Hd`y$jea5xEI91@wu{Qdpb%#M zoV3p-Kk@skI0Z_@vO~Nq6LC7=bZP-I$ z@!`RQ7ETyp!||So7j(HT6svWpG)-fePWl=i2<(T{7%L(@r)F`z6s`O4#QDgxx1?6yzk$?7)KCl;_&nG@Ue%MA?9{@92*J+H7dnSE>3@w5J{fouw z-6NF+w!at64~CY}Gn8Z_z7yPnCy{CO(NMwjuOhZo^Q΀C{cJmr5#+`Q&9VS3gw zGfE0516c9{n-ZEAsI5pblr0m9H5w@oC5|7LQl64;H~$P!-8{v%f@T<}L@4?v(bTei z$C5o*5a_X4kt*X3S1cFU$1NYZz)D1sihYUCRmw}%G3UO`!t@Yik>G*j zD^0ir^!|l5zucFoq~%ZM_6jE|xt9JGF&LYBLKK}l_V$jQA13Q-wvp=GU1i}qdMJo5 ztA050jfTfqv>CVHeot_7s%A3@E@)|{s+wY%`}yRaeVEYPixSr)>xE7OTN6lE=uS9y z#`V^59SiQDq&N{v1`8PtJms& zmZ#qrU-zPjH&0XUJusq?(veM+=9X&YMw-1`SAlPruiP?^Y`lscAiQ!hnkI|AiSxPo zm*aEpy+Nd-_mOo5sh}~yELPKXYo2=}%Odk8D)Jm53Uome15T&_a!LMq)35%C>9nXl z(jTd1xa~0yLa~vWIj2Nsp_plcrI89hYHXsiiSqL#o2haRIDdpMdyQ^bGq#s;Uak+1M6Lrvpi zfYYXCcRl4(Ci^kM4h!d$DEtZflkHFnCO*~bTHp~SABFDC`)B%)7X3iQO41F@j0HJK zOA|%J_++a2(;F#kc5+%1Msv~37JxTA5qx|3WbyKhiJ9?af5%9(=enAYtyC61)7x{W zfe~wZ*H~kk#TkstWh*r?=$vSDkt(;EO%x}<#R+YSyf@Bk;J60bG0?XQ+)un-CqAAB zKAtapd_3{vk(vPYYoDcz&SuZd-xgp&M>A zyg)0{^b}+`>|ZHWi?q+lSCj>|U2784HUG-K;mL${Cx2xmiyyF|H_vSjD=aqE-@npG zNfu}!>5FD6rjLFIMdLzJKohH^jtU*iqfH^o^Y?+l6&m@O5P`jF5+dZGxmt#&f9^hd zao-o+nb{NrO(aO#8|l(Ld*7PIdDBws(4x4M&}KX7v*&|q%(_bW^#&UtcWa) z6#FKMV+$n=b?JL%;S@RGd~bAJf`Y3THVX2(obybXy)}Og zdGlgGtQgYAz+7pbJF)4DxEO1!#qUAiG(bG-PMoZGLG^+-RRHrq48Kt$nA8V^5gyAa zpK9oIKa-|p&>Gg3&1+8i)rLQ7nNg&MUzPUxxlkJ^f#+ArHJc~@cbRS{cAw60=nF$* z8)XBw86qBa7w5GT?L`hdIkHhpL(I)!WxYYQ`}Jeb=+5?;a3^qA+ekIoHqnEfo1gt? zV#`~RCQNe)$m;6t~ehyILgs4L~vvgLgfMV4o#fzTgo=G8*;%Z>hQq6X&qATXJr zl>Zk8GF=I6cq1hybJW4MjnW2^mN;@wBsqPCge(K3LOM>sQVb!~nzWKU7@_4tGfPW- z@N3B)e*KH2U5EZI{Q3X-U;hBZF5-W_{lMP4%VgCi!CFJwY_+-NMvAfR_O&~agPlSL zRgBAx(NR=gQY1rOaQW8hiemUQ55;+o{CbUnel?ukvZJ=vaEMy|Q@?=T!KXmGU^K-x zoRTI0Z}tSV&4YD1``eDkFK>8zf5+P|Z`j}8@c3oJ{?_q$-?6`S?C-m8t9JD5VL$F$ zQR7rCwfF@QAz_ZZRA3FW^^F`x>iB_?8BXe^Wkmy%YX{T|Y;FE$HB@~F6_HWBfhaG& z8x?9vW7jefk>w_wsB=xK&(umY=qEm_iOh}z5|bAAML!^OmG)T?V`(z%KlYXPum`I7 zqb`39mxW$)GF$NP3&AIJ(lkk3nzFu{Lld+@%;nO=f@)FlW~2P_IYlK3PG6t z;F?>XYlsatL`wFAdc$Q+lrjl+7iH+ zDfAmRbhSKo` zR6ILnaXyRl;?>Dblr>OlN!c##GX%GSeJf!=fRML@7f~zBT-Q33%F@10ZuKXA#IH&F zYQ!fk#XMo1sVyDda5*+NG%z+aY}~N(jz@cUY2O?6zN5E2<@=oXo0^o*n<)#Qbx47V zS819gD5aH9IbW2UN~$V!T2dJ7=Z9jDU|^u>z@-DtB}-egvXKGETZ765mA6S2_JQ$m zAjufw(mYN8)bzmyrR}q%&wvp$5zns5$;}T4LJUQB5Y{c;B(}F&3vFwnG=fmIG*Nos z!fTNSGdG7DOo^L;!g|Wg7w=ZQ@6`OonbR~qN|Pj*-nu#c`C238q1dKc80Lp}N~Gpy z%BjsN?#=Utj@=@nzKI$m<$a-=2+#<(iNlUYy!R>X3*||?53LbIn%f@)7u3D~iqUKu z!7bnu{@J(9WPRLocnWohQ@3F^3&3sIwNJd%K9$u&@W!C}z(vXe%8 zJLBhxA_UvE?#(q+B8&VMm;3si5Otk0e)@&j;fH0d@m{qz1kreapj zzkO`=0!VTh!RI&~e|k9%eFgznA{Q@Bl!x=V&xvExRvu`!LG_6x#oTjdC+SV5H-il; zxsT^o(qRrD_*ARAi97rT?m6&nRy0=EuqJ}2S!PrY8N~73_UH8GgOJA9oOowmy-AA9 z>(ZN~F~Y!WbiI)Fp)}G|4pE}d`q8_kJ~JL@dIjQAV-<@5*yd*V@%F$qE^Nn%qaSG8 zFnAQHl9-3j~OHgY2AEd-RNCD{rvEsd0VBjt2eGU@V^|J7#7pC8Z#)}Y>qr5~wsJ7d@) zNF9Bzum`e(HT)?QLU!SWQW;rHT%HA7Hn8vswfem!-f)j`*_jdk#>o0Z=7;_ilM@?c zxIaot6E&rM_a~QTp_OR!1ESQ;_V9+reU0bf@^>%WCSrdd zdH5FA8!QJf)E(XZRln~!3(<0xOi2^_rkA||X}-k?vRG%u)_|=6dnX*-fC5`9c|c4h zD~J|iX>6q$)tsS<%bwsNjwwCQ;MDNur0O_jpzDy5@Gn&IS|Z^l$E^$-3h^!mx^1di ztJ1!D`GRt)V2FEyrgWAPsA2cmHH-BxMMg~6gx!E*68+^q~jvWWzS_4P!XOZ@4 zGq02iK^okX7VqzElNTP^1W}=cb1Mp)V=4;Z{@L&~1!O`bDUf<|$Y)4<$?03$4gHB% zz5t`bPZ@wDjR9ryxn1=!pK(gFYLk+y7hoPvifn=1l;#o%uQ1s!>&U1i=%fiu5C0re z>EE@H`nP|)V{6_tmDO$Q4Sj1F=oRM*k)f1#-$?DdP2|Q9_o~=eD|VdF^|G(m2>p6Y zuYFw`xgY*x6+h%Q26xlvKlYBdw;hjt$NSrk_xBzD_HV!7-~as`zy9`)fBgF|c>Cpn zeedXdN8cOvy@P|&=Xh!`Tc*>#nJjTvG!`%kI5IO(29l%TYoma2?I%ickFuFUOvOtP zXPxwAT4#U7rb9{widR>N0iSK7WHGzLFHrkT=Eew%AC!u2b|tT69?)WC(nMj?LWL>0 z1UFaJ3`Hy1EIOauMpco&XzMpzZerEc*No!=#=y7^jPtcei5B&Q-{vE&l z`WxQg-|*OXG*(07!ZW@gDW|<+@G&GkN|TVy77+lW`RM=K-n%ZzktA7qN7dXtE(uf> z$p;?Umt8JN+2!mmr;riurph0BF*VObH?n$UT4W_mBsx*m1t2rR-P~05obPBg&KF+@ zH*hnIsyidVKB?lAND?lWE3Wt5Bp*8PzmY$-Kb|G)%QLEcN^ZrJg(ts19@w4_Tw?O( zQqFWm1(KV1;-^f{H2NBW>Nxt4E%+gu@*qEe3X69|nb-+N_M)cN!B%E+I3FX`~lnv45CKZWIv$F$uKea+^1`8f+5P90y(V zJnE=rl5>Dk1{YIWI3mAJFs)f}V2`3+Moj0_y~2PGOwFpA(S)GDN**+Un&&bqmyuWG zB(pd~_LxLT7=6>@LpiM<4|Vl|xdmK7H4mvV4L7xcLO8mU4=qMPS;MKf#ABa1*kC%U81Q77p!FJM{h{in^JkIg36YS~?# z1f4rNnS4&U`y`SfWLW<*`8Ur5R8%frWLS~b-Hy=nyA|2LMCj3w_QBO25nZCApi;H@ zyJ#Ptn+SF##&dt-c|35)6J2-DISZ6LCPhfd7@676f<{kX7n9_fkKf=qOJgU?vc2{l zy)pX!bcr=3Wzp`j21;*qD2F7QE;dLYFdxI-psDD$XFu6>0vI?*eUK4R76F3M^3Uqi zqxeb=Tk!BklbD~Uh2vq8lu|0YUu4P+=_bZ?$fMv651lF%9yBLBiORURQWw>;Iu+ey zztIt(2BJ3PhW?xq5OerOX(MhXc08Tb_1q6UALiwFJa#er016kv=3rUbV^k`o2AO`@iKwh0B;Jt3Tu;1pD_te#iIk-*A7s<9fYf+g886kz_&cDZx0FC5z;?hzqVDT)b zqO(cqBs=-4sU0eTQAEY$fEnxcf@|LsZEkp1o4Os39qYE?IF1ZVPBXH97RO;S#^>XK z{b91s_}nYzxsnkB4x=Y9_g4xjC~)+nOr&BO6={c>2kR(`n2TsBKn?8DnP!NDlFp+l|{dp7EwRneqvf9$B!B zU~?mdKKI?GQu0KXJw(rbrlBz{D)6V)80)rRTNkXYVO^Se`CWk`Z09;vMQ;t=DbNQK zx=%)?d{39LX2Bf3Gf_L6G<(1Lq5k>PxYo7E3jI0*2<*xboV@Kx91ZY-7g1W`?GWTT zDVa`Xy7`F`I=!q+?BUKO0-hALuH0Ja@dmKmGN#JoX0mhbW$W6X~%*Mb?RpY?<7x>K>U&h_0g~>HxQR8dT#ZLI+uSomU#Y%4WRx6~? z4`@Fi$37-fj|Uz<-tl;UV1F6`GNlJJBp0nUzb+P^NIABCF$ri;l*TArN}y~^q>yqc zE2R3EqSgX3C4T$?e+t;xKOiQ{b}@O{<$^7yQtO(9l=!u?ABV~Oo)0fFY_ne16|HUd z^Q;TDWx?gL;EzAPmxJrTTr(eRLp+hO9=8?$p62wm(-K`-y(-^HfU<$+0$9bm7V;rOr-YX7{d@b`DhY9?gn2n@)K(Vi{^`3V8e&RPw6A6u6yxx zcyZ7jI3w=A5T@}E!f5=&aWg1LktHQH!_ONs>cxeU-RB$+HZs~1lc0I9C#w}v%$$38 zFH+c&x@*@bFsQHoLS$LV@FB+J*SvbZKjvpMLKk_GThXbjKMsRZ4a)_IH^k67!%_;| z1WCA*BHbLQcXD5udUI7=sMfN<7PHt9XMuoRV8AJa&WP!gYWj<<)eKFAVmkGxo(@)8 zc#6%x%fvWfK^w&s7BpSK6cee{5AtIoH5d_P8pTHVB919u_@n|XL0RZacCkn2Jpe;A zACh!vjM{HlJQSU|jj%=j>5kwJE2n?wmZggS(KMi@&t zmQN906;zTiiXg~Jx3iB%jD_T0{FSCCq?}*e+M&`t`{_Ve`+2$^_H3xvus0|6{lI>> z1;qdR$UGYpDUY0K4Q*j8&7Sd>ZM9s_ko32Zn2zsJdb0wmaelR7q8;oQZGe$D2>xkw zGSAK;5HX3FUNVa{?!$IlhYb!VQ@@uqCs9VM3lJ6(gHAYS<&ZzD z0}G!Si0-qcDnaHHRn4Adoq=L`k0^blMaLA72Xt6@d?%V_#rPQ3&hq-p% zV76x+u?;xVI#BTIgbHK(50>_Q`|XC?l|5%;pgxz?4rNDpCl2ukoO&4V&=eng$Md-Z zKM02xR7MoD*Pnmd*Q3{5kB~Uc?*}tsYfise7<^r@-ZosmU2yyDhWqzBzWw%w-+zC@ zZ@+!R?|*#5@4vs{kKe!JcHgk9&HPMWOc6zn)xhEgInT_)p?Q}ijtt9yn1Nix$C5nc z9^O@fDjM5?DH5-V0&}{UW)>|srI@Hdj+FOFKU~@<+4<5mKx)buVk)Hq^8pO8H-v_KBsmL%M&p zLMq@2PY-r|KFlZ8{W=|Xm`FA^57Dyt)N=8HYIW0S;|U*}VX`ri@(o_7<(x47E>+A^ z!F+iKt72pcU|!Ic)hADvA@4g)q*_}byZOHM-Tj1)G*>me>auRw)`sh4!}Yr1dNKLy zAHRRY_wR3bd%NR$yOU1DN1QIq>9gSD0XB{RHJi{fjb%d=SbLaNg! zq%_%5QBZLq^o8849?JYyA}X;+Iv(6kZxS65q8&hUZ94mBXH6$-4unkG_AZ|`7VtqKFr8U#y^2D&$d@v^uyUrcsth*FCv|^Gq>I`^wqCDY76M z{z_La^!|kSeJQedv*6eE%sP>ZY)0wB(qWS*E2z37(Q_|x3H9ODkwdIuZ_++VS)%1a zS#b2aS}~Q=sMWqMK9O3uO=099f}RU0Z~EA|$lrhBalNy8&Jo7c&iy>?(A+0;nxl21 z?oA~qI*W(c7G#Xp5Mr4;a5;4F{<^X^^jIp|s}8t^Ojb`2H2-5Vp@!?DzY-p!xSr8L zF^0gMwV%m(o6}J}v&|6jJ~8 zZ$Hov8x6m|KXDvpC|owi^}e8Y9@M?waw?VGS5+JyZf6G@$1*>5yze`nhwI**6lGo| z!V(!we!%6$Zh~cHT&^1~Hv>XkZX53RE8f1{@cp+peE(+CsPFOm?T)waSKPkaA=3UV zpO!709g7}REgye}kysN?HVAW^c*fGb)b1O%q#kyVqK3~1VTyxKtW=@w87@8Ye7xhD znF%E`;;b0YBFgcP8Zd-zB$FWDh!~$pebmqFp^>y z3Oy_rI@L$h^B1)Mql+b;THbe)^X&)r=fhr)2cG`g_vi3O5j-f@CsB(RP+JsFD^?Fu zUY7;bmV&08<$VTXaQOjRK{snoYzJoDO_=%cY*=KO4uOH{w&~Lqm+KYR>kZfI4Y%7J zxBCsZ`yKcD9gQ3IWyiiZEN#cWGkPQRh0)iBzHN|x;Iggm9lYXxyWoDi;M>~`-@jdP zzu$4WTpaP*hI}&$oSdk+uS8@_8-0V6D#eP28_hw`aj;d#!^iaykd46oD+8-6fNmZD zYufPJ{T;Xa9q)g><9feiKMp5fAtie`Av_Kz_&u$VdU_#v^z^N9poQZgjybY4XBA%p zPn^YUnlO~`GiglEjA;~jk|FC8HB@$2^RzT8h{rT(03w3YcaXY{T+O2LagB!Y8E;krDTuiWdhmntUOjrZCHcrrfG2uc~%R#X8q zpC@cGPgkWJy`z;ZnAr7d=wOA-3{jy^p>#~-5(+<^;9_cZ&U-5dwH)T0v&4Gl{-x?8 zCLhwXVVS<AP<2?nnFcLr@Y3BOriOI#(Rb8K{n+~V}&+C zSBA>rDw9zXH(orWD*V2h5=m&?|K^F9`n(FP(|LgeS=?1Bd1N@79No8s>AtPMleGHpMidRZ z#55WhsVnx;Q z?06PD`TL(w{M(No__sgb@o#_rz`y-?$NS@n=YHT2F(Ynk#^Z170cm!~-L?hm*05bx z|GMDjMb!0v!S!ZAxwqQ|Z{Kd_k9xbhPwI-d`vtfAigjbO#X3MllJqqZK|l7SeI5HT zA%EjxXlQ3K4+wG6Uxv4Wk`AMVVy&gqm-06sixXS7+F%1n#ODl`C=3#wGc$r>Q@v5+ zgmYM@iR`5wN3F2Dq6ke@@1Vp4ky36&shA1BfE8mVUD2E(*hU)Exw#`KsnHJ7;6yD< zNd|v;^Nj}~U_pF2f!ZIQmGqAN`M@JykB4WD*?Y{yXfD|^-`whuk)>>|-lbZX>^7N} zy%ZV}zMKOjPC=tEM6*qF$`m9X${{5bS-h(inPMgnXv>0Y6i~N2uJb6mVj)n}{S;=N%5xhfNqH`-RQA84SrG+kh|rqahUpk6**xX7RstcZ ze7u?j(Ik6fmTYPw%Ds`~WWFSg*7qSM9)y~Io85AFc%=9wcN}oNKu0Ock_IP(*yL2x zDHOf1#iwkR@EVX;RB@j&B}@gWj*xP2^TVYp8KPKXU>LeF{7_RN6;ms5ADmdX8AN~w zOGg;#&k$38l|DtqE0D76MDEY7tNdr;(!u8JmKXG^=PY&5v8Tt3yFbr8Kwg?%?~oD9 zOxe1psbriJwF>&qQOc&7YO{Ck4np)lE@=k!XDZ3L+yN8B{PcXhyrV`5T z)0oMIP;ksOyf^`cRzNEey4tx1J%BsI-$FXwP=y+)CA{7tO<|ySLRupO{m`-kv_aMI z?nFaOY)wN`GLWBP&Ix7&DpR6S95IEIN~cYZXiVw={)>fvy<^aRz zKB+fMv>7>aWysng>jB+5k&Fzap9r1mnZkH?ysX*3(&S(!LEs|m%B&dNd z45Z}QnGU~b4zn3@Ve?B-8`Me0pkrEQS~P7Lc>EJ4QvdgV`EO`!2Ud28OC+kd<TZ~`?%>%3oey6pDUCzje3RA3 zUu0K-y6H{PQ~p%_4>lc$j1QXub7LPQ@D!jH-91Pjl&c^U(wWa+I!qWcu#i~*OF5_$ zKFpYZY1u6bGPYimFcOmz+L!8ibp z@X*@j(mYtO>)VBWuESwY_McNxyairBQAkQ+Lg2XN#qPt%owJGD4Ue}sy#M%t!zZ}M zap3uUIwsa0pcdM_qdl7!QBR*pS{Sn|M;2izS8=HvmA)#90smK5`>c=(5?*#%E0iX6 zbvYTD{SSkTRD)_Iz%%LF_s++{_TX0WseyQCFQ(Ar+Ab%TH;%&t4~+eU{GO!s6P7kb zaX$U}hzHO~dtW74@n=g+fs)i$Cj=$UKBR3L+LgdSd)Hr!uV^X}G>dmF&7AS;;UCC| zgX-Cp)7LDLO2Lv|DFw|iR68Iz7(^UT+}%kY->*PsDryPT9J0Qzd!$C5OAxtS6+v^4 zd-jvnOc$Q~L1QZKgiq?Z@A1;jHz%$rfvG=8hPPp&I+h7cQY0~gVMkSBO)XH3m>3;% zE23z`H~9A_Itr@v+Z`@{uP>1i4z7caslMlrZvT+KazTwpj>E79JAEw<)*b-e{XEb@ zI~#{#43BY2q^RKQ^FwNr!_PcmWbqc4S+ zU4XqgM&beW{8S%kwDI#LoJnJDIp9UaW8Z`PlcatB z=YRgMAlX3#T<%`Tyj^g+*(j8_jF*Wl&!+FdpMs*~H;OIKUW>%+GMX!>5=fgWv7S5I()i5i(?bp|F{x?-p{tH{9O5 zi27#JsPDi1hIMW3i5f4Hg^o#rs^WUxaKBygcDvzzzu|tnn0cC7gpKcEtXB!i_R6Nh3N5I&Hyv4e-iDr>d!-)dVtCxmdlKXCi;j^}g7MXIg#33`nj$h+In8+Jms0q{`D^%3TPBkRiP6Phaq#tCkJwx z`m$lJ6;zaVEu_3636iw`nuWpAtox|~YGj#w{t!Hz@?>!jCgnsvbSR?$#sM0ILOO8X zwe6f8MU|A%%}t)tz(U<)2H|1^JbM8ZsPo=u-&#O{xD@+dOeuwvtd)Xc6jbW|&MC~Y zJP*o_mZV4^H@_?RN!?xHbdvau!Jx_ZR{L56#pE&FL_JaNC_b+sfd*k0 zYK83Fb%jH<2<$&A(g%>okj7Kd{l<<`{Sn=G{ zysS+ax*pJN2et!pJ;#fqtJ_q@h^;L{@G5KsbnAm!e5S)ZZL z1&0~}{!p=kY*}H#3m%k7E^!)3#q>xGf1UvZ?f(?mgyR4Czy2HKc!C)C{@V?|{eA;A z#pPl?j@D>;__j{?@lH@SOzr#gfgeBK@#l|s^G3Zt@Z<47-z5)??CkwzvuV=pe!>0i zirejq@4wyf?fVVC|MBLYryK5XSKQw&xP7;;*Zajl0SluoJ~3;Qx7)}SQMdL4bqUaZ zn6z(q3DEn)14|pU84QB7(D>XO&fJQw8AJ74>H>0nustY?-$cd7)YfT`snt+2Bxs)4 z)Laov(mpNHj8RCiMu@@{DKpb~w)c;}nTO-hqaI;b$y$eTU}6BsB&r*0S0%vk=N z68=h#lsc&{JW{>y`8lRP%0UB9|7wNTcqp=akpe2Q1PO0*kJL6Yq%3zP9?qi^qVCxM zbwJ2veIwlTa5;%;>8SbN-=cv@;sL;meLro%x>+F=6RC^)qwYSDx_^7ax3@Q}+hWDj zv7>kUI>zr~B!9bJaKGR1cDv&3e#7l{1*sW%2kFoxkC}1xnY0g>L{oYASDlNg$Uo9y zevNC1oj}J-siCfiSOo%cRhsFH>-B-#?PfH-`#bjiz~k}2`}+^HWwD2bjDvW1sP^lZKRS&L7vQX$ptWNHMmj z$-Ey9GO|~?c%p!aI0ofBaKx_@*`5r(kJ7$MX|0mvA;TUXNG)`%ETo_xwyddsK24?K z03=H282_JbCRWJw-1kRAO+VWc34bvQg>w27d45WgA;?L=D?i8xp7lbTYJS`ZU+o0X z|6xEnnl=xCP|t)h^ZoeEC9CS>{HRC9?4)}!Ws}nYI}Ir*oazT1RI{nEQ@u0Yr=8O@ z${R2q<4JzQQGij|Q`Hk36q)6L5CmzB>V%Aqdc+ARHdoJdX6O84fvKQ5*g zRzYRAs^n+dL_Q%CWU|I>U2)k=&VIRU29sJ_dhyxMr01|~f`{sAg!Nh>Wssc3{Aaw3 zy`9E1TM+xP6;doVk&<upI^5br#YT4ZzC5ldmg%dO3AfP`N^vx!{A*EHy4yj3f-505F zkDZ^zTcv2Ks3t^1KyP4u|x6LO|7kq!a;qA8J zwk^1<%_dEZ%q$^sDwaL2EJr1$DkDHv`#|xEs2#ALJNkY=j&26x(2#n7VOtmQV&T2^ zvV%v;ETRzkv=%rg&}+eb${0;!KG&;G*e*<5xR)oz+!SU zfCj;jxuAAY1Gb?bP6^ZQQjEwLBf3d~QhQC+#OPp;ImY*wRzC#K=i%3Ahk0^>)MaY2K&9Lx}f1q@0!$mc+%SB|5!&+Il~dsL4K^Tztap3W59z0DI#v z9S3@yQbfL9DKzyg9m0$q56?xfbfz?d4-9%hF{qfniA-9@Lf{U|5K-}oRCg+4Eu_qo zMPi>@fGXJIE+PkRqTY!~f_1QZCv`5B$+%_tOI$99FR^jl~nud ztY(>mO=uc=*F$za5LfRMrm4m#qJCaT(Y)=~VyY??m@*_?onUGAU`^e5;{X6407*na zRGX3_a++DBg)>p|tN8QrS!FZN{kipXI}814&_M+tWD#nde{?|xV?^mMu%D|-!@>$y zLyk9fz=~6Gr<;Msv`TXdWoU1u#r=G3_=9zX(%MtL+aopJHyQ7nAmRX`A;Bi4q>=5& z9^2AfdbfCBC}_x_xJb*!F}WgT3Qdwj;;bcNLCYjmQu4`Qh=!!i`It_D&1|m1mK=@B zkh~3pJxD$K%fDwVo~xuO_tS))x10_t_(ZfEdvYyuDIp>D`zfMXzBrMx_pqM?5ZPsP zDk$guOD+YMLOINc^HQ!W>mB6kmTzonG z^eCW}JVn&i9e%y3!xxO|$o~IHDWt}o%*<#oX=1}_fwnkCR%&O}JlAQEX(RrlWxzy7m zdUOowBe(YNuA_H69+sZI?>ipP67c+ZS^@Rku^$FJIl7rrn{R{-fg|P&=xd8jg}9ZX zxh?4%3+apKke*BuAnNXE2{1KJlA0_sVEI78>{1CBF=#d1`LeFKT&~3vb;sMcH@tl_ zV&CO^}kDWIBD23wPR*-im#`Sv6qyn}oO%>J% zQl#7)R0P`XTC+k*Qf4X;_ff@U=luDoU7w+A=kXMXX+X%!;sKoxKt1_vvah10sEW~C z&6gTkTS0uX#`z%eVbb`Yyd-(Jn2WjUr-$!(sybD1#s-XFdi(LoR#`13Dh?-h>}KHrf)Oq@eX{?k#|nP8`Y zicb$H3aO4Rj7~svLS=O4EdaNJL-8Rl*`!ev7(;x?f-0naRLzGZ{=AV9Kr2x!k3G9$ z_0%bJnhClBZmyJB6id1^T=v#wOvyfYdxJ36$IhGnA#YOh3<^M?%rPP(EblQ&Glrlc z<9aY;(NuZ_wb@Bi=}YO;iZehIdLVJ{q$CZ`gOBVL(UeRl_Vsv~m(^snMCzFA(AUVx zY5Mo1eR4WAX*KGV+&_z8VO1>c^vv>!N9;EpWQEjX*@;VA#&h&({sWkSn zku)bmDjy{zv(H1qBgJUse)_b=lb7iM-h8i_ug^r-3zHfd11DLblZ&RvTUmr zQd^lwMOQCRII-hNsiAc4$EPe{(BArS07QNewN8h+X`PhkF**j4oF)Pu75zLVi7&$4 zCaH>dSy4OMnM6;eh^o(v65v~)7uB>_qb;Y26xGgA^*KoT9Z)1Tzf?#aG772Zj(0Dh z-rt{iyg%I+v*Y*dH7%i9{3DxLlv)-X@N`Gf{9;G*K*_Gl4yJKlKAoYYo&|%EKZfgU*`!fRJbv z^YE0d6^(`31zxucF4wC|ag0iMcxF~V4#RR=`uVj-<$aQo zJ;q>=9EN!J#%2T1C`38x`Ay~>nH@pw2K5ERM(I*sGNVcZ0WLJf*^TGxAT3teY&5_O z)Z*SIY5;c+$u)Jt(zSpl#0>5s&C@!K(-Ck`Gh&IovF3+H#w5+kBC257MeT4ZY7+wW zCRo6sLccU(C(f{RpA_We7Dk&*-v~zG<_Am;l3NfG>nN0I*6&g|l_!?GjiyQX4*?ME zzg$S!_1!;8JZljZA@=1kJ7k57)|Uhsg!AqIVf880x~|xoN%kY&!~j#P$@F3)<;Y@^ z5!HY`5lBHXt>;e)1^dR?qa-xKj(Nzl&3;9|xmpy|Ow;E5ZAwQxH}a+LVQ*$B5ZI7|hHAI-CoyXJ}pl8jY|fO~PGS6n=^XI-4$*BY3N zE@XAL2f+3Yf#A>%IS9v-JR2$|Qtpv*?QZi;Jo0X|BxRUCL4lFo#Zy^2d(TFe@WWc@pv3rNY!bSy=+PqvcBUm zViAcB192wId%?5}iC@cURLg^ONW?Yt;9}Gncy5;@iK=-3o}P5%_IC|LGv#KsUN&r( z3oe%{w#&u*P;d7#jrxYSZ*REWuDISVxNO##SeNE#_Az(fiM=a zLdv~Qa_o3M)-2GY)448-qy1g(BRzA_h|Gw*G$(;!iZU?@-xmw1LyFHPrcW~DVfvwd zpbufAgSVxLejWddGK^ivl-c$;l_Kq+~`R`>O zhjpr04YSm%XSG=|)y70h-M^{q@6pLRolq;9WRU|Q?#~^@$ghk7&i`JdZ*KU%suxT1 z9^{^_AsiDa-+LG%MBR!hN^&cMeHyW*)U`#< znE!oUnrE7tJ%b#h-jn2Kg>(1wVdMjf=TI4uS)9rWXw&js9QQA?DKeI3?~fHyn$xJ> z1f2!U{X^ksh17zzH7iuvCV!1wR`2NX@LrFvcF);4$<9+D3noJ^jyT^aqQuWqLSs7R zl3El{k%-kE(GX?EXfX7DQX$1;n9K|djhvO-pu8LJhZaK;_w+_B3q!FO8j|hLVlsK( z?RvxQa>M;*UG{ZZteS0vrMccz4oKf2$J4%cua3pS!H!6+jzVf7mt>8k@eDYtk`5^k z^yv5mKGBiOZE0CZwbhHLko9p)=*Ttv1$XWQq@>$mbjzqb$iftV97i6YS}mZ&H~JzC z86E&TMJY*obzG81 z!;0LVomfAFuzc`{1yH3;1>^X9w(U}I=NDY>cMIKpd&B+PkoH}+%{)#Q3k+VC#d9Yj z%fkpde0ote!Iw$thp7?e5g1^NVwuVm$h(smq$MJshNahP$i%Whp#cz6 zF?pE^nZGJJ^5h{^*~KqV02=-zC?>B`6jJGp6v;vc;g7+r*~cflNh0dY!+6ReE54E8 zm!cVNH$ezA&CrJ{C~6V)6NChQu0lS|;1x)zb!e1zgska|U^9)Yq26Aa%TN;)?_~*pkGTnL?xXC+# z0!c0%pH9k_Kw?rYnn3}j8;Zaag@8G!vDuaJf8u}gq0)q=|XB-%0I z4+afp=0wtGvOby!i$Iwa5s$K#zL0vV@b{o#tqsHga`vQbuKelsy32NWQa ze@gT(qNy$GU!5rZBZU;FK%n@X#5P9Bb;>RTGDGBf>3a zeCI($lHSovU<-M{6Gc-RzhkmOJa3p(aK?TR$|*z{|Dzc19giPB@O=Eh`;R~Ie7xiR{XLP_y+@aV#(}aF z)af1RGJ-#~_zJj9T`($v$~eYbA=)g|j+*(W*ri=pI2_#8R-*S_K1}o$M2lXdXD|kd z9=(ea1fx&1(V0XGqYPuzkSLL8VZ!KTf)LTWM32sKF8}Xb{`+}v_QigmwQkn4_g?Q> z>ru*E`hHb)9t=tEPs*Uk_T}I^W$NcNp6%ahmtuGfEYbsYk4vY+dsc-a-8UPm#e9MI zCQ?t4tUp}D1Cj+L$YtjIYze*&3z~>En4RD3XF!VTQ~SlC%tuNs$BvkKH0V{0y6J=D zamp2|tX~Zadz+t`OlFn1cBJ~l`_jtxh^Y=P#aE;IZe;ZB8O>ES-Vs1oE;s78s!qqN zi~vS~jeUF;gJZGrYvsU-*EZCe&Ic?;2R!084LU|~f9s@82U2U&9gMli61C5j{KIJ1 z)VyRM@Yhg6gRAv-TA-~06eIIg$0<_SQJmgO;;uA!8Uh&~Fw|nh1t@}HP-+%q5 zHC(5JVdgZSyB1*D8uD#5vS`y?qDGVL&2~CBrutUQsHkNN%4g*1t0QHm`BUqlj6v*u z5xsk*=yAOx=$LOOhhaDe_54X{sJ<`r81v!#pG1`1U-bZHXHYqWdPBmWcnKRqjO}&0 zwg~fSS+@|lLwcZs{ISsCw#Mgjc-hzTQw;`OIewdZAZCK*V^8NFrTj}djQ^5QS0RWd zE;aQ-^r4&ghx?l$CH^;-Jddfq8r{WtqMf!Fj~K3ac*5jGfC!_;>b!oB_)2{-`UnTB zcy6kCYAFj{l$FMZ{9#C{NmA)SMBz6Uv)k0lm-z55&Z0(3rBe^p!gx;lZvPHbmfW-1 zqnf^qotul@@#$V0;2jlmzY`9zV46khSTIK(GJp_SpU>$&tge32^(G7H@Oz;C7dBGm z6D8MpICtv?Hj{~IQbS5paaNYvv9S*K04^g9SR_7py|Py;Hv1>Nt@n0xpyo!d)s8bU zWgBB5AKiGZ3A3-JWN-Oh3k#W#UE8W3Rf!xMj~^s@qQ1A0YiDW8vIrb+U2&_|UpP`O z2a$&z@1;H@AampMq56Iu5(8c?d{S9K=}T$oZoFEu^o#<1d!u%%mUGUj;|=M6knRzt zpXw5j@ddbx(96LpRq6M5t+lKM(f7#1V?LcjYtNg?oYLK-Qz^uzp7V^Pxq?^>E70PW zC0F6CvJ?gF5LeutG=J_+W;(z1h9%*>cgGJg3uTT12A!<@aZeiTV?~GDi?U)W>1O6U z&Wd-i><8m(c9fhL)8#~62YDnvVr(e!o#WHHN&;n7uXyIJq4Ld|q!*zW-p4q6f9tk=`UC3rl(#$2&&JmKWRN;rKC*Blcr(SC3Mu zu(=XC_leh>iUHkdG2N~_R|ve(g=@0km;xQDiA_CfCryEXjmQAe3MV&+pubJUTIZj) zkh5hpsGhL1!39HvnP{Mhb&u%)969R#5u3nACU#PpHuSYK;J!SHa1_=jgPn$nua&$G z)$}uJKBVNMqEtbCiEuK`+NWQ0q5ULzg18fsT4a}WkH)=iHv-)`z;7$)O?p6r-+6q&QLQcc0v z!SasI;GPU$o^cZ2-me2U0s8Bx*zm0kv^(dH zxjL0egCr=jb6|DhC{{pa#lq3^+k^(s{Xdw0tZM;!v>6jK6i{5Bp>!mb7-+pB=sf@p zME-m-zqG2m| z7aE&s0gMMQfYGsufk%Lp_mDg*4GuQdjdtazvnGpZ!gEN4k~XXj-;3(}W*!F=&@<EW!aQ`5|@NaSBqMFe!CDgthp^;w5jh4S=ZX0NvGSigjJII$RhId zM$*OK3QojRimpZt4Q5>Xzf3m=A{lW^w%&RYYc3?(f-hy2NQsD5kaa$mH6d-*M!4UR z(qp}I!!Y5|h&H=N(^A$Y02;3gpnus11I^`b97&8j$L?oi^B{7e)bv)2)-;O1CZ&|n z&<6A+P|;#izIeg|{LkrPu}+#-o;ybx_T*1nT@M5W>)4Bz%Ap1)n}w({D-{VMLNQjz=L(9QOIO3F2VUPX$?=FdlipxTWRZ|t5F_@ zv{ZrYla(PMEeE4G?*%2Q8RD6~CDl@b-}%v;rByBU*=>^$wI6mT&gqISI=IqZ?O34{-5?fkj}ptZK@q$&Gx}pR`y- zJPSve;ra?(^}rQHL}VVk5_)J6(6qcgLaBc3)zNN@GQ}rHXFIZM z{C2Qiq&vq;AkOQ35hD{TzKc|a+}&^1HxBOa@UUx#!K!J6D*4yfV!YHRS9u zOXqfn?^#-+KglMFyZ6Ng*SKHCG}pM>j#-nOl)T2qB0v@3soW$F_#>`KW2fe5_FdKc z1~a5FUBk->JWkaIo|x=LjAJ4l-ICpNs8Z0Nw+g0c=0mtpTW&Sj^$f8RG5SD@I~t;Q z_pWYy7E7U?<~eHM{npPSueK7ojsbsGHVJnwBR+~R2aWP3!{eA?7lE(b zYMDToZg7D6-+e{}8>a$#e`_5#8Ja@QJRe;s0a1=Q)#lf&eO7BrQjH!7TH?R0bZwlQ zKCdZIuuyz;0=k76ig^*}p{HCYkeYpNHCex8SW_|8d%%fMv7IH?nKg1#c*VJPc1`_c ziX-o3-_uf|x;Qhk+yZj+^Mypf-EgG6+lzWtEd5dc@zR@6uThe9J2r>;sJ#Z{;;J^uGSqBB_%xte916CvVc878nr4k(@QAA2<1s? zZzC=FTVxWqjisc?BcM&^t&D^3Dt*+QB(k$_tBHT`5;|~+gS@f{Bio?LlW^j(l=3$U z?`&Mx^z6R~3xUGu9sbUM@SP?o(Mbu$%=9sN=|ZTCkgx{vEE1rwX{zwgf3)dsGMFYm zR-A=rrD){}?1f2nPGYmQWG$q|c3mF0L2V6=8Z);?;uzC;2FfFLQzf7jpK|=kNp*A> zj$izU-?r2}qO7{gHv)00u+VRpEGw$PuQx_GJ`!Z1xytF9jV!x_m|U4kkc8am)(vRY zjTz|rRT}!Z#XJf&E?l(hQoi&E;Ph?~uVMrTePdH6TOij&)^ll0+z1eF4uhEH;ozXoy(Jf>$`YU6{aA2){`4|tlQ zpLRb}$b`k~GMZiBF928%H+M~47cr4Ce#iQK)fk0%9;0Me*)LG5FS^Yw&GNEqtCsE| zjVF9c@AC1_v|JSpXHO*Fnrx5-oP{P3=n{hr?)h-A%9jQ3oUa~m&Q$nanAf3%QfpB*G&%fgvm*(~mUN$`iHRqk2jSf9fy zMpM=K3CEO2KU!>|-1sfT7RJV|>ZRU1F8ir(o_!|4bW!gz2>V<8X?P$%o5qaX)sF?x zsT^zkl5?K3*-X42l|S$e8&~T#UW!h43Lz)1pi4Qw@)Q&u$=imzqb>aD&rf9i3B>=c!3_?Ra=( zrT?7~kYM z2VL3-pdnVVF?!&%#oklak;d9PSF}~kVu~tfM=TkAS+JoyFYzd;EnjUrz_L9%F^dMw z$Lw!wZ|^ROvmYVL%GUMl@2&)vwaTztT#0HruA^cf*$Xh5icL(0impAa*rTS+kMi)3a3*q9>3G~TZauju*!_T=WZi;(qGBwnZfO0 z2WBsu%RU7A9M5=e24{|Wmcg6&J)|hm0Y&t~5No%&Ih;?$au5f1DuD3C5lBptbWH}N z2q%geNedFALJWlEgf4cQ23roQRD{8*o{Mj0)GFA%w9RTI3#U|8Nh?|~;@}eU-#!C@ z#!*SD8v#&PkA>G@C1a$~=sS-v^_6`{+mIs+=ob1*4RAyc@Brb_$@ zoBDuX>5I+zVVLt;<@(J)u}LIf@pAfw*6itt0uBir*K`={d=Bkbt>_mD zuw38A$wImbLRp?=Y8BnDJo_C_`L-OQ}M2f)?><-mYf}24y8ppEZs(2{sDfoFg9I zRb=9cIQ1$TacZ0Kw{ednwc-=GqNoeA-l&bgV=;TxzYiWP$RZJIi@K)Gs66>(DyThf z0vtZ;OBk>$yZw>+PiYh7qWXfsE`&~ia|7pq)<43ggA=?#x=hz$N z>tmT~D>X@{ozdH9n%lgQ4g71-*f$N>Wkx>5JW%A)D6Vt=)U9ZQ$ zsNUwzajhRW9^Q}IQMp|$i;WCsy^KQ??t4v-jCKG@J0Zh_XNmMzSQY6wvS|+%hd-r7 zFYzAgt}iZVXB{4}_qheLd^f!(QXzca2l@T;O*g6#i+D_PGi9uYxH&>uFXVm9D5y#s znhPzvw)P!axG@^(5Uo1JyIwt=MBUb)(h-q|a+557d+O);j=y|5VEkyP z_K#B7ua-^UtCXmX7ZjiPuYun#u^-ODHB#NSjO*HuihxmDS?3`SD>|b9oAozD0I4Sj z=j|5&bG_@C7;%}+5JMl_E56k_vt-H4Qp>)Y?Mb2HS8*I&H+;&}Uq8t0PWWxcQcL1| z`DElqhK-l09U7_SXN6ef+Sr+I~?T}$)8c*XUZxXo0rKs%uGKmHA zb3XJ>XQXZ24(#%}$d=fDzVgCr{!z{7aKp0k?j^&}pQxj*`5@Z|co&W5gOTk%&uVt! zr=Dgn_&du|U3q1U8ZblQFcz3ikIImuaH^l@V8J9EaL(q!ZyCjCeX$6)jQUdQry4(t_+b> zfk^)jH+mT!>v=$^R5VJ`mWT~YkLe&G9qqf!DI)L9`Pr4kw7clpsvMeSvo4JcMs5hx z!Ej?p`3kfg-&UnD@w)~#W!r*R%g*uYy@cQdYL&hK z;gG&%IcGUlRkDc$CRUaSl>`qMK+JS&r^HuKAyd}r;gYnI@JD|~ZotRC#%M5+LTPoI z^UCr%Ct4b0=^o;qc78xcl#bn(&sSt$Bg`6+MC`@T2do;|rSS9G@3#*P%T#gYoslx~ zVSPuNYcnWxot{B;r_9WCLmv}u@>WmK>(S2}JTX)mv9mcE)q!dsBZU6;$=#-ClazyL z;u_UA)pdIud5h^N|nHu*Q`gHbmU7?&ez~A#D4{c}c4A_EgLAsAGMq~2-?iKXa zdtDY(&P5i{qjG2ZWM_Z6O;l`@F&B!NC&m-PwB`$#Oi5iyKbfgXu4`K=#05s#rGWFy z8r%rzI3wrv>8W@jUhjO@UKeYK1nKFJ|4M0PWOk#tZ&3ke2Q|nS zd)mI@YCSX6%kr}YamfHKsiM22TF;^g;@2t9^=@`-F@@imN>vSQ^XBUjE_7+pf{;~s zxS*qBp(2TC@w)~y#y;+u&({&9MO8W!t+k6&o5o!oDEB_3>a;U0Z|48*< z^xK)EV(3`6GZ*?zz5AC~2R5)s=4M3^QMG%ry{&|N+v0?Z#TPp9>ANDZDt6Nu??-FB zj!{q3e@ddfJh@l~AZx?0;(P42*iih{aiY~vf%nE;ACWtNXu!OWF`Ef+$EU%wo=0Xn z(eH0APY^e2&u`WiT8J*K>QDZK&aMmN+by+-O|yd?pU~Rru>XsT*)S&X{~WTa*W7Nh z+xZvF_1dSr{ta^UzbV6NpK<@&iaVo3kpDw?$N&H1{Fkr)-QJ4U-2#TX+CEa_mhABE OmcF);R-LAO?0*10${;rY literal 0 HcmV?d00001 diff --git a/open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg b/open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg new file mode 100644 index 000000000..1e727a053 --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg @@ -0,0 +1,1809 @@ + + + + + + image/svg+xml + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/open-machine-learning-jupyter-book/llm/image/implicit-order.png b/open-machine-learning-jupyter-book/llm/image/implicit-order.png new file mode 100644 index 0000000000000000000000000000000000000000..e88de5d79fbdddfe581a3af53019d68e640d0d3a GIT binary patch literal 34807 zcmdRWWmJ?=_wR!!n4p3nAl;3mba!`mHw+~WA|fE&-CfcQsRGgsA}ug#_{iFf&h_eSWp~5UQjg`Rpsjzq3dX&*-E@~j0hdAZ5t&#%03 zBR+(Ykr3uH5V|MTn70d74!6v-%g(5e9Q4ImX0#2ar+9A+&(M0j#ui5rc}V*4xrh{P zVI`{2f4<#h4De+8@1v#0i(j7q`!?DpnTs4e?ICGC{`$~=p7yZh)BmNX;cfDOn^Uc% zH9}j$%*a+(!`{%y#=UW=Elu7xd!%d@Bvr!!KLQsXya`L+Udw8Pow%$6Y9|`(2yNb_5*isEUESMGSBOKT#l-*Sc;rS z1&gc<4a#&`zZ^^}D1*!zD>9>7m>*v9`7NVDEPWnh)~-VzUWQi%yP~B~k*-X~)>D1t znCH5i47YU!cg$79vL^shzZ9E4Gg+9?IU1w!o$%?*F0$sRw5z8oat~Wo&tWlR7d1p@ zBzbl|=3!~6jV!S!?L+vyhk)a{UekT3k{ritC=~-rf0C5)H2jZ#&zoGN3{DxN?^DQ4 zyxS{zj^m{^ho42F1M29A{guJq6ZQK#<7{H%P_oslGFO^)604QvVKu>yP3A2^M~eEA zA<^wDZf=hjyqEvBV$wSEL+{QGaGSZ9jDJa?lJ7=77umciNc+qF&N(Si7$x9fW58^r zxqf?g#(&KNB6*pBn-M-C>~p|GT7?5;G7)9?{$YV0dbVySB~?j!#rTtbd?+g0VE-OB zCfd|)>vlaWgZ7W^{g0bG?BFtZ^3sln+=9gsL2yM|_yau&4FAieD)A3=4ql%St}+xW zdL8PpEgN~2*y2P>s|_Q9u_1GWl8Yi{vz++K#E%I}`Go~t#Voy;=wz2eCwuqEXkF7^ zvpu;LSv`nHg~JN@X4^lB#p~S?JPu<0+vhK^!iO3TZC78gN!)Z#;k>fkb<$X8MG&vlLIGoG88Z*AruzmHGo`xP02|Dy#){z2yRRMc zR@8+kvNo!dW@b3IuXIl@q2CAxGxW!VD)S_@9@ML22SkrBw_q8<)_Azqg0oAOwi# zk~w~%X-HP~0b1we#K3W0NyU#AC6m8p9SSZ?29*N?rYtEEk&&Fn`g)R4(akKBh@-F- zCYKeaM=~<^r_|AK?8wp4(>^=wzsn5 zJ!6B^DESp{VHN$|>J$`F;irtErz2-gy{W?~j@G84hBF(&x+YJWHQ*xZyEvnGXFC7c6^xcZ~w={xJQaa81Q) zIooON62^S+0J<}u^TqYMw))wc+q3}vX*9D<>&~(ue3#xTf&2tx+o<%t4Mh*QGNn#f z0I%8-=oLrT-AQ} z3BG1XoL5H;__J4CN&cz#S$0k?tKUU&>fla|Valxg=cd<|D~R8{(58Kvl)bPqGHUMV z?n1D-NoJeAECA*9PTLe0PXVUU=YFw0iBV#K$-1W$7V8_BS(~D&`6)2aZc27Yd+0{( z#rWW$KKW|~G_LdEjt=%iU8Je}QTPuR-h$gO6bs{?s)p@f3D>nGXY;F3k}4$88UAIX z%KD0L%<=K*9A9#5rJSRk*iX3_{qeAs&&c4OcI40G?)I3$Kj0>&jUp!fILa%>b#DFM z@7?!rqj=DT_#NNpty`m&y4bmMwOe&$6mReM-;>SyEq6rE7W~IN5BVSn?euy8({|iI z2B)&GRpWWfE1^wJ8|U84;|%u<4p!@D&b1^g>_7kL8fl4^j3T7I-95N)b=&y3ggGuzfyA?D=d*YYI#a#&Nl>*o0YuaMb!R3_ z-Nt^$|HD|Vwn=H+`OfD>l$KfvUdcu2A4*I(Pdl;OIX~%SmaM^GkVV=eB^Brbx zbYWkNzVvws#(kenn^BLB*k=RGeqxTlsU8`?M7c3jaa(WH{!9zjK7G^ssd=HSD0&1P zlA(~%VTnVOpDj>(@O#m(Oo}dD67ArYDv=5GevZ{AFPnsn19uL3s2d!qvWQVYGlOe&or~Uj-BMWbVUFRp`Z%X7)&&O! zcYPairYiovY9Hou!18U>rN4BPe2kWQ#lkZ$?2=KeDwhGj$5 zPna}?1}A*NW0%zb)^3Q)%;aaK?+IUR)SH40rBlAkR>?B9$}bmhu_!mvo42S{ni`st zGcQ(JBFu98yAuRuvT@H{`;KN!hIf9Djpb=)MQ^P|z3lDk&)?XXw;yR@t&}_|m-A9! zu4KPO3Ek6}Bw;P6%-@v>iL}(i<0A7;q~^m3?>6IMKA5Wi+6#_UTe&`Ey=7*y*ryt` z!R*QBFO3{-OVtc!uU3-!QR#z}gLz#ihngWP1c;^33{Uw|)F|7GryjLe*6B$<4daXt z=<#N8#($Kz+5apo_wk6qyVd@88$X8y@no4l=N98GIlR~3J>r;A_ilC#?%z9f-FmFj zw7X31W#T3?`+zwKoWT`VVJz+J1Co0CwNl3Q^z*NRtNl%X)>L3uMEq?zHAjE1M+~ZZ zYnc1Q`rK(|cGnt2#S!~YVqMy&Hc%p3DTDV&tJ~VJ3;=k$3 zH8(?gD)y8ju)Vzgp6XJ3n!5>Uja)WI?qqoE`}EK9&Tf6*o z3aGhW^Y7Oi&C1;Vkf0?)LiW=4HjpEDb4lG%kHT$DZ=8_nsb8kaMMHyscn2XLKd1r# z%(FKuv6MY%#Ns26xz=dM3$^mp7;1GU=!yNAy~&I-+4R>zYj8_Dd<#vkuhB%wGhkcX zEHy&eV=KF4R1Hff{JzIfP@ijr=M7cN><2RF`P5W5CLsxqX0lG>j^AFadW#U!Q8(Ca z>+34ZTuW4OLqlXIO29n1r0qcFJ)xE?zq+l-n0Eik!J&eiUyP6t%H498bS@GNaw6a^ zU;QOd*ywPS!limz=vbFML_iVL_|(o8ZSkC~b|v}nu^uq^`dNs8i1QsUA~zP!cx;jS~siLH&L z8^I;xJ7MgUaUT=}5mKIl?m+;jp^rTPdqu9g8k@#tUe!4aDEfR(f=A&8lS!^wZIXRR z#EUx5AhRbP1RP(~d_AvGN8JV0^E(@ehvpYH{dq7VsEGLrhZ^m66cs%-rxwOLsG-Ru z>{!|l{d6t8((W;FIwKjB+T@A$HxYLFP{6rLh0EeU%Y+Tw0rc(cuS4y{L~pnyx|xRi z#c~teJHIRG$BL{s7L~tLXg8!}s7Ge?{sF}KV!7t{Z|3offWT~l5SLS+ETc-sJycS? znot6oG`fvj+M7PkE0m6{l()Di(pg(s3k!U&l&sU=XuRjY4srizR?(2fbTD&nt z-{xS*=IjIo%%;DAi~@4piXv4g?L~NE4e_skpkLd)>Oe^iQV%&(QGHrbA#BoFisjQC z6cV;x?rYZR+MGPjq^LJ1)V?&BZ-vH^IY$M-l9Cx~qw5i?6LxS4G}PXI?F@FTim9})BOf6Hi;zdKhJH^DTGL}rxx|xfV#n6SuiN*Rl=_wH9!79tLrU1{ zes{*^KTI8LY=(|E>baQ-Y}lqQMc3x3BbfzU8t6{RAneRJ?e`8vG1j(ey)-+~Mj@C; zt#n3;Z-X4B>J< z5o7s?!f;fmu(~=o7F!7#Rnkznwf?4&vGcf$z7L)J2{UE47~j@1Z)zG3fkwrjSD1+0 zu}`JUjItM282nyACHh~34ZX2OtP-o87yruB-RV8Tk_`FW#6~&on6T6`0(Z(xb7{%p zEW0>bTp5UNM)BEjJh?NGu0PZ-%}do&JnzRunboolMP1sQ_-cI9+8JR9?9TFY;Q1UGU1^^i!SZ9 zEne_>)^(zNCM*-W>2&k}qzT5{Qj*-c?%#^@bi5(LoRV-=7lz^A?XNCLiq(8O%R*Zu z)$zY-xhvC_$iET_w>CT6cBHA6axL=6n(eoQjF!jRRx2}p5zfX`_1{Tbnaw_iN9=Cr5EN5f*Iur-9gKl3{>rdPE;RU?5s%oZIbI((UWxAZeaFSw0am5kCz3#Z zgN%UJ7YRM6Bvx?2wU-~?i7-8)(w&0iEGoV#k&~hz z=DBkgVYNnsgx%bZ8FBw9V@33l2RaR>jgMbLP{Da5)v!Lv`7#nYUh^Z)KXzpI8zf3k zKT=`~k=$(#*IfAX&wvPpcF931nT^JJdH&}@+oSl+Q+u>e8XCU>gHnB*sfHsfc%sk( zR>IMr45t@SIH)~WQ|HfQt7_m2wYIZaUV8+c-5{f@3>;?3II?2XJcx$yqgkkV%!lw< zb%gAVmE~o0z#qaQvD)PLmrn|oGIFTgJ}|cxICVKt5?;#Ax<7XZI36psf3@f18u48?Q@Y2s+4_D0WHjd+5SYSAU7h@?ikKK- zZa#ssGH#Pk;b30`Gjw7Qqd<`tqK@I$_3r7n`?_AoWa zdtMTBNEb?iQkwmvfe6D`m4;>gFH5b}Rhv~E6cZCCS?<_|{1;w1FB=0U%6`hqs={#+ z34>6w$+ql%M(Ur4RSO*=CPsRTU%2x(+Y$!w%>8;>8dc)P3(+WEUv>Mp=M`!4%lf@D zX20BhYs)FRY?_K6*TQU-e|e?~bQ~aDDJ>nJH6H%AH${XanhI1S`M+M_j!xjo6(Z&D zf#UR5;99kQty2@^H=p><@gbyEq~AU>khC{7FgX{}XEKG0t0};*h~emNyee{bkUMej zzjehpC-FROaTLlbas_pS0K7IDLmL86#nwowZ7V+EXf}6x&5|--^a$}2>hU9nKTX1m&kh|L9|30Gyu`R~5Zlr4aXB$(q5&Kqc}!Foh|*@W7V2+!Vm-wA78AKlU4)_xGj$^5 zLQZAz(iVHeJxMnPwNy{qu_&LOv$>d5 zO)pO^CxC?Lf(Jdr)e%Dp;?qi7 z*hX7eQOD!i)`?2p*U;;)j9z*jUX1x|*#aGbS$*)^jwiT;9@n?7x7gK9r43{yXI?KU zk%BBN(K;N${GV-hupb!V&J3CMkEeI;K-qMsCJ9~SkxyA3@z-UgbJ@#5bd41~UCtF` zovY~2O3VC|!9wfCO8Kv~IK|y<$UAuo$Ikl-s}9v=m5z8X4wml`+&{e}Q|8`4GM~6M zDu+kt>$5g6qGrn+5a)??Sj|!Zoz}MT8s^w;_x7~F{6Y0b6D_UVgW%I{@pvUG#2>4% z!U+AW%!+2WgtkoCP%+BJ{luk+jl#mPCLeNiZ}&N^b$jV-m%CW;k{G#<=+^eIzo@SNWoh{qG4&`x*O>xEswz@7uPH-oL4H<1Qjbc zoSqI2MPy{`<{`OHwn#6@>S75R>4_=t1(Xzp%xx~)IQUS01nE2c$>{^}cqub8#|>`t zd%$nu4U7!Ov=~28$XGKs>vz!lZy@UCc_~6P-PJwQHPu~d8Kq%-=T(o+%*%D&J2pGd zCv0AjQ(qsPRi%HB36wA0%a^t1MA%#p8LsE*v}>Zj@G~I(4md2WS(t_HY(w9y-?Qtk z^W(G2P?RbtEbFxo%gfr*Fq-2Nq&2;(ovW|i!4q~6oo0e&?91K=-l^Xn^uVpRK9tfW zhL(iJ1-WTyPA`wecJS9(g~0mupPZx*`L>!fJ4cstIeWMf^UAz(jf zrM6APZ$mq5U5=lYSLpbkkf!_?7-r;iiA+M)R5m4>Ti9%`uk)D{+iyhA{`e7J)!|4} z_0?*ka8J|FX6!#jFyfm2KaHakjnOsl1|VlZ=fq3b=>J7y>HkSL>;Kz>o++#ik;QCI zmZ@K*-kf5Vz?2)dydne665M6udk}gv z27)6-%Jm4*mzs*NkAB3gb;T$|@v{5pPbvzEO`hn+&N;V$_Z~;lAChDs=UmT|F<43Mx^pt)D%6M(lrGZQv&F>FG(x>#o=4b=op1BThEf zL+n4FCmGe>FQeacq@%4pvpJRrPG%uF#b~cgbjhZpyxi{MaG8M9&T3F@vEZ@F79)?) z>rJPb;NajQ#VnkM_htIu@$jHMcyM`j#g@{q;uDrzD3=nOn(DPah{wjp#(^rOs3_>V zsbpZVSZg_A(C&kWg;i@i&GbwI%JH9T@iotT)&%gd|g9%Eo&z#!uL4SsNW zykXenBs{%b?7lvjlAivC@&WsHY9t}=+0mNp%~mRvTyk`Le4WDrkBiIk+CUQ5aymx= zj5jSU4gLPVFD+eN`Yd%jyWRbTmXQqJ=??!pst9Folj_${Jn^boz1AKkK06H=S?;5a za#}YVK?a7wk&&ow6WXkV_9=l9uOt7{wnpwgUH5=&*9}Bp9102w5fRZ-w4lhIzP`|t z*xTD%PEJm<+{VW9{sj7o@p0@So2#?E_4W0?fB)h?a5~+d0?R;vj~`nwn91KVw)1h( zw)4}o&h`AR&JfHLx%Dr2Z14FpD@~PiG^-5cr#Mi)f%oCi@p|GaFOuUl-{cHA?f$N& z%7P=kPPf-~cIYp|KP(P|(=aeFpfEOJd97L~7aAUp6)FbXh)YcT)$C>~-{|5(&%kg$ zAj|(&JFmm0dKUB1BU)No>!Xx1Md`h(t1Dh!-nd!IYAx_u<(i}|@6TwbQ6rMbLIUr+8pX7YPWG`Ki9 zeVug2c^n)X8ah!ZPyPUhNss1g0}l`H=g*%40s>(TE^Nv(X|yA0ZU$Ex8XDjhc^O<; zj$|HVtuO!YSpYW=4{Gn3nOaNpJ3Z?9DvK%l{gu$#hZ=84PF8N$ z1@Erod@fEV6bqQV@CgVoiTK`khrNQnt^NhOmHM(r>PF;zNpF zwxdCaHya*HzCCL&T{@aQ4(Vyv!#~2Rs;cngY5Qq2d;2!WB>|$S^PQ>=&Lx$}n8ZX0 zfs@TKQ5)V`t@1yA{$%ia#QHnT)D-0AifoTyzImr)Xej7;WF|fbPhqd&n(*9HOlBeF z<<%Xr^*{Xc9YHHW6opGjSV3KXr=i(~;d?e?IW;vU^LP~f^QTXr5+BQ$Hjkvi_4W<@ zZyrGI?(W~ee}@g+M?;In6&FiQ zllJChDO=~))u~;6dr0Tl?X6i-TjlBH_3w%+D?wQ}lf@L9-bjD+235H}YR z@q39$NH{FD>rZ$)rQz_m2IcPA+1TJee=a0&cRV5(Yxwu_`tGXZ4mH5I^Am^V7Z`gX z*F?xHF`LCx$PT=$(L1+)wy~k%SL3YzT$us&g-?6`+u%*quD&>GT`D@pO##Pt|2u&h zYjlh^#Wc&=jc(S~B?#AQtyZ@k4d?<9r-(gtceFMjkF%pIE|`|#Cs7b19*(ocbDzIp zy}iNj2KnvVx8seGr6%V*xTwvS^L$&K8Z&7LiJ#k3Wp1xg+i1zi$zLVpD5`=wl2#D( zMY?)=UdB?&iin7$GK;|RC(^b=_s+K_EWwrrRolFPwe<79r^?PF6rClCls@-l^L11$ zQpmtgkXBHjBF<+r$S*3gG&5sK8C-1hrbo^J0Ob^pcTC%vuA-)-+~Rp`&CSj2^*TWU z27|f!X{$Pxz^kgY%E8Tq0x|`;ZZcivDV8q|{O@!P3@YOwF|x5X?{nze@87?tr!^f? z`uqDoe7JWs)7K|`L_Zz|z-oHXSjN`Ou@JL<#Eez&=0tULE3;T(Zv%nYRNw1BRjdtQ=lkY=6r|5!$`-f} zTMec_q3i`Zc6Rp3>1kaptztbZZEbC3n*@Ndd*6^-(Q7qNYBjlo16mEr?a>mGjorT9tVq*#Djx_6jkeSbXL=q8M(ts zD&S5yI50u)xVdNM<^~o57)R3{-Twecf3q);mXB|Ka8Ry=zuMd;E4SFO<0jF^-NR$L z@{Ie9g0OI)6a~!H)z!tNth~IuwDjoYB%bc)>fgstp6u-GB%HWK@I`Z+M`A*kmzRhm zO(CHGo~vLz{51*$w7j~SXXTvQ;0X)(-C$1uw<+>)y^tdlSI#GllfvdVm&U{=#>Vbi zgHd**@_TbaV3}plsxaQXd;aQ`_;g)fo@p1sKoWBjogAI@+3pOpVY^(*hU}xr^6kyl!u&i$ zLP3%5xYQ0{hm@351#>Q}du1q%tI=U$c4Wj6>?(fmGc(qdz1jM?Ht$9)hQe_0fE{=o zHqApuCbA#an~<3)@cz(ZzEfXrCk-F@@w)hjHl z2~f3?l05(ez}=S~u4rm{58f-**F;Mz8AtgAsNy_{2tb87IXQ7u^4qiZWuc*crRrrz zM@O5i04Yz7kAbWq$1?^@q@yznP}APduC=XAddLB+JIV86_~^>Y3g9ByOpF-42PHTh z*5ef8>ELn!W3;uk`TF`+l$Xb|7XeVlqE`6$>C-(Su#3>2;oparmzQC44WJ-BzkhhJ zzrSym0N%U(=1ez7NlJM#hFlWtrTO{!-MziOWjmc>78z;5 zngY-l$gD)Kb^Scwi1#O(FKG4j^()pNH?RN@0ma+jkS{*XZqoG-(yFzH*xoyG6c!eS zqBAoqw}{8POy2qi>%(vpK?QVnc1ouJO(W~)wl~WGCG&bz#)^Ly%e}ce_t>9L`EK;_ zJHIMoJM}pv9jSOTe+HtWQ?$IOz?2JXX0EX9$j%n(z#Udwpyg?Fz2jstu;(b<{jdvyk+>|X#iSEoC;FJ2g1S-r->BG-OLNm&3c4+t`4 zW#up=4`ZnTjPKc=yALACZ2C>{iim_+@Vl`IO-2=i3_;HRum#(m%#pAA@qT zAtFhZEA%5G0cH!3=SMpkK|w+I1F*l-Gcu%8B%_F=eF<=IewtM0(}lkCI@#O+0_$=k zE3V@i2FC66vzcl;DS`}c7mF76JqAWb!`Ao0%4+83l-y2d`wP0IK4AM^>es*jYrkZ(r0;DM!1_37nE2~OP8D-Kq^k-|L=tK-w`N57UESHX+zOA)Y3Y~gp z!Rg(bm!cz?{C#o)mk2CY^UI?({D{fkWhXs7JzzdcREsu40eX{NyZRn>#knaElfd99wbM~F&u(-CDefG>T0BY{9x91BdN28 z4<0-Kg`Mv?h6U6%1_Uf!F8Pax_YvtDN=ngYMMBZU^|KMEiYjI$g^D1JNMREub1X#1 zoc(Hys=p2_sTmm_qp%G|8mB$k5*_h(<(TVhADGJ=fJVxq@NSbDTA*BPY;Cu;0B0g@ zuos%$2zXqJ5hIW|M_DXSnb461;A*rW(o0>#F%j@>-wC+W`Y;}AWq>Q&jWi|diF3)I z94P5EX|Nj{7Vx2b2_b6v;e`c#R-I-)^EC8=#dZmD)IaHh{stL8W@ctwKy?)+;IUbZ zc1TQmto0|H7uk{>KnDO@j{<1JynuQIC2)JO8iho9h2{1>u#J6-jf-2QQ*Sd_tPCVM z-G*pfO9mjef6x&Zhe<3jU#ZVKZUy!g_0C1#h5jp0pc60cvag!)o)kd(CaqZ!{xVhrtx zbg?5ShmSf(Q*SKz0WG4nJr(^kPT@eSHpR>889Sx^(NPvACWZa1xaAu%GO~DP@f*&^ z5YPk(=?YO_X}E8|GUf|&v9n9z3=a&D2T%FjB`VR>)zsVz*xa=0+cMjvhYk-9v#G}w zsq|al$7D*~QMLHrwNDNb#Bs$Dm#DyL#g(ZQ(xK4S6gzb2Atq+gfK()$=5j9H%4|{Y zUTuAS0)K?0VN zyDh6Y#O3XrPbdl|=-2UF;YZA~m#>#i;&VS80Ge`2J1jrT4=H`gw^UHM^(_z;`h!5^ z^BINZZq7Z!WzZ5*o=jxaO_rk)R|f0h8U!#ewEuEc&A}2asM>u`r*El~9q?K|4N{*1 z`|-`qjc{RXCO_t@C=YdImJs6;OwlJRkg2x8U1^F9Oq5t z4a`cc=dmQQF9qD-W9+PBbt@je?{iX2w2O-i8f}~kZI_oaovky&!wO^Vc`nU9Of2N& zd4+|adPdiW45yywYgOl;EEQ4>?szGxm)bd31)ML@bhNCFJ&6QJbd<9qvd zY-T29UPU5gV}Cz0DJdx{YpLE=GeSe-5sY%Ptg4-rlXHKn4A;~71$2Q#qTRngN-)2) zwDj$p5Dm6WyN&1LHrmm(7*ZMfv47b~O@h_$ z`r_o|1Oz-l%O&)gxo?hsUtGKeB@m9oSXxr@=;_mR4qF`l!g?NTd$8KZ78VwkmeX(- zOJv2m$|jM-n2R-lJRr45CVITQynqs-qM{m3vtd?_zyfaxngVuZacK^4PVrvOk}P*4D5f=)6Rorh&n z{_Vy!Y-(zkq8f7%O!kjE7iLSKKsily_$ zNES75?}QP%t*z}0x8q!cJxGa$hlfq7FF=L>L^wb(@P**wy0o5C6SzL?1S+2o=-A(1 zjh~`{q>c#0cR+`(ettm?1gtRb@#*QB(>sxzQQ)JRn|VOp=I4LxTpq1bvYEyG!TPt@7p05TKuLx%ORl8O`a(=lKmP5ehKyl~l?w(lTfs32; zpE`)Ape&QiicUnzY~RR%Q`N0D5eISruv%DH7%Lmwz`(#3Rnc{Z;x-7ZBX&g+NDgN0 z@xZ0b!(9-^SEC#;F}htQ)mjW10!?8-L7iZm`&}%D02l3lI;Dwt*9=?h2XUuDGK+9u zfGddD zM@3PSvy8?7C2K#|zzBpuE)8T*}OhKCJ+ z`vMC7r06W|kVHjSw+tqbx(Q$=j69P3*qg;#b@ zm1*MAIw5*rxLJs2C8=-3TRJ%G16;T|**XT&p4jgKX!5|G)wT6?&I#FAfT=+KvjHc6 zc<6-m@U*Zftg5O)&~5;S*VNL&to4)){TEpTlJ~@B!%;h&1VFX-mL|HcF`^F;bA13NYjp=YjOJ zSV-x$D|t+2W;=*~z&7sc?vB7?`!rAX1|J{!9i8|M*w}eVMST9Z={}!7e^$DyvO0-I zgmKnO_}a;yS;$xZtsgOB!PmYrOJ_Ybb=aRp4{+H4V%TGU+R#6P-8vSrv%k8s@-NBM zhJhjlwXx9Q?+@&H+_~G3{F*S>$i)+}97EFi7|-R?;-ZNek)8pgGnW=bQIbD(xL8<5 z_y36b<=8kMj2$j_ayiV?oWnX~%#=i-;;O2odCYnt%;w4?RwZDGqKNpxh5t)XPPf%r z1tJSI%C(Fg9NKNB=n^QwNdPBKrI;z#T-)5-3|5ZKa=2WpI=8tQ2{sw^YWp5As>A8r z&>`?VRfZh~;nSa?Z}qm*|58o`wUbAW#5bVLEO>+`N`(6-Ok%%3-%vP4o=I&jv~Y3T zcdq{R@$pIJVP|xP>y0CR+ns?Ls>^spv5iXt5&)7L}AR^ z+DYI9dR+t$6sN_SN$Bb66TF>4m=7)#B^$ITG%NMsPRl=l@ge4W4@xPfsvT*wJy`-6 z0uYAHeo&+15*0D>Y_ClD+sEMG2bbB!ib@?(_REWMObfr-tEwEyZ94!QGBGio?=MWm z%`Sn80MLkHw9*xd5Cx{+#AGGO(9aW?ZeThY^_wS`ma+!cfbZG`D$tJe>iY5+i0<*+ zGdHc%QCIXAjIh|B#3a~+JOlmxLVKuUPbRb%<R|V1XprY#~^uD znXcWe{R%w?@kE{^CTo5{IcN`YG)@5&-Ct~z6(kjQxW;@8KFL8p*W^s@IgIvCx;=yd z3~^>avOy^X~F8=@m}#x`v)! z`LH?*{AU!gVoe=boj!Zc)6-KxQjF;;pfeNLs%<}0W3J+xYZk`~L-YdYu!WLlMmsaz zd+-Jw9o=n3T2WEc^;DY>pD-EZ*?n7yK+vxoUuG|2DyUom_5l(ysk+)3oSv_5TTagM z?o6#WSPVec+uJs$6Bu5H41JbACM6dK(?RI4`6cg200)whl?7=BAeARTvu`z=wYYBO z!f5+#1!-uCL44qEbJw`!M_x6T?a!vGs|ym>4PbT@6cj)NLQg-iv~>3)_zB{M7KHL~ zK6KdZTRW~=r`bP+DZmdwVz_2Y2=<%M6pDza!aW(x46|v~V`%J$t9usOjh9CW?dHJ-Dar(f$z#Es;f~g}_nwlK-Br)Q|HP*gDyo00EzxrN&9Pr{jDR>D6Af zY0vfbH83TB#CkVlLBAnECiT*%9U=NXma90%TZ?jx8>nPZc@q>)_@1gMLVzt8cVGv1Qn-B+Y z#6uJWvJ_LD$*Z=xu)t`rv$Hco6aZUYrnH$EMb=1Bk3v!zXvjc9g3oj`G=45SZ7!+b zzx>WP_h)nGN35b65Kg9`{{Zqm#8hhR8t6H8T&#p^*IJZd`GC$HE~CzU2=oWCmwrmB zsjFA$x17Z(3Xq!W!<(D$33+ zA5Vc3j`#G_?jNGC-?Otk*4>MX26<@+u-2hwEGj`6edcCnVp+2vxr(0ly;6t1e*GG) zy8t5;^h)qX5um{KF?Sbe9-n$?O;q=ZdzxK}`BP63M!U@F79;d5S zB_*VS(?AP@j+QVL$f5yhNw?d5aj*n7q)3wEZo8}s3P_hSWld4cAjqRZpl~7Uy76vY`y~WAchun*g=~4PnK_V<1*<< z4LgmdeePL}DYh6+msVC27Y_yv^Hk**j~+dekdSzawhA^)P1Sim`nRgeNP$zDr#2%AJuW^L(|d-Jok$9#Kpw{ z$(dGbf&TQ^*cj-g0&*ia%n=H~TQV`QgnETH4HT$JTU#$5gM$rCW9k(`jA z4oE0V(4R?r_s{nr3VSg8akH5hB_b*+Dk1_}4#2zhGU`ExdyPvnw|xLRv>9;X`rfb@ z1!QC_01bp_N2>hF6%PkZsDB#K)|LQ-!N37L!lHWr{=Jx_q-^^ko1mco^~GN~s<@}P z|9Z{L%vGm5;1gdfAm|Bor3aLhc#ZY)Ro}Rj+G&>Nym)F88rc>OjPTqvm5 z)hNN?p`n6-bOB$T8nCb_pz8qSoGP~Xba_I;s}k*h;bJOHLMET5h(hx@@LHfb3VgV0 z_~S>9-hvF8i;LQRZge!7y$BqT_4QNAp%QR6z#THH^TfoA0+jnHI2d&CO70pcKL=j2#Ng1qmX^Uz+qPbMJ}d9R6EC>PpbG0a9xTq|#tu0$ghOYs+WyX=k)643#FOaoMV!>bwzq{Zv9stPAiVXcd4|kHdE@ z6UiMqi6UaIprWQGhLHhkRwYD%76Fl7+DcZj(8!}jj&Wg&g9NNFKmQuU(FkrsaUe9~ z;^Lk?eL4#Y=i|ZO(yFSF{(d?pCX3n`FsDLy(4v{2lN0!bM6@!X*S9T2iULfssDn`~qM9=a!53j)t z+q~3ORDM|0^7He*ijba{>oRe+v9Xbs4h11RV4z974rzHX#sV}AxTe|$kXJX?{{kUZ zd{UC7rKJm855S(Gp&{~92v}B7nCCCk`h6otHW~r!XJ`KbJQx!bGt_dA=LpA59*oCa zgDnRP^Z8nmA?MV%?l^!}z$yIBD_vk}hUP%--5UiBjYQC#g~4KZzrk}C%j@dmM>bkO zqBNTZPCll~9GDrH=(zpfllvcl(gG>k!QKK^!lWMz#_Bo16W(14U>zpxLOU1s2@e(y~awut;Gl3XDsERfdr}7M7w1fb9=rN1~{W_4UG2)uU|$#o+0ZeUbCIw0 zqL={T-Hahds^5N|FD;kM64+<aAehRu-=Nb`SyQhm?2*zC50&(=Su8@QkP=_y`r*1=Z>CZ_$Rj*fzY zH82O_Mkt$ieco{gTKvfUmV*-)-klfPnu`1}feE z_L#WO_|tO{%^6tQ6Cjurw8w>g5mW4u6C_u_8=6-a&k zfRAo%<-3_5Mp^^Ec`xt}I0daMDaBSs^v@ooj%WeuWf}k;0a{Nm>q6aM@+4$rCO61aU?r7VQ<-yPW|!K$ zx%i(S=svT(TR3cZL$Vmt3Z_7|z(MZ?8iBO@eIA<#vJeERAXNo3RQw+M>0qo5{7MgQ zV5ERU4GRq&E35o>svv690d)a-vN;EA5==}ERBa}o9CEOn4p z2f<0jne)s6CMVtBw_rSXJiqv#U%68M#RDPm41Br+4AG{(`oI2*SpJWrt5OG`Po6-p3F7Ue z%WE86nxug}5buKEVa7yOY{(dN!2mqv(#U{8y6XJ=e9$-m&g0)iD9Cb7&(8c#w-YK+ zg&nJNbIH)rLmJH+IY!-}holsHVB!E!rlOP-_~3zUbubJ*5Bh|lhrwY%Wm28o*LD1& zncZ!BG9VxziP>=fKbB&XC%y3d8Y~3#>K*3ad3$^Bt&hqvvXPUMgIS%ry2+_2^%m}E zwXKX2+WZI9)I|W3qVPHBSXi7uYpxkX^1;6kTgm@fA4*H(a@2q|1bjCF-VVHvq?p*p zkdUUpc@@yhBd$dhGdWvLZ2TR{r2{_tr@-_nvl z=*WUM1CwDlfF(;irNEmPU5k~2z5s9$X=!uW!a>(JH?Lj|3=I+Tz5fc3BkUDD`!gXy z-F5KPJ0ST%dI*Lxz_1g@^nv-*M-O(j26>TeB14Lxe|uC^RIINs@xN(mn|qm=ng6oK ztdXS}^8t{yLq>pM!T(f6P1~TWNKX%R)$Np~AqP7L&L zg^ZFdqa#^S8QGDjtc=J`iX z|7xnS5x)Vh9dlrRv9&7Uma_l(?+0@Bf^}V)Zi^L8!~~ z$N15QBM^bx7?bHlW_EdMs4m)TQ>4m^P{1b?8SH0O_JtgYWsGQ8oEh9i*b&RY&Ms_W z0}Bs&-spE|8(B4hz@%U+0O3Fib-appyq+OE*7E^w26M4ntm;QQaVv$D{Y^E+SPQ=M z^??^!XM?%FjW?RitnqEIHZ6ch?re(lIyjLaQlR3ke}*0o>MDx{9>b6A?d~Wvi1& zUq!gqy7cbdivdxG=I~$8?bcnaS=3IraE!ry_{B!6u)CbgW+KZR@f2NZ6E$@z#(2&y( zj(*fb0~%$e(+Da^V>wX&@?}SD)Znh=_a@_~jXcnEN$OWVpgw_{d=jal!gY8(NpLVu z#4wUa86syVmz1oGOuXeCOkc6w!j~)kev1C_Mc4ds`dY?c?o~AGFkH*HYpqR9a1u_w zFU^Trmnw1%Ai1Kn^j@Q|u&C%m)K416sGxjsf@pH+=s7 z3nRA6tBaPfe8OIqDJ&`|C@3tf)$TRh+1VKqvH?#7BO{E6*}1*YA7QJ{@!DO6<_+-* zbRzYpO&AyoK~_TQ01`E500M=>!`L`JCU+@d6_=#T+7qn?Jmz?sjlW(eR7Ed;i|-*BvGRI6R+k-n;<~ zeeL>nqo7rs;_b%U3gOIDRUL%YbWGn5xM6hEZbvLA2Q3Jp;t3k?N7urLtFIqfTN{K& zviai9hX83fP7g~<>t4LLbonw%`QdNNH1B_X0CoZTFDi=sDuZhQZv3JD)@n(L@%iYe z9t#VJnTA&-;8sEc%vu3-JF8J&#ekGgsKM-?ZKpUpYkz^sdtPOA__e{{Z4yKWpBnjCW`T+H2(m@s`34>o}LxVjDMv^ zO1xv!7Iqt3TPauU9ZpV`+KYjbwhUF50fpX5h ze9SaipC!Hb-GC=Ri-#Szf`7oYCg8E1wnugV{E|ojpi^Rosi~y4iRy z0v=m9e{nW`|N3so73ZQ4z|4g&>o5p_h>n^BoSPIYiB!gz9bQ(50#Ac|M>q zz>*_&{As0i?FCGQ=n*%=!=uG+$Ha6%t%6*hn=7oC|IaAdZhtKz0*6lAjP1mU6P`wJ zGaE3!L@Oj9rJ~mMz`^dV+|T}6u@cj-pd}-#A^xtV#jzz$#KU68%|?r=^EgN+jvRSg>HZ1BA6_cIze2N8%!@Jjluj8>o;yF1@PLs2^zAn z#3PpCrSl?n-H%fQ!m6&5(-b_O%}x|6CXpnO7!lOn(?hW|zaFm7F_U^`37~iE80-0g zBPJ%Q?jsut$%hVYA!})BhAnjL{7@TYjp!frHo8h~nmDS68~Kedb2c zAJD++Uce1@0ca9@+0v`}<{1FtbSrZH0A>ETMR{Glsd(=5p^-smX67SKw%?%1+JmM= z=wj(HrUY(zHv-8H#vtPp4YuxR(JK3=;X9{=g6LY=)G7&d)wmkIPA^RDWvyLkl#d)c>Se4hUm0D2-MDlBYMp+(1!&-yak+LJ*C zopyHqFw9q2T->Yhz`-wq(Vl!OF})}-722r3^yC=LVGWutm-fO z8XCre&A4_==hUek5+GBe->I_rnlxpUmv`0Gsq5(#LJ=^Jv=Ym`chEx&0fnKlPTaeD zJpDo^K_(8?1`YQ1-g<6xWbhePFrU0UJBeZ5;ZdWYJ*ex#a;7sLOnvyr& zVe6)ewV0?|1WHU`IT@NfNHB~8jA2bSQn&Ffw$;#VwH(I0?jez`wZ3`sHfFn*V`Vez zAbI{h1mE_V-prB7y}#6UO&TYGW6MF95WU!Vq{{YRTc4>#$8o=5d3m`dU;|Us<{|+4 zBdwP&U2^kOORwOiPC zVYvQ+#6LN!1Yh z)`X)U-L%=o^3%>(S#b}*aG2e4;L z`q@Ya%6Ju^tHJ3D4GM-mu=$gdlOQrH%F6K9*J-NOf=pYv70V3wd<3MJdnsAFVmmK~ zF9T**g=Y>>8eR^#v_XstzS{t*Z6x=^9k0g3^@XRCZr>i>%5(;PQE=&FU%m*0U_?6; z_No05%v2ILr@qY0xH>p!i$7PtqNoL@Mr5rzo;Nb56yVN5;vLnX)V`9jbC~;{4v&p# zh`YMFHgfv#5J;W9y;tROixVZjrWwsDb^_&0nyL{Uz}_t{8))9>EP(`Y$k@0GKa2$Z zow@OLfv-nGc9QZQn2MzgA(NU3yJTGKrtz7X!QtOWm{96N7(iuEO;l|+Wg`-v1>IH& z^h0I`4i5afQg%%Y7+d}#z$VP+3g!efJ#?2-sJ#?*-k!xLcmDP4&tp0mpejfCVLOVq#)Y zi8&cVkC~~dO@z{wW;;3#Dd?_YqQ?_=2Q1BE%mU$s);8GH>V;B+2e-1EG$1i8w~4*w%wjje2#I-5hoZtG3JJd>(?nvf|*!1vI&Tb0_#GD z9DGjTjQ>4^`fW=Z#{X2K-{0Y6FxL+`3KSR}@K| zdh7O(+9F0W&fg%SKEPR(Zlt{K&_A))Zrs>P&V@$P>J+Vb<*+ud(=wET7R^|NPp^Ydxe zB!9@Nxl$8`)=r)}wE!}u5PTi3c&~Q`045h*PTg)c!Oxwz7V(u5kz-;qkJp!sYjx!_ zYLq-XFSXsMf$_5SIDfR(W(B1UYu~j)lW2nN>O_TwpG9skdXRODF+#rNz1HL&Xhm5f z@1Tx6ONu=r{0fx@GUA&zU)MHB#Z-e?0&X5FuIcY`qD7dkKU!b*SsW`p7@czQL#hed1NJ+~Es+qVs7w$z)>&XEpxVOe64LbJPny zu_1?uF-b;t98Fc5&Bz^aUOwYQ`2=HBR^96unYgbd>`>1`Q!$Mq3zSSoMt3l6_yvIX zfXusFTMH^Hm)h2|=t{w|k5mw@A#OT%;t-Z~Yinal#>2xegJ&9V3_1%o<@293;fg$lOe?^C}jyYrXfS-9f7XJ7PstA8!e+q!wAyd+}+`*49=6b?Dc; zJ-9yDU{MXFaStFZgEiASs~W9UhNgg{rE&0}f#y~hu`0CYCr{4yR}?C}l$zs=?lKq^ z`hZlAe}0QlP(zJc_3m%{RMi{=0^-kY@V~;;j}#`w^Z-H(uEMA&2H9lp?b{vg>_%s1 z)_(R_>(*>)u-)VSGOOl2%O;Bo!8JO@u^VEo!z{g;IN@t{CT+LimDgO4r)it~c z8VD}9uD15@oDV4cmbSJADolI+U|cc9qaAAl{v8cOi(={26%`f5oH#WxAyO_rYgC+_ zeJZS|8G@Fg$0#w9@Zb7 zvhK*~=eyoTQ_-G*$~enQfjEe<=kmO~V~e^UQ#xvJ8=1-1WzHC?09@MS1}SqkSX@Zz ztz#oLbkHJAMM}!bz&vGNS|D#n>qNnR*hl3}Z|0zxdniSI4!<)fp+F2v(2JJ>E}tf)ADS~+=-;EKy>?m4JAx|YF8wB;sIdDn}`7xsEx z&C3%Rh)PMD5X;=#WSw;7*IXlSNg#!!Y593LbGC|KNT}h}tNWPHb#GE(Y?HKbHq#cBTZLQ zM|XGjP-m?mszT$-ry#Mk3!_ldaCe7&o20nS6mAIc^MiVfWZ;jG0fN%nmOQD&&A}nU zQ5NF(>exyVN=-G~L`kH8l_m4;)6V~S0p>uL5YwTcBpFHj@0O`HgCCPC@4?Q?VtKQT zy~vGpray!00%*SZ)vNmH9t@g*R%g$iO_g=o9#)5>#}fGPtc4)$tFtY7wR|C%>1Q<= zyS&M)onasP`tm=pFp?=hb-^h{vfw_tc#x4nL2Kwq4R7*n_IeS`au?yV)h$GbAlqi4 z9tlp{oc^XJtvtFG7?2tUy?}>yI(U0}V)~RV*AM$+81B9C!U1?D%I3lCpIG$#z`@Mb ziVj_`opkqu=?-5S%)Ua4m@6Vu)mT8F7R6rkB4an^@f zS)q#b5i?2b4H)np97Hx>|A=`^1pHVw3ng7J?iy+kX(T^DOvtaUzN4k_;#>5%gXsmc z)`KNqDDf$!xbsLs!o~)E?73>m`K!gC!UM761cSdwrDb@fW^B`#&PDDIV`0Y8&UZUI zjgs09AA1wmC%%n|=|SFaxU6CX#^83NOvp>F!X1YW5H##8o_LB(`+(btR+!HLHMcqQahA!+f!-&V@-V+p4mC+`5g+JRwW{aZe7nm~=(5HqTT9IUn*7aH(4I?g-a-2sNeEkUcp zYoPdrgd-Q14&F-k|K+_dZ63nH@?i7F^aem~CS@7EA! zmN0v`;V{MQ5P#}HeFaDJm|pLysMCw1tnvl4CF@Qar%rA+J$f|QPCj;soV2R}qWSD6 zbNsGtZ#)1vkZZpd5P}kSy!OJ!w{IWh<~Du&c;`uEO-6<-+za67>)ut4@?Jf4`{hxA zw}Xy)=g-S}b+OerF|lP}1r9cXY!`l_?&h~!6eKr}Q?FmnT_dFWwKP^Inwih}`7OiK zV*l6yk#B%oCa0_so6zFARY{f%x&HqCUGDPKK{hrU^;u)KLM?E$x7R2XC^#KMQuEq~ zUwaAK1S1(dk~|GfdT#DGZW#u|Yx2~8Z&~wmC@~2@+ZZ<3c3sj9qYnvUzL(s2-_T~gvXLHrvBSmrTTWOEQfR^zEf z_nGxKn9R*9EEM8(Sc*3VG_%V3v*YCZ4<7=6;E3sz+s683C#(kGIf>~YxD8Z!6rC*< zitK%8j#^k$l}5PxUp@QgJ{93vRWs`gPTyTitqr-nS z8Sz-k;^pQR_maHX`2EL^14JwdfLmn^t_n!?FP- z8Qg(8Qzg8efPOJ~+TPxaSa1OrVPyT-7ux#Al?-Iu;E3>W7_BH*a;^waEw-{9t~WSt z+D^gOgh8nEy|b9?z}Tlx>Uz5=TX|B9-aLOk3=N5*#G$g+tjy~wDJeBe`+GE@a=x1a z46KDx1F_TD`Tph0mti?W7QpQecpS`f0wF(s{`>?XxTOU^6x|$_A7sl*Lg1cyZj+*L z?%<>q=q9B0qN=7DzN){l*^OZdjmpBnASODx1PX$H@ewNn@_lt`u9yO#|`HdYN zMJu#{mirL2(-LAJ&c4Hrj31!N0r>DO-Iv_7`}+Gq8|=_e$z7@w{E8MzQA?I<1q%xc zaSYl_tOGUL*hjv8eF_H?+!dl1c&PM|Q9Qjfht7cy#T_2_7nH652P!_lR~a~C6(o7--VOejG0MN$xk^|iEo@LM9D10R?TTt7A!^vJ+% z$95t36oh(+856TPGSQD+CFh z^oX987D88L(Ap}P)Uv;?NG=8Pz%vX>^hDxaS&8H}{lkYPVH{gi5gdA(a)+sS=ZUK< zrlDI21T_H#g~_W8Ghpev_(Dy1UZ5tfv~#y`zyWPpk&_?r@Fdy503fMxGhFi}8a6y8_tRFw7=KJa84M-(Gkq9>-H%UQLOE2+Wf1nE?`j@%dv^{K&G+vDtuL5E8b5&* zglB)JDuoQXee0HwpWg*!kvbJoQ^B|bWMUFdPk-Lo>4~T!=xuNb_}^iB*g}00?E@ZxH+#7+2vS0FMs_176@De2cOy4`^m+ zfDq;^C3@gt)iOSM)E)E%xC&_38);dOWIX})#pMIQMy zL<&5rQrUv6JRkZD-urJdAzDk{b6JbL%A-le`i6KE+j^FXAv%-Op*jL)9A_U={9s@XOA9V|1(t*-;sRggD#?8rt_<6{`QF99HC)c;`AG!;u-z+hi~E{_N& z_2ZASIl(=})FHyFE!yUsoGE=d=Ka2z)6B z03mS1T|>`E#%aK?jDIvWESy0@MH%9&L-97JPo07%+fy{z-K0HzV9QerX*oGCU$miQ z4W`8(`WSC0fP_c&g5?CuqmMVZ00HCfg780l*cA&kFi-pSKh)Z^lDv*EE)J{jtk1s5 z5hlWsB=}OWx&i9WzPvvYObiTni;H>L*cK}`uDh73CdNt$bHW1?9s1Oj_U+$)EN1#Z z4U_~-S}o!)sR*K`AGgrZ`0aYfW<`cViPNz2bDXlf_vhwAIcKjzmJ3qOi_!yId_2kP5L}$2`+G;U;4y+XR z`wJmmZruby6z})m`8-TC{Rta=Oc1`^uHcaguP1C7GlvwiShIzD~?w+kn2H?mDIomab#heArBtBQd_GkPSSdG=`^(xr`VQpmlZsmh&uq>d%3y! z^gs0EzzBkl^L=qK2oz#}!dTF=aIIY53Buh{oD-CQdoaBU3XZz8cI;sYqa$1#(87}m zgoH&T11GQD*Aje!+69cGoozp4MWIl;ySrze_+Jr39v3D^5MkrSI*#}?v4{_oqY~$4 zb^;M^LE&7yyaIP#5vQQXzpuA@hH}>Z@6X)pDaY&Y%X7>B@f8$}=I=|b@#4Uie}5$W zKmX#{Vm=`udz30vNmq*~eFWi`Vt!-F9TSP z{)hndBXQlGC>*a*CkIBiP?E{4-tqegZ^xALG0jrQO>ls4B3I3peGHlGXZ`zYWk$hp z$Hc_{g*WNB&G^(5mIgL8H#=Yu2Cwzd5vx2dz`%-7$1-NXD$U+$vM}?t0~xP*#z5$x zj`a2Q;R^v_u}o%hFsN3ke*LPDr_;^J{zo%dz(O$zV=J6hwO`lEm)QlcKXCuMozgGZ@?;t=@BJI;vEy@25 z+gpFe@B^rDL=ue9a9{Tw$7U4l+j;-u1r-U)@t&j2n}?$HAO!LX0Kc&@M|1Ss2Izx) zd}I!i#vO-QuiizY2)CDNuS(0bF#?Iq_I_dZZu0eUXD4giMTA^2L_eq@?8?Vb*X!6bSl; z^+F@=XWg$@eQbAwZynuxp;svi?_R%_P*GWRq2=H9?`>02G>+HkMd%nBzC--)Dg$ml z1r}s=i|pD3I9k__s+u{CI0Z7Pa3wzHZ0ff{xj|FZe0&m)gI;{Bg!$5<8z}O5?yiFI z@=GH6y1Ee^+z)9_NZkg!I;^ez7{XJ3f`%f)VM;|9SHOW9hYM+lN&xfjml|lS0y4*_ z$yY<n-mfm!Vkdm4{oyYi!0>$q9Oz;D8a_AfIrJJiLEmD z+tKLL9?0_p{(u_CCZ~y>;+WglgIJY zfZx3npR<#Ja$(gDr{B9*ig41#U1^G5G;7Al-fB38VgLlO&e{`x{d(p!-n9;>ZY*Zg zbZ-d?kGZZcZBHJoRhU_@Y!u!OtPcz6;adeEM^48IHV-^20Mww+AO=IB1PzQWJL|rM z@J9;^3f5x?>+ADFp#j=7I(qar*5qN((h}tFLUkzd7}J?t&d1cN_@@nYkV6h~AQUu8 zS=~wLr5WsCp5%{0bjjh-J*J%y7ssXS*NL|M;RDyMAXxV@j9VROV zwxJ=6!Ar1l8l=^Vg^*A9_$Mt*yJ{<$K{%Lz~1lS4@}6 z{s~4?+1+vUVReXV1B^kxBh)2~pFWK?XTF6&U%hu6#dCk*0J``&+&FE7lO+-e6{7fv^{Hh7&_HZeLwL zdLhP1JVbF`Vsj2P@FBUF{9mm&`Qr?ToIv)Hwr^n$ScQ`hwlwSxvb-(thxEfBN(in# zj9BQ7f=p01w~#WCyRo)DtUQc~4z~d}{OXHDjD~FmFntJgfq5Po9MT1py%Gms%-hk- zi8lN&nLJYvm?NvbrV`Zn?(*oCrBK{QK0c#gzC;|z0F%Xa>&=G`DcLVx{?Ej6eWvC& z?0Ix-I9rUQ%&e?@4JbKEiqSSFUF|WcocH#u7oa^__BehkE{YeLQ|0)y-!(fUr z2M5Oz3zPg-ks1hji}+j9Zf!k1K5cIwP3BsOL`Jc^Zg?4e z)k6^XX@fKO2k6u(C&Lr!Y<8b9SE*6TzqW%l0&bfUS#>4)RJp~UKjY7di^`wni)hHW zzz99$hlT`B9G<$84~NBFoO@SoAu9xe17C6Wc+b8>QC)pVLI&sfn3LuEhnxn=nJp>p3aohaXqSXEnxNzU20$; z)g1QRHMn&@eFH&;n+QYcIeD%Crjajw#_Sv%ZMXIasHmj2jsYvIX9Ih68p%A!$=Q?l zac%1FOH6Uzte<mh5T>B?MF(s@6z1n=d_$T9yyKxA?|OSH>%KG4Z(@zey6TL3 z1-OdcKTr!nEmIbmCnlP>yYE@Z8^_1g?puORi67wF21YjG28#;kX86*v^NyPLYe_fw z;YQAwboxD++hUS7n1y-DUHg%H5aAs&2wlg7)yaIvSy5UN}KVmok6_>}EE zkp+(Nb&%UqS#p2FrTY&-K5x-V8s8s7p;04Q*`x}SDVOy##^FYo6&JEmN$iLhx`Lqk?w?<)gR zT0Jo6(g{(zRq}YJiPOz>{}tT+{@)n0@N-}lAAk$u3FuBROWg<$cW`l0hCqd%S?H$U z{OIQH6s#VeZF%wf(&=j7eFK~=#YB`C)4jK(>{|eSm_6yD?I4Ql^qavefA!`=rl_Bh z^)~mp1nIQHBA$QY?GnD_i?}{fV6*2tLqk>LYlHTkZ#lAxtp#vNtq&8pX1MYQVr;-bMQ8wT;z=T^RVuLfeY9prd>J>k^h0Z3kI7vk3M63DcR3f zwW>=_;hLpto4(SQr`S)*+rq2D^~E{>D+fDi(Bgr13^qP{u4Y%rexd5WUk_x+FDOIE z_+1L}cx*Y64GQuV0HCC#WFHB^Krz9L?AO!Grh1a1qLb6_U3hMWC7ut2kRZ5@lnoES z>)=W$V@?BYIhdg)VJLGWD|}n__f!sHhzGaD~MI zpKCDJxW!CP79RxRcn8a}%~~6@jP^5$Q7N^Ho@FY+`=X$;U>DI)B?o8&lOmaL8=bQ3 z$Vj^5nLglO^LSNJ+KD9m`tvUD^oD+Cofnlz)P40*C*@lh?*%ntR%T{YMYabA8zdL@Op`1qshY9LYp76d^n6!z(B@LkpE5n9>D3##DoHT(2isWvn^cmyGf+| zdV1n}_P|SdPg-NQs3>KLDaHzQ=9_d$9La(`&s1cgci7dW%Ev}EF~I-=s)9yE`Z_*u z0z^Ze5GD=@Vc{c&hL}GLC>N}y{Bx3w6DW%TP)^X~zKgZIeq9srONx93?WXh0Ssw;% z>STc%MOqJBwpXgh=o?#D^!N5^eKX=-5H=^qp_2{ik4;bO(^6vx05jTr#MGhlq!8@F z$%|g+&*-}IBMqwGY_T22cMLp_-zqUCanuZM+Zd%Qi#MF zfhF1{`nYaaq59#;L8%N_p2%7I^(N?n`Zv&E-Nn`v z!2blz?J3OW*4As7=mGywaB)~r4q#I@CCerw284BLm+7&vSN7)1^An3LM+TyN&;$$SNuFPIG+B0iun3H<^ z@L&0X@OBscuG=m@N9J=ka)N-unDAs|y1Khh?PiB|qm#l=#pT>zq36+qpa&?-=1|xh z0TI8$0E&Kf^SR9#2M35$Fiz8LmnPD>L4pG>hp`NlD*zj|39)l>I-NOl3kK+IC|u@J zS8>)(TAjw55J`APbH}qKrd$YSOwx18V%h#X(WZ`M2y$U{=}CL@ZZ=`L7v_SdCgPW$ z*c}3xhEoPN5KnA#XM1}Igz%D*J#umsG7|pX)QYE1xo8jZ`L-)*0FAdbV(Q<4eIduh zMMbYoesG3!6&7fGdIao5)R07XLkA=q9c&Kk&WntWin2L%s{6x-6#yHlba`2PU;*08g9q$b!IGapi9Ul}E%=}Wv=rb^SZNVP zK=j(-Mn4G#hD<}SB~cbpUypJ}vQ0YR+^FBVk82LzCRmK)Sn=^23|I^?9XF~>t^VOo zbP0vXA96t=2oW?43{ZdW|Cry=z(^AcYIhCsP4LrOwr<5Xk_QLTuZ|fT%V76ObHET> zT41bET1-q9FVm|;y((d1W;U|2qJVbYYFkudlr|I)&@#A{PiA?r^gMX$D_kho7|t2o z8{89IV*FkDMie4JCb;@<-`>s12`gP3p08b9#ttj39gU!Rx#+@L$F{SxQ$DcgLeD;) zVtkAT)>MeGA{zr1dIV0fy|S{D&rPx`tDHWgr@=tnfEX<(YRYlLOGf5faFK2nG&o7g z!^Xy5O;Myf9v7qEVE&y8#H07Vudn5rOAqb{wl;$C6A~2k?8>TQ{l}69i2*q-E-ntj zLj;4%K!sQfpEXu41TlfA#4ZK&eV`hUU)}wznu`Az;x$u*PM~Eq*p`e$l=-pXe_i9j zhbn+&`1#`psEPEH@|sQKS+L-!uA8K;DgohnVS?Lv)G!KoqegoK6=xV1Kx?S764`%w@aHi;JtD79br}I zI;Y%gcpmFtOqE~!pXRZRhX>OAFjU%)6YHy3aUUEjXrYC)<7$Bm!#Erj6$LG%@qhoo z1U)1+HI=f4qgFkM^0^ln*qROK2y6QQ>VP{wPfqGEsV|(O*tY+^E=I0XgA>o3R<#mvdUSJ|2A$Mb-A--YkybVOWA$#d4 z$s&FIMZJ&s0B``QgOHHwlCdAHP!XRo-Ia~JkdR*X{(B)0x|Q1^Aw8FZ=0|3z zfx18Eo^Co%s*px_;zu^QPX51PTs6U{faP;5k|+co7j=N8H4vNSh(GcK2ne2*26&E} zK1j3ue`6k$0zm!W5$Uf51pVJ&pd|)t{ojZ|S0E(*zp)8Qf!q?*)y@9z<*1QOcnXB8 zFZw8P_*z;RcQS}@UhWo12T}jM{A}6mfy<9~(oIe4>QYhgKjUntP`O7!z6!(}8)m
    \n", - "\n", - ":::{figure} ../image/warmup_loss_plot.svg\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/cifar100_example_anomaly.png\n", ":::\n", "\n", "In this tutorial, we will use the CIFAR100 dataset. CIFAR100 has 600 images for 100 classes each with a resolution of 32x32, similar to CIFAR10. The larger amount of classes requires the model to attend to specific features in the images instead of coarse features as in CIFAR10, therefore making the task harder. We will show the model a set of 9 images of one class, and 1 image from another class. The task is to find the image that is from a different class than the other images.\n", diff --git a/open-machine-learning-jupyter-book/llm/image/attention_example.svg b/open-machine-learning-jupyter-book/llm/image/attention_example.svg deleted file mode 100644 index 45fd2897a..000000000 --- a/open-machine-learning-jupyter-book/llm/image/attention_example.svg +++ /dev/null @@ -1,9628 +0,0 @@ - - - - - - image/svg+xml - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - diff --git a/open-machine-learning-jupyter-book/llm/image/cifar100_example_anomaly.png b/open-machine-learning-jupyter-book/llm/image/cifar100_example_anomaly.png deleted file mode 100644 index 7e06e5a5d24157731d1976feca48d9bcb492e50c..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 155392 zcmcF~MNnPg((QqRySqd1;O_3ho!}na-QC?C4(<-Y-66O;!QGudckoZ{o4l%5uXcA2 zy1v?b@~!Ib)oVv6%1a``;lTj_07PjiF=YS%_|E~&Fi`)ZTHJI#0D$J|sjBI$Z0JT} z?`UUcX>Cg4>|t+8V(M;b1^~FPRcBZ_l5;19e2vh&!mJP=5kOb2Zsf;pZ^N@RvQ+*u z`#q)2B8`C?4cN%*`SzUc_c`P9*{3m9%s z_i^!JzV`GZqSSfxsBUNfjuR=jRdDNiAJ5_Q^L@U%7WBwJ`xHV3oO=1t_eCOGiDrHj~4^toB62lHj5=xAnf=HV(TZ}=(vy;R#b&d>e2 zl~L^NZLjCs=R?1sMzXJ;;iCOlL(bJqSI4H8KnH%Y^6aa)s=KeMOw{3dUH^gC5D~g~w*tMx2&x_5)ix%!Ed2?^k(~_P->3Fe{)eW9SJit$ zyW5Ss=@`8l=iF=$o{yih-nHx7XVoihx3L77WK>a|fBimuKWnkSlCF-niqVF!>mRBy z&tyaxCKQkEvx+RFI=8!5b#{oiw;#jw{foyK2ufHuIk@`YvyHUhfjTc2PKR9kSLU>P z`)^+0GK0-ALrmb`Lh%}h@PbMYn3YE)f}rRW+CkFLg<5O$EX*YA{JA)|HB(-E-6NE; z>?f9O5fV7_2lpjM!i_S_*Zai*1%|7U8EQ^eaW44m%gWVDr)m_Od|nQBrCxVb>w4&rto-hft4p!2DD#;1ULrehG5A`pJsbvc^XmE*Gs9`d^ao~Q zxl&G+Q^{)ktx9KFu2l)LhNx?DJTs_TYt8rm(Q&z(DzkKd;z-&iRbctJal_+}RI0OQ zN7Qx;i>%aHwz9b?m`zShn}D^6Ex=)^I0Bn`>NjLw^7y3cU;9s}Z z@4h(A7w4`5AvDJwl^K7&R?NgR#ZGx3@ zlqUFUnZ{(7^?54X-}XS&4ZR>kEr+U9#wB*K*zF1=^=>AJkRKBnp29WbhS;AC%R z%A_Y6=rUeR{00^Ib@C6D#2jz0?a24L1^FW`i3!)SR-V!lK9~vaf}nSnlIfv-_%|QP z%*>1Sl12ASLre?tHq&X9-g#iv2>gkV;k?2(waGM)b!A}$w=2DAIh}h zsHhED9RI4evHN`q&lRAen7An0d+1~oZ*l)M`neiK&#WzTm^$F7m;G672he$hPFaT4 ze!o}i&0s=dwO+hn7X5{m*qUejUfr^LvXX{}svTzB6i4j`L0;tw27B?)M~`nj{}>T- zq)f-k9BLy2Pzc2h-|>nXEjX5+bVawaF!Sh*b1;-h`E_N z4mXFwX1B-HTsmA74iR=GyiydA0&J+uU4QXj8AGaJ^_r>Mb2JP^;(Vy%MCBS_yJ<50 z@oOEAKpUJxb~iE`_n**m?4z3I4RfkkM>gT$a5S+>>mD$%7||jn5)1S>$w~7_zI^ zid_a=*rty5F5{K%>i@}1K&=ST);1u=)bfwyjdoIAF*US)^vDO`vK$(v-N%|w@-Of% zIOiW-eFH;xh2}9B)hiJm><8zuHI7IaZ;URKALIMG{L77IAcmqJoV{^Rc+0&6aV>ox zfPQ1rh6)*By-H%FNFTmGD@W&7&Nwk?z-kS0M#@tHc!)Y!�)u10{q&+5K^ouC_28!XT1-mR^-CVAP;2 z35;+{m`)5iJEP@6io%6K@zsDibDKMR!8xCbpfilF7l-EbM3fN8JVV&9zGc4Uvylw- zgoe@on#BC=b!Z2rtr>k}c0;cCkcz!DLR!ac@KaufRjFWNEq~PqxD177^CuPx+(dkj z<>Qj`P} z$+{h%^q^tSpp2jPNuBcaC_nsizzT67S$}eg?`Gnn4@vP+DL9o7ex<@<;&0L1j#e3c z=WdXfsgY~~q+!!@{xm0BRR6o71Deyi;AAB6+`X?d22YH(=wZy(j@13Pv& zg%sCMN1Jw3G(C)iO$-~XMN;3Enx`5qHAJ2^J&u3pQX;thX>AgV^hQ({)3Fjwg&iiQ zgs#l91QVFi zJO3kNdO^)5)l+FTw0C2+Xv~IG&4Y{?OlvI`HrRjs9c9rBXhFpG> zK$amf6P8hZCF*W2%qt^m3;eell5SHfJhgQ+%;SJ#|3m>~dR+i8%*Xi%5wK6_E^4j@ zGrL>mzu!A4M$L$iX=^V7>ivx0p*4V^J6e_{y@=kvjK?VFjRWhwasLCDC&FQNPraY4 z8i-}GRKK$4jH{~d8$HMwvbZ`EA>aoWL0QwWy+>k-zvB@mWHF^4Q;%Mg@+E{1@$oq@ zwIK7&y9Q43wigm5!jk4r2|uBbOBJ!1Ft1HuC|fVXN}F>|#t_!uAgq>X?s6*|4ny+X zcKIkuGKf>k*>l6(ho5DkR>GnuaYS}ds5=MnuEPfeTg_G?b?pI74&p8)a7PQ=i=YRE zn=)5D`A)+@aZ~Ip@DJv7I9SnG!V2F*ryI<44=1MZBSUG2dclbc!QCYBk-(~RFPL|jZ&BK_SwN@ut5wA!N_q-$SNg#mzZn$>{DV$ zjqcl|kkT4d-^QF9W=hB3cxl`*Pn>M>?<*K>JO0rt={!2|40Zk$T}anLNal;^9go3_ z(Je(&dytSZ>-%1h09Q&)C{VIr_tG5MX++{+a!NPNhlP5_&t zKMRF&?OGf6z%@0PtBh4uY+c}?%IBwB!V40 z9*AokNLK{OUP*-Y%y5Cv-Z9H?55bQLvu6NTeZ%xE#-`dMnxY2O<&IQ`Eum~Np&xkf zwP>*-bs>6Y^eH!N&MhR3DLYd!ZyhcbxaqDVcc{v~3QU<{JPAJi7;;$o(6Ck;FlFB8? zq#MtDB`M1ghIGA)Cs}1ihi3-8lBtCEv{E4=jCRLR3Ktz|fA5v+P0pEPC5fISeH@V zQB;V{c>aXZ(ugY&9tJkgMu=q_20}*-u%8G_uhh{>G~Yy%%j1xv(CiQ25;YvoR>yra z>?J^}93Nf4CAS7+=EG$Qv_H@5g5O@c`fg~cA=9QDJgcijKH_}iQr z1Qf4jOsD9&!>Iid=~07zgsp$(Nx5-|fv#9^)iU+|Gkzd* zz;|a3K+J%-A++dPov|D$I8K48vRInVPkUUI7sk;oS7;C zC8O7(L2^pTNk;7l_*0C#eR-Q1${T{v`*XFno>`YVSd$%>O9gzqs6{aPl^}{cDyrvI z!g@_<%)%>$F|7>VN>gms;nRHg_ZB3zvmH~DVsJ1js7hvfjqz?&6WLXJ2F{^Ur6&h) zGk+SRw4A>xhd{@K#m70aHtE2yIf5cPd&^+p-`u~mTirp{vBQLHeBW&-&rv29=D|Dk zHn7kk7|6c|eBC(Idi!mC+x9h&1ZsO@Z4AkbZC7^Z&Fh^{4~g6fal(wby;r)ObxCol z88q>DAg>0>!b+EPjNEbPw#Hj{A4hf9h*d2gk+0ZB>yhfHc(DL}mf*)oX{5-aT~C-2 znhZ$pgC0b(>~T7J=>r|;XAISL6lbUZR>9dWUY8^ert*0%R6wvO ziRVkTxEV>MiT8*>VIDr?b6^WOM)5S)W@dfHvNoFSLYN!q=M?Qoc&`6E)___-No$5f zGDs?h33wuptvlhqS#lqAfvt8i@o^6xJrNM7InV|6?8y6MZ|C`eT_fRKVHa~$0svqk zmZGAH(xRgO%Rm7D)CoQb{89r#1Vj2N#o}~?=Olaa1(D&9N~n_6BB->%+68LRNwR{$ zl!($%QAIo2_SeSsr#pL9l|hvxFuwj=+&nbDr6Rs$;l3O3ab1pdc*S+N`71IqPo3fb z*}9BM&>WbPbWwq|#}X)p(FU4ksmyfJ((bXfUeC{-*!x;rN1p`zzow1+6)18IL$yft zsB+h;aI&8w$g8L;2ag7eD0SnX6V4(h?6TpJ;Q3gMN>rQ%QH|>V49nm*7M7zbph>Fz zDD@`YZy5K8YE-XiIOKnU7v=Wxh&W?|j_(D>A{7n}q~%Z)fn?TpQKko32wv`9xxJ&c zV>sIQCm0i%TEr6yJ`=xRE{!WhcSsj-{dqB3Ck_%az_wVhetTPK!Kd>7!fl(=`E(>q z|D@s9Lv$Y&2vTq08^W!$q@FwP!#Lob)%7hkZj;_{R)F8OchD0lxn8L0>4KA2T9^hMoG!?N978mkbt4Fx!qT@sijCuv}ItKUD^_vsB5S_ zb-6+1bFP@EnI`tZAiMFp#u5T^_qqRX>@zXwG|KON#Bbr~#69-tzYJ$Y@MB(Z_U7u@ zL;L@Ho3Axrz=9Ls$bT8S6#svxoc=!?9Ql9x&U**Wi=>~h+D*^9!WZjSw=X0B#veR{ znnY3RO*-(G-=^Nl?nn2XQv~%VtQa6%=yOd!U?49p4-p{q;Dkx?=B@SsR4St#{XDsr zGY~`)`RIH`Z1N!hBdHm&iF;lXgo6`m>IqdAB*yD`d;Wf^kg&$cx<12%p6;Iwb2JgX zJQ0{S5uG*>IByA+3`Q|9lUa8aIHSLP{r3#NyR&QmE{zig;*7^w&z!y#c7jJVLNT&1 z+<-BXEYb%3yyLmiXdBXLJl&~3lWKBRFX>07>xJ5f4@e*?CurC6!R}4}tMv=hbTr@)#s97YwAc1{mHL7(;3%1lG`Y*LliRD4lzW;(4Y zOlU#RioZ2XTIEYmEi40gag-h|oZTz;nO&EwPG5>ydKdZ7wC60af9p6rF#*X~i{06o z?I4iz`f>QQ^Ke1v_j#9h!}xW}_?+hVg8lxI_Z3~M-?Qe1wB2=R7mh22ke_>YJ`yjS z911D2o`G8S`f{^;;~T!6xAtREUg`_B$L?G@%&$kE#rZjh=kxRGi}fA(Yv1on_wLs9 zE%NK*>scSUQtwMWPF?Kwi+QoRrkfIFo%O9>sJm`~9o0qPrN3y<{|gDAsLO=pru(yJ zxM$&-E!WDY3F_4fPkp!Bw!!>*ZM*!@MZNih_pN#ez`}aSXE?sPl$y-^H`=nH+3Z$Q zH*Y_;J}N(av!{dhQQ027PaUpB*`5wEt5$r7xC17VR_uZY_Vvsc5?Lb}$^4G1nJs(d z_fgs7$B6#AXMyW>*~&K1^~)lU(?K>`mmJ0U!N<*|67Cd&Wa3lLd|@7ivdhM2WgH-Wg` zZM+Wm*k!N||5z<#cotHP3o90#43q6Y2{HZ1-C__fN%^Z$jtYgv9~Oj5vJD&~u5-!d z4xk}bYG5B-zIgltzDO8J*o$rdGz>cV)rDP01gIbeI!N$mK{GAooz1X zA^f;Z%PV&NCR4x6K@~Nn_1$&rtQsX3j3lK;GSTn3#^WWg<}3Q^?(FNZBV@`YG*G$^ zLnHTfyKN+4L`4~*0e;+erh%?0oZl5JvdO3ElWHvwe(=gCPDQn#P+$s!M8pV+*<+(O zJvG`ghq;>da6Z^FqnA4XEdVX(Lu+t*AWs|+{4wV%lmH`zKmnfaJ-%cxx9qJJH?(Mc z!_hN(iS75}+C%u<2?pgm%6IC|D(dd;vf6z(%Hd@512#h$bjQr^ww7Myn{F=pMr0_B z7fKJWiW_5Z3@gjZq1B%p*bzABKhl*Y4x=Nem`05xXgI=TZ>-2sdbA7YVxO~#H7Y-o04%T;3=)sJ^E7ND7S0^f89o8*4YJKd>}5@>R*38AnFVH1(XR*1d_3; zVgq=Eg1d0SD`wUyca`d{BEOp^(}>Vs^khl0l7v@<&I6D$d{BW+u8Y zI`wNWcv6)WAB|>^azV*ZpnsmLEc2LMid0DOlgY^|?op5IVy#JB7vjy;1;xdMqs~zkojqg_B~~aBNvO%(s~cWQ3DJnx+HPh|bfbH^A5F6g zi>W0gzsnRn!e1?3H9mh7!h}o?Vp1+0jkk_CK$WcT5*ZEtT?{@L4$hSc4hCEV0nTUZ z^&v{=lxzLems2bz?%I4b|HKfe_;5TDkX(%mm%uXl@S2a3g6tn@0Ym5KWNQ?5d>Hm! z5rg{T5g{NRI+42Nundd_9vqeytE`?naZO8ACU^DDL#yu#0HT+oum;!wwgK%TCs2J* z!ostJ9p@_`cY~BQR;555Me?j`UpoBrYJ+j*O2|KBHT97fewcFQ)xucF&kxmOurNZd zMCaZZ)UMDft*)1mw$z^Dv-qxE!fNr{Umrmth`Jk(%Xb9jA2x4|qIGJCkFn>7B)j4A2ibkJq4BeT%T^j4f zX~m5TqEF`L*TdEK3wCxCQ;R}MPM8STOq}sOK@D&VTBwRqea>UAXNqQnEwkp}hxyCg z9v^P|z=##aOq&da5z&$Y}0m)$)+)sG%#+2cBF!B3ml40xCKZ?Ca;lAag^@55*!Za1*AGxR8u zOYd>%S@*}AJs!G#Z<}A~Ax*VzYc#JoDj%`%U;W}^(B0P+hScmP#Pj16Mo{?vwo=jP36*VOyvluV7Fa1& zz(GC@D|u;f_IpPA7}KvMMb+PIz}zkvL%H{9jE2NuX51|-b_G7RawBuc?@?!d(-{e@ z#+hg(lj%pn3bk~;FV6jpdFt=Cp8bCt0OPsU4G0St>b+wPkmB%LcAV$M{^%OfP&7uP zldR-peQE9-4axBzZqhO-IheDQQzxAA@qN!f#6ZFgx0;2I{Kg$N?r*n4tlB&>Z(`Xo zj+S{VXF2G~Yd_{%&68f*H)N_Fn);C#4rNVBoCM)g#OB~AL`%Kj`L4Ma*9 z)Q4z6^Fav@-4Se9J03%+@E>V>D>CCU(YAn)jR|oqha6fyo>R7^H%87Sw#i=zG&zwo zj8`c0CDB@_a}BuEr0qwc3kgvyP)$MZt$|aAjPjSU+=jK*QluTyYW01mT&G*fvDy4N z?grn`owdELv_+L$Or%*({)K+NqR=N!X;^QYVA(t*cnSBm5 zw7FUrGb@Wct{r%%sMRoG=Y&&Xw0*2TI^6g5kf;k6M`$er+cFB4|Zk2{>`Atw`T z=3$?Zkr2VG0P8fZY34P_?G>-1IDRSI0(n{7tG`aUpdpAyNkC}oT_9A!&?<*>)+_ZR zE~+AzWYVd>!K7*;*khWj8vQ>gs_n_WF4doYewfE)WQ133w{fnBemX;nw=vUMy?*#0 zTp0xg)cAcT`)(=3M3N%($GVAIrPPmoR2jiOeA&VIu+p5GE}4C z>r;PphnqI>pVZk&u&I_rL?pP*^yt_O>BCw%?N`Ule#{BN#evhM&P{6Np`q zk;2%D zT6^&CDL{1J6gM7ZUXmlsa841l7R18D*!-YIP^Ecf$h+}Jr$woqKVd7#adOQ!Y-l1Z zv_8Xd%TEFUjf3%MW)V9ssnsg>drPL2pt$NMwpNybJk$tfMwHNDU~&OMC`_>7yd?E& zOAIp{?GDgtaJQb!mOB#CuWRUcIBJ^(>=Lj5nI8g}UzG@TBGr_X8-j$eCm$PlkMgqq zXb=8*WcjK9BViikL{2twxF%vc5lHE{l);)(O%cgvXOwSxvaA~$Y=9x&xpQdnr3%mq zX-ktQBOH4;9>r)yz|%bRy+{1)uPROkN2;?V?N?M;A#Hh4=)aMBb&Mg4iNe)H5xP)P zNF@trFaH!1o}~ySk1uS2Gz7gF+dA})(DaMN^A{@xi0253pi~@=V)RUBJwCHbJs34T z58|NkIUqL#VkFqxJD~8rdlpa@{_0&D9I|LSwwr{!GyUb0j{UHu)Fny!MHH|w5~2M{ zNv>uE6i9A+SIl+-m6=L?L!m*_LA5d)PR1$q=(o&q5SZesW|f~XF|b^>U*|l33$BPV4QU7IF?QQv)gAZCD2um589} zkOZ^@>KDVf(n<{~%8SNmZBmb%t<;8`N($|)@4I&Fou$oOJR$hsdj^#WhJNV+WvhwC z!Bi{obSwW_!rQuEa+6%(E3AHxIx?VTsOdrq6WTi>NzY|7CHcLe*aQW2JX8Y%mf;E| z>Zjr?J>cE=`**NqZhaN&()>Sc!T-b7_3H$ko?zGB!#q{CJl&GZ&`Hx?Q%(9&yXByj z>kLk{wpFU1oiX)AH>zrh@CV<2`u$8zp&xub_;oGa|7eJfexTfV*cE2`5L8VyJPH~LC*q`?Co z>B*ZayCZ6x(ys;`BnkAXX!_r=G%*JB)D&zK(xX;Mr=F9uI|YF>^6{_ysyr}HnNQj( zY)t$Q(>KNiaC7PFuYD$Cu?5P?yloNR5L~)g42t<})Y0GLp;X#31YB6@Vaxa%4rrLv z3ZI1@i63d*zuV=Ais2w?E}-$|4kieBt15=KnQ=5v&%RSN^oJe%?fF4cJjJ)0Ki_6v zxE6SG6Em!oxEJR&UNa5PBh`T{f_^KlP!y!$^rCEO%Na2Sa-=cTMVl@i%f6nbn?AxV zK3}ngU5=^WglWaM@1ohN9b#zHEE(Ofub}9kt$ze0gz`Me->5_P!-ym<3o+9uH;8na z%KxiwdJOOkX_$EeIoED}zjp-RPawZ`6&Rm81aodq=5Z^6$skZgl=_!j@om>bTc7Cu zae&5pl-9*fdoC`bKq%60S=SLUL-C*$^6-izO^E)~P16%Jfk>@e(O$^~!Udz9+kxnX z7Dx%WOHcMKgZNJbMyzi}iZXg!c_VTgp*1>TEJHz7QUSse;8mfOKu3-`?b(-oYgL!9X_s&1`dB6TeUo!1$ zpifW3J!>!+$JGX>BMnSub##3i821%p!W7UK6_USH6*oX`z!77~tmTV;d;p(|gS>LH zf*@jbkHc^yo*?rR&;mUpTIw681kOUtD^~eIDD&52^r}7o(K|C+1_hQStXm{tQUoBtJM)>WtLRpkfTif{8?d>mxcfJ(6L_h#>mahW-FKx!e;b`#Rb=_-6aaUJ!>k;m zwlG^)l4iV%@=u8`nOm0D*s4lz*0%bk*7U>e8Br@SyLj#b0$3rr9pCC*_-! z>Jzk31td6_+-`y*wSQS0rT~8ozghR$!BsvjGbz<@Hm!GB&VV84&_csVe+^pz@@hvWI9F<-nM)pC0K^z7;Pbnoi- zbnZH5)?W;nI_Q=_*R8Igky43FM=N7+J5*-i^(EWH{lt-iORN}j@qmJj#Uda1@n{~n z^x0Bf#tM@r`n#2NM(w|$U)iD-V$~5TIHIG z3WTa+tAhbu1V0#uwP~WyJ>R~WNle4uu^94KbLq?5I(-Hzt0Vi> zS7#SFRi8{C#U5eps4*6(43otxd9lE^gCJ!&BQL=iR6xQ;iELc58uS+}i5+L0e`9IY zuaaf@Q@5p8%4X>2PNa61TELom!nl*au%` zl#W)>ksRY<_2pJ*`5ZwOJ1cu@<`OM!c-O|!rH=xNCHmx!dl;MAQde*B;(*djQ^Cs) ztOd-43^L5JwsmagARV*cD96iIrD}KjX?Zzo<9^$zs~ZPEn5sQLhYsw#N@$t^K4i)z&!rGV$E z;u-=8xDl+hH%pT&(u$$B@Tbhw+EtZ?*hHJtqeqHTk8J`5@q)2K=0!DSs~6C&>`T-E znO1hq?MnfQ1*m`}t0Ifo;1P7JZ>e}kS;LjH@U2oEfIB;6P+?XiRC(eYN+-KFH^%so zPhK}HXfKa01=e)C?t*t~ zyVl*SXYb-8^yv)q>MCnd4)YdT5iWD9^MOgxotYkwNqUkp&G98>tORDYPjwm!q^G|O zm9M9-;r;47d}i9mLd67Ty6)iBAri6KkO(G2sXrO*lHtI^=*L(r_X3HYJDmzi9 zSsQuY1;JrmHCp~$FRjDGCiNbUquhwDHXn@?yU31(3ByD^ddP&8<{P)wa}Le(J{!0U z#I#k0OS^MmQQYYzF->|?gI{duSPgzkVexWGOB_US>f$cEXIT16_(i}G3R5VtL?%=| z&@X;ZZCOx2HkO8P8THDiz?Gw(J)@3pim*VZr>F1G-nrx5ZJ;}o?r3{A69^#U6GaHS zwXpENawx&8)KS}a6)x~jAlm4_AOQlDu%F9H!Hnyw%vRsQI_D0A9*S;4wnirVl8XcD^ z#KNdcZ4m+|H#1(lQ>mOZ`|qM$Ju zCyzgW$ezx(R->eh9!(1<{itZH&GWv>^Cv-7BSi3SYIyI~Qw1_&-Qr-AxO~SB;SsM+ zoDtA@LG~AZc>*gNaHjNL={Y%87Y&c0plc384p^vnQ-x5m>Eq!5aOjR|ssy_0TeL}g zb%(BMrg>!AYnx@KL8i?|HUi7;%iMx8BXGXK3K{qwm1OFUWy)O2zVDr@@ox zC49ng@lKG+kTf1QoQu!>bI@YoW;Tu)E$K@{Vv`2n!r{_&;RX${Zk`ys*A58UfRwYU z6dVRM8L0BsLjV2*I-r-!r2-`@buu4x|0B8r8?II+Vtt^UThF@?qAq-Hj(rB^ite2! zk!DE^N;*uTw(UXuZKWZAq6V8gK0{sX&YukKANP^^U-xtZAGc9p<=Ctln#(H2ED~;< zwhh6*x#ole;s1mwlk;4P6f6(eZzd04p*7ZHXdSKyo&s0U{n0n8O(Sc*y}?|+cfJf7 zD~vPay11DH6@ZB-#2R-%AV!qG8BLfOU&ScVg^yeBSUS4WvIgbwQ*s!Gi4AL;$+sXC zA2d_XMXS;W>>3OOEavmseJIX4fl6T_ zkD|g}5K2F=s&XzqmEN2lcY3CNNK^_MRrtCUVD-z(z#E-rN;XFzX2%#I@q_*lieB~u zfuwkZCs0Qtdrmq60OgIP>r{-tskLsHlC^%#gPL}Vn`nCro9~_;LdV2^4I;g3OOKC_ zMpR?+qwDYnE5XEO3)JC|`U^Mp^oyHj>A~b)#7os1MGgI&Fy&k_C`4{vYctRBe%bsY zlu<=dN~??*ZJQcRe;3zb+4m6*B4}fjd>5so3~`KBuk>i$o*iWpGbTlTWxsbgC#m3| zh!vt89fd`4MDE8(A!bYG@@<_*OtIJgc-9aAq# z7b0AGkeLwK*iYj1_`6+w0LuTt4iXegLMtO**mUK;h8#*6mJ}sw@W34J>C9IV)Noyc zjT|ffB}&JHFo}WQem?YXYAc02a_RHPaL_v%mie)&6^#J0W_x+N2z`3O->SWG@4%yT z>A>vNhg{a_yU@1pP?K400GXy8Rw--Yl93z*o9fj~q91xVL}23kj|=6(sBnb$Xr6 z^Qr-7X(9Q*WLFcn_W8QHMJOkc7999SY+)ks)Pto{nSw?x=$Cb508KSR#8bhQfgv8d zF`|Ly5{tQ0U1KImCE_@@@ittG&4BBAooQuEQB3ASnIbNh3&Ck(Ea5LHOyU>xQ_=m^0ER~Pv_ncZu$HP0rmCefpJ!k zhP6}ceC$|)qzisUR|^SDMo_jGw9a~c-_*X-iOQNV2t8uRoykCK2)hTG4dz8TE4L zhcsCT_L=8%i#rr`0>$o*eZq?_>kjK2Kke{qI(J=-H%#1QW%_*4kGorG<0vP>9}G9% zHfbuQG1RI*L~gayL8If;lSOntA(SgbcU1Yr?1UnTAlffCN{Snl=T7S=7L3Z*7dQ#y zS-tp;WxjKDR1(3mEAyn!@ljV39F`DiuH+fZAxepdPU8Ox#ZpA9;uIP~p_N+0kC-F5 zGi1EIl0c{quz=aZKN~H$0JD0o`L5QYSkQV z8;lk5kk4uoK4)Y1F8ve#gNHR_jv|O%saVNZZHysux`RBv0x3{z4ySJkMZ@bmwugEsd_L3~=cYv`4me;qi)D9E`Quu)dzgJ!MM+Zwmoec~K90jaL$DV|rW#u} zYZDiDCdW?ANX}L}X$25j$2jrZ05@;CjBtpM!E@#*nxN%|vY>1|CWJBe(C87rq+3;hCSnMVXX)uWf% z6>*l%S(f~Pd&allL?6^-mqV3(2s@F;JcT=-$vlH=X|!PrmAh4Q6_exXw|(>2vU~A+ z|H9{A4MVH9*{;6|^r4GMh;%(t`L{Wss+tIN(XD5IKm50rGt7ihfhx(cZ=Y=YY1Y|Y z^)F`2qaU;s{j^)D#cw%7+JrArWAYeGb%aV_wwqwdXtHY*D!?!<-14yc97stxofF`~ zPpL*mZ?rNm|K0=wCk?O&Y%XfpscJ7-xLhviRZ-)&E7lviBNvj%1TIP%cmWpI2wTH( zX{ZipjQ`28&Z6f%`x_Y2k8sy*MlL+>&JUL;qRpIUkZ-B(9qOkCIu~bW4};D(+(zUI z?D-cK>#{;X*>Y>G3~fKB-F}+{#`ib%fXq7Yv;bK_@zD+s4 zN1FatG-6~sMH}_X{g(m@g&VMQKB;`9j6ZLGPm4lMyg)v$Dgl+By)*o(-1By@Bx+F9MK)uOmOCG~4Q(RN5haQjJ=KS-Or;hDLRVvN6J$iQT zNBAr?qXzb*2GC+r4|(g5o^+Q#OSa0l8VV*6Di{hPz6)UrtrGlGl>5dEQQnm77x4>J zpyzemu$v1Yvr&K++8`-f82(ZCza}%5dsKa;`SuCD`~ups=?Jdwf`)wcBb}*#LK8Kr zAcvez%Y9U-2>J|4)h|DCG_->ATZT%bwP7UyK%gg{A{h^93-Bqn1>isFc7C5w>$Z7L zHO+4OY1l%LZ1ioundGIls*TldiAW0UT!Bj<(&N5%Kq8h~LDO=xdobPKDNzQS<{uV2l2zBHrR%X>0wvBSkeNDWdvFoYhux{(# zvvvL0y?eHfsK0&w;<=f0i8JVUo(1{aut)S#5Dk+EM5GKr-R)M0Yn0c>feqV^IL2(LG^ z#cWY$wVHUod!PSD`h2h)Ec!PJ(=)0jcxO6lDe7O%s)wMH03?h=g7i`(P}TPU!bYZ` zRX`^Sr%JAGl3S(hS;~QO=^8QZ+VGpF$sxBM^61#KT}lJRjI~$H!R49c%?zm+D{#J^ z#>MJ!nwkfpX#eL>L4ku1&-C0$Q0VNU%bEE;pk`BBl)l#1S;OjEvU43iWw@*(=*Le> z9zi!p^gs@d3Tzd1Np`pglR-17g*Z`Yl11dI+i+B=;-cWA0>AQJV;uNc{9B@m~~WkHdgZ!#m3m3 znSzoBy(Wn&TI=CL4>}GaVeKtQMASPGbT+KI4y$&1xhjcd^gc}N4Y5IV@Lv15SxOJ~ zPra8ndHqklG+rm^^5Mon1&N=@{1j63fipMtq{86Phr1qmj~?IN3(#9k)b+)iUXP5_ zpwI3lFqAxb;IBFPo3dY0b|BQDD;+qWc(CM+s8f@F+FYq~vbLnDwM7F1I996l*V-Ec zH(_6^wYRmgdPyFoeRP+O4vhEE`XP8Ngt)-!_INaW@z|A=0nK2kujzp^<6j9#X6t;=)OZ-!lIs~tQL&~!jva)V&o}fo_DK9IqkSyH{7nBi7LfvZ=^&E zc3U`e&RCZf>tezEvP@WJHc8*am}Uc1>nUShCoF5mG&9oVQWF?Mv+pKJOWA6DYQGXg z=@j}N^vpwVKwHeesMbJKIbyY;4Gf1ufW6j2jTx?5+$iDJsgi)f-Hpb+LqAd@b+^DO zqKJE3bbS}+WOZoEvS2-(a5|lFIvaqhua|j2S$d6B*$ZyB4PU>0#n&&N@%8JBX`XPq zy`$`#)rsso%Dz=&+Jn?oA23&s3JCPWf1%jOP9IBJHCQjF%CNf4`LeEfdAWF_U3;8P zE0$%!G|gtX!L35Pv3;fNDEmG*A=W^;X8XoWx_)PW8y#w&@E|-SI)WdvI<01btb_T+ zhj?x{y$||)eV9J{9big6X44E0bKI5Cl4O1%y|?IKIo)bwG&Z`X@l8U*&1T?Hn>*Lq zCIA}&U_lPTC5095?rOH-MUB;|4uq|ZM77RkMnbmiHz&(MassD}XbXBSI4;G+n&qExOUzSiVENShdwwo zk}Tb8!1pIT*kCDwgi}ALKl+{|s0PkgTw3S$-rDOHTQIK&6MGNBC?--VB*kj2iXSuk z&sV7*QJ3zTDx-p?(bBl}bkZR0yIVi0fuI^(4Q3J_rL9!Q`bmm)rP}MKq7^zUTi-3f zV~?7+5c}~{MW=^Mdj*mVQ0i{3#Bx^_(+cMUz^M|%psX8uHKIDtB{JtZre^#B);`}t z?sSyG=g+STZj^u_L%qb|h<%f3iI{-CE?Z&I#EZm#?OrT*;vPW-l ze{{n2g;sGk$ae5zSXHx16)D8Cze$58KWa44vw!68W>TNWP|y9b51$6qHMd-Dv1!Wgq_ZNcPFL*Npx zIZs$->z$LmmgjaWD7Ota7Hr#sZ7;zxSCKM-Gg)t*Gyx3k`;PN%$L3CW+ve_XyXkYw zCe2+&wKir{3x4Mr^PDj+R-Z6ABb8^y^DeO;#&(zFcs*ycnQ58zZ%zPgs{V=+7<4+qMm#KYzyOe_gSD{)~B^ z@cuU8dVR<3dd2k}P}!j+aYZm&wd2|uNj@@Nga{`%x=`fy+>He z$n%V8DLC!s==Zk1UO>1?du!9zQ5zE5~zq< zzSnHGn=Pc|K+OujyFI!yiPWvHp&xjhaIDSX@Fe%1A|E6lz6(f^^lPSKPF}vJwal^} zg8${Au%^5M4}@h z9X$D~0Y(4L;!K=0R;nH!1pwC)%g3h4$JQQ^y>tt9u9oGFzbVx3Hv4RE$AQ`q+k0O1 zenj0#6`&5F5-f>OO?x|O*PB99?~51{-$%QXsOr!f%tKJXNMbTd0V8?4Uv4($-h*tGprF@|VO3FQ;8uuuS%p7$PWf)2t*dP1KMLlHNY%~iko9egWTWl5EpAG&H&Mtu(+6!}4w+#(GEonHt{K|I z^nCc8xNz1W&;nC#S;|_7pOd+_&C`r|UaZl2xmiZCY(UxEBtm;JH#6pw0jQiZ<~3uQ zQ)6>P%+0D4lZZ$KQH^@`B4+PZ#7vl{j5H@qQ?j7H0Vp;wH4*Y`(mD$WJUf`2DUcnz z1F30V{Btb}SnVP2pWTduWs2 z%!wPA`Vna#-PINyGmZ$_d~a}^67oD_Syr4*CtNNUoX#f)Q|HFskT<=n8sD<-Sl0#f zJYkv>I03mzkQ`WS8wLCSkumx+1!tuNMPR=L^m!bLgAq1)L@%2U1;fC#`wszV9RZR~wnO!rpPSiC~KEYACx4QYY6! zrTbkI???4#-3qvwLO!w{rbEzqXy%B>aECzaj{sB}vxYIz;~PWG9&#%^qv70{lILKT z(Hdz|D^9P{i_S3PDCBRwku3PJAc53sfD(5%C1nB&2d6t@r3O3TA)t2|o~OK%&w<5j z6-p^smc^1&^E*~Iv9RwhDBFg25XiI1Wq0SlvV)jVX6t39MWA!G>~2mTr#BZvbB9OhtAntI@fu|Ixm=~1?w^+dE0}?O1;0o<6p_L?YiwK*UkFRh}D)Bw^}Dv zBxF#kJ%tE|&cpUBy>*MlL*xN_$d-!qIuv)KsuU%W-ibVa2uN|y{8Q}xq~1iALy!#} zoO!!jFNXkbG*qdE1*#TYiFbxl^m^dhjLC?b-c%EnfvSHFCL&$E(V1C|yb$RSjP>PU zRJgRnLbYwCLtB)V{dd4q!IU#&M4{H!Wrh@8YZ{QOa^;ZFt?$jDkt$oe3s>)o2?%L`?fA9Ex(!La?CpxS8Gf<*~U zvOWlWj~U%ZEYx?@G7O7AT*NSX_A!NKl_@pL@p+!H%nL}(we5B**xuia15Rb&zLMB*1TxwV<0** zGgT5qi9soWb24|OFrWVL5eo5B%3dS3;(WGdw5QV73;cM4OO06)~1w#WnC~$ z8Pk+P(_~x3x7!t0w!UF^VCvfyP#*?F)a$6-&bkNJ8rckKJ^PPtw)f2H*PJwFCGf{l5LYQAwUC}#Qs z6LIGR_&@>l2oyam>cw|BhF zE8dqkye*5l00P)=8*cAcYv)-Ma$?MuN#fl|2sG zd{-P;{j0fhu8TJaqo^6Nh|7Q3 zqpT#|z@oZ3v}V37=n>p?DUNa=?lmX1SW2T=!#7P*erqlCT8OITAgfg)l9uJQ8YZ-Q z&|r2w0@q6GRLx9KH%Kk{As2VI#;VnYu1m11#0*-Xx;vN}<-(7peFz9ayDX0mu7aQg zGu0cPc9t&4`bgTdVzgn`)*|&zYY-7Y2P_9>s(luB;AYIOI56@SA-?7(N+6n`#1PsI zfRYZicu^&c=9EpKOr4F2$$E_dsj`4xNiN-Xqq-fppL2&@4(1XG450*}?8z7nvNA#X zoorNhxx}5sw6y)8y7U!~tBNN912grYUt6tlZncME_o|TAvd4ju>NU{^>vA2{PD8Il z?fuxK*m@zif^sufinAKt1-a?ZXiCyu9)speJE&^4{vb$`advYTV~*I`T2R776DFE4 zrx_DZnDdOpj7@J~B_w5VVHDb14@ooPNul1>lw1`^vDYyUvOJ}pw5)7Yb*lFL`?sWh z9L^&Jm6O#OcSfq4wp3#pZkp+;DAFC!qFqM)w*XSdX-Yo$y|f*rWQCjafWoszusGonovsuLCK61{5iKJz=(*!aOD9nUR-dOw^pZm8;hjl~TZ| z_Q+85?(E&@p=Y;353jiKsUM3JfNC$6&Pvrz+D1LwO(~)%qFtgF^yz~O)mT30fm{{X zT3tnx_x*@v>UaKVqqyCvuhvLebJ`2m(+P_MsAX9(ud6eYRtq&xGeo2pLRwc!uiF)< z6ztm#x9hv#+@-Ox_892hZ$|sw!Hv9GwCm3kH=_VVt+(_BrS-E^)!8VwJHMQpr0;UE z+NgD1F)s^NokDBvXCnY6PKMhDpllIj>y-O&*cTJ(#v>D%jJ-_)G-{p5PHU<8H zRMPi-G11ZTI(%wvU}!y(9CR_8q_0UhLOlY7785e4QG9cum}Cybyf*Uy6@_qV+4L$A zYPPU!n@{4(NbT+Z2X)KX%GAP}S#z>S;?#8awQm^#R)>_#@MNAcGBczU+^#q5w;SHS zE-2gPh2kbPY%<|epoM_Kkllb$Cuz*w0N~!Yh&K%S- zeD%VS8sH{payiJht)9i@RG7e+!J4hE#DPt5wmVAuDDJ6b+P5*+686ENV^C z*Bhw|Gn8!}S^vNZ6Hl0U#>^9DUN9#EREZOkSbtAeOLB|sZf3!)F0wIF+`$w`iHaPF*AOJ~3K~%OxP#RlDNGA%k zP;997Hvm$xlW~|hTEsRqQrl>?+E>#ztX#i8_IaG6{>Wxp9~O3xz-uhP?jWM~36YR} zs5aRafP3RbNMn7j*DJ2qg!!6qeP_%IW17s%zk(=c@ocIuVB!)Av8Jvh0@4Du*3yQh zgAqo~#!_W>?aNC~F-hw{LEy>2*0g5K=LuymSWXk>b;7hvnC1-f8iQKESK7i;>s+K| z2dic1F4km|l3QImBLxO>@g~EGkvR6Sh>N1OVs`@_3>0hNNYyeO3|0K$g9b}zm+EnE z!aIBasHxQmrPf!W7Wy(LaL!-=%X-53d}~y)F;X?jEP(8IzrK6(-hxuLwy?JzBe?fR zQH+ja%vCySE{jvn)+9AqfT}gaR=c=Y&=We*pVtMK^NN@A2``s3USADF8GxGI7C02a z`E_2TJ|qUDSqUB|LcrHfw>-&Dvwwv{bN4&;4=D%7^4P+yCgzAyDzFk zzdfaHnt&#g@L(^Y*R-g4j*%O^RNx zYM`ltDCX?X+!{5rJGSL1VV1eiYbH0cu;dSC!s)!W+NrmpUGAryI|j)jlJ#0 zgn`Y?a9K?%4T7l4`;vp%q9`|$a?`fEd^vfth=UY9)cp;;$vT zWORRPn44i~Ol4`n&iRGfnA8m<5*xQrc?b$_V#Fp#CwH=|0eO{<+HrDk%2^%7T4V3N z7ui0Md`}K03^*pD^kCp1=(gyBd*hx8o+20gC;0e zCnLpd{<#OIEjh#DPJkK>(L$%{dI~Fu+Fg3tec8m%J6FFpa(0sN18SlXO!be}FyXbk zV1~_86-^amvC45TLQq1*8ov>g>2JiP^bwJOuGSn9XmZ$7sLQME5&Cn9$I)IlE6!q( zn&jD93owubn!LZAs{LJb_H#KuIMpE1l1MybrWsROG4qU>7po0TGp00Q=D8(mNvQSL z6x!{|HjD9|pkjhEQtYz6-7Lq|8PnR}xF*o61E4!ks>(pDj2w$MJ4at1A*J_wtNWy? zyYLk^m0)W&tPGN5fJ#i`sI|$Kd{l1nSCsZ?Fn_2;cxLy}n|cvfKN-67$A5@_?zKAy zCSLEFY7TW1hZ?9mn_wR_rZT2w0z5-o*&h)reyq$w-_T$Tm54J^yJeZ`4eEMX5k)S7lI+1;Mzv9JDYHhN z)QXO6k?`icqE;Xa{1BkJQ-lr)K^-^F)||FB;hiuq3+8pjG|$M>1Ww6y?Nu&hflx{r z#DwLvLc{{VY9rsAG*tj8nBFpOx2wxS2^-zqGWxz2S%*!bOEpuX*8gd4wH+1^CeP@% zj_sv%U=_gB>9pc}I^l9LyYtuA7rb1~IGs+I=NUPtUdIs|3m!n!Adr==3AcV)rPWH^ zJNU))#Yb85gMDYC-RZzgK4P!+upa0`GIKb`gRD;;Bwjw+T7M#gB?@}LfYDujXHUMH zY?v*n!re-nf>eWi-6aXmaH^)UM>h+Q2Bf%&l8C7F@=*2`qYk2k6(Zbfe)eK!IXP{; z#=_T*#zLjsdi7Km%4A9Yb0FktG9Z==ROKmKK$+l%IT@GB8J|CY!Iv*z@cFZ~oY?m* zDhfTxWYRtZQ4Dsl!D{OXCvQ4Vj5+1j*nC|V2T=Z4jfF~7Szfkmr+6L1W&mp6Hk8{2 z6a^Dw&J)hdYRUrZF^p{&>|_$iQWT}^wrAIUBWIH^Co|5d>kw6UwQ7yth)46aVkWOO zDQ)kfz*I+*1pE+?6702*esI|2CW&jwCzl$bvNi#b7>UhhnvlHTR;eZ~#?sq+MX60s z-XNTZn7~xEpEYSnz|#+_^Ex%I#|t9KQX9xQrObJpw?R+Fy<4pZH>8Jg`O8KfD|1QoU|MhjF5r1iRT z5XG_0i?LQD#qWNyuZ7;|y2`^^%~su$s^Lj&H0zD33{1HVw-SKV$|bZWXov@^|L+Z2 zdutS5J~@~`1owIkAgbvJ&NGy=+|qE zl&g~b=TKEjr0(KeJdscx&#ehLF++0(iy|eDPhAlJt`3^6R^cAA3X(BTykJf%W?rzQ z)m=(wdrlJ))ph}dqzOqf3hjf?PT|RYVkAb3QEeAVAT0o85Q^21I@A_bx8rY`cLy{y zN^od|sBO=yQfKM%VIi@5(jA&AP}2&cgaeRr0^1bKO=|tO15&YwzXnRwQyD^|ZG;Bq+Nr*F2Y-tIQ!X|aqpf;;Artldaft7Xhw@vhr3;QVPNvcvYB5F(w&|ga}Lm{v7(C(1QuuVs6eSz0L7$e2l%Wm#X zdBav1RIJ~UdJ@E#qTJTSR+P+c9z)fHg^(zN7{(Au!89cc!V|SQeLkIhj%Q3cAv18j z-P{$jH0d#FNxK?5Wk8C+Wa|g1g?5=4OUgJ+vrBl5iMpK6IIkz1PiHLi94v~xAi7i_ zRjgj?x?$gTs3;N$Q<`y_orNljZCA``$HqHKvCQf2EdH(vo)U5v@RX4!n`4~VuTc%= zorS{KFRYeqOUR&ii|?Ng1a>LdQMyF7?niK=3r`HiM!j^j)hFsLZ9EAMHBioMcF#oK z$hJ1eM6$o5ba1EZeXA_tWHQ(0w%xj~QoARrK}~B;TxE*t^0`{;RgGAPyh&#Uk=Z#P zffNy;ICvr@W22PX+N>TF-u0|ICh)50rPZ*=kPxX7DIj;_D8T%$xZKvF6l-=lJx6gHU65uAw?wzH` zj^`&BsoHO-iS}9@SeK3}Xh=y(ONQHb-a{NCI<`Qm<_9#yj|a9%4=T=Dk`fdJAIV9yKP48;BxJ#^Ba-EFIv=@5fQUPuh3v-kU(7HpqjF2l75{u0-rEb(Vco2=)SCQ-BNRE93o*vCQOs zo`6JxluXu_S&)3ON!6sHx|&D<8jzAfJDog3kaI$IAe9%wv{;U*)a+leAha@4^C@FK z&me^Xsme&r6NJpz0e6g4koE-zse&kWo6eM6?v@yw8M$PXRIC<(ckEHt7dMlpIaZ+f z=AB&D2)pTrIrfLj`)HK*rHoYjg)LTa&MCJ|wHEb*YVDRu`|>oIHFQo`mlMK(d&`C#?G#{$@u0uTOG$~vF51jikuU+ZF3_Lv%0rn zQ@bs{2+k*$_Pw0(`f|qW>nkoV7pwnTOv*?UYItf5__5=<9<1zJQ^P9lD=r7q#z=jC zFR8ztv~Mg7{uXG`KjveHM$t6bS0A&rEkET!gLtnjZJprJeMhP}Kyg-TD?9SOIU8?s*z62_%9dO(AmywSCkIi~CDB!`q^VU& zN}0X3bi(<3!sTVf%a=1wrwQwlF=s|*#oO06Y`d9HY}vRk^7C9{Y6>VU_a zr?2yj^Sa`4GA8zNz8H`?pRMk3FJ?#V^$PGlFXgsjzi!yx&0q*cnzS!@EnC?Yw_R{c z8?J1}r@Nx;8?M`qTiI=$nFYB>ll0`Az>{CUgeH#C(BDa;E8}2u->s0T)d&699xJM? z#sj-|DTlYU|?t@hu(Rhjn1KU zUlI)scbk)JXrkLXTgUdX+E`xwj+pUSW8#TiqwRuB{0I-WFxQ^t-R zn@%XS_d46=Sg39`5u3eBV=T$%o@fE-j-q?Jo`TwhTe?{W_es}#?Nv33Cz=3Hz#t?H zH5jHjJ1m|YY;06+4Ga&UKyP&x5>m(xXAVZJ&Bn`HQ4+_41vxxlBauvM& z$Cig<51&U}wwo=mYG+^E_B*@Qvjz^(Y?8g&-$kpH<>s|88D+m=-`I7>0#rtdtA#wd zg{f^x>OiVKsAhA$p1{DAB4sn7ND1IvHS;EkQ_o_hiIJC#`7~j@neKHtEtt;}rX?dS zwop@|-Wfee!KmGg+s$y2xWJ(v6ncvG~5pDSh$#XrBF0ah-mv zR*Hg4V>WqTHdd~qh59Ia|I^xk$7FtH@th|s z{;w9K5+trw7SVL(!}bySbx3uZEJO3sk;(e&4tnHOa`2bAlZZ5F+&v4XA#>TX^*;!v z26r{;W$y8QPzr&VAQ%&pG?M|8w`Q>#^iEw%`It zMx?ZR-v2SDPq$rCMv%by0e}QkD!Wu&p6+S;3#Y2QFH&X%$%Sx)RTOG|xtBwu#EAVcNYGA@n=l4p+`u3|rn)>=Kr?tnIPbK3 zFIOa6lEkx5U!9@fN%3a6dkBNO9gt_oSXpW$$Q<1WajIeX86&k-WxPnf*S_fhR?>c? z%bFpY?Hgam^7Omodive5erEk?17t}Um}@2?ffakWg-TMUB)3*DQ4#@GUSJDWmqFY8 zqg2+v7h~8MY_aCtmS=Ci{p8ZV1FalrY**Zeb%vVo+%Mu`vU8wr%~b@Q{{2Fj`LHVx zXD0q&Z@qHQldjO5NG??>FJ>HPH>tQ+DzA+zwu}PxyP}751t*a>xwE>(xAv9)Rv`7y zP7M=zi!3sg_8T_nQ2laK{~18H_cr_eTlDVd^@+m^d5NuTw^a*~ zK=MEFEH|{x-wp%aZhqesm{Ph-ss~=iSZ&x8|u=f$K$|nzdi7H9C&{_(R<%}E`}^d)a7^pP;2!PW=l9bj%NOj(?HPU z@jxkr=dcFpG56HMc=qM!kI4eHA z?E3Sw;`yx06Oyhd+O5@?=Gq$G9|wLtPP^V7cz?5mzVkTHs`aNBBVmWXDA&F38=ucl z{Px>#_}hQ}4S)aJ-*MmF%E;ZjAgU(c8&;Bg9tXHq`yL{|C1h~q`pWKL>>iM#Lc3ig z1KK^{hLO6b{43$wPJqu%9Li!UhwJ+WJEl3eh=^K3o9_64T{WdOwCXj^weqsD8Oaf8zTc@PN&1AAX z_ruo)j^T%<_fvGr1dR<%gA;%MM`#68X)-HK{}WdNYe7LpqXUgj9Q;5lhmGgORdxc9 zkcCa~VfP!#VsbR>N*+=oioJt7QW=+&HnT#@^Thi;o1~*r{LSkVm_@x`8p&xD%?I7B zR5p+Eik11Ag)n99{`0YrE6e;iTGru``F|jgT75RVP9^u;(2G-nGF$W-u}btJ#rZwo zjBefr+8&~P?(nK9pOL{~r4;j;-2KLNU3fk(P%9?UV?r$zh1f+heIBrw)r;v|C#oF~ z(`!(+Sw-aNH0ryILvrb0M8<}J*O;&prL``l$c6W!b`X+EC}+Vj4wLyYVLSw_zM;hd zx_JUv!JKqxDoy2JYFfh(xKyY3Lh@BdoD-aKAD8-=Y`&)U@T{Fy51$b6gev?QB=vG{ zk!~fm+pkJKsW8aTj;LT*{4gb|en4v|^~PQA`7;u;d#4z!9XQ|K@ZK+o+#ra@jyDG| z4Ihsa{k|bl%shsLMdZ>z*GEU7#JbEw@;I$L^gPd$nB!!E42(f9#lsGy9%i&Z&jZKN zEMTJ0><$}!V5p+I=c9*2Vswj-P3eUpKM?E;l79Ezur-O-B#Sg102e#;1&2)_^|RI2ZpGaFlO}kCRDQYcdrFhw*|YOH zf!F=s@KYl^TW;I+piTx?tP`*Hds^{(5&P-vD=$MnzYY08|&ap3A1LJ(iRq#d^?) z?SQ#~Cs|(I{Y{hbWy|bJ+>chNisrHP-JgOB2q)gt*TbZ!93L=i8HVEL^`4oauLaE=j1){43XND)UJ`?8jy2>@W6YS?yrQw1A(lzK zv&uX(k+e6bals2mAl8Si3WXo(Ime;XE5ktG=W~*ob`(!7SZl z$r^R~o!C2VOs73W(7UXk!QfvG2RzA%vPI0+-YZGWVE4o+kqs7yC26Vn(Mn?|z{@N@6J@=; z6ZZZP1WXp#vpS=6UPOyQicUfcKgldq0dcTY0i?bbW(A0nED3}$Sdu338<2{FDRwXw z0UL#yc~ZEd@`1)D3Y}>DK&1m+EEP}8cxTc3m*BpD=jXuZv*Y=^fO3P10V%4sk8=t4 zngeY1i05QkU=rIO$oy<7G`%~T=wf~GpUnio8DeDNQxQ0@z?4-s%t-xhBbFb5r#wWb ztWwk^jg#TZ5q}-w+0~UXh$MLa)-mhB8^0Tfw0T?C8lK8#=AnNX!Z8`C;c@1aR0iZ~ zt?6rH^7!t|1icneM~DhTt3`Q}uqErEqLTGOooxQ6)%7wMNHOArhzJSMtc&9*Ga$EtsqMr3$DWMuihJ|Ik)?fC zzj56cp3f(0u`)?2=39B3)fp6p^lmJ`MB0{C0_i4Lmgq?0H7yG7!4RMi50}XD4L8O4 z7ffJOltMr&kjDWX0HJ|!0B8Ue&=^zNr<@q6P1<*u2f#a( zXzK*S7nX-o0o?EsPJu$k(g<6>NJh|#KYC#v< z_`4`Xt!#O8!951fJqp#x5VI)e<@_MV{9IXbA4RUbshxj%o)A@x4RDz(!RR zc>07(Grz21peaqujM0yF>tu8%D#kd@=flA*TZyM7dlH~e3zcz+ud1ZSFLgAzTnXXR zbD6h-ml0g-UL=q6whl1sLgVJIo|5MQ{ZLnZSg3+{*P0{=DNjJfGa)bmssL5ISgF~J zn^MA*R-g+<%=ZLQG<}b=fYd!13fkp;{wJ6p$`8Dyp|Z)aDm_s7fkr2Y1cHvP9T)`Z z9sOopm*Vp|@bPiu<9UJ04Xzzj2Y4uo0HuZzF{bdr-q0Lw4Pv|3gr*&P5~x8?*sC|FY8R0fwk{~jQQE||eQmV%i+T2A-8_p1HR^hnJ(=NFVYk-WNG zf!5dfvcvvic1+V_OV~f~uT;hQcZ$@5H36SJpk2)Oaa|WaTg7oOq?x>?RNJfpe~`=4 zlpYb;Tl|n7-%~sg!vWH8aM2aogGrOkM90G0MyJ~iBuf&POn4LF7t^DIEiPS!0BwLa zi&I4bSMP!;E0Hc=g~Kx>u05x%&t}OYUpvBRdZlDalys-+{lKa(o~18Ru76Qb!K)U2 zrY8C3x(uOXGA(0t^8)pb-fy%T6B17=gIZXHT3<(H%xKVB8LbkoqvGCx)Ma0fF*|z# zNVV2*oQr)rjsv9{6V^N+Lq+VLYr$QM#YvML^lNnclrnu)=g}<0q7>I1(~L-$K_=51 zCMOwV;C6jCdr$yz@nJyO!Z;>~ngEKnV2@WuXA|Di?|Il(a$-a=&vY_c9);Ium;(r< z)W`^j$iNCO5&Iln7_>UPCXk{XNS){Tqx-FZl5sbuqfwgG>PtwLj~ZxIIIJIEXUI(W zFV}(?gBxAV@v**WIvD6WaM3JIWn;iWUUciR;;_^&(>`}-_v`MEdPCS+?zI%C*EoQ8 zc~dZJulK;j0KC?3TVpNMhfqOiP;DS;01hw>lnT_Fa92XDgu55i+R%>E3U&|cg?oED z@pv;PHr8j3^_j`bP**5H$YcXhs~ZSpt_9-KB9rUYtIc7E`Nb543C)uN<;@sfcECP{ zcT#naX&=r$+|lcm!qZ%al1W6OxN z?Y!_@At0^>gM=|qAj3SmItUWwn9iQmLZHUpNMjPm2u$I`J{gEIiJ>9@RfsYrCIT_+Mpr&l43}pU+vhE5+0%eR9Bi|d}x&mDj%q9tU;p#l@1g%kXDd* z+Fd7%TXA0l_jBO+d2iR})2s$u2a1Y^KDjrHy*5OBAVLu@k?UV%mXbztB#4|{rIc{D z4yrUkkgVyyzRaPvv}Brc@zLwU%P7O8vz5q0ANzUX9r-w^4YB>DPy~0*dQdm&#wvzywr&3i(1SFXBwbvzQ-XPFFv#z+j{ZZQ!aN*SIhK zyxGx4nHj`nt&JjywL^R_j6!hv9}1-Ygi>(B2tP5)ejx*w(G_1zg5O!y=_>*4#`}== zwd;xVDtJ5X1LI=-O&q_~8W%xmcsvT;9+nby)QZa9v!G^~8Xo;XdJh*bFwe80%SW|H3vs8cNVP{yzc zVZSkYo0Bi1V1Nja8!8RZV|q)76qN}_VcbW0$#{|NOQrItmy$l61%p$3(+IeAtpo z>p^J35HVoI6=5u<5-H2UCr_*2pH&Evu-2l_a zh0otU@!Q7}A0HP!KA!k|UikdH@cg{+yeK z7p2^7o*gmC52MqnfatoKpw-n8B#VNlEa{V<-Xhc%g}S2QUbd=G%^eoReK z>Dw7VcTd!7h{kK+AZ3>n?b60|K3eRyx=cDGp`4Ti2$|nYcj1z8mpN!Bg@f{n#s-|u za^m!R$^s`0q;uX!wv^00{V-HQL^KEA5ri=%`l$UclH3bG^ZOcp2NEImM@BbmXxosKxZf zyhH;u7&Outl6XZz61iYHCT(nGeh#Lf)qqqnAQd2rxq-NY%UI=_?$1!xEgCiMbpsT% zEeuoROevi?4299h6^80Kv_>m;`s9CnB*?nzO=JB^@25#m>XjnqovKe=Vv*!GoRzXf zKR*<^reCF@iplIqty8KnFbdpYYjftRS;6p{#@7kNB!wBMr|iYVeX(#Z-%vXogg1zb2ak9e?SAetn|lK@EVd6>pmOt%WP%Pl7Q2Eo^zwP!YAA&8DdnEmNQLp9TQ?&vQ8<7f#h))K1&dC}eV@@1Go#LR)`sN4sPaaVy@X_? zfU^88GWI0C6sYx@2)~7~o_cvnO1L0#8 zkcae^Ng%|jcQy@3ofGEyS7w;&BSb2Ge>1@1;Y{$g!Arpd;3Zc3uNv0>>9C27j--EH zoV(yt7|>`->-ZW-PXAPnus(b@jU{6^Z=n)2J6Gbyl}L^ICV|c(ZHpsv& zH+exJ@Apw#*ile;ppf9GWLG1!N~lbW!cE3c_~sGo{r+D1^>Tg-2OaoCmDpVU(8mKa;R*&_r#!8h2E*;F>h2Emg-vEICzq=z|;ucxdW-;MUQjZOtg7s;=Y#zqO8EzOy4n0KhDDr za_4E2^;(UQa%zYDP^eRH9;Kum{T?3O%0M+AOD)ag_)X#`s`J-}v@aG)o^reZ53lX< z`XY%AFdYFP?rCrj+v+Ey`MN-b%VE>KKY>&j%7baj)PDw29)n-KbhgGMoN3Db;_v(^ zBUQP)tan-5wgyDx9t@S z#l1T`578#`#eG8qKoMgU`^>agXssr3i_aa)ccaYyeN*c3lF3Hdf;QY6YSTq!MJZN7 zZvK~wQWb5Gv!)f@(;{nasK<%(d~+amqSnK%R#7UWaCpvS72Jxwhb8q{9Y@9K$(g4! zQulq&;Q?-V=aeVB-WNVUKdspEJbjNaAZ0@V03zNmcwyX6A7*&^)QGIJb(;(YIDEFn zQ~W}{me$+6M*L{WOwzr1kfu~e!AahenBR%==5UwXe^Eiy{IF+W!un~dB`Z?Lt=*-J z9pe&=0gPZ3hMmKifZ;%js`;8K3K}<0S7k_q8aR*=9T<9p#LGYBb_a!fLA;Mg#41bF zJs1%`4T5zmI#VCf?kbU3*H=HDqz*O*Q5uYtRxp{Ne;#^Eb0(@y<^Ujc88ex0+yl?g z%K+5J2R?rLz~BG=X>8Qh@%i*}=i7R$h%Eq$E2D-t%g;K6;q@a6L|OZR?DE911~eaPI8-Sk-H0zQmvMkkKmPhk{H<0?PWqn@>s~<6AInsQ}aB`tr z0#fl|P%eVkT_6~VG<^`Ro4~BN?h`}C#z?ipBs+}+wFoL($UtL7D=?V|t#Y_vpsciX zX_R$DgL95vR&Z=CvPWrhgfh6i&(f%?dkj`-oIB@1TmV#vG=QtOJXhbCze@Y|t9XI1 z^;PK9awH9~>et9?Q1Zs41&T1eFC2$uFRP6eYVX&ff6aaOCH8bh-6DaxIXRj#0tZvW zP4-c)%GNW5TIUHvRDeR>QeQ!GfV4v$K}?P8?HT%eb)=`~6= zZOZyh0LT_gzxR6?Q)E|sy_fO@G0PjnxVp_&WIg3{F2CO>G9OJhrBP!PrRp(Sin_n#u{}6 zLoo)=GU(y!YHbHvGf7`CQgbaUbpe{VZ**yYvb0ZX!BLBcyV$d2KE>+86laB0A!2Jn zFjA#fa4{e4aW=fYz2P_ys0#Xh;raZ)^Ya7OXSMI&8G|ty-IX9~&qa69| z^QEwe+`Fy`NNuJD+PObl!^3Y^?Aa|3m`_WOx9VGD8I{UKG>FX)OveJR*9im&PR{Ul=}NF|W+J{Bi^69q3&# zY={TCExAH4FaVUHpr+trT6!*Oaj1S@QbdwOFNe>kK^0udlj<63JzNHMpw!y{&dNpq zQZD|xA`x1CKW4T~CrScIUrC6UAu7F;sQxAL)o(!5GImp@x_DhdG4{SSJ~Zo0Th1X`e{^{Q<1JCufN7Pd??DF-u}BLLszTfTMXB zLRN1cmm(OlC3QAi!WnAz4zkFB))~&h%BH9KNoQ|6sC5NI>WNVnocu#9ErA zsg>qwYIa|m-|H}-Qx}jb{^zevUbhJ$Xa!Yp;T6r#D_M+|4#WXeK}$f&fY*}B1duw~ z@iS>(>YZ8}&WofkDJ_(~(d1`a7`@hIEg7vVW}Qi!`7m3#u0_sk(vEOS8gQdMhwO)@ zI!33Ic~2d*K|;Z3&6wID9w5Q1gC(TUgsoxcCcliv=Ta zF+kQT<2VY=qvCN|g6I3&f$Q_cxF6<490PaAdR1pQ0xLjXRzzae8kWB5nxEpQBhi&g zr9boZo4lsXubsbHp(wepY$|pl$F6{>q7u2J$PMpgp$Y~Z+Ps(rPlBy44xDWb8IZ1C zqae76QOiKT26`#z)w;xl6cqxGur-8n~VoCh_t4 z#7D^bK0feyb=*U6-_ex$d13wFUK6(tAzGey_d z7fn`$={cKwioPguUIr)n0-*2$#5DQPl7tbF(CqgrDK^k!UMbnZx?~BSvCR|th$axM zn~!dOk3O6-f4?VvQj+KM>IPHD|0W=%UnT*$c+eYcwunkin()Ge#_SQ{EmskaKPf7KKXz{n|zu!DK)*OxjBd}0sd$y$VRXQy3DSG9zG5jT++eoB`xEn3viXJdbohvmaEFp z$_^j^77OJOw)bupmWRoGf~B!gpdTQhJd?^quu7$Shx7rxO(w>0FjX0kRsoU8_8GXZ zCjEO@MElV!p|e#&t(Fy03(y+jI0_zT!~6RKqu;oD$9;E{(b4aYAv!~NR+5|prjSuk z4ji)Rbc*}thyP(%`qvs{t?sCm?zrTS_Lwa%R3(9MnMq-0fVYWwqn3cw0nk8WL8Sqv z?#Z7~@@i{p9|NTm`pvj2qZS{Iy4<&yj>-%vY$dclFtMx!@5NzH1yf_Kz9gh+dcLOg zc=h?LV%6l4_j3bNfpUeqq)^xvhlgsu#Z+v`S z_;_CCi+Vos@%hBv2fAPeDr|D#J@?}!YGZ*Y3*+e1u0QuIn8c9?HVuA>=~{y+XS9nY z0JAcxnmmd&&tZ~&y~@_0KUy<6MZNYEsq*E|nm{W#Zs~>L3H6UGYsCigbZ6x`4|`t% zCXfIFy(n-vGy9p1B7AvoH@}f!_Nz`1=RqaLJQ)jhaKq6KJkEEV#~Y6JKr1H-HxP>T zrgs79?H!)pvE@}en&o&|s<%nBmZc$NefkAR$=1U)hpXSDCM=KJ0#oKgjDd`I&9S7w z`WU_rS1=HO?weEUZVL~*CoIQ+3?5o!F7kK=;+`t={r$dS?7tC6=|A~r`sVff$nqr9 zA(|f?Q?`@ulob(7H(ksmTnk&0jUQ-@nzuK}9i8TR20!F>GIjL67{HaLFIxs0HJwK5R39bTsqvwc|iL zt1(jt<7gJiUL8c00<_9_92M8wfpMRZapSr=u5zPv#uMy5Tr5vn0+;r=xtC{%8K{ab zAoVBuP%zJcijBs!Aqh0}nJB4k{{WMwi>? zU_-4NofP+N?_Y5_Zed1e#wZ1&3}`h6AQxUePUI$bHXj{JTkDgj?pa!53SD48M;PLh z*Nk}2LX4Z0%kqDOG?&@@QGPKzE!A1}=KEdSe7z8o9KnRiVpp*U+J*N>&oIS=r^oLhCzK?o@}C3`>& z=;A{OrQ>AcbGM;`D2erzo$+O3qAC~ET5;3^M{PLkfwP(i^k@$QQuRck1f+C82Bq_g zq~Mg;7hmEEUoDmhELX<>03ZNKL_t)V&w29l&#z$V6>?(hgraoTrAX3A@;<1oK_-;& z^A5+LolKsmJy`;1FTqatw0XNM>4Ip$vDHjx)ko>|0(h3x_g$s^)UT`-%sCN zMALvf1wwHsV$6rWR+Hq|Wiid!h&UhUTxjD@3cmy*SO*Bt^rjMrt5(u$u^8QuTzTQ0 zj<|*$pu6C@2A-FJspsWDYHmoj5S9Y8W>*9^6bDhkMvd-2KN~Y;hX|nFjCzPqf(L+V zS^HGLJj@rRx=Kai8%@EKQPpXs z;gQ-W$=*^QU|nc>6Q*>J_Ggahg`$e*g^>~uzk;bbxYDH0+uk+zXt6)*tWdTtADiIH zN_QNx4kl{Pcx?ux`rW-xDLp%rp)BjOqE~CXeqpC3BjqWMN@}4CWQQRVkLwk&HB&I9 zn+uTo<~p@KW>A&^NcS#3_oCWbV4f?H0qr+P1_}hN5RPWar-dy+uvNz02bc?Li-A;` zopHfPO(4}6=UMS~ACP`Rb>LZVW6>`AzPQs$AoBx{=E=6CQ<>~w>N{`89|BS__M(v4 z@GP|r0o{uFj^g(ce*j2flX9{!g)`VGGEY-=Uy$-Z<$+2Ag>Mjbmz1u`k={F)fX0rIko!L_{rNMWZ)^poo-ui9G_Yaz} z=t&mmEGb}tAV8UM=e0Bony8kbCeDHk-xuyX&ITQy&yJ4?raboi^NEkoPfI&=*`@kG zLSLGjL|01OOlTvQHDxg6N}2F{i8FHy-lFv5pL1}PZ!a#{&iv5Qm6B#ZO|TyF?fLMZ zt(gl*<(U6(4vpJjZZSL9dXpU64c!3MChgM{ytB!VmjZMPTX@AHDg767Xf^4sxq)O$ zgXRwr*%L)GCAZav^EmNnCmzR%vp#UNH`D~A7}eKg!cc+Az4qu4gIq~*Z~LA?c#(B# zB=+eHo(PZRI!j|kzGF!-HHCx_?dL>GZB3txhf=Asof8v-IN5g{OF04Ypp#W9KhH+N zL`j^Fy0cQEYPCbq;a`WhL}zq3W;z^@ z4wR(sfxwi76lb+o8k)=ds>%9lD<~Kq7824v1%{X`#4ouAo|oYHwAasT;J$Th+g4am zz$u>C<%jIyQvERkNMZJ5NF+UxqIEWQ1%d`UXcS`829^dY{vgou?wGU>;9`=T<_Dn~ z-Z17ZK3r2$*tgQYe7NH^8)8-v%beETX=Rl`=-E&oGHy zmM~4Ola-3;wob{`79Sk4($xD!Iv~Cy5&L~sc*Y$sMqDw(4pvj@l@^Lv=tyNH$`QQ7 zh3)ff2BM}^Eh5Epx@8op`qhS-C!s`augf)G(qUGYhlfmf5$R;Q{6K5J&Pz9)#h}R~ z+Xh)(+IO5M9&Zo4zrW*rd$a5Dj`Mure4GZj;#mx7AMMh<$v(#rB};qL#=p9%c!H@` zZTdHXR4q%gl35=B_w)M0K+1ezJpXPEk9BeNVY$2y z>mH)bT$?A^eh-LR2$=xF)=sBGb_O;vQtHW#?L5&OL`mus%hEn(o)PC?AM8BpO@|M^ zw^c0hFNz*DEQk@5;ovV0d=p5;VL(krsvbElT`etbD5RVGdZgfrA;C!X<&&z`om1TV zbhqO509MFdngv>v!e}AyD~pwCbxHeloPdhUQ~i-2x{W@;=td!Dq_~9W)B(Deo)d60 zqudUne2utEL2q@c1=yz>q>O?(`)C+yzE3CyZro0JDF%k7ULwu_74y$BhRbDwL=M$M5ln0^%pQnw9RgO#$rsCa$i^ZE3s^9u6LF(h_7t6C&6 zQb}Ww`nCpS0x6jRQL{QlO`%|C1c7TOU6va@Jo`L?p~_yE&99`C_KBb0ms#4US(;7X zcc3xKK-v7W*y?jy zH96mTKJa$F;j9m|`amfsT796F6HqeCh_7K*CE%)&F(#gi&`V-6&2@9Kmm=He zIm>pPedFaK{CbAV(ms5Z_L2H9Q~azL%$(EUI)mc6|5;^vTcr|YOxk|s3(y`KpnJpk zk{S`<@){OG5r!zEM7bJe5c_*nZe}pruTS!$FJ354 za*>%rl%jaOv~?C->k6C5<0{gZ@fFd_DMeEWcVw&k;i^PgYOzC}UzD!dMK?n}BqeGn z_6`+I-6|IX)B*-pFp_OVGam@w@Oct%FPC~jks4{jsuKVtgenH(Q z)ry7n9IZgK&Y_aM2LYn&XFkr;82+=G%#RhLSeolm38R_^XcS{!YI=sJAJu&mtk|T~ zR?Xrd$_s^AWtC6;=wLwGkuldmkPlhd56_G&AGIRch_bI-XE>34E8@~nUENrz115%f z5pbcAN;(7`9fb%NmI@%AA*Ka{+HvAgLUmtM3F(g4GuVKs&MrCYbYeeTxsP`;xz_bG z6CSBfZ>2>v`@W_rxFYs%&97elz#i*NF;IdrRuhYnc@Vi;jQ~}cT8G|CfVuwTl(zgtaD;65pIg|yXuc+q z_eZYF{pgVt7ojiwX;oQ{Zum#LN! zu`@^j=hvjy)jP&>&!A+3SOvTe%;tLS_7sZJU7a5J9O9Hg+Lv>kCKK?Mp5$CSN#q}6 zs29G&QjmiQo+nbdoo()3p`mGT=oXj6Fy}~o{R?o+e;G*W57ukKtF&v;s+99)wusdg zeMnC4ZKyh;z9;UmLp&8rJwl*nNkyPRNjgq3;AHZ;;&I0Yq`XUp%be`*Obu36{Sat3 zHcD>m;L)q2Ujv`dfzM~h=hGOe>-GZMv=Fz(eh3X#%0N|2&Vw>jRg3}L1Ln6HmfojK z;HD@I5V6B|ap(AwDI^eu6~%6G*EjmGOPms2hh6fb=oN*Usr!&mGUenw-;^D|MzLqf zU~5Xol5Ap@`2BS^WWMiWvA6fw<$bd^siWWbTy&z;ca`j|pEn}H4S;oA*KLP^=!kMK zg)#h)Gr%4;Axqk8hY^ZMR9&G4N|0g)t?I8*N(4(mV4fO8nonU=FXWBbaGr-V>k%T_ zlf9IZgo123T%47ub)XhTFE(T#7jvsRO&ke!D<9R&OuTBnj(dB1~M%92)#&`=N{(q z=kYaVqvCE(NTDz^|E@Ktc7Rkpn!Z(kZQxqd6@bcpD@@xEXGvRhGeIxVkUIc*+YgX? z=`f6poLEYb8_Ucm!Shv>JUKP16`T$6vhDls?v=a%=^zph7b^2-nbGPZY~I4PXGV;De5O{gOn+*WxgBxl+?jwu=IJV>0vKBaa>xUjt?w#oKPHe$ud@s^`C}%d8xr-+`JKiTX84;v7;pt~ z1-Jp_29;)CN63r6mk3$i)_Rb^)RL;@P9<^`&Sa}J=cncpXH5*Y_x-S^T-w-gS`k$D z$a~e~Oq!CoNRMYc8oH*KvR!`J=qI$3dqk7!F%N6}z z$*O_1zKQ#oW%&a)2{p~Lf~kjizu7{Xj~9l@yXW$H9CiWBXU3j0P6fl;*+1<+N=f=Y zYl(&jlWI5~=9zYd!l-$uJX``^9v`1$q@WtWD{Bh*7yjo@0jXD}R)6Q~e~$R^^1IkJ z_|VU*q*%&aJV~=_38_D7$^D#>aoL^7N*W{DH)1BBcsfui$>8eJrs~twi*C?Fo2h z#AVX0!)JEaXaT=6JShMmrTg~(&-KK<$Y2&$ zngm6<%*3thdExndx*qw0j2k*UM%m4eG4zsU;+DY-PK=Vm?E9!!x-p8A!3B$}P>VqE$gJ!z8S=ppN14Kc5nf?hAHiExjPO zR`X_^1IHLHUu4`DO|z@Ll{rj7-S^aR9tR$e2i||ZXh?-UDHs^F-i)_q|`uA;qAM!LeN>VbN2aVA$OYFRF zJg?g&z9w5`3{-#rYqL}~1Ke$bC$8&foxZ~Oj8xHCakHW^(4C#?MNvsnNaq%BEL0Rw zuEPw?3GM<=Y&upS()@|Tfu>HN>6=VlZA245MVAfK%YogaKr!oemBtu8IC20*B;c`z zQwBzs%qDKyIr#AVGKPH8OQuxjVbOpTPcJYP_xrE~fRs|5MJO1hSpm8nP?z$POZh3f z;;2qQ#Z;C=77>(^0%LSd)WxNVn*5n@AI`yBq%JZ5E%rQe1xmFp$9e!&K&1l2z7wpy zDl@f|K-=;dF90>%heg>VE{nZdlPuH5nx7Thwbtl#R!P8dmN0KY4>X@eyKjxPCDClg z`FYp`>nu`dGAZhSJu?CYt1(llV#sf87jcm!&s5Hm`)Va2-NPp+z>g2!C%firBs+kL z()cpv;F5q;gx}B_#M8plSpsLisD@T8Xz4uOa0VMyPBf1?w`aBh!;`49UwejotYl)_ z_xb|Dl!%Iu;TMy3GRwDTDoG$pR9vCG(hEQWb_G-^e%5xCS3dAlwu^++FAkXTAq-F2 zb01&y?b&0**+~|(0A{KC95rLIPz>GEdvjW${x?ed^qYVB`oF)!4(#)=q_ z(QdND7U5MCh?o}gKFOK@u%q`IyV7Fqqq2>hq+O#S^Sfc^Y_fC`V4%k-F*zI!G?r35X}#Ypx0wj#ynZHvP& zUxiAavOWtH7+`a@+nkm$lfzVa9tb(^aSx!7VjyZrK^5^Cuh*CQ0k}nZs9)uUiHJZI zPTlpv%prhcDF8bYHRHb-qx0+uv4u*P{4)-}ey>I+S!)C5u_Rm?4BqINCcw;g`tx<1 z2hPWdx3@REzrW+{*E`<-?HAtP-|={VU&m@n>%~i~u?-p=NO>Pve8yu4ga>HQWWfx? z)%mq0cGlEW^$Mcw{<%x}(Onrq55capFZ~q5eXK}&CQB6J08SAeQM}H;zvjYIex7lM zY+81#19?27ktXGb^Vz|XWqsQM8;_$8tsFF$xH2R44J$Q)I=hZuBt_-!=Iy!9=ecl5 zsz}%TRu{`Q-+*Z&)IIEl)DLBr_NQPoV5f<#^|##^_l5hqa6O;q#qswtBG5}^>u5U< zoTtb12Y~W>h`D%2i_XKx7=q4<=FHTc7^&(#guzOwKdS+zHXnx{V520NA;vO#IwAc~ z*|%_7N*$Z`SNNb(A!v>xP?wBr3@v@gARaY5hE+L0FcKrRJQO^Q>T@tT1Va^IAMz1l z1oIUKWz{7QmRY(T;3>&bgON(frCE)! zsxFW`nXVLxd(0VAa74-bCRfyS_@%j@c>vs?RG`Gu>#1ZNlbNO6JyKa>zq3-Jt8Yq# zW)Ik6cHRFJz_I|PP zyu#W+)C^Sk-2p4TCGhB2YEy=g;4S@FWr2EOw`5}Ra~}ky`eiz3oh2X68$TF z7y&6nEr{a271vWys^UH^)lUXPq;g~YZpzH93Jq2_-NZOHz+O;SX#!*(C;ce-(3_w1rEVvxOq zE5+5r4@NHSyKZNsR6x=Jy-laWyQp6C}+Q%j*tK#JBv z---dLqZJ&rnmJ!(J1K2gRTq$2G2jIhQmxvpN}v>s%Bx{~PPzJGw5I1R{80yvq1(nQ zBA&*@5YjD`jbf@-D-LI*-rnEv_Wp)n@9%j3^~>3*_mnZGEA&EES+dTl_Lx4lAt?=v z2L`VIgOFbZkP7K;$oe7`aF!3Qyjzr?0x8um=$?J?Uul}oN%~q=3W2IWKP6Z{2U6h! z-GG$NDRxZpB2)#@`Ql%b=nO6+)pT&+Wvq`v%p^V`&<2zut!`e~XcJVHJvYtA8&m|V^$YxW*MlMF^?x$bZ<)yIW?KTVpON*oI@ zS}S;aG&~*$-rgQ~yuEEeid|AW*Hpjvx`<2q(iKcpx;XLi6ktmAi;83V-5f-jbn(Z6 z`}kWnIp_ozHDpvv-ZZE4QOJv8J`I=q&i55ScP5G<7&I_k9isEYD=UpI1;nTfvpM*7 z(9@Jb`LQOe=IygEN)ll8&Zc-z~;56$x`6FjZp7EVA)Cj+ID zpS9)lx_U>ajy{Yz%XI~Ff=ihOGKrtML~%*`5{SwvZ^^iyeNju|7yq1-WC>f;b?i56 z(qIdpGHIVW)uxhT)|0$ecljc1a?KRBAest(|9C!sR3^m1x;UUMehF+ZG4ZBAvu0^! zf;bDon@@+lH*ZPxG^O;VITF1DP!O$-x7bIbRwOQ$-%mfLl0Ba+UP z(*FXG$~QyC_L?TInpup01fY`knmwIoF+vNptb=-D6yhKS1rIB)<(raxh`#MV&Wx)O>Gun=fOfvWkLIJ!Hq3U>xG%`{gxq#%zkqRD6svX%q6se%d>x_a1M&mPly=hU%W`HVA6R~LYi4OR z7DE(8yo&)EtI&Q5&veES$Mjo=F-epYQiwo<&q-iUo9uj1 zNVVaZHvf5w*B+Er(o2;5fDu5Fq@QN{;i(E(>#^^-e$`f&Y%IJws!18L9!~N5tgt1d z5;4x>#MvI()%ZZ=hC+2Rzj~%35m;J2q4wKyT z0GAWYCqM@v3>9lumjP$bGLzi0+BoT@s0v19S0*Ro<0Vw2scK&m~#5jo1C_okaE1^yq z)Act103ZNKL_t&_74lY(KW+z($BA~v>xtuRwrm5*>}I412Faw0Ip;Tg~OuFH4R=A<_|lL>Iwxn=Lai> zG=U#Cv|rHs#<*_CsMDek%MP4e-eC{ zM%qE1CT87|rWzT8Fgh8Cx}NCQh5LHqzMkm!g)wd~Pfei^b128o7_9d-r{DswZDQdR ziGwWrN3-^Ku-|jAy$LD(FPB!^kT^9j#J0_=kIx>)*z7}u9_!vu<FLosh%z>+Qu7k)veF@Q$#4QsB4htLc@tmX+YL*&L=|F85M_-c zQk1<9T4ln$9~)zwfF012^VZ1Hkm;(AH@&jgtQ+N) zrndowO?0aVGSu48nwli0G}J0+O~72HT=dyKQA#z6J|#biELI6=6AK_wJr$|O2KAz` z4k=Ac%ZOx`VY?xpxmf%@?+jE-Y4b#Po0QL?k8CZ5ME?b$0Ii89aR4KDunZ>Sz%_tW zo`kwbsSvo*T&Du+Dp|CDAL1Zt0E0H44b%+UjTwy@2QdydCaNM`y1W6r;FJZ5JPsU> zM*>o(RYbIE)7t6N`@ht^>z3Tea;3S=+#Lv3&-D90Xseh3xSP!nySO2kM@^R0>-0z~ zl&n&b$&3KtcG>&eoBE_d)YHvlrkaYWrQ!7Jv#2Te9-GX2VUj1OL?NaU(}G7MdslY_ z9o}Lgv7dy)d~DN#)(L$yEEyJKAj>DOE)^7UnRmw+LFvNwVW>A$Lbi!oH`eurTo>m1 zjdks<)VeR|{k939+*WjBWi!=$8)et?rm1oF>!8;rYHzXsG&>@S-<0Uq>6kaJQp0!r zXH@5gs!zz5hjNFc_JQg3n|dvF6^8m!a2xq!%MzOw#91sRk#2LLPq8LyDegsvMAiny zurrR2Ef?f1N+KWr#5)sPs2-6Ku1wkg{j%no{4;ggBLq~T-R%Pfnm)tWCHUx zY1JgiBwIu-!n|(G`;GgyChC5F`BrLr1c^T6WhmRI#}==^KG)alX)xAHHfHM$O`p9)XzREYZwRSxffUB@aU>6QHIyOXM$zJx$#k+K0Cx zrY(xc059>VMasb?oQ^&oHu!Uy%5x)Cw+P!j`)Hzoy#`+k%Vid_ULF+NaU3UJKYoCy zZ9MNZ$wvIXdEdDB?L~%#Fpjg#e=g(!G%C_I_k@;SvA>#J5u7 zQ+Wav`dr_Q6zu{=2@ot9rW%AsHz`h4jkINb3Xy({dMQ(;k`_yxE6l<9G z3`DrzZ(P?K{pUz>8;LySx}RI?GGH;RtPNLc^FmHgR8cpw+Z2q}Zr-hN;`BL)QaSXW zjr?;n^=y{tp>g=98!3Hm0jhs6HchQGP(Jyr8!2*`pLwLh@LbX1?zdaZM()dt1V(A1 zQrU*i4Wl&-Za7-QAU3(5I_Fc1Nw*kw`#5mC;&o!Y>PE@}YjY!&SqYT-eW)by@mmgV zRYU3qMx+ESA~RGIsm^Yi-w>cUJ^A6UNeN0wC)waaAVM3AzC5gS?fY9}(;_S7=N52H zk8HLL62O-A&C436%U*I{ko$t(mu;-pGQ7HO${hzbJhoD>VDm(hXI|nyJGtST46xrU zSIZ_No2fv2C!JNY3Fd61*yNDWNUfD{7wB$`XjzfHq*=C+7F2ZvW7o3* zbv{7i=$ToPmTQH&?Chzsl`3h9_A_&(A8shtcFS{mUb1QbC!132fcM7DHc(=+y)OD} z6x?$hGG1-Q^oT}^AlW=+K<|nnr{;zSs`>s#^c$~xK$dpsvucSc`30&#kIG%$?{B5>(?GMg z(tCiQQhuyxst`?7-AExL74rGb$eIHv*;G`&Up*Ewu=PgD@)G*S^V2W(VzW~1m1U=Kt4cZ9!3QSF^Rci%`k!Rt5%$CgoHmSf$w7fXBmQq`76Z}Fs$^)om zekBeRk6^k-EA=Oh6z}U26|t#hfA&b*cFD?Gdn0wF7f+VlpU(G8Hd&zrKFrc-wZb zHNy)Fixv5WE|jMK;+IN!pEqn=4q_t8jTU7iMb=dL-!p=jxiMJS(XQ;Fo4N3CbjZ`#`Zo6iHmREqDSwJQG1qw9y z7PEDuKDM5XRI_0%T1c*R7ijE0p>Yn3quXnAlP1UCkq{?xA;QH})8YUido#6VTu<7b zliKY%nXr>9o<#I(|i{gjX>HV(-TbapQuu~AL61~~iuZE*+IC=gJx%-my( zk7=)}OZ#qXoN^Q8P1L-$N6LXT;g4#h=8JO4V1$^54829O-FR;%)BOHlSdrFEMy124 zWN~R>+{?BN>RUZCA)1;&ma7^(`aq(IV$f<(>E8VK9*Q}`e?z*()Ud!ggpFp(8d-a; z*d~e|O_a&{#xuP7v5^WgB4zVbbG+!GV6a0he5i=n0A^4v;))Mu|9fjPJyzGHayEF= zrQS#Z^8)T7>ASBl+*jF1t%x6?LhTIzRV!eOVo~5?Svm z)LNF2*h&(HXWsE{a>&70TI7$X`aXtu-&l5^^%C^sGpxR3&BjBEY9O|PV2fg4 zF%S4sI{=YDZof}LkgX}jQ@`fQE~#MWH308nIo8IZQpM$xTAmF`_-9|>wa}!Y%feEFHT2F&*X!N# z2Abc1Oa*V+nolD)xyK&nVu!H08Bum>cpR*AE?n1*_xl2A_n0P=Juc<`yK-l2!;TfBrP z$a5p%)Ivtmm`D#>c<6O`pxnV`kryvy{05vJsOmu)0M|vJI;Y>l!7v8JCiRRZB915Y-K-wUV z+*M|%7l`1B$}eee-bjVy?l{eZlzMtU;N}__!{lCl3^YG*L>44Mj$;5)$dUJbVcyf^ zRdc3P0g-tXL4W@H?b@TqCTUCe4Ab1&o_tY#j(v1dn|effPtZ3P;-5Zzm7W_n^~M;P zHDqt>6U+P&ikbkET_ElT7``RgTZ4ut{gh}+a~jMS6|YV%b<*H_`jbPq#Dn7ScaqYa zH|XX7fdbfs^MT%PEPH1U^oWjuoQaUyB)o6%zE} zCEh5zxV$$9jo8^m#oYTmWB;(hmj#vS({lF_uk8!0Udp{a7Wc2MMA1LIPZVoVlZ$D{ zEwMAvxHFnG4=lFWgzhh03ro!xt=w!ozyYD^-u#c;!kce~6EFRcX?Zikv?-{)mB%D+ zAVrleB``}13`uszB^Oy=HOo^uGcv2L56nd%w!qM}#0RohHc{8xUUB|d6J?n~tqrt( zpdbDojRRvee_5^5yZv0P{9IM8Y(of5xpMVL>BC1HR2JQm-gpMV0*kx*bs7z$z48E) zhYOcxiAW#CdXu)N?7y`s_;hpA4xRED z)K5yHpM9>aGmfLpXkj#m)?dC5xg0Yy;(=8m7`gWtI4+N9Y4m)kmv2d z;DPSIrah4I|EsMuIa>a?f!hwxjqJ<%0Y`mTyVe+V>n|!ZD)i%epZuC9`zYFIgDW&^ z#;0;kKb0EEFLXEx1dNWxg+>UmVM39(2I5>x*gPt--G_~$l=r&mj;;DPHd5qduwNUg zAhi+X@!=5>e(8ZL)#sy$+PNdXnhh;(n`qK7R4_F{x-^@o{c1@bs)2#6Z6csO=&0PJ zZmt)_V&B#zV%g+M^ImvgH-3IyHdP%AjFyPS&=Ff&ycqJ+P0S67K@EbAxd`_J+ASdV zq;)iiGL{hTMYtznPBMSkWUOWWBujagRE!2jb9-lrk5G_l_+8>xb|(w&hd zE}@JB>o_>!2Rt%(x8N~#hroRzjsqRtytr*>jz!f+;3j>yo%-MW&o~Bt{N;_5lj3G# zh-Vpb*u-2#^C2R#Fqda(%^NH7pHLbpAJ2(+fyWG83?akLs~Dk(9j_a(6k$KPv3*H# z=my2CefYwxJ0FgovXIz5HC4~Y@EW!Jcd_O84#O>CpMMloRb#Dxm_MPsi(K8I zJPDnsp}UV}Dk)-d9hSpFkt*=IQDhag^dM_4Lnw`T)kGDI(90`4>-HZZD=6 zPD`H|D35^KEf;B3?ismW#{Do~om!?_=91LSST<3T1zF)wJDyx@!{jJuYoHjbDTY&2 zV{k$enPS+5h@<4W*;fRZCfdS}n1|oKE|)dM@Xt%UdDKOZ^F4-Q^3kMhKbxeq)Slxe zTQ@*;^G3?IDh^TU>I0S|o(~6lKdfCr1*^{QX9%#xDd(uXzS-`mU)wZNuM>QU^vqG5UcqUKstbTFD2Blo?6i=&}NS z;gdDyjx#n={lsYANRfJwvie5Hn}mpTAbo6))Ze7X5mlQgeZrpfshG}%MZ#2q-dQx# zs&#Wkzcx{&5`9v=E4b zavlGx=BfYEMk*h)v9a)N6kO&N(;$=w{Z?O=06=Vm!lR*T4FpAZ01P(*FLt!3V(11a zMgx@%RcqTg=@SQbKS|KWY?4oC9Z6ejK})-Bx;u42opZjH;$EH!)o#!7=vI9Als?R} z8mt_{{eW6SkCmoQ+*8o{LTd}mw|i-9TaZ+r6@(vaqUN&0(>6uXD5*79b-GZ45CZXw zYA7ry?`rn13KiXfeoHj{IFT3)_c-kkNApeTbo}z#0dVScwUADjLAaYynE+*ZtaHEN zbpBo=H3t6r>n{KXD7o$nbl(|Ly1|&mMdeUK0DPNaz?sQ107z=N`Oy?*=N-?7f3$Ow=nd zVp-uI*XXI!Xa7M$7SRHGPL4fy&j>%7 zs1I~|dX^Ss-U*Rdvi4-S?VOsf-LIvVLBi2I!AWK!W%KBv*fSbx-%Q^3og3%Y-4EDp zLqSzuM3i#CwG73bazPCmsT8?yw2iyiHu*BAv^Ack8rw^ak1q0+q9}2X#gc3icQd6$ znquFvc_SqZEcSv=?S8bwj$KY>en|8NDuWS?ZB^Vu|B&NrP3Q7wr1KEj&mww8jYt+j zHI!4*Gm(>aQZKA;FZQ}MQq-*ZaDSzuqmNad5%v;)??k3;qKK+Pr-@>DYF z_ElxlrW=IVzjpRUiaQz&G#(%{D0P?ZHHUpv8uFfSvHT&mFFP<(4u#*THhwkx>?stB z0jd2yVNFj|!3#e^ymJt^Y8;2;rk@*O_%_O34PzV@mfI{icl4JH5#oFK)(a1RXUSnL zp;v^4!~IdYk)m!s#OBXl(uD0`8l-wWIqxOKnM**8*bbfO8{N-f_?NpsbQx<-8W{P<-XRg3`Tt34<}N+pj_w+F&M znl9f+l?dHTY3FP*0wNHoW0MZfxg3LD5S*mwWh`X_g;u<4`n;z9pknP~q_l6XnFVUd z{9;-X`OHF}(mwY{Ic2H(lggUTjh5wKq~SZ-~n{=iJ-K>$-7W7q0h(`@W%o$+_HVQ1Pz<0j3ks`uk zYg$I*O&^1L=7w(-#5Ue&-Gf{0@2PBoq;>H1Pvp=VpGWvOO_S-Hr5FMf-$}fey6nBw znc?45yoB~Xc7y5A^ugK6kWY>c#}E&4j!@?VZ91TdJA`j1pf)Sw`@rY}V;ni4Y3UiF z5<2c)JcUiTi4qhLzP39~%S;G(GHV$ZfaZk~iUv-DUtrTL95JJ-PS1 zaNlJk_4UU4>&v%M*o{G66@Dab1n`Y**)i{jKyWkFR zlA;2bkin=G3Kb#6jT<@-YhZa85!g45R`Oai#pUea3MQNFPr9fFoWd=P$f;8(`&ho9iVe#x#CS`|%#*Y%appA0#BntHIu4WK^%u_ZmwB3W zH&Uv*k)pDlqhNMto|B#Iy`R>Y^#h{~8}=9PrjnF=zLw-V8)h_r&e=ddo(c4jt?x`f zBK9V6OS82o$G)q>^gmH#5US!wpRb3yNWZC5Yv!5P^x-SUmZT~qWt%CPS9$nwNwvrk zt@#v}Rg1G#B;#Y;bH@LUMyim0qP@!pO?gXxW%_+Q#Pt&e5h zc${hNdi6IlJ(;*U) zTKO%vbfdI{oWNc!33I?ypsQa;`h2`R%)*iDP2|Y&mLZfor8Fdd9n`u7RjNr`Q38GQ z@Kh%!YL(X#GJqufQbE*fgf75P#XEO8drPvXm%(=G~snRJ`xY;sWNq zWnC%HTq`r6RG;#t&*r91hnYFOZXp8;EUXO*)}s)Oze{Z1GZa&)9AMbfyx*otCilE} zo{*Bg_;=x(X}oD%ZTDN{=-E`!Br{_0QJp%w)}=IOd_3%I^6yx%vj_l4`a z03ysWtua50G8!^%19P}{D4+I_WPm(>^y+A=|X`256$`lWMxDW8`P6HBuI-fYzYgd>f^Y>1*{Y{PuuVmif7& z=w$ZufUOw~e*!%cU+lfq3DF7FmBhkL(SkY9T;3;`F8%UBlW)zt`%I4GQQ#PJ001BW zNklvVszK+P+o)P= z8StStQlSphR0RRdV%ehUk1A0#_z3U?ej)8XB#IyG?&^YiR&x3wKBf;Nu{Y2fbw3_q z$>M;D_`ecB7TwDWEqzIm^ElLT+gRwo#U1ruXr#nZr@)p>iEg;Sceurxz)*iO5tEq_ z5hhJsI{_531g!-T&uCS7V+L~530BPmO?NXy+sqvwpa>SoR-znuMi=7|WBTMP?!=g4 zL4e^zG$UQLqRWoeH#r7YHqIo8R1T5wk;;7w-q*rC1-AIGK!N*GeEpmtyd6hwQ@^#W z0GM-PoQ@an>{BQQX*dpe7x(t_2U^pnE)#3sxThB?*BjS$rJTkxq=MQU?V6cA;KrQ~Ia8^Y362ua|in?%Rt>WWEXW`UrqkjN`;{ zoH#;v-@DW0c5@Xd+LqoRhAs=_w4&Fw_91&(F;%~yVuj&--_{^30j3}9ZZx&@n~tUp zQ*8p(WFvtY%~fEWFp(^b{hFnZYdua7O|P2l&xd?#m0mu-Tp5ODH!XxurYZ9-;9E5& z@n*^f2R*XmNE6z^%$Q@~@FwawPn@rpQK{}X?)%!3#EQDE@QGys$RYWnvbnY<2Pq+4 zYlB7VvxmxLlLO4+jfMETPan;? zZ9+%4+`o}Ut{;LPevIi7Lv1SRS z4e#l0do!Tcw$_~Z?45o0F&xO?(~V?P>_~+iN_}hE{>_55>23^Jpfn-Ix`5Iy$dwlG zjh8^b5e-nO(w!z7UkH10Th!*w)VgK>4nyqe{djgHQi^Q3Y^ z<{N7c%&X&mGcH~jY*Q#=fu^~7Z%OUU&X6$&1c}3VoRr!XaG#7gy^eg7W7pSm7}0X+ zSMz5l+)rY4N+-7GXd{R+e*86YE#MYQn$pZuWFJznKUbb3v~isN3>_W@ZK!|SNZWY> zxXcl+d4_0&_)1*o zE$=BMTZ#~t-!t;du$g0{_Q1%Bk=DqOs^Nsgq zejS_EPQ2eYe*BLE=WF2jF>wAEMk%DWn?+p^y?vv)8>tA7HR+V;vfuA$q_!VLR@F{> z5S)q^Pre0Hs$isT6{OvKwWeu&Ut&!Zd&uCg*W)bkB^xOwpf|!XOd7{-T#j%ap8J@L1p?>J#xsbQEg-sy5=Fr zh9|dXip82KJIbutimhqDJxtEvQ(155_m;TiY6NaQi?KZo*2o%(?SlfO9V;5C6+R61 zLcvz(tc<2ciCR9R)qoMFB>UY9-u1w!5rFrZ@R)S5slOX1FIX1l5hq2T&e3nYNQCj5dh$l04I3jr^t`puWiQ%V?_Ga^a}y4cN>T`Pn4Q0VaHhsvh@QxXyq zKcxK2y?DeH#v6@PWB>VFYPiBTt2>lDxK89%%6&& zOfOQ?zbMX~Hse|p#Ia4TP4d3p%Mgp6-eTHTyJr^LA2d>|@{ph|KPzc@P_L70AE@Et z&#K*@qTV(kj1q`HDQsKfNNEuCXizyEr`>Uk7v4w5_0{nn4OeGe@8)ocZp1w-V7(!e zv95;sCR`n8m}c}F9#+j!G9f+J_N?4kZz7w@16%$RA9j%4BCig*G4SvW5l zC)u&XNQ(_-UiwRHdoR?0c9`dl4Ga{6%Q@F$OJV2QksE!3_cYtOK`cIFT@ZbP=>qt& z>?+-jlqh@lSJ}Gfvvvam>>RL$q&cC_8mVp@DB7bR((WD{F4SfE1#IwWokV~8(hPKX z{F)`|NA`so6$K8gX+{|)()QqPTGDeH1>KZjlfzlVsor!>nV528$&D%YZ0W;mz!4kL z6V1I$jXGFcW*Eu8YNTS27qW8MVYfm){N(NJ+_!xUQ6y=9Et|n;q!=n@DX>_OvQgD4 z!J`dp3i+G+{UoWiVar&*G*S?KNH8f9cYzOZiT<*bdLVI%`f0NhwuLqGvs`X1Bp{uZ zry7uR-SsDHAy(6lS3ss9vV*QE6=Q^dQzIp7Jqii=D0Mw$OL%5#9R_arkkFZWRjWN8!02B6@p*cUYL<{U-*`F zxiez4oZ!S>XghpaWPu1+Os*N$lwM|rPpnZsq za;cCtQ@tN1zi`cc8gt{oBt>UncnC>uNZAf#@~QWlF=-!AiT8Wrz6JO6>u-lsI;$MgxpxQ zh~ji_6qmgN<tadgHpLBckouH<#G85@EvX@ElRQ zCL8Ba0f8(=6#h=y)E!U>wJn*j?}t3-T170QPux%IlS;Hye+|He$lgZT7W_lL z9^viLMA0YViqb&6y;uQb+5t&g35-Sh@=s(c#P(cpt)gE)9&SxLYJe4w5bG!QgdQgP zI*C`c;!t3L!x35=z?kkidyP7JJ1`Hi>!|OXuGWy$K8}slx}cLWZ)>FP1~e?t(ZJL# z&Y`zb>^Z{T8>#S!V3AW9u^U`}G>cdBnzB~Scmb6K=-d)l34o6}bHxT+;(!J;^-Yvf zd$kWJH^+FBHQs7xRpAlkP|#o#fr>XUkV)1jL~eUO_yT(5lie<3BHL29L=ho7Q%(EA z)`L3Y-&GJy9Syc^Y1DnJ*RcZ<+5_@a57+PpJ!(MFKl1dGmLzT8&}-tnISF?AGxnbW zlprh=ITw8fygtHa&5R+wnOiEj<;J}jQwY2~CvsT$Zq$H{4+n2_-$w=5YfD3=+{YC6 zkq<9diA;zyTb=ePA};aV03SQ|xX9as_(us8(Z*2iX0k`rz2D*GpXF9;q^g&z2L9$S zL1bq>5!jVm@k8TNXxb*k3PY@r!g2M-@cksA_j+? z8-4SI?6szOID?QQP?f_ zI9=C``+eiSrr)G38z~Xsj;3glZL36aEn6WknNXSd`FF>>Ca$+{q`q#f+b*Or2LAei z%*u*ov8B!V11IEtU_%zm#GDuI`@;Ke=GMBAnyWTbv0dsm&8mfEA^kk+R+pi0Yk9(( ze3GomSOkG28S5$SqeOf^9=Ct+y>7)-uEV6-2V-*P|2V$c6*cEXkPK?3$AxbFrg@O|PyQ zvRj}jm1Ot86Z=DXSvVx98Tm4Ndfx0>lM%R9_hDE7hhIOmr|M3@TVOO|bh2DDux*!F z3^c*qu;#>Z44mhI^Yy~(x^R8DS^B3pQpCG})Igz2W(lQF>Yjd2@*{Y8c@St8u>nS* zVn{Y|3!Xx&cwV)H1%Nb%JKHOz5C7CZ+ei@x%0|i?C?n@lZlw;F^_jFUUO?BK??fr6 z(T7(xiGF_nh$T2fh_|ea$Pv`V6YAGdPdsvIq==sKi@ND^@PS4n)rVyQIYF(*c11Q~ zT-U;SzcGJ);r{z?y#M|i@1H;M^>yLzpI`X;y72So8`pJXy77C7VlGANBMDe7Ou|F3 zG7!`}M%&=7JEBlOD03=lfEFJE8d0!=mIA$7|HW1UaZWv{WWW?)j692tgd9nzHFh%_ z)m-M&1pSS8%WrGcL6P#sly(0z?Kp&gsEVfM=cNBV#Rr-(owvNJhRk=tGx+yA*g84L zN62Zf-pD4Xp{WBAOz|PyFgigFf=uA(9q(cArqO_RL&1;RMoQ&^Ov1XoiDG-ku?}=T zu((@u6Afy_o%C)zofQ=dhD|{O~=on~pctP5{k>Zq1KDnlTJw$$+l-LvfLuuctO;xcIgb&v5 z{Un}dxbs$RBy>?8>9vml<~S7uJ1lys;+hw(#kiI=QUXa*WN%KQjYf((8oCX6f(%z; zVYd)_i?T%}H*e9LFiU!!n(uSBlE6VpN&Dl+mJtttQv7@?$-J5K_R&OVHvuKlKDRMX<2&wcOQA?xsl~9X*Vi1JhsMU$BkA(xdNmlWP|)9 zvPS)c-Bd*ch__vV@A`Lmq(0G!l=|uK8mW(XujhNI0uh_!sP$CfP>1?Nv317mjJ0}p z?6F=fP>^k8Ti&GQr0|&!k*mr0Go0R%%fHyfprw&?^p4@{-0%W4=8WD|CxNUfSkvf! z_hqPb3)yvKxiUn^65!_!kDQUAGH=XzW6cXPT@&sX5sBg$4d>bM*Iy^*wIEZ>H0ZN4 zBA0HHnUkY>0k}X2Lh_ulPD=*mSN&FsNT0!F7?1J?PEZ@T^sQpMFVKl z(%2qO@O44f!abu=6-UWeM7Lpek3FY95H3^UnSG5I=FCpBi;qZq#l5 z*-2`IGjaeM5@Jo18`{}qUs^mKn#KI~+BB3<))XH57S}g^=W*gbUwFSxeEn_E37ast z0&>|hsc3%ft2I;1j5X(`hUT=)fCrHp4|-zI^9ZAAGPcczh=Wteo;hLKPw0s3e^%q3=8*GP1(DXVXmG9Ml0{O3x&*yWWEo`QePO-7Fu&fIUtgF%zcAky zR=lnYGM9Tc>^^M)W_Seh7H??%d*e|d><^#>Hs`%kq(>wh+t2RXt!8ja`I9~*IeS2- zk0=QnHu|;c5=-j?$v>a6F$~mrJZynS=2tYnLYq30fz%OnTsBhl2OU)|x_~T{WJR$J z%LUQ`|0eo{L|J8hQ4etW=g7Z_0D0Sv8(9uaCQa`RIy$iITJKHKNYIV^9*pYTDie8R ziF-E5$!bI{hAv*+lt8=WpZJN;ex}^e11=Ll1AF?el=Ybp+2BITuKEy{@Mkcu?!hiG zj?wKpAd99@g~{pTYyvQ^>6v1ZjSl6ufl|exuq~Eq3|}u`Ezq(MXi@{{)i7fC^ugZ>C$gX`&f5P5DiiInR06uSyw(# zWRGQuf47gMUFRj8@zG3iay3=i&H^$@ULepFYTLm8LbQQS0}ZYW+DzRo8Hd(=)B`pV zlD~~Z{~*VfP*1HEl}}uM;sKu-ip;is6sOuoHr*$APwNVVe)U-D<5?G*Gq=VNXXa&Y zBSX0)0Q+X1g0!QQ(2UPO-WqGRr4>dKrbgi|EoJ8p2+!|}Pn5fEdN%f%vfC(t{#}jK zf4*k=m?}X31x2WsYr=z5{`>gEW6QIGz}MEMdW5fdYN3tzR!{}=TA0_wycX7N@*j4Y zPfX@}W4y+J*Ky!=47^XB%Agc-3C10{>;kb;r=Ayyixkf)(&X;m~+%m1{I zVzw0d_!Y`E8E!!y2+!lH`Y^&yUHkmW!K&{)hNLX`)42MA2hREmK zYgs0UEZlb#RWQ9eTYVqgbJwwV&{O*_HZ5|h&;;HW7?Cmw#2MT?XgzEN`Hzqw^DN(5KoIE{X6z|Ho@TbR*NIc`?Y=NcGMbOk38}0%o%FS5wxP61H0K`9`-iPT_G2Fhs6t zBBazee64*pB^zpjPf$<^<4(DnPY+22SktaE$DCt2$?m>!|NM#d^@hCP(CY@~1dC!2 zL&pGtutcHMJVU5KsiBX7<7L;)`Fi0v5A#{lQ}&Y5EbKtl945KE54v-P;goN`7(c)Kl3;``fxuh1*vVh z|8+UJP&Y(LUFN&W=1HeDPIB9?^#=6zt&1x!mTg7jSsFQG{_cXrMjJ;1FbMkUkof|u zHGap@F}+R+YO$<{Hi!+HdYuryTuJyBdE;(=qUOVh4TVS`RFwR*Av=4kV`rreDjY1K z9V#C9UUuRi;~e*z4)`{O!YI>uaSTRu0nt9o})FI4PHB3j_*afZJ9P9@FUVQX`pZ^T4}vZ9TG~N0*(?D zyQyV2Hp|sR%LC2p(qK#X*yzL#ZD=v~ESYE>`}!2-V{PeniBe*0Tzz897aVU(6BtK`O%iCbPlyH(W2qdjL95v~!^A zVIm=BFfFGR>0{dV_!JdSjeDJrWI(7ra8=1C4`p(ZDYKFZ+4>Lyp%k^8$Vujq4VECo z6hiqrbGjsu9E0tj8-UiUZ}cn zgnZS#)-+;ZZvXrm$@jK{TQ@n>Fx@*dfjcLGZ~K$BS5^!tWrwfR#G2FDxZGXYMEOCV z+H0?t5?Nd}5Fm!O21qng?zW)@(1b2_@WpbCvEl34uZ^^h_%kxHw8rQ-&J$}byk9TN zLOJ{(Y9ljyVyd6W^I0}i%czSqXP!(|`>ph3mer}8U8j;{BqU1S~cMMN}cdH z80tn!{$L};O!$9)oH?`&nL=|(ardTaFrl~IKv~U6*+hk{uI6e~6Ec-kJMzR(Hb>c9 zF_(OplH2!?$WTsnfp0dRWkdO-WC_UM8=a0izvJtT_4S4I^~Sm`Z=@DjfzAx-CL>c~ zP#>T^!2Y#4{jqz|JdR<(xepg}$w?_23{7fZX{Hh^BEdVff}urLlW+4xTzZ2Xj79Mh z4SuDd1;{4PgBU9*;Ku=OB`hUO5T<(!&0?d!?X=yaeWYES<#)~B<$ykdLVjtX*Lf6qyus~D!t)#p19}A=E>xtf;n#=EL{#VTDT|j zI?qN*tQijFO9n6Lvx$N?dA3y+Z~8n7GB>>r_Hv1oYV(JB4BvJqP4T_r`($>EW*{{l zZo}WaHa{%{Wh)i67scV|Ua0UPvF^x-9V0=G$kx573m=~nRvGY%H^8A9X-=%|+*95+ z+Av>~aQ;0aDn?6SFab}6M1%;3AiD(JrtP5&Mk3xq${F>b)vi^VLZnoay*6Rqa=*xV z4P}zW!eCo0%2KgXZ;72sTw(zpUMOtIpNQ3Sv7L_K04}j4Hmu-BWWzV4bgS<(J3=9} z90FoZ)bvIwsSM#|^xwnu^!#%5YbpJyd*r`)BlU|gTEG2|=yL%@RoYkD&PPM_iB?sp zHzV_2XuQ0m1+dgG;a#@;CRd(YCKL(xHpyx}0k-YZdGyCN>NO7A4g?vHV{_FoxZ_A< zu!ld2Zlm$E92}X;R{%mszL zZ1Eq+)g(2#JSTqb&K2KrW+ZliQ7)CZ_Ol3M*1vf&%~nmYCERgiu_8h zeOKLBKaE*wQ7q7xbFI zy$r?9Y%)h!6V=WW?KsfR6MUTL$Kipk1O2c7*Fa51!xCm~@y)Bq1_KC8@{~T=x~0wR1^`_6{YHxVGq1D@p|7QHEL-E**_x-mpK zv7#~AdCVPy24oQCL6}Fwx)03j%x4$m_6?Dwbhx#aE7p_g001BWNkl=pzLj~}@_m@Gffa-ocVC$6TUUzAik%W5}LNNd307?_-PgoyJd#OeUMGl==CG;u_*L1(R%3BTb>xG6Y%+ z$m*u+Na#j20ugN!0yQ|p(YkM-YzuWpBXtaGq!`7^0}O8HY?&+FiC4BbGC7jCA%5N7 zL`{KCYXW61%*((N@5}r$p~tri8g5dRlmak3&}~lhj$W1xvrfkSlW-k`_aKaxBXg50 zHbNWC4uQ4ON46iJ&zFA4{1kg?K+KlOA^ZQkli;g9*QfpT;Y%nFk$k3P$l8_?nSK*a zs!_rJ1&6Geo%&ScV%-b6&pEj#uDpoX1%_J+T?5KQ~p!r7X7z29R=lS~aN{Lo? zu*b8J61$HwFnVrjUu;`K&Zm7h0Ik`wjdbU1OX#9{P=KJ`Xw%lo!eDWvA&S#+|rs3N94 z&pjJ1nra1Q!x4Hz$?dXjqm~e+ONgc+V6Y-+lAj0UW8>(XC@N!iZa#Qv@wwUW`&<9S zBjxgHFRnt?r{x|fI**k0>AskEQSDRa6oW?-_1H#zY@}GxyG2TzZQ$%DbPdQFko&OB z{=97?BokP-H#SlRU?xZIqkCZOiE(sms7A-=#(ap4ls$;8nL^6U`d zKn~@l0=TR%fDPgfqULf^*6G3MiXNXA6OKM`^nvqK{FuPCfEo2fnEaRE@1OsHpMU=c z8Y_qv=KM++S4jKjT42w`v_{GRSJ~X8Kcx6YSyOS)sDpTb`2f=>COCKcI6*DqGIBDj z?BAmS+D+aU8!7imW_`zh&`2pB(!Oj2BGaw*X1a0drY>+ChH6GkG$NykHpiTkgJE3F z5mAdVIN*@>p-B6f9W-IvsL==5qigm+-K(e~&!Y^74==UL(?#~rNT1Ta{Z@&qVnW># z^7pm2^gRY(=DTeUT4L1fX{g%9)N6;3&_~l2pbUzkcPfa21HfCY@xeRD4VAbniq$3OgHSo`KLtQ9!#7Za65Rjwk#2s{OgRyJ5zYQ}hpDZR<3R^{)d*_x;AjaojH@un!gBFI9^!3~~?;-sWK zfre)mx#^aDa!mqO0f`>~^M>35GACqp*L2H+P($BHGPfT3(=+vy|f6_^{*c8HKIIVT`zOaVpd5Pv~I~sUuZF^~$ z?a%18Ng)cL!7~3s2zA(D1*}=D5;xll!LC1AMG8h+XG}zBeK_DEXn}`^`tBk-0)7c~ zb5u0Qp7%l`P7qWNEi6da^1qRjgd8wP`Sr#{+uyUYOGin=!kt{i|&H7&Dtr zJOhIn&3#e|wahLLKkj|_`%IbjWPZY5At>T{ds+e}3)7}0X}l+VLMAx>-0g?g2la+( zCJz-F)F+ngn_1BxHTVR)&i$z6aj$EMB;5`P9-;cE@>d|#03jr;4$P1L+6*5yN|X}^~p$G~V zKnwJ{Z>8e*{Efw_rl()#RQkQsCKv8*Y$^xc}5wHv9}sJ4bSZ2LXVVU5&rxbINW*`Q41I!&qs$H>SL%BCSG z!Vw|iIVGhD-!JjZ9;^ky^lge@tp?%=o^0EmF5qtTSndj-Tl_5(bXY^UWap19jqJk- zT{q_a#=LF2#^pt=fmg_Jo0Ij(GkBb`r!|XgQ{?OxcbXYTPG{RhIDJxjNhztoR2wO2 zAT0^r48ZUerA(G^@j$6W4uPuHF?g&y@}I znhVT~(QPnE0H{C7$Kmm!$vKdV+6KB&;1Q3awGb!lp&ICk2i50LF#sOq&6<&0-ODYm z`Iqo4X$kCLd?I2Mu^IRKO9YDqQx}#lOf|ywa_V6QZ^Li{(}{QL+TWF7NTKn#x+vjJ;(Asx<$$T6x^YirZ9uR=; zTMnySA+tydrxd!_o}6P@$YGk@noB6u4Ei@S<_Z9X!6pE8c^8~e1-y=i*V*vnw6Evc za2$r;?k!}8iYjm27SLyoc81ltX}NjRkkE*lg^F~)5PH}H6(cDDYrQK}8`}v^z zhBcb>9e%P$Be#Q@yg8$sBF53Up%-O9(42Rqgr%#<7_7OK?%P#@`&--`ew;YQY2lx~ zm9l}!miC$cd(+Tr8hmXe@_r6qVZ2q5=f$TWfNSu+<-EsOL|xs_w-?>nMyNYo?vHO1KDSfW?4pLGFQ%k zU6;!>w(PpYAU5OZ=0C)=ZLeoYDoO*DpS@(vgZg$nNNrLS465alOKF*}GqCJEkfN-~ z&0hoeH(DEDS0aErBr>n41JvEau?dvS=Ev?{ix`~9wr}PQvuEkTi9YOJYK_egL=2*y zGQ9_tFhDi(y?^HkQ;V93H#JZFNv1{G=Mf(1x%tq%hz^MS?jtj>`~DX5lLAdHnGRoP zLL~~NF9^gza~1IgYLIw!4{@VGuuNr(od8=iYMU+FSdV_7x5Io=ykioim3Rf%#l8ti5QYVk0%@Lhs#bisRhh7xa+r6YT}TV-OB| zhTD7WO%9_diD)Y!isvinnV*f~uVsh&_2Ri^vcYl&ASty?o#@ohF_sqOGJl;|uIpB>HItxuaj{e`*jS)@c(}+=5wZJfFsd2 zc!?@`@*66F5BatiDMu~Xswt4F5xO|y3SLKwXRV@gGj@{puMNf|={X}+{F8La-aDEH z@>U$NQII4N^%0U6+oD*z?+3+hU?d1n@B)nN*XmxT*2UCt-F4JTiQ``DNnwtEIFFG{ z)Zt;k9GNSd(dPwU&?$G~y*S2IRIzfSVg#l|0uGTdSBEa1>+{&i#X~01u-Mpp;v^7> z?t(%TSP(Q7izyJ3DUfGL32FM&V#}wl770*5Pm_4h@K00|Yz^6!EyW5Nq1H&yhyMI2 zSBpXbxIjn0PJK(ZJ%ZmAn>FH=9-AvF5>@=+!nN{mBWioIxG!T+w??swFHU1&rQS|DU<6^vvWlu zdA~*Wi+U5qw4bN6*Ox^H_@`?89c8j)lMHLlRqCeT5QAJ^a(QNYAc8ki>9O(VKQ>XM z3yM2)4s)lF-$u|q#CC84v1R%-KdTxKhckq~uNBOB&0AtWPoqf_{~oeG|GJgZO$kms zbB#<9WTPGA)JLE##KS_Jwww5Jr#cd$^#iS$zpwQJ!~JH>x4s5Et?ZKg6Rqz`liYc^`JB zCznVC0GPA?b|V#@#Y9xZ42gCv63x>l^X=O%w;Lkq<4L-(YH4~lbDMY95o|edQDcDF za-$+gifxFZjll@nFqh}|^O;{aHc|#RT0l*JWs8zuUuy}j>y4j3zwq_-g}=YPaEJ7L zMKdKBJv{NqfOTemoLfOmenwktEWX{3-avx(?`JBWV)KPl38EtUScmu-{TK>{lBUQr zts$2W5D67T|Felwvcc90DY;rhW%*a$z&lQ)ViW6i!$D323jX~6GLKXZ^ZMm4_aQBL zB$VjCU|7WTlQqLtd6~?D;^Ow-FugOsr%QuGe6wIOm&SxeMkb0L!E%8NH#~a7adf<1 z2VSoO$6@k-^B5S%z`PbNy6|2XJ-e^l z^t$(k^JM&}8>yptBW2B2(!B5_lhCJsq7+_Syd#n2l&HY6It;Nu36 z8qiGoDJ0tEsy9-)6d!KGKf1*G7&w2t@?w&Y>@ZP$JiN3>821A~#B}-MSo$f;q$e;5 zis}Tqk(QtXOx6SwHHfe$&J$*HBy{nbFfq>4;0xzz;k31pO5g~VQSxMNSxw0_TCt%j z7|%FdJOtyDh_ta(2+%hR{JEAY}T$z<8h)RWe><$TX;J3LZqS#>1etiK6(Kv3!NJj=Wf zmyrm`pEptjj6tcNqHx4UBb%S4fA&-+bo7nO@QxU9rX9Hy$#T)nAny2es8Yw~XR@osfV)7uS zfyEZ6s@=9y=qISzMhatqMhB0>8Y;2_maFt?PAK2jO!0)$4V4a(2_lzOJGda=6V_ab zzi&R!fo~|tuX}HzM9IRcy}TlJ2@~(A3~*_t!T{$JXxm6RRaUKuV)s4pny#iWb%jl= zSAsNzG&fX*8I&HZe^L*K2$o-KL8gqrV&cl0j;)P*Y=XiS0}D9q4W6-w66UN}k%z``3^ zpgYen(Tfj3fT_ zJ1KmH)}$PK|4h+Dks>{Nl#(SX?^W4oiRNZl$QvlQv`?o8fA4%t+MM%!t*80}C`af2 z|D=8YVuAIGVypuFC?{rqh%CYK9aN2yBfw0+ZUnxg2x|;!4E2sixP;o_We1fh*FKmk z&FUleuuhlKvf~eG{RV~kn;WUuj~|bf@-u@XpbfjRN)2%|`Kp-rg$c!_Grs<_fMsLDG3ix zU`~7`8>-;YJ#M?ysGEy6q*rN9E`BJ|k7=N4N|KfWO9?Ct?&R}4$tj{h3kgyU-CtzP zL5tFCrj`eZmqzNgM(TQBxUV;^_l{%u{A&Seu{wNaP-()8+y?p$@Iem`=g*82aLd$;vQjwD&v+#@nUQkwUF z&ONFs637Vmec^}j0IBX~tyxPm{b5K+RXhNRNI#hUHgUZSaI=NLi${$xbIN47ZkehS zPoMG;OU<=yw1SFA^mO0dpsru@=MqpBT!FA*#w1Acez2M-+GKs1%SV1T!bq$`ho`jZ z=g@F0bpcg6alM!RuafpXy#e<1@fT-1RnP?e+iF?!NnCIDS&IyP!c6-|WW8c!7MmPt zO8T1nqXL6|JCC)IIy#Qt?eOvy^}6tu!tc4yKZFHrm~m z*d01zT7;D_6_$fTpXiKr>P5PrB>yLuG`oy!H$EaIaq&B}ItikEgO?k5otSb#hE4HR zEkIdiK;!`^;kXV@rXN1p7MB;;CZ;!^YBiUq_(btM51ft!KOEajQC&Mx3Z82b&wm!% z`4WladKU&gB!x<30EqAgx<75hdDq;V@?|TK1fC2!lCmF&18?|YW{)m*2qCpDjgDYmKL^P@^&OA{8wW&Az{ zGa27F6SZu^Jf}*5~fDB$XjRf+r{u%Sxu6u$#pMsX)Z0>88VAqgQFo zcS_gN*Iw3tdu{TQOedP2D4slXEW9?v?w?$ttV>S5Ah9r@xWQiq2dwD~u+(@>29jyUwiQ_Oys$cKZ=(1B3 zG2XRT$>K~NkB2?(7aorXRMZDY7kWQ&^nrd%9G9YB7y4tM9}l$S1joRjpCInI-ygW{ z9~k3dZ~z3rh*m>|1KeTn(pg^@iu2(6hUsh6gy88rTree?vbHj`a8`Kef7CkuU$Bpq*NTT9Cb>&mpf@<1k!Te z`2Jg=*4(0m%BQfHp0-sb0)Rf}BmS@3NPX4qDeY80O3~8apZj&~ZZd@Z*H$$=+j2Iq~*lTY}bGl1^kUW1pmVV`gd@Gl9C{MjOEtlJ1(M%0pD!^Ka+K1q3EV{RKN&Np=t7UaJ}+gD!RY z$KL%0FcT-Nj+~dEJ4V6ec_Eqaf^y?myRk=*CVQd40Z&XLwfSPGD0iql0vEQ@sUxTuu_2%+)&L)q&(OXTFQQE8!;*_^V7LiQZ{8EN#wf9Ech{R36`joDY zIt}#qeM$tM8Zj%X*SQfg24)K;W6H9gf1LZH(;&yI6zqn5%($Tod zjBvsf_C_k2T|3XlAk0CSlQ1SQC*d&_kD<7G82Mx9B3Ai(jfF4|!=M0w`Si~tkTU*EdZc1vEv3qi|5oj5ezveTeo{Wq zKNs7_GKA6%PQ}mwJHTQ$(vTHC$=no#U(<<$N#yjw%5R*RhDdI^M^<^y*77Yw<_l3N zW=gi;a+MXY!xz}gkWiX@N`iIm@QemN8@M;{;TxyEHc#Am`C&NDCgfI}Qyv|;Q^Qf^ zvpn&!c?zj3`Jm_NB^Qsf2%qf~%%QoRwSS%%N*L-N1sI~Z-yit6KXAWaxZf`$mL3yF zKXIOlvkQ6`wEIAN?`Zu!y-!>p?ZC%1alZyWu3`QdZ*Zu@Fq9T) zdv0GuEMg%VCy`BN8TUHLds%$_fiW-eZF}>= z_#XnG_y5VJNlCY@RQ;G+jM9NGdCq5p)T?Z^EdT%@07*naRCBWw7iVmv^6$6i8>yAQ zkmgX@3#zz@vwNm`CmfDF-%`=fV>Jq%ZIf9x84&7Hn3xhbc501yGi93(j;=$OYf_bQ zC_$7lRnv@00e@fyHJ2RrXnH&|B$_F2plX0ymSdZ)xz2QATS9I~40LRm5V4xun8*d~ ztGrNJ=szJ7q-;<^CMMgPZzHc!Arraoc{9oBZ(fIm7PPIKYL3=Z(Mg5+ljK4%sZt*g z%&TE|s=36(dJM=E9Os4WdSHwb^BSoYXP7w}FkEvqP=@=q4b*uw$DemESe;fVY}>r0i8I8=PJ(1k3HUm{YV+hXwpH9!(|j?{Baq_FZ`3{oA~lHONae$3kJ@`A8ekny5iWBE5hyfX6^KQjbgV7&U{-8il5S4kOlK2zc0QOx{4TKN<}qrj~F=N&NJ5 zD&abta|=};-eg7dG>1=bhc{9wbH%4z6RJ&AGSJM}EVM*Ut8XffJK67hS_!8y-H-Q6 z6#70;q2`8c^+o`MNV*IOCbZ&zzk6ny91Dmo#%k>CySeWLeW9^wpEPfA7=X+4o^I;8 zD)2SzGcuyGf_k9b{l0cEsPk%~GAnD5XS0cl#UK0mr?Uj+`a{>?a}Q50O_K+vO8|BX z`FzUC2Mp4#Ty7_bjj!0_F$C}T3m+fWOugSt+IKqz=c%|2#qCh^Ho-yKvp%0bNb2>a3-2l- zb{`_GI!NN(;YBSwaO;~a6edFNeFKwe$#||H zC_nFByDd=1rperl2j2_ons4H_y z<}eu4bONu0Jl$(xhf+);5DP>o@2e|I3d_$Gq{Dx+f6Bo4a^EV{jRK;Od_5DalnrWu zmt~E}zU^$*Oi@F};pYvPArE`B2<6V4HxOIYHjDE2yEW(kb!p%Bv1oOdn(rxD>gffl zGPQgWYZzpbl@u(z5a|+Oym*o>t+Oj`g+50bSJ&-E7?SyCzd(FG|Bbx z(FKc}@wKd-sU(ZD-)E!vsm{-Mj>v*e4{M@Y+qY5Vja2jeiSS7PPew|J7l^Cr;~)S< zpkdX`CI;Q#m{WX)vBnMRA%64(uF#2u0uVjDwz7V%FsT|H#^*~Kw%B0g7!LrzFKUkmG@Ibqs%gCTLv;*9FGJ@p#z8=y73Q4|@*$`!ouYR7XF~ zMAi0g8vov0qS7rF#HP+& z_$oe0_F^Mdb6?``%hT6e##)zMsR0-u0*U7oz#Q#c=i=H(MWGrJ6xj;cnAqp@N(G1# zTPUJ!8x=PN_4lW!VyNPpwt;%T5B&VN@ayBkulEZ-KQ7!aLF)s(PqgDi>xzCT&Kuzv z4d*#<$^jx?g{JuRG8xn5P9~-1wUHXY7?h3FeZU&2Q5&fxw{1d@YkZ-2R=$zmwHqts&TSgs^9_T$hm}pQIotnEGmRPwX<;y@a1~ zXt)e=s1DNTa{;0u|HJZ~urBC~gl950RBM|;Doz5#g2sGvy2`#`u3(|HA%!(r)NPnU zhtsiKo*JcGDeP?M5HSN<1~$pITWE3j7AdJi8_73O)rHUG30`iBms8Q0+?Oi(nzz=_4k!B^ma`|+F}YzWvO__kodmpQ$;N(3D(TJ|+hR&Fea$v8cpc2&&3p^bD~l z-YKO2=|-w|E~NmMplg1joyPT5l{l^rq2k9YF72we0L)MqoNfjnNg!n-Wm~B==!rRHKAX`< z^$zkJzNi(Y%JKYnMbJIVK=&(eQO(DqlGPXl*`-R;O^va!c>R58eOACJY&FHyyn*Tv zI-nR1fO2W3H%AM6Ly-oO&D078E{)XuU)D%%%@onsKB)v1=<1FBg)E^WBa2H8GbY6> zJ~b*U*67$$49&^CXNLQuY~yntwv~Fh9eBB&=)K{(hN09wgJ_mqo0xi%0{|@Rg3+4e zG94meYsH<8lZQ<-FdpU#rZs7ZX(Q9xEO+GQi9LjI1@c9MP+}9mn8;H~`^bygwT((C ze<|oB(8fV(;u=&FMW;7r+}sbt`|JKps!?wM0t?1P+V|W={RYKkg8{g?G@%=LBlbg+ zIi-$kv!>QWxrc|jW%iKxA$MH^q{DQ}*OdiX>%+Cwip~JMVd$rsjF0o?%~anE#JzbM zJWoIQle*l^*V`=sqff$6SOu|TV_suAMPlZjff{@@&(AX{7dzRF(3P(;BthP;%&j9V9|EE4KhaK-Ju5Uu}vNc=utFp!dgxAMbbk z`1Qb#Uk`jdYzuW9g4PB70FDDVXTwbn3^6PKvyY&>iAucxj2m~z`PRoV_l?xYgvtEo zYIH+0-vpo`Xx=cjDVeRrQ_zN>T?Ta+QILpllAOF24O2{^w~z7q&gzY_K;U@qUZs&* zO%f%UE*q%)@4g9OpLTlEz$WcO6akd-n%t0YgT)~3gz_dtDJR}xO%dG8~Z4$!o zx=Uj)vSugs+C%~bpa%$tmkd_0N8x1znk8QZ1*`$(=UTqlkR(fNuh{>7C(yy5qk%ea zO;lpe>mVU1v@E8*xkAcTE z@G)%T`|IPv4{xM?eq0bGJci`m(*_-oW1weE6vY;bx_yprqCt{1;4gB% z1eQRyknn750#`nkhbbr5yzun%X^V9Q!M0?)E%?b&V{8fzF<4WZc2<#)L7`~oMk-=S zq^nC9>NZH$epz>WO)NI3s(@Ux^joR>p&Txu zp~+-xrj!cGQ_^1&A%b?3=o#2384BsT`3{%zIY#~H;KQa>{dS<=PPFr|kjie`sg**E zC(n~?jaBt7?M16((P1m)QWVh6)28WfK31QcjNIZqPid-GEhnH-tNfh=_}%{#$1vBO zrj|R1h3lF8&ZuPOYg4G-08pM6pK41y!iy2Ur8Tj%{4y^BA$ff=FQat#5XIwkYLn9< zJG03Ei{~Or8Bzt>lJSxH`&}giaqV)abrAVY&bx7dZE|J_^An5#Cf-D4tC+I`C9dl* z@_vRpJZUhg7G9J$LN)fPU5fr|^nTzxPuy-VczJ!r>)RiIDjsd0is!b9_Mqq01eM0T z==o)O&IJe)(+UUCoYKu{VZWI(rezaF`|6(%fyz?0fA(iv<8NJ?8PvEvU%H`uQ&y?G z$X#5*19@d-VH;{JQc)JK)!SBlkv+0@1ZLdtJwL`Hjs5W&c=1aMa^;EhT(+9JKLXe7 z@e^oP7_X#tzkjw4bR(IVm#--b7u!IoZ>OFbsd|?A zAZ)m4dyEU?x-7>DC72sj?mN}6%-%$KVUGwlZ(yGJc?=)+u+tsJR5L2<-%H%ZwzpQg zC~~{w@BQ##d6V`#@TnG3rZ1X&gz^9#gwq~$WFw`1ri9;*9G;?SaG>-gQTWM%OOz2< zHM6P|UWt%zmbb)#scoww_xo06?Lna90gMmmd_cw>Ixo*STUn3SHSpoyrH}iC$9*{D zhVs2*&1<;_B*~A{`>GtoK48kkHMPiSZ6o#L849%V@RWFb%~nyBnXl0 zMt{yc9#8V5g*Uf;l|2Roph;lret-VPMv5vpiq@_98H$9RJeWWVVlOp3e7p{n6Ybr6 zXkyn^ZliqQni5=Z$TYlkC0B1#M)-3+y&f22dT{Pcd0uQb(<9}Ot~x}VUTauZ>D-2V z+#0$p8Zn1=w+_5*A|T+o*p z*sIT!j%72sTg$-zma>oV9?T1WUM!k54?fvLqoF`q_ZBEHH#dh_tpj+a}n|;45BTpdB`AhZ7+ZGs2jIg9-b_ox>9eX;=pENOrC3E z>_%^WYNqzwFx{peYokC}G<Jid5aQ-UbWu-qJsV8t;SI}OI<13JyDULQ80 z)7G{f2g&IF1v??ZA0EGCz;FA?}~ym$dDQl3DE}e+?C3p;nvY& z1JNkuUX~MLlhtcnU|x;X$n8=@qt0dWPMP0K0L4z{ljV=f$IOkCkV#po%7cq4DTZQ+ z*e6EgUMQ#?tx3V~H{YW0g7|ICRLPOZOg=3zTT0Iq$!e$}^wgvg^f#tYoSbkh1iuz1 ztP}efCzx(vIYsHHQ z+rcmt+B7|QQ<@j=|0%Vv@4lHWqe-8DYx?Ot4H=**k+kf#7G;yMXnh;Cq(Id*sfYUI zL(4E)Rm$elGfjtm9!-*X6oqcVFqCSO3P#$RD7T{|0j~@fml6os^k^YBXG9254gu}Y z&rQa+)J(EH_xP0Ih+XYe>9Ww>)}c?4I1>2abuTByCelo^a1i`t)w1f_MCRsl1CUBjxR)mLSTK)pNww=8Tw{EfG3b_P}MKjeL zP2Si&QohnAU-g|w$IFY;_+C%^`Ns?X>(4j*`Nu2%{PPum{PBjjw;Nt=2acnm(R$I= zHcHksYlcbOF3uK#yU|1i{#=1_T=~sNTt?xNy;vpt%~nH)<+&6=drBE!x6GFoUYqlV z?t!N5JQm%kH~*n*%9}YrNq3QLdQy$pX~GV3WZg)qrras~FGL0ji0Qp(7$ggjLw55G zWg)g?AbSY#zK;5atlFm4wUMH|k&0`vNw=(-)*LhJ&8OP7FdNAcjnqt1lIntORO!bM zMVofVqozBeHo^&AwqdE`$R^6memv&8}4dsmMNCp87r7zl%#l3 z90d@w2Q=D^C;E|Su2t(6b9o)0`z%Q_O`M#z6h9&3I1b#-8(v;)8})X-yY`t&#*3o$ zZpQ0!gV5rU6kUab#t@+VS=wlkia9TgYZcCRqu6t$(l5upnf5KY`ZcZ|==gnUUoqcS zNkj!nWJAnrQL%+ZQ#N=N0D_velBtS5H0X-1Kv&?D10!TiAgA-)(2komQtgI*-f)~Z z9Je#6lFU6N)1u8L_bycwWz5L3f^4{tka|*E7GZGT_kdu1?z7jwjx=riBgZ6u%XX7OR3_rzUeHbf~OA3o5r+U z?Vj7HTb>iYfceFmB_6Apg30J4?wc-mB5}ou(2^fEtPLt7rIQQ&KlhyjF~CO-5W)*+ zzRsC2335M_%m?JU=SFHg9tM5*xa0jJKDXxP*L~pU`-S(rP2`Ve>XqKuVo^Mo=j;{& z%$?KTYd;DD-uzD*7{a!d8iM;{;^RK>^M2vy`#m9)ms$U8BUKtH_{ZMJ;!eiIIR^Uh z?Xdy}7)?DSwoRZ-lY_GuT!Lal>oFZFS%n}RBU-|9E54qipe$Hw5{Pe;?Zx%$ZwB#I zi>^UjQYfjHRGyGA&0hy#%F8z!DW^#W^h!`N_i9{4b$l$v40#g;TKC61V}mM4twa|z zSMzuj!6pIKEs%LNJDP~|l=OMy6dS2AMv>jEj3|Ag%Suufqc-z66W{(cZZQ7F4uTNX zEqFPbA@@|(p{@;0e6TztGwW#s6;>G~&j*lOcK%Y0OEWDo}6@zQgZWM}EQXKRYF$JowSkZrIBUSS-|L)wbH~;9C z1B$@WNa@-}#f3z3CgOZTQ`cBix+~xynkkbNc(|7rDaX-qJ5Rj6-0;WS3;yd*YoPx8 z^9_IgVGY&W+beE215Ct5${-7xP1LwnBNbUxvHeQeM26t42C4<^o_(_CD9|W~?0IoL zuIuNOG?Sde&%=R5Kdg~zr#<@ZM8Dk}6U+O9E2>5}9J)Mjr5xuj1*9QU*2g%EVkOIz zIRI27DvR>>dBeC}Fm)sK)fn|REmCs~T-T%EO;HGgl-G-jnR?Z4D5jxqpk~)wZlUT% zYEC~m*VoLv9+&rZ5pb;COYTp2*&?8@H_l0CG~#w(d3dzO6j3+h_abA3q^+-pOP`{T zbCRSv?G=pNcoNGtyV-p>T0D+e+bDTj_oos^H!U>j^6SuEc^D=0JaKz@!Ryu=a6Se1CrdZ;+Z=_aJ+42T4hKKyRoNJv?YdEuc!F*F3lu3gwh({P8 zGp4@9KwnC({W%XZk#E5EMrui!?Ly&VQ0AwMl@i?R8_(VehhiD(y+H2+;BN2}Bln#? zpjlzb-8WJ*3uL?20n0aHkg&%M%hTUql3GV>ue*Xm}+5*JR5iz z*=)LJGd^xjR9xdh<&%-0XY*8tMt(hv7?O$XH?)-VlmbBHNoirlSoin!(Y*oef-W7c zEn?k*FHquxi(WnZpb8;cc~vMrfl_3Y>tF#(z6qOPUIfZ08sWZ)*hq0hYbQG2(Ata3 zga?Fe8zQ(YS8Pl#iZn6WsZ3?hm<+2QoEk{g4@e8DGzKv5fN!E? zSOX>Vf?Rhz9(Vlu^?_f%KJeqGeg5VBq^Tg}z0B$Fk z2?v@#*G`9R?ziFz0b?rRnpnVzhkK>oFWXW`$olN@sgY7|qy`)C#XlYjbPy&QIuUxv z?=ZcQ_P{ljCA-vFJ;16AZpfSs(X_qvN8w8J)0?RMBiv9WKY?eCz` zrkmX+4pOpLvN(B`^)1kbgmF5nWG?ir70>%jQOmUK3TgS&Ea?+ygnql364^IWM0-Kc z%uq4;P29A}9}>aJN>Fd2W;9ZcN8Y4u_$=)k(`0?Ecy4yMhxt<)HvT#tK<&2J z&qjq)rehwdTQ*YHxH6M%%*&tinMUi{NSTMBOtb<8ENd%84uwbpo;JJ5J;}MMf2eML zs=TMQZa28_1ic~_swaMxE>SW zW*e#2PqgMTzr(y4@e$mtS#{Jo`*S*{lSH{ARHZa;-r%X^MoMIC`G5OHYHn05$>}YXvkmLIkF1p_#ITz1qa8Ss z7`um=o9EUVPf$2AsV3;<0Bw8CbLvb5gU{ZemV6kB2`p`L(eHSqGO{2{-j-uw#cMQD ze4w`zy}dwrLa9TWZKcG)@3xhq#4)$jDKDuu>B{y?r3u!sC;UjO5DFL*X9J%18c{o| zuLtyc_eScDIUbO4LB<`Ak2~IfzT?M_cl`C^9sm8~7k>P@@Ynl=zdk1Z`ho z#hAX4n&=%kC*znC*JPZ+IEir(L#Y`#w7KlC338M|YbymTX=@|SLfi`0koA>EX;6K} zo!d|IQ;9dkelHo>`}CS&!L}H6M_<#MnYHo9v5;%7krE-94Pa zQPOsSISm_h35F{85cHd3424WzKRd#f8>PgL`;lb2Fwt7TZeUy|rc0g7ggdi&jCFNh z{MC5Pi*~Z|dKNE2v;;dQkVT6U7<=0KC*Bl24f4O^krJT5CU#$mA09pl4fRYu!n5of zsc5367x(L7g%`mmPpO4g^my)ElEUeMYsGk8LTSaGo(1y0?9JZBu z^AOerFtV`f*Vk9PzP;gmdBJg>)>QG+hN^gk#0{$y`BHAkXP?+zS0N|OS+cQo+j6fB zl~2T!Qer@~wCvyZq8z{JeDixtJQRRKK0ez!B4TXP0ja1MK`0cG> z2*AyTFq~isLhI^}32v7KExSQ?LT%&PE@)rZ2*W6qvlGT-%+WAschq&qtkcPmOvbg% z77uXr6RjOM4%>_!$G^u!`TzhR07*naRLwu0IQk95{G?teKp00yt6Kis7#FVjF#6_r zSOZN1G~E%PmJqF~Xr5S2an>fO>N7?^fjP9^XGD~tk@Z1JTW(4Es*1mgax=5)H{Eg| zzSvB`Js$;s5z<=SJgEQa&3rF_mhvEi>;0b91l$GOOiHG#wmoNpsBJk^4aDH0jg@gF zCJ*+tlx=lg?q^$zfhhw_14I`@F3fpBbYQAA#U@KL0$GHC>vcgM56Ja^KJJj~!y2gR zk89v@39g|gty6}=0z8#NEdYxb2?@48xP0Dn<9I;HHcz622|^o!Ngab6LCy!Xoe*?P zWlXSdsRqt|;OK|R1T(jgJ<6gLetyfztVW~TY-5WiAf9&plA+@X)8C<&$t;7a=oG^> zz_l_d>Aw^bDbto%oxtW33{XW)mex4D^b=u#v}{uw(lfc=-R1Xe_yzFB9v7q7v!S40 zNm|O#q72KPcM$sJ)GObRxhhV7bWyp`(|xE)0c!)MPjuX^y2leWSL-%y@x7SCAE#z# z@rPZfYtyt7YD>0i0sZVlx8>s&ZRo0{455vDt4{;fR2_-h_A%(0WuOMuB{>JLfhckn zTAUbR7oZJtCFT4jay4ti8g7&}3T42!DYQw=Pnp?R(u*e3f-tZc8*$TpFo3de0&x-dw#g{(O7M#~*Ju{PB9j%XwJH zEmM+160|^Ac|k+W$U6NI$@bIrzYFt{_h)GvS6)-U}I-tL9wL4R5Ao zbVJw|KkSiCZ!OR z%u<6-jNze%8WA81gc6MuVg%O{2`t;@qfDK)45tO;N$F>~DX7afj^1&*y(Dd~9^9>Y zIQHucZnqco<1`d~jSQ$#zan#pLz+|}EWY_x-$unILQ;;4f_Eq41OwD?!R;FTjV6kS zz~@oWzWxf*dj6{rfpP;}Xs1-rzh!NxNx58J6Jl%98d26RM&~kU&Dqqvq})NRJ2{j* zgHjDd!Nb4L%;K6tMArt;hx-e~h^37nlQ4xLvzg4L*cO5c83C7p8n(dd6TO`tT5TJ$ z<1opgeLUSqs%22)5^VQEsn`^LTz6dl7^CBQG>mqE`2iaK9EqDwViT-O7-{hUyS|@d zSd?3pTAa~hZFp82ZPeOzzgqV%2P7qVQI_w4D*2^u=u6=hQ6BmIxX63IE7K2YBr+R{ zSo7;KTN6cng4(c$7f0IcT9btD*pD!qUw5)RlqLyHcE7_j%mXH3;Dkw+fs4eNsUa7p zTzezs!J>dduYq}8kh?Wgay^XD=ThuB;y!83o>B+e3F>U>fTQCW1E&sjCY%T3I2w+F z%`erW@mUXL3j$SQ2)XQ)4+wWqKSBKj_tQ62=6#l_xL;RZ2lkOujUh42Q_qNuP~x_- z2u-pNa92>5)lg9sgwthx1B#Jb+Sph!Yewg!4I!nK)=q^aol`y)`J3KIgfWll3shUY zE=!uDbzA*y(W#crWG%#<$me^|$QfD|hU&8}Bw5|Jy=$XVhY*F|t@I@GE&kIQ=vso) z75Fqv%m82AP)z?CJ`lhRUA+L8KT#JH!VQfjUq>o_t4*yD3jq3LL;nJ1p+zNObwZGs zg(x#rV_WFXyp?^})cyU1LBu6Wh5ej`mL_Knk|^y9lwPq^0ueGALQ|HM^BPjS(i{?@Zh~b)v$QmfQEIVv0UriX($26~z zc`S;lbZRyolv~1<8{CiteKmNavPH`sgHNv|{C|Y9F|pU4hz)QN4Vg<~7g}`bt|cS^ zF&k7_8`RiLi9q|bHE8S5)}Yi@e!OF46Q^Gf!=xLzM5x}tX!+&1G5UGpMKqt2Xp(yG zxZO-*cAh6*+_Te;BhQk6LYNF}O#sOQwC&6^vwK_@_PWiaD4~&KSlvb7GQa9cN?*%Y z4b?YweSACCR25FV%0^bxYerGA*T}qLB(8E*&Ac|14I){iq0{4HR-G!kzvfItlTJu3+=E5j=RZ}+X?nY zirdYas*oIZgJBRC6mkr4JS=pI`x#~6dVJvV_`u`wj>qG`qcvQQyAQRjNQ9e=$$fOx zJoFm2vCqv=gkbAe8}p}TM@#maM#0z6GMw8AA8K{`>4j?K>XnSd?dz(htS<<7q`YnX zcYz9hr(7+Y0!L|=EmdryS{vYYf&1lF8_Ixar9Hc+Pr?v5QO}E>X@wE76Wzp#engnF zf#}L%#*l|S=9Lc{7OqWz>co6pn2!hK@xZ*_A=ll)Op84Iy3|N{+Msj*4cSC>891dw zXo3m29fUKoB|CfbDLHBW-ag0t(OsnYjn%R)9ds0% z(o8S&@a;{5AL@o!i%{7}g^$8z<2e|J$OEpy&C#g;=a$liOV=fD3o3>p2M`2V-HI3Y z>wEfrQ&$??wzcp8^sVrSKdU9JO4mR*LbMlfgL7k7skWlaRw2)K40`gls(#X^#Bsnb zMq`(@HZ@7ERCN{;TKK`lYB(GIRuG0MG3w|x$~&vk%l(Q8BRt+rV9!yVg~XRs!P4m1 z;Dx}Q0zzzmUl(zbiITc#()oQ}Bqcs!^OcF$+`=0u1qFY<>>>JzjH-V9&5#;b((ymi zNX14<DK5=f5c|w`62zU;sXMo` zn=k6^W?QM(7yR*h;&yhN-BzK#Ix%^{bX0w8!~7uD`Lv~(Vp2cRjL(EiLbQ;Rs&|TF zBgMTZ>v&K`Yd<*PQ@xJkl_VD z1eRKKU?#zv$#EuX+_uQNrXqk#`Yeo4Hj<&s!^SWe%8VrO*htBOQIw4oyAFGk3_O!< zP%+qXT~tCF7jH(;tk@+Ow1PBaldg6XRRmSYn(@fNnl9rrtnbQn30WVh%UYZHWYnQ3 zTIFquPcV^erV5qO^!w=q_J-+z-h8+ahb6QJP3rnQ|zdB za67G;>IYgo{GU%yJI%M%4sgF21fy9)MV_^Xf_^tYr^~>GeSF+c^m|9|E-9yVyG3M; ziOay5p)IyjvTC*&WWo|Apm<7+4>bDTL{;20mkoYP40;R$$%lzCvc(rdU)?{?tyI}a z#YXKnRDE&xwq_1V2wl`Om-Vp~?zWLK6n#W&FxzlNCZUVT_N3bpJ9$Zdr;7k+wgpaM zC%@7#4<0clQEs88Z=}K_7d{_*?+WI1VLYzYP+fP->F1p*EKJr%5M?NJyC%D!$;2TY zr%nh48Ube~oSkrVLht0~kDo#Qu&vV#)GZ)asDb+l?l<(?3(ngI+Htxp-_NBna2(Bw zS}`R_+$+RonxY<^(a7^4EfOH4x8y1wZoZM?3Ge{u0;-%3Qnrkg+hi+n3+20j2Xf2f zQ6^HhvnwNNY0Q+Ds-$jZBQ!O7Ud^6Gs{`a4&D7V@ccz7MkKw{9D~jAaVq3o4NYN6= zl;PD5)RYt0;uQF^tglM@;-##ftx(*2GBE$YY@`~kMyh1ZEIt1+PIEDEH(qcE zZk*dFEdb})?L~=BTgxPZ!CA;hE|0C?J8nnA?d-UnhAVx&9Y*SVeZilvH~jH-!`sV= z-q`(7rURCl)Nz|idWO`zd_&{%KE$S`G*_`%(P*hz^Dv>JJ&jP!lyd~-j@BD&MJN^7 zj#{GnbqVg*#N(RaW8n5NaXkcM*v~H##a3*k4mWhWcHO^&xJPREM(TP561w=KW<{23 zF#)2W=Ud4;T3c@&=1Y-Wck9>WvSOx;Fr%+!eSm&T+UJD9IB>IEjk9ptv@Enn4iE51 zgUl$1ME7W-!ZS6K{AMQPKxtcUguPaxaRcHdU&?F>+QZ?hq^P(a=xHnRuO2A4=J4h% z>^Tw@9yt2xNL6vj29tY=W45a}$-24lh9SQD~s`CLb&NV7909z_)VI@CI1G0Omj$zf88ZQQ-o$ z*|F%)ewihj1>d;bM6m5LG1?O z1H?CLthzN-++O_8^aPf`<+28XxiH)hHOGZ}KP<0}?TiokB$1k_-(`SQph?!iHa8%% zZ>N1Yp>@J(=LVN|+#cY}WitAeKZ>@3CGH90nMxD`in7j6z(l68Ia<6Nz*MBu^g9wN zk}W72Tny?gKxv{(4qApp&EEa3DQ0ty4V$-0{Jo#_F*WI^`J)&l3L%EDBkO6H7fvor zd3Y0LuWtgGG@>WeJid>C`M5A150m-DHeBqwoGv#tPo9DgMTg+Pv}`z@pf=G+aW>%A z$Ug#Sc=V3gN*&+65vaL@T6~|s<&i4FW!rMMJlE}KptXf6Q7ein`@`Nup*A;~o~y0$ z##fO7^yFXK-*Hks!W*Yy_jRp6Tl%tDjIW(}j4rJVuX}UaX85Ftbn{4Qp)%`SpOgEk z75juRfKyQZ6tr6vKOvt_HyLlMJ z1aY(Wr%v2^6q7!iz-xwjMuTRwGn=yZZtM8l(QrFDUeAu(dEoWsz{`2yj4YbiCczGP zL+p|joa}|?v~=8y@dBTpxGzgE2cpMuW`eMmll>wjDHcnxyZ#rhmN$ z?)S@0_>S&Ogd6+~C5yNgA;?=Zh~>vATY(H+i<@Uj;)X2JPbR`@RSy6DMrxb0Z1`y$h)$n8*{j@+6HILLaa9tOmit8A-#>5fLRN&D=7wu)TxJ+Tr3Fcu993m@Cm!IWR zK`$$4+NnF6yEeir317^(${7+M1`Q9CzeaOFF5ALhx+DSUhYhZ}ZA`SS4JrM*vEs8i zKLQOs)WDt}O&EnyXPxGE^Y^-Y9-nEvgfAq}MME#5F2O6rNLgg1Pn)Cy6|ldnApcm+ zv?-7D8%f+SzjI|$<092wsOP%Y-8ZzWLyDW|c8~CzY&Bf7?h?xg&D}D-ia=hme4QDj z7{RcSZKTWFmZ>p4$XsMpM3I~f2bl>_Cysgp6<&{ipmp0uaX)M`#U}Mb$dS3P28{jh zQD0d?yf*N}0@iL@G*ADHrRpN{0$Q71YoRX^fRH6`~f+PulZ zA#!1i4_x#PP`8s(-Jc|;u9%{xB2o6Ox!+@(U%LWm*8r8GO~pAC=McOMvBqW!8Wf!r zM*~h)bOPR9PrSao;JjI8BliP*9N^;wA1CNIfp+4Ez;TXDNZ@|Iv6|*e>Wh@ z{lc9G2$yA^Pl>y9K3lyL+QIA58Lb0+*d(62f_prtx(%v`)$5s+cWE9+_fAoBNjoph zcRuOZU{LhTdf1u+LXFv=LTrsFhf1RSt^f0Agl&hj#ipgSzT5HtC20a$eyTnT&2(}fLq6)_smo}}=ug|3W9wk9q9obIj&Xs9<5XjX==^=GRR zynrkPAg65IHXpTWDmaEWNAWR+kV`c~n-?@|6n~y$pc_H%QwS_3xv&21g_18@MXKJ~ zy{D}|rwCG5-e0olD$dOXef1r_SrIA_*=&n{JNuWX)XyX@_k`YDhFhW|? zubBa^>!t!M;~OWr#r%w&2}g5H^U)1odOPuQJMs2%;5@ter(g|}j%idj_fptY%m_r5 z;nUEjy|AM+QYM3Q4E9`vJ0n;Tgh~VD^6w}*#KU>RV}kI>_{Fydp5Ad!SaEy7(%A2o zyng?9;NyOw9gN%CiN~)K*M~ihlhHb1o((#BHd46?_C{)4g6kT1JT83PANaUG@VLx7 zA~B`ll>}}|8lNg;=E9mNC@SK7hg9S$*_#m5M(Qc#vI3<3PcP{jZn_yJE8rvb$@J7< zmB~gW^@KdkWqpr_HB$W;I8TQdOwS$*1ai?+~x$11LV^l_V8A685ptb8ilW#sQjp#=~g~b zRF_7anGLl~HjVA)>DxLfT6if9M&0BUVw65@#nK}dSm%OXe!?inq>b6KqRcZuTLV=a zDHZuvhff906R$&{h4sn(yL?%?B(jnGUOLRUo>{T!?d9o7S1CgZT{No*-ApBZW;)d2 z-U>wzNJ(CI((-c&w5=BGF7IY}=Ds&j-YkK7Hd63_%m^%J-+Um8p?Gh&yvC(-$_;4h zp}5vat!7oX4HPqQjgB@hjL|TzXr=_r1I)G=;bHG-VIQuEU1-fsuUr}_Bh&6A#Tb@Y zvBzHIaY`#Qho$65#b0L6U=o-y8u4Braa#&6*_3-3^mk5AnoyoaK&YD*fMjA|q%Uo% zY#ZUoZj^b)oOjH5SuRp~b;RXhzICD6G;s;Rc5g7dd<+v)9=PTmj9*YXAm~QjLxU>2 zq{vCD)Zhjpe{L+WQ(=8Y7vKKJjOWQw1gKP~G{fpm` zpN7Bqk}9YRN0a$$%(P_@QtgC+?cNn?3|^)ve@<4O)fzfV#)QOUkBQLoyP2#|*lGsz z#ZkE?jrMB|6dCM#>XrprH`ZmLxSemi)KN>r7Z3?QYoJdAzLs8UWEAY=YQg#75 zaceHkxLq7Oep#kkns8o3Wm@Hv&?zQWQ$cFX*H@6r_6fH$bd( zn=jCL70c>Dc0(~P#kgO1eBAMPzcQJ<9fbQI2k!4D?!RuhK5iJdhITMybja)mTp%pp z0z7Sd3_QG%vW?UokH^3@XHu17vH&h}62I9|X{gfBkF|OY#AdwNq)7LuG*RjW9zEeW z{~gojKWU;^pBGTHrYF^36nQrOSU`=6nqDdiDI2NB{Q`92I8I!bZKY;-q*4>Tnl zMyf^d9s7`@>6Rsm67Z*+%ug0QC_Pgi#D`iG+jZ5*wi5k3{W>xUmU;F3bI+Nv>uRPK zD#0aBH@ek8Nlu?OIen1eA6^qQ+D?Jts>$pHoF%7}PY8Q;j|-IUx=%S-9HO zNY$puD5B9o`BusssWfmHn#{IR5#PY>=Sn?1FCStwH9n1O{qRlHL>ZTq&8{_4efUP| zYPdRsF9u(;e7^zakULv5MN?c!;2S94N_7u!&cY_ALb??*0r{#@AH;BkvBgxxbr8+e zlZM&Ee!K>aIhFmc4~4Cf+MpT*!%Ps%I-2kR?6JcLe3L11LDhm~#mRAYZ;xh5QrSQr z2hyr!H|782h0(;E4-9;Ow7Cc9@DuH@W@^Ncz#%wvc#w8nKeok;Gufo60ddjWbEV_)@$_R>RtPPR)LPlUC0n-E|>x@kN5j zh7ihr7PJIO=A9(TbF^hcW*B@~W|B{}ayz?}zJJ?L{JTbqsKS_#kOIsSAAqq_{Fh%~ z+koz=@p$rXt(?1nx8&bY-D!74g|oSxOEX2N^~V#0)4jstQdpx=@+T?IqO$syM#+%& zv;uZBGGl|(#LqH4pnPIZwrkxz^X$0Z3=8(Gyx2mGVd|S`q}F{^N{wXzwH9Yup)^KB z1&Lw`gC<$&!F7mf$hlOs7mA-gHBz+E5j7QD5I@IMcMGelGSWv4d8!S3|IYtpBh}YM zj^TIv2G?8f(mo3^TkIfU8wXUvpC>eF8wY8B-Y2WR@+wF$8ms;9nT5p-U(BS^C=4mVl#h4a~8pU!2X>I{vp2?5}dt|9B=_0#c zYIuPtLogqL>+!(#@xbHbj*nj-_;`PypX}1U7hE5AT=zT1{XoBUZ}Mi@NRi{^4OlV8 z+DP4f({)`&o|7o%Lc-;*kDa%i1xDmg$X2|5u`-Q>^Ax*cf2UNr-;=bDzP43;ZzEL@ z{5x)by({;%F+BzQ?ZlpFhT?EhnK-Z5Mnxl~C6qI+3)cXPVrHTaz$9KITQfzySwZ4) zA)qxl2nm2@P8qglr3QO`!4q>DT$tv?3V4y+7K6yHZDO!kpztKi6VcY(M7~gv3Y(D=RRU=|jPN5f@u&Jp@==N;_XK{vmgtjgQ*w%H zrLvF}R{`qyw2@Nz-gQ|vQfA8Z9igBQFnvnrKUloaIHta;`?__*z#Fwm$!=C$Xi!;j z>DnTPE044Sa(tTCyy$3@=TuVnD#jnx;4Hb7j%<Q4>fB5s+a{m~pt8{Z;rDSwFSBBdhITQoM&J?tyICR=$llu|5K_#D z9|&n*P&rF3RE9efwt+%O`@CO5faGH!RX0+RUA7fvHEhPW*D>S*FVpsOwE4HCeFeE3 zmCS_s9)!e!4_67=tYOHOt!LE9nOo#&qHw~)3%{U2S+$~Nc-I9*AB;>H-q5> zF_4Hb?(euBZLiQ;ZKP}C~2JFX@l zfU$3f-S-yTCapy%oZ_4EU@zXto2e9_J~h|+6n9ej4Dsx0*{;UWn7(~kJ~dL58F&d| zN>D(YXicE+Nh{^_^AAItk4nK$;zruvhp7m`7Er%Q_^49qzIqPW1wiCEky3tq zErUW71x9Z(w;2EcAOJ~3K~xOncX{869a>wKw2FkZ&x6qufGM^W{%Uje+)TxAyw+DX zWOI{BDz9Q_qUp;g#Zsk-LMHE?YPf8=HdD{izUp5na(ttI`bMit`$*RY&s7OuZ4OQF z@()S->bjCVa8{L<`Aca}S%#JZggh~w^2VK}#BHL$1K9oRabukRyNgHT*+R!81lFKi<^+3C z8A!L#jLz-i56WNJN^MlW$X5zB zxU045MyeiwzrT^9)O)W`RVv#i{v_qoP3Bd=85=agNrz)%__ZK&xL;~X`w}?eRHo_8 zmlZl{nBBpcq)yATkBtN(4cZ%k9wsn%f@@{NrYxvz8|?>-r|XrF2> z9IiK0RNsG8Ox;CB9Ycd?z*Q4Ykwtjj+|jd)4~evF9HAiOLGY zb~pq)H;S5X3;c7X?)1R6nxlZUG!O7~DO?6(DkZs)QXXS6!jSY;(8S8qv7>$ryVlxZ z@YNhoJ`9j?HJ9uOW}BG9-p7+;BNY;7laZybktzMb#I(zCqY%mhv&3NdbDrLkYDSc9sBO@@CyhdfeKIj)1>LprqC{_g_OMbY70s zzkoSB6t@B0GZ{mE%8PbM5gici2Q&Jd=*NlkIB`2qT(=XC#|`iIiK7d88+O*I`3RMo z1@AIj^VG!xvAZ}5eL7%M9sFV0QB)+)?5{z3Vc7@LJ{-vC3i6FqMu2eSp%zgYKM_j_ zPm#}s1))!k`LbCDO?Ic${>QFT-?%O)3ksk`SysrZmUFGr{3EM@D3ZY&$APrt5B@Kq!X7fromgE1oaIt$H&x zu3_XGH_)*XhEhtS$PH4YamcG5wH~6{oYX|4yO@2iXG&U3L?O>b-dsdb| zM`BYj_}{UxT`TObIdDzJH3{z@1AqPZPyG1n7k>Qp3qODS!q1=Y_;~;D&D69ez>ClN z2-LTdXE6DjVAkTPNcR$#?k=F*TGb7eu1P>_Rp<+8%6CxH>^UuIR&6~4F>R_{bp55# zkBzCg6veWmF3;Jzw)J+AMk7?OP8eb_`I?180o)r}$*g-v`;z-H!- z&n2=;B7|AOc+Pp_5XI9_#0yk4d1(IV+#K!@P0ngyXp_}NKyZ!-TcWAVpIg#hxcl&mHzDObEu~(qAL!~q)1;a2#us}-)o`^5?iX4g z2A}Y=RHBg@j^B;1@5+nF;tAkH_@;WZXpV^eXx(VD+*dfgrU%{|qJ*U^_z1bLDza5{ zTcv9bFYy~?!4v~2Wf2`gmp%?&e&;1q68H57(vtQ^ji(@LX|3OkxJ>)s-wz9 zWBTqJy#njsuk>(h%23^ufQD5!rSjluW0mjv zRNj0Q_wk)Glm3MyFq&fXMlLE_MWENtC5^NjmWrN(c*_~A#g9{EZMnG--`vcSH(45w z`p~}t6V{(oeTtGfPtU(Y5?=KQkAX!5Eb{MqK8LCzXp7{wj|4u2UrTpFpQ@sJ5N~>N zxwPG~tUGqknWkjn31yhFqb=4)ZjuU>Z-*_?2Aim+_vh_rrB3aeyV~UImhBby9!_~W z%)0PTG*XwZ{GyT4Cvbpj(HNDYMu4Q`vEPVUi-vVSLY-!2p|;IEQLPbP&l9gVZ=zmr z_~Z44+i~DL8ak6r?dB+`Sf8TNn)0cVgCYQ$9?nW+a~=_l$4E?gG*S^>J02!ELS#M( zImw8UN{6XV0JH*)5@G9&R1~x=H12>-)0o#riXrF>oCgEia61@BCp3cLs%5MuY8YWq zu7T@Pd^{BQ%O1aeUHI$&`wM^l_YeH_*H8TQ*H8TX@e99x?a+zDW7kGXm;6H&gEI#n@VR<6Q7~UK#x(-89u^yj)Zlw6BX0`Xk7$e_*;~Ef= zHC>GjMLBfYQMJnU-)AJ|IyYZsC%FkXNNqgV0frZih zVH)^ny+X1U4Q1t5jI$nJ$PocE%krA(Sik&{zlPP{g0 zDI7dhhJ3uy-+q*S-K*qbS;4XiGbW}JkXJ5w8~?RpQHisYswn<-Dd zUw4Zq`BrJBT0e3-64zJj9c?tU-Z30J(fdR{299gsI)*n-K42Q&xND(RrSCEtsrB!> z`-rWHT9F?S-NDT^eUYn65iX5Fx{QsKh=s1JXXp6fFXZmeu#)c>{O25)JXT>;0`+1b zmXfNJgLzX^XqcA#;qei$L81hdgk%qKCEd2Oy+Gr zDcQybG|~J%Z(Csxu4j`illzj8zmr&_;>V+5;MbI&*v=pwmxqBXl&+FdXT!F~{R8hm ze&Oej5B&Q1!2A2auluz7@gQ8nnA#y54v34F$A`WSick;T6&;v5KstQeYR@bAP<2Hr zz>(7n%4fD8&97^X8Dr|@S)J@?YX6V+LS`-vc^ zMa`ps<8h#;d+;Z?5mn)k7@S4Llvu=8dpb%ANyavUZe}dTb9fw?(MZ)ZTcz>Mw@&_^ zQU_2-&?K8ncA>>@r~5pm8a#ZhY4^Cb4W$AFqe?%0$||dwUi*+{q1f4<5`;~EnzoYM(?y)#~%%=dab@pe1$`f|hRxl%pT zo#!${XD*AbX_F?bGY1tCrF1q@vTvlkcpc+{j0s|3c*?vTWI>t`pjn|q0nUj35kq`$ zAb!J0`L{Pwwq?3XBc+f=z|kymy%D-={%sp&(=eUlvcQ4sA$WWU-tQCd9|J#rUik6X zPyG1nCw~6*6F+|Z!q1-{`1mMZsS+wX3pUz(a6vz_^(;z9pY%Rezh{OxjZ=cPD8H-&nbqGQLq`A*-7T7~UfLD(VHlkpFU<6ZrBNe}=xGYz7F$77h(Q@e! zRZ?VR?SMhr9RDDgw35uj6#Jank%eu+vZm~bunl}fqF6Id%YStw;=D{5b_v@j}IPechNgGF1(v%ur_!NB6*Gw8&~& zu?J>I=ap+x?$SyjMJb!8O*T<9?q!B5#m=pG+D2u7qptjnijL4KHPWYMDt@;>qcK?aM zs>&n@AmR-I&z3PBS8r(Jz~}@08W??|A3oXl0Ox3ysA^7YP_v?LbkOdFQ){%c6XR!d zgxPZGykbdXZorhtPtr0i;tiQkzs;T!v5q>5uN$t^$Sgz@=_sQp609+!Q`1#|BkM|4j9~1AV;Nw!9(+Vlk z?)yI%$P*Pgy{5NJ$wGq#7f3IgBIO#dIc;JuXh}`QBU;%BYu$+{-_+^EXj@_JK&fS# z$ZK@@-zSCC7U2EKbpNG((&fEG0e#?QK2`sDQB_5~ zfbzl$8x+IO!~Q}VGqklVJfj%os_FqYML@{%CeiYh`u$9F^XiH&jJ`xTa`8V#K(9AB zRx!Wd57i@_-qhj?t`R3x)KU@fZQF4FX9}q?Ca!A~Yu+N05P19WvtudtF#B|IbvLl` zbE>Z22pa~K=p2VkJ7|7;JQ64o6RA!GT&ZGbG(Wtl)-fLyDG;TxyP}(APF&}7PfVFi z#rW2X6*NkOItybovN3aX*ru4*)eW2(%D7m6YZIO12Upi>&*9Q6MPuO6ZBu?6jKigU z3_FbK9Ka|v!E0hX2cADK{Cpqy@$2X+gi&H9c|N zWPDNzg#dxrVUs~~mc@W>qEuhVy%08&@*3&RnzHR9;iZ4aOQ70UrF~II(PsX*Z&ti_ zfMoF&Wm=(ZS4^a=sIqY0xu$CQJE==4?YU^@3I2G3xTE)Lc~)Xdq&Bqy!r&~~BzSZy zqPlmk4XbRQJrTK2r}(c_+yg~Pcry@-Jq@e#?L}2U3+D8xClY-MRipQf10ohk{87gC^Zxnx>pLDT?NBSQMGp}nEahCVw+Orz$7F)#Bv9@Zp0 zhLf&pAvMH_vZauAQf3(YUqa}c2TM0zO_6r<1C_wYJh%znve+n1>AI;u^rhSXC28Lr zm~?H0)DoP!kJyp#N(+oKfkYd7quU`meweR*lJq;2%IV%ImX!muLa=MTE=;}hz6-yDc^2Il ztD%%Blbqc4Vrn;CI1}`NmXp8Oewky!P2TVEd(i)#HKP3Eqt*1Fy95|tpGe)a0`E69 zIemi58Cu|H3#?{mOe9R6UMRN_8IqF_zImsM4@`^4M|Z#M@TpXy!b>3)QgGEzFfhe) zSW=_Pv%MyMbdye2{rU!$`b65NWx_ROT60y&KuSgG$VJx3bf3!UChbcKVaA{F6Y zg^I??nTxQ@z`uZ=q^5dhMp3ECgPf%#Xlxz&M{h>vdvttzJn-#t;O%kP_s3B(av})U zizpP@)y6+TQrc3U=-1&nIl6&?akRaO84h#bkBFE1=BY ziI)UxFvO%@Udgiu1h&SIZhz&sw~oid3d`nlEsEFe1s|RlKHg9K{PDz(A1D6${={E@ zec-P@zT^8}Kk)PWJKldj@$vD*H3VZQt|1HIU8Yen(iQhOSYYWgekQ)1s=jBx!aud- zeX=AO(XA<6hZjK3-t9M|V!8f^_2( zDLPM3`!Et+Z|PNH^+H2J{^pkmdLtaYVMHO-+2ui244KrAy(6Y6!Gvnw2Q7UT1VTq6 zhc){|q4{KL6?CgWM*1;4dnYy^6n|93UEel5TE`o^r&Y2M5W^QFf|OlW;6;GHFPR;M zf{tZ^3MZkCZkX_mBo!V_-eql>7M^~+((x}CTwx;?s`>=slf$a&*9-htq#`G7eE)vG zw2pmZ^DiC%5?y&44js|Ewu7)uWf5BV^gXoZrw#)N(IZ_;(NdNHC)rBeSpze4OZclxW=I2-%m!9j@Bk-pJ-#E z&nTc=+ULboM%C0Xaa#LVe7}(1J5&L@u){%@ir-?vVob4NTiGf`$%(VfloHdB;cOE> z`pRBzhn~&zn8<+z9uYO4ObyD6p-rgrq!>CF#kW{mTahbRbSFN!@cm&bgpIB!De=P= za33gd0bzPJ&6qClqv;j%SgkAW29#24unH~LRK`%k5Lh8K*B>X($I)p8+chT6dE%PD zbqUVv!g-pP_B>C!p6`Br;Cx;<&w-Ci@f?aV?dOG0o3|-3_UH?CS-nr%1QU!|o=4sQ zF(JB$2pV=63o&iMru!#Q9cX(_;L2@^7zJ$JC>tWCV7jOj^j0K%)g$!_rD~B9M>uul zsF89inuuk~M`BPm3o{E*Z<8FNO&Yecj)K0aNwupPOIQ7R4NQ;bSkhP7d|y<3oJEC` z&RDw1ChuIqvbNVL`2EUar1(ii1?)0ERx5JIiKtsJc*YShw0BcNY)~)S&QQyq8c`K< z5qPVNPxj{9Ru?L56y@lFYo0gjKDuPovARh3ZmCSvU9KF2&QOIh8k!pFztKN5k=n`< z-6>K;wB8VX)q)1Tkx;d)QZ@mX1~}sNXpBeicpTk}C@Y|j-f?gVM6AS928%0njcQtGk46>ovF>$D;OJkKRQ@|@vxx4 zH&1!yZd_aSlxPLx+kbw;fBxq;eEW}Y`1Z}h0$cA;pA_o*2@JL3>CX?FME&)FKmL5j z_wVm`|LYyk_b28x5jGZ?;pHk6zJf5n(LwzNcfUoB99Q^mMNdl}MqDpEbq`3b2pw9! ziQfUKFgLXR*!9i2%-eKt;S`92&SA*im`1sjb56SgBX0mR|GfQcEU>Yd(serFsbLH& znAwX8%T;j9t}f*6(HdrNkfWPNX*7(+-qq$jh)tDzFQ?dU-<#*9vH7g1n~X?XB^dgI z%Y9hdJ7(mo$(l%ow}Vp#8JQ#NstfftZMsoC*=1&lf8`QXToT(&vu64nnG>L~$X^bP z(F$z{WOD0iSfy zD(KfT%~FX&3ul~yR!?0PX>qvcMTP_Klco*ZcY9rRLi|Y~m0Am>WtQj2q$&9GfCo+kkHiS=4fH&y_#!=RiH%8qGZ{rELDX z4Havy8spQtcpXw{K}U${bWdNs881Pf_v4lRVt4+V(mp@b?}ZdnYRZ++CF?Cr?-YBE zGMwD!(23msqv{GTqU^FYbdO?)Jr;uI2Kd$pD=uLLF~`~w*-&#ThI>`R|1%|K!7pD7b z$1w8lm=o7jTvKq(PyuMYs?1~F7JR7-n0x=zB&+4P) z7tTCLzJ2PqQ*Iu^rF`ALFX52u%#|5+zrSLB_8-)&Ad_bye;0)XKWCB0E)``5A+WwS z-)OBD8ERe{l{3$K(_s>}$6wcl^Ew^=WY&n&zn6LjiXX<55_`nw5EDtR6{9!U%YL(z z9AQUr^$j;`5QI{s2K2`Cn`sNkjY%1SFfjl$U-OD;C*n59qK;}|x5*-J>rP?)IO&x# z#|PxNz`;NQgn;BaS8WbWAiZv z__|Uq6`h@&JjRqp@0lfn*&622pyPn_4(UBJyX+upFI);OXo>}CK6=OO4M*?Lql5IY zDW*6Pk{PUO)7@?`2|b!^xUN;sLaKM0e))k{oIz(W#&9Y~@LKQ^bLn`m4JZqkI&CUp)f-TSzYA>Ah<$$~E=YQ|GDuXfU zlR_$E7Re_Z1Z~Bmz|Ko)3wgZ?6Is&D@JmgT@kr^CGn@=64b7+04FYsce`sq`8&J-a4dV@R8J;PeBWY=HWLE8agY{Cxkw z&-W*Oeth8P`v=bR#PfOLe4h9?pSUiAKg}S#x*UA?p>UQ2gpfWjv-Y&|k0gtzc>lR^ zA!``Wz-?p)zdrgwAVxGq&^htG#*-abrXOI_9tdQmj3R2=Z zm=>r$C4faC5W&zUG=$HgteBz>c2>bZ2AimzI4u=lV2k6C}f-4B}I$ z!*b0I^Dk8maIBRwCx#4P+y{t&c^r_(1Nu0ebV=y5uLN>0q(UMnjyN}qkZYyAi`KeL zq%s6`!(NwS3U#c$o|oDmjgDv>C%MRSIFFb$=fR|vn7j(97b;OSA*@(hk^wVhVFmyI zAOJ~3K~(iF^KByKg;ebf!bycuL@m^7)zI7bTUcoAaVdRq$)pXuKsWe8rJ}{=x5cZ` z4tCM>k;LY=x?52w#eHgFz_$W04qwU%&YmzrD z#RH681E`a-3oR)dRy0|Ys0{6;1%_Jk+_t1b-k34!xnhV~6LLr%q_cTR3>`7~s-%?n|M3&wfBeAn`Sk0=^Z7J+cX5)^*3#ugykvYL z-I{VU$OkDTV`U8WoeiIH1>Yc0Dx5v`J;lLPps`pW`PNJ4!Or`^L1{Jh_aHE$v4>k9`$E#S3BHQ6w6~&yI$KR`+EOCy&B! zk@~LiMFpwP@updnkd1;vt_0U!&pAQY0H~RH**AG23wvA>D6uJ&I3nEte{yulaX7~Q zfFymeC?{1?Y8V%jLf*bP?eT%;o++7!PvK%pW1cH?3;rePcK?Q~FEGYM+Lr)=oJieu z*kx+u^17TFr6*=de#7$t-{dMp!BFheq43xCGnS+yd?q%L;?{HG2&ZP+!rxw`eeCke z84J40g0hGb*<@5k0$57iFA%=gF-|>MR%A(lB3ofW*vN8Ql&d=fn3h1 zo82n9?fl79tp2-OfOVdn_hPCH%={TH8!|5mA8|~i+M;NpPNk}@pI(Zn9dbb%DNNJv zhXOG|ar-Q3oo%nvi+owxIkNEb2EmDgXUovt@-P#6BT5v3LT2m6WN;`@3(_J}S~a4Pu`idWXyHt%y>vupNfM&^|3^N<8eY>1* zp9IU^ye>Yudr++YJ9q0%;;N-7&Q4_v${@)0wZLZwm{4Q7aWA3@4N$P4SaBuV++}@H zJU!2+7gJV9oi6d)Kt$m=YK*3g)>fICivg&|r6ku{&4 zELj4dcTkT(R3r|>;%jn97KMrYVGifD_dsQ$tAb2JYdEfBN~=<$J< zu*1yLX@og#L@rcrOwX{fZgxq7uhgpY_b}Pd8UtuuF(%NQju$egFoSYyXosRpIoIx~a`5+?uh(1|K! zxQ}aE!AE2!@XmH!t>GBm4{fE8qJ?|beprW`$+|=7MFQyP;NF3D0Ii!*+jE9cH1#j= zxff~}r}S6vZ5G1HZkMQk z?(=3EZd><~)vb_E(CdMC^IqM3S}(o`y-k~RYlz&*t09}xPwNiKhTo#*S{|GD>9+Yn zbWNXI2Jddacrm`lp;8@KA|4bg@v zdBD(fK-&dvCxlOzZTj!z_szUK+TFe zsgDngDd_!y)&*T3anP{L%wmIJUJ}#zKBRdZ*!2XaZ>JIv=SFw5%eZJjbs< z^UtKiD;|Tv>HSLh%&p0_I+puBV(*L^7Tsx)T0tREU{yOuWY-nw7Ny*<+=}dStX9Y% zvvkJAJ`ZC5Tu`NthF{OWhb>MB-JV%4^+NpI8WqE2(F9X>bkT|+@`5d?ef-Z9QZcMZ zvW^)3(xwzIpvOf^i!n1661I1;oTzSIp`#x-I^#GF9L)%Q|MATL6NZK+N1|Ji6DkzO zGPlJznX?s#Kwd}RC91EV} zZEiuE0f+ELKy;^bQ}QWa6jwBDGNp}^a%b6TH^Pt8l`a?^Ql-e%T!$YyXV-CIM7lbb z4x72#g7mP($Yhsxhzx@)9p6C58|d){9B-iG0csB`s;DJ)Sj)Yi#Yi+2(EL8X**)`F z^GR@?jVo!0ZiSPk1asl%=WA9-nOdokNtS4?LO+5L3smAW8x!4$RfWh_Q{8`mGF5+9 z&O9m8p*P!STFH#iRZ#84Af?}`>H+AEr_HW4HWA*v{Hip=I%Qc-Pu*lEDD*%r-*3OE z7P(;1lQLB)euA+0Vg!CkbK zQb#GI+I>nEg?bW8RW{Pq4nD^%q_Vu$^4edqPrvuS)(8i;au{8mZ_AL<^uNf6pxDKb zP%IBeG)ETNF9?mRCmiT|h+QlI4TA&-8CacYf@xy52^s_31{l){pm+|c`i0^iZ#kv1 zhzDTGg^3Rg0nQ;f&SA1;mjmZ#H8myN0Z7UK@9@%y*aX4@+e5Js>k-8s1V7Il-Jn>< zf!4Z5QFwEsu@_8YUYBOR>fYU_+JSx$j&2QBwCTTRblUkP&LEl=x&@|6mJ3PG<$cnr zw2Gj;sTDHRxf+Cy{WgN-PIz5(TBkB04!iBka%dt32RJI$Y*!8-$ieKHU>X$5I<;D7 zjgb9iV7u7ldKR>O6j0)^B6EmYCA>g7&w=N8;(R{w@qFUrdE(>aiSx49gfUGKH0DyS zFjfD~+S^j8N_DLek34YMQsV!2FCSqa^WFC{t?>c3z6e!vaYcvcI_Ms5;%zd=;`dtx zXxy4;M1G$}W?55!#R^~zUBL4Gl4l~QL)3KIxtxlOHCY2m#dZi3D1jD6NX3en=!Dh@ zqSnNo@tmi3Z>RnJ-}>KgPNZ}@5Z6#P#ci{p@lRks0^XE)GMO4)#u!OW)m!05d!|uoC#GK`%A2jc@QE}cV^z$qb0`Qk#K>#rszA^rGdYN9!LTG@0E$&lwoh21AR8x^nq z9^M>U`l4t5snb1*sepgXdnztE|uoqTXw9wn!slrwC=8@W_RCFtVUz2M7W+BDs z*^%|)M;E4^$$Xt5fySIsBN>>9>KzraC%o?S46JA1mLmrBtfL9&PzX&|Xc#qdvB+OvTU-TvO1miT1o4oV2pXD(x|__FIOl-y?HM0tJQ> z+@?3I!XFY{^4=InZ#a&Q$Kx;np+H{0`F?twc%jOhXWlh&DHYHg&^vH+LO&SPX72Zi zcdwB83Xx&js&mdM(hTf<}B?~*xfF)mb++x@pFvEe1QVrs6YOA`CYX^So8YM)O$ z|HrN|wk%i+@0D59YB%U|Vx*CSOz&GwF&0mHL+Dnkb<#FECAfzTK9&6LuuagVfZ9E>p; zZFnFm*N`){RCg_1QW1!@LV@9807Mp6#6UY0$51>5aEWa!LJeqaCex;=U9FvE?gTPj zV~*p7D4y5E`^SmzKR)p1pFi>a`v<=Nc;fy2#K*_L^Az;%@-b*?OruL-(@Qsgc(}>i`S;!yX;Qdoi_evys6NQX&aLU=#sUJPlj0c-_9Qt`PIZFQME3w{-D zw;edP|J=3z;jJhWARSOy+uZfUm_sYkh-oiktzKs3JSzH5+CLFQb%1t~^7}P>V3{GLae{XA+dY>pGLp`0+fwkb2@8!|e;}GPJo( zQ?xXSGB$>(18JWX1)4Z>;hUzjd}s^9$tBaq3!~m#&Q~z!-`?=; z@rJkFEK902JR0LDvcBdOKS$+H=?pMr=2$ZvrK8D-qW*k!s9IMYGLg zbzH}k(#2UtNImF<^rd>+wRi>SQ@_X87T=(>K?uTPU4HiPIc7H@`m$4Kg6ZvwNz@#h zUy6t=BP{M`Zh0@StL49M5oRYV$Pm$qc4?n_hZiy+a^$_i zTjnn@=3o$C77Xaf-4#sReDU_FT@dTkd=Iu+0@~m5lH(2NkL8teX&)RKAz}EglBA60 zO>EO`g0rvz?5&qYELy?I(-#?$B@>Mu)8#cG-HFD4t_gL`Rm{m#1seEsN#cW67|3EtpKS^?n}Ejzb(25$W{3Hd zO>vn>@uG~13dV#%0^FlZJOMgwD&_~m>TfgkN<(u$a%(2V3z=X{_Q$lq^(&=)HkA_b zf+D>d))lvjl+lKp1Etc};ba)%xenDnk!sDG%!lGQ)QTt~Oy&|!N_$pZ5YN0AJK1(euI#L!v_kSU?m92lGpjNF;m)S=a=hp8sB;4H0pUi@o}Db|9E1| zQPda?E7hPf7NAm%x8Op&-29b5rB>!DNmbehZ?Kne_S=vno+Y}$OQ~eN?hUq;k-G9x zZ5jod%gW&Ly?W*$T6;FR98X;-rP1v7{iC@zTGMOY7J{GbKu^uyKyo__WiHuj*d#12 zfN|bz?yccy4ZSmX6117Zh%rDIc^_`guYXqBha8*cW{`+US_MFu0!cHuol3x8ya7ja zZw||hI=cP4M`s+Ji=5A7LF}Q2cS>(eRTzPrRJ2au?h90pMJoQ;E96>8iIZH0U|tiQ z88fB>Nh(_hjp_cV%!c5VqHIJq@;nQ|*7%-eN;FHx-`FBiO0j1m35FC{hYdLI$pa-~o_RF0?B zHQU^RV7G@wbtge=9=Eqw1ns=b<)vfr6!b!?RX-1Jd%U!>uU<5JZYkSz=(+65RGXEylw>c+1p6@u% zrxg-&tk2U8_pv7T-pnKASMP^mtK%T;6DfO7#EZd+{w8?5DIO2S<4_!jwNR!{E2dBR zcEkexl0l*+!K_L_SMy|{%hzOTq7vVhSDMN*5uXda6!NN+d(N-mKY!zhSzXfzG(!3S zLdu7wE?M&KTh5iA2ko`-0$_QZ&APmy8l=9QNR7;k3b4d=4VU#9i8H-)YG+H3owxW! z0=IIIYV$|jlbWE10)5ab=d1o2?S@z?{;b#Es5B$7Pii9DOhrh)wOS+YkK zVT<>u+gWjW#d$0&E2R;F7$i{0r&pe<`G|GTn+8Fb${fY*ChRFkD31v&D+ zTS&bi6*aR?L@gyoAZBRy1(>yl584}K@;(a^eLM_t{^-jKWwJfZuMSNvM&DcJMOoaJ ztdD$5ZiQ3x?|TE<;V4wmj6%?980C3Q%+@g4#AHJEin=9kadpuwS=|EnBAgroNVCH1 z65u%$y#bFC_!twTGv?7Tn+2qPn}VZ9ZWcRiU@uYxfERWPfU0s-)WzP|QfcgQn!>3Rk0Nax%>kp^w(Wj8Mp#)2UUufA&dbT?7=^-rnR8df| zs1;KYW-NY17MDc8vAz~1>getC=;#h4C?<&2<--_x6JUtfyQR5@l0Y+M+m8ZLDkJvEn^q zDnyW=6NlW$wc$}fML`t--;$Yc(|sy6DE{vjQ47=(68|Mn`}dlw7C<~x+NYvGb5yjl zA2MPODKMfiIg*wL6KmK30-6L!1HsqW)C-x%gasaM;l0jBi{%MT0z8CK(Vd4 zy>}e#z~}=823DbbIfzdkJ>|p9%M7i2RY|VRr|B2`G z1K0WVX_Txzj{~A*XBq=dctO>Vlq6E)3{<5i4xE^!kP;kmJ*Jg5Dj3SXN8cJhK3ReS z79fFAM9!Rk=G2qwA(}m;Rlqv!9#&d8eiTIrL7IT4SRqUHlVUQG2i1r+C{qW}I)8k@ zhHt6_;1_w+RH(yRJhesAt75)yo?HH-?eFmXF)S%xhVJ^6unG4^U8l?Zh8NN?!Kun( zZ!{BN^}NaaG&JNz-iere>jcg9coAh*?*>yX9sF&|z3%M|H?h_Zye_eNKgVyRzRJ^$ z%r|>|3w2J}EJC3qM#hg;xp&1wDs4HGHM1(4Gd0A#G>*8fiB!!TjkP46^WjziBNq#5 z&LIMfr?Rbq|BOefnv?xQu(j2AUFro-1kWzA60Ol?lKTipZ+JU8j)xafZB1oj!j!>U zph6$H=SG&I%X26aF3}17h@#4ise^4IC64zHlZKN3qcw~^(U~xa9c)o2VJ>sMvC4PZ?i!j4f^QN&d^80rHY3GH6DjuL}rptqo)Y^Bus6XxG-^{ z3*&rR5p`w}Ww{u0+B8J5kfc?rp{g62)Fgjl#9hs=8m-15+H%wPbNf7?e?v-dqZt$dOoMe(72QgcoqH(TA}B03ZNKL_t)d zadTY1Z|p}$KN@;(F6RJvG|&B5Nbef@Cv)PwMowCzkb1iwm~S)n-@SE%NO*W1HxHz4 zE2GB+uT~sJ!h1QKhy0^4Wq6?)V8@HC1K}hCCKd(G?y@g%-(m?m?qM32{uc4tZNv6iSxTE;lUhBW!xq zEx$>MDPI#KU0u+b_*5h>me~@ZfnRu`^p>fG5(IpkL?NWd1q34G$Y(DAmG;?J;O(`- z5vp4&>MyAPh4r~f<_a$4@5)0A4^+TkEL>V-=Cyo#yEvcLJNw!^BJxQq<+^D~P5orD z`1L_4k(SU@+rNwrP3cE~@|p-}Ns*ORplSpV#t>5h5(cy_@~pBGxG3{}YJ6aP&VeTm z>CToHL^1X6n60CE<8>Sd=9oEUSJljcvw_rfPj8)W4=<+712Sg^w}$Brwi&RBVXBU6 z*hK32{J=G?B>UZcl;-)ieIN>|espVy6hH=dfZIgtg4QRv3;JWCKPHa1iQ_TR4r^|b zar8aln3nV_Z%=Vp3>o&>pI=ceQ>Owcp#aasU0tV9<=M#fN71O6Oy0BAm}KSF)y4)^ zG4-iO^KY(~BXSV+2I?uk&Q$mnSVDxm!{0aJy=iL_KC@v=m-DUtY>|5-H|p|y$?IH2 z)Ta%^JRfwB2&OXFrIlKsMZ{#lu!vvriCAQvQTT-wTN*_5G^xHyhT~1TTh8leS-s1v zh$_>rQn+N18e@Yk7Gz=NL$oP$hy3VH5z__Oh^#4Vbn_~tN;Cx0eo`Pwbl(M{K^&f4 z1K5b4mvtEra&NLGk)*sE+oR)X zrI-pCAE{$M7Y1P+*7ap7Ma9q1dgzS_$AcYeu?nhHNV%k70yGt~vw2Y(V>le6PNS&E zw5ABIDLAL5e@@&pHf98{+2M~#@piIJq*}vZ#_SE}RD65ucsn}gn_5?uG`%5X4osOi z1~8=IkcKyH=#%mEiPZTqah`(fRE(jRlZq5$g>Ak}q_EJ_HSkWAHbW=*Q6R!36}pFC zZpD-=jc}P_-Q|3Nu-qazb$^z>P)BXCHwiV8g?P6~WiuZ&1LP#Sr1oghNl(eCOLz2Z zykG1Ej8P|TqwL)#QLP^->zD+L-Y_rTCsO`%h8Jq-=b2)eCg+Lk+c&50c_-ze9uAD0 zNP*RHy2~f9CQ?)u`87@tA6IsrWK9f|dKOb)n@WX-oAN-id`n|aT#myH12hUGr3U4# zMNkS*BrDSl{nex$sm;Q2D&@(b+xN7Lqw4x`pVY+oU++d3-6v8Da6-H6ERD&rX-!nD z@RwvN#YbfH#S&sEHGbiKLvz42i`+Io3t3rIX_&U(U5LQG$rGm2KiQkYl%A)J&{T;?HUGR4x3o=A{vZA^fSMm{Lzb zrJ!C14Q92&0@cGrIk*w6P5u{!)b#aw6;h*|26ayw*H4a_IfcT_V~AakNZ5&v^z}lD zS#oM>(@yi460@)K%NVXOba*y#m0SxyeF#9=SY~k#1a%9i?XKCi+}V`@yJf0{Yhp>3 zp#e{3m-Y25qSYKz(YsGILe{^YCfZb&17#jOpaI zC$8(XwOFKhzsY5NUPK)|_ksR6>^+Dgio013j){H#EVpNPqg-uAS+r?SK;#LwXZtVl_L zHC9S%E=wsvs7Tgi8s(I!NoOwlhxCkbnPAH1gUxoA4^v``;yTgF^Mxk8QN1vhjd)sA z8{&;ZS@**~DD8_Er&wFEW`_d=m-$htnAaN|Sn|MxK8}t@?>HPl5gAeJ=)~m$sge%e z3#zmhXyQd{F?DoCKfI{w!0~Vb-L%XJOqYU8!R3&Pi_N3eh)R%GB&*ni8=Bg42!u7I z3c|qPlCmsTOnn@Tc{FIZi4=~8q3)l0gKMBoFFP^}&Im!*BusQn>QFv#Xvg_6Y$|n5 zTu;Hc1keB~RDl|?_X&;KrFTUTiNoZ?; z(KbnWp$Y8QmI-^Cl5WyId|le7`~6HI5v@>!pi_NPaM))H&os*Mj&}y>7NG6j05a_U zA{=2&W(hCSKK;~~c$-EQ?Cut6AmzDTTwZ96pVB=C3OCOb@L9IUzn1+GRr+BBcv=U` zdJWN#N>KWmno0THZ(cJaw^IIIO2uCn(4f*G-wUaksBoEeA|(Z#zW+x0w>m_61NXe} zOMdHrOoqEgzF);=meV7XLR&$lOG2B)W|79bN0WTriZz5<&&dWw)(n^5hxQyallGNJ z6Aft}FMm|`f~rSB)e{)d)xE*tnVM8&um0K0%PzUr#Igk4j;Xk=iSrVSVG$!EKoeJZ zq*55V_Jf#6r9#1`Qp8?RHv7JJpg+W~D5x6RVGUoiBG@4xwDP!O*{itgXk3JEN&WY= zinoAP=suCaTHACQgh2tOzWZvmLAYlD#Ys54O?CrCUy@m)Ak!~+r4&orM|Vp=G;Jyv zMzrKr`ovMkr%#wgbolud3eQfX+f``h7R@qT78RetxlU#2b&QPY32 zkh*9Rhj|@W%-p3_Ti# z0OJx|=V67=WpueYubfE99O$!Q>cEf-Q=T|3!u!t?&-V-0b7BrppO=+zZn<5$48GMY zT8bRvdN9%#!}X%VET2a;v{x%R6Rjli(mXAVnH5^+XwLi$e^d9rLojMSOwF*#dIH+a ze^NvJ5}ptSqi0{5(C&Iq*mX=<#Kln=q>hi4JgJWeCb(g&E)}ci_JUbY4c5ImLK{WB79Ta&c|a`@lVp-%3z$&(ZIR2^B~63Aqe5UJ7(|7_Xg4J?>jq5(t%_Gwi7 zS)}26p`kfqKHLvQWPYW_4b3N8A*Y_rsCY6aWIUiL?ek_&3kmXyrg#M=gul)GRJx4# z*YE;rSab3kMb0IXIX=-;BWkWvXi0FbR6%Q~MTmO`H=s4Kt20^`^llSErkMquO(Qt* z>`XB!T*1ys&eSE5rF+#LMt)wDwYAp6!90?Hn4Cz#B$Z6!pT&f*K$_mxgP0^z!A;}0 zASiqDiIP@Vl>SO1`;>l$3)a;cqF{4*a=dV;-ok zu;P7@vIDZJqKW&bqo|R{A*^8;^e^rJg4P2_oHcmw4&{1Xn)H=WtG7y#w7RHxKj-Ue z{@m3f|I3Bcw$tYBvjL}Q$MFU}e06PjT~Rc3#=%}lbuU)f9*YJqrhG!R_HmzqiK`Fd zrn-w#ax-W5E^!+4RF6<08S*xFYFqxeB=leNQKpM-V| z@bdyXLmQi^^)dqAI1L)nm|V&wiF|HTFvSX~Ar1W^{QU96^W7#=GQ>m9Vp%h97T62J z?CAJIk_JB;b;BiQAvda9clCK}LCm@WQc-Y~wQ#EIZ7r;cum$o$ccRwU3aPcB1k^39 z-MVOU(Zpa4WEE$ZeGVVj7pl*;K$Zzh5c`hatax&1U-v6KQrF{wJ_lOAz}@musEcVA znn0U%u4V>&noNTyECXds&^7E(QuW?UR=m6K366iS0F9}d0m^k&^E( z>P-sT48;i{AjlSVM%g@y+#1dMi&hU-Z^5>=rXbyat)P{DR^_PP6KIDAqK(mUih60BfKf`-OMOKpB{2ot1{liT*lg)q*C zA|WC@#$~eoYnWFmNpBkH?3E?hngfiufm@ZemAwZm_lnlgCYwd+Z_P%Sy>`uDV*_}E z&Px5Wh}^T!^6vEZ7&9Yt4-0GKNQJbI`yJdxyul&(l5P2o^mAowQYDyXyXBn+XcXCIpO@fF&3D4F(>08>w zU4l-&|76LWa40A9>Sra?KK|F5|K}aIY*^I{?^{nG)nP9-d|F|@P&)yKwGdJdum4`x z{{IeaVg>Rp0L1Dn?GOm>-1`*DJS8@wHvR|0dh{I% zdV51ZOU%M$>!7OmAyQW5trO5G<-oQ^h|Q-_N3&d?gBuP~9Mo`;pqJ2Hg}7OAjYDID z9Yl6GJ_oef0&k1fy2U(AuXb$-`^Uyc2X^I1F7^u@EPPF+v| z39UF5lG{zF~oJ6$gSQ6Kz^EV@~UMDyhp1 zO9xNIWRIam_5-^VqBLqDH5Ygy=zAoR84#+p=tD%CsEhRGjYD*a)Zr3hYP2{EDMQ=p znA`<|Ycf?}gbJI&C8)tOsB|s#dOMwyPa5qo7Za+ro>?KqzgI}4WAM#S*dDj&oZ54Yi)-q!!OO=&mznU~PaqEx&mT_Cj&d zuJZe`FIM1=2I^PK-&QlEO5{TQ{$CYVUs*Lt4zwtwa6q&}VV9t-iIl{}i#A%DC~WHy zhkV%EpF+Hf6#;1I%8+uRe)!WYgq3cYd^U;dQAAPO@aMgEXrGXyL+4@m>OL@sHE3m; z@%%dpq(ly@FT`pkrQ{ITY6vhfZh)VQVqJGnJ{6Poyljk9SF4DFC8);aU?g z92hiL^Dmpeyy%5!*$h=7%XD{ADd)6_{TY`vr6zTcFsJQ-8?U`LCmhoX>17@SX-n{r z6;uZ-+He6A>fV^MKc(L;py=i6`n%Nwp8+wYhX4LxhDysY?H6cO@i^L0f^xqdQ9OI` zba_r#_|&7AO8R37LC>?nOJ);333V?j;{pg08R-#BIL|hPqF+DfT7D)+nwxr$&IzQl zpbWK|kQ&Gv@hP&TZ#I=DY!PuG(RVqwO`|N^G9V|MV4gk1p0tu?%k(jPlfE%)4gIdt zz*<{y+z(ZWrln=2ARp>(qc{I9|$^wy;-Umwr=Z+!4|pcAFgk-uA^7<| z@bP}(ItB9zDv5`c(Pj-;2EdRK(Lxqe^^*ALlkRm;K_73@K9PN-U9#Bp6UqOb+P5!s z_$~9a`hvI$sTi!$E>CiKgN>z{KXhv|m02?;WM$=O?I^E;rhD+>&!tp*1E_>KN#pUgAb4^TX6Cq?!{g1$Ca8pEVYVX@N}i6EwMaOoETc z*!8`I8kU8BbQh^&XcwAW^Dr(6jVOGCjbIn)?J+rLuCnCyn&(#wDZz%W-7|7trF~S* zzqf*Fg&n^JvX&BTS*GIC4T&G!h%d!&WL;@m#SvAq9epy1+T|wLxE|%XDW7i^j9xHh z(Ls_OWGYiD1@>^s__m_X;qo(5_eint6Pl>xEIJjR1laD)(vWSfQy{(2ma-Jnh5SwF ztIErP1sl1Nc5l<)@Hy3@DM(YF7E)i4_RRy3H-8qog{k6zqFWK*#S>{;GSNwpVG4>d zE{IGtHqUNj0<{LA<_}i{@p?fvu!=IRM)#VHSBfa^=zRDzsxSPxPo$(Dz~PFB2!$UV zM<2MF(@Fi@A#qV;LN1tvh+(h;Bw&}*`lfxIQ*Q?O6$%{rQLVi8iUHRWN}T_F1haBI zFU4zO2cB3SQD$l_bX3VZ_E|# zk^k&51~Iw8O`UR|)zbz=z@j~4c>rgT_NltGFSA1_i=Zl)QX4-IX6?WSFd?3E9!k@IYBp7}=zKy3s*6i>c;5sfNxIojhkMa2x)24+uuwqyZT9 z4FzLDpYVgO;yNbADY%ZA*U_8%aBPCd%xGhx4@Mgo`laA5cz;gIkvgXrfwYR5O)u@z zDcriIc_UU&$EM+1^u3KPRx@@a@Nanl7W(|QS!d(bk-K+%e=EeVO8d0rwQ5anp&Q*V zJW|-gc&p4~3z{s&;DSv=j?`XAwSGVW^y7g(23q%NRCj6L5DQmklYUM$L^6frdqyE8 z5E>9D4s8$s2d(6E3H`X~4K2Jir)-vo)n-HEiIInjX;M9(YDhVtuOx9>t#-)G0yc!d zAQ0$~Wjdd->m%CIfVRw{n=C5)6H(Baz7No*o18%lo>3`=q*0_uuFuQM7s065 zG2lgy5r?e3klM2)`K!{tPsTJ_Bz~KJ>OOg(Uw^;p^tWesF#zHjweEOXt0M#%MU(z1 zDUJ$};WJ~;4x~u$nckaQ$Ra7HPdPn`W``7_MJXyjD-<-QQ5I~f?!u4%OK+s!{#tw{ zyW0zqq>35K5CJ?GtJ~;qDM> zCM8qrGK&{fP=-!GF2%eAqMU*nssk4>`I z8bhXerO%E%>`8`jI+)ScEDNT#W7kIP`x5($B^hi+@QjJoYt_sL- zU-*p2#nCy>ukh>|-3RP?_w!@( z2}$(KL)w=RR7x_Eh$}Adqe8xk_f5-0UCV?g z08}Bb(`b}?bO9HyEJ=;@khSh^9;rD0epX1)T~gph>`n`eb0WQT&|f{ATD@>})4CHz z?#gIdK73|M1izxt9Mjn1CmNF5Dldh+`cF-yNMB7YWzvD-yNPf^L!Gk|&OEHniD%tq zb3w!-pFm}?;}SAoF&42)R`8le_8XfAgh4LRi}Co%Mp@k#a($@ZK-2PV@$k$ZMVj#=lAaP^=G~(XT z)He;!p<+dwgg${b6>TWmA^122=jpNpS%kLeGBeTMv8I(;@?sJ%sNZ_{+XyjId{Nm*GqT%E2e&NTzR|>k!r}jc@%W#*hf)NDO)=}vyPUR!ML{| zG*$HZ!(bDK!zQkYJ_h<@pbtUwi`(=dMh{eUse*W3k`QAOVSvHPKAAO24!ie<{R*>a zr2*#UUA5-u)yQZAtsuV0y4Z>yUZKX=I-6P}i8gYRZj&JK*pn!VLcZtXdYS3Bhy^{5 zrYWlpiJ{RJMcF-5ny?+;lJ_Gqp|!r#wt9alq~czPPVV2X zq2_$0Fl?}in}4d1D;K@!m#U7&x&+3()ol3j=P?+#*pVU?~VKcy!BmI@8 zNpKiT?vpRFEt*)+SC{t9py>OD`y-adLTrj5HLS|57{q%S;Z^X?LJhTT$Oz2~t?;?O=f z+Ew6~p|XP4n}wYT5kz?DcUR4|HhgbUI(<4HsB;LOW8&i)ct0 zAB{1k;q6%R1RgeLmuV^Ebd$%$&y54`5Cso$`JfA}{DqD-R{<5ploGlEU5yLSMDV;6 zr!W4Ja=r2}@-3+v_fEEcVrAMTeAbDv?zu>TOOz=SVehQhb&RRzw5-lxW4;k=Zu8Es z6jEx5*6Stc)_somfQwcji`;EmA;m|dTb)=Ztuc{`B>P3|>u5(uKMp7ey$>(02KpiB zeW3M;)-Dhm3YXlk2NvIeCA*0Nw)@%nkQ0oiCixJTZi&;XrrMRpS(z9jXmesT%N80z zq2rzlRP}M-M%fIMwX%7oxJY*Cmg%tGQ$Hwp$JKfSQDpf+c3Q_EhsLU9vDHn!pv+KK zGzTr%qNwE{v~rI`ZF3LE#w-~j0c`;o)YLp9okz6*03ZNKL_t&&4(YIUsiKRIy=bH* z6&a1cbPSLw)hc`&nHI@KnO>2<4@$rG#8&%{e#%6;xA$n%qv)N$$3=1{nmFAGcff6O zy({-L0WIX~gwix~|71st*NIYTym>)&cX68}BZz#GolyJPlGDd1q%t2Ui>QG0gvVj# znwI-ZzI*(dC$=?9YoSxt3$0bojx{ zBO<`w%tU5G@g(SEh%jh2uyn9a7Z6W$I?(xnqrKs+e>38q6sUyA3bK#1234Qnbf0*v zyVZu!RyaLs5yh6t$fQ|JH755On;)uWrc*R=ZMa_*5VgR)`}w2*2-YFnzP4tE(C!UT zqhI;~x1yYr9Jd~I5~mdUe>*2YbAskDzbp+jE%9JV9hREtao0LH-#iO;Gbg z0SlC;H^?psE(iuhuB|b}DJ3TlOEv(L3cz@P=>YQsj0Xq@8sHwQdjC&*)`mT65OM0F zO%It`?%|X8cdMwi*4+UwdXxMWP|Ze>rd2Xde-8uWEwhk%N8KwV0Yt6&c@k7Y)O=m? zCZUi`a+;rOb1(obnXv?sxb<6h$}?Dg#vyJFnuf{`t{$)CLc^Fho5Cq77)c8gBLqt_c;yK9uO#H}WinxhN#iLU41LpTZH*(S@o-)bqm6=ZW`^6VG#ETy`*p zcg&8)e)BhD>5z3Sa7|3IC5&`05*AI%J6~#=QW65m^ez=)PN1`Cr{kT}ppgyv%C2hciQy33)8DRSTNeX-dD1kF zkp&okWg=y|{d$ z$7A9eg2M}`c1*NWz-^#;7m}KtE@Vcoi%nU@(ZLgJ9w&I1VB}H3^zrNxDDo-P(1|XB zArpNLj7De^7(O9pLLQb7;mN4}wlzA3Q?HbfUEPh!$+YhC7#@>?FpV_mCS?Kpb38IL zOJ&xT6M^B`rd{NwQ3qHM!@@Sbi@fT9Rj3(#X?PCESR}^i05?B0QzeP60_r#(IF7!e z7IZJ9T5DfgGi&Y?6{rMlp@;kK6+zZprpG6?oqCqYX|$I46jiC)g3DCTRn3SZmlF}| zKCaT@XWZ&ON-3Ynl52qqwSw@^CgY$-+%<&Wu{HYoTowS@TN~rCag2Ntkm@DF|!*d`h)yzv}sLAo(()ZSSFfuYg(Y8(;FV`fw%UCNB@T2Ixr__ z78QvTcu{;Kh{F_fY08eNhMxmkg6`LuRmVuI!g;uG+T z=(PQTt%YXAls!wV6RZP_0bx@%P!pq@GP`v9!qf|*Cj=MBp)7XT>?y+v-v~_yx8My- zZ&pO{0m7q90!kxN24oCqv#*eu(msy0+yC%I{PpUoG=C}27L1+vSt1RT%UL}&%sI_k zUVdHiATf0u;g5=puuKH@aNXwg(?v%W_tLM46g;}e8oL4=hZmNiY&izWeuzx zK~2u>k`>3FnW&_?3_54Mff=HyuBmWX#h`4^99cnS#axFMA+&@{I~oMO4~VRNE%va` zBq^pBpFWXl1j(}NK&SAL`uXEsep4m*Dr}>YDHxiA3-ez;qSxmQm5-tEQ6*O@( zsoZQ5n7Z7diVF@FP`|m_utus`GsUdv4c17BqKi%Dy9nCw45~r6#2(sOy&lY^k*e!( zTb0Gupjan9NfDMjnhd5bh;*Iv@VpMv=R!-yO%|@DiCVP&cT@FOHd0cgeRXZU4y7_x zpgtZ%tGrL<7Va(y1TDts*huwl^Q}}z+h&%O&^6QQx7{~V1Fc_ZZ3DFv%-so+;u|T4 zBCtIZLXKhvRo_SjAfw3n$nS+^BQ>mb?5lQ-*9NLwM>osCo!4GzE{h6Y5S03oQXYVXg+=?|?ijhoSZ@m|Gyb0hTWeCFUL zC;oDi^aTo4qmJeqDTBmlQu;6y9Q4r3GU#XkJj6FrY6Be7+eYePjZ~!E!`^c${s!`A zTH2R+X4U0JimVyhTWq9&&UOwmVk?E*mN$n>^vJ68<|+MAHH)&G2ieHRx$FahXr``^ z2VXvfqy}PZ5IM~9C97~s`wB%=yzul!3N(x`OT+R6uM16mBgH3(Piv(31mkiGfHzX? zL{ooRQ+2ug*~~HO{wvYBks5j$HbR@-3r!|{Pz@+zo2iCI8yYsSZeZC!yxDe%8LfvG zjBF^zC*(SzeA@F|SKmoIuJ|j`K3%2Fb;G2kG0{&ODInT)U_GxXexAaYG$A7);WWsE zN&BJ>h=F47uAUk5c_dDHPCu`hcPy6!WPd6Zn2^=2wA8S^^5ufCJ_6SEJ2F`|pEI{6zOKsr0U%4q7~WCElUWbkC;e?r#V#W6I{mx@yq$B0I1;N!4H zYFtvLymr%U&9Ugr7+tXqg%II{;w1x5VUQmnw$=19#UH8zFoY~?sy0%* zHd5Wg(Auy`zX)R}VQ3TsxP_}yPSz&@Cc-CV1w+(>Jbh}i$oxKs@aob;E#bS-Ncmbe zrF}G|eOWv&VRgt*z(3nag|sgyZb`QAEmE4z%fll=>*}gqOX!~7^oDaA6R-&{z-xw- zZ={46eH;FYjnswKxAX^4N}AXRec+_cxoytKpbBrK$UkdSatu{)HW)X5l#JCY#U}5w zBF5~P%=fV2^}f|ogvc${fGJk(NPV}GH{HY#l6jyh0Wu($d6mXxA=8sG)hvU;|4hIS zfH0S`hgGUk2bpOl^Ofs!jMzw>NfoSgA`3Bw%QkqC8%HzMd-vvv7jrHE#|D30+E+lx zw`?Mv6W5Zjqxgv}NvkKSc%W#F7A}zT{L`pzq*f&%N{ZqFSKzX34*BGrjj;Nod@4y| z&G#77MXPhrIkU#fpUe75$C%LtN$Lt?@w$-`r(cOT>*9@6Y@(+BaX2kXmSNFsr1)kE zr|GG~jm)Q|Sb%PW-1!bcsatvntOtJ_D`+Wj&FqOqKpcdL9LiY_ARJY$%3(In5-{z22+AM>3mcD;=U+UZ>v%lF#Poq?rDD9;sNx zEO8mQH&b!HH6Lbh%5{m@x9V2dYiX885+6ejeobi~1I^Ei0OOG7^2HWP#G+S}kk=qH zVYbtDb%#I7q@$9J%R02Z!+f|5sd))vsl$`-x2+yc%E*`4B+d-135qN4W%8&evTl1WCT{AM#1+bP`b1UGC$D)}gJ=QH4o^1YnO z25UY~{W}^d`d1hF8q`{kj%5;+9tY=VxczAaNHesht})Re1#&Qb+I3UPPz;I?Tx%Nm z1bSzjjnKP*FZTK;MQ$Gn3+V~rh1uO;t%gA_C4rP@}Wve z>WRYF0_Z96_E2CWcD=rUMA2jOnetA*2ntmyQR%I@Ppg4i_kV6>({uw++;cG^(2OWRua zS1YtFN_yQ255Sb+0r&|&`@f#9ij-p!D~hNbK9SRI@ri8KvK<4k^WX+^A`8_EbS#Pog zTZU97d{X)EV;dzuX{6laOAfT4Y^pq7fHVOYaWhh(_od*aLXP9!%It&>>(QAUOeP>A$kUW=MX$f%0rh*&0N; zPsisL@*EfX5HMfl5U`-VC_$)KeGot$QWDcNZZudk>n5>FFLL{SDL_Cf~*ntu|8gC#xK%2iZwo;e(_CMZLIR9XqLJ|KyoqY0RbIG#IMlqQ1keng-Es2aCan#I`}(m|ynm34r|Ji+-k zaPmJt03mc!LB!&SU7qKkrG!*4f2riZp#<`yNZ2tOw6dklMr!;`jnq${+i^jtu7_?J z&BbIO@)9O?+8#tqie-_YQ+TghFjRF^yg z;{2R#l5SJ~Qxu0Ea6w^)lFPa&jp~tPelc*8;(*voO*1I|(O)u~o<&?1)bgadtRO9k z#!BplV4^Zc6iQiUsA4tH;dXN`Q}nqV6>kJ0b0Zqg zrMZX=OGAb=1}Zli3lV`_1LHbVN-*53raHsoRAK0qS;F_65` zG&TQm>-qD!9V_4GLbfV7M)}_*Q$>r}clF$e=tRBxp(Q2sSST7O+ssD6AM!n2WYi@L zIkeV!&)6nn#Cv*Po^zufkXgcvUCI~tE;nzaP$UjAK9SZcq~uEzRhg&1g)OQluVcy2 zQXqpnNOgv5mg_jSyLlVqCxgs8*1bvtZ)=iD_FLp^C_!*b)Z0(!a_Jkosp-9mLNP-} z$Ro0uLTL&oj6o(foe&>cZmtvscXe zZv^b5*f31~CMPg>_Om+4n$4wr-LoiVG>90|J<}Xj(8q-~Mj52Uuxt1~`(P@d66x`v zo82Wgh%2vFm$21n!)Crt$X>09x)K}#TrhKh(e{8N?<+BmL{aP1W__$Mlo+q-fT%Y& zkOe4^8Hog_+BS;mfz}>?cJokq79ntfwWp_~wH>WLT%vVBX}7P#R0c8*$Y}QWaIX=4 zX(M&>Cla!*`G-qwOK+~e`M)BxG#*MwTD7u$73R4^qBRR| z){j*>jyb@kG7xW8^QKyai&vV;hAU;yKD=l5V=EQp`7LKJeX_GDqofOK@&#+NR08UlBh81a@d}9+U z+vf0$-H2`6%-nji;01q)>7j?iB`V3Ke%MkU3_~36Y&Tv@&VRwYOX!+?LQyU>@*6%j zf027J-9bFH;g|&w>7<3}EsmUg+hvCUPoGsh*l*~{H1hezqA|**W`;A{Mv7OaScOh} zVbCgK`492XrXUi5UDju^xo9Bm?RJV+=4Q;lOKwq0U~j+*WCS986lu@Qmtt>FnqUSr za_Hy1fgR#olfxP*mE20%L1~iB?#cG5>{D3sY&Fhe$QlCQUPh2&Y^1Vbm>VbdEfiX} zhN}0nk>Z_%^CaT_hNX6Z_6+z{uu@?fBx}nAeXWD z1;VH+{(IUss`_={#ay|Ly||4Dq)Aocp*V9%W?s~SjVm%iKUnf}qMKxJ{xkUoh@!FK z3HhkLo+{r_DQfpd`ky3ou$ThfFDWAg)==I_^^@H83nHtoG*pps z79^~qTu2FZJ{0ArxO3G%f2?P-=u>}T1vN%!BDxdumHwux+G5iD0<%lzEq8jG6G*9@a>8+ep1Nm-OjClM^J(d@3D-;%u95q+DLjJsT))3c8}18mP&+>axF-_KB_}V653`BATBSr2<9JNI-3v z^tE5^fc34E%lbgCN#s<@f?4m-loHpmY^-t}TAN-EzUWf!dDV->tI~@lBM0gvND1u8x71SuOED;YFdJ81xG{P7B!TX9K0;->q3W1L0$e0B9nA_a59V9Cei7p@mwSylnk{eNF~;ie^w0kO`#MF{i+3~ zpKF4b9MI{<vOh20m{m12rKYNXmq)4u zWn~DMmcgag$s}$~)HI8sES$#h(%iEdQoupFO6t6bsK!urS;(d3lm*aELf;7ICFqyf zDt}n9+N9W8X-+egW!*?cb0(6jO%AycD_Upld&6Tl$iPX8rw%;Fz;kpw$AzK5HWXJA z@QI(!R4ReY3NP|B?AewI*zlr4Y4l}XVJDL+SDyV1Hv@GEgRA)#GNAyULduAi&oTUP zF%7I4J5-GF^?;nO2s>q8ne&`1-V$%VLAg$o5-h=Zo`qnqmYcDlQid6q0B{quE@-y{ zUHxA31V(GBT*4O(RBf(eLuEN6T!?RT7llRQH@9z@iv6@DnV zM3IRdn~p*QD*LeF(F%&3mW|W|O&GOLVV^;35u)sno7QiKLTI?-0FVh1RJs8dzA(qq zxskH-Yl0Ryq>kgvzbFLdHO>EXw z9c0E_k-n}nn<;l&ruaOk=5(JT`zFo?9xkWCn8S=Ba~?V;>j|4sEe+JnC|kD738$z< zXMg~T20u5g%`w;XNv@7+_`U~}qaxAe^QP(6K*g(Wn5u+_;`6HrX#I;oiqc5MMha@1 z6HRzfkqi$4vslhl^UA?r)X46yB(i#@iz5+c(Nh{Jig4ZV%2EdCKDHfiz2mJlyfNWH zNU%ak`;TkI_f!UruQa>kHKyrcf0_*SwY5*?nQyjk0v0_iuX?wA0NWb!Irbfv@47`6d z^#q3;s1pzU`?4G#d;T*$(A%3qsMI!52I;xbV^dBIt!?Q24W9aLDi2X8pCEJ$^(}~u z93Fi|U9v)_X6Wh?4Xbp2!^;-_^RPb#d$ARYja+P{!XIVwzRT4&c4nty!8cJv@S!Kb zLziYQwE;bh3B~N934Y|a|q7MrED}->PUg-7fNSB5j1qRVQ8c zz*#iYPrjEZRsu3|T_?`-#CaY#4|^TQ;ajK;=-x>6Zm%GQas#y~GlCZ!LL-L`c8yMK z>{shn857E4k1N-5m!djS?jt*j9q~;L$8gajq;ZmoW|LG<=_!%;Je|E0EmyzPZ2}tq z%2$FZ$BU_M21-gHPThbDD*idj1igqq)5Or(PHLXssHF4w_Bayi{XQ7cNY!nQmOx3A zAnrmu)KYXs0i{m4RMniEqDohDAJAMLO{IyfB*%D$jD;jvH&QcqOFq;3{zYz7eiSaJ zo@mF${uCB8dmbry zZdF64o*9xoU)BIUwjF!#cyElI36GT4U5@CEzoT_@9yjA>vIXY$#*l79D3tI~_?fot zn+2u!3$35%m*8r!VN0{zMs5OblAbAwb?IU)kO`d7;;+rKMpKGMBn!P3L%`|Z8}36J zP)v!R;`HZmn{h4fD_5yHF83e~8cv8cAG!@MTWijGz3`lc}eQToP zFr8(gE`!PWO!-%@di~koIRhXIL~r>WYk(7J9tT>&$rl1(syIrBdO(BN6;hdrO2sDP z&r{U}a+G_+q_HQ4HjQ{ed z|CTiSkX4YOw=F9NsfP@tYDiz!kt&MOQp;*`N=}H>jeV0`b`8Qc7)=;m9RE4YJXym( z*$u1Rr2*U4ux%aUxfaCJA@M1o%#CjGdu*NY=nZEV%NA&MSX|)^qL96D(!;pqEQOb& z`>VBPI`l`!s3R~_*t+m001BWNkl8e@n=-bI99iE5Y2|CddrYv zqLq{_RGfVh@?>Dh=1T5-s!f@Led0 z6|)eCH$=lX8B)>ux@P(e-*lYk=|%brM}D=J^5#v+4&oE%Ib6PX&1sjv*9MrFsPgbi zjL@#@fNFzqgJ^>chBV68)d&KUr}ZJ&6i}C_Fn4g<(E5(PJFs?pzNBLTVL94Xz?sqIe zk9-fR)UN!0Wu1tJ2SVBI#Q^bCZhQb}BEKxCfQ!UkSGY(1<6u!-f`}y1i-xMS(cC z=4m6KK@96XqVb4xIqdjR|vSqj>kT5UBG#%M}~BO4woLe zoYaT3z1yZDknMB$xdt;=uiGp?NBQ-H&)G4w43}r1ta)Qe*!_Xs=y_=FRq6?c09^ow z&+4*lDY|SBRADF9=>WPzr9-G;5Myi%AJ-We_k;FFe6Fqy>LI z{ryuSJray0EQw(cN;Sd1uiPw$WN>@}<^2sFDJ%)fR9h%7Hkj6Cs#3<%xbGkQpsHKc zY7@&&Q%XN2g}$RDyF&%)k3b&O$+weW1BIuk^#N!W&X z7_-y)`UdX1Vc7Qv`u;FaQ1MDBHK#^ea5ZG0duSdU1#S1+D4P43cQmgpw!+>$AG4V% z4=JSuH=IpW6sI(&R-rRbLLJPw9*t~-Cm!d4j~^fS{^L77KJ4rBdEh)u=YAX)j?<9y z`@Xw3htPx8*LWU0TIFXU6fijrI5KYt5=ulzYpYD3pzq&Jr4B3#1{_qu5B1Z`Ecn zI7>(rhaapmyra`o`OGdGqbU5P7BGITL?OyTwh^uyDO8|9 z4cC-q`-L0kKA(eH;VG^(@GYcNqT?-wp-Sg<@g@I-vs_vZ2WHM8O}MKJ2;1L{X!w?D zE&OfJJ)3FCn1GNuBS))^I~sLt+|jvV^G0eTvPLSumPSed8U-;AE7SlBp#>ZWq2xXz z0kfd_v$7XXrfsCGff`D<)O{4Rf!fQS-E=r5(w=AMd7-t-&eK}JE}O`T-XBW$cF9+q=b{0I7%B~d2GR|R9ikoJ4H4?g7zP1xe--P*s0#v!G*bFZ%24v(3mrhi zr53U-Owb;|-8NM2t|}1-dh7OC>%h1Uob=(b7MO4<5y&{bKG0j?pkFx3Db*+;;R|KV zIy|3px?E_KiK-UWLex06d9nfM^Z9}2^8>Hv6R+dId0si>RBbs32uYjx6O3Bf4u_8? zp$PsAYwl<705xwo+UygufnuF!OO=IwIzjl(_Ssr>&T-B(1-ds8MEtT)ECnPl$* zz%sz@oluzpWh3_{? zMA|oRp8WOpz~iyoMvEJ`d1^VgoD_1Ut<2}0NrYrh|EJWY4o_*xFR3?d%CgiOsv%_) zbi4H{GWj-e3A)j^#6wz#`DXO5Y@}A{A}Xpok{Co&f&;b5+VF5#amn5?QH$cGWE0so zs^fSa-c0#xe*!HwQX8lpI8ROU>UDV|b-I+GG@;Ru#&&GDw33-2V}O}4nq{I~o{w@h zy9W2ijyKoq3$dA|+Xn6y3cSTe%D0*(?c*$>hnuIx2LhIrGVI_Nc>I7A!ePloVmAAg zPK4GNHQkwRUt-E0kty==bh&HTU&qTDsP8}UcEfxg0`$#TGtFW}}*Yb&91}3g6hYx8gI(iC>sLF|#+S zDbMUN(&{7%J3n)ZN!dqw`Q2V zi4ySKx+TfU>8R}fz6i zrR@2>>5>JbQUc0pRaR2rBc`G}Qrn61I;5a|EuOTMjJPz8U!VEgZz6SpdDv8;a(#D>zaVQ1-r^{(Wo zKX-N7pWi-R(Rm(|uFoMqlNjxj#B_QmO$l;drzuHZFT7q)+ep1$CI@$!{SZ^FggKuY zgF8PtNq;>)+;)(z~j#QsUX|pLb!OZ>8eE z5=o*@eSg&f3zHBRn$rogfPB`LPs)*RvqUij=HnTJG2rq(m#+oFn!~g@rG34zOZ%+3 z+4jveox{QhYZzHQha8}@zo zY>u^gdV4(Zc-!skWA~zvQ@7gO{2$hwkYcRmpDu-n$o%m>H`; z!g&D4(Qq6ulLs769Iq#i#{+1)O*eT5_0t=v!#7gbP1+}moNp9J!tkKlAOJEm=yHls zW1Qab3lxuQ;2eShn$o`3!M>I1o5}lH=2l*#=J#K@DCYD)=JZNX^44x7C6%vA`{u9b3UH`0F{kHbh~=`W zb7~(@nkWx1_UyZy!Sa$iUSxWKI>roBjF8>>{oILziafkl(&zN#g&BH*N0d`si<+P~ z9&9ct;Td(XgmT6%q}(ukWKu%fwdPw*3hF|FpgWAAHh;1r%TWk~34NfPj+4w{ah(cb zIgiSic#%?jZ#e-v3AIoUmkx@o;f(1PKAO#*%F-Nu`q;i8^CN8@qfo+!eWGbObpCk| zLuiyn3VHuiDNxk?vxW<4Nzf&Dj$Z3%)XLS=H2WOCuFS8{bHU zBnfN)CMX)KdD`Qd=pGfLP8yL;(V)};zM#}?L$48RPnjtcd2{!S!|8o1llqW$hOgTY zdtNxx>Nk#t0agBPo0C5PeJ{Z*B$^q9=$CupVGJz;r@zVR$~3$mt8T z259rI_V<;IOr;|pTFVam_vcG<$OLjDbwYd5sPZfL_7q_(zS|TB zgh$!8TXy`}KvY?QrVgH&4V1DsR6+a7jbUs|>^g@qFgPW^LCa8PhpAB#IqPUloMsz{ zbHM~%1gI(`>4YgwrTx>DzNk6H+hD;64{D1 zu;gkJo*jU4Gg+1`-7(N`fzeJu+ZuT5Xj@0yn{A)oc08hKdfP1I@cm&8m2aOOkB9wu zFB_>=vErP`?Zk7!kecpWBZU%hTt^?74^(_kHDe|kDVv~bd62FDs}cp6O?dgpaZ=m)7^#kXnOUPf9^v%ucZEY^IkrFXlR!j(`nQoq$ z(8RWn*ANVGG7m?j6c~+H*9F5jti_9nnm6ArIRbRhuypk}t zNYpC80+Z63fFTQ6<{A>2Q5QO~=29A-X}c}OQ&a<-#Z~N|qNu95QZrv_Zk1#V|Ni`p z6Po7aPwO9DzRoMj8({M3YYBr~Ep4mAIHCfG@)N#!6tqnH^~+nRPmNTzb}_c*eCFHz zW1tfQo{XS?&-10+6Cra?VkE-9Ym&a+Hfy5VhTR*f4q$`MMk>O8UE)jiigizv6@8LW zDnlfc5(GT%%oxo?Vw z`NMobqaInZRJo-wML=z|Nc`9&eu{1|2i<-6+rY$zE@xoK0q_CR!yBs8evU4HPTMYe z{bF(|oBk^^bP!_65R$Jb^5L&|4yz1P;n*+ohR35lKt{aCSeCi5itFU7L(%H9ilNDx z%ob$WC9N%+fH+sjuo~kEkMrxm%g^rDap1fThdzx-;$@TZac|(w8+BzHKBQ7?^+{35 zLNb38{e+LUvp zU4270{7kw`I*Xyc;hOb=Nr1<^Pos{wk?N)=Fyc7E8^y_4uuZOQYg_aNS|#K%}Zb-`)=PtyA-LMEFW(@cAwCysOAI4_*%zddy-83xVSQks>9W;k&aboV}VGksXR_{<=cb ziY=PV!3By>F`B7~M8ClRidUDa_fLxux`a{Bee4ACM$E(nM(FB+K#Bh-J}D#^LzYHL z|2vIT9X3Tipwc4i&cRv?K77(F=@A+jdO=_>rO5K@U}6JHV8_lIHgBRfwuY(^&{UjB zDUQCcBt(A(C;Vnq&a%niX!4vjrm|&&n~hbl{ut&jZ4BA$vl-XGxp!RG<{5MY`?lfe z{(WXlht!;xg$8z8QM$LWN;0*M2yWfDP9Q|3(M7paH*WzcsGd`pjMAFJ4Mb|*bB+ax zKgZZ?3FbjDl={{U7R00`cfU@1N;EFg;>fIHA27y1EE=f>W=(l2B{{a$uH9vse|7*> zJb{?kz4$8jJh3h;`jwmx2mmM&{E{E^kEEoZy z;Ndk_qXbDX$7iFN-$|rCXvBngR3#o>TOBY6JP2*5zpUovO~GA3T|iyz$2^A*A-A*TW;EwT z?&v16fRHsdd1(t!S+VZK=PiETsXs;1e~MBEpuj8^s1N;*B=Y{HM(V9MNbgf%C|+EP zeY>3)of{g>7lW)|h>d`hErNPzEO9nb23_zaZ9&JH%#W7P)ENvp@+uIYJfGaq{9

    1H-)#H+H9YU4d@SE;Rn3=iOy`$ce(>0f9%tw zuz6C?$3@H%IpGA4VD@uyBnFf`ZS2`__|FG5(xaw$A6VqXiHl_*&{gZd|Cq=o)Ornk z2s){qu8MVBIqcSi2it>KzB^yKeA_PwjTr1d!uPs4z7}}}UyHFR5`R7h^m7@54Tg(s zcgaw0Z{#G7!S;y;x!-f>>03~uPK>b_Q5h1vc^uiO$cPj87dMt%V%AnFE4*hAn?mr< z?u+7RDm4bsY&8~Q zk70x09wi~Qk}hxofEcBhjLk;s^1nU-#xlN7(5@!$tt{bxx8>=1|GfzciHpSlZp|>N zoiaO_hx|8ZlA1ioAZCP$kiYSl8MSK5YxclEzG#5lsr$=Mzf}puy%>91t`+dE*nMiGgE$iCeo|CfD=cK$4IOnX+7wDeJ~NT&%6zoc0I# z2|1<~4>RhT^GBBHm^k|Lh=#m^#KvcOpi)mKxiW zFQ8D5R$h$)w|5&mMp)YEZ>+}$wx8kl#{yY$a#bez$f^3KYMVW=1>k*LY+h@^NU|#z zU9CI(0;vYP4WyF-s8LSSge51vhD*fV5UhUYa|z_VzkS`|X0>S}`33?nQ||eI=7n>$ z6vtQ+t0ksF(g1~EMcVE#>MGZM*WQfzeP|Il01LK{H79mc77E`Nur>#6yoeTOr2gbH zLZDZl00anp3?18aV7m4lfeAsmt{l;O(^Y4+|J9h+I+Fb`8C2hazSno1&kxe<^(?mv z)o)9oBG>uk6P1x#1o97kX@+%UT*PZ1Q@MgF4S2#rqm?#OuueaqpZg8}@qCVbiYGV; zqFksi8+?BKK(~zQ1SLg%;76sHh}%l>uGRe9c#XgB^><&MnSb(C+@)CDz%@saK_<&) z{F-{(3bsnK)?NTlC-9yC&BK+GGy!@T;At1!TH3pwkh=N6qIts^{CEGua%%%Ktr5e= zFBU&?sG|V~+g!A9z&1a=58RRGeRvbqaDt8YDaHbCWJseMimq0{_x^X`tE?&5o48nl z)O~HCk#e97DW#dVeGE1e`d|U%I%+LQ9VlEVZ74Cch%{byIk=8`>Km=~$<)kOM0q+R z*m62ilm)f#*BWtDbv}Ipl+}(jdt0ZQYF^*cHGA1r?gd)J#P@Fx>?q1*7R)b+H3ubW zMTN-qwvv>TBKgy&In?p0BycOs+zX(d|2DALoattsLfxCH_&whd{6$NOdULVQ%cDoi zY2oguete?Xi|H^+zlrALvkuQv>}$>Da|{B=W=|c--A#6lcDc+;!28?=C{Fd=N#oYp z<9w#d>y@ZxwOgRz8tSagLfU(e;D^Sso|Ki#em_4_tj=w>H*LzU=b|JbczK=2f%C~B zbRJxt=+WBrZX%BVo3;3!%erS*k0Qjb zLr=BwAx{p%-TKLEIe*PvG*Bjc^%ZxvJ-OR>Es z^b@hUOq*4;X!UXUPNd2JFR#E-Fhd&~kxeTxqOHAl@B1`5QDvOaHLQSdsf4fmGMq}U zt`Do)Bx$euu)wGepP=`Fv7nhATcICL`rWFTg|_)Wr&%ITkq}}k-Px@|7_8{sC2mR@ zCpjnRpb{Y`v17RiSw);q?xbTgdRvCS-tVG*$15^%thWI%+2MCI%yB!QLwjMZCVL}i zbK!1*#tF0Gbe_-}RIpf`TcOsff%SZs@6 zXqG%wXRo=Rtgl{KY>~Xv91?_X9jWo% zXUsMMF{x^9J$NpU>w7gq02u39l70M#+~KRei7zjC;!c$X6hwJo0J!=OAIF$azg4Vy z%3UYW&C&ptQ+p@L;BZ4xY)?he${X%?KRcE1JE^W;J)E*CAm9U59Se7YfeHrftW-fG zFf6Ngt(SVk!R4IXzVO^4&{B5%xAB#KlDBzbf7B)a1m%&VtThnCXK6l)dzhxQpmV>i zU|8qOpjW9!ox3=zaC}fY*#NTZzG;df@(+y^M2mfj)GkzGV86 zk@m-B|G+>t;XgR!|G^&_c6P`sy>+u)?YQ=((X__2gH5(gv&dvH>hAAh4PTJ8n|QH? z=qC=6_@3Kt2I^IDJoaa`D4~kyv^eiPQzM|0SffLj*0&lJn2c)AeuS#~ng_ZL@R??R zx~xyCj(RT`NBzJ+p2UN?b>;DK8gIV!JuJbE9=tbQT>&Kc%}F-6oByoYD(!^IzY%~# zH}R`zX+DIkkk@0jLoN2vk=~mn3DiU@FTrrOK_K+S1mygXKNEkIGRIf0`L3|pFC73n z(7EQs)Zs_##Olv`Pg+g$l2soJOXb+6n`-ScM2EYmkdC}?8B)NZnjlRdCK*5}{IE?{ zQsDga(cj(Xv)HOO@o1(QZL)lURW8;Y^H0qz$ubFRS>D)jS{N{ab`jk6`K#p^jDIMD znI>bDmIF?!F!9C6r(=CHM4BB@9}QxJ?|S#37WfceTB6A55C&4sGJIz6e8@_Sj*T!I zzhVSz^U_JjzXUBp^c?}|6E%!J>rZaw(wW;YwfP?bWzA5=PtVBxTp02PL+GT1KR|mhbb6@#lZ<_TeNgDO*5SDDjrm}% zCi&^%mWhJ{OX|ao{A7_ro!v;roC>uV=^&RyhveJ2uZc@d=~U_A;o&5_GK3vZJDlFm zy6D`L2L1C^x!m>p*3Xa>vu*f3(fi*09u_dhFoxayE*u0V+;i0fvo+>CW6OT;T*98-vuMhB;zG^W(5X`CYBRTYWn?~^d7ml+bzd6U9 zl!Ryo*44oFmYo`s!;UcA!EQ9Xa@zGg=%^!l!o$0p?9Zq)Hvzieuk_sJi@rvY2tVHT`APh_(wp8|-)Bi~uy;>#OHdD`J6 zSY^tsQX-t~L>~>S5!WAjdA#_Y2=HWZwKw@f)NK`#olV7K+8jENNVPRC{ha0!7!-t9 zeC(^kIlN?Ash;oe=CmsHxxs8|EJUxo_bdZ8H1gq#)z}&u8bj%vWWn%z2h%qH$<{X@ z&_?PYHWUJser8LzTtkm8T_Q-gwkdzXl5}7nU~dFQ;YPMoC5d6e{0W0|w9}MpovsKF zNyG2jbH*Gb!=%6%m#<>EK29@kTm0Pi$QYwy8Khb4g~!?x!zZd zj)xnnk47>2*dcShF(g}B;tn}7i5$%5+nE-_myF<;vqmVuSn~)h8(Y*8nRf!H`GH>R ze}k`AX=+M{VyvQZWJ0V&s2ro*C@ZDehw9;qg&61`cC!8tY{+Y zK6M@c2A?_4V-`IIfBqu{pMr^rBpSFU>`*a{7*KHM7%vTniQtfp3+=kz#dYC``YSLDNWj}gx3RHdwemJBJ|5e1_ZN5Oc)oP9 zTj!HLAtse$P?o6Xat(@H6ix~+WTD<+J!gQBxn_Z@-RpXQv&wYIlKIJJ$>FU!tG#C$9< zKM9p5ya-q36Mh#xOWIVisQ`~Wtd@!qUg?#ud;|PBE{)N{PiBusneAF^kf}J7!W}WH z(ud^QX!ZB^=TwNKz}ZLr_#-31p|rB^kdyvShQV4^%v^BgsCOYqDV_~>md8L^R=EujzZ;gtPYObw>jisAKndXI-C)o- zz2J@bfaTivLC-ryoPpDs;wfTW6(epO1tq82vvOqPr%P}+F&ZybPCgJbVQS8tOmvG` zuR=@Dye;jt^O;oqTiGC!!>2!b>)(0?(Wh5d^*7&LtE zZ%^yp5p`a>8pJ%NhQWl`q+Bu?Tt+2SdY^=xpa_xWKWYf=mJ<+!k38sg_(w2SaARYY zmHThh3<2bX-WGm^>?zI4xPSAAI6wt~A7!V6cZYaDgpf`1@7CJ;B7%NzLl!bzrH@qO z!?{@)dUk~b9MrhewrzDhd|PV?enwj7aP9LNwzw1jqA^Q~XVtfwJnHsXNMsDs#9kQ! z({f&~v7bHpt#v(|bPw)2N@==uiBot7-Y)u*1L`J~OjG_6rhDh{MwB9pno87@F2KR8 zE0MIXP6>x^xQ#R!(#a*%zesNb4mD-1cKs|R*~UMP=&e&&s;Oel-M6=y-V{UV=|i5= z)ds}>Jj`vHH#kg>HXkWA7*D{uPKUGNU1_N3S!R$^tAJr4y2e~@^|_g2wQrMf?bLds zwzs(cE)jK){D%syoY1F1-AV;KokKYE)^t!K#>Yd&Q;Bx@c+jUUD2_ zj98yf$St^jq^Ki6bsw%ZMr%weI_U^KrFnUNYUkpuhwYVD7BsYKSJ$0A}LyS2_2Y)UueNA!XYkC}>r; zSTPY+LeDTd8y*LT!^?F)CpoLu3qMpF*2$hQIRnG$aC)O~<-XbeEV-Z~C5=eJ z8_O28IazMG+sSt03%qj&(>EvY+?0S*RRqok=|U{eH~Xi0KqLKG{5C*sa5_i&@@HiP z3XrA#NB4OP>Eu`WIWBi?`b9(o!i!MwZk|XsdkNG$Xh7wJ$4zx8qJWX{E%R`Dl?(ohO7L?FIq!hZjOxe;ekd zTMTt=a(tS55W z7fw5e>!wOn!rD#jyKP%JBT5HxPq_{ywJ`DCd*g`L+oc^V;vKWSfVPZM;!m?&!A8~s zFNH}*sUd|=OG7?r0)j&=a+R%~?LL>37mqEY^v-2T)#epqI`Lb+P8M_u*xMP`tmH@P zl_=LjjoWLfCDA9FofzRD&op+qxA#rL6-SD67)2}iL4SGGR?@NUL=2QV)P1yl9XNB& zNkx~iyF_Oa!l-YrX74;K9c`GvADh%ctHtbX4CAuHyV@BUVmzb29Kt)lBR1KR}{w!XC$;k`fgGx~2D zyDmPFO@$sUK7QyAzthdt_74}S{1|$BPQsS80!zB(N8?%iayJyOyvvtG-pz7R%*ha@ zaGQvTlX7EOggkw3V1Xh9!tZw-XRAIQ4U)gO8Tlk!No`R)*)=|ncf8az=(ic!{%(%P zsgBL)TZyNrj+RzT@@GTB@^V2Nd^5|U$6rHYC}r>lH@TJY$SGO`n5jn9>sgp3_ca=j z`ld{|MR2k6LwVHm93qp8yPKyYWmS^sEe zBW^Y?p3>x)57Yo30E4t;9~4~Okiyu=$25(JV+_xE#fD>ke(=!&vAeucjGOqq7Xt@3 zMLnw!_Khi|kY((Py!u|j>2<4;-_ge-@{hDzW}Vy+1G*1ffkU#R6@Ow9?Sr6q^x&*! zrQ{1D;c~J4Iz9ZuZ9*u7U&7}+71I#ZpC2cges!|)W8~^H;3skf%!z*;=Q)W+0-u-| z;NajuQPSqN_L@R@G+6lEth~3kkYW-w`y55FjCBgTE=!4KKCQXA`4pwc?k^gOyU{HV z`(&du3@UKJQYtQxdDLEh1eqHbh*_$dssB$$XLy}QkV;<<&{k5`V^^nd^u=6yUF~N2 z064^4`WfV#?_Au;pj&D`48_tNxPc-MRXO30{~qdav))=vx^1vo?CS{Wx#4YaZE{}9 zn{R#c&7OBl&7ZxNM=<9(G(yO)T#2nri{S2=3f)r8)A*si$>QyiynQT}X0jWMbU!~R zpxa#K)G@4(%)Xnc2mO69Et+mZS~)px()&Enbo;bc;DC4<#pm=mWiW{Jt=jbZaIbt! zzn8LX)hZaR@A0m;>F!dq5Q&1_GXkx1+8SD`n4w+>R&dxk`^=pcYU#p zm{@dtqj*s&qB7I`ApE2!yuq8#-DV-^`>PIkU!xR0Zjk*o4DT6hK6oo%sW8cArOu({ z?M-Y?M|fL%4u;7f7R2P&kzf#eJ~!ZJZ;?WL$p!p&Eo8)HFo`BrA7GMz9Pufhm_58L z;Ie0WTKBaDqNI+-$rp7QM$J~6sERVu@-l$g3~LKaSy5iRm~3#;bpo#o#skB?DkCT$ z8(Uv*y>ib3$7GTKOM(8i&QP{Qe1ehkXxy0-;?`?Wakzk>f|`>rZz`)yo~bg{7Lo{; zKB;hqOKA@=&9bNaewQ1x>7BgOq9b0N!gzKqXKtL6@9?SC_Wos_MoAY(Xj8=R(n-5< zb4}pvud)7ZH#f*}x-AMP!3rStbVH8cJGI_G4+ZkAM|JqAZTVO}IL8MZSE%X#V<2QZ zp=Nx!>e1y)`8GFSP1$M-DWfoYXkBaB&c-8(voVFs-r1TF8Q;fY!fJPprs>)mP_#gw ze_QTPl?p_K5@C+&$Zt#83?zJ8LDa4DCAZFvtNm%1ntNluKEPNAOsy<3AH3M$Brqh7 zKsW+H9A=-yz#wZGn3Za}af_!NMO)zAnN^F%H!CkcMzNRNFM)Jb<1$mAsV+MzZEb`+ z-*?Uc17uH_y)xe8)SR|^&8`jH_#PS-66kT!|M;rzd^MPO?JWZq=1-KlOK0Mx$a%N% zCNMM-7Z>UVw^ozHhwvr()I$98eh|3fqn%;s$jmXdIEPH{<=VCKlrb>9(3}A}@cyIe z>lhxc5h5mW47&q=FU&_;8?k0vj}*N}Q`zH=35b!^*4$(^v`INpwjvM4XV;g_ioYvt zuK($cAz=D58WMqq%jdc*rdMUCZBJ?Z%{JEA>*mPl@p4>RK&0SJrM}~8TZ=bc5eZ^H zG~hsq`7%Pk6Yc5uIm-jd29xPB`LXUVwtN3xAC~rL`XSCPkJUs0=o{TiqG5J7TB~FV z1e!JRo6X@@&c!1sk11PhBBHJLpsN$6%_FwYGsWua>P7!B*;RH|oP>myTL>QU@-kA3 z=nSl+OO|qz2yM8N>YeJQR-lP4j$hWxHAjm>w%GLm;D!REWJ6)0@!#PQbr4)` zj#lrlk=ht(zGcLuW5S5H{2@|o_Y%^01S#5p>_4ZkFM{8+?J~FjheZkm&zKPao{&PZ zrZY*+M^hjq1Su=W80i4hfch(h=D3HrxaY6Gp{TgHg>J0 z42_vu$}Zkc0nI<1H7C6BXE8?y`&0afX+@ha>lX8lW!{?8f`(jfEt0N(Yd_2IR<(Vn zd7#bh^B9kxzdj@1SvHK1`?1I{cyYq^`iW3QGe%}o+gQkgJn`^Er9~gH-;}!noQu($ zE8}2!LR4nDJT13kQB2mnyi=7g%CabYG0DgwKnyBT*9O<1dUrBgtJ&+{;-KijSJ7Jq}Zgz@ehBsuS zD2LM+F2}r4p|~2P-a18jl?t*Kp6c5x`OvE>cbeTS@LVP@FvWM+O+x(I^iB!gCaqVJ zcwx#AhC+x&WmGVM`!iUoi5_2EL=SXNlUsW|oBDg28~|y+}N(kZIKz@np*WD2S~?qn1utlwR zc>l3CBa^Y?fEfP%{`4Pij+8TbXk1rXuRZ>tiYj5FnzQi@Kt+M!)LGvpq1*ll?~A?1 zo1mHI0%;D#bpw?3W7M_eKbGrR-Bv0qCYMfMksOzG1A=iK>i@5)9H~i`F9fzM5ICf7 z8k4yd9HgF1PHJrwN zhag~oSmVjq+g=qF74f+&3JE#S<5%dE$|9(od3K3LuCab1w@D+{a3*iE=k_prfM%)M zM*#sj!`)xGYpvd=t7iICi2eC*+@0Rpf_qY!l+JqbqzgQ@1}65?lFvi>bHu%D?b$fq zng*{3e7JE=ppnp@twpfSYS^tFLJ|rkWLBXMfj_#L#UVP`_2HaJjq-oZG4eTF69$At z0gC0%t%tH0q`sQ9Gd4=T)DsE80gbp3*w<4aYLWtC8k41}WOok_Cbw-A^78WZ0s=Ze zKX6bcDh}K)BcN;b2tt`GRsDphPX7v2$e;uRrlJ>DIAh`&reM!4P0WW-w5_tS!6=&O z4I1m1rC*nra~~}Vw7#Qsb7>_=D`#*~sAO=(7SLzh>ucTl!X% zJ8J8pAQ_`^5m6RaCDmlzUcG0eB>D0{??UQVGIEgAnET|h_M86p!~1^AyLn4u-qT3W zH|=$fMq)ioGhX10&W%)N%veNo;!O&;+}S`FMRaH{so7}F--@Vj^eT;x9bz1=l7`t` znqYuF24>h{Wx7}|>A$LpQ`&pq zQ~0o8pm#6w!|NVz)yRN#nsD0S(C%j@qVrs96z;hu&))aP6Y|;oK^o*!7~xRyOmu;r zGnvM=llgjHtXJ(6^*fP;GMK!!LnldB3sPEzQFt0TQ!1W?1!rFnMNI<{px}qnKO*}5 z-dbm;S`cEr3w!tA%J>iwP1uV%S}iVgl*$aKEq$#is4j0>g&F{6v%nybO$jU%L#=-~ z@c~6!L-uQ@T??}O8i9q-@_T`N5o?GFT_uUMl0Qa+{flxXil;}*a(&dOgpgi6dQr9Q zI0kzeQAfHK8NWG#C}tunV=0my_4=^>t-`ft{J-=e{2BD*DFl2a^_|fap(r&U$=0cD{qsEmBqT-89*a@;%#MX%kKUGU;mI zwIsu~E&N8!=`p@)7!eVh)0{GcYx}W)jmfYHb+$rDm*>mB?69*y(-u#KvLtT(D)I*7 z^NcrIcU@L5@#OYGzhcBF;rBV5n_`fuSdXqq@S&o05fcJZ^Lf3aZwFc!BXLjI)5aVR zmB*1WWlOY*7;8Et7_|8~<58;fDb;}uE=!EgH#qXx3er>KG>cRHmTH+V2in9fIn0Rp zX|iQQ-~crPo*up+4C|dF`J+dEK6N_4@f17X{R4wa8^2KeF4Ox_w!MfayUH{Y zud%-prm@f2feEtTEOEBTngT;2pA#ndh!m4>2i86=Zz{{OfCw8;V!~7FX@4KyBT#rE zP|JRd1PH2B4)3YlYG=Ys4j*gY+(aKl2Whfhn|2;n_D5!$TQ88i(e(>YV36fIiY=d~ zj9@ldv~I;`pQN1MER6hM#JmW)pv4+X!(8+E$LtVdjolYjMA1QRCCVl+t-(Q2#e&gu z1Mczl>bbm?@EG+|V@gLy6O?EtXy;MA9AkX8yevhKJ}t?Hy&uG4tj6N!kkCzFKr2jD z>VgX3dK=X}Gx@^2c*|Rck6S0Gz{oNO(CB}@;n>yKpBQ!(lwHSG@b&i4j%onrt%0LxP*Tt*gJll9q~y){TF`O`Hj7{4U!Pbg`m zPI{ZipXP$jI3~77^fpmo*q`+yT+s&SH4-rFG0v^yo^))-U0rBIj>FCLxC4QgmR?l!TFVl+S-#i3f(2?lTJ z+dg0cGR8SEWP<@1llkHdh8JF6YCA`lGArCysbKidQWctHlP9-@s_7K`KXpLQekzWK zbBZNvO!68O(OG&SJz3&{P*$>pZB=^ve%Y)Wm&iIBo4y-oWUo{G2|rhy-WH~qjsVZjQ@o{M$?6Yev0x~G<5%nx3tefo-V-KSf)`SbgsT&snprWj0 zjjo+z9I>0yP=Hq-33)iLF|hA1JCLOh`Te0om1A!sS{?PAehfQTnDrOsa7e(ot7Gaj zi?S>w`7?eAy*irw^&$FJq1lAp7=~^u4&QSsMU-TVRn{xaoy1DZF9xVPJKrsx&55Eq zXb28AWA@mIEIIP&_yg*Db6zuD1$>jGLey!xm zYd1e6DVB)UxJi;u5fQvlN|Hz5x3lz#wNIjMJWb4i=a+}G)v2_2h^uAuUkCNA4Gp#$ zY`mcdqolu%=sug3uN3T(=+ns$#i}q-=DNxC>}#{jhV;;ddWGZv;WX6acwBkFB|G-s z6LxH9tBp#`=9znbJaj8O3W=?sqXQFqj0j@Ah|-KvB5XrV$*FTsT-L7vb4p?*DWvMt zY`;&2Gr3PoG3FhnFk&Y6D-!xok9`uxMhKI_Q7>t}V1)N5p-xAD_-LOg2W;|t{Wrb# zW`U8D$2x(inxVJ*QqqFIRs=U51^*nG5t%-X@i7YtAmr7UmRO<((#=euG=tZ`%n)yy z*6Yq-@CDPCG~V#eXO$$`d#|qJ9a=D!&U3EpNz>0^h2qU*gxW1X+8x37oBW$beMH@;e%gD$Gm;OuS*o zw)uL*XZI_Cd_aJB2;0VSDD@NyKvXU6#b;{9pmBPYnB&wEUXh-Y+xKr`H~D!OhEKU9 zPmZGNds+N{P|J9C1puU8VVOi?$P{Ersqb+~A5TqiF;@PoNM~Bt(dE_ozR7kH4J+ zB_PNu{kz#J9fKC%V@mknLteEkJ`mi6*c|tHkS_RiwS8})p7s6h$$ht#HKS2X)(y(EXpHm*IN(X*2uSNEL_&dW>Sv2V7GT*d|2)&y zKvaZ}`6G~T&zlz>s*Ol8T6*rYVDx0lc7&dn(_?+srO%*nBK#Af-Nhjm}Y3I6MAprdcUA^wI|FUK_P|GMrIJ@BX zR#ESlBp*C${yaZFHjw zsyufNs!Og)41s)44!a=z;@hdu;f*GuqFBH=HD?r$uYQ zPez1YX(23fOFP0Iw+Fs`Z~IPdpMSI`#MsE+=k*sEMWl#~S+GHQC0czQL#-7&i{(=! z%Yj4NzXjhRx9e^`S^RTFzpUH!8a;b2(4sRqjP8f^l(yFfJlK#?7W0#)Xxl>~vk54i z6xBwz4%KVo`{!dBhEmo9lN@bFw!%`j%=w~JEH8J)1mYRMJqW|VV`r2CBqLkuIld5( zy`*Jmn3W^>hxOJ=B9n=c(e{XOC89aEm1U_;Q${A#8}nP(F@o&)37YfIL)KHY--P^$3!MT1H{D(BRJ~chRKp^7KiaQHn1|j z*SRO*i<$!bnQ|lSQk8U_uwjJcFuJ+g@%xwTOCb0TXTVDwzw)v{#2*I+@!(Y^{O8yx zw>4}rAcH4R^OU-ZixFaj-Y>=vtw+(PPZ3i-Y%`a@Ww%p>m_d(1$CWNb>ZA^lSmPg( z`TbW2fEdqSHyJabyO;IGyH;j^DxvX8=dHU);o|LX<6d0jDdN ziTS)mzo^3+-NAkE8&PXmDTF5y8X~~--aTXwviGclYxYUnid+O^fFhM3m4*2#9O5*X ztmOS1%x6E2RicWg+u9HV&L?H~&B1W7iMrPogzxi+kMka(JwYc{FnGex)H6= z^={DRXU5tT0LNk`uf+hALGI5{8+S(NEULZuBcdbL+KrI@`ISqBkfD1KQbr1|#cN_V zEu@tnewj-G_Mz_UjuW)G6xJ1r<{UcL)mrVAiC|NI7(vh*EE zCb*T*-fOej*@mK**ezdu@5}FRE)IdWD}M9sADTP`b?)xyiT$Q?lg^dH?@bj8y#oxX z{v}_sa^Dlv6@z}(otIvXzyO5_;yI>~k}dEQ!RLsFM#Wc!37$J43vC;F3#I%Wj~Cc! z+;5PA@gU!ZC@eS{voCN{v;Vjy-7QZvF!U+)z=A%NO z26g1toOJgOkQrp9Ij&qvqLA0@}t2@8EMm~}*JaFRiiR{WoiOK$# zbCwP;y3ssHp6AmG-O`H~a4sRsx#-;PXw(qe_X3|bl$F@tiL@GAxjaUU-w?@MV3w=R zQY-t#Er&%J6c|-Wn%HAt5fp2>Ah=Q1YoFEcvw8X+a7*L&FTf7wS`eMOZ2+d5%Z9Hc zZ{=&{&I@nxaBe@+ua{?4?krU=6c-EP(uFZrf;3SnnX8) z=I8m{Fiq`_Ze2DICDHxy#WB#R3A^l9hb2fh*AbB=R}%xTA(4`O-wl8pouPH;W*<=L zO=vYM)*Z%A!)SwE*r9H-fsl51Un;>j&6P99WKpK#$J-Pn`TeZ2xLOsPiZ!HApu)+K z^nEFiU|uD{z1dU&I~BcZV-iGuMD$ml)Vq~;ppBl$ZQ&eF10D)=trBGdCk8f?#;OE% zW(LDr%P{BZDkH%Khl804&s=`A8BR*!x4>95O!@9`R67e}5<*5kKCJ`0qij}9TBns- z^kh{xl9PNvha@28GyH)U#;VEVV;X7DD}ias64(o|c1{G>Vq$j3s``}*XvQOx+{HJw zt&>{I+;a$TC%c`x_K`t=Q8m9pUtW%;0b%Xe4y&XF+_R=WHxLnyM!GL~sQuO{P&vDs(o!NZaYcrbx@X z0gdQ(5C2@!k7=44jDM%LztF<(Th+vtApLYl(^!)qmyuYB84kzQAEjeh8h)+GjHw~! zR+{^%2L|=;CU*a{Nce-udDpAT`?Z=mMFA=ck zZMC*V7N}-WE(%o??oX0vPqXx7@i5^`moeI}I`7rnMcNMo9hVxydF-(IO2DWtEmIRo zREb01IScn^NGB<#d*w&uQ+i@zF+vz%T7}M5{wq?U$nD0%*JwV&Oft}jAQSsSk9HL{ zBQ9Q@Qne%sq6d1GKZurP0m0daXUu7K(vAUI9#M@1O4H+hh(8LjB<}MJvYScLwVh%W zDXT_29-~?j&kZx}N=-zY_AQp$qzHac{PQ4+ePX&2)haxU_3bYiA|Y-c8((XLy$5Ic zEA6?On4wChr$@z+?~V__2(HCCksfpY;zcxCv4TX%{wwb# z1w4&fI>{8ID>{|jeaDRN~}ZC`vc}vewihl z(mjPhFVWG1sYJp#X*5EPP)RyMe0!4y^yH;tB|LEvFv?1zG9meqnraLJ)wF}d4oIz| zwjwO`s(&NBc{e+_>GPx3|5!uFd6v|n=7-gMGXtT@W=+-FF)=jarH zX%)D8K(gr5t+DE8m1>FjNmf* zHZ&AWn9AGvs6)H!)I-IIa9A(GSygGJgWKBp_gMW7X4roYxUIA@Z~vkpYra>A-I8tu zxZgaV-J6t~7Mm|;eXQT<+6i(fxi}%EjIRrq-~Nek!b}>K7Oy_mxnKJ(Ur+z+l*{k7 zDcSCuZF-(KzxXlv)_pIh9%#xu@9CThS|&vSlG$N<~!$5C=kOg zF$2(XL~bI(oOt{8mor7lH{n3b4^HY?{1wSS^l#(S%Hdjn$gd9kU1>;vQ6~KL;h_JQ zLG?e(NUrB_(Z4SNhf^4b&{2d!p9f;wC z$&_ej6G}gxRW~T(9Iq)Nbbw!x#|ed*T!g+9i=mf>?fz+!pF_Ot6=z4Cw$DTUo_`q4 z=f`;EVM_I}HU?n_leBpYQKoVgm5CaDEgmuOaanL{MLnT-~P(Y3Y&l)liq{v z%_KJKxV-9Ii7{@Vov|J)^5S=(`{orkX)tkZ3)Q4uG!ICxEjK6U4}?{Ub1zb-NFFm_ zOsl3YZ1|N=K7!tpAhL_l9uWiqGm8+Tz2@@I8L!(O)-PwbnL(o!IgX=%fu*yTaI|YQ90U zOvYuYv`H_pg8Wn}0QWnUNM_Jf**HWniKqFP1eLXkkGjR3h{&D?l!u});kSp7T$4zj z@)DxgorC>R`m>y8#>UiFi7WN1dqR<4*ivdKE*exB5-_V~q&cU*_6HIIG1Mmeon_V z@RW>XQB+KoWrh?>4c5>bfsP-0uR#e8%3`haOB+(*-{r`yELfv%y*7 zx8eM{J5p9}>g+^SpfU(kxLDS?gs6xHYkdHoKl$E-!gfsy9T)GqKZ4KiUHs<=-t*pH z{N-Zcf1&eP|426HA8E=8Tk^^%>D(PhB@3Es4z6-5>m5y*Ev1Igz}M(xGF#8_)WPbpTSZw*p)EYpzx zQ*u6$&W#>4wPvaONf!D7gbxEMdAF*?h`HmJk^jiK7^lZ{(s_=BOJ&h(8=e?305K|GqFoOdL;kTEQo{wS>rOwC|v#i%wCFg`h4Xh~5ETly{+)S?0FI#Qev z{~7$DaWt)a6dR%wa?nNCytWw6CA*+`afydPEN{Df#&hv0`_)(0wv&^=nL4RbHL5x6 zcn7@hQE(8yufGEXTlV0f0Xy;FjXWYsUNQn}gCErhDiE4;mcOw0@l}Zkgb!D{Rc@u@ zM?G*Md+~PW7IZLr;xUWDV^F;n;kD}{<|9I>YY$OT?K>ypB{OJtSrU1?n${F>|M>a| z5#H3HSZnp+EK~c)nCXI0o;c=Zn&^xP%kNSpwdTtZUbLq)iT~GWm<%EnftDYk2gf`x zy#ed^s;l9fPLjIK63?@ze&IUp^4D#sO#l?)4nmxAUS2BssaZXXe z{>0V;8`w<~U*z!g%5|_wDrb{Yoh}W)w-n>)v%1%RD%<%v%QKvL0hiIBk72T;J@;iD3QPoJKcf{6c(Fp)WGdwl6_yH zw=--3FB+T|bXLiMgbgZue9_2~h-wr8Bmfe%87ZpgadHDB(P0n-?5lI#)m}cI>SB^4 zt!7!#=GQYuCRW1jN}qM7?5JEdw3i9*S{y8X2h1UNmPDK)AM9v`-L^Ca(Q4iFcMQ#fX0Q5C9%{T74emo z2@^!OlzY^p)th&luSuon14askgqZQ^lh z1e9Qv@{b-8^H7^OVfr(B;0M9)wnTHJTbI8y7;PiSjGn>R=IdldoSdkaz5Q1d?%Q)E z&8*6@6UktL>)z7|<|piwl~$&=r@@Z;JhE3bWcPs+ajTO}ZnE`Ofc zum!n|pi?CH&ool$=c^at%SqNfLtj0BVPRtxI&cQ&>$Q-fyR1^L=&ife$W2Kz+u;>+ zvro#K^e^&Uc4=4#eD@iKrfIT0Pj$q>Z)$ z9HztA{lfvpcF{E}gu!)YltUO9-k#817iST#wvy(`LrQ;h3WxD=|Eyg3^z1ePYG(p!Jk zL6q}1bw9KcaQy8L^lr|{fN6(EyN_&YPz+p%E4K}8^H>=MYnlNACNGNl8IYi(qeIO1 zD9c-7bBr(G!nVO>j&|uLH*Iarm-f+I(tzW5H6>mGM70>tc%v^f@`RN6%70Z#tKASn$8k1MXk(kdhvOVt#{jDGxq_2nci0hfO<|9!4 zVDt=!pmX{In`Di<3u9bv4{I1Y0leI+fXuV1h6jwoeAxPB&p!~5^xm{97aDyQizQPl zLQI>|(M%DDCvloD7xCt^mSY0la8kXk6o(WKRyOZz<}k$LI`}Gi`Ah$grLzo*tLfHg z@WFz+!$9!h5Zqk@A-KB}++lEsfgr&p$cwwXy9G&bcMTTw9=^K#qWD48oasK@z4v-n zvp-#psa(D23J37nr8^vSI;+#aa{rf-e4O>?S7;DSo;Vijl|XdoYR1T^od1j3H3^K; zg;c332PStNC1K`o{U*C|S8Zb(4gCv*vRRAoDuQ%kwE{5uu3*?(ir&88oEyTvgs@2O z#kJO9K<8^XiDb((c1L`YPSQzHohMkR9jcx*j(k)$JA_vJZsDC%q6gtpMTMc$Ay=7R9{k=G5QisXO=uvX@WDV_>S1ErRn#W z(PWY|Uu09~;c$w>OqMI@f0dj^w<C4VVCe=99eYPS1*jnF=md>MP}yXAuI{$mq;|yF`|o=`sQ0i zj~ZntALH-n=r*N7N6$^aUHBS5T4KTZc&URoy^MrlAG^&nRIM$IoLl7QLrrod5NS~L zDZc-wYa{D=ET91m=^c>QT@2rhYm&a*^~i4U>=gJfBYajRw=I!q+ zO%L6ZkIqHRbc=dBK30;ks`FI+0M{8iw`g3pN1!p-s}L{W z>4lWf!z|s(lJ*lYtnaIgF0N*_XG&BjWOOhs#?eD_0?-;!VcC5A3H8o^we83Ij7BN> zXz?y8vDx_uKUtJN?#Kx#2-q>fGWq@7)@1i^2&PonWwqrO7MDGVA-oLd5b(vI5;Fqh zcV9$3mGq|1gC@+Y)iR6^;#1PG{3)rhaQwNKc}3C;0hso9TZ2ayHhGM|b6b`^dY9{di1$zJ+YnvByKY zn(d5E>&m-T&UA*{@)37f_cK<~*b3chDBmN~0+FSW`t|nK`=!Z0@$$5sGt;{gC#JN9 z54mGT`<@I875PBJdB$w8uOQr+QNi5~XR7;I{5^6@HZ!gG!?oCbOI-vzQU71saQrj3-UgbolOu%b=oe)Pze>>A*HQJn=#+D!LN)$zRi7Wq{DD?Ly zLJ(2oo)Sp})w7Q2!~I&yv-4uiZssBOkc-uFei2{szh*bZip;Aaw`Eg+{#5s#8UO z4*YGnGK0yrWIR!@p&`7qn!p$C2S^#fW_FN!r6(1#zFi9?=So*(J%=^i?B&MZ+~{Oh z32gU|a7n+>l01!E&Ljzl;U`AtLDsXl{=1CS)k&(;j_ZNSCzfsd+SuStQ_)s%6a-Vp zRDt|e*(Ci2y?KiAwuS|n8o5w8;oWZ)&|@Ozs2joMqwbJVK;!rROQ>Qf$(U`tDPhGB zc2rlWRg`Vm5ET9w|D<{)RqU&Fx?D6kwV#S^C*O+cK}1Xz517NA2l&A}S+EI=bprp* zU-tBcvoE!YpnO90q&8GjIr(Etg-p^Yxpl=&_JDyb-dcXf=fo#0H=LRAJAUPI_;lXp z6NB#`PfKu*KtG7UFxXYZ%dlTccG^c2`{354=eIhdbDfo5-=6(UiIjjwiH`1qkkPwa z6_F2O+bLvVs$3@QMHPb<42&6%o&Mh(x4GU~(zWQEV7GH#MZ`cC-C47MVT}^?{mKMz z-g!ce69(si%a}>-{rw@TTA4f9A(QN@@E&E+D@7e1+g#6>8JZ`#+(u2#uJmoWUwN@{ zz!~zcwjU79reE52zkn+TGKh^sVbsfi{=2udH#gJ&DmBPprmc&RzMvsldBZ((cLKKa za>8`L`lx#KYM&Ug51@$n5&Nt;9qq^5K`$V;7KDnNc;DNxeUZDrbL1zK-(R(U$D3SS zO8;W9><;3tCbEnjbDE^{*{#^x+W-EytG$DE` z?hykJTX}l5<3PK6CG&_mAq9#culSICoYK0TWj|fu)WR1WFff3jr0WJnY44c!i;$Yv z`kR-b`cP*v6&fdmDEl+uBCWJ}zqoH@H{;?TfuZ1x$6mhMEPsvlIl~T%s-#5_qi$(f zj3rPT&!(iiaH4E4#$3>RMo8#4Gm>ZvNDik=akF7Ndgcb{QuM z$Ph#ZS*v=qk`G)hFSDKgK~Ze31=Uqn4ceq;E$zFbbp8;v9U}1obx#HZ+|7WtHm{z>>m#Vd|w(OcNA>e+5b9qy!S*_|};F zfR(f6w=Qa|^x7(utW|YAP`atbNGA<{jq-z1E03IRA<7n_wTU56ymn0DB7fA-)Hev( z*YyW>z+R^dTkfen*oRsE+g}oqoD(ilNsB67YGXxQvDsZj6+a_`$B{Or zd>!4g$*EMY&Y&M?@on9h=5a5m8uiq-b?Th(dn2pwv8FPp#v8P=d_N3{;+)#9tz)87 z%mg2cS7XWpOt;xhZ+`^bTq{Qo$6VrfrmxC-Y!=2iMYMk}FU4d-@Ylo>Rb;VO;GH+U zH2%|+!S^t|Xz;r^Q1I}bX}H7G#woDPq2i-LMn@fmkxAoe^5acrG$e8(<{4VU`AQpU zWcX2xymIYk=$Q;7-;}SQdjoGEPoh(J?Eq8qjrg;5c zEZ2rZLP7#)Z=>~>=UP6xx>Bd!5N#tqxmxGBqWdITU{1D!$HfU+T(!f0p!+5;CXsOL zUWi_PGbmXf2gdJXftZU(a>Z6?%%>G|u?8DAKXYep`K-atVcmZ9lfR;W%Foh1@!}TU zeL4K>S#_xlK$fP!yu)Rz%e*PhxBZA_*2vJRaq`KO?KY~)cuG=}0SZVIJj#UBruqx!=d zlfVAfa>f7x6Ck8)&=G&+MiMEUGjLP!xgZ;D_!cex`%Y;+B@ocq=p17OIu`Wpx`eO@ zYRwF|^w^;u;LpxJZ40w^$RAX)lh2Ymg@v60cETL=TdA43aI*y5!qhMW`;aj5r5lcF zPhUD7RVf{X?R1)8eB93y8K^MGmpNeh4?+h znkliI*<_--x`0{u5lK_0(@8se4GvbVwU#$3tswArSFgRw`H{x<3tN`*M%kGJ+W zetMnwAU6>-$7^OM_0l_F?t|~Aa5d(r)%FYmnt{~Ek|!d5FctjnsRO$}#}%o3-A%9c zLzc7aKm>B##WRM;D0}*;91vCWKw;c>y@~Xv9PT}BVw&*?9o%F|o?lF>J%NMYF4ZFI zQV)L*p>!4F%^MF#2#>gl+uCQ}KFI5Xf<1k_jelJv9#A*N6R~4kD(4JFZ1$bEfoph;*Vj_b(X#svOV&B)k>6Q!&z-B(tW7%Rxx%6(V#cNTsMjlW6 zO`1tJfu{=+Ebv zZHa#)qKEzxL)$=>OlHDcY4-(z3E+&nT>mk0R0c}z2tTgBvN`R5YsiDj?qJO>+m(D& zXp~xGWRrN{+$q|a^%DsTs<*4*{8fJF0T1yGw8bACTSI^Bs9sXY7^*aRo%&H1f02xA z)AQyB7zvDAq1)KID!{7M>WO@`SQc(76Q6Gs1Y4roNID8cGbS>4jN?4xc1NB+-U5yT zWC<8Co=k_8?!>Ie!dE@uebbP^F&d1Ke7LOD1b5n8bGm# z^rbCYLrTD0b&mHo%!o$b9j8} zu{qP1RB)bBvB9?v7wfny%-rMBch!n+BuK*b8_Xya^>k`SjY{ zl{7te&7{pF!9_}07LiI?UF1M7qJJ2R##xIVh^n0SdDl#=PLSy~+4TU>w=7%QYAicS zjDnFJ+y}`6$nUOz0Bt#n9b3}AelxQ*)%7CXF<9ZW%!JY;STNmb zyM7POb8&1j061GY?4@{ckk|wfr{zEjoTa%U;*iGw61-Hh@5Ln~23&aYE$XxpDO}Ql z7l?l+=>7W?b_Ca-J3ELyD%i-kyY5LN;Ki2l&J_wmLi#!JzjOX8_n61^M})A<^4CL* zgKU@jcw#_yh%gt>T_Ud2==d>Tv!CZ2pfRJE3C+AlSkB8CCAbecJgc_)e$jKjOeKnv zB@T1&WhYnbJ04J45Px#Rg zE6O=6xJU)6ju84a9C#l&BB;-dwp;T+ENdczqWE9V z;(_I?7K%7KU-ookG8#p7u+usl{F5{Zg}mbGg_)PDk^X*l2raNp82?lFW;E#iYoz`k zwm&0%yDhjKzr_K^XKBG2Va}#u_J=t}%X+1Xoyiv<`-UAvjqFI&?v>8t+8ctrz>b;F5V?)CkPISTDQ@TUca?Irc_PU6APdiQ$#s8?k~45S3TB8 z#LaSuv-ceO(WGK%B0iX8*uMt9qX8ffzWp6mPV(QkiDrT`DiISuWzU)@bZ>oTS0k3m z9DGclcG8UO9)ZH+|>WtldG|5^xYMKJb{G5^AWvXX{^g5pzY8kGd%TlBucW*#~!?0mey zB8NR&q!bOjvO8vH-zLekc2itN? z3Bacak1uAry6myAV<>@Dlom3n-;9`n@4?(fk_f?YRTJl;Vk4wBj-oZE%)ZJrvX8+C zrCb6Z{n+5K8T@A-y!fFkV6Ni;AmPN`;5GpdhnVmcX^gYmXia(d4# zUd0y)=_Ls|CCXg-5XuAd2Q`L{>{<>A+%VFGB!^X{us2xuraeZAhCs*`!mAaUs|_}WBI(;$tSs?s>8{r zohLuXUSMcWYpB49BA(_py{}@w*R;B#LP-$o^e49%An|47_^f$BCs(3fN7w3p%D5T_ z1lDQP&uUc}BpcX zvvSnlB@hXW*Vm1X!^1;l8JVHx>;8{${uHeSHH`eJcm`-dA4SaRHa*nK|Ml5JqfAq& zwZ`{9PeVD_0ik{{T&lKQgH$wriC#->X5Qwxg$f}+U&R4^4LNmIiXT(o z9ER)9EA9b{G;HCK+bA4HxXY`K=Y8MjY@PO|KWB>6>XEuIo#`*L#`&PtU?EqK0#5|> z!rRQ~8!%fExotJIpBw>Bhhr~rA@B_+9$n*scFS*MHgAD7U;UrWD0Hy+-B0Mh+w?NK zCaM)kci_qyts8WBZyr~5rJQ#mq7!cbQi`jmH3V5SS$|Jf*CubG>X#AV^}4hEQ#=#} z%1qz>2IdDlIXNY8&a_t_)?P96Nn@X({LGZx^ut~hh2shv!VrdjdhmBxYbOFi6;Lr# zt~oj&pApA4jg*{DlZ^-s(cP4;|3ncc_y_^4ys;AwI)$s`fm=wyn?Pn7(rX*8)Kh~A zr;rI_48dZb>6u3ThUslby&_{7Xgf9KT+sKmdF_Um4}CI{22Q>}H*0pO(FcQi;Izg0 zujcrX4i2l(=Z%?ekr2ok^AJ_GYM~vN>gb2Ypd+%lYMBnhKjwe5Bk}ZU9|b=De1hXH zOX&qjaOWv!q@x9T93Yqr=sw%^-uJ8Tz>$-Spzyz!UgB!cM zp#Y^gOk}NtBkY#r9!rT#-ox%mc>faF##be~I#&Ukn3(9~;gR~1tH>_1uG&h;KKxIc z4OGPTAPxR@?)MUX>kxtkS1FZ0J7c6FKn|-iaol3S*Cz@2PjY04uD#K4;8YYF-r4a_ zITBqGIdK&eXhi_z5oXpuz+<>=HPltzDPb^vrh(@ob2qdwalGlYivkOQHRehLX=M}K z3AH=9eg4vXtAMEA&E3!DlNcBq(odBKJYbVz)$!7LjS4N?L}5!d=lY*oW&q}cdE^EA z*fiBcF%r<1YUm5IjwCKN_etnXCDJ`)sw22o<;tvUb9Q82yw5e|+x8QpBGH@#X6j!y z5=WDl>!ACs=qa^!Cx_oi%^wDU{Sja;veW1MZQ@)_0r(U=U^~CQDPrJKh~lUJ`@&Rj zyQmCBguWXjNRV{Ok7AXLaWUJY(ZP%&_{e&L{`d%4q`HX*TAaKND+xop-0l|}VuZ2M zhexhAM_)>BD(wpi9A}bew4@6D@0>psd1B2+DXA+Q< zH?-jR)8v@rc3h0EugFYYZ&4c46_0IiDd$OU)%vz$a52RpOuC>3Q+8@*D46us&g-i2Q)S5A;Vb(Z9rbBCeplJDVSi2h0G)4UUjh0H5m16uLo+m6t z2zUtuOZy&K4syuTSdl^fz3R#w1!x#@X|j?=0-POoi;LM{FiBgr<1usT)9ZcTol{rG z*H*Lv^{s-+o;mX_(qOaR&3;A&JIEUH;$(SWOr&3-!Yt6-QEeaYs$K0(5AOG$89WnQ zU?z{P3s05vDD>QBf?jAEx3TI2`DR6FFgzF;@a3tX)V;lWVwCi$!JtT1qOauN1{Fgc~?XSqR&`6wO#hKA@1*G?+2}6KZ6Pj?%h8PSFDGaLo z0+u;Zu|(0}XROcugE7!i>G3PA{$ zDE4!yfiWBN4SK#=G)&{No<&`c0pb+~$zVIrk^xcz^q4JL_APHPBL%8hE+y*}>|edR z)8((i<^MJm6baRPQejyrn5zqbNeaIHNC)FjG|W{bp4Q6sYYaJG5K7A3(|l0}!3bgN+|T)@oOa@NoJG*Qo(>;l@R!*VfW_E%?C}9vRcr=4p}| zKtN1#hT98edzay>2rRhsIg|-|`)UC{5L-mcE52bNUCUyl{wIprFQXO(B2w)&AjHC9zs zfw{oriiG070kEGm$Bo=m5b>j7tE-#^O{RcH<%-H8lQQu|8G6jh3To&^NFEOvXh=cUA`85c}etRI9>;_v&M!%@QQKV%$=UM*)98i%chCG>x zmv?um7zVWR!gQCnw<$>+ALFz>&8ej3s@}O={Lm$Tc@B`APFUYq5<;+WFO92?ptsfS zzR6>1v|mOP5edk%(rVOB4nTFDmWg__QTOqTnGB&{Oj@`PlLWzDqCpdwWa!Y&mWjWQ z5hR9e-gioUdyTtM&r!y1ega+{f3P^|t`qG!%oSr@WKvKEUg;D!Q?X71lo2I3=Msev zU`@oy$m2f(8mlH0=Ly1CIzBzu?&J~m8mmoeNoh$P5Anh7MCNLzKegs#g}`2(1Sk;< z0D2}ibvsL6h!Wv+yL|m3in!=Nx-)uaVsSuc$JG<_^c($Q^_>#PJu)r=(=9R~A)#4I zs56Pr>pQ^g0|>^)D^DIa%r=!A;aQgdT=pkx{ledq?SL5pd6Bi05i<1E8jO~3ar?&fA8P%l=Fjh+CG74sKrN$`dq zBFq$Bh;Fr9Qoq;B^2|5mtkm04Z3=B8Ohh!hLnh}oPAqDxSc+H-uF&Y%TUlX~$e<}pF z4?I-g-7hDn2hy;ph9Q$iN?GB_8kG(FLiw@m5%@dFIV%ERE|H<pX)#H$LXWv^Z4l5qq z&$qpG`GruRy-rWKB>?M9{|KzD;&CI}X*87^Z)X2z-e?GKwTA+{4`|C+;D5eurQzcN zD`~|*MD}ds2=wTb@$NImsBo>0<|E%e<7-eD)n@_^zn8CU(`mn-(h7qSR)H<*2Yp{p zM@)^;Ph%#lOOiSXD| zp#*(0a@fO-9?Z+ZN7f`pHdDVZnqSR+J(l`e&a{jABtKUWNBQcL{|qRN63O|sgiXr>W%%C<8)dRQ8iiY|;X-ghX&va*uns~hS5HO5KjqRLk>5c~TSFHuHz@C4X zFG~V%o^y?tE`0gVxg>J%+q7pR%+lWxYAEGEcl@EaBLPEq>wdvEW7-pj?Ex`O)m}G} zXori^YKI zhV60Vk}%tm4q*E=3eZR`R!Btb&C<2wN}AQzYg*o`$x>S(R9DP>GlE;aZH{%uE?>Xc zju(-?qizMWkD$g6)expB!HO;x0_8blz_VSu4wR?NeBoJ4^elRwAU*|UNZIw_AZNb-ksh;;GRG`8CIB?726z?{T)qSgq8l4v5ZM)uJI^CgFbV#^!i_CU zU9$?z7lyOgNN%bnV_IM7!m!6-S5Rv$3Oila&kuO4dF_{-`~2&${Pu*3(q+q~v0$xZ zuwZ)~MMcJ!RC$Hj^(3R2@*WN?ZA}sU(YB1G^~TTvmf+bs1)T98vlP!h4U(<(3~F8O zUZumFaYp|joA-ys=ZRGM@-c*K2-~4CI9s0X9j-#u|C~BnkX8k_q1;$MxXWYQ%IHjf zw;OY$eR^X`w`~4P`_qO{MN-!TdKFgKOsu8#EwaN}jNR22r@F__vl(mrA>)5h!A@)t z&?J5l)#toyPu&dJK@2%<-6_tj#3m?7FO+cb!_GX}85dafwfeGs(vI#*W+F|k>12UO z+nDDT=pdBYVyW2yw6KS?$w7?Qid2(YaINPvpSzb}id1)<34O<(!ITD1ODk&s51+|r*L6Cw1PtHj=0CmoL zQmiS;Pfb8w-HyPl8?j?8FiYV!-}Yap)sJEpd^FvtXV5qh^2fVojk{6H9MCobOoj2p}jZ>%h^S>pqOtvnB5A0!54bwe@l zFF@}da)P-P295fnE@%0>8OJ%c69%!w;^Ve9;##!YTx0C|exQsi)?3fRGiM2)rK(ia z#L&uyt6_!;WmuX#MO3EIMD)K@)H(m1wLlfCW_U75ms+@W$e`WE0BFi+d)p7D&$Z5c zeon~La;kRr+mgV(3E67|8Hy?BgwW2>7`}dO|G6~&7Ac#v{5WOXmo`LX0EBvbXBccEP6ROw3>@0SJ?Q$^3eSfH_v*LuI2={ zjb9(s@2_+Ma>S}ILp`G7!@$%0(|KkX$x>xa&kAz(>T7tn&~q-Y zH0ZNY2v08=6^1h9#)%%h!!i2QuG?l;83inBMG)OaCm?u{J#YX=yxV?n)Je0`D3wQm z@|yHGAl%3vW!_3dQI+srLCm0O4^%?GKNW%Hn%rA$+Q%6am3vyCB;jgXWy759B-Np-9^Ny~WG0AxL38 zvg&<4%GkJTLIx4wrTD9r(gtZo=~XMjX<9TXv$^GzQ5p1-AB0v&kr|G}Dvk zBa6C_(3}$mz2<)k$HO%sNRu+Tzu26ej|FqHi`W2r%ON)Fz72t&?_d$1V-!|DQ+8Wp zTt9%U8FK$Ki{tLiMa3KKK<#bM#9~bL?V$P$4FRbj-A}L}uL~RY1?U*uS6P7!#Xzw8 zxZ!ur%oo`SIBR7YG$*a9C$Q&zl*0&?LEH2);?N7a5O80%m7vm6SpQaw{eVAEm>@op zgA`LD2-7U|*hgq7&*OLENmCll;(MYQAXn*iV5Q}w`yQ!;INsD;JQpAcqTYt?Pk*@} z>mWDy_t?!T;d3AN``x1WGv<|dEW89tVT38ca^ioqOI!hV1 z0cViWTwDiB%xD(MLGo+MvaeOy%s=|Xsjl4=&^CKE544Jqe(c#J4gF1--5`k9livT< zvAK^$xm=!J3sbPH3>mSbBITX@WU3clYDM=j-OFa%4O`E!fG}~@Ff{92hPgw+yj+o9 zgLPBKJ+vStu*2VT;P)_D1EnnZizDEH5Xy@Vh5Tu-RR}P*@G%NQTFbcptx@*TND;T`_neWe4KZVo*4W{-&|pBOqz8)p2H+52IeiQgk*gG^NBVE&aZ4i9*g1N0;{(M=SkLl6gr;06Tt)cYN+(T;!9()6B69FC zjer7hz(iDQu)e{M6wT%_nN14iXGTyzGAF2Gidq&H@TKSmk6;D|!i;TZ)!_IdR`qs- zZEt3@lKC!Ah+QA*g4Vssv(3YMY0Pc&PJq``gY$*L46(4=AeG6Tlw{U7NR`&D{5;~C z{@%9nl$6tX4W`&*h;YLT{K~hxK*to;i#6BTZ=!7q8e&xJz_`>B4CaKErww7RA@}_D zEy}9JLo;R70~F)E;?VBjij0hkp9X$?f&Xr!`|lK}gpJr|mQn@D6822mugEOv`Dj)d zJWC2B&4kpD(dAus**ooqc;;#+CV=g-yWsGki@oi6(VN?6AA(c*f@#(ujqqv#o_0 zZYHKpfQx&7Rr*q z_GbV(_5z$)C8AEZIZ+6fRG$iR2u~;ifp!*G?zsMPkrIjT?QcttwQgz5h<@4LI?0-Pp*Ve zi6d1+^FCRgLS({!Us$oD2T3f5`V3?Ju@Z`5u1wj${-P=y;Rz1fEz?ehL>U1C~A?Yj$rAn7bvU>Nxb=u+e>~Cb zqunGr>?U^CYn1njpb7I+YuH(%xBZb>#l|seD90J^sthft3NFYSBfklKzBdyaSTz7( zWN(3*iobuQ26*D0y?vdQ!>LgmN!l$AsPcFtEP0qbNG8~Z(lEglwzS`iWZlo`6yA8-*NXVl`quyfhH(en~5hT~hs~7Q1ag z9&>i#N#6LE5Mwr)EO%=v)e&MEju?hYTaT)Pe*&VwFl>pnXC-Y3(UjuPn-oz}<8~7o zKj1x&DEexhzc?llEA!D74bnm(RGt;B}^tN|cdj z@gPmOY<{IK2diF?qN0g<8toqJc`{~?qJs45fQDiyXo5hz=MbvyGpTc@C8~blw~v%I zZOu=->k!Mj>tnoF+ulj$0QiaTta85ZII3HG|6R>; zM~W-JuwWrm#DQ@Dv_%G>HOvGSBJXZaD{|RZUB3}v=u-6BTL6PcKZ)<3V>2Os3`+uVSFjD zsAW0NzNS)(+p9xSAOo{!%Y$&<2G+&`6Tb4d4S6*Z8eQdt zV}r`IG$ALLBXjF89;r5?#Vfz1Ml0SQFU$DowT<0-EdplM80^hw^f6WftmJZ9B5rjn z1Sb@bvoDVY*4d35B=O=~lrVq4-SiwZCaKYy>k|d53Mu(9lXfbeS&~JODct3H0hO)~ z!+lRw3Rrj+kr*>l$H39zTJn~@D9_#?H`oTI8vD4TFK_ z(eyT}|2TLhK0<66b(7GE1y-M|cjiEm{>e>62jy)R13$uO?U-_&>sUKDG5;|fD}1Fc z7|Bmdn9dYA2~?{X8Y|~l3aS0nu`k%X+SsqeCeZzyQF)-6FEHqq9z&lu{71I8cBagE z@lrUfi8TD^kXZifPpu+Uby?gz7~{9{`+q+x<40C4Hm}!Q4h%iCNadCUJ=!SI(b-y8 zC0E?*aVNLWx3~Ag2MSUP{0^tG$sELrjVEML9rQbC^oRGBzRYQQ`b2OIn*6Dt zN9!UF8WTh03nR?Uv!=qj?jQ|{1KP{X=mC!C-TogtYU$o4F))gA@9AYBJo3>!O40>V zLATA{dS_baVflqfSN%V?_6)n_HF%0#52Q#f&*ZBj)ibT(g*TEWow4=!_smjss<&j9 zBvuAx-jM$`fgQK_ZoO~4v2MS-wOW!{`yM!8d}x|sWNW>V9H+36Q)qJ^rMoWqW!OrB zy)OCybW6v)BcA(AlELVzf$q3%Y+%Dt`&SKwTFN<{+2puI!GqQ&4!=rREJ9_3yjLy|YaDWdDYx+LUUVs7jF^q$?XZN9_IoI6hzV*@g z?&|qKrxQa5?+9#JvZ~Ll{UgF>@w}jgl9)Mnq3KrDF0)AqJHaD1HVUxg#v@-QNN!@(WCt^MJ5SlS<3oBj#riiwF?5(%RvyK3TmFVp%Oo_iBe1meB(fy6Aq{zV0MZOU;jyk1b zr>}OAk>=H2$W#)i<>YW7o_zAt*K~GMMClm9Ab??F@QMOlK0A^1C+DOVmobYzFvOOW zTeIO*#Ca0c?Cx$g8qPyHkc&t| z3zPQC4bdT!fH#PYuz<~{8}<;t&U}+K1W_B-VVygT{XhLScaEhD?Vp6AYrs8oJ5xD^ zn$zaD#RM?w6Zjtu^ej?|1wE_XjsrrbVX|0Pk~QJ+%j&48%f^N0>qg)G8u^6_ zcN8mHbp$t~pAhO5jK8_}Q`Z?UP>$LHLn}o3lWx9gI=NL^!S{MD`gRhwN~2W?mRg59 z86GjhGikw&i&fPjS>J{Qf}D!m9_!av2Pr3acj?tuH*;edgID>rOtY~Rbl_HmUVtTx z0?R=Guw(QJAm7J)cmRpaRqN-g!i($-4)dt-tNWL3G{UFA;WBYnT@exZ%@zX~*E4*k zh+f+P83}Aqg>?8oiIidOQx3VZ5y5Act0tL=>sYm;MaGhhpG5AJOSbmUgg;8GPINtF zH66~h-7v3P?+VCa%hbO9ry<993phC%92x?ngpajCLPE3vyK0@~REnuH{NO~CyPeuvYnE6M>=7?B3gHy-+L930T6p0Bk04S5A`BUWa$#4(m0naJO1 zFI3<|*=pxK$H}WzD;?A8)OE0}*kvQObyY^tq;Fh!WZJe?ot>W5?tPGk{jl;930KUL zDLp1?eCGXICJ}YUUe_z03;pqD-*Ax% zbYt68Hcr~d>MHvj(jGd_w)%jC&|c1b;0Oz=-P zs2d}BW=tp^j}JeOhtRG}`7fOE80+f360$GM4I|xME|JFjZJjnj;kW11cbVFWh2^0} zaCjrb5%gC1T+Qwe{13m#A>T&+EAV=`1(q)`uxAP?@>AH2 zgM>$M@~IA_Z{i@Of#y{=mClI-%lQCX z!CDbHL>4?|gCTE%^6UxEP{2I44U}GFI=pYpzS}R1G&^mI=Zk5Q-*u?6TBEyG&5BfB zXh#G@fGuJkjUHyT{Rnb>c3&c~tJRk$^UXJvmujt$=-jg6j!H?4?3RbvL9B6i;Gkom znw)>SC2=6L3Xp4uke`KlPg#J0)8_g3$%BoKO42bdnyHx-mqrd2E=kEGNu_?kdxJqZ z60ul(=F-Gd?-3g!8RHfO5i>_ne+wY65_0kZH>+6B7gP)Xf*_zeJvOSA;{>ux0^R%G zw+`Qmrpf=&bd~{CMcWpZlJ4$0(hVZrb!ZT!JEc3MyX%02Gzg-AQqm$FN_TfkcjH^! zdmq2~v-eqh&9&wn;~P(^LHrNAe#Yi#|D>_V_1=b& zVpeonN^flrbU90aj3X?p!FOHkt=fFh?$0d~?=zRN980-8A007A_7*BVo}(7dl{m2v z;;=9*N+E@*JQ9*?8RIqZXumH?RI(K9;Hc?4+b5Sn#j0#(1Gr-qi#tXF~#{KEK1#$oR^+JC0Q`*@J1q`|R%qY(nO{>9Y z=CF2?DF&5~`EM)q@>(X!bzeW$)m{RE%;auYmPRROYT$F)s}_pj?yKBG@#aqHNHgS6 z?t`@F^WCl$uW6mJ;x%pqM+s_qN&VF{ix z(SwWz3(%W_03DIezbVx<@V9u%YFDc)QEZU<{R1_nbi3ay`$M>6waFHZ4LoNTBdXcW z-%O9bJVyNmmuhP$_GfgdWgK(xGh3YiHye3jN72d=DNXp}$q{|VXsN?|-DyOA;bScu z$~oo3(j6 zz@VT!Q)kv4Ge&p9#i0{@`l1}AtN!zn+8O0NLF9C`DDt@&^6eb1 zf^m1gq4DR51`8W?RXGd=Z~w}JJ=T;`^ zpRs;^moX`KBk)A)7T+SBy)_%CnpykSO(zNQj16FXz~mC`#Q63UsdP*Nz`3wAxYl~X zyX97YU%$J9dY;uNQPpgL;3ssUm++NaVH7vbrHP2b_uQYK9FKuyP>HAAUXkA^2?Q>g8A{^a4?w&IWvR1FQp%wsQx~VV}z0?cI^J#uJDB$ zMqZi`@FJf}--*B~_B9y_k_FQ>DxtK4=#THnp?0D3hDyoJqX&##^r&;CHUp%!I2CAK zA>%$`Is{ABOnU$MB~Eo({Td;Nbr=eM;wgy74^9Z|&)s=%rI_XyRQCdbJEsaLx4oC1 z1IMp?!oHvdk<5G4Ct$EET~fa~3aGS(D>AefApmc+PiaqtcCRKLkO(OJSdl4Z`vk7E zP2_+{$XEC=VXonw5A3aoJm1*GAjPudc@6 zm)R6D$Gc8?a#i?wJKGPjRsQ_->zBQ7;G^GcVpu4%Hh!#pCGCVb!OyQ)GWEY7oIU>` z={hvZeT*1$bsk@>iauLtN4I%h&a3#Ki8ML(s%nM@r_qh;_0~mKhfG1^LMGo6KA?)Gt zk2%c5`FY`YaiS(FZXI2}zo8uYmg`@yADf+=(;XUtn*0@rlwL9)(R4mplg(t!-Rb~i zKXZ^pI7S#H`5T)mid@eGSm|7Qf~};79#fdqzif9F0)NVBMDg-1C@8Gyf<`7LG;hBd zFh7uyeo}^2c^w(OdztV18U!ScG7M zlOUVZA)P)I3(z2NKYo~Rbh9S4lV26iJTkI+RU3>ZUe7(06x?q;oGS~Wlau)!^aS52 zJcSkM;q1Jqbl=q!Ozyi?oWRYP2*U?)TgvKEbdtj%-XKx?nc)gjt(~qCRhh<5wByD= zg?%NiPPHrgf`)m8j%_LJov~IivgxugBg3pd@8>Ygt9jol=PeTvfPEGfeg!l4bkxRn zD>?67%{i+CJox*|y4+iJO8>GmsQY~tbT&qbADW{YEtty#lZoKHf|CMSvneC z@n~eE&{RPVYAyroeS{jek(kx3Er!s%M$RgNQvR^#VFhXz?Kb*_td2%iI#_1s+|F(% zZ7rs;qg>ibnFUd#pE!U^g^Ge>9tS;w^QmWjGNOprQ=1nx_k~;uz6!MY`1(fi4wO98 zneCpMtfJ?{Lo~LNW->!LF;V^W;c-$ME$^cAY6v3!Bralr|1hHP!{)841n*iC5OVtl zqNV6)Xe3mz#{6wSbs!i0_{>k)9?L-_hE_$c;BADQ4&QrSgJ(A@a`ZqqgpXk>fame}js@EB zgLj~&9Ms1l9J`3suLvEt9+FQHZeR5%f(c`v62fA&Sm^3LB;9pHUIedR)~siMwY?U4 z;V2?1y179So(9jbUDB< zx(#{^i_5dK*lmUZOZ_0;yKXnyI!07I`OFWc9yNb83ywy=Br999#C~0+ddp50eKU^S zj^l@mo$a}5#|lO9cHiKZ%tz^1@xOYur9v5GQg0qbC}Tb_qIxa7Nu8!u>fHHW5-;Fk zlQENj@}ftTgzg*7*ft+sxfx0$Q@=c5^&W$qK|1ZrxV~)Kmn>fJpJ5a;60y<` z_Uv4ZcNmCwktnCf@5kS>wB)TQC3TN9|xeY8IMT8|K18@ zgOPv`Txi43jbA9_6Zl(*H}l|I#Y09&-n9Z$`IniG3vZcH4&};2Wv=o8y|h5?f^28J z8Jpi}j?ElX=PIg0M`}vjMA2ug3OvOJj#?+nXv6&T{RI-L z`YT_kRFe# ziT4Ky-To2tyB?WSTr~_L3VQ~7#`4IICsD0VM-2a!NZ#fj2g<=XYVe2H-wq4JG-9EE zmDj(VDE`d@0#>tiFoYK5@Bl+}y4ESs{j=aFa*5u|RGAld3~OyGt9R6H#c`ViodSOj zJ>fs#B>wY>K^}{`9V84}h#;U>xRBS-aOVaSW(|XxX@{yr+VpiGb6G_a&?BbNSpLB+ z!Ob+;ypsFzF-S6A-ppbz`biXKYiS#Hl2g)ta*in@D+3*8LG6sD{EU>m6Rt-}X1i8AZyK% zDnchmHAYZ`fO{wbg?x#a)`7fZ=dwZQwmITpPEmjxq%Vq@0z+=~G4ps@HIr@RC8dQ5 zB9ZQ&Wco<0OFIxXQ$q~n8o}cNVW_$VnHP8pL@|V{4CC26Rk z-fVjCV0pj#7tI$SQHg4FY%LuHny*sel=~aCd-uHGe6qTR(frt2A13%V?is zqyPWA9X(gpz*w~jzBp%K9Kojos|Y!pq#~rh7Pb*{yB`ae#0QIGad|1&seWo&$!cD^ zONJ+WlN%V2w1e-5xASnGQf{RADcU8sD0{sl7ecckVM59($&`XUs{e}S@(YJd6^tum zAD+ROXrM$c9lgZT3N!tQD)9c|?^CF(#m8RYLRNf*R^q1ok&hX;_Q?lf&l0~$ zwon7nyHHoTOu4npm_&yJI$RkaZ}VuSX(xo1hI@qvy^M!`Y^^&M>0evT->)mW{I!RDI4dtvpT=@Ae~kGX2P)k+nZ81YWYb?Yu>6%KnFt z&6}0jgjqp?CANXqGEM}3a{VCT(WB4mz(G?zGejxCGKDdOcEJt5EDwJU^&PV)= z1@pT;LGE)s?B^IHG}iH71~hCqtf#QTUm=g@sB%#~if5?$CF~&mqN!D(C7~7*G*_ei z*dxB^Cc}CCYZuAMgUILpcJF(!nVNd$=a(j+lfe^yetCpL_)4B*K&%=rk>!xgH{k`+%=G zxC~XBkX6`5VpI13Z^u)&mE9IH*uv;dStADz<#=62CFG9s6es7ekfwVTMVbOC00_WP z?Tu4dlY0Y^q=kC|TzJ!>IP`sj!;li3^g(e66fL~M!>)7YzW>u}%Js6mp8%cSusQDS zYBf&E%oVpv$f;!pyFS0)#iX|8>Yr>uUJz?ToIa$sdOB+LdzZ&p+(kX-HmPKveVXDX^Seker*Z;C zV*jztD0E>3el=aUY7ea#Df1j^2v@4`A#P4_VHP_t*1A22o%QpoA1;v?& z))n{B{N|p>2p!3fgSe|)K0Z=M&Quz9>3@FDd{ikjZ3rM`Fm}q&joLH+D5>r(n*T(v`XXws27^$DfcSjsdYJpe%@at+bQQoTsrQdAT&rAl#$^_}ltJcBoq z?0RYu=M*^Snhl{M%ehxChVvT$`tSwX-o!{Da)M><^$9gPAZE8bu^9oY$B%k9cW}Xl+vwynY zVJa6obA_ye*-!eRgfjXKc>c5(EdLMa+Go%64Ov$$-v7E@1bd3 zeNG@N#wPR}%PXYP_8+}64R?kwrSz6@r#fxtmT@|dS#Xk%neIM0gIMH{N7qjGbd^aN zdV=9)M4orlbM|6ep};qMgF(H8+0iOATU!H7s(M@+QE82H(2?7fvlH`Q0Snwpk|eAxG8Cd-Px0>BsIrE5$LX5j>+^^Y`=m} zV%ZNOhTRCl+HvvI1Wo^}B0FEjCskokqi0-11bb^izT9hQXk9yj3_cTAM;3MzflI*< z?7;BxbYZC??3z07xXK+JHt~0D_w=^{y53reRl@E}+*49`ifAx(yQb%nIbBK*NR^;G z3V6|nbfx|JjtTsHR;^9`cXb}D7IK9k(B+TT=Qw7Q6-IMi1quLKVPMDGmrs95J`>SC z!awp?{i_zUk9`yK&#S2T@W)eG#8>4ZZDxY_Nx`5ly+A=gUq}{+x@ai$hA9vPt$y@A ze468nAc1ngB!cbq0Wvdu-b=@)ndoob5(YbBgSJK8yb3!$YpTVg5c4MgmVRWuo~EDC zGJ|_}wa|5NhF$soj2$C)kEtJVzWT^U5;AoZ_)KkCp+Mh{f(VGO5zayjKn@YouTF83 zzlRGCB!E?1CpJ|=j5Ac!fU@JwxNaNs`DX- zd`*`+kN7ZL>QnI4cl^(XbCRGCB4^!oQWlU=dtXwEHq#fZb>Dp(w>gdD-AU+t5jOoH z(pkV0S>x5BN$XKGZ||YmeqrwcGllsCyu7{e-lZ?MqRC&p%y7#6Carv;(9qnhl+H&~{Pp$c1OqF@$p}pH z-yQy-8v#2&>3Qjl@%D>}hFOq;Z!@X=1ps?kv{a*U`+8-GC6!h^3IZS)}0G^2huSN%NaQ?k?_*db(#)|;S^1S)pNvDHm1Jd?iif8FM=59 zL%Q=K(K?bOW{!*AoB5Vgnfb-x8}U{25X*t}e}z@5apY!i94?D1g(^iG09VPWGXfLF z*VNc(`jiAo9{L9a>cxR!up(Wr0IaKRIpGK5KD!KXLvdpNcUJ)!yR}6#?avfwt!`y& zK9=F&r&r8F(F$esLu-<<3KL1_Bi?CMn9nG|!*KAv^j@y#lKkwG!jT*J6|>9w&Ah)S zOQ*VBw};YxnJx9DwKf#T**FOS5AnAsr?< zwEc$A$kqa>iaYOe)NObS9d}ML^rLr`y?e8WecERNnZ8A?0k(isI zy%Ccmla1UT6vd3s3kL)cV&bRzgwPHpsfRr-joQj1cQrL^(m@D$nyv+ZQNEs~<=pry zRI3QH$ultTl5TPY_)skH{`^d(y4O50r0N?}xKD!okvw4V`Z4QI_(j(?hgv+BdBAsy z21qKUKe)Va0ZxZ?v?E4jn8)(B2^i5&v4fZW{pq+0w(u~BXFwzw$l}LgC64a^9G`CQ5H68W2-mln8&ec3 zE_Gd;uvPSL=CeKI4FR`Qe&57HrY|BJT(-or6Pjpl#|%lFnowO=VrP77nY`P*_i+6< zn%1r*i795b<5D6W*LL8n%NO6{o2M>i!O1W6cD|fS#IFz^m)hRbHBFS~gTnn@;`Cg2}x^ZFum+YfrdBQyBU=t*zw3Xi$Mqc{T zFftZf%&_pH1jh~~?4%2uE9V;iR%MyqOya#v966n5%Z=US^fWODa8FefM)+ln#1?Tq z_-c6t%*N9*f1~IzgI(oCKU}XBnI3s?hM-%?!rNj+LPTWbKLQ_MFO+^efIc|vCAWRP zKG~juGj;Tnq@YY1@Cw7XfzD=RF{m^8OvZs8!D% zOa!UNMp{g-QF7)dJM&vBGzx8~L>iclV$9-zfn^|SD%1J8Fj0NONk{V|8muy4o*mJU zkrcIfm?evlK-UiwOC3ePc%y}Dn)wkRR|9;Bx zx7@AmxYFB(owo>AI5-p+ZwX>wq2@kN6ijT-+I=Z0BNloWzfM~UNqy*&faMea`yosn z5hEXZ(w6|?6UUE;_$+OT(?!#R;a5LIJTEE+vUx?Z6N`Fshx%Gug--3iQpyvZI(`jZ zAh(0ARy+9Ss{BZDkhYvy5TWdK4Dfc8qlPz=R%`d(cT=K_#m}SNx?biI^Q*2jh-*kt zTjVy|k`&#d6;%9FpDQ+4TaYV}Fc8Fk(t31|JgOCXO8vyc zeA`j7rZ(aby^cvGqLVQA!0$!-_SaS&XbALlC;lfhIsPd#SyFP1)B!1Ryi|jJHus%- z=At5KE>8ZplqI1+1$@0ptyB(K4P&w?`B$5dT5&mBhH*B1E&pZQbk^sxI?RS@cJPM#)MRJVRP=9T<73gdZAI-MA&%S zw!emY@B?`S)z!pXV(j?YEjf1apX6dpXgDLCQ?vPUyDy2-8f_0bOJ8qx{7}O-5#2hq zn8oovD6f)MPa;W*+4XmvgTVv3!wbswxVepi7-!MBGG@G+DKt8m0Jh zpKC>8*ZRyWgsm&{E77Wt#CofQ8jjA;KWyfD#i*+l7b&8=Po4Mp(bpbQd|y>#`z!AS zUp^m9@1o<++09`lNeR=fu_Cn{Ujy9+CgawY0X>!W+__+2^?@!U!NS17?e}pl#d_KI z){wF?*7Ayq7n1pc_Us82u}+;2K+{}7PeH5pHs-j6e%nt>~ml zh}i!~GF5zF=dw7Ko1*qhmZ_fo|LhO@=b+2BoWPgu@7Az1+JBQnM2|@6D4SY+@gxVqTkBZ0((g z4{fsD8w47$nc?MPLe&RCchuU%?v>etVYpe*hYHVZJiC%=LqSN1l{^G*LaZ|IEhHFg zVmaWhqTISeq?pWc0H@BL)u5r0FIlrdsyjvc1uBb{MR*)V`fe13-3nG#p}6W z-!t)pD8vt?kmh$qq$H=T>klQk=)U@HY*8mUvm&=_KG9ox^i?rYXl3FAP>+($8>}ap zT`SHQZP_F}(jd1Ps-u|>S97f{)W~>CK^EwKnv(y8kEPMqwU^GZG*oC2;|z}3Mz6SP z;)6(f)s&iGB{Jecpc%ewxRyO~MPnNo6Pg;Szc6WSUqCZRW380daPFv~QYI$($~HTD zd#seH$e3gh-lAWZb9Q6mA$=Yufd1EqAZuyc6D$^kzW3Wz8Rn$VH(#7tbOSX6ZZLvc-LapiW?1OUl2d z2hnYaaL4{DVaU@79a}xz-a0chJ?;1|o>U6e7eQ0)=t3W0pu>Vssi?OKDtKpu80liTMP}g><1> z9gi2#4AjI+!d8U~7z$=zjJIN%nwn+^d3<%x_aiC>7y4`PGBm)JZT;59l^cxcJ1NV& zb@OBE@HQllQi9OGw;#FE8eFy7+V~^JC*4p!XgUwEkP=iIt5zkAwsxO46bt))dzq6eWsv0R5B& zCH{uXez^kt-I>TOgZ*Rwf2;g6Eo!0I%me)6 zbp=5y$1&2O;cTw0<3yngCM0}e!j3-GW$Yc^meY?kte$gd#*ar@EQ%kIpTil~DE;$` zQIk_RYcItVymulh?}m=+{804I^rJjaWeRf`VF#l5Ri5(IAIM=2B1tPNnA>+oGgZrQ z?ZqW7QE7H6pu{ISd5$P1v^e$+dsC_s7&mco2t^^NoS$(X+g?~(_mU{OmQJ@Y^ z-j5h~BUq~uNdfCve(w!xlv=-0NbL?H<*h?PukrW=q+qUAhoVLy?GqjAi2G;Yv2^9& zor0*ZI8A?}WMgCFdN&4{RmB?&^B|bWtW}MEE0c#9qKC}TT{Nq%rIlJvKf2P`Ga3&> z$839lrz%);Qc2fpHH#Ii6OL$J6375-*$bcT=6=_QTaVeu0PLrzNq;S{;QdP>lHoDs z$6Anf_+&Bdk#CliyD83l9aCB=@DBv}9ty$TjmtKqN8eBA0w4#TeOP* zV#bCS1}#H^(>0G4=J#1CzmD~YesJ59D*3B@ob z1E$x$e?46K<2&u|qjw3m$U-LRof^`S1Crw5dNmhTa_91;f~;7V0lG)EXFi{;VuY`= z{{T{02(Y{qK7|LKe+D89XBQXf1vLNc6by_utZn;CEluPcZfVQ)>s4E&m1BBZkP-dy zfQXLnj@KCej%51hNe?+sIkQT}lL6+Prc7b`-QSUd(EUy~6k9BUj&{u!Zx>P*DA;Xg zBmY}2Gm{ae1tzYR0M_=M90Ub^zBbvb%DmIPuf%`O zB~BqkO@U&H+M_3R@;$7r(95i@?MKaJRGl+nx`!+*o>g;yp}^xpsyd5vjm;3p=^%sv z`==nDKQRd$^Vl4wFSed}HO=!F-g86^B1c~o{cZIWtVAD7Ph4i6LXB5v3ie*U8?Tda z9iKB6)zaWQ9Uk}k&FjdJtD>ll{T})9a3oV$M}y8vY5rnXy@0NsmTQBGspM<<#zIOOS3=RaH$li9q6y)Qyde z?L$MVR`OO@p*uS^J_Dt)tE;Q0ymvg()rpDN6ciK?ZiZ5%l|%9NGaJ>eDk#4~cKn_< zRqv{`xcDzEe7Z+V>U9^v@7$k5PJDWz=bu*dOD$Nwl>Y0t33VYwlM|ErcbK0`>?y>pw4{k7Cv$}w^;B7mnDt$_&EQ*P_|d8PLEt%L^kI5 zv=;d>C$}2&dj@2nc#qj+v1k1KBUYUg3Sen{C;HzzkV#3x3n0f8)7F$Ol)EtsXOKd< zm-El?nO(dkF`%ZBGdNZe4d${8W3cI7=7OpepB_bR6jH{a*X00)n9M+Xe zVBkv<ejbg^)Dfi zYrPxO2P5P-eHY_fdi79=5H62FX^O^umatHrolka7OV1oqcklZub8F?_7D$Ct;YMp@ zxH(aN-FLE6k>Ye21G%d-Q(to5t{on$>OG>HJuyqO)JXJ=h5)Pp7DI&SLNR z-jayfTyAQ2n4lm`J+pM#J$3*bSjQArdUZIUeC%D#cl5K8Fzflxw6#1FqJ&X!NwZSE zc+=97zs~wY#@$s#K_Lm(8AYg&=N5l{hVgTo-RMo4hC2^DrS)uN_|-L?L`A+E$om|TX>De~%@_k4?BWhYk>k_=R>ilY-%d3GUn|M?V zJH^iylJGPDwSJ{sV)$ie zB=#SHJ3%-lC8bQheV`ha2Rdk%!v#TUCB214H~B5hnPx9eKyQw&UwZ%KZGKhz6|5}T zC2;F=eZ~UqTlbcPNCwLlg-1FeVi`DhFE~W_A+sXEC0VR!a3B>|CRi5kNQ;MMl2e)N zar=t6V@R|>P<#*|i1-Z3Nwa!D*Zh$TUz8D-l$6=L;gZOUE6z}w8z+LeMw)RBO!+$_=i*X_Av25? zMzl{)Yxi_%^BnP<63&wf=eax(ro8AGu@(({{b=bI_}D`l?DAOlrYr)uA*>YeKtG!S zbug3H+CtYbL2 ztnPAnt+5g)1nInPK_Os;!bNvxaGP&%`|E%~2!@@V9dqr~-ZVCF6wB6F4L-q@1p|(j ze)8TMf@KGignH_i1F~qL`I)%Ajov4P`)aT!$Xp2?J|TFceP!xUDJ&XcNW1H}x5$Es zB+(xge2c}<3G?x7CfOIL!=(KMoEZ?N5n689(4Y|;u*y1i$q#1&7EoCS_bc5$i>B?@ zc2+Oe&3N}=$#XQV)P3L-LFm9ukE{jfI2_Pt3*15^uB27RcdnrA z%;*jbCTM39Ozg(329Hz|e4}3;KIa8=PzT=A_CF~(E$6{^3$*1{U*buGJd`n6Xbo%H(>-Djb}z~r!SRfOhZ$M;57e?m_!JEjsF$GDMZ!#Cd8KlZh^of`6O1lneFJ1vPCWRAQ5n;S@K@Tyh(p#pAb5J1-aP}0Pi&Y5pjYq zLkhg5r*MX&&A*dnUm(F<46l;H8f4h%%wI;ujnmLVC+$J$UF$!VQh+~@+V!gg#9?Wb zOGz^z`gwkn?)mwt-;IBsFBys{Ce|CiE*+(li5no5WcwXPz*ja7O3|UPOM#FUNZlxZd!98N+AxX!neb2 zbV34*cSru~J(#;;nd9Mz!AHbi)^4vMMgDh}&~e7re;_0!d3p3t$WDXBxxQWWZ*z{K z#lq%m#iTHEuLoR$%%Ey%PO}pY>jgK|^Z+EHZ@4r5ljt|i*==1p{MUu!s9F>rsMhQV zqo$*jS|kCu#1$dOjipC*_f)E=t{QG*FWrsa8z*7v6CCr%o@Pjy7_ZtLE!{y(y&ClnUa)L-c7mfZ7)O}%xu+8(j z6DK$e>jbpsP(z1Uft`%d!@c%=zS=9c=38bN>UZIg{5`nf*v*05Z*h^dlOM@!Ii4x3 zccUV1?`d<_-4KAT9L5()mDJ3F4Goh$3CkL=?R!D`mCMnR=uLVGH@!9vff328JJ6nn z(QerUD)vfQj*7YO;M6%Tvh1O9a8ZPQ3Y*pQ5$k=?uwH9DOr@LJ9l50jBI3GE)_XSv z4Wb%Koi*}B_bT-3H{fMF_}jZ-2o7|@_Xx3h#5H2s`5y`u!F)*nNLJHpZ3 zYT5K^QNWQZ{YcXhNyVt)b@_nc?}gO#g2}%)lzAax5KO$3AE&%~u-COwQ7UO3$!ktF zS}>YtP^roZyOyb(i?-f_L|SEvn%r$t=`Y*Hb=^^TS^52!>qAFrXTsBHL_@tX&$rKa z$6@N4>z`zZ*XB$d9TPxPvu_=sXOwPvj?X|u5ySK|TPkm5DaGJi#E{-|Lj~+_@>%>2 zuszJXkz^q{I;mUE5$;S6NDYJmaP3n+Su8ZT@UUK|eEjrC+<*izo$q!YSR#&NF<(T& zt(0pm@$8%KSaLVj7nLOWI9o8x61d z==<+--I|eNC1RipMFrr2F3=bHlt`u$#xznhf*3W2_>Gw0u<>sAq#@0D+`2D`*45sT zrWjfJt1Cj(1N7P3avj?5N_h=ndH>mCnS*@XJ_*o~Pbtanz`PT1H{6OfC4bNeD9ti5 zH;Ez1U>TJCVjfz(){>)5he7{z-0r5|V1nl759WB=;+3MVNZo-M>7Q;1xXIpfYaOA( zHMib~%M!%_JG5gAQ~5XC#Ejt(KKmJUhF=={lDprZq>23q^NoAGfoF2EeEY$3XQcfTs+1fg zSR)oV?B&XY2v6LSsXuXcoFIj&7tW}az9|+_Wv76=M6*;R_&^pOT66nx;{RXsHbv_Z zMn!*PkX6cg@_RyDoJX@7v|oSoo!XCz7DrvGu*rr_-}8i+7?91b{ozQMan*Q%1~vwT zSc(^@3*n~wwja;b*u-khk}}Ah>+@whJNvA%tK<=q3b` zypt-`5!%mmdMVjq>4A?o@?YD&MCWHld@<}E;6wGbK9@|U3qO``G}95TL$6cYjgFWM z#Q(JFJFq`bL7avypVjn6W2f1P>xcGhQmPLD7(M3yU0(>>ltakb)&kTc_Au3H7Of2e zKk3)UcT8S>vy-S_Ojh6`{LyM?2pCG*4^tq=leW1~0)+m(u znkXd)+GisZrD}i3u1i+xnJ)h%Uf1jSOgxmq2e4l}MatRLeltWOPK$4w5F0#M$6hYf zjgJxd>5Ni7_6{FhT6t;S4WM?tGgdkK(DED)sr6qrenVsawbU2&la^7d;P2<`rQ)6F zgIn)ohZzw6F|vm+-SdwbVt)7v<^Wn7B+wNs;RA`rA~j@VA|WPC!{I){r|IKEDeg+) z9}7VU62bL5SDC?QiOiL}{dmx4gTXfDiZBWM!sBD-C#0n7>ee-c6oSI{ZcX(U{9FbL zI8Y2wq+H3-6(u2}Jwxm*>K&fHHaqo2d9-tui1;tNR=jfrJ=EH-_k@4N+90lP}0}uxbzz3L~fAkG4{BM88v+r|RnFF7|-R8ym-p8qJp=~{Cq{}p4;E*{%_cOuLA02Vt8OU{v z8$5S=;P-NHDk7MkXE{1BYC18t8aa=GE|+2!QxiJO?+O=7b)O-6-5KLcJ}4K2A9)lVYo<%^DEGToA8vQTmuRWLXi@6MYsy!d_T=-MUK|6AAw8OgX{htlTz@trwWpZI^ewsQS=#mgL@m{P`{u-Tpq0=QA(u4=KNI-ieA~>d%Ya0nnC4^ z;xY&Mv{PHUBA_u*-y{E4)cNHW?nyl10WSjfAGcH8TB2e?x!c3ED^sEXbR_#*`sJSt z*YlPPekq-p)^L;}Z)ul9FAD?7l`qAE)KxH+0%izr*yj71{D|1QGuCoirU{H^&pV?_ zksGM?qNw*HcILZ0ESej(e6gDgQXg!}QN$Pz8J#SB0`$1%;(64#Fs$WJ>%r#rwv>NK z8tv|=)}!qszG>g|Oc*CTG`RF%Z3?|2rl)Mb-V z4nQ190nX_knM_>ZicR{>iO_TjRvHjg7|jl=HG{Uf|7rF$xo*aYBz{B)*?ujs?F}KX zMY0M+dr$E9I#YOoK*tAbxL88&vo)GB;=eKXDB&i8zdLqqceG%0bYrxArJ&k?uNu;B z=$bI>Tapi5O`m=tSdb?x*F0LH(DC5|{pe>O6|K7pz8){Tut)XO62Rg5QHf}~q`nI0)tF+k5mrF?OL;sOn!sfC3Zu7(E&0=Uesv1X z*TP37Njj!-R%hj7X^(B)!U{?w`~!m1K(B1XG6BwOhVY`gJqq1_A1(D4;wF*NqQ-jG zh3DH($&ehLug1_&j?N)UO1$s4^J9CgyD|u;G9n-8wquvOI|w!m`M!S@`+x$aLyPsb z+o;2tM7p`2XsU1F*elf-*=)BTRznkGjDx){Q6{}<<~hy8)6@-*vBtb=gFvweV3m)l zG^h~7o{&}njV(6Sdxt5^B zF0}=4+7#ge;w8oSQj}tVDc4D@lvei`+&z|%oAScS^_ew{%%u|@;Pas)IGkn5@1}kI2Qw@P)*5D>;i1(yfsaURK*#)8w*|af$P5-T~P?<-{+ML5s6rV(r~8 z<2GeS2A}hcQIHo5X5$KNbiP?pe*Z=~3JK4R{Gsq^Ta;vNqJ#-58&8Y-e#o7nZ|YU| zS#dMrM#ubfrGkkIMi0HhKeBM8OWAIPw@!nZ#j2y7aMhc=5?T-p(&Aweoto(=|8`3k zI`~x(sgZ4vva0NhhiocgIubY`VYV$W-J>rGg6kG1lTNtsE|0nXd2Q(UCyA$$v!pq) z7GB8`erx=K^|^>a$yR{n?jN-7r8|76QIN2nBC`t7Y1D>=S1iSG2P8IN+CxWla)}0p z93!h`pKTI-OSsH=o_1ktu#rDH#qYe3IB=VD5ei&mf1-`;Hz-w-_*<_lWoPYCUnIdl zw(N=IioJU*_e^r*?MX;DO;csmvpY;0*#VqHAD}}ME$!?+0HjC#_v`&}dUJ9DD9V+SHb00}-n*(m zR`Tf52}Aw}H(0IRMza>A-gQI4(!p5`-Exz9L=fhTdYgNk>f#}eaU1$@7Sm0QEC5Cg zsl=7Ni%KY73`2ZT{%Nn{ZvYZat;4q z-_Imz^qAS3H>t2dw+pMBjqtbeOfsn;XZ-3vMpRx16_mp#PsK=UB#_PQ>iKTOToCd? z;N0)DnMm-X8t^J6NxpqJ4OH!+eYU4S1@?@XqWg}$shHfQ?d{~`Y_3>2drP+{(fRI% zEnCw6uu0^@Lofkx(|(B?Z=AWo2BqSKsxcyI#InF0s?Lz}f0{F-`BhGfIpzd;W*!l7 zd4l}+WgasEF_KQJr`tQ~z61AKlQ|;KHUHzroTTURHMl1yp;eB@C{mb!u%=Ld0vpgg z?GWB|)U3bML64yudXiQRaX+fIugIZZ0Gmif8;=Xe{X8TQXW(ISBnpz0fJ*WzH*vNN zP&_bM#CPDKBQj{ub|>z#Wr&X%w&3rlQKd`E6eUpJ!}~@^A#xbdjUJD&B%3*@XiO@< zk!l$p4Z?9{lhvgy3ug^~PW2)+{8=SPiA3*Ls`$FXgUJ3hA{yiC%w@z})HKK{cWUIl zCw292eg%x9cV@er8u>Lo=9K~XGh&1RxaiBaEoVmIKSny%;~SZ8PLxDq&U94I5+0R> znfLjY$TL9B$dY)7&oKrPkQ|m>Ve{PAW(l4P86X+qYJr=%%Ay6G?%M`F^REB}I^DdH zk=apv-V|Qx7lgCFKHSX~G<~b~S?i4bYzCqUcY1@r^9{>$1SV%aU0yFkAeMH#hoAqPB%^EVTvP;#PcaLeE0Z0m!T13A{mL9Uy*~&a@`fFP z`Gt=O-IT>&4IzF>Whursym*ND{vsPus7#&58q2Rt8PQ+uJ~WrblgHYJ{`kKOO8Zd* zeiU&-^qOxYs&Uep)PT0Z`BnS9CN>I#O(QN}EqMga^-JBe%XSaT25dys8C}_}I!CAf zU=|cF$tw|f35@tM&=VmUK&EId8`j4edG0(tUm>kIU`B_&1B0Li!v2+fln-WJ>MI{2 zy}Vw7NflJ8sJ{w+d9UQU1<}0`Q_vCN1ny`uPr3)@{<|`UJO-#O=SIo$1(JuKWx>_8 zH_N`S1f$Id24cIo6WWvbW(}GozOO2$g3@DWz65S_cv^_pZSqp!ZgrO%Z0REvu`#wI zV5;u}^Gw!_UFrdYRd0ucQ)idl0%V4+HXrp%MTHkj{_95-Ay$Yq)mA%)F!3RigA zSF-Bbn(jcty_Y^(gs>L9%eD9NR!0zU!hjYpG0Wx6Tf|U65?RJJXt2~|0l@PuQgAz| zMCyfP@G~092g@B-jMFa;oQ!tf3bT|MnN`;u+dOe%1Mgyc?Mx%GC04S5z?>Xt5j{$q zW&6A2s#H+#e78_;vzIH~tu}C&!d^(ZdPL|3XD1BM8vZA0lP--=n8D#!;P_RY&a^Q= zUl}nfaIq1$+0>ifeep}q0yTU(FGCXZo_m31W)n7auO&?f^q>KJ*7MJ4u+06>No7Q; zqWu<)trE10Zui2nupb+BBvj2wacJl|;fTn;A1``8Jg1aL^r8ccBtG554PY{}lZsx` zzhmWQLRUCaqEn+t__Zmd-QU>#wtq&f$UetU&5yM)W{{i>hh*&mPsA?- z-FT_sI4_)Sd9yA(3)D7?#70%3eWG;f^V5!ZG*}nE;-F9WJkz0E_Q=#k+y39GRRX=s zvTlCZaT=f$Wk8~Bh`!QbC)^nRScsl^h0YTNk(3*qNx~p{CL&Vol%hv3)v5t*FE&B5 zSn1xtwrXL~_jIv27IYSJCf$p}00)M=J=V+VG#cC&!BEOO3*nCBdwk`XrFsF)O_&4e zh4Vl+9+ZyQghGC%gmVZ>o%;j#oepMIq-2!^B9h`MU+_<~J0h(b)gw+Q%C4W=_~|bl zFu7OAAK`tOx38SVmBmRh3hcv8Ye@g*+t*{k23J)nKQgH#HjSSVruX0G^g5WT`cqG& z0y+-x3;*k%14Z<^5&6UmsfGdI{Y2+${mO^s7+vS3orcrgLOLZ@Ao z%^XCr6`=W*m)qYt4T-r;>V`DIngQr=5Z6a7weDMLMn2 z6e=sBQwavfH!X4mUlmR@Lsn_z0le`UJIOw9GZdu=RKcyIjYM}jBS8u~2^87>oFrMV zqKFo!CiBOWI*_2k{_4gqjBWG?tsfOe%Ca0O+f#u5F`EZ!SuUI#>eahLIoXlei}0Um zu6aiPZ3elOGa^eq+fFIrqKLcK+Mj%tT8>>gXPb`qngc~0spq>hT&;B~KEjJ<%a?}R zJgt+J%7_DdQI|y9o&6%&#^!E2aP6FDq?uh5oRf%U5VdZ?z&qIX#dEPib}ky2a9i0A zAmI*QNX#(EfbR+7zK2UkfHkpQdP%l3pV@IeO$Gp?!Z_8Vg~PqPsw@!v(pma-YbQlN zmx_F%P$DZux;R{F{v%r=O^2SSCIcQoH}yob&o({&>Fu@S1`ek1=hPqVRs;%}boKwk z*h}cAqFp2K>!?nzG>8C&CxA$VKPLxBl5mR3%}a&ZXn&JALK{$s7>>MFYl1gBXa;a4 z_RA02sNoTMY|_VREt7i2t>X(&B+kjh*A<|{W6udpx}(#9woFb1B4#-Q!^+mW?aKTcbxUJ*9aM3EsV6@NfczH5^FO&`)m5$^7aYTdrh z=^4UqnvQsuLVEFjh3-ZJ5C1~=BY(6E&z7+rFAo`7>JK!-Jr6s=2r=}>7}i^K zCkgUVwvo$N=?U7sT@|57>gCn%EZtB6JV9JcO$d{?yq^xxTgS@E$R zmt!-(BD;s3$%yepZ^gjjI9+w}Tiy&A40r6dnQYqtsKngJYN2y^62ar z=<9d7H+S5t-Jt2mNLjFOpZ9d|^W*iOItgU2I)VIVkrZ#&jQ|KicM71coUpkYs1OpO z_vcV7vnkRj5S{S;`H>tuw}-bKv{LY_l}BZ&s!^7VA_dlZ9s>f3A*a4hj4Dv>Ros_-7HdOM6~F_vHp_naCifrt!5Pa?1YI$av9Dq1OEz-E!T5 za?_)IQb2ngLFw3P2q1J32UkS!egxHFfoCZNzD@j9{qb(S*d&Pi)i2{^O$DFFR^I? zZzk`umZ+0qr|f$2vGklyi{$;NMB;;DJubB4z!NhmG!{Vfec{`z(Iz0+9`O)YPxttxZ{HTp$9QFGyB@lrlV_k1lF}{ zs-+AfWbu8-k7bRl(MlufS6}p?^SK>%N*z9WgLK2gq=0&2giWvT-hQKthV3pezSncB z9x-H#)Vy>}K$KBEFW!(MQ5fR)K!hL_T*6cwM-_6oop>7~>oLJZyo2NU)0+bvn#_|; zS|O#hP%BLQS{b&sx_7jA<#8@hz*Atns-7n}VHWK6ghyCM=A7v*T=rgjyo!swwH`2x zbu7Hr@|LCn<@%*ddngqSN=$Kf1Pm*CsKmNp{9E03;Kl+1T@xuqbdskU7US_Wu>~IK zSz6ePt#SkBb`$BE0;m~QXQh%U^!UmC_?y0W+>%e?77D6<+-a>gYiMavV-(|m`{!AK zfPs2mFYd{CL5nS%7no3piqj)Wu=2XDNKO6t{|2v6)~fo+jI9TjMzBmJyp#TxPSqH+ zY78aTN!Nu~DMMA#G7QZTzqj%4T0QtL$B;)bsRF?t8I>}ZWXkmlFXGN{_aawYjK3CG za{Fa6Xv+RA&3kOhawmCdH@=~?$pLx9Erg(i3&k}cQ(f%A3jc1ra9J~P_Rd*Y@)tpv zru@DWb$El?;gEi}adwj!+NhP5I&nI(Go~_z5y(?ZEsmNk>9FB8s9kbUoA3RDn85ZF z`EkQX7H!#FY}V->2tv!nF6DH!0^tU$+We>NJUO*g#VG`;FtSt_PCtL2Z3{gZ+YM(n zG--GK(rc|d;GFNVmDD;HCOKvmn)|t_@NFD88@7y+8>{0zASf!HAq+nW-^d*V75SL% z-0*YhFIV?eav`hRh%ilnPLi%Q!7{bz&xUQ@50}OsfcVO{8nTHzI6H?<_20Gp3hyuO zt`o%kRZsFd95)aDJy0_dDo^PxXyLw2gvmEQ+{Q!$usH+^ckV-u&Dx|>Ewl<5FoE|R ztC)De1nvC7#aip!ksC2d8PV@?H~tQ+1oIeV_D;vkE*=QzlMiYadzEDsn4SuWUf!fm zLHNqRc$;lU#V|2QJ~rkZQ0*^(PnrgE4-|G_zm&2HXs+m?MaRD>ZzfoCvPsFqnR$9s z#uUDzs1kQmGrv3_fF6k?a<9RW|MddKQRopu z^B+{~)|AxzeAvZ3SDF4kr=xjy*#f7}QyK1Z=6-x2a-NNJ&!)el8;1ba3q>11`AkX4QHHX_RE31+9MuYxiEA6t3(;s{)75H&eqY-Gg-h9fC?M|2(ihrqi3#Fv!(J% zUy2$>v}%;Rba+i&RvVnRP<-lsq9Pq64JRwuH+0|O3Hlv+xIHsY?)Rb9s`!FIWny)d zT?p#&&ZT9t3L4uDvV8SXW(-Ttgr(A+G3Yn8vO&2Nhfb{QoH#i%liSL(4ZdM+X01*4Sr_W;!ub;kX% z`BAL;N7B)x4WP4WqEO$pz8ec~Rf2yV)$*xXpgE{=gQX?_xq()U#LMt03jW7e9U@g4S_pd$wLnG0m#!_r#`{Ys43)bdx z1_2tWr0IP-#%xkp93H{C4D@!2;JtqF^AN(TKyek~YFUNA#1;J5qPvbgs>~$0JbOJX zoTU$&kMXo`v0wIx=`*n~UMOxSWKerX0oGGsA1Mf=^ua3?^X_u;w)2o*kD z?I5tXw{Q6R_2ce*c$}3u8Hz@S-_^EAw?lG)q&Xi|DKV}E9h$7_<9mxH2i9ks3UA1pT`t=h%T_2}`Y>v?lpz z2^uACV6ey@%}9+j7h=#lSH@GJ!Bo6|yBVu8oezNvS(TWvSDM%rN*G3AY^@HFwwN6z zz$6O(*_MPKl~pT3kRMi3M~I93yyOp}B-;-tV}`z#7UspJ0eAAbq&W-JvfDOVz=qN~ zTB$76-PM(Y@1!X?>K&Gq;<Lp=#&ss}@qf0wD(g8si#7|L1j4|2AOR(}Q!~0O4F+GJln6b$Ra_*d;hw(lLnPajFjGiPNp;NP%C;#ytZ10rI_i=?Dp zgn1o#np!6PMk9o)T&C@`2WS{OWKkM6Uw%sl-;E1T67I}Lkb_2*E460I*+wnTNk(N1 z`0-w}9%-c6q|ugJB!$;2?tPm;C>ZZ2%z@vpi$d-`L8!3K6uH?T&04RE7t1Vat*R}m z8IR4oi^$u2X3if|(jt3maJLqhy0WDlSwDvw)2L&79ja5fx24&dj=IlgSC)x5F}8C4 zv9xtVtteO&MjiT3Nj{eRQT_!FpT^!F5cK{v{Tm zwyVD6aa+MA$H!Za)e5Gu5KXtflWA6BGSpGSCQ8NO zE;~1_5}s^~8-NxK>!stp20i{dKFhwNHGZn7q-Btg69gwNStq*IouymCCmkcaDJ#Vj zM=HO<`f;(=g!dbr0J_xUWyS(x zDm&aXwwUJC!H-%KX5;`scA(7dO|RP}o4p+BxYA{EFBrU&s9sAtpe6Wu3{cLZL&_3GHadsn3^vlSP zbt$&h^{dWIY@e;-<=X&5i1RoRBwXx*a2&9ix;p)B4*chgMI%Ys_i&y-INebuze0M@qq6j zPE{>ZPG(i1^5Hki|GpG(TfL?S+FNr;(?D9WRe?ouhIqfkzJ?x>5LspV&{O!+@yp}Ah)&BR#i?#ZY0r;XYE5-e8HJQR4 zrC}iTz;=Nlm!1_^HlTiZ6L`5tch85#N|o>QS~|v7REbY*ACT9w zwh~;r(<+N9)B5Aozdu?QU|nE?gzsqP(1}69+u6otUIyIq0Y^D(yx&xj5eF$H>YwTy zb=ztjPvSGWrb&%y2F8|fZHzSr!`~^Dqt8t<+E-thQ>}{B_+>I(v9F{MkBDt8^LT&C zc~L24Yy)v*EJh8*DtKOWj^_$Tr{9d*Zxz(9LKS@Si4%}ntHqWGLK8j5zDg9FSt%(6 ztxn*tU;Ad^r~FmFke$&FImOne$^izFG!ZXsXz@-M>k1j`_}rY%qUU+JlQvM8pu?<% zaR#KVWywnTJ#`Gtn4cDH|e-?kWmMnnZ|Y22TNJ#zbJW zG$R1A!K-q@%V8n7Kq_Q2)WOl_3uPA$hpNwDc{1PGM(2kws!)yaZ6uQY^3@I<_scJW zcwJeCa*oAMq3@cn`_t2E;GKF7P;l(Fm`2H!VUp){e&HUT!?n3=-kh07&s1r~H6n=PE|<2M}75x9D|GVkxhmxC$(a$Ep-jW;eSEN$r-aS_!WUR<6R)TAe= z^N_G*WM`D@k6{z=y^NPo=neuvoayS+D^FLnsrA7dD!(_b#zzm12K39{y3rQ^%TBV) z1<-X0!Dde^lqV>^+%N(r4Yebs4CY0<$kq&@vI@K_isuDIFbrrgsK`Yb85zcG+KAkj z*e!lHF->Y3N;uk<3o8u8mIn#~;d?QQ!tH0(hwpt3&=;O~IiSKVo215>d$9@kTs65M z;91ARklCn&0S^RjHEX4B&a%by_XHBI5Y}&ii{dxA;(9x?f^0gDE3v=LzX=aQ?Iz9) zyzKv4Tuebw%k{6uVjEx6j?ogerc!}9{7e{lDrl=kEa&)be*XMf9w#oZ<>;W(4CwLt z7Z(lEF`{KEY#M>Qg-N^qsG=e#9b2rs*W56$dK9NcFZ=!ME-m2oz&LVoxMWD4ksVXe zg$bR62N3;?xt4}m#;dUyaeI*Ek{rCM$!q)BV=eVGwN3~W6b`oq+^yotD@(KNU>hec zBj>8nMA+FFY{Qo9us%sC?r71frA0yu>`V7&T&=v7Xg;T1t7dRG8!eO)x7wPLl6L@y zG4Jmt;1?hvRZs52R#yRVq{2`#%pXZrXN)$oEh=Td&QAC>eN6wdMTVHnN76RSHo+Hg zqV7BUh4lVk31_MRE@W)V^;y!ePSed^cmIU z^`>fUE3`MQ)e0PBBzFWnTRG|Xkg9=z*l+^l7RT@mW=(QueXWRZYFAjYC0m}S9k=gc zsB7-KVZRI+&XQRXPaz6;QAMcjv>JH0z?ogGbvl-+QTvDWKwD`maB zmA1W`_gpRPi}=5u4r3@B+~_3*&D#imB%UMAnlsAp!@sl!>1N)rBPw4j&I_u|iiJt8sZkdy@9QlnjA@+|!I zifegBCWY>hi&E| zow>b5otJ&q&3umYIulD7lJLlY!;}5)a{G@*Kbbs>atR@yPKPZ>_Z1f`tsFIfkY{F^BRuJY0YS%5WfgW_AXo-|VJyDD8M!=OGkCWnYQ(&!h36`fdP3V6FnVkp?bz(q zqE1NrD+?2+n1(A#wT+$~IhenZ*|Liw;ho=rFVa-RZz+EbXv?m18~ z*p0~de&%Djm3{rK8`W&dEumX&V=Y9~Y zV%ll=&c4h*k90UB3**168Fl$1igxz3ea?71n}6tY&# zh<1t2n_seJC*x(!du-7(9Jb7$b<(xtsQ)=*fxDG`3H_PaF&=7WU%u0?b>;=z!Hq}a zE2Hdrtx7-3&L!km42$10P1D2fR zmnG%L!!r0kM?61#f}{K?cUTT3q% zFuoV_Cz*JDcf-3U4J>zF9H8xEu-|xy$Gsn%UDV4FOrqCqs;r_sI6>tAZ*viP0|Z>h zGOTEBY?{3VCVV_#&ddU(O`IPdiwth%GE(Q746KY8iuN(1!%FB+B47l5imdJ254(V& z7Mp{M>qe^HxK$-L65znbv})-2GO2d1l(R`-k4jWj4HqgP$ztD6<6RY@>>z1eSPc+NJRoqHwMl{Dl!5BwZ@rciW}-uBZ1rp73?>F z2dtcS#_S2m=$%koekq<2yKi;=p5FC9nUPleGZd(~^-{xl0F+0HKpLE#IYIK+S-oio zy?U;HL7VGJB~W;^3#j|Gft&}u&wA+7)6;XmS^3gm>=}g;?^V>%9$}6uL_`&e0sl?K zsxJ%(oZ{0JudDli@dq|Qe1U~HR+KsfQEdlc${E&z0)YTrcz&El#i}Ecb6kIx(Y@u4 z7>37~Eq%U00pXY2(1~%$(#+9pxrYh=LoX4^B2{>qi3YtWeeGv6g*6Z|O|GVvzY}w0 z;YTXMNep6kmJ7OaArfsX@)6XoO8gzCKt1nxRFt}Fr57HYpXoiMWS^3z=c=R*8=%mX zHbXd_CXfOo_OO6=4& zybT0tL%sTE)N}dMY-YzckG`&F6AqG7#v|jIBVIx^{AFGsYgc;5kaepTsrXqVY>QxA*Bbvsar%krkFJW;xg{(UvycLE=F4Mq&v zhS0tqhRQtTk~t$kLWm<>%*@pP)rKisCqc(m~EBKU2^#KXO?l3#|v~9>eD!UR*y-wNCSu>5xRAu&|9U zje06$Ent^wKUK{ef^mNNqa*+yigE9u<_vap8&3Q~c>#zxK7@cFSPPP;l1m!TFaAQn z(p^FD4@mpcIk^Y?RZZ#&Zph5skC0J^Scuh}B*$X@4?1WnXmS4xk-76D;O_?Lpcs67 zaNNr7vDHYsfa9Ac2K#?zv3Q}u{OOT5^6g6jmIn`Jj%VJ?m-P~qCcDR}uKSxid&vmT zI7%5s%htQreBTSjNa$XhW3D&Y-YAvZc7YU~Hc7G~N8b8c{<@ND4_U{TnKjM`etwId ztdbVFsSuXtxTmKwkfQCeuq${^+8dYr4KSWmkWD31Hr_b0u@IDuRy{?cYE)KKJ0$6E9+RUZ&_}3w}|d%ha(qa_hp)jqt5Tsesuf+r?Ng_37@n#-An7=SZfJN0P*yx z5q3UedSSaTD$|gd$|UNRgD>`C|Ezq`voYmk#s*pXgDluyTe_TP5O5Er0Ptsx@499z z!`HG?0}2ANth~Q+r8uU$DR16K;N!a*`}ggW#(Ye^dba#Gr4|vv^l6Dk^hvNwB?mOQd z%sGNt9%k}Ibgz!S*L=JTDE8(FX5cjIjO6OQK?Dxf4BIGMz;Fl{JZB-o!oteTPU#>p z37Bf6+zB{jn%sf#)3Zer_e}tdn!65n#rjr5?-!P`B_LM&pqey5MH1YTQH0Z{2`@Rt zqhg|gOV#&UiC!)hD};wg*DxBl-yGmkR%9?KS@B3)mg8*Bav5age*7PE%B0p6%!SZ{ z0xHpuc{OQ{hw~h!L=a?KPV6u}W_?ttyxL8~#l-(_V!8(H96Rd?`5$I$roH0swZ&j6 zN|~qiJ=s!ka4(rCT6pMww!TM;SJu1XsSLjc+lALYPt8I*xTdf?hZ9m`QX*#L_XdYb zc}>4O?0*Ev)y9)xr-OE-F&;bWQ(63@dP&oz#oRGc`0+P68{?UQ1p*ipvtDmdkYL!uO==8ke+xJReGw`9gte&@_EOAC+G+3m7i12&9JOpvqJm^zA^KDc z>L<{|yvn$4j?q~BXIe9)Sz33*OC>HB_DiURg-2&x2Uq`gANoRDCZ2j=FZAUC zG`2JUr++#6U+dkktw<1+Z>~aFAeF^~=_tX!yt~cw@4`@az26*=VJ6G#rpM|hr(kyK zMliZ(92^ZH)c9INxbwFBusPrkC(%uS^*jF3Q!fL~7qR%t^{->&I|Pk^gD8n_g^1Kc zonm^}a0bplucd5GkK8++dz%2F1f~@IZhwFp_Vhr_hJJlXcb3!n>@Uh+bdX^%u4&<@ zsP_Zq_NjH??bP8Z9TYiEhDzIkUW`uw{XjEjlP?{Z{{aJwf21zlfN5oelk>uo?z6vsXEPw9bxGURu(CR}|ic+#t`S|7&#Om8*0f1V- zCUZROXOJ(e1z&xas!7>Q8XXvzgI=LuVGl%;nau8AukD?_)uc2IB~%hDl5XCjYVe*g zUs4)L0L0^Su&%L}AlSAY!mXR9PyV99H;}C)tR_Y&gK9E7Sl@)8T!385Nw>=>)X^5T z`K_iKqWypT(5U!bWJ&imI#KztgQBitWh#5jbnd{9#Ce=#_skxgD6*ygUnI*kUm6A& z{0|m6_S=C>L?b7mc7ULq%oR}8IJ8~U#9YRhBa0J5QOiy@gYTre6#Z2?=1o0v0FBC< zcl+}g)SJ8V(j(}vmMJEs`=+cOZMY%&@wev<6rgY#um0J3)*vctee$6lkDqG$w0#lPIMY6;`Q zRNb;I*#XKi<_A99ehVt4RE(0`+nBl)N&|8DNaO;} z5HTri7I~eXNlz!{Y$^}!68L(l0%p+p1oOFRIXB>FQuvQD)v7dzCHqQ>@NB@RCFD=)Xz{BCS}a($?{0%n`dUB{t`lent_AT=c`$GQk*6{jPNp3!z;1HKuki>OysuF9P_A&C||pFJA9=9%FSQ_AgM)-6A=R zD&eYs>2I3ID*>D`6ZNbBLSIhaj1wRwC}mO^W>QW3EFym3)ragWQ=52|o z_P*2%b5Ha%!3Y{59KZObaE4!GmWYNxRrqF?eRQLpEG-K7t6M~N#fGohgQLxfG-LMs zIgF>T16KPspsErS?2=#@&0S|PX|url8gw!J-;y9D$Ht9n^LtXxo#ACZ0Ty~RnWP+R zNvwqlG^KUixy|c1>3tsQJi9yjx9_J~Z(O@^)^T1DQ~j85REFU4;W*>b?O4v^5tJ#K zLS{)~{gA-o!w%Y8zeXI0&vHTo9=b8cG%pDgl?neC&un*6^068{d?UMGE6=6S=HA8O ziK32XWN6>^zCE|=c>iEQ>EGU^S;F4cHLd12czK-YC5=#ot_^ffo@JlTyf@JMNocu$ zD!=&wF`vhAMp;%`YcF&=NeDuBetrGX#$zno#+xLYoCr4dVlr8<7FRWGr;LgEHrVKE z1A)Q#(J&T^&2^;+S@)xgfV0vTnQPL59Rdpfmq*X |MJq^haqQN>qRlT>}KY9?h+ zhNm6&g|$@?QWnNkMh{m2ppK5Cp)wXa0)stwo@<~?lx?Ix)XPMdUv5|ylM9HW$xN~~2P9c!5qyY1(^to?n!}{oT z3X!9L@u5iw9!I>N>{z_S?AKB5L*ph~{E|S#-VKF+qv~UQ*tw;i3BJNUSC!R2lws1=4FM}A(Sm7s zcDO;78M))#Q&H@m{>oOLe5qtv828d>Gv<8gm&dG%x+*XW!j@qaICtm?A!rBF2U-E8 zI0V;!@rY)yNvLu;=1OmrH(~~)pQnsvm_XWSuR81IN*A#>IF0y}5>pdKZXrYuZ^6a` z)Yh(1r<1mUzFRQdR7zjnpf-}(* zU20M|p434q&SudrE-mf>a%EUE39#ZuMKKIj{=T!}W-pXgX~+Ec()^1!^mA;~lExWd znck6e_|&c&`1I!4{!u?5j>NL*r!NKN)-v2Wh;sF_peQB{N>zNUoWq!B3Ur);BmZ@u z1#62d2GS)fAm!lP^r@)UP%g z+#a4)D)sPap!BVX?@w1n1i9aR&TPiEmUq7-5X{pVt-`M{=xAw`0eE14(Ya8wZF@r z-Z@-WJ!zG)BMEZA&WaeAB#M2e|5X2{c}So`?MQ;b^Fn%HGrlYS>~tj2A0Z)@5~+6R>Bjr84jv*R)49z z;vp_;-ER~-N?5C%zdjvqlI16E<{_rm8=z=RkbQ`5yJ<|pysP~;_DZ{M$*#Tjv#dhh zu&V>r!_Yr(l@?lj%w#9HcLDJ?&9_zukB)U#XnNb*lXQ1ZgHd?FBwx8)N2=)x;9q8? zNF^JPQ2seZSRpOjJYNOKhLM!hjmnxDrG_QOJNgCqlRgHv$=BGodIYV58}~2~e(70c zMaIAtS4Y-dz2Q2SG(Wpq+d~|D9L;cb$F2w3-s$4+14>eZANuioA4vl-?M)oYtbRE= zuC=4wT{}mKef&zde1aiZTNruw?z8pz-Z9#xC5z2`k@#R3)uwRbY&I{L{OfQP{LxMaT00TlF0nu!)6r2 zm<@<-esib5KVOmpllh|(IfbGSZy!<22-{$zvl-+NS;pV;1MMif+;ho*Q6rP=YRx{) zRREBo{3692dywcTo7TI1(13z(D6|}zNhM%imk-IL*SYc;!|CX6Z&tKVna~*^WWi6H+; zo+Wz>jD1?D7C-t+Gutv!p-F?~z?iSBdRk;+YIbU5FdOdniRNFcKau}F*~@S)x)})u zVlRIwN-N1pEfX@ZXWl9$nt5#Hh}@kY1V5U2%KqhE z;_Q`)X&Db`kNMm1Zk;tTpBt$w#dP@8bAFieB04C#?L-Pm?8MRMkmuD&{I%sw<2PY; zG`f}CmcL;X2FKar@6ofqEQNo_yDF7;3eFy-evBR%;KF%~P8@)JF!>!|$M!(C62fVK zcqQ4G@=Hl06~tI5xhvDVQmLk(FHRvp;?!`U0>Zd{mxZ5nE-(?v4v%|td23Tn!E}r^ ztR5qL6X)=vz<4$NCd*CKrE_>9A%MWAefRT7VuUff$fpk?gxQceKBJg|AFUqPa80j= zuDg7k>FDjrVkj2Ayrp3LlO7sl{9|m!{m0F#?^MFx9wJY*+UQ5d3o_|=y?jJB*io{< zU0Lsb2Z_7iS<)VdPlUvVhH)2t%NFir?tR$O^*LA|6{> z_TWEyWf;?6>Mw4`GDo48yx)J^_Vs^4`c|y&JDlm+Gb^1MIgk30j#F*b5QRwzcsS4(y|xuR^Zs>0N*a<}331znG7mgi?Gt1GmW5nV0#USeNB(NDMwl&4()yBGpd z1^K12vg_OrT=-Lg8NnnNJ*HT&X* zX)Em+R)wb#v-RS*SU2Z+EYTGidjzDzzugBM;`_L$Vs5*&?xmjzu1Y?UM&YHanJ17# zLVjEl(eg0l&~9#@^k=U<^knWT&ys4T+UrnB)^}^J5v1Hw0-lUsh|ogV>6<^<`vP*B zvS2htrrbpvoZQDb8By*SDZkv?OG@kl$Vhl|F>HsJMy|cXdRhYV_6&G623GdGGOJ&+vL#uqpx1gN+TCB1%C9& zzmi7#bjAN|YV6lrnu0`bdOs2f7Gjdf>VjqIb%*cVgaN0%_$yKF`mSPyqb1P}!9dVH z+%_DCX!HOcnql7D&;Bh~$oXv7mMXBm>Vmi7ISVIL4Zhi~u$YtetHAYi>W9TfMvBz% zG&W!4sS7UTO+vzo^-R`MwqP%=HDcYjf>6i9VKN-+65`M#^j9lJxDIN0a1wWnH(s-c zR(Fy-r4nhI-xS1VH1UeIX{>*Y=G<&NUTf7&a`{{KnygQUkej%e2w`t4`6{KZh)6*3 z9{&V8VP9988m3|g_mB#`ne18+jlJTFF(`)nAxCy!5;LU$g^#=5*}QE+)70jyb#$Zg zyU&)!(L7LayRIriQP&n>2w|zTRra7Ll|VuMB^??1A`o%rl>G?sTx=wgi@FU!(h`gi z%_&e`ChbSDwECD1M#^fA5D_@VUde~LQi5$}le6WFll3GA5duXdVZ0-3umCD0u_dUL z>>=UTMx8!RBAkBtODQ{?ps9M6FrDl_I0E&$H3l3@AAiS*ZFh4bt1FD#Ueefb=YIl! zQl>+fxR*L_-uQ9&J3Vn|%-Una1IxvQVC4CJFRj;$14rQCzSIN#@mA`?xokyW@W#PM z@8!tEm#7)Vgr=`w#{jOC7M;)!*YEHrT`MP{v9dU{?S?8X@yxG`IDF&!KOSWa!qYdU zpGe}uWWhJ@FWy&qXkIW%I;hB9__3%bb#FHrU*U45GnkwL}(ke{Q@%uXmQF;jYV~V-l zdam4T2T+0JCJDuOq6#^PvsYMFJwgC7iu(KS80J$!n#}H(4{z}*uk)FMN%gvoY;D~D zl~Potf{<(XpJuaeTe2nJ9|JV*@z=k<)t_&}J4`-7N*o*0g@cZhebg$S!@Vu($-g?X z5f=(IE*+LnPOCZC?_aV@;*v_r6n^#iZUCj7TOYb$LWk@mTFW1;0FiX_lS{7)r5*;; zJ%=Q5P~8>3MLJVVD_P$EQO@tBVHm1%E@T$dE7=kW)BNHWYcTmn6r{L!3YRN7>0dL9 z@v`x`u@i>B?IE_BHU9L*QMl+$c6gw7Ys2{*9lZhSTg@jTUSE2u?0-vn^Kxa-fkLyq zKM%LxW~69GPjZdgvBj=sU*UR)Ak~=I!{t}Q>*t*k$(PBnpK9X=9+wB_kK4_|01k5s zGjFzf&ElrZ8TnJN|8~-ke-Ih8YkpSICI`B|k0!{!L;L2EEW@Je)-WtO#LzKx z!_eK`IdqA13y4TdNT+lR(%m6~q#)hhNJ%%+-Ej7I{_u+nW`@Pyd#(3*?i(xjbM2z- zatr@#Rca{%@y0V=_leokX{E~}O(yW|3Jla^dd9%nU0k8Y z^n&Z`T%$l~g=k+6U`I82k2ot500M;cXilyn0C~H`SFZbmnGlb_lEwlRooCQhkU~nj zgWP8&VQVn>}}Yqw%N zZ%VjQw-l(shYFm+I61KNDyCfRqK%4in{RDMEdsp^owp_`>oAzKrv_*T!zjpHwSPSs z`<}ibBUbn>!$Pm1+S5}ehbX4eyZ5#6%V87{5s)c~6nnpV+HI0vZ zo5)XMr_-VcAP>HE?a8_(R|0qj@68Gq4q)4VAJ$FcuozrVH(G#w`B-qt=i9bawFTg=&B<4_FA zxzPVm+;j2a#Wc^z`3($PMIPljCAKQ5`F?met}8@y`E1;@ef=;Ysr1&6$!j^H;Dh{O zVLLq^0%RjJH-`q~n{`XJ_64&y@;M+uMp#)@;kin@|LLO$_J$V%T@+!0HLU-JLk>Z6 z+&RIXZ7}HQsP*>yzbY-v4L%s-r9rs&0xftcut=NR5iXtj8aEi&7%C~ePe*q42)8Eh zE0&0C!WorhWYzTC5v`*~&|J&n1T#=`#ung0+b zJxwp%pj9iG7bhF&zFGHvw~~_&kCz3;kx3dX4taz#9!{I9GkPM)=G-j!1W;BI<{&2;l7F_YAtVF(r~}} zu)&IxJCg--mIhilgmO$qrW$a_f`CY=~ep`UxuKbQ!0Rp(Zw2_}+V|`XaQXRJ$ve zTl8Nzm`7v#B)##=D&n#EZ=KiXM%fIoHIZ?l)CK2SnM@$ol(*3yK(FX*QEm=`3bS#D z$;AYh;(<0`NvU;yw+|exO+J5ywq;|a^aR+Ye>Sw%624b{i7v+YuFSB*PEM-Sw70mA zG2iHdyYl4%Bbs~?V{xEaI`srP8bdHn92>Hhu{TqRApUZQmnqSJR7-!;znkO7N?xd( z#7I*$EKD4sh#ew1S9|(X2+8ql;%SWtH6JFL4HxI|%TK>v&C__=3jk~QYAp$53E^*?;JXgY zpB|bl$2?DY@h!G!8eCIm=ND4lcPmPMss-kBQN=Y)D+Old+#zZqj(*f!vn2Nj|FD>s zn#q{!h>#T^L^1d(loJ;nRhd^JpA&|V)nCV8(?ep1q}NEBmCzw?ZpwwfK{TVmSPY#J z*C2XN8mcK@rg_dMv$M@J|XvXVa*Hj3}f9l>{x!4KH}^XW@;5e6O;}NBR$1E zUhc@w7wGr5UZnHFpkd_xeFoRK!!0^yS=SqtOEwpXZJn-Qm=K z1S3A+5aNA3igM5EJ`tO&k+rEvYe5KXlTw7uzErMv;trE$AOGmjD$loT@wXo-Wtm${@Abo|e@p$oVcC?H-Gi!581} zyB>ay&%MxfR!g!HF7w+Tm@J(%whp|`JGgcBbNIWC3&YQw#bf=0EXmwCT@6nqnytFN zWX*2smstJzZw<}eAI~jsPFoaT?^?9+__W`u{a0S;y$Gsp*R|H4C8Y&Bn~(_@mUjL; z;$@hcqbu6uJV_<^Wn3e$py^H|+}NJ!G5qbGSg~R1WQQ~n^h3qi4(Q1+IL=OndZQgOK%~U7#&QH+Xj^XRJH!=y!Wj0;lyG-08@|Gn~`9g zoEr3)R{BzvovE<~5Tnnq%zTGc;7RHdcN+K1sX43B4&n6v?)&25Qgr=2gIKE^ex+=Ck-uhn+k(Pcc>*zo-%{2Vz;gF7 zRs9DFZ&RA99+h1&{E9idmko=up71sFcq-rZtrMyR8h@~q2=Z;TRF=y0jjl%ljpxB| zzzB%)*(I0jz7w*cmds`%pZbN0&wPuJJz|+S&y5H+o$vIU98Ger4wZq!9f7!ozR>tC z=)+V)_8iZ%$$Z}l%hN3?qXIp>U9Z=d3}MrDtLO|ent?!u=ko0+YTDq*iXZJzWK>PY zyYGbZv1ow>+PJfyBI8NEu71Bq30h;RJlm55Ww7CNL@XeAJF(JSE-KZxaiX1jn?ist& zbku4cKf@KYDx0BqVVr(rcPTIGx@*3f^L|KwB!0vp%A`;WpHK2sz$U3WL{OfypQpIe z)?R+PZ87iXOqA;|7zz2Qz8K5(=XqR7*=knM3P&`^hZ{A8Po1Vw#FgNVt2<*|X5LIcP_$G9@ z`<3Psv)&+0#aRDvh!m8@r_!t^H`WCm9-!w;o=)_E4^!)+<9=+zF`9~)hv6n=^X70$ zXGLA9hkU_*x#p~fl9qsuLHyfI8c}{+SI>b~)K9-V&{A^^24E9P$|#k&ug9BfgFY)L zoQ>N^;p%L$Hiii|)0=lptmzy{HM0kDL;h7Dn%}m~7pM$FQH%Vh@eresBN}be!d(eB zoenS~MhjA8<*ucucjQpGpG^=_7kx2F2pxh)T|KDindwJ;PkH!I`WrH(Puh|5wmGtX zx^K{f9u7uhQT!Zq>%b+kG-wERa={uFu89$%QfSAlZdI+r;RMTy1F56#P!^ARdr2F^e z^+P4!Cb+KkxR29Fv7QEM-B#)TCB-njH^1z`7Ohk3d8AwCZeglNgkm3F^qY#`wnkzh z#h`>G^6-&{b?R&h&95}1h$CO2HgxRi;+gyHBuA@l?Qey$^gHAN{YP8l-)P0|88MPM zb;M=baIK$m@lm=(BAd5k(_seqO&Do+P|gjFdwAPZn=7`1o>%c1lcGPSvh^ocrDkWZ zS!f`M^!;V3l1`~~Z*54;@g((M0YPGIcjWbPpjSJw40dd&>G%FbeyWOPqeHwKh(r3u z>>w??2huug(mNd2fx;#l{IW?OO0Pvvn!)$sRV3jP0%H43xxxW^LaK)%=s_W2Rcctb{)b^~)z8 z<2AGjBf*^nSo{<0a&Jl2q}KP~cdAfzdvuMgc0sTR;ob$0tf864D*r9uc`cOBS19S5 zkYAArqGI2acT4=-;?5z(9<(%OaTs3ddD809TE@pM7OJgUtOrUi5S7FC8zYKzDt6;Q_g z`^`%4R|lzlpS6)ewP?!>*>gWW+p=@Uvf8K(!FH9sGftw`)|M$7a>2q8 zPkC=UU3Zn=BN3mDI)|*VX8jrDEG2MhN%|lkepcX9rF2oiJBRNpXx(^2Clp&@DI#$p zJVXSDuu8T6ZPp|(NXd;+aWW!4g@HCxYO?oM7_h6I4UBl=x=YVG^;*SoG1Xx6)eI!N z?7iz7NuI>#-=Nftdcn&N`I>Dt0)DK=5i4(k)-1kd>C3IW@x`J@HMBe_57ra#<#Ffu z`*pu{2Z*ChgjqUSZ8|ria6qvP+MgV<{U570WQk_8F%^Q-OoCp`r_m!9D`|VTEDRyMnR}pRU95{sL&Y!CP?RoT z`$5>I9XFj4z4+}*c21k>tPM9k+Fzd5$_U{iD?ayh_$YQMvkTIkLr;Mb9mluZKWL=C zw)?MsY-~$O z`_SmB&j%d;LCW19KhoHomM=Zil$m*I5L!$+h~@f3-@TD;_QLGKmrVyVlZDbDWRzVq zu#HDTIZ=VUdRnLnkUa`TIb;8%P8F4bF%Akr)Om_9w8SEn8O!P zMpeE}EJ&dGSz$yNIH1V(7r5$j%Zt8V-I5Vd{KZ)N`#pDKLJ*0fJ^5^t<|ECm2#i2 zXlheux)pr#MVQu9TMAbtA>iDxjdAd^_NZ)n~x-%2dv2a=>Ff6*Af2}e_HCj&mY4$r5l~+8%49Htr4^rouubufUXo zsIU{Sys*f3Pa*8I_}33>dYOO~kl;X)wfP`e&co+tUl*k}d64t80P0EtSW(Mb48IC= zeax3geHux~5|*(efL{MXuHKE4ik0WM+QwbD8cQyMUe6(<1y9)6i4JjPz-fNZmT#>t zeVa>xLOMgPSxZl4J#E*MBOV6vChZ2lApC|%W>BFX(PZ1K^M!C04!61vHX(B{PW?qZ zX{b@S5L#j6nNdwgPlG(?6!O+~BYz_2d~phohF?N8&i4TU2}r??>x(xkk|PW!gbVQ^ zuGnmsk4@S#Ad26BthJ=`U#)?gQC@G5rNM2;zk2ja1OwTq_MRPOgG;@F(AFw`SGeTF zWmu4_)|#?OP>;+g2hFCJBycZ^z6G)I+sF2-e)cIG#L0^H|8W@eMb&4rw1_et)kWib zAQtoEeoQ%^FR;pV{}TSOyUR7AYn6D7S9L`}1oBpR`#yax?@wd|D=Wp%tB}z`09LhYJhCLuOIn75(}1Z&|C#J4i5FI3_;r)0f(h{u+ZBrbQ%jcq6XDB zEb466=^vf{!lJa3zjV&dDl>!!3;N~ulH2^4;q?UV*RoG51d=epJNb5b_v+Q6T(RnH z>t&`*05|r~PGCFsLmgZA23`85Es%(x9^^_8V!2$bxIfckt1lQ9 zgDKTqiD0|fXNns>Oa6w_#V6b?KuvMzqsWj2b>T9p{=Kq(BTvc`qJT8YU9BAMmpDaRx@ggDhhg5&^ccW4RkqRZ={N6@c})#KNFcIv zr5MRHY4h=V1g5NE=_#b?uS{jth$Gev+G!gN^g(mC7@d5*Hp%|Xdr>YU{8Esh@^8{T zr3e%95_zDmtS3PRs(-PaYJv=?-J;?kEt(z-(NKeZp^cs=TZmcsG*ZLf&jI8Z8!s3S zwRYF@!x5Q=&2K_s_KT#Vb$dsj>1F*oaQ=kgo0^FwOvi|JSIjW01JfM9>{AYi>j3A6 z1yF#poybMFzd3QVeS6l@nPa;|LH&qqheMRt%_l;BUorjS1PsH`=SzEwqe#+d*V9{( zuEjsRr&ab~>x;S&shYBQ#=&^bN(!Zyl4Jg_Rv4dJn)r<3a6IUAp_sfQj~+)tUXVmD zYt>d=y%CC{5(uP*&|~aQas|Xt(WfL6=jyT{xn(Pn9@8ggt>1d_6;DUBd=+lx7%Vp4 zV);~`^PjUpf+WBJOcwE))M?m78)__>9$1MaV0;b+z=o`mFAFYH)oZX%%j366$jpoe zj)OOVLW4iuBR20!c6g>ZfmA!8uPkPdX_-(ieV*ET{h;#XS;{%=kierL#t^K7H>#!B z;yEDRPr1H7AuH1&gyQ2gis0Q!hWRDW6-C^G)tS~yxt|FnL2AlASDG==d|)=YGMlJ$ zZPDsmMaKM1ie(Af=c7iVKqQ_PjrT2I4^zUZ80E0z!-9NnL#-32YbXhkE# zH9Rh^A1KwG0V~xKh>&ymhOm9I;-x8xoy7|=@ZSRmVLu1I;DF2|-^o_#M>ak+V|$nh zSj<*Q_MEn%R2qY&{tQulxwhJ$f9z2_pBpEqN4&Gqup0S{cLk;6l^CL`8H@_3!d^7w zxSUK3>T|#KC?{l&ai7b;VS3$W`(TraEE>ek-&F4srLfB9x7Rv_b@varjsr53!gjPV zJ?qLCIv9f#hkgw-mQn&UMqXW!-*JUC5q!MbPWhh~q&NVu8g)hDv#hPJQ$JuIA0JQ2 zMqHtjk`eYxLOXyO>QBG~m70;!JvoU3)DnWSvdDqOLVuv1BhFUu0gQdS0aeqs&@YGO zE}&^x4d@I#y}gHp)5Slnwi%mrudJBh5;aWoGV}84tXw3qyvrTk9J_Mp4DbhGV}m`T zw*Q1CyCBJeae@KS)gQ~89=A+^r|6=}Z~s?3i`s<$;+uTODz9>!Fx_m$Q6B!1=-#`i z*X?nN(l!~f=eJ1wVwgV%Qk2Ru08!KyG9WB4S!Cw$uk=%YWmsy}Nnlvp+6qbM`=|t; zs4|scs@sJ^BtT&zs!Cs)o^svNWuYeXak=dvAx>NDQZ|%NZ`HAbwiTT}b6@&|Kq%FF zj_1SowlDH9(D-J=siCmO0>Yid{a2$0f>A=UBeWxtL-$ZSCs|)qUsPH0ml~{tI`{kE zm##Pm^L<}0MKr%9&Cb(h?-?ROa}as8+w58qQvJ`_}HJ*!xO74$@nhxRPV*>~`f z@fn7k(f&3E4!rd0ZPAS=DC8~nR5L|?qIV;^cZ|jGtZzEKZWb7Y`^sY;hx}3Q>l*zR zj(y8UIVu&Qi{d60g^wnmXF8soDq2ivw7{^QHz}I*px8xspyx9&rg;*tVs3(Rb4hO7 z-t&%QPoO4A2v53zTOeV|)=lpVE9fMGt2h-g@5W9F5?znO8gs~bqA}YO+kF0hI63Kr zY8c4c#R7S#2jaJDZ&;0g>tuWHj1Wf?^T^-RYa0Iie4xAV^=NW(@)r2SP8UB`!G3_i zh<{ND*;#iZ0ZiD|ZC_Vy0O0vt{N)%2^j6>ND~9|BRT0oG{Y)nX+$xhREHlbnrwE@N zW=P5u(UPDolAy2>#g|~;odjE;y_1j1g>WrE_x^OE9wW;Z?9ro%fjO}Qn42jqW>q=tYr#F%-(N+y z?*wj;o+xXbCQ#wkUmFx$IU~<3I0nZhN4Y{?pAl(t_Gjc@3LQo7g)G&seZM`)_38#_ zKwDkoa`EY*nUWS3`AhYzCM}H+ID{uEg+LgiF7P42=b8^{|3H-RCW>mW3N_q7bhjNR z;4lD6eb0{D<*?Z?=EWHQ;(F(;PPY7}O|2w>%#ZX})lmnbsSnMd^v?$`vV~TSYx6Ym zYRbpuF)Z_-;UAAh56TDk{uB5hs7&9m^-H~|t_p(9NGf$4pz~wp!$8Lu1)4ZajofaWt7QWN zKs(P14`8nf0lbNC@-n-Jhpgizvr=BWfT32iT!-l!yRkt!@zLL3>eqRuAU+T;av==D zNBmzUCmpdtJyQ?|Fo6yx<{5sd>3xZKQE*4)JoLrzqbX03I1LqUnZZeKO#Q@0p*Kc5OGGlg3i|} zUf3zG54bK(rE^dDxQrxlPHvwn#Q*q_0U8x5S>dsi6sbD37O5ckCC4o!%@{jVdb2PB zuB?}NDZH+JK=I<|EmE7l2bT49XV~|0PyO6OG8bwc74mEW0(K)tL>g3~ zD%oO3RnaP<`>X=G6GYVC7r_esPj{Ypp{czVPtTKBFJ0 zq;v+I{p%3sCliC`6&uW|iId0a97&tFzN?oY{_{#++DHs+*MPl?eFn;n^TE%OHCF|9 z3em1F|J+AXXR!L?esoSeAD#A*Ru#BKN#ELmD!dr?LJ6#M}fs)EqeD zNW~lpf~xSL$s1YBt{(W#fu0tb!=<*XDfE1JA$KHEmhWSJI0h3x02YI~-4YmP7UC1OoYO}Lvzq6R>S z(vd)vrUpZ?P>qa?pi;T~!F~XB-fRpP=u}e?Ed`LEw^aeY8eoFe&Ow{6c{&<(!l7Cw z{&cB*=Wngs_HZn+Pbuw>iKM3?8Ptn`e{b`6KFmQs&d2_ETv{|4j8;RXi>{o>njOd6 zglJ;6-fElga_RdP)`iybGX_D=PZR-duMw z!{QtN;RkZOTTT{vhWukW^b9auA?`Ori{NT?^%~Xyy zJtyQV`r|kTR`Lu~mQn&@zi&={{?u3W@Y~bAIX%Y9fEswX^IXW1>|F(Nl^F)XGsgbZ zx4#^8NzXcWrWB~Vb|VjE;7-qbc4BE7J2`sFAzsEo?Zd=ejVul(^U+-r1hRjWw%4qj z6))fV;JHx38XYLVasUDZW5V8D8nH~S1IchVD|UeqjN`yG|0?jHFkm0~zH0T+F7D*v zx7ls+E1zLipCckTA6rUL0H(SVA8N%>%I805u2-B8yvmg$TvJR5v5%0}mAk;vPls@~{FH z>e1I94ArL#0sft8I_AF!zJ;kkMAXtH**uHP^CL^e)6FB>OPnQcGj}V zK@@%yuxCo17RGKnwNmK1gRQvwNH3Woc8I!&kj31e5g2Cyy7RB}*g;0dt`)LsZ>`jwNP*A5(f!i_NXv`j=)FaSjIbVtPn7hEOp#i(iSVI|uR{7es<-Im; zBtK!+qh+>3DA|>jhgJAc0Z<7aN;Ut+lUWbf5PMTaJS%ip#V=7*71#E}re!$k@6qvn z)m2L2+UXInjV?rbf9}0VL;pLnV7bUqrV=n;zb34OoJs#X#5v+H`ND3$!9&>FgI%5b z{+(IMqy(9q^F-W-@;K*p>A^i}urA8AVBBpk>HR^UL;$@s>wNLppu3kWF5|#2(VbiY z99-!7(|dvHjg{J%{51lCxLJi>p6WGWzj-GthRhu9p~y%ha5|pTERG~}0HbZ_CgB7L(J2(F zVJIn;|^(=H@@tPS`IPNKl%5MSru6yB9&|v&~WQ!*XD7O{mmHye19T| zJm>E(SIjLfE$D1tPQ{t@SPa0~julS>9|xkLoQ{a?3eWxkBMYuz=e@TnWuDU$9Rdvg z{o?5jc!xJ`96x?!IZ%m73xjg*DeBO34_`1ZtYkuNk|l`&GF4olCc<947GNDCrptX- zWuk>B;jq;B(&AZZ+HN9D>)gMl&k_4DT5zhVpim~i?8o%D&x=wVnR?r2kM*tmk5NR0 zVh|O`=x|4kG-RkfC7J(M)?Yqs|Cheogk|UZH5XqLN$p)`Io9y>@MgVV_OJb3>Oucr z3=y;W`?)35S#z@+`X|diy?HdKB>Jzl!aw?PbCo2C%fd;onW{B(9IL(gYUp@BFUY)Mo!Hp4+oloqic0KWtCz>>=v7frlzLu)Od)bqQfO=HYwYYU#zw8s&n{gtAch<>!yzv zHRa~Z%4*Bq)cxCZ=X5x2R(o5yAsU6oX~KfQPJ5=6bCw;BHcL`yo0`n%(+Vq^CCs>;Y$J{(ZVJDP$#2eB279maSlW53(3B3@Z2(MIYmM;b_!KXbazILgoT(|by-$D z@uNvB-S~+n|6(^zejoQ!aAuWZU?B5*s&zJh4ICI128t9KSpjHcrccGDLbjWEpL1d$ z@L;qm+S3t{`7hO64j@iSDFZ0o!5Vpf=*NzN4f^nIjIUNRIh0ZUAStsl5e|agirVV+ z!gvt1hU|lh+*NtIQs;K*f<>!SCwU+ABX7AO8$heSk(3v=@B{#U-U--W(54wKJ8QgX zd*#TSw=a=|ZU%*vjtzFi;gU0MnfaCu=zg`R)RcU$|X;dHxC9Ua|T^4rgI#uQ+siNAG1XS+6)#Fh4z zDzAw4zult>OMc3-60QWBaBxi1U+9Qn-)- z>Bt}CNS6+5{?mOJ&+_Q1F&n)>j08i4KMjE}B8|Iac9H30qlOEi%5eyoppw(!dApiu zxc>~|$7YDWia&y-5EXoXahrcur6udnUgQ6yZ=xnBgIqV=9&hzvk=YY(s3-0ZAs8dM z2lO01UTuvHUDsX}OHnSSL$QIZEtT0&JtWzJErMYkRq2O<>(e-;0C`bMg->s)72*R5 zW&?@*`-p#u(*8UTV;B<-GrOh}IP_hMBTg}L?xV*3dwka=&Pl?<@w;PW3d@FE`saFs zz-)xHY!97f01@r#2n*5YBh5Z_eCC?hS#(WK+?1~cjI*~NQ*3glr3o2#V$iBOk1uy^ zt4~aIct6I%qH{$o2n`K`YqkedC$W%VO>% zt0PG6Q6LH-MxIut5Y8H}xBK?tWeG(a8c5p6W>6PDxX(B-T@j@@-{>{MImtVoKL|yW z;Q5O9Lk$U?h^oJUX|BcV!b%cp$g9t?WDdG#aiaXgWVrZVHhDzT<)Dk;(V#8WGe^9m z?T71?LFf#hgfULY)9G<2Pzoy%jXiG-YXANeiM8TS$biNAXBbau6#|rDL{v?_7K%?h zzX$khmbtCun@AVRP!^ROF;0GdU0lQ<4(93fca=u^^fPNRvKQ1}o~9k*F~xZGX=vY* z885!zs2Iq?CEr5Gzu8X9C(^5dKu4>4_C4gmK=E|d>jA~e&Y}K~x0U34w zh%y98d8(>NQtlIrA^L6q-srssa=1|k_nd*wfdZVaF8(?=bHz-MZsf42-n1Rq%i|TP zwK{QtFds9_qQ?Khr{KX~7MB9lmpj8Rg-fs`sed-Nl#8!8YwJA1jy*jlZ22AgvYE1` zTR%l(1g~%{NSl*?`KF3bZRmV0u@2m4JzeC#|HOPO=wl4kL zNO~=lL+~SwW8h|4d%`Uun@rIkq$z(qw3BHsi`&?}_sJi(-cIM6x$^h7OmVuScIHm2 z{3sW{8F$qj!qG`5q{gyXGZmUY+lLtZd5Jez2>bkfaCxli!arg0Ytfp3>3Uz2s*A() zxOP(_6g##pqEzh9!5PR_UAt$|urv-0sp_U*KrE|B2-o+`+bxqq2Jq6%cdepGUP2mh zt^_k5;yuJg*`Kfor;d4^CR z9A$Zqu$@{Lx2flrpJ$2ZyuvJhQaD16$)(9ALBc#JnGUWRwiT_e) z>*hp|xIMI)$H>)jBteU&vOYpapN?;>D7Ru||(y^41uZ|1`r(0tugZ%Ssj2qT*VvriTcAoUXD%h|koh;9egm=U z#qq7|tx@tbu4DTXy%Ii0^pB6l8vylMzSHqShP3PvYa^6D9@HE!_}Fle$ID~qHA6vN z4XeoxS5-<_SL)t`TlM5aeqcs08T3R5&;GqtK()Y*vCqNM{FYoz2XD+hX90#>0!lV7)C)(FB<+&dyNtjwe|RqSq!n;uw2 z9{fFy)@)XHZ?QxFA&Z#TE+h0nd&RGp=?kG<%mav@-XWLZf|3_qEu3dICY!UF0G()89LLF={HQ#ZdwU(PhaG?DVE?r`iW`=!B46^N!_BD^7;1a0ldFhJ>hm(&ymM-ySL~nvBxP~LZ=@(!DfNTUgiyP6K0VeB$PTJD>-fa=QE8v^I_vgeS!H%fShi&BV} z29HFWoZU>>YW!l(q$GoLD`ur9{rq^L0%59HohcTgI(Th7o~K{nAEI){j&x}kireeg zv-;LA^eD?0eu=gBhXg1g1YHf&qnghmILs<=S?j&qx6yBy8J}-(NY|tyOoEDr0q(|P zaZV*}YFxx(&Jr0)G_kp~G$)`MKGpOd=xfx(MyNkO-I$j7pGerPG}=J2xEwoK|3Uf`fk7x( zsSdErwAfU@us`25i*&AtY({5mR+FeKl0!Fz1TuDGUeczR!<|V+&xd?5M}XoM^ds3~ z>nl^A%5fw+KR?$+cY-r+r@YPKv-^XW=?B(%3Wa^RM4sKY%Q=}HSg`NUG-+Xh$sKq5 zfU$3z*$mMsVi7H`Y0qm@Uv`tOd_6($)E4@H!KE0=vA(MHdD%SGnjtvspCb!)DBa#q^l5@sJW zZL$@d{aKN0`87@XMFNNI9?{B+%7eYmVuE)pTny>9v>RzmMn5f7wz!7RlMKt7Yp}-=4ww`n^xu zf>poM!%D1$aVMKRM>ncB=*p@1kUiKP6C;9w)2iy5PZ=zs9@fO$XiZx2(QMyK*1EWS z!&KUq2Gd47H23G3kp)X!HN(icp>&d*P}Sv;_s-5Cc|rpWzt>V zq^J!q^a2RLBCt@wyD7I!Qs||l85C+lOADVU@kU!Pso1+A;w!9Oswni4aRF>Aya@H5}Q@b z?MkW$X+l}y>{&+061zN%3`D-2j_vUDkcT;UMVeF9K#uLJYr3dR866XWcf?itp|xdwm|c%U1eN-d_I5M7VE1G8`e+Zy?eOQNdDapS4=6;zPE6 z_QQg7zjr(Qm3(xZ#PDhO$FI9|m9!(Rjz({O5vvqp=qY;qpkO&7_pYV&HcZ?F-Y8!S z?DnF@^mu^y(Mgw?RK{bn9-M)=p7F=*e&SY8g5ol)eT~o%m^`YpdSiv|5oh} zt50te2(mTX_B=k6LtjBr@vp5^2$;jAYYUYg3^sG#7_Z}9g=o408B01cA6$)(!%1rtSNHF=>ms)QMz)>D8BsUg6nOI26Y4az zVKAv@0ErRLLEW7lJR2a|8hKkifU5^7a{I?RZ zsy{93IA5p{iBx=V5ojCOjjiZu*$*y!TDx6e-fE;p_6SOO0>fB6$+p~APG5lGZ-}-n z*>j<35cR^sl!*8#N(qx@5bp-xnYDCwp?LAFDutnnBAv)B~I#fhV&lg~9`ijD>& z7oURojWnDJR3Ive3+-(!xDiq^wR z%LT5yLDIOk3sSWXe%4gPNKxkTh;qf=xK6IARsjZ~0JU2AtF_73MV+DOeb{UAln0dc zoR1HUILorZ+KV>~&j?FLv75h4l?TMTkw4Ecgs9fRy)qh`U~tJmQ+P<&7r4{vItnv4Qd7?kc#J^I5H1+2GW(1I~H$E}Z6 zEa5?H4W^N@qf8(A{YpMjOhbQ$Bd=VoYoZ_jB?v_&kZfIiTZkp~H~I6LgR4%1cK1)W z6%=CAM&*aFQqz+183<=}#AsHl?bMp17rVz(l3pKrmJ;`WOsF#B+v|)fjoL4A!1MF0 zslcq3u#`Z>J5_pi5`p|?bZ6bO8!aNf##$)H4B7muL;!)RT|U6`(8$*_HZciQZ385fMs}-Cx>L2dDg3G9Drb_hu9j zTlL*#+us`*M~A7HD%}$t>PZ;WD{VFY`TkbYsxcJ3t7}ur*FgEM*dWRDEOt-C+VkBv zW=+`!!0kk7K)fbL2VQvHt^dAZ%|#%|?z1IhI_S7qoXytyo3Wee{@1xQ!odHLbd~{8 zb?q9Klx}d4hM^mz8>yj7x?7MEl$P!uQW_**y1Tney1TpkY`?>AelWw1XRUSJ_jRa5 zl_R?GI#Ehj^u_T<39b3_jni~ulv<`Dwc3YQ(tCuoY3InNv&2##(*$3g>J z6IL~m&2esbbj0^u?0&>*Dtc!k@!Lr~8DN$qdSX`+TZSVP{*ocxFBp~ALr!x8JUT2}ZcG^W{=p30Nup!|=XE}`LH2q$G!Zx=f+K{Xoy6M_tXSr|7Hi;zID zkU}65!CM%vwy~X`Li}F4i0JADVc#QQlT)xssf(j<3!Zm-_$hCvx}XCvbxvAp8&cQ# z!r_vZBdgOJfnvzt0Y?sX;HqF`EdIo2Q00FOrqFKL98;lhAc8W5>()rksyG^v#YR*T zsJL1dh*&K3kk=y!co1Tx60;(UiGE|t!QrIYR)R0pQrpzOzDi1Cbjk^_iA~O(K$G@T zR^Mic0ka)`REPwfUZ=aPioHHRv}iX_m%ZJ`Qi=r{%CM0CDkJ_KYZJEI z_-L_6Jmd|1XXu&obC#h%t62Lh<9GNScnZ`B&ydtqv`iLr#7RZN-rLf{9J9)+zpEscx=)94JeWi4Iu6PxuF}&)K+?7BI>2teq*_PzN z)AFKaCS>~pzu;>5z#M_~LP+zacNi#;qM#ECXW{9(n8|qOX-?qN=|YK}V&}BJ=48C@ zz$|Y4L46x$qCf%{6^sle+oRj1q$m-bqukww=Mlr|x}MX>rcCk$`~J#W*4hX<13LGq zYLTxQzvK`Lq5nINMy%VXVasqbAQ2tL^$z5Vuwrf7hK_w)}f zGLIin&80ZBT1j6KDifsO%JIC##nB$bu49PYk1^-U_1Dw@Uco!VLxl%71H&`>QgyJI z97vw>kR$-b$*-g61~|c?xU?VLh4`#2w6nz#4s2wmsq6^Unspl-Hfq2WoYNvGhoKs( z%^fkv`uv~%rivEm;0T`O0CE;YXZ{zP9NO1@-tj*8&W2BTwvi{wwk_{>iP8G3DUQo| z@q=%TNZ>kiRifz2`LG53-1&D7AhS05~4V3oc{w&c=aTjOR9lBBUwLi>B{h z;bcPt`db3_)rP<^N+^h5vsf8PGWiHTi*kz7z>TrFL(@rIt`GpOfeA9pTV)x09Vb|6 zU*sxbCo)ATc=zw~qcoRGYziGH#VAfF876~ENlD^tfirx0>C8#Kobi~OSGvFb+8LZy zpG8ld|Efa#AP!T_hlIY|(kB<#K-0TqOh6|*AsY)AFI)oK%X`6@%*oEJU~s);bH~rS zg`gT9vhONH%!q)DG}y6cmE<9KXo%G9vlPn_nb;D8R}EjHnCcYmFCzE^Fql#oS?H;xy!P#(ib4oJcn{->D) ze~bum+v-DaBv6YxpH>nPB7|g(=X@gk{)q#QWAdyK4&JbWm3!( zj---~G6E*9WIm_-kw3j5QfR(I?5?Ch;lS0`3!UhZNZ=5Bc>C@*QJ5@_0spd7XuqaD zN%bs^S;61R-pe@Sd&?@?%bEJlMb2wUHa)ufjRYEGgCJH48{VKklV~rt{MB=bel@>$ zq!p4gD;k9@!zeSO-zM%u;2sTu9l)oyhJ8ZcefS>*T0l<-15&!yJFeF@nV&V%VFd!t z7vmiwjDt!d(Wz-`S4P>~5>iz|L_mw*)ff7#Ml32Kl>tj4FS)0iqu}zkM_HWr)*suR zFXadkLf!CPN4RO0^BPq<{SbB<<#Tp7C}iSbP$Qq-RBJ#yOMY(&Z*7F zoCTw0WMUWMOW{n;J48O-c2=H;KZ0{uC`Bj#{FQnZ% z!V!VO!+lQFNuYW6MAsMf151|s&7tZ-!4Z~P&Xjw{q81pgBu}|akMEQ9UnF3EHj%3w z$f*SZSh*!`z&}=zhpKCE5HhPjpyH?vfylcs7d7KiW(WG;2;xef#VTn+#((!VVOq0Q0{>?9 zWxE*S+-TB}Z9^UMHVf(=O~&y1a>jWBZ++~GnPCXT`U_dn+cUyRr_e2xJjsJl9yE*Y*5g+13H_h22sO)*4D|@gi!~A;k(YV#UgZ|H>A1jgtFMZgQ#XVE zhpZi~3;_s!)Qe9BwlltGk&Q^3?i1wz%u4cEjvXLm@bD?!n^v*WDrA;vbb}d$UE;bF zzY4`pu3^T$!USNt)(TvBff2t8>(93S@z+OxU0>ceX%cFqSA{$a|I_cONMwoB zd>ACrrO4KxrbviAAN?Rxc$cSezj)!5J`VmWW2KI&%}B85K}Y`SE|pqL2Adi=-et|} z10X&~C^`<^-Zo>>Yf!eJ{!cj@7PvM2b8~GR-E~R{02D$D7@w$Pja0bv?peaaFbFv$ zbSUFr%c+n9eSTnkM5>yQhW?arcXTPZnI9D>t9##MWc<->f_eCJoGyBw!1u}{Bd4*wf|gW%HO=029eqH zg^}Z~8ev1VlR^uf?cu>`zQCaN2^MFZQkd#pVzZzRV0Oyt*grxT9a9+4X-4}UZh`tc zw1jIoYoQWb%Ei)y5#I$Th50CE{+I~-_x1WCA(r=+O?yIMdKYVL@}pV&pddvi25 z93cg8qFumT$p-M%HTG+yhF&-Lq#nnW#4o!LK%NBNFc7z_>EP1}et54sVdst~d`KA3 zuZ8IwIQ{Fq%jv{R#Y-6}s4TJ{GR|anz;p>}X4J=q_d1N616_F0B3?Ws3sIht4Y)pl zI2JV}EasZT50^!)sfu;ItW4o5hYK(*(qBNFR%#g1a8lh?O#;8dfS5NO z+oLPF*OUzSZBd*dH8j$ia-+HY+vR74%F?p=Ih>;IH(z@2z_0kQCG?lZX&LKjZ$rZ^m27 zy$`#{K*>`Tn?q$o-VoO4H+Pqp-g$n**1k#F+is-0oh#`Zo}+QpKgda6)o-yxFg&f7 zBUteyd1BB3Z%j&)jb;z*C0b^0R84-W$$?%poKyH?8vZ6;uB38Fw!?dxVmh{nf}fP7yKw_nsX53HjQe^B)e%OX$OOtk;cRviXQBXw-`so{ZcN(b{vxv|cS+@h+>tOrA!NF}Q zgWZRc6ecU~GAO)gna*^pBwem5UOmB%*U3;w`3pT%pk8vf1X8i6#`s@@P=ysVM)=WC z4259{sjL1AmSHQ<;u^|+cIl?Y(epAM8egGfJaYu!cm*XOJ9$kPkA1}~62XU(BZ*sY z(=(;J3=#&$l5zKF?qF+ zBW@dYqKpiWiBcy@3*TyB@BQP>fptuK{$DhZ z;S@(5nzi}U@3o#>N&EiSf(W8U-DFvM|R=NF@)l2z z_fx`1PMpHWl-a<^rO$MsZ_M0uOiXAmPj}KloC^G4x{Lp4gb8AMx3Tdsr(dbXRHbta zBqBnUgdZdjojBO0**^$kgFqn}fZ}crtxmV`yE#TL9V4S$M(WeQhDc3=Gp>?Ht>l%I z&W1Y*GX-zB(9t2PN=Z17(1C%|`&a=Qv7<1PUedTK;|JFVVeWIb=d^m$zcFRJ-e!&J zV@q`<_Eg9pHB*bRR$@k0t6EFRcv8c$>Zc`T9KQ@HF!Vf?G*kn}rYK`1+BEg(Waq(x z{bVGmp{UpW4}TaP9G-{+Tu(aZH@r6n%G@?@3|tKt{E~gTD{^WhV}R{Q(6U#uyCXJ=*7^$k_DzBa^Wex%o&4O!&JEAP`HM)M1B#%Za1A|q?h0+;(w;sIPTU%8=Ddpg~yO%*-=4<<4_;t8}rdqEI!HiF3`+=dNQ zDl@aJr5a#%Si-i(RL9B#!IEtj$Kfs`#z>%o?SRC1^w&Q6)wHUUJNniFdg3(%*3*J{9=VW z)*b&7A*I0y^k{GC7B$xX3BlQpEnhE4co^ZSZmf%ztd(&E8Wh4k`TBnQb@`uD)2O4N|jkHM|zsxs1qEK`j8BP~U|IL)YQd7~9G2HI&oN1$FI2*!pkQ2a# z4e9HLj@1^UxlK^Q|r!vorv-BkObwcmW}_5py(ap))7?DgekmHlWaFj+usdjHov zjWihG2T)PjfWg~!J63w4&Q|wn)&tdM(>Qs|%Zv)SKfKfVoI=awXE@0%A_niFagUM{ z*S(|S>y&jqV(Z-g`o~|mHcF&`DoP9sc)_Zm-9Krf;lgLBK4A9zxDV-EfeUsSi(B)0 zpZ#1QkuFi3zb4Zqitu_hlqpk%b=;_{LpJp5Ipg`1HsUZ;YR#MZNw3v49*i)R6Qc`{ z-53R);OKOP6O(y8T{W;P5PWrIZ+V`&n(8t-8sFdq1mwNU|Gr=p6OKVGw$CwsI(1Bh zLjm!b6W980jJdB%B>4_s2pi_uQb$$k!=s|=>^l`SSN`|l@bCQD8~!0lc4aD0u3VQ- z!_lS!oau8hQ9S^9qrqu27rnHUA$wpQ0l#xWPGZp?qCp<1>)KQr8WLHFwPH7Eb7i<1 zxe_++f`5wb06JpL1fcNLnVxp%8QFn3z$QwS?cM20_N3KzV_!5`G?2(Bb%+g^XJ$)x z*Nq7XHs-$t$xD+4Qtw1~*()A9a&M+yXcr2wco%Wepq5&Ebr`v#nSBt1h_h!ei%@Q1 z{HIy!^F6c0k_tJ<2Au7+bQVkc110jKC$oQK^ocgP=Dl2kC#u~C?v5Bw?498@)2TrG22d*}Ezaihwn^M7wUjnVcSY_u86)K2IJg;pu8Sa3WOl;6HS zXM8YI@?&#A7E-wF?bV~@_3?;fT28#Hc5r!Rg#-vk^rgPoKc!p> z2@0C6n%8ZZEf;ID{??{oWfQFdLbDvbIshZGBg-;f;wYZZZYuS&=&mK}-xG|ehNdEn zE&-Ooh3C7piX*Ppbw$9a-vOIu37P-T1t*z@KZrFRKNpF%;mK}bLtDMXvSZ+*qy=0~ z2l*#WT;ud%-3K{ zGL~@p7a&<47y6E#G%#O+t_J z9^7b4xs?>dFw~3DOz^}|hmD+zAN2`JM3H6;mOC0(+kX+kl5g1=er8r-ERc22!{gKN zJR292+bM;5eS*@}Ftrb5yE=N0#L61r9*%e)Z5MV4U3!(=jv4J^laO;mLJWw_;FvH7mKjh^E zly@uyiFuTjnK%r8^21Co9Gfd)#ey5kj&X#!mbTxiF!oP5&IJ^HNEe2Y7!aPc#8j-{_xz1cw(QhOVw_y6jpV_lGf3tA}VU* zDus=BEDf^r!}@=0nqte#%VrnoO`giCs;arL{ol~Pe*LO?pVE(qhjwXVD46K&L^k}g z+&?3LQO2W+MSbm_i+<>ozNrMrPoki2_sCzn-(tb!Hq#$-l?eUg4bo%(K6<3~W{dqW zvo&B23F=%moMF0 zHt|6UuS&>axG7`GKQTv+5RffF#EHakNC)GH=}f4Qqfwy~e_~?cFZ7eAdcDY~C7xOC zRC)PyK>Mz5vHv^s940s$1$H+mIXS1>qIEd~llfbK_3z{OoM0;jYsjboK(?g@8> zj>tnMNEa{d3*@i|J~asj@$K+g67b83H=&Ls6?ZcI?;0vQ_5kRY0?J+oAvexsCYd?C zIJ8oP&;tS{k*}hWlibkr+WRG&S_14)o=`GJa%2u4-q#jqt{kRhb{RC=+k2SPxhCVi z$FG+j&y`=ic+<7RQOt+e{W+-3!v12evWCtXdJszgWEC|TQ1QGp{<^^Z*M_wZ?FltH z?CfbxxLwA^=XhFWzWqKG?-y}vbA4QYt<5ynBUlWjVeL5b>(;AO_nGzjd70Y)mYMO` zUvKlnzf59G@xr-y05NhrVt$Vh6v{W#wt>+_OAdcjM^@2V8N&BT{~6=C?S3WJOog~z z8b!Y~{v?d|v^^C{U*;cZefBPl39_EmKZDzpsvP=1@N6NSv6FcD_IQCRp`Y7!aUfV$ zSPP)F@f;9f5`8=-;LF@HVPcZX9~<3?=NiBqt|_82);oPDOI~ z!puVq@iWyEwVQy7ny>@82fl;LWFUkHnQ~zS`#$kGR-?ncefPifudW7x?cPl}2tj^DhAPF#K#0DMDLy?iBC6ndtBI%!~5sO|fIf>^@#vYU8B6 zq(UCDuBE&QxHj~*8PCOOklztvg;uy;n0)0jain9ob{0~V@iSeo`tv1psS)WjDudb{ zDZjH!lr}Nh*M#-w{03_VurEt;AdH5W;D{Y_w!kwI_wftT4bPj8y~`^`y@a7hI1v@! zv|lID4-Nv~cF2v2HmE+_=|&*~I}nVt!^-yz-xs~_TfgdwZ{K!-^2enan|6JUatPO9?*wFId6H|Ll(_d&r-=>llQ1d-y6N~=k>N#kUp zXKfF%xAfrvA6=b3trEcTpB!&=BqL4MViNwgV}v6RK_mWj;RmnfS@G@Fp+&Qa*T8dw z@6%T;+-ll5^0Zp|swgjf3TVaC`d2fvRg!4rKa&Bca5O`f;tWfE@1UT3>C8mmT}QFt zSR6-j9UEYU5!T~D$$gGhgZyE8?cE&7rb|=I)l-`;uxF9B4nOdY5hG}1=NPFeGUihZ z7)v^uGy#r$5FWo(7iRqP4wR8g-Q?h^%tx#ks5^=e+Z$$WPiyT;XDy*I8-0TbRVYu0 zF96rt!?rF(Cdi2s1Yq-AzQ8kV{e%8|-B??%q0d6BdJW`GE>1Y9LLPO?8p_a)weaF^ zo)h@Rn6zSwSv8LnIUT0Jox8zW(2vdq9Se_F57wXw2Q71={#K&Z46ZTR ztY0Nd4;?yAQn z0VjzFG9!1_5j@vt;U_qKP|Lck-{wsa0Huf1FZVYV$tEy-7fj`K%+WJaQO%lqiM@mW zFa$atkxXiTd(~wrbBqnZAtPgb+9Dp}M-$8lryb(eE;kT*Xk+X6$WM6XqS>XHnzAvi zP^#kSrFPBFMF2z3qqvzp1DDE@tZbWma<_gqX>RNsTCg z5e1V>HK-)Ruz*&_H!w%_DFy)Cc^R+IFZjiqsV+d*^z+HJ0SYK~9Cjyl(XeaIyg7NP zWn-^NEx*=lf+4->Z3$6XYc$n76DXn% zA_=Y-DN#kM#GPWvEOOVe1;bDr(czP8&OM_MqIG~%A0^c)qMIs6_y4qFzQ5;2rL}q2 z<@`)-wc7&hAzh~Hzl}ZHkodYxMO>i%dN9Es_{?!8987M;? zE;c1gbn97~)jjm|(!j_-uR|=@_X0^Ce{f4SiG`U~5rKyi8BDC#>TxrpX`bq|H#T?z zMs)Vw%vHR41AN%_Rx{^H)h7QHAXIzWF>wZte1K+uz|~%cAvbG_apQ?;^s8lo$3Z}D z`wZ91?C+}{7zv)p$-2N6*}r+1aq>@N{bDNS{2#L8frZ(@6^-|x_*?j31*Mqt*9IqK z8Wi~5+C3Du%Y~?EH^bm&k&ZVUivja5aq1iOz_koYmNS5IN8E;fDGA9{yz;tx-g@b~e-< z1ipI{HAecKIi9}q;P)L|okT^CX#4W+AZD5i;*YUk|B};i!T=lMDjuHDduHd$yR2fB z=mZ@J>b0JkkO7i7WxQ=Eu)k`6OxNuT?I{N&*$Mb+66EQgAldyBbvrKKR||-K|K5&S zV}E8dHp48@7Sw6Mh@Z_F7kL2*YX*|ad>`OuN|CNO8tN_hF6N$wt% zNM5J5s7&l_lK|fCiKkaHQ)bk>xYA)aJ%&l}+62ryjP2@72Z{f~#bo5Fnd;S4c6{XX zVkCSP92%Mwa607$t_upZ?||R58Oa}>`9^b$8x7Q>f6aZqz*cofqUMTDuw`vHcKRD*sYhv0Xaw)V|!!1?$|2<4=qiKUtk=o};#U=35<1 zk}aBudZgX9-?rwXUKo}V+Nvbcp0!u~z{~UI<7RaxnZ=0u8Z;Zl?2}e(LKvUxsk?c0 z7s=(t9mqTUe(EX|Rt<#E9LZ}r&rl9@Dd>St$z9tS@3{SS@Q)J1V8!LDW(%pO?u1_U z)#h97gg!0K!pgO};LVGT()WNh|HHqdLL3WP8M)J!^mX7kMENzpf{#xeBEAzjdcM$? z2y(XLPjt2GN(9FAt!7lUw4^^Po!;kzhZ^_RohX0P@d6i3uP^JhUCMlNOX;8isiCqs zml7FZu@SXddD;?v4k?y4bvjMNr08;-tKjGe#^3y}_ImtnI7JtG*qe<@r*_T;WBQLh zfTWo>Q$JY7Fq<{xOj!7y&NID^V&1g=TfgiI42ysng>W6__ciKzc4ihvi)Q_KE5-`x z9HV+1gr<|888w@A)+o~o_uI4ofH$4dA>`GnhK4Rnb$o{Nb!WTqxfM8^#u<2#eUxu% zooEeV97lny2Mim1QEUU_T;o!UWx)>NC#6E+tH(0ZH&k>*)5uZ!TE~&~cS~9&6|Jr{ z(r&qoPxz>~4H_y-MC^O$$#Bv9*zmZNAM%^AA+Vt&{;ZiEr+h`%S9fa5k?|eo(lG;$ zWcqDZ?Bx6Yay-;c$?R;=FgwqY8l^haD}Hzre--^8j4`kbPse9h<047ID=5iap>v(_ zg*6Tp=v#((Q(|5PAH3W-me>#jW3Rh*K z1Yuln7@mPEtcm!VcRroJ0kjug*70D1cNrVW=;wjFF;aebvxfgHPc3^x;W&J$!WZ zVqt3I-rap{Jeo5`2UTYttJe`CvBe?HL`3w)S3~$tz%><^>qs6 zr@VIZ!D6`Z(~WmP3EEK6hu*)j_1xG=7QI=K{OTmJp*p5^Ooq`zc(}2MMRyJeSM5G) zLi&Cna6v9{o*|PyHPWn#k*x450b0qQ-+C8swDd+ZC84X4Y0yOdAGAY8!@7tDq;|c%c0S1bMs8=E zA4qVD5AUX+*W|EG-B~{>f14JzFvR-bx0KK!0K@n?mZ7fxWm<&dFUHma$}@n;#w0 zCfjBM;qUsa@MT3N-8v=@v@Mgq$Wv+jxo$I=PVNFny6TLf0^8chxF6_~JM(M})c2yM zEw2w%TFfB>A)WL3gnl5aD!60_fkp~^cMq0bdhI(u&+=c6+}j1ykkMGOto=*V5DnuI?1tjRQgm7BG)So2kgJe`7>;oS9l*O#P>ns`Et`yDRkWd&(uhEyxl z!;O?gUhq;rl~6NdUGjf{#^B1&^s&Y*ki~Z}lpU)M?)`kM!zAHXXLvw+9tsCLT!Leb z{!@QU;G9N(;lrQ9CYAfm|KmTjXOkiMJS!C-0heb2$*Uv>cS3j3rWaKCAYBJgKk&9~ z5Df&>B$RLE`pm+q_iI{P!z1HO^kc~%@e4t$C-@T4ca7!i!dDsRon zP7bLpOJ>Un_N;rn*?-5pL?sVJv^!e*(|Q~p;B)5j@<4_7`of6uEtUEGLDAOZfMuJ` zOxR|FsKf(VPjqQpHTpWnUOOQqM2Tl>D?8S8Kgp=ZF#cWoMT#E@mSxYLsmaB}nX&7B zxs@lsVRYej8cKsr1cJE*k&0&+$i?q;^lIQjkpg+-WB%qWcH-|4jVF)6goAJ9PM=Xf zEli_V>uGLT^|&>lbgjqk2T!YLR?OP-yb7|z&g_S~W^DW_s6k;E}`q zHg!Zw6L1hE@}$($qs_R!c|iLE<&ew`3LiF!3zV94F^Vs+zTO_~PgU~GGLwY6X1B=M zaCuI4kT%H6-3p3fgbCe(>dJnDR{|49iK1lM;i?))m*SZ&@<)PtziFq{1J+39z5gaZ z+vXZij!>Lq1>CMo+5Zu1wGO#libi_-y#Fd~6G5~}^5XI4N1N#?j;LA;rXZUjk-Wex zI0}I+W1H^kNU$t}#g*3`N-(tG;rD)RS(ak^a03)q~^1s!*%jJ~+S_r*EBvspuP4|f zevI0SB>(~bdLnjE^ML1;=WEKVGODhILO|kc=I^oZ>)W>shVe<2NXv?Me~#}1zcsQ$ zl3keK#_ytf)Vu>I>0TdU?n06ObY6%w53&;9P;6kek}RmeWs6h1pqm; z`ca^?OidI8d>%x9!{~s_q}MZbZu}h4yM*XD?)WgPKl=KU-;`VqZZ9uB+S&Li@onO%G=o9i&R9awlDyx*db_AF1EfseschsH@Bnyc$ z2IP)Mvc!{kg&o?Jp}^TAXkr-f9S{yWoviJB;3H$N(C%BSG^tD8Ql(NF;s%U=1^kZc z<${o|~iPliZ1Jj((XJr`TThMBS)e*gQ$?PuzM`eB`NSB!sdomI0Cq;G|ci zntNi#U&2=d_Obp-?iJ{mqefmo6w&l*j1)oCS5KZa1~|Mag+RtW!6ih7>{Hscq=bs?o* zb<@XrAk&>~ubZ+z@0gjkIWGg@WxD2ob`{9T`9r#l1RzE`s)&JKUZylieok3k@VTW4 zg-c*SJCYGGM$im2#?qOANS49}nyqY>*!_BXLY?j#JO1HxCc0S>1k5+mFqiidn-zOE zZK(bJqpnW>k*<|+Q1U*kHat#z+ylv%Z_XN5>NxT~6MsB82n!=hk)0z5WSVHrt0I4& zaOLoWzC9bpx|A?5G6}WKHEZ)Nc)>(WUPJI%O&lwQAW*)0lrwn+Z*g6)FD8e{mBqP4m5D676c zb~x-iig)Pk#VmL1%5LL{zfGRKU=tWU>F}Rx3gaBw8z8fKbQ;Gh@i+h|cb!}lQXR4y zU+E=5_6sEN%)0QN89O71duwItOh!)p+k`Q1x7l8j&~B;nv7JO)e$257)DTsYHfSDFMm3+2rC_~to7OdpY65eh7DK=K;*{aOy}9T9TAb3N35Bd~RUOu1*>w{qkyEqA-`pN_m_Z*p&T z(jbp52>%Adettawk2ZYIRNildRXg3f+x}Hn_-=9m>?@`hq5JP-hyQH~E!)83n?%Rz z*$=Q7-`)?X2t;ODN5%0w3hCj;V4r}a!C#^|`1bN#n~@O2PpP=U(%zWe^=hW&RC}+; z1b_97J50N;=MUZN0aWoi>G_lnj1F)CG~vj5rh4aR3-pCmMO?z#OIQi;##xfSZKrr` zW#9o=eLZ1=C!wZH;=ocd=RaztNRp&MuIv^LX@hxe1YeA04~cqA@ws`S@EQhox5FaI zbwT4V zBOWy*YSCu-Wc(s9yiK%t?^O6wYwOXEp-#uH{@4Vj)-o1 zeK0r;ojd#b-dT{H9G{}`dm(3h-i*{AG+9z_*_b;~64Ey+Jk~pu?qjM=p6{an5$sp= z1_Zy(mY*EPG7LVC$k`gp-lVD6@72OJ(gcNbs8kp>M+NSMMdhr^;7m<%R({kVTNf&)8gr0X>s+7Sjb6RhYZv#S{0KK8+cyGdk|FP zwdRUDpxaSnwX$ zb^!=Vq0p=p7ASMV+JeAD^)#I$CuiMt!W3s;WFfz|UiHF$xgybOn-S5pa(pwAs(BaU zcmCRZwtk<1U}C8CQ(ipN?j$Sx!?sT`k2CxCea`og`^)yYfFETREesj7vaZ%+q0V`6 zl34I?46L7quJfHGKh|&|W4?S0P?DDH=40CX&#~Og4}8sI^dJOA4p8+<%E5c5ez6vt zr4Zx+C1dGp+K||$CXQ|iL1kxgP`H+~av$@vWH`mGIM3mhh~N~bmv7gD-+_Xzv_xo@ zrQE}?h#_6+eF_lRz*(Dq9pNyx+fiA+O(P)RN4ldcll6RO+txb3VHEWmLofCfCl2?%}29?Ivwm&bzd4X#~GB%6FY zj!!KS)pi__i#FUnQgKaFZrKO{h2rIi}Q28G`v z;Ku3jv*v+zN59KhLFIx}pooiBAE_w`1Lr~Rqlu?@zHFf|ZbpWISkD0?X6%*7&@3$+ zcV}Ch;_R$@AO0{<$Zx_>AZ|ZC6t>kvp3eIozYr4SwxfS`53%Z6r{ZeI= z1;FBV*c$#IO|{G|Ktrs z{qElTac)kSjT`d8u7I#KwPljFx+|tU1QaC=f?Q0T&TG7OvTZ3~{3UqKvq<*% zp857p!e76AtC-o=c7|&1hg2k=hl-^hqA*EGZ_rhrV0A?yU7F0a?na6gu!r@Kze(Lg zj>uxw)#FRbgPj5_?Z1;yBZhgDC(CY2;%pny#NRP}aMfRkXw|hCQvY5;>~=aYn=6?C zO@w*W>O-#TKpB$$Iodz?5)SN%_G<(Qe}EVv?5e#_8u%G+bf@nZqBrO^_H8f|jFtW~ z`Fub|LxnqMhX(gQX0DBFW2yKw3km146r_MzBeC|(BtrmVt!ch(LoJC3!Qj(m{ zsR~kHGlD3~YT%?73a(fuD~|nV7nL;bFN32+fHWi=y1bDvP0hO8qnWYav;t-Lfad1z zU2xafW^;PF`N!5}4Fo6xLGXs~gU-O7pZWlIYU)Py4T-XOr-{4>adbrU)pMb;j<-;)K}@j^7MGXP}&a9cvxzb4);d^ z@uP})C2|tg6|F(#R}1Hr_|RbY-1Te&Kep>JiH8MuJtS5^2%Z+&d_LI1w7YJ9!gLwHC-n$0QQb{F5L%*~cJRv-|7DH`6@tEKtu zStv48Zo@PPQ;HlFKvB;|=0g43x*Xf0@tf#>8s=pdi;0ZlFBLhZc9u~?WRN)^3~YK%E?~v5mhous93PkEsq9f0heze8DJ~00^&Ko zzRU71qleE7r>t6^K7v}7vjx0}-3#Q{XhScgC1Zn%3&`q@`NFP8mXxnH5hsq&p)sxx zhzs0ozdjF(yFX~>qNawpDcH2%3cLd@R4Rwfs2Q+H=$^%=i6{~UyM34 zWH;?^m_-IJFAH+YL`|Cnj&%aV-Xi><0nH3NK=vNw<5|4Ut{1}aZCwi&7uO%?*q5{E zdQax*qu=~<;Z7Ko#kB+Q(omS?5=S*Gy03W;aVe^C**DHXyFH|!+p^$acl`TzCH}hu z7-o%nel#qy`GqvEsd0puECKW_rGY%94SX~>DMCxJQ-7$Si(FEwM~E`tUYVgfUbJhp zE_UG~)J{fd=e1bIERk+<+<{om$hK%RC@OpHp5pQGbU*-nRG znXVRT!XLpl(`MAl_zWmQK+_a>dEjQxAOv*Gcg|V50f$5#@{pHg01YH316R1xhzJz# z$9-DC%R&BvSlTaX^=gZ0dNrP9uLIN9hj08%3nM5fv*U85zryp49Zg}ld$EQC=a-Go z{g!~y-?cCkW<@huTv`;o@$g`*J8_!L6ldejVLH7;o0BqTnb$@>9HJSCE*?Yj>@iE3 z!ly?6^A}5*=Qczf1h;`Sf!<%aVL9-Kcs4csgsC0myGyJ(+R~j?L#+(pj9+88ddDUS z)&#SR5nX5+1T>6|PVU+NK;{t*#Q#2Ka~@=~hY@Lp^yS8ZyR3{6sS=0_>BK9n0{m14 zWW*owQwe|5qmzkNpm_C1k(kT={RioI+#Xk+HMWcK{MpO!zyQ$LpaXt$@((2h5De!> zG_#TnnC^PA&TECQ$u=K?zsMqMS>umWpLH-v7&haLtaYbUQK#Hy+aNCP(B@WSAUly zE-BWzS6?xl9{7*tS9Y{C3aO zUJ=Oukfp3tgUgjbTuBByWC}M~!9y&X7BbU4!XxPaU}ci=-Jde1nrDmMq%BE4Z7bRu zp&2w3MBqJmgoNjBT8t3KVEAMsx=%-JTtcb;mhbZ@=A++mTR zEDgv)pK5Wul_pM~d!8}AS*Q9Qi5O^A>iP^lTS2ix^5b?dK8U;swN|^KD!~!)64g!z-Fmx! zfOI6_omidKB14=nH)#C`#9%0ZlGY#`YT1Fge;S5FR>NQ2^C>kopX4)z^_vJT=K+U} zK|Of~w?86!SDcbC{n96ZCwgBL85Nbd1$WnaH71Y@u<%pUXK826VO{@V?|(*XIQX){ zx7VjL*K1%sb3LtOy4>J+N3D`|c?)co$njA5Z|t&INVQ`8_L%*w^dV>QtV8vMWZ%2U z|M~}N_CgI-U;9VY%k!WbJeeaHWbx5WnXd-e^o*9|HvR4(k9t1{7GfMZs=F9d*Xs#d zaVNCCxeyLwWQ*lBC27{1o{%HQYcGe%gy(!ZO>Gdfq)_ChOA;}NmLW{LWVJcIU>(ML zY)Qt%QCKq>{@TFY97&Ei#7Bqd$jo;KOk_r*VWNxYEFtA8lZqH^F=iBA&kkCumvs(o zTd@{r7@*~zo3~7@8Y^s4?`KB{8Xj9;Fqcf3L2KCmx;SltZ0T)l{d@Hxk87Z8TU_b3ti=6RDp>9{`(f`OnM8U=j?=%y-NThRZ%rQ6ca)X4wRco#kt^z_K>FSn^R zbjp15tsXohOsg)L9cyc3{|qg5QtBlru=)~g9NtBYcyIOJF5b@=`WBfA?F*L42KH3g zGuK28bt|i3A$?V#qn7Dr3`r9rT|^@=#wl&8#e(-`k#_#V0Q1=&6Wi-CM$n+?10O&* zYAX(e@cGBWH6u;k=Bf|1m1z)WB=Lm5*o9mR_*g_*|4|R+adiDZqTVtr%I|H%m1Y11 zh7QR=8U~OM=^k1@x{**iB&18aySqfXyBnmWySuyhPz#`)=Qe~=U@Ft5K#-QM14F9p@cZNqhrf{i)pl8L0-|5eT;bADq>D( z>b!q7jZq}cb8-xl1~llQ-|i*U)(M>f_TPhoBA>{cCXak!7u_0F{+d=#zB|0r66u3( zRP(z%R}jMl9;0fk^sAuG=cfmIIzn;0$5J4YUx&+l#L&7~fm(yi&C`}>1r9+_NtXXDG{3+r2-EZ7Iwo6R$k z;Tma*dSwo?D``%ZfRmZ=CAuLbRlo{5ofV`#Ribr(j@WoqNq;7x@Hp3iCr_qsp**i4 zq)7LC&bks}#9=7EaieD$toy*V67mP33>y?oK?0;k#-vs|<|Kgel1g zxYyF^bYpU(b`r$IC{bfwrNF=EZ4z}ny!LKI$5>oIup>bsyMlPAc;Lw~OGgys^qJRg z6E&V$>)lP1pX2VBaJ1jskB(A4Cj(vz{@2B%{P?e#!N$pH)Fa$l0&E7m1*$|U9FH{T z!h7(IjErc`vS>=3PL}N+T4+J4cbli4J;a%lUntz4wReYNf0pjkj@6c;cZI*6Xi*>_ zL+#r3Z~s+P{yPb7Y56r2A{OEdd>R97 zN2~k8He;#~Jmi zaD(DUMMYini{ogoy=61>bI4mi`HaO)&4AIHi}>Yv94kKAx>Hsyo43cm8NvRFTmQBN zE_B%kwfyM(19Zk+&4hzp&6Nm7-6)zth>3o4Z9!AJz(E13m}UX_QV?(d+=@D6jy+xk z!a(}hy=!R~r@DJi7QstCQA!Or&D9|OrdshIf9VO8S*hQxV)-mOQO>_mgwiKKU{W{_ z&%3kQWRlr%w>rwF^ggI!_hj9_{Bsnuw26woI6mZX0>k#;iysDKf^7!>Zh6lKXc1w~ zJPT9wTEHVM)r5GF$EweJjwM?_=I?PKI`%dV6!No(+@4C*R9rn9$pCaGQy966%UJ26 z1(z9#XfmK#t`9dDOYaN75(Dnq9cogYfaq>*Ym3GR_jXaQ-UhC$Aa6u5p~p^ml}1={QNWfl06u;+Wt)P0?NcgJ+m ztldpvB?0Dn0}#lFKzWe!#R&IVCpv3??LVA?-E#iD%VP@!bly-!m`T8j@$NOo`d5xF z#F7A7hnlvZcRB0Osaw+g6Z0;pHSr}rj)Q!kf#&-++ea4$%N3~kG)7s1o>;`9Klya6 zaUFy&P1j@0Y2PxO|NM`FB*^q>*MyfHNZKkIOv4Q;V-5*TYFM_!>71)Ft*Uxll&x=+ z=xMi&4h;#3DfVA+ze!6yz2N=xsW7&op(9_S*?^X1tX{-#q+0K+cS16R}WUpS_3NN>G z6ZPXBOV+{X&Bf1}9)fbWC&%!OB;L=Hs|bj{(-=qNM9Bx_z=x{ddJCM?nF6)j%e(UQX8m#%r_p%UI!!azzeZh;7Q*mPuWjtGDvqooCy z!PT!$xsC7~tgWpZvPV5&LfSey#w2;YC{+$@9)X> zrQkF0z-G~Ya0tj!A%5$`=2y_ zN;*^%7(OYvG+x91pir(tQpK-v-w}C;Ci~RmKGHh+OA^56(ftL?n^kIO=P{-k?yB<7 z+?rwAo4{%*^$$F9xz0kRwJv?clNi(Sc^<(q%%>6&Z=zs4^?&VeO7+<#;@aK{Ejm&j~FG9Qx&pRvl(NC@DQ#i8@teM$d99W5}um~K)tVkXe zXvOZ{#+I#k*40-dE=EEk-B787qsy)Kcv3$$lh*w5WLRLy5QJwW2<;wULN9lj5);PF zpmT#zI#SCPECX#U1_1kj16li|uJmW_1~{~u3WMK$$vVD$%$uHNs!lBGFu*)}BESNA zqG~BRyShXfF8!!siiY6b?Owbm;F4`E$uvDGn2$L~US2*LVzcI50dz|7*sQ(+iV6eN zbX~;LH^R7=+jqxvXbjYtKp5J8UNbz0hV3pkaJ1g@@mnw2U)a1LGBuSFzE4XnXvo$&@D7=`faAgS<$C(t1Ge!%s|4{0QC zdd>!=65O5j!G+?od3ec-=uTNMYN_l5ck+z18d@V8ZXO<2Z^9T&=rPz~gAUp)nM)O4 zZi+BGwcOtotq7hb?lCC5!LI%|-9a%=0>2nF*8msI#q9kj6TuMP!0L%+nA2Vp8`q%3 zzrSr>xbI~0jBz9cGK+VAZAmgX`zm;-;k%ud-7Uz$?13No`hYIZ^;wm*<#sU)=Cd1- z6dWqD7jeq(^*Ka>l5&&o`RJlp@khwOEJ`nZ`H2om!3KRJ2_9E8R^8kD$(|~h&s{&Y z&>;Q8a_a2&sla5T%G0aeR3$AeUu#6+%+v2O!r<^_c$j+4x6idr*uwr=F6j;5MsMMx z5gF!Hnm^P+E%$-r4ul@4%=LEI%AxT-mf-vb@X2DpEksZDN*IZ>NG(*uV3#EiZ$OidCpmD$81sMP7Ls;ait>_@$5U==8 z*9BibC`rK^(H+M%K|VXXYcsL$|NXgl1s?nSD{%PfVF?SOhHziwoR%UozcK#m!|R_7H+ zxp_rujc#;~`rqE7t|MDaMLlnQABtUUXX-zg$R+iRxB~?9DPjK{(WK=cG$vz5VfBD% zz%DLX`n?*wMvL=$cMy-`VIl-O*j!lm3C|Jez3%ZYPH$e$mrKE))!EVYMOy&2TVK+D zH2N%7|FUCAM1)Dd}6?f}VaXH^Iih;{vb8l+3A)HC6DbLYpXS5q+ZTl~D@`OPad>3KG$6iM!F= z$CmSR^mX*pKN%20zEy`G%%!QC)Xdlr!0qs`>gd70>^fsoa$9#Z$}T$br&AgntzhEVh#YjUMurw9Z65y{Hu7TI9NQXTfKLq)v$ z>D1lK^A3b;shSS)-aw7{Xfy7;g2VWiL*92KN2JpV<@TfPqoFM5<(xEC+0D%j0&B)Q zbuv2TD@&NB&)>;2XMXrX!AsC+h4;S&rYn&G*OyAs_{swZ2Db3s43mP#`=UuK^(Hn` z7wxHALp>0O?S*0CF1grUdn!k1)j5<(+#4b$EmHI=}NxX}gklwq4gS zQyp5!a^PR;CWBwk`HK7YjU>i0BTo9?PO}tMoMh@Kt9I}b7gT*K&N-l-R}8W>vDJki z{EsDm^5$V(-dMArp64G+;AX9sB?`i`^1UztP&JDo{2OUw!HE}&Ww$3|w@cx#S+`|T zH*pY$iCd2{0HE4HeUx}u7>r#HeXLF$mj<2 z0$-5xmQqZ<#2_Z|7nOhE9>hXBoYHKM3W#*uB(hjmCaq!#6^?$sH#Ec#!IPNeWJ@aP zpU|i{^n>G+SdbC{kfPl>)sXH>pN6E<2IKJxV0NN02-t+G5PU%lF4|&dE3N$hsvThn zC7LS3M1n$X9=&v9%z%hbYAPITP=3nLq{3^mn{WcMwcOGhWM-yV*dkwc0*y|{>h1g!4lFJMRP z>=>oEvXQA1YV%nzVU#ta4FUh_k1C$uC(S69WdeX?1%64Ayz_kd0qXfEaH;1Gdg?cT zMv$S|=QKj=?%?vbSA?6aC?_zFd|E9z5?h%t?HeKtJj&izPqq0`GIH$Le?`G_U^s#} zyOk!!1sh=y1er}ydf$4sKi{Rh8_`{Bn542_hPnf}{7dtl%Z`)F|uo;X(Ww3x7256mtuT#=q z09?6j-hlQ4lfIaE>@nOhqi(S(GyS5ZQb&;QNzz#%kkOyH(!!O37e0Usf-g(R4U8Eq zkymlT8U8q$i1lm*aje70B1vR6F>78lxWCs?H{X^66Q<~wF!Z;_dRH=VOv!NQ%^2k8 z)EO9Xp~vwd9EEZTgnU_B32)z&h_S`0^vj1Xp${4DE;>st{Ey*4*eKq=^3%Y1;I4aL7_ zQroNiTlyVg&jT0d2gcKndDSrbz8jWU@u|wP1KnsuWnv~WOYFuMTiN-m55KIx{rsgh zzhNfU;c=3vg8r zo$&dIWE$*~3y{9DTSJ|v3ask)S3esD3!pQ<(cDw=`*|yIl07^-PF-Bo1)d4<*>p){ z;=F3@X%oNCm<}klC*wm9-QE>i#Prk0zh$_5nWBlAR?vi)x#YXf#4#FTl>0tNVZd$) z>pd7(uTnoc$%Ek=+gNlO|F)vA-$@9goo@Z_IJ3|n0vStd-$zK~%aX&2A;&Xl<^Iu` z%yP{^+nCgmsNI6q>Fz)e4YGvbEsoycHbfkCn}@W#t*el`&SkxKI#giARX5J)=yJq( zcxFZWVBa)kUU4(>T(2WZleFM&X=z6;xrUZMr{!Vg;*lo$v0?PtWlQg&X5gSEBg9<> zd&hvJ@1P9?u$lAqwni}c-KrsnpO;&Cab86zcgsM8IeEL`o8%;)+Ku{1Y_X36zhNY$ z8pyusEgl~5`#6$mI+|`r6RZlBe=j{|WB^YqmOQ=ZH1fIrWgo3WAwMKc#Dv3yc5uNs zumQv}o-Kr36GFV-CYrDS+9!`H`MZUblE9g0xL`g7I ze_1)RKGFN4f6@FE0ct-XxI5YM_Fecs=q7}u!f8krWB-N((Du9vqj+WO6Iw0+U(6Zc zd>H?9MeuZWa)Dl3lIC-{t2~pQ0SsVJWez1|mtWp8SegACOqtK@ywAVgQ`94ks5i5Y z;*RFeK{~{D>xmso zz#2ZxkU{x$OTRlJu?sdbwyj;%f5lV6Pudkn$l3V)jK72(u_Pf>d~F?EB8wqKTW=<8 z{)R($+%oP0NyiJdw_r{3&o@k4p^wv&Y6;X2oja3hU`V zira5BgGdLYg#T~-A%%(De=fqH4M=OBaOpDs(fL970dUs{vRQVnsPj4z!fQMAr|z03 zgmwVA52&4u8W0fCY89wAD#eh^Z94L|6^`{V@J$Vg z`^MMp<*=kfMyE>7-9Cil`B6}ACO!_UIKB&V1cFwUej=ACq>kG6Dy)fYgysq16 zeQC38Qf|k8)>Cz$;&f_!;VUc4BOI2kQ-fD4eh?eZWE}WOG%&`y+`V{J|K`fAvBlxv zE=TX{?3EZS^iS*8;e`w@UfN@$!n@Z* z8Vav}DV`D~L~?KmGdvomY!`E9&Q+;4l6D+{Wd%P6luqn-{#^d3Z4W zCRobjSV9{bDK;bTwk6_K9{>xF>w$xXij%OwN3JT91~rIdSvCmi)Bb@hK$b0XB0a;A z_~c#ETIwnMEZ=540LN-331HZP7{^uHPLQIxJ|tJ7=;3bke(8Y#zuwwQPa0L}>g!n2 zW&itcSTS*4I8h1E)m@XnsaWAZjO6B2{n}=}Mddf>QmTBa7uqRK6n(&+s$*4RO%?w+ zglES4h;d==^SkXFU)DO_Ypn${a;_MqQVK)Y9ij1@T$uY*kMJ26vUa7GRiuWJq)gj& z`Q!vmU;!Pt`tZck!O2}Q^lLYv5j`*WC$}LcgMgGtVrdmdB!n&=k!Ch!gOWgTH@N#r z`)a3In45jNFVo+9GM8||$` zqCQn9dt(T7Mn>s=m~HP$ARu4AUL07omCy9&M~n$1d53ydQDUHP>+Vgo(305`}NkZb)R*I}t@E_a_K{|1Wwi<#ke$s!%+ER(LqAj{YHS_-~u zeJr?Lh}Gw-kVZwqlv0CF3^}1x|1b?{_ueQ9sDGtq%=~bo7$14@w zZ|Zrj=_}1H^_qt;zv;s8zh^j6wH|m5Iej9^XZDHdo;dMp(o`1$TJfgA>H2o9Ft81qkyM1iTu$KaRC926`l{L^fJ*y)TZZu9A`5l?gU#wS2`q znur2Z1cn%=`FV*fg-n0@-L?7F3B?!@`Y0RUu;DS^u-W4(4=VA*vrl`)Yao}!(3LBY z4fzou_E9uCvM>8Xe9y=iB_=v*0cl$d(p@5R$O#eOi0o$zg|&)sTV{;U&+ru$+WV&;>3ZxTB`ULS zoDWQ$KRuk_^_|~!u^FzCuh(NGU8HVi8{V`+Hi z5n9rSFWg}DyG4vkIOA0!ozUHAvB)pwo;2+ZbkT@DrJo0bacOCjCr>ze*d}U1R_GBf ztkll9MZckef8O|H%^!%2kA^^E#lDTGMR}N2g+Xo`IwxA=!yubhT2&C)@4?i+?$iR& z(d}}PX8;|tR&SoVQ}x^m3Lf|Sp8o^p@Q^-bqJv=wdXs~SE~55dzuy|#+pB1fMHo)CM@$(kRV&{C znVSy;G3r%5U+t0%f{kWx&o+$pAc)$zhVHMC6DECx2q^Pwx3u58-P*Va&_pKtd9xXX zCgjwlglMORphhG2DgvTRy;ik$!VnS-igTa;X6FlvQa*|%#%s9~CxODNpAJ`eWf*4X zaH=}q(%vxp?BtD~FMEX?)#8VAhw<+>HmP^vj$!4gx z9zN7NyT#3JhoEM-PxR0&L}XB*3hn46XNM>@&y4~X;oet6y&bW3uttt?{~$lF;;VJZA+r)@m~`=aZh}uu0?E1OTH;IBJM^ns^A!TgI7hx)eP#N8wyeICXu$XKT2t;VU}8J!7ei1R)aYnb zqE9ncQ%FMv3i@P3fS>Cai%NZJjM}zfmE~D&O)}!ceZQDBtFhuNRB>edvthHzb=&*m zeVzwJHfMn|rgd^xyaAeJ(e1LdD7J6!*=a+4YhgxazuZ7ahjaajJWA0>N3pQ}M6NeP zM;?B)7D#3R7(&~dpP(3rC^{v{(3a~rvSny4Ef5FmHk7A7I_x;ARzMV-<{c$>{e(hZ zI@-sjkowhJbuiE#<>EJO9E~{rH>>ic3sbP)BMC_cF34Z0>}ReYQQ<#rJi)NDhbDUB zUEl)@9BXgS1c7B011oC~I?RXK>`J%6E>AYXl~EDqgNElsf(G+q%@2Rjdh?1VxGjjW zKy^X&@B_Va)O3A6o-;B&d*-+tC5Tl3GoyH+k{N5|uL&`ClV`=;!H;Nt+y$cd!&9!B z$aIe6J8duD0_d{LxC_E#2FU=#;dQpTqSvW1LKp-9c7%EZWKE3IQ1aekEwm+Y>%a}zTw3kd6ItiQ^J*oJ9@q=Bl%3+|chtt!C zx@jlmq({E_9OxO+S`%so?qG`?CahxzPfIpqP3A>SII|bPzO%bc(t3*cwAn8Vv;qIV zWTj@~1@S8capgFkH(HUnluKu zh(ar^XutVMI`6fZ6N#HM-j-XG7>DRUXFNz?HAC&z1F~79ga#FJZ%YFpAcNWpA(wY( z#Zu7}y(J<(g~QGkrwfb&odqOXt2s3*Ig5=Fg%nZ6rxo z$}aR8&4xXi;*UzrF2ChD-6!Fp?j>6>9HqD(NQitMJ{a>*flF4kjB0Sees90i4mCk! z=Ua?CCie>6RWz%$5W4?M-#=SVT=a4&4oI8E2w0IWWXig3_qCwSmkgJz%{~sj?;&{| zW~~Ar5Y(3OoTz*8Hjxpc_YAu$GVtMgiQMk@3X~`d7U*H{1n*IMqJ^T{d){cK-KAfI zZ7jCC`0nJ0d+Q4e)aEo{BP7JZyJ&|}eWIwyJawz~i=l&Bd7o*HBy)FBReXrv`qz5>OS6HLQXZ%y)7gTZQ9i3UCkA?z@U#7Qks?{;^8 z>WLGM1LL~b;qLVJk6FL3NkWk!q$@$QZ))^^-xDhySZ12QDDx00v7{x!ob&mDNCQ;2Yo)X&hei?PMu!diWPT=t6`*B)4GOe~5Q43Rp9$Dc~^zTebT-t!ZttWjIHJSL%#3V0x?(P|?N|;8bNR z-T$tEI^Rl^F8JL`d9L!*qX{=#x;fzQIyu$FlU={r6!KT}9qN%iGzSf0#TjnUx6%-h zJcWvzZ2nJETEHno>Eg=p|hW7@P(UF@T@{jHu<&82nuoWV2HwNHT8$7+kuW z0oPOeIhKvr>7K_#)z%j+Dz|x`+c*99HeMuvRwRoV=F+{>KY;WwI4M5aO^@T@M{Akh z+w=saxr+NvnaFT_x&6V|P}P={+jRKco;)B(Fo+uljNh0KFZTC19i4K%O)Lf45}o-h z4yOnx9+#=+!TYpaOVt&TLe?IU8hpBjd1RF(6R5ZIOnQTXGD@-ng^WK7;+x3&SR`uZ zyB`M~oI)k__f?YPG`!U)e@ienfiS zmx#fp*DVbsUO^eS67&nQMboPy^PzyR@kbTWuVRn&%QN|!*}KsH`%D1dwMKk6h%~x9Ax@?hg=eNgZITA;y0MT z$A~bb!cfJ7Gd?P%u~KgQwaqoyzgh}2lhYskL^kx#h>34Cv`Py@{-x+QlaaZ1d3S}Pe&Mww>p7Qi1VH z(zq#P?nAAUx5|E3^UeB@f?(9sc@?`6_#vGwi~K(5+TGPL(z_QT(wHkV5e$vd2v%RWIA=~{kvAUP8)yC!8qkyoJc_g@>F&cLSi+=G#eO|tt@b96i6C+ZT`89CzOp- zYN&)`NNWL(RMZS;$ih%7Sei*YL~Tz_gq|)`geoR24fgLCSgdF^}sGytBU8LXt zJZ|mGR@~1`1+Id5LCF3#P}Xy3f7tEz_K~QYzxefKfXQF=-CuwGYL)QJia-12OE7;? zDf8Z(wlqL})0~6;rVc*ZTdP8Mh57$KWg5?9+I9qb>*RJ${c_=l*x@a&Be+i>?Wv6( z$-)L}FK>2S^B-hGc)#-tXR5m!V|4sOdMe4=Xm$JwDL6{Kzj$r>OrwC|O_c5{J|pZV zJxN|FsAW{|PF6B7Zop0S91PEL>@MUWkm2tb^N9rB2plf(k^rH#>;>`?8kbYs`FE3f z=koxJ&Lr`X6@WY#CNqCl{PFJntc!9Zc;4T&JHjNrp1dnUQzjY6)e-LxiW(Z$&T zAkOH@7k4ew0drF?E}h@(NF8*~e@Fm=k^!*7t)f<{)NK>c0UFzA{nJPDgs`|B#tRP+ zV*L^BCPe{_@PL%qiz4#&Ij35`22Yp<&F+Cq@#?QPRY&Dz3IY{TR3?&!5*V#J!GE7# zfsv92?(XP-Q}yMy=x1zZS0 z$MXH57{2eA#br;q>?qW3>`G37>X`Mm4yts9nDJmlLDtk_9rTCp#45f1<-UxcY7uRj zm{%h4TyoWwRS0mcTpGjB-S1=Xda3Q7cyry9um>V=MbC2-C?AW!QuEbsl4a9K_5SW~ zdj5YOq+-zwk0^EQ1Jf||qp_ftZzBp;=$YX# zBA;0Gjb)L;qHM-SW0N3eKY`%z569O(AK+5ezv7!pVB%v_RW#y!$1Sv(oajo%8J?!3 zy>vyzH|9j&>iWXg2`LqCR=EWKi?V|PQHH6=_*!~=LJP4FOEo|zdpOwkyRWM0i+s8# zJ%kvdIDUDZVJLMAqQcxhL7G`=z@&8*U7yRk>&r&)&pF~0daJ@UhTM~5wgk7y;xE;SpqvK6c} z>wbqJJ%3`|FV6}NB7V=c*RJz}e3kB!DlnQ&bk#Kfe%<>amO!P6JgZA9r63JBR*w|# z^}l~u9SemMRDeLuP;1O}1~%vA#AV;Fd!8Y_XZt4ryN%r+?)9l4l|1N@Nv*{A4aH)Bu} z;UmZHx_>}RKUiiF@Vpe`$6 zd&t~~Ah7y>2fGHC50&;O0#CdxwbnZ8p)-M->&0`u`K_9jPqADL3%sqPPrPNd;p_FR zZNQ)OCvWRw5P52yaPwaGctfDjhEF*Xb3GIFXn)S{RE~+ph8@0xKbGyZgF6}@1x)UW1sZ8_5Th@o66KW4j;q@sKO39+L9$?vgTqwr(|4)F= zaLfg$H-9Z^0KQe=QmZRL&m$>KB76`rWF``#Nc`T6LMI||gqDd()}&_j7Z)wAapziY z&wxMueqvYh-NW9u?v1Y>6wg0B+}{uWmSlk4o+!rgfcbf;$tbBL}<@yD@bmh_UQj>8AlaKw{zQRYoTV|uTY$W*@_&}f~M~Ws&u|T(GdW`D}WNgY#4v{(&UR-sML^3 zH-#8$-B$+_jegNIRB|k!0%bW6cn?s2*Gn&p9$A4ww{&IOZsx!t;0i@_d7y`o{ z-Bce!nQsB{U|x?GBwM9OsZv&;ig=vnXI~{h$18gMUdVQX3GqijWyK5r)sLxiPo$8G z^#klKnce<0tWkDong(V_VhAR~XVnco_Wa&AH)#qr+J8Ts@ey+ecvIV-5jLM+A53Rm zSf^zU40rw};y9Bl;Z<+_QY1UV=lItwl#C7hx$O~58$sT8Td}d~HU(0qRp_Y9<%Yd z_TPgoHRO_>&m3qBY2o zF%n_r`su*;?N9c~io3yk!rZ*919BZm&~1ydsVp%!+0*@F7@!sfO+XPcK0m8O2X2KV zWNIapqMl>6rzTaM_K|z`U(PTwNkpH@EI95v$6x~g8j3!+O2;i3xnKzg>XbwN^UwF5 z)jJ>3VW*B$q*4)Ws7Sw5D*?MhDf-W0+m7KYU7lmyu*6NKukMPr8UN(^5&)X)k6x%! z6sv@*qZ~c(!m&HP4qwvWW{qwBQ3}qj@ z;`8%!v~yK$s0h5&qK0LB=4XS={;0Vx8R&4~rI5ukP5fZwcov;tn4xS<;EKh{a*KMi z88+V%|LMVKn||N*5#d;yphN_je?!$y<~XL{YMEeDmTR3Adk<&Ho3_VC4*3jTf36bE zoH7c0qP@NBBa6LShs)hhbK#}tX&y&(y(Hhe2hg3oAc3sOJqE|z^sBGM>L}a#9e(p+ zg&LqQvB_e2VPELufzliAaEmn57W(>=(uwe53+zh8*uYX?d`befO!I z^g8Kq^D*+bs!`~Q?VS7|!sbJon^Iu9VTW#zZ-K3(-Vcl6DYM2^3tavC{3fc8enl3JTI|ySK*J zzIntL_S@du`?_+z>SL>|9ZC7=^71?y4_Fn3EvsiM)oDCQ8iVcl-tk;35i<6&B^X_8 zGp88$7$DQ?H9CNR#BdJ~%Lk0zzTE9t5A(iA3PGB3YR9@d{+CuS@BZ!%HG!`wLc68*}3O_ zP9tgJ{~UJgb*LDflf**G;r+Vor2qYMScIbX=V%)>8f7dF(dI|cuGPQyQ`;q%vFlIT zM`I`no)2I+%>g(f_Vy#7U&PK80Mb%z)L}`*u?WZ*kybEaVPPQB>Qya<%W74OTGH<; z0j|%rqaP1!W68^Unp3I%nAS$)T`F~A5!BU-#mpYM|1CY(JX1ss=tR52cmm1+*f}e} zqTQzQL6k}mr_c6z%r6u0<3M}HTzFg)`ZqaLRMd7r8=vkT{*Rcs{$9{-VaA8)Tl%@M zex}|X;%&Te#b!ulVq3s?lZ8s-o57lKeu>Y9dgqsK&`WF$^m40FrTBZXyQMRll-P;3 zZwY;I4p_aU)U#MP_Rb7>mbymm=snW)q0ly0AmWw8@}D9*JynS9aaCEAq)6gIGv*wU zSyY3dxhz@#S`@%uvCU6$_Mq-@vL32YDE3yu`@ouolnJbLga91z`(3lmng~(KSoK%s zB;L3=C=i18<}{z~|JDuxL*QMm@=(!>`RSy_G@yT|af*H36K=(Oa}}J^g3|uvtvzGUq+b@QD=llUm@XPzTd^~9_9)k#mZDVfuP~f@ND`N&L=M;gH3TGy zP3w#d;RTwIkYe79{pwdo&HvLJlNsmpCExYt_?j8muYFt;h}M^`lk09!J2|EuxWvlf zu($VQf_MKSERNO8qB!u$!c6Y_x?wxUm4If}7;toGY)!Q(= zib-fRNaT^X6<=TJs7fo>58>`y=jr&BU;cn@NX3x{r)jHT_ULz8Y$qw^RyY5Ogx4ZIn$`o#X(GK8DiY4 zQp08XtPcWIR3Z$@buzmrI%C@Fp{%pSzq+sp*l1BP?Bx3txXU3bf{B5cj@<^?rtduX zQ-T9jdsZD%b+52ks4`u3bZ-oF?HJ>R?^4UNxnmIN2KoKF)7-I{R0?rfbD><`MCPO& zI|cyD@!_5#%X9?^X-OA1TIeQradju#*w>1Mjg4*Mm-F?Z#@*tWaqu*%$z&E@Gz*8U zW4opF>qw)%@PSbgpRS|BfDI%a(mgKGA)2KKGx{vX+zrn1@K>WFgPYY345M@rIW#{@ zHLB&yF+JXRkRy&dJwLj*q^`&3h=^MOA4)q+T@(9UvGjysZD&$)LGYB*EX;(IkHG&E zI-doL#DzB0q=K;*y~nc?Oc|nG?o{ycQb2?f0}InhKU5Ar+u$rjq`k*xSaU0=!#<5B zi;!(pBNE+Im|YvH(!Aori$&~v)B$uG_LO+xtXu_`D8=qr`QQk|p&)-|r%1L!cM^Gi zy3vD~NXc3yjFkij$Ai1wsiC{|$7>KK%&}FQYF$n2Z=|w5lD`z4yIbY5G=HN{8(K!< zdwhMTvKZsFr#pk&q_nVcw$PxGsFJo->SN38P-J8voR7yGI98TYeCM(h-~w`Qx9+(~ z@z_Q{dDAV!+7er`+IG7hqC~Gc29(N?&V|=n%o(m2AuJ)sxeuNJRdjLUwP?YByPM(k zcT@IXRRcR;0MOO+FzNj;&KY$q>K0$m({Z4$6Sp^6umNDDaVJR+8PhnLEJ+ecPf@2f zzx5Y!B8<5>dkroDS`JJTZZQ12*Etl37q0`rRRY9kh0QYo;8by1TyNsL{fd%dg}3gf zJRT^AF;?1q$MU@zg>;R;veJR&dLt!rEdA>aru$j#g_ROk@8N|SDBXAXdf2cVL!vaX zRh#a*so8cvkvyF2y65_6nWt@(L^FXx-Hn>&L#8?cmBBfhp4gUx9{?nTm)Ww6eS;?>1fQL?jDfn8 zImUYK&N*iUX>{?nihx=DPCt} zNn=TtzcV|=zx)N~b{000E^#Ufe_<2WY$+UOZ<-nL3uY>gnPg@ydw|xRwnE?ec3>>HOzz5JW@-Bccm2@ibNVaJa?ML(_jiFvln#^#)_%;v{GA!!9_Fbm#`MUo8~N ztfd-_-79|DD1~@`bHW<2;rQ;;syRM!j&UCFs+B8%B2Gz8oOSSX>0VnQ3_;bL`)5N(jp^|x1{WLf3N5*;Ewt0c%RT3FyxWXe-R3LjP! zDv@eX3#pfA00Pu5D#byKm~V&SIs@H%++m~rD{yGWwNmWJ{0L9G_mSrCMVD$$zbDF! zKfHIhLQpzjxE5cn4^>3|H~Y14L)tm_@=Y8fiaz)Dq=+h{4i{ZJbVyjP5MN>sKfk#L$Q!bFNd~41f2|4AD_&S*%wm3Q} z?w`IF?)NM)-_B3hYQqq9)sZi_H(!tzIZMka3rp}qZ_)gb{myRhn{9){BqwTOdnxJ z)x42F^G}ButPx{&qxJx;r>McmK0R#SQR~zJB0FHj96g>;AHLlsUq8goO={$i!lYPC z^FPHpnZiH1Q;9_)stNOHvzdi+#2=kYh@6keJu#@XXXAX~6_sjan4@aNES9c3LAuw5 z5kM&Bo{xW|hI6kLKLsTo#rPY;5#Pt`vXNh+AbdBA_DY{j4>51(AqIfb;het&9Lh`~Y&e00@10A6b3;$Bb9Z-} zz8TgnMhSWx{!OS5Ed<|j5h-RvvM_nne`b$|m*@kZxzvOJl#)`rKSFcg>;_dEJVmnE z!&go-`K|%|kQ3-+C^_pt7TKbOdR)z3{+uJ{4yR3%w|zGY?IUx=wbc468918(c6vjN z{5|S{++)`OZ1SC?S5X9^6`Qs^6iDtR8zeA_odcopfD0SJhjHl7@n>R$5 ztFT*ZqHKNrEbg|}>a*LGvCD5wUVJU?KSB>&uTQ_`Mf{_>5_@Wp9^Q(?{N$`Ev;6b- z84A&>#6}Y4=|yADIeaMkj)3^%@J|B7Pr}sj8ad-IbHm^w6>6Wv0ndVn+9S9~=1V~) zfK31JGo*|T>+7*W|9Lo80lMF~78YU!=5zMz1!QrGKg<{*+|CB4e7$t_hJ_Sp%BoG2 zP^OnvK|H5ds*IsOa4t2FcR3{p{dU=3E8Wf)a(7uU$QHA%Ykq-NQxs~X|JuaYDWpKZ zbXQxtKE^kD%|n)xKonx-Cp!Lv>3XzNdxeQK;V#JHoy=NV~fqV0+hZ| zE_|Y-h23r?=xqkKMF+nI650Mg)VeHGZkk!@z_w?FV^yJ{cirGLL`E-ZC&iB|%SLm*3}JKfeXN zZWVExiU2A@`PSP?18<4(ux_Sk1ItwLVMi%c?w3Nl;pBc(W(e%pOJOXeT9nqu z33@+-51k~dUxb^l_hr^4+(k++5h!_FzYyuV+#&s|GLe6)ZD^06bf1~n40TlJo*U(^ zY4c-QYaW^EUkwWPaj<#F^k^HfWP=Mh{TzqpEX&tlRq)7oPcR6ExA~+kc~-X~R1|H! z7EOU|xZ55Ia{sh-Vds2bM0@sC+}3y8hZ5C?Z3}H4q;$4uuaKi&(U;o;CxQ8JrL6#V z6IEvOAmYGjkTUid;3d%JU-QRc6~=<6rV%bj6i4qRd2J*FiC%)oa-734P;16Mz(o{Q zj9s+GJNg|B5u?T3G|%h`KlFQb<|nH8{N<1&R1~+ zNHAxVzH26|!PcD`0mN~sU_Srgt-*eeUd)p>fTv?2xNWx0zUzI%6QBhUb(^&X!dhCb zrg}SXSP5nzv92X46t5)DjD}7u+(UOM|Ez@pm$tn?KKQjkE63=ZSB7v zwU=CVc}DEiamxas2K-?qHEBatf27Xk>;PkVG7#rF2NQ(hbt18%B#Xf=CERgOqfGLAR73C7tj2zCXYpu$`TAKF@RC z*R_Sm>INX1UK@==$_8v1_pD*2zBk|QKGLI;Gj0Tm!r(WztXt+)$S}wOe+Ky^M@sXbME{FN`+3%6duNY`6p?hCme0aZ*oAWhxR3h^R z0~ezDq0o|xXf0iqZ$4*@HzEcu@bUhik$oiNMsZX41AlGAoxx{w?lr*&_8O9)udXZN zxn~*S13@nmP}ewJGTS*dGDn-MH75Rvhw$PR*{6zpj&RkMOn&`>;#||S&xE*>Wyo1O zib9#NnB;L$J~cvZF1AO+LbX8_7Xib!Na)`5k}L z41WfNBUypE4PWYo9FO$QYxlI~V zs}7cj$oy|lIW2>q4cW9UPD<;~ztsD~lp`OJI{U#)3-Pi_0Q|>L=}*z>u&Fq?F_j^c z(}YghR8Cl#ZxXRZ4Kw>PnAN1fe~Rxb*OnUe`Q)L+IOc_g^($&Q z>)O^2^{ZM|O~03cb@o_q|BRIO7ZbaiE}J_RRjb4B3l>=KX3|Pt`PPJpb9sL{7G!sC z?dDLt?IA!8Y`{y)y#@JA{%CA0hc~H?kmZM&DW>o=JEPXT`$@7Z%>YLLZ3&}pXD;)% z`vILN_g_SV9YN9eOBCf9>i)Zb5oze3Q%MueMi((J869!W_ydz9e2-6~?sA_wE-PT5 zU`I^~L5(NnhKVBS7bgfe5L~txk-fj^P^v6iuQ~eA9dMM|9ghQeu=D(Gtj=`&W>^KO zXVf;($=4Vy;dOsp*3t%-c5Xd+qvm>?tmuz5^~Cb&pAjE<2>X?kGN>j@WSnBpNXLmX zhpc@cGfRWMi$0I!qRtP-7zwNd2|LGLOpO(>Op3c#Lbc&g@-^!CCcZ-{;7z*c3iY$^*?3YH)KD`P zFt8Aaw%QPcIIS67UQ}HNS}r-uwek)>1kh#|Vq>f_4u@Fp@l)+jjh9Y5F_nZ~Qqj2c z&E?p#f}+vy{IWz0oN<-2XS%Lkyy@-T5~JC#++Gi(*J@#2hO$1-A*dTH$jnPp=1ewM zmFwPT*b^0aMx%mJ^i=)wy`N@x7a`db=)G!gL_yaA$0JoMK>C08xUW|Pyf8<4Z0GTa zzwaQkdS}o+D~jOUNu*eRjG+rLDmT^dv9!S4;TTHS;3J6OJmZMRtY%@FRVGjla_R8t zag|g3GmY1d7HFBA6s=zsuFtkQ?yM+pdG4vQINddDDs6Gpq0dKOP<7ww5DiF6TiBIu zpmbgm!$M{|dj!hlLiU>+iLHS2jz_sZJjUbV2j%|L6TqXg9-nsIELy*ZBYu%-TO(yU zow^d4EoHEpr_!n()Sj?fmX;HM(2%jmq5S`QKTp-OLr84ZZF5zi+J1kgtI+oOAMZq@ zyf-n4%0(m)EgvTR46J^Gk2Bw5ktf$g>YqPop>Tsx()M#2ekpLAHOw}c4PrOA{|QT( z4n-9C5s#KR(c={`@L@(Xsiz|JvSs&z6|4x7k?h)bC!UrtaQ`J4)e65F^;F**7|7d> zvik>_bnlG=Halq@j=f84BWG|_FKfZvI@b~NDc_o9Q~yUnwd|52b#0vyza=_0(gzKw zH_LS)e#}zWvq-TadM0fqv@C^un~h*eQV>IB9h7@>RZDsJr{i^8vs`xRx|aZHZpq)3 zC^YzVM`vpdki&V^29xsnpp#210kjY$%Ci!(sA+Fwu7LR^Dnod{~#X4v2iM#AE;SV}`OAOsH*D_wA_awBp$fgu>0gzU|@yYxLr01UmStomvgvMcn4Z_h5R}!dZlEA zVb}^BsyK(Jl=f_m|Xash#SwARcqk`9|@Ta!e|T;bwTThJef_H8XH&e{$XX79Y0893{5IRHNHG1Bs`$)yi=VMF7&vun+p?3p+Y6nf)-B@Oa|C@KheGLL20 zba4QeuDV?uc(YA~P3`A)mvQ$2H@A+>kucE{Usa>q^(>cKnXm(PY9WD=?%WS=-y3ooP_RcDlm$!JG!@P_;dhzFR7wu z&$zOD{{h#`0&J4r^J2>B41)c=Xd2eevk_e z{jWp>)VeKyZfc+SErc7IhiFR|0u+vbVjWGxvgP=WL8U=7{6iPx=9kEZZ~R`&FRb>5V73BXAZakvO=^xd^vre3sV#pfE+j`b%tyV| ztX`UJZrtmv2`Z68UktHe0<4w4EJaH&{UO1I9&v%!y z>Dk}iNXu5FZw+XD7cds2?+yixql}b@!*Uii3e2n|PtXM_zJ~V?lBG@82{*=EE>bO1 z{J*7UA4bV%8HSfyF-E+lj6E1Q%`tEt6^&0d`rxh07_hhSr2iR zMI0A}yC>J(U2&8cf*z-Fy}1S)&AT>h(kD z@Erc>^84W_A1V0sAI*7g(mn>S+BRx@amScM(h1*e+i{IsHEKAMG%=o@B0?fwrs{ zXNjLi>!SNZg;fqMg#g0u6^|-uB? zWw>r8y0yP$6g?Y!EvvswJwE&`E|+t(lkl~EQQy$G07fO|1AW9AjiH2bT6VyB0)n^Q zpenPfZM;jtDGY2m;Ux3++|l=^9V0NG&4ax(GRvh1CH>D%`tiriQ0GT-!oZi7O?#H& zH;fuX3BBHB(DS?km7T-M*`__MzBE{!z(p)E6AWZ|$z?)_ZP0Gt>R?0sMz1Xz1z=D_M*sX!ih=UM@O&9zUhY8b9 zN$uK0@tUUpMUBf5E@?Z+=kGF3lT!%K0sH+$pi^6e8~A}{t%b-U=n^#AX^0we>jTCK z)z2U=3r86nB+aw#C$qT~6O5R7<<=E{+*QL9r`(pZmVQCn7kt^vL7K322Z589n+Qp# zkH#qN@?@3*_M&0%C<$&e`<&i(!ddf=>e+YhQt-O?P@XOjv}GJ$;ja&VbV4_|)@>gp z)Sl41ys3u=K?)82R^$VUoMD<*rtSB2DxxP^rUOT@?Ye6i@5dFFd}aVPVsAF|O?}Dl zDzPEy@ViG@bc?vb1%209 zd0E2}$8K0I&PCtb7iELC^y~8IsFN!6f^g2LBUNK|lh?WK3tPOTqaB%)LagxAHTHo& zUNrDj&V$+QT}xrOWqxWB=V0mSmCy&LBJTqmiI8K`nMZLXeKT|1ZVk|ufq{#V=-@Fbe-9{e2rYO=+UUF0_x+AVZTLwf?=!GI+{dO95%Lg2=@!&%0 zGr{m@@If?+bXkm4Kz4@?!-nHXusjJE$hHf|!t<&wE7M9ePqd9I!f>S5{<)b>TiFi2 zuV?2h;ePU@%f!Rv`cC%_U4EgkhA4gxn1=-f=eP9!o3&qPiH>Jq0&Uy`5J0R-l@Ln; zHZHp*w!rCSNEeBboA%n!?rHNM0A)~or<`yRqre~g3!Oq`JwA6VzuqUC{MX-;p0?5g z1T+#yfQa3O5S0gW>z+8&_H86Kl8C_xrWl}TGholFVRjThqQ25FGFY@QX~dl|QFhK~ zsWYVsz7EEc(ozN$4!yu;qV~V|I+)!_49<(18p?~ep|6Gy; z)nW@zZYjxRHCD|oetWh&7o~H3`Ren+@?eft%r5co^}VB;l5S5=8L~k%i5*kU)NFD% z(Z3`Xr5MZN>p&WOGMv!zmuD_6ImNLa*cuU(b9XmX-oX7OtSABg?g!GE#JelG!+^xN02fW6+h`3pYwk%bI9d$S7uWa@sP`}T^04_cN6JhBG)~cC^TnV7zS1O4)t2H z{f2eNhNmA!;gZ{8m^`kMnn3n;tFB9{>cZJYDCU>iyd3t`%`)UXKepm)Z*Z8kH&(~U zzITNB8eq!IZQ50R|- zABc4H0mcEn)^S%wbs<3#5A72F(@lZNn5eUPVG(#yV4A#p=AXq*hnw4Q7Ld-zq6ISW zk`u$d`bbq$gGH_RV#>%$VXCurYGAjG#3h6I?n2h%_-Gq-ibv32|CQ0pAXU^2NJdy? zl_XzK^2D~|Z%I~TY?t39Rr8%I7$z3Kw6NtO)rqoPPY16*Xo-f7)7i^Wa)oQ3bW=v{ zR|WCMR_c+9!YQ5)(WGa!eT$<=eTg0Kvx1il3t!9YMN4qN&x$wu9^Dk#&_-rbyle+E z&p5J8kuk?JmnUw*X7un}W0X9NbXl)Rh`0IlD&CkaJy2w7oc6j+K3^@$z`5du`1d~? zcR4uO5Qc490RM7QtQn_~?L$x!3{EX5l2ZN=%$wy@_&pi*1 z^dNiaR8N?_62KvqH}M3xLeQ;m)o_gf4Nm$I(FDrb(UZN>l=Y6c`X8uB%+lro$W9XO z?0`@j^h|tbm@!tqp8N_uPK^Lr=UweB0cd=QdNz;eb(mrTjx!saWQwR|ELcwcsRfGpfUJIhMGFs|4CbEV)P9I*TK z=0&b=Uy1jgq<}0qlE)^3i0rcgahGnVkvVzG%Vi0~QJ)jU z_Xr<1V(3QK-pL!nAY{5dZ!zK}D3y$x^tGQ`!6n~H!DCIVfO{|uq*>umd^yh*Bq$wP z$)qZ;F**U4k0GG>Ev2ZA+6zUU)NOk66M3o~F9AkNtK9o1bYMHJBLW}`U+Y&~$T5Im z^>~&n9(K>9SJvj~wx~%vTm4=x~Nwna7c~RS7+b%HMV^1|J0xsyXqoc?c#r z5qj_T@>{)|aRXydK)xz$NRPx~fTXi_-8H}S7AZd4_kLTWr!3a1a!WgW@fCNbZ6u>f zsOwYYZb7igVBH~cSvU@J#1h4iNqwh+gnH&V)a8I6T3D|&S$vRy=@lj;KbJn+9SB>L zB)n|+Ih-08&s#tJ#%*m_!3uHsMzmEJdtF3o9F_YQjYpZDhtIg3!E9v*xX3^a#9Mjv zoX-WddjGlenfl%>P5D=x(s6?juvGeK{XJrSBmIZY_js-Zo@?CQLjm70t&e`7hz3{^ zkISQU6T@3QJ$(RcT5GIjiW#sr#|CtTA7SB3zcka1G@L)Dq7PEcaGYe&vvZOsH)1d* z{59V69~7wLb}E@h0w7xxhJ&A#j8`s3(&j`|OS{{bPx4^NQlXE+Q*^v|UITuCTiP>I z@ecjP6TYzGz7*GwY_>B7rw=&GiLgj)#I!z9#MQu~edt}!$SeHnuiuNh#aI{QA~4|% zJ#pD2&&^WzWbTWQ7|cmbHP~E@AJ{mRDDLs_Zf%8`cCuOF5e^HuPwR{rYT zRY1iVw`+xyYYkjj*{BnaLV~y#=)~8qn?g6g!s&mI=~y@?s81Eb+&eAp-V74BQx`WC z0l;&C*n%dd9OI-b2F}-BGByw+ioD*7>W~R22GaA$=4FFt%303$iWlk!D94(=tcq~y zr@k?|L=gk>US>da68joQym<|;cWB`j%Ml$h4yGpdL>&+KVVBOw*84OQ)uwr%kWOfGaki* z_=F+D;h)yG&E%1=X`gEKA;^Lnuh-+Nns6iOP*EW8kg$_(odxbMT&V0@vu+8Y!?8V=qR1_R2vTIU zH2`LH?R7EUuL|a957~tW?0S1K^{c}F%8kJfr%rPFXHUC4qz72L>)2}CWK`*vRazMS=nx393fg>>$;wUN&_cVgIe-v%G+ z@@wb;*ZH3HSmLEag@J&y_32@+K-7VRTQ^YA%^K^1vkg5{Av8*JR5Fc6v3T4}_H;3wX-y&QcZp~mzlY2FClW~Ycgfl5G@Vo+BQ z9D)D#D+p^cQca|I1DC?kQawp6%~iEcaiPM9Mj0EG-ztvEm-b5OtpTxM8V6AO^ZV7v zDe#Sybr1966T`GiMI&@NGiu-c169to7KNKEMlE2ly|~Au$e2tNPLVw0dt-2?*WOwU ziQi9K`8U&k({dIBTvxG0$<4c!IG2-L)Z25}++pm={s3kQVajli> z>HqIjoczs`>MstfmnWZ;N{j4v7)x2pU+g7_tobV)EwN^tXe+dj%q8B;H+j6fpUmv? z$LqcV%*u>Qsn-B~Ssx;Cf3LgLJYsV@jQ%XqgU#?VIQLCc$k)#=0p3re(~9gWmbe^; z(Ozn$o^elVXZWsiu45J$Zzan(C&VSK~`jvp#5NP&RvGLv#xUa+| z<_J12)|DUb0)U}R3~<0iqK7q$Af`|X)yxWHx^@{WAT@GcL$GUsBteCse3L;0x8S7N z8j#Z=ZN8Fm-`)=Eef)2)CP9*Gydw#hcP>FevyWbE_B-w0z-8xy4#2D%dmJ}m^m&!7 z>gzZHRXE>yjTmr&KzSy{g5zcP+LtB1h#4TP$RS8sIk2loY8)H>{r58T)+S}^EWG7p zE9>pYj#3hyp|&NeW$0Ue%qz1hg(PiZbyTyoTeoROg$jaUjNZRYph3mr41P2s)~!3e z>l-`~T%LLygV%jzYf!GmjRerxGM^^{H_qH}m^^{)hL8zaS?e$!JfS&Y3!#F3EuS!H;vX6j!(7IaGA-;6XP+B1)mC9v>N&K` zTDI%O37C>c_5+8aoc*21q%gi37$`Hu^V@O~|MVZOJ#JN0-148qu#Nfc=VfJi(++BU zY9jV1Lag`F%ed*;G1fo&_)e>1Y<}H3_A^&5BpnGG}mS%Hx_Q3#-K0 zA1mnKn|pW$A-Ca1FE_Lee)eCIbDV-!G^7YmdhLhFr~}ojqMNRx*zs60L5s&x>2%y9 zUxDIn4Nc$o@<}tsM&71`U*XK4|9|sbWgkvGauPG+ZbE=<^~^Qst5iipE*%#i1CJ01 zYptT?PglC}h4nZ%=L(;19MvF~_t0LHEA;C;28ltdE^?P-Q*0$~`F7FfSS=HP zPVQ%S+6#mi%9e)63qu$Mbv*2Hj+AVoeZJ`;h};|Ow_Vz+Me|1@0EO$wx}p6lKZ{uy zu)&5(9P$?HwU72$Nq5V_d#S@tP*cFV1*NNL7L@BKHVe@K%^ z5X6GdXZQxtmI++oq>49iReaR_Qr5F^x^Ukw?_CP!o>ve9qt(0|L^yY+JV zYYMz|uH+tbp?e^dh<3~tE#Pk{Y3uO;9AO;f61?o=zmo%bQ7BmCR763c7TG{EZld2( z#X8#JA@`Suc6j)C>Dq70p!q(Gi>I?ec*XNWi9CeB#Cz@f8qQG>sfY&rl&$pgUG3&f zdVKEJQ72(}p;=#50Ak9x?qsznvwGI}2Eb#?woq%3oy)izXy}>heO^Hqbb?|~V`8`V zX|r2ynZ$sfo>Z;4;KY)x9W$LEv;dm?0z>TDT9;Ff)MTW)o0bB5*Wb-C(S$#1$uxw^ zpg7oALshWMDm`cD)wr5mem&(Y$-Y*uAPELsE6~+iUr&A?@cWKrH)EUavYy@HSPf<| zUfuk+^7fCYJ+s=_k3N~_#JiCV0eeBP%L0fe2P)Q-@_Mx{={;gI?OlPs_eeOZ%6~=k zw$icQcF2_UU88qkTKE{m4XF41eP6aO+}F&5P-KHKu&k%LFdH57B&IRsy=n&O%%=-pV~vvzPLAn2 z{;{GqV5AC{FGe1~<>Mz2mFAT9M`gWxvP`pQMrI{>)2CyVPytrbt*pqcPY&Y3w6WQ> zO1cKheXK@Ipl54@`DC#Sx>Aw;R(l$ucqBu3xg^o`2~0%9rS_ujbJ#dI@+q`ZCr2H} zJM_5Yfs+~0i4JUkY{QKWq0>uc@N~w)qw`fc$#aO@pNll>BM>S0<+UO?$Btfm+RBq+ zc;Sg!C+nZ%L#|kPFE|lfrtoux8W$SA{XgJCe6eudv~2h1o083-{Gz%2nQLb=EWO+o zOdr-k97K`Q1i$gT_Y9lr(8E*y3eSJtoE)Fc!4C>nr_pT&WjG;o&X?3S;$ zkaiPQgM^a&r}!nggD`MiWB=FTj?gsTe%JfuUB(zvk&Q%{)A>%+CH)lh>GF}HFvy7V z9hXS8u>fXwvs2>lesu(tTKS8zkL@2_h@o!0&*^o@*N|p1bT2)EfApY6Z1*OaYn|uv z@2%H8_=qS##<8a_kUT0Hc4~}7{`_i~VlkphV~978h?g%B==`e5qjfG4bUaHI%fFd# z>n|vFs1vYTIx_E!mh*s@;y57LrYD(ma&@Hw#JRWMbd*woISXf{Sb-Wwtr|pVdLniq zF^4hZr>XwEihEaT0kOuwghwe~d}}ioVP44DGa`&l_iM2qE04GwZmr6s(rfczvN@5= zzPUnI$8em3JXO@wfpW-okQa%b@pCL8R*cePac;1np45?n`yCt+e<16Jk>8uBkp$Ww zFa4~Vg_3K+bZ%X^9YRkJe4>4sL0>z@-+l`-7{t?n&Y%Uix2ROvuUY=N?k((f3bA)Q-j5%0t6SWo$?EWF8~n*}y=R&n0qd;e zOlvKk>qCtK$@}GQBm1>_W0L-`cL^|}Menbs7k&%dZiJ$jnHNMlCEquEB%uW=5q5)& zDQI%;D_gu~NhQG~WeGFImE?9V7*Dvw4cvnC^)(%h8fAS~dO0xi{nNR^tI04JgEBKn zgjA~#5pdXFvT{c=4$OM%=vpfnWlz-d_9=%#;K*i)}JX-6%1 zMyZYy=iXWI;oEIdzl3qU;2!kd@j>|60i}Aw$xV6I9Cn0%O)H}|wBzM#1zo+xAJ3j> zSCb?0e!dH=0;F~{k-fHZ^6i+?43(mPsZ-2Ynd&wyiGKa9K)3!@EPE(P6D`PddP1nK zpcW0WS10xr8`T7pS{1Pe6|1*~?S&~c?2+k~jRsxoZBq1)&SBRje14FM2pI@$wxhaP zU4u|posU@>blFdHUa=2xx|IFa1goy*G-MbjmV5<&U=C}-zn_W(ZA+7clQY`YKZh9F zEX#dVPfS!e0soe>=fhD&thMn)2KJAQj+6fiU>q02^pImektU9(_qbpABAo`{EFsJJc*VM~fEz07|qkz&_f>xn)Fd#3&*gyN28U;|V$L_a#d;K^wPeajaXsa>Dyeqh0 zHD^FD$eyqoH?Ir^@h7$4TZpo1Q6tVsl{&Xc}1| zwD|WKem{{G8}6xOnv!$HC4Ltr19eviNKr#E#Urb(NW2V^$nY$56GFB|SgK;s*Cwe& z>b)cea(hN|DmNLEff~SQ>Gv$L`NimuZzH@Ec`1?Blge}&M*HDu38wtE1%lDzWCBAV zgr63@6xX^6$!HYjJ>FKIz7@5xL|A=d;DoA($Id_QEXNDhwEt znyzdz4a+8uD|-j^>c2Lgw%hvKyK{_n zEMya8%`uBzFqP?e#-{CAvqU{dhT%>LA*u-3tc*z7U z0nWAP?Jf7mJsiNcQP=c}ALZ4Y1~T`~mWF;Z@KF6;6Gtk#v}1RyOSl%A0vwG;4qfJU zI9U5JX$CjU{@Kdp8NTQ{(iF74eN9VF=lOgNbv^{hs7RRGtJOSMDxeQtyRB}ACr-NlPNUzKP5w*v;p3#Uc zM&gBv{4A&p0`oh4B?S5!3L1gotc|H-mG}%N4y!-X(85Y7f^|O)mXeKz3$}`YFn@E zvD>r0J&-^?rXNsFtUugGD05G57Eo*2hG=c=AanD!;fvxlEB6ludde-pjUkjF4r2qyCmI5M>kGd%F6pnkdc; zk9TnmbO)Goi4Z3FStbvmNfOt%uGqQx9XGNknRrZi9VJR|WN>B%A+=%2F|tk=Rgh}G zZy>^di!7x5^EFPC*G(-@OFfLU`nSwynu-chrTRLIfjie`Jt9t+X!e_>ihIh;=dV;8 zT0&*o^F*`5c%Qp_qTbBC5_Du`pBh->CN$f=RLKOV^>(V{wMX~}pJNj8S=o>sgBT-? zb^)lN&ojfYlb(e5zjQ|@Jq})&aNc+<1lD)pUf0N0r%IL5?NGPpGFm65km*dkrh*dJ zg+qpO{#GiB%NPM<;E4fEkzTW>VWkdm@SM{+vf+*nj05^d8$S&#GLLi!5qa9O)x=Jd z4wkE@cwAEXC{H;DyW2gAoLL=4e8&wKuR6I@!MfkMfq^C|;7u}T)4oqSTnoMY9XgP+ z$z45x@pR4j*PZ^tozRs%!(C&LDwqFeus$euqXXwgyhs6)1D`Nv-4HYqNeMK59TvSfu;(Qet@DT zZ!Vp^<))(>PLZIbCrcbt z(altoWA>8Oj_4UxtRUe*M1`NeEut|X=~b=>H=j84wCuh8VK0v$0$`cZc58N&$6 zv!pP&;Z-SVXvgplB14mOrwyBUhz2D827=YYSQIq9sgO-@L$j(bzst7i_dQmw@mbZd zH(^Os&y$8|t?NY^%0!}qCE1Rs9N5vSh>AE4HKA6ibud9VytQFM#e^KJC=szkYeJ;A z3@IzG*qD#{Zq#Z;#=`YJ$>W~`sp=q3ZmXkkmWwgg3mLjhzU7Km_YG7au2>MkrI9C$ zk7>;ObQFG-9TS5k?t2)^q>?<}vf4d=@JVigK^sZkDZ>ZWbQmaJ<~x1NChA<^1y@bl zuWNk7pd05e?$=;qdCzZi#4H%H>dJ(7VeoF9!m!6dfE->25Qq(13d=SvD(|t*L_V80 zF#>d*-fH-hT6WnhX3&&WAR4l6KpDImaNy&-CW_zvv}lTqlna(%EZEgqNzz+pCdEhg zs13#U`Ekn2$Y9b7$#sXSsuSN)B*I$EvXx}dm0C(AEuR~E+`r*c z<}FDeUWqp4SG}hZox_OOX|)`eC#RO!1C?4piJYP<00O}Tv70QBrwW@ zBWE8iR1yBND^qSgyfP1AS~4O^Ye#>lH-@73X6wbG6R?ge4AuZPUae}9B*^AWDUoW% z?3NOESr7#f}vOJSG_$%~6z*8@hPBi@yGj^#=WixS%H^mfb#wx)>qk`zwb z+P&v>A{-(&n1whuB>}-f3!5FUvZe++wkdxh6~7ITaPTD&0@QE^-oQHWbQipZlq5-NX+I) z=*J>995x(O990ojzX)$Xzc0X_-;gixZ}RZyX3fjNKG)25e>E^9XnZ{Qs_W`<+waKm z=O?}XVw!*a&X$1zTbbPd*qcA+B`bZ2E`mqM5FdtM{wsfq!C4Dtzzx;^bj^M1?UZI0 zypWC#k?U}TM4pcMxGL)!7;jvIRM1}qF#H<9+VC|$;Rot6n&hx|?btf&Of?|)Qcvt@%Bo#{>O*oel-@EksaGxcrDYqyPT4heLuvl3h|Du2a8Zz-)(5!6m zWKUSMJEqwjP!EH zez`Qj^h3Braq&WgG59ns{Vsd-xlI^hb9HX~{ms9Pt50MbR)M7ZCC7*CXjWcc;FL>O>~_e!dm5i1!D@A?3YI zE~&X&8Fo!*z7wisWC>LjJm$rF$Ty8r{sr+sa)wXcl)bnn=E<^4oN3s?jG%5DPyT*A z6;YDN5>v-OdSCy}$76I{3;a*xDY6BVDD$3otKD-_k~EnHPNzG=Y`+h-BCarR)cne- zTYslTOEMv6EITkq2(1kVfj!M>rFQw=ba~gcmWaU�|xwcwN_lP5;b>22S*LIL$2K z9Jb+b&DTNP|5i~#3SzHG5$0Vljl@8r|6!c9k^93tSrW`o`Jij&n({Uw8OG;>xJO4hfd$LJ2s9fxtzPFDI4pcHANXHFKZf;zjXlWRZG%K~o67Yc&hC5D3;~^Q z=4z7wBvdI-H_Z{S!`T_nEj>^bDjc~f&11vt`X4M-^FJ$<VA@PMYTKF6<6NE&9)Cei z%|YHQj2CUc*eP4T!l#Vvms)Rgp$8%G&|3@oLW3VOGh4GK=-9Ep<>Q(CUfjPxu-(UE z7pnl@NetkohAS1E;w1es9GZR-FOkcC1E^h}ta`_L-38WCAe!F6Z0}R4h3nAX3023H zDB*%uN~Om?yDFsL3JKgczJ6@ML-O3+R6s>a@@jKlDgSf;E)L4s{wC&>{R4V=i=;1P z;z#KRiJ!}u8I;)`nPu*c<0M{ludK&$ z2HB0Q?s1&Pmz>cYj_01DP8z~0F>)7AzD!9gq%kV}R#-ckF6VQ&1+ZI*GLP42T3TwL z=a%C+{MYLdgaV|&z=J0zHy3U_z{PciRyBC$-!S|4bnAUop!5moL}6&-p-iXnzboNP z;hW6q+7FKtLV{U0I77>po~%HK6!D|1NU@f&Dni$rL+y~DPp%L~(K>$GU5GbCS7H8C zqFv?s=42-p+PI9we)>3$pA% zC(s$ZS%-wo$)|HCa)H=4HpT3{pavqoeQVnpi4>`0n_NXno%mSarBn$zqWp`qTaD?E z1)yK@Mx+xfw|vfSf1|fAXkt&==yl5}Z(piXwz_sK1=?~)KC6Fh?WuBa5@*Zy;u2&# zO@FXP`wZQOWh|?MRZ&Th$9_y7n9R9C9eG_?V)$nTEnN6hBOM>;uCD!sR+7NgwrOdd z0sl5{b1zy0tkU(CLq48l;?Y}wYG}cM^t5l2^iMF61}2Zh!5p3It+rvLJogYv+?_(w zx4I5+BAC_W4>@s!mFV6=D!Nw;U!u7)hr?Y(Z9`uYF2QuGKaiqMlp;d@A;wN}pw~f1 z!ES7h`d*`6*Y1fQ-b||s`D;Qb2U}K>34Zw{>$MK|a-ChF)F6+@w=%MwoVH(>mq);p zitxRt2P$3g`HWOQJu5@n$t9$WFMA6BU$GfCz;Ga3*j`>9Q_$YjPp2W+czBAy`H^|V zzkr=)CAf3i)!mJv?mh5DiKuJ;ih(wmW6$FGgGi|K_|Tn#dP=a%gUVw4^8= z2n3R&7ai*0PlW{8XYDqW5eV+AQN)HDjHXn$)Gq&7*t1WP^T^k_${&GU!XH^HN8xj ziqh!F;Hb^NBgR3oLH2AFy!rf{PHbk}#w}<$AUW}F`XOO=vUTbqa;q}bvznTh^tO7e zO^atsxRtxTjp8bf4EEFVXS<501e{0&M;D>J!AjBs=E@&6$t)Zjvtz>Y%No;wBtx)_ z3o+xvt20W*87DpvnjA3?F z-I2D-FAkL(-8Dqj{X8P3lI*aSCgNW{Cw8-kV&YdUu=8?J(*?NW4->$2ADAa?ku{Y( z4nO7O;W?=QXI3UrwxQN2RHmas}M z{{6W*0jF$S$&XhpC5oJ$*SqHvBMP{?+xX0Eb!%|yO!#7GPF?6@yB6-+9u`~T6*^CM zHL?S^$6DY>Lk|-qS2}Hj@z__sp4>oXs=G@P<`!H_zwpdp)XaOIve%~F>k57BGDkZe zzDrd=q8GrjJ~A-Dh8)rxU*1&OQ@u4(77qFAiarr_{HBA(u^fUC_7LVLC^Lk>K!d*MY@ zXT8Bzrj+1Be;`Z>Z&e3VYh2j@CkAv$Fr0|B|8dfsLiOCQ=94NZESxnaEAkb&Fh%~! zAsi8<3mxPU`bn-?D&Q8if%pP)5eG9{@k`yXPoi6+ry{y_`?d4QMpZjPR(0HNfH3^l`I}&y5;v;~F8U z)O=dS0DM!YlC|Cu0NubSVbnAdz8N7sz>qb2%ed~7<4nn!X`}0kNjnQO3^q)C0dt3u z|Ew}pDVC3_3;1VGA#w&>mFob<5vkYcwq{<-%K04Oh*v0NT+Z|G;bzN0*xI0sW8$EA zJM2g_=!g`F)JsB->S!e^?|m^O{{i`e&ud2dSkc+zMUMf)B){Lo-!Dk?&D||L(_}~_ zgBdb1?%Vef%QJzkW;bz;n%T`Heo8#44+Hqe&+#M+;u!^n{GFmP%O8zNWjumX zOjsF@!=8~siQ|RyY&^aTl95clKIztmJJy*c{Kgva;sCmT;lB`mShuP=F9Y5@Bgh@H zInEb-ZvBwRV}5P_Ugqs_;z0J;^J}{#OvBDuVrBwd1u?2PuSszVsT4=zG~Md%Rb(oU zw=~DT%d4O1$6t1moEXGLj%xFAiR`n_8l@1F-A}MPK7%N91~W`42Pu^+H4_ALwcB9^ zV29-ldSx&;$f(42S@~Z-SX%n#I=HQRMkZ9%@Qb$%XV$b?N23$l${*$cev=k0Qh4g? zp|jmFR>J3K!Tx+>31}1B5qB*LT|oi$-mBPf0w8+AT5Td}fJyYAv;T_XHy0dXl*OWj zhZeC{OG+1n|1+e9?U#kl5Dxy*csOjnRmF1ije2gBs)3sM>nIQ>H@`y-GY_qwY-27J z5?w4-#80V&9%bZ68krfEDO%bWD|Jk&y7Z(>BACH7E3?lCVYGPe;v+~!s7r!3^0s{80V1h>1 z)mquVW>Hi1Z$hCrZ6j-rixngU`C`eUr+$22(h-RNvGi2|QFU$Gk}@iZt-Q6jngmi--EiL~0dB1<3bKu_2z1Dpt0w|o6-8=s6zg6F&?LBl% znKx)5&xihVq$}B`n^nr=7xbM5Qir7C~A~Sig zDeDE-6X)LQzKp z(1v?$7JgQkd}>JiQXte3vz*MQ^=Z1!dEM(vipRRM(c&~k6&06a;wov^lNwL^^F@uk z?@}IZdk!QP$?NwNDgn!LW|fqXwN!u#$q!k0_&4Q{aDmAYP{T#kvubO}{niN_lbgBX zfLe0$Rwim=U^a}>tMvQvTSiWFu1lZgBCf7Pvy2Wp^g9ci~P1za_( zF8}=b!8Kj5%nayDU+x226Ht%YZ3;Q7*U^NY`ZE(gFqWf*!@>cLlG`9tBV9IOQwv3d zs&tPg-k2yS(Qp}~`LjWi4m&m4LC&|Emk@&}KnmH+1Oj3;pB{^sX{e2T+wWWj?$r8Q z@sU2umuxP&**9*7CAhxVR5TV0ahavzm&i-#i-pKN*)V`+a#EgvBUE*WlD&`*lOykY zX5$O{0o%d`qru?t6-_kGX|`r|?Z##=t(H?35ennN@y3@pr`{L{T9D6ex-|2Ru5|k| zbv@3=>UXFMBj}sP*v)gky3}Ae%N_f-LWtZ5qtyw>H__m$a+uO^a6=q}&sn2fWOYoU zJS-DY8goAC^M`xdAJebnGF=o~h=;;5BQfDyDLQvj`lf#d+6YC$i|Ii`LT(#`?qAgg zFc8+QwY0Q8Yfa|7pNZD1(2(9J9)Yh$WZ6w+(>6-*!1DN>N4Dy(@9whP5p5L;x1Zkm^CDu*F11sL@K*X$IVopPMJ9ek;gLY@Ybl^;GC>UTPcp%aX8?b?+- z1u}+zR2;y7ok9`$vW)q$`tAvdc>QuNk@$3twjWXKAi@}0G-y&C2{f-Dhz+*kl3My) z{cDtDqUsNMlk>y8)d@7XoG)(ezmq_fPBw-=Kn*($C@w$m;}TvTP)6p1B9$_Dj?~qP z0w0sFpB9iaglh6u_Y-+rrAtm?X^BFW7&7Ds8>%=t0!hh68hWMU0x8^#ot)TDPftJN z1Of2JY=tI06g!@ni_@;@0}^>cN{XNUq2+~t2_bFB%d^(89StuO;oEENPe4QUJ5xa* zNL7ky2wysejwDD954;LCrJ)WS*GSPsCN#WAeoMOg$3jgx@l`` zPvr}1gWYY3ce}^zDh6a}RJ%4S0!Q6%GkdnUO1M{5IqT?fIo=Fyg~$$S*bR4wJ(_AB zS2PL}x~i&=WN8mv^qv*o+i7y=BcA`zBy&=gF03D~!x=y52~vc^h@7Ln<_?1)(zke< z>ePl|DUyLWor2RNiE)4TpDJDNGA)DD;1CsxAKJR-XLDb^7_0vk`C&b0M)Xv`-;w1s zDF=o-?|ImsVczwGew0U_^v0%~H9hlX40@LudB_0(*8`&5WRrIymZu(I6gCxN$o4Fr zk-n*S)IOo!eQrDbbjC`uS2M1dYvv<;?{Kp|E&ogM z-!r`xk)JqVUA|Vg90-2*RVQ^R6hY8stq+LkWO`aaPyORRLo}t40gr5o^z(c2-)?C} z;oRs*CO`1q+CHh)Mj!5E`YC|7F*;z;IzYX69c0lsrbzTS;9r8%l@fyicE5Yn@e}U@ zA%hoyZ2nCvo`4)FM~b5Iz&CRuqfhSND>U7_yyV>6xZ!Zqko~K;KB#cDhLX07soGpL zMmX#P-MRO$dV*!Ds~3dGc%tflq13711KCyLLhp_Jid{ueR&MIInZKq4UY1VAPpt?k(E#KP@ z=LMFr7#&}i-u-#}3T(qh9*iKCFNVh4IDJh&7PBi-RI^*=DLfaAEM`zW^Tiz^2<*5% zw|}7Tn>VtR{Mj{%eHPAcZ{oFxDyki#Vqldp7j9oS-vB|I6y87;bpAUuees-*#p>nt z1Binj7C$4c5>l~w0p_703JFAoAyG+$D-aAu{h~Ej{!ci<;MnnHnAogXDh#nf(k>lq z>m`#}QzqGFLU|@%oK;bmQN56T$%1{2s_{bO_!*GITB8ED3cIP)-U(%CG#6T35Gdk>ig5bL+z&a zm_?DW3&(@TVzq@An&eIf$iD@zSyM>o!b-0dEYoOERI>hk+Z2JO zED68^=Y&!+nQ+ey$=HhIT3fPURH^dw4AuncqSar9SAHS(q4XHKX?|nbj`HSrL&SrP zFS6m&&M7Iiih(oRIKmR`dX}!?)|N1_oL!pQl|k6&(BSPZA>k*bEIg}O%AstC#+Sn> z?hIeqEl!h2V`F3MI+k^5A*3l{wsl`L)aC{-*Se}uG@mB~b+y$s>txr@r3lgwGl)Iu z@H|YWkiEm3VQ$2+s|F=oJkA5M7|j-7LOi85V!5vxwH&#;x3XBwFOiLz=pBAF=#wa) z8mULJ-T%riBDxh(p^0c-A&|HTBL?G&CKL48VZj^OqR{tth$aChCscLywyVhF($7T7 zCb#j@>x&U!3RJGu_YC<~AQ}V2%t9;K8yiTg`cb#9lhU{=^=dP}IWIDNpt!0jzNpgh zmj+*{tByDvb`rfha6fJ`iwMN$KMFbJ{BVK!qj)$TcklQps*KU1;XW6ku@()^{VoE3 z=1b04L}Vf4Ge%I#yL$2$ZIcb^kfkcRn=C%JdeQjbCuN^pew)p3P|EBV^`-?e zq74^>jj1ih)LZ@aBH++cWE>$CG`7RC->6{hI5vqV`~KI&R$YXN;F(DEJ-r4lc!V^O zdc|bXC>Fab{)kyQXD9uyprZ8F^}4 z&27J&G%}G>-C$;SvVPY%5OBk?wR}10wP$(a9&jI{H5U}jmztl$gvULVevLwJ!hXXI zi%cNqn(%q_U*ISQ-tl~O8DwhV&qPHKBItsGw7 zMDsw4fyN==dNELC&}f|>+O<984@JeM6oK4jxM(?rP{5@^yovUA{35uyB(>db<$YF? z2%vuQ(vB-LHhmQ#-n<>_CZsDw%e@Ld%9yHQe+EuL(k(KrpB{Q*d0FA57r(;4iKy39{Y-(32rdtYKRRsRj7iYI7%`l${l(KlY_j2k}h)g6i^ z#4inIX(an?jHYYY(nnFBUhI;)Q?rNxbOsvn4KhcBEM8td;fql7|^ib#{cs$ zX`!=~;ziRgP{WP12&chJzntW-Q5l@U$G*U@fX!Ufqvg-p`HusB$4~CNBfK`&SXs z%$Q=dSPZrIDylo=UvPGI_89w{pr_jmE9w}C8}+g7&?TzHHvevXFY-4&lY+!2O^*Pd zPkKZa(tY#wA{gO3(%fT-!Ck_*@5ur}h(9?+LTKkTQzMrUbh1BeF05bL`0)Ab&r-sO zkQoWh*NC+-iPG-Q|u?dzeTgVNvrDLvG-^{0Vz5c0N$4une__C|&R*Buu;gy4TRfIoG0 zmh*SJLDtYvBrw9#Z7-gQ$fb2~FC$atk_mzP^V@DVkt_%slyM|ZB2CZ0az#xvzx_CF z*6q>6&ht`IcD6Lez@eg#FFTzJ(~y9I!jI>~{M1D*s(=gJ@j@Z&feh^GnHg>mU<)a; zYQyF|58W++P%9F*sVr;|8r!c1#xQr>^$6hGIVDmSui`!Z1wbYJM%$#oV;-U~=M-6R zGL<;%DA{E1zn1`Ymmotvk|J0$*PZG%p&SgS&fa!@j+3EZ^Ma;o~Xlm+yG6F`lKC@Wg??e69Hp5h)URg#$1&gIG) zBztX+j-_*TDI3yt1wT;?r#fblaV47A^UQiH&|F3@)1||1B2#~tNS=A#GMt~Z2uDsd zvkh}s=HT)9a&K%KJzhD@6Ocxzds;u-k8xIahbe13pP0lPF^n!ruEnqRAvG0jg1SdA)ipN5x z3PkmS?QZFM#_TV*B=AvpE`%Wc8q1UVLTMxHw5Z(8j8MGRoMuGg!h%3tnw;02bs*V-4jZ$WGmM$NEj4=1PK)Qk@@v)CXBtS{+jEynf<%ebNcVDW zw?5i;L=00z@eu0-y#z>?2$IW&m+XDS)DfY_tKW((9}oGs#H&Gxl-yHg+R}{BHvd0gy=qGsLhD~Q z$;{e}{mixHHF3N8D4$?mO!3!L(D^51HHWQe$Q-D>B}E7Yl!H?@mfKa7W-<`a$Vvj* zhyaNq;`G1#%a1v0=D&}AzRSW3b$Ofe*g%J{{U*sQG93Wux0ma!=I3Y2)v2K>4Ywt; zzhv|{qVzBgr9*K*aN!Pd?uKa-5ALrV++C&SDMmJei}DvjxagI_;mm7XZ^Ckt4j%h9 zhNyWa;x8s5bTuWcj?~77K@C z4at_(EJ7j(!zbHJ9>4J};N7LlMP#CZ2~ef3ZD!n`KR44mqAxv#amfyDroFpI+*A3u zeWzy@lrc`=iVLTX(!-@i=({cLyXWkiCFK=I#|Y{Qa#C9bbePa z6JoF!gM?GSnA@B{GLfnj^9_5)x_-@lai}9Yge+QejLLrLhvO&c@G@Z8Q*2c?R3}AN zb!@owgr!NIcU>n37Q*$NR$Eld77se3zqzACWnruEQD+ql{AG>*a&P1mNq4uW+s#A+ z+DXc(8D|(E9^7TUI*wJJ9}pBmXgN#eZp0aOHc~=Up_Da0!2|WqB10H@NiY!cz~FG; zNLNxuJON+GkC7mB;LiPs4I9Rkv)v-=)wT~VCoPN&jTeI=Xdlg?_ZcwVN&>heP?q*m zCEjOlHFJNxFM}wv8H_rtiJ6IRNF$VNW&f*ezkhR6(u&lsV%g>5-{GUaeL`_1{`zO3 z%#+WAJDR4y5=+hT1S0a@sSQji1VVW~rIjQ}|4cm8AWb&4;0}G5aNp_^DKHlC$f6;k zWiVvh#ebfWAGlMIW3({3m?jx|ViJF$m)$OXRpct_dzL=INw~hYWqTu3OqlacRU+Vl zs(6x#4?-%*=U@{Al-}D|Vh~C0c?JSJC9iCH13z;Yg(4DIJkn#j*u2DAxpJVAdCeD9 zHQ?BLiD}RZhU`wW01n-IBa&4D<4-LBibVTrU+ zaK+;4eVK6cs}l&Y0w%BIIoG`}V!JK4j&7&H;)!D|f8-lRT5DHKwr4Ci%IJ@mvA`U7 zOC1!6fP^|0DjwcUYmZlw0EXg4eBq=0j$LPnx>m?aGUZS6E11QL&=j59Q#oGag5_@E zO1_t0RYXgBo)@%AAnYeD)%#o&@?yZqvV|Ns5YfLZU_1CplvTUBK}Mn)D|+G^YKcv> z{Dyv7%pz(>zK2JwoXRlpujb}P6*?nJ^UKo&O_AnPemWuGVDt*v@imlh&^<3r;oT+V zr4>dO+I%gvfF$&tVELKDT#KYlOn(!A!nbXr>+|sJ#xnBecEdQoh-Iq0@81J^88NtX z>x>8z5qVp!lP?w*ztflAjtXr&&qq@yJb%>F8tRc8US5351p4WAiELAk$~v?1Ak%$7 zrz#|Dguesl7YbUXzJ4QWNdXO|d4;cs?ol^Sk*1q_{ELB`lGTx(y%jdel&5dia6ipD z2_e8}UEax!;?gp6s~T{O8{#fmsYSWfdChr4cz8{k-Jy*VaVLQksgKydV|poua{3jK z!nx|axVZSDA-rBKpMcr4t4$@cIPj>5N@!e08L4s6{VK}cNd+d&#({4PW4~zI26AU^t7epB%I4wyMz!o#X}vImMKu}& z{r6$VmJd;)2#U(@!cx8=CmV$Yg&?ekyi)VV?QK0K7DKVuepR_IqUGie^W%#TIIzdx zx+MsT>D~VfK{cRIhM)PaKy_aNLY^=kMGdBGs~Ir>$yHlRvGJ3VN@78v}KOv2Rn}Q z+FM5^0>+@ywyFUiUHHOJdWM}$7s@KgXuOBP)5L`5%&bEEU%=!eDf>f2Y4P_2tB=3s zpt^2L+SAufSKZdMleu%r(XyW>naX(1@I)9S*4ZY)+L>%77uB2|dM7{bL_@z9KFg?u)J ztV=|wB?pU^3{CqTnkSQU*Unrs9CLH5oQSxV{7UaF;%FX{C@UhPWsqJpQvXgah;WH` zToF=R+i5=Q4d`ez%T!2nv^3)2Ys_6Q$4}q9Mz|uousi=pS0UlLS; zMQtDOkkG}BP;?3fws4UY)X3AZ8>r5*9PJQbkrH{*Q>8#4hJ8PN(L`g~rgKj4~JJI%jFy+ z*yo&C|L8sggdhFXb(f!3+@&xK*{Bt5YV#KaE!=dRc;Nj+6RwJhLs;5dr_`N2hID9% zeSTnx_brMs66|0n#fW%~JD@s=!iolzXaJZC5S4v}`c}B)oTR&}3PDF?1Wv56{Sfo@ zw|!#n%Ix1>JKYm3KhV1sXU?FzDZyeIO(MOJW@6Rhar8-He}9BIZz4Dy+0>q7bIJ&21hr3$Qi7p9W|Q z38kK{LQ(3#!y(Nxk(;t$c+}PAp)UXXeDthMPsA)pg7*}m+?KjD_~Q*EY-l@F9UPuE zUyQf=HhXyMw~2IJ!u;4idONX7^~|ehK)7jVZ9e7fCkDrLjQ_09=fs^8*}BOYPbZJQ zJ#hJBx%&gr2B#@9bBKUq2x8VKZSbc2#D3DGlk%&1&B}+^{kk z2QH%nd?y2Ol;92GE$bF_Fl?SdGJ5!jvy5;n7}7_D6m|9QOmvvrL51sHh$<%LH=YW? z-cAJg+#Dcz=Fwre-mgzGRmad`f8DzVK_WN5r(y|Nq|*>b+oUPn$)uk%v?ct3o#son zpcmebx*g2MCrWg8@JAd-WJboeSvyJYM_q=~9Hj>;PZ`@qwFA>Gpg@4Argfw_?N-s9 z0LbQw=-f%sv6EchydZ@18^X{%Rq(8Y3;U<>k=seQB<{>^q?`u+A~_Ql;^rdL%*!A8 z$lZNvrbfIT{w-|)iA3(xC&Dmbj{IzC<1P)!jc=Vsr zmKlyi$DrgU2IRgkdE0+H~yn>ted6fG(K)GSaa2)t+ zq?Q24$AN8) z9!uguF{{~pr3=qP_um>G{=?HC&W5b<){m((g`I%hBRK_it;cX<^^^1$IW|9Zlp^O+sZH8D}HA(>iQl-E9Vt zmr&XLc`BfUD#o;Xl9bgpXXr*k**NOo=P|gLx`kKf;fEBW zL!ouzDS!`1jJ92 zLN@yjCGh^GQ>(ql&8ZXuvcoyC-GhQHpKI5=$>)vm#GJ4{1sA+;z%hxLZ_oeOJQ()- z8ge|A#Z?QT4eD%#wVz17)FHzMU4AXnJ}Nwz4ES!Q%LhxN0aj23pLvwbaERCWvRnt9 z`&1j=x?Jb$#p>)(lRVpx0NT0#MvsFO!}&X9dvYY~L`Jx&uf98BzvHUsfl-%}ias_r zY#ET2g5k|VesjcDnfhbJyW^GBZU>+7!4!e8v$GT(LCMcTPc%?im>h=S=Vd?I3?gJ= zGkXfLRO|hCdu89DkPLILPSin{Dmr6xC)^7szLS6L~k(3jN&6h3Ok_wz7j?4e2K z+5}p))W~uvuu|4D8VUwM-&W7Rk6Od7dsF`ga%8>xM`h}JxS+GkI|mP{%y%{{D|QoY z8&VyOE(Y-&oK$8J4K@-PlP-muvD9pV482QFCxZ1as~y3XK~22Rf9!19tK1{VESbdJ z4(Exc3VCrEF-FIHz<73(5dbON)#SqeSl;231=In);#hopb@_YAX0Kr)W8x%}^j44@ zzC{hE!6;2K8$opF5Q=iwX@-?dmnNf(GMe|Hfdwl%ao|w;aFZD|(A0*?F<0=Vd zLE&{|MOiO$oK{n^Ia<_{=afGeV;C`@6$XFp+7pR@ME6xy1r74ijNVP*&U6k1V|F8s)&o zQRh%D5U8>aJ3FY>em*wb1I8;8wVXj4-NXon=AvnAvs;WVZn&2;jUwG=H29X)8Otu| zNKKov7}0rML`)H@**|tzekSH7#7uwKpr3B$Id0g%nBM(M`0Th>pm3M-fB(e@s6aT% z;A}eYUSBZwJ!GfOmF_f2E7$fd_})2Jm7Pd*h+eITOrVoVaG_DlwRl#Rwsw~A>hN0% zA`1GLrQ4rhKV&<7e*dTYGr$OE2{uUSr%}f*35N*1E>-+^!5c>hswFPW?M$uQm;UT< zstM>XuTXz2bTepuIf3^&HXUoak-9;Gk{P^jV&pVL19-*5-+A-z#iQN`GPx2y& zcoK=%6Z|yy-n>p{R-Ad@kWMM%S)cjV)zZA^jU}ZE9KxNa^*){D@z$PUG{f)QnPRh5 zZiMIi7n64_Hn2!!<$EEDQPIcj7#B*i<3+Ap0iGLFHEP2ePtge(KoC5@?UIwDv&U51 zG;kKAd-29wfwdY5g64vt=!p}$xK2$bXohSk)W7f@om0>W&+>ON0gg1)&=voj9vo$k zYkmf!T>Yn2zpbNzc@m3`ooij#CmIO4I*!A!gPW?yu*T;0{mLJ~bl(R}#SiJef2GIN zQ&G%#xv?r=g)SHeGx+bE;6~4lC}ibmKe7UP%H>S7~3g&R=`P zez3XKd7F-T~0pbRW5XfJwHXqRKZ|6V~m zF5MIO05PC(D6X+&cY3;-;?CcC?+8f3Z%oA4E3W;?i z8U|zLe^p5|ivrFvX71PBEY(PRgga+6qq?`WPP;TQ+bhrNHX-`s@2&T?b z>!`c6_YxxBW|hAm_3-gbA@|rU5~~4lFH4H?BzS-Qg%#gX%YY zxMX$-f9-#qx=I_Lp3#C1uq}XGCmlGz1XQ^n=i|do?4@a-v+ZQ1ACPI@0ctV?38KZi z-eANxOI&obizpz@9z~z4Ocu=rYeO9{h8aZUVQYwmIMKv~vyxZc%n>QT7kCLoinukS z8&W9c=a-UCGGkVHXs&46cjdIM=q$%NL~Jsd=pT!DwORf<<~Xy1>@z$Od&Ee>i;|{LO>;aECCMLo>~$$w&ZEN}+0zxvGc;XH zi_*lUA6F6G`5^L=ymlIMy#T1l{_@6h2n`T}OA+m`X?;dRCuEHbN5PmFSN=u)U9bWD zBtoZB7YooWjRFEl_vd@}d=YP@#dz~+G;}C(qqAlrNKW547hu#$0PKzv8rzLjt+7Y? z>bd{vJU*|4Mup2cV?ty-;y&OZd`=OJY<^d^a%H=P*`4+fEQe|uR;$q$gYzpi`p*k4 z+fDe7qypacV?DFPb?NWtekt#}%Qw|DID-n6vS(&3%E&ERgNj;@ypiL@XIb<_F{oZq0!StsnV4axgJ z)=@=pC;ER+t!TE=9Z>l#vFC(67sIG96Tn1fDSd`Zp0E?*hJo_x|6CI@*IIzzRLzS6y>E&?|QThJq)aS2poe3sX3b-0;zm$17Y@&&a z2+B7qbhL0u{==-y-e+EUH1(y=;!ZTOU_?qJmE_hO3r|{H4(daAlVt)k{~bm^t`0Hu zkCl{#A=vzzOO<+N5dK5jc2_XZokHh6Yi}=&8O#pp$?dJQ~uc0E0HFcSC7Mc3x zX<2k2b)-0lIDLX9WckyL1-9!Eh*aRu`L7xT55G4Rq@R4F(xrksTQ4Q&d|E?vEnx)& zBY7?IbBmwEwdPHWUYGxOKN$EF!GnHqaKJnJ0od!0em1lHY;{d(u$kP6z>$Dm0R)h; zvYgsp7*RayyGqOPUmO<+@{B0!9CAo0+^GB~66>hDhpJbxYfnOIt=9z*QIasN^%35XaC(;*enISuPl3Gc*hdMFkt^~!P#52bj zLZnu$_RET16O$qB)JZ<4niP;R=b&t-0TE6ALpk8)NcjEx+LF7!UZz9 z#SbAb?7k2T8JN>9c2Wd!6V;UIUWC3hLG~H{4_OLz(x$3O7Xs_;Xu{Ph9v5Mxl6EyB zgT|q&0-uXpy|77jBx#oZqrD$UAF42Ru=41%u9ng3St0}93}1fvn^keFMcf_uXxa9U zIq`-%Vq+snbF~>;!t)vN;WQpfmnzOPM%RL(IrD|ic*o?A(SJu8aHwGh1Oz~-s;aAP zGR(lSfFXr=Gw>R#=jG8c@OH(HS16Y#R`B{@IFd90H`^@>qT%j~t~i86PVFS_h8i=o z5#H{oShlZwwk_m7jQ*s)*@tCB-3A>f*1qRhA8@~4qjZw=!_JysAL;<_k=$`}IEm4_ z-Q51EIDs)5f{~OR9%UrM&951C6WoCOa8sV505aYWWC$IgoaDgh7f^k6Hvx{ za_YsLddnH_;E{b7$W;j?>orr&uMhZ>Euytcr}S~cxqa# zbGip}%t`9f&HsO?e*8j0LStT)DuX}EUCmB~J%~0mk4Y!m^&oZrC(_RspExPSN>yWA z>b$VhCVIMEDLU>b6XP?!7OOtF++QAkZL>c>cjY_boox$i8!)P z6x|Iwel#(vQseQOYro;+@aD*b>eCiGS{#os_t=RXSf4lo2#q^gT58>s=*ov|mu3V& zWwhdnVz>gVrnEc_R8-5PO>Ck(%PC<~QhS(SDaSDC`E{?T?{Ji`**b-A2_$%pLiqg+ zCLrqJ8mVncMYi~;e>jw3D-j&{KF7f0OhcYMr13iq?)4hpwihSc3#(!p;X<~!0k42kzQ7Vff zY|12(Ag5c>{4qU9*f^o%BXu=^Ed%JXb*vI%KB9NLut}(kDDqa(2!}?myE60@U{B!6 zcvKBg8(?52J@Neo?wT@bL0EF?+P`D0{@E|i5Aw{a8NHKyp&QNg`-jwCWpIM+hOIJC zxU0u)ePgXs*iDl0^$2})v1S$5c1D`CMtNk1I^*l+pdR&i=+74Cgfi8}v%l?sQ;61hDV^V)LWA^}UJ=e-rzR8mg}mvu zP~?i{XPWWzmg5-g+{TYhzTSUPW*K7fwZ!HfcEy12qMZ`>ojZNvqR#LS57XNWx}$vK{1e&vO=Qea=n9okaszOaRDpyZ?_@uOl}U2iy&}<` zAnp6Rjpa-t>2;(5`3yv`vqH!yOSEZ*tBh6!#9_#-*e9_g>cL^#?@s5T39vPGK z1*6pY&CkG*>XK|6n-R)OXD~=faGEY$sA|LiRhe2CFJSm)P~Kg0W^!5&QRpCeh_6SA zPR`gR<9EiwXywjtw_rk(=GLF^4jZ>6V9~{_UiFSD)z74a*4?3Jod+r_E8pMWkFDGf z4ao^xjQZYGBLAXbQp<;lD7y^*#}pJhr5~=*-5G)_I7S#DR^NoVOep%%l~wTMEant2 zoR*zqk~*v1PrAgLR1$ujML^z~8H=2R^NmvC*>EU^clI_hwi%(AUlJKj8MsCnIR0@_ z@Shy2ucV4q_ZMEUzbZ2W3P=MhbgK06wzsz%g2v$t4b=c2?B?T>>CB(eFos!;BW!WD z*6u;UCvE+DqX|$njRc%`Gd*FE@hg^<(Ay_Ff3HZzJl)}CDz7?JdJP5VBSY)PQV=uG z&Fc=iA!UpcHZ0*Jr)cX*(x$TJq;@&dXzW(iJl{PbKGCNNGVjAr^YU{ew@v9UGc%Tn zmFuTZ<(xjHSJQy-CgA`3F?98Nebmu16SS^Q+afLG97i-s2>dwz-`vnBmXbE+^~{)f z*4s>f+w=TT?=p(F=8g#yppv*h!l%I9qm$x(Wk~;Q84BeiKG~=qWK#5{uUUBKblA}h zr0kr>w4xJ^{QMnbL*8nmf~%-t|Ka{30>_%$Cyi0n+8Bio`9CKlpvq5Aiy~0FpKI(h z|KscJoncU-$hfn!b9M>V@Z<6Gf)^9&@NMWaPn8)E94(n3OBc^8Do=*4YKl42wS2gRpM9{1iJkRXCrcI8J=n;5(*k1?bqFyZ9^HjF?>~4MYNk` zWF8J0H8h{%Rj%5wdR=7By767tOihE-?tQE8#)r3HRN$@QlvobOTkMV`H6-E0rZK5c zoCl&^(AHaxVv|svlrEFQC>ju?jZ1M! z?0K+e6iI`pu1lJxm+Io1=eh7GxCN-RgY0&QJKqc8d#f!mWB*-=^}L@h6R~}qR#5Wi z)mqTT`w}9Xcp7h{hXHjYfSS8eDgY6Gtm9SKlMnih?w#3wC^4j3@cU60=AuyEF2Dee0(_t1xrAo zuHWiXFy%1y`#=#47MtzUgdsN~-Y=Y~_aodKAYKTS_jAk<5`W@UzN15;VuYCm zKggGX_eG=9lxy)D=t21_ckzm{mA59Em1;<9O?O#~2s0PNY_}~C5D>4P1__G~j001T zWurLo#3C=cNQP4fe@%>H6Qt)h#eHoEWOSH>!|M*0385nB#4N0vwNGlsouhCm9?zS> zc*4n2!G7Rs7qihFpo}5u9r3f-aq^CD^V|y>d{W)YdlUHtVxomIX4AC*xxsLaoAXWT z{}gjWmV0BVT1+%0LVyt`Q5h2Er=K0;sPWHO2>l#wzIBayj>X|9P#oO`Cwj|gxstF# zb?J-$pI;V$J+NW^F5&`UDvRfHS=FSJVqsgxG&=)x=>CO1k}bK(a)3{B3W?jXAcF?| zU>@F=UXiA2VD;;L!O1K#DqVuQED>8Q~IrL=R(uR;Ka;rNSir*@OQrbBpBLjydBQ6-yZe0UIs42A)-NbzI8 zGu*z)xLwPTkwoqSU$;duZl~mW@42rB9~qZ9lS618 zH+-2FV&(}~*4Ir~FT`-Zlm6!(cLt*A6IFo$m*!x$+T=iQg>hVV1E^#t&=(UWAvxxM z(OjN_2V}f;!bzzg>d>G{gY2t9fpu?>0&v3qZb0wP+?c=SqopD5zC*N<6XH?c5%9Y( zH>lw1l;cGeqk?LT;f=G9W&cD0RDc1A2DY^8F%S^0btuYGHscR{FXS+kg{sd1b8Y(v z`prv3J~aqCL*2$JiD$2E2|s$M?H9eH1LU(Djg+bU7b9i!A(uPyE51NP$`!(o?f_p0 zHK*mek}B456$Rl`fGe{BhF zzJ&R~g{ZG~v3F7k*-Cl91kd@I$%W>Ggv|+SIldB4PhNXwXj*iRJT1s}hwu6u18Cf{ zl13>f3m+nl!ucN~B8D?75T#@T%#1zJ2qLFN(k=Ge={&(d-KNeYX~@KvP=?^c)C6XO zN6+uiojoePKYJXh`5k{y`FIiM@~Za`6W4AqT7mEXW|{p9)st&bNmDu*EHBECIV?n) zt^Q`t|9(*cRk(KD9Zh-BDJH^|;;3%8>aAJE6Sf}vdZV|XKb)Wf1OmZTE=<(VUjb$G zYdq0maVkJX2oS0sl@^&#^Ou|4AKC3 z7f$3Xp+i1wj*TS5Jo+K8+Wo47_*sU2$gra6VCu~D&FrSe3)DgQ^gO=y`^ zi#uaX7!>)hENSg*o4XZh+mKJ$+fv>_Z570Yuk+Qh)?tu-bG1Duq@~Rg+1C$wI2o_u zk$yP*O%BOoOV_(;qBicrGH*=*gwWsTcj{L1Z-i9~Qs=m)8~%Hyhu!zz1tQBMZSqT2 z=X}kjv-rdGo!N;VIc4UI&L^GIc^RM@naIE%5oL#^uCD%t{Yuf9LuQAM?rl5+jY>WS-jOO7Nz-p16@}*c1XsK(r~Ue@OZhh@IZb*p@3s&; z`El0uB0FFt$mk`2LfADrgse3kVfGK%o58;APYh7~*1u@^Ln4iyg8qfxU`58GGnKz# z_>q8t2=<8ZC;qLwpb|dU)t?$1Mo^+jwgY*LtOS%dxa9H6#alYiceA5|LDE3A&sya? zXjZJyaFVegz#dB)(tncWpnqKBjBAKkNFXMGnu4M9^88Th9k!RSNM*$gznFqWFArG@ z`A6-Xs&Y9A|B6I0a^~?)DpAa3D68Tb{phRZM;b4FFH661UgoQPHyimT+E9*mFrY3;u%DkbEJe?j$g4YEozaLwCxc!*o%c8;y;Dk1;H0X47Y2O1_fEcg5czBj-%9#dL1D z=@8kL%OKBTtp&->f$s3o%;`(*#*N*THvHT%v4<`|sGxZtEZxy}on@ zKS#$O235~M72YBg*qS)!m;I3(P5u{;*85H))+88g@6)z|0g@TMH!m`9kMTak^bcY_#`$ z$g4}PA1>>wzECsALWRe9oq|>p2Dh1D@UQe>sKdDaX)D2pngdzrLMLrVNJfy!!fxWbToICJuMj#J*tWAX$9sxXjpw<&3p3{6XUM4Vj<$)HG9a-VHg*7m@c zM78H%E(USn$lUjN0+m8xdFvf*T#2ts$nUj*WDmvi_)6u$(1X4u#i6KAc)kkQIY|Uw z#VkoRvZ$y^vsiI|Fvd)i`Xw->Hh_^D{VtSI7L_XPj}`#;+XdU9^AjoT$`;6lg{cRE z-17KG-9>NgAO_}U$rjCrryd-3jG(C0Lx0*pk4|ADd~= z6W`un>NCfmqRH%H;JKv)3Bmh7%KmEHW$?#1X(6bRI%dR~6)<4R)QdQGg|K+-W+-wH zdjNB_2V!PXzEi0ks!2DC%|!ZmtgxTu-1*Z&7^86(nZS%v?7k!9Ilw7$ge~1np1|}& z8wPx~PYD)T84f{Xkofka6?XpK{l=Rjv?3i%XjpVx;YL-Oacvv-Fm!)o(!#|WIu;56 z5nP&$_!dCQ8Lhq)fi~LU)TL+XgFiF~+p7M9YO4B9g>>HQ*Ii3hA#q?L$drK)FDEa#4a}rvGcPfvh=qv| z_aGJ0_DRJ{$V~6B2I7{hWb|A{T*xqLKN#|6b4R-<-Tl9w&N3jX@B8{nw@4$>LrQm- z#0)JUlF}uObayx6fRu!w(j_6?-O`Onr*!@get*w1FXrV8bMEEbv-e)>v-*i-5Tv+{ zPoM~eBYm^z=AN7-+=t(zR_yUi1GZGZ(0@JrP7?&t134<+V<7~MN7SA|Ctc6u1TV&O zaQYUm9=ouReNNUF*5JO!*G2r%tL?A2Jp3F&L-$;%SchhKy@AoC@OjDq4)=;JhGzM_ zra#DZK2V}WC*-zuD0-b7ot+PtsBZUTg7FoF*3NQG(n?Py9j*fT8g*?eV!8B5p*PDe znD9&INbVpbCS!LVMTd~KLa8icfRENI5)!a4knT9F**he;%7RqxK7>3|JGeVbDUb9D zYRIwsQcn_topqlM+?WTma~HqOwLjF}kuKJ<%(69w-0L$(dcTRJExk|*SrX$@kmsdU zGh{KnS7`W{)02Fju%Y(&Lp>8zDx3!Wp^Y<)pqQ#eX`qx2p`%$VlePQ+G#K^GP?ElW_mHye5&(`KBP@Ojn zTY1glMw%q`?;9|nO8DPZb*>Z`zDW_heGB#o$zBwvmXi~={rO&Ltmq?>${ zgj$UH+a^3k*TVH~?;`ue;c|Q8FHBmCsKd{i(?uUc%K~rrL@%=2-nch+2m?d^dZQ2E zB|IvoLXxzzz@_d%8zbj5k@9Oew{v_jSV7ANedm4FN1;-xRYcu`Yb(^#zRGA3O5L15 z)e8Aaw&Nj*!lsLs=#WjdSeudz`3(t)KTvyh<5c)e)3Y?>eL5zM8Vzm|Ud-kPMkIQa zHu2skt+3XM+cV0Z_jjx}2Mr`{Z{ve^`WB!MT?VTlkKYgKp47cq&46y?N}AiVIeRWz ztiKa47#m_A$SkjmVK^yti4jH>3j4;<8*+H7tWW>A;JDm(NT_OS|iHu&nYQ9q$ULL)6u90ZAe}wqq^kV1CSud_60?@1kXCn#&$z=SrlZT~J`|TD|Yfaqrx+IgSlH$>>UbLxl2AM~CVW zb+mtUKAG~9Tu*B@$-I`j@vQtrg&J~*0&A;7V0?c|Hjm>c`rVruoKPBAHWz-@ur#Fd zdM)M#9+|Vy$kfo~M-F`JiVbWl#6sC**QtV<`{yC=86A22!YD1fJX$YOTEJ zKtlVGNp&^8C&#!8QjA;Mv|m2m_mKofa!YPY+5;-y?bsW(DO!Si&B43+Fi4(kntb77 z{p7qCuX2)|P2Mf++CPXy=WS!Dl&a{E42Y%IMH0W2nUSW^i(%Vtapve3^_MBXyapvK z>OH|k3b{Xjy=RU;71~nl(Mv~3qQ%Xv7~|FV2Yc>Fo%w6szjO$QI}Dc{Sieoa}^1*p2aqA38q7pf!)4DgPJN$=$nM2L|dFV|wqbl6j`at{mQ1-3Cdt@t#! z+$}^ec?=Orbm}Irq#bU*xzKnx5@<`D(-7qu^U+Ri`RV(dhtJJxfEQQecwC$R7uiM#`u8_jjpZ_93b)6!VDeXdVneudO$OYm5y0u%zO+R!hHNGH z!yf$;|M!rCe!mED-)r=mA^S01{(Tj^Dl3UB)|H;|Ui_@MlXhOX^wxI@ykePKnbw}A zZq$&g%g+P5S_E0if#iH{>}LNAzZ-h z@@Q@8$qSDm)2BZ&UIYgQa@=fiXb@LEG-E#>c6?5I`nUCba}<%2lk;k_0CDo*5^y0+ zTUGv4z3FK*q^Cl;tU#M7{CK-y}J7Eh$Ksm7s(kHW6ZFGw?*{_m3tP+j{lE;1@XP!Cs z>o?yt-r{{gO$bgVUR5n>WCqQE>8?Xo$N9i)i(Ene7&3LR`%jbPv$jT(>S2T$T+uwId}H|pxRO9Ule z_tGV#BuhK4pJ57G_-KuYXyADyWuIRN@UxqyBM_cuNF`XvLQE3l2|V##-o2JdLh)-` z-R866z1Z$o8#C)Yo3!&j+Y0-t3GnW_aS`Os?cw)Az2&-7%1Lv^k4_r4f{p58sOgL=rZ2ZD#8`{3(mry+4wpJISNBNNn-Hv(u$XD9GTix5bcy?l%7U#vR-1 zz9DYV@M{h<`cDNKV*nngXH zXC4DX`zsdr&BQCWZPfB#yj0SxXr@QlPo*RmQ-W`cdi)?jCDT}=yeX(pE zZ}qzKYD@FCnV*IZ_q(YJnX`N8-S!}8pn)m#v!ZB;OyxT*pVLJaOh+b(m>6IL#i!Gc zjl@1|ecJDZC|Bb7WhEv7mQ#w7?I>UAJsJM1!?^IA@1$ZUPLpVFw5hm+ z;$Ox@-$5+|W+=mer0eA1mri0B_3a$(tM~{U+iivQ^dkh``wMN~fOH8pK?xsIlR{T~K#{g~2u)=`{02FiHOqo~QpP=97 zDU6Bxy-sTOaiY z$HYjwBq0fem}=ha1+U(ZBIxtwmR&ffKHh3?RgBw?Ku66zX>>*-?*F+ro-=^QT5`=u zs1AO-B(+-*sX6FyQXxhvDJo`uOMB%?!f%)E4A0*gEmX5nTItOOs{ld&fTRGy#B0$g z;4Qq5ejNCGAXNx0V~$s93Ma;KZ@@G4kMl-os?%keBzc+LB@$1G!ArbQzZfol?LYh93iWiM{Dx|vyF-u79{c3lgP zKnGiyT#s#sP2>9oHoy-DGu|tp;s4GyOPaM*t@#AO1~PrQ#%4c!A2>q9N5RD4b^hue zZY~Pj9}Lt~S0zrA31~OfGAwCDt<#m?MmE=QW%tqL7Be$1O!Fw8BOZJVjKd$!dR3ry zG$V!0=4DP#_HXaRsKbtpjpb^`ippQtAU)GIS?!3Zn4SePr!U7}53gPV#E%jliUtQ(&rK<*aZKFN5X1u9>6;%4s%Ax`bY6Qi3Lo#8S?o5coC3l6Y8Bo$1x6j5nG3 zSsT>zXD{u=EfSZXe+qp@#?!;ijQmnLfA#wtp^ppl?ZTDSjC}e*xyoByh>4uavyhw~ zzsEh|JV-ug9V>}Hi7-n69NQVsS35T#D?vu;;r5cG<_`$T0>VFN0!@nG`+P;eE1FiV zaKLhYi5CK}BRjf;Z{L7s4-KsC1(RS>u&J$(Y`QWaVh1uRjx4v=t-eoJrZ-TOk>75_ z@Y8w5ajAuMjD;bY8L;B+=EQOwt0K`F0ZZ{-StR)Wzw3w0@=3EqPf(*+hY*lvuO^8x z?d|O+zQ4uHA4dj{0tP5N!9ET`U70E}gE2bZl^Rhf zbJbip)O>;#QI86F;slA}y2w^l?cR#`8l*;V!K=d+RW%=44q%nhHPtBiY)Hhece(ZI zoncE?PA!9jN{PXx&39LI<~nNbO@{GTADGl3FdqsgHv?G_nMB+K1!!B44~>5Nm}4g` z`D4x}rJJZS1uK!d%+HZW9~wqLMO6GbZ4?YgRcjrL6jL}P&G|iN9JVKz9!wlx4Vs&1 zFiM$na&k<|M1~#hI>%|zX8|*fWIip!>Gl~c5q|eX^S{q=p9N>jcnkq+8i&a{vrZu6 z`XW3508W3^4hq`VT8D?9fJx;`plCj2t$U&p2g$PzLD(zAM2O||5v6M-g~p8ExBj=Z z7%KSX<>e$|-U59en1aCYeOuSyZ9HTA$iz2wR4_ZZnuI@rMZn7H%)&6k*w!XJ)RC@3 zH{p!GgM)*SJbYEs1+wiuoCLJ%m-~*aYkjxrZ{`i|3^xnp@WP$3)M zxJXxdOm0*U%ZnURfL%{&yVRs#Pny7(SJ3?IV9umFc4b)nzT?ke+9@mhlN?&(GW%af ziu*#Cjqk9fb`FGd)qgFL%s3%vD(j*CoMY!Jwy~3tcf!Ca6%W`F-N^P^$i6u5w%a^> z{{+ic2t`3QwdJPbOVdVwSb+jw_=yIUQhZ3&hbaY1rdLS+?qSK5RR4YQ2N{(-oDKWM zi%Atv-j)<}3k!=n;81J}Y~%d4rpUuIM?~(o6(Wd8Y`js?sKskWj>7*eCe%!4tvNK; z!M)wlecG$FbppMq#CPOfmc7i*6uE+2k@Osf{%DS#m*k|m@lH<0w1CBnP+rQYKL#~& zaodPx7M+q8wHGjjpwbIPWp6=dFL9rL1J8fwUo{0m<8YWiv;JDjJne&UP+ol(uw}NO3_nUEdE@4Kdh3Ep+ zf75+gF;oJ#7B+G&;d`7Fok+DhKdE`B6!SHeA(-7#i~&i+D4j( zRHH4$D8l&3!|2yL@%RAJlY4q@&IB;yHkSgwKzEI{Kel;ABjItLpbdPdGJI0=@Sqzz>dmnuuKo}!#PrJ~}H@rYS z+*;&|IkLOsI(A&C-Ba{K`%T!Tmo2R@aP^E{kSi@xmmoet?fhFD8AuT z+wJCS6WrOG5g8pc^K*U1WIq~uArzl?II=&p>T?y6o67j# zBi2QN#91tV&y-Z%*lYal(v>%`)k=~T)1tAkeAFio(4!7~PPZ|H_djV920O7U+hQ2s z`p?fkU-ktmj*vppWL$pbVMB=^l7*uf$A^@&-b=^~Zs{`N248&%9Cqf(6*qbG&#(`% zKmxV-M;&f6+IIx^sMk}^U;W;rb~xR{KKU+&dn#rZ%lg;V_$ZvvQa?LM!NZR=-?n_Cnl3nQ0z+PP;+6!A+t6 zMG6v@y%@z{lMQNAq2@@QrK9s-*tg{*$cX;e6^3F*%fIE(hNdWqrO$uj4E9weJ?^hU zsJ<>Qj8@{PCH4?-&oRizCi&$WxO?AlL;`8046jRySuKJKyqCUuD`m6%YZp!i^He;u z6|AWPlR{%T+hNHSGI17)aP-7h5b=P4{B^N=mU)L!+{BV>pmZH&AddnX4N%-|v z4DC*w0vtm5Mm#mM(s725WZsm=#$KqUa|rwPo-~y7_75$KfL_TPYOevZDkN05#*ei`B{IJOK!JmFUDU0JW+w6ss{{f>B*h zuo}nZ-5Z(32K$^2FZ=#=U{WFgZeWQIOJ=|~N+=UW&^9dY8+E+&A^BG9_BSGr?IFXW z$4`<^d_@t{{x6KZh@5`j4KZ+nG@XhiU3gR9@`-mSol<@UoI5BO9`?!McD6l0Hu$_FgIhf7nEGjt|xZyWXL`_1{)ELSC(k`kM zg#Qu!5a*Xfu2OTsSfFgYXU(B=x})+-cPXKhNrwOKs}P1qgyzMXQ9V6Spum&J9i`#~ zcypCiHw^XCKfm$ss7&AN#>_><#)cLa76Mm(dOC|A_a5E#d%+qjmaYs7_^`f5=x?ud;2^ zHEHCBl*P$+LA1gC>h9TZXwDnfjf`%w!19RTb|aY~ASl?HDNqM;v#G+536KolZf+ zWv=U2Bu}``mY6Jc&J3`=3MoNYm zt{H5is55IJ4+D9f?QKvAI@ zd%JP)4(&()6Xa%$Rw{5y(zBVxBl*f+=zUm&(Zn9;((J2xfnXNdv_CF~f70wD&vG6d zeF71&t5|cZXPtvzHIB#{j=IP$a&|2oc;*uWVEszIH%M_48(r;=yW2ST_w`rR_^LC+G{ zwiXz<7K%-F2(%+PHZv6$KrHp@r>pJ5aTT0vIsp~sH&@eNz|qO0JUyfu<>2N0bwBnr z5m4}QfDDLHJs->SWHSJKLXJ-LJD}EhuAC`?&d$zWM|t^oL#{Ot`RMk1X$?d}$`mMP z7~0yhCXWD)S`NTLJ{%f%lh_T!z(iE0(8q`29qSiZY&>O`g*o2)2cF&W)cj;@UI8zXUHy0p-l$q%N|9^Eu5- zll_jKhuTBVm8!Y04S>Pj*=Zd-VaX7F4)|}lWJfK47|RWn^6zJIL=`ez&*2QuB})Rd z$)jeP?q^0UX7s8^@0x7Z?;x#A5w>n&YqV$3E4PGnRk|9pBRaVD7%5)1H z*F)NWj@G&rkkjSwfQF6X+Gepqo6S!P0)ZF;*-#3nN$AbsDW$ltUP+i!8V9C?A1RUVK~^TJ8BU(8ns7IQrRC;afP4tq4&Q zg)eDv%Wfj_)eWy!7v1!tC+Ca{8pYB~rA7bRHAo`f(RBfCzAT9D#?V$|b3%0E{oQp% z%!U!@pC4Zw3r$yAJh?jCdlAQqj)-~gm;og7$LT*TM`^mh&s6CFax5aH??tvqSQiyK zEo@V*TIB~M;BjAhLixo;JCT;4PloT7zQ8BJe)ssRRgtpzj8i3H$kq$CYxGtA`a#Hz zehxa@^QJ^cR&jO_K^ShN+Y$w9!ucRS@+ph>e+pm^8F;72mulH z^~j435~|moH`@67;H>ywGW}vuOlq(R4uNPrju3)VoAky)+~+9UQIcRUl{yD2_hb&C4TSi z1sMlbqax z=^BYGzw}_7Js`zzeD7{NtVEX;`D-k!teB#D9DOw2Y43zDyIKWM!5*-4Hn}-q=X!Oy z;2(T}l&V>ISHNoYeyxN?{O7$C2x8}N(C=uH>*`5C@`@TreUC=0!0&vKjY@=N5ZL60 zEt!UqQSNxA3E3QXtvGf9XMf}MJQt@w*D|2i#zw{WpsCUqotl$;|ATSuF2Sls~0SxtH#=GgB{ZqEE$Q*D1+ z*L!5N!%1!4bQb-Qzr&J=EF7%ce>;*uLoV0+RQ_0E>aMZ^OQ%TzPjNH{W?NUJ*6E~1 z${!yrFYX%zdS5G322F5B)G9_OoT_eU7bc|>ovK$LIVQ8T3k`%*p*wTd`xxOr5J%av zcs;0aY_h`euRj{^eEQ~ouuwyp35ky&`({dL=k-5J9?9Ohl=hT;8!puWYD6V`2 zm(E^(O^JM@{_-$;@=go+cyMOacO0 zzGG$~w*o-}sT%<4OkALn^--Q;htGLu5*q}gbUY-gfcm>X)vvyI1VdsNo%vc1RbRiB zf;nD0+5N$k8%q}rL+C{%oC2{^@;SJaes9o+VrQqAd zFyxo;UI{ZYbzQKyC7SyeHB$U~fen%z7#6|8(nq7 zLTVNmnpM?yYEp9|I>z~BhlS^=VNkU5D3cc)UB13mlEZs;`xiCmmJKuUv-&8Gqdk@O-ux z>k4(o?rb!X1e#yMj=_R!s9Gg_)vYNxNy-xU^XO#HOvk``_{mtQyc&~v!>xs>rLYuu0X z`+8y2oJH{dUj&JquC?n$uY)aJ=We*2+0Ac9?&siKK(Pc)1bCQ4Eb)%ikSOPIu2y2qJ^3?wB>VV#7*LcIk}sPi_0LxlQYr z%in6ru*s$+nOB#5)QO&o5p%JiTaB7Cs&AWo+XdWco#??`yAFS&yjje>+3o!+#+ig* z9=V&Ta|7?gyPs06>};%$0dc+$%$Uynk1%W4ZMxuQ3VmV=Lc9U=v(JSmhutj`CE(vG zLjRAZ|HP5Wn-T+O-V}0y};|$ z*Tf`ryiO;rznguN&quCmFR-gb{uKXPe{e@iux*rU_MH8#ssG~%MC zvt*k@#MQ(6we$Gk)JEdZ``3e+KY2HLVN!{_`+75x56BSIxvM{H98tK5wkt+%8b5qa zR=fZlUMVpkNcc}bA8D;3SHfpnf>flE6kZR1o{e9ne}$#yk*yWV1CnNZ^~CB>h*i)8 zn^0PFM5%v=>QN@&VRmc6xLRA*e<^gy7s~2wDxGXC;vr-F?w+h3@2K}FUkHcP&OIkG zdA@AgBpw9m;k&~Q6hW%XkPPzlqms!ghtC5lBosUq+ z&#cT0-?ug`T2}BNl}DrwqUZZsG4jzO^YUBc5)|mg{_p-k(KrK*X8NUd4dyM)59DCq zj(Eh4KdKaBClfl21gzP&;Jgk?Y?<**e6Rsq(4cFuU8dQ=uFntPw(0BnApy4YvcTyA zx%|A^dz(|yTN|E!_!VNLQzLkB9Yn7s;4Royl@WZ_eA2SrOJ=58^ms~nF}`xf&u^M! zuhZm|0VVbh6p3Z}vf)zEn? zNWezIzKN@=YlXdQk363HTd=jiU`765-+W619DE;0%iM8DwT5iecU3Bu_xDQ-#`Sk~ zDkzKl-9ApHL}XZecAWXa74%`<%(`5opnoP!(9k(8;_F`{d>+ObjP_4)oo&?pqm@?@ z&liiL3c(kUT!@t1YHX>=DvbPls>^>D{Wwg?WU;GVxI7VH1_BlF?`&T`G4Q@4B^oBB zFGdSsE8wt`cJqkuk8~|06Ru=jlIl38y6<;lxpus{%q{#J!IvB?&3xbq!}!VcvG~I+ zl<|)eWi)QHf`HdF(q4yq7o{a$ak){aG>_dOg9x%!1X?5OLaac(Lv|pSlJ(bV3dB#O z-h*d!@59!Pzg2Ttt1s=dn$kn?B7XY4KZ6);{UT@(i{bWmH8mqsiC%I*RkZxpR+nl5(dnh zeRs67(LD}6Mz}VOo11*k^y@R^HBcOPv6RyIDI zodWv=LT_EaIa1Uq@}N^}-6IbI+ZWEQo0Cap)^)K1NS;X9wwEOWZWlQ>AYs+nNFLu0 zRLx$aCs98@>pV^CAPQYCUrHIyvbF{-BSM~YFY(YruYXU90B2j2agL68$ri81KnfVu1M3KF5@GK!yjRN<$fNck!1z9;`$_Ux| zvn9NFp`fC;%SGGHb<}X^3`LM(9Nmab`K;Sb4^$_j4-<9F4;k~#8f*&q@=b%q3n>Ei z3_>;5jNCMw@d9F-I7|(cV6)2lf{r%qYpx$ldR=1*(H>h*4#pS0XOR(MK8LJw!?VpyH{QJE;9C-9wC znaskrJ^atw7vw;XtF4@gVjXdJp+)Q~pQp=?<}rcnO&3y$Ld-`H6%n!Yy#gr}r|$Q? zKOWc@IdV05MazObT4{_BiqPHHg01m1$vbpUdvUJcJeTUX7h0W&{mF?5+O(w2MEZUQ zVW8yAR$pIC;dfwaNcZ{egxB}pi!G;6UbuGM`H* z@qZ=TTDIAX{9>Y+J$K;4nLWMG*b;hS%W#^Fes_bX3f{~ZyE$7xE z?~gqK$-}E`k2;*`#$5^7z2J=qFoMLA>h~{qS*}h32$C`u9wfS)mXF`|R3RWk;k+*;$A{oJ~mW(948(@MN}hI$Ysn<+-!!_A05sT7BAf~V^JuEJSK z?SGFdcijDSe)yJhs*2y~Z)~;JbJw$gtTgfGgBgxCe_MV@o*a4LTs z;`0BP>S34BNQ+s`B=YQL{?<%U?sizm_-^^jv4E%GWsmVYJ68&UpBpsvm|RnTu7^_Y zm()@W2GTtu*24HHsp<37dvW^il_^g$-DGe|%gx_tW~_qVSNvQg~B)*}u- zi4^P$sMp1=#MX%EGZb5vtc_o}DvR9dB?ZgySA`Ge5B=)}gEuMHvU+pc4+G_vTKUN* zew33oMH)KSQ>39kmfTv49%Gso1!^W~GohbhCcF+x^ZkX>U6V0d!z_XW?a2=}nYs8o zX?d>h2Bz4WI@p$t*&*i8wb8PABU)DeE<;`rP!2tg#gIDZTIPgG84}&!%rgomA<>A@ zkxvF})vjdV4EKb+InkF=cU<5mjHR}MPBI7nS0{cxv~;NX+fvs^YqoeEBoorH+O^(aFj|_G_JXilgwro999?Jr`&E>C+oyRx}h5 zA=pT8RCAAf0i#8s#0CmA%)LC}?tc;l3_)rpPW0>mqY4D>(aOC`WxA^h&9oM zbpa2YF1M6OS8>;*4MW+3%R}eG$vEF5V5E=#ga)bKNj4jwXQ;~Q5Js4V2dzH0~bl!-C^igjRAxr2(l?jC#auz + + + + + image/svg+xml + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg b/open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg new file mode 100644 index 000000000..8542c6e3d --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg @@ -0,0 +1,351 @@ + + + + + + image/svg+xml + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/open-machine-learning-jupyter-book/llm/image/scaling-laws.png b/open-machine-learning-jupyter-book/llm/image/scaling-laws.png new file mode 100644 index 0000000000000000000000000000000000000000..c501579591b5f507dac8a7b9ae3cab964b9d48fc GIT binary patch literal 86847 zcmZ6z2UJth)-_6z9(sV#MU*DJgCX?Zr5AyK2uK%^-g_5OI)oxkkt$8;9YF#jO^^zFAMgKt@7?zpgKPb^Bj&T!cIb95MXhwwS)A$r9!>}aFcSas z^S?utx^r>u|NCnuT{$JPkpDhom)dgM^?%RCo?(LSKF03n8I9Y(fDE*4wtH}`me`I8GL8? z7_%I97xJ;JY-Ep9J$1M%f40_E{=(|Nul+LtI}}?R8^P5p^)xmKIPUe$jrZAKKTc`s zIk5VF|7D7VZ#ih0J?uInP5yT7`iJD$U-RfcHIWwca zL!@9V_;9H8#qG~^lmvV_aKT~LLj)eU_9o>^t-9_j9rUU7c!-6EQ?-K(Y|rX&bVkN| zP15?sH`{XGR{6b?4Hh>yPni)}dW4;4LomCWz^qYUzDMgS;raP_HtgxN{>ztqC8ecl zWwIt}P$)-9NeLArBX5-1zrZx6NK&!&`6(gF)^sv9n^B1QSRqz$9^7`h`&q zmY#3%bjH;Cp4Ud@3Ef(LBePn>f18;b`F7ZR9|MSWGT5j;coP`AVqg$0d-mM_o{C+I zA5VKEXT5Wt>=s+t!-rfPCI#xzvXnJ@v@|K_(JIvo2T5UO-=K}(y8^nb=?Ub03$F97 z<~Qjgy?g=d`83lv^9TkCGbB}i|09+>cmiUhq?s%sQMvlh9VvrnwJ|L`JuJ;)Gj?;_ z=4FB1z0d33R_jdCX$ZGAd55*OAR={W{WrD>{D@a_?X}*!AP-3K+&VKCqR#>4S;*O< zZD?d2Tw)#f9p(Z#8#s&FJH9OaE#ZGJ_t2u+GHYSDc^B;8R0* z-4(UFggysn)^}|7pE-7gwX*ar$>#fy4V$maB{qC7EM%HdS>K4Fgb@4r`Js(I1r`&W zCvX>bD^tonOuu)!BKCT&4*GNjgNDlw1tB})7;OA;nWa3$Ym5IaU#mJbJKmLr=F9gqnhahmdGnQtzdHYbuNQ>J^Mgn5} zpURp!GKj8+Wa_ldRWXh)W$`{9;%lLGM)^_kWBsPfD$lnTPW@R8zbEMlDQIZ1k!q&} zO&%wW*1rk^Y_+M~mz#Z_uQ4pBhWIB{BpPE$seSaV#}aGqMC);iO%a{YV=Vb|#Q z;NA4AUQq(@UHsSJwoFVrO*Mon$>il)z@z634|JYcg^o~L#9eK4`h&Io^%P%1(p)uN zE)0D=kEfdh48k609XLce;n|BGZ<>!Tme$*ci3hJWP=6>)6Seu8(^{=GCGla7LVv51 zQq9R;J)8y-Fq9%*wOe@aGkU}KfLuFD-I$A1>+xaK(ALEkF!zvQwf;O5l^dWV2p_Hm z{poE?`pN&^b**EwBViO=&+4EDtqQn1Rd#6jxrbW*@%Bt3gyO+;#}j))Go`7zfQ@wZ zIzOZh`txqgfjYCq%ZD>|iJ?Ig_kRp$n0z+XB!Az~AbO3N(zlnRASM>BHKDEo!7_O$&6xy*SlktCrZon^NeT9VO{Tr36YL(n7f8^MN!Cn_@FLU9oE^}59s6+WzI8z#^`lJmhZV|rxM^gkK|eG*mpwf< ziQhN>%Wig%($lmAHPXPYbN=UiM?uCn2*m83B_i3 z;+@8nl^7H_sb~o*W&G~$-j(}lI=fzb$gq3{^tHWSZ_xL@`|%^~M}YdLAdk*`+7%h& zrc}k;3$YhSpAr}U-|r{ZKJ^1gPr*? zAHH5Ep|habXG2rzn2*@SwBx3y*80*tRq)rplK%T1xz=`z=Q7*hZ@yw(pICyFU=IXD zzAHqVx6^#jEIDv?5!4ExA4__-SD17}x-Ls(kY9v}$QHbbjJn9}Gj1UO> ze5MX9_lb|!eV1+qkoTns{%KU?ecyn@E%8>TDH)L!*h(B)Z($xLOh2-lpAc{J$eGuT zB<7OzOoRUo8G{IO={;Ej-ZR?FSLP~DxNAIjA}YKXBEHPj=j+onfsJJHQtoQQT znWb3qehW=nNP{c|QrpIG(}s{qB=WbnKT@3ODXo&cyga*5=XuF*-eNPDgccv|WwV!$ z4{fo&P?vzcy}imt;{mUc9nba2|IB!>N5}Q?=(|u0tI7YFb+zqA10dR3PEPKp58qJX z1ov-IE0Fc$iDy_*4J$h0sxX-Up$PGOGa;ae|o^tsBkD#3Kq}J zrMko?qEPv+zc8}g5qKgiP`upc<0NAii?Di^l@*sG=43zomc#ki$As4s*gBEByY{7J zsdsH4>1XXr_J4piH@B#VID|g+$M7&!M6jqa7luUjiQJ!4rJre;ENVx^lRST>>3GBl zo7CT*3-nv2fmC=oUXmpmB^n=Q+4yL0kCfl9Aj)GQ^D{&A_>xm=)w_|(z)kHt48RBq zB8P{;1t5f8DUK%*!FDeGWcGJ49a zXC7u3IYM^Gl%L#DT*ducS@@80ie95lHr(kop=3}_`(><1I3exm-l~Yp+oGdrHN2X> zN`N*#K>}P@)xnr+JiwM`1)X>hbW754%eAhh%9&UWefn4FZlx#cmrL|Cnk43y)o;8_{sbBQ`Sa(1aqsNM%nS_NquH~y|B!o?-tOlKMC|klbk1zQB+@XpXoh+Y$UkRK;OZ<>(96b zL6KKMCnrcSLepMBxcACM+|C?4MaMJtMQ(GEot$*W=B3f|NSaSCgK7?0Ga|_2><5Iq z(Rn*8$pyRSYM~vD=V2u@Ei{ClP~*Dj*r=a2-qKWsTo3XMtkGbX`zw%Vc&*5NQv3W# zf>-49#b>}ftB|!-wp>0Xd0P^%L|+l{VJ0pB<&(a%f%uZc0;rdWaDSWcdDkQH`zX2M z)sGEG?ko`{#RqI-vz0Eb#uXDzsUf480!KfkmhIq$ax|fLI~6c3?JnQFnbqzH{4I$C z>@q|3Q99gQo@~i9L$@J5&Q1TirD@F&;HCFVowGAOp=;@BI&{ABZ~u~2m1_|`eh9=S zdD89X6EQ9rVP zSRL-=hSsMwbSLZM^tiZRK~-j6ZU%cWk*!{=gR5;g%Bzy_dfZ|B3Bt(u7pTWNWNV+6 z^jV_Cl*o@w@C<8mx^+lmAgNtOxR8RsC?|k|Q<77E%OdEc%t~;3;vB8F+v(Xr{uzE6buDu+)JBd#8Xf_HUIDAQdI}#cg#>=#J>~6A zF#G}=^}k2DB=bah;7T7UMTR!4tk?jKbO9&C?`MNiL-Gb|x@K3aZI(*45CpRcYDd$F z6xZi0$3uANkuew@S0YMPlWAeY&3TU}`yi$V%~WthTd!U4nd8%i5P@td1t_#j2yDXI zW#8E)z=?-s|9l}v8cvM|MDj0Y75afCx}SqaKDb{*jV7oO*ko)5OwIN9>abm{qXk5I z@5tAaP3KdED=6s-#ufcz8B7RUF}uy zsr>|a4(3V`ql2jD5~}KwgqZfS%LRTlUhdlc^mziyRqQXe8D+*qM{B6wjCZ}xoCPp^ zKbf&AOQCygEq`~7cAhLD@c4cp319=8&(v?}#Dr>E$mIX_3;i*J(@Mz2s@+qSHW8#M zM31y^bF0So&D`jIq>8zDcCh-s;c|57OFdjm@`dVqw=cn^{kQ8{+hLM#zB5Ugup~{Z z+ppW}w+UCPO}$<5nfylF5J7+WM%tv<_0N9QQpJO2NI(i@D*Kn~ybv z8RBFrd#VwFE^hOY40*iJtEU)!k~Nd7lDb7XyeCob^eVfBVkw&ROKZX<<}bKvZwn`Y zBWc_~loaQ(hho>EPN=kqgfvGThU`9=*&vnmFU5IBN?j)pE-wJrp+5vi z{?@_`wijqcuv8q`&D7fK7|3cG)dFI&(UthQk$IQqxzS=w8T|uD1q+e=E9#x!lyfLP zs(NDeA0NVvKcy4Qf!2N+gqY@3cx}2ixvm;Km*sjwsL?9p<2fbScfT6PltLG4n^4a+IyGBO zF}oEx3#fX9`>r6jqeP>UKfe#*kNK2$@%B;r;p)#V2`F2rU%=5#;(Q=4Hh9zk#@}5$ znTkoGm6e>JeHl2iCTDi`OW|_svwK@AE+=XBS(i2`P3A~wYiilhQbRLt$3c1i9j(tv zTUb}C+XY}Q>MWO#s9eCDp7U$$Al_1#4QDAowr=FoTu%SYA-)MdVRbDdY|}rLX8ZQ% zwu_agVfWmST|-|#%gZE1&4GQZ{KbnGxwa-BA*3Pi$_Iham;Bu#Tag9D&PxvDON^Za zha4kY61)F!rXJmHT?(_57Sn?Q%FHSV8G~RywyRBp%imdNet2)3{6mg4yL0%#;I*Dr zxs>JF!GS>2+doaNabK)jet>Uh#9qByz+waPXWw2|p%!1geyxHfJ0^>jiTEuV`oU)^ zUnk_r$5pnilx_x5H^X18sa>l*^n2*@-zOaq;Y~Cl4PIXzx?lgjmE=GC1v;Ww=7bzm zMZQC)mEpRbX#<78aocEl?vkzzzMVN2uxF4#1S}=0@MAkrd^H}lNt?$(jSp7uZOn|veKg2;W7i!t=u_GgSs2i{)yXf1ekx=8 z48IiU{JTg*QEBhzXadN_naW_a+KSS$H!kKP)D_G4|`a-8kR}AngGpAM0ZL&30^~wp5@!%BkGmm>cz3cAQYKwVNse;s%;qU7yU68~}pk<4P z0Gy3&nWmY~W2s%H^L&jAgIs3U%@vOb3-8*P)-RBya~Mh;X!Txyc;bDIT5b5*?7r|Z zZ*nY0tUmbSSPSN4xeX(9(%xn^pkk&~bx62`)GTZ~l&Q@$x=Zs=YU^{2m;VY<)C5$S zt;uqzx9-Jb*0PlfO0w26hl;lmk~1IHdClDeP8ccR z>pmei@|@Hx(-%&>=?C=rk#vWKuk?sYe+@WzY%CpXJE`!29{gn@eX~SnLsWV%{wt61 zf=VTNqpi^v(s$0t0qy1?O7yp3NTH5K!fkwo7>^VvT{=r+iRQW(>Rwxm_u=KCB{7{O&F}3+_yX(Cg?vmH4gpnuQC3*V zpn^=!Y-V8h{AR@|(N}W(g`wDChH6nYQB$Vy&BF%|)}mQl$DIO6&!7s$y*`L<`TR){ ze@bMdAxJuyKA6&BEx`CqN+#--kdW|46CJ~Xe#;XpC0O()k=3mAg30PVzGU-*nBdPx zX3r`F$wETeu$;Qp!N5IULS7*G`Lhz7XH73nVvwL!r;H_!(Pe08nl$`;{L#aQ!;pDj zr_=rU2M<3-F4Hw{lo{M0yY5z2SG_H}elEr?hY1M@$%KW%+2x@Q zv9iKJs?7F$1$P1J>Q6_IpWAZhGaDHHv+r*{jH^bUUG@;MgEQBh734;|^Qq-f2QABN zv>3u*+g1VC)JDhIQpVZ)bZryWbC=>UrhNBePqNTrih#ggEV1_3tfsfX?{5lo?Kdht zW%hx|xs#&mGMutEj@nP?S4&$8BHu>7S9(9aHspTFASB*_!~`4te)VJcF%9*1DRET* zF0dm)%ixftxcna-fN+1`GOl@2GO;{dQ`F6rAgx~QWE4;=4IiKNG zOq8ScI?{goX^Lerbzvp$sSm@uOB^Dy|77qtJw<1Us-k8!)-Hu)`*Zd9a4;HR6S(u} z=wS+r%&+zh%<3(&4Z(H+X>nhmmxEvoDB>u4cu9ldbi5m!%;?VNZ^ zt({mc0?kAxlcxlGUUg*GH9+(m?e!=$f6KD=_;jyf!nWQ%M*0XmZQk{*Na3UUe8Uxa z7sK6I*d6y!p{QeGKuyAO=#_3~OQwxS;^p_lZ`zQdtJ8zN0>35RB}e+Xj*G3l&7hUt z-Iq33Dh-W|lb=i;yNqhz^dqkZ7;N2I@bw+t=IRt{5I-K*ipwZmok#4YMurr|+{tLbjX3^EXuc&LrW-BJRF{!mD@q?Iy9+YA44Y4_U z_xmu9=mxQ`ABr!4<-kqgM*z%;TAGPck}Xlr^mVGq7e)c9OnaIZAHP`5c=?zoDa62L zOrlQ^eOsx9B;Pr+m(0SHMQlH9sg_*Fc{JM_HGo09*k3ymA|!6YTDK*}+R0?85&R@(8V+-kIpJ5Bre)vkEK-q-jdQifGLGbr)Po{m zHM4lthi_U@5tn+FxLmeZ^Ih+db=B#>xFf}4-a-`gLPYN{&VDKLC|v2L5e}i9?6I~; zv{j1GPNRq61C`PsmzDmA?7mCvM?ZWCWU;oHzcRuG>|ksKEYuR5u{T>AhYe}a1FQ%) z<~sLcr90xE>beP`4cy!=AR{GxH#>vPKKk1RIeRS0+(9)1RyStd1MB?{;J)eK_%TEP zP&l(ktD@{5V+OfEOt(XtoGHib;_@4UG7-DruK>N?!z!6(A0ep zmVT*41#8dt=Q(GuuqbTU6(37u?+ICc)hLuoHWx0k>pU7{u7ADr6wNa1JB3p0 zD;N2Tl}u^K6`X}vIsrVGtXgvq>{!kaeU$RzU6J|R1g%)&B32PXNdk)lQ^2T`$BWoD z=SKWH%}Zl$UAzRNAr{|5X$8C1v=;sM3bs*yDZLm^SbmtwMcE@@28E3At!}6O@T!%+ zsb)&P#je@Jj)^anprebg>Bl*NEx;eZJj^#{ybnvxRlT(;4U>(9>3F~|7&S2!SjP#Y zrKc!WHh>)XG1NPUh7MayuZEZnl4G+w&N7!mM=yX!W6zckv+fU-Q$v$v(;S01%W7~` z+Z?~k6*BMPYY4U%EQb^deJ7~PE?jMLDi#b};61$%Dy0L_K`nM}v;FnP()IA8ha3X` zJz|De)1oO|2$O`r?L2O{DtPw`iio?wv=y{^aoEP>Vf-%aR;vN3kLxQ!-6Ic%vG@?*-qzfb)nt58gWHeiko z?k~{PVeAd1y^6*Y_NE%2lOTzR-d}f`_ntTt+o_WWj-@CTmu7IXgQN?a2lZrlsRB5`J}wfTt~1I<$l-yMJPB{Gie4^FS2iKlobO}oY??aWq*T*?!nZ7^O#tpR?(2yiAtxxcv&#p!2J{_NqE+>q4s$}k>WVp{sdOP zo;9W_NscUzTZY5oV8rjD)e@|6VW@YF9JJXdGRX}Kwl8#ugnI_|2#~W8QL+JxJ!+p2 z5eUQLGEtTSDNNS_ zQwboEfYR$%-~7WYhEoil{Q5q=7K{y67|Zpq$AX2=9{~o8JSiwE%q2{c;55|CIqx2~ zT>!gTM7*p@b}q|GW;4^4HYnjvEUp5MvWH-(YdTIX8T*;sACU9V@Q!cS zm)&=0m6RLjU+Gr_bx2@BVE{0DVDNd~%@BK7Hlh%W2^G1{Ky;(KZZG%rsg_voH@u}) zHsOurJ>KbAYqPW`4$)47K&1s!7yt>5N8I=nxiUr1hd<$>EoHnMdA@D-M^s@wAMs~c zoqTof{+sc985;YSBq=!U^nr-vSs)e5+gHnrg+s*(D!WR=^ZI4b z$MS_vecy<0e3*-yl|}^iF~s5Zl=4E?sYl=4UDrHqot-NK4%SguzDog!0#xuiB8d%3 zs$7V_utLMm&VJEp5qmLJq)92k8AjZd5v+%597BRSFglJ^I{czcac~8u2N$eoCnq0} z(lFP|-i7dG@}RYaaJVXD>PrkHB})9hIkVJxE?k9EX#^zYb=~g-Oi2m22(Z7mDO#@E zsK5rh$gJzcK{BTKh`RmxsyS1^<79F}FVy@NAyG3P(JruXjp!Oyqy!ZdzMvQfKaH_L%iJ4vLgRVoZwmy8c?{UNfd56E$W?}3@t zX>6}!KQXH!K=kA8JBm)jvgn2R@34#8WTv7ba$QMZf|X`WvsK_oj=r7)&KY!PbqfV9 zbiex@Ap9+S&wDwUuc#uo>7H}7*VEbVahxN18`ngqz$NA1kb{^nKHcz;y2EhClTbB5 zRJOQIT@eR-UQ}pV+IET3VHwPZ#iKtkdTQfz8sb*m$*c zJV@GCgzDnzK?9dV&PEgtjf-1^ARIP?T`X@!7H2HCeH6la2pEh}zwe%G*)nPn-eaMh zZ05$upfCj?aSyUfHzL6MmytDkE--zk3H>?bB!;*9NhMdi7uPf?Om7mQ#2 zXJX1Zdgj-Z`s#Qk9#(oGaA*}K7i~YAA*gc(IWM+)$5`Ue7{^UGb&9DzfhR>4R1pQ+ z2K+od$5J+`ZXM@qNi-oUP_a_4bJOXedRGd*CSGWM0U6^0mkHKQ+GEJCN^!HI`UdWIOpbQ!wv-%3{My83+G}tPb zQU_cNE}y8gBO~>K#JSz1GSAUoUZ+0YW*#KNhYa}#bJD~wD>t{7PHdiK$6Co!++L&g z+ql_-Q4t_kfsXCyCBH+&b582>F-K!sq)3{R+t69kRk2}@|4%2e*;2#qa5bF54Btt` zPReV!^~Pi*^_0kd@qbWFOy{HE^*dw)(PrPTja25kjpf?(%=Fj?WvaBL6ddA$Bo{9n z7-P`iqC}qR!Uixrfjh;LB=BJBy7ISI{4PW4`cee4S!FDOz%)74hSi-+rB^RyvJ8P! zM&mE9fuq1|UY1J8|zPox5trXd2y`(IWiUWEb22!VgVPKRkbgzfT}6|2}0^zBIJ z^aazO95(+#N>^gxd>Ahda6gmHFT>3Cr@mOsmY8Kv=2>AZ)l%9V;7aw?2H5Cl$mR4A zU(^moyV@QTN$zDF{Uz7A^56q0;1x^s!jBuDBmM5j_>wD*5(lWJ0=Vfy)u<9Ippbv} zx#tA5yC*%})x%S?mqlO|9IJ4P)uFazxXgJHyD~5}=6#CdS@nsHir9Xv3?>4-P+5EZ`H;hiyvIziKUtHVcgx)+VHoWz$rY5!&*Q!%u-71c}T_}0ZvdwVLp z`yzTRyHH9Ss++Fh?aImlrKI@i;w;jB5)4n-dm;FV8gC3;#QjdVH>wReQ$>g|J-T=i zyp(d5j~F?NE({IA20r5A$YK=EUMy_qEF|x93ZZ5Bh(79VHNvo=ks>hfPX97~;Ov5ElT~XUp zu45Ef2;2u&0_Unf!s|Ru%PUsOq#c`r=!e!tLpOM29o|cqGvt;`3rT86-_j$cpVI-g z^<+BM_eb@iT0&h-#8?l3fAjbg`5Af0cZJkGykOw(YTj2w}==bSS-m_ zrIqPcOZ1S_dz^(ci`Zra4+s1k0bseJKGGgGYez!Pv?ygn5BTW3(gSKs!|aa+IMII+ zDi0ut@jnh=pO$TBQ&U2j&lI1~{46?DcZ|7cO7j^UL2TxJef^%^g z%k7HDMhY~Qf4I-7?S!(z@3rjGUC`sg-8NG`%&dePNeIO~Y3Zt+s!zDFkAm5^r>m=- zGjQcTT4${;Ue5;FXO}oWa=;Ib1^w~Zq2|CGPT)H}ZvPPFGm0xk)B^lYMxT2M57J$~ z7b8wKJ-I%B`>82wjH!@JS=i}k?xQG~L|g2K8TBmDf9s2>WMv~arp-2xNI=exrFE** z6VSa=cD0q~`4EuBrV#k^{B}}9A&n}ACrSSwcR_cmtS?^+i$Z3^M0t`D25w#yNVXeZ zFI2MVI^dP^Z?k^%TM~w#IuI)vZ0>57AoK;KejK+i)nX;hzC$$1(3d#G3UzA{q$p6V z4ujQV+4o#Tc5ib)7>EFxv`g`~<;CbN75Lha()#YX>+2mUJQ8k>*q3C@(U4^_rW1d- z>&vyH{Y{O9$Q_Am!G*ncI}(^LMi~9w-gV6{5$_L^+8nCQ@nUh^CrN2K-eAoqkyj`9 zCz1e$OE^Yep&*s0E4p7mEja_nn-O01%Yn04O*sERz#J#7%e^P(93S(=4H7S9{V0q% zjfNO~Ip_bXBlX}~?}+6=z_}TwIV0r`dp1A7X`k4qoeZrRqodf2mh;N)*h@)1(WSu;rx-I z!<^(rq7JF1>8}coaQm#Lh-kFcxgGFS|v7g(Wc6 zn`~LTXDsmWs~0u$3x|weXz3<)8P-sP%#;fFfKaG5=fIf zD|nkDgx-NlDebwO*R^|;3d4NOa9tnfDjke|3+s}IHB?A7$}v$es+d%|Qvs3c=;Md& zyR>hnbv;VwjXT2|of05`8;u|H=}NW&w@cz)rN)P+ z5KhriF9T@;z=jbx=caA?YO|x}(L?7i5XInq&E0t<`Y*8)ndzvK#!koQk?;hVw;In1X|Ni2- zxg6PfRnzM?5I(34!A6AkIu$aCzUJ-GlTVMBy_7!&qd=&YUtzIgUlKp{^ zQjhS~#@xpBSy2yJh-Llx>I)7+{4EcV35~lUb4;gV>gCZwor~Dv+v^*(Rs~(%8rSFw zSSx#JxlnIm9bjycIhbm_LdH^^4HR&FJpR2t+~LnTXu z%Dx`5$9FWtZ`rO9YJ2cZUfIY3l(v4a@UdJni&LFD%ktyBI)hM2Uc#MX(R;N5def!) za8@$A0unGgDNUcVp;;wGF?^Uvn?nWSc2f*82K3=C+d}*Rr8M%jV~)+@UKi(=!&kbc zF#_W;6WrB%n4$MSTUR%cN845hA|a<>mhmd*hhN-(s}}21j}0@aaZOUGaKIA?Ta6M( zD@mDtf>Mt5$9>Qg@a4|hKDeBPBdB!M0g{X#O(XYWBWguCX){@zRbU^S= z4V?A)0lYUlP9o7A)RORKZ(u>Y8vWgKt%)idp=H-4F(y<1<%RXySqj`^-U>?{_@(Ou7sbg%UtQPfg>HeYvY6zF$G0Wa!w=4D$B$9 zsMn|U`W@PPht3Roy{0=yogK-$&^O`R`6RX))30^L!~d5yqV2m<{VUt|UyH5EZp&u# z=?-}hbE3NR6KvDsUym)aJy6{S+3OeU!KDf0d@iK|2eqIO`ms`IcyG?qd6ePK1HJh- zaG;!lpi%4}qNHWRufMeQs6C`-dA4~2=L=Iyasn>$KD1U8hbt{g%LiUEmV5Ybo`YI;;L|+bxG?fJN=63<{KdLBA4mT}%g|({bFa-0hY6X9rC{*X1ieC- z)T`0!DL*d6<0n4DhjE%^Hzb%**c}0TjP5w@P}#u*35x1;V_aV$f8EU4;wpJ;RI)?X zcU(%3G9%+wdov#-#g1`|QYOwJWu1f?KE=PQBA=tVWHB&w{=K21-{KewHL2ra9!T*u zNLpu9*y+eQcXwX)r>5_}IV3TSY?TkSy7zgyf_w z8kXu5pgI(=;|0KH^Ws?6IouA)-yHA2R`rO*XTr+VMBZI_+X|8MR2bXxa5P8bJF7{d zI$kG81SG~Oq`7`l+ zgY7wOYB@6{n6SC`pBLx*+vxA(aGV%%S3eLfOzs0+ zHkh^2v(f@UJvPfP=|lo!=06=W&5cEsLo@M~WFN&~$+^*s{7okqoR!H?fP9C4X2d=y zb^_t^Vf98QWDRn`TH-yX!7fdAzW;nxPtEzr;zvTdL9&}Lfc!dZ(3vf6$&W2-QpIWa zCjy-V$4sRtNt=l~7sDyJ!o+j%2bIKvTaP*1cmB4hKBKy%;~i>nOq%Xq54JAu%r}bl zRSEto_2Ro4bV5r==M8E9n=TdVtEm}61*rJ6F%z?E^uef|aJh>_F? zQ2Q?@CA1sHqit zJdvMu^}#7jZn(}nAx?$zlGG|8apYSZ;FD9BAl1kU3FH=2bKl!9jk{q(|7DLi|7%Fp036XCmdM z5N7;q*Nrj>^Z^~ZD;#OnQW3dd%SnwVEOTysv<%FVR%ieyKLSF>LL`o{*kQwN1O=LG z?fyzw_}B$**0z;=qCw7vLHje`*lzQ~(D zdRW*A$unooQ{$S!UG17V%Uai(iLQf4lBcrR9yg9)(thM?BCw(DcBb3k*rhY@MJOcY z4Qp86mM*jYm5)^cHhaPmX;A*M*}tzEQcq=!c*b~Zup%T^?M3p}N#%nJA>||9H@teY?4l+!44&VD9uB%; zXw$nM3tI>z4JgeD^S)QFmv?K*0^!r|lV*fW*a@3{fq%L~zlpWqZN6V^a5c9^CsgP) zUWyAXY~z;aRwGCGi}_7xP;n^PSl(2flob@vth@d`%F$ZE8ihH|GGlJuHRK3+b_ug{ zlFq$(JL|-l=;^<-)8wRuf95XmjdwplpMQ$VKNRd#PT&P0$8|Guw%~t@G6xJHbfg%qnH<#PgU9Zxaq?;9*m51>1^EUK8CCQ8Euk=h~HGE)2)tOkId*BkL#c1 zaUWLCfq&c=n(5$lVOKeR9=y3k0I;{$Xr>XS+0b8v;R7^p(H3qqu)MrD#}0*dH<;MX3PDM14g)P}-+o^w30VJEk8A}!^V=4!yqO%uHQ1hVxt zV2*~6=nVH%a7wPsm_9@8sV}JB(?&68TD~cX2^kMLA#YJTDfRuCyu)uF<3qM`g^hm(c8}x@)KPC;SbOOx-2+ZE+DZ|F6_#W{ukzl5mdjN{ z`=0TN_r3G!m!^96p3~+u!3s~MsF6r+0BMSpmcD$*bCl0z6R5{O=Y&~ANQ#4mpBJ*7 zbVu_aHG86x>H6>pq+D@@{R)S+j#`D70IdgyZ=cI#Quo4r*4ie%kI%emF1n93V8?el zR)TBOXMbRs-R;P}m?3gPK!Q}@9PdCHK0c1B>=ca-BKvVV z9^qmIknbz$K})P51rU`v2{@&~kFu2N`@|Mm?txYMsuYFzl|6G@M;9U5@Z1bPa3OLp ze3sC?pY24aD&^lwT{fflvWH5KF6{x1KE8$IwYIj#3ri4IiBF(g0SDK&SIDvxwiGf< zwH0Y#;C!~(kGJ~T+x45L$aZup@SrRrlA|KihClR8TiB?zx%M`0DN%Oe;#7p{fq8KO zqK64@@TQR;XUHM>#feo4mm}LhKoZY3`vUYpZ@nN;#3aC56=Z{KiA^xHYa8fV5?Hwx~k4(wA~4 zGIF!PQ2-hd9NRnv>UPFq!jojPzo$?Q!gDr=H$ynbuR91J8{CtKYScwMF*8z~y#xdM zOm8`dHbl%};nb9wC3y%Acl_yVQFM5E@kNc+{e2SM^tXX+w%%iDwbCp08@MSr)kXRa zr0~hs7x`Qbo8l^U8>?y0?y|}xRGZt2|V!=e5@SC8hc|{UH_d_`JF&t zH)V9$+CPh>8`9St>D_Mc?_OLJdpP?r8CW;_ZHx`G-+1{o>nvgVGkJ3U)E@x+$gO~z zlhE8Yya}+&g!OgFNx#W%767IB7RC@Q-I23~9Y#J%VwiWYFwOY4;o4Xai$qx#>kaS) zt}{U)hI^kUq>5IS5$%NXZ>|ASORpQ_Z%1o72#{e{aT^;*fL=y^K%&WMpVKg;;}VF# zSWoTgVu`>p`oMnY&+12-hMJz>zANSTt!6$a&If5MgiY#c!9U&^K6 zi;Egeh_b*6&joC}u^Z+FbMZyYL+)FuNWyz1GK<}7yoH1M@$iqhm99`lu(q?{id@NTAt6mH%14gL z6woTPL^fK9Yn9!8Cqry>mN)y9QHOfv!h%^q+?N46)0BxErn`>g{2gC(HbtvXpu1O? zk!^SX>Z_MyJ~izxw6SeP=(O;0Y7>ZWG^#n4Ew8bF-D(yZ`dHb-eRF5#4OXhTlD^<`FD2UI)zYQX0{&8 zC&8Cn87TkSzpOvf+M@t9Ez*6a@f6wvNt93+W<%L`*hs2lU-&uETj?M#otU3~ z*k|h&&`O~SMxd{Wttdplg`Xu*lh!i=F?8e%{m#!cloP9PV7vmoG)(*(xzi$iR|1%#KQ@>fUb0g+&9s5eGCFdX|xdbw@8+38={+WO49^dueict@-pWwfJ zXH`rBn9oFNkkDJ0*ex>e$oXWNeY;}Zw=a-rV;1GJpn&}^0OhZk+9#ZRpEZBA>trU| zua76UIx;qOtHs)4Ic{*)GMfhXD4#vBzkBz%wr^qiv^zY9xxQZlWOe=Y!<_zGa8Tem zZ}JYoa@jrMr0x0@wmwWw1^gm^o80!}z4KPx+Xwx+H#->*8NwR<+T)#?hf8^0R0gKf zv;)5{5}m0YX;F{;I0Am#Pl2;D#kSKj(ea#y*o&48EW2e<`3gc&24P~0%fdk6*RX}M ztvQAk;Nl_!Jg15{+T>)%#pxm5Az zzN@3bs98q)9w`_Z_WX#EsTEZNkkUQ5Hy1fco!o5qGS(2TVC07-nZ}?4&{L7}IhZiC zxBk@W{}6T7K~aD2+ov0mZcq>bX=#C_1e6dFr6rX`Qbd}iOHxEaV(Ahjq>)aQSh^cr zI(Nw>euvNZnR))kapryY#2wf5y2YxFOv<&V_l^v!*rlyG7waG2td(umU>Rii7JYV4 zJFdri628aAOu+h~8ii*1!y+Z`o|Hh2?s-hfkl4q2O_CD%7NoHDSy{cjH)BS}MkRhb zN6!Mjd}NsyrK$9-Cv|`A*G%O5tX8a^BI9C~-lgWuQu{I81vQVUR>cg>YB5|K{2`xhC*>6jjl4r6ugLnRDKwg*%hF0p> zHK}Z03y-~$yetNqJ<-U5G@`b@tr28WG7Djzz*$#4^^ow{de4}`#Q$Hwf;ve63*sB*J9=Bfoyk?(VK({!4@X@iOsn2yZ+f?SPY*fFAO}w7P~db zJdlDV>K1HXHZ7!!&j8VuGA}Sk)=#2hL>uAVkplAvJ+fOD?ZUws31?G$8E{yyn@{I@ ztdo6*l?W70fJ5RNdhAl~itpPl>o?C0j}KlfZmE~oAkokYxy842r_}_;7WRX%I53XH zL~`lg_}Hi;SlD0B@02UBlO72eRS``g;q@-*%KIKCV>TyyZjbNj&yhAaJkcJxIs7C7 zXuD7Wc!h`Sptabwt6=j6gberms4ege6_=Q3ZJOhi6_7aaXnZt#@y71=CWXPGqo#0b zY6|Jp`w7$t!p?=z{a?Rh@9M-aM$L$lJ4v?eCWl=Q+c|a?@BCV_hTtdvLY#ZUz_ znY-6{95ZHXi)dzf3qN?>DcH0__5&K?gX{$wnXn8T1EVpnf8jQ@Xlv1oG*f5ftZ_|? z)jFis%Fm;{9OktQtV08Wt^BSmIbR^kYta{;B%WS`_T8Ku4_;Z?YE}yl{dR}Mqf`lpFLXZJ7Hljm_fU)$owgTF6f-y! zM^Q5L&l`nJUgZo;h9Tlh?BnAGz;%ic^c7*WksN1?bu|t^cxx+M`(BE8jmi%%5AXBq zamy0S^!%e=-2xu@pUK}J2p0oS8B^4tB#)5l)Bt;&^d99Y!Y(Kd*RN|QdQalh6>2~& zzQ?qkJ_u>P&SVxf#S*Nu1wCMhlok5kulPo zr)fHn@_(%XDdvncll=6}k+Ggj6+8axOc zI0qG(ydxe=R^_#9*ZWuH&`ThWA1w2eXEb0A*|pb64TdZ&irl$_1>8&&=O=ryJ2SOK z-~LTfe|Z2qHrrnFcYquqtL<*l#pIntu!+v|-T2CA3%-7(Ig6QOUbW5gJAV2{u#+8i zyfn0Hw|~&8!XRvqM?kS1TWvR?PEG^pHJBSi8Tqqyz8JdPZ1f-zO_>j!T5V(O!;Q5%AOv{7bSRu(*!pi~$ zN=)P&N3JjSC;RP1{Y#$@uMc5h-617%g(;bP*p3tn7z!6`un1l?X z7lvJ)S+Kt=W&d0dEVJaxGw^K<;gTN^pqmo?2)kq)0msO4zpE`6uZ^xs>%F`AaKmtp40JX_M}pEYf~Bm-xoI5M)JilIqmVD1UXD;|XkE1{FB93%uV$GN`7&=g&6ZB!{#RX+uR6-G|Zq@`!R1`V=^3QIZFo_Kz6X6qozXZ19 zbar(;^RO^j`>zhYsCE=;@#lcGhHPx< zjn-$f?vlV1$*p^r-8uY}jcEj7nneTE552b}pjAGXS|&gxxBRcAXan~&cTEay?DmA% zS3AmJE$E+jPPw^$a&WVtexP3;fqazd9`~}}FSBf#(V5c6GEEqoNkQyS>i$3HOt6&_ z{1>v24pIz!IH{<2SmuQ?oLJ-DoxeYj=$Yb z^(*j;%12}G3F4+T{=S~eVFd+B?PT>FIOmiFWBtUrKXE*keo-e z7>n@V%kcFM9`8$fU%C2RPAH+xmO}6ybzT%YZNP~r&qdwvB76K! zZH4%CQR}YI3}3pn9EN^{W!fgShzV%kPyhsrcQ}59WrOVpe^(JtQRjqVdt;M z_%D|)xnF495ya$jV2M@7f+O+o-#_juYepG2{?cE&k0{q__74xgKwBF+)&(WpcWbqML74~znA+q3v+?%QQk{J_6i~Db~R`tR?Q6JYW9V7 zvZ~9Tr;QlU-|DGi6dfvaVh6ULg1{kzoS`(Cn{`&EKTl-o*H&%omf)G6W951B*)u?~ z?Q}cx_~D@ZbztQFy2f6n@&d6z6*o!otxof?W*$m(-voR4#U0fW6kQzRGYq*adJpb^ z8oNV@W=~uI!dl3m*b#rSbE)xfbh5#|akC7OJYQ2s6tL#{iHF!^R@#j}spBJG@Ey#0Ri5!&brPpJ2V^A0qU;>Uoa^P?lZP zrH=)gw$#2VKvnFK7rN$^uIzk=BQl6jf#dt3qJ(f$w;b0Dan(-$9PRcB$W`!K+t;8U z^9PpCGxcZv3*Nr05)vkb>ivU~;EZ6|$eN4DGf~(xGTg&ac+H7Ak7n`PLgRR07-`MR zy0Mn#L>jLXH}GrBDg5T`c!do&b^_ja+!S%UP&IZKjjyjSl8d|j?>?|1 zs)pTS1sgZyth8>KWY%LP`ic5OHC*0X>=1rR93a5|Y=l+)dY$^#b5Dd}@_t!MXxki2 z=)S{o&x&8EO4z|bxJd2QDA>moAVj=1PR-OZq%Frhu{csOV6e6v@J{!BmQx1ApxSnf zc*qhgIr4u?ZtRRE`fHi~x#dcERnUtpq;xLNgPm89x2-KGZ2yq2EPT=F=-=MD^cePt zd+<|;T+>4y)xOpAyM5ozRq_&za$(5$4Zm zw(bX*@Z%W0!0x+yNwb+hc=`BXQxrOHvjVk!>TWqBgR3`RkXg{?KmW1afp>K3!Fz3+ zpR41QK!Je|aXI=H>4a{An#!q2k~7HqnWp|Axrn;A=*EBAB@u|8H(r~r%x;b&dHyY2 z(v{&iU5COe1_4r6i{^$K6E@Sh!D2-F1@IlvB4oRM$w}Bx-ci9ka^H(A$QNu z&$FCY?sFWM)tx^ggEwJE&RLaz@QXr^v;X|7OM%E6VUN;G zDtDt>T6w67FI<(*ksA0dWRv0WnTVFQT+=3cwX#6|+O+t}TCq7@aXTm@fUJ@p9&%F0 zP6FM+@eIy7?fX|i2wua`a%&3+(>;STKflK*Gy|ehgq<{RKXXZAf^~@Fcs-XtK?Sf8 z>mN@ah?w)&2H!=~oDOZM?-#p7s>=Zy#wF>iI?7}3P={K%eTD~mjd38m1Q z%gPUK=Hf4L2MX&&pw5s$RFJz&8#-sa+A|S&-(uxR=+)t>dy?fTBNAXvM8n6hxUK2K0Y|-=q+x`>%-!*F5R-`u(-%ts}MB~ z2vdObA{o)4#$)@4wp7v!uBv z?Q*&iozDOzYFKiGs$+HWk>m5@{hmK3TUv9Q4HbUmKX31SRvkrU4uOo2*l1=Z!Bw&K zz>goGiWR!g+>NUw(7}Tn0c9d3(j{0ADaZ<_#^yPF*I52aIQGXLLyZ%IP?Qah#=R5y zfC-BQ(u@Qr+ghlnz-k@8mk5TmpfsVD!6SdZ z#Kqm+dE~(ONT8@pDdXN-+RWbVCXW}NE^9+&mj4KnzF)w+9?b91PL^@M^(9_wMGait zmmpgt1SP+iLl62@0WUI7yZOiOM!m$>3Ykum%zqDBe4r4Ly7m!kZeeVQP6^@?Y@Vb| z3uT-MIshHXKXVy5co4p$GW|ffWrklR78|}e{=;ht&uv|4X!P|$bH1R^=kbA>)BMSiDkzvMtU}y|?S{R> zu|Mj0aH47o$yy2^2p*kXA;1If$ib6v&G@6}V{g^)mH-BU(Fz`P^dp?{wfpr)=yv_( zF=&>;IHCDp|F@k10t{S>1p|^Q^fbkyFKY*(Y+NpDRx6hMF8)I~YLRUNLw|HkU9fuD z$>@qiPl32C*F3$g^Aq*KVh{y~#PO*wKTwe_+~MNFue9!O&&0kI-d_0Q;Xe34cB&WK z#SKsYY;MFxOuS?2_PL{@V;wLIh#Y3bA*GKG4vrK-Zn_Wt%Vl5$?EoIIn7Lo>sY&oB zFuL=5GA#@lF;jw`7{sw*EcdlZNP$euw6uM4$vim?B8$#DOPd-zqw zNim?~*YDRCQW3}KwxLE3!^e?sv8J%c{O=n!j=LGey=f)R^DK@6q$n?FPWqOhY_iI` z`g^g76Tfh@F88D>Uc-abN~&6OpfXY-##iFgM?*ZVjTi4`0`K9N$EQg4a^dyPa#on` z#P{;-ehYI79?|U-<?-aCA9gDmy_qO; z`_rkWRe#H0=k2>4Yk#$Rk&80&xRv;v{&||Hc)&c2j=8VKrW&fy5&9z#mY8OLCD%o? zNBE1)z>R8f)bzXYIw2n3lj;jP0i#7=UFZbQWias8P>DhZ0AdhQKuK8#Y?^QVZH0jW z(hC1f$~kjuWyQ+Wi{nUfomy2y>d_<1^~Hd784s!X4RPjJSF<3$aj)XN!b>kEa6YW1 zbhFG1UWV)UxrPytF_^>cn7uB*P3B8^YG`QqWfL{$zOXbtOyikzqS9e%+&)Uq_%y8?x`w@=Xzn^6+zqZJVRBt|NkAD=$x}pJZ^ijvnsJ1e!4O& z0)OJp_(yADVWJHu>z_~v84CA#)E!@pH$kZ2$2Ol@2L35(E=yRYd)jK7OZGZQD%Y(8 zq@V3(6>+5Sy5)Qo&QD5Px-@QFt{CGsRXSt*Dc{9QWG9zhVe;uh=T`IZ-`H1I->$XW zv~sXwEB=VyBFuT%nfh$WQPUpLgK~a#{b8Y`8-Ww>wL_L2o@!>2Z{Ks_gR6^?N6xU9 zZ6iEOSX%>qXb#dmqS}Wemd?m(C6CpDunQjYX-Ia-tN6UlI#a!pi^h3%MmweG|BgIy zBD1Jz^l=Qtc+2^@`Vc!6)d!`z2juSC(S7?qxX1la25WQkJMHc54}uS7vYUxuxbwx= z-7$(~^RR>CvYf$^-iK(T)vRXBBZ^l!(n)DWg9Vh8pvCHOanQr3WyT@q2rM!o20!0? z^1PL+0^)d65_)=iR9eKSdzQAfBWb&pUrD#soc@ZlxcFWL!FB10n%B66{dBqMF7~gA z;Ly;NS`g4fK|sF)dZX{&SFya+VSWxu;^>A(A~jO2gtOkScWf|2YCVk@{kiv~rWbY8 zS7{ZrAcfJWfE>3iuzukwp0fL!lPlSnngj=Tf0;k)ld=A;9cE)O9 zn+*AF6M($(RQd&}Sy?&8N7nd;?n`Wv3d?2$Lgv&Ht^1_thLS0b|Y%S@46HCX}Z@5QCpNI)F}y4UQ75f8$DsETwx&CoQiclV;Qd6xsqgpVsEk0Ii+zMV=S?Cj zB`v+71Y^@1&NIO+6&I3#+$Ej1(9_7c>oxm)+ zOJQqWEM#4krU4BWJA#3GUPoaZkfsPDQhMJitNvLI!zG5I7nr>k zWM0qG6{Fl30115D819m-wNNN%mrc6uDbWn^pf(A5@i{)>QwS+txJXEm<(Hx;Znkx`WbAVIX~r_ z&|*rtEP`TcWN-dBkNi+VzI(qYF8nZH-orGie3X zjW~nZtVu4vE1kT%%^>NRM#Hf)RapQk)|3Fb)IvS{4WJ!3(`CR_f`0*A$jEx~$jdBV zkG(R1FD>%@ybLN_d6Qc;%t3?{?fXj|&Wo+sH_xlI#{aHx#4WYMc%5y?ZO=${wX5mc z=XCW$jd(l)0$r_bLCyt4ylH?Y|g}2vwyh}u9PeVxNHI*v zPZ4O%pHjwNN#X8Y?eHwT1`(fVTFfA{KRRD4)pEYkbN}dwR_Ce}mtqkFDT%jzyhDzp z^xni={Y~KOcfkTk;55Ihvt$UY0Z^=&l-7OU-YYzqC=;FAH&3Enf7HYJuD~T<^OG5I%zMww>qhxTae^5d+d&I&om$L^$fuDV)L#0=5Z=i zw@x+de4WD}g&&}paRh@Nicie;U(pMV9+ z1lTU)iI~@rD(=-{u-p~8g?LgE+l zJn4>z+JeA6l_=cV`}cU&4p(@8Gk(yHtn@aBKyOKU9lAgwcSKI$Y_OWWoUE*>i@MEx zMSm1onM@Os=6qJZ>lcs#ZDn#VP=)Tg7#a*)eBybc*Hejn(|Fz&vT|S}To_@=L_wuB zM4B@FiSFE$RP}h9>`8Yam&_1s@k!%;n`GUOSM*ia?$jK*M<<#&MUDfx3>gQ%`UWox z;9lxV`vObJUlNX^MOFbu_dT#GJ|X7_DlFuoyuJje3`7xt-Stc%sbeL>9NS%*CYH{c zw+`MQTx|}v-dxPP+#w)m@$XNS%qe^_K{agiER5hTS}lg<6Tl47)y{hrYf8?2jHl81 z6cZy+=F9=KI-F{~I_tBH^`z{I2lKvCU(6~_)fFu$3CP(ZxZxfqc-Lq2rh(Q!9xYB7 zIwt>=V*!veF8*|-#E{dd&t6It64BT+b?B%{TM$?cxwO1@m$Kd^YID2BBVL1^`Xy5x zX%bGdrg6YOkbH>v=v8+n>3!nDzWoou;iJk9qg&5uGe4Urno4xVA&biP>6?%gCO)<2 zLTh0P;sQ6YfXyh1&fEgR7SxT$gl|CKO67G_pB9IcK$!{`3dQfmSjDba$<}9ZSC-}@ zK?`^H#S4>Qa%PFloeBAs#H8l#-Z|@{wWmKli$KnP3$U%I0YCT8LS)7%YmL8_WaMmnY&CAVf=FMeJ6z!)drf-A~Nu2c}JkvAu zo{?6)IBVI|Vg(OOLX9rHN<2GwG_GTv=RWn9JVqTH-@+N9ojtF<`qDtJh8P~;HiUPUp1fpSi_o~S{9w1 za3r#pB6wLi0zQ9$4f+7>zBexb_n{lG(uO@FCMNy~7(=j!rB%S!hAH`Z4XRfwNoKxr zN1PlSeHz-}YLD0n}u9; zz09;}`eJhxlx7Lc7cSSMInXZj63XueaZrH~k5zgWQO9Q4Pe<}OPT3vLhzWiiwlB-Y z8T{l{YDCd=()?zTOybk&FF!TsYL0gy>15U}gA}hatDujITNq0-><7vUgbPk)ewMTG z>a}NJtZfy1o!{YW%PG&l<2TQ00dPKAyy~K4vqT1FvZ&j>s!YPH<^T-PkMmlSe9_s% z(V3qiIT@x@3gJV!u;w(h2;9psdriHk@=E3Go{xM$*43kuI*IHM-^)9ka^x%A+G>@O z&8=DjGC6B~JDc@+(RLs#Y4`Af-DhzkCl8_ ztTj^~{J3@2d_#<@1$EPeA+YJ#$*wO?*d-jMgWrwhKW(Gc^C=^{DW@WL>De!s)gI#h zI~w}J!XhAFA#CEK#9`EY6&iktbn*+Z;G?k2cj4&<<=9Q!p%-^zEmO*wl(Y9duT{4oxJ-PYH#q2fTvw1;t4V%`X18P((e9h@ z&Vz~6gZ;r)lau*lC5Dsd|JC2Go}21L2pF>-c>Q`-cRVGuZ;>HQ^m0MxCPoE&U^g!1 zwg1#ikZ^Y!A)Bpzjg3BoDO0&=BZhP0oZ}HM^)q8hjo{t&p`1L8KydQY2sqKHV($WX z=YI8eRC}V39sq|G7Z+z?x*`=56Jt);5jlCDJhNB0|5YafX;8B( z&$nwq9iI|Y20Ny{??l=qR&b}!nK&WaD>MwQBsI=owGj;sXxdTV7JcJ4*Tx`qe*XzW zbB=Z((M9$!C)F)&9ZjctE_8|uYd+B+j{=2^@)2l{#Py>C0EiW!;D0E^c*CXjGYGJh zqeKP#XAY(K3D#L`9wis|NsjUuvpa*1UREY?q_o zawsk?9!+Kv`ZyMSNxsCDru# zo6cvs@gH4F1k1Z%i1}88++$z%#46gWswl#fUlCo`*}wV6ep1N&mOi}V&Rsq;*>kbm zliyo`GWRU`!4bRD1k^luuLr6pn8M(y&=L6dO@N%{kK>@(nc}E;jX{UP;?*YzrmkO3eK9vU&gIRIN@O!oQoCEoeN*D8yu{#-@{{VG4OVxI(5{~)d~zndIxKqK zE>I_C3=;N*G*jat=VB=lquyBvA()BDEV0=Y0)ipaN|jL2OWc`#0(2^JeRDz|E~!0x z%@jNv$^bj;%a<1)&o`hZK7Y8>_s#qfGAC^jmla?-hQbTIEG8a$^?8`VhnV2sKy>QV z81Bb)Jf-Frso&^vVMr4ja>kRwTgqF~k$_c8vu*F)BMb8y-(Zvs8Lym#zf%|}40!9! z@usDO!kl;uFFdX_4T=c#MVccAXxJ| zHG~v$vzi}{*?o}W7ftq(w|FYd+*ZR^b9L*O9CmoV7>>r!9`gR)m@Cltb29kuam1*E zQ2Iz#34QGUw#%pey&DM(`OD?HFLljM4_R=!uo>@9yBn8(+b>f1;^&9Zt2Omk-kcfp zM7?WnG>gm>U7xbrm}-8lc&R#FdQ_*26MxH>@EKRd(qNElwfiY}0VSo|&9Mk1Ako@U zAiWG#1T&1T;8KK=@+iub<|$$g){}2ZfRnGQNP2lr&MFyW>U{b>my_+}{h55#dK2g;_yM-|^^)osFaEQ|F@j${ML^=Pr4=1?G5B z+6l9>CI_>AV9w5Di=X^0)txMjEv0SNn z?SY=#d%vTbk+~P#^_#C`TA^UbCI-X+9d|UMv+mm;c933<=YaOesq3~n85;ygnSi`A zl$;su6rJA&Jb1otRfI+mpJFQ8B=b^@;1^tr(|Y5ZnPuh zr$F&Gs&1V2^c`8IaDk#fraG*M$ILD7?1DhkJM+L_m8!p94QwVNTG!sX zEF!9!VfVq`jSHi!_P$&CsEo4o*P4p`^j1ih)m5vf?HOTNo>hC|EJOBjI zl>Fi3T4A@{nrCu$H+eEv`$?-q?LrJnsFCOg1j9B`Rx4_oihzo1l6R8#>gk(J3t*26sPwZ8 zvB9#E>uo;iD=v^i3|aAi(E$2ysHMCl2=!eS=W=&bv2&MV&1jXH4R?LH)ruL2%DDLes__x4SDQ2 zsmYGRXyT4fLBzyuPbHF0=UmCL~1bnUHgxM8W&r2>eDzqt|QDCn2Us9qIV?N|W zz|>=&+E?CY$N%X3PMSphZ-vp0)zI!fwbumq2&QbHi)gDLEFB|lVvs-HA8&)RFsXTv zZ>(GrPP$nOV(z*B*k{b?9T~6l67(3Vig=2Zr^?$TU5ekxr3mb-OzJ`Ametv<3 zNXPcO+$mvKxZ5v$otuVJ9n1eq?&uqz)87h&X!V+I_CiBB8|jfqXUorQpU}8=y(M`& zo6qK(L4tqXcY#uWg_$WRaRJ@+HnuUW?CO-%nfMQq3lQ=_8Y^z9vvKNt&L%tTDh^eKtt9?BGg&Qklh+(`X)#IpTYnad&7QaN7 zav!*QGGJQ(<{a)1RcwwntpHA!n>NzYI`s{Lu!|Frqa(C7d;IG!zjiD4_(WD`EtEvvWfPpC&c-RkON~hj zmP@y~o@8-K;l!ikCepk)%$_d4vJkzG4x3Z-dpp?Nf4Uuwe7ILR@1mJHiUl^1TntiG zW8tzrPy3GB(w4{K0N<($>YOmceCnAWOL*UxeYd@2P5W1YaEtTeW0kfgb28)}Nrw_w zAAkw8@jb%j8>7Hgk&HYUVkl1vx%e`vCa-q?80n4{sb0(RScC9hd}F$uoJ=03EW*&# zMqxK5vR+gW7qIN*{yrkyS^2l_p6;Tkq`YVu9_a4<)ZQ|^2=2y-a0CW)fJjQagXZlv zjP(x<`}<&vIS$|cK%Z8Vw(!A;*-bybOk`2td()PF_Udn~&1^R7{eHuU>}+bgPc~xY zj9^zKcIN)eo-?4AoqsE`ds$BC=+z*Q(tN~u>|F_~RMqt>qU}0upH0jYv@2_G%ICUGu8DhcjUQ) zR444sRF&8o%~$o0*jr?i4`@6}ZYglJr(xd0um5;#zB4huhpS8Qz)V*+9jr}-b&~eI z6u)u83E6|YzT46)6&21m=S>xOJjCbL>NOCTQ17x1WwU72m86!=jHYwT+AYDaiF#~H zg+rsSY(oYM(C=N6gT|d&0w(4?*s1Ac+Gw)O0(D-?cSJh*h5jIHfAdT$3!E{e%`=+k zUxSAab58g6OzR9jUH)RX3Ek+Gyem8H z>UC02LD#afV3I+AYpevndo3Gk{Ql?5<$~L_xykJ&Q6%*}WW+U}A6#y-l*|xL2Trta z$$xVd9|JNZCgaC>tBWJeBky0KV)v&D*L(OUlSWRz%mP=(%H$b1Ib?W5NP(_K#X!Tfk!oc;WH;igM?rba z6KVc3-LL-EY;sLkG$Xyos!vGeYidL@4WM<9+MEjiashU}4SKp%uT#;SZp3Z&YqyDapb5Um9q#0siF96l9kqfV z)$R_=pOUR`(+mO*>E%$N`BUp=UXzULFC8j*mQT!CvAnM30L2~4f>lK! z>I$RDn`y9ClQ`IG_NrJ8E?JBr{Jvw@S@y8b(cteAb*ZAR3z#XVt*O}=fJ^x1@1H*+ zh#x6nGHI^_=44Op!_Q0GeNd)m*Kj9T4v0NpOoUEn+bvjx$2@cgNjTFT;K2JH80gzp zDl9t#l`~y`NZoSG6xfD~vD%rg{@q*L_^LuxSHF-;spWQLH(DWnX0L1cb=*U!dfBm* z_NRw`ru0Wf_O1h01F8Aq^npx#Lh>(Viyxa~!|sP!w;lXsv4#S41%jWaYku%$iOiD< z_JDdB<%8HDp{u!d?s{~|AR^J5|L5Aa(`K7?o5r4m@!DHjnhr@p-4FZZdp@=?MgFWP zO#G;q{Y353b4?K9ML09zG4Prj#JY@~)6A3YK(}PQj;4$kn7o$i1>}xZcUxRb=C>tiC*&}UkoA31x82zg&VLOd zU-Os1*j(hBo3=t?Q0?NzZtwog=T8h$nL_uUrTK@46QP)xwC3{kh%PVft~+SgbOy~r zRDN0M(EOr5G4mE-N_Uqwg<+IsC$q!&8))d#p|+d$+6%4n1oEx*N0NKh3V@=X_g^#w`ko|uv}_RpF3x%%Gmm@mNS zhB>3~7JVX4A9A(+Hxp;zIpu07OF>Pn&eljV!UlVwU{Zp%xtDgSS%wFHX6%6Xx4<8v zhaq!4#XAfZcjB-;X}ve0Ye~uNCu1!Mac@myuk}J7KPa?1WO6ky?$@1DlIQ0qJDK+u znw^^i(Np^+XbqH(DkEHMXX`}VcBbFLt;cHokK)r_2WNCUbViq5g>x@# z_%l|n-Q||)ZG4&YjsaZi(zF+r4+pJBC1mGZ1RA1WN5pZeng3pFd++9UQ!Q~#F}i~$ zJ?P=q=38>5^ksac(S$QgM?ru0Sm{@@dFIEI1-ew% z$w7>J8IRVKQ!!r1F5vR8B7YaC=9;cOG}O8ycr@JcN1D$2WRMZFmuX1CLVwsRM?+;~>+*B{A`VD|}*_QF8fmP76}{pQH1 z)=h>I)Aj4*b&wsK2O)9&{j7)iy@VWSH6i&sWjah@?z{B_wGvK|Rp0Hx(rJNJa)IBl zfee85`5l{%=$=xZ9bbl3SOd7IGewWY^_!AedM>$Rxg62l_4Igc95?*?PYvW|)>I1C zz*M47RDb)5C|=@ttYk@n6gXThBM@%t7rH3=CIkN}arcu{mb+Zi$emW$kj{XFo9dAK zaVgMGHXZ#a-a?6!6P1)YgCNR^*mqW zt=q}Fb-ImNsDPzXt8X`EfXEWS2>Ul$$2%{RJQjg1X~^?j@FR;DEy8z_QvTyk5Ks%{R4IzfY9ywTL%Q1@bKAxDgQ=9MgAbRZ_oh|u*mEri zcsS(q^4>mq65bK^L#qiqql>a)P1lN=Miu0wb1R*sEp4KD!4S+pr=yrK{g3$<{rg^9 z5vMe*{F6OOp6F~{*^w2qY`iMt_Yu=(bw+TVEF5`GEFLtucFW5S{pB?21N30!Gb~+# z5@pm=0?b~qi$<>iRRzKZ%Pt6OO%yaPo4J7PA}%!K`{G&3hilh)C#@H{N*OyLkKDGy zLHGZ!h?SL9zs57bp7eEM^WGNm!R*w6PXD9n8VA0t3xLBEEqfc-?mPDV70_L<+?4x4 zkL31s9ZMBs4P9;X8K0>GM!>o+xoCdvCF(`=f0IFwAG2_w!`6Ocdlhur$0lK6e#sFt zrW@gI3vS)h;>CHVCE#U|I#If|5%l%i!WL^d^p!g6>gaL4U)q#O>F9fbdoM)>Pg;oS zE^N^3v@qk=+(eq|;UTKR@#vw?aT_0pNs8PVOEbGHe#iLfVg`AD!8Y9|xvV0p_7)-j zSHXf!N2=^T8vG$LUD9$po+^3a&w3Liya6hQud2$b%bVmQr%;CQ#|3#oW8%FRe^3j{ ztU?>u3!zUnrZkXVU3m1wewP!6XlD1V}Px|`x z4zR#6OW3yp)NKR+di{1hy8Qh4_F1#+#ePT61DX-cb)P8WcTFOkVt1mqmvn3V2Q_CP z-MP5N=91-ww385%pBRyLBZ~Z-f{8@ljD074c+_|xbg&>ZRoDUt=vl1by3!B`6R*9{|5&i`xoydH_A9tmj6BRNu_J|9B$L<6!3Q`c$zb0r@qdU;j+RLF(nwkMR*3 zThZz}YU~C_du0#J8N(sQmw-$dK(V8f-RC{IRiha0zE(R;OBL-Ai@BLRSeXESm?Ck?YJ-_HZ}k@&;I4=ji1$HkJS-L#hGw!z%k{CR=59J5)9C&jm4`6L}XCnx)omn+hd3T(#VUITb?$!t9tydnt=uLe+!;J2b z*2lU9{@DPc&2^=cm z^2D7maUp-9t4&5Rp>#dFvYHyMe_rxXH2UgS-JWN7JeJ5A4>qQ8ju7Y??IngP$^ zv@8%27obtwMU80L&?)$AUGXbD;Hkj|TVAlYI=;@cJ&zQs!6>Y7b@33zN~j$E5fBIP z{_eS)`tE_)S&s(0ex_0?lWVv}I~|ww^40RFWO|EH*?0K4CC&4m3PJwDA_x#Q3z;>u z05g?X%fqGhb)Hbc4!*8^%viBLC(pXakz38;3f4s7PlUu{p~n3{nuAJOy3!EeojYP1 z!p)~ngjZ#kPi9ZN2*qIkz>u0P4=JyyvOk$aqg+6afja9%=K%3$4k#C{j}<3*esY+s zSoOQcWK>y-ns<)eh>I}gH=CMWRWbCwwN(aF*<0kdY{^{9B>KxS3%j|5;b|KRZeWSIbkV7EpYK1{1Kf z>c-0R1g;HhT!)pN$0*0n@VDD&a8+s3G_RV9UR+x|eLDD39>h*lgxA9wLMp&ZRSsr_M6t`8Mj$Kd z;5r^WQc?!Zgo#<_2gC^Q!ve@<@ocL*F5t@VSZ_!WX$%0 zw^C|VQHXPn`%i1*hDTvwi7A*NWEp-}^$e03K8H`yyK^&*of7;H|E_U&H7UBe)jXCk z^g4TL_-*5)9*Tg2_)IZEy%$JleTi56z9-?%@6{4fG@__WUm*~Cm>5MH%`YWMZgsaS zY|~yQtULzdrJK%8)JezQ&O)23{)IL{eD}wxG0AMLo_QjzpL0p|4L1qy^Up~^uTZKh zUr3pk{PaztDH>u__qui&8oJq0uRAK9Yk|@OS1dd*^)4DcTe$$iY*lJ4=w)~mjcbcE zoBWWO;(hKw&9!yUnpT}bR(U_?>t}f|Syvp30lT%v`=|ENw78TZQZ4r-&E!Soya(u3 z6rHoD!pmxY@U-$?@uJ@uloLi`4o+p+nvN6NcyNwKCzdec%C4lBhn)Xjm^j@GUpwU1 zO2MP2Pa^qobSf!#&OZ))YHgkW7_Xd&mT(-n27%`d1phk$xP#9|Ookc@aN$=j_`41> z^tEy)BIgvW0foPbP(~n2TK)JR+bS=fNJU2XEJf1EG;eTk{0FmO4r_| zMv;=5h{BhU5a}4*DIy)CyJIl6-7|mpzV1J;ozFSXdCn8>=l%Zu=j}LWw?XWo;veH} z7pvYYUXG?<5i_G5NEv{#X6{%)* zZainjeOBgdNcQ>Kxe%PX$PwhE{i}RF%h{3nB%hn;^5!TeN8HK(>R;csCO%%h7W9O@ z^T5Dkphh3#Nnw4n9%11hwthG#Y_b3QX?&LF%0YRGveI^6>#Ewz2VSs9nm=;ala6b+ z{{sB3BQQz_hladI8hB;9*;dd+7kUW&DDb_~pO z865t(oQ<^abLL%1S3aZtyL+JW$bz`1NBbjOmeIdTq&b>yhNl;!>nZ8{giz8fYo|@g z^w~LMKJX3LUEkavYkI_5?Lt9be9#KIK0;p6pi>T}3$F%*tx7zDgK_2k4T{3E>ow}* zj2SUC+FS0{jWDyUDK=<a#Daqc*ubH)~Hqx1u?ao%nVDl%0hh`qM)~b6!y{l_zJ4ezpeuVxUXfAawr$FPN zaI9hdHccIJJvV*tkAZ&oT)K?2bcqT53JU5qVRLOG<=gk~k?7r#ym$bg1xX0PQQ>M# z-R#*F8OR8;cf;s)I;v0M&RkDH91e@6@nHf>Vt z=hu58R{B6Q#2}M(ZtnL_KH~-E_sk7r^Y8V;=X0E4&Bz>0o}oMDsyX@F1oUxeK$I^9 z1qD4;*YK+mE+<~Or6ZR8E~T$deM5t5cYBKES=921Ty6nYXsF7nl^l#HfSexE-F5dt z`bX2@rA40V{m)+2C+l8lw;`b=7DMH%-){<{-9?V#nbj9#JB;Q}R+Tk#l{zg=6lZHr zUrZmUv48vfxF8ywbC2MiI~^-OOkGbKP8WkTV#U@ldHbvN=4(Geg;aq|n_by&`Li<; zBBU32?h-S{@y@rdrH!a502`}aukS|F#sNq<4w#OQ+9+;k@35rUoy-bo2`ZwpmT!?h&a zM%Y-fIe*NYU`E}`PTGKjZUIVa>Q7@ul{g&5?jjg!QcB)A7#)M6fWCPpCX)ctSzyfM z8Ggcvvxux}T#qc|#$QF?#Tgn=MWczr!ynl6^jvq8&Ot-`aPLa>!t5A*=F2)FW4ko& ztlBdM$-r!-dk$Lqgj&sY5)u;E#aNk#g7EU>?LBp5Va zc|+WllR6>tsu&Y>M9q2b29|zdJ5L~U)@P*~fVqt}Z}!x3eOw>+>H5X3K~yRT)jHUJ z?gP*l|G_-`ihZ_n`cW8~kt!xkyUnsSS;QlmioNnnV^$Gxt>0GSG|$LKh~K1Iir?2e z21U5s-0mLHjv>QYpCv%+bo40;z0x@+=6xb`lmBypwu5BFXNha)AO~v&b14GWSP&s4 z-npQs&*e=r$a|C&b?$$%^>OSbkNAT&NR{18t<-G5p&t6|>~l&MzzrX>YdkD_zyROG zq5av4;RV7wzKaTq!*=&(xf-v<%~tZIN~M+S33m;Pwtv{QV97qhbn(kRc+PRVErxh5 z&Fw3k%Qnk|-~r`K>gxmd&e@%NY!j7EMUiNdX|FlHHxLhE4X4Y^yj5Cut+o+ZxXk%O z5ine#kXVh_WxG}44Js?L%IbdhkFu9mM+BS3Lbcm=ujnOdV;x3eZ8yG8_d53`3LB;C~v>M7Cm8kP-up9nE${mEPhP?VEU|vPd!JD{ZEv2Kyw{4 z>MveOC5Ku>y;pk9av{doxcs%$Yke#o;+G#6%k6y+l%aPs>VA=L^hmhXvtKzlD?hXD2-?Qkoy_tdUaVY6bhkC1?`udACi<2<|a*!Z~bG$6tdZA84cvY1(n z@Byoz^RsQBaV@G;rF9u^_?Jtxpu_T7rZ|A{eoLkdd>Yq8`e^NMzEYO|`9n#Ezcl(vQzTID9a5u`@|VX8_s=FB zO0Mm<+ZQAeitf4(DUX}Hw=G(iQ!Ii23nLX=*CKrXkyYv%YY=$-#4OJq+*SBmN0de1p4T7CS==L*~4c23Ov<7Z}2rT=6SlP+}E~s*T z3lZWS=G+{(jek&hg0vn2)oUX}|A8Cla8AAw*sgp%u>knr${QZ(n{f|KgNM9x`QG|2 zgB-7nQ$s)Pea4p1VH((t@61%##YCoS3of*(VR|IrzE$5ZK&&if;pW945bl!3^8!WV z62Kf$^rE5%q$PyQO#3}m!n>F?rNlArF>JJcr0-f9aL?faMOLLAPGo0$&> z1Eg!kE5}f;tBcZS(WTHrHKoFBeIpe(&~AUe9p9pT>=&mK@oY1%%Ql2aKm-aGe6EH& zR~5H!cw<)s=0syjix#*)+FzbHjqPP3B|6+2aob+qD3>onFzo44{72)55jhIrw9O-=g&wu+}~rsW>GMrQ8ESJ0Ka9ETYOwrfM#Z{`}66qS@*_uE+< z!0(8&K7x3RyG#cx9`mAitFgEGUMOdNCpxl_!h=)eU(j;^WVpWbs+aZ@QLi`)IadL1lOLY4{#?$m_@)lLf@+MtJkC!AlD0APYC(7uYhWTOF zzvqeB+Q689m!IaisrB26-)F`Z-l0Hf5HnNtN#@Z5JU=)LdvA{|_QhHULpJ*0P5Ha*>+)3b0L-$4i_VQ(Le zxJQ3XG#9i+-lNxdR(9VjWH>i_+p|6>H8`9h3{s{)&uK%xXZQc-R9(`_S`YLr*G8(i zy=rB3F1f|7m=PZA;wHcDoafbxC(q4Y77d(SIyW<2=DbJ5Ue-yc*$-sM^BmdIqMx=6 z*FoZB2IYKZ$LC4mEg@6~lT%ZsA&m_j*t5gc=u>>M>Dy)?+1=+3o?z7nR>Xd%>%p7F zNaB=?G0N?!hG?leY}u^BCHRlycA$m~NT{MANke$M6|N(T03vKMS7HF{vj6in?yAp0 z%4~OMlE%ShC}Kb#?HH_$v?6C9#QrJ<^9-psvoU7gGYEZ_bsc7 zqH$;H%E_ElggMcjU=>6@(Z~E6FX{`sM}V6}s^asnR(00WK1Mp6B!zy!fk5N#^fl_0 zL3KaBTHh;RGM#WoQY_1sJL^h1RH1Qg+ro`a$AiCGtAcIw`TfMc%SmiB3~|f3ee+p0 zcxv$N+R#9zf-5SN#d>GD=3p{w>_PaBLjfbVvuOR$*U*4s1PsS7WYI`e=O5H}jP_P}0dfUslmVBhs*3vA@X;rKPJSA{e)Sh_gC&ORm<~h^eGY&% z0iaLQYSrr5;v1wqB)Bl{oLpyi%x`^F$ZnuZtYws znV?PGV0|Z^9xU#WloEkip9&Hmy{+YTB;ZKJGzbaBs~yXzD^s5z8~YIP^#>u8&IsfQ zjTba$h_97NLGaB${%QD1oF)C~RAQI+(Zi-&gRMW)GoY15^_W!_ZPo z;>*<3Ps|+aTvEYKro9u+&wZ4htNe+j+tw_XTBO72R4GWmj;6KgS{CaMGI`Y#6GOQ3 zn<$^LlsZiSn3G-Ep<&#`?~PjgPID%(<~2dyx)a@!OKSu0W#L#dT`J2moR+Z8G7nH&pD2_&l@eE z#s7SzA9GuLpx)naWkM72HLy~L^41+IXUYKP-SjF6CIFb@4}t&jgG3Uc{eM&yNr0+O z@?BRX&XPvjzw_ZuB0i#Nm8WeI`l2v14dp1Ec&+2Ws6d?Q=Lp+z<*~vdC&bDhg0~-= zcXx_5Qk>=ZNp%{i%&{9VaUDQ4g2OwBO>&2y2>`3zApKNHhC)GE7K%@67taNz=1;~N z!xt^M30*b&fykjb$v`nK|1~Sc;d;3=BXOnUrB83%Ha%^Hj-sQYhQLHQ)tqe?46u0( z|1g!=M^6ztnsTTJhWjXYlp>DQ40g6Gi*JeEN6{(mr!cIj-i(?;<*z)@UQ7 zV@`_}a*#~UV9bX2&l`2NChA-Ht=5e=lz7V%B2jM+;vVYKnThNDiA{^P_+ziRmw$3^ z90AgUEFb34ocUSKjF{N)g>OBLDGZxJEuFA4rcJZt4@sigClhDZDx zoQ8hacE(;TJk3wjUwq*5H9v`~_2f2gebiR)x|O(cluxm3i5P-u*I+ZGApe({|IPq8 z;RNqVGCi7P3t5>-UdYxzUpU-_w||Rdyn5}LjfHBCTphvC>9&sa+X$)zhg=Q>#zn=0 zL&xdA2b$AI&)5SlHXno1(6^2NSF~G(#3XJpgev{w$y0SSj+;bxcv4x12BVbG2WNk zF9Wro2}YjKqg1*&39;*O%KJ}U5pr0X;W<>wjk6QtjKnp4$?&=IJZ05mJ0e+Y(DqUh z>ub*Yp#OYxBFGBkINbSy_~Z*Wpb$6a6r~nl_mvkor6Fu6RI9766G2To-^^``>ejDC z<;~%mK=k?{Dc@Af?q8(0!3f&R_pT2S&7+pXSNrm>jGla7&@ltn83lkj5MR~9-s z!zyzkgs6~l;&Z;S{CCLgTNtOv;~@NkPDxYmw*1TKLr!X|}Mb*#C7a>mPZ=QACGuj=?AjGB{j5NaLrFS_7Ps~CoAepDVTy4~#yMEQ0z*p&lesjf=c@iZi_6_k_kW}b zQWYP6O6nJ7DKMNu_CENa6;3!Khtq?^dil)`PE0}(R%ZjmkuFDiP1aSE>)!`e0(l<3 zE@?~?M4yJ{J#j<|0n#&0f+4;(^R7d|@pWW+ZyT9NDK3JB3r`o2KeAJ6g()Qx zU4Q@vZfBlJ$Up^F=`ILf$D~&>3AreMkpA}t zUSYKC_B~TZ2AeY6?FpWD!G@DaREApu(fSDQDF2mxQI9K~fHnBDX4!zWZ@>vla&olG z9PbbG@9*k5Z=YOWkT@)pk-S7PS|8F!^6I~4S$79)(zxv!q;RL)oO^`q4aa{pE9Zx) zp@ytvk|%_FfUwGotDvynbAwY%pRPX+K#!KN?w`)>?e9yMWDjQmAq59E-tXVPkFVF0 zX4ajwVVwhlF^wFdaL#}ubRqHo+j_6d)Y0<`3qdJRd!8Wtwm?bi^k?|X{j4`Y_lO^R zU^975=ehbNi*fFeS9{=`u3EG{T=SHxq13h2 z59>ReagVU>dmrghta2MrMc@{zw=;LpZ|fPYC63&G@jL~CLcS)q0U_mfSPF-do0{8+ z{kOehn9zX^*haaL^&4lw5*Kb(|%+?63?9z1HLfmvb(0*)?5i%iRgh%QTr&dc{ zEzeLbwTCm&s(nL!V8Bjc8N}g6sn0fS>y-V@CN$z_{ZC7u;l63H~9;hLe55z+zvQ{L*|@xlN+Te;ekr zS<|`iw0qrtJM97C(_W5ljJ|O*GS~+|iuQCD3poyUC6#%m54WrlHRw;7B;v1gARW$B zO+aIfe~yyZn?y!y8HmCZ8dS%cZQO zQieciOp^i>7sM5K{xUL{5ugE&CDaV~ z!7)jG_}gKT6~1@K9q6hVJ-7!)Txr(LBT7u(w0)zB%AgAL+t=ysNgNPWn>!iI9=@XA zEOT2w@J=s(XIrY^R?f5$YBHJ)^_k8;J2wbk0o;IkkZP;TsnTiGxHD`_eO$aoSX%Z+ z=a%`C@wKFPfT>%(OVo`z1$q7FZa&$CGXD`6qp$`7d;92eAB4_*UN4GrQ8_Oir*0LO~g2#F}5TJNK%UJf{*o8pDcufvBwg-2(`Qw)R@W6 zo0CPFY0SxB^-1L4H|IER)^a~2$Ok(Jr5wr5?k=rX_@`VHDjBhuI|fs$%8%>nx~u+^ ztz7-q>UhlfpFea1ABRL++hCloyI8vUVNLF4y)lb84HkE-Lg}U1V!aa!F5+$hAh`TPR9j%XO-l9!7JlxKE^+ER7#1tej zg=vd^t4yfx(rb-=`0G4-+q4OWTmJ+hAGMT1)x5o03an zEudf?V8htwep$>JN+`$Y_r7}dI@@Shx%w>GbQ$A5ViW5!h&@HMA}GH0{t;%a{7VNI zP%S%7Ea_KauAj9NS9$sK9!KJy-CmFh^uOwLiJwPaN&n!8r8-RK0VM3)pvm2Q1zktZ zA3gl8{bKFRmWk-_L@)5Hd>N`iT*?BO+7jP|lxX3BeO3D{|sss?3i}CI|&H;s?pEf$vo@3^qwZ|#P;pt$jC{$6!oV?MrTi@iqJpUTLnm?jx>^7+Y?AHzq%TyJJQsnSk8n!G=G zTQ%xLa3Mw|fEhtZ5DinMmMH5Yv&ySZ7WH(o#q|_9dHsYsUD>PdgvC#Ok42$1OkO$1#GWZE8hj2yHS|+@_@z+m{&ydPH z!N@GklD{uyE3)aJ;UlGuLY0?p!W64VT=|K_pQiX-Mx?+GYjtO(gNK< zvC|UJ$^mF$=|N%@Rn=t)>9^Ce|GI`Ha-3!tjC;Bz6+ZPCw(B31fBVwzz*ddwF}?zQ zpB3m#K7U@3l44^dY}(1hJXg>y?xVIJg8=VU;@S2xGk3l_Yjdb*%09+;+r~o2fcw8o z?dUjb-Lb#b+aFKqKik!@&eZ^pF1x+fBfAGWoJn3fBB~i zq@cRY3KaIfWpXtf&kkRA*^79e!?1DG5uPW;4Kclpc3p4_TOO$!EC50I#KQ30{PnvQ z0lTR*c8i+LyG6IE=3lA(7bvD=2{%qV8XH@-d=&RTV(hN|ektTVar-}l(=M%7BI2u{ z+lr#1B|)7KK|w8$GEbEEy8{4rbouA&y<7St=znJb%&8a{RLQg*5&~ zZU0<6sL+E6wCkB7yn%o-TgWZ?g9~me;1f@r;VGSZt_~&S3Xx`Cp!h@(cpxheayk)N z{vvdbdOmUlhV36hbBVG#3tNg(eJ2~bpnpPm=*BUD<|7BTV?&bD*tl-@rO0=Z?HGr-P$my&F zoY^ERWbHPxdRLxbsZsV?H{N zz>`rKx8n(~_?9g7lb;9C^aX7Sh5Lu$ytk)rPL3Z$9nPpE*vy0!kt8N@t3R{Pin!3N z;2g7wQ&zeF)oE?w`+B2<84w+vPc8u~Ge9L^5tN9cd!g~lUb$9pDsF1-DdGn+;>+rXVJ&U#d!nO~&p@F3<`J#@n^}%} zdhaUss1DJoUMIFEA!PNR6A~C`xD=V_l#awiYUfEL<#!dQ@muL=o3Oj*B}eV0qZY5Z*!sIaWf(@hLvwH(P(L^&=4-2i?}jUhIj7V0 z9KHPY2#k*7>YL{xQjsfhD=t(6Di@8{s!xJ6suxZ=#pkTpZ=K*DCBuKeO!kf50KGvy z)i2-vyUImzvXFT?o3cB;t+(6SIfGLv0pwdN9AQa2A5#I1WGEbT>;P*qqd`Vc>9o>i zTJ0rK*XXMbF1i|xlzuBSDTzsZ$1d4R>m!BGB8w?DXNi07HeJm1mI)HlLPOpYHYrwJ zA9w%>0tLl*f0|Ue$qy*_Z_k7*)1U|Ll?#Mf_&*7*CaEyqI!%+k_Q)65JfEiJ8!n_I!-@juMKA==+n551IH<#UM(mZMbc zkJGsjcNM;OV6Pnd!rcpCxRHva(28p6Sdlt*_C%K(g&N+>!~6YLM^tMSXXyH!Ysv7Z z+L}v=y+!u_wuD!SkI@{Czix!v4KRbPtE2Z0MTcy@lHQUHwaRk%7IF|jvlh=&s3fiy zi3N(Mr33-rsdUiJjrI!HF;!aGDX)d8^;qryEkAEGb-4^GRP9sFfZp-a)r?m!svZlE zjOW#cDpQod?2H3G@xc>V`!%C4P$XpcY*AO)o8cgUcO zB>~7*Ze{$1`@{!2~0aQFpo~q{2U?-7ZLU7B*r4nXf0GdJh0=en>{|BH!UZqf3)YjlVBNA5WP=X}> z(YO<~^J&cL{N%B&`{kDEegH`LAp%I)6|0{%2^dHhX2c`QjzyADzXXLwr zN`9JRmC5`+i`yXi{Q2{mj2rLUfPyPKfvYS+(|-4#-qY!g*E-70J#hyBmG9l?wU6m8 z+pW7i-yc)qGc4=>9BE6|HphtAbjbF-ludd51&Bv)WJ4}hllI5Ul>6H zuOR|Yu22553udwKAucp3kuD)_>*!bj#Hj56X0a?qQBF!hn`q1Lzek|v@BxiX{;IjG z4?qvgV8n+!Y_VY%6XA_)H^P?Lz$4ynP7ukOXCu>)UdN!^K2O~cQqS)XvS%5YDI?Km< zrT0`L$_RnN_VvnCO!X#w4RBoSRD?g@hcG3HY-1#^<>zx?V`6CNv97MJyh}fb>~YdPTTVWsJ&!J~P?bq98v$<6)dK>JZS_ z(h=q{!04V^59a)2%Gq^L>2y5sFX4(igoJ1=H0=5@Es!^$_fW2o^w zLrj9(c5FdBAWyb^(f|a2*nxRT;nr~B{b_tBU&8Sn?9OdKxuoAcj**#}TvJo?m(vvU zOs%KH;OLtTa);?4SMPM?>f?!c*VV$g`NonCtG_2#51;BeeFNZ=m1XBN2%B&gj3i!Z zpMzY_>^;NTws3PP_grhRens1FCySr2!7y;6#!Fu0m#e+ZgX-lNw&;0hKX1m1D$CpC z#V6|V?criD_jkW~a%>3%k@Y9dkrN5X_UC>~hc%(!7rIyF&)O>~bRB!E6lB#^&~$~z zQk}qOs9nJ0eGS*e_ZKF6xFVH=F?q+^4?yIE9`2r;xhz zlxL4%utugm3@V)4F-%fZUEK-*;1V{zfpD|9IjDE=$g7ETE~fQ|sArQ@|I~c6#p1`T zOa}{9gDEh_I!87oZn)LuLFOCo9(TM}{E!gl+(}6P9MWixWG;ixWg6uyV}1;~q7`Y( zT(7D=$5;Z)ZqY1v?DEdX1oy$AiA@n=X1d4L`0(qEPyB>Hg(Z92zSx~onWN^)F71ohDj4g{>)+3FR&K>O$@57H(@}WIHrYYHBwR=Vup?W4q*yqnx1_NJ zJR?ETgU%rutC{h6VERNvqVgrsC69Tr8y#s%`pUXjX=P4=2_k!2lT7(=PsXp~p}4YX zBBxx?USey6R(BFNQ!Ev@0BPm=KE{ING+6fds`Ja<^>ge_xLGn={7v(#zqnj#RS>gY zeh0!R8158_Z*FaERa{jQ1Vtcx9s!}xbHsaJL!6I=l1fuzr$T_NHgZZVXBdl zh2lQvxokfYfl!;qOT@+2tF4*sAS0;1d1FZm)St0&O_^jdv*r6aE{2omca-xF&0ir0 z5*%{|dH{I1Vs_yZa%Ufr1mz93k1M3s>1iL~PX8#{@g8vN_%%5Z=mc{8WCLU=qEm#L zJE8a(%pCSoKP)X2rbEXgEj(BsmMh{`tAg*BnN`E%u>B>vVhCJr`dbxk?Wp>b)6>of zdcL@Zf&PAVfIR$e5Ls1K^=$t1=qQodzpt|j%x9#XtIe6jRi7nsyYxyp4IecC)MM2p zm;JH8QwLH&yd!YttXgg=Y7~UBRy@|e8!&$G&DO{jw?FcD8?Wd@F}x#Hg$Y` zrq3Y2dq^67N@R`af+F8JjCvo%DLA(ljc$s0ISAsW7;a(#-x9^k^;yaym-xgJzoOrO z)1&dVCyaR``Z?)F*DtgE1Dz?Q^3yp};gOp2TLfD!hzD~^Tkdnwvw@h!Bz+ht(p?10 zggJIRt)gT~S+f}S{0JzT1J93|fxt=_;Fhuk*c{$@Ftdic{6hl}EJDRyoaY+NEz09@ zp)3{=xAfA!0J_D4-AwP<+Zb1O_fMWh|7y&s|8)e=;TxIorw`$`|9CG_L#-F?O=*be z^*^N!fyrrj!i>1mGVw2N@SH2Go_4mgNVXUM$6?ayIU#6Pl#Imncd=2Uz}uN4XW}+g zc&K^6X)&I2J-H9V-tgL~8Dtr(xgvaF|Hf0ZX zvSj-f_W0|=H9(V12&d}*1cZfi%K1HuI=?o&3AMj-pg9PE%8>or7R2~CD-aF64#kdN z(uWOm)8T6=A&=%rk*PEwo2YGBzZ}Rs^s?}A9G+RkECTS|^9T#~*Z)y_{ybSRJ~~=2 zyZdMhfS-HSoEyc<&KnzSGyfTbqaW8~?8N)d+LC5f&Q2Y5oGAjss@gfI{x)l143d)J zpAAij_)BLVUx?ooZsU^fMst??EqsWOij~U)JZU zk%i|q5hW!3+lH_7#nlJ*e)-_3`&?){pPs(yo1_l{%x`5A610H-+ZZm!EKy0(`%5Jf zk44!%eFw+|^9Q&y-(Ly0{(g$1MXITN`@M+Z^C>W{rK?B{aG7#T6oM`GqApMQ+kTCwd>5a(36CI0QA+ZbZr;GIiy%{#W z{$^|YtZ-VZKV9_b@1~;l#~4|~Nhf~gGj9s7|K1lBb3VoVd(Wya;Qp(`%)p-;9Y<{*Bhe4HBKc3qe?OqVs>W1$W4w-6y*;QwW~(saSo{hKdYwv-K-C6emy+?y#?Hu~Qa9>#D17(P~*uB4e`9dW5@x=ox&vsi0 zsT&>;YI3-zdIPzB>!#K_<8pXAnd0v~8+2%|3jUy&|IPCfF6B+T&Y@JpEmF?bb(d265L+Idw56^r`0msvzx zvxB(xO?dh*QQ_U>42R2S0@>VG(*XyX>LPbsjj73+gzmBG=vaz~NemV93VP5{zV@?k zcHzWp-2>U!2r`ksMxU&}-%AC@QVcG}&UK)=0XtZ{_@`m-G*R+Fr6461z9gm8f^Qm3 ziwcXIQ35v~3?0;nxR`7Q`##5KySCbv`5apKh`8L(vo+NU%9b;e$uzKda$GuOo9fSp zdQL;5iM>11dpxvM*0HAKNb?$<6^a=a^Wmfpn~OF6m;-?_oZmdL`-PAv5_|F=)*cO4WDJ2vC%Ka%pcx6qrV3E&J=;-=N^L^j(C2HNuw2(;Y z6V^Aok;S>`YGsWTn!)P*S94dIRS;J8wC|rpqOD^1UNnx8ZYyIbN^lCKP!lIrOWm}e z6b6QhSF^cc+cbau{91RnuGsX{a|l(iN9Ib3R^3D!{(uFJ&c%NjLx1kd@6e(H`j771 zf+B6W?X0z<1BnF-9MIVm^IU%3H^m)uO*2S+xPdQWe$M5NE@c2)1BBT{=h-5XE0ebk z_uLL!(E8rkSMYH};H2~5@*AF`{3pe*XKpjhasNK#!KKM!i-ss5gk4&eMW!I*eW4U6 zFe^R%&gziXwXBhik~%Of-((zP)>KGpyI4~~zPKKb>xf3CTlG@OC8X!eb43h6t)w2y zVj(nM_a8!iTtzM_i#z5Zut*kp<>$`XJ=1t4xo}|(&NqYAck$2l0 z!RQuy1?78lWMkV#AJX^*SYEHJ9BSc7!$Or?{|HUfU7F|2iCrr~k`LmoBa$=Wo$X)=4qBXQ!D8IgByxECu_E{(FBkt)31ckFG zokt)(Q1Z=VZhh}HqHy&>lshk3o@#;`m$F{7cNav1GO(bOJ3S%R)NbN;t#MHz7eitl z{GKe=Y72v_@r$ZsGqOsZ1w;WFaYgU`mjtAK4W|3EuU22zwZo&alLZs-Pr0w}w|_g? zougxSaOm^!3R(S2%38G~I4gbhVD7WORn?A=-{5#Ou?R&e*0K8FPHyeL3r3yB>pK~C zh<2KMd-Tz9nJC@HXXBHy&w4K|b+?g9uE!&(54+ra8H>*-aIy&C1!>dz`uf+eJKL7v zdK#)WE(zB4|6-B;vc4F*-%f|LT%Xn1Q>5n!QI7qK_@svUzIT&na=FF7`zzmk(H}u>eA%N&G`lgkb?wxh?Y2B``2>I8UGZXk%38mviG8IY$ynF)sMk@vRv;+HC0Moc0C=HSVUS*>}CVl@{v9g z4|V&ijvdKp&7-f8WL2sQ*hYT9hQx5t@9c`o{oUXsJA(T?YCPz!UZ=I_hZGfErh|9w(L@c1{$H3gvug~jlNid}N6<&MCv zJUMhKriGN@JUG;SZlOC4eV^eEKs2J9Ojo_) zM#SCy==NSYEiAs}^fm(j)aRbcHz!o&XB9Le7WZEu?a)n*e6uNJ_^Gf>6f8AbPl?hPIR0(xL!80dLFXGu|5I<8pQ)jx zYh*Um?xKp1v*CH$c@Mn3zksqL<&`Uj$?rh7hLE1#x^=53?SQcUjb_`$TLFho-t|X6 z6?buU+yOyJ3USseaqKSR9#m5U>Nk#_lLswm_1OY@dqdx$AC#qm)+cZ8%{jG?)i|-^ zG+Z3)xqJWpQq0(E!eaWvE-o9%$arjuv+f>cz2iD*Lej^V@x^?ev8u8sz2~nOptJ=zY%%<-;(Yt zodHR`3YO2Q{A6$yzwMHx9>P0F-Cpg2Q{>2f6xy);%5kEMz2k0Nq}*w2GPkB1xlwX4 z?LTX|C8>BB8UITK>^v$I(UE;Per?VAXQZt}jVjlFL-;3CM}CG~3H&-@uez?*14&)6 zd-oU`cGHBwJiRpF#s=lGGTT8e8{t;5(a6U4#B7j*|4JFTJ^U{BrpUE=j(PfUP>1hM zP&&$~;=^CYLdB#rYj5A#AL01>EKy(G%YrA!`8%I-Y$#0ceW>zb=~Wh%@b>5w%LOv` zBI6o{PvN{cw^keIO)U8lH0Q-yoR);O1dKGeyI#Y-VC`Sr2=EFG9IhQ5GO|+XLnOjs&glPD(nx z^M>q@f12P*MQ&dHi>Qw@XWZS%&EC^0zTrW&bSp z(IZ-bg1%Jv@RL~U1l4ou!2NwP=*2xxfdTT)yZ+*a#DoVNYZk5BJSIK0kK}P5-cRRw zgCB)moEXzn!B?N`z_R>5qB;o z2{uTVb#W0O^rD^>8iuum@pJ zyK!)M+ZGW5i9IqHU@_b&`7v=$u+kH^Z7^xBAlLl%r<9PbaHlOZ`~LS)=`F=hp8pIl z5!TF}K@-tB{=IYm{YMsyvsPUO3_QoY>H(|lDEp;Qv!x?1uwn!&W0*wV;ElD{j!6o$ zOU6%Am5+-G4}sDzb+`mSzq*ORzoT-$k-f87M+NcRyeUtG-=5*r3G}kuO)_?rXB1hP zK2>BETA71-%07O!459a+_7~Me4OOs;Z(H1VypzB`jNq-Pe<7`Tm$0aroBRSVj|&V- zD>EoRbq}CB`bfD`G1l$G#a+;A1Q$}P>Zr}!SUDR;*C&no(JV2HJB06TjA6^HdTE)M zm%xckf`|esPen_s0jPA!94CGzd<^Vqa0c?;9WmGWk}LiNk6hxYvEl9V_>ghu12kW% zr4%tmv6@zk@{5=)Q3&h1(eU8XW1%0>3x?fGQP?kuR6w9y6Pqw@F>iBX8$qI zj-4`gq?wAh4Z*g&Zy-Pv3Ad*x?ub}Fqt(W9(^ZG|O{Yyr^Z^gDLHpoAbLyrvMeger zH0;l_dO5EvG|w~Cto(c&;#LURAJ4t*63F305UHVJ5V=y_2F3$M6|zxGLM zk1F5ME)&74lxNptlkOTlJlzoBO~f~hO*XlA5TWO)BSx(3#yxMdVT#ojU#o=u*9I59 zqpn+(^6^vrw-~bwni!TOO#YKjqS!ryNARD&ob`eUXQ}j`@gCr;n-}xEjN>Q@A3N7; z*EgtK^mrWflfu1147ZQuIAw;{q@s5o!8I&S*@~5 zNC`_*xIy|>uKV)wri>ct!#(yBKhCLu>C%-Z8^?yL;ekdkUH4z}J|dq{dG<8FsjsV_ z1<-1NuvU9g1COAI>iqF1jYYYaaGWAfo&Z6SH$1rO(zGQws~CN~`qFK)Z$Fc2*ujDH zwW_~(&Xrp-9%PH;cn5r|iYOyXfb(MXj+X-J=k~qJ56>VWr9)%?{F)MwenE}zFQS)5 z2HV$gjPZX|!&l`O%DHhZJ?6G$+X0DJ&!721Zt?5%l_Q|WAu$;w!)wHkZwL##TKaO| zuOy9}aH*#r`=R2tr^K`eDKlYgaWdfp?Z2_af=f{s-ULtND=QZ$MFyn4lg zqFVaZY-$@M$>zQ$$7R&3EOXl50%o&~RUgEH%Zi~zoID60JV%YfXpL*|wLg#l|3BZ_`u+CFf4kjz3YKf7c+16FlFj%_ z3-90=U1LbL+f{czc3Vd&v7z#YaYq~j!zBd6@7sQ;8_}8Yy?3wMnRIw(YM52dR$*bp zJ?V$8%|A~ss>u)8B|Yr-c{xtvfwmWX;eq2s#{E>aM7PQm45~mW{<(1p@7aJ~&iG$l zGw^R#NS*^!{q$0Vg z(Ef}TnB$>K`T8qC@>8pRURMy zRRCmJH~fz`LPZzI9>f@LRI1BLEgCjgQ~YPVs1-X1aU9pdY`hN z6^evo9iXJCeowFQX()FSfwuf2OlwYSR4I0i;NDpaYMy=jBey@{N#=YZ%OpIvNo>%` zuchrPn|sgZ|E{~UD^6p_Rp*tz*Mo&x?ovEje5;x%SRy;}DPhVJ%P26ylV~Gq-339myG{NQtw5%=6_da#e zMLYz&=MVbJpra}f!~etFTZTm$b#0(3N=dhLN{WDlbPwGiDguInq)1AmbccW-QbP@m zAe~auFtl_E!_W;wojtzq_kHKj`G2m9Kj7ueexAMeiu=CTS|lXmA9r}3Om1ZTa z?+@c-Mzyucd4muRG1pL2f!W>DLmYPts8b9l5`w@t%(9I%)p>N8G#+2cR9^oOq5b+^ z;l1zYinw94zv7$j?j7u`?A+WR>uTD+q`WHccIFPpwVk(9JH1bV5<{P5*Z5g~a|Kq5 z6_@8+(R6dsfAIZQRzHvbaTdkw3iz`4I1CkdM4R(>HR9$l2kRD&H^~K4jX?K8sas(K zl_}ttuou{w%G45+AdoD8RR5@{F`Vc=+Mb~tlD=s~V8hYNn58ki#Mr-*ArR^fv$b$<)F<`ta0qH2K6M&A|ECuF2pBs*-dhkDs`+g+B_}h_cN8{!21c6G8P~hWT3O{+f5Gy^eeX3fPHb{=e(`N5XXmB$bu2s^0^j<7?L>p|CX7t<-Pb!ov7lfKXzmI zLD;5m$9a2^BNNYUVwD%z5ktQQ0`YPz{)VYu5{4rIuGY=D;M*RIZ!VZYPgJi^2S&v| zFPD44A}@Cn#-1*tuD&5CJT*1dmaZ#y3mme#kKOd5WqXPO9kOEMM)APMH!^E1<{iM_~W zWo6ATieLOp5E9VuC<6e!+^lUI^a9V>Z(sBSec7cTrk39y4A4+m!IL`VfE2y8b)=fc z_uDB0c!rnf7T`;pW(Yw$rsj(IXqleVqJjTV1?+FHQQN-7^3>(1rB1K|^*fF5D8M8j8_vYZggKIv|O7 zDf;es#-XM2q+`apOLmGW;cM@J0@h<1ipSZ96 z^hNAhn>LPk;d)2k<215m=Q(oyh4w9)=A%xIoXq4Rr|NPiDK*4DyU3IV)4MwJ)n7%Z zCf@aPQi;$UBnd*bu`V$rr37qU%Dr8qSoD48W>5t;ZE$sDf9anVu8uVh#XgGM#Pq9d3r~)P3hIgsv#@9_oLeLK{q0?tQmFNfjF9D$k`iY*t#{k9J{jzX6qtMJWThORO-xon|?hnglD`GAYZsbdqBRA?dNdTOJP@blWh@l zNz4ti!X`S+3^4<4XN}#vAXI7BzxS>TbLpa$!u_(ubg%t~%F}g*JiCH#|MQfBoC&FK zmgp(T8OQx3N4%Srf|cHNY(EL{TXY3k97X448|jm{L_4^Er0sl3Q%Prq=wwdT1kr+Ix?8dP_`_xB&%Gw*Ho`gavh zBfc5&E*YJB4?dz{^U%WVs{hE@srE^hd1ogW-b42_7CeLHJp>)ap?V5$Q z_ICLxPp|IktK&)2v%PjgFboe`{D=^J^~A|2$<0;iVTeeBn-dH-Fi*KA~{979u&)zrC2~f+|y{PjV9$HC>?l%fltZ1H+kjV%%4m+1-hhB9x?b+4%?M+K-{xJ13 znR{g7D!5kChLx96M0LCk(FolRt0J;T97s-$U|+KXzT4 z0ml1BlmWwqeGKkZArBrFOGJLh8Q;ACD{kEm^mTF)XlrXD0zAmCUo!7!Yv+b2$U@<8 zZNOs9>V$d?$cvq{(Z;4~Mb>>j+m?4cyu^GJq0%*uK*tTxd=gj}+TmQ_y9(-lmRk1I z7l1rEYX6S1-T#~afU3aZ&buyafWLZ|RJNX~A>!xQhMPSOEiIzlUKtzjgBinNEd|*M zN${A^P$FUsFcmih@fP<&7mJ1E3nH5rVQB}R&b@i6A=ANhLXOIm39v-)DZ{)3r?Pb60i|tdP z2c(TcfZfO|9wrzUVx;l;+e=<Shq$=7h|t6+#CQ#Y;bIG_yiz&3Tp;rgYTjUN zyt=yOvc-?nqBR#i0CV{2Qu&Mxm?Z5uI4lm5;ud6!pJ?naCPHM4k4;3tmD{? zYXV*2o3Pl386w)G@U|JeWb~cOtFLu=LflSqwiGp7wBK`!8;|A*{r&YJ((K`?ATM&% zYAPX}7@0IUtU=?4oa0+SkJla=$I@{OW&=~uVSvsQ`Kz-T@^9bfOoU<+%mLITBJj-U zdA$p#CjoVoi0@<<4a-+^gtv4mcCA6KjviQMrW#^2iI|oo^61FL0>7Ql;2wD(F32d0 z>LP7%E&hIuy`(*BR*StAMx$-7fustXW4Nd4nt}s9h=>o8MH#L{-_6Zp6b#zW6kS>r*;jSPjWsB+B_Dam=wJE)>yHfy~>xNY_d?O8mCUg2o9?FA|APs8saP^a$v z^)0@0FH6iegu;MPU_Idc*bK(tvay^C>S{5Y2$9~NdK@k|di<)8!^QL;F5?cG6t{w& zuvFsn*wWD{{vVGg7Qgq1^I2Spq5*I*nu|r_>|?4XM7QKn>wgi8wN|~Ppk}c;%DLE` zYhEWUapf#&*%wA=;@FIyY|7)_b6-ZHjBjOKYLYyAdJDuK+|obCE;3rgq;$>U`hd>p z{De|bBcq{GIG5=0j35K_4uiwmyrQUu+JGFqqau84On*t_dcTXo3d}rsR{Z>P1m$Bc zqrQJw&ii~8$I-kfG&B81oqZ(Q|%`L#No{d)48kdz;CA_3vM^rM4hu zrD0&>L2_m8@c9ZUo1?yhxQ;j^9KDNoO2{z^gd~NaPd&H64NfSrK!;&? z)~bbn`P+MR)cejm_pA{%jRXu@e}G&;Jb1JG#;LO6bguQE)7~Xqf1FBrC$^MEPpBWX z_5kYLRH9kdUU9dr=emMn5Zbk&jK|l_q@~6U!GOs{0Q>^G?I+1oZn*aMYnc zz-WrwIK&^Bbb-vDj|>u{oy6E(E7j)Qi#vO*XXEbIA^c}`yB~N=%kT84g!R%31$aIB zZ8(mOXCU{;-A$Wp*rYCeh{U{~kQo2w?YgjRKI^sNmRsI=q!A8w7g+&^nQVgcq_Fw* zLIndjWYUU{yj;`;L4+PZ3|d5EI~E}M%W$oZz72kF+r%cJ^E%xj&+b)%b7sfKymLk;X`RTej;pA_$N_w*K=sM4t`Lo6@ z%s&v77OsFyh3?6dyGYkIG;pg4yqJ0+_X)fZJ)Y5z=MzQ_VVkmkEf+ft%i_;6zJ@ig zpaBOnGc$7!Fpz>-LYI;-9&X6Qb-2qA)W{u?D8zcS@D^;UdeqX$z*{_*DfYZ@&x;E` zGaJzd*M?~m+g$%BRo;=|`l$xn==mG-$#N1wfD>$S2 zHvz>LcR;uJ+cY#Z7AnDc_{iz1qBHMtrzBouB3fD=F?oc0%jJPu`TMB~3v-5g*u47+ z{=P`L-;FQR<$c_$`{yA#2-41bQzKDd*spnUp9hF!c+qxgrBd{v;TAl-kKKdZ9a&8x zaElHJ7lbfT%Z)g0vr#p(R*!{W?7skCn1qT9&(=#RlE&dKi2PMC*>?fcC6*5WcDX!I z;7T1;s>&f9)z%|aUQt1kA?-H%oT+}}Ru^sem8+nSqCV=4f;fY zjV%Krk%j@2k;Ilh{?&w`8Cx7aB-ws9m`dC?#wiQEK+@sfZDNR; z9_80jy-k;ayx!m2BU*^yrT2ttk*zOpic-|mN)FbFF+8fHrhE`s>>U7cT|Y+vvQmE z&ta6$#fkORk`8~ZW|~+caX#w=x*!=%!=KWIJHfOK)Ntcq#Bc5M?}`WgS|AW3Okqg6 z%VUg%6ol#rdVM|^tHSZAmsP+Kp?!h zxR}W^M~`}c1BBvX`k1Q6=o*+#Wyx`)Ga&F+6;nqzgB^Js9m6cdq53Ja_eSK-`_PNP zEVg`%AQLrLB7~TpK8_-*cKsa#>TwVwJmCG^1@?PURRI?@Q>OTR2y1D3=jQbWm)3@F zS&B=qeHBEgGk+sT5mU}m;dR2yOy^m?4^M&Jd z&3ezSK`MEfk5?&Ac*IZA4yP%5PIDs*rq8N!+D1E#569wC zTQ+ivlggQeH;mS+Ne43kxk(f{#SBxkPh6~$l{6PDEG`nUtM0SoS8X=`i)#_}9#OCT z?q1jeutRVD{4jedu95#F+wbO@x$MR*=(-Ze2lXAYTyQ9BjQElXo4!-n=tZ)aa3It2 zqSp`Bdz#WEpM9vGP*wq7AHMKD8dl0G$m_~p4suCrzB(}~S6@;f+tKP$NmD8m4cPG1 z6T9t-3wH?ArPbG7$EM@XCDmqER4kz8IkEu60G^&v=I~X}Oz!$yLT8C{U&2Vfaz*8P zot8y+{&_@lj*)N3$T(ksHk0b~0z|TzU~U&z?D|2b@kTZ8!%lNAU68K}uz1PB2QhaYx8Y%!Mb08y@3p%fy$5wr|p=;-+lsYkI2HcaqK#`bAN1K9SA*#%ZSgw0DM0$JDZ+@{x3&A+_p@Rm#m%ogFLXPGnZ!RWD;1b9C&cm7WY`yvHqkR@Q56_ztuOV?Ig`AD!b(l%;X8 z15jrow$db7wu?GY$`Oq&&+Rn12JHad{sQxyY}09U_+K5tAWQPg@XUM+FE1~Pf|Flz zrf4N4UsYUjqg?+>FRG!B3`~$yjqPj9!m{rP$My z>nOP~YwEgp89vpovN_jW40_l4VnuzKEHcGO7W3 zkaKs7!n{8qCA9yxgQCQjb}SrzL2|30T{EXb910=yw-TPpSeXVL+T?LW;Tv1 ziMFS_1?aQU3FqM7C+{%c z>>WN;67*UB=sjgf7^3xTcFC{oe;=iu>xgqKO#vg!nd_m_x@^?-1YQA)qvCRMTd}JU zuF53&j;8CaG3R;foGVb}KvZ{uJRFagnXj}aMaQgPIcGCEJ{1)eb??K74~H6jqKhU= zam!UXZ<96q?k&pl$A~YjuCj*T=ny{*f0l{dO~{jhS2_wlZWLoZ_%3M8?9|Pv`YHNz zDMnKnQY^MMZ}6T$;q5eTpS=1^SUrZIsl-689}MoQ=Y!=EnXs-oX>`?)=?tyMEEL`5 z)cq_bn|~M--5>zJ(9{29L)|!i7ud>w7urq0`EaSyiAK!jf^u7WC(Qc3MBF#ymsTgf zXs1Gj2@bIiWsp?Z^OT>dkR$8xpP=r{pfZ|HJY4k8t_W+dQY00Wr)O)v$q4%mZZ7Q7 zjilQJ&qX}}S0FwKQ|{Fbe>fi|b4AAC^#8kH^r!3RQ4WvnDbN&OWTLG_S9$>9ga~+de6@J=-M1 zvLGKqKz@!w!Udw{_168Ho24v=(jP&78<RwJx&_P) zp{D=AQUU4byfLGO_a%tunOXkG*t6VCnV`S=lb_peIAQ6fq_T7@m4rA>{?ReGw{)50 z3zfC(K_Hn}dRc~}h^-xkL9%_+Qd9G>X8}37F}$vu>*<86e6+CfQE@V8ZJ{{o2R!Lt zm}AT=$&Dpky5oM(o(`C^&(CY$iyYi{FvY8>_`g$3U2i%UOU2IQ$FBXnWi7PXX5Nhi zMYdcjnx*C3Q3OFVQ3rFJAsK;Z3VmVbSMAuYZrbdjT;!S6AVd3i>s{k~x(1ipTYhJ5 zh_@^;{G7s$j5MsC{YQP80iGi;==v()Ze6v%Pdy{Cmb2oB_x|0e4@F>p`2}3K*8;|Q zng5(lmSta%ODmS65~OU$o0FXf?fqbcr2AbRB$7GMzJUlB-Yd{^peoZhence(jFUB|bRx zKj#8XoCg4G^xyjEDJm)i_#M7sncUPrPDgnY%~OdH?tXjUc9BQpc$V_2VcqTj;|rtC zO~m0RE6sqTV{k!=+}oupZof1(Oe`TsFX-P1biwiQ&ofd}{kr}Vpi#(7f+~O&l2OdR z@U=8X3w`(qY1Z4n><=O=+}CE7+}Adt?zX8|!fCjA)zAK|jtayy1eIXD`W)_yX1brd8Z~r1`9x~aw*F&z}Z|ZKFCVQAy zoW6Mbca7VK``^Zc4^Go(MYoy+a!GawwZm*}YZOflm(|!e3Oc$gc+46ppfcs<<@rIj zeu>&F%klvrIhCZRpt-rrdb{qset=5sft)TAxMsF(Ff^?VNERx69nfE0N6K0flf25K zz-pPbN02v*hJW$eaN=3a(*knquL;a^3hgB&q@NHPGaUh zi~XUcsrhW8TZnw#@5aY&x>Dso+QW8zv`z+KOQ6-(70OlO>Sz=(S;pYDJCFZZ>+$8T z-wi3W0I(UoCV&53T8$-HvXv(yB366wzI5_yQF~SDKhl)9O(Z@26>mM$2NQ4a}Z!O;E6&0+``^$=g0FQ1XeTPn(G@zi1 z!rvLa(yO${wT!2Za*-2y2a3}ZQG55d@Y!(Kc5*|X1PIqRC ziaF7Yh?x!UIOoy&!1oZKR|x$^xD&Y49`a_jJFfSO+f$eS*A8$TaRweW>^94nnKV^C ziwDi1gtoS~74O-QTu6GIbbu*LFW8M@=q2H3VgL4w;bn<^M*RrFD~F+>+TOKf*y9>( z>&u$)T*&$vpEu0!>X5`N`(VszuPp35oZ&A?|RofRr)^&!Z zDx0!4Rl2gz=irB*(@85!@}#*Dvqh*nACZNhmJM^KnMK34Twc5SmZsAt!{@W_8%pg7 z4@4Cf{30kG{dhko)_E`{VP|hY43!U=nI9rF%3kPiUoIL?IN>|*63R806k@-StNJ=S zJ4@E*>;M|AXf@^Q^MtRpPGXYBadB`EF=wpBnN-_70ELCm*=NGZx%c`Ww=70HZYJ`x zoq_RI=Bg!yKcGZF>)rRwOD@KOAO@o<`I5(DdOym`TY;#$a>tO8I*mqD)umMMj>dYa zeM>zYAu`kK>bhFm+Vdgpp`Hw4r7}y^D z^pDQ4oZ#o@C!SQBkvg45h^Q<)n4+_v;T}bJB@0@UfDRNn$6oMp#HBJe7*#QM@)2_J z1shiI*j4frA!(Hp((G*C!Z{7(`gWJ#6JM6WKp>W^s(087og6;=n%-b=EIC^`%&*gB zt^w+ZxNEKm7^)!)-x3qar>krYIlp&r&(q9h=)2dHUUUf+3=6@m!_bK2qm#I1!xVvY zU+onTAu$xzY9*dIbhVSzYiTaz>trY7*7pHeKw;3)I7F<##A{pV?Bask7Rmdz*2Lp) zqH5IBNw-2VmyGaLAuIST(9)l){^Kt1S{J*S65$PyLq+)HV|>z0SH3Kg*F)-%-7=Uh z!!jay~L4S76sbYsMFZR?uQNvjCn(chpWy2Hh*de37nVTbu*oT-E?)oROsUdJ&6BMw6Qf^bsr!NI$sMhsp*)EydK9gRDc;f76k{fZ<1rR zn}2H1kU^D)g~?@u!Z*Yx6(NvdBID`q!9fZ-`sqyb^l{>#jY6FAa4!}J}5Db?WGYj)RdwRi&P4C9V{b#)H`=kW6O$5A*z^3ox2_vvvv-mtm zMGO?Eal-po7kEKuy3y##LB!P5;b8tIf4CfzZ#`q2Uqkv{{jC9+4-Jje^|zuvouN=% zYOa({GS)6%oQgKZSS2>neW_^gP&>xN#29|1%0od1GXc`WsfMS=f7Ga|NZ#OM6fY;wnMu2jt=N%F z_F=JY8pbN^urhXxKObTE-rk2-wI6N+$)WB9U!+^N=J2QcKHL60;GZas3i1L|ylg;}Gh(Dun3r;si}Z=W4^KR4Eg z#PYCePAV6)XlR0O_-lRz6_BF|PcqX)~1$pG(A{+?WUS7_!D=~r#njNKG zgN_>JUonx3`4Po#rcqi_=L%#WQ2?n?fIcR-N4DfSHw8BR?2axi>5KuTZk zwTCcAeJ=3OW)dmfeEs$<>-9p5UrR-hS+5*RmHvmQa9jf5u=ImV?FMeIEd@hl3;4Km zlL4I();(K~EE^-Uq7Vr8b^Gvvh|?zl;(oIA0{vjx{I9sY)#!N81Nz{}LP69t^;+bB zHkX~WVk?0S|B&=Loh?6u$?5MlUzB2(uqhMiwzgmmtEGMU!t5zt0DDC_)+%cdZ(WJf zjM?vwW64SiW0aNk&+uF=qH!Y2CvjI?{!WX9yO}qyL|xvjmp>W|$)^`Qy2zTfLzz-C zL7Dzkc_ktG>qkF`)CU7$KdIlv`FYsoHP*0^nq=~rGc6fe4se1&NCq(%S;P($cP0$?!01Jv+gjt`UWZcCS?2dhqS9CYR^~pl>VGx z^&*xq7r$^r;n-byOpm<;b)`C7QmC&zq5^2*^})m$naL3Eao~S5a8C^MKR+WC zkYL?1SR!{}(YCpDWts+$QXix+=zO@6k1vpV^eH!lE6fGSS#QLzCfy8NB$wJamTO0( z2vrQa;w{d~UfY3%{7x2UUO$LtTZIdyyxFO!s9taVDQ9j9$f6JsC~8>=7JDLvLa+$h zj49g^a)o(vB~Rq4QSu9n04?v*+l38`*b~^M>Zqg+tGuiH9skmgSE%F81eWPc%{)S> zGnsxIRmn7$mU)dtmtOho$q6!$$=yM6W|nDqXuDV1y`}3U?p2G5TY6_kH?+EEU6i#)hq%E^wRRqERe#i`pS!dXFdenqGL~HoY>J! zR<|H(4!-!{iWiKzFg`GX68nTp6d8Qq{7*Wj}C{VBOpxc(}7Yr_fc_0Yr%O8`ztZhCd|e8levJXl6##m z@>1!LoD@W%QG61se4=7vNI`RhdpL5`1KIPnUG!e+MR3@UzhLTg_^`A;gwT4>4mr!% z9joh3FaRs^1w_z^0-!g0gC^oa=d10%SLjQ|BA}WNNOD#=V)sS^=^v7%XmpkO2YLcI ziz+i+zH`8!+Lc6sPI9vi7k|B`i}6OQ+{|wP8BZ8enwXd@OCZFgQHbo9LCvQx9VGTU zNZH2!6gU^4C!)uMXJRpG1VNy2$aR%8i81a{f8Uy;Upbp9P`PoOt5RRo20I>&FnaiT zPatTJ5ihiCsRy`%NFyy_AXZ+GIg?R zYHI3SSd|{eridY#u(t~Ql74uk?zI}HR7vqUZ<~y|)H6EnMnfvfCNntCIVVo{;80mx z=iL{sl&_qGIBu$rj2Yi-609LQu~#sH*|=;~EcfomOeHeIEl6$lesF5CT#*KOWYo8E zTdAmnV6BE42=UB!j})KhGaAcdlobwYBRLqQui#7aXeEy#|9N9&5)--ez zwQcT%G2yl_>Ezo>*;0Ro4iX&4zU%GF%L**ixdOCaL@N97A#1ieq17?mQA~t{{Uf!M zPg(5K`vh_FY8#l$>oTMyNY<3GPC9Zt)^l-4s$A3)6KmKafYU(#jV$whw)@{n;1YN>Nuu%iLAwlr`)Q z(w!M9)AT(>8+Fh1b-_nBZ6r>VOyLLtVo7Tu1LdKc?TXCGeu(39vw6+T8f_OQ4AtuT zQ=rE_Yz8HulMeHA1)0-F4AH@(hU^6Z^ zXFhhZWwe_)Z%uwd==_07kIM1|Y(rQGs>wn^h^!8!ec@?R~el8s$-b3sADGpN>~>}brZpGvV1+y&2z64g__(RM8hZ_P9*HU%rq_s8P1 ztOS4?BEciWm1Ai@#)h(%>+6&@2zT$ZKrb1G*^V}9^AF`Di@z(d=JHtKJ~W1(sz^E& z?k3n!$F3YOLTEYqb-u%4-?AJ<{_)RJ3G9|WYpw{*Men*-kooF)II@7# zKEV&U6^Ud60dwPiL_fVi?P1WN~qHDvZiz%}b*+^aRx`GUF&v z#_)^yc&#PK^g(T5N;v2gK%kq3fBZ_KUPNW<%WyFdYcS_QFobtQtbk@33vD1>rSp@L zEE}|y$iuQ#6``R(2ahZE_?Un=xB#y{P|B5d*y@ht${4g!GrU5F94~8~o_x`$&MCLn z)PB1+#N~-A7FdZlPs(8Of&Vx(1^?|z_p@P#@U~;lg8;GNjgRe1zX%Dz2jLIG9TfCx z^$-HNWx07d;fQ`|uS-rF1^6%WaqZ>7?U`&fgq|Ri>#yNVe_ZOjObaiYenwxa7#|x^ zZ6g3S8l*acw=;F{ zaUDLoIzPf`awkInZzq*3?Vs-SPBOpgu0akro3hht#SdiXse1LS5Q*hC6!Y zvU3Fm6j^=|R#57JUC_4gbsohl)5Y;Y5*Ong(EvK63EL5t?lOQS>#7q6XNUa&{Vo3n zO8q@5vQgK^BpGzc&@I;>VP$=9bEk$BRXF>Myt0`0r~o5C3`2iCm)?AUFHshJa2zf2 zax@Z?<+IihnL(|TBqhQ@^(Je*0ELj&JB70ysr-oFRgg*&C*oacm|g=y!%0WMecWjlIT~srY@O)4C2o3O2SLBd}?MdlLTjvQ_3&S_X%4K_7iPkljh(Ai!5)-6326$M#+$;fdibGClq{`|#(Q?E7|D*${soQF$aMuV2j0y3~7nFa=DdF!_J|0nd4+|?IPHsS_a{5+~)UZa;Hl~3A zuc5}MVtyV+QPg-G*m&G^<`4s)6pobe$c_q2LMQuAp09dt<@-5s<##2@ebScCt2FbTIT$6{Tj%+DJqAmTbeZgMp9L#-j6q6%vTmYmU(rm&ga2 z?NC%O9W{(>=`(XMWauhdg+KF8rxfg605D>Czl67TsFbUm&3BU+r((`!WpV=bmPcH0 zm^~7*w)8nGfP*VWSJ>)Pn68i23!j|y5iE3yk6D6EU758o5}+PdGN9cpc}^f`DxO<8 z;u4f1gr5%~9U#jq+?H?doY`$Ow(R&;AW^vR1i!ctaI_Aom`lpp&tJRwc&+ z?fHo0<<#wCu`?D}R+ApY^MoKc}t~Ako*+3g6InnQm)2~bW13D=UZB4 zVly9@8pQJDVoQbR=c=x(MgUBj+vuBkkfi_s+cWlxx(V}Ma;NC64xh?NVduUZz7afw z73Q|iimqnp4!=7vd3q~>QU>8Uy&GgE5KX}_ zHhVosxFQTW=h9lRu4c?QiK*|*JA#o+U#wpo@l}Y0ieoHEfiLV07FLMB;nN8U4%Q5a z^TuWV@Is2&_TA%KqaQUPQDRs&Mq_tyL{X1M_D4OmcNttE))Z@uSP>9aM2)SL)$`v= zh4{eB4nD0H*^t;U$<^c-7{nEGK>2{#ugdB*O4)Y^&FkAQa5U8NP*H`=Z|Xk^dChXB zMl>50mdg&Sq^|eQYIQ%;4JR76V`2}=3eC#ON+=O<$r$ODQmh}odO4#TkbHsVT0P>R zaCMq>?v+VkJq;P+LH<~~1I3#nIxyIHp?OVprJ>!I$d3(#Fon zPkL7+I3U`?DSW}7+z?_%7v_e-*UN6Hnvz!a%aS!55v=0G|M@$32as}rvHADo4AfO< z1;BrPlEnbq{6D`?1sBNw|NkHJAiDa=F&(v3fgvF*63n;-4@5b5p9u?~Nf@C{&=N|c zo=uvijU_fArF94!{cJjbmeb4TP}cfY&o+qIADu+NkvM>$Av~s*3ZI-mC%Oub>|6_= z)Yi_~y42QN@jvKPa%jGi+0NXXvu$yI-*Vz&hUif&F_R7=$o#{ z*5{J)JZ=l2u?W_`}7pLB3 zC|olx5JkH0LkpV^2hPaL+ZJJ^sbKY&CB-Fmcj=B5Lh_D5CmQi$F`~@kCj7ddWQi?9 zLEX}Ym4TOw4>leQLw&?0CX{_4WAW+40S|d(zNR>PSxkCMXwD#|!xBUM6RLoA_iePN zy6l_3CT00m-YB7`CG$m%;Y?qR@JS}~+HIwoSffG$>CwG{g#7#esGaT4MYr)uPG_<3 z&?Yf49#9+{-j7X;yW6r&f0lQ6KitgWt`F?EmzzPj;`N@9p(9{xLRwy=HOZGaY+B$qgAz_leXsFx?1NI!7$;dWC0-L6=BAa01 zq5I}}^?~HVk!USLZE53|7Xr|hT3-9~8vQ%Mx;5DmGx=vfjcnXsXMuRR!C%S_f_VLTjI2yp>n9~$XO0OM zLYR5aMn4?MX{%g&xh#70r-o1{fL6* z*`r_XP5e}_cmCwIH7sJB##0SsEswbQ&HwNvi}%YrxxK(@;S;)}$Nu0=2Z!X=(+4~g z+eaVTllj|+&pTmnG3NJ;BWYn+Z@xbObzqbiM4dtBwVrFYXdiwHd$ouoJh<0D^vmNV zo?cV}_EoS}+trVq5BuctSbZ1cws+Wf<9lC{E7p2k-Ff!d=`;NljQ@?--8 z5qL2A!zB+Nf_H^AROPMSIC)d=8&dB1hTo!;-D!O1qOn*%RiYMS)Sf-O(XeU%_w%it zfM3|#9-EFL4!^p5_#=OBhFz7w`hT~^C&iVtD^srhUA!z$`%%UB#cqh|xmH@h=wzFA zj^+GFR%zSs*;_Z4<8(11nTNh)3&&@afIx3gNqd*rR2fSVsU_Z1y;~K7C2#xez2+U- zN5{i32U99d7sYU=)N+yNR=X{$NdDA7XK0wYu{T`Bnr_NU5@Bt9#KTwaqM)E!V9(|> z692=Yk2*Pufjupo-jer&Qs&opz&M`<*TZXG`f9%@lJ#Xq`CV81xNpC3U84TZT2>Qt zkai0Hofo%R%kPZ`=B>R|dxQ~sC$T+R1OeN3PO! zQ+k}Jw;R2ZCiF7AzeCxbH$9C1OPTQe+dJNFw2Ge1(e;lM6|w8RX`kdh55LlQ_~%X8 z+ZgBZ)P{(QNm)}ryk?IZS4Rzc+iP0PNGc4;m9Sc(l;SP6`1tflA0tUsyQ#?YB4X{q zujH>+4qi}+y?shi7T*%EsHS8l(t0cSt20s3BR{vPcYRcd7?Ccnp3^9Gj@=7^aUdjE z`@ACRn#`xXN>o;-asAd)j%+YhDo77D$ShVq_KD20>rLL2w#3x3&Uw?=%bze4|HQtR ztNAaMiCPTY6{b&~7e({&)5n}>BwlLf3Rz9lM;E!R{1T&dvAc5rW`9gw{mWgcj09+p zT(PFLORWPQCO@6KC*8yzBUVsDQ0D9JidD){7VG#E!B$;$w(t>Kjo0?iVb;*%_=!fx ziAo&b*}C2o)Gz1Q_K!CzuVU%s z)l`+t;IME=SW1--j+s|WEb2|@IF5zNTOYv@>{$mhLV5t+Rdv!x{HlFM*$*s->uB^9N zZ)Jp=SqEzQZQWlNbzWMwi*RV@ZAuY*X}MFcIK_Z1UXQ7vhUezva-D^hnlt%y=G)To zyO-g|8ZWD5zkk&j-i~Sg@Lil?l>K|wsPo&#O{nAdkB&U7291UlH}54L$#qd@z8HOI zHG+JB_arsY<8a+9>-u%21Jg$`L5GD0UXMG&CyDMqJR;vTYB`r|)9WRuS?@}fhY?i7 zOL}ygmG6wrM}hjM1bV!ou;ZcT>s*)q_sK17G%82h9O|DP%Z64@y-!>bFVtJTH>e&W z@F7^aKbJ#f-MV=$bU@kk?L=}^cX(?VL-(BA?+JQ!AYg0_bshvYqb3t zo4E1ASMU=@QGM5=O4pI}dw-hWv!&AU>`%u#tuJCNTwzd(7#7dhYPtQX$b#=anWB&y zSyee=u)Z@{cep5KOYOV)kW$8{a@**si!uE`o$qV%bnif^hTBr8WY?z-xaDVb*1W$yHE1@aX? zXP$yI`3>v4C=1Kplz;u;nie1l;~E@asI!t&aV(`6a`6vpf0f!y_NhNLyyz!t_eG%P zQ9WdP(L>+OO^>~o)8ZU46b(X2Z2VP=sYv&5l*>8=(ei;l-;$1j$%J&&pDmi->W zTwkB~Fe%dr`rVs(tPpY9O(6X_DZ^X6KIHq-Ppt)|(XN-7 zlYAKcDUw*ZA&Q*#T|-6ad#;0P~0WK-Q8N;-L-ge5AN>nR@{OI zx0CPRGS1!3$X&+BN>*mpn)1%)O}ZNnHtnkJe^wSb%6*RMw^?eLeHF%p@q1Sh8XQyt zEcj84ay)(!@{v$1$iJJH(`R)&K^{*Q>Us{XYqopPVsEg%`E1~IK`5WQ{e5NU_~DHr z`O+>TUC4az9>c_V^=PXfWTxbn^dd*UZK%ja9G$wxje9DbWTRf^a-N5aYJ9GT3pr(4 za!isTlz zZsIAGd6A9-ha16?|JHlU{x(cA%l33_b>iwXA{q`|23XYSA!>Qu5;IpG}D|2Eb6io_(9jItn53qaLTWqP9^{^wZ; z;ahi%5rS$xlIKvS^OC{KV`u0tC#xt0$kg55H+Df1RLLj>$Z)7`l>pA$| zKh(d%${ctJpyz0GBJ`WP?Q#J7%oydb|%QXrI zEoXUcdQ4|G3LQb&HG)*0fg*~hgg8<@{V1K}qOUOpe%Nc7T`$;nl%ur`?lh2kz@@k4 z!t3nu?JUayT#M%I(#)i^yo`v8OM}NZNvR=nUNWq{WP_zCuQdx^Nhn~4#+@7jM=RQO z>ug97_eJPBVY}XoRd3?}a6cm_tso;QuJsKSyoqqJ?|j)~%@J)N*nT_KoRCCW*7|;C zy;^=Q^^}V-^8qQfLe$8F9Anpb+y4s7&E|GMZH~-C)WwS{ATU%9eqpMQBzCrT$(!gE z0(N-w!egs*lH-HwK(FHyD;W+4{@lY&MNScOSmGzBCZbpJ?9Q+{N{6+CbP30q%Rl;p z^{%`EzHg7I-|GS%J7pECOTH31;mhe|v;3X9@Hk#6nQo5P+YQ#K*C1vcbR$zqR{NOG zCi=eE_-iiC7{c<6l^$k~j5bc-jE)SI zn<+OLyn(}`M8NpyP}h-GBaJh7BKqEudmFF3NGv7`L1EBl!M!oOGz_GG5k*M zX1f*fI5#of$s7E-)aUP(bB#0~^cGuSHhij92)ldpK9ct$9FRfzKt%P>DNLWho-p+J;r5iktkz_t=GKe5u z*HP|DO41QnYE!1ov5KQN>%XzY%R^Nw`IpwI=KVs#u}&D}b|g1HFukNXJU)%Nr81u2 z`VPV|22J}D-GKB|7DxLHH$y zbSeo0DLq;WbbV7U#(o73STGnpQ`7n4;N9tr78aue&7q@YkzU(9ATN43IOnCB{$kCu zn%YUZ5w4M0GdHRBS1zYCKf6Rq)6^TA{AU7BG^(VJFl+M;Syvn=TRYxyNNm@oy}Ify zB)Al)AUCR_Gw*~eRVq+(A@ugU$if&$PlO$`QE2=dA=9<)Ql?i}<_Gc4f!q*U9U-rI1o=;G&RzoIo z`eto$lNj0Fym9G>NpEMGuP98Xlo|%99!p;*aOwXi8h_$GLFw|6P~_<}RLj~T_{y;= zd?Vm@bcMFAVJj)Cs7?Bv42l-dh#Al%?T8#ACG#$^6h1ZX-SGu(B~1HTzDmQLsdAUl z1Q~%ESzFzoKTMGLi&aRc&fbohrjHE>liN%-pppmL*6yqraST3)Hh`k+bJQhvXj0~R{u`b;2mYP?`2iY0zo11}dl2En3pTIe^_d|uS$=Qi zEDNB!>H4(jPAidS`l_hu%6-e<2xt{zznO~EtFCb*Ww;vLlxs`YW%w#-v3V`y@(STc zCSv=9NqMz(;#g-0y=IG>IbmtE@ctP>+nBCWK;p2XfJ$FAta(k%AV1^3I(#PWEF7;k?8-B_Sk1qYGBob??_-%UVaV|kCHcKX z4N*8e!#JHEiRg$TS@bBlfqJZ83TtH$?hLp&egDOk_g=m5-7_1JGjB^3 z9V)?!pH77Y zLUcwz@>Gzm2S=lF)B9Idzs4@>3Z=ZVBnU(Nh^=khh{$# za@$nv^ml&3*t+b;(Nf?*z_lW|qfv6@OSjbvhMQsX&EX-vXxD_U@Hg_tiraBy{us@i z1oN+dqoCPad&s?E!a%P-$hweoV4pN>6tlLuS=wqx2^v6Q&;Uv#H2Z#6O#;c{j&K!` z(kMt~+j(=Hy5hmb3ttydKPoVR@fu-wP*Xg@?FID50)6|6r@HOly!xrl|J3{h@g9Hl zmgW1t5ZsAahf8+_O`B|lv(Dr%LsX1P0XMic`#wa71xnldpt?%7OsU6rf=Lfl-d!t@ z)Qjt(sDGc0FYByY*A+Xq!IT2~ZGY|0wAbEj%>lj*(JpvvB1w8+KKfg|L9vEyYZ5*G z;Yg!4i8sONOF)rahT#E`v(t_2WaMN#AkAA!|48?gcCZ9R)0j8A{;fa>&|AS7rYKKe z>)x_0b(Q~DCSU)EaB72Yv9x~Dvte2CZO7vDu^B%#D&SL5J?lW%{C(M{Jf-@wmwEN7 zrR%Zf71}tP($+Jwt*RPnA?_1+@Lhe` zkKmBef%=TjOR6dwTHYtjXS&l?vTvO>xDrQKt{&H{Pgzd{qQXeK{@zOMRi|iQtOg!b zSg&!A#ny)OW4;nmMb=_)o)n~^X!u>^6L3a5WyJfou5cozoo8j37yDFhSoYX zfnBH!xs+X?73XC~5sJ}LS{Mqy6nfdN*Kz}yLu!OqdxK<&hi|u*Cx|MSTkhHYPfT60 zQlP;JQuj&CTFYoz@&r&tP9PM~(hJDpcH;Ol6^Q2jr?N)j)~F>Z>p~m?CJ)W@yN%VC zaqL!Ms~ubrsS`A=buqRSO!`Ify=@(l9!)K7^^b#BU_FMD4q5?GdvfExHkakER0&eUh` z$q+hgXdR0Stz##TZ%5-{UxiIbkNvk$8Pg5aBqH901j@y~DO4m)Gl%qDo&$BC-Ti|l zE0y<#^t$<{Mk$`e%0!Rka28taxofL-pM4t(^>F|fs}gwhk8X1Z&q>Q$R1r|0=HLtT zd<`Y_c|!C(URgJZ%V5bQC8WUIY|66Llq0h7#}0rXSaLitL*I^jD}Zyq0}^;)vjI%&|`gx({)U zi#$_bU-@#oSdE0ubQUEhZBF7i+-cCmr)9#A{g$P*?Dzr=9^3=L{bTrCzgO!AeH1ZK zV8U{pe{iD;+^=HE6-~+U+_g?TCGG;EvXGiI&Ca0=C=E*vIwyc>iXMuGI-jMb;X!dQ z+fPNRS^{%lvJKUG<*0Ox*{lPvW&e#Ar(674AFE#)bCWE_8{(SiV^6Yy!0T-D z1%ZrXD=CO8kJN1tx0)Wd*}&sMGwTvfO4KLoAx144G2JXICx#*I{#6gWc2YFuaH(U7 zZHr}UoHHN{)ZIOJ?|Uo9F3Dv@R^>fcJUZ-ix?I}v^C?qx*Hal&Wf)ZmC@@f}jb053 zP9~hmwceDov3`=*B+@AUhK*$gJl;beWmVxYRfqTe_p;bxL8c;18aOk4YigwoOLE^+ zK3a>WTn<;`b^WdOJRXO+&v*-#Qu>aQI~QzavdJ}IcXi&ac zs=5E3ZS5B~E)i zF`iH^Um}_RT@B@QYj)j*)RrQ=o!%4X*pTz@E39@zZ=}ldv&{A>{uyI;GI+(b<=@E# z>>62B#8z+l!<^-MeVND@8C;Xx{n#{mCtRCD5L^fHcvO8DlF_7O{#cM)dVeDP^tto7 zLs(v3fz_uw*Wx~isO+w%+KWI{4Ehscdn~7JmY@)F_EN+OsLZMjB>okgZD2$uTKZ8b zEx=|xxs7R>vpb|C7`hoK**!mx_xcWw!HG@jyu0FJ22>2#^uwclN7{W)y6{)`z8O;P z)i~GZEA7YkecWKCe75*Z5y<)g&Q6s>WL7?f%uH!D?qlUwo%g(B9o7!0GUs3XQQo-; zeYwHS@vckPdLN~E{aP$9mywDUA!InJ3=^FJ>Z$JdBJ-RHbjZz}IbmISb0xJL%j-QIL77thgFD7{xEYdklO5m*r! z;oHMg^f^A1ot`?umCYJ>Uhp&MoraG8W%rP&b(gd=Eg_|ZPd&bz^qz33J>9 z00l^d{od?kEchDD-|p$MQf$NwWLbnfB<{4Im1eP?$Q~*Z!KrVitwabXHqu+2Y=I(9U`aN+Oj zQRIbROiNCioMg@D&oqRo6^$cG1aCI0vZV$4>913+Pw+^KBn*k9SdLyW1}D=vmARhS zBycdvjcm2JP1}u!t9c7^%)mz`-XT?q{C9))N1%6uPZ`C*vkL2vWax5g+eb)Ab?>J& zWse=@1^6y-zZ)2W-;OqXb6#G*<^09KVl6+aeq-C;u*ZE3$LD$b)ME4G*y$$LY%tWQ zZnMjybeG*?tra`R|3XSHHgUn*?OH*eU;)WCV+6IjjMJ}z^NqkRJnl17STD;$k0dTd zgKYuUpv@BM+59SWe>S*j^PQ+5s>0zi>#pXuoc8C9u(2})0*Rg|#!je328g^m#OpR~`H=VU+rqC_oc*$gcm+&Z5ruxdK zS7q$v*u*d7eMrTjM&Bz{d1X9tsUC~YLrS?spd3%I8+-Ns5`_~x)@!gpV!enO2(^jI zHUN32@U(j3`#rQrx(McokEK-R2^VmbiQv8sHRWwF8~!m(lo~T2EamiTpc8tk!UYl0 zSvSF;w(KTzG0vQ{G}Ruv1Zi2EKj?3D67$6_2-ey#$(`q-g)`K;i_ zma3-}FP#|Cp*a<~yi( zx8L5dWPyVKf_vyzR$DK-zaS6R58yN0OY!m&qc**@ahaCOG(c2|4z2SUePyl5T1fgz zpT)Ei)#>J=2HLS@{KM@W_jHA)T$HoQ!J=k7I@3y=`!pmR2Ty1~ANpq?juzVL2rZt- zF#cCqG%A}OP{G5Ki8e6DxZgvjZoT|~;BrwAN*_-*n!`5GGNQ$?0B_&*mJcDa3It7c zg;$!BCs+O$e(ZR-DQDWZ(H?WF?lC*(bi0cG*MFYOro`IYG41 zF8u-fIXv!+pJ81na7}loahQ4dCc6jqhvF;8wdJvJQ&(xfSEz|9fBeKu+LzB;Pr1f} zblP+#*3Z~NA}ee2Rb$e<2YHFA84r(Y5xoUphR4-}dtJ{=2DFSHnd1Gg-36I1Vl#3I z{#G~;8%ZztOuGS!v=EkGepzx00HETP~t(^WHUgZo9!@%a~&X?pm{WlL9@Qohbk2uX(&HP0v|CaahBC z*%1&zm)$GZXNh4b+vm7es|)zCcq$92>*jaw68U@7OOq2270Rvt9$m@*m9&vyL6`gwNMu?; zNJ?~~hDvhSs{cpe28D7$0S)UEbRXfa?#76xD@^(2br_VkhW?K@?&i;b5UTsIthwCq zsPS>x{f+-*W&iV3g91yT?12A$m}1ob*y8{9m?vNVA1_*eY~j_0+R#bJ%F4E+WG9w@ z)nyG}pGyiUDFi8y21YyG#r*8JpzY)&s^Vg0sJ<;KN*-%tEY$}N>)?=Fu%4}}6lh{@ zMov@xLz_WV!|V&xYH`mLD)gZp*fE&c2NwJ%u!vp<%a6oPX{MAB)a@pqrBpu{VgFQ% zqiF~jjb#`f6QO>7ZN$KN;>;np(dNbMlZ*|^@2j#S z^ZKNc*K#v@JbZZ=^&-LE(;6k9>R%fS_UoH!ENB7Ezl>`doPN0I+`=DC&*=i}*wK!g zRG$X4wf0ORU7xGN2J**wc-C}wOnxS4nL&vLbiKj5a1Q5^Np;X~R1C~{6}REX&#y3o zp272I@`v%$1hihjUSb$krpz2u{ z);e19K&*AkAqF2lexvXDz`L?CjKc`GqHWpi;zY^-kurW;g`uJ-)79kZu?pd5yPm&i zwry#7OvdwFm~h;ZNaOM4M0zEV1>aIY&Bvnt&Y$4>A_>(Ze8H)Ychk+w4R2AU?lqp0 zsMShJ3_!^_huVlSr9iQXve&hv12}R`(9P_NU%go4Ahy9`W06rrXjmL<(gvhPq8wE5 z=QC)sO5i&*W`3ydNKL|S4Q~UHdWGDDsRzAH$VvTcyp1cAPFwu~Jw6)3D_vTrN2H=7 ziH8rSWK3iKgc@&@e?M0jk`mX6<~;bau^%5C#6$j>L=&nS&ygw(TqEo0=77^WA{hS#KM^A4j#p+bg9<(7dHLd}6pgXyXR+$*yj7(x9 zZn9$~&h$W?sl7_W=$ofHe_3G;ZBN{gRz+gg%sVbu-?QEzsWaURv}`gpX;90RaFg}H zA#Ph0!QLM`OiIvj>u7-8bfz}m^b)Qu2QT)g{|J=~rd=%7~KkX+A8(00?g<(`T;(prq zVca($*e_ZNsZS*T*J8s&oslW?jNj8gPgIZ740WcEvzd~K4!9_;3{#xM88aQ08|utn zL&r!JLAqRPAwdOc?D$4SDX?s~;)?<=eEPYW1;3Q1Fa`Tte0HleL^zefG_o-Ipo&Ca zvE@nBLV(MDJ%~S^z5fC8s%5aI&b5CtQIm>YPxfDWhgJuGn>>KYc$>)lxb^4vATk?b zeMeRvjg`~8py%A*ne~EE(M3~iY`?R2ZHU^4ifyQzY9^dv6}RArN;?Qy2L@I6?ygvB zYkDS)mwv{j{wAK~RnV9mF$rLF%e~YzppQMu)R7m_vLOO_Fbz^7IpY!F_pZ<42=T{? z)T2?*eg80aIQgrzloc7J$hpb6`;H=Y^wqB}fqc8x1eH21Gvd{zSRV5kHft5J4}r-_ zvk5;Ja=_& za%bc*3w9}WNKJ0bhd1e!WpIHPZsv&ak!fsoQS>MMPE&v|q zP9IaB^(~uLM$Aw@9ea&aBiy6`Gs<=>Qb&}8GZDZ_oDLva)8N((>N}LXpvZ=PtdEDU z|90EpMpmvjC0dS`>t~3$QVv*HI09ONsc>5DB}I#6{~Q8J?w!EG-Y~eYZbqwEBZ_zg z5$5CV$%J1G>vH8?rePVf&fuNdF-JD>fdP#jChknBwG&Votw9yZ#+O(|ICQ^VqJx=w z!>dZLIYW2j7yh+js^m?YTfwB#i{_| zduwTOohTi-(Sb;lU(|$L!);bJ6|6_s>Y`K#u}Qs1$_0JViSYxHZ_H(KO#`kH47q$K zoy3dw$f=J64&)9^%z%^~p?Q0ARSApbpKBQ)xORLYe+f%i)@BT@y-l{300-oHNRgU% zsPNxhS2l1mdEwP=_y(Gfb8^-dYU$pda_@foa?_j^q|GKxg}iH4DO}p)M09+#SFxx zq_Ih{aj}Na`n5>6xMI@lfsF*6IqQ!9P|-LR7on%v)046h_}+d%RI% zd^E4x5|8o7b*RYU$EN10l5X$I|HM1eQ^Q7tr0XqJ0a*#w9XZbbyDibhh{x{p8AQ~U z2c`>uDGvp3D5BAGpNWZDQ9ISCI-U=6rJbDiINxiVGXGnzTA1eM2rjrGA`agc6jp>XwJ;& z>g4vnhIf0GeAen&bnR}`_Q62}_9j1iZVH{`gyMyMS4t{c#3Zf{dUq*lOh^sY^e4( z0UaA^G8SpQx}L*WX8qlM0jx1!hRHYW`$Ufjm44lfxA3^p;jPTxDul;&Uu$dN+4CZW zkHVB_c?~mUnT3x`%6-Mcj%{Bj8WG`vSxQ$-1Dc}p<4#f-P$0^qqdP=P7v3$b=Wg{G~6*S0)T)Amec^cmhH73knnI zWk}9%1fp!cnu~dL6B$2guTjPd!<{hYFCh5T&ktuKp%>TqEH!B%&fLA&x^6JeR9k(L z*S74*2Ge0<&j2%U= zwm*Gp$SX3<(K0RBS#3vM!2F)D$J)HZ(vd^wjgfgJ!sxht`yT~%=3~_H%I$x{l)T2v zyp^~_3*hqgSj-}emn2L2e92m;el((9b~|c!5r8JIWRFI=f>fmAA2F#&<<)5rU@f(T z`o3t);gGYi8C;(_p*zo!Mw%%G*6*s}rSYZg9T~3H6Qr*g?Pr`%DB`ISOvEPRRDbsk07 z4$h7840q$j?z2;DHB-h^l5$ZmTHIJ_b;FPcF2P!Ys%&N{T8)D+*6SPFRaWJifQi;K zX_e~rl4PKb0UOFGP{LU!*v1?S6_ip6~KyaosL*txNEHV zD)P&I9K3WT37eGm-?N;^l?UI{cOTQT@jhiol--(H{fehe@p+0Z>m5X?u4r|V* zTY{e|@Wr(3I5Aw94&*-q3mR)QWxyIRlO!zBrjEC9c!ug-g0QJQvE3% z@Y2z($OU=ty>92LPKOgLQ7O}}J2nj2LLZr+r_PmfNKY7gud3v6PqdNn!p}3*iwi#^ zxnTiZ%YPP6<;$O0LIt$InnSa7_2DMtxI;4Ld2*es!XjhUX$>$XA7Kp4@3Kch8Nk}b7 zd?m?-8b^uCbn{5Rz`>n|TOQ9)jO$v!jN0!%bu zUNUB=Y7nY_TrBHvCzx0MA(S>n_NY}&1PRG0g#pfvdC9}#vpVGPp!PwE$rTa&JL?GD<(BEhrH=?CuJu*;zR1x^FGC zDGy85gQCOfV*cXS14q*2vrWQPdnh4@HLIQ&6=qeEP8uuc!R_sOd$=jsu1yp)+8A5A zw~j|n+RBQ0>m3AUb17lFO2~ZQqs}MP7>&CVKDst?_Kn&e_V0KeyHdK1p{(P|imq)U zZ^Dfdcq!BYnR@z{2}ES}Rq+rpV2mv70txetDdCX7QLOfrX1`NDHOnn+E}YiXRpjHH z9<0iQyftV=APGdo2TG;1EEeqnS0Le9t+x!mv|xsD$Gwmg#iSQ;o;9tVU(J5n4z;WU zW%`Hvwgg~(c0w<+`l61)(9PtI*}&y5+m`$O#-63ztN;kSN2=NeGd<*$i{aPGJk>r4 z(Xc{DJtZ3mYNO4FWi%j9$Z#u2XHPHoQQ%OkXH_B`nz>3w3smH=!7L)vkX`-&+E#C+dh8CO0WlRyONc$->ew9dFtRP{8zF;&`3Z86=iIOtS$s6PF5 z@mDg0<}c_ed}+p;d{jC=qLXTzS=buIrQ3;C3Ecc>B%?WcM+nO-Xzo z5cqw>q>&Xe$MB>Z&^on3*`<^Fxk>-oPUfw&HXb{l?H8fcVsF+9g=r>VUI|xUuU-7t zCFN9iv8m}V=*RoM2kY}JT2TbQ8{tMY88Nac>^9-{qY1KBLC6JzD`bi-A*MloZph%M z+uB?4N7Te_<4hwCPt?C?+`M~1g)r0<@};E3ie)CJw~{fvdDg2hY6nnCZ`U9NlCQ6` zcST~5ceq6=lumZ)c)845Xp-6yPAj!FY1`3E=9(C~+s19YU#6bFYjj(dtkre)E6t6W z{)9YAo|^1u{gT@D<7RMHe5s}^| zc}n))XV4qb8E%SyH!WD_?9(AtorrLkb+XH*Tzq*;b;xuqQp;zwW3MV4Vz%wle2zrP;BRVU1h4Fu4cT`rfbQ0N9VCj2pMQLRtY zVC77D z)pNOhxlB(irpQn+G2sAiy~|lTorSSz4`zTc&o#PgRSCLuej{c|oY*c^^I@cc_~IDV zm1{n3RZ=npO$#$3yzaQ{L+4>qZ_`h(HQ5tT|47xCd>aP0EH%la(45sz6y@5})2CDq zwtC3kxEZpXNL7gwHCDTd3{Iedf4>83eIgHN6+X7F+Zb8xepXVtRkU>XkRD^VnE>NTha~=N8^rb(LCX^pAd0ShRdS==46@CvQ8nZrd@;+<}Y7%I=u zOIe?lz4oZ&>*@w3Omqy9FL=2e3>q~z|BCpj{Dp56_GibY;b!jE!nF2C!oPGLB0y7O zaH$;{c3f=wT-H6XyQI$wc6X; z6>@@wE|(=`j@rUvB7cqyao43syD(N+0Nk)_q8qy4oc#gNM5>g>#gH(WelRtb9nQVi z`%I^0&Zv?b1ipfC-7E^t4X$lLv9k)=E&szp4~0d%J>olR9iRMCrWopVJK-4DZ$=i% z1a0?k3)ZBW=(FW3Q3hrGO6T2|nxRMfC$xro-ZmDo*be5<6Y#{d)?nyS)0K-N;3Qxg zQ?&?lR`3p46itF2#vB=1bPt=+ey-wyvWCg~R!KqE_bqhc|A>lQ`PCT>DT@ z=bWefUKOOerqOFx%zqtG$1*keK}?bmF!$RTzQg6~t|Vdil%a2-AnAEMHDH#58)}T4 zmnC?$QiBc`V;FaSA;6F+DQY+uWBH%04tl=WORep$PV=No(98oXOBM_EVIa?5BH-kc z=uw+TlK)*flCMnK?AF={+O(*M1|>L3-jP}Wg0qBSU(%`UP#b9o`K*aI`FyJ$oX2WZ{fMHcN8@ zH2JQ$ptmX!tu=orh(2=^Bv+QV%qM3i`<#-J_^+B8QYz(^*GE|)=t3M^3aF_EN&REw zH3~2OF*3@)uYjoVh=WauKXzqIb=c5AYhVgh&e7bs-as_)J-kEW+Oy1FCuaDU`PDS z?fjI#aQQVqcyqEmmX+d2YRnt%UvgGjv7QdD!~CXMN%7>fP%2d)%xSuN)Vd0HPKl9T zRVmMrYUVvLaO@LqW=2(han{rHtPaRMMFP^;EYP27HPMXP=@xdT-G;D?Pn~FZ7>j8u zY?`(3OINmmM4L5VjiXD*^ZPw;ddDVkZaNyL5EN8Tu(9H<#s_9CttLSP8KS~8^q4J= zCs0NiwnESpPzp~mBie*G9v<3tB7v3}O<9y%6ljsfy1)g+GrFK-^h2XU;g2G8o|nSy zASWmpx~`hDNiU-?s>XEGWiI}n0aD4jG9k73_v1n<8#`NHv@S1Cqvyv8Gf#TmQPIw( z6A!zG%Fy3MaxJaY;*J^u#TiF*omO0WXAG3W^*(}BJA?7Jc=w|)WSAu{j&R0c*>K^Y=JgIYbw3PGJ9O52 zsZ@k4_~LOc7Rnz9Z8PJ?j!e=!_{8ECPw5&uDQY*_b$bO+F@>h$mOL*tcUw*Bc#5!D ztR&oMnkjlRLT?(0+T#j{z2Y!pM2oaFQhSuLYlEkNd)}5mV;zXEz2nmITwDp8)h`wt z`0wGEw&=9i?dP`Q?;6vNQ4b};i(ooZ0^KBR3d9i`eubg9$VXL%1}xe5yRonyj#C+` zV7SsR-SZNiI~R3(Mk_APQJl^n2gpqZ^?qORz+@pZ>Hdsu#&!|qtNN|w{zIsmALnTY zfR?q|N%VxH7&gVwX0FojiHE!m^g%92hE_x5d0idz%1pyyL6zZTZdpuQ;hJeARM#xr zPp~p__~cUlH;#=->FsK)yAhR{?lSt(&d??m-wFyrWA^7ZK|NOMeWv4TQJJGHJ8?se z%S;@FK0QX>LQC{ifQ3=O>CEcf7?c+wxUO+TU+SgS;1SS=K}OEW3$kR5vN@0?L2$VJ zb1#Sp6(!|xvNSp_SdHCzv;vA%A}LQug5SKj36W>&MH*6CB~#HndMtJrl#h7sRzKq5 z=}-t1wO@Ge|I^XpWx_4}j6aiVMEPh`A~uRPKLAV8clicWYd|;HwC0P$U|z9^@lhK- z7gm+Urh*dEh_zcDHI2I@bdN(c4yko&JBZQ^UZ#7z)r+e3`mdm;!&1K`KL68uxo8?w zljCb&kg@c_Wi-x%q=&3t9-xyJoo#@`;jzy3ZS*mr@)3$ z%mbZCbbMn(oTJ7ELhvZSZGSYt)hxu$XYwq*;r94s*}2$oxPTjW@PD5l*UF#@3-AZb%Bd1G`90P8{vHLQe*7B7jU#O^s7w z@?0w%VJ;X3Q-*BTb4$60Fmh-QC}tFuJ%T+PF*G<5&Y<+$DkaN4TD+N_;sZ(hbJ zwKZnTrjc?w!xS>*!_S}`aW(LTC~)QoFbf0BfV`GY^yy@U5}(&6A6xAucKNx$NJeuF ztgr!UV3(WRNRTV_;tHeE2K0oZ919?h{{p;dB*+xculpZrDF8OdUs@um4GS}-U1in0 zN6l!Yj5-UjLteNMPw(mHPGv;>Ta$TX=tbX$pFC?EXA(+i5PWIqDQwig@8g*D13{Je z)V$-N{#wXv?Cf71w&Ami*Njh0BAV3hzDDJg6OB=ftHAX@%fQG|h<{QL25?R!79@T1V@u1kge zfp$cp+gdP0=mp%5T5Jn?a+T=XM}!1QwD>b!Uc-bGw+vJw)tevnf>il6`}&=z6TY8c zKv&stegu0|nvcQejD17!H9uatC?CIzPsI8}XzW|D?9?Ft@ksi;fQ5h^M>l5d@fq5O z=l8&Qd+; zrS%eLTdr<`%%BrX%XEBDU1Be*{S)PO;3H9z?^sgU&8#aSoT2coNEuM??k0HTVFu5f zr^7`2{aOfZdE9^z?3IVd!q)#0`dML+E6r1bP7Q@SVT~o_hMI2SG3MCfGQI}dZps;5 za1b9nN%D+SJgKApPvOqRoco)%++AJg=;F>La4uhAKl&j1XI7-|X-fC>G%`|Zbx3L| zoL&}qTfiRW&i1x@9>4CJTxwK&WvLMLDjy)a1XqJe|AX$@F^}U2X}LaipTjKjh@~pp z7TjI)R6qQ#rSvp>TNIig8vTgJ@-n}1V<^pHbM>z?aaEb-77Fd0U)u1m+KW#$F3OS* zu5lFj$CitW&EJYcz9ZfGDY|LvC!`uOgg?1PnoAF^Vnn94_01Vmff**K$arzL5sHTV z9J{!STU)J(XAHTl*|;8)Cnu^Sd%=_y8SL00)G1nZ2@#u>y&P*YX%>e8nc-#a^EMS} zmXRn@RrF)uBWf&5+1s)6^ki8@>a->os&)a(E;<&j`s6#+i$c2Im5x&*8zbZ;iLvedv+zeI z#k8H~A5U3~HFY_WYWrRXMG}Dc0u+XdVqh{v(5piGH{GoJc8e*qhIF-d`S-z8SEQ<0 z^K1keoH0tCES&O#RCO`0<*97D=nZo@=bO5@of^M|Px2hxzklbUSJ-4bzF5J>*cpV- zWfe@Z=(Knqvs6q(vuI4=>Aqk+tHk}TEZ?BaH2DoHrzyi_giT~a-3x>^#s+D3x>!(< zO0m56{oU}^pzOE`_;@6DTz*D3 zWa;Q;`0^1|xzoNqbE7k$c@%d;cdu!J@AY;6y7rurja9JmQo~^iw4dF^(_%|Sg!^dW zc~@eTQ^4s)r@Df!ycBC(XVPbreb7BOhZ%X;)@{*_F3?1(2}V*FYe9FbPZ8VaO8NHK zSgt>EXMvh;K?9y81B`*Q)j>8@He;X^KaGZWbSDW$0 zI5-;d2px3^EpB%Ljpy^g!(RNFjGfQ!Xv`G`EqItwD6Hm1%=0uOvJuUQ-blvsb*fNb zE<8+yC|2_p`*i0Ch3H<<>z<0SCnE#hd%&ZORD~+i(!Fpeu~|q+8n1or-jXL^rd>PD zMwrmguZ`;6QBo$R^F1CVp-PLUv!EBuGhWx&tIP z#wh!4y#yRZnI+Hww>j%wxmJUJNOhH>skQiPaGfXOk=`zG^c%6aq8ln=V-lDdI;>Y^ zPtN=Mfg6SiIM>=Z+fp?&!@d-!PtC11bozPVzfHQ%Vc`xzC;|n(#S4dUQ7(7C12t4h zU(_t-i8Eo7Idroce259zIM^`rwKGEu5bKp-b{VbAT(0-6hx~x_Rt3_;C*;kvk^cc| zF4ef!qX~4h*FI0nP!^3eVsgktq4rc&5*EEK_9*e4N41mhnNm^k@Z^uf8z>t<(Fpt@ zuCsxC<~OLHC5O?1mcv;3spxKIWw(^8Tyn}T{(duT?bPu<8~}E`h(10Y?3<*5Q|Vu` zR8K{7Oj-u#-cF-+cQDMiwS zs{4D%=eg-ba9ywb|EPF(2%Es-M$N5av}JU2d7tez=W?a`3%H|v6!6Q=M9)ZN}*djouk=j=~Ff~~{yoxufE zAo*+4(Pg%gCv2Q=G&HGovSU9$QH$2Ce@vHfDVWHyYGu-AMGkx4dl4QgZhLidCz@R- z^4G1y+bcM-_(5I$Qk@>V#Tvp*w9H7FQ^eRp%OMl|=vbu>M%#OV$&%F=rA_2D_9fMO zG+w*iGxgFVOmV0*#3CYqVf9Vj&f-z`!UsFpRUyr5`}$)qX@Ax1%8GIMV}XbwDf!3{}ZF^3$kWe!VC&DJHY0wQIR@m0d9MG&KG0`bI9X%nD^~lJaaZ|$BY;v=Pl^o zyBd;ZlV~kq;`_38Mz86@WSb&A?dZghA3vpK`~+_<i0wG-NhC5d#jM)qoQZgcPL)u>K)k`jdJL-(E>TQd>u<*UObunI3uflDK|OBSV$r3|ER!N z(0=U*WC^yTOsbw40c}20lxnZ>0-v{?_2f!3{>Q-Ztgso6cST!TNl8k>?U9K0e@Vs; zdPG~E@$otGbHm$Z7wduNX#Wqv*#A#zHu>R&)HYx0TOlFb=MT__jD(_im8em`{|ANw Bm6iYi literal 0 HcmV?d00001 diff --git a/open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg b/open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg new file mode 100644 index 000000000..cc73d30b3 --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg @@ -0,0 +1,118 @@ + + + + + + + image/svg+xml + + + + + + + + + + + + + + + + + Encoder + Decoder + + diff --git a/open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg b/open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg new file mode 100644 index 000000000..d8d59e5df --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg @@ -0,0 +1,1579 @@ + + + + + + image/svg+xml + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/open-machine-learning-jupyter-book/llm/introduction.ipynb b/open-machine-learning-jupyter-book/llm/introduction.ipynb new file mode 100644 index 000000000..a53688e89 --- /dev/null +++ b/open-machine-learning-jupyter-book/llm/introduction.ipynb @@ -0,0 +1,155 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": { + "tags": [ + "remove-cell" + ] + }, + "source": [ + "---\n", + "license:\n", + " code: MIT\n", + " content: CC-BY-4.0\n", + "github: https://github.com/ocademy-ai/machine-learning\n", + "venue: By Ocademy\n", + "open_access: true\n", + "bibliography:\n", + " - https://raw.githubusercontent.com/ocademy-ai/machine-learning/main/open-machine-learning-jupyter-book/references.bib\n", + "---" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Introduction\n", + "## What are large language models (LLMs)?\n", + "\n", + "Large Language Models (LLMs) are advanced artificial intelligence systems that have been trained on vast amounts of text data to understand and generate human-like language. These models are typically based on deep learning architectures, such as transformer neural networks, and are capable of performing a wide range of natural language processing tasks.\n", + "\n", + "Key characteristics of large language models include:\n", + "\n", + "1. **Scale**: LLMs are trained on massive datasets containing billions or even trillions of words. This extensive training corpus allows them to capture a broad understanding of language patterns and nuances.\n", + "\n", + "2. **Complexity**: These models are often deep neural networks with numerous layers and parameters, allowing them to learn intricate relationships within language data.\n", + "\n", + "3. **Versatility**: LLMs can be fine-tuned for various natural language processing tasks, including text generation, translation, summarization, sentiment analysis, question answering, and more.\n", + "\n", + "4. **Generative Capabilities**: One of the notable features of LLMs is their ability to generate coherent and contextually relevant text. Given a prompt or context, they can produce human-like responses or complete passages of text.\n", + "\n", + "5. **Adaptability**: LLMs can adapt to different domains or styles of language through fine-tuning or conditioning on specific data.\n", + "\n", + "6. **Resource Intensiveness**: Training and using large language models require significant computational resources, including powerful hardware and substantial amounts of data.\n", + "\n", + "7. **Ethical and Societal Considerations**: The development and deployment of LLMs raise ethical concerns related to biases in the training data, potential misuse for spreading misinformation, and the societal impacts of automated content generation." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Narrow sense\n", + "\n", + "In the narrow sense, a large language model is described as a probabilistic model that assigns a probability to every finite sequence, whether it's grammatical or not. This perspective emphasizes the probabilistic nature of language models, indicating that they can assign a likelihood to any sequence of tokens, regardless of whether it conforms to grammatical rules or not. This perspective highlights the fundamental nature of language models as probabilistic models that capture the statistical regularities of natural language.\n", + "\n", + ":::{figure} image/implicit-order.png\n", + "Implicit Order\n", + ":::\n", + "\n", + "### Broad Sense\n", + "\n", + "In the broad sense, large language models are categorized into different architectural types based on their structure and components:\n", + "\n", + "- **Decoder-only models:** These models, such as GPT (Generative Pre-trained Transformer), OPT (OpenAI's Pre-trained Transformer), LLaMA, and PaLM, primarily consist of decoder layers. Decoder-only models are designed for tasks like text generation, where the model generates output tokens autoregressively based on preceding tokens. GPT-X is a notable example of a decoder-only model.\n", + "\n", + "- **Encoder-only models:** Models like BERT (Bidirectional Encoder Representations from Transformers), RoBERTa, and ELECTRA are categorized as encoder-only models. These models focus on capturing contextual representations of input tokens without autoregressive generation. They are often used for tasks like text classification, where bidirectional context is essential.\n", + "\n", + "- **Encoder-decoder models:** Architectures like T5 (Text-To-Text Transfer Transformer) and BART (Bidirectional and Auto-Regressive Transformers) include both encoder and decoder components. These models are versatile and can handle various tasks, including text generation, text summarization, translation, and more. They combine the strengths of both encoder and decoder architectures, enabling them to perform both generation and comprehension tasks.\n", + "\n", + "These models have demonstrated impressive capabilities in understanding and generating natural language, leading to their widespread adoption across various industries and applications.\n", + "\n", + ":::{figure} https://raw.githubusercontent.com/Mooler0410/LLMsPracticalGuide/main/imgs/tree.jpg\n", + "An Evolutionary Tree of Modern LLMs\n", + ":::" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## How do LLMs Work?\n", + "\n", + "Large Language Models (LLMs) function on the foundational principles of deep learning, harnessing neural network architectures to analyze and comprehend human languages.\n", + "\n", + "Trained on extensive datasets using self-supervised learning techniques, LLMs excel at recognizing intricate patterns and relationships within diverse language data. These models are structured with multiple layers, incorporating feedforward layers, embedding layers, and attention layers. Utilizing attention mechanisms such as self-attention, LLMs assess the significance of individual tokens within a sequence. This process enables the model to grasp intricate dependencies and relationships among words, phrases, and sentences, thus facilitating its ability to process and understand natural language effectively." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Architecture of LLM\n", + "\n", + "A Large Language Model’s (LLM) architecture is a pivotal element shaped by various considerations, including the model's intended objectives, available computational resources, and the nature of language processing tasks it is designed to tackle. The overall architecture of an LLM typically comprises multiple layers, encompassing feedforward layers, embedding layers, and attention layers. These layers work in tandem to process input text and generate predictions.\n", + "\n", + "Several key components significantly influence the architecture of Large Language Models:\n", + "\n", + "1. **Model Size and Parameter Count**: The size of the model and the number of parameters it encompasses play a crucial role in determining its architectural design. Larger models with more parameters often have enhanced capacity to capture intricate language patterns and nuances.\n", + "\n", + "2. **Input Representations**: The representation of input text, such as tokenization and embedding methods, directly impacts the architecture of the LLM. Effective input representations facilitate the model's ability to understand and process textual data accurately.\n", + "\n", + "3. **Self-Attention Mechanisms**: Many LLM architectures leverage self-attention mechanisms, such as the transformer architecture, to capture long-range dependencies within input sequences. Self-attention enables the model to weigh the importance of different tokens in a sequence, facilitating robust language understanding.\n", + "\n", + "4. **Training Objectives**: The specific objectives of LLM training, including pre-training and fine-tuning tasks, influence architectural choices. Different training objectives may require adjustments to the model's architecture to optimize performance on targeted tasks.\n", + "\n", + "5. **Computational Efficiency**: Efficiency considerations, such as computational resources and inference speed, impact architectural decisions. Architectures that balance model complexity with computational efficiency are preferred, especially in practical applications where real-time processing is essential.\n", + "\n", + "6. **Decoding and Output Generation**: Architectural design also encompasses decoding mechanisms for generating output text. Techniques for output generation, such as beam search or nucleus sampling, influence the overall architecture and performance of the LLM.\n", + "\n", + ":::{figure} image/scaling-laws.png\n", + "Scaling Laws of LLMs\n", + ":::" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Constructing LLMs: A Process Overview\n", + "\n", + "Building large-scale language models involves a multi-stage process, consisting of pre-training, supervised fine-tuning, reward shaping, and reinforcement learning.\n", + "\n", + "1. **Pre-training**: The initial phase involves training the language model on vast amounts of unlabeled text data using self-supervised learning techniques. During pre-training, the model learns to understand the structure and semantics of language by predicting missing words in sentences, predicting the next word in a sequence, or performing other language modeling tasks. This phase aims to equip the model with a broad understanding of language patterns and nuances.\n", + "\n", + "2. **Supervised Fine-Tuning**: Following pre-training, the model undergoes supervised fine-tuning on specific tasks or domains. Fine-tuning adjusts the model's parameters to better fit the target task using labeled data. This phase involves training the model with annotated examples, allowing it to specialize in tasks such as text classification, sentiment analysis, or question answering. Supervised fine-tuning enhances the model's performance on task-specific objectives and improves its ability to generalize to new data.\n", + "\n", + "3. **Reward Shaping**: In the reward shaping stage, the model is further refined through reinforcement learning techniques. Reward shaping involves defining a reward function that guides the model's behavior towards desired outcomes. By providing feedback in the form of rewards or penalties, the model learns to optimize its actions to maximize cumulative rewards over time. Reward shaping helps improve the model's decision-making capabilities and adaptability to dynamic environments.\n", + "\n", + "4. **Reinforcement Learning**: The final phase of model construction involves reinforcement learning, where the model interacts with its environment and learns through trial and error. Reinforcement learning algorithms enable the model to explore different actions and strategies, gradually improving its performance through experience. By receiving feedback based on the outcomes of its actions, the model iteratively adjusts its behavior to achieve optimal results. Reinforcement learning enhances the model's ability to handle complex tasks and adapt to changing circumstances." + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "open-machine-learning-jupyter-book", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.9.18" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} From 9f47300c515bfd415deeb00d135df94ede770fe5 Mon Sep 17 00:00:00 2001 From: peng_windows <2686728826@qq.com> Date: Tue, 26 Mar 2024 13:07:32 +0800 Subject: [PATCH 2/6] fix --- .../assignments/llm/basic/transformer-architecture.ipynb | 4 ---- 1 file changed, 4 deletions(-) diff --git a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb index 274f60583..4c24fc64d 100644 --- a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb +++ b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb @@ -48,10 +48,6 @@ "Overall, the encoder-decoder architecture based on the Transformer structure allows for effective semantic abstraction by leveraging attention mechanisms, position-wise feedforward layers, residual connections, and layer normalization. This architecture enables the model to capture complex dependencies between words in the input sequence and generate meaningful outputs for various sequence-to-sequence tasks.\n", "\n", ":::{figure} https://media.geeksforgeeks.org/wp-content/uploads/20230531140926/Transformer-python-(1).png\n", - "---\n", - "\n", - "width: 90%\n", - "---\n", "Transformer-based encoder and decoder Architecture\n", ":::\n", "\n", From f44a878cee4f49e8e8069e02eed9f7d95ea36e1b Mon Sep 17 00:00:00 2001 From: peng_windows <2686728826@qq.com> Date: Fri, 29 Mar 2024 06:11:23 +0800 Subject: [PATCH 3/6] update llm --- .../llm/basic/transformer-architecture.ipynb | 294 +- .../llm/basic/attention.ipynb | 1685 ++- .../llm/basic/basic.ipynb | 2 +- .../llm/basic/transformer.ipynb | 10 +- .../llm/image/attention_example.svg | 9628 ----------------- .../llm/image/cifar100_example_anomaly.png | Bin 155392 -> 0 bytes .../llm/image/comparison_conv_rnn.svg | 1809 ---- .../llm/image/implicit-order.png | Bin 34807 -> 0 bytes .../llm/image/llm.png | Bin 166613 -> 0 bytes .../llm/image/multihead_attention.svg | 288 - .../llm/image/scaled_dot_product_attn.svg | 351 - .../llm/image/scaling-laws.png | Bin 86847 -> 0 bytes .../llm/image/transformer_architecture.svg | 118 - .../llm/image/warmup_loss_plot.svg | 1579 --- .../llm/introduction.ipynb | 4 +- 15 files changed, 1663 insertions(+), 14105 deletions(-) delete mode 100644 open-machine-learning-jupyter-book/llm/image/attention_example.svg delete mode 100644 open-machine-learning-jupyter-book/llm/image/cifar100_example_anomaly.png delete mode 100644 open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg delete mode 100644 open-machine-learning-jupyter-book/llm/image/implicit-order.png delete mode 100644 open-machine-learning-jupyter-book/llm/image/llm.png delete mode 100644 open-machine-learning-jupyter-book/llm/image/multihead_attention.svg delete mode 100644 open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg delete mode 100644 open-machine-learning-jupyter-book/llm/image/scaling-laws.png delete mode 100644 open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg delete mode 100644 open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg diff --git a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb index 4c24fc64d..208716078 100644 --- a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb +++ b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb @@ -74,12 +74,18 @@ "import torch\n", "import torch.nn as nn\n", "import math\n", + "import copy\n", + "import time\n", + "import torch.optim as optim\n", + "import torch.nn.functional as F\n", + "from torch.autograd import Variable\n", + "import numpy as np\n", "\n", "class PositionalEncoder(nn.Module):\n", " def __init__(self, d_model, max_seq_len=80):\n", " super().__init__()\n", " self.d_model = d_model\n", - " ## Create a constant PE matrix based on pos and i\n", + " # Creating a constant PE matrix based on pos and i\n", " pe = torch.zeros(max_seq_len, d_model)\n", " for pos in range(max_seq_len):\n", " for i in range(0, d_model, 2):\n", @@ -89,13 +95,11 @@ " self.register_buffer('pe', pe)\n", "\n", " def forward(self, x):\n", - " ## Scale word embedding representations\n", + " # Scaling word embeddings to make them relatively larger\n", " x = x * math.sqrt(self.d_model)\n", - " ## Add positional constants to word embedding representations\n", + " # Adding positional constants to word embedding representations\n", " seq_len = x.size(1)\n", - " pe = torch.autograd.Variable(self.pe[:, :seq_len], requires_grad=False).cuda()\n", - " x = x + pe\n", - " return x" + " x = x + Variable(self.pe[:, :seq_len], requires_grad=False).cuda()\n" ] }, { @@ -173,11 +177,6 @@ "metadata": {}, "outputs": [], "source": [ - "import torch\n", - "import torch.nn as nn\n", - "import torch.nn.functional as F\n", - "import math\n", - "\n", "class MultiHeadAttention(nn.Module):\n", " def __init__(self, heads, d_model, dropout=0.1):\n", " super().__init__()\n", @@ -189,9 +188,10 @@ " self.k_linear = nn.Linear(d_model, d_model)\n", " self.dropout = nn.Dropout(dropout)\n", " self.out = nn.Linear(d_model, d_model)\n", - " \n", + "\n", " def attention(self, q, k, v, d_k, mask=None, dropout=None):\n", " scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(d_k)\n", + " # Masking out those units added for length padding, setting them to zero after softmax computation\n", " if mask is not None:\n", " mask = mask.unsqueeze(1)\n", " scores = scores.masked_fill(mask == 0, -1e9)\n", @@ -200,19 +200,23 @@ " scores = dropout(scores)\n", " output = torch.matmul(scores, v)\n", " return output\n", - " \n", + "\n", " def forward(self, q, k, v, mask=None):\n", " bs = q.size(0)\n", + " # Linear operations to divide into h heads\n", " k = self.k_linear(k).view(bs, -1, self.h, self.d_k)\n", " q = self.q_linear(q).view(bs, -1, self.h, self.d_k)\n", " v = self.v_linear(v).view(bs, -1, self.h, self.d_k)\n", + " # Matrix transposition\n", " k = k.transpose(1, 2)\n", " q = q.transpose(1, 2)\n", " v = v.transpose(1, 2)\n", + " # Computing attention\n", " scores = self.attention(q, k, v, self.d_k, mask, self.dropout)\n", + " # Concatenating multiple heads and feeding into the final linear layer\n", " concat = scores.transpose(1, 2).contiguous().view(bs, -1, self.d_model)\n", " output = self.out(concat)\n", - " return output" + " return output\n" ] }, { @@ -287,14 +291,10 @@ "metadata": {}, "outputs": [], "source": [ - "import torch\n", - "import torch.nn as nn\n", - "import torch.nn.functional as F\n", - "\n", "class FeedForward(nn.Module):\n", " def __init__(self, d_model, d_ff=2048, dropout=0.1):\n", " super().__init__()\n", - " ## Set d_ff default to 2048\n", + " # Setting d_ff default to 2048\n", " self.linear_1 = nn.Linear(d_model, d_ff)\n", " self.dropout = nn.Dropout(dropout)\n", " self.linear_2 = nn.Linear(d_ff, d_model)\n", @@ -302,7 +302,7 @@ " def forward(self, x):\n", " x = self.dropout(F.relu(self.linear_1(x)))\n", " x = self.linear_2(x)\n", - " return x" + " return x\n" ] }, { @@ -378,21 +378,19 @@ "metadata": {}, "outputs": [], "source": [ - "import torch\n", - "import torch.nn as nn\n", - "\n", "class NormLayer(nn.Module):\n", " def __init__(self, d_model, eps=1e-6):\n", " super().__init__()\n", " self.size = d_model\n", - " ## Layer normalization includes two learnable parameters\n", + " # Layer normalization includes two learnable parameters\n", " self.alpha = nn.Parameter(torch.ones(self.size))\n", " self.bias = nn.Parameter(torch.zeros(self.size))\n", " self.eps = eps\n", - " \n", + "\n", " def forward(self, x):\n", - " norm = self.alpha * (x - x.mean(dim=-1, keepdim=True)) / (x.std(dim=-1, keepdim=True) + self.eps) + self.bias\n", - " return norm" + " norm = self.alpha * (x - x.mean(dim=-1, keepdim=True)) \\\n", + " / (x.std(dim=-1, keepdim=True) + self.eps) + self.bias\n", + " return norm\n" ] }, { @@ -470,11 +468,40 @@ "metadata": {}, "outputs": [], "source": [ + "class Embedder(nn.Module):\n", + " def __init__(self, vocab_size, d_model):\n", + " super(Embedder, self).__init__()\n", + " self.embed = nn.Embedding(vocab_size, d_model)\n", + " self.d_model = d_model\n", + "\n", + " def forward(self, x):\n", + " return self.embed(x) * np.sqrt(self.d_model)\n", + "\n", + "def get_clones(module, N):\n", + " return nn.ModuleList([copy.deepcopy(module) for i in range(N)])\n", + "\n", + "class PositionalEncoder(nn.Module):\n", + " def __init__(self, d_model, dropout, max_len=5000):\n", + " super(PositionalEncoder, self).__init__()\n", + " self.dropout = nn.Dropout(p=dropout)\n", + "\n", + " pe = torch.zeros(max_len, d_model)\n", + " position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)\n", + " div_term = torch.exp(torch.arange(0, d_model, 2).float() * (-math.log(10000.0) / d_model))\n", + " pe[:, 0::2] = torch.sin(position * div_term)\n", + " pe[:, 1::2] = torch.cos(position * div_term)\n", + " pe = pe.unsqueeze(0).transpose(0, 1)\n", + " self.register_buffer('pe', pe)\n", + "\n", + " def forward(self, x):\n", + " x = x + self.pe[:x.size(0), :]\n", + " return self.dropout(x)\n", + "\n", "class EncoderLayer(nn.Module):\n", " def __init__(self, d_model, heads, dropout=0.1):\n", " super().__init__()\n", - " self.norm_1 = Norm(d_model)\n", - " self.norm_2 = Norm(d_model)\n", + " self.norm_1 = NormLayer(d_model)\n", + " self.norm_2 = NormLayer(d_model)\n", " self.attn = MultiHeadAttention(heads, d_model, dropout=dropout)\n", " self.ff = FeedForward(d_model, dropout=dropout)\n", " self.dropout_1 = nn.Dropout(dropout)\n", @@ -487,21 +514,22 @@ " x = x + self.dropout_2(self.ff(x2))\n", " return x\n", "\n", + "\n", "class Encoder(nn.Module):\n", - " def __init__(self, vocab_size, d_model, N, heads):\n", + " def __init__(self, vocab_size, d_model, N, heads, dropout):\n", " super().__init__()\n", " self.N = N\n", " self.embed = Embedder(vocab_size, d_model)\n", - " self.pe = PositionalEncoder(d_model)\n", - " self.layers = get_clones(EncoderLayer(d_model, heads), N)\n", - " self.norm = Norm(d_model)\n", + " self.pe = PositionalEncoder(d_model, dropout=dropout)\n", + " self.layers = get_clones(EncoderLayer(d_model, heads, dropout), N)\n", + " self.norm = NormLayer(d_model)\n", "\n", " def forward(self, src, mask):\n", " x = self.embed(src)\n", " x = self.pe(x)\n", " for i in range(self.N):\n", " x = self.layers[i](x, mask)\n", - " return self.norm(x)" + " return self.norm(x)\n" ] }, { @@ -510,12 +538,13 @@ "metadata": {}, "outputs": [], "source": [ + "\n", "class DecoderLayer(nn.Module):\n", " def __init__(self, d_model, heads, dropout=0.1):\n", " super().__init__()\n", - " self.norm_1 = Norm(d_model)\n", - " self.norm_2 = Norm(d_model)\n", - " self.norm_3 = Norm(d_model)\n", + " self.norm_1 = NormLayer(d_model)\n", + " self.norm_2 = NormLayer(d_model)\n", + " self.norm_3 = NormLayer(d_model)\n", " self.dropout_1 = nn.Dropout(dropout)\n", " self.dropout_2 = nn.Dropout(dropout)\n", " self.dropout_3 = nn.Dropout(dropout)\n", @@ -532,6 +561,7 @@ " x = x + self.dropout_3(self.ff(x2))\n", " return x\n", "\n", + "\n", "class Decoder(nn.Module):\n", " def __init__(self, vocab_size, d_model, N, heads, dropout):\n", " super().__init__()\n", @@ -539,14 +569,14 @@ " self.embed = Embedder(vocab_size, d_model)\n", " self.pe = PositionalEncoder(d_model, dropout=dropout)\n", " self.layers = get_clones(DecoderLayer(d_model, heads, dropout), N)\n", - " self.norm = Norm(d_model)\n", + " self.norm = NormLayer(d_model)\n", "\n", " def forward(self, trg, e_outputs, src_mask, trg_mask):\n", " x = self.embed(trg)\n", " x = self.pe(x)\n", " for i in range(self.N):\n", " x = self.layers[i](x, e_outputs, src_mask, trg_mask)\n", - " return self.norm(x)" + " return self.norm(x)\n" ] }, { @@ -573,12 +603,6 @@ "metadata": {}, "outputs": [], "source": [ - "import torch\n", - "import torch.nn as nn\n", - "import torch.nn.functional as F\n", - "import time\n", - "import numpy as np\n", - "\n", "class Transformer(nn.Module):\n", " def __init__(self, src_vocab, trg_vocab, d_model, N, heads, dropout):\n", " super().__init__()\n", @@ -606,50 +630,107 @@ "metadata": {}, "outputs": [], "source": [ - "## Model parameters\n", + "# Sample English and French text data\n", + "en_data = [\n", + " \"I love coding.\",\n", + " \"Machine learning is fascinating.\",\n", + " \"Natural language processing is fun.\"\n", + "]\n", + "\n", + "fr_data = [\n", + " \"J'adore coder.\",\n", + " \"L'apprentissage automatique est fascinant.\",\n", + " \"Le traitement du langage naturel est amusant.\"\n", + "]\n", + "\n", + "def tokenize_en(sentence):\n", + " # You can implement a more sophisticated tokenizer here if needed\n", + " return sentence.lower().split() # Simple tokenizer, converts to lowercase and splits by space\n", + "\n", + "def tokenize_fr(sentence):\n", + " # You can implement a more sophisticated tokenizer here if needed\n", + " return sentence.lower().split() # Simple tokenizer, converts to lowercase and splits by space\n", + "# Tokenize English and French text\n", + "en_sentences = [tokenize_en(sentence) for sentence in en_data]\n", + "fr_sentences = [tokenize_fr(sentence) for sentence in fr_data]\n", + "\n", + "# Create English and French vocabularies\n", + "en_vocab = {'': 0, '': 1, '': 2, '': 3} # Initialize with special tokens\n", + "fr_vocab = {'': 0, '': 1, '': 2, '': 3} # Initialize with special tokens\n", + "\n", + "# Build English vocabulary\n", + "for sentence in en_sentences:\n", + " for word in sentence:\n", + " if word not in en_vocab:\n", + " en_vocab[word] = len(en_vocab)\n", + "\n", + "# Build French vocabulary\n", + "for sentence in fr_sentences:\n", + " for word in sentence:\n", + " if word not in fr_vocab:\n", + " fr_vocab[word] = len(fr_vocab)\n", + "\n", + "# Reverse vocabularies to get index-to-token mappings\n", + "en_index_to_word = {index: word for word, index in en_vocab.items()}\n", + "fr_index_to_word = {index: word for word, index in fr_vocab.items()}\n", + "\n", + "# Model parameters\n", "d_model = 512\n", "heads = 8\n", "N = 6\n", - "src_vocab = len(EN_TEXT.vocab)\n", - "trg_vocab = len(FR_TEXT.vocab)\n", + "src_vocab = len(en_vocab)\n", + "trg_vocab = len(fr_vocab)\n", + "dropout = 0.1 \n", "\n", - "## Initialize the model\n", - "model = Transformer(src_vocab, trg_vocab, d_model, N, heads)\n", + "# Initialize the model\n", + "model = Transformer(src_vocab, trg_vocab, d_model, N, heads, dropout)\n", + "for p in model.parameters():\n", + " if p.dim() > 1:\n", + " nn.init.xavier_uniform_(p)\n", "\n", - "## Initialize optimizer\n", + "# Optimizer\n", "optim = torch.optim.Adam(model.parameters(), lr=0.0001, betas=(0.9, 0.98), eps=1e-9)\n", "\n", - "## Training function\n", - "def train_model(epochs, print_every=100):\n", + "\n", + "# Training the model\n", + "def train_model(epochs, en_sentences, fr_sentences, print_every=100):\n", " model.train()\n", " start = time.time()\n", " temp = start\n", " total_loss = 0\n", - "\n", + " \n", " for epoch in range(epochs):\n", - " for i, batch in enumerate(train_iter):\n", - " src = batch.English.transpose(0, 1)\n", - " trg = batch.French.transpose(0, 1)\n", + " for i in range(len(en_sentences)):\n", + " src_sentence = en_sentences[i]\n", + " trg_sentence = fr_sentences[i]\n", + " \n", + " src_tensor = torch.LongTensor([en_vocab[word] for word in src_sentence])\n", + " trg_tensor = torch.LongTensor([fr_vocab[word] for word in trg_sentence])\n", + " \n", + " src = src_tensor.unsqueeze(0) # Add batch dimension\n", + " trg = trg_tensor.unsqueeze(0) # Add batch dimension\n", + " \n", " trg_input = trg[:, :-1]\n", " targets = trg[:, 1:].contiguous().view(-1)\n", + " \n", " src_mask, trg_mask = create_masks(src, trg_input)\n", - "\n", + " \n", " preds = model(src, trg_input, src_mask, trg_mask)\n", + " \n", " optim.zero_grad()\n", - " loss = F.cross_entropy(preds.view(-1, preds.size(-1)), targets, ignore_index=target_pad)\n", + " loss = F.cross_entropy(preds.view(-1, preds.size(-1)), targets, ignore_index=fr_vocab[''])\n", " loss.backward()\n", " optim.step()\n", - " total_loss += loss.data[0]\n", - "\n", + " \n", + " total_loss += loss.item()\n", + " \n", " if (i + 1) % print_every == 0:\n", " loss_avg = total_loss / print_every\n", - " print(\"time = %dm, epoch %d, iter = %d, loss = %.3f, %ds per %d iters\" % (\n", - " (time.time() - start) // 60, epoch + 1, i + 1, loss_avg, time.time() - temp, print_every))\n", + " print(\"time = %dm, epoch %d, iter = %d, loss = %.3f, %ds per %d iters\" % ((time.time() - start) // 60, epoch + 1, i + 1, loss_avg, time.time() - temp, print_every))\n", " total_loss = 0\n", " temp = time.time()\n", - "\n", - "## Train the model\n", - "train_model(epochs=10)" + " \n", + "train_model(1000, en_sentences, fr_sentences, 100)" ] }, { @@ -665,40 +746,59 @@ "metadata": {}, "outputs": [], "source": [ - "def translate(model, src, max_len=80, custom_string=False):\n", + "# Test the model\n", + "def translate(model, src_sentence, en_vocab, fr_vocab, max_len=80):\n", " model.eval()\n", - " if custom_string == True:\n", - " src = tokenize_en(src)\n", - " sentence = Variable(torch.LongTensor([[EN_TEXT.vocab.stoi[tok] for tok in sentence]])).cuda()\n", - " src_mask = (src != input_pad).unsqueeze(-2)\n", - " e_outputs = model.encoder(src, src_mask)\n", - " outputs = torch.zeros(max_len).type_as(src.data)\n", - " outputs[0] = torch.LongTensor([FR_TEXT.vocab.stoi['']])\n", - "\n", - " for i in range(1, max_len):\n", - " trg_mask = np.triu(np.ones((1, i, i), k=1).astype('uint8'))\n", - " trg_mask = Variable(torch.from_numpy(trg_mask) == 0).cuda()\n", - " out = model.out(model.decoder(outputs[:i].unsqueeze(0), e_outputs, src_mask, trg_mask))\n", "\n", - " out = F.softmax(out, dim=-1)\n", - " val, ix = out[:, -1].data.topk(1)\n", - " outputs[i] = ix[0][0]\n", - "\n", - " if ix[0][0] == FR_TEXT.vocab.stoi['']:\n", - " break\n", - "\n", - " return ' '.join([FR_TEXT.vocab.itos[ix] for ix in outputs[:i]])\n" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "## Acknowledgments\n", - "\n", - "Thanks to the awesome open source project for Transformer learning, which inspire this chapter.\n", + " # Tokenize the source sentence\n", + " src_tokens = tokenize_en(src_sentence)\n", + " \n", + " # Convert tokens to indices using the English vocabulary\n", + " src_indices = [en_vocab.get(token, en_vocab['']) for token in src_tokens]\n", + " \n", + " # Convert indices to tensor and add batch dimension\n", + " src_tensor = torch.LongTensor(src_indices).unsqueeze(0)\n", + " \n", + " # Initialize target input with '' token\n", + " trg_input = torch.LongTensor([[fr_vocab['']]])\n", + " \n", + " # Initialize list to store the generated translation\n", + " translation = []\n", + " \n", + " with torch.no_grad():\n", + " for i in range(max_len):\n", + " # Generate mask for source sentence\n", + " src_mask = (src_tensor != en_vocab['']).unsqueeze(-2)\n", + " \n", + " # Generate mask for target sentence\n", + " trg_mask = torch.triu(torch.ones((1, i+1, i+1), device=src_tensor.device)).bool()\n", + " \n", + " # Generate predictions for next token\n", + " preds = model(src_tensor, trg_input, src_mask, trg_mask)\n", + " \n", + " # Get predicted token (index)\n", + " pred_token = preds.argmax(dim=-1)[:,-1].item()\n", + " \n", + " # Append predicted token to translation list\n", + " translation.append(pred_token)\n", + " \n", + " # If predicted token is end-of-sentence token, stop\n", + " if pred_token == fr_vocab['']:\n", + " break\n", + " \n", + " # Append predicted token to target input for next iteration\n", + " trg_input = torch.cat([trg_input, torch.LongTensor([[pred_token]])], dim=-1)\n", + " \n", + " # Convert indices back to tokens using the French vocabulary\n", + " translated_sentence = [fr_index_to_word[token] for token in translation]\n", + " \n", + " return ' '.join(translated_sentence)\n", "\n", - "- [chatgpt](https://openai.com/product/chatgpt)" + "for src_sentence in en_data:\n", + " translation = translate(model, src_sentence, en_vocab, fr_vocab)\n", + " print(\"Source:\", src_sentence)\n", + " print(\"Translation:\", translation)\n", + " print()" ] } ], diff --git a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb index 2c4b9c098..792958402 100644 --- a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb +++ b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb @@ -2,6 +2,7 @@ "cells": [ { "cell_type": "markdown", + "id": "dc000dc9", "metadata": { "tags": [ "remove-cell" @@ -22,324 +23,1556 @@ }, { "cell_type": "markdown", + "id": "8bfa70ec-5c4c-40e8-b923-16f8167e3181", "metadata": {}, "source": [ - "# Attention\n", - "## What is Attention?\n", + "# Coding Attention Mechanisms" + ] + }, + { + "cell_type": "markdown", + "id": "c29bcbe8-a034-43a2-b557-997b03c9882d", + "metadata": {}, + "source": [ + "Packages that are being used in this notebook:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "e58f33e8-5dc9-4dd5-ab84-5a011fa11d92", + "metadata": {}, + "outputs": [], + "source": [ + "from importlib.metadata import version\n", + "\n", + "print(\"torch version:\", version(\"torch\"))" + ] + }, + { + "cell_type": "markdown", + "id": "a2a4474d-7c68-4846-8702-37906cf08197", + "metadata": {}, + "source": [ + "- This section covers attention mechanisms, the engine of LLMs:" + ] + }, + { + "cell_type": "markdown", + "id": "02a11208-d9d3-44b1-8e0d-0c8414110b93", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "50e020fd-9690-4343-80df-da96678bef5e", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "ecc4dcee-34ea-4c05-9085-2f8887f70363", + "metadata": {}, + "source": [ + "## The problem with modeling long sequences" + ] + }, + { + "cell_type": "markdown", + "id": "a55aa49c-36c2-48da-b1d9-70f416e46a6a", + "metadata": {}, + "source": [ + "- No code in this section\n", + "- Translating a text word by word isn't feasible due to the differences in grammatical structures between the source and target languages:" + ] + }, + { + "cell_type": "markdown", + "id": "55c0c433-aa4b-491e-848a-54905ebb05ad", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "db03c48a-3429-48ea-9d4a-2e53b0e516b1", + "metadata": {}, + "source": [ + "- Prior to the introduction of transformer models, encoder-decoder RNNs were commonly used for machine translation tasks\n", + "- In this setup, the encoder processes a sequence of tokens from the source language, using a hidden state—a kind of intermediate layer within the neural network—to generate a condensed representation of the entire input sequence:" + ] + }, + { + "cell_type": "markdown", + "id": "03d8df2c-c1c2-4df0-9977-ade9713088b2", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "3602c585-b87a-41c7-a324-c5e8298849df", + "metadata": {}, + "source": [ + "## Capturing data dependencies with attention mechanisms" + ] + }, + { + "cell_type": "markdown", + "id": "b6fde64c-6034-421d-81d9-8244932086ea", + "metadata": {}, + "source": [ + "- No code in this section\n", + "- Through an attention mechanism, the text-generating decoder segment of the network is capable of selectively accessing all input tokens, implying that certain input tokens hold more significance than others in the generation of a specific output token:" + ] + }, + { + "cell_type": "markdown", + "id": "bc4f6293-8ab5-4aeb-a04c-50ee158485b1", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "8044be1f-e6a2-4a1f-a6dd-e325d3bad05e", + "metadata": {}, + "source": [ + "- Self-attention in transformers is a technique designed to enhance input representations by enabling each position in a sequence to engage with and determine the relevance of every other position within the same sequence" + ] + }, + { + "cell_type": "markdown", + "id": "6565dc9f-b1be-4c78-b503-42ccc743296c", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "5efe05ff-b441-408e-8d66-cde4eb3397e3", + "metadata": {}, + "source": [ + "## Attending to different parts of the input with self-attention" + ] + }, + { + "cell_type": "markdown", + "id": "6d9af516-7c37-4400-ab53-34936d5495a9", + "metadata": {}, + "source": [ + "### A simple self-attention mechanism without trainable weights" + ] + }, + { + "cell_type": "markdown", + "id": "d269e9f1-df11-4644-b575-df338cf46cdf", + "metadata": {}, + "source": [ + "- This section explains a very simplified variant of self-attention, which does not contain any trainable weights\n", + "- This is purely for illustration purposes and NOT the attention mechanism that is used in transformers\n", + "- The next section, section 3.3.2, will extend this simple attention mechanism to implement the real self-attention mechanism\n", + "- Suppose we are given an input sequence $x^{(1)}$ to $x^{(T)}$\n", + " - The input is a text (for example, a sentence like \"Your journey starts with one step\") that has already been converted into token embeddings as described in chapter 2\n", + " - For instance, $x^{(1)}$ is a d-dimensional vector representing the word \"Your\", and so forth\n", + "- **Goal:** compute context vectors $z^{(i)}$ for each input sequence element $x^{(i)}$ in $x^{(1)}$ to $x^{(T)}$ (where $z$ and $x$ have the same dimension)\n", + " - A context vector $z^{(i)}$ is a weighted sum over the inputs $x^{(1)}$ to $x^{(T)}$\n", + " - The context vector is \"context\"-specific to a certain input\n", + " - Instead of $x^{(i)}$ as a placeholder for an arbitrary input token, let's consider the second input, $x^{(2)}$\n", + " - And to continue with a concrete example, instead of the placeholder $z^{(i)}$, we consider the second output context vector, $z^{(2)}$\n", + " - The second context vector, $z^{(2)}$, is a weighted sum over all inputs $x^{(1)}$ to $x^{(T)}$ weighted with respect to the second input element, $x^{(2)}$\n", + " - The attention weights are the weights that determine how much each of the input elements contributes to the weighted sum when computing $z^{(2)}$\n", + " - In short, think of $z^{(2)}$ as a modified version of $x^{(2)}$ that also incorporates information about all other input elements that are relevant to a given task at hand" + ] + }, + { + "cell_type": "markdown", + "id": "fcc7c7a2-b6ab-478f-ae37-faa8eaa8049a", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "ff856c58-8382-44c7-827f-798040e6e697", + "metadata": {}, + "source": [ + "- By convention, the unnormalized attention weights are referred to as **\"attention scores\"** whereas the normalized attention scores, which sum to 1, are referred to as **\"attention weights\"**\n", + "\n", + "- The attention weights and context vector calculation are summarized in the figure below:" + ] + }, + { + "cell_type": "markdown", + "id": "01b10344-128d-462a-823f-2178dff5fd58", + "metadata": {}, + "source": [ + "- The code below walks through the figure above step by step\n", + "\n", + "
    \n", + "\n", + "- **Step 1:** compute unnormalized attention scores $\\omega$\n", + "- Suppose we use the second input token as the query, that is, $q^{(2)} = x^{(2)}$, we compute the unnormalized attention scores via dot products:\n", + " - $\\omega_{21} = x^{(1)} q^{(2)\\top}$\n", + " - $\\omega_{22} = x^{(2)} q^{(2)\\top}$\n", + " - $\\omega_{23} = x^{(3)} q^{(2)\\top}$\n", + " - ...\n", + " - $\\omega_{2T} = x^{(T)} q^{(2)\\top}$\n", + "- Above, $\\omega$ is the Greek letter \"omega\" used to symbolize the unnormalized attention scores\n", + " - The subscript \"21\" in $\\omega_{21}$ means that input sequence element 2 was used as a query against input sequence element 1" + ] + }, + { + "cell_type": "markdown", + "id": "35e55f7a-f2d0-4f24-858b-228e4fe88fb3", + "metadata": {}, + "source": [ + "- Suppose we have the following input sentence that is already embedded in 3-dimensional vectors as described in chapter 3 (we use a very small embedding dimension here for illustration purposes, so that it fits onto the page without line breaks):" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "22b9556a-aaf8-4ab4-a5b4-973372b0b2c3", + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "\n", + "inputs = torch.tensor(\n", + " [[0.43, 0.15, 0.89], # Your (x^1)\n", + " [0.55, 0.87, 0.66], # journey (x^2)\n", + " [0.57, 0.85, 0.64], # starts (x^3)\n", + " [0.22, 0.58, 0.33], # with (x^4)\n", + " [0.77, 0.25, 0.10], # one (x^5)\n", + " [0.05, 0.80, 0.55]] # step (x^6)\n", + ")" + ] + }, + { + "cell_type": "markdown", + "id": "299baef3-b1a8-49ba-bad4-f62c8a416d83", + "metadata": {}, + "source": [ + "- The primary objective of this section is to demonstrate how the context vector $z^{(2)}$\n", + " is calculated using the second input sequence, $x^{(2)}$, as a query\n", + "\n", + "- The figure depicts the initial step in this process, which involves calculating the attention scores ω between $x^{(2)}$\n", + " and all other input elements through a dot product operation." + ] + }, + { + "cell_type": "markdown", + "id": "5cb3453a-58fa-42c4-b225-86850bc856f8", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "77be52fb-82fd-4886-a4c8-f24a9c87af22", + "metadata": {}, + "source": [ + "- We use input sequence element 2, $x^{(2)}$, as an example to compute context vector $z^{(2)}$; later in this section, we will generalize this to compute all context vectors.\n", + "- The first step is to compute the unnormalized attention scores by computing the dot product between the query $x^{(2)}$ and all other input tokens:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "6fb5b2f8-dd2c-4a6d-94ef-a0e9ad163951", + "metadata": {}, + "outputs": [], + "source": [ + "query = inputs[1] # 2nd input token is the query\n", + "\n", + "attn_scores_2 = torch.empty(inputs.shape[0])\n", + "for i, x_i in enumerate(inputs):\n", + " attn_scores_2[i] = torch.dot(x_i, query) # dot product (transpose not necessary here since they are 1-dim vectors)\n", + "\n", + "print(attn_scores_2)" + ] + }, + { + "cell_type": "markdown", + "id": "8df09ae0-199f-4b6f-81a0-2f70546684b8", + "metadata": {}, + "source": [ + "- Side note: a dot product is essentially a shorthand for multiplying two vectors elements-wise and summing the resulting products:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "9842f39b-1654-410e-88bf-d1b899bf0241", + "metadata": {}, + "outputs": [], + "source": [ + "res = 0.\n", + "\n", + "for idx, element in enumerate(inputs[0]):\n", + " res += inputs[0][idx] * query[idx]\n", + "\n", + "print(res)\n", + "print(torch.dot(inputs[0], query))" + ] + }, + { + "cell_type": "markdown", + "id": "7d444d76-e19e-4e9a-a268-f315d966609b", + "metadata": {}, + "source": [ + "- **Step 2:** normalize the unnormalized attention scores (\"omegas\", $\\omega$) so that they sum up to 1\n", + "- Here is a simple way to normalize the unnormalized attention scores to sum up to 1 (a convention, useful for interpretation, and important for training stability):" + ] + }, + { + "cell_type": "markdown", + "id": "dfd965d6-980c-476a-93d8-9efe603b1b3b", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "e3ccc99c-33ce-4f11-b7f2-353cf1cbdaba", + "metadata": {}, + "outputs": [], + "source": [ + "attn_weights_2_tmp = attn_scores_2 / attn_scores_2.sum()\n", + "\n", + "print(\"Attention weights:\", attn_weights_2_tmp)\n", + "print(\"Sum:\", attn_weights_2_tmp.sum())" + ] + }, + { + "cell_type": "markdown", + "id": "75dc0a57-f53e-41bf-8793-daa77a819431", + "metadata": {}, + "source": [ + "- However, in practice, using the softmax function for normalization, which is better at handling extreme values and has more desirable gradient properties during training, is common and recommended.\n", + "- Here's a naive implementation of a softmax function for scaling, which also normalizes the vector elements such that they sum up to 1:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "07b2e58d-a6ed-49f0-a1cd-2463e8d53a20", + "metadata": {}, + "outputs": [], + "source": [ + "def softmax_naive(x):\n", + " return torch.exp(x) / torch.exp(x).sum(dim=0)\n", + "\n", + "attn_weights_2_naive = softmax_naive(attn_scores_2)\n", + "\n", + "print(\"Attention weights:\", attn_weights_2_naive)\n", + "print(\"Sum:\", attn_weights_2_naive.sum())" + ] + }, + { + "cell_type": "markdown", + "id": "f0a1cbbb-4744-41cb-8910-f5c1355555fb", + "metadata": {}, + "source": [ + "- The naive implementation above can suffer from numerical instability issues for large or small input values due to overflow and underflow issues\n", + "- Hence, in practice, it's recommended to use the PyTorch implementation of softmax instead, which has been highly optimized for performance:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "2d99cac4-45ea-46b3-b3c1-e000ad16e158", + "metadata": {}, + "outputs": [], + "source": [ + "attn_weights_2 = torch.softmax(attn_scores_2, dim=0)\n", + "\n", + "print(\"Attention weights:\", attn_weights_2)\n", + "print(\"Sum:\", attn_weights_2.sum())" + ] + }, + { + "cell_type": "markdown", + "id": "e43e36c7-90b2-427f-94f6-bb9d31b2ab3f", + "metadata": {}, + "source": [ + "- **Step 3**: compute the context vector $z^{(2)}$ by multiplying the embedded input tokens, $x^{(i)}$ with the attention weights and sum the resulting vectors:" + ] + }, + { + "cell_type": "markdown", + "id": "f1c9f5ac-8d3d-4847-94e3-fd783b7d4d3d", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "8fcb96f0-14e5-4973-a50e-79ea7c6af99f", + "metadata": {}, + "outputs": [], + "source": [ + "query = inputs[1] # 2nd input token is the query\n", + "\n", + "context_vec_2 = torch.zeros(query.shape)\n", + "for i,x_i in enumerate(inputs):\n", + " context_vec_2 += attn_weights_2[i]*x_i\n", + "\n", + "print(context_vec_2)" + ] + }, + { + "cell_type": "markdown", + "id": "5a454262-40eb-430e-9ca4-e43fb8d6cd89", + "metadata": {}, + "source": [ + "### Computing attention weights for all input tokens" + ] + }, + { + "cell_type": "markdown", + "id": "6a02bb73-fc19-4c88-b155-8314de5d63a8", + "metadata": {}, + "source": [ + "#### Generalize to all input sequence tokens:\n", + "\n", + "- Above, we computed the attention weights and context vector for input 2 (as illustrated in the highlighted row in the figure below)\n", + "- Next, we are generalizing this computation to compute all attention weights and context vectors" + ] + }, + { + "cell_type": "markdown", + "id": "11c0fb55-394f-42f4-ba07-d01ae5c98ab4", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "b789b990-fb51-4beb-9212-bf58876b5983", + "metadata": {}, + "source": [ + "- In self-attention, the process starts with the calculation of attention scores, which are subsequently normalized to derive attention weights that total 1\n", + "- These attention weights are then utilized to generate the context vectors through a weighted summation of the inputs" + ] + }, + { + "cell_type": "markdown", + "id": "d9bffe4b-56fe-4c37-9762-24bd924b7d3c", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "aa652506-f2c8-473c-a905-85c389c842cc", + "metadata": {}, + "source": [ + "- Apply previous **step 1** to all pairwise elements to compute the unnormalized attention score matrix:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "04004be8-07a1-468b-ab33-32e16a551b45", + "metadata": {}, + "outputs": [], + "source": [ + "attn_scores = torch.empty(6, 6)\n", + "\n", + "for i, x_i in enumerate(inputs):\n", + " for j, x_j in enumerate(inputs):\n", + " attn_scores[i, j] = torch.dot(x_i, x_j)\n", + "\n", + "print(attn_scores)" + ] + }, + { + "cell_type": "markdown", + "id": "1539187f-1ece-47b7-bc9b-65a97115f1d4", + "metadata": {}, + "source": [ + "- We can achieve the same as above more efficiently via matrix multiplication:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "2cea69d0-9a47-45da-8d5a-47ceef2df673", + "metadata": {}, + "outputs": [], + "source": [ + "attn_scores = inputs @ inputs.T\n", + "print(attn_scores)" + ] + }, + { + "cell_type": "markdown", + "id": "02c4bac4-acfd-427f-9b11-c436ac71748d", + "metadata": {}, + "source": [ + "- Similar to **step 2** previously, we normalize each row so that the values in each row sum to 1:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "fa4ef062-de81-47ee-8415-bfe1708c81b8", + "metadata": {}, + "outputs": [], + "source": [ + "attn_weights = torch.softmax(attn_scores, dim=1)\n", + "print(attn_weights)" + ] + }, + { + "cell_type": "markdown", + "id": "3fa6d02b-7f15-4eb4-83a7-0b8a819e7a0c", + "metadata": {}, + "source": [ + "- Quick verification that the values in each row indeed sum to 1:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "112b492c-fb6f-4e6d-8df5-518ae83363d5", + "metadata": {}, + "outputs": [], + "source": [ + "row_2_sum = sum([0.1385, 0.2379, 0.2333, 0.1240, 0.1082, 0.1581])\n", + "print(\"Row 2 sum:\", row_2_sum)\n", + "\n", + "print(\"All row sums:\", attn_weights.sum(dim=1))" + ] + }, + { + "cell_type": "markdown", + "id": "138b0b5c-d813-44c7-b373-fde9540ddfd1", + "metadata": {}, + "source": [ + "- Apply previous **step 3** to compute all context vectors:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "ba8eafcf-f7f7-4989-b8dc-61b50c4f81dc", + "metadata": {}, + "outputs": [], + "source": [ + "all_context_vecs = attn_weights @ inputs\n", + "print(all_context_vecs)" + ] + }, + { + "cell_type": "markdown", + "id": "25b245b8-7732-4fab-aa1c-e3d333195605", + "metadata": {}, + "source": [ + "- As a sanity check, the previously computed context vector $z^{(2)} = [0.4419, 0.6515, 0.5683]$ can be found in the 2nd row in above: " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "2570eb7d-aee1-457a-a61e-7544478219fa", + "metadata": {}, + "outputs": [], + "source": [ + "print(\"Previous 2nd context vector:\", context_vec_2)" + ] + }, + { + "cell_type": "markdown", + "id": "a303b6fb-9f7e-42bb-9fdb-2adabf0a6525", + "metadata": {}, + "source": [ + "## Implementing self-attention with trainable weights" + ] + }, + { + "cell_type": "markdown", + "id": "88363117-93d8-41fb-8240-f7cfe08b14a3", + "metadata": {}, + "source": [ + "- A conceptual framework illustrating how the self-attention mechanism developed in this section integrates into the overall narrative and structure of this book and chapter" + ] + }, + { + "cell_type": "markdown", + "id": "ac9492ba-6f66-4f65-bd1d-87cf16d59928", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "2b90a77e-d746-4704-9354-1ddad86e6298", + "metadata": {}, + "source": [ + "### Computing the attention weights step by step" + ] + }, + { + "cell_type": "markdown", + "id": "46e95a46-1f67-4b71-9e84-8e2db84ab036", + "metadata": {}, + "source": [ + "- In this section, we are implementing the self-attention mechanism that is used in the original transformer architecture, the GPT models, and most other popular LLMs\n", + "- This self-attention mechanism is also called \"scaled dot-product attention\"\n", + "- The overall idea is similar to before:\n", + " - We want to compute context vectors as weighted sums over the input vectors specific to a certain input element\n", + " - For the above, we need attention weights\n", + "- As you will see, there are only slight differences compared to the basic attention mechanism introduced earlier:\n", + " - The most notable difference is the introduction of weight matrices that are updated during model training\n", + " - These trainable weight matrices are crucial so that the model (specifically, the attention module inside the model) can learn to produce \"good\" context vectors" + ] + }, + { + "cell_type": "markdown", + "id": "59db4093-93e8-4bee-be8f-c8fac8a08cdd", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "4d996671-87aa-45c9-b2e0-07a7bcc9060a", + "metadata": {}, + "source": [ + "- Implementing the self-attention mechanism step by step, we will start by introducing the three training weight matrices $W_q$, $W_k$, and $W_v$\n", + "- These three matrices are used to project the embedded input tokens, $x^{(i)}$, into query, key, and value vectors via matrix multiplication:\n", + "\n", + " - Query vector: $q^{(i)} = W_q \\,x^{(i)}$\n", + " - Key vector: $k^{(i)} = W_k \\,x^{(i)}$\n", + " - Value vector: $v^{(i)} = W_v \\,x^{(i)}$\n" + ] + }, + { + "cell_type": "markdown", + "id": "9f334313-5fd0-477b-8728-04080a427049", + "metadata": {}, + "source": [ + "- The embedding dimensions of the input $x$ and the query vector $q$ can be the same or different, depending on the model's design and specific implementation\n", + "- In GPT models, the input and output dimensions are usually the same, but for illustration purposes, to better follow the computation, we choose different input and output dimensions here:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "8250fdc6-6cd6-4c5b-b9c0-8c643aadb7db", + "metadata": {}, + "outputs": [], + "source": [ + "x_2 = inputs[1] # second input element\n", + "d_in = inputs.shape[1] # the input embedding size, d=3\n", + "d_out = 2 # the output embedding size, d=2" + ] + }, + { + "cell_type": "markdown", + "id": "f528cfb3-e226-47dd-b363-cc2caaeba4bf", + "metadata": {}, + "source": [ + "- Below, we initialize the three weight matrices; note that we are setting `requires_grad=False` to reduce clutter in the outputs for illustration purposes, but if we were to use the weight matrices for model training, we would set `requires_grad=True` to update these matrices during model training" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "bfd7259a-f26c-4cea-b8fc-282b5cae1e00", + "metadata": {}, + "outputs": [], + "source": [ + "torch.manual_seed(123)\n", + "\n", + "W_query = torch.nn.Parameter(torch.rand(d_in, d_out), requires_grad=False)\n", + "W_key = torch.nn.Parameter(torch.rand(d_in, d_out), requires_grad=False)\n", + "W_value = torch.nn.Parameter(torch.rand(d_in, d_out), requires_grad=False)" + ] + }, + { + "cell_type": "markdown", + "id": "abfd0b50-7701-4adb-821c-e5433622d9c4", + "metadata": {}, + "source": [ + "- Next we compute the query, key, and value vectors:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "73cedd62-01e1-4196-a575-baecc6095601", + "metadata": {}, + "outputs": [], + "source": [ + "query_2 = x_2 @ W_query # _2 because it's with respect to the 2nd input element\n", + "key_2 = x_2 @ W_key \n", + "value_2 = x_2 @ W_value\n", + "\n", + "print(query_2)" + ] + }, + { + "cell_type": "markdown", + "id": "9be308b3-aca3-421b-b182-19c3a03b71c7", + "metadata": {}, + "source": [ + "- As we can see below, we successfully projected the 6 input tokens from a 3D onto a 2D embedding space:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "8c1c3949-fc08-4d19-a41e-1c235b4e631b", + "metadata": {}, + "outputs": [], + "source": [ + "keys = inputs @ W_key \n", + "values = inputs @ W_value\n", + "\n", + "print(\"keys.shape:\", keys.shape)\n", + "print(\"values.shape:\", values.shape)" + ] + }, + { + "cell_type": "markdown", + "id": "bac5dfd6-ade8-4e7b-b0c1-bed40aa24481", + "metadata": {}, + "source": [ + "- In the next step, **step 2**, we compute the unnormalized attention scores by computing the dot product between the query and each key vector:" + ] + }, + { + "cell_type": "markdown", + "id": "8ed0a2b7-5c50-4ede-90cf-7ad74412b3aa", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "64cbc253-a182-4490-a765-246979ea0a28", + "metadata": {}, + "outputs": [], + "source": [ + "keys_2 = keys[1] # Python starts index at 0\n", + "attn_score_22 = query_2.dot(keys_2)\n", + "print(attn_score_22)" + ] + }, + { + "cell_type": "markdown", + "id": "9e9d15c0-c24e-4e6f-a160-6349b418f935", + "metadata": {}, + "source": [ + "- Since we have 6 inputs, we have 6 attention scores for the given query vector:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "b14e44b5-d170-40f9-8847-8990804af26d", + "metadata": {}, + "outputs": [], + "source": [ + "attn_scores_2 = query_2 @ keys.T # All attention scores for given query\n", + "print(attn_scores_2)" + ] + }, + { + "cell_type": "markdown", + "id": "8622cf39-155f-4eb5-a0c0-82a03ce9b999", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "e1609edb-f089-461a-8de2-c20c1bb29836", + "metadata": {}, + "source": [ + "- Next, in **step 3**, we compute the attention weights (normalized attention scores that sum up to 1) using the softmax function we used earlier\n", + "- The difference to earlier is that we now scale the attention scores by dividing them by the square root of the embedding dimension, $\\sqrt{d_k}$ (i.e., `d_k**0.5`):" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "146f5587-c845-4e30-9894-c7ed3a248153", + "metadata": {}, + "outputs": [], + "source": [ + "d_k = keys.shape[1]\n", + "attn_weights_2 = torch.softmax(attn_scores_2 / d_k**0.5, dim=-1)\n", + "print(attn_weights_2)" + ] + }, + { + "cell_type": "markdown", + "id": "b8f61a28-b103-434a-aee1-ae7cbd821126", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "1890e3f9-db86-4ab8-9f3b-53113504a61f", + "metadata": {}, + "source": [ + "- In **step 4**, we now compute the context vector for input query vector 2:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "e138f033-fa7e-4e3a-8764-b53a96b26397", + "metadata": {}, + "outputs": [], + "source": [ + "context_vec_2 = attn_weights_2 @ values\n", + "print(context_vec_2)" + ] + }, + { + "cell_type": "markdown", + "id": "9d7b2907-e448-473e-b46c-77735a7281d8", + "metadata": {}, + "source": [ + "### Implementing a compact SelfAttention class" + ] + }, + { + "cell_type": "markdown", + "id": "04313410-3155-4d90-a7a3-2f3386e73677", + "metadata": {}, + "source": [ + "- Putting it all together, we can implement the self-attention mechanism as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "51590326-cdbe-4e62-93b1-17df71c11ee4", + "metadata": {}, + "outputs": [], + "source": [ + "import torch.nn as nn\n", + "\n", + "class SelfAttention_v1(nn.Module):\n", + "\n", + " def __init__(self, d_in, d_out):\n", + " super().__init__()\n", + " self.d_out = d_out\n", + " self.W_query = nn.Parameter(torch.rand(d_in, d_out))\n", + " self.W_key = nn.Parameter(torch.rand(d_in, d_out))\n", + " self.W_value = nn.Parameter(torch.rand(d_in, d_out))\n", + "\n", + " def forward(self, x):\n", + " keys = x @ self.W_key\n", + " queries = x @ self.W_query\n", + " values = x @ self.W_value\n", + " \n", + " attn_scores = queries @ keys.T # omega\n", + " attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=-1)\n", + "\n", + " context_vec = attn_weights @ values\n", + " return context_vec\n", + "\n", + "torch.manual_seed(123)\n", + "sa_v1 = SelfAttention_v1(d_in, d_out)\n", + "print(sa_v1(inputs))" + ] + }, + { + "cell_type": "markdown", + "id": "7ee1a024-84a5-425a-9567-54ab4e4ed445", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "048e0c16-d911-4ec8-b0bc-45ceec75c081", + "metadata": {}, + "source": [ + "- We can streamline the implementation above using PyTorch's Linear layers, which are equivalent to a matrix multiplication if we disable the bias units\n", + "- Another big advantage of using `nn.Linear` over our manual `nn.Parameter(torch.rand(...)` approach is that `nn.Linear` has a preferred weight initialization scheme, which leads to more stable model training" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "73f411e3-e231-464a-89fe-0a9035e5f839", + "metadata": {}, + "outputs": [], + "source": [ + "class SelfAttention_v2(nn.Module):\n", + "\n", + " def __init__(self, d_in, d_out, qkv_bias=False):\n", + " super().__init__()\n", + " self.d_out = d_out\n", + " self.W_query = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.W_key = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.W_value = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + "\n", + " def forward(self, x):\n", + " keys = self.W_key(x)\n", + " queries = self.W_query(x)\n", + " values = self.W_value(x)\n", + " \n", + " attn_scores = queries @ keys.T\n", + " attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=1)\n", + "\n", + " context_vec = attn_weights @ values\n", + " return context_vec\n", + "\n", + "torch.manual_seed(789)\n", + "sa_v2 = SelfAttention_v2(d_in, d_out)\n", + "print(sa_v2(inputs))" + ] + }, + { + "cell_type": "markdown", + "id": "915cd8a5-a895-42c9-8b8e-06b5ae19ffce", + "metadata": {}, + "source": [ + "- Note that `SelfAttention_v1` and `SelfAttention_v2` give different outputs because they use different initial weights for the weight matrices" + ] + }, + { + "cell_type": "markdown", + "id": "c5025b37-0f2c-4a67-a7cb-1286af7026ab", + "metadata": {}, + "source": [ + "## Hiding future words with causal attention" + ] + }, + { + "cell_type": "markdown", + "id": "aef0a6b8-205a-45bf-9d26-8fd77a8a03c3", + "metadata": {}, + "source": [ + "n causal attention, the attention weights above the diagonal are masked, ensuring that for any given input, the LLM is unable to utilize future tokens while calculating the context vectors with the attention weight" + ] + }, + { + "cell_type": "markdown", + "id": "71e91bb5-5aae-4f05-8a95-973b3f988a35", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "82f405de-cd86-4e72-8f3c-9ea0354946ba", + "metadata": {}, + "source": [ + "### Applying a causal attention mask" + ] + }, + { + "cell_type": "markdown", + "id": "014f28d0-8218-48e4-8b9c-bdc5ce489218", + "metadata": {}, + "source": [ + "- In this section, we are converting the previous self-attention mechanism into a causal self-attention mechanism\n", + "- Causal self-attention ensures that the model's prediction for a certain position in a sequence is only dependent on the known outputs at previous positions, not on future positions\n", + "- In simpler words, this ensures that each next word prediction should only depend on the preceding words\n", + "- To achieve this, for each given token, we mask out the future tokens (the ones that come after the current token in the input text):" + ] + }, + { + "cell_type": "markdown", + "id": "57f99af3-32bc-48f5-8eb4-63504670ca0a", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "cbfaec7a-68f2-4157-a4b5-2aeceed199d9", + "metadata": {}, + "source": [ + "- To illustrate and implement causal self-attention, let's work with the attention scores and weights from the previous section: " + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "1933940d-0fa5-4b17-a3ce-388e5314a1bb", + "metadata": {}, + "outputs": [], + "source": [ + "# Reuse the query and key weight matrices of the\n", + "# SelfAttention_v2 object from the previous section for convenience\n", + "queries = sa_v2.W_query(inputs)\n", + "keys = sa_v2.W_key(inputs) \n", + "attn_scores = queries @ keys.T\n", + "\n", + "attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=1)\n", + "print(attn_weights)" + ] + }, + { + "cell_type": "markdown", + "id": "89020a96-b34d-41f8-9349-98c3e23fd5d6", + "metadata": {}, + "source": [ + "- The simplest way to mask out future attention weights is by creating a mask via PyTorch's tril function with elements below the main diagonal (including the diagonal itself) set to 1 and above the main diagonal set to 0:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "43f3d2e3-185b-4184-9f98-edde5e6df746", + "metadata": {}, + "outputs": [], + "source": [ + "block_size = attn_scores.shape[0]\n", + "mask_simple = torch.tril(torch.ones(block_size, block_size))\n", + "print(mask_simple)" + ] + }, + { + "cell_type": "markdown", + "id": "efce2b08-3583-44da-b3fc-cabdd38761f6", + "metadata": {}, + "source": [ + "- Then, we can multiply the attention weights with this mask to zero out the attention scores above the diagonal:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "9f531e2e-f4d2-4fea-a87f-4c132e48b9e7", + "metadata": {}, + "outputs": [], + "source": [ + "masked_simple = attn_weights*mask_simple\n", + "print(masked_simple)" + ] + }, + { + "cell_type": "markdown", + "id": "3eb35787-cf12-4024-b66d-e7215e175500", + "metadata": {}, + "source": [ + "- However, if the mask were applied after softmax, like above, it would disrupt the probability distribution created by softmax\n", + "- Softmax ensures that all output values sum to 1\n", + "- Masking after softmax would require re-normalizing the outputs to sum to 1 again, which complicates the process and might lead to unintended effects" + ] + }, + { + "cell_type": "markdown", + "id": "94db92d7-c397-4e42-bd8a-6a2b3e237e0f", + "metadata": {}, + "source": [ + "- To make sure that the rows sum to 1, we can normalize the attention weights as follows:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "6d392083-fd81-4f70-9bdf-8db985e673d6", + "metadata": {}, + "outputs": [], + "source": [ + "row_sums = masked_simple.sum(dim=1, keepdim=True)\n", + "masked_simple_norm = masked_simple / row_sums\n", + "print(masked_simple_norm)" + ] + }, + { + "cell_type": "markdown", + "id": "512e7cf4-dc0e-4cec-948e-c7a3c4eb6877", + "metadata": {}, + "source": [ + "- While we are technically done with coding the causal attention mechanism now, let's briefly look at a more efficient approach to achieve the same as above\n", + "- So, instead of zeroing out attention weights above the diagonal and renormalizing the results, we can mask the unnormalized attention scores above the diagonal with negative infinity before they enter the softmax function:" + ] + }, + { + "cell_type": "markdown", + "id": "eb682900-8df2-4767-946c-a82bee260188", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "a2be2f43-9cf0-44f6-8d8b-68ef2fb3cc39", + "metadata": {}, + "outputs": [], + "source": [ + "mask = torch.triu(torch.ones(block_size, block_size), diagonal=1)\n", + "masked = attn_scores.masked_fill(mask.bool(), -torch.inf)\n", + "print(masked)" + ] + }, + { + "cell_type": "markdown", + "id": "91d5f803-d735-4543-b9da-00ac10fb9c50", + "metadata": {}, + "source": [ + "- As we can see below, now the attention weights in each row correctly sum to 1 again:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "b1cd6d7f-16f2-43c1-915e-0824f1a4bc52", + "metadata": {}, + "outputs": [], + "source": [ + "attn_weights = torch.softmax(masked / keys.shape[-1]**0.5, dim=1)\n", + "print(attn_weights)" + ] + }, + { + "cell_type": "markdown", + "id": "7636fc5f-6bc6-461e-ac6a-99ec8e3c0912", + "metadata": {}, + "source": [ + "### Masking additional attention weights with dropout" + ] + }, + { + "cell_type": "markdown", + "id": "ec3dc7ee-6539-4fab-804a-8f31a890c85a", + "metadata": {}, + "source": [ + "- In addition, we also apply dropout to reduce overfitting during training\n", + "- Dropout can be applied in several places:\n", + " - for example, after computing the attention weights;\n", + " - or after multiplying the attention weights with the value vectors\n", + "- Here, we will apply the dropout mask after computing the attention weights because it's more common\n", + "\n", + "- Furthermore, in this specific example, we use a dropout rate of 50%, which means randomly masking out half of the attention weights. (When we train the GPT model later, we will use a lower dropout rate, such as 0.1 or 0.2" + ] + }, + { + "cell_type": "markdown", + "id": "ee799cf6-6175-45f2-827e-c174afedb722", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "5a575458-a6da-4e54-8688-83e155f2de06", + "metadata": {}, + "source": [ + "- If we apply a dropout rate of 0.5 (50%), the non-dropped values will be scaled accordingly by a factor of 1/0.5 = 2." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "0de578db-8289-41d6-b377-ef645751e33f", + "metadata": {}, + "outputs": [], + "source": [ + "torch.manual_seed(123)\n", + "dropout = torch.nn.Dropout(0.5) # dropout rate of 50%\n", + "example = torch.ones(6, 6) # create a matrix of ones\n", "\n", - "The attention mechanism describes a recent new group of layers in neural networks that has attracted a lot of interest in the past few years, especially in sequence tasks. There are a lot of different possible definitions of \"attention\" in the literature, but the one we will use here is the following: _the attention mechanism describes a weighted average of (sequence) elements with the weights dynamically computed based on an input query and elements' keys_. So what does this exactly mean? The goal is to take an average over the features of multiple elements. However, instead of weighting each element equally, we want to weight them depending on their actual values. In other words, we want to dynamically decide on which inputs we want to \"attend\" more than others. In particular, an attention mechanism has usually four parts we need to specify:\n", + "print(dropout(example))" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "b16c5edb-942b-458c-8e95-25e4e355381e", + "metadata": {}, + "outputs": [], + "source": [ + "torch.manual_seed(123)\n", + "print(dropout(attn_weights))" + ] + }, + { + "cell_type": "markdown", + "id": "269df5c8-3e25-49d0-95d3-bb232287404f", + "metadata": {}, + "source": [ + "- Note that the resulting dropout outputs may look different depending on your operating system; you can read more about this inconsistency [here on the PyTorch issue tracker](https://github.com/pytorch/pytorch/issues/121595)" + ] + }, + { + "cell_type": "markdown", + "id": "cdc14639-5f0f-4840-aa9d-8eb36ea90fb7", + "metadata": {}, + "source": [ + "### Implementing a compact causal self-attention class" + ] + }, + { + "cell_type": "markdown", + "id": "09c41d29-1933-43dc-ada6-2dbb56287204", + "metadata": {}, + "source": [ + "- Now, we are ready to implement a working implementation of self-attention, including the causal and dropout masks\n", + "- One more thing is to implement the code to handle batches consisting of more than one input so that our `CausalAttention` class supports the batch outputs produced by the data loader we implemented in chapter 2\n", + "- For simplicity, to simulate such batch input, we duplicate the input text example:" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "977a5fa7-a9d5-4e2e-8a32-8e0331ccfe28", + "metadata": {}, + "outputs": [], + "source": [ + "batch = torch.stack((inputs, inputs), dim=0)\n", + "print(batch.shape) # 2 inputs with 6 tokens each, and each token has embedding dimension 3" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "60d8c2eb-2d8e-4d2c-99bc-9eef8cc53ca0", + "metadata": {}, + "outputs": [], + "source": [ + "class CausalAttention(nn.Module):\n", "\n", - "* **Query**: The query is a feature vector that describes what we are looking for in the sequence, i.e. what would we maybe want to pay attention to.\n", - "* **Keys**: For each input element, we have a key which is again a feature vector. This feature vector roughly describes what the element is \"offering\", or when it might be important. The keys should be designed such that we can identify the elements we want to pay attention to based on the query.\n", - "* **Values**: For each input element, we also have a value vector. This feature vector is the one we want to average over.\n", - "* **Score function**: To rate which elements we want to pay attention to, we need to specify a score function $f_{attn}$. The score function takes the query and a key as input, and output the score/attention weight of the query-key pair. It is usually implemented by simple similarity metrics like a dot product, or a small MLP.\n", + " def __init__(self, d_in, d_out, block_size, dropout, qkv_bias=False):\n", + " super().__init__()\n", + " self.d_out = d_out\n", + " self.W_query = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.W_key = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.W_value = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.dropout = nn.Dropout(dropout) # New\n", + " self.register_buffer('mask', torch.triu(torch.ones(block_size, block_size), diagonal=1)) # New\n", "\n", + " def forward(self, x):\n", + " b, num_tokens, d_in = x.shape # New batch dimension b\n", + " keys = self.W_key(x)\n", + " queries = self.W_query(x)\n", + " values = self.W_value(x)\n", "\n", - "The weights of the average are calculated by a softmax over all score function outputs. Hence, we assign those value vectors a higher weight whose corresponding key is most similar to the query. If we try to describe it with pseudo-math, we can write: \n", + " attn_scores = queries @ keys.transpose(1, 2) # Changed transpose\n", + " attn_scores.masked_fill_( # New, _ ops are in-place\n", + " self.mask.bool()[:num_tokens, :num_tokens], -torch.inf) \n", + " attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=-1)\n", + " attn_weights = self.dropout(attn_weights) # New\n", "\n", - "$$\n", - "\\alpha_i = \\frac{\\exp\\left(f_{attn}\\left(\\text{key}_i, \\text{query}\\right)\\right)}{\\sum_j \\exp\\left(f_{attn}\\left(\\text{key}_j, \\text{query}\\right)\\right)}, \\hspace{5mm} \\text{out} = \\sum_i \\alpha_i \\cdot \\text{value}_i\n", - "$$\n", + " context_vec = attn_weights @ values\n", + " return context_vec\n", "\n", - "Visually, we can show the attention over a sequence of words as follows:\n", + "torch.manual_seed(123)\n", "\n", - ":::{figure} ../image/attention_example.svg\n", - ":::\n", + "block_size = batch.shape[1]\n", + "ca = CausalAttention(d_in, d_out, block_size, 0.0)\n", "\n", - "For every word, we have one key and one value vector. The query is compared to all keys with a score function (in this case the dot product) to determine the weights. The softmax is not visualized for simplicity. Finally, the value vectors of all words are averaged using the attention weights.\n", + "context_vecs = ca(batch)\n", "\n", - "Most attention mechanisms differ in terms of what queries they use, how the key and value vectors are defined, and what score function is used. The attention applied inside the Transformer architecture is called **self-attention**. In self-attention, each sequence element provides a key, value, and query. For each element, we perform an attention layer where based on its query, we check the similarity of the all sequence elements' keys, and returned a different, averaged value vector for each element. We will now go into a bit more detail by first looking at the specific implementation of the attention mechanism which is in the Transformer case the scaled dot product attention." + "print(context_vecs)\n", + "print(\"context_vecs.shape:\", context_vecs.shape)" + ] + }, + { + "cell_type": "markdown", + "id": "c4333d12-17e4-4bb5-9d83-54b3a32618cd", + "metadata": {}, + "source": [ + "- Note that dropout is only applied during training, not during inference" + ] + }, + { + "cell_type": "markdown", + "id": "a554cf47-558c-4f45-84cd-bf9b839a8d50", + "metadata": {}, + "source": [ + "" + ] + }, + { + "cell_type": "markdown", + "id": "c8bef90f-cfd4-4289-b0e8-6a00dc9be44c", + "metadata": {}, + "source": [ + "## Extending single-head attention to multi-head attention" + ] + }, + { + "cell_type": "markdown", + "id": "11697757-9198-4a1c-9cee-f450d8bbd3b9", + "metadata": {}, + "source": [ + "### Stacking multiple single-head attention layers" ] }, { "cell_type": "markdown", + "id": "70766faf-cd53-41d9-8a17-f1b229756a5a", "metadata": {}, "source": [ - "## Scaled Dot Product Attention\n", + "- Below is a summary of the self-attention implemented previously (causal and dropout masks not shown for simplicity)\n", "\n", - "The core concept behind self-attention is the scaled dot product attention. Our goal is to have an attention mechanism with which any element in a sequence can attend to any other while still being efficient to compute. The dot product attention takes as input a set of queries $Q\\in\\mathbb{R}^{T\\times d_k}$, keys $K\\in\\mathbb{R}^{T\\times d_k}$ and values $V\\in\\mathbb{R}^{T\\times d_v}$ where $T$ is the sequence length, and $d_k$ and $d_v$ are the hidden dimensionality for queries/keys and values respectively. For simplicity, we neglect the batch dimension for now. The attention value from element $i$ to $j$ is based on its similarity of the query $Q_i$ and key $K_j$, using the dot product as the similarity metric. In math, we calculate the dot product attention as follows:\n", + "- This is also called single-head attention:\n", "\n", - "$$\\text{Attention}(Q,K,V)=\\text{softmax}\\left(\\frac{QK^T}{\\sqrt{d_k}}\\right)V$$\n", + "\n", "\n", - "The matrix multiplication $QK^T$ performs the dot product for every possible pair of queries and keys, resulting in a matrix of the shape $T\\times T$. Each row represents the attention logits for a specific element $i$ to all other elements in the sequence. On these, we apply a softmax and multiply with the value vector to obtain a weighted mean (the weights being determined by the attention). Another perspective on this attention mechanism offers the computation graph which is visualized below (figure credit - [Vaswani et al., 2017](https://arxiv.org/abs/1706.03762)).\n", + "- We simply stack multiple single-head attention modules to obtain a multi-head attention module:\n", "\n", - ":::{figure} ../image/scaled_dot_product_attn.svg\n", - ":::\n", + "\n", "\n", - "One aspect we haven't discussed yet is the scaling factor of $1/\\sqrt{d_k}$. This scaling factor is crucial to maintain an appropriate variance of attention values after initialization. Remember that we intialize our layers with the intention of having equal variance throughout the model, and hence, $Q$ and $K$ might also have a variance close to $1$. However, performing a dot product over two vectors with a variance $\\sigma^2$ results in a scalar having $d_k$-times higher variance: \n", + "- The main idea behind multi-head attention is to run the attention mechanism multiple times (in parallel) with different, learned linear projections. This allows the model to jointly attend to information from different representation subspaces at different positions." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "b9a66e11-7105-4bb4-be84-041f1a1f3bd2", + "metadata": {}, + "outputs": [], + "source": [ + "class MultiHeadAttentionWrapper(nn.Module):\n", + "\n", + " def __init__(self, d_in, d_out, block_size, dropout, num_heads, qkv_bias=False):\n", + " super().__init__()\n", + " self.heads = nn.ModuleList(\n", + " [CausalAttention(d_in, d_out, block_size, dropout, qkv_bias) \n", + " for _ in range(num_heads)]\n", + " )\n", + "\n", + " def forward(self, x):\n", + " return torch.cat([head(x) for head in self.heads], dim=-1)\n", "\n", - "$$q_i \\sim \\mathcal{N}(0,\\sigma^2), k_i \\sim \\mathcal{N}(0,\\sigma^2) \\to \\text{Var}\\left(\\sum_{i=1}^{d_k} q_i\\cdot k_i\\right) = \\sigma^4\\cdot d_k$$\n", "\n", + "torch.manual_seed(123)\n", "\n", - "If we do not scale down the variance back to $\\sim\\sigma^2$, the softmax over the logits will already saturate to $1$ for one random element and $0$ for all others. The gradients through the softmax will be close to zero so that we can't learn the parameters appropriately. Note that the extra factor of $\\sigma^2$, i.e., having $\\sigma^4$ instead of $\\sigma^2$, is usually not an issue, since we keep the original variance $\\sigma^2$ close to $1$ anyways.\n", + "block_size = batch.shape[1] # This is the number of tokens\n", + "d_in, d_out = 3, 2\n", + "mha = MultiHeadAttentionWrapper(d_in, d_out, block_size, 0.0, num_heads=2)\n", "\n", - "The block `Mask (opt.)` in the diagram above represents the optional masking of specific entries in the attention matrix. This is for instance used if we stack multiple sequences with different lengths into a batch. To still benefit from parallelization in PyTorch, we pad the sentences to the same length and mask out the padding tokens during the calculation of the attention values. This is usually done by setting the respective attention logits to a very low value. \n", + "context_vecs = mha(batch)\n", "\n", - "After we have discussed the details of the scaled dot product attention block, we can write a function below which computes the output features given the triple of queries, keys, and values:" + "print(context_vecs)\n", + "print(\"context_vecs.shape:\", context_vecs.shape)" + ] + }, + { + "cell_type": "markdown", + "id": "193d3d2b-2578-40ba-b791-ea2d49328e48", + "metadata": {}, + "source": [ + "- In the implementation above, the embedding dimension is 4, because we `d_out=2` as the embedding dimension for the key, query, and value vectors as well as the context vector. And since we have 2 attention heads, we have the output embedding dimension 2*2=4" ] }, { "cell_type": "markdown", + "id": "6836b5da-ef82-4b4c-bda1-72a462e48d4e", "metadata": {}, "source": [ - "Below, we import the standard libraries." + "### Implementing multi-head attention with weight splits" + ] + }, + { + "cell_type": "markdown", + "id": "f4b48d0d-71ba-4fa0-b714-ca80cabcb6f7", + "metadata": {}, + "source": [ + "- While the above is an intuitive and fully functional implementation of multi-head attention (wrapping the single-head attention `CausalAttention` implementation from earlier), we can write a stand-alone class called `MultiHeadAttention` to achieve the same\n", + "\n", + "- We don't concatenate single attention heads for this stand-alone `MultiHeadAttention` class\n", + "- Instead, we create single W_query, W_key, and W_value weight matrices and then split those into individual matrices for each attention head:" ] }, { "cell_type": "code", - "execution_count": 16, + "execution_count": null, + "id": "110b0188-6e9e-4e56-a988-10523c6c8538", "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Device: cpu\n" - ] - } - ], + "outputs": [], "source": [ - "## Standard libraries\n", - "import os\n", - "import numpy as np\n", - "import random\n", - "import math\n", - "import json\n", - "from functools import partial\n", + "class MultiHeadAttention(nn.Module):\n", + " def __init__(self, d_in, d_out, block_size, dropout, num_heads, qkv_bias=False):\n", + " super().__init__()\n", + " assert d_out % num_heads == 0, \"d_out must be divisible by num_heads\"\n", "\n", - "## Imports for plotting\n", - "import matplotlib.pyplot as plt\n", - "plt.set_cmap('cividis')\n", - "%matplotlib inline\n", - "from matplotlib.colors import to_rgb\n", - "import matplotlib\n", - "matplotlib.rcParams['lines.linewidth'] = 2.0\n", - "import seaborn as sns\n", - "sns.reset_orig()\n", + " self.d_out = d_out\n", + " self.num_heads = num_heads\n", + " self.head_dim = d_out // num_heads # Reduce the projection dim to match desired output dim\n", "\n", - "## tqdm for loading bars\n", - "from tqdm.notebook import tqdm\n", + " self.W_query = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.W_key = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.W_value = nn.Linear(d_in, d_out, bias=qkv_bias)\n", + " self.out_proj = nn.Linear(d_out, d_out) # Linear layer to combine head outputs\n", + " self.dropout = nn.Dropout(dropout)\n", + " self.register_buffer('mask', torch.triu(torch.ones(block_size, block_size), diagonal=1))\n", "\n", - "## PyTorch\n", - "import torch\n", - "import torch.nn as nn\n", - "import torch.nn.functional as F\n", - "import torch.utils.data as data\n", - "import torch.optim as optim\n", + " def forward(self, x):\n", + " b, num_tokens, d_in = x.shape\n", + "\n", + " keys = self.W_key(x) # Shape: (b, num_tokens, d_out)\n", + " queries = self.W_query(x)\n", + " values = self.W_value(x)\n", + "\n", + " # We implicitly split the matrix by adding a `num_heads` dimension\n", + " # Unroll last dim: (b, num_tokens, d_out) -> (b, num_tokens, num_heads, head_dim)\n", + " keys = keys.view(b, num_tokens, self.num_heads, self.head_dim) \n", + " values = values.view(b, num_tokens, self.num_heads, self.head_dim)\n", + " queries = queries.view(b, num_tokens, self.num_heads, self.head_dim)\n", + "\n", + " # Transpose: (b, num_tokens, num_heads, head_dim) -> (b, num_heads, num_tokens, head_dim)\n", + " keys = keys.transpose(1, 2)\n", + " queries = queries.transpose(1, 2)\n", + " values = values.transpose(1, 2)\n", + "\n", + " # Compute scaled dot-product attention (aka self-attention) with a causal mask\n", + " attn_scores = queries @ keys.transpose(2, 3) # Dot product for each head\n", "\n", - "## Torchvision\n", - "import torchvision\n", - "from torchvision.datasets import CIFAR100\n", - "from torchvision import transforms\n", + " # Original mask truncated to the number of tokens and converted to boolean\n", + " mask_bool = self.mask.bool()[:num_tokens, :num_tokens]\n", "\n", - "# PyTorch Lightning\n", - "try:\n", - " import pytorch_lightning as pl\n", - "except ModuleNotFoundError: # Google Colab does not have PyTorch Lightning installed by default. Hence, we do it here if necessary\n", - " !pip install --quiet pytorch-lightning>=1.4\n", - " import pytorch_lightning as pl\n", - "from pytorch_lightning.callbacks import LearningRateMonitor, ModelCheckpoint\n", + " # Use the mask to fill attention scores\n", + " attn_scores.masked_fill_(mask_bool, -torch.inf)\n", + " \n", + " attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=-1)\n", + " attn_weights = self.dropout(attn_weights)\n", + "\n", + " # Shape: (b, num_tokens, num_heads, head_dim)\n", + " context_vec = (attn_weights @ values).transpose(1, 2) \n", + " \n", + " # Combine heads, where self.d_out = self.num_heads * self.head_dim\n", + " context_vec = context_vec.contiguous().view(b, num_tokens, self.d_out)\n", + " context_vec = self.out_proj(context_vec) # optional projection\n", + "\n", + " return context_vec\n", "\n", - "# Ensure that all operations are deterministic on GPU (if used) for reproducibility\n", - "torch.backends.cudnn.deterministic = True\n", - "torch.backends.cudnn.benchmark = False\n", + "torch.manual_seed(123)\n", "\n", - "device = torch.device(\"cuda:0\") if torch.cuda.is_available() else torch.device(\"cpu\")\n", - "print(\"Device:\", device)" + "batch_size, block_size, d_in = batch.shape\n", + "d_out = 2\n", + "mha = MultiHeadAttention(d_in, d_out, block_size, 0.0, num_heads=2)\n", + "\n", + "context_vecs = mha(batch)\n", + "\n", + "print(context_vecs)\n", + "print(\"context_vecs.shape:\", context_vecs.shape)" ] }, { - "cell_type": "code", - "execution_count": 17, + "cell_type": "markdown", + "id": "d334dfb5-2b6c-4c33-82d5-b4e9db5867bb", "metadata": {}, - "outputs": [], "source": [ - "def scaled_dot_product(q, k, v, mask=None):\n", - " d_k = q.size()[-1]\n", - " attn_logits = torch.matmul(q, k.transpose(-2, -1))\n", - " attn_logits = attn_logits / math.sqrt(d_k)\n", - " if mask is not None:\n", - " attn_logits = attn_logits.masked_fill(mask == 0, -9e15)\n", - " attention = F.softmax(attn_logits, dim=-1)\n", - " values = torch.matmul(attention, v)\n", - " return values, attention" + "- Note that the above is essentially a rewritten version of `MultiHeadAttentionWrapper` that is more efficient\n", + "- The resulting output looks a bit different since the random weight initializations differ, but both are fully functional implementations that can be used in the GPT class we will implement in the upcoming chapters\n", + "- Note that in addition, we added a linear projection layer (`self.out_proj `) to the `MultiHeadAttention` class above. This is simply a linear transformation that doesn't change the dimensions. It's a standard convention to use such a projection layer in LLM implementation, but it's not strictly necessary (recent research has shown that it can be removed without affecting the modeling performance; see the further reading section at the end of this chapter)\n" ] }, { "cell_type": "markdown", + "id": "dbe5d396-c990-45dc-9908-2c621461f851", "metadata": {}, "source": [ - "Note that our code above supports any additional dimensionality in front of the sequence length so that we can also use it for batches. However, for a better understanding, let's generate a few random queries, keys, and value vectors, and calculate the attention outputs:" + "" ] }, { - "cell_type": "code", - "execution_count": 18, + "cell_type": "markdown", + "id": "8b0ed78c-e8ac-4f8f-a479-a98242ae8f65", "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "Seed set to 42\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Q\n", - " tensor([[ 0.3367, 0.1288],\n", - " [ 0.2345, 0.2303],\n", - " [-1.1229, -0.1863]])\n", - "K\n", - " tensor([[ 2.2082, -0.6380],\n", - " [ 0.4617, 0.2674],\n", - " [ 0.5349, 0.8094]])\n", - "V\n", - " tensor([[ 1.1103, -1.6898],\n", - " [-0.9890, 0.9580],\n", - " [ 1.3221, 0.8172]])\n", - "Values\n", - " tensor([[ 0.5698, -0.1520],\n", - " [ 0.5379, -0.0265],\n", - " [ 0.2246, 0.5556]])\n", - "Attention\n", - " tensor([[0.4028, 0.2886, 0.3086],\n", - " [0.3538, 0.3069, 0.3393],\n", - " [0.1303, 0.4630, 0.4067]])\n" - ] - } - ], "source": [ - "seq_len, d_k = 3, 2\n", - "pl.seed_everything(42)\n", - "q = torch.randn(seq_len, d_k)\n", - "k = torch.randn(seq_len, d_k)\n", - "v = torch.randn(seq_len, d_k)\n", - "values, attention = scaled_dot_product(q, k, v)\n", - "print(\"Q\\n\", q)\n", - "print(\"K\\n\", k)\n", - "print(\"V\\n\", v)\n", - "print(\"Values\\n\", values)\n", - "print(\"Attention\\n\", attention)" + "- Note that if you are interested in a compact and efficient implementation of the above, you can also consider the [`torch.nn.MultiheadAttention`](https://pytorch.org/docs/stable/generated/torch.nn.MultiheadAttention.html) class in PyTorch" ] }, { "cell_type": "markdown", + "id": "363701ad-2022-46c8-9972-390d2a2b9911", "metadata": {}, "source": [ - "Before continuing, make sure you can follow the calculation of the specific values here, and also check it by hand. It is important to fully understand how the scaled dot product attention is calculated." + "- Since the above implementation may look a bit complex at first glance, let's look at what happens when executing `attn_scores = queries @ keys.transpose(2, 3)`:" ] }, { - "cell_type": "markdown", + "cell_type": "code", + "execution_count": null, + "id": "e8cfc1ae-78ab-4faa-bc73-98bd054806c9", "metadata": {}, + "outputs": [], "source": [ - "## Multi-Head Attention\n", - "\n", - "The scaled dot product attention allows a network to attend over a sequence. However, often there are multiple different aspects a sequence element wants to attend to, and a single weighted average is not a good option for it. This is why we extend the attention mechanisms to multiple heads, i.e. multiple different query-key-value triplets on the same features. Specifically, given a query, key, and value matrix, we transform those into $h$ sub-queries, sub-keys, and sub-values, which we pass through the scaled dot product attention independently. Afterward, we concatenate the heads and combine them with a final weight matrix. Mathematically, we can express this operation as:\n", - "\n", - "$$\n", - "\\begin{split}\n", - " \\text{Multihead}(Q,K,V) & = \\text{Concat}(\\text{head}_1,...,\\text{head}_h)W^{O}\\\\\n", - " \\text{where } \\text{head}_i & = \\text{Attention}(QW_i^Q,KW_i^K, VW_i^V)\n", - "\\end{split}\n", - "$$\n", + "# (b, num_heads, num_tokens, head_dim) = (1, 2, 3, 4)\n", + "a = torch.tensor([[[[0.2745, 0.6584, 0.2775, 0.8573],\n", + " [0.8993, 0.0390, 0.9268, 0.7388],\n", + " [0.7179, 0.7058, 0.9156, 0.4340]],\n", "\n", - "We refer to this as Multi-Head Attention layer with the learnable parameters $W_{1...h}^{Q}\\in\\mathbb{R}^{D\\times d_k}$, $W_{1...h}^{K}\\in\\mathbb{R}^{D\\times d_k}$, $W_{1...h}^{V}\\in\\mathbb{R}^{D\\times d_v}$, and $W^{O}\\in\\mathbb{R}^{h\\cdot d_v\\times d_{out}}$ ($D$ being the input dimensionality). Expressed in a computational graph, we can visualize it as below (figure credit - [Vaswani et al., 2017](https://arxiv.org/abs/1706.03762)).\n", + " [[0.0772, 0.3565, 0.1479, 0.5331],\n", + " [0.4066, 0.2318, 0.4545, 0.9737],\n", + " [0.4606, 0.5159, 0.4220, 0.5786]]]])\n", "\n", - ":::{figure} ../image/multihead_attention.svg\n", - ":::\n", - "\n", - "How are we applying a Multi-Head Attention layer in a neural network, where we don't have an arbitrary query, key, and value vector as input? Looking at the computation graph above, a simple but effective implementation is to set the current feature map in a NN, $X\\in\\mathbb{R}^{B\\times T\\times d_{\\text{model}}}$, as $Q$, $K$ and $V$ ($B$ being the batch size, $T$ the sequence length, $d_{\\text{model}}$ the hidden dimensionality of $X$). The consecutive weight matrices $W^{Q}$, $W^{K}$, and $W^{V}$ can transform $X$ to the corresponding feature vectors that represent the queries, keys, and values of the input. Using this approach, we can implement the Multi-Head Attention module below." + "print(a @ a.transpose(2, 3))" ] }, { - "cell_type": "code", - "execution_count": 19, + "cell_type": "markdown", + "id": "0587b946-c8f2-4888-adbf-5a5032fbfd7b", "metadata": {}, - "outputs": [], "source": [ - "# Helper function to support different mask shapes.\n", - "# Output shape supports (batch_size, number of heads, seq length, seq length)\n", - "# If 2D: broadcasted over batch size and number of heads\n", - "# If 3D: broadcasted over number of heads\n", - "# If 4D: leave as is\n", - "def expand_mask(mask):\n", - " assert mask.ndim >= 2, \"Mask must be at least 2-dimensional with seq_length x seq_length\"\n", - " if mask.ndim == 3:\n", - " mask = mask.unsqueeze(1)\n", - " while mask.ndim < 4:\n", - " mask = mask.unsqueeze(0)\n", - " return mask" + "- In this case, the matrix multiplication implementation in PyTorch will handle the 4-dimensional input tensor so that the matrix multiplication is carried out between the 2 last dimensions (num_tokens, head_dim) and then repeated for the individual heads \n", + "\n", + "- For instance, the following becomes a more compact way to compute the matrix multiplication for each head separately:" ] }, { "cell_type": "code", - "execution_count": 20, + "execution_count": null, + "id": "053760f1-1a02-42f0-b3bf-3d939e407039", "metadata": {}, "outputs": [], "source": [ - "class MultiheadAttention(nn.Module):\n", - " \n", - " def __init__(self, input_dim, embed_dim, num_heads):\n", - " super().__init__()\n", - " assert embed_dim % num_heads == 0, \"Embedding dimension must be 0 modulo number of heads.\"\n", - " \n", - " self.embed_dim = embed_dim\n", - " self.num_heads = num_heads\n", - " self.head_dim = embed_dim // num_heads\n", - " \n", - " # Stack all weight matrices 1...h together for efficiency\n", - " # Note that in many implementations you see \"bias=False\" which is optional\n", - " self.qkv_proj = nn.Linear(input_dim, 3*embed_dim)\n", - " self.o_proj = nn.Linear(embed_dim, embed_dim)\n", - " \n", - " self._reset_parameters()\n", - "\n", - " def _reset_parameters(self):\n", - " # Original Transformer initialization, see PyTorch documentation\n", - " nn.init.xavier_uniform_(self.qkv_proj.weight)\n", - " self.qkv_proj.bias.data.fill_(0)\n", - " nn.init.xavier_uniform_(self.o_proj.weight)\n", - " self.o_proj.bias.data.fill_(0)\n", - "\n", - " def forward(self, x, mask=None, return_attention=False):\n", - " batch_size, seq_length, _ = x.size()\n", - " if mask is not None:\n", - " mask = expand_mask(mask)\n", - " qkv = self.qkv_proj(x)\n", - " \n", - " # Separate Q, K, V from linear output\n", - " qkv = qkv.reshape(batch_size, seq_length, self.num_heads, 3*self.head_dim)\n", - " qkv = qkv.permute(0, 2, 1, 3) # [Batch, Head, SeqLen, Dims]\n", - " q, k, v = qkv.chunk(3, dim=-1)\n", - " \n", - " # Determine value outputs\n", - " values, attention = scaled_dot_product(q, k, v, mask=mask)\n", - " values = values.permute(0, 2, 1, 3) # [Batch, SeqLen, Head, Dims]\n", - " values = values.reshape(batch_size, seq_length, self.embed_dim)\n", - " o = self.o_proj(values)\n", - " \n", - " if return_attention:\n", - " return o, attention\n", - " else:\n", - " return o" + "first_head = a[0, 0, :, :]\n", + "first_res = first_head @ first_head.T\n", + "print(\"First head:\\n\", first_res)\n", + "\n", + "second_head = a[0, 1, :, :]\n", + "second_res = second_head @ second_head.T\n", + "print(\"\\nSecond head:\\n\", second_res)" ] }, { "cell_type": "markdown", + "id": "3a819062", "metadata": {}, "source": [ - "One crucial characteristic of the multi-head attention is that it is permutation-equivariant with respect to its inputs. This means that if we switch two input elements in the sequence, e.g. $X_1\\leftrightarrow X_2$ (neglecting the batch dimension for now), the output is exactly the same besides the elements 1 and 2 switched. Hence, the multi-head attention is actually looking at the input not as a sequence, but as a set of elements. This property makes the multi-head attention block and the Transformer architecture so powerful and widely applicable! But what if the order of the input is actually important for solving the task, like language modeling? The answer is to encode the position in the input features, which we will take a closer look at later (topic _Positional encodings_ below).\n", - "\n", - "Before moving on to creating the Transformer architecture, we can compare the self-attention operation with our other common layer competitors for sequence data: convolutions and recurrent neural networks. Below you can find a table by [Vaswani et al. (2017)](https://arxiv.org/abs/1706.03762) on the complexity per layer, the number of sequential operations, and maximum path length. The complexity is measured by the upper bound of the number of operations to perform, while the maximum path length represents the maximum number of steps a forward or backward signal has to traverse to reach any other position. The lower this length, the better gradient signals can backpropagate for long-range dependencies. Let's take a look at the table below:\n", - "\n", - ":::{figure} ../image/comparison_conv_rnn.svg\n", - ":::\n", + "## Acknowledgments\n", "\n", - "$n$ is the sequence length, $d$ is the representation dimension and $k$ is the kernel size of convolutions. In contrast to recurrent networks, the self-attention layer can parallelize all its operations making it much faster to execute for smaller sequence lengths. However, when the sequence length exceeds the hidden dimensionality, self-attention becomes more expensive than RNNs. One way of reducing the computational cost for long sequences is by restricting the self-attention to a neighborhood of inputs to attend over, denoted by $r$. Nevertheless, there has been recently a lot of work on more efficient Transformer architectures that still allow long dependencies, of which you can find an overview in the paper by [Tay et al. (2020)](https://arxiv.org/abs/2009.06732) if interested." + "Thanks to [Sebastian Raschka](https://github.com/rasbt) for creating the open-source course [LLMs-from-scratch\n", + "](https://github.com/rasbt/LLMs-from-scratch). It inspires the majority of the content in this chapter.\n" ] } ], @@ -359,9 +1592,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.10.4" + "version": "3.9.18" } }, "nbformat": 4, - "nbformat_minor": 4 + "nbformat_minor": 5 } diff --git a/open-machine-learning-jupyter-book/llm/basic/basic.ipynb b/open-machine-learning-jupyter-book/llm/basic/basic.ipynb index ef8e960b7..c4bdd0379 100644 --- a/open-machine-learning-jupyter-book/llm/basic/basic.ipynb +++ b/open-machine-learning-jupyter-book/llm/basic/basic.ipynb @@ -27,7 +27,7 @@ "# Large Language Models Basic\n", "In these sections, we will explore the attention mechanism, which allows models to focus on specific parts of the input during processing. We will study the Transformer model architecture, which serves as the cornerstone for many state-of-the-art language models, and how it has fundamentally transformed the field of Natural Language Processing (NLP). Additionally, we will introduce generative pre-trained language models like GPT, delve into the network structures of large language models, optimization techniques for attention mechanisms, and practical applications stemming from these foundations.\n", "\n", - ":::{figure} ../image/llm.png\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/llm.png\n", ":::" ] }, diff --git a/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb b/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb index 4143ec705..0836e4464 100644 --- a/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb +++ b/open-machine-learning-jupyter-book/llm/basic/transformer.ipynb @@ -110,7 +110,7 @@ }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -155,7 +155,7 @@ "\n", "Next, we will look at how to apply the multi-head attention block inside the Transformer architecture. Originally, the Transformer model was designed for machine translation. Hence, it got an encoder-decoder structure where the encoder takes as input the sentence in the original language and generates an attention-based representation. On the other hand, the decoder attends over the encoded information and generates the translated sentence in an autoregressive manner, as in a standard RNN. While this structure is extremely useful for Sequence-to-Sequence tasks with the necessity of autoregressive decoding, we will focus here on the encoder part. Many advances in NLP have been made using pure encoder-based Transformer models (if interested, models include the [BERT](https://arxiv.org/abs/1810.04805)-family, the [Vision Transformer](https://arxiv.org/abs/2010.11929), and more), and in our tutorial, we will also mainly focus on the encoder part. If you have understood the encoder architecture, the decoder is a very small step to implement as well. The full Transformer architecture looks as follows (figure credit - [Vaswani et al., 2017](https://arxiv.org/abs/1706.03762)).:\n", "\n", - ":::{figure} ../image/transformer_architecture.svg\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/transformer_architecture.svg\n", ":::\n", "\n", "The encoder consists of $N$ identical blocks that are applied in sequence. Taking as input $x$, it is first passed through a Multi-Head Attention block as we have implemented above. The output is added to the original input using a residual connection, and we apply a consecutive Layer Normalization on the sum. Overall, it calculates $\\text{LayerNorm}(x+\\text{Multihead}(x,x,x))$ ($x$ being $Q$, $K$ and $V$ input to the attention layer). The residual connection is crucial in the Transformer architecture for two reasons: \n", @@ -4033,7 +4033,7 @@ "\n", "One commonly used technique for training a Transformer is learning rate warm-up. This means that we gradually increase the learning rate from 0 on to our originally specified learning rate in the first few iterations. Thus, we slowly start learning instead of taking very large steps from the beginning. In fact, training a deep Transformer without learning rate warm-up can make the model diverge and achieve a much worse performance on training and testing. Take for instance the following plot by [Liu et al. (2019)](https://arxiv.org/pdf/1908.03265.pdf) comparing Adam-vanilla (i.e. Adam without warm-up) vs Adam with a warm-up:\n", "\n", - ":::{figure} ../image/warmup_loss_plot.svg\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/warmup_loss_plot.svg\n", ":::\n", "\n", "Clearly, the warm-up is a crucial hyperparameter in the Transformer architecture. Why is it so important? There are currently two common explanations. Firstly, Adam uses the bias correction factors which however can lead to a higher variance in the adaptive learning rate during the first iterations. Improved optimizers like [RAdam](https://arxiv.org/abs/1908.03265) have been shown to overcome this issue, not requiring warm-up for training Transformers. Secondly, the iteratively applied Layer Normalization across layers can lead to very high gradients during the first iterations, which can be solved by using [Pre-Layer Normalization](https://proceedings.icml.cc/static/paper_files/icml/2020/328-Paper.pdf) (similar to Pre-Activation ResNet), or replacing Layer Normalization by other techniques ([Adaptive Normalization](https://proceedings.icml.cc/static/paper_files/icml/2020/328-Paper.pdf), [Power Normalization](https://arxiv.org/abs/2003.07845)). \n", @@ -6821,9 +6821,7 @@ "\n", "Transformers offer the perfect architecture for this as the Multi-Head Attention is permutation-equivariant, and thus, outputs the same values no matter in what order we enter the inputs (inputs and outputs are permuted equally). The task we are looking at for sets is _Set Anomaly Detection_ which means that we try to find the element(s) in a set that does not fit the others. In the research community, the common application of anomaly detection is performed on a set of images, where $N-1$ images belong to the same category/have the same high-level features while one belongs to another category. Note that category does not necessarily have to relate to a class in a standard classification problem, but could be the combination of multiple features. For instance, on a face dataset, this could be people with glasses, male, beard, etc. An example of distinguishing different animals can be seen below. The first four images show foxes, while the last represents a different animal. We want to recognize that the last image shows a different animal, but it is not relevant which class of animal it is.\n", "\n", - "

  • Yp93GSfm2}3)4rkY9h=kl9*=$cpo+||yH6{j6WfGQ1bu9wDA7!q#$C_-shOfu z6wyk%qayjpe^=JLshh1cAml?JaJ|(EvPgtvPPxY=kZWX971ziAb^H!|JpY{DznRyp zl5FZWEoGP@ka1z07KVEs2d?9V(_hzdU|d%Q9wG(+TEazT68w(i!0Y)$-*!;n&7ZVA zz{kTO36>8+(4uUMrnlhfxxU}{ghHOCh{pr&u3+6HQyHB)TOl#tPxRDjoU;d~u< zJ?-!LdL1~<3$NqEahy1>fnR@p$G`pdD`fm?eic%?e;{Zy+|Tke$j1s2iTtNpDp25Q znP-q7QvybmD2a)3o4${6;m403`2PKSe*OOa8@_$B*Y|JV@W;15a2%&&&1VSlwrzf3 z9vGJ&nuDjY7R8`UUWoffjXiWT!YpT~=NC3oYS}fFVHn$s*-*`brTY9R#qTA5&l?@8 zC{6cOPzv*Xj_<%2fn&oRuq;K+XIS8}k)SE6*Lx$SKiOI(78kdvX`TDg0vhou!=*!9 znkX#3EmYrJ&5p|AisqKhKRYFw(m*Bi{uR1kg!o1?rAZ$&il*J$pBt&%DCuhMrh*8r z-X~IBwr?ii3Gy3HBzqw`R+{}y`?(O}bRpVe!YRJGpPCiAeX?TF_ z&=Uh^HdFBoFnCkPMmsj%t%-_es&(^1$*_DVy}WVqIwA4NBWsKa65t{g#@xz=J}OCI zlv1x2_cswn2*cXDf5|B!nXVe%N6qHH^3Oapj-bg_JxuDl!APO}fH+c&h z^;dW@H2mmDwflrzrtT2zt3W{r0?z?L(xqxxm&8|sxo(c2$?;ryq;PUIPEBXcSsc5I7 zoj}_Z?Lpv8(bRbx&7&1)vDD4efe9q+YJ`~$7yUpGuG>#o;V~u=T<*bjWlnDG3dAB+ zRlGt_A-yWd0+0S#g%>Xf8m*SU8a9o1dB@iWRq`ZOZZxbY8Pzl>eS>x2vy1^&f78vyK zc|;ylWHc4?Vp(Ob0cqi;sd}J^TFwiwlnZlOBqe;g2Gr0UAmef?T_)ujY#6?^ z%eH{=%{~6l7l1$6NHL-Nb@@q`?IhYT2FtiG#^GCv@=!f&LpAcjDvPAlwo)`Rpu8D7 zjwiO`fp)yW{eiw6xc0-0%h`~5S;AB_?z5q)8>yMeRHUs$=`E{^^K3HoXzh!z|0?@LE6~^BSNM zh!B|+NneWpk1}SgvmwDl2QMzIO4Pa7G%=)n-@kps?|=M(Z{PmNhU&-nANcm|8?Ng@ z@0$-AHuOGl?z@K``=;@_%!E#}j?~F&UL+2Zp|LjW5-*y+)JVay=%!j(r^l)6AHzcm zxW|!b83v@>s$i0H)=r$wRNW>-KKbwg#&BT8P|YorEQ1uSA8c;3CM6Xm%ZO5SvhWh9 zV9hB|y>%;9e#}!!Rs*o_`u=F7q(ZiJ1)bMTkomD398zVpiSnOYsTqtr%5~?`-V&iu zG1k>D8l>9Dotjtn^cYx;KO1AdWge5py)xSu)Q!d+Kb!*={e(aN8VQ}#i&8Iv>ZOaK zn8^QbEQx`J4oGaHG&WKr&GbQ}?Yv=YJN9Uxd?VGNU^#8!B0-kZnVluxNEOMO<$ZZ< zLnuUuGGbxg7qOL^G+}i=TK5pz81TsOP1No(=WAfwwtSXEc<}IPcP&jbI2<$Vnb4Mj z<(jXipS_uJHrL|z7IGbs3t}mtE-KUZ1}@KKc8Zyh-78FX_EqFG;BXJ8qLG;7T|7H2 z7~Ym;L_X7|>8+%ez!Aoj1Ul);vpGxC{uL|!g8RQUm(Wik;5qcD?S?9Js4HFkygm+G z*+`wZu4AFn#-Ovc89`&&gN){0W{!;Ze3x0Y@ckqhbyw|Df% z4&EMU+XHR4?Zl-pYM?4kF=P1t*0DWyY>y55+uTTP-bn5Hj=uQ=yx*Iq8;y@D3C4BJ zF&U#=e%>Zxvp7%bNyU?cmO#7;u0*|@RX%eZl|Embce#kJ zw6^2e-_Xw&fbPGK6cc3m3_YJB{9UwUj(iU4WNumXrmTXR`LeQ!3TfX4;SQmWOAb7b zfp6cR`2C-M;M*U+VF001BWNklqfrED4|JcO2QpLG#!RvgSTpPptTmknD?MX^fzMr?*j zm*LJ5z!~p>>XP%5o2ifpl`OZ&fx1R*q^beF1T+_&|1?JPGCiWQc_ihsDG0#CI`sLu zv`!;5oxo|u3fbc1_QC#9c<-UCm zSv(g?sy9-^-<8|Oj*NB3WE&}QpQ1fInu8R2RhRm=YyMqOc-8#>M|~5!p;Ucv)HKK{&=y{MlV2|Dc@!{0cJcJD86KbRW6#+OC{RKrN`o)8 zz6mfYVThj(LerArTRCDy+thXjr=oR5f3w^+MUbS~-j|>(a`+kj+GUvMTM4X2#ckTa z==RUCk%Dm<3S%1s*F3Yd_?vZ#sqZ!Ilmz`n2+IFtBlX)a@7Vmp9hYL9itDtmGLQkX z7C_5I=jegAfh6>QjtjA6lQ93jx}GI-gAp$sPCRq;=rEs67^D|!iF-;~YXA?lHf-)9 z7H~;w>qBwbg!+oN^>`SR6(KYBU*E9*`V02oe!>3RJ0Ab`j{bPV_IN{oyjgb7zMH(u zum553>)0OVd)mro$_l`q{--dwQZub65iirPY=M{-*~QdtM<3C*TPyqhYo%YBAlt&X=o2gyv*ORe<4STYa@lv88f-X3`}iql+Kj-g_lN0;Ig$) zs-^^tBF|=+x9jxZC!rF){V@8Zq;8yfW-q0zuY}T0&3Ob0t{bVFh;9ZNrwZOec7r@M z#9%AODfoDu`2O+2AK!l9_up-cV}XSBI?ofmG2R|Ke*5hWuh%=S>$K?*2#o*ex@}R*bd;ocslL!hecDJRyQv5(hMW@F5WhA*gjP0D z^o3`p&S8}|P&v@fxEWaqqB)$K8*4jHMn;#|`F>^0h>Wmp%=eE8Q`YBAQ*5T<=huxD zd&s-xKDK6I&@F1;Xa+Jfls$$-orn_66rjT3VG`h~{E1Bzbk=tUve5+EZKZI+%1>l|L;bM0Q|52 z{0FY%0aeB8HE=u!Ue621bKp3{vLfOFi(ICS(06-SZ98Gx!sO^OzCHu#%o8D=d!RNY zEXZ=(7)73p-5~ZV&;l8_q~Q|NJRd`B8tc=;!3^qxL5dEz7sM{0iwM{{+S>yU5TtLW z2kwTBwi~M9m*3v-{_7il`?q)e^4lAJ`RyJ3@j!n(u0WQ>ozXTzqbVs;r{UBs(Bu*Eq}0_-Qpp&=vi_x;&w)Ov>w}UQVQl4ex;0XIlL?j) z;iu+nEuwVUR$*o~+`)~1wvj5ZiJ0nT@g-t3ESG?eF(0beb=ulLg7_TYn@Ie8_{~?Y z=Vu3CMVd=O6Q8d0LCB5JbY4H%D5p=V6$ux6BWDv$F3lCK*esVCT9JHRg7Xp_=fLZ@ z&|eo?Kk<5=`0;Vz_FH#6g;?X4} z?Jk`hs<>)HrO}|!`W`1PU9yCgn=i@2YF&!xO;vN*m)YGJ1MxhvqB??sNj0fuTs&eN zDBs&Nfz|=@@{&08OChy1QiP@8M`^OP1WZn(cx({&+$b=l=>}@aNn7HS0x7XnPvAv`Dk1RPRzUf5<+2 zVw<=>_zjQthDUo~?+@&4N9S&NMC^HXqEI#*H^jRHZC>>>*NeuFTMmpu#+om>`|u{S zk+OmC;etO`n%^7dBf8LNU}M3fD`W$XN5lEnasKjvjCZJeV$bXHyrI(r$Q?YegcMJ8 zh~+z)SAqwz3z*H6IAK*F5VBfo;?XsG@5}Re4YI{h>40=VEThd<6-5uJC0Yj#NK0C$ zD_k-U)jtDiF6Z^3j?5^JumGx)Ll?qd4D7?AAb&{<{XevkBH%y&^B<7w9jc0t9|wMX zKk?(+3m@%7;}h(`dqyZBv<~!*Jm8kFJ=hwm7LAlOQ*rRxDsg}c9y^*S5y?Q#QNlT| zGbRo;#g-vBMQ{##?U!Zc2-y=&d4Rh>SkOtFGN|YtYD&GMgRr*-JRVl~Zf0EX+lIDn zc>nc*_g^3Q<<|$^etqEm*EjTiN8ca$YTks+<-c7q7Q*X=`9?0`GeavkLTi&r0^HX8 zH+{9iI@q&hW+He<8$LBI^^UB*f}c@yZvUdX)2e@`Hd`y$jea5xEI91@wu{Qdpb%#M zoV3p-Kk@skI0Z_@vO~Nq6LC7=bZP-I$ z@!`RQ7ETyp!||So7j(HT6svWpG)-fePWl=i2<(T{7%L(@r)F`z6s`O4#QDgxx1?6yzk$?7)KCl;_&nG@Ue%MA?9{@92*J+H7dnSE>3@w5J{fouw z-6NF+w!at64~CY}Gn8Z_z7yPnCy{CO(NMwjuOhZo^Q΀C{cJmr5#+`Q&9VS3gw zGfE0516c9{n-ZEAsI5pblr0m9H5w@oC5|7LQl64;H~$P!-8{v%f@T<}L@4?v(bTei z$C5o*5a_X4kt*X3S1cFU$1NYZz)D1sihYUCRmw}%G3UO`!t@Yik>G*j zD^0ir^!|l5zucFoq~%ZM_6jE|xt9JGF&LYBLKK}l_V$jQA13Q-wvp=GU1i}qdMJo5 ztA050jfTfqv>CVHeot_7s%A3@E@)|{s+wY%`}yRaeVEYPixSr)>xE7OTN6lE=uS9y z#`V^59SiQDq&N{v1`8PtJms& zmZ#qrU-zPjH&0XUJusq?(veM+=9X&YMw-1`SAlPruiP?^Y`lscAiQ!hnkI|AiSxPo zm*aEpy+Nd-_mOo5sh}~yELPKXYo2=}%Odk8D)Jm53Uome15T&_a!LMq)35%C>9nXl z(jTd1xa~0yLa~vWIj2Nsp_plcrI89hYHXsiiSqL#o2haRIDdpMdyQ^bGq#s;Uak+1M6Lrvpi zfYYXCcRl4(Ci^kM4h!d$DEtZflkHFnCO*~bTHp~SABFDC`)B%)7X3iQO41F@j0HJK zOA|%J_++a2(;F#kc5+%1Msv~37JxTA5qx|3WbyKhiJ9?af5%9(=enAYtyC61)7x{W zfe~wZ*H~kk#TkstWh*r?=$vSDkt(;EO%x}<#R+YSyf@Bk;J60bG0?XQ+)un-CqAAB zKAtapd_3{vk(vPYYoDcz&SuZd-xgp&M>A zyg)0{^b}+`>|ZHWi?q+lSCj>|U2784HUG-K;mL${Cx2xmiyyF|H_vSjD=aqE-@npG zNfu}!>5FD6rjLFIMdLzJKohH^jtU*iqfH^o^Y?+l6&m@O5P`jF5+dZGxmt#&f9^hd zao-o+nb{NrO(aO#8|l(Ld*7PIdDBws(4x4M&}KX7v*&|q%(_bW^#&UtcWa) z6#FKMV+$n=b?JL%;S@RGd~bAJf`Y3THVX2(obybXy)}Og zdGlgGtQgYAz+7pbJF)4DxEO1!#qUAiG(bG-PMoZGLG^+-RRHrq48Kt$nA8V^5gyAa zpK9oIKa-|p&>Gg3&1+8i)rLQ7nNg&MUzPUxxlkJ^f#+ArHJc~@cbRS{cAw60=nF$* z8)XBw86qBa7w5GT?L`hdIkHhpL(I)!WxYYQ`}Jeb=+5?;a3^qA+ekIoHqnEfo1gt? zV#`~RCQNe)$m;6t~ehyILgs4L~vvgLgfMV4o#fzTgo=G8*;%Z>hQq6X&qATXJr zl>Zk8GF=I6cq1hybJW4MjnW2^mN;@wBsqPCge(K3LOM>sQVb!~nzWKU7@_4tGfPW- z@N3B)e*KH2U5EZI{Q3X-U;hBZF5-W_{lMP4%VgCi!CFJwY_+-NMvAfR_O&~agPlSL zRgBAx(NR=gQY1rOaQW8hiemUQ55;+o{CbUnel?ukvZJ=vaEMy|Q@?=T!KXmGU^K-x zoRTI0Z}tSV&4YD1``eDkFK>8zf5+P|Z`j}8@c3oJ{?_q$-?6`S?C-m8t9JD5VL$F$ zQR7rCwfF@QAz_ZZRA3FW^^F`x>iB_?8BXe^Wkmy%YX{T|Y;FE$HB@~F6_HWBfhaG& z8x?9vW7jefk>w_wsB=xK&(umY=qEm_iOh}z5|bAAML!^OmG)T?V`(z%KlYXPum`I7 zqb`39mxW$)GF$NP3&AIJ(lkk3nzFu{Lld+@%;nO=f@)FlW~2P_IYlK3PG6t z;F?>XYlsatL`wFAdc$Q+lrjl+7iH+ zDfAmRbhSKo` zR6ILnaXyRl;?>Dblr>OlN!c##GX%GSeJf!=fRML@7f~zBT-Q33%F@10ZuKXA#IH&F zYQ!fk#XMo1sVyDda5*+NG%z+aY}~N(jz@cUY2O?6zN5E2<@=oXo0^o*n<)#Qbx47V zS819gD5aH9IbW2UN~$V!T2dJ7=Z9jDU|^u>z@-DtB}-egvXKGETZ765mA6S2_JQ$m zAjufw(mYN8)bzmyrR}q%&wvp$5zns5$;}T4LJUQB5Y{c;B(}F&3vFwnG=fmIG*Nos z!fTNSGdG7DOo^L;!g|Wg7w=ZQ@6`OonbR~qN|Pj*-nu#c`C238q1dKc80Lp}N~Gpy z%BjsN?#=Utj@=@nzKI$m<$a-=2+#<(iNlUYy!R>X3*||?53LbIn%f@)7u3D~iqUKu z!7bnu{@J(9WPRLocnWohQ@3F^3&3sIwNJd%K9$u&@W!C}z(vXe%8 zJLBhxA_UvE?#(q+B8&VMm;3si5Otk0e)@&j;fH0d@m{qz1kreapj zzkO`=0!VTh!RI&~e|k9%eFgznA{Q@Bl!x=V&xvExRvu`!LG_6x#oTjdC+SV5H-il; zxsT^o(qRrD_*ARAi97rT?m6&nRy0=EuqJ}2S!PrY8N~73_UH8GgOJA9oOowmy-AA9 z>(ZN~F~Y!WbiI)Fp)}G|4pE}d`q8_kJ~JL@dIjQAV-<@5*yd*V@%F$qE^Nn%qaSG8 zFnAQHl9-3j~OHgY2AEd-RNCD{rvEsd0VBjt2eGU@V^|J7#7pC8Z#)}Y>qr5~wsJ7d@) zNF9Bzum`e(HT)?QLU!SWQW;rHT%HA7Hn8vswfem!-f)j`*_jdk#>o0Z=7;_ilM@?c zxIaot6E&rM_a~QTp_OR!1ESQ;_V9+reU0bf@^>%WCSrdd zdH5FA8!QJf)E(XZRln~!3(<0xOi2^_rkA||X}-k?vRG%u)_|=6dnX*-fC5`9c|c4h zD~J|iX>6q$)tsS<%bwsNjwwCQ;MDNur0O_jpzDy5@Gn&IS|Z^l$E^$-3h^!mx^1di ztJ1!D`GRt)V2FEyrgWAPsA2cmHH-BxMMg~6gx!E*68+^q~jvWWzS_4P!XOZ@4 zGq02iK^okX7VqzElNTP^1W}=cb1Mp)V=4;Z{@L&~1!O`bDUf<|$Y)4<$?03$4gHB% zz5t`bPZ@wDjR9ryxn1=!pK(gFYLk+y7hoPvifn=1l;#o%uQ1s!>&U1i=%fiu5C0re z>EE@H`nP|)V{6_tmDO$Q4Sj1F=oRM*k)f1#-$?DdP2|Q9_o~=eD|VdF^|G(m2>p6Y zuYFw`xgY*x6+h%Q26xlvKlYBdw;hjt$NSrk_xBzD_HV!7-~as`zy9`)fBgF|c>Cpn zeedXdN8cOvy@P|&=Xh!`Tc*>#nJjTvG!`%kI5IO(29l%TYoma2?I%ickFuFUOvOtP zXPxwAT4#U7rb9{widR>N0iSK7WHGzLFHrkT=Eew%AC!u2b|tT69?)WC(nMj?LWL>0 z1UFaJ3`Hy1EIOauMpco&XzMpzZerEc*No!=#=y7^jPtcei5B&Q-{vE&l z`WxQg-|*OXG*(07!ZW@gDW|<+@G&GkN|TVy77+lW`RM=K-n%ZzktA7qN7dXtE(uf> z$p;?Umt8JN+2!mmr;riurph0BF*VObH?n$UT4W_mBsx*m1t2rR-P~05obPBg&KF+@ zH*hnIsyidVKB?lAND?lWE3Wt5Bp*8PzmY$-Kb|G)%QLEcN^ZrJg(ts19@w4_Tw?O( zQqFWm1(KV1;-^f{H2NBW>Nxt4E%+gu@*qEe3X69|nb-+N_M)cN!B%E+I3FX`~lnv45CKZWIv$F$uKea+^1`8f+5P90y(V zJnE=rl5>Dk1{YIWI3mAJFs)f}V2`3+Moj0_y~2PGOwFpA(S)GDN**+Un&&bqmyuWG zB(pd~_LxLT7=6>@LpiM<4|Vl|xdmK7H4mvV4L7xcLO8mU4=qMPS;MKf#ABa1*kC%U81Q77p!FJM{h{in^JkIg36YS~?# z1f4rNnS4&U`y`SfWLW<*`8Ur5R8%frWLS~b-Hy=nyA|2LMCj3w_QBO25nZCApi;H@ zyJ#Ptn+SF##&dt-c|35)6J2-DISZ6LCPhfd7@676f<{kX7n9_fkKf=qOJgU?vc2{l zy)pX!bcr=3Wzp`j21;*qD2F7QE;dLYFdxI-psDD$XFu6>0vI?*eUK4R76F3M^3Uqi zqxeb=Tk!BklbD~Uh2vq8lu|0YUu4P+=_bZ?$fMv651lF%9yBLBiORURQWw>;Iu+ey zztIt(2BJ3PhW?xq5OerOX(MhXc08Tb_1q6UALiwFJa#er016kv=3rUbV^k`o2AO`@iKwh0B;Jt3Tu;1pD_te#iIk-*A7s<9fYf+g886kz_&cDZx0FC5z;?hzqVDT)b zqO(cqBs=-4sU0eTQAEY$fEnxcf@|LsZEkp1o4Os39qYE?IF1ZVPBXH97RO;S#^>XK z{b91s_}nYzxsnkB4x=Y9_g4xjC~)+nOr&BO6={c>2kR(`n2TsBKn?8DnP!NDlFp+l|{dp7EwRneqvf9$B!B zU~?mdKKI?GQu0KXJw(rbrlBz{D)6V)80)rRTNkXYVO^Se`CWk`Z09;vMQ;t=DbNQK zx=%)?d{39LX2Bf3Gf_L6G<(1Lq5k>PxYo7E3jI0*2<*xboV@Kx91ZY-7g1W`?GWTT zDVa`Xy7`F`I=!q+?BUKO0-hALuH0Ja@dmKmGN#JoX0mhbW$W6X~%*Mb?RpY?<7x>K>U&h_0g~>HxQR8dT#ZLI+uSomU#Y%4WRx6~? z4`@Fi$37-fj|Uz<-tl;UV1F6`GNlJJBp0nUzb+P^NIABCF$ri;l*TArN}y~^q>yqc zE2R3EqSgX3C4T$?e+t;xKOiQ{b}@O{<$^7yQtO(9l=!u?ABV~Oo)0fFY_ne16|HUd z^Q;TDWx?gL;EzAPmxJrTTr(eRLp+hO9=8?$p62wm(-K`-y(-^HfU<$+0$9bm7V;rOr-YX7{d@b`DhY9?gn2n@)K(Vi{^`3V8e&RPw6A6u6yxx zcyZ7jI3w=A5T@}E!f5=&aWg1LktHQH!_ONs>cxeU-RB$+HZs~1lc0I9C#w}v%$$38 zFH+c&x@*@bFsQHoLS$LV@FB+J*SvbZKjvpMLKk_GThXbjKMsRZ4a)_IH^k67!%_;| z1WCA*BHbLQcXD5udUI7=sMfN<7PHt9XMuoRV8AJa&WP!gYWj<<)eKFAVmkGxo(@)8 zc#6%x%fvWfK^w&s7BpSK6cee{5AtIoH5d_P8pTHVB919u_@n|XL0RZacCkn2Jpe;A zACh!vjM{HlJQSU|jj%=j>5kwJE2n?wmZggS(KMi@&t zmQN906;zTiiXg~Jx3iB%jD_T0{FSCCq?}*e+M&`t`{_Ve`+2$^_H3xvus0|6{lI>> z1;qdR$UGYpDUY0K4Q*j8&7Sd>ZM9s_ko32Zn2zsJdb0wmaelR7q8;oQZGe$D2>xkw zGSAK;5HX3FUNVa{?!$IlhYb!VQ@@uqCs9VM3lJ6(gHAYS<&ZzD z0}G!Si0-qcDnaHHRn4Adoq=L`k0^blMaLA72Xt6@d?%V_#rPQ3&hq-p% zV76x+u?;xVI#BTIgbHK(50>_Q`|XC?l|5%;pgxz?4rNDpCl2ukoO&4V&=eng$Md-Z zKM02xR7MoD*Pnmd*Q3{5kB~Uc?*}tsYfise7<^r@-ZosmU2yyDhWqzBzWw%w-+zC@ zZ@+!R?|*#5@4vs{kKe!JcHgk9&HPMWOc6zn)xhEgInT_)p?Q}ijtt9yn1Nix$C5nc z9^O@fDjM5?DH5-V0&}{UW)>|srI@Hdj+FOFKU~@<+4<5mKx)buVk)Hq^8pO8H-v_KBsmL%M&p zLMq@2PY-r|KFlZ8{W=|Xm`FA^57Dyt)N=8HYIW0S;|U*}VX`ri@(o_7<(x47E>+A^ z!F+iKt72pcU|!Ic)hADvA@4g)q*_}byZOHM-Tj1)G*>me>auRw)`sh4!}Yr1dNKLy zAHRRY_wR3bd%NR$yOU1DN1QIq>9gSD0XB{RHJi{fjb%d=SbLaNg! zq%_%5QBZLq^o8849?JYyA}X;+Iv(6kZxS65q8&hUZ94mBXH6$-4unkG_AZ|`7VtqKFr8U#y^2D&$d@v^uyUrcsth*FCv|^Gq>I`^wqCDY76M z{z_La^!|kSeJQedv*6eE%sP>ZY)0wB(qWS*E2z37(Q_|x3H9ODkwdIuZ_++VS)%1a zS#b2aS}~Q=sMWqMK9O3uO=099f}RU0Z~EA|$lrhBalNy8&Jo7c&iy>?(A+0;nxl21 z?oA~qI*W(c7G#Xp5Mr4;a5;4F{<^X^^jIp|s}8t^Ojb`2H2-5Vp@!?DzY-p!xSr8L zF^0gMwV%m(o6}J}v&|6jJ~8 zZ$Hov8x6m|KXDvpC|owi^}e8Y9@M?waw?VGS5+JyZf6G@$1*>5yze`nhwI**6lGo| z!V(!we!%6$Zh~cHT&^1~Hv>XkZX53RE8f1{@cp+peE(+CsPFOm?T)waSKPkaA=3UV zpO!709g7}REgye}kysN?HVAW^c*fGb)b1O%q#kyVqK3~1VTyxKtW=@w87@8Ye7xhD znF%E`;;b0YBFgcP8Zd-zB$FWDh!~$pebmqFp^>y z3Oy_rI@L$h^B1)Mql+b;THbe)^X&)r=fhr)2cG`g_vi3O5j-f@CsB(RP+JsFD^?Fu zUY7;bmV&08<$VTXaQOjRK{snoYzJoDO_=%cY*=KO4uOH{w&~Lqm+KYR>kZfI4Y%7J zxBCsZ`yKcD9gQ3IWyiiZEN#cWGkPQRh0)iBzHN|x;Iggm9lYXxyWoDi;M>~`-@jdP zzu$4WTpaP*hI}&$oSdk+uS8@_8-0V6D#eP28_hw`aj;d#!^iaykd46oD+8-6fNmZD zYufPJ{T;Xa9q)g><9feiKMp5fAtie`Av_Kz_&u$VdU_#v^z^N9poQZgjybY4XBA%p zPn^YUnlO~`GiglEjA;~jk|FC8HB@$2^RzT8h{rT(03w3YcaXY{T+O2LagB!Y8E;krDTuiWdhmntUOjrZCHcrrfG2uc~%R#X8q zpC@cGPgkWJy`z;ZnAr7d=wOA-3{jy^p>#~-5(+<^;9_cZ&U-5dwH)T0v&4Gl{-x?8 zCLhwXVVS<AP<2?nnFcLr@Y3BOriOI#(Rb8K{n+~V}&+C zSBA>rDw9zXH(orWD*V2h5=m&?|K^F9`n(FP(|LgeS=?1Bd1N@79No8s>AtPMleGHpMidRZ z#55WhsVnx;Q z?06PD`TL(w{M(No__sgb@o#_rz`y-?$NS@n=YHT2F(Ynk#^Z170cm!~-L?hm*05bx z|GMDjMb!0v!S!ZAxwqQ|Z{Kd_k9xbhPwI-d`vtfAigjbO#X3MllJqqZK|l7SeI5HT zA%EjxXlQ3K4+wG6Uxv4Wk`AMVVy&gqm-06sixXS7+F%1n#ODl`C=3#wGc$r>Q@v5+ zgmYM@iR`5wN3F2Dq6ke@@1Vp4ky36&shA1BfE8mVUD2E(*hU)Exw#`KsnHJ7;6yD< zNd|v;^Nj}~U_pF2f!ZIQmGqAN`M@JykB4WD*?Y{yXfD|^-`whuk)>>|-lbZX>^7N} zy%ZV}zMKOjPC=tEM6*qF$`m9X${{5bS-h(inPMgnXv>0Y6i~N2uJb6mVj)n}{S;=N%5xhfNqH`-RQA84SrG+kh|rqahUpk6**xX7RstcZ ze7u?j(Ik6fmTYPw%Ds`~WWFSg*7qSM9)y~Io85AFc%=9wcN}oNKu0Ock_IP(*yL2x zDHOf1#iwkR@EVX;RB@j&B}@gWj*xP2^TVYp8KPKXU>LeF{7_RN6;ms5ADmdX8AN~w zOGg;#&k$38l|DtqE0D76MDEY7tNdr;(!u8JmKXG^=PY&5v8Tt3yFbr8Kwg?%?~oD9 zOxe1psbriJwF>&qQOc&7YO{Ck4np)lE@=k!XDZ3L+yN8B{PcXhyrV`5T z)0oMIP;ksOyf^`cRzNEey4tx1J%BsI-$FXwP=y+)CA{7tO<|ySLRupO{m`-kv_aMI z?nFaOY)wN`GLWBP&Ix7&DpR6S95IEIN~cYZXiVw={)>fvy<^aRz zKB+fMv>7>aWysng>jB+5k&Fzap9r1mnZkH?ysX*3(&S(!LEs|m%B&dNd z45Z}QnGU~b4zn3@Ve?B-8`Me0pkrEQS~P7Lc>EJ4QvdgV`EO`!2Ud28OC+kd<TZ~`?%>%3oey6pDUCzje3RA3 zUu0K-y6H{PQ~p%_4>lc$j1QXub7LPQ@D!jH-91Pjl&c^U(wWa+I!qWcu#i~*OF5_$ zKFpYZY1u6bGPYimFcOmz+L!8ibp z@X*@j(mYtO>)VBWuESwY_McNxyairBQAkQ+Lg2XN#qPt%owJGD4Ue}sy#M%t!zZ}M zap3uUIwsa0pcdM_qdl7!QBR*pS{Sn|M;2izS8=HvmA)#90smK5`>c=(5?*#%E0iX6 zbvYTD{SSkTRD)_Iz%%LF_s++{_TX0WseyQCFQ(Ar+Ab%TH;%&t4~+eU{GO!s6P7kb zaX$U}hzHO~dtW74@n=g+fs)i$Cj=$UKBR3L+LgdSd)Hr!uV^X}G>dmF&7AS;;UCC| zgX-Cp)7LDLO2Lv|DFw|iR68Iz7(^UT+}%kY->*PsDryPT9J0Qzd!$C5OAxtS6+v^4 zd-jvnOc$Q~L1QZKgiq?Z@A1;jHz%$rfvG=8hPPp&I+h7cQY0~gVMkSBO)XH3m>3;% zE23z`H~9A_Itr@v+Z`@{uP>1i4z7caslMlrZvT+KazTwpj>E79JAEw<)*b-e{XEb@ zI~#{#43BY2q^RKQ^FwNr!_PcmWbqc4S+ zU4XqgM&beW{8S%kwDI#LoJnJDIp9UaW8Z`PlcatB z=YRgMAlX3#T<%`Tyj^g+*(j8_jF*Wl&!+FdpMs*~H;OIKUW>%+GMX!>5=fgWv7S5I()i5i(?bp|F{x?-p{tH{9O5 zi27#JsPDi1hIMW3i5f4Hg^o#rs^WUxaKBygcDvzzzu|tnn0cC7gpKcEtXB!i_R6Nh3N5I&Hyv4e-iDr>d!-)dVtCxmdlKXCi;j^}g7MXIg#33`nj$h+In8+Jms0q{`D^%3TPBkRiP6Phaq#tCkJwx z`m$lJ6;zaVEu_3636iw`nuWpAtox|~YGj#w{t!Hz@?>!jCgnsvbSR?$#sM0ILOO8X zwe6f8MU|A%%}t)tz(U<)2H|1^JbM8ZsPo=u-&#O{xD@+dOeuwvtd)Xc6jbW|&MC~Y zJP*o_mZV4^H@_?RN!?xHbdvau!Jx_ZR{L56#pE&FL_JaNC_b+sfd*k0 zYK83Fb%jH<2<$&A(g%>okj7Kd{l<<`{Sn=G{ zysS+ax*pJN2et!pJ;#fqtJ_q@h^;L{@G5KsbnAm!e5S)ZZL z1&0~}{!p=kY*}H#3m%k7E^!)3#q>xGf1UvZ?f(?mgyR4Czy2HKc!C)C{@V?|{eA;A z#pPl?j@D>;__j{?@lH@SOzr#gfgeBK@#l|s^G3Zt@Z<47-z5)??CkwzvuV=pe!>0i zirejq@4wyf?fVVC|MBLYryK5XSKQw&xP7;;*Zajl0SluoJ~3;Qx7)}SQMdL4bqUaZ zn6z(q3DEn)14|pU84QB7(D>XO&fJQw8AJ74>H>0nustY?-$cd7)YfT`snt+2Bxs)4 z)Laov(mpNHj8RCiMu@@{DKpb~w)c;}nTO-hqaI;b$y$eTU}6BsB&r*0S0%vk=N z68=h#lsc&{JW{>y`8lRP%0UB9|7wNTcqp=akpe2Q1PO0*kJL6Yq%3zP9?qi^qVCxM zbwJ2veIwlTa5;%;>8SbN-=cv@;sL;meLro%x>+F=6RC^)qwYSDx_^7ax3@Q}+hWDj zv7>kUI>zr~B!9bJaKGR1cDv&3e#7l{1*sW%2kFoxkC}1xnY0g>L{oYASDlNg$Uo9y zevNC1oj}J-siCfiSOo%cRhsFH>-B-#?PfH-`#bjiz~k}2`}+^HWwD2bjDvW1sP^lZKRS&L7vQX$ptWNHMmj z$-Ey9GO|~?c%p!aI0ofBaKx_@*`5r(kJ7$MX|0mvA;TUXNG)`%ETo_xwyddsK24?K z03=H282_JbCRWJw-1kRAO+VWc34bvQg>w27d45WgA;?L=D?i8xp7lbTYJS`ZU+o0X z|6xEnnl=xCP|t)h^ZoeEC9CS>{HRC9?4)}!Ws}nYI}Ir*oazT1RI{nEQ@u0Yr=8O@ z${R2q<4JzQQGij|Q`Hk36q)6L5CmzB>V%Aqdc+ARHdoJdX6O84fvKQ5*g zRzYRAs^n+dL_Q%CWU|I>U2)k=&VIRU29sJ_dhyxMr01|~f`{sAg!Nh>Wssc3{Aaw3 zy`9E1TM+xP6;doVk&<upI^5br#YT4ZzC5ldmg%dO3AfP`N^vx!{A*EHy4yj3f-505F zkDZ^zTcv2Ks3t^1KyP4u|x6LO|7kq!a;qA8J zwk^1<%_dEZ%q$^sDwaL2EJr1$DkDHv`#|xEs2#ALJNkY=j&26x(2#n7VOtmQV&T2^ zvV%v;ETRzkv=%rg&}+eb${0;!KG&;G*e*<5xR)oz+!SU zfCj;jxuAAY1Gb?bP6^ZQQjEwLBf3d~QhQC+#OPp;ImY*wRzC#K=i%3Ahk0^>)MaY2K&9Lx}f1q@0!$mc+%SB|5!&+Il~dsL4K^Tztap3W59z0DI#v z9S3@yQbfL9DKzyg9m0$q56?xfbfz?d4-9%hF{qfniA-9@Lf{U|5K-}oRCg+4Eu_qo zMPi>@fGXJIE+PkRqTY!~f_1QZCv`5B$+%_tOI$99FR^jl~nud ztY(>mO=uc=*F$za5LfRMrm4m#qJCaT(Y)=~VyY??m@*_?onUGAU`^e5;{X6407*na zRGX3_a++DBg)>p|tN8QrS!FZN{kipXI}814&_M+tWD#nde{?|xV?^mMu%D|-!@>$y zLyk9fz=~6Gr<;Msv`TXdWoU1u#r=G3_=9zX(%MtL+aopJHyQ7nAmRX`A;Bi4q>=5& z9^2AfdbfCBC}_x_xJb*!F}WgT3Qdwj;;bcNLCYjmQu4`Qh=!!i`It_D&1|m1mK=@B zkh~3pJxD$K%fDwVo~xuO_tS))x10_t_(ZfEdvYyuDIp>D`zfMXzBrMx_pqM?5ZPsP zDk$guOD+YMLOINc^HQ!W>mB6kmTzonG z^eCW}JVn&i9e%y3!xxO|$o~IHDWt}o%*<#oX=1}_fwnkCR%&O}JlAQEX(RrlWxzy7m zdUOowBe(YNuA_H69+sZI?>ipP67c+ZS^@Rku^$FJIl7rrn{R{-fg|P&=xd8jg}9ZX zxh?4%3+apKke*BuAnNXE2{1KJlA0_sVEI78>{1CBF=#d1`LeFKT&~3vb;sMcH@tl_ zV&CO^}kDWIBD23wPR*-im#`Sv6qyn}oO%>J% zQl#7)R0P`XTC+k*Qf4X;_ff@U=luDoU7w+A=kXMXX+X%!;sKoxKt1_vvah10sEW~C z&6gTkTS0uX#`z%eVbb`Yyd-(Jn2WjUr-$!(sybD1#s-XFdi(LoR#`13Dh?-h>}KHrf)Oq@eX{?k#|nP8`Y zicb$H3aO4Rj7~svLS=O4EdaNJL-8Rl*`!ev7(;x?f-0naRLzGZ{=AV9Kr2x!k3G9$ z_0%bJnhClBZmyJB6id1^T=v#wOvyfYdxJ36$IhGnA#YOh3<^M?%rPP(EblQ&Glrlc z<9aY;(NuZ_wb@Bi=}YO;iZehIdLVJ{q$CZ`gOBVL(UeRl_Vsv~m(^snMCzFA(AUVx zY5Mo1eR4WAX*KGV+&_z8VO1>c^vv>!N9;EpWQEjX*@;VA#&h&({sWkSn zku)bmDjy{zv(H1qBgJUse)_b=lb7iM-h8i_ug^r-3zHfd11DLblZ&RvTUmr zQd^lwMOQCRII-hNsiAc4$EPe{(BArS07QNewN8h+X`PhkF**j4oF)Pu75zLVi7&$4 zCaH>dSy4OMnM6;eh^o(v65v~)7uB>_qb;Y26xGgA^*KoT9Z)1Tzf?#aG772Zj(0Dh z-rt{iyg%I+v*Y*dH7%i9{3DxLlv)-X@N`Gf{9;G*K*_Gl4yJKlKAoYYo&|%EKZfgU*`!fRJbv z^YE0d6^(`31zxucF4wC|ag0iMcxF~V4#RR=`uVj-<$aQo zJ;q>=9EN!J#%2T1C`38x`Ay~>nH@pw2K5ERM(I*sGNVcZ0WLJf*^TGxAT3teY&5_O z)Z*SIY5;c+$u)Jt(zSpl#0>5s&C@!K(-Ck`Gh&IovF3+H#w5+kBC257MeT4ZY7+wW zCRo6sLccU(C(f{RpA_We7Dk&*-v~zG<_Am;l3NfG>nN0I*6&g|l_!?GjiyQX4*?ME zzg$S!_1!;8JZljZA@=1kJ7k57)|Uhsg!AqIVf880x~|xoN%kY&!~j#P$@F3)<;Y@^ z5!HY`5lBHXt>;e)1^dR?qa-xKj(Nzl&3;9|xmpy|Ow;E5ZAwQxH}a+LVQ*$B5ZI7|hHAI-CoyXJ}pl8jY|fO~PGS6n=^XI-4$*BY3N zE@XAL2f+3Yf#A>%IS9v-JR2$|Qtpv*?QZi;Jo0X|BxRUCL4lFo#Zy^2d(TFe@WWc@pv3rNY!bSy=+PqvcBUm zViAcB192wId%?5}iC@cURLg^ONW?Yt;9}Gncy5;@iK=-3o}P5%_IC|LGv#KsUN&r( z3oe%{w#&u*P;d7#jrxYSZ*REWuDISVxNO##SeNE#_Az(fiM=a zLdv~Qa_o3M)-2GY)448-qy1g(BRzA_h|Gw*G$(;!iZU?@-xmw1LyFHPrcW~DVfvwd zpbufAgSVxLejWddGK^ivl-c$;l_Kq+~`R`>O zhjpr04YSm%XSG=|)y70h-M^{q@6pLRolq;9WRU|Q?#~^@$ghk7&i`JdZ*KU%suxT1 z9^{^_AsiDa-+LG%MBR!hN^&cMeHyW*)U`#< znE!oUnrE7tJ%b#h-jn2Kg>(1wVdMjf=TI4uS)9rWXw&js9QQA?DKeI3?~fHyn$xJ> z1f2!U{X^ksh17zzH7iuvCV!1wR`2NX@LrFvcF);4$<9+D3noJ^jyT^aqQuWqLSs7R zl3El{k%-kE(GX?EXfX7DQX$1;n9K|djhvO-pu8LJhZaK;_w+_B3q!FO8j|hLVlsK( z?RvxQa>M;*UG{ZZteS0vrMccz4oKf2$J4%cua3pS!H!6+jzVf7mt>8k@eDYtk`5^k z^yv5mKGBiOZE0CZwbhHLko9p)=*Ttv1$XWQq@>$mbjzqb$iftV97i6YS}mZ&H~JzC z86E&TMJY*obzG81 z!;0LVomfAFuzc`{1yH3;1>^X9w(U}I=NDY>cMIKpd&B+PkoH}+%{)#Q3k+VC#d9Yj z%fkpde0ote!Iw$thp7?e5g1^NVwuVm$h(smq$MJshNahP$i%Whp#cz6 zF?pE^nZGJJ^5h{^*~KqV02=-zC?>B`6jJGp6v;vc;g7+r*~cflNh0dY!+6ReE54E8 zm!cVNH$ezA&CrJ{C~6V)6NChQu0lS|;1x)zb!e1zgska|U^9)Yq26Aa%TN;)?_~*pkGTnL?xXC+# z0!c0%pH9k_Kw?rYnn3}j8;Zaag@8G!vDuaJf8u}gq0)q=|XB-%0I z4+afp=0wtGvOby!i$Iwa5s$K#zL0vV@b{o#tqsHga`vQbuKelsy32NWQa ze@gT(qNy$GU!5rZBZU;FK%n@X#5P9Bb;>RTGDGBf>3a zeCI($lHSovU<-M{6Gc-RzhkmOJa3p(aK?TR$|*z{|Dzc19giPB@O=Eh`;R~Ie7xiR{XLP_y+@aV#(}aF z)af1RGJ-#~_zJj9T`($v$~eYbA=)g|j+*(W*ri=pI2_#8R-*S_K1}o$M2lXdXD|kd z9=(ea1fx&1(V0XGqYPuzkSLL8VZ!KTf)LTWM32sKF8}Xb{`+}v_QigmwQkn4_g?Q> z>ru*E`hHb)9t=tEPs*Uk_T}I^W$NcNp6%ahmtuGfEYbsYk4vY+dsc-a-8UPm#e9MI zCQ?t4tUp}D1Cj+L$YtjIYze*&3z~>En4RD3XF!VTQ~SlC%tuNs$BvkKH0V{0y6J=D zamp2|tX~Zadz+t`OlFn1cBJ~l`_jtxh^Y=P#aE;IZe;ZB8O>ES-Vs1oE;s78s!qqN zi~vS~jeUF;gJZGrYvsU-*EZCe&Ic?;2R!084LU|~f9s@82U2U&9gMli61C5j{KIJ1 z)VyRM@Yhg6gRAv-TA-~06eIIg$0<_SQJmgO;;uA!8Uh&~Fw|nh1t@}HP-+%q5 zHC(5JVdgZSyB1*D8uD#5vS`y?qDGVL&2~CBrutUQsHkNN%4g*1t0QHm`BUqlj6v*u z5xsk*=yAOx=$LOOhhaDe_54X{sJ<`r81v!#pG1`1U-bZHXHYqWdPBmWcnKRqjO}&0 zwg~fSS+@|lLwcZs{ISsCw#Mgjc-hzTQw;`OIewdZAZCK*V^8NFrTj}djQ^5QS0RWd zE;aQ-^r4&ghx?l$CH^;-Jddfq8r{WtqMf!Fj~K3ac*5jGfC!_;>b!oB_)2{-`UnTB zcy6kCYAFj{l$FMZ{9#C{NmA)SMBz6Uv)k0lm-z55&Z0(3rBe^p!gx;lZvPHbmfW-1 zqnf^qotul@@#$V0;2jlmzY`9zV46khSTIK(GJp_SpU>$&tge32^(G7H@Oz;C7dBGm z6D8MpICtv?Hj{~IQbS5paaNYvv9S*K04^g9SR_7py|Py;Hv1>Nt@n0xpyo!d)s8bU zWgBB5AKiGZ3A3-JWN-Oh3k#W#UE8W3Rf!xMj~^s@qQ1A0YiDW8vIrb+U2&_|UpP`O z2a$&z@1;H@AampMq56Iu5(8c?d{S9K=}T$oZoFEu^o#<1d!u%%mUGUj;|=M6knRzt zpXw5j@ddbx(96LpRq6M5t+lKM(f7#1V?LcjYtNg?oYLK-Qz^uzp7V^Pxq?^>E70PW zC0F6CvJ?gF5LeutG=J_+W;(z1h9%*>cgGJg3uTT12A!<@aZeiTV?~GDi?U)W>1O6U z&Wd-i><8m(c9fhL)8#~62YDnvVr(e!o#WHHN&;n7uXyIJq4Ld|q!*zW-p4q6f9tk=`UC3rl(#$2&&JmKWRN;rKC*Blcr(SC3Mu zu(=XC_leh>iUHkdG2N~_R|ve(g=@0km;xQDiA_CfCryEXjmQAe3MV&+pubJUTIZj) zkh5hpsGhL1!39HvnP{Mhb&u%)969R#5u3nACU#PpHuSYK;J!SHa1_=jgPn$nua&$G z)$}uJKBVNMqEtbCiEuK`+NWQ0q5ULzg18fsT4a}WkH)=iHv-)`z;7$)O?p6r-+6q&QLQcc0v z!SasI;GPU$o^cZ2-me2U0s8Bx*zm0kv^(dH zxjL0egCr=jb6|DhC{{pa#lq3^+k^(s{Xdw0tZM;!v>6jK6i{5Bp>!mb7-+pB=sf@p zME-m-zqG2m| z7aE&s0gMMQfYGsufk%Lp_mDg*4GuQdjdtazvnGpZ!gEN4k~XXj-;3(}W*!F=&@<EW!aQ`5|@NaSBqMFe!CDgthp^;w5jh4S=ZX0NvGSigjJII$RhId zM$*OK3QojRimpZt4Q5>Xzf3m=A{lW^w%&RYYc3?(f-hy2NQsD5kaa$mH6d-*M!4UR z(qp}I!!Y5|h&H=N(^A$Y02;3gpnus11I^`b97&8j$L?oi^B{7e)bv)2)-;O1CZ&|n z&<6A+P|;#izIeg|{LkrPu}+#-o;ybx_T*1nT@M5W>)4Bz%Ap1)n}w({D-{VMLNQjz=L(9QOIO3F2VUPX$?=FdlipxTWRZ|t5F_@ zv{ZrYla(PMEeE4G?*%2Q8RD6~CDl@b-}%v;rByBU*=>^$wI6mT&gqISI=IqZ?O34{-5?fkj}ptZK@q$&Gx}pR`y- zJPSve;ra?(^}rQHL}VVk5_)J6(6qcgLaBc3)zNN@GQ}rHXFIZM z{C2Qiq&vq;AkOQ35hD{TzKc|a+}&^1HxBOa@UUx#!K!J6D*4yfV!YHRS9u zOXqfn?^#-+KglMFyZ6Ng*SKHCG}pM>j#-nOl)T2qB0v@3soW$F_#>`KW2fe5_FdKc z1~a5FUBk->JWkaIo|x=LjAJ4l-ICpNs8Z0Nw+g0c=0mtpTW&Sj^$f8RG5SD@I~t;Q z_pWYy7E7U?<~eHM{npPSueK7ojsbsGHVJnwBR+~R2aWP3!{eA?7lE(b zYMDToZg7D6-+e{}8>a$#e`_5#8Ja@QJRe;s0a1=Q)#lf&eO7BrQjH!7TH?R0bZwlQ zKCdZIuuyz;0=k76ig^*}p{HCYkeYpNHCex8SW_|8d%%fMv7IH?nKg1#c*VJPc1`_c ziX-o3-_uf|x;Qhk+yZj+^Mypf-EgG6+lzWtEd5dc@zR@6uThe9J2r>;sJ#Z{;;J^uGSqBB_%xte916CvVc878nr4k(@QAA2<1s? zZzC=FTVxWqjisc?BcM&^t&D^3Dt*+QB(k$_tBHT`5;|~+gS@f{Bio?LlW^j(l=3$U z?`&Mx^z6R~3xUGu9sbUM@SP?o(Mbu$%=9sN=|ZTCkgx{vEE1rwX{zwgf3)dsGMFYm zR-A=rrD){}?1f2nPGYmQWG$q|c3mF0L2V6=8Z);?;uzC;2FfFLQzf7jpK|=kNp*A> zj$izU-?r2}qO7{gHv)00u+VRpEGw$PuQx_GJ`!Z1xytF9jV!x_m|U4kkc8am)(vRY zjTz|rRT}!Z#XJf&E?l(hQoi&E;Ph?~uVMrTePdH6TOij&)^ll0+z1eF4uhEH;ozXoy(Jf>$`YU6{aA2){`4|tlQ zpLRb}$b`k~GMZiBF928%H+M~47cr4Ce#iQK)fk0%9;0Me*)LG5FS^Yw&GNEqtCsE| zjVF9c@AC1_v|JSpXHO*Fnrx5-oP{P3=n{hr?)h-A%9jQ3oUa~m&Q$nanAf3%QfpB*G&%fgvm*(~mUN$`iHRqk2jSf9fy zMpM=K3CEO2KU!>|-1sfT7RJV|>ZRU1F8ir(o_!|4bW!gz2>V<8X?P$%o5qaX)sF?x zsT^zkl5?K3*-X42l|S$e8&~T#UW!h43Lz)1pi4Qw@)Q&u$=imzqb>aD&rf9i3B>=c!3_?Ra=( zrT?7~kYM z2VL3-pdnVVF?!&%#oklak;d9PSF}~kVu~tfM=TkAS+JoyFYzd;EnjUrz_L9%F^dMw z$Lw!wZ|^ROvmYVL%GUMl@2&)vwaTztT#0HruA^cf*$Xh5icL(0impAa*rTS+kMi)3a3*q9>3G~TZauju*!_T=WZi;(qGBwnZfO0 z2WBsu%RU7A9M5=e24{|Wmcg6&J)|hm0Y&t~5No%&Ih;?$au5f1DuD3C5lBptbWH}N z2q%geNedFALJWlEgf4cQ23roQRD{8*o{Mj0)GFA%w9RTI3#U|8Nh?|~;@}eU-#!C@ z#!*SD8v#&PkA>G@C1a$~=sS-v^_6`{+mIs+=ob1*4RAyc@Brb_$@ zoBDuX>5I+zVVLt;<@(J)u}LIf@pAfw*6itt0uBir*K`={d=Bkbt>_mD zuw38A$wImbLRp?=Y8BnDJo_C_`L-OQ}M2f)?><-mYf}24y8ppEZs(2{sDfoFg9I zRb=9cIQ1$TacZ0Kw{ednwc-=GqNoeA-l&bgV=;TxzYiWP$RZJIi@K)Gs66>(DyThf z0vtZ;OBk>$yZw>+PiYh7qWXfsE`&~ia|7pq)<43ggA=?#x=hz$N z>tmT~D>X@{ozdH9n%lgQ4g71-*f$N>Wkx>5JW%A)D6Vt=)U9ZQ$ zsNUwzajhRW9^Q}IQMp|$i;WCsy^KQ??t4v-jCKG@J0Zh_XNmMzSQY6wvS|+%hd-r7 zFYzAgt}iZVXB{4}_qheLd^f!(QXzca2l@T;O*g6#i+D_PGi9uYxH&>uFXVm9D5y#s znhPzvw)P!axG@^(5Uo1JyIwt=MBUb)(h-q|a+557d+O);j=y|5VEkyP z_K#B7ua-^UtCXmX7ZjiPuYun#u^-ODHB#NSjO*HuihxmDS?3`SD>|b9oAozD0I4Sj z=j|5&bG_@C7;%}+5JMl_E56k_vt-H4Qp>)Y?Mb2HS8*I&H+;&}Uq8t0PWWxcQcL1| z`DElqhK-l09U7_SXN6ef+Sr+I~?T}$)8c*XUZxXo0rKs%uGKmHA zb3XJ>XQXZ24(#%}$d=fDzVgCr{!z{7aKp0k?j^&}pQxj*`5@Z|co&W5gOTk%&uVt! zr=Dgn_&du|U3q1U8ZblQFcz3ikIImuaH^l@V8J9EaL(q!ZyCjCeX$6)jQUdQry4(t_+b> zfk^)jH+mT!>v=$^R5VJ`mWT~YkLe&G9qqf!DI)L9`Pr4kw7clpsvMeSvo4JcMs5hx z!Ej?p`3kfg-&UnD@w)~#W!r*R%g*uYy@cQdYL&hK z;gG&%IcGUlRkDc$CRUaSl>`qMK+JS&r^HuKAyd}r;gYnI@JD|~ZotRC#%M5+LTPoI z^UCr%Ct4b0=^o;qc78xcl#bn(&sSt$Bg`6+MC`@T2do;|rSS9G@3#*P%T#gYoslx~ zVSPuNYcnWxot{B;r_9WCLmv}u@>WmK>(S2}JTX)mv9mcE)q!dsBZU6;$=#-ClazyL z;u_UA)pdIud5h^N|nHu*Q`gHbmU7?&ez~A#D4{c}c4A_EgLAsAGMq~2-?iKXa zdtDY(&P5i{qjG2ZWM_Z6O;l`@F&B!NC&m-PwB`$#Oi5iyKbfgXu4`K=#05s#rGWFy z8r%rzI3wrv>8W@jUhjO@UKeYK1nKFJ|4M0PWOk#tZ&3ke2Q|nS zd)mI@YCSX6%kr}YamfHKsiM22TF;^g;@2t9^=@`-F@@imN>vSQ^XBUjE_7+pf{;~s zxS*qBp(2TC@w)~y#y;+u&({&9MO8W!t+k6&o5o!oDEB_3>a;U0Z|48*< z^xK)EV(3`6GZ*?zz5AC~2R5)s=4M3^QMG%ry{&|N+v0?Z#TPp9>ANDZDt6Nu??-FB zj!{q3e@ddfJh@l~AZx?0;(P42*iih{aiY~vf%nE;ACWtNXu!OWF`Ef+$EU%wo=0Xn z(eH0APY^e2&u`WiT8J*K>QDZK&aMmN+by+-O|yd?pU~Rru>XsT*)S&X{~WTa*W7Nh z+xZvF_1dSr{ta^UzbV6NpK<@&iaVo3kpDw?$N&H1{Fkr)-QJ4U-2#TX+CEa_mhABE OmcF);R-LAO?0*10${;rY diff --git a/open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg b/open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg deleted file mode 100644 index 1e727a053..000000000 --- a/open-machine-learning-jupyter-book/llm/image/comparison_conv_rnn.svg +++ /dev/null @@ -1,1809 +0,0 @@ - - - - - - image/svg+xml - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - diff --git a/open-machine-learning-jupyter-book/llm/image/implicit-order.png b/open-machine-learning-jupyter-book/llm/image/implicit-order.png deleted file mode 100644 index e88de5d79fbdddfe581a3af53019d68e640d0d3a..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 34807 zcmdRWWmJ?=_wR!!n4p3nAl;3mba!`mHw+~WA|fE&-CfcQsRGgsA}ug#_{iFf&h_eSWp~5UQjg`Rpsjzq3dX&*-E@~j0hdAZ5t&#%03 zBR+(Ykr3uH5V|MTn70d74!6v-%g(5e9Q4ImX0#2ar+9A+&(M0j#ui5rc}V*4xrh{P zVI`{2f4<#h4De+8@1v#0i(j7q`!?DpnTs4e?ICGC{`$~=p7yZh)BmNX;cfDOn^Uc% zH9}j$%*a+(!`{%y#=UW=Elu7xd!%d@Bvr!!KLQsXya`L+Udw8Pow%$6Y9|`(2yNb_5*isEUESMGSBOKT#l-*Sc;rS z1&gc<4a#&`zZ^^}D1*!zD>9>7m>*v9`7NVDEPWnh)~-VzUWQi%yP~B~k*-X~)>D1t znCH5i47YU!cg$79vL^shzZ9E4Gg+9?IU1w!o$%?*F0$sRw5z8oat~Wo&tWlR7d1p@ zBzbl|=3!~6jV!S!?L+vyhk)a{UekT3k{ritC=~-rf0C5)H2jZ#&zoGN3{DxN?^DQ4 zyxS{zj^m{^ho42F1M29A{guJq6ZQK#<7{H%P_oslGFO^)604QvVKu>yP3A2^M~eEA zA<^wDZf=hjyqEvBV$wSEL+{QGaGSZ9jDJa?lJ7=77umciNc+qF&N(Si7$x9fW58^r zxqf?g#(&KNB6*pBn-M-C>~p|GT7?5;G7)9?{$YV0dbVySB~?j!#rTtbd?+g0VE-OB zCfd|)>vlaWgZ7W^{g0bG?BFtZ^3sln+=9gsL2yM|_yau&4FAieD)A3=4ql%St}+xW zdL8PpEgN~2*y2P>s|_Q9u_1GWl8Yi{vz++K#E%I}`Go~t#Voy;=wz2eCwuqEXkF7^ zvpu;LSv`nHg~JN@X4^lB#p~S?JPu<0+vhK^!iO3TZC78gN!)Z#;k>fkb<$X8MG&vlLIGoG88Z*AruzmHGo`xP02|Dy#){z2yRRMc zR@8+kvNo!dW@b3IuXIl@q2CAxGxW!VD)S_@9@ML22SkrBw_q8<)_Azqg0oAOwi# zk~w~%X-HP~0b1we#K3W0NyU#AC6m8p9SSZ?29*N?rYtEEk&&Fn`g)R4(akKBh@-F- zCYKeaM=~<^r_|AK?8wp4(>^=wzsn5 zJ!6B^DESp{VHN$|>J$`F;irtErz2-gy{W?~j@G84hBF(&x+YJWHQ*xZyEvnGXFC7c6^xcZ~w={xJQaa81Q) zIooON62^S+0J<}u^TqYMw))wc+q3}vX*9D<>&~(ue3#xTf&2tx+o<%t4Mh*QGNn#f z0I%8-=oLrT-AQ} z3BG1XoL5H;__J4CN&cz#S$0k?tKUU&>fla|Valxg=cd<|D~R8{(58Kvl)bPqGHUMV z?n1D-NoJeAECA*9PTLe0PXVUU=YFw0iBV#K$-1W$7V8_BS(~D&`6)2aZc27Yd+0{( z#rWW$KKW|~G_LdEjt=%iU8Je}QTPuR-h$gO6bs{?s)p@f3D>nGXY;F3k}4$88UAIX z%KD0L%<=K*9A9#5rJSRk*iX3_{qeAs&&c4OcI40G?)I3$Kj0>&jUp!fILa%>b#DFM z@7?!rqj=DT_#NNpty`m&y4bmMwOe&$6mReM-;>SyEq6rE7W~IN5BVSn?euy8({|iI z2B)&GRpWWfE1^wJ8|U84;|%u<4p!@D&b1^g>_7kL8fl4^j3T7I-95N)b=&y3ggGuzfyA?D=d*YYI#a#&Nl>*o0YuaMb!R3_ z-Nt^$|HD|Vwn=H+`OfD>l$KfvUdcu2A4*I(Pdl;OIX~%SmaM^GkVV=eB^Brbx zbYWkNzVvws#(kenn^BLB*k=RGeqxTlsU8`?M7c3jaa(WH{!9zjK7G^ssd=HSD0&1P zlA(~%VTnVOpDj>(@O#m(Oo}dD67ArYDv=5GevZ{AFPnsn19uL3s2d!qvWQVYGlOe&or~Uj-BMWbVUFRp`Z%X7)&&O! zcYPairYiovY9Hou!18U>rN4BPe2kWQ#lkZ$?2=KeDwhGj$5 zPna}?1}A*NW0%zb)^3Q)%;aaK?+IUR)SH40rBlAkR>?B9$}bmhu_!mvo42S{ni`st zGcQ(JBFu98yAuRuvT@H{`;KN!hIf9Djpb=)MQ^P|z3lDk&)?XXw;yR@t&}_|m-A9! zu4KPO3Ek6}Bw;P6%-@v>iL}(i<0A7;q~^m3?>6IMKA5Wi+6#_UTe&`Ey=7*y*ryt` z!R*QBFO3{-OVtc!uU3-!QR#z}gLz#ihngWP1c;^33{Uw|)F|7GryjLe*6B$<4daXt z=<#N8#($Kz+5apo_wk6qyVd@88$X8y@no4l=N98GIlR~3J>r;A_ilC#?%z9f-FmFj zw7X31W#T3?`+zwKoWT`VVJz+J1Co0CwNl3Q^z*NRtNl%X)>L3uMEq?zHAjE1M+~ZZ zYnc1Q`rK(|cGnt2#S!~YVqMy&Hc%p3DTDV&tJ~VJ3;=k$3 zH8(?gD)y8ju)Vzgp6XJ3n!5>Uja)WI?qqoE`}EK9&Tf6*o z3aGhW^Y7Oi&C1;Vkf0?)LiW=4HjpEDb4lG%kHT$DZ=8_nsb8kaMMHyscn2XLKd1r# z%(FKuv6MY%#Ns26xz=dM3$^mp7;1GU=!yNAy~&I-+4R>zYj8_Dd<#vkuhB%wGhkcX zEHy&eV=KF4R1Hff{JzIfP@ijr=M7cN><2RF`P5W5CLsxqX0lG>j^AFadW#U!Q8(Ca z>+34ZTuW4OLqlXIO29n1r0qcFJ)xE?zq+l-n0Eik!J&eiUyP6t%H498bS@GNaw6a^ zU;QOd*ywPS!limz=vbFML_iVL_|(o8ZSkC~b|v}nu^uq^`dNs8i1QsUA~zP!cx;jS~siLH&L z8^I;xJ7MgUaUT=}5mKIl?m+;jp^rTPdqu9g8k@#tUe!4aDEfR(f=A&8lS!^wZIXRR z#EUx5AhRbP1RP(~d_AvGN8JV0^E(@ehvpYH{dq7VsEGLrhZ^m66cs%-rxwOLsG-Ru z>{!|l{d6t8((W;FIwKjB+T@A$HxYLFP{6rLh0EeU%Y+Tw0rc(cuS4y{L~pnyx|xRi z#c~teJHIRG$BL{s7L~tLXg8!}s7Ge?{sF}KV!7t{Z|3offWT~l5SLS+ETc-sJycS? znot6oG`fvj+M7PkE0m6{l()Di(pg(s3k!U&l&sU=XuRjY4srizR?(2fbTD&nt z-{xS*=IjIo%%;DAi~@4piXv4g?L~NE4e_skpkLd)>Oe^iQV%&(QGHrbA#BoFisjQC z6cV;x?rYZR+MGPjq^LJ1)V?&BZ-vH^IY$M-l9Cx~qw5i?6LxS4G}PXI?F@FTim9})BOf6Hi;zdKhJH^DTGL}rxx|xfV#n6SuiN*Rl=_wH9!79tLrU1{ zes{*^KTI8LY=(|E>baQ-Y}lqQMc3x3BbfzU8t6{RAneRJ?e`8vG1j(ey)-+~Mj@C; zt#n3;Z-X4B>J< z5o7s?!f;fmu(~=o7F!7#Rnkznwf?4&vGcf$z7L)J2{UE47~j@1Z)zG3fkwrjSD1+0 zu}`JUjItM282nyACHh~34ZX2OtP-o87yruB-RV8Tk_`FW#6~&on6T6`0(Z(xb7{%p zEW0>bTp5UNM)BEjJh?NGu0PZ-%}do&JnzRunboolMP1sQ_-cI9+8JR9?9TFY;Q1UGU1^^i!SZ9 zEne_>)^(zNCM*-W>2&k}qzT5{Qj*-c?%#^@bi5(LoRV-=7lz^A?XNCLiq(8O%R*Zu z)$zY-xhvC_$iET_w>CT6cBHA6axL=6n(eoQjF!jRRx2}p5zfX`_1{Tbnaw_iN9=Cr5EN5f*Iur-9gKl3{>rdPE;RU?5s%oZIbI((UWxAZeaFSw0am5kCz3#Z zgN%UJ7YRM6Bvx?2wU-~?i7-8)(w&0iEGoV#k&~hz z=DBkgVYNnsgx%bZ8FBw9V@33l2RaR>jgMbLP{Da5)v!Lv`7#nYUh^Z)KXzpI8zf3k zKT=`~k=$(#*IfAX&wvPpcF931nT^JJdH&}@+oSl+Q+u>e8XCU>gHnB*sfHsfc%sk( zR>IMr45t@SIH)~WQ|HfQt7_m2wYIZaUV8+c-5{f@3>;?3II?2XJcx$yqgkkV%!lw< zb%gAVmE~o0z#qaQvD)PLmrn|oGIFTgJ}|cxICVKt5?;#Ax<7XZI36psf3@f18u48?Q@Y2s+4_D0WHjd+5SYSAU7h@?ikKK- zZa#ssGH#Pk;b30`Gjw7Qqd<`tqK@I$_3r7n`?_AoWa zdtMTBNEb?iQkwmvfe6D`m4;>gFH5b}Rhv~E6cZCCS?<_|{1;w1FB=0U%6`hqs={#+ z34>6w$+ql%M(Ur4RSO*=CPsRTU%2x(+Y$!w%>8;>8dc)P3(+WEUv>Mp=M`!4%lf@D zX20BhYs)FRY?_K6*TQU-e|e?~bQ~aDDJ>nJH6H%AH${XanhI1S`M+M_j!xjo6(Z&D zf#UR5;99kQty2@^H=p><@gbyEq~AU>khC{7FgX{}XEKG0t0};*h~emNyee{bkUMej zzjehpC-FROaTLlbas_pS0K7IDLmL86#nwowZ7V+EXf}6x&5|--^a$}2>hU9nKTX1m&kh|L9|30Gyu`R~5Zlr4aXB$(q5&Kqc}!Foh|*@W7V2+!Vm-wA78AKlU4)_xGj$^5 zLQZAz(iVHeJxMnPwNy{qu_&LOv$>d5 zO)pO^CxC?Lf(Jdr)e%Dp;?qi7 z*hX7eQOD!i)`?2p*U;;)j9z*jUX1x|*#aGbS$*)^jwiT;9@n?7x7gK9r43{yXI?KU zk%BBN(K;N${GV-hupb!V&J3CMkEeI;K-qMsCJ9~SkxyA3@z-UgbJ@#5bd41~UCtF` zovY~2O3VC|!9wfCO8Kv~IK|y<$UAuo$Ikl-s}9v=m5z8X4wml`+&{e}Q|8`4GM~6M zDu+kt>$5g6qGrn+5a)??Sj|!Zoz}MT8s^w;_x7~F{6Y0b6D_UVgW%I{@pvUG#2>4% z!U+AW%!+2WgtkoCP%+BJ{luk+jl#mPCLeNiZ}&N^b$jV-m%CW;k{G#<=+^eIzo@SNWoh{qG4&`x*O>xEswz@7uPH-oL4H<1Qjbc zoSqI2MPy{`<{`OHwn#6@>S75R>4_=t1(Xzp%xx~)IQUS01nE2c$>{^}cqub8#|>`t zd%$nu4U7!Ov=~28$XGKs>vz!lZy@UCc_~6P-PJwQHPu~d8Kq%-=T(o+%*%D&J2pGd zCv0AjQ(qsPRi%HB36wA0%a^t1MA%#p8LsE*v}>Zj@G~I(4md2WS(t_HY(w9y-?Qtk z^W(G2P?RbtEbFxo%gfr*Fq-2Nq&2;(ovW|i!4q~6oo0e&?91K=-l^Xn^uVpRK9tfW zhL(iJ1-WTyPA`wecJS9(g~0mupPZx*`L>!fJ4cstIeWMf^UAz(jf zrM6APZ$mq5U5=lYSLpbkkf!_?7-r;iiA+M)R5m4>Ti9%`uk)D{+iyhA{`e7J)!|4} z_0?*ka8J|FX6!#jFyfm2KaHakjnOsl1|VlZ=fq3b=>J7y>HkSL>;Kz>o++#ik;QCI zmZ@K*-kf5Vz?2)dydne665M6udk}gv z27)6-%Jm4*mzs*NkAB3gb;T$|@v{5pPbvzEO`hn+&N;V$_Z~;lAChDs=UmT|F<43Mx^pt)D%6M(lrGZQv&F>FG(x>#o=4b=op1BThEf zL+n4FCmGe>FQeacq@%4pvpJRrPG%uF#b~cgbjhZpyxi{MaG8M9&T3F@vEZ@F79)?) z>rJPb;NajQ#VnkM_htIu@$jHMcyM`j#g@{q;uDrzD3=nOn(DPah{wjp#(^rOs3_>V zsbpZVSZg_A(C&kWg;i@i&GbwI%JH9T@iotT)&%gd|g9%Eo&z#!uL4SsNW zykXenBs{%b?7lvjlAivC@&WsHY9t}=+0mNp%~mRvTyk`Le4WDrkBiIk+CUQ5aymx= zj5jSU4gLPVFD+eN`Yd%jyWRbTmXQqJ=??!pst9Folj_${Jn^boz1AKkK06H=S?;5a za#}YVK?a7wk&&ow6WXkV_9=l9uOt7{wnpwgUH5=&*9}Bp9102w5fRZ-w4lhIzP`|t z*xTD%PEJm<+{VW9{sj7o@p0@So2#?E_4W0?fB)h?a5~+d0?R;vj~`nwn91KVw)1h( zw)4}o&h`AR&JfHLx%Dr2Z14FpD@~PiG^-5cr#Mi)f%oCi@p|GaFOuUl-{cHA?f$N& z%7P=kPPf-~cIYp|KP(P|(=aeFpfEOJd97L~7aAUp6)FbXh)YcT)$C>~-{|5(&%kg$ zAj|(&JFmm0dKUB1BU)No>!Xx1Md`h(t1Dh!-nd!IYAx_u<(i}|@6TwbQ6rMbLIUr+8pX7YPWG`Ki9 zeVug2c^n)X8ah!ZPyPUhNss1g0}l`H=g*%40s>(TE^Nv(X|yA0ZU$Ex8XDjhc^O<; zj$|HVtuO!YSpYW=4{Gn3nOaNpJ3Z?9DvK%l{gu$#hZ=84PF8N$ z1@Erod@fEV6bqQV@CgVoiTK`khrNQnt^NhOmHM(r>PF;zNpF zwxdCaHya*HzCCL&T{@aQ4(Vyv!#~2Rs;cngY5Qq2d;2!WB>|$S^PQ>=&Lx$}n8ZX0 zfs@TKQ5)V`t@1yA{$%ia#QHnT)D-0AifoTyzImr)Xej7;WF|fbPhqd&n(*9HOlBeF z<<%Xr^*{Xc9YHHW6opGjSV3KXr=i(~;d?e?IW;vU^LP~f^QTXr5+BQ$Hjkvi_4W<@ zZyrGI?(W~ee}@g+M?;In6&FiQ zllJChDO=~))u~;6dr0Tl?X6i-TjlBH_3w%+D?wQ}lf@L9-bjD+235H}YR z@q39$NH{FD>rZ$)rQz_m2IcPA+1TJee=a0&cRV5(Yxwu_`tGXZ4mH5I^Am^V7Z`gX z*F?xHF`LCx$PT=$(L1+)wy~k%SL3YzT$us&g-?6`+u%*quD&>GT`D@pO##Pt|2u&h zYjlh^#Wc&=jc(S~B?#AQtyZ@k4d?<9r-(gtceFMjkF%pIE|`|#Cs7b19*(ocbDzIp zy}iNj2KnvVx8seGr6%V*xTwvS^L$&K8Z&7LiJ#k3Wp1xg+i1zi$zLVpD5`=wl2#D( zMY?)=UdB?&iin7$GK;|RC(^b=_s+K_EWwrrRolFPwe<79r^?PF6rClCls@-l^L11$ zQpmtgkXBHjBF<+r$S*3gG&5sK8C-1hrbo^J0Ob^pcTC%vuA-)-+~Rp`&CSj2^*TWU z27|f!X{$Pxz^kgY%E8Tq0x|`;ZZcivDV8q|{O@!P3@YOwF|x5X?{nze@87?tr!^f? z`uqDoe7JWs)7K|`L_Zz|z-oHXSjN`Ou@JL<#Eez&=0tULE3;T(Zv%nYRNw1BRjdtQ=lkY=6r|5!$`-f} zTMec_q3i`Zc6Rp3>1kaptztbZZEbC3n*@Ndd*6^-(Q7qNYBjlo16mEr?a>mGjorT9tVq*#Djx_6jkeSbXL=q8M(ts zD&S5yI50u)xVdNM<^~o57)R3{-Twecf3q);mXB|Ka8Ry=zuMd;E4SFO<0jF^-NR$L z@{Ie9g0OI)6a~!H)z!tNth~IuwDjoYB%bc)>fgstp6u-GB%HWK@I`Z+M`A*kmzRhm zO(CHGo~vLz{51*$w7j~SXXTvQ;0X)(-C$1uw<+>)y^tdlSI#GllfvdVm&U{=#>Vbi zgHd**@_TbaV3}plsxaQXd;aQ`_;g)fo@p1sKoWBjogAI@+3pOpVY^(*hU}xr^6kyl!u&i$ zLP3%5xYQ0{hm@351#>Q}du1q%tI=U$c4Wj6>?(fmGc(qdz1jM?Ht$9)hQe_0fE{=o zHqApuCbA#an~<3)@cz(ZzEfXrCk-F@@w)hjHl z2~f3?l05(ez}=S~u4rm{58f-**F;Mz8AtgAsNy_{2tb87IXQ7u^4qiZWuc*crRrrz zM@O5i04Yz7kAbWq$1?^@q@yznP}APduC=XAddLB+JIV86_~^>Y3g9ByOpF-42PHTh z*5ef8>ELn!W3;uk`TF`+l$Xb|7XeVlqE`6$>C-(Su#3>2;oparmzQC44WJ-BzkhhJ zzrSym0N%U(=1ez7NlJM#hFlWtrTO{!-MziOWjmc>78z;5 zngY-l$gD)Kb^Scwi1#O(FKG4j^()pNH?RN@0ma+jkS{*XZqoG-(yFzH*xoyG6c!eS zqBAoqw}{8POy2qi>%(vpK?QVnc1ouJO(W~)wl~WGCG&bz#)^Ly%e}ce_t>9L`EK;_ zJHIMoJM}pv9jSOTe+HtWQ?$IOz?2JXX0EX9$j%n(z#Udwpyg?Fz2jstu;(b<{jdvyk+>|X#iSEoC;FJ2g1S-r->BG-OLNm&3c4+t`4 zW#up=4`ZnTjPKc=yALACZ2C>{iim_+@Vl`IO-2=i3_;HRum#(m%#pAA@qT zAtFhZEA%5G0cH!3=SMpkK|w+I1F*l-Gcu%8B%_F=eF<=IewtM0(}lkCI@#O+0_$=k zE3V@i2FC66vzcl;DS`}c7mF76JqAWb!`Ao0%4+83l-y2d`wP0IK4AM^>es*jYrkZ(r0;DM!1_37nE2~OP8D-Kq^k-|L=tK-w`N57UESHX+zOA)Y3Y~gp z!Rg(bm!cz?{C#o)mk2CY^UI?({D{fkWhXs7JzzdcREsu40eX{NyZRn>#knaElfd99wbM~F&u(-CDefG>T0BY{9x91BdN28 z4<0-Kg`Mv?h6U6%1_Uf!F8Pax_YvtDN=ngYMMBZU^|KMEiYjI$g^D1JNMREub1X#1 zoc(Hys=p2_sTmm_qp%G|8mB$k5*_h(<(TVhADGJ=fJVxq@NSbDTA*BPY;Cu;0B0g@ zuos%$2zXqJ5hIW|M_DXSnb461;A*rW(o0>#F%j@>-wC+W`Y;}AWq>Q&jWi|diF3)I z94P5EX|Nj{7Vx2b2_b6v;e`c#R-I-)^EC8=#dZmD)IaHh{stL8W@ctwKy?)+;IUbZ zc1TQmto0|H7uk{>KnDO@j{<1JynuQIC2)JO8iho9h2{1>u#J6-jf-2QQ*Sd_tPCVM z-G*pfO9mjef6x&Zhe<3jU#ZVKZUy!g_0C1#h5jp0pc60cvag!)o)kd(CaqZ!{xVhrtx zbg?5ShmSf(Q*SKz0WG4nJr(^kPT@eSHpR>889Sx^(NPvACWZa1xaAu%GO~DP@f*&^ z5YPk(=?YO_X}E8|GUf|&v9n9z3=a&D2T%FjB`VR>)zsVz*xa=0+cMjvhYk-9v#G}w zsq|al$7D*~QMLHrwNDNb#Bs$Dm#DyL#g(ZQ(xK4S6gzb2Atq+gfK()$=5j9H%4|{Y zUTuAS0)K?0VN zyDh6Y#O3XrPbdl|=-2UF;YZA~m#>#i;&VS80Ge`2J1jrT4=H`gw^UHM^(_z;`h!5^ z^BINZZq7Z!WzZ5*o=jxaO_rk)R|f0h8U!#ewEuEc&A}2asM>u`r*El~9q?K|4N{*1 z`|-`qjc{RXCO_t@C=YdImJs6;OwlJRkg2x8U1^F9Oq5t z4a`cc=dmQQF9qD-W9+PBbt@je?{iX2w2O-i8f}~kZI_oaovky&!wO^Vc`nU9Of2N& zd4+|adPdiW45yywYgOl;EEQ4>?szGxm)bd31)ML@bhNCFJ&6QJbd<9qvd zY-T29UPU5gV}Cz0DJdx{YpLE=GeSe-5sY%Ptg4-rlXHKn4A;~71$2Q#qTRngN-)2) zwDj$p5Dm6WyN&1LHrmm(7*ZMfv47b~O@h_$ z`r_o|1Oz-l%O&)gxo?hsUtGKeB@m9oSXxr@=;_mR4qF`l!g?NTd$8KZ78VwkmeX(- zOJv2m$|jM-n2R-lJRr45CVITQynqs-qM{m3vtd?_zyfaxngVuZacK^4PVrvOk}P*4D5f=)6Rorh&n z{_Vy!Y-(zkq8f7%O!kjE7iLSKKsily_$ zNES75?}QP%t*z}0x8q!cJxGa$hlfq7FF=L>L^wb(@P**wy0o5C6SzL?1S+2o=-A(1 zjh~`{q>c#0cR+`(ettm?1gtRb@#*QB(>sxzQQ)JRn|VOp=I4LxTpq1bvYEyG!TPt@7p05TKuLx%ORl8O`a(=lKmP5ehKyl~l?w(lTfs32; zpE`)Ape&QiicUnzY~RR%Q`N0D5eISruv%DH7%Lmwz`(#3Rnc{Z;x-7ZBX&g+NDgN0 z@xZ0b!(9-^SEC#;F}htQ)mjW10!?8-L7iZm`&}%D02l3lI;Dwt*9=?h2XUuDGK+9u zfGddD zM@3PSvy8?7C2K#|zzBpuE)8T*}OhKCJ+ z`vMC7r06W|kVHjSw+tqbx(Q$=j69P3*qg;#b@ zm1*MAIw5*rxLJs2C8=-3TRJ%G16;T|**XT&p4jgKX!5|G)wT6?&I#FAfT=+KvjHc6 zc<6-m@U*Zftg5O)&~5;S*VNL&to4)){TEpTlJ~@B!%;h&1VFX-mL|HcF`^F;bA13NYjp=YjOJ zSV-x$D|t+2W;=*~z&7sc?vB7?`!rAX1|J{!9i8|M*w}eVMST9Z={}!7e^$DyvO0-I zgmKnO_}a;yS;$xZtsgOB!PmYrOJ_Ybb=aRp4{+H4V%TGU+R#6P-8vSrv%k8s@-NBM zhJhjlwXx9Q?+@&H+_~G3{F*S>$i)+}97EFi7|-R?;-ZNek)8pgGnW=bQIbD(xL8<5 z_y36b<=8kMj2$j_ayiV?oWnX~%#=i-;;O2odCYnt%;w4?RwZDGqKNpxh5t)XPPf%r z1tJSI%C(Fg9NKNB=n^QwNdPBKrI;z#T-)5-3|5ZKa=2WpI=8tQ2{sw^YWp5As>A8r z&>`?VRfZh~;nSa?Z}qm*|58o`wUbAW#5bVLEO>+`N`(6-Ok%%3-%vP4o=I&jv~Y3T zcdq{R@$pIJVP|xP>y0CR+ns?Ls>^spv5iXt5&)7L}AR^ z+DYI9dR+t$6sN_SN$Bb66TF>4m=7)#B^$ITG%NMsPRl=l@ge4W4@xPfsvT*wJy`-6 z0uYAHeo&+15*0D>Y_ClD+sEMG2bbB!ib@?(_REWMObfr-tEwEyZ94!QGBGio?=MWm z%`Sn80MLkHw9*xd5Cx{+#AGGO(9aW?ZeThY^_wS`ma+!cfbZG`D$tJe>iY5+i0<*+ zGdHc%QCIXAjIh|B#3a~+JOlmxLVKuUPbRb%<R|V1XprY#~^uD znXcWe{R%w?@kE{^CTo5{IcN`YG)@5&-Ct~z6(kjQxW;@8KFL8p*W^s@IgIvCx;=yd z3~^>avOy^X~F8=@m}#x`v)! z`LH?*{AU!gVoe=boj!Zc)6-KxQjF;;pfeNLs%<}0W3J+xYZk`~L-YdYu!WLlMmsaz zd+-Jw9o=n3T2WEc^;DY>pD-EZ*?n7yK+vxoUuG|2DyUom_5l(ysk+)3oSv_5TTagM z?o6#WSPVec+uJs$6Bu5H41JbACM6dK(?RI4`6cg200)whl?7=BAeARTvu`z=wYYBO z!f5+#1!-uCL44qEbJw`!M_x6T?a!vGs|ym>4PbT@6cj)NLQg-iv~>3)_zB{M7KHL~ zK6KdZTRW~=r`bP+DZmdwVz_2Y2=<%M6pDza!aW(x46|v~V`%J$t9usOjh9CW?dHJ-Dar(f$z#Es;f~g}_nwlK-Br)Q|HP*gDyo00EzxrN&9Pr{jDR>D6Af zY0vfbH83TB#CkVlLBAnECiT*%9U=NXma90%TZ?jx8>nPZc@q>)_@1gMLVzt8cVGv1Qn-B+Y z#6uJWvJ_LD$*Z=xu)t`rv$Hco6aZUYrnH$EMb=1Bk3v!zXvjc9g3oj`G=45SZ7!+b zzx>WP_h)nGN35b65Kg9`{{Zqm#8hhR8t6H8T&#p^*IJZd`GC$HE~CzU2=oWCmwrmB zsjFA$x17Z(3Xq!W!<(D$33+ zA5Vc3j`#G_?jNGC-?Otk*4>MX26<@+u-2hwEGj`6edcCnVp+2vxr(0ly;6t1e*GG) zy8t5;^h)qX5um{KF?Sbe9-n$?O;q=ZdzxK}`BP63M!U@F79;d5S zB_*VS(?AP@j+QVL$f5yhNw?d5aj*n7q)3wEZo8}s3P_hSWld4cAjqRZpl~7Uy76vY`y~WAchun*g=~4PnK_V<1*<< z4LgmdeePL}DYh6+msVC27Y_yv^Hk**j~+dekdSzawhA^)P1Sim`nRgeNP$zDr#2%AJuW^L(|d-Jok$9#Kpw{ z$(dGbf&TQ^*cj-g0&*ia%n=H~TQV`QgnETH4HT$JTU#$5gM$rCW9k(`jA z4oE0V(4R?r_s{nr3VSg8akH5hB_b*+Dk1_}4#2zhGU`ExdyPvnw|xLRv>9;X`rfb@ z1!QC_01bp_N2>hF6%PkZsDB#K)|LQ-!N37L!lHWr{=Jx_q-^^ko1mco^~GN~s<@}P z|9Z{L%vGm5;1gdfAm|Bor3aLhc#ZY)Ro}Rj+G&>Nym)F88rc>OjPTqvm5 z)hNN?p`n6-bOB$T8nCb_pz8qSoGP~Xba_I;s}k*h;bJOHLMET5h(hx@@LHfb3VgV0 z_~S>9-hvF8i;LQRZge!7y$BqT_4QNAp%QR6z#THH^TfoA0+jnHI2d&CO70pcKL=j2#Ng1qmX^Uz+qPbMJ}d9R6EC>PpbG0a9xTq|#tu0$ghOYs+WyX=k)643#FOaoMV!>bwzq{Zv9stPAiVXcd4|kHdE@ z6UiMqi6UaIprWQGhLHhkRwYD%76Fl7+DcZj(8!}jj&Wg&g9NNFKmQuU(FkrsaUe9~ z;^Lk?eL4#Y=i|ZO(yFSF{(d?pCX3n`FsDLy(4v{2lN0!bM6@!X*S9T2iULfssDn`~qM9=a!53j)t z+q~3ORDM|0^7He*ijba{>oRe+v9Xbs4h11RV4z974rzHX#sV}AxTe|$kXJX?{{kUZ zd{UC7rKJm855S(Gp&{~92v}B7nCCCk`h6otHW~r!XJ`KbJQx!bGt_dA=LpA59*oCa zgDnRP^Z8nmA?MV%?l^!}z$yIBD_vk}hUP%--5UiBjYQC#g~4KZzrk}C%j@dmM>bkO zqBNTZPCll~9GDrH=(zpfllvcl(gG>k!QKK^!lWMz#_Bo16W(14U>zpxLOU1s2@e(y~awut;Gl3XDsERfdr}7M7w1fb9=rN1~{W_4UG2)uU|$#o+0ZeUbCIw0 zqL={T-Hahds^5N|FD;kM64+<aAehRu-=Nb`SyQhm?2*zC50&(=Su8@QkP=_y`r*1=Z>CZ_$Rj*fzY zH82O_Mkt$ieco{gTKvfUmV*-)-klfPnu`1}feE z_L#WO_|tO{%^6tQ6Cjurw8w>g5mW4u6C_u_8=6-a&k zfRAo%<-3_5Mp^^Ec`xt}I0daMDaBSs^v@ooj%WeuWf}k;0a{Nm>q6aM@+4$rCO61aU?r7VQ<-yPW|!K$ zx%i(S=svT(TR3cZL$Vmt3Z_7|z(MZ?8iBO@eIA<#vJeERAXNo3RQw+M>0qo5{7MgQ zV5ERU4GRq&E35o>svv690d)a-vN;EA5==}ERBa}o9CEOn4p z2f<0jne)s6CMVtBw_rSXJiqv#U%68M#RDPm41Br+4AG{(`oI2*SpJWrt5OG`Po6-p3F7Ue z%WE86nxug}5buKEVa7yOY{(dN!2mqv(#U{8y6XJ=e9$-m&g0)iD9Cb7&(8c#w-YK+ zg&nJNbIH)rLmJH+IY!-}holsHVB!E!rlOP-_~3zUbubJ*5Bh|lhrwY%Wm28o*LD1& zncZ!BG9VxziP>=fKbB&XC%y3d8Y~3#>K*3ad3$^Bt&hqvvXPUMgIS%ry2+_2^%m}E zwXKX2+WZI9)I|W3qVPHBSXi7uYpxkX^1;6kTgm@fA4*H(a@2q|1bjCF-VVHvq?p*p zkdUUpc@@yhBd$dhGdWvLZ2TR{r2{_tr@-_nvl z=*WUM1CwDlfF(;irNEmPU5k~2z5s9$X=!uW!a>(JH?Lj|3=I+Tz5fc3BkUDD`!gXy z-F5KPJ0ST%dI*Lxz_1g@^nv-*M-O(j26>TeB14Lxe|uC^RIINs@xN(mn|qm=ng6oK ztdXS}^8t{yLq>pM!T(f6P1~TWNKX%R)$Np~AqP7L&L zg^ZFdqa#^S8QGDjtc=J`iX z|7xnS5x)Vh9dlrRv9&7Uma_l(?+0@Bf^}V)Zi^L8!~~ z$N15QBM^bx7?bHlW_EdMs4m)TQ>4m^P{1b?8SH0O_JtgYWsGQ8oEh9i*b&RY&Ms_W z0}Bs&-spE|8(B4hz@%U+0O3Fib-appyq+OE*7E^w26M4ntm;QQaVv$D{Y^E+SPQ=M z^??^!XM?%FjW?RitnqEIHZ6ch?re(lIyjLaQlR3ke}*0o>MDx{9>b6A?d~Wvi1& zUq!gqy7cbdivdxG=I~$8?bcnaS=3IraE!ry_{B!6u)CbgW+KZR@f2NZ6E$@z#(2&y( zj(*fb0~%$e(+Da^V>wX&@?}SD)Znh=_a@_~jXcnEN$OWVpgw_{d=jal!gY8(NpLVu z#4wUa86syVmz1oGOuXeCOkc6w!j~)kev1C_Mc4ds`dY?c?o~AGFkH*HYpqR9a1u_w zFU^Trmnw1%Ai1Kn^j@Q|u&C%m)K416sGxjsf@pH+=s7 z3nRA6tBaPfe8OIqDJ&`|C@3tf)$TRh+1VKqvH?#7BO{E6*}1*YA7QJ{@!DO6<_+-* zbRzYpO&AyoK~_TQ01`E500M=>!`L`JCU+@d6_=#T+7qn?Jmz?sjlW(eR7Ed;i|-*BvGRI6R+k-n;<~ zeeL>nqo7rs;_b%U3gOIDRUL%YbWGn5xM6hEZbvLA2Q3Jp;t3k?N7urLtFIqfTN{K& zviai9hX83fP7g~<>t4LLbonw%`QdNNH1B_X0CoZTFDi=sDuZhQZv3JD)@n(L@%iYe z9t#VJnTA&-;8sEc%vu3-JF8J&#ekGgsKM-?ZKpUpYkz^sdtPOA__e{{Z4yKWpBnjCW`T+H2(m@s`34>o}LxVjDMv^ zO1xv!7Iqt3TPauU9ZpV`+KYjbwhUF50fpX5h ze9SaipC!Hb-GC=Ri-#Szf`7oYCg8E1wnugV{E|ojpi^Rosi~y4iRy z0v=m9e{nW`|N3so73ZQ4z|4g&>o5p_h>n^BoSPIYiB!gz9bQ(50#Ac|M>q zz>*_&{As0i?FCGQ=n*%=!=uG+$Ha6%t%6*hn=7oC|IaAdZhtKz0*6lAjP1mU6P`wJ zGaE3!L@Oj9rJ~mMz`^dV+|T}6u@cj-pd}-#A^xtV#jzz$#KU68%|?r=^EgN+jvRSg>HZ1BA6_cIze2N8%!@Jjluj8>o;yF1@PLs2^zAn z#3PpCrSl?n-H%fQ!m6&5(-b_O%}x|6CXpnO7!lOn(?hW|zaFm7F_U^`37~iE80-0g zBPJ%Q?jsut$%hVYA!})BhAnjL{7@TYjp!frHo8h~nmDS68~Kedb2c zAJD++Uce1@0ca9@+0v`}<{1FtbSrZH0A>ETMR{Glsd(=5p^-smX67SKw%?%1+JmM= z=wj(HrUY(zHv-8H#vtPp4YuxR(JK3=;X9{=g6LY=)G7&d)wmkIPA^RDWvyLkl#d)c>Se4hUm0D2-MDlBYMp+(1!&-yak+LJ*C zopyHqFw9q2T->Yhz`-wq(Vl!OF})}-722r3^yC=LVGWutm-fO z8XCre&A4_==hUek5+GBe->I_rnlxpUmv`0Gsq5(#LJ=^Jv=Ym`chEx&0fnKlPTaeD zJpDo^K_(8?1`YQ1-g<6xWbhePFrU0UJBeZ5;ZdWYJ*ex#a;7sLOnvyr& zVe6)ewV0?|1WHU`IT@NfNHB~8jA2bSQn&Ffw$;#VwH(I0?jez`wZ3`sHfFn*V`Vez zAbI{h1mE_V-prB7y}#6UO&TYGW6MF95WU!Vq{{YRTc4>#$8o=5d3m`dU;|Us<{|+4 zBdwP&U2^kOORwOiPC zVYvQ+#6LN!1Yh z)`X)U-L%=o^3%>(S#b}*aG2e4;L z`q@Ya%6Ju^tHJ3D4GM-mu=$gdlOQrH%F6K9*J-NOf=pYv70V3wd<3MJdnsAFVmmK~ zF9T**g=Y>>8eR^#v_XstzS{t*Z6x=^9k0g3^@XRCZr>i>%5(;PQE=&FU%m*0U_?6; z_No05%v2ILr@qY0xH>p!i$7PtqNoL@Mr5rzo;Nb56yVN5;vLnX)V`9jbC~;{4v&p# zh`YMFHgfv#5J;W9y;tROixVZjrWwsDb^_&0nyL{Uz}_t{8))9>EP(`Y$k@0GKa2$Z zow@OLfv-nGc9QZQn2MzgA(NU3yJTGKrtz7X!QtOWm{96N7(iuEO;l|+Wg`-v1>IH& z^h0I`4i5afQg%%Y7+d}#z$VP+3g!efJ#?2-sJ#?*-k!xLcmDP4&tp0mpejfCVLOVq#)Y zi8&cVkC~~dO@z{wW;;3#Dd?_YqQ?_=2Q1BE%mU$s);8GH>V;B+2e-1EG$1i8w~4*w%wjje2#I-5hoZtG3JJd>(?nvf|*!1vI&Tb0_#GD z9DGjTjQ>4^`fW=Z#{X2K-{0Y6FxL+`3KSR}@K| zdh7O(+9F0W&fg%SKEPR(Zlt{K&_A))Zrs>P&V@$P>J+Vb<*+ud(=wET7R^|NPp^Ydxe zB!9@Nxl$8`)=r)}wE!}u5PTi3c&~Q`045h*PTg)c!Oxwz7V(u5kz-;qkJp!sYjx!_ zYLq-XFSXsMf$_5SIDfR(W(B1UYu~j)lW2nN>O_TwpG9skdXRODF+#rNz1HL&Xhm5f z@1Tx6ONu=r{0fx@GUA&zU)MHB#Z-e?0&X5FuIcY`qD7dkKU!b*SsW`p7@czQL#hed1NJ+~Es+qVs7w$z)>&XEpxVOe64LbJPny zu_1?uF-b;t98Fc5&Bz^aUOwYQ`2=HBR^96unYgbd>`>1`Q!$Mq3zSSoMt3l6_yvIX zfXusFTMH^Hm)h2|=t{w|k5mw@A#OT%;t-Z~Yinal#>2xegJ&9V3_1%o<@293;fg$lOe?^C}jyYrXfS-9f7XJ7PstA8!e+q!wAyd+}+`*49=6b?Dc; zJ-9yDU{MXFaStFZgEiASs~W9UhNgg{rE&0}f#y~hu`0CYCr{4yR}?C}l$zs=?lKq^ z`hZlAe}0QlP(zJc_3m%{RMi{=0^-kY@V~;;j}#`w^Z-H(uEMA&2H9lp?b{vg>_%s1 z)_(R_>(*>)u-)VSGOOl2%O;Bo!8JO@u^VEo!z{g;IN@t{CT+LimDgO4r)it~c z8VD}9uD15@oDV4cmbSJADolI+U|cc9qaAAl{v8cOi(={26%`f5oH#WxAyO_rYgC+_ zeJZS|8G@Fg$0#w9@Zb7 zvhK*~=eyoTQ_-G*$~enQfjEe<=kmO~V~e^UQ#xvJ8=1-1WzHC?09@MS1}SqkSX@Zz ztz#oLbkHJAMM}!bz&vGNS|D#n>qNnR*hl3}Z|0zxdniSI4!<)fp+F2v(2JJ>E}tf)ADS~+=-;EKy>?m4JAx|YF8wB;sIdDn}`7xsEx z&C3%Rh)PMD5X;=#WSw;7*IXlSNg#!!Y593LbGC|KNT}h}tNWPHb#GE(Y?HKbHq#cBTZLQ zM|XGjP-m?mszT$-ry#Mk3!_ldaCe7&o20nS6mAIc^MiVfWZ;jG0fN%nmOQD&&A}nU zQ5NF(>exyVN=-G~L`kH8l_m4;)6V~S0p>uL5YwTcBpFHj@0O`HgCCPC@4?Q?VtKQT zy~vGpray!00%*SZ)vNmH9t@g*R%g$iO_g=o9#)5>#}fGPtc4)$tFtY7wR|C%>1Q<= zyS&M)onasP`tm=pFp?=hb-^h{vfw_tc#x4nL2Kwq4R7*n_IeS`au?yV)h$GbAlqi4 z9tlp{oc^XJtvtFG7?2tUy?}>yI(U0}V)~RV*AM$+81B9C!U1?D%I3lCpIG$#z`@Mb ziVj_`opkqu=?-5S%)Ua4m@6Vu)mT8F7R6rkB4an^@f zS)q#b5i?2b4H)np97Hx>|A=`^1pHVw3ng7J?iy+kX(T^DOvtaUzN4k_;#>5%gXsmc z)`KNqDDf$!xbsLs!o~)E?73>m`K!gC!UM761cSdwrDb@fW^B`#&PDDIV`0Y8&UZUI zjgs09AA1wmC%%n|=|SFaxU6CX#^83NOvp>F!X1YW5H##8o_LB(`+(btR+!HLHMcqQahA!+f!-&V@-V+p4mC+`5g+JRwW{aZe7nm~=(5HqTT9IUn*7aH(4I?g-a-2sNeEkUcp zYoPdrgd-Q14&F-k|K+_dZ63nH@?i7F^aem~CS@7EA! zmN0v`;V{MQ5P#}HeFaDJm|pLysMCw1tnvl4CF@Qar%rA+J$f|QPCj;soV2R}qWSD6 zbNsGtZ#)1vkZZpd5P}kSy!OJ!w{IWh<~Du&c;`uEO-6<-+za67>)ut4@?Jf4`{hxA zw}Xy)=g-S}b+OerF|lP}1r9cXY!`l_?&h~!6eKr}Q?FmnT_dFWwKP^Inwih}`7OiK zV*l6yk#B%oCa0_so6zFARY{f%x&HqCUGDPKK{hrU^;u)KLM?E$x7R2XC^#KMQuEq~ zUwaAK1S1(dk~|GfdT#DGZW#u|Yx2~8Z&~wmC@~2@+ZZ<3c3sj9qYnvUzL(s2-_T~gvXLHrvBSmrTTWOEQfR^zEf z_nGxKn9R*9EEM8(Sc*3VG_%V3v*YCZ4<7=6;E3sz+s683C#(kGIf>~YxD8Z!6rC*< zitK%8j#^k$l}5PxUp@QgJ{93vRWs`gPTyTitqr-nS z8Sz-k;^pQR_maHX`2EL^14JwdfLmn^t_n!?FP- z8Qg(8Qzg8efPOJ~+TPxaSa1OrVPyT-7ux#Al?-Iu;E3>W7_BH*a;^waEw-{9t~WSt z+D^gOgh8nEy|b9?z}Tlx>Uz5=TX|B9-aLOk3=N5*#G$g+tjy~wDJeBe`+GE@a=x1a z46KDx1F_TD`Tph0mti?W7QpQecpS`f0wF(s{`>?XxTOU^6x|$_A7sl*Lg1cyZj+*L z?%<>q=q9B0qN=7DzN){l*^OZdjmpBnASODx1PX$H@ewNn@_lt`u9yO#|`HdYN zMJu#{mirL2(-LAJ&c4Hrj31!N0r>DO-Iv_7`}+Gq8|=_e$z7@w{E8MzQA?I<1q%xc zaSYl_tOGUL*hjv8eF_H?+!dl1c&PM|Q9Qjfht7cy#T_2_7nH652P!_lR~a~C6(o7--VOejG0MN$xk^|iEo@LM9D10R?TTt7A!^vJ+% z$95t36oh(+856TPGSQD+CFh z^oX987D88L(Ap}P)Uv;?NG=8Pz%vX>^hDxaS&8H}{lkYPVH{gi5gdA(a)+sS=ZUK< zrlDI21T_H#g~_W8Ghpev_(Dy1UZ5tfv~#y`zyWPpk&_?r@Fdy503fMxGhFi}8a6y8_tRFw7=KJa84M-(Gkq9>-H%UQLOE2+Wf1nE?`j@%dv^{K&G+vDtuL5E8b5&* zglB)JDuoQXee0HwpWg*!kvbJoQ^B|bWMUFdPk-Lo>4~T!=xuNb_}^iB*g}00?E@ZxH+#7+2vS0FMs_176@De2cOy4`^m+ zfDq;^C3@gt)iOSM)E)E%xC&_38);dOWIX})#pMIQMy zL<&5rQrUv6JRkZD-urJdAzDk{b6JbL%A-le`i6KE+j^FXAv%-Op*jL)9A_U={9s@XOA9V|1(t*-;sRggD#?8rt_<6{`QF99HC)c;`AG!;u-z+hi~E{_N& z_2ZASIl(=})FHyFE!yUsoGE=d=Ka2z)6B z03mS1T|>`E#%aK?jDIvWESy0@MH%9&L-97JPo07%+fy{z-K0HzV9QerX*oGCU$miQ z4W`8(`WSC0fP_c&g5?CuqmMVZ00HCfg780l*cA&kFi-pSKh)Z^lDv*EE)J{jtk1s5 z5hlWsB=}OWx&i9WzPvvYObiTni;H>L*cK}`uDh73CdNt$bHW1?9s1Oj_U+$)EN1#Z z4U_~-S}o!)sR*K`AGgrZ`0aYfW<`cViPNz2bDXlf_vhwAIcKjzmJ3qOi_!yId_2kP5L}$2`+G;U;4y+XR z`wJmmZruby6z})m`8-TC{Rta=Oc1`^uHcaguP1C7GlvwiShIzD~?w+kn2H?mDIomab#heArBtBQd_GkPSSdG=`^(xr`VQpmlZsmh&uq>d%3y! z^gs0EzzBkl^L=qK2oz#}!dTF=aIIY53Buh{oD-CQdoaBU3XZz8cI;sYqa$1#(87}m zgoH&T11GQD*Aje!+69cGoozp4MWIl;ySrze_+Jr39v3D^5MkrSI*#}?v4{_oqY~$4 zb^;M^LE&7yyaIP#5vQQXzpuA@hH}>Z@6X)pDaY&Y%X7>B@f8$}=I=|b@#4Uie}5$W zKmX#{Vm=`udz30vNmq*~eFWi`Vt!-F9TSP z{)hndBXQlGC>*a*CkIBiP?E{4-tqegZ^xALG0jrQO>ls4B3I3peGHlGXZ`zYWk$hp z$Hc_{g*WNB&G^(5mIgL8H#=Yu2Cwzd5vx2dz`%-7$1-NXD$U+$vM}?t0~xP*#z5$x zj`a2Q;R^v_u}o%hFsN3ke*LPDr_;^J{zo%dz(O$zV=J6hwO`lEm)QlcKXCuMozgGZ@?;t=@BJI;vEy@25 z+gpFe@B^rDL=ue9a9{Tw$7U4l+j;-u1r-U)@t&j2n}?$HAO!LX0Kc&@M|1Ss2Izx) zd}I!i#vO-QuiizY2)CDNuS(0bF#?Iq_I_dZZu0eUXD4giMTA^2L_eq@?8?Vb*X!6bSl; z^+F@=XWg$@eQbAwZynuxp;svi?_R%_P*GWRq2=H9?`>02G>+HkMd%nBzC--)Dg$ml z1r}s=i|pD3I9k__s+u{CI0Z7Pa3wzHZ0ff{xj|FZe0&m)gI;{Bg!$5<8z}O5?yiFI z@=GH6y1Ee^+z)9_NZkg!I;^ez7{XJ3f`%f)VM;|9SHOW9hYM+lN&xfjml|lS0y4*_ z$yY<n-mfm!Vkdm4{oyYi!0>$q9Oz;D8a_AfIrJJiLEmD z+tKLL9?0_p{(u_CCZ~y>;+WglgIJY zfZx3npR<#Ja$(gDr{B9*ig41#U1^G5G;7Al-fB38VgLlO&e{`x{d(p!-n9;>ZY*Zg zbZ-d?kGZZcZBHJoRhU_@Y!u!OtPcz6;adeEM^48IHV-^20Mww+AO=IB1PzQWJL|rM z@J9;^3f5x?>+ADFp#j=7I(qar*5qN((h}tFLUkzd7}J?t&d1cN_@@nYkV6h~AQUu8 zS=~wLr5WsCp5%{0bjjh-J*J%y7ssXS*NL|M;RDyMAXxV@j9VROV zwxJ=6!Ar1l8l=^Vg^*A9_$Mt*yJ{<$K{%Lz~1lS4@}6 z{s~4?+1+vUVReXV1B^kxBh)2~pFWK?XTF6&U%hu6#dCk*0J``&+&FE7lO+-e6{7fv^{Hh7&_HZeLwL zdLhP1JVbF`Vsj2P@FBUF{9mm&`Qr?ToIv)Hwr^n$ScQ`hwlwSxvb-(thxEfBN(in# zj9BQ7f=p01w~#WCyRo)DtUQc~4z~d}{OXHDjD~FmFntJgfq5Po9MT1py%Gms%-hk- zi8lN&nLJYvm?NvbrV`Zn?(*oCrBK{QK0c#gzC;|z0F%Xa>&=G`DcLVx{?Ej6eWvC& z?0Ix-I9rUQ%&e?@4JbKEiqSSFUF|WcocH#u7oa^__BehkE{YeLQ|0)y-!(fUr z2M5Oz3zPg-ks1hji}+j9Zf!k1K5cIwP3BsOL`Jc^Zg?4e z)k6^XX@fKO2k6u(C&Lr!Y<8b9SE*6TzqW%l0&bfUS#>4)RJp~UKjY7di^`wni)hHW zzz99$hlT`B9G<$84~NBFoO@SoAu9xe17C6Wc+b8>QC)pVLI&sfn3LuEhnxn=nJp>p3aohaXqSXEnxNzU20$; z)g1QRHMn&@eFH&;n+QYcIeD%Crjajw#_Sv%ZMXIasHmj2jsYvIX9Ih68p%A!$=Q?l zac%1FOH6Uzte<mh5T>B?MF(s@6z1n=d_$T9yyKxA?|OSH>%KG4Z(@zey6TL3 z1-OdcKTr!nEmIbmCnlP>yYE@Z8^_1g?puORi67wF21YjG28#;kX86*v^NyPLYe_fw z;YQAwboxD++hUS7n1y-DUHg%H5aAs&2wlg7)yaIvSy5UN}KVmok6_>}EE zkp+(Nb&%UqS#p2FrTY&-K5x-V8s8s7p;04Q*`x}SDVOy##^FYo6&JEmN$iLhx`Lqk?w?<)gR zT0Jo6(g{(zRq}YJiPOz>{}tT+{@)n0@N-}lAAk$u3FuBROWg<$cW`l0hCqd%S?H$U z{OIQH6s#VeZF%wf(&=j7eFK~=#YB`C)4jK(>{|eSm_6yD?I4Ql^qavefA!`=rl_Bh z^)~mp1nIQHBA$QY?GnD_i?}{fV6*2tLqk>LYlHTkZ#lAxtp#vNtq&8pX1MYQVr;-bMQ8wT;z=T^RVuLfeY9prd>J>k^h0Z3kI7vk3M63DcR3f zwW>=_;hLpto4(SQr`S)*+rq2D^~E{>D+fDi(Bgr13^qP{u4Y%rexd5WUk_x+FDOIE z_+1L}cx*Y64GQuV0HCC#WFHB^Krz9L?AO!Grh1a1qLb6_U3hMWC7ut2kRZ5@lnoES z>)=W$V@?BYIhdg)VJLGWD|}n__f!sHhzGaD~MI zpKCDJxW!CP79RxRcn8a}%~~6@jP^5$Q7N^Ho@FY+`=X$;U>DI)B?o8&lOmaL8=bQ3 z$Vj^5nLglO^LSNJ+KD9m`tvUD^oD+Cofnlz)P40*C*@lh?*%ntR%T{YMYabA8zdL@Op`1qshY9LYp76d^n6!z(B@LkpE5n9>D3##DoHT(2isWvn^cmyGf+| zdV1n}_P|SdPg-NQs3>KLDaHzQ=9_d$9La(`&s1cgci7dW%Ev}EF~I-=s)9yE`Z_*u z0z^Ze5GD=@Vc{c&hL}GLC>N}y{Bx3w6DW%TP)^X~zKgZIeq9srONx93?WXh0Ssw;% z>STc%MOqJBwpXgh=o?#D^!N5^eKX=-5H=^qp_2{ik4;bO(^6vx05jTr#MGhlq!8@F z$%|g+&*-}IBMqwGY_T22cMLp_-zqUCanuZM+Zd%Qi#MF zfhF1{`nYaaq59#;L8%N_p2%7I^(N?n`Zv&E-Nn`v z!2blz?J3OW*4As7=mGywaB)~r4q#I@CCerw284BLm+7&vSN7)1^An3LM+TyN&;$$SNuFPIG+B0iun3H<^ z@L&0X@OBscuG=m@N9J=ka)N-unDAs|y1Khh?PiB|qm#l=#pT>zq36+qpa&?-=1|xh z0TI8$0E&Kf^SR9#2M35$Fiz8LmnPD>L4pG>hp`NlD*zj|39)l>I-NOl3kK+IC|u@J zS8>)(TAjw55J`APbH}qKrd$YSOwx18V%h#X(WZ`M2y$U{=}CL@ZZ=`L7v_SdCgPW$ z*c}3xhEoPN5KnA#XM1}Igz%D*J#umsG7|pX)QYE1xo8jZ`L-)*0FAdbV(Q<4eIduh zMMbYoesG3!6&7fGdIao5)R07XLkA=q9c&Kk&WntWin2L%s{6x-6#yHlba`2PU;*08g9q$b!IGapi9Ul}E%=}Wv=rb^SZNVP zK=j(-Mn4G#hD<}SB~cbpUypJ}vQ0YR+^FBVk82LzCRmK)Sn=^23|I^?9XF~>t^VOo zbP0vXA96t=2oW?43{ZdW|Cry=z(^AcYIhCsP4LrOwr<5Xk_QLTuZ|fT%V76ObHET> zT41bET1-q9FVm|;y((d1W;U|2qJVbYYFkudlr|I)&@#A{PiA?r^gMX$D_kho7|t2o z8{89IV*FkDMie4JCb;@<-`>s12`gP3p08b9#ttj39gU!Rx#+@L$F{SxQ$DcgLeD;) zVtkAT)>MeGA{zr1dIV0fy|S{D&rPx`tDHWgr@=tnfEX<(YRYlLOGf5faFK2nG&o7g z!^Xy5O;Myf9v7qEVE&y8#H07Vudn5rOAqb{wl;$C6A~2k?8>TQ{l}69i2*q-E-ntj zLj;4%K!sQfpEXu41TlfA#4ZK&eV`hUU)}wznu`Az;x$u*PM~Eq*p`e$l=-pXe_i9j zhbn+&`1#`psEPEH@|sQKS+L-!uA8K;DgohnVS?Lv)G!KoqegoK6=xV1Kx?S764`%w@aHi;JtD79br}I zI;Y%gcpmFtOqE~!pXRZRhX>OAFjU%)6YHy3aUUEjXrYC)<7$Bm!#Erj6$LG%@qhoo z1U)1+HI=f4qgFkM^0^ln*qROK2y6QQ>VP{wPfqGEsV|(O*tY+^E=I0XgA>o3R<#mvdUSJ|2A$Mb-A--YkybVOWA$#d4 z$s&FIMZJ&s0B``QgOHHwlCdAHP!XRo-Ia~JkdR*X{(B)0x|Q1^Aw8FZ=0|3z zfx18Eo^Co%s*px_;zu^QPX51PTs6U{faP;5k|+co7j=N8H4vNSh(GcK2ne2*26&E} zK1j3ue`6k$0zm!W5$Uf51pVJ&pd|)t{ojZ|S0E(*zp)8Qf!q?*)y@9z<*1QOcnXB8 zFZw8P_*z;RcQS}@UhWo12T}jM{A}6mfy<9~(oIe4>QYhgKjUntP`O7!z6!(}8)m

    1H-)#H+H9YU4d@SE;Rn3=iOy`$ce(>0f9%tw zuz6C?$3@H%IpGA4VD@uyBnFf`ZS2`__|FG5(xaw$A6VqXiHl_*&{gZd|Cq=o)Ornk z2s){qu8MVBIqcSi2it>KzB^yKeA_PwjTr1d!uPs4z7}}}UyHFR5`R7h^m7@54Tg(s zcgaw0Z{#G7!S;y;x!-f>>03~uPK>b_Q5h1vc^uiO$cPj87dMt%V%AnFE4*hAn?mr< z?u+7RDm4bsY&8~Q zk70x09wi~Qk}hxofEcBhjLk;s^1nU-#xlN7(5@!$tt{bxx8>=1|GfzciHpSlZp|>N zoiaO_hx|8ZlA1ioAZCP$kiYSl8MSK5YxclEzG#5lsr$=Mzf}puy%>91t`+dE*nMiGgE$iCeo|CfD=cK$4IOnX+7wDeJ~NT&%6zoc0I# z2|1<~4>RhT^GBBHm^k|Lh=#m^#KvcOpi)mKxiW zFQ8D5R$h$)w|5&mMp)YEZ>+}$wx8kl#{yY$a#bez$f^3KYMVW=1>k*LY+h@^NU|#z zU9CI(0;vYP4WyF-s8LSSge51vhD*fV5UhUYa|z_VzkS`|X0>S}`33?nQ||eI=7n>$ z6vtQ+t0ksF(g1~EMcVE#>MGZM*WQfzeP|Il01LK{H79mc77E`Nur>#6yoeTOr2gbH zLZDZl00anp3?18aV7m4lfeAsmt{l;O(^Y4+|J9h+I+Fb`8C2hazSno1&kxe<^(?mv z)o)9oBG>uk6P1x#1o97kX@+%UT*PZ1Q@MgF4S2#rqm?#OuueaqpZg8}@qCVbiYGV; zqFksi8+?BKK(~zQ1SLg%;76sHh}%l>uGRe9c#XgB^><&MnSb(C+@)CDz%@saK_<&) z{F-{(3bsnK)?NTlC-9yC&BK+GGy!@T;At1!TH3pwkh=N6qIts^{CEGua%%%Ktr5e= zFBU&?sG|V~+g!A9z&1a=58RRGeRvbqaDt8YDaHbCWJseMimq0{_x^X`tE?&5o48nl z)O~HCk#e97DW#dVeGE1e`d|U%I%+LQ9VlEVZ74Cch%{byIk=8`>Km=~$<)kOM0q+R z*m62ilm)f#*BWtDbv}Ipl+}(jdt0ZQYF^*cHGA1r?gd)J#P@Fx>?q1*7R)b+H3ubW zMTN-qwvv>TBKgy&In?p0BycOs+zX(d|2DALoattsLfxCH_&whd{6$NOdULVQ%cDoi zY2oguete?Xi|H^+zlrALvkuQv>}$>Da|{B=W=|c--A#6lcDc+;!28?=C{Fd=N#oYp z<9w#d>y@ZxwOgRz8tSagLfU(e;D^Sso|Ki#em_4_tj=w>H*LzU=b|JbczK=2f%C~B zbRJxt=+WBrZX%BVo3;3!%erS*k0Qjb zLr=BwAx{p%-TKLEIe*PvG*Bjc^%ZxvJ-OR>Es z^b@hUOq*4;X!UXUPNd2JFR#E-Fhd&~kxeTxqOHAl@B1`5QDvOaHLQSdsf4fmGMq}U zt`Do)Bx$euu)wGepP=`Fv7nhATcICL`rWFTg|_)Wr&%ITkq}}k-Px@|7_8{sC2mR@ zCpjnRpb{Y`v17RiSw);q?xbTgdRvCS-tVG*$15^%thWI%+2MCI%yB!QLwjMZCVL}i zbK!1*#tF0Gbe_-}RIpf`TcOsff%SZs@6 zXqG%wXRo=Rtgl{KY>~Xv91?_X9jWo% zXUsMMF{x^9J$NpU>w7gq02u39l70M#+~KRei7zjC;!c$X6hwJo0J!=OAIF$azg4Vy z%3UYW&C&ptQ+p@L;BZ4xY)?he${X%?KRcE1JE^W;J)E*CAm9U59Se7YfeHrftW-fG zFf6Ngt(SVk!R4IXzVO^4&{B5%xAB#KlDBzbf7B)a1m%&VtThnCXK6l)dzhxQpmV>i zU|8qOpjW9!ox3=zaC}fY*#NTZzG;df@(+y^M2mfj)GkzGV86 zk@m-B|G+>t;XgR!|G^&_c6P`sy>+u)?YQ=((X__2gH5(gv&dvH>hAAh4PTJ8n|QH? z=qC=6_@3Kt2I^IDJoaa`D4~kyv^eiPQzM|0SffLj*0&lJn2c)AeuS#~ng_ZL@R??R zx~xyCj(RT`NBzJ+p2UN?b>;DK8gIV!JuJbE9=tbQT>&Kc%}F-6oByoYD(!^IzY%~# zH}R`zX+DIkkk@0jLoN2vk=~mn3DiU@FTrrOK_K+S1mygXKNEkIGRIf0`L3|pFC73n z(7EQs)Zs_##Olv`Pg+g$l2soJOXb+6n`-ScM2EYmkdC}?8B)NZnjlRdCK*5}{IE?{ zQsDga(cj(Xv)HOO@o1(QZL)lURW8;Y^H0qz$ubFRS>D)jS{N{ab`jk6`K#p^jDIMD znI>bDmIF?!F!9C6r(=CHM4BB@9}QxJ?|S#37WfceTB6A55C&4sGJIz6e8@_Sj*T!I zzhVSz^U_JjzXUBp^c?}|6E%!J>rZaw(wW;YwfP?bWzA5=PtVBxTp02PL+GT1KR|mhbb6@#lZ<_TeNgDO*5SDDjrm}% zCi&^%mWhJ{OX|ao{A7_ro!v;roC>uV=^&RyhveJ2uZc@d=~U_A;o&5_GK3vZJDlFm zy6D`L2L1C^x!m>p*3Xa>vu*f3(fi*09u_dhFoxayE*u0V+;i0fvo+>CW6OT;T*98-vuMhB;zG^W(5X`CYBRTYWn?~^d7ml+bzd6U9 zl!Ryo*44oFmYo`s!;UcA!EQ9Xa@zGg=%^!l!o$0p?9Zq)Hvzieuk_sJi@rvY2tVHT`APh_(wp8|-)Bi~uy;>#OHdD`J6 zSY^tsQX-t~L>~>S5!WAjdA#_Y2=HWZwKw@f)NK`#olV7K+8jENNVPRC{ha0!7!-t9 zeC(^kIlN?Ash;oe=CmsHxxs8|EJUxo_bdZ8H1gq#)z}&u8bj%vWWn%z2h%qH$<{X@ z&_?PYHWUJser8LzTtkm8T_Q-gwkdzXl5}7nU~dFQ;YPMoC5d6e{0W0|w9}MpovsKF zNyG2jbH*Gb!=%6%m#<>EK29@kTm0Pi$QYwy8Khb4g~!?x!zZd zj)xnnk47>2*dcShF(g}B;tn}7i5$%5+nE-_myF<;vqmVuSn~)h8(Y*8nRf!H`GH>R ze}k`AX=+M{VyvQZWJ0V&s2ro*C@ZDehw9;qg&61`cC!8tY{+Y zK6M@c2A?_4V-`IIfBqu{pMr^rBpSFU>`*a{7*KHM7%vTniQtfp3+=kz#dYC``YSLDNWj}gx3RHdwemJBJ|5e1_ZN5Oc)oP9 zTj!HLAtse$P?o6Xat(@H6ix~+WTD<+J!gQBxn_Z@-RpXQv&wYIlKIJJ$>FU!tG#C$9< zKM9p5ya-q36Mh#xOWIVisQ`~Wtd@!qUg?#ud;|PBE{)N{PiBusneAF^kf}J7!W}WH z(ud^QX!ZB^=TwNKz}ZLr_#-31p|rB^kdyvShQV4^%v^BgsCOYqDV_~>md8L^R=EujzZ;gtPYObw>jisAKndXI-C)o- zz2J@bfaTivLC-ryoPpDs;wfTW6(epO1tq82vvOqPr%P}+F&ZybPCgJbVQS8tOmvG` zuR=@Dye;jt^O;oqTiGC!!>2!b>)(0?(Wh5d^*7&LtE zZ%^yp5p`a>8pJ%NhQWl`q+Bu?Tt+2SdY^=xpa_xWKWYf=mJ<+!k38sg_(w2SaARYY zmHThh3<2bX-WGm^>?zI4xPSAAI6wt~A7!V6cZYaDgpf`1@7CJ;B7%NzLl!bzrH@qO z!?{@)dUk~b9MrhewrzDhd|PV?enwj7aP9LNwzw1jqA^Q~XVtfwJnHsXNMsDs#9kQ! z({f&~v7bHpt#v(|bPw)2N@==uiBot7-Y)u*1L`J~OjG_6rhDh{MwB9pno87@F2KR8 zE0MIXP6>x^xQ#R!(#a*%zesNb4mD-1cKs|R*~UMP=&e&&s;Oel-M6=y-V{UV=|i5= z)ds}>Jj`vHH#kg>HXkWA7*D{uPKUGNU1_N3S!R$^tAJr4y2e~@^|_g2wQrMf?bLds zwzs(cE)jK){D%syoY1F1-AV;KokKYE)^t!K#>Yd&Q;Bx@c+jUUD2_ zj98yf$St^jq^Ki6bsw%ZMr%weI_U^KrFnUNYUkpuhwYVD7BsYKSJ$0A}LyS2_2Y)UueNA!XYkC}>r; zSTPY+LeDTd8y*LT!^?F)CpoLu3qMpF*2$hQIRnG$aC)O~<-XbeEV-Z~C5=eJ z8_O28IazMG+sSt03%qj&(>EvY+?0S*RRqok=|U{eH~Xi0KqLKG{5C*sa5_i&@@HiP z3XrA#NB4OP>Eu`WIWBi?`b9(o!i!MwZk|XsdkNG$Xh7wJ$4zx8qJWX{E%R`Dl?(ohO7L?FIq!hZjOxe;ekd zTMTt=a(tS55W z7fw5e>!wOn!rD#jyKP%JBT5HxPq_{ywJ`DCd*g`L+oc^V;vKWSfVPZM;!m?&!A8~s zFNH}*sUd|=OG7?r0)j&=a+R%~?LL>37mqEY^v-2T)#epqI`Lb+P8M_u*xMP`tmH@P zl_=LjjoWLfCDA9FofzRD&op+qxA#rL6-SD67)2}iL4SGGR?@NUL=2QV)P1yl9XNB& zNkx~iyF_Oa!l-YrX74;K9c`GvADh%ctHtbX4CAuHyV@BUVmzb29Kt)lBR1KR}{w!XC$;k`fgGx~2D zyDmPFO@$sUK7QyAzthdt_74}S{1|$BPQsS80!zB(N8?%iayJyOyvvtG-pz7R%*ha@ zaGQvTlX7EOggkw3V1Xh9!tZw-XRAIQ4U)gO8Tlk!No`R)*)=|ncf8az=(ic!{%(%P zsgBL)TZyNrj+RzT@@GTB@^V2Nd^5|U$6rHYC}r>lH@TJY$SGO`n5jn9>sgp3_ca=j z`ld{|MR2k6LwVHm93qp8yPKyYWmS^sEe zBW^Y?p3>x)57Yo30E4t;9~4~Okiyu=$25(JV+_xE#fD>ke(=!&vAeucjGOqq7Xt@3 zMLnw!_Khi|kY((Py!u|j>2<4;-_ge-@{hDzW}Vy+1G*1ffkU#R6@Ow9?Sr6q^x&*! zrQ{1D;c~J4Iz9ZuZ9*u7U&7}+71I#ZpC2cges!|)W8~^H;3skf%!z*;=Q)W+0-u-| z;NajuQPSqN_L@R@G+6lEth~3kkYW-w`y55FjCBgTE=!4KKCQXA`4pwc?k^gOyU{HV z`(&du3@UKJQYtQxdDLEh1eqHbh*_$dssB$$XLy}QkV;<<&{k5`V^^nd^u=6yUF~N2 z064^4`WfV#?_Au;pj&D`48_tNxPc-MRXO30{~qdav))=vx^1vo?CS{Wx#4YaZE{}9 zn{R#c&7OBl&7ZxNM=<9(G(yO)T#2nri{S2=3f)r8)A*si$>QyiynQT}X0jWMbU!~R zpxa#K)G@4(%)Xnc2mO69Et+mZS~)px()&Enbo;bc;DC4<#pm=mWiW{Jt=jbZaIbt! zzn8LX)hZaR@A0m;>F!dq5Q&1_GXkx1+8SD`n4w+>R&dxk`^=pcYU#p zm{@dtqj*s&qB7I`ApE2!yuq8#-DV-^`>PIkU!xR0Zjk*o4DT6hK6oo%sW8cArOu({ z?M-Y?M|fL%4u;7f7R2P&kzf#eJ~!ZJZ;?WL$p!p&Eo8)HFo`BrA7GMz9Pufhm_58L z;Ie0WTKBaDqNI+-$rp7QM$J~6sERVu@-l$g3~LKaSy5iRm~3#;bpo#o#skB?DkCT$ z8(Uv*y>ib3$7GTKOM(8i&QP{Qe1ehkXxy0-;?`?Wakzk>f|`>rZz`)yo~bg{7Lo{; zKB;hqOKA@=&9bNaewQ1x>7BgOq9b0N!gzKqXKtL6@9?SC_Wos_MoAY(Xj8=R(n-5< zb4}pvud)7ZH#f*}x-AMP!3rStbVH8cJGI_G4+ZkAM|JqAZTVO}IL8MZSE%X#V<2QZ zp=Nx!>e1y)`8GFSP1$M-DWfoYXkBaB&c-8(voVFs-r1TF8Q;fY!fJPprs>)mP_#gw ze_QTPl?p_K5@C+&$Zt#83?zJ8LDa4DCAZFvtNm%1ntNluKEPNAOsy<3AH3M$Brqh7 zKsW+H9A=-yz#wZGn3Za}af_!NMO)zAnN^F%H!CkcMzNRNFM)Jb<1$mAsV+MzZEb`+ z-*?Uc17uH_y)xe8)SR|^&8`jH_#PS-66kT!|M;rzd^MPO?JWZq=1-KlOK0Mx$a%N% zCNMM-7Z>UVw^ozHhwvr()I$98eh|3fqn%;s$jmXdIEPH{<=VCKlrb>9(3}A}@cyIe z>lhxc5h5mW47&q=FU&_;8?k0vj}*N}Q`zH=35b!^*4$(^v`INpwjvM4XV;g_ioYvt zuK($cAz=D58WMqq%jdc*rdMUCZBJ?Z%{JEA>*mPl@p4>RK&0SJrM}~8TZ=bc5eZ^H zG~hsq`7%Pk6Yc5uIm-jd29xPB`LXUVwtN3xAC~rL`XSCPkJUs0=o{TiqG5J7TB~FV z1e!JRo6X@@&c!1sk11PhBBHJLpsN$6%_FwYGsWua>P7!B*;RH|oP>myTL>QU@-kA3 z=nSl+OO|qz2yM8N>YeJQR-lP4j$hWxHAjm>w%GLm;D!REWJ6)0@!#PQbr4)` zj#lrlk=ht(zGcLuW5S5H{2@|o_Y%^01S#5p>_4ZkFM{8+?J~FjheZkm&zKPao{&PZ zrZY*+M^hjq1Su=W80i4hfch(h=D3HrxaY6Gp{TgHg>J0 z42_vu$}Zkc0nI<1H7C6BXE8?y`&0afX+@ha>lX8lW!{?8f`(jfEt0N(Yd_2IR<(Vn zd7#bh^B9kxzdj@1SvHK1`?1I{cyYq^`iW3QGe%}o+gQkgJn`^Er9~gH-;}!noQu($ zE8}2!LR4nDJT13kQB2mnyi=7g%CabYG0DgwKnyBT*9O<1dUrBgtJ&+{;-KijSJ7Jq}Zgz@ehBsuS zD2LM+F2}r4p|~2P-a18jl?t*Kp6c5x`OvE>cbeTS@LVP@FvWM+O+x(I^iB!gCaqVJ zcwx#AhC+x&WmGVM`!iUoi5_2EL=SXNlUsW|oBDg28~|y+}N(kZIKz@np*WD2S~?qn1utlwR zc>l3CBa^Y?fEfP%{`4Pij+8TbXk1rXuRZ>tiYj5FnzQi@Kt+M!)LGvpq1*ll?~A?1 zo1mHI0%;D#bpw?3W7M_eKbGrR-Bv0qCYMfMksOzG1A=iK>i@5)9H~i`F9fzM5ICf7 z8k4yd9HgF1PHJrwN zhag~oSmVjq+g=qF74f+&3JE#S<5%dE$|9(od3K3LuCab1w@D+{a3*iE=k_prfM%)M zM*#sj!`)xGYpvd=t7iICi2eC*+@0Rpf_qY!l+JqbqzgQ@1}65?lFvi>bHu%D?b$fq zng*{3e7JE=ppnp@twpfSYS^tFLJ|rkWLBXMfj_#L#UVP`_2HaJjq-oZG4eTF69$At z0gC0%t%tH0q`sQ9Gd4=T)DsE80gbp3*w<4aYLWtC8k41}WOok_Cbw-A^78WZ0s=Ze zKX6bcDh}K)BcN;b2tt`GRsDphPX7v2$e;uRrlJ>DIAh`&reM!4P0WW-w5_tS!6=&O z4I1m1rC*nra~~}Vw7#Qsb7>_=D`#*~sAO=(7SLzh>ucTl!X% zJ8J8pAQ_`^5m6RaCDmlzUcG0eB>D0{??UQVGIEgAnET|h_M86p!~1^AyLn4u-qT3W zH|=$fMq)ioGhX10&W%)N%veNo;!O&;+}S`FMRaH{so7}F--@Vj^eT;x9bz1=l7`t` znqYuF24>h{Wx7}|>A$LpQ`&pq zQ~0o8pm#6w!|NVz)yRN#nsD0S(C%j@qVrs96z;hu&))aP6Y|;oK^o*!7~xRyOmu;r zGnvM=llgjHtXJ(6^*fP;GMK!!LnldB3sPEzQFt0TQ!1W?1!rFnMNI<{px}qnKO*}5 z-dbm;S`cEr3w!tA%J>iwP1uV%S}iVgl*$aKEq$#is4j0>g&F{6v%nybO$jU%L#=-~ z@c~6!L-uQ@T??}O8i9q-@_T`N5o?GFT_uUMl0Qa+{flxXil;}*a(&dOgpgi6dQr9Q zI0kzeQAfHK8NWG#C}tunV=0my_4=^>t-`ft{J-=e{2BD*DFl2a^_|fap(r&U$=0cD{qsEmBqT-89*a@;%#MX%kKUGU;mI zwIsu~E&N8!=`p@)7!eVh)0{GcYx}W)jmfYHb+$rDm*>mB?69*y(-u#KvLtT(D)I*7 z^NcrIcU@L5@#OYGzhcBF;rBV5n_`fuSdXqq@S&o05fcJZ^Lf3aZwFc!BXLjI)5aVR zmB*1WWlOY*7;8Et7_|8~<58;fDb;}uE=!EgH#qXx3er>KG>cRHmTH+V2in9fIn0Rp zX|iQQ-~crPo*up+4C|dF`J+dEK6N_4@f17X{R4wa8^2KeF4Ox_w!MfayUH{Y zud%-prm@f2feEtTEOEBTngT;2pA#ndh!m4>2i86=Zz{{OfCw8;V!~7FX@4KyBT#rE zP|JRd1PH2B4)3YlYG=Ys4j*gY+(aKl2Whfhn|2;n_D5!$TQ88i(e(>YV36fIiY=d~ zj9@ldv~I;`pQN1MER6hM#JmW)pv4+X!(8+E$LtVdjolYjMA1QRCCVl+t-(Q2#e&gu z1Mczl>bbm?@EG+|V@gLy6O?EtXy;MA9AkX8yevhKJ}t?Hy&uG4tj6N!kkCzFKr2jD z>VgX3dK=X}Gx@^2c*|Rck6S0Gz{oNO(CB}@;n>yKpBQ!(lwHSG@b&i4j%onrt%0LxP*Tt*gJll9q~y){TF`O`Hj7{4U!Pbg`m zPI{ZipXP$jI3~77^fpmo*q`+yT+s&SH4-rFG0v^yo^))-U0rBIj>FCLxC4QgmR?l!TFVl+S-#i3f(2?lTJ z+dg0cGR8SEWP<@1llkHdh8JF6YCA`lGArCysbKidQWctHlP9-@s_7K`KXpLQekzWK zbBZNvO!68O(OG&SJz3&{P*$>pZB=^ve%Y)Wm&iIBo4y-oWUo{G2|rhy-WH~qjsVZjQ@o{M$?6Yev0x~G<5%nx3tefo-V-KSf)`SbgsT&snprWj0 zjjo+z9I>0yP=Hq-33)iLF|hA1JCLOh`Te0om1A!sS{?PAehfQTnDrOsa7e(ot7Gaj zi?S>w`7?eAy*irw^&$FJq1lAp7=~^u4&QSsMU-TVRn{xaoy1DZF9xVPJKrsx&55Eq zXb28AWA@mIEIIP&_yg*Db6zuD1$>jGLey!xm zYd1e6DVB)UxJi;u5fQvlN|Hz5x3lz#wNIjMJWb4i=a+}G)v2_2h^uAuUkCNA4Gp#$ zY`mcdqolu%=sug3uN3T(=+ns$#i}q-=DNxC>}#{jhV;;ddWGZv;WX6acwBkFB|G-s z6LxH9tBp#`=9znbJaj8O3W=?sqXQFqj0j@Ah|-KvB5XrV$*FTsT-L7vb4p?*DWvMt zY`;&2Gr3PoG3FhnFk&Y6D-!xok9`uxMhKI_Q7>t}V1)N5p-xAD_-LOg2W;|t{Wrb# zW`U8D$2x(inxVJ*QqqFIRs=U51^*nG5t%-X@i7YtAmr7UmRO<((#=euG=tZ`%n)yy z*6Yq-@CDPCG~V#eXO$$`d#|qJ9a=D!&U3EpNz>0^h2qU*gxW1X+8x37oBW$beMH@;e%gD$Gm;OuS*o zw)uL*XZI_Cd_aJB2;0VSDD@NyKvXU6#b;{9pmBPYnB&wEUXh-Y+xKr`H~D!OhEKU9 zPmZGNds+N{P|J9C1puU8VVOi?$P{Ersqb+~A5TqiF;@PoNM~Bt(dE_ozR7kH4J+ zB_PNu{kz#J9fKC%V@mknLteEkJ`mi6*c|tHkS_RiwS8})p7s6h$$ht#HKS2X)(y(EXpHm*IN(X*2uSNEL_&dW>Sv2V7GT*d|2)&y zKvaZ}`6G~T&zlz>s*Ol8T6*rYVDx0lc7&dn(_?+srO%*nBK#Af-Nhjm}Y3I6MAprdcUA^wI|FUK_P|GMrIJ@BX zR#ESlBp*C${yaZFHjw zsyufNs!Og)41s)44!a=z;@hdu;f*GuqFBH=HD?r$uYQ zPez1YX(23fOFP0Iw+Fs`Z~IPdpMSI`#MsE+=k*sEMWl#~S+GHQC0czQL#-7&i{(=! z%Yj4NzXjhRx9e^`S^RTFzpUH!8a;b2(4sRqjP8f^l(yFfJlK#?7W0#)Xxl>~vk54i z6xBwz4%KVo`{!dBhEmo9lN@bFw!%`j%=w~JEH8J)1mYRMJqW|VV`r2CBqLkuIld5( zy`*Jmn3W^>hxOJ=B9n=c(e{XOC89aEm1U_;Q${A#8}nP(F@o&)37YfIL)KHY--P^$3!MT1H{D(BRJ~chRKp^7KiaQHn1|j z*SRO*i<$!bnQ|lSQk8U_uwjJcFuJ+g@%xwTOCb0TXTVDwzw)v{#2*I+@!(Y^{O8yx zw>4}rAcH4R^OU-ZixFaj-Y>=vtw+(PPZ3i-Y%`a@Ww%p>m_d(1$CWNb>ZA^lSmPg( z`TbW2fEdqSHyJabyO;IGyH;j^DxvX8=dHU);o|LX<6d0jDdN ziTS)mzo^3+-NAkE8&PXmDTF5y8X~~--aTXwviGclYxYUnid+O^fFhM3m4*2#9O5*X ztmOS1%x6E2RicWg+u9HV&L?H~&B1W7iMrPogzxi+kMka(JwYc{FnGex)H6= z^={DRXU5tT0LNk`uf+hALGI5{8+S(NEULZuBcdbL+KrI@`ISqBkfD1KQbr1|#cN_V zEu@tnewj-G_Mz_UjuW)G6xJ1r<{UcL)mrVAiC|NI7(vh*EE zCb*T*-fOej*@mK**ezdu@5}FRE)IdWD}M9sADTP`b?)xyiT$Q?lg^dH?@bj8y#oxX z{v}_sa^Dlv6@z}(otIvXzyO5_;yI>~k}dEQ!RLsFM#Wc!37$J43vC;F3#I%Wj~Cc! z+;5PA@gU!ZC@eS{voCN{v;Vjy-7QZvF!U+)z=A%NO z26g1toOJgOkQrp9Ij&qvqLA0@}t2@8EMm~}*JaFRiiR{WoiOK$# zbCwP;y3ssHp6AmG-O`H~a4sRsx#-;PXw(qe_X3|bl$F@tiL@GAxjaUU-w?@MV3w=R zQY-t#Er&%J6c|-Wn%HAt5fp2>Ah=Q1YoFEcvw8X+a7*L&FTf7wS`eMOZ2+d5%Z9Hc zZ{=&{&I@nxaBe@+ua{?4?krU=6c-EP(uFZrf;3SnnX8) z=I8m{Fiq`_Ze2DICDHxy#WB#R3A^l9hb2fh*AbB=R}%xTA(4`O-wl8pouPH;W*<=L zO=vYM)*Z%A!)SwE*r9H-fsl51Un;>j&6P99WKpK#$J-Pn`TeZ2xLOsPiZ!HApu)+K z^nEFiU|uD{z1dU&I~BcZV-iGuMD$ml)Vq~;ppBl$ZQ&eF10D)=trBGdCk8f?#;OE% zW(LDr%P{BZDkH%Khl804&s=`A8BR*!x4>95O!@9`R67e}5<*5kKCJ`0qij}9TBns- z^kh{xl9PNvha@28GyH)U#;VEVV;X7DD}ias64(o|c1{G>Vq$j3s``}*XvQOx+{HJw zt&>{I+;a$TC%c`x_K`t=Q8m9pUtW%;0b%Xe4y&XF+_R=WHxLnyM!GL~sQuO{P&vDs(o!NZaYcrbx@X z0gdQ(5C2@!k7=44jDM%LztF<(Th+vtApLYl(^!)qmyuYB84kzQAEjeh8h)+GjHw~! zR+{^%2L|=;CU*a{Nce-udDpAT`?Z=mMFA=ck zZMC*V7N}-WE(%o??oX0vPqXx7@i5^`moeI}I`7rnMcNMo9hVxydF-(IO2DWtEmIRo zREb01IScn^NGB<#d*w&uQ+i@zF+vz%T7}M5{wq?U$nD0%*JwV&Oft}jAQSsSk9HL{ zBQ9Q@Qne%sq6d1GKZurP0m0daXUu7K(vAUI9#M@1O4H+hh(8LjB<}MJvYScLwVh%W zDXT_29-~?j&kZx}N=-zY_AQp$qzHac{PQ4+ePX&2)haxU_3bYiA|Y-c8((XLy$5Ic zEA6?On4wChr$@z+?~V__2(HCCksfpY;zcxCv4TX%{wwb# z1w4&fI>{8ID>{|jeaDRN~}ZC`vc}vewihl z(mjPhFVWG1sYJp#X*5EPP)RyMe0!4y^yH;tB|LEvFv?1zG9meqnraLJ)wF}d4oIz| zwjwO`s(&NBc{e+_>GPx3|5!uFd6v|n=7-gMGXtT@W=+-FF)=jarH zX%)D8K(gr5t+DE8m1>FjNmf* zHZ&AWn9AGvs6)H!)I-IIa9A(GSygGJgWKBp_gMW7X4roYxUIA@Z~vkpYra>A-I8tu zxZgaV-J6t~7Mm|;eXQT<+6i(fxi}%EjIRrq-~Nek!b}>K7Oy_mxnKJ(Ur+z+l*{k7 zDcSCuZF-(KzxXlv)_pIh9%#xu@9CThS|&vSlG$N<~!$5C=kOg zF$2(XL~bI(oOt{8mor7lH{n3b4^HY?{1wSS^l#(S%Hdjn$gd9kU1>;vQ6~KL;h_JQ zLG?e(NUrB_(Z4SNhf^4b&{2d!p9f;wC z$&_ej6G}gxRW~T(9Iq)Nbbw!x#|ed*T!g+9i=mf>?fz+!pF_Ot6=z4Cw$DTUo_`q4 z=f`;EVM_I}HU?n_leBpYQKoVgm5CaDEgmuOaanL{MLnT-~P(Y3Y&l)liq{v z%_KJKxV-9Ii7{@Vov|J)^5S=(`{orkX)tkZ3)Q4uG!ICxEjK6U4}?{Ub1zb-NFFm_ zOsl3YZ1|N=K7!tpAhL_l9uWiqGm8+Tz2@@I8L!(O)-PwbnL(o!IgX=%fu*yTaI|YQ90U zOvYuYv`H_pg8Wn}0QWnUNM_Jf**HWniKqFP1eLXkkGjR3h{&D?l!u});kSp7T$4zj z@)DxgorC>R`m>y8#>UiFi7WN1dqR<4*ivdKE*exB5-_V~q&cU*_6HIIG1Mmeon_V z@RW>XQB+KoWrh?>4c5>bfsP-0uR#e8%3`haOB+(*-{r`yELfv%y*7 zx8eM{J5p9}>g+^SpfU(kxLDS?gs6xHYkdHoKl$E-!gfsy9T)GqKZ4KiUHs<=-t*pH z{N-Zcf1&eP|426HA8E=8Tk^^%>D(PhB@3Es4z6-5>m5y*Ev1Igz}M(xGF#8_)WPbpTSZw*p)EYpzx zQ*u6$&W#>4wPvaONf!D7gbxEMdAF*?h`HmJk^jiK7^lZ{(s_=BOJ&h(8=e?305K|GqFoOdL;kTEQo{wS>rOwC|v#i%wCFg`h4Xh~5ETly{+)S?0FI#Qev z{~7$DaWt)a6dR%wa?nNCytWw6CA*+`afydPEN{Df#&hv0`_)(0wv&^=nL4RbHL5x6 zcn7@hQE(8yufGEXTlV0f0Xy;FjXWYsUNQn}gCErhDiE4;mcOw0@l}Zkgb!D{Rc@u@ zM?G*Md+~PW7IZLr;xUWDV^F;n;kD}{<|9I>YY$OT?K>ypB{OJtSrU1?n${F>|M>a| z5#H3HSZnp+EK~c)nCXI0o;c=Zn&^xP%kNSpwdTtZUbLq)iT~GWm<%EnftDYk2gf`x zy#ed^s;l9fPLjIK63?@ze&IUp^4D#sO#l?)4nmxAUS2BssaZXXe z{>0V;8`w<~U*z!g%5|_wDrb{Yoh}W)w-n>)v%1%RD%<%v%QKvL0hiIBk72T;J@;iD3QPoJKcf{6c(Fp)WGdwl6_yH zw=--3FB+T|bXLiMgbgZue9_2~h-wr8Bmfe%87ZpgadHDB(P0n-?5lI#)m}cI>SB^4 zt!7!#=GQYuCRW1jN}qM7?5JEdw3i9*S{y8X2h1UNmPDK)AM9v`-L^Ca(Q4iFcMQ#fX0Q5C9%{T74emo z2@^!OlzY^p)th&luSuon14askgqZQ^lh z1e9Qv@{b-8^H7^OVfr(B;0M9)wnTHJTbI8y7;PiSjGn>R=IdldoSdkaz5Q1d?%Q)E z&8*6@6UktL>)z7|<|piwl~$&=r@@Z;JhE3bWcPs+ajTO}ZnE`Ofc zum!n|pi?CH&ool$=c^at%SqNfLtj0BVPRtxI&cQ&>$Q-fyR1^L=&ife$W2Kz+u;>+ zvro#K^e^&Uc4=4#eD@iKrfIT0Pj$q>Z)$ z9HztA{lfvpcF{E}gu!)YltUO9-k#817iST#wvy(`LrQ;h3WxD=|Eyg3^z1ePYG(p!Jk zL6q}1bw9KcaQy8L^lr|{fN6(EyN_&YPz+p%E4K}8^H>=MYnlNACNGNl8IYi(qeIO1 zD9c-7bBr(G!nVO>j&|uLH*Iarm-f+I(tzW5H6>mGM70>tc%v^f@`RN6%70Z#tKASn$8k1MXk(kdhvOVt#{jDGxq_2nci0hfO<|9!4 zVDt=!pmX{In`Di<3u9bv4{I1Y0leI+fXuV1h6jwoeAxPB&p!~5^xm{97aDyQizQPl zLQI>|(M%DDCvloD7xCt^mSY0la8kXk6o(WKRyOZz<}k$LI`}Gi`Ah$grLzo*tLfHg z@WFz+!$9!h5Zqk@A-KB}++lEsfgr&p$cwwXy9G&bcMTTw9=^K#qWD48oasK@z4v-n zvp-#psa(D23J37nr8^vSI;+#aa{rf-e4O>?S7;DSo;Vijl|XdoYR1T^od1j3H3^K; zg;c332PStNC1K`o{U*C|S8Zb(4gCv*vRRAoDuQ%kwE{5uu3*?(ir&88oEyTvgs@2O z#kJO9K<8^XiDb((c1L`YPSQzHohMkR9jcx*j(k)$JA_vJZsDC%q6gtpMTMc$Ay=7R9{k=G5QisXO=uvX@WDV_>S1ErRn#W z(PWY|Uu09~;c$w>OqMI@f0dj^w<C4VVCe=99eYPS1*jnF=md>MP}yXAuI{$mq;|yF`|o=`sQ0i zj~ZntALH-n=r*N7N6$^aUHBS5T4KTZc&URoy^MrlAG^&nRIM$IoLl7QLrrod5NS~L zDZc-wYa{D=ET91m=^c>QT@2rhYm&a*^~i4U>=gJfBYajRw=I!q+ zO%L6ZkIqHRbc=dBK30;ks`FI+0M{8iw`g3pN1!p-s}L{W z>4lWf!z|s(lJ*lYtnaIgF0N*_XG&BjWOOhs#?eD_0?-;!VcC5A3H8o^we83Ij7BN> zXz?y8vDx_uKUtJN?#Kx#2-q>fGWq@7)@1i^2&PonWwqrO7MDGVA-oLd5b(vI5;Fqh zcV9$3mGq|1gC@+Y)iR6^;#1PG{3)rhaQwNKc}3C;0hso9TZ2ayHhGM|b6b`^dY9{di1$zJ+YnvByKY zn(d5E>&m-T&UA*{@)37f_cK<~*b3chDBmN~0+FSW`t|nK`=!Z0@$$5sGt;{gC#JN9 z54mGT`<@I875PBJdB$w8uOQr+QNi5~XR7;I{5^6@HZ!gG!?oCbOI-vzQU71saQrj3-UgbolOu%b=oe)Pze>>A*HQJn=#+D!LN)$zRi7Wq{DD?Ly zLJ(2oo)Sp})w7Q2!~I&yv-4uiZssBOkc-uFei2{szh*bZip;Aaw`Eg+{#5s#8UO z4*YGnGK0yrWIR!@p&`7qn!p$C2S^#fW_FN!r6(1#zFi9?=So*(J%=^i?B&MZ+~{Oh z32gU|a7n+>l01!E&Ljzl;U`AtLDsXl{=1CS)k&(;j_ZNSCzfsd+SuStQ_)s%6a-Vp zRDt|e*(Ci2y?KiAwuS|n8o5w8;oWZ)&|@Ozs2joMqwbJVK;!rROQ>Qf$(U`tDPhGB zc2rlWRg`Vm5ET9w|D<{)RqU&Fx?D6kwV#S^C*O+cK}1Xz517NA2l&A}S+EI=bprp* zU-tBcvoE!YpnO90q&8GjIr(Etg-p^Yxpl=&_JDyb-dcXf=fo#0H=LRAJAUPI_;lXp z6NB#`PfKu*KtG7UFxXYZ%dlTccG^c2`{354=eIhdbDfo5-=6(UiIjjwiH`1qkkPwa z6_F2O+bLvVs$3@QMHPb<42&6%o&Mh(x4GU~(zWQEV7GH#MZ`cC-C47MVT}^?{mKMz z-g!ce69(si%a}>-{rw@TTA4f9A(QN@@E&E+D@7e1+g#6>8JZ`#+(u2#uJmoWUwN@{ zz!~zcwjU79reE52zkn+TGKh^sVbsfi{=2udH#gJ&DmBPprmc&RzMvsldBZ((cLKKa za>8`L`lx#KYM&Ug51@$n5&Nt;9qq^5K`$V;7KDnNc;DNxeUZDrbL1zK-(R(U$D3SS zO8;W9><;3tCbEnjbDE^{*{#^x+W-EytG$DE` z?hykJTX}l5<3PK6CG&_mAq9#culSICoYK0TWj|fu)WR1WFff3jr0WJnY44c!i;$Yv z`kR-b`cP*v6&fdmDEl+uBCWJ}zqoH@H{;?TfuZ1x$6mhMEPsvlIl~T%s-#5_qi$(f zj3rPT&!(iiaH4E4#$3>RMo8#4Gm>ZvNDik=akF7Ndgcb{QuM z$Ph#ZS*v=qk`G)hFSDKgK~Ze31=Uqn4ceq;E$zFbbp8;v9U}1obx#HZ+|7WtHm{z>>m#Vd|w(OcNA>e+5b9qy!S*_|};F zfR(f6w=Qa|^x7(utW|YAP`atbNGA<{jq-z1E03IRA<7n_wTU56ymn0DB7fA-)Hev( z*YyW>z+R^dTkfen*oRsE+g}oqoD(ilNsB67YGXxQvDsZj6+a_`$B{Or zd>!4g$*EMY&Y&M?@on9h=5a5m8uiq-b?Th(dn2pwv8FPp#v8P=d_N3{;+)#9tz)87 z%mg2cS7XWpOt;xhZ+`^bTq{Qo$6VrfrmxC-Y!=2iMYMk}FU4d-@Ylo>Rb;VO;GH+U zH2%|+!S^t|Xz;r^Q1I}bX}H7G#woDPq2i-LMn@fmkxAoe^5acrG$e8(<{4VU`AQpU zWcX2xymIYk=$Q;7-;}SQdjoGEPoh(J?Eq8qjrg;5c zEZ2rZLP7#)Z=>~>=UP6xx>Bd!5N#tqxmxGBqWdITU{1D!$HfU+T(!f0p!+5;CXsOL zUWi_PGbmXf2gdJXftZU(a>Z6?%%>G|u?8DAKXYep`K-atVcmZ9lfR;W%Foh1@!}TU zeL4K>S#_xlK$fP!yu)Rz%e*PhxBZA_*2vJRaq`KO?KY~)cuG=}0SZVIJj#UBruqx!=d zlfVAfa>f7x6Ck8)&=G&+MiMEUGjLP!xgZ;D_!cex`%Y;+B@ocq=p17OIu`Wpx`eO@ zYRwF|^w^;u;LpxJZ40w^$RAX)lh2Ymg@v60cETL=TdA43aI*y5!qhMW`;aj5r5lcF zPhUD7RVf{X?R1)8eB93y8K^MGmpNeh4?+h znkliI*<_--x`0{u5lK_0(@8se4GvbVwU#$3tswArSFgRw`H{x<3tN`*M%kGJ+W zetMnwAU6>-$7^OM_0l_F?t|~Aa5d(r)%FYmnt{~Ek|!d5FctjnsRO$}#}%o3-A%9c zLzc7aKm>B##WRM;D0}*;91vCWKw;c>y@~Xv9PT}BVw&*?9o%F|o?lF>J%NMYF4ZFI zQV)L*p>!4F%^MF#2#>gl+uCQ}KFI5Xf<1k_jelJv9#A*N6R~4kD(4JFZ1$bEfoph;*Vj_b(X#svOV&B)k>6Q!&z-B(tW7%Rxx%6(V#cNTsMjlW6 zO`1tJfu{=+Ebv zZHa#)qKEzxL)$=>OlHDcY4-(z3E+&nT>mk0R0c}z2tTgBvN`R5YsiDj?qJO>+m(D& zXp~xGWRrN{+$q|a^%DsTs<*4*{8fJF0T1yGw8bACTSI^Bs9sXY7^*aRo%&H1f02xA z)AQyB7zvDAq1)KID!{7M>WO@`SQc(76Q6Gs1Y4roNID8cGbS>4jN?4xc1NB+-U5yT zWC<8Co=k_8?!>Ie!dE@uebbP^F&d1Ke7LOD1b5n8bGm# z^rbCYLrTD0b&mHo%!o$b9j8} zu{qP1RB)bBvB9?v7wfny%-rMBch!n+BuK*b8_Xya^>k`SjY{ zl{7te&7{pF!9_}07LiI?UF1M7qJJ2R##xIVh^n0SdDl#=PLSy~+4TU>w=7%QYAicS zjDnFJ+y}`6$nUOz0Bt#n9b3}AelxQ*)%7CXF<9ZW%!JY;STNmb zyM7POb8&1j061GY?4@{ckk|wfr{zEjoTa%U;*iGw61-Hh@5Ln~23&aYE$XxpDO}Ql z7l?l+=>7W?b_Ca-J3ELyD%i-kyY5LN;Ki2l&J_wmLi#!JzjOX8_n61^M})A<^4CL* zgKU@jcw#_yh%gt>T_Ud2==d>Tv!CZ2pfRJE3C+AlSkB8CCAbecJgc_)e$jKjOeKnv zB@T1&WhYnbJ04J45Px#Rg zE6O=6xJU)6ju84a9C#l&BB;-dwp;T+ENdczqWE9V z;(_I?7K%7KU-ookG8#p7u+usl{F5{Zg}mbGg_)PDk^X*l2raNp82?lFW;E#iYoz`k zwm&0%yDhjKzr_K^XKBG2Va}#u_J=t}%X+1Xoyiv<`-UAvjqFI&?v>8t+8ctrz>b;F5V?)CkPISTDQ@TUca?Irc_PU6APdiQ$#s8?k~45S3TB8 z#LaSuv-ceO(WGK%B0iX8*uMt9qX8ffzWp6mPV(QkiDrT`DiISuWzU)@bZ>oTS0k3m z9DGclcG8UO9)ZH+|>WtldG|5^xYMKJb{G5^AWvXX{^g5pzY8kGd%TlBucW*#~!?0mey zB8NR&q!bOjvO8vH-zLekc2itN? z3Bacak1uAry6myAV<>@Dlom3n-;9`n@4?(fk_f?YRTJl;Vk4wBj-oZE%)ZJrvX8+C zrCb6Z{n+5K8T@A-y!fFkV6Ni;AmPN`;5GpdhnVmcX^gYmXia(d4# zUd0y)=_Ls|CCXg-5XuAd2Q`L{>{<>A+%VFGB!^X{us2xuraeZAhCs*`!mAaUs|_}WBI(;$tSs?s>8{r zohLuXUSMcWYpB49BA(_py{}@w*R;B#LP-$o^e49%An|47_^f$BCs(3fN7w3p%D5T_ z1lDQP&uUc}BpcX zvvSnlB@hXW*Vm1X!^1;l8JVHx>;8{${uHeSHH`eJcm`-dA4SaRHa*nK|Ml5JqfAq& zwZ`{9PeVD_0ik{{T&lKQgH$wriC#->X5Qwxg$f}+U&R4^4LNmIiXT(o z9ER)9EA9b{G;HCK+bA4HxXY`K=Y8MjY@PO|KWB>6>XEuIo#`*L#`&PtU?EqK0#5|> z!rRQ~8!%fExotJIpBw>Bhhr~rA@B_+9$n*scFS*MHgAD7U;UrWD0Hy+-B0Mh+w?NK zCaM)kci_qyts8WBZyr~5rJQ#mq7!cbQi`jmH3V5SS$|Jf*CubG>X#AV^}4hEQ#=#} z%1qz>2IdDlIXNY8&a_t_)?P96Nn@X({LGZx^ut~hh2shv!VrdjdhmBxYbOFi6;Lr# zt~oj&pApA4jg*{DlZ^-s(cP4;|3ncc_y_^4ys;AwI)$s`fm=wyn?Pn7(rX*8)Kh~A zr;rI_48dZb>6u3ThUslby&_{7Xgf9KT+sKmdF_Um4}CI{22Q>}H*0pO(FcQi;Izg0 zujcrX4i2l(=Z%?ekr2ok^AJ_GYM~vN>gb2Ypd+%lYMBnhKjwe5Bk}ZU9|b=De1hXH zOX&qjaOWv!q@x9T93Yqr=sw%^-uJ8Tz>$-Spzyz!UgB!cM zp#Y^gOk}NtBkY#r9!rT#-ox%mc>faF##be~I#&Ukn3(9~;gR~1tH>_1uG&h;KKxIc z4OGPTAPxR@?)MUX>kxtkS1FZ0J7c6FKn|-iaol3S*Cz@2PjY04uD#K4;8YYF-r4a_ zITBqGIdK&eXhi_z5oXpuz+<>=HPltzDPb^vrh(@ob2qdwalGlYivkOQHRehLX=M}K z3AH=9eg4vXtAMEA&E3!DlNcBq(odBKJYbVz)$!7LjS4N?L}5!d=lY*oW&q}cdE^EA z*fiBcF%r<1YUm5IjwCKN_etnXCDJ`)sw22o<;tvUb9Q82yw5e|+x8QpBGH@#X6j!y z5=WDl>!ACs=qa^!Cx_oi%^wDU{Sja;veW1MZQ@)_0r(U=U^~CQDPrJKh~lUJ`@&Rj zyQmCBguWXjNRV{Ok7AXLaWUJY(ZP%&_{e&L{`d%4q`HX*TAaKND+xop-0l|}VuZ2M zhexhAM_)>BD(wpi9A}bew4@6D@0>psd1B2+DXA+Q< zH?-jR)8v@rc3h0EugFYYZ&4c46_0IiDd$OU)%vz$a52RpOuC>3Q+8@*D46us&g-i2Q)S5A;Vb(Z9rbBCeplJDVSi2h0G)4UUjh0H5m16uLo+m6t z2zUtuOZy&K4syuTSdl^fz3R#w1!x#@X|j?=0-POoi;LM{FiBgr<1usT)9ZcTol{rG z*H*Lv^{s-+o;mX_(qOaR&3;A&JIEUH;$(SWOr&3-!Yt6-QEeaYs$K0(5AOG$89WnQ zU?z{P3s05vDD>QBf?jAEx3TI2`DR6FFgzF;@a3tX)V;lWVwCi$!JtT1qOauN1{Fgc~?XSqR&`6wO#hKA@1*G?+2}6KZ6Pj?%h8PSFDGaLo z0+u;Zu|(0}XROcugE7!i>G3PA{$ zDE4!yfiWBN4SK#=G)&{No<&`c0pb+~$zVIrk^xcz^q4JL_APHPBL%8hE+y*}>|edR z)8((i<^MJm6baRPQejyrn5zqbNeaIHNC)FjG|W{bp4Q6sYYaJG5K7A3(|l0}!3bgN+|T)@oOa@NoJG*Qo(>;l@R!*VfW_E%?C}9vRcr=4p}| zKtN1#hT98edzay>2rRhsIg|-|`)UC{5L-mcE52bNUCUyl{wIprFQXO(B2w)&AjHC9zs zfw{oriiG070kEGm$Bo=m5b>j7tE-#^O{RcH<%-H8lQQu|8G6jh3To&^NFEOvXh=cUA`85c}etRI9>;_v&M!%@QQKV%$=UM*)98i%chCG>x zmv?um7zVWR!gQCnw<$>+ALFz>&8ej3s@}O={Lm$Tc@B`APFUYq5<;+WFO92?ptsfS zzR6>1v|mOP5edk%(rVOB4nTFDmWg__QTOqTnGB&{Oj@`PlLWzDqCpdwWa!Y&mWjWQ z5hR9e-gioUdyTtM&r!y1ega+{f3P^|t`qG!%oSr@WKvKEUg;D!Q?X71lo2I3=Msev zU`@oy$m2f(8mlH0=Ly1CIzBzu?&J~m8mmoeNoh$P5Anh7MCNLzKegs#g}`2(1Sk;< z0D2}ibvsL6h!Wv+yL|m3in!=Nx-)uaVsSuc$JG<_^c($Q^_>#PJu)r=(=9R~A)#4I zs56Pr>pQ^g0|>^)D^DIa%r=!A;aQgdT=pkx{ledq?SL5pd6Bi05i<1E8jO~3ar?&fA8P%l=Fjh+CG74sKrN$`dq zBFq$Bh;Fr9Qoq;B^2|5mtkm04Z3=B8Ohh!hLnh}oPAqDxSc+H-uF&Y%TUlX~$e<}pF z4?I-g-7hDn2hy;ph9Q$iN?GB_8kG(FLiw@m5%@dFIV%ERE|H<pX)#H$LXWv^Z4l5qq z&$qpG`GruRy-rWKB>?M9{|KzD;&CI}X*87^Z)X2z-e?GKwTA+{4`|C+;D5eurQzcN zD`~|*MD}ds2=wTb@$NImsBo>0<|E%e<7-eD)n@_^zn8CU(`mn-(h7qSR)H<*2Yp{p zM@)^;Ph%#lOOiSXD| zp#*(0a@fO-9?Z+ZN7f`pHdDVZnqSR+J(l`e&a{jABtKUWNBQcL{|qRN63O|sgiXr>W%%C<8)dRQ8iiY|;X-ghX&va*uns~hS5HO5KjqRLk>5c~TSFHuHz@C4X zFG~V%o^y?tE`0gVxg>J%+q7pR%+lWxYAEGEcl@EaBLPEq>wdvEW7-pj?Ex`O)m}G} zXori^YKI zhV60Vk}%tm4q*E=3eZR`R!Btb&C<2wN}AQzYg*o`$x>S(R9DP>GlE;aZH{%uE?>Xc zju(-?qizMWkD$g6)expB!HO;x0_8blz_VSu4wR?NeBoJ4^elRwAU*|UNZIw_AZNb-ksh;;GRG`8CIB?726z?{T)qSgq8l4v5ZM)uJI^CgFbV#^!i_CU zU9$?z7lyOgNN%bnV_IM7!m!6-S5Rv$3Oila&kuO4dF_{-`~2&${Pu*3(q+q~v0$xZ zuwZ)~MMcJ!RC$Hj^(3R2@*WN?ZA}sU(YB1G^~TTvmf+bs1)T98vlP!h4U(<(3~F8O zUZumFaYp|joA-ys=ZRGM@-c*K2-~4CI9s0X9j-#u|C~BnkX8k_q1;$MxXWYQ%IHjf zw;OY$eR^X`w`~4P`_qO{MN-!TdKFgKOsu8#EwaN}jNR22r@F__vl(mrA>)5h!A@)t z&?J5l)#toyPu&dJK@2%<-6_tj#3m?7FO+cb!_GX}85dafwfeGs(vI#*W+F|k>12UO z+nDDT=pdBYVyW2yw6KS?$w7?Qid2(YaINPvpSzb}id1)<34O<(!ITD1ODk&s51+|r*L6Cw1PtHj=0CmoL zQmiS;Pfb8w-HyPl8?j?8FiYV!-}Yap)sJEpd^FvtXV5qh^2fVojk{6H9MCobOoj2p}jZ>%h^S>pqOtvnB5A0!54bwe@l zFF@}da)P-P295fnE@%0>8OJ%c69%!w;^Ve9;##!YTx0C|exQsi)?3fRGiM2)rK(ia z#L&uyt6_!;WmuX#MO3EIMD)K@)H(m1wLlfCW_U75ms+@W$e`WE0BFi+d)p7D&$Z5c zeon~La;kRr+mgV(3E67|8Hy?BgwW2>7`}dO|G6~&7Ac#v{5WOXmo`LX0EBvbXBccEP6ROw3>@0SJ?Q$^3eSfH_v*LuI2={ zjb9(s@2_+Ma>S}ILp`G7!@$%0(|KkX$x>xa&kAz(>T7tn&~q-Y zH0ZNY2v08=6^1h9#)%%h!!i2QuG?l;83inBMG)OaCm?u{J#YX=yxV?n)Je0`D3wQm z@|yHGAl%3vW!_3dQI+srLCm0O4^%?GKNW%Hn%rA$+Q%6am3vyCB;jgXWy759B-Np-9^Ny~WG0AxL38 zvg&<4%GkJTLIx4wrTD9r(gtZo=~XMjX<9TXv$^GzQ5p1-AB0v&kr|G}Dvk zBa6C_(3}$mz2<)k$HO%sNRu+Tzu26ej|FqHi`W2r%ON)Fz72t&?_d$1V-!|DQ+8Wp zTt9%U8FK$Ki{tLiMa3KKK<#bM#9~bL?V$P$4FRbj-A}L}uL~RY1?U*uS6P7!#Xzw8 zxZ!ur%oo`SIBR7YG$*a9C$Q&zl*0&?LEH2);?N7a5O80%m7vm6SpQaw{eVAEm>@op zgA`LD2-7U|*hgq7&*OLENmCll;(MYQAXn*iV5Q}w`yQ!;INsD;JQpAcqTYt?Pk*@} z>mWDy_t?!T;d3AN``x1WGv<|dEW89tVT38ca^ioqOI!hV1 z0cViWTwDiB%xD(MLGo+MvaeOy%s=|Xsjl4=&^CKE544Jqe(c#J4gF1--5`k9livT< zvAK^$xm=!J3sbPH3>mSbBITX@WU3clYDM=j-OFa%4O`E!fG}~@Ff{92hPgw+yj+o9 zgLPBKJ+vStu*2VT;P)_D1EnnZizDEH5Xy@Vh5Tu-RR}P*@G%NQTFbcptx@*TND;T`_neWe4KZVo*4W{-&|pBOqz8)p2H+52IeiQgk*gG^NBVE&aZ4i9*g1N0;{(M=SkLl6gr;06Tt)cYN+(T;!9()6B69FC zjer7hz(iDQu)e{M6wT%_nN14iXGTyzGAF2Gidq&H@TKSmk6;D|!i;TZ)!_IdR`qs- zZEt3@lKC!Ah+QA*g4Vssv(3YMY0Pc&PJq``gY$*L46(4=AeG6Tlw{U7NR`&D{5;~C z{@%9nl$6tX4W`&*h;YLT{K~hxK*to;i#6BTZ=!7q8e&xJz_`>B4CaKErww7RA@}_D zEy}9JLo;R70~F)E;?VBjij0hkp9X$?f&Xr!`|lK}gpJr|mQn@D6822mugEOv`Dj)d zJWC2B&4kpD(dAus**ooqc;;#+CV=g-yWsGki@oi6(VN?6AA(c*f@#(ujqqv#o_0 zZYHKpfQx&7Rr*q z_GbV(_5z$)C8AEZIZ+6fRG$iR2u~;ifp!*G?zsMPkrIjT?QcttwQgz5h<@4LI?0-Pp*Ve zi6d1+^FCRgLS({!Us$oD2T3f5`V3?Ju@Z`5u1wj${-P=y;Rz1fEz?ehL>U1C~A?Yj$rAn7bvU>Nxb=u+e>~Cb zqunGr>?U^CYn1njpb7I+YuH(%xBZb>#l|seD90J^sthft3NFYSBfklKzBdyaSTz7( zWN(3*iobuQ26*D0y?vdQ!>LgmN!l$AsPcFtEP0qbNG8~Z(lEglwzS`iWZlo`6yA8-*NXVl`quyfhH(en~5hT~hs~7Q1ag z9&>i#N#6LE5Mwr)EO%=v)e&MEju?hYTaT)Pe*&VwFl>pnXC-Y3(UjuPn-oz}<8~7o zKj1x&DEexhzc?llEA!D74bnm(RGt;B}^tN|cdj z@gPmOY<{IK2diF?qN0g<8toqJc`{~?qJs45fQDiyXo5hz=MbvyGpTc@C8~blw~v%I zZOu=->k!Mj>tnoF+ulj$0QiaTta85ZII3HG|6R>; zM~W-JuwWrm#DQ@Dv_%G>HOvGSBJXZaD{|RZUB3}v=u-6BTL6PcKZ)<3V>2Os3`+uVSFjD zsAW0NzNS)(+p9xSAOo{!%Y$&<2G+&`6Tb4d4S6*Z8eQdt zV}r`IG$ALLBXjF89;r5?#Vfz1Ml0SQFU$DowT<0-EdplM80^hw^f6WftmJZ9B5rjn z1Sb@bvoDVY*4d35B=O=~lrVq4-SiwZCaKYy>k|d53Mu(9lXfbeS&~JODct3H0hO)~ z!+lRw3Rrj+kr*>l$H39zTJn~@D9_#?H`oTI8vD4TFK_ z(eyT}|2TLhK0<66b(7GE1y-M|cjiEm{>e>62jy)R13$uO?U-_&>sUKDG5;|fD}1Fc z7|Bmdn9dYA2~?{X8Y|~l3aS0nu`k%X+SsqeCeZzyQF)-6FEHqq9z&lu{71I8cBagE z@lrUfi8TD^kXZifPpu+Uby?gz7~{9{`+q+x<40C4Hm}!Q4h%iCNadCUJ=!SI(b-y8 zC0E?*aVNLWx3~Ag2MSUP{0^tG$sELrjVEML9rQbC^oRGBzRYQQ`b2OIn*6Dt zN9!UF8WTh03nR?Uv!=qj?jQ|{1KP{X=mC!C-TogtYU$o4F))gA@9AYBJo3>!O40>V zLATA{dS_baVflqfSN%V?_6)n_HF%0#52Q#f&*ZBj)ibT(g*TEWow4=!_smjss<&j9 zBvuAx-jM$`fgQK_ZoO~4v2MS-wOW!{`yM!8d}x|sWNW>V9H+36Q)qJ^rMoWqW!OrB zy)OCybW6v)BcA(AlELVzf$q3%Y+%Dt`&SKwTFN<{+2puI!GqQ&4!=rREJ9_3yjLy|YaDWdDYx+LUUVs7jF^q$?XZN9_IoI6hzV*@g z?&|qKrxQa5?+9#JvZ~Ll{UgF>@w}jgl9)Mnq3KrDF0)AqJHaD1HVUxg#v@-QNN!@(WCt^MJ5SlS<3oBj#riiwF?5(%RvyK3TmFVp%Oo_iBe1meB(fy6Aq{zV0MZOU;jyk1b zr>}OAk>=H2$W#)i<>YW7o_zAt*K~GMMClm9Ab??F@QMOlK0A^1C+DOVmobYzFvOOW zTeIO*#Ca0c?Cx$g8qPyHkc&t| z3zPQC4bdT!fH#PYuz<~{8}<;t&U}+K1W_B-VVygT{XhLScaEhD?Vp6AYrs8oJ5xD^ zn$zaD#RM?w6Zjtu^ej?|1wE_XjsrrbVX|0Pk~QJ+%j&48%f^N0>qg)G8u^6_ zcN8mHbp$t~pAhO5jK8_}Q`Z?UP>$LHLn}o3lWx9gI=NL^!S{MD`gRhwN~2W?mRg59 z86GjhGikw&i&fPjS>J{Qf}D!m9_!av2Pr3acj?tuH*;edgID>rOtY~Rbl_HmUVtTx z0?R=Guw(QJAm7J)cmRpaRqN-g!i($-4)dt-tNWL3G{UFA;WBYnT@exZ%@zX~*E4*k zh+f+P83}Aqg>?8oiIidOQx3VZ5y5Act0tL=>sYm;MaGhhpG5AJOSbmUgg;8GPINtF zH66~h-7v3P?+VCa%hbO9ry<993phC%92x?ngpajCLPE3vyK0@~REnuH{NO~CyPeuvYnE6M>=7?B3gHy-+L930T6p0Bk04S5A`BUWa$#4(m0naJO1 zFI3<|*=pxK$H}WzD;?A8)OE0}*kvQObyY^tq;Fh!WZJe?ot>W5?tPGk{jl;930KUL zDLp1?eCGXICJ}YUUe_z03;pqD-*Ax% zbYt68Hcr~d>MHvj(jGd_w)%jC&|c1b;0Oz=-P zs2d}BW=tp^j}JeOhtRG}`7fOE80+f360$GM4I|xME|JFjZJjnj;kW11cbVFWh2^0} zaCjrb5%gC1T+Qwe{13m#A>T&+EAV=`1(q)`uxAP?@>AH2 zgM>$M@~IA_Z{i@Of#y{=mClI-%lQCX z!CDbHL>4?|gCTE%^6UxEP{2I44U}GFI=pYpzS}R1G&^mI=Zk5Q-*u?6TBEyG&5BfB zXh#G@fGuJkjUHyT{Rnb>c3&c~tJRk$^UXJvmujt$=-jg6j!H?4?3RbvL9B6i;Gkom znw)>SC2=6L3Xp4uke`KlPg#J0)8_g3$%BoKO42bdnyHx-mqrd2E=kEGNu_?kdxJqZ z60ul(=F-Gd?-3g!8RHfO5i>_ne+wY65_0kZH>+6B7gP)Xf*_zeJvOSA;{>ux0^R%G zw+`Qmrpf=&bd~{CMcWpZlJ4$0(hVZrb!ZT!JEc3MyX%02Gzg-AQqm$FN_TfkcjH^! zdmq2~v-eqh&9&wn;~P(^LHrNAe#Yi#|D>_V_1=b& zVpeonN^flrbU90aj3X?p!FOHkt=fFh?$0d~?=zRN980-8A007A_7*BVo}(7dl{m2v z;;=9*N+E@*JQ9*?8RIqZXumH?RI(K9;Hc?4+b5Sn#j0#(1Gr-qi#tXF~#{KEK1#$oR^+JC0Q`*@J1q`|R%qY(nO{>9Y z=CF2?DF&5~`EM)q@>(X!bzeW$)m{RE%;auYmPRROYT$F)s}_pj?yKBG@#aqHNHgS6 z?t`@F^WCl$uW6mJ;x%pqM+s_qN&VF{ix z(SwWz3(%W_03DIezbVx<@V9u%YFDc)QEZU<{R1_nbi3ay`$M>6waFHZ4LoNTBdXcW z-%O9bJVyNmmuhP$_GfgdWgK(xGh3YiHye3jN72d=DNXp}$q{|VXsN?|-DyOA;bScu z$~oo3(j6 zz@VT!Q)kv4Ge&p9#i0{@`l1}AtN!zn+8O0NLF9C`DDt@&^6eb1 zf^m1gq4DR51`8W?RXGd=Z~w}JJ=T;`^ zpRs;^moX`KBk)A)7T+SBy)_%CnpykSO(zNQj16FXz~mC`#Q63UsdP*Nz`3wAxYl~X zyX97YU%$J9dY;uNQPpgL;3ssUm++NaVH7vbrHP2b_uQYK9FKuyP>HAAUXkA^2?Q>g8A{^a4?w&IWvR1FQp%wsQx~VV}z0?cI^J#uJDB$ zMqZi`@FJf}--*B~_B9y_k_FQ>DxtK4=#THnp?0D3hDyoJqX&##^r&;CHUp%!I2CAK zA>%$`Is{ABOnU$MB~Eo({Td;Nbr=eM;wgy74^9Z|&)s=%rI_XyRQCdbJEsaLx4oC1 z1IMp?!oHvdk<5G4Ct$EET~fa~3aGS(D>AefApmc+PiaqtcCRKLkO(OJSdl4Z`vk7E zP2_+{$XEC=VXonw5A3aoJm1*GAjPudc@6 zm)R6D$Gc8?a#i?wJKGPjRsQ_->zBQ7;G^GcVpu4%Hh!#pCGCVb!OyQ)GWEY7oIU>` z={hvZeT*1$bsk@>iauLtN4I%h&a3#Ki8ML(s%nM@r_qh;_0~mKhfG1^LMGo6KA?)Gt zk2%c5`FY`YaiS(FZXI2}zo8uYmg`@yADf+=(;XUtn*0@rlwL9)(R4mplg(t!-Rb~i zKXZ^pI7S#H`5T)mid@eGSm|7Qf~};79#fdqzif9F0)NVBMDg-1C@8Gyf<`7LG;hBd zFh7uyeo}^2c^w(OdztV18U!ScG7M zlOUVZA)P)I3(z2NKYo~Rbh9S4lV26iJTkI+RU3>ZUe7(06x?q;oGS~Wlau)!^aS52 zJcSkM;q1Jqbl=q!Ozyi?oWRYP2*U?)TgvKEbdtj%-XKx?nc)gjt(~qCRhh<5wByD= zg?%NiPPHrgf`)m8j%_LJov~IivgxugBg3pd@8>Ygt9jol=PeTvfPEGfeg!l4bkxRn zD>?67%{i+CJox*|y4+iJO8>GmsQY~tbT&qbADW{YEtty#lZoKHf|CMSvneC z@n~eE&{RPVYAyroeS{jek(kx3Er!s%M$RgNQvR^#VFhXz?Kb*_td2%iI#_1s+|F(% zZ7rs;qg>ibnFUd#pE!U^g^Ge>9tS;w^QmWjGNOprQ=1nx_k~;uz6!MY`1(fi4wO98 zneCpMtfJ?{Lo~LNW->!LF;V^W;c-$ME$^cAY6v3!Bralr|1hHP!{)841n*iC5OVtl zqNV6)Xe3mz#{6wSbs!i0_{>k)9?L-_hE_$c;BADQ4&QrSgJ(A@a`ZqqgpXk>fame}js@EB zgLj~&9Ms1l9J`3suLvEt9+FQHZeR5%f(c`v62fA&Sm^3LB;9pHUIedR)~siMwY?U4 z;V2?1y179So(9jbUDB< zx(#{^i_5dK*lmUZOZ_0;yKXnyI!07I`OFWc9yNb83ywy=Br999#C~0+ddp50eKU^S zj^l@mo$a}5#|lO9cHiKZ%tz^1@xOYur9v5GQg0qbC}Tb_qIxa7Nu8!u>fHHW5-;Fk zlQENj@}ftTgzg*7*ft+sxfx0$Q@=c5^&W$qK|1ZrxV~)Kmn>fJpJ5a;60y<` z_Uv4ZcNmCwktnCf@5kS>wB)TQC3TN9|xeY8IMT8|K18@ zgOPv`Txi43jbA9_6Zl(*H}l|I#Y09&-n9Z$`IniG3vZcH4&};2Wv=o8y|h5?f^28J z8Jpi}j?ElX=PIg0M`}vjMA2ug3OvOJj#?+nXv6&T{RI-L z`YT_kRFe# ziT4Ky-To2tyB?WSTr~_L3VQ~7#`4IICsD0VM-2a!NZ#fj2g<=XYVe2H-wq4JG-9EE zmDj(VDE`d@0#>tiFoYK5@Bl+}y4ESs{j=aFa*5u|RGAld3~OyGt9R6H#c`ViodSOj zJ>fs#B>wY>K^}{`9V84}h#;U>xRBS-aOVaSW(|XxX@{yr+VpiGb6G_a&?BbNSpLB+ z!Ob+;ypsFzF-S6A-ppbz`biXKYiS#Hl2g)ta*in@D+3*8LG6sD{EU>m6Rt-}X1i8AZyK% zDnchmHAYZ`fO{wbg?x#a)`7fZ=dwZQwmITpPEmjxq%Vq@0z+=~G4ps@HIr@RC8dQ5 zB9ZQ&Wco<0OFIxXQ$q~n8o}cNVW_$VnHP8pL@|V{4CC26Rk z-fVjCV0pj#7tI$SQHg4FY%LuHny*sel=~aCd-uHGe6qTR(frt2A13%V?is zqyPWA9X(gpz*w~jzBp%K9Kojos|Y!pq#~rh7Pb*{yB`ae#0QIGad|1&seWo&$!cD^ zONJ+WlN%V2w1e-5xASnGQf{RADcU8sD0{sl7ecckVM59($&`XUs{e}S@(YJd6^tum zAD+ROXrM$c9lgZT3N!tQD)9c|?^CF(#m8RYLRNf*R^q1ok&hX;_Q?lf&l0~$ zwon7nyHHoTOu4npm_&yJI$RkaZ}VuSX(xo1hI@qvy^M!`Y^^&M>0evT->)mW{I!RDI4dtvpT=@Ae~kGX2P)k+nZ81YWYb?Yu>6%KnFt z&6}0jgjqp?CANXqGEM}3a{VCT(WB4mz(G?zGejxCGKDdOcEJt5EDwJU^&PV)= z1@pT;LGE)s?B^IHG}iH71~hCqtf#QTUm=g@sB%#~if5?$CF~&mqN!D(C7~7*G*_ei z*dxB^Cc}CCYZuAMgUILpcJF(!nVNd$=a(j+lfe^yetCpL_)4B*K&%=rk>!xgH{k`+%=G zxC~XBkX6`5VpI13Z^u)&mE9IH*uv;dStADz<#=62CFG9s6es7ekfwVTMVbOC00_WP z?Tu4dlY0Y^q=kC|TzJ!>IP`sj!;li3^g(e66fL~M!>)7YzW>u}%Js6mp8%cSusQDS zYBf&E%oVpv$f;!pyFS0)#iX|8>Yr>uUJz?ToIa$sdOB+LdzZ&p+(kX-HmPKveVXDX^Seker*Z;C zV*jztD0E>3el=aUY7ea#Df1j^2v@4`A#P4_VHP_t*1A22o%QpoA1;v?& z))n{B{N|p>2p!3fgSe|)K0Z=M&Quz9>3@FDd{ikjZ3rM`Fm}q&joLH+D5>r(n*T(v`XXws27^$DfcSjsdYJpe%@at+bQQoTsrQdAT&rAl#$^_}ltJcBoq z?0RYu=M*^Snhl{M%ehxChVvT$`tSwX-o!{Da)M><^$9gPAZE8bu^9oY$B%k9cW}Xl+vwynY zVJa6obA_ye*-!eRgfjXKc>c5(EdLMa+Go%64Ov$$-v7E@1bd3 zeNG@N#wPR}%PXYP_8+}64R?kwrSz6@r#fxtmT@|dS#Xk%neIM0gIMH{N7qjGbd^aN zdV=9)M4orlbM|6ep};qMgF(H8+0iOATU!H7s(M@+QE82H(2?7fvlH`Q0Snwpk|eAxG8Cd-Px0>BsIrE5$LX5j>+^^Y`=m} zV%ZNOhTRCl+HvvI1Wo^}B0FEjCskokqi0-11bb^izT9hQXk9yj3_cTAM;3MzflI*< z?7;BxbYZC??3z07xXK+JHt~0D_w=^{y53reRl@E}+*49`ifAx(yQb%nIbBK*NR^;G z3V6|nbfx|JjtTsHR;^9`cXb}D7IK9k(B+TT=Qw7Q6-IMi1quLKVPMDGmrs95J`>SC z!awp?{i_zUk9`yK&#S2T@W)eG#8>4ZZDxY_Nx`5ly+A=gUq}{+x@ai$hA9vPt$y@A ze468nAc1ngB!cbq0Wvdu-b=@)ndoob5(YbBgSJK8yb3!$YpTVg5c4MgmVRWuo~EDC zGJ|_}wa|5NhF$soj2$C)kEtJVzWT^U5;AoZ_)KkCp+Mh{f(VGO5zayjKn@YouTF83 zzlRGCB!E?1CpJ|=j5Ac!fU@JwxNaNs`DX- zd`*`+kN7ZL>QnI4cl^(XbCRGCB4^!oQWlU=dtXwEHq#fZb>Dp(w>gdD-AU+t5jOoH z(pkV0S>x5BN$XKGZ||YmeqrwcGllsCyu7{e-lZ?MqRC&p%y7#6Carv;(9qnhl+H&~{Pp$c1OqF@$p}pH z-yQy-8v#2&>3Qjl@%D>}hFOq;Z!@X=1ps?kv{a*U`+8-GC6!h^3IZS)}0G^2huSN%NaQ?k?_*db(#)|;S^1S)pNvDHm1Jd?iif8FM=59 zL%Q=K(K?bOW{!*AoB5Vgnfb-x8}U{25X*t}e}z@5apY!i94?D1g(^iG09VPWGXfLF z*VNc(`jiAo9{L9a>cxR!up(Wr0IaKRIpGK5KD!KXLvdpNcUJ)!yR}6#?avfwt!`y& zK9=F&r&r8F(F$esLu-<<3KL1_Bi?CMn9nG|!*KAv^j@y#lKkwG!jT*J6|>9w&Ah)S zOQ*VBw};YxnJx9DwKf#T**FOS5AnAsr?< zwEc$A$kqa>iaYOe)NObS9d}ML^rLr`y?e8WecERNnZ8A?0k(isI zy%Ccmla1UT6vd3s3kL)cV&bRzgwPHpsfRr-joQj1cQrL^(m@D$nyv+ZQNEs~<=pry zRI3QH$ultTl5TPY_)skH{`^d(y4O50r0N?}xKD!okvw4V`Z4QI_(j(?hgv+BdBAsy z21qKUKe)Va0ZxZ?v?E4jn8)(B2^i5&v4fZW{pq+0w(u~BXFwzw$l}LgC64a^9G`CQ5H68W2-mln8&ec3 zE_Gd;uvPSL=CeKI4FR`Qe&57HrY|BJT(-or6Pjpl#|%lFnowO=VrP77nY`P*_i+6< zn%1r*i795b<5D6W*LL8n%NO6{o2M>i!O1W6cD|fS#IFz^m)hRbHBFS~gTnn@;`Cg2}x^ZFum+YfrdBQyBU=t*zw3Xi$Mqc{T zFftZf%&_pH1jh~~?4%2uE9V;iR%MyqOya#v966n5%Z=US^fWODa8FefM)+ln#1?Tq z_-c6t%*N9*f1~IzgI(oCKU}XBnI3s?hM-%?!rNj+LPTWbKLQ_MFO+^efIc|vCAWRP zKG~juGj;Tnq@YY1@Cw7XfzD=RF{m^8OvZs8!D% zOa!UNMp{g-QF7)dJM&vBGzx8~L>iclV$9-zfn^|SD%1J8Fj0NONk{V|8muy4o*mJU zkrcIfm?evlK-UiwOC3ePc%y}Dn)wkRR|9;Bx zx7@AmxYFB(owo>AI5-p+ZwX>wq2@kN6ijT-+I=Z0BNloWzfM~UNqy*&faMea`yosn z5hEXZ(w6|?6UUE;_$+OT(?!#R;a5LIJTEE+vUx?Z6N`Fshx%Gug--3iQpyvZI(`jZ zAh(0ARy+9Ss{BZDkhYvy5TWdK4Dfc8qlPz=R%`d(cT=K_#m}SNx?biI^Q*2jh-*kt zTjVy|k`&#d6;%9FpDQ+4TaYV}Fc8Fk(t31|JgOCXO8vyc zeA`j7rZ(aby^cvGqLVQA!0$!-_SaS&XbALlC;lfhIsPd#SyFP1)B!1Ryi|jJHus%- z=At5KE>8ZplqI1+1$@0ptyB(K4P&w?`B$5dT5&mBhH*B1E&pZQbk^sxI?RS@cJPM#)MRJVRP=9T<73gdZAI-MA&%S zw!emY@B?`S)z!pXV(j?YEjf1apX6dpXgDLCQ?vPUyDy2-8f_0bOJ8qx{7}O-5#2hq zn8oovD6f)MPa;W*+4XmvgTVv3!wbswxVepi7-!MBGG@G+DKt8m0Jh zpKC>8*ZRyWgsm&{E77Wt#CofQ8jjA;KWyfD#i*+l7b&8=Po4Mp(bpbQd|y>#`z!AS zUp^m9@1o<++09`lNeR=fu_Cn{Ujy9+CgawY0X>!W+__+2^?@!U!NS17?e}pl#d_KI z){wF?*7Ayq7n1pc_Us82u}+;2K+{}7PeH5pHs-j6e%nt>~ml zh}i!~GF5zF=dw7Ko1*qhmZ_fo|LhO@=b+2BoWPgu@7Az1+JBQnM2|@6D4SY+@gxVqTkBZ0((g z4{fsD8w47$nc?MPLe&RCchuU%?v>etVYpe*hYHVZJiC%=LqSN1l{^G*LaZ|IEhHFg zVmaWhqTISeq?pWc0H@BL)u5r0FIlrdsyjvc1uBb{MR*)V`fe13-3nG#p}6W z-!t)pD8vt?kmh$qq$H=T>klQk=)U@HY*8mUvm&=_KG9ox^i?rYXl3FAP>+($8>}ap zT`SHQZP_F}(jd1Ps-u|>S97f{)W~>CK^EwKnv(y8kEPMqwU^GZG*oC2;|z}3Mz6SP z;)6(f)s&iGB{Jecpc%ewxRyO~MPnNo6Pg;Szc6WSUqCZRW380daPFv~QYI$($~HTD zd#seH$e3gh-lAWZb9Q6mA$=Yufd1EqAZuyc6D$^kzW3Wz8Rn$VH(#7tbOSX6ZZLvc-LapiW?1OUl2d z2hnYaaL4{DVaU@79a}xz-a0chJ?;1|o>U6e7eQ0)=t3W0pu>Vssi?OKDtKpu80liTMP}g><1> z9gi2#4AjI+!d8U~7z$=zjJIN%nwn+^d3<%x_aiC>7y4`PGBm)JZT;59l^cxcJ1NV& zb@OBE@HQllQi9OGw;#FE8eFy7+V~^JC*4p!XgUwEkP=iIt5zkAwsxO46bt))dzq6eWsv0R5B& zCH{uXez^kt-I>TOgZ*Rwf2;g6Eo!0I%me)6 zbp=5y$1&2O;cTw0<3yngCM0}e!j3-GW$Yc^meY?kte$gd#*ar@EQ%kIpTil~DE;$` zQIk_RYcItVymulh?}m=+{804I^rJjaWeRf`VF#l5Ri5(IAIM=2B1tPNnA>+oGgZrQ z?ZqW7QE7H6pu{ISd5$P1v^e$+dsC_s7&mco2t^^NoS$(X+g?~(_mU{OmQJ@Y^ z-j5h~BUq~uNdfCve(w!xlv=-0NbL?H<*h?PukrW=q+qUAhoVLy?GqjAi2G;Yv2^9& zor0*ZI8A?}WMgCFdN&4{RmB?&^B|bWtW}MEE0c#9qKC}TT{Nq%rIlJvKf2P`Ga3&> z$839lrz%);Qc2fpHH#Ii6OL$J6375-*$bcT=6=_QTaVeu0PLrzNq;S{;QdP>lHoDs z$6Anf_+&Bdk#CliyD83l9aCB=@DBv}9ty$TjmtKqN8eBA0w4#TeOP* zV#bCS1}#H^(>0G4=J#1CzmD~YesJ59D*3B@ob z1E$x$e?46K<2&u|qjw3m$U-LRof^`S1Crw5dNmhTa_91;f~;7V0lG)EXFi{;VuY`= z{{T{02(Y{qK7|LKe+D89XBQXf1vLNc6by_utZn;CEluPcZfVQ)>s4E&m1BBZkP-dy zfQXLnj@KCej%51hNe?+sIkQT}lL6+Prc7b`-QSUd(EUy~6k9BUj&{u!Zx>P*DA;Xg zBmY}2Gm{ae1tzYR0M_=M90Ub^zBbvb%DmIPuf%`O zB~BqkO@U&H+M_3R@;$7r(95i@?MKaJRGl+nx`!+*o>g;yp}^xpsyd5vjm;3p=^%sv z`==nDKQRd$^Vl4wFSed}HO=!F-g86^B1c~o{cZIWtVAD7Ph4i6LXB5v3ie*U8?Tda z9iKB6)zaWQ9Uk}k&FjdJtD>ll{T})9a3oV$M}y8vY5rnXy@0NsmTQBGspM<<#zIOOS3=RaH$li9q6y)Qyde z?L$MVR`OO@p*uS^J_Dt)tE;Q0ymvg()rpDN6ciK?ZiZ5%l|%9NGaJ>eDk#4~cKn_< zRqv{`xcDzEe7Z+V>U9^v@7$k5PJDWz=bu*dOD$Nwl>Y0t33VYwlM|ErcbK0`>?y>pw4{k7Cv$}w^;B7mnDt$_&EQ*P_|d8PLEt%L^kI5 zv=;d>C$}2&dj@2nc#qj+v1k1KBUYUg3Sen{C;HzzkV#3x3n0f8)7F$Ol)EtsXOKd< zm-El?nO(dkF`%ZBGdNZe4d${8W3cI7=7OpepB_bR6jH{a*X00)n9M+Xe zVBkv<ejbg^)Dfi zYrPxO2P5P-eHY_fdi79=5H62FX^O^umatHrolka7OV1oqcklZub8F?_7D$Ct;YMp@ zxH(aN-FLE6k>Ye21G%d-Q(to5t{on$>OG>HJuyqO)JXJ=h5)Pp7DI&SLNR z-jayfTyAQ2n4lm`J+pM#J$3*bSjQArdUZIUeC%D#cl5K8Fzflxw6#1FqJ&X!NwZSE zc+=97zs~wY#@$s#K_Lm(8AYg&=N5l{hVgTo-RMo4hC2^DrS)uN_|-L?L`A+E$om|TX>De~%@_k4?BWhYk>k_=R>ilY-%d3GUn|M?V zJH^iylJGPDwSJ{sV)$ie zB=#SHJ3%-lC8bQheV`ha2Rdk%!v#TUCB214H~B5hnPx9eKyQw&UwZ%KZGKhz6|5}T zC2;F=eZ~UqTlbcPNCwLlg-1FeVi`DhFE~W_A+sXEC0VR!a3B>|CRi5kNQ;MMl2e)N zar=t6V@R|>P<#*|i1-Z3Nwa!D*Zh$TUz8D-l$6=L;gZOUE6z}w8z+LeMw)RBO!+$_=i*X_Av25? zMzl{)Yxi_%^BnP<63&wf=eax(ro8AGu@(({{b=bI_}D`l?DAOlrYr)uA*>YeKtG!S zbug3H+CtYbL2 ztnPAnt+5g)1nInPK_Os;!bNvxaGP&%`|E%~2!@@V9dqr~-ZVCF6wB6F4L-q@1p|(j ze)8TMf@KGignH_i1F~qL`I)%Ajov4P`)aT!$Xp2?J|TFceP!xUDJ&XcNW1H}x5$Es zB+(xge2c}<3G?x7CfOIL!=(KMoEZ?N5n689(4Y|;u*y1i$q#1&7EoCS_bc5$i>B?@ zc2+Oe&3N}=$#XQV)P3L-LFm9ukE{jfI2_Pt3*15^uB27RcdnrA z%;*jbCTM39Ozg(329Hz|e4}3;KIa8=PzT=A_CF~(E$6{^3$*1{U*buGJd`n6Xbo%H(>-Djb}z~r!SRfOhZ$M;57e?m_!JEjsF$GDMZ!#Cd8KlZh^of`6O1lneFJ1vPCWRAQ5n;S@K@Tyh(p#pAb5J1-aP}0Pi&Y5pjYq zLkhg5r*MX&&A*dnUm(F<46l;H8f4h%%wI;ujnmLVC+$J$UF$!VQh+~@+V!gg#9?Wb zOGz^z`gwkn?)mwt-;IBsFBys{Ce|CiE*+(li5no5WcwXPz*ja7O3|UPOM#FUNZlxZd!98N+AxX!neb2 zbV34*cSru~J(#;;nd9Mz!AHbi)^4vMMgDh}&~e7re;_0!d3p3t$WDXBxxQWWZ*z{K z#lq%m#iTHEuLoR$%%Ey%PO}pY>jgK|^Z+EHZ@4r5ljt|i*==1p{MUu!s9F>rsMhQV zqo$*jS|kCu#1$dOjipC*_f)E=t{QG*FWrsa8z*7v6CCr%o@Pjy7_ZtLE!{y(y&ClnUa)L-c7mfZ7)O}%xu+8(j z6DK$e>jbpsP(z1Uft`%d!@c%=zS=9c=38bN>UZIg{5`nf*v*05Z*h^dlOM@!Ii4x3 zccUV1?`d<_-4KAT9L5()mDJ3F4Goh$3CkL=?R!D`mCMnR=uLVGH@!9vff328JJ6nn z(QerUD)vfQj*7YO;M6%Tvh1O9a8ZPQ3Y*pQ5$k=?uwH9DOr@LJ9l50jBI3GE)_XSv z4Wb%Koi*}B_bT-3H{fMF_}jZ-2o7|@_Xx3h#5H2s`5y`u!F)*nNLJHpZ3 zYT5K^QNWQZ{YcXhNyVt)b@_nc?}gO#g2}%)lzAax5KO$3AE&%~u-COwQ7UO3$!ktF zS}>YtP^roZyOyb(i?-f_L|SEvn%r$t=`Y*Hb=^^TS^52!>qAFrXTsBHL_@tX&$rKa z$6@N4>z`zZ*XB$d9TPxPvu_=sXOwPvj?X|u5ySK|TPkm5DaGJi#E{-|Lj~+_@>%>2 zuszJXkz^q{I;mUE5$;S6NDYJmaP3n+Su8ZT@UUK|eEjrC+<*izo$q!YSR#&NF<(T& zt(0pm@$8%KSaLVj7nLOWI9o8x61d z==<+--I|eNC1RipMFrr2F3=bHlt`u$#xznhf*3W2_>Gw0u<>sAq#@0D+`2D`*45sT zrWjfJt1Cj(1N7P3avj?5N_h=ndH>mCnS*@XJ_*o~Pbtanz`PT1H{6OfC4bNeD9ti5 zH;Ez1U>TJCVjfz(){>)5he7{z-0r5|V1nl759WB=;+3MVNZo-M>7Q;1xXIpfYaOA( zHMib~%M!%_JG5gAQ~5XC#Ejt(KKmJUhF=={lDprZq>23q^NoAGfoF2EeEY$3XQcfTs+1fg zSR)oV?B&XY2v6LSsXuXcoFIj&7tW}az9|+_Wv76=M6*;R_&^pOT66nx;{RXsHbv_Z zMn!*PkX6cg@_RyDoJX@7v|oSoo!XCz7DrvGu*rr_-}8i+7?91b{ozQMan*Q%1~vwT zSc(^@3*n~wwja;b*u-khk}}Ah>+@whJNvA%tK<=q3b` zypt-`5!%mmdMVjq>4A?o@?YD&MCWHld@<}E;6wGbK9@|U3qO``G}95TL$6cYjgFWM z#Q(JFJFq`bL7avypVjn6W2f1P>xcGhQmPLD7(M3yU0(>>ltakb)&kTc_Au3H7Of2e zKk3)UcT8S>vy-S_Ojh6`{LyM?2pCG*4^tq=leW1~0)+m(u znkXd)+GisZrD}i3u1i+xnJ)h%Uf1jSOgxmq2e4l}MatRLeltWOPK$4w5F0#M$6hYf zjgJxd>5Ni7_6{FhT6t;S4WM?tGgdkK(DED)sr6qrenVsawbU2&la^7d;P2<`rQ)6F zgIn)ohZzw6F|vm+-SdwbVt)7v<^Wn7B+wNs;RA`rA~j@VA|WPC!{I){r|IKEDeg+) z9}7VU62bL5SDC?QiOiL}{dmx4gTXfDiZBWM!sBD-C#0n7>ee-c6oSI{ZcX(U{9FbL zI8Y2wq+H3-6(u2}Jwxm*>K&fHHaqo2d9-tui1;tNR=jfrJ=EH-_k@4N+90lP}0}uxbzz3L~fAkG4{BM88v+r|RnFF7|-R8ym-p8qJp=~{Cq{}p4;E*{%_cOuLA02Vt8OU{v z8$5S=;P-NHDk7MkXE{1BYC18t8aa=GE|+2!QxiJO?+O=7b)O-6-5KLcJ}4K2A9)lVYo<%^DEGToA8vQTmuRWLXi@6MYsy!d_T=-MUK|6AAw8OgX{htlTz@trwWpZI^ewsQS=#mgL@m{P`{u-Tpq0=QA(u4=KNI-ieA~>d%Ya0nnC4^ z;xY&Mv{PHUBA_u*-y{E4)cNHW?nyl10WSjfAGcH8TB2e?x!c3ED^sEXbR_#*`sJSt z*YlPPekq-p)^L;}Z)ul9FAD?7l`qAE)KxH+0%izr*yj71{D|1QGuCoirU{H^&pV?_ zksGM?qNw*HcILZ0ESej(e6gDgQXg!}QN$Pz8J#SB0`$1%;(64#Fs$WJ>%r#rwv>NK z8tv|=)}!qszG>g|Oc*CTG`RF%Z3?|2rl)Mb-V z4nQ190nX_knM_>ZicR{>iO_TjRvHjg7|jl=HG{Uf|7rF$xo*aYBz{B)*?ujs?F}KX zMY0M+dr$E9I#YOoK*tAbxL88&vo)GB;=eKXDB&i8zdLqqceG%0bYrxArJ&k?uNu;B z=$bI>Tapi5O`m=tSdb?x*F0LH(DC5|{pe>O6|K7pz8){Tut)XO62Rg5QHf}~q`nI0)tF+k5mrF?OL;sOn!sfC3Zu7(E&0=Uesv1X z*TP37Njj!-R%hj7X^(B)!U{?w`~!m1K(B1XG6BwOhVY`gJqq1_A1(D4;wF*NqQ-jG zh3DH($&ehLug1_&j?N)UO1$s4^J9CgyD|u;G9n-8wquvOI|w!m`M!S@`+x$aLyPsb z+o;2tM7p`2XsU1F*elf-*=)BTRznkGjDx){Q6{}<<~hy8)6@-*vBtb=gFvweV3m)l zG^h~7o{&}njV(6Sdxt5^B zF0}=4+7#ge;w8oSQj}tVDc4D@lvei`+&z|%oAScS^_ew{%%u|@;Pas)IGkn5@1}kI2Qw@P)*5D>;i1(yfsaURK*#)8w*|af$P5-T~P?<-{+ML5s6rV(r~8 z<2GeS2A}hcQIHo5X5$KNbiP?pe*Z=~3JK4R{Gsq^Ta;vNqJ#-58&8Y-e#o7nZ|YU| zS#dMrM#ubfrGkkIMi0HhKeBM8OWAIPw@!nZ#j2y7aMhc=5?T-p(&Aweoto(=|8`3k zI`~x(sgZ4vva0NhhiocgIubY`VYV$W-J>rGg6kG1lTNtsE|0nXd2Q(UCyA$$v!pq) z7GB8`erx=K^|^>a$yR{n?jN-7r8|76QIN2nBC`t7Y1D>=S1iSG2P8IN+CxWla)}0p z93!h`pKTI-OSsH=o_1ktu#rDH#qYe3IB=VD5ei&mf1-`;Hz-w-_*<_lWoPYCUnIdl zw(N=IioJU*_e^r*?MX;DO;csmvpY;0*#VqHAD}}ME$!?+0HjC#_v`&}dUJ9DD9V+SHb00}-n*(m zR`Tf52}Aw}H(0IRMza>A-gQI4(!p5`-Exz9L=fhTdYgNk>f#}eaU1$@7Sm0QEC5Cg zsl=7Ni%KY73`2ZT{%Nn{ZvYZat;4q z-_Imz^qAS3H>t2dw+pMBjqtbeOfsn;XZ-3vMpRx16_mp#PsK=UB#_PQ>iKTOToCd? z;N0)DnMm-X8t^J6NxpqJ4OH!+eYU4S1@?@XqWg}$shHfQ?d{~`Y_3>2drP+{(fRI% zEnCw6uu0^@Lofkx(|(B?Z=AWo2BqSKsxcyI#InF0s?Lz}f0{F-`BhGfIpzd;W*!l7 zd4l}+WgasEF_KQJr`tQ~z61AKlQ|;KHUHzroTTURHMl1yp;eB@C{mb!u%=Ld0vpgg z?GWB|)U3bML64yudXiQRaX+fIugIZZ0Gmif8;=Xe{X8TQXW(ISBnpz0fJ*WzH*vNN zP&_bM#CPDKBQj{ub|>z#Wr&X%w&3rlQKd`E6eUpJ!}~@^A#xbdjUJD&B%3*@XiO@< zk!l$p4Z?9{lhvgy3ug^~PW2)+{8=SPiA3*Ls`$FXgUJ3hA{yiC%w@z})HKK{cWUIl zCw292eg%x9cV@er8u>Lo=9K~XGh&1RxaiBaEoVmIKSny%;~SZ8PLxDq&U94I5+0R> znfLjY$TL9B$dY)7&oKrPkQ|m>Ve{PAW(l4P86X+qYJr=%%Ay6G?%M`F^REB}I^DdH zk=apv-V|Qx7lgCFKHSX~G<~b~S?i4bYzCqUcY1@r^9{>$1SV%aU0yFkAeMH#hoAqPB%^EVTvP;#PcaLeE0Z0m!T13A{mL9Uy*~&a@`fFP z`Gt=O-IT>&4IzF>Whursym*ND{vsPus7#&58q2Rt8PQ+uJ~WrblgHYJ{`kKOO8Zd* zeiU&-^qOxYs&Uep)PT0Z`BnS9CN>I#O(QN}EqMga^-JBe%XSaT25dys8C}_}I!CAf zU=|cF$tw|f35@tM&=VmUK&EId8`j4edG0(tUm>kIU`B_&1B0Li!v2+fln-WJ>MI{2 zy}Vw7NflJ8sJ{w+d9UQU1<}0`Q_vCN1ny`uPr3)@{<|`UJO-#O=SIo$1(JuKWx>_8 zH_N`S1f$Id24cIo6WWvbW(}GozOO2$g3@DWz65S_cv^_pZSqp!ZgrO%Z0REvu`#wI zV5;u}^Gw!_UFrdYRd0ucQ)idl0%V4+HXrp%MTHkj{_95-Ay$Yq)mA%)F!3RigA zSF-Bbn(jcty_Y^(gs>L9%eD9NR!0zU!hjYpG0Wx6Tf|U65?RJJXt2~|0l@PuQgAz| zMCyfP@G~092g@B-jMFa;oQ!tf3bT|MnN`;u+dOe%1Mgyc?Mx%GC04S5z?>Xt5j{$q zW&6A2s#H+#e78_;vzIH~tu}C&!d^(ZdPL|3XD1BM8vZA0lP--=n8D#!;P_RY&a^Q= zUl}nfaIq1$+0>ifeep}q0yTU(FGCXZo_m31W)n7auO&?f^q>KJ*7MJ4u+06>No7Q; zqWu<)trE10Zui2nupb+BBvj2wacJl|;fTn;A1``8Jg1aL^r8ccBtG554PY{}lZsx` zzhmWQLRUCaqEn+t__Zmd-QU>#wtq&f$UetU&5yM)W{{i>hh*&mPsA?- z-FT_sI4_)Sd9yA(3)D7?#70%3eWG;f^V5!ZG*}nE;-F9WJkz0E_Q=#k+y39GRRX=s zvTlCZaT=f$Wk8~Bh`!QbC)^nRScsl^h0YTNk(3*qNx~p{CL&Vol%hv3)v5t*FE&B5 zSn1xtwrXL~_jIv27IYSJCf$p}00)M=J=V+VG#cC&!BEOO3*nCBdwk`XrFsF)O_&4e zh4Vl+9+ZyQghGC%gmVZ>o%;j#oepMIq-2!^B9h`MU+_<~J0h(b)gw+Q%C4W=_~|bl zFu7OAAK`tOx38SVmBmRh3hcv8Ye@g*+t*{k23J)nKQgH#HjSSVruX0G^g5WT`cqG& z0y+-x3;*k%14Z<^5&6UmsfGdI{Y2+${mO^s7+vS3orcrgLOLZ@Ao z%^XCr6`=W*m)qYt4T-r;>V`DIngQr=5Z6a7weDMLMn2 z6e=sBQwavfH!X4mUlmR@Lsn_z0le`UJIOw9GZdu=RKcyIjYM}jBS8u~2^87>oFrMV zqKFo!CiBOWI*_2k{_4gqjBWG?tsfOe%Ca0O+f#u5F`EZ!SuUI#>eahLIoXlei}0Um zu6aiPZ3elOGa^eq+fFIrqKLcK+Mj%tT8>>gXPb`qngc~0spq>hT&;B~KEjJ<%a?}R zJgt+J%7_DdQI|y9o&6%&#^!E2aP6FDq?uh5oRf%U5VdZ?z&qIX#dEPib}ky2a9i0A zAmI*QNX#(EfbR+7zK2UkfHkpQdP%l3pV@IeO$Gp?!Z_8Vg~PqPsw@!v(pma-YbQlN zmx_F%P$DZux;R{F{v%r=O^2SSCIcQoH}yob&o({&>Fu@S1`ek1=hPqVRs;%}boKwk z*h}cAqFp2K>!?nzG>8C&CxA$VKPLxBl5mR3%}a&ZXn&JALK{$s7>>MFYl1gBXa;a4 z_RA02sNoTMY|_VREt7i2t>X(&B+kjh*A<|{W6udpx}(#9woFb1B4#-Q!^+mW?aKTcbxUJ*9aM3EsV6@NfczH5^FO&`)m5$^7aYTdrh z=^4UqnvQsuLVEFjh3-ZJ5C1~=BY(6E&z7+rFAo`7>JK!-Jr6s=2r=}>7}i^K zCkgUVwvo$N=?U7sT@|57>gCn%EZtB6JV9JcO$d{?yq^xxTgS@E$R zmt!-(BD;s3$%yepZ^gjjI9+w}Tiy&A40r6dnQYqtsKngJYN2y^62ar z=<9d7H+S5t-Jt2mNLjFOpZ9d|^W*iOItgU2I)VIVkrZ#&jQ|KicM71coUpkYs1OpO z_vcV7vnkRj5S{S;`H>tuw}-bKv{LY_l}BZ&s!^7VA_dlZ9s>f3A*a4hj4Dv>Ros_-7HdOM6~F_vHp_naCifrt!5Pa?1YI$av9Dq1OEz-E!T5 za?_)IQb2ngLFw3P2q1J32UkS!egxHFfoCZNzD@j9{qb(S*d&Pi)i2{^O$DFFR^I? zZzk`umZ+0qr|f$2vGklyi{$;NMB;;DJubB4z!NhmG!{Vfec{`z(Iz0+9`O)YPxttxZ{HTp$9QFGyB@lrlV_k1lF}{ zs-+AfWbu8-k7bRl(MlufS6}p?^SK>%N*z9WgLK2gq=0&2giWvT-hQKthV3pezSncB z9x-H#)Vy>}K$KBEFW!(MQ5fR)K!hL_T*6cwM-_6oop>7~>oLJZyo2NU)0+bvn#_|; zS|O#hP%BLQS{b&sx_7jA<#8@hz*Atns-7n}VHWK6ghyCM=A7v*T=rgjyo!swwH`2x zbu7Hr@|LCn<@%*ddngqSN=$Kf1Pm*CsKmNp{9E03;Kl+1T@xuqbdskU7US_Wu>~IK zSz6ePt#SkBb`$BE0;m~QXQh%U^!UmC_?y0W+>%e?77D6<+-a>gYiMavV-(|m`{!AK zfPs2mFYd{CL5nS%7no3piqj)Wu=2XDNKO6t{|2v6)~fo+jI9TjMzBmJyp#TxPSqH+ zY78aTN!Nu~DMMA#G7QZTzqj%4T0QtL$B;)bsRF?t8I>}ZWXkmlFXGN{_aawYjK3CG za{Fa6Xv+RA&3kOhawmCdH@=~?$pLx9Erg(i3&k}cQ(f%A3jc1ra9J~P_Rd*Y@)tpv zru@DWb$El?;gEi}adwj!+NhP5I&nI(Go~_z5y(?ZEsmNk>9FB8s9kbUoA3RDn85ZF z`EkQX7H!#FY}V->2tv!nF6DH!0^tU$+We>NJUO*g#VG`;FtSt_PCtL2Z3{gZ+YM(n zG--GK(rc|d;GFNVmDD;HCOKvmn)|t_@NFD88@7y+8>{0zASf!HAq+nW-^d*V75SL% z-0*YhFIV?eav`hRh%ilnPLi%Q!7{bz&xUQ@50}OsfcVO{8nTHzI6H?<_20Gp3hyuO zt`o%kRZsFd95)aDJy0_dDo^PxXyLw2gvmEQ+{Q!$usH+^ckV-u&Dx|>Ewl<5FoE|R ztC)De1nvC7#aip!ksC2d8PV@?H~tQ+1oIeV_D;vkE*=QzlMiYadzEDsn4SuWUf!fm zLHNqRc$;lU#V|2QJ~rkZQ0*^(PnrgE4-|G_zm&2HXs+m?MaRD>ZzfoCvPsFqnR$9s z#uUDzs1kQmGrv3_fF6k?a<9RW|MddKQRopu z^B+{~)|AxzeAvZ3SDF4kr=xjy*#f7}QyK1Z=6-x2a-NNJ&!)el8;1ba3q>11`AkX4QHHX_RE31+9MuYxiEA6t3(;s{)75H&eqY-Gg-h9fC?M|2(ihrqi3#Fv!(J% zUy2$>v}%;Rba+i&RvVnRP<-lsq9Pq64JRwuH+0|O3Hlv+xIHsY?)Rb9s`!FIWny)d zT?p#&&ZT9t3L4uDvV8SXW(-Ttgr(A+G3Yn8vO&2Nhfb{QoH#i%liSL(4ZdM+X01*4Sr_W;!ub;kX% z`BAL;N7B)x4WP4WqEO$pz8ec~Rf2yV)$*xXpgE{=gQX?_xq()U#LMt03jW7e9U@g4S_pd$wLnG0m#!_r#`{Ys43)bdx z1_2tWr0IP-#%xkp93H{C4D@!2;JtqF^AN(TKyek~YFUNA#1;J5qPvbgs>~$0JbOJX zoTU$&kMXo`v0wIx=`*n~UMOxSWKerX0oGGsA1Mf=^ua3?^X_u;w)2o*kD z?I5tXw{Q6R_2ce*c$}3u8Hz@S-_^EAw?lG)q&Xi|DKV}E9h$7_<9mxH2i9ks3UA1pT`t=h%T_2}`Y>v?lpz z2^uACV6ey@%}9+j7h=#lSH@GJ!Bo6|yBVu8oezNvS(TWvSDM%rN*G3AY^@HFwwN6z zz$6O(*_MPKl~pT3kRMi3M~I93yyOp}B-;-tV}`z#7UspJ0eAAbq&W-JvfDOVz=qN~ zTB$76-PM(Y@1!X?>K&Gq;<Lp=#&ss}@qf0wD(g8si#7|L1j4|2AOR(}Q!~0O4F+GJln6b$Ra_*d;hw(lLnPajFjGiPNp;NP%C;#ytZ10rI_i=?Dp zgn1o#np!6PMk9o)T&C@`2WS{OWKkM6Uw%sl-;E1T67I}Lkb_2*E460I*+wnTNk(N1 z`0-w}9%-c6q|ugJB!$;2?tPm;C>ZZ2%z@vpi$d-`L8!3K6uH?T&04RE7t1Vat*R}m z8IR4oi^$u2X3if|(jt3maJLqhy0WDlSwDvw)2L&79ja5fx24&dj=IlgSC)x5F}8C4 zv9xtVtteO&MjiT3Nj{eRQT_!FpT^!F5cK{v{Tm zwyVD6aa+MA$H!Za)e5Gu5KXtflWA6BGSpGSCQ8NO zE;~1_5}s^~8-NxK>!stp20i{dKFhwNHGZn7q-Btg69gwNStq*IouymCCmkcaDJ#Vj zM=HO<`f;(=g!dbr0J_xUWyS(x zDm&aXwwUJC!H-%KX5;`scA(7dO|RP}o4p+BxYA{EFBrU&s9sAtpe6Wu3{cLZL&_3GHadsn3^vlSP zbt$&h^{dWIY@e;-<=X&5i1RoRBwXx*a2&9ix;p)B4*chgMI%Ys_i&y-INebuze0M@qq6j zPE{>ZPG(i1^5Hki|GpG(TfL?S+FNr;(?D9WRe?ouhIqfkzJ?x>5LspV&{O!+@yp}Ah)&BR#i?#ZY0r;XYE5-e8HJQR4 zrC}iTz;=Nlm!1_^HlTiZ6L`5tch85#N|o>QS~|v7REbY*ACT9w zwh~;r(<+N9)B5Aozdu?QU|nE?gzsqP(1}69+u6otUIyIq0Y^D(yx&xj5eF$H>YwTy zb=ztjPvSGWrb&%y2F8|fZHzSr!`~^Dqt8t<+E-thQ>}{B_+>I(v9F{MkBDt8^LT&C zc~L24Yy)v*EJh8*DtKOWj^_$Tr{9d*Zxz(9LKS@Si4%}ntHqWGLK8j5zDg9FSt%(6 ztxn*tU;Ad^r~FmFke$&FImOne$^izFG!ZXsXz@-M>k1j`_}rY%qUU+JlQvM8pu?<% zaR#KVWywnTJ#`Gtn4cDH|e-?kWmMnnZ|Y22TNJ#zbJW zG$R1A!K-q@%V8n7Kq_Q2)WOl_3uPA$hpNwDc{1PGM(2kws!)yaZ6uQY^3@I<_scJW zcwJeCa*oAMq3@cn`_t2E;GKF7P;l(Fm`2H!VUp){e&HUT!?n3=-kh07&s1r~H6n=PE|<2M}75x9D|GVkxhmxC$(a$Ep-jW;eSEN$r-aS_!WUR<6R)TAe= z^N_G*WM`D@k6{z=y^NPo=neuvoayS+D^FLnsrA7dD!(_b#zzm12K39{y3rQ^%TBV) z1<-X0!Dde^lqV>^+%N(r4Yebs4CY0<$kq&@vI@K_isuDIFbrrgsK`Yb85zcG+KAkj z*e!lHF->Y3N;uk<3o8u8mIn#~;d?QQ!tH0(hwpt3&=;O~IiSKVo215>d$9@kTs65M z;91ARklCn&0S^RjHEX4B&a%by_XHBI5Y}&ii{dxA;(9x?f^0gDE3v=LzX=aQ?Iz9) zyzKv4Tuebw%k{6uVjEx6j?ogerc!}9{7e{lDrl=kEa&)be*XMf9w#oZ<>;W(4CwLt z7Z(lEF`{KEY#M>Qg-N^qsG=e#9b2rs*W56$dK9NcFZ=!ME-m2oz&LVoxMWD4ksVXe zg$bR62N3;?xt4}m#;dUyaeI*Ek{rCM$!q)BV=eVGwN3~W6b`oq+^yotD@(KNU>hec zBj>8nMA+FFY{Qo9us%sC?r71frA0yu>`V7&T&=v7Xg;T1t7dRG8!eO)x7wPLl6L@y zG4Jmt;1?hvRZs52R#yRVq{2`#%pXZrXN)$oEh=Td&QAC>eN6wdMTVHnN76RSHo+Hg zqV7BUh4lVk31_MRE@W)V^;y!ePSed^cmIU z^`>fUE3`MQ)e0PBBzFWnTRG|Xkg9=z*l+^l7RT@mW=(QueXWRZYFAjYC0m}S9k=gc zsB7-KVZRI+&XQRXPaz6;QAMcjv>JH0z?ogGbvl-+QTvDWKwD`maB zmA1W`_gpRPi}=5u4r3@B+~_3*&D#imB%UMAnlsAp!@sl!>1N)rBPw4j&I_u|iiJt8sZkdy@9QlnjA@+|!I zifegBCWY>hi&E| zow>b5otJ&q&3umYIulD7lJLlY!;}5)a{G@*Kbbs>atR@yPKPZ>_Z1f`tsFIfkY{F^BRuJY0YS%5WfgW_AXo-|VJyDD8M!=OGkCWnYQ(&!h36`fdP3V6FnVkp?bz(q zqE1NrD+?2+n1(A#wT+$~IhenZ*|Liw;ho=rFVa-RZz+EbXv?m18~ z*p0~de&%Djm3{rK8`W&dEumX&V=Y9~Y zV%ll=&c4h*k90UB3**168Fl$1igxz3ea?71n}6tY&# zh<1t2n_seJC*x(!du-7(9Jb7$b<(xtsQ)=*fxDG`3H_PaF&=7WU%u0?b>;=z!Hq}a zE2Hdrtx7-3&L!km42$10P1D2fR zmnG%L!!r0kM?61#f}{K?cUTT3q% zFuoV_Cz*JDcf-3U4J>zF9H8xEu-|xy$Gsn%UDV4FOrqCqs;r_sI6>tAZ*viP0|Z>h zGOTEBY?{3VCVV_#&ddU(O`IPdiwth%GE(Q746KY8iuN(1!%FB+B47l5imdJ254(V& z7Mp{M>qe^HxK$-L65znbv})-2GO2d1l(R`-k4jWj4HqgP$ztD6<6RY@>>z1eSPc+NJRoqHwMl{Dl!5BwZ@rciW}-uBZ1rp73?>F z2dtcS#_S2m=$%koekq<2yKi;=p5FC9nUPleGZd(~^-{xl0F+0HKpLE#IYIK+S-oio zy?U;HL7VGJB~W;^3#j|Gft&}u&wA+7)6;XmS^3gm>=}g;?^V>%9$}6uL_`&e0sl?K zsxJ%(oZ{0JudDli@dq|Qe1U~HR+KsfQEdlc${E&z0)YTrcz&El#i}Ecb6kIx(Y@u4 z7>37~Eq%U00pXY2(1~%$(#+9pxrYh=LoX4^B2{>qi3YtWeeGv6g*6Z|O|GVvzY}w0 z;YTXMNep6kmJ7OaArfsX@)6XoO8gzCKt1nxRFt}Fr57HYpXoiMWS^3z=c=R*8=%mX zHbXd_CXfOo_OO6=4& zybT0tL%sTE)N}dMY-YzckG`&F6AqG7#v|jIBVIx^{AFGsYgc;5kaepTsrXqVY>QxA*Bbvsar%krkFJW;xg{(UvycLE=F4Mq&v zhS0tqhRQtTk~t$kLWm<>%*@pP)rKisCqc(m~EBKU2^#KXO?l3#|v~9>eD!UR*y-wNCSu>5xRAu&|9U zje06$Ent^wKUK{ef^mNNqa*+yigE9u<_vap8&3Q~c>#zxK7@cFSPPP;l1m!TFaAQn z(p^FD4@mpcIk^Y?RZZ#&Zph5skC0J^Scuh}B*$X@4?1WnXmS4xk-76D;O_?Lpcs67 zaNNr7vDHYsfa9Ac2K#?zv3Q}u{OOT5^6g6jmIn`Jj%VJ?m-P~qCcDR}uKSxid&vmT zI7%5s%htQreBTSjNa$XhW3D&Y-YAvZc7YU~Hc7G~N8b8c{<@ND4_U{TnKjM`etwId ztdbVFsSuXtxTmKwkfQCeuq${^+8dYr4KSWmkWD31Hr_b0u@IDuRy{?cYE)KKJ0$6E9+RUZ&_}3w}|d%ha(qa_hp)jqt5Tsesuf+r?Ng_37@n#-An7=SZfJN0P*yx z5q3UedSSaTD$|gd$|UNRgD>`C|Ezq`voYmk#s*pXgDluyTe_TP5O5Er0Ptsx@499z z!`HG?0}2ANth~Q+r8uU$DR16K;N!a*`}ggW#(Ye^dba#Gr4|vv^l6Dk^hvNwB?mOQd z%sGNt9%k}Ibgz!S*L=JTDE8(FX5cjIjO6OQK?Dxf4BIGMz;Fl{JZB-o!oteTPU#>p z37Bf6+zB{jn%sf#)3Zer_e}tdn!65n#rjr5?-!P`B_LM&pqey5MH1YTQH0Z{2`@Rt zqhg|gOV#&UiC!)hD};wg*DxBl-yGmkR%9?KS@B3)mg8*Bav5age*7PE%B0p6%!SZ{ z0xHpuc{OQ{hw~h!L=a?KPV6u}W_?ttyxL8~#l-(_V!8(H96Rd?`5$I$roH0swZ&j6 zN|~qiJ=s!ka4(rCT6pMww!TM;SJu1XsSLjc+lALYPt8I*xTdf?hZ9m`QX*#L_XdYb zc}>4O?0*Ev)y9)xr-OE-F&;bWQ(63@dP&oz#oRGc`0+P68{?UQ1p*ipvtDmdkYL!uO==8ke+xJReGw`9gte&@_EOAC+G+3m7i12&9JOpvqJm^zA^KDc z>L<{|yvn$4j?q~BXIe9)Sz33*OC>HB_DiURg-2&x2Uq`gANoRDCZ2j=FZAUC zG`2JUr++#6U+dkktw<1+Z>~aFAeF^~=_tX!yt~cw@4`@az26*=VJ6G#rpM|hr(kyK zMliZ(92^ZH)c9INxbwFBusPrkC(%uS^*jF3Q!fL~7qR%t^{->&I|Pk^gD8n_g^1Kc zonm^}a0bplucd5GkK8++dz%2F1f~@IZhwFp_Vhr_hJJlXcb3!n>@Uh+bdX^%u4&<@ zsP_Zq_NjH??bP8Z9TYiEhDzIkUW`uw{XjEjlP?{Z{{aJwf21zlfN5oelk>uo?z6vsXEPw9bxGURu(CR}|ic+#t`S|7&#Om8*0f1V- zCUZROXOJ(e1z&xas!7>Q8XXvzgI=LuVGl%;nau8AukD?_)uc2IB~%hDl5XCjYVe*g zUs4)L0L0^Su&%L}AlSAY!mXR9PyV99H;}C)tR_Y&gK9E7Sl@)8T!385Nw>=>)X^5T z`K_iKqWypT(5U!bWJ&imI#KztgQBitWh#5jbnd{9#Ce=#_skxgD6*ygUnI*kUm6A& z{0|m6_S=C>L?b7mc7ULq%oR}8IJ8~U#9YRhBa0J5QOiy@gYTre6#Z2?=1o0v0FBC< zcl+}g)SJ8V(j(}vmMJEs`=+cOZMY%&@wev<6rgY#um0J3)*vctee$6lkDqG$w0#lPIMY6;`Q zRNb;I*#XKi<_A99ehVt4RE(0`+nBl)N&|8DNaO;} z5HTri7I~eXNlz!{Y$^}!68L(l0%p+p1oOFRIXB>FQuvQD)v7dzCHqQ>@NB@RCFD=)Xz{BCS}a($?{0%n`dUB{t`lent_AT=c`$GQk*6{jPNp3!z;1HKuki>OysuF9P_A&C||pFJA9=9%FSQ_AgM)-6A=R zD&eYs>2I3ID*>D`6ZNbBLSIhaj1wRwC}mO^W>QW3EFym3)ragWQ=52|o z_P*2%b5Ha%!3Y{59KZObaE4!GmWYNxRrqF?eRQLpEG-K7t6M~N#fGohgQLxfG-LMs zIgF>T16KPspsErS?2=#@&0S|PX|url8gw!J-;y9D$Ht9n^LtXxo#ACZ0Ty~RnWP+R zNvwqlG^KUixy|c1>3tsQJi9yjx9_J~Z(O@^)^T1DQ~j85REFU4;W*>b?O4v^5tJ#K zLS{)~{gA-o!w%Y8zeXI0&vHTo9=b8cG%pDgl?neC&un*6^068{d?UMGE6=6S=HA8O ziK32XWN6>^zCE|=c>iEQ>EGU^S;F4cHLd12czK-YC5=#ot_^ffo@JlTyf@JMNocu$ zD!=&wF`vhAMp;%`YcF&=NeDuBetrGX#$zno#+xLYoCr4dVlr8<7FRWGr;LgEHrVKE z1A)Q#(J&T^&2^;+S@)xgfV0vTnQPL59Rdpfmq*X |MJq^haqQN>qRlT>}KY9?h+ zhNm6&g|$@?QWnNkMh{m2ppK5Cp)wXa0)stwo@<~?lx?Ix)XPMdUv5|ylM9HW$xN~~2P9c!5qyY1(^to?n!}{oT z3X!9L@u5iw9!I>N>{z_S?AKB5L*ph~{E|S#-VKF+qv~UQ*tw;i3BJNUSC!R2lws1=4FM}A(Sm7s zcDO;78M))#Q&H@m{>oOLe5qtv828d>Gv<8gm&dG%x+*XW!j@qaICtm?A!rBF2U-E8 zI0V;!@rY)yNvLu;=1OmrH(~~)pQnsvm_XWSuR81IN*A#>IF0y}5>pdKZXrYuZ^6a` z)Yh(1r<1mUzFRQdR7zjnpf-}(* zU20M|p434q&SudrE-mf>a%EUE39#ZuMKKIj{=T!}W-pXgX~+Ec()^1!^mA;~lExWd znck6e_|&c&`1I!4{!u?5j>NL*r!NKN)-v2Wh;sF_peQB{N>zNUoWq!B3Ur);BmZ@u z1#62d2GS)fAm!lP^r@)UP%g z+#a4)D)sPap!BVX?@w1n1i9aR&TPiEmUq7-5X{pVt-`M{=xAw`0eE14(Ya8wZF@r z-Z@-WJ!zG)BMEZA&WaeAB#M2e|5X2{c}So`?MQ;b^Fn%HGrlYS>~tj2A0Z)@5~+6R>Bjr84jv*R)49z z;vp_;-ER~-N?5C%zdjvqlI16E<{_rm8=z=RkbQ`5yJ<|pysP~;_DZ{M$*#Tjv#dhh zu&V>r!_Yr(l@?lj%w#9HcLDJ?&9_zukB)U#XnNb*lXQ1ZgHd?FBwx8)N2=)x;9q8? zNF^JPQ2seZSRpOjJYNOKhLM!hjmnxDrG_QOJNgCqlRgHv$=BGodIYV58}~2~e(70c zMaIAtS4Y-dz2Q2SG(Wpq+d~|D9L;cb$F2w3-s$4+14>eZANuioA4vl-?M)oYtbRE= zuC=4wT{}mKef&zde1aiZTNruw?z8pz-Z9#xC5z2`k@#R3)uwRbY&I{L{OfQP{LxMaT00TlF0nu!)6r2 zm<@<-esib5KVOmpllh|(IfbGSZy!<22-{$zvl-+NS;pV;1MMif+;ho*Q6rP=YRx{) zRREBo{3692dywcTo7TI1(13z(D6|}zNhM%imk-IL*SYc;!|CX6Z&tKVna~*^WWi6H+; zo+Wz>jD1?D7C-t+Gutv!p-F?~z?iSBdRk;+YIbU5FdOdniRNFcKau}F*~@S)x)})u zVlRIwN-N1pEfX@ZXWl9$nt5#Hh}@kY1V5U2%KqhE z;_Q`)X&Db`kNMm1Zk;tTpBt$w#dP@8bAFieB04C#?L-Pm?8MRMkmuD&{I%sw<2PY; zG`f}CmcL;X2FKar@6ofqEQNo_yDF7;3eFy-evBR%;KF%~P8@)JF!>!|$M!(C62fVK zcqQ4G@=Hl06~tI5xhvDVQmLk(FHRvp;?!`U0>Zd{mxZ5nE-(?v4v%|td23Tn!E}r^ ztR5qL6X)=vz<4$NCd*CKrE_>9A%MWAefRT7VuUff$fpk?gxQceKBJg|AFUqPa80j= zuDg7k>FDjrVkj2Ayrp3LlO7sl{9|m!{m0F#?^MFx9wJY*+UQ5d3o_|=y?jJB*io{< zU0Lsb2Z_7iS<)VdPlUvVhH)2t%NFir?tR$O^*LA|6{> z_TWEyWf;?6>Mw4`GDo48yx)J^_Vs^4`c|y&JDlm+Gb^1MIgk30j#F*b5QRwzcsS4(y|xuR^Zs>0N*a<}331znG7mgi?Gt1GmW5nV0#USeNB(NDMwl&4()yBGpd z1^K12vg_OrT=-Lg8NnnNJ*HT&X* zX)Em+R)wb#v-RS*SU2Z+EYTGidjzDzzugBM;`_L$Vs5*&?xmjzu1Y?UM&YHanJ17# zLVjEl(eg0l&~9#@^k=U<^knWT&ys4T+UrnB)^}^J5v1Hw0-lUsh|ogV>6<^<`vP*B zvS2htrrbpvoZQDb8By*SDZkv?OG@kl$Vhl|F>HsJMy|cXdRhYV_6&G623GdGGOJ&+vL#uqpx1gN+TCB1%C9& zzmi7#bjAN|YV6lrnu0`bdOs2f7Gjdf>VjqIb%*cVgaN0%_$yKF`mSPyqb1P}!9dVH z+%_DCX!HOcnql7D&;Bh~$oXv7mMXBm>Vmi7ISVIL4Zhi~u$YtetHAYi>W9TfMvBz% zG&W!4sS7UTO+vzo^-R`MwqP%=HDcYjf>6i9VKN-+65`M#^j9lJxDIN0a1wWnH(s-c zR(Fy-r4nhI-xS1VH1UeIX{>*Y=G<&NUTf7&a`{{KnygQUkej%e2w`t4`6{KZh)6*3 z9{&V8VP9988m3|g_mB#`ne18+jlJTFF(`)nAxCy!5;LU$g^#=5*}QE+)70jyb#$Zg zyU&)!(L7LayRIriQP&n>2w|zTRra7Ll|VuMB^??1A`o%rl>G?sTx=wgi@FU!(h`gi z%_&e`ChbSDwECD1M#^fA5D_@VUde~LQi5$}le6WFll3GA5duXdVZ0-3umCD0u_dUL z>>=UTMx8!RBAkBtODQ{?ps9M6FrDl_I0E&$H3l3@AAiS*ZFh4bt1FD#Ueefb=YIl! zQl>+fxR*L_-uQ9&J3Vn|%-Una1IxvQVC4CJFRj;$14rQCzSIN#@mA`?xokyW@W#PM z@8!tEm#7)Vgr=`w#{jOC7M;)!*YEHrT`MP{v9dU{?S?8X@yxG`IDF&!KOSWa!qYdU zpGe}uWWhJ@FWy&qXkIW%I;hB9__3%bb#FHrU*U45GnkwL}(ke{Q@%uXmQF;jYV~V-l zdam4T2T+0JCJDuOq6#^PvsYMFJwgC7iu(KS80J$!n#}H(4{z}*uk)FMN%gvoY;D~D zl~Potf{<(XpJuaeTe2nJ9|JV*@z=k<)t_&}J4`-7N*o*0g@cZhebg$S!@Vu($-g?X z5f=(IE*+LnPOCZC?_aV@;*v_r6n^#iZUCj7TOYb$LWk@mTFW1;0FiX_lS{7)r5*;; zJ%=Q5P~8>3MLJVVD_P$EQO@tBVHm1%E@T$dE7=kW)BNHWYcTmn6r{L!3YRN7>0dL9 z@v`x`u@i>B?IE_BHU9L*QMl+$c6gw7Ys2{*9lZhSTg@jTUSE2u?0-vn^Kxa-fkLyq zKM%LxW~69GPjZdgvBj=sU*UR)Ak~=I!{t}Q>*t*k$(PBnpK9X=9+wB_kK4_|01k5s zGjFzf&ElrZ8TnJN|8~-ke-Ih8YkpSICI`B|k0!{!L;L2EEW@Je)-WtO#LzKx z!_eK`IdqA13y4TdNT+lR(%m6~q#)hhNJ%%+-Ej7I{_u+nW`@Pyd#(3*?i(xjbM2z- zatr@#Rca{%@y0V=_leokX{E~}O(yW|3Jla^dd9%nU0k8Y z^n&Z`T%$l~g=k+6U`I82k2ot500M;cXilyn0C~H`SFZbmnGlb_lEwlRooCQhkU~nj zgWP8&VQVn>}}Yqw%N zZ%VjQw-l(shYFm+I61KNDyCfRqK%4in{RDMEdsp^owp_`>oAzKrv_*T!zjpHwSPSs z`<}ibBUbn>!$Pm1+S5}ehbX4eyZ5#6%V87{5s)c~6nnpV+HI0vZ zo5)XMr_-VcAP>HE?a8_(R|0qj@68Gq4q)4VAJ$FcuozrVH(G#w`B-qt=i9bawFTg=&B<4_FA zxzPVm+;j2a#Wc^z`3($PMIPljCAKQ5`F?met}8@y`E1;@ef=;Ysr1&6$!j^H;Dh{O zVLLq^0%RjJH-`q~n{`XJ_64&y@;M+uMp#)@;kin@|LLO$_J$V%T@+!0HLU-JLk>Z6 z+&RIXZ7}HQsP*>yzbY-v4L%s-r9rs&0xftcut=NR5iXtj8aEi&7%C~ePe*q42)8Eh zE0&0C!WorhWYzTC5v`*~&|J&n1T#=`#ung0+b zJxwp%pj9iG7bhF&zFGHvw~~_&kCz3;kx3dX4taz#9!{I9GkPM)=G-j!1W;BI<{&2;l7F_YAtVF(r~}} zu)&IxJCg--mIhilgmO$qrW$a_f`CY=~ep`UxuKbQ!0Rp(Zw2_}+V|`XaQXRJ$ve zTl8Nzm`7v#B)##=D&n#EZ=KiXM%fIoHIZ?l)CK2SnM@$ol(*3yK(FX*QEm=`3bS#D z$;AYh;(<0`NvU;yw+|exO+J5ywq;|a^aR+Ye>Sw%624b{i7v+YuFSB*PEM-Sw70mA zG2iHdyYl4%Bbs~?V{xEaI`srP8bdHn92>Hhu{TqRApUZQmnqSJR7-!;znkO7N?xd( z#7I*$EKD4sh#ew1S9|(X2+8ql;%SWtH6JFL4HxI|%TK>v&C__=3jk~QYAp$53E^*?;JXgY zpB|bl$2?DY@h!G!8eCIm=ND4lcPmPMss-kBQN=Y)D+Old+#zZqj(*f!vn2Nj|FD>s zn#q{!h>#T^L^1d(loJ;nRhd^JpA&|V)nCV8(?ep1q}NEBmCzw?ZpwwfK{TVmSPY#J z*C2XN8mcK@rg_dMv$M@J|XvXVa*Hj3}f9l>{x!4KH}^XW@;5e6O;}NBR$1E zUhc@w7wGr5UZnHFpkd_xeFoRK!!0^yS=SqtOEwpXZJn-Qm=K z1S3A+5aNA3igM5EJ`tO&k+rEvYe5KXlTw7uzErMv;trE$AOGmjD$loT@wXo-Wtm${@Abo|e@p$oVcC?H-Gi!581} zyB>ay&%MxfR!g!HF7w+Tm@J(%whp|`JGgcBbNIWC3&YQw#bf=0EXmwCT@6nqnytFN zWX*2smstJzZw<}eAI~jsPFoaT?^?9+__W`u{a0S;y$Gsp*R|H4C8Y&Bn~(_@mUjL; z;$@hcqbu6uJV_<^Wn3e$py^H|+}NJ!G5qbGSg~R1WQQ~n^h3qi4(Q1+IL=OndZQgOK%~U7#&QH+Xj^XRJH!=y!Wj0;lyG-08@|Gn~`9g zoEr3)R{BzvovE<~5Tnnq%zTGc;7RHdcN+K1sX43B4&n6v?)&25Qgr=2gIKE^ex+=Ck-uhn+k(Pcc>*zo-%{2Vz;gF7 zRs9DFZ&RA99+h1&{E9idmko=up71sFcq-rZtrMyR8h@~q2=Z;TRF=y0jjl%ljpxB| zzzB%)*(I0jz7w*cmds`%pZbN0&wPuJJz|+S&y5H+o$vIU98Ger4wZq!9f7!ozR>tC z=)+V)_8iZ%$$Z}l%hN3?qXIp>U9Z=d3}MrDtLO|ent?!u=ko0+YTDq*iXZJzWK>PY zyYGbZv1ow>+PJfyBI8NEu71Bq30h;RJlm55Ww7CNL@XeAJF(JSE-KZxaiX1jn?ist& zbku4cKf@KYDx0BqVVr(rcPTIGx@*3f^L|KwB!0vp%A`;WpHK2sz$U3WL{OfypQpIe z)?R+PZ87iXOqA;|7zz2Qz8K5(=XqR7*=knM3P&`^hZ{A8Po1Vw#FgNVt2<*|X5LIcP_$G9@ z`<3Psv)&+0#aRDvh!m8@r_!t^H`WCm9-!w;o=)_E4^!)+<9=+zF`9~)hv6n=^X70$ zXGLA9hkU_*x#p~fl9qsuLHyfI8c}{+SI>b~)K9-V&{A^^24E9P$|#k&ug9BfgFY)L zoQ>N^;p%L$Hiii|)0=lptmzy{HM0kDL;h7Dn%}m~7pM$FQH%Vh@eresBN}be!d(eB zoenS~MhjA8<*ucucjQpGpG^=_7kx2F2pxh)T|KDindwJ;PkH!I`WrH(Puh|5wmGtX zx^K{f9u7uhQT!Zq>%b+kG-wERa={uFu89$%QfSAlZdI+r;RMTy1F56#P!^ARdr2F^e z^+P4!Cb+KkxR29Fv7QEM-B#)TCB-njH^1z`7Ohk3d8AwCZeglNgkm3F^qY#`wnkzh z#h`>G^6-&{b?R&h&95}1h$CO2HgxRi;+gyHBuA@l?Qey$^gHAN{YP8l-)P0|88MPM zb;M=baIK$m@lm=(BAd5k(_seqO&Do+P|gjFdwAPZn=7`1o>%c1lcGPSvh^ocrDkWZ zS!f`M^!;V3l1`~~Z*54;@g((M0YPGIcjWbPpjSJw40dd&>G%FbeyWOPqeHwKh(r3u z>>w??2huug(mNd2fx;#l{IW?OO0Pvvn!)$sRV3jP0%H43xxxW^LaK)%=s_W2Rcctb{)b^~)z8 z<2AGjBf*^nSo{<0a&Jl2q}KP~cdAfzdvuMgc0sTR;ob$0tf864D*r9uc`cOBS19S5 zkYAArqGI2acT4=-;?5z(9<(%OaTs3ddD809TE@pM7OJgUtOrUi5S7FC8zYKzDt6;Q_g z`^`%4R|lzlpS6)ewP?!>*>gWW+p=@Uvf8K(!FH9sGftw`)|M$7a>2q8 zPkC=UU3Zn=BN3mDI)|*VX8jrDEG2MhN%|lkepcX9rF2oiJBRNpXx(^2Clp&@DI#$p zJVXSDuu8T6ZPp|(NXd;+aWW!4g@HCxYO?oM7_h6I4UBl=x=YVG^;*SoG1Xx6)eI!N z?7iz7NuI>#-=Nftdcn&N`I>Dt0)DK=5i4(k)-1kd>C3IW@x`J@HMBe_57ra#<#Ffu z`*pu{2Z*ChgjqUSZ8|ria6qvP+MgV<{U570WQk_8F%^Q-OoCp`r_m!9D`|VTEDRyMnR}pRU95{sL&Y!CP?RoT z`$5>I9XFj4z4+}*c21k>tPM9k+Fzd5$_U{iD?ayh_$YQMvkTIkLr;Mb9mluZKWL=C zw)?MsY-~$O z`_SmB&j%d;LCW19KhoHomM=Zil$m*I5L!$+h~@f3-@TD;_QLGKmrVyVlZDbDWRzVq zu#HDTIZ=VUdRnLnkUa`TIb;8%P8F4bF%Akr)Om_9w8SEn8O!P zMpeE}EJ&dGSz$yNIH1V(7r5$j%Zt8V-I5Vd{KZ)N`#pDKLJ*0fJ^5^t<|ECm2#i2 zXlheux)pr#MVQu9TMAbtA>iDxjdAd^_NZ)n~x-%2dv2a=>Ff6*Af2}e_HCj&mY4$r5l~+8%49Htr4^rouubufUXo zsIU{Sys*f3Pa*8I_}33>dYOO~kl;X)wfP`e&co+tUl*k}d64t80P0EtSW(Mb48IC= zeax3geHux~5|*(efL{MXuHKE4ik0WM+QwbD8cQyMUe6(<1y9)6i4JjPz-fNZmT#>t zeVa>xLOMgPSxZl4J#E*MBOV6vChZ2lApC|%W>BFX(PZ1K^M!C04!61vHX(B{PW?qZ zX{b@S5L#j6nNdwgPlG(?6!O+~BYz_2d~phohF?N8&i4TU2}r??>x(xkk|PW!gbVQ^ zuGnmsk4@S#Ad26BthJ=`U#)?gQC@G5rNM2;zk2ja1OwTq_MRPOgG;@F(AFw`SGeTF zWmu4_)|#?OP>;+g2hFCJBycZ^z6G)I+sF2-e)cIG#L0^H|8W@eMb&4rw1_et)kWib zAQtoEeoQ%^FR;pV{}TSOyUR7AYn6D7S9L`}1oBpR`#yax?@wd|D=Wp%tB}z`09LhYJhCLuOIn75(}1Z&|C#J4i5FI3_;r)0f(h{u+ZBrbQ%jcq6XDB zEb466=^vf{!lJa3zjV&dDl>!!3;N~ulH2^4;q?UV*RoG51d=epJNb5b_v+Q6T(RnH z>t&`*05|r~PGCFsLmgZA23`85Es%(x9^^_8V!2$bxIfckt1lQ9 zgDKTqiD0|fXNns>Oa6w_#V6b?KuvMzqsWj2b>T9p{=Kq(BTvc`qJT8YU9BAMmpDaRx@ggDhhg5&^ccW4RkqRZ={N6@c})#KNFcIv zr5MRHY4h=V1g5NE=_#b?uS{jth$Gev+G!gN^g(mC7@d5*Hp%|Xdr>YU{8Esh@^8{T zr3e%95_zDmtS3PRs(-PaYJv=?-J;?kEt(z-(NKeZp^cs=TZmcsG*ZLf&jI8Z8!s3S zwRYF@!x5Q=&2K_s_KT#Vb$dsj>1F*oaQ=kgo0^FwOvi|JSIjW01JfM9>{AYi>j3A6 z1yF#poybMFzd3QVeS6l@nPa;|LH&qqheMRt%_l;BUorjS1PsH`=SzEwqe#+d*V9{( zuEjsRr&ab~>x;S&shYBQ#=&^bN(!Zyl4Jg_Rv4dJn)r<3a6IUAp_sfQj~+)tUXVmD zYt>d=y%CC{5(uP*&|~aQas|Xt(WfL6=jyT{xn(Pn9@8ggt>1d_6;DUBd=+lx7%Vp4 zV);~`^PjUpf+WBJOcwE))M?m78)__>9$1MaV0;b+z=o`mFAFYH)oZX%%j366$jpoe zj)OOVLW4iuBR20!c6g>ZfmA!8uPkPdX_-(ieV*ET{h;#XS;{%=kierL#t^K7H>#!B z;yEDRPr1H7AuH1&gyQ2gis0Q!hWRDW6-C^G)tS~yxt|FnL2AlASDG==d|)=YGMlJ$ zZPDsmMaKM1ie(Af=c7iVKqQ_PjrT2I4^zUZ80E0z!-9NnL#-32YbXhkE# zH9Rh^A1KwG0V~xKh>&ymhOm9I;-x8xoy7|=@ZSRmVLu1I;DF2|-^o_#M>ak+V|$nh zSj<*Q_MEn%R2qY&{tQulxwhJ$f9z2_pBpEqN4&Gqup0S{cLk;6l^CL`8H@_3!d^7w zxSUK3>T|#KC?{l&ai7b;VS3$W`(TraEE>ek-&F4srLfB9x7Rv_b@varjsr53!gjPV zJ?qLCIv9f#hkgw-mQn&UMqXW!-*JUC5q!MbPWhh~q&NVu8g)hDv#hPJQ$JuIA0JQ2 zMqHtjk`eYxLOXyO>QBG~m70;!JvoU3)DnWSvdDqOLVuv1BhFUu0gQdS0aeqs&@YGO zE}&^x4d@I#y}gHp)5Slnwi%mrudJBh5;aWoGV}84tXw3qyvrTk9J_Mp4DbhGV}m`T zw*Q1CyCBJeae@KS)gQ~89=A+^r|6=}Z~s?3i`s<$;+uTODz9>!Fx_m$Q6B!1=-#`i z*X?nN(l!~f=eJ1wVwgV%Qk2Ru08!KyG9WB4S!Cw$uk=%YWmsy}Nnlvp+6qbM`=|t; zs4|scs@sJ^BtT&zs!Cs)o^svNWuYeXak=dvAx>NDQZ|%NZ`HAbwiTT}b6@&|Kq%FF zj_1SowlDH9(D-J=siCmO0>Yid{a2$0f>A=UBeWxtL-$ZSCs|)qUsPH0ml~{tI`{kE zm##Pm^L<}0MKr%9&Cb(h?-?ROa}as8+w58qQvJ`_}HJ*!xO74$@nhxRPV*>~`f z@fn7k(f&3E4!rd0ZPAS=DC8~nR5L|?qIV;^cZ|jGtZzEKZWb7Y`^sY;hx}3Q>l*zR zj(y8UIVu&Qi{d60g^wnmXF8soDq2ivw7{^QHz}I*px8xspyx9&rg;*tVs3(Rb4hO7 z-t&%QPoO4A2v53zTOeV|)=lpVE9fMGt2h-g@5W9F5?znO8gs~bqA}YO+kF0hI63Kr zY8c4c#R7S#2jaJDZ&;0g>tuWHj1Wf?^T^-RYa0Iie4xAV^=NW(@)r2SP8UB`!G3_i zh<{ND*;#iZ0ZiD|ZC_Vy0O0vt{N)%2^j6>ND~9|BRT0oG{Y)nX+$xhREHlbnrwE@N zW=P5u(UPDolAy2>#g|~;odjE;y_1j1g>WrE_x^OE9wW;Z?9ro%fjO}Qn42jqW>q=tYr#F%-(N+y z?*wj;o+xXbCQ#wkUmFx$IU~<3I0nZhN4Y{?pAl(t_Gjc@3LQo7g)G&seZM`)_38#_ zKwDkoa`EY*nUWS3`AhYzCM}H+ID{uEg+LgiF7P42=b8^{|3H-RCW>mW3N_q7bhjNR z;4lD6eb0{D<*?Z?=EWHQ;(F(;PPY7}O|2w>%#ZX})lmnbsSnMd^v?$`vV~TSYx6Ym zYRbpuF)Z_-;UAAh56TDk{uB5hs7&9m^-H~|t_p(9NGf$4pz~wp!$8Lu1)4ZajofaWt7QWN zKs(P14`8nf0lbNC@-n-Jhpgizvr=BWfT32iT!-l!yRkt!@zLL3>eqRuAU+T;av==D zNBmzUCmpdtJyQ?|Fo6yx<{5sd>3xZKQE*4)JoLrzqbX03I1LqUnZZeKO#Q@0p*Kc5OGGlg3i|} zUf3zG54bK(rE^dDxQrxlPHvwn#Q*q_0U8x5S>dsi6sbD37O5ckCC4o!%@{jVdb2PB zuB?}NDZH+JK=I<|EmE7l2bT49XV~|0PyO6OG8bwc74mEW0(K)tL>g3~ zD%oO3RnaP<`>X=G6GYVC7r_esPj{Ypp{czVPtTKBFJ0 zq;v+I{p%3sCliC`6&uW|iId0a97&tFzN?oY{_{#++DHs+*MPl?eFn;n^TE%OHCF|9 z3em1F|J+AXXR!L?esoSeAD#A*Ru#BKN#ELmD!dr?LJ6#M}fs)EqeD zNW~lpf~xSL$s1YBt{(W#fu0tb!=<*XDfE1JA$KHEmhWSJI0h3x02YI~-4YmP7UC1OoYO}Lvzq6R>S z(vd)vrUpZ?P>qa?pi;T~!F~XB-fRpP=u}e?Ed`LEw^aeY8eoFe&Ow{6c{&<(!l7Cw z{&cB*=Wngs_HZn+Pbuw>iKM3?8Ptn`e{b`6KFmQs&d2_ETv{|4j8;RXi>{o>njOd6 zglJ;6-fElga_RdP)`iybGX_D=PZR-duMw z!{QtN;RkZOTTT{vhWukW^b9auA?`Ori{NT?^%~Xyy zJtyQV`r|kTR`Lu~mQn&@zi&={{?u3W@Y~bAIX%Y9fEswX^IXW1>|F(Nl^F)XGsgbZ zx4#^8NzXcWrWB~Vb|VjE;7-qbc4BE7J2`sFAzsEo?Zd=ejVul(^U+-r1hRjWw%4qj z6))fV;JHx38XYLVasUDZW5V8D8nH~S1IchVD|UeqjN`yG|0?jHFkm0~zH0T+F7D*v zx7ls+E1zLipCckTA6rUL0H(SVA8N%>%I805u2-B8yvmg$TvJR5v5%0}mAk;vPls@~{FH z>e1I94ArL#0sft8I_AF!zJ;kkMAXtH**uHP^CL^e)6FB>OPnQcGj}V zK@@%yuxCo17RGKnwNmK1gRQvwNH3Woc8I!&kj31e5g2Cyy7RB}*g;0dt`)LsZ>`jwNP*A5(f!i_NXv`j=)FaSjIbVtPn7hEOp#i(iSVI|uR{7es<-Im; zBtK!+qh+>3DA|>jhgJAc0Z<7aN;Ut+lUWbf5PMTaJS%ip#V=7*71#E}re!$k@6qvn z)m2L2+UXInjV?rbf9}0VL;pLnV7bUqrV=n;zb34OoJs#X#5v+H`ND3$!9&>FgI%5b z{+(IMqy(9q^F-W-@;K*p>A^i}urA8AVBBpk>HR^UL;$@s>wNLppu3kWF5|#2(VbiY z99-!7(|dvHjg{J%{51lCxLJi>p6WGWzj-GthRhu9p~y%ha5|pTERG~}0HbZ_CgB7L(J2(F zVJIn;|^(=H@@tPS`IPNKl%5MSru6yB9&|v&~WQ!*XD7O{mmHye19T| zJm>E(SIjLfE$D1tPQ{t@SPa0~julS>9|xkLoQ{a?3eWxkBMYuz=e@TnWuDU$9Rdvg z{o?5jc!xJ`96x?!IZ%m73xjg*DeBO34_`1ZtYkuNk|l`&GF4olCc<947GNDCrptX- zWuk>B;jq;B(&AZZ+HN9D>)gMl&k_4DT5zhVpim~i?8o%D&x=wVnR?r2kM*tmk5NR0 zVh|O`=x|4kG-RkfC7J(M)?Yqs|Cheogk|UZH5XqLN$p)`Io9y>@MgVV_OJb3>Oucr z3=y;W`?)35S#z@+`X|diy?HdKB>Jzl!aw?PbCo2C%fd;onW{B(9IL(gYUp@BFUY)Mo!Hp4+oloqic0KWtCz>>=v7frlzLu)Od)bqQfO=HYwYYU#zw8s&n{gtAch<>!yzv zHRa~Z%4*Bq)cxCZ=X5x2R(o5yAsU6oX~KfQPJ5=6bCw;BHcL`yo0`n%(+Vq^CCs>;Y$J{(ZVJDP$#2eB279maSlW53(3B3@Z2(MIYmM;b_!KXbazILgoT(|by-$D z@uNvB-S~+n|6(^zejoQ!aAuWZU?B5*s&zJh4ICI128t9KSpjHcrccGDLbjWEpL1d$ z@L;qm+S3t{`7hO64j@iSDFZ0o!5Vpf=*NzN4f^nIjIUNRIh0ZUAStsl5e|agirVV+ z!gvt1hU|lh+*NtIQs;K*f<>!SCwU+ABX7AO8$heSk(3v=@B{#U-U--W(54wKJ8QgX zd*#TSw=a=|ZU%*vjtzFi;gU0MnfaCu=zg`R)RcU$|X;dHxC9Ua|T^4rgI#uQ+siNAG1XS+6)#Fh4z zDzAw4zult>OMc3-60QWBaBxi1U+9Qn-)- z>Bt}CNS6+5{?mOJ&+_Q1F&n)>j08i4KMjE}B8|Iac9H30qlOEi%5eyoppw(!dApiu zxc>~|$7YDWia&y-5EXoXahrcur6udnUgQ6yZ=xnBgIqV=9&hzvk=YY(s3-0ZAs8dM z2lO01UTuvHUDsX}OHnSSL$QIZEtT0&JtWzJErMYkRq2O<>(e-;0C`bMg->s)72*R5 zW&?@*`-p#u(*8UTV;B<-GrOh}IP_hMBTg}L?xV*3dwka=&Pl?<@w;PW3d@FE`saFs zz-)xHY!97f01@r#2n*5YBh5Z_eCC?hS#(WK+?1~cjI*~NQ*3glr3o2#V$iBOk1uy^ zt4~aIct6I%qH{$o2n`K`YqkedC$W%VO>% zt0PG6Q6LH-MxIut5Y8H}xBK?tWeG(a8c5p6W>6PDxX(B-T@j@@-{>{MImtVoKL|yW z;Q5O9Lk$U?h^oJUX|BcV!b%cp$g9t?WDdG#aiaXgWVrZVHhDzT<)Dk;(V#8WGe^9m z?T71?LFf#hgfULY)9G<2Pzoy%jXiG-YXANeiM8TS$biNAXBbau6#|rDL{v?_7K%?h zzX$khmbtCun@AVRP!^ROF;0GdU0lQ<4(93fca=u^^fPNRvKQ1}o~9k*F~xZGX=vY* z885!zs2Iq?CEr5Gzu8X9C(^5dKu4>4_C4gmK=E|d>jA~e&Y}K~x0U34w zh%y98d8(>NQtlIrA^L6q-srssa=1|k_nd*wfdZVaF8(?=bHz-MZsf42-n1Rq%i|TP zwK{QtFds9_qQ?Khr{KX~7MB9lmpj8Rg-fs`sed-Nl#8!8YwJA1jy*jlZ22AgvYE1` zTR%l(1g~%{NSl*?`KF3bZRmV0u@2m4JzeC#|HOPO=wl4kL zNO~=lL+~SwW8h|4d%`Uun@rIkq$z(qw3BHsi`&?}_sJi(-cIM6x$^h7OmVuScIHm2 z{3sW{8F$qj!qG`5q{gyXGZmUY+lLtZd5Jez2>bkfaCxli!arg0Ytfp3>3Uz2s*A() zxOP(_6g##pqEzh9!5PR_UAt$|urv-0sp_U*KrE|B2-o+`+bxqq2Jq6%cdepGUP2mh zt^_k5;yuJg*`Kfor;d4^CR z9A$Zqu$@{Lx2flrpJ$2ZyuvJhQaD16$)(9ALBc#JnGUWRwiT_e) z>*hp|xIMI)$H>)jBteU&vOYpapN?;>D7Ru||(y^41uZ|1`r(0tugZ%Ssj2qT*VvriTcAoUXD%h|koh;9egm=U z#qq7|tx@tbu4DTXy%Ii0^pB6l8vylMzSHqShP3PvYa^6D9@HE!_}Fle$ID~qHA6vN z4XeoxS5-<_SL)t`TlM5aeqcs08T3R5&;GqtK()Y*vCqNM{FYoz2XD+hX90#>0!lV7)C)(FB<+&dyNtjwe|RqSq!n;uw2 z9{fFy)@)XHZ?QxFA&Z#TE+h0nd&RGp=?kG<%mav@-XWLZf|3_qEu3dICY!UF0G()89LLF={HQ#ZdwU(PhaG?DVE?r`iW`=!B46^N!_BD^7;1a0ldFhJ>hm(&ymM-ySL~nvBxP~LZ=@(!DfNTUgiyP6K0VeB$PTJD>-fa=QE8v^I_vgeS!H%fShi&BV} z29HFWoZU>>YW!l(q$GoLD`ur9{rq^L0%59HohcTgI(Th7o~K{nAEI){j&x}kireeg zv-;LA^eD?0eu=gBhXg1g1YHf&qnghmILs<=S?j&qx6yBy8J}-(NY|tyOoEDr0q(|P zaZV*}YFxx(&Jr0)G_kp~G$)`MKGpOd=xfx(MyNkO-I$j7pGerPG}=J2xEwoK|3Uf`fk7x( zsSdErwAfU@us`25i*&AtY({5mR+FeKl0!Fz1TuDGUeczR!<|V+&xd?5M}XoM^ds3~ z>nl^A%5fw+KR?$+cY-r+r@YPKv-^XW=?B(%3Wa^RM4sKY%Q=}HSg`NUG-+Xh$sKq5 zfU$3z*$mMsVi7H`Y0qm@Uv`tOd_6($)E4@H!KE0=vA(MHdD%SGnjtvspCb!)DBa#q^l5@sJW zZL$@d{aKN0`87@XMFNNI9?{B+%7eYmVuE)pTny>9v>RzmMn5f7wz!7RlMKt7Yp}-=4ww`n^xu zf>poM!%D1$aVMKRM>ncB=*p@1kUiKP6C;9w)2iy5PZ=zs9@fO$XiZx2(QMyK*1EWS z!&KUq2Gd47H23G3kp)X!HN(icp>&d*P}Sv;_s-5Cc|rpWzt>V zq^J!q^a2RLBCt@wyD7I!Qs||l85C+lOADVU@kU!Pso1+A;w!9Oswni4aRF>Aya@H5}Q@b z?MkW$X+l}y>{&+061zN%3`D-2j_vUDkcT;UMVeF9K#uLJYr3dR866XWcf?itp|xdwm|c%U1eN-d_I5M7VE1G8`e+Zy?eOQNdDapS4=6;zPE6 z_QQg7zjr(Qm3(xZ#PDhO$FI9|m9!(Rjz({O5vvqp=qY;qpkO&7_pYV&HcZ?F-Y8!S z?DnF@^mu^y(Mgw?RK{bn9-M)=p7F=*e&SY8g5ol)eT~o%m^`YpdSiv|5oh} zt50te2(mTX_B=k6LtjBr@vp5^2$;jAYYUYg3^sG#7_Z}9g=o408B01cA6$)(!%1rtSNHF=>ms)QMz)>D8BsUg6nOI26Y4az zVKAv@0ErRLLEW7lJR2a|8hKkifU5^7a{I?RZ zsy{93IA5p{iBx=V5ojCOjjiZu*$*y!TDx6e-fE;p_6SOO0>fB6$+p~APG5lGZ-}-n z*>j<35cR^sl!*8#N(qx@5bp-xnYDCwp?LAFDutnnBAv)B~I#fhV&lg~9`ijD>& z7oURojWnDJR3Ive3+-(!xDiq^wR z%LT5yLDIOk3sSWXe%4gPNKxkTh;qf=xK6IARsjZ~0JU2AtF_73MV+DOeb{UAln0dc zoR1HUILorZ+KV>~&j?FLv75h4l?TMTkw4Ecgs9fRy)qh`U~tJmQ+P<&7r4{vItnv4Qd7?kc#J^I5H1+2GW(1I~H$E}Z6 zEa5?H4W^N@qf8(A{YpMjOhbQ$Bd=VoYoZ_jB?v_&kZfIiTZkp~H~I6LgR4%1cK1)W z6%=CAM&*aFQqz+183<=}#AsHl?bMp17rVz(l3pKrmJ;`WOsF#B+v|)fjoL4A!1MF0 zslcq3u#`Z>J5_pi5`p|?bZ6bO8!aNf##$)H4B7muL;!)RT|U6`(8$*_HZciQZ385fMs}-Cx>L2dDg3G9Drb_hu9j zTlL*#+us`*M~A7HD%}$t>PZ;WD{VFY`TkbYsxcJ3t7}ur*FgEM*dWRDEOt-C+VkBv zW=+`!!0kk7K)fbL2VQvHt^dAZ%|#%|?z1IhI_S7qoXytyo3Wee{@1xQ!odHLbd~{8 zb?q9Klx}d4hM^mz8>yj7x?7MEl$P!uQW_**y1Tney1TpkY`?>AelWw1XRUSJ_jRa5 zl_R?GI#Ehj^u_T<39b3_jni~ulv<`Dwc3YQ(tCuoY3InNv&2##(*$3g>J z6IL~m&2esbbj0^u?0&>*Dtc!k@!Lr~8DN$qdSX`+TZSVP{*ocxFBp~ALr!x8JUT2}ZcG^W{=p30Nup!|=XE}`LH2q$G!Zx=f+K{Xoy6M_tXSr|7Hi;zID zkU}65!CM%vwy~X`Li}F4i0JADVc#QQlT)xssf(j<3!Zm-_$hCvx}XCvbxvAp8&cQ# z!r_vZBdgOJfnvzt0Y?sX;HqF`EdIo2Q00FOrqFKL98;lhAc8W5>()rksyG^v#YR*T zsJL1dh*&K3kk=y!co1Tx60;(UiGE|t!QrIYR)R0pQrpzOzDi1Cbjk^_iA~O(K$G@T zR^Mic0ka)`REPwfUZ=aPioHHRv}iX_m%ZJ`Qi=r{%CM0CDkJ_KYZJEI z_-L_6Jmd|1XXu&obC#h%t62Lh<9GNScnZ`B&ydtqv`iLr#7RZN-rLf{9J9)+zpEscx=)94JeWi4Iu6PxuF}&)K+?7BI>2teq*_PzN z)AFKaCS>~pzu;>5z#M_~LP+zacNi#;qM#ECXW{9(n8|qOX-?qN=|YK}V&}BJ=48C@ zz$|Y4L46x$qCf%{6^sle+oRj1q$m-bqukww=Mlr|x}MX>rcCk$`~J#W*4hX<13LGq zYLTxQzvK`Lq5nINMy%VXVasqbAQ2tL^$z5Vuwrf7hK_w)}f zGLIin&80ZBT1j6KDifsO%JIC##nB$bu49PYk1^-U_1Dw@Uco!VLxl%71H&`>QgyJI z97vw>kR$-b$*-g61~|c?xU?VLh4`#2w6nz#4s2wmsq6^Unspl-Hfq2WoYNvGhoKs( z%^fkv`uv~%rivEm;0T`O0CE;YXZ{zP9NO1@-tj*8&W2BTwvi{wwk_{>iP8G3DUQo| z@q=%TNZ>kiRifz2`LG53-1&D7AhS05~4V3oc{w&c=aTjOR9lBBUwLi>B{h z;bcPt`db3_)rP<^N+^h5vsf8PGWiHTi*kz7z>TrFL(@rIt`GpOfeA9pTV)x09Vb|6 zU*sxbCo)ATc=zw~qcoRGYziGH#VAfF876~ENlD^tfirx0>C8#Kobi~OSGvFb+8LZy zpG8ld|Efa#AP!T_hlIY|(kB<#K-0TqOh6|*AsY)AFI)oK%X`6@%*oEJU~s);bH~rS zg`gT9vhONH%!q)DG}y6cmE<9KXo%G9vlPn_nb;D8R}EjHnCcYmFCzE^Fql#oS?H;xy!P#(ib4oJcn{->D) ze~bum+v-DaBv6YxpH>nPB7|g(=X@gk{)q#QWAdyK4&JbWm3!( zj---~G6E*9WIm_-kw3j5QfR(I?5?Ch;lS0`3!UhZNZ=5Bc>C@*QJ5@_0spd7XuqaD zN%bs^S;61R-pe@Sd&?@?%bEJlMb2wUHa)ufjRYEGgCJH48{VKklV~rt{MB=bel@>$ zq!p4gD;k9@!zeSO-zM%u;2sTu9l)oyhJ8ZcefS>*T0l<-15&!yJFeF@nV&V%VFd!t z7vmiwjDt!d(Wz-`S4P>~5>iz|L_mw*)ff7#Ml32Kl>tj4FS)0iqu}zkM_HWr)*suR zFXadkLf!CPN4RO0^BPq<{SbB<<#Tp7C}iSbP$Qq-RBJ#yOMY(&Z*7F zoCTw0WMUWMOW{n;J48O-c2=H;KZ0{uC`Bj#{FQnZ% z!V!VO!+lQFNuYW6MAsMf151|s&7tZ-!4Z~P&Xjw{q81pgBu}|akMEQ9UnF3EHj%3w z$f*SZSh*!`z&}=zhpKCE5HhPjpyH?vfylcs7d7KiW(WG;2;xef#VTn+#((!VVOq0Q0{>?9 zWxE*S+-TB}Z9^UMHVf(=O~&y1a>jWBZ++~GnPCXT`U_dn+cUyRr_e2xJjsJl9yE*Y*5g+13H_h22sO)*4D|@gi!~A;k(YV#UgZ|H>A1jgtFMZgQ#XVE zhpZi~3;_s!)Qe9BwlltGk&Q^3?i1wz%u4cEjvXLm@bD?!n^v*WDrA;vbb}d$UE;bF zzY4`pu3^T$!USNt)(TvBff2t8>(93S@z+OxU0>ceX%cFqSA{$a|I_cONMwoB zd>ACrrO4KxrbviAAN?Rxc$cSezj)!5J`VmWW2KI&%}B85K}Y`SE|pqL2Adi=-et|} z10X&~C^`<^-Zo>>Yf!eJ{!cj@7PvM2b8~GR-E~R{02D$D7@w$Pja0bv?peaaFbFv$ zbSUFr%c+n9eSTnkM5>yQhW?arcXTPZnI9D>t9##MWc<->f_eCJoGyBw!1u}{Bd4*wf|gW%HO=029eqH zg^}Z~8ev1VlR^uf?cu>`zQCaN2^MFZQkd#pVzZzRV0Oyt*grxT9a9+4X-4}UZh`tc zw1jIoYoQWb%Ei)y5#I$Th50CE{+I~-_x1WCA(r=+O?yIMdKYVL@}pV&pddvi25 z93cg8qFumT$p-M%HTG+yhF&-Lq#nnW#4o!LK%NBNFc7z_>EP1}et54sVdst~d`KA3 zuZ8IwIQ{Fq%jv{R#Y-6}s4TJ{GR|anz;p>}X4J=q_d1N616_F0B3?Ws3sIht4Y)pl zI2JV}EasZT50^!)sfu;ItW4o5hYK(*(qBNFR%#g1a8lh?O#;8dfS5NO z+oLPF*OUzSZBd*dH8j$ia-+HY+vR74%F?p=Ih>;IH(z@2z_0kQCG?lZX&LKjZ$rZ^m27 zy$`#{K*>`Tn?q$o-VoO4H+Pqp-g$n**1k#F+is-0oh#`Zo}+QpKgda6)o-yxFg&f7 zBUteyd1BB3Z%j&)jb;z*C0b^0R84-W$$?%poKyH?8vZ6;uB38Fw!?dxVmh{nf}fP7yKw_nsX53HjQe^B)e%OX$OOtk;cRviXQBXw-`so{ZcN(b{vxv|cS+@h+>tOrA!NF}Q zgWZRc6ecU~GAO)gna*^pBwem5UOmB%*U3;w`3pT%pk8vf1X8i6#`s@@P=ysVM)=WC z4259{sjL1AmSHQ<;u^|+cIl?Y(epAM8egGfJaYu!cm*XOJ9$kPkA1}~62XU(BZ*sY z(=(;J3=#&$l5zKF?qF+ zBW@dYqKpiWiBcy@3*TyB@BQP>fptuK{$DhZ z;S@(5nzi}U@3o#>N&EiSf(W8U-DFvM|R=NF@)l2z z_fx`1PMpHWl-a<^rO$MsZ_M0uOiXAmPj}KloC^G4x{Lp4gb8AMx3Tdsr(dbXRHbta zBqBnUgdZdjojBO0**^$kgFqn}fZ}crtxmV`yE#TL9V4S$M(WeQhDc3=Gp>?Ht>l%I z&W1Y*GX-zB(9t2PN=Z17(1C%|`&a=Qv7<1PUedTK;|JFVVeWIb=d^m$zcFRJ-e!&J zV@q`<_Eg9pHB*bRR$@k0t6EFRcv8c$>Zc`T9KQ@HF!Vf?G*kn}rYK`1+BEg(Waq(x z{bVGmp{UpW4}TaP9G-{+Tu(aZH@r6n%G@?@3|tKt{E~gTD{^WhV}R{Q(6U#uyCXJ=*7^$k_DzBa^Wex%o&4O!&JEAP`HM)M1B#%Za1A|q?h0+;(w;sIPTU%8=Ddpg~yO%*-=4<<4_;t8}rdqEI!HiF3`+=dNQ zDl@aJr5a#%Si-i(RL9B#!IEtj$Kfs`#z>%o?SRC1^w&Q6)wHUUJNniFdg3(%*3*J{9=VW z)*b&7A*I0y^k{GC7B$xX3BlQpEnhE4co^ZSZmf%ztd(&E8Wh4k`TBnQb@`uD)2O4N|jkHM|zsxs1qEK`j8BP~U|IL)YQd7~9G2HI&oN1$FI2*!pkQ2a# z4e9HLj@1^UxlK^Q|r!vorv-BkObwcmW}_5py(ap))7?DgekmHlWaFj+usdjHov zjWihG2T)PjfWg~!J63w4&Q|wn)&tdM(>Qs|%Zv)SKfKfVoI=awXE@0%A_niFagUM{ z*S(|S>y&jqV(Z-g`o~|mHcF&`DoP9sc)_Zm-9Krf;lgLBK4A9zxDV-EfeUsSi(B)0 zpZ#1QkuFi3zb4Zqitu_hlqpk%b=;_{LpJp5Ipg`1HsUZ;YR#MZNw3v49*i)R6Qc`{ z-53R);OKOP6O(y8T{W;P5PWrIZ+V`&n(8t-8sFdq1mwNU|Gr=p6OKVGw$CwsI(1Bh zLjm!b6W980jJdB%B>4_s2pi_uQb$$k!=s|=>^l`SSN`|l@bCQD8~!0lc4aD0u3VQ- z!_lS!oau8hQ9S^9qrqu27rnHUA$wpQ0l#xWPGZp?qCp<1>)KQr8WLHFwPH7Eb7i<1 zxe_++f`5wb06JpL1fcNLnVxp%8QFn3z$QwS?cM20_N3KzV_!5`G?2(Bb%+g^XJ$)x z*Nq7XHs-$t$xD+4Qtw1~*()A9a&M+yXcr2wco%Wepq5&Ebr`v#nSBt1h_h!ei%@Q1 z{HIy!^F6c0k_tJ<2Au7+bQVkc110jKC$oQK^ocgP=Dl2kC#u~C?v5Bw?498@)2TrG22d*}Ezaihwn^M7wUjnVcSY_u86)K2IJg;pu8Sa3WOl;6HS zXM8YI@?&#A7E-wF?bV~@_3?;fT28#Hc5r!Rg#-vk^rgPoKc!p> z2@0C6n%8ZZEf;ID{??{oWfQFdLbDvbIshZGBg-;f;wYZZZYuS&=&mK}-xG|ehNdEn zE&-Ooh3C7piX*Ppbw$9a-vOIu37P-T1t*z@KZrFRKNpF%;mK}bLtDMXvSZ+*qy=0~ z2l*#WT;ud%-3K{ zGL~@p7a&<47y6E#G%#O+t_J z9^7b4xs?>dFw~3DOz^}|hmD+zAN2`JM3H6;mOC0(+kX+kl5g1=er8r-ERc22!{gKN zJR292+bM;5eS*@}Ftrb5yE=N0#L61r9*%e)Z5MV4U3!(=jv4J^laO;mLJWw_;FvH7mKjh^E zly@uyiFuTjnK%r8^21Co9Gfd)#ey5kj&X#!mbTxiF!oP5&IJ^HNEe2Y7!aPc#8j-{_xz1cw(QhOVw_y6jpV_lGf3tA}VU* zDus=BEDf^r!}@=0nqte#%VrnoO`giCs;arL{ol~Pe*LO?pVE(qhjwXVD46K&L^k}g z+&?3LQO2W+MSbm_i+<>ozNrMrPoki2_sCzn-(tb!Hq#$-l?eUg4bo%(K6<3~W{dqW zvo&B23F=%moMF0 zHt|6UuS&>axG7`GKQTv+5RffF#EHakNC)GH=}f4Qqfwy~e_~?cFZ7eAdcDY~C7xOC zRC)PyK>Mz5vHv^s940s$1$H+mIXS1>qIEd~llfbK_3z{OoM0;jYsjboK(?g@8> zj>tnMNEa{d3*@i|J~asj@$K+g67b83H=&Ls6?ZcI?;0vQ_5kRY0?J+oAvexsCYd?C zIJ8oP&;tS{k*}hWlibkr+WRG&S_14)o=`GJa%2u4-q#jqt{kRhb{RC=+k2SPxhCVi z$FG+j&y`=ic+<7RQOt+e{W+-3!v12evWCtXdJszgWEC|TQ1QGp{<^^Z*M_wZ?FltH z?CfbxxLwA^=XhFWzWqKG?-y}vbA4QYt<5ynBUlWjVeL5b>(;AO_nGzjd70Y)mYMO` zUvKlnzf59G@xr-y05NhrVt$Vh6v{W#wt>+_OAdcjM^@2V8N&BT{~6=C?S3WJOog~z z8b!Y~{v?d|v^^C{U*;cZefBPl39_EmKZDzpsvP=1@N6NSv6FcD_IQCRp`Y7!aUfV$ zSPP)F@f;9f5`8=-;LF@HVPcZX9~<3?=NiBqt|_82);oPDOI~ z!puVq@iWyEwVQy7ny>@82fl;LWFUkHnQ~zS`#$kGR-?ncefPifudW7x?cPl}2tj^DhAPF#K#0DMDLy?iBC6ndtBI%!~5sO|fIf>^@#vYU8B6 zq(UCDuBE&QxHj~*8PCOOklztvg;uy;n0)0jain9ob{0~V@iSeo`tv1psS)WjDudb{ zDZjH!lr}Nh*M#-w{03_VurEt;AdH5W;D{Y_w!kwI_wftT4bPj8y~`^`y@a7hI1v@! zv|lID4-Nv~cF2v2HmE+_=|&*~I}nVt!^-yz-xs~_TfgdwZ{K!-^2enan|6JUatPO9?*wFId6H|Ll(_d&r-=>llQ1d-y6N~=k>N#kUp zXKfF%xAfrvA6=b3trEcTpB!&=BqL4MViNwgV}v6RK_mWj;RmnfS@G@Fp+&Qa*T8dw z@6%T;+-ll5^0Zp|swgjf3TVaC`d2fvRg!4rKa&Bca5O`f;tWfE@1UT3>C8mmT}QFt zSR6-j9UEYU5!T~D$$gGhgZyE8?cE&7rb|=I)l-`;uxF9B4nOdY5hG}1=NPFeGUihZ z7)v^uGy#r$5FWo(7iRqP4wR8g-Q?h^%tx#ks5^=e+Z$$WPiyT;XDy*I8-0TbRVYu0 zF96rt!?rF(Cdi2s1Yq-AzQ8kV{e%8|-B??%q0d6BdJW`GE>1Y9LLPO?8p_a)weaF^ zo)h@Rn6zSwSv8LnIUT0Jox8zW(2vdq9Se_F57wXw2Q71={#K&Z46ZTR ztY0Nd4;?yAQn z0VjzFG9!1_5j@vt;U_qKP|Lck-{wsa0Huf1FZVYV$tEy-7fj`K%+WJaQO%lqiM@mW zFa$atkxXiTd(~wrbBqnZAtPgb+9Dp}M-$8lryb(eE;kT*Xk+X6$WM6XqS>XHnzAvi zP^#kSrFPBFMF2z3qqvzp1DDE@tZbWma<_gqX>RNsTCg z5e1V>HK-)Ruz*&_H!w%_DFy)Cc^R+IFZjiqsV+d*^z+HJ0SYK~9Cjyl(XeaIyg7NP zWn-^NEx*=lf+4->Z3$6XYc$n76DXn% zA_=Y-DN#kM#GPWvEOOVe1;bDr(czP8&OM_MqIG~%A0^c)qMIs6_y4qFzQ5;2rL}q2 z<@`)-wc7&hAzh~Hzl}ZHkodYxMO>i%dN9Es_{?!8987M;? zE;c1gbn97~)jjm|(!j_-uR|=@_X0^Ce{f4SiG`U~5rKyi8BDC#>TxrpX`bq|H#T?z zMs)Vw%vHR41AN%_Rx{^H)h7QHAXIzWF>wZte1K+uz|~%cAvbG_apQ?;^s8lo$3Z}D z`wZ91?C+}{7zv)p$-2N6*}r+1aq>@N{bDNS{2#L8frZ(@6^-|x_*?j31*Mqt*9IqK z8Wi~5+C3Du%Y~?EH^bm&k&ZVUivja5aq1iOz_koYmNS5IN8E;fDGA9{yz;tx-g@b~e-< z1ipI{HAecKIi9}q;P)L|okT^CX#4W+AZD5i;*YUk|B};i!T=lMDjuHDduHd$yR2fB z=mZ@J>b0JkkO7i7WxQ=Eu)k`6OxNuT?I{N&*$Mb+66EQgAldyBbvrKKR||-K|K5&S zV}E8dHp48@7Sw6Mh@Z_F7kL2*YX*|ad>`OuN|CNO8tN_hF6N$wt% zNM5J5s7&l_lK|fCiKkaHQ)bk>xYA)aJ%&l}+62ryjP2@72Z{f~#bo5Fnd;S4c6{XX zVkCSP92%Mwa607$t_upZ?||R58Oa}>`9^b$8x7Q>f6aZqz*cofqUMTDuw`vHcKRD*sYhv0Xaw)V|!!1?$|2<4=qiKUtk=o};#U=35<1 zk}aBudZgX9-?rwXUKo}V+Nvbcp0!u~z{~UI<7RaxnZ=0u8Z;Zl?2}e(LKvUxsk?c0 z7s=(t9mqTUe(EX|Rt<#E9LZ}r&rl9@Dd>St$z9tS@3{SS@Q)J1V8!LDW(%pO?u1_U z)#h97gg!0K!pgO};LVGT()WNh|HHqdLL3WP8M)J!^mX7kMENzpf{#xeBEAzjdcM$? z2y(XLPjt2GN(9FAt!7lUw4^^Po!;kzhZ^_RohX0P@d6i3uP^JhUCMlNOX;8isiCqs zml7FZu@SXddD;?v4k?y4bvjMNr08;-tKjGe#^3y}_ImtnI7JtG*qe<@r*_T;WBQLh zfTWo>Q$JY7Fq<{xOj!7y&NID^V&1g=TfgiI42ysng>W6__ciKzc4ihvi)Q_KE5-`x z9HV+1gr<|888w@A)+o~o_uI4ofH$4dA>`GnhK4Rnb$o{Nb!WTqxfM8^#u<2#eUxu% zooEeV97lny2Mim1QEUU_T;o!UWx)>NC#6E+tH(0ZH&k>*)5uZ!TE~&~cS~9&6|Jr{ z(r&qoPxz>~4H_y-MC^O$$#Bv9*zmZNAM%^AA+Vt&{;ZiEr+h`%S9fa5k?|eo(lG;$ zWcqDZ?Bx6Yay-;c$?R;=FgwqY8l^haD}Hzre--^8j4`kbPse9h<047ID=5iap>v(_ zg*6Tp=v#((Q(|5PAH3W-me>#jW3Rh*K z1Yuln7@mPEtcm!VcRroJ0kjug*70D1cNrVW=;wjFF;aebvxfgHPc3^x;W&J$!WZ zVqt3I-rap{Jeo5`2UTYttJe`CvBe?HL`3w)S3~$tz%><^>qs6 zr@VIZ!D6`Z(~WmP3EEK6hu*)j_1xG=7QI=K{OTmJp*p5^Ooq`zc(}2MMRyJeSM5G) zLi&Cna6v9{o*|PyHPWn#k*x450b0qQ-+C8swDd+ZC84X4Y0yOdAGAY8!@7tDq;|c%c0S1bMs8=E zA4qVD5AUX+*W|EG-B~{>f14JzFvR-bx0KK!0K@n?mZ7fxWm<&dFUHma$}@n;#w0 zCfjBM;qUsa@MT3N-8v=@v@Mgq$Wv+jxo$I=PVNFny6TLf0^8chxF6_~JM(M})c2yM zEw2w%TFfB>A)WL3gnl5aD!60_fkp~^cMq0bdhI(u&+=c6+}j1ykkMGOto=*V5DnuI?1tjRQgm7BG)So2kgJe`7>;oS9l*O#P>ns`Et`yDRkWd&(uhEyxl z!;O?gUhq;rl~6NdUGjf{#^B1&^s&Y*ki~Z}lpU)M?)`kM!zAHXXLvw+9tsCLT!Leb z{!@QU;G9N(;lrQ9CYAfm|KmTjXOkiMJS!C-0heb2$*Uv>cS3j3rWaKCAYBJgKk&9~ z5Df&>B$RLE`pm+q_iI{P!z1HO^kc~%@e4t$C-@T4ca7!i!dDsRon zP7bLpOJ>Un_N;rn*?-5pL?sVJv^!e*(|Q~p;B)5j@<4_7`of6uEtUEGLDAOZfMuJ` zOxR|FsKf(VPjqQpHTpWnUOOQqM2Tl>D?8S8Kgp=ZF#cWoMT#E@mSxYLsmaB}nX&7B zxs@lsVRYej8cKsr1cJE*k&0&+$i?q;^lIQjkpg+-WB%qWcH-|4jVF)6goAJ9PM=Xf zEli_V>uGLT^|&>lbgjqk2T!YLR?OP-yb7|z&g_S~W^DW_s6k;E}`q zHg!Zw6L1hE@}$($qs_R!c|iLE<&ew`3LiF!3zV94F^Vs+zTO_~PgU~GGLwY6X1B=M zaCuI4kT%H6-3p3fgbCe(>dJnDR{|49iK1lM;i?))m*SZ&@<)PtziFq{1J+39z5gaZ z+vXZij!>Lq1>CMo+5Zu1wGO#libi_-y#Fd~6G5~}^5XI4N1N#?j;LA;rXZUjk-Wex zI0}I+W1H^kNU$t}#g*3`N-(tG;rD)RS(ak^a03)q~^1s!*%jJ~+S_r*EBvspuP4|f zevI0SB>(~bdLnjE^ML1;=WEKVGODhILO|kc=I^oZ>)W>shVe<2NXv?Me~#}1zcsQ$ zl3keK#_ytf)Vu>I>0TdU?n06ObY6%w53&;9P;6kek}RmeWs6h1pqm; z`ca^?OidI8d>%x9!{~s_q}MZbZu}h4yM*XD?)WgPKl=KU-;`VqZZ9uB+S&Li@onO%G=o9i&R9awlDyx*db_AF1EfseschsH@Bnyc$ z2IP)Mvc!{kg&o?Jp}^TAXkr-f9S{yWoviJB;3H$N(C%BSG^tD8Ql(NF;s%U=1^kZc z<${o|~iPliZ1Jj((XJr`TThMBS)e*gQ$?PuzM`eB`NSB!sdomI0Cq;G|ci zntNi#U&2=d_Obp-?iJ{mqefmo6w&l*j1)oCS5KZa1~|Mag+RtW!6ih7>{Hscq=bs?o* zb<@XrAk&>~ubZ+z@0gjkIWGg@WxD2ob`{9T`9r#l1RzE`s)&JKUZylieok3k@VTW4 zg-c*SJCYGGM$im2#?qOANS49}nyqY>*!_BXLY?j#JO1HxCc0S>1k5+mFqiidn-zOE zZK(bJqpnW>k*<|+Q1U*kHat#z+ylv%Z_XN5>NxT~6MsB82n!=hk)0z5WSVHrt0I4& zaOLoWzC9bpx|A?5G6}WKHEZ)Nc)>(WUPJI%O&lwQAW*)0lrwn+Z*g6)FD8e{mBqP4m5D676c zb~x-iig)Pk#VmL1%5LL{zfGRKU=tWU>F}Rx3gaBw8z8fKbQ;Gh@i+h|cb!}lQXR4y zU+E=5_6sEN%)0QN89O71duwItOh!)p+k`Q1x7l8j&~B;nv7JO)e$257)DTsYHfSDFMm3+2rC_~to7OdpY65eh7DK=K;*{aOy}9T9TAb3N35Bd~RUOu1*>w{qkyEqA-`pN_m_Z*p&T z(jbp52>%Adettawk2ZYIRNildRXg3f+x}Hn_-=9m>?@`hq5JP-hyQH~E!)83n?%Rz z*$=Q7-`)?X2t;ODN5%0w3hCj;V4r}a!C#^|`1bN#n~@O2PpP=U(%zWe^=hW&RC}+; z1b_97J50N;=MUZN0aWoi>G_lnj1F)CG~vj5rh4aR3-pCmMO?z#OIQi;##xfSZKrr` zW#9o=eLZ1=C!wZH;=ocd=RaztNRp&MuIv^LX@hxe1YeA04~cqA@ws`S@EQhox5FaI zbwT4V zBOWy*YSCu-Wc(s9yiK%t?^O6wYwOXEp-#uH{@4Vj)-o1 zeK0r;ojd#b-dT{H9G{}`dm(3h-i*{AG+9z_*_b;~64Ey+Jk~pu?qjM=p6{an5$sp= z1_Zy(mY*EPG7LVC$k`gp-lVD6@72OJ(gcNbs8kp>M+NSMMdhr^;7m<%R({kVTNf&)8gr0X>s+7Sjb6RhYZv#S{0KK8+cyGdk|FP zwdRUDpxaSnwX$ zb^!=Vq0p=p7ASMV+JeAD^)#I$CuiMt!W3s;WFfz|UiHF$xgybOn-S5pa(pwAs(BaU zcmCRZwtk<1U}C8CQ(ipN?j$Sx!?sT`k2CxCea`og`^)yYfFETREesj7vaZ%+q0V`6 zl34I?46L7quJfHGKh|&|W4?S0P?DDH=40CX&#~Og4}8sI^dJOA4p8+<%E5c5ez6vt zr4Zx+C1dGp+K||$CXQ|iL1kxgP`H+~av$@vWH`mGIM3mhh~N~bmv7gD-+_Xzv_xo@ zrQE}?h#_6+eF_lRz*(Dq9pNyx+fiA+O(P)RN4ldcll6RO+txb3VHEWmLofCfCl2?%}29?Ivwm&bzd4X#~GB%6FY zj!!KS)pi__i#FUnQgKaFZrKO{h2rIi}Q28G`v z;Ku3jv*v+zN59KhLFIx}pooiBAE_w`1Lr~Rqlu?@zHFf|ZbpWISkD0?X6%*7&@3$+ zcV}Ch;_R$@AO0{<$Zx_>AZ|ZC6t>kvp3eIozYr4SwxfS`53%Z6r{ZeI= z1;FBV*c$#IO|{G|Ktrs z{qElTac)kSjT`d8u7I#KwPljFx+|tU1QaC=f?Q0T&TG7OvTZ3~{3UqKvq<*% zp857p!e76AtC-o=c7|&1hg2k=hl-^hqA*EGZ_rhrV0A?yU7F0a?na6gu!r@Kze(Lg zj>uxw)#FRbgPj5_?Z1;yBZhgDC(CY2;%pny#NRP}aMfRkXw|hCQvY5;>~=aYn=6?C zO@w*W>O-#TKpB$$Iodz?5)SN%_G<(Qe}EVv?5e#_8u%G+bf@nZqBrO^_H8f|jFtW~ z`Fub|LxnqMhX(gQX0DBFW2yKw3km146r_MzBeC|(BtrmVt!ch(LoJC3!Qj(m{ zsR~kHGlD3~YT%?73a(fuD~|nV7nL;bFN32+fHWi=y1bDvP0hO8qnWYav;t-Lfad1z zU2xafW^;PF`N!5}4Fo6xLGXs~gU-O7pZWlIYU)Py4T-XOr-{4>adbrU)pMb;j<-;)K}@j^7MGXP}&a9cvxzb4);d^ z@uP})C2|tg6|F(#R}1Hr_|RbY-1Te&Kep>JiH8MuJtS5^2%Z+&d_LI1w7YJ9!gLwHC-n$0QQb{F5L%*~cJRv-|7DH`6@tEKtu zStv48Zo@PPQ;HlFKvB;|=0g43x*Xf0@tf#>8s=pdi;0ZlFBLhZc9u~?WRN)^3~YK%E?~v5mhous93PkEsq9f0heze8DJ~00^&Ko zzRU71qleE7r>t6^K7v}7vjx0}-3#Q{XhScgC1Zn%3&`q@`NFP8mXxnH5hsq&p)sxx zhzs0ozdjF(yFX~>qNawpDcH2%3cLd@R4Rwfs2Q+H=$^%=i6{~UyM34 zWH;?^m_-IJFAH+YL`|Cnj&%aV-Xi><0nH3NK=vNw<5|4Ut{1}aZCwi&7uO%?*q5{E zdQax*qu=~<;Z7Ko#kB+Q(omS?5=S*Gy03W;aVe^C**DHXyFH|!+p^$acl`TzCH}hu z7-o%nel#qy`GqvEsd0puECKW_rGY%94SX~>DMCxJQ-7$Si(FEwM~E`tUYVgfUbJhp zE_UG~)J{fd=e1bIERk+<+<{om$hK%RC@OpHp5pQGbU*-nRG znXVRT!XLpl(`MAl_zWmQK+_a>dEjQxAOv*Gcg|V50f$5#@{pHg01YH316R1xhzJz# z$9-DC%R&BvSlTaX^=gZ0dNrP9uLIN9hj08%3nM5fv*U85zryp49Zg}ld$EQC=a-Go z{g!~y-?cCkW<@huTv`;o@$g`*J8_!L6ldejVLH7;o0BqTnb$@>9HJSCE*?Yj>@iE3 z!ly?6^A}5*=Qczf1h;`Sf!<%aVL9-Kcs4csgsC0myGyJ(+R~j?L#+(pj9+88ddDUS z)&#SR5nX5+1T>6|PVU+NK;{t*#Q#2Ka~@=~hY@Lp^yS8ZyR3{6sS=0_>BK9n0{m14 zWW*owQwe|5qmzkNpm_C1k(kT={RioI+#Xk+HMWcK{MpO!zyQ$LpaXt$@((2h5De!> zG_#TnnC^PA&TECQ$u=K?zsMqMS>umWpLH-v7&haLtaYbUQK#Hy+aNCP(B@WSAUly zE-BWzS6?xl9{7*tS9Y{C3aO zUJ=Oukfp3tgUgjbTuBByWC}M~!9y&X7BbU4!XxPaU}ci=-Jde1nrDmMq%BE4Z7bRu zp&2w3MBqJmgoNjBT8t3KVEAMsx=%-JTtcb;mhbZ@=A++mTR zEDgv)pK5Wul_pM~d!8}AS*Q9Qi5O^A>iP^lTS2ix^5b?dK8U;swN|^KD!~!)64g!z-Fmx! zfOI6_omidKB14=nH)#C`#9%0ZlGY#`YT1Fge;S5FR>NQ2^C>kopX4)z^_vJT=K+U} zK|Of~w?86!SDcbC{n96ZCwgBL85Nbd1$WnaH71Y@u<%pUXK826VO{@V?|(*XIQX){ zx7VjL*K1%sb3LtOy4>J+N3D`|c?)co$njA5Z|t&INVQ`8_L%*w^dV>QtV8vMWZ%2U z|M~}N_CgI-U;9VY%k!WbJeeaHWbx5WnXd-e^o*9|HvR4(k9t1{7GfMZs=F9d*Xs#d zaVNCCxeyLwWQ*lBC27{1o{%HQYcGe%gy(!ZO>Gdfq)_ChOA;}NmLW{LWVJcIU>(ML zY)Qt%QCKq>{@TFY97&Ei#7Bqd$jo;KOk_r*VWNxYEFtA8lZqH^F=iBA&kkCumvs(o zTd@{r7@*~zo3~7@8Y^s4?`KB{8Xj9;Fqcf3L2KCmx;SltZ0T)l{d@Hxk87Z8TU_b3ti=6RDp>9{`(f`OnM8U=j?=%y-NThRZ%rQ6ca)X4wRco#kt^z_K>FSn^R zbjp15tsXohOsg)L9cyc3{|qg5QtBlru=)~g9NtBYcyIOJF5b@=`WBfA?F*L42KH3g zGuK28bt|i3A$?V#qn7Dr3`r9rT|^@=#wl&8#e(-`k#_#V0Q1=&6Wi-CM$n+?10O&* zYAX(e@cGBWH6u;k=Bf|1m1z)WB=Lm5*o9mR_*g_*|4|R+adiDZqTVtr%I|H%m1Y11 zh7QR=8U~OM=^k1@x{**iB&18aySqfXyBnmWySuyhPz#`)=Qe~=U@Ft5K#-QM14F9p@cZNqhrf{i)pl8L0-|5eT;bADq>D( z>b!q7jZq}cb8-xl1~llQ-|i*U)(M>f_TPhoBA>{cCXak!7u_0F{+d=#zB|0r66u3( zRP(z%R}jMl9;0fk^sAuG=cfmIIzn;0$5J4YUx&+l#L&7~fm(yi&C`}>1r9+_NtXXDG{3+r2-EZ7Iwo6R$k z;Tma*dSwo?D``%ZfRmZ=CAuLbRlo{5ofV`#Ribr(j@WoqNq;7x@Hp3iCr_qsp**i4 zq)7LC&bks}#9=7EaieD$toy*V67mP33>y?oK?0;k#-vs|<|Kgel1g zxYyF^bYpU(b`r$IC{bfwrNF=EZ4z}ny!LKI$5>oIup>bsyMlPAc;Lw~OGgys^qJRg z6E&V$>)lP1pX2VBaJ1jskB(A4Cj(vz{@2B%{P?e#!N$pH)Fa$l0&E7m1*$|U9FH{T z!h7(IjErc`vS>=3PL}N+T4+J4cbli4J;a%lUntz4wReYNf0pjkj@6c;cZI*6Xi*>_ zL+#r3Z~s+P{yPb7Y56r2A{OEdd>R97 zN2~k8He;#~Jmi zaD(DUMMYini{ogoy=61>bI4mi`HaO)&4AIHi}>Yv94kKAx>Hsyo43cm8NvRFTmQBN zE_B%kwfyM(19Zk+&4hzp&6Nm7-6)zth>3o4Z9!AJz(E13m}UX_QV?(d+=@D6jy+xk z!a(}hy=!R~r@DJi7QstCQA!Or&D9|OrdshIf9VO8S*hQxV)-mOQO>_mgwiKKU{W{_ z&%3kQWRlr%w>rwF^ggI!_hj9_{Bsnuw26woI6mZX0>k#;iysDKf^7!>Zh6lKXc1w~ zJPT9wTEHVM)r5GF$EweJjwM?_=I?PKI`%dV6!No(+@4C*R9rn9$pCaGQy966%UJ26 z1(z9#XfmK#t`9dDOYaN75(Dnq9cogYfaq>*Ym3GR_jXaQ-UhC$Aa6u5p~p^ml}1={QNWfl06u;+Wt)P0?NcgJ+m ztldpvB?0Dn0}#lFKzWe!#R&IVCpv3??LVA?-E#iD%VP@!bly-!m`T8j@$NOo`d5xF z#F7A7hnlvZcRB0Osaw+g6Z0;pHSr}rj)Q!kf#&-++ea4$%N3~kG)7s1o>;`9Klya6 zaUFy&P1j@0Y2PxO|NM`FB*^q>*MyfHNZKkIOv4Q;V-5*TYFM_!>71)Ft*Uxll&x=+ z=xMi&4h;#3DfVA+ze!6yz2N=xsW7&op(9_S*?^X1tX{-#q+0K+cS16R}WUpS_3NN>G z6ZPXBOV+{X&Bf1}9)fbWC&%!OB;L=Hs|bj{(-=qNM9Bx_z=x{ddJCM?nF6)j%e(UQX8m#%r_p%UI!!azzeZh;7Q*mPuWjtGDvqooCy z!PT!$xsC7~tgWpZvPV5&LfSey#w2;YC{+$@9)X> zrQkF0z-G~Ya0tj!A%5$`=2y_ zN;*^%7(OYvG+x91pir(tQpK-v-w}C;Ci~RmKGHh+OA^56(ftL?n^kIO=P{-k?yB<7 z+?rwAo4{%*^$$F9xz0kRwJv?clNi(Sc^<(q%%>6&Z=zs4^?&VeO7+<#;@aK{Ejm&j~FG9Qx&pRvl(NC@DQ#i8@teM$d99W5}um~K)tVkXe zXvOZ{#+I#k*40-dE=EEk-B787qsy)Kcv3$$lh*w5WLRLy5QJwW2<;wULN9lj5);PF zpmT#zI#SCPECX#U1_1kj16li|uJmW_1~{~u3WMK$$vVD$%$uHNs!lBGFu*)}BESNA zqG~BRyShXfF8!!siiY6b?Owbm;F4`E$uvDGn2$L~US2*LVzcI50dz|7*sQ(+iV6eN zbX~;LH^R7=+jqxvXbjYtKp5J8UNbz0hV3pkaJ1g@@mnw2U)a1LGBuSFzE4XnXvo$&@D7=`faAgS<$C(t1Ge!%s|4{0QC zdd>!=65O5j!G+?od3ec-=uTNMYN_l5ck+z18d@V8ZXO<2Z^9T&=rPz~gAUp)nM)O4 zZi+BGwcOtotq7hb?lCC5!LI%|-9a%=0>2nF*8msI#q9kj6TuMP!0L%+nA2Vp8`q%3 zzrSr>xbI~0jBz9cGK+VAZAmgX`zm;-;k%ud-7Uz$?13No`hYIZ^;wm*<#sU)=Cd1- z6dWqD7jeq(^*Ka>l5&&o`RJlp@khwOEJ`nZ`H2om!3KRJ2_9E8R^8kD$(|~h&s{&Y z&>;Q8a_a2&sla5T%G0aeR3$AeUu#6+%+v2O!r<^_c$j+4x6idr*uwr=F6j;5MsMMx z5gF!Hnm^P+E%$-r4ul@4%=LEI%AxT-mf-vb@X2DpEksZDN*IZ>NG(*uV3#EiZ$OidCpmD$81sMP7Ls;ait>_@$5U==8 z*9BibC`rK^(H+M%K|VXXYcsL$|NXgl1s?nSD{%PfVF?SOhHziwoR%UozcK#m!|R_7H+ zxp_rujc#;~`rqE7t|MDaMLlnQABtUUXX-zg$R+iRxB~?9DPjK{(WK=cG$vz5VfBD% zz%DLX`n?*wMvL=$cMy-`VIl-O*j!lm3C|Jez3%ZYPH$e$mrKE))!EVYMOy&2TVK+D zH2N%7|FUCAM1)Dd}6?f}VaXH^Iih;{vb8l+3A)HC6DbLYpXS5q+ZTl~D@`OPad>3KG$6iM!F= z$CmSR^mX*pKN%20zEy`G%%!QC)Xdlr!0qs`>gd70>^fsoa$9#Z$}T$br&AgntzhEVh#YjUMurw9Z65y{Hu7TI9NQXTfKLq)v$ z>D1lK^A3b;shSS)-aw7{Xfy7;g2VWiL*92KN2JpV<@TfPqoFM5<(xEC+0D%j0&B)Q zbuv2TD@&NB&)>;2XMXrX!AsC+h4;S&rYn&G*OyAs_{swZ2Db3s43mP#`=UuK^(Hn` z7wxHALp>0O?S*0CF1grUdn!k1)j5<(+#4b$EmHI=}NxX}gklwq4gS zQyp5!a^PR;CWBwk`HK7YjU>i0BTo9?PO}tMoMh@Kt9I}b7gT*K&N-l-R}8W>vDJki z{EsDm^5$V(-dMArp64G+;AX9sB?`i`^1UztP&JDo{2OUw!HE}&Ww$3|w@cx#S+`|T zH*pY$iCd2{0HE4HeUx}u7>r#HeXLF$mj<2 z0$-5xmQqZ<#2_Z|7nOhE9>hXBoYHKM3W#*uB(hjmCaq!#6^?$sH#Ec#!IPNeWJ@aP zpU|i{^n>G+SdbC{kfPl>)sXH>pN6E<2IKJxV0NN02-t+G5PU%lF4|&dE3N$hsvThn zC7LS3M1n$X9=&v9%z%hbYAPITP=3nLq{3^mn{WcMwcOGhWM-yV*dkwc0*y|{>h1g!4lFJMRP z>=>oEvXQA1YV%nzVU#ta4FUh_k1C$uC(S69WdeX?1%64Ayz_kd0qXfEaH;1Gdg?cT zMv$S|=QKj=?%?vbSA?6aC?_zFd|E9z5?h%t?HeKtJj&izPqq0`GIH$Le?`G_U^s#} zyOk!!1sh=y1er}ydf$4sKi{Rh8_`{Bn542_hPnf}{7dtl%Z`)F|uo;X(Ww3x7256mtuT#=q z09?6j-hlQ4lfIaE>@nOhqi(S(GyS5ZQb&;QNzz#%kkOyH(!!O37e0Usf-g(R4U8Eq zkymlT8U8q$i1lm*aje70B1vR6F>78lxWCs?H{X^66Q<~wF!Z;_dRH=VOv!NQ%^2k8 z)EO9Xp~vwd9EEZTgnU_B32)z&h_S`0^vj1Xp${4DE;>st{Ey*4*eKq=^3%Y1;I4aL7_ zQroNiTlyVg&jT0d2gcKndDSrbz8jWU@u|wP1KnsuWnv~WOYFuMTiN-m55KIx{rsgh zzhNfU;c=3vg8r zo$&dIWE$*~3y{9DTSJ|v3ask)S3esD3!pQ<(cDw=`*|yIl07^-PF-Bo1)d4<*>p){ z;=F3@X%oNCm<}klC*wm9-QE>i#Prk0zh$_5nWBlAR?vi)x#YXf#4#FTl>0tNVZd$) z>pd7(uTnoc$%Ek=+gNlO|F)vA-$@9goo@Z_IJ3|n0vStd-$zK~%aX&2A;&Xl<^Iu` z%yP{^+nCgmsNI6q>Fz)e4YGvbEsoycHbfkCn}@W#t*el`&SkxKI#giARX5J)=yJq( zcxFZWVBa)kUU4(>T(2WZleFM&X=z6;xrUZMr{!Vg;*lo$v0?PtWlQg&X5gSEBg9<> zd&hvJ@1P9?u$lAqwni}c-KrsnpO;&Cab86zcgsM8IeEL`o8%;)+Ku{1Y_X36zhNY$ z8pyusEgl~5`#6$mI+|`r6RZlBe=j{|WB^YqmOQ=ZH1fIrWgo3WAwMKc#Dv3yc5uNs zumQv}o-Kr36GFV-CYrDS+9!`H`MZUblE9g0xL`g7I ze_1)RKGFN4f6@FE0ct-XxI5YM_Fecs=q7}u!f8krWB-N((Du9vqj+WO6Iw0+U(6Zc zd>H?9MeuZWa)Dl3lIC-{t2~pQ0SsVJWez1|mtWp8SegACOqtK@ywAVgQ`94ks5i5Y z;*RFeK{~{D>xmso zz#2ZxkU{x$OTRlJu?sdbwyj;%f5lV6Pudkn$l3V)jK72(u_Pf>d~F?EB8wqKTW=<8 z{)R($+%oP0NyiJdw_r{3&o@k4p^wv&Y6;X2oja3hU`V zira5BgGdLYg#T~-A%%(De=fqH4M=OBaOpDs(fL970dUs{vRQVnsPj4z!fQMAr|z03 zgmwVA52&4u8W0fCY89wAD#eh^Z94L|6^`{V@J$Vg z`^MMp<*=kfMyE>7-9Cil`B6}ACO!_UIKB&V1cFwUej=ACq>kG6Dy)fYgysq16 zeQC38Qf|k8)>Cz$;&f_!;VUc4BOI2kQ-fD4eh?eZWE}WOG%&`y+`V{J|K`fAvBlxv zE=TX{?3EZS^iS*8;e`w@UfN@$!n@Z* z8Vav}DV`D~L~?KmGdvomY!`E9&Q+;4l6D+{Wd%P6luqn-{#^d3Z4W zCRobjSV9{bDK;bTwk6_K9{>xF>w$xXij%OwN3JT91~rIdSvCmi)Bb@hK$b0XB0a;A z_~c#ETIwnMEZ=540LN-331HZP7{^uHPLQIxJ|tJ7=;3bke(8Y#zuwwQPa0L}>g!n2 zW&itcSTS*4I8h1E)m@XnsaWAZjO6B2{n}=}Mddf>QmTBa7uqRK6n(&+s$*4RO%?w+ zglES4h;d==^SkXFU)DO_Ypn${a;_MqQVK)Y9ij1@T$uY*kMJ26vUa7GRiuWJq)gj& z`Q!vmU;!Pt`tZck!O2}Q^lLYv5j`*WC$}LcgMgGtVrdmdB!n&=k!Ch!gOWgTH@N#r z`)a3In45jNFVo+9GM8||$` zqCQn9dt(T7Mn>s=m~HP$ARu4AUL07omCy9&M~n$1d53ydQDUHP>+Vgo(305`}NkZb)R*I}t@E_a_K{|1Wwi<#ke$s!%+ER(LqAj{YHS_-~u zeJr?Lh}Gw-kVZwqlv0CF3^}1x|1b?{_ueQ9sDGtq%=~bo7$14@w zZ|Zrj=_}1H^_qt;zv;s8zh^j6wH|m5Iej9^XZDHdo;dMp(o`1$TJfgA>H2o9Ft81qkyM1iTu$KaRC926`l{L^fJ*y)TZZu9A`5l?gU#wS2`q znur2Z1cn%=`FV*fg-n0@-L?7F3B?!@`Y0RUu;DS^u-W4(4=VA*vrl`)Yao}!(3LBY z4fzou_E9uCvM>8Xe9y=iB_=v*0cl$d(p@5R$O#eOi0o$zg|&)sTV{;U&+ru$+WV&;>3ZxTB`ULS zoDWQ$KRuk_^_|~!u^FzCuh(NGU8HVi8{V`+Hi z5n9rSFWg}DyG4vkIOA0!ozUHAvB)pwo;2+ZbkT@DrJo0bacOCjCr>ze*d}U1R_GBf ztkll9MZckef8O|H%^!%2kA^^E#lDTGMR}N2g+Xo`IwxA=!yubhT2&C)@4?i+?$iR& z(d}}PX8;|tR&SoVQ}x^m3Lf|Sp8o^p@Q^-bqJv=wdXs~SE~55dzuy|#+pB1fMHo)CM@$(kRV&{C znVSy;G3r%5U+t0%f{kWx&o+$pAc)$zhVHMC6DECx2q^Pwx3u58-P*Va&_pKtd9xXX zCgjwlglMORphhG2DgvTRy;ik$!VnS-igTa;X6FlvQa*|%#%s9~CxODNpAJ`eWf*4X zaH=}q(%vxp?BtD~FMEX?)#8VAhw<+>HmP^vj$!4gx z9zN7NyT#3JhoEM-PxR0&L}XB*3hn46XNM>@&y4~X;oet6y&bW3uttt?{~$lF;;VJZA+r)@m~`=aZh}uu0?E1OTH;IBJM^ns^A!TgI7hx)eP#N8wyeICXu$XKT2t;VU}8J!7ei1R)aYnb zqE9ncQ%FMv3i@P3fS>Cai%NZJjM}zfmE~D&O)}!ceZQDBtFhuNRB>edvthHzb=&*m zeVzwJHfMn|rgd^xyaAeJ(e1LdD7J6!*=a+4YhgxazuZ7ahjaajJWA0>N3pQ}M6NeP zM;?B)7D#3R7(&~dpP(3rC^{v{(3a~rvSny4Ef5FmHk7A7I_x;ARzMV-<{c$>{e(hZ zI@-sjkowhJbuiE#<>EJO9E~{rH>>ic3sbP)BMC_cF34Z0>}ReYQQ<#rJi)NDhbDUB zUEl)@9BXgS1c7B011oC~I?RXK>`J%6E>AYXl~EDqgNElsf(G+q%@2Rjdh?1VxGjjW zKy^X&@B_Va)O3A6o-;B&d*-+tC5Tl3GoyH+k{N5|uL&`ClV`=;!H;Nt+y$cd!&9!B z$aIe6J8duD0_d{LxC_E#2FU=#;dQpTqSvW1LKp-9c7%EZWKE3IQ1aekEwm+Y>%a}zTw3kd6ItiQ^J*oJ9@q=Bl%3+|chtt!C zx@jlmq({E_9OxO+S`%so?qG`?CahxzPfIpqP3A>SII|bPzO%bc(t3*cwAn8Vv;qIV zWTj@~1@S8capgFkH(HUnluKu zh(ar^XutVMI`6fZ6N#HM-j-XG7>DRUXFNz?HAC&z1F~79ga#FJZ%YFpAcNWpA(wY( z#Zu7}y(J<(g~QGkrwfb&odqOXt2s3*Ig5=Fg%nZ6rxo z$}aR8&4xXi;*UzrF2ChD-6!Fp?j>6>9HqD(NQitMJ{a>*flF4kjB0Sees90i4mCk! z=Ua?CCie>6RWz%$5W4?M-#=SVT=a4&4oI8E2w0IWWXig3_qCwSmkgJz%{~sj?;&{| zW~~Ar5Y(3OoTz*8Hjxpc_YAu$GVtMgiQMk@3X~`d7U*H{1n*IMqJ^T{d){cK-KAfI zZ7jCC`0nJ0d+Q4e)aEo{BP7JZyJ&|}eWIwyJawz~i=l&Bd7o*HBy)FBReXrv`qz5>OS6HLQXZ%y)7gTZQ9i3UCkA?z@U#7Qks?{;^8 z>WLGM1LL~b;qLVJk6FL3NkWk!q$@$QZ))^^-xDhySZ12QDDx00v7{x!ob&mDNCQ;2Yo)X&hei?PMu!diWPT=t6`*B)4GOe~5Q43Rp9$Dc~^zTebT-t!ZttWjIHJSL%#3V0x?(P|?N|;8bNR z-T$tEI^Rl^F8JL`d9L!*qX{=#x;fzQIyu$FlU={r6!KT}9qN%iGzSf0#TjnUx6%-h zJcWvzZ2nJETEHno>Eg=p|hW7@P(UF@T@{jHu<&82nuoWV2HwNHT8$7+kuW z0oPOeIhKvr>7K_#)z%j+Dz|x`+c*99HeMuvRwRoV=F+{>KY;WwI4M5aO^@T@M{Akh z+w=saxr+NvnaFT_x&6V|P}P={+jRKco;)B(Fo+uljNh0KFZTC19i4K%O)Lf45}o-h z4yOnx9+#=+!TYpaOVt&TLe?IU8hpBjd1RF(6R5ZIOnQTXGD@-ng^WK7;+x3&SR`uZ zyB`M~oI)k__f?YPG`!U)e@ienfiS zmx#fp*DVbsUO^eS67&nQMboPy^PzyR@kbTWuVRn&%QN|!*}KsH`%D1dwMKk6h%~x9Ax@?hg=eNgZITA;y0MT z$A~bb!cfJ7Gd?P%u~KgQwaqoyzgh}2lhYskL^kx#h>34Cv`Py@{-x+QlaaZ1d3S}Pe&Mww>p7Qi1VH z(zq#P?nAAUx5|E3^UeB@f?(9sc@?`6_#vGwi~K(5+TGPL(z_QT(wHkV5e$vd2v%RWIA=~{kvAUP8)yC!8qkyoJc_g@>F&cLSi+=G#eO|tt@b96i6C+ZT`89CzOp- zYN&)`NNWL(RMZS;$ih%7Sei*YL~Tz_gq|)`geoR24fgLCSgdF^}sGytBU8LXt zJZ|mGR@~1`1+Id5LCF3#P}Xy3f7tEz_K~QYzxefKfXQF=-CuwGYL)QJia-12OE7;? zDf8Z(wlqL})0~6;rVc*ZTdP8Mh57$KWg5?9+I9qb>*RJ${c_=l*x@a&Be+i>?Wv6( z$-)L}FK>2S^B-hGc)#-tXR5m!V|4sOdMe4=Xm$JwDL6{Kzj$r>OrwC|O_c5{J|pZV zJxN|FsAW{|PF6B7Zop0S91PEL>@MUWkm2tb^N9rB2plf(k^rH#>;>`?8kbYs`FE3f z=koxJ&Lr`X6@WY#CNqCl{PFJntc!9Zc;4T&JHjNrp1dnUQzjY6)e-LxiW(Z$&T zAkOH@7k4ew0drF?E}h@(NF8*~e@Fm=k^!*7t)f<{)NK>c0UFzA{nJPDgs`|B#tRP+ zV*L^BCPe{_@PL%qiz4#&Ij35`22Yp<&F+Cq@#?QPRY&Dz3IY{TR3?&!5*V#J!GE7# zfsv92?(XP-Q}yMy=x1zZS0 z$MXH57{2eA#br;q>?qW3>`G37>X`Mm4yts9nDJmlLDtk_9rTCp#45f1<-UxcY7uRj zm{%h4TyoWwRS0mcTpGjB-S1=Xda3Q7cyry9um>V=MbC2-C?AW!QuEbsl4a9K_5SW~ zdj5YOq+-zwk0^EQ1Jf||qp_ftZzBp;=$YX# zBA;0Gjb)L;qHM-SW0N3eKY`%z569O(AK+5ezv7!pVB%v_RW#y!$1Sv(oajo%8J?!3 zy>vyzH|9j&>iWXg2`LqCR=EWKi?V|PQHH6=_*!~=LJP4FOEo|zdpOwkyRWM0i+s8# zJ%kvdIDUDZVJLMAqQcxhL7G`=z@&8*U7yRk>&r&)&pF~0daJ@UhTM~5wgk7y;xE;SpqvK6c} z>wbqJJ%3`|FV6}NB7V=c*RJz}e3kB!DlnQ&bk#Kfe%<>amO!P6JgZA9r63JBR*w|# z^}l~u9SemMRDeLuP;1O}1~%vA#AV;Fd!8Y_XZt4ryN%r+?)9l4l|1N@Nv*{A4aH)Bu} z;UmZHx_>}RKUiiF@Vpe`$6 zd&t~~Ah7y>2fGHC50&;O0#CdxwbnZ8p)-M->&0`u`K_9jPqADL3%sqPPrPNd;p_FR zZNQ)OCvWRw5P52yaPwaGctfDjhEF*Xb3GIFXn)S{RE~+ph8@0xKbGyZgF6}@1x)UW1sZ8_5Th@o66KW4j;q@sKO39+L9$?vgTqwr(|4)F= zaLfg$H-9Z^0KQe=QmZRL&m$>KB76`rWF``#Nc`T6LMI||gqDd()}&_j7Z)wAapziY z&wxMueqvYh-NW9u?v1Y>6wg0B+}{uWmSlk4o+!rgfcbf;$tbBL}<@yD@bmh_UQj>8AlaKw{zQRYoTV|uTY$W*@_&}f~M~Ws&u|T(GdW`D}WNgY#4v{(&UR-sML^3 zH-#8$-B$+_jegNIRB|k!0%bW6cn?s2*Gn&p9$A4ww{&IOZsx!t;0i@_d7y`o{ z-Bce!nQsB{U|x?GBwM9OsZv&;ig=vnXI~{h$18gMUdVQX3GqijWyK5r)sLxiPo$8G z^#klKnce<0tWkDong(V_VhAR~XVnco_Wa&AH)#qr+J8Ts@ey+ecvIV-5jLM+A53Rm zSf^zU40rw};y9Bl;Z<+_QY1UV=lItwl#C7hx$O~58$sT8Td}d~HU(0qRp_Y9<%Yd z_TPgoHRO_>&m3qBY2o zF%n_r`su*;?N9c~io3yk!rZ*919BZm&~1ydsVp%!+0*@F7@!sfO+XPcK0m8O2X2KV zWNIapqMl>6rzTaM_K|z`U(PTwNkpH@EI95v$6x~g8j3!+O2;i3xnKzg>XbwN^UwF5 z)jJ>3VW*B$q*4)Ws7Sw5D*?MhDf-W0+m7KYU7lmyu*6NKukMPr8UN(^5&)X)k6x%! z6sv@*qZ~c(!m&HP4qwvWW{qwBQ3}qj@ z;`8%!v~yK$s0h5&qK0LB=4XS={;0Vx8R&4~rI5ukP5fZwcov;tn4xS<;EKh{a*KMi z88+V%|LMVKn||N*5#d;yphN_je?!$y<~XL{YMEeDmTR3Adk<&Ho3_VC4*3jTf36bE zoH7c0qP@NBBa6LShs)hhbK#}tX&y&(y(Hhe2hg3oAc3sOJqE|z^sBGM>L}a#9e(p+ zg&LqQvB_e2VPELufzliAaEmn57W(>=(uwe53+zh8*uYX?d`befO!I z^g8Kq^D*+bs!`~Q?VS7|!sbJon^Iu9VTW#zZ-K3(-Vcl6DYM2^3tavC{3fc8enl3JTI|ySK*J zzIntL_S@du`?_+z>SL>|9ZC7=^71?y4_Fn3EvsiM)oDCQ8iVcl-tk;35i<6&B^X_8 zGp88$7$DQ?H9CNR#BdJ~%Lk0zzTE9t5A(iA3PGB3YR9@d{+CuS@BZ!%HG!`wLc68*}3O_ zP9tgJ{~UJgb*LDflf**G;r+Vor2qYMScIbX=V%)>8f7dF(dI|cuGPQyQ`;q%vFlIT zM`I`no)2I+%>g(f_Vy#7U&PK80Mb%z)L}`*u?WZ*kybEaVPPQB>Qya<%W74OTGH<; z0j|%rqaP1!W68^Unp3I%nAS$)T`F~A5!BU-#mpYM|1CY(JX1ss=tR52cmm1+*f}e} zqTQzQL6k}mr_c6z%r6u0<3M}HTzFg)`ZqaLRMd7r8=vkT{*Rcs{$9{-VaA8)Tl%@M zex}|X;%&Te#b!ulVq3s?lZ8s-o57lKeu>Y9dgqsK&`WF$^m40FrTBZXyQMRll-P;3 zZwY;I4p_aU)U#MP_Rb7>mbymm=snW)q0ly0AmWw8@}D9*JynS9aaCEAq)6gIGv*wU zSyY3dxhz@#S`@%uvCU6$_Mq-@vL32YDE3yu`@ouolnJbLga91z`(3lmng~(KSoK%s zB;L3=C=i18<}{z~|JDuxL*QMm@=(!>`RSy_G@yT|af*H36K=(Oa}}J^g3|uvtvzGUq+b@QD=llUm@XPzTd^~9_9)k#mZDVfuP~f@ND`N&L=M;gH3TGy zP3w#d;RTwIkYe79{pwdo&HvLJlNsmpCExYt_?j8muYFt;h}M^`lk09!J2|EuxWvlf zu($VQf_MKSERNO8qB!u$!c6Y_x?wxUm4If}7;toGY)!Q(= zib-fRNaT^X6<=TJs7fo>58>`y=jr&BU;cn@NX3x{r)jHT_ULz8Y$qw^RyY5Ogx4ZIn$`o#X(GK8DiY4 zQp08XtPcWIR3Z$@buzmrI%C@Fp{%pSzq+sp*l1BP?Bx3txXU3bf{B5cj@<^?rtduX zQ-T9jdsZD%b+52ks4`u3bZ-oF?HJ>R?^4UNxnmIN2KoKF)7-I{R0?rfbD><`MCPO& zI|cyD@!_5#%X9?^X-OA1TIeQradju#*w>1Mjg4*Mm-F?Z#@*tWaqu*%$z&E@Gz*8U zW4opF>qw)%@PSbgpRS|BfDI%a(mgKGA)2KKGx{vX+zrn1@K>WFgPYY345M@rIW#{@ zHLB&yF+JXRkRy&dJwLj*q^`&3h=^MOA4)q+T@(9UvGjysZD&$)LGYB*EX;(IkHG&E zI-doL#DzB0q=K;*y~nc?Oc|nG?o{ycQb2?f0}InhKU5Ar+u$rjq`k*xSaU0=!#<5B zi;!(pBNE+Im|YvH(!Aori$&~v)B$uG_LO+xtXu_`D8=qr`QQk|p&)-|r%1L!cM^Gi zy3vD~NXc3yjFkij$Ai1wsiC{|$7>KK%&}FQYF$n2Z=|w5lD`z4yIbY5G=HN{8(K!< zdwhMTvKZsFr#pk&q_nVcw$PxGsFJo->SN38P-J8voR7yGI98TYeCM(h-~w`Qx9+(~ z@z_Q{dDAV!+7er`+IG7hqC~Gc29(N?&V|=n%o(m2AuJ)sxeuNJRdjLUwP?YByPM(k zcT@IXRRcR;0MOO+FzNj;&KY$q>K0$m({Z4$6Sp^6umNDDaVJR+8PhnLEJ+ecPf@2f zzx5Y!B8<5>dkroDS`JJTZZQ12*Etl37q0`rRRY9kh0QYo;8by1TyNsL{fd%dg}3gf zJRT^AF;?1q$MU@zg>;R;veJR&dLt!rEdA>aru$j#g_ROk@8N|SDBXAXdf2cVL!vaX zRh#a*so8cvkvyF2y65_6nWt@(L^FXx-Hn>&L#8?cmBBfhp4gUx9{?nTm)Ww6eS;?>1fQL?jDfn8 zImUYK&N*iUX>{?nihx=DPCt} zNn=TtzcV|=zx)N~b{000E^#Ufe_<2WY$+UOZ<-nL3uY>gnPg@ydw|xRwnE?ec3>>HOzz5JW@-Bccm2@ibNVaJa?ML(_jiFvln#^#)_%;v{GA!!9_Fbm#`MUo8~N ztfd-_-79|DD1~@`bHW<2;rQ;;syRM!j&UCFs+B8%B2Gz8oOSSX>0VnQ3_;bL`)5N(jp^|x1{WLf3N5*;Ewt0c%RT3FyxWXe-R3LjP! zDv@eX3#pfA00Pu5D#byKm~V&SIs@H%++m~rD{yGWwNmWJ{0L9G_mSrCMVD$$zbDF! zKfHIhLQpzjxE5cn4^>3|H~Y14L)tm_@=Y8fiaz)Dq=+h{4i{ZJbVyjP5MN>sKfk#L$Q!bFNd~41f2|4AD_&S*%wm3Q} z?w`IF?)NM)-_B3hYQqq9)sZi_H(!tzIZMka3rp}qZ_)gb{myRhn{9){BqwTOdnxJ z)x42F^G}ButPx{&qxJx;r>McmK0R#SQR~zJB0FHj96g>;AHLlsUq8goO={$i!lYPC z^FPHpnZiH1Q;9_)stNOHvzdi+#2=kYh@6keJu#@XXXAX~6_sjan4@aNES9c3LAuw5 z5kM&Bo{xW|hI6kLKLsTo#rPY;5#Pt`vXNh+AbdBA_DY{j4>51(AqIfb;het&9Lh`~Y&e00@10A6b3;$Bb9Z-} zz8TgnMhSWx{!OS5Ed<|j5h-RvvM_nne`b$|m*@kZxzvOJl#)`rKSFcg>;_dEJVmnE z!&go-`K|%|kQ3-+C^_pt7TKbOdR)z3{+uJ{4yR3%w|zGY?IUx=wbc468918(c6vjN z{5|S{++)`OZ1SC?S5X9^6`Qs^6iDtR8zeA_odcopfD0SJhjHl7@n>R$5 ztFT*ZqHKNrEbg|}>a*LGvCD5wUVJU?KSB>&uTQ_`Mf{_>5_@Wp9^Q(?{N$`Ev;6b- z84A&>#6}Y4=|yADIeaMkj)3^%@J|B7Pr}sj8ad-IbHm^w6>6Wv0ndVn+9S9~=1V~) zfK31JGo*|T>+7*W|9Lo80lMF~78YU!=5zMz1!QrGKg<{*+|CB4e7$t_hJ_Sp%BoG2 zP^OnvK|H5ds*IsOa4t2FcR3{p{dU=3E8Wf)a(7uU$QHA%Ykq-NQxs~X|JuaYDWpKZ zbXQxtKE^kD%|n)xKonx-Cp!Lv>3XzNdxeQK;V#JHoy=NV~fqV0+hZ| zE_|Y-h23r?=xqkKMF+nI650Mg)VeHGZkk!@z_w?FV^yJ{cirGLL`E-ZC&iB|%SLm*3}JKfeXN zZWVExiU2A@`PSP?18<4(ux_Sk1ItwLVMi%c?w3Nl;pBc(W(e%pOJOXeT9nqu z33@+-51k~dUxb^l_hr^4+(k++5h!_FzYyuV+#&s|GLe6)ZD^06bf1~n40TlJo*U(^ zY4c-QYaW^EUkwWPaj<#F^k^HfWP=Mh{TzqpEX&tlRq)7oPcR6ExA~+kc~-X~R1|H! z7EOU|xZ55Ia{sh-Vds2bM0@sC+}3y8hZ5C?Z3}H4q;$4uuaKi&(U;o;CxQ8JrL6#V z6IEvOAmYGjkTUid;3d%JU-QRc6~=<6rV%bj6i4qRd2J*FiC%)oa-734P;16Mz(o{Q zj9s+GJNg|B5u?T3G|%h`KlFQb<|nH8{N<1&R1~+ zNHAxVzH26|!PcD`0mN~sU_Srgt-*eeUd)p>fTv?2xNWx0zUzI%6QBhUb(^&X!dhCb zrg}SXSP5nzv92X46t5)DjD}7u+(UOM|Ez@pm$tn?KKQjkE63=ZSB7v zwU=CVc}DEiamxas2K-?qHEBatf27Xk>;PkVG7#rF2NQ(hbt18%B#Xf=CERgOqfGLAR73C7tj2zCXYpu$`TAKF@RC z*R_Sm>INX1UK@==$_8v1_pD*2zBk|QKGLI;Gj0Tm!r(WztXt+)$S}wOe+Ky^M@sXbME{FN`+3%6duNY`6p?hCme0aZ*oAWhxR3h^R z0~ezDq0o|xXf0iqZ$4*@HzEcu@bUhik$oiNMsZX41AlGAoxx{w?lr*&_8O9)udXZN zxn~*S13@nmP}ewJGTS*dGDn-MH75Rvhw$PR*{6zpj&RkMOn&`>;#||S&xE*>Wyo1O zib9#NnB;L$J~cvZF1AO+LbX8_7Xib!Na)`5k}L z41WfNBUypE4PWYo9FO$QYxlI~V zs}7cj$oy|lIW2>q4cW9UPD<;~ztsD~lp`OJI{U#)3-Pi_0Q|>L=}*z>u&Fq?F_j^c z(}YghR8Cl#ZxXRZ4Kw>PnAN1fe~Rxb*OnUe`Q)L+IOc_g^($&Q z>)O^2^{ZM|O~03cb@o_q|BRIO7ZbaiE}J_RRjb4B3l>=KX3|Pt`PPJpb9sL{7G!sC z?dDLt?IA!8Y`{y)y#@JA{%CA0hc~H?kmZM&DW>o=JEPXT`$@7Z%>YLLZ3&}pXD;)% z`vILN_g_SV9YN9eOBCf9>i)Zb5oze3Q%MueMi((J869!W_ydz9e2-6~?sA_wE-PT5 zU`I^~L5(NnhKVBS7bgfe5L~txk-fj^P^v6iuQ~eA9dMM|9ghQeu=D(Gtj=`&W>^KO zXVf;($=4Vy;dOsp*3t%-c5Xd+qvm>?tmuz5^~Cb&pAjE<2>X?kGN>j@WSnBpNXLmX zhpc@cGfRWMi$0I!qRtP-7zwNd2|LGLOpO(>Op3c#Lbc&g@-^!CCcZ-{;7z*c3iY$^*?3YH)KD`P zFt8Aaw%QPcIIS67UQ}HNS}r-uwek)>1kh#|Vq>f_4u@Fp@l)+jjh9Y5F_nZ~Qqj2c z&E?p#f}+vy{IWz0oN<-2XS%Lkyy@-T5~JC#++Gi(*J@#2hO$1-A*dTH$jnPp=1ewM zmFwPT*b^0aMx%mJ^i=)wy`N@x7a`db=)G!gL_yaA$0JoMK>C08xUW|Pyf8<4Z0GTa zzwaQkdS}o+D~jOUNu*eRjG+rLDmT^dv9!S4;TTHS;3J6OJmZMRtY%@FRVGjla_R8t zag|g3GmY1d7HFBA6s=zsuFtkQ?yM+pdG4vQINddDDs6Gpq0dKOP<7ww5DiF6TiBIu zpmbgm!$M{|dj!hlLiU>+iLHS2jz_sZJjUbV2j%|L6TqXg9-nsIELy*ZBYu%-TO(yU zow^d4EoHEpr_!n()Sj?fmX;HM(2%jmq5S`QKTp-OLr84ZZF5zi+J1kgtI+oOAMZq@ zyf-n4%0(m)EgvTR46J^Gk2Bw5ktf$g>YqPop>Tsx()M#2ekpLAHOw}c4PrOA{|QT( z4n-9C5s#KR(c={`@L@(Xsiz|JvSs&z6|4x7k?h)bC!UrtaQ`J4)e65F^;F**7|7d> zvik>_bnlG=Halq@j=f84BWG|_FKfZvI@b~NDc_o9Q~yUnwd|52b#0vyza=_0(gzKw zH_LS)e#}zWvq-TadM0fqv@C^un~h*eQV>IB9h7@>RZDsJr{i^8vs`xRx|aZHZpq)3 zC^YzVM`vpdki&V^29xsnpp#210kjY$%Ci!(sA+Fwu7LR^Dnod{~#X4v2iM#AE;SV}`OAOsH*D_wA_awBp$fgu>0gzU|@yYxLr01UmStomvgvMcn4Z_h5R}!dZlEA zVb}^BsyK(Jl=f_m|Xash#SwARcqk`9|@Ta!e|T;bwTThJef_H8XH&e{$XX79Y0893{5IRHNHG1Bs`$)yi=VMF7&vun+p?3p+Y6nf)-B@Oa|C@KheGLL20 zba4QeuDV?uc(YA~P3`A)mvQ$2H@A+>kucE{Usa>q^(>cKnXm(PY9WD=?%WS=-y3ooP_RcDlm$!JG!@P_;dhzFR7wu z&$zOD{{h#`0&J4r^J2>B41)c=Xd2eevk_e z{jWp>)VeKyZfc+SErc7IhiFR|0u+vbVjWGxvgP=WL8U=7{6iPx=9kEZZ~R`&FRb>5V73BXAZakvO=^xd^vre3sV#pfE+j`b%tyV| ztX`UJZrtmv2`Z68UktHe0<4w4EJaH&{UO1I9&v%!y z>Dk}iNXu5FZw+XD7cds2?+yixql}b@!*Uii3e2n|PtXM_zJ~V?lBG@82{*=EE>bO1 z{J*7UA4bV%8HSfyF-E+lj6E1Q%`tEt6^&0d`rxh07_hhSr2iR zMI0A}yC>J(U2&8cf*z-Fy}1S)&AT>h(kD z@Erc>^84W_A1V0sAI*7g(mn>S+BRx@amScM(h1*e+i{IsHEKAMG%=o@B0?fwrs{ zXNjLi>!SNZg;fqMg#g0u6^|-uB? zWw>r8y0yP$6g?Y!EvvswJwE&`E|+t(lkl~EQQy$G07fO|1AW9AjiH2bT6VyB0)n^Q zpenPfZM;jtDGY2m;Ux3++|l=^9V0NG&4ax(GRvh1CH>D%`tiriQ0GT-!oZi7O?#H& zH;fuX3BBHB(DS?km7T-M*`__MzBE{!z(p)E6AWZ|$z?)_ZP0Gt>R?0sMz1Xz1z=D_M*sX!ih=UM@O&9zUhY8b9 zN$uK0@tUUpMUBf5E@?Z+=kGF3lT!%K0sH+$pi^6e8~A}{t%b-U=n^#AX^0we>jTCK z)z2U=3r86nB+aw#C$qT~6O5R7<<=E{+*QL9r`(pZmVQCn7kt^vL7K322Z589n+Qp# zkH#qN@?@3*_M&0%C<$&e`<&i(!ddf=>e+YhQt-O?P@XOjv}GJ$;ja&VbV4_|)@>gp z)Sl41ys3u=K?)82R^$VUoMD<*rtSB2DxxP^rUOT@?Ye6i@5dFFd}aVPVsAF|O?}Dl zDzPEy@ViG@bc?vb1%209 zd0E2}$8K0I&PCtb7iELC^y~8IsFN!6f^g2LBUNK|lh?WK3tPOTqaB%)LagxAHTHo& zUNrDj&V$+QT}xrOWqxWB=V0mSmCy&LBJTqmiI8K`nMZLXeKT|1ZVk|ufq{#V=-@Fbe-9{e2rYO=+UUF0_x+AVZTLwf?=!GI+{dO95%Lg2=@!&%0 zGr{m@@If?+bXkm4Kz4@?!-nHXusjJE$hHf|!t<&wE7M9ePqd9I!f>S5{<)b>TiFi2 zuV?2h;ePU@%f!Rv`cC%_U4EgkhA4gxn1=-f=eP9!o3&qPiH>Jq0&Uy`5J0R-l@Ln; zHZHp*w!rCSNEeBboA%n!?rHNM0A)~or<`yRqre~g3!Oq`JwA6VzuqUC{MX-;p0?5g z1T+#yfQa3O5S0gW>z+8&_H86Kl8C_xrWl}TGholFVRjThqQ25FGFY@QX~dl|QFhK~ zsWYVsz7EEc(ozN$4!yu;qV~V|I+)!_49<(18p?~ep|6Gy; z)nW@zZYjxRHCD|oetWh&7o~H3`Ren+@?eft%r5co^}VB;l5S5=8L~k%i5*kU)NFD% z(Z3`Xr5MZN>p&WOGMv!zmuD_6ImNLa*cuU(b9XmX-oX7OtSABg?g!GE#JelG!+^xN02fW6+h`3pYwk%bI9d$S7uWa@sP`}T^04_cN6JhBG)~cC^TnV7zS1O4)t2H z{f2eNhNmA!;gZ{8m^`kMnn3n;tFB9{>cZJYDCU>iyd3t`%`)UXKepm)Z*Z8kH&(~U zzITNB8eq!IZQ50R|- zABc4H0mcEn)^S%wbs<3#5A72F(@lZNn5eUPVG(#yV4A#p=AXq*hnw4Q7Ld-zq6ISW zk`u$d`bbq$gGH_RV#>%$VXCurYGAjG#3h6I?n2h%_-Gq-ibv32|CQ0pAXU^2NJdy? zl_XzK^2D~|Z%I~TY?t39Rr8%I7$z3Kw6NtO)rqoPPY16*Xo-f7)7i^Wa)oQ3bW=v{ zR|WCMR_c+9!YQ5)(WGa!eT$<=eTg0Kvx1il3t!9YMN4qN&x$wu9^Dk#&_-rbyle+E z&p5J8kuk?JmnUw*X7un}W0X9NbXl)Rh`0IlD&CkaJy2w7oc6j+K3^@$z`5du`1d~? zcR4uO5Qc490RM7QtQn_~?L$x!3{EX5l2ZN=%$wy@_&pi*1 z^dNiaR8N?_62KvqH}M3xLeQ;m)o_gf4Nm$I(FDrb(UZN>l=Y6c`X8uB%+lro$W9XO z?0`@j^h|tbm@!tqp8N_uPK^Lr=UweB0cd=QdNz;eb(mrTjx!saWQwR|ELcwcsRfGpfUJIhMGFs|4CbEV)P9I*TK z=0&b=Uy1jgq<}0qlE)^3i0rcgahGnVkvVzG%Vi0~QJ)jU z_Xr<1V(3QK-pL!nAY{5dZ!zK}D3y$x^tGQ`!6n~H!DCIVfO{|uq*>umd^yh*Bq$wP z$)qZ;F**U4k0GG>Ev2ZA+6zUU)NOk66M3o~F9AkNtK9o1bYMHJBLW}`U+Y&~$T5Im z^>~&n9(K>9SJvj~wx~%vTm4=x~Nwna7c~RS7+b%HMV^1|J0xsyXqoc?c#r z5qj_T@>{)|aRXydK)xz$NRPx~fTXi_-8H}S7AZd4_kLTWr!3a1a!WgW@fCNbZ6u>f zsOwYYZb7igVBH~cSvU@J#1h4iNqwh+gnH&V)a8I6T3D|&S$vRy=@lj;KbJn+9SB>L zB)n|+Ih-08&s#tJ#%*m_!3uHsMzmEJdtF3o9F_YQjYpZDhtIg3!E9v*xX3^a#9Mjv zoX-WddjGlenfl%>P5D=x(s6?juvGeK{XJrSBmIZY_js-Zo@?CQLjm70t&e`7hz3{^ zkISQU6T@3QJ$(RcT5GIjiW#sr#|CtTA7SB3zcka1G@L)Dq7PEcaGYe&vvZOsH)1d* z{59V69~7wLb}E@h0w7xxhJ&A#j8`s3(&j`|OS{{bPx4^NQlXE+Q*^v|UITuCTiP>I z@ecjP6TYzGz7*GwY_>B7rw=&GiLgj)#I!z9#MQu~edt}!$SeHnuiuNh#aI{QA~4|% zJ#pD2&&^WzWbTWQ7|cmbHP~E@AJ{mRDDLs_Zf%8`cCuOF5e^HuPwR{rYT zRY1iVw`+xyYYkjj*{BnaLV~y#=)~8qn?g6g!s&mI=~y@?s81Eb+&eAp-V74BQx`WC z0l;&C*n%dd9OI-b2F}-BGByw+ioD*7>W~R22GaA$=4FFt%303$iWlk!D94(=tcq~y zr@k?|L=gk>US>da68joQym<|;cWB`j%Ml$h4yGpdL>&+KVVBOw*84OQ)uwr%kWOfGaki* z_=F+D;h)yG&E%1=X`gEKA;^Lnuh-+Nns6iOP*EW8kg$_(odxbMT&V0@vu+8Y!?8V=qR1_R2vTIU zH2`LH?R7EUuL|a957~tW?0S1K^{c}F%8kJfr%rPFXHUC4qz72L>)2}CWK`*vRazMS=nx393fg>>$;wUN&_cVgIe-v%G+ z@@wb;*ZH3HSmLEag@J&y_32@+K-7VRTQ^YA%^K^1vkg5{Av8*JR5Fc6v3T4}_H;3wX-y&QcZp~mzlY2FClW~Ycgfl5G@Vo+BQ z9D)D#D+p^cQca|I1DC?kQawp6%~iEcaiPM9Mj0EG-ztvEm-b5OtpTxM8V6AO^ZV7v zDe#Sybr1966T`GiMI&@NGiu-c169to7KNKEMlE2ly|~Au$e2tNPLVw0dt-2?*WOwU ziQi9K`8U&k({dIBTvxG0$<4c!IG2-L)Z25}++pm={s3kQVajli> z>HqIjoczs`>MstfmnWZ;N{j4v7)x2pU+g7_tobV)EwN^tXe+dj%q8B;H+j6fpUmv? z$LqcV%*u>Qsn-B~Ssx;Cf3LgLJYsV@jQ%XqgU#?VIQLCc$k)#=0p3re(~9gWmbe^; z(Ozn$o^elVXZWsiu45J$Zzan(C&VSK~`jvp#5NP&RvGLv#xUa+| z<_J12)|DUb0)U}R3~<0iqK7q$Af`|X)yxWHx^@{WAT@GcL$GUsBteCse3L;0x8S7N z8j#Z=ZN8Fm-`)=Eef)2)CP9*Gydw#hcP>FevyWbE_B-w0z-8xy4#2D%dmJ}m^m&!7 z>gzZHRXE>yjTmr&KzSy{g5zcP+LtB1h#4TP$RS8sIk2loY8)H>{r58T)+S}^EWG7p zE9>pYj#3hyp|&NeW$0Ue%qz1hg(PiZbyTyoTeoROg$jaUjNZRYph3mr41P2s)~!3e z>l-`~T%LLygV%jzYf!GmjRerxGM^^{H_qH}m^^{)hL8zaS?e$!JfS&Y3!#F3EuS!H;vX6j!(7IaGA-;6XP+B1)mC9v>N&K` zTDI%O37C>c_5+8aoc*21q%gi37$`Hu^V@O~|MVZOJ#JN0-148qu#Nfc=VfJi(++BU zY9jV1Lag`F%ed*;G1fo&_)e>1Y<}H3_A^&5BpnGG}mS%Hx_Q3#-K0 zA1mnKn|pW$A-Ca1FE_Lee)eCIbDV-!G^7YmdhLhFr~}ojqMNRx*zs60L5s&x>2%y9 zUxDIn4Nc$o@<}tsM&71`U*XK4|9|sbWgkvGauPG+ZbE=<^~^Qst5iipE*%#i1CJ01 zYptT?PglC}h4nZ%=L(;19MvF~_t0LHEA;C;28ltdE^?P-Q*0$~`F7FfSS=HP zPVQ%S+6#mi%9e)63qu$Mbv*2Hj+AVoeZJ`;h};|Ow_Vz+Me|1@0EO$wx}p6lKZ{uy zu)&5(9P$?HwU72$Nq5V_d#S@tP*cFV1*NNL7L@BKHVe@K%^ z5X6GdXZQxtmI++oq>49iReaR_Qr5F^x^Ukw?_CP!o>ve9qt(0|L^yY+JV zYYMz|uH+tbp?e^dh<3~tE#Pk{Y3uO;9AO;f61?o=zmo%bQ7BmCR763c7TG{EZld2( z#X8#JA@`Suc6j)C>Dq70p!q(Gi>I?ec*XNWi9CeB#Cz@f8qQG>sfY&rl&$pgUG3&f zdVKEJQ72(}p;=#50Ak9x?qsznvwGI}2Eb#?woq%3oy)izXy}>heO^Hqbb?|~V`8`V zX|r2ynZ$sfo>Z;4;KY)x9W$LEv;dm?0z>TDT9;Ff)MTW)o0bB5*Wb-C(S$#1$uxw^ zpg7oALshWMDm`cD)wr5mem&(Y$-Y*uAPELsE6~+iUr&A?@cWKrH)EUavYy@HSPf<| zUfuk+^7fCYJ+s=_k3N~_#JiCV0eeBP%L0fe2P)Q-@_Mx{={;gI?OlPs_eeOZ%6~=k zw$icQcF2_UU88qkTKE{m4XF41eP6aO+}F&5P-KHKu&k%LFdH57B&IRsy=n&O%%=-pV~vvzPLAn2 z{;{GqV5AC{FGe1~<>Mz2mFAT9M`gWxvP`pQMrI{>)2CyVPytrbt*pqcPY&Y3w6WQ> zO1cKheXK@Ipl54@`DC#Sx>Aw;R(l$ucqBu3xg^o`2~0%9rS_ujbJ#dI@+q`ZCr2H} zJM_5Yfs+~0i4JUkY{QKWq0>uc@N~w)qw`fc$#aO@pNll>BM>S0<+UO?$Btfm+RBq+ zc;Sg!C+nZ%L#|kPFE|lfrtoux8W$SA{XgJCe6eudv~2h1o083-{Gz%2nQLb=EWO+o zOdr-k97K`Q1i$gT_Y9lr(8E*y3eSJtoE)Fc!4C>nr_pT&WjG;o&X?3S;$ zkaiPQgM^a&r}!nggD`MiWB=FTj?gsTe%JfuUB(zvk&Q%{)A>%+CH)lh>GF}HFvy7V z9hXS8u>fXwvs2>lesu(tTKS8zkL@2_h@o!0&*^o@*N|p1bT2)EfApY6Z1*OaYn|uv z@2%H8_=qS##<8a_kUT0Hc4~}7{`_i~VlkphV~978h?g%B==`e5qjfG4bUaHI%fFd# z>n|vFs1vYTIx_E!mh*s@;y57LrYD(ma&@Hw#JRWMbd*woISXf{Sb-Wwtr|pVdLniq zF^4hZr>XwEihEaT0kOuwghwe~d}}ioVP44DGa`&l_iM2qE04GwZmr6s(rfczvN@5= zzPUnI$8em3JXO@wfpW-okQa%b@pCL8R*cePac;1np45?n`yCt+e<16Jk>8uBkp$Ww zFa4~Vg_3K+bZ%X^9YRkJe4>4sL0>z@-+l`-7{t?n&Y%Uix2ROvuUY=N?k((f3bA)Q-j5%0t6SWo$?EWF8~n*}y=R&n0qd;e zOlvKk>qCtK$@}GQBm1>_W0L-`cL^|}Menbs7k&%dZiJ$jnHNMlCEquEB%uW=5q5)& zDQI%;D_gu~NhQG~WeGFImE?9V7*Dvw4cvnC^)(%h8fAS~dO0xi{nNR^tI04JgEBKn zgjA~#5pdXFvT{c=4$OM%=vpfnWlz-d_9=%#;K*i)}JX-6%1 zMyZYy=iXWI;oEIdzl3qU;2!kd@j>|60i}Aw$xV6I9Cn0%O)H}|wBzM#1zo+xAJ3j> zSCb?0e!dH=0;F~{k-fHZ^6i+?43(mPsZ-2Ynd&wyiGKa9K)3!@EPE(P6D`PddP1nK zpcW0WS10xr8`T7pS{1Pe6|1*~?S&~c?2+k~jRsxoZBq1)&SBRje14FM2pI@$wxhaP zU4u|posU@>blFdHUa=2xx|IFa1goy*G-MbjmV5<&U=C}-zn_W(ZA+7clQY`YKZh9F zEX#dVPfS!e0soe>=fhD&thMn)2KJAQj+6fiU>q02^pImektU9(_qbpABAo`{EFsJJc*VM~fEz07|qkz&_f>xn)Fd#3&*gyN28U;|V$L_a#d;K^wPeajaXsa>Dyeqh0 zHD^FD$eyqoH?Ir^@h7$4TZpo1Q6tVsl{&Xc}1| zwD|WKem{{G8}6xOnv!$HC4Ltr19eviNKr#E#Urb(NW2V^$nY$56GFB|SgK;s*Cwe& z>b)cea(hN|DmNLEff~SQ>Gv$L`NimuZzH@Ec`1?Blge}&M*HDu38wtE1%lDzWCBAV zgr63@6xX^6$!HYjJ>FKIz7@5xL|A=d;DoA($Id_QEXNDhwEt znyzdz4a+8uD|-j^>c2Lgw%hvKyK{_n zEMya8%`uBzFqP?e#-{CAvqU{dhT%>LA*u-3tc*z7U z0nWAP?Jf7mJsiNcQP=c}ALZ4Y1~T`~mWF;Z@KF6;6Gtk#v}1RyOSl%A0vwG;4qfJU zI9U5JX$CjU{@Kdp8NTQ{(iF74eN9VF=lOgNbv^{hs7RRGtJOSMDxeQtyRB}ACr-NlPNUzKP5w*v;p3#Uc zM&gBv{4A&p0`oh4B?S5!3L1gotc|H-mG}%N4y!-X(85Y7f^|O)mXeKz3$}`YFn@E zvD>r0J&-^?rXNsFtUugGD05G57Eo*2hG=c=AanD!;fvxlEB6ludde-pjUkjF4r2qyCmI5M>kGd%F6pnkdc; zk9TnmbO)Goi4Z3FStbvmNfOt%uGqQx9XGNknRrZi9VJR|WN>B%A+=%2F|tk=Rgh}G zZy>^di!7x5^EFPC*G(-@OFfLU`nSwynu-chrTRLIfjie`Jt9t+X!e_>ihIh;=dV;8 zT0&*o^F*`5c%Qp_qTbBC5_Du`pBh->CN$f=RLKOV^>(V{wMX~}pJNj8S=o>sgBT-? zb^)lN&ojfYlb(e5zjQ|@Jq})&aNc+<1lD)pUf0N0r%IL5?NGPpGFm65km*dkrh*dJ zg+qpO{#GiB%NPM<;E4fEkzTW>VWkdm@SM{+vf+*nj05^d8$S&#GLLi!5qa9O)x=Jd z4wkE@cwAEXC{H;DyW2gAoLL=4e8&wKuR6I@!MfkMfq^C|;7u}T)4oqSTnoMY9XgP+ z$z45x@pR4j*PZ^tozRs%!(C&LDwqFeus$euqXXwgyhs6)1D`Nv-4HYqNeMK59TvSfu;(Qet@DT zZ!Vp^<))(>PLZIbCrcbt z(altoWA>8Oj_4UxtRUe*M1`NeEut|X=~b=>H=j84wCuh8VK0v$0$`cZc58N&$6 zv!pP&;Z-SVXvgplB14mOrwyBUhz2D827=YYSQIq9sgO-@L$j(bzst7i_dQmw@mbZd zH(^Os&y$8|t?NY^%0!}qCE1Rs9N5vSh>AE4HKA6ibud9VytQFM#e^KJC=szkYeJ;A z3@IzG*qD#{Zq#Z;#=`YJ$>W~`sp=q3ZmXkkmWwgg3mLjhzU7Km_YG7au2>MkrI9C$ zk7>;ObQFG-9TS5k?t2)^q>?<}vf4d=@JVigK^sZkDZ>ZWbQmaJ<~x1NChA<^1y@bl zuWNk7pd05e?$=;qdCzZi#4H%H>dJ(7VeoF9!m!6dfE->25Qq(13d=SvD(|t*L_V80 zF#>d*-fH-hT6WnhX3&&WAR4l6KpDImaNy&-CW_zvv}lTqlna(%EZEgqNzz+pCdEhg zs13#U`Ekn2$Y9b7$#sXSsuSN)B*I$EvXx}dm0C(AEuR~E+`r*c z<}FDeUWqp4SG}hZox_OOX|)`eC#RO!1C?4piJYP<00O}Tv70QBrwW@ zBWE8iR1yBND^qSgyfP1AS~4O^Ye#>lH-@73X6wbG6R?ge4AuZPUae}9B*^AWDUoW% z?3NOESr7#f}vOJSG_$%~6z*8@hPBi@yGj^#=WixS%H^mfb#wx)>qk`zwb z+P&v>A{-(&n1whuB>}-f3!5FUvZe++wkdxh6~7ITaPTD&0@QE^-oQHWbQipZlq5-NX+I) z=*J>995x(O990ojzX)$Xzc0X_-;gixZ}RZyX3fjNKG)25e>E^9XnZ{Qs_W`<+waKm z=O?}XVw!*a&X$1zTbbPd*qcA+B`bZ2E`mqM5FdtM{wsfq!C4Dtzzx;^bj^M1?UZI0 zypWC#k?U}TM4pcMxGL)!7;jvIRM1}qF#H<9+VC|$;Rot6n&hx|?btf&Of?|)Qcvt@%Bo#{>O*oel-@EksaGxcrDYqyPT4heLuvl3h|Du2a8Zz-)(5!6m zWKUSMJEqwjP!EH zez`Qj^h3Braq&WgG59ns{Vsd-xlI^hb9HX~{ms9Pt50MbR)M7ZCC7*CXjWcc;FL>O>~_e!dm5i1!D@A?3YI zE~&X&8Fo!*z7wisWC>LjJm$rF$Ty8r{sr+sa)wXcl)bnn=E<^4oN3s?jG%5DPyT*A z6;YDN5>v-OdSCy}$76I{3;a*xDY6BVDD$3otKD-_k~EnHPNzG=Y`+h-BCarR)cne- zTYslTOEMv6EITkq2(1kVfj!M>rFQw=ba~gcmWaU�|xwcwN_lP5;b>22S*LIL$2K z9Jb+b&DTNP|5i~#3SzHG5$0Vljl@8r|6!c9k^93tSrW`o`Jij&n({Uw8OG;>xJO4hfd$LJ2s9fxtzPFDI4pcHANXHFKZf;zjXlWRZG%K~o67Yc&hC5D3;~^Q z=4z7wBvdI-H_Z{S!`T_nEj>^bDjc~f&11vt`X4M-^FJ$<VA@PMYTKF6<6NE&9)Cei z%|YHQj2CUc*eP4T!l#Vvms)Rgp$8%G&|3@oLW3VOGh4GK=-9Ep<>Q(CUfjPxu-(UE z7pnl@NetkohAS1E;w1es9GZR-FOkcC1E^h}ta`_L-38WCAe!F6Z0}R4h3nAX3023H zDB*%uN~Om?yDFsL3JKgczJ6@ML-O3+R6s>a@@jKlDgSf;E)L4s{wC&>{R4V=i=;1P z;z#KRiJ!}u8I;)`nPu*c<0M{ludK&$ z2HB0Q?s1&Pmz>cYj_01DP8z~0F>)7AzD!9gq%kV}R#-ckF6VQ&1+ZI*GLP42T3TwL z=a%C+{MYLdgaV|&z=J0zHy3U_z{PciRyBC$-!S|4bnAUop!5moL}6&-p-iXnzboNP z;hW6q+7FKtLV{U0I77>po~%HK6!D|1NU@f&Dni$rL+y~DPp%L~(K>$GU5GbCS7H8C zqFv?s=42-p+PI9we)>3$pA% zC(s$ZS%-wo$)|HCa)H=4HpT3{pavqoeQVnpi4>`0n_NXno%mSarBn$zqWp`qTaD?E z1)yK@Mx+xfw|vfSf1|fAXkt&==yl5}Z(piXwz_sK1=?~)KC6Fh?WuBa5@*Zy;u2&# zO@FXP`wZQOWh|?MRZ&Th$9_y7n9R9C9eG_?V)$nTEnN6hBOM>;uCD!sR+7NgwrOdd z0sl5{b1zy0tkU(CLq48l;?Y}wYG}cM^t5l2^iMF61}2Zh!5p3It+rvLJogYv+?_(w zx4I5+BAC_W4>@s!mFV6=D!Nw;U!u7)hr?Y(Z9`uYF2QuGKaiqMlp;d@A;wN}pw~f1 z!ES7h`d*`6*Y1fQ-b||s`D;Qb2U}K>34Zw{>$MK|a-ChF)F6+@w=%MwoVH(>mq);p zitxRt2P$3g`HWOQJu5@n$t9$WFMA6BU$GfCz;Ga3*j`>9Q_$YjPp2W+czBAy`H^|V zzkr=)CAf3i)!mJv?mh5DiKuJ;ih(wmW6$FGgGi|K_|Tn#dP=a%gUVw4^8= z2n3R&7ai*0PlW{8XYDqW5eV+AQN)HDjHXn$)Gq&7*t1WP^T^k_${&GU!XH^HN8xj ziqh!F;Hb^NBgR3oLH2AFy!rf{PHbk}#w}<$AUW}F`XOO=vUTbqa;q}bvznTh^tO7e zO^atsxRtxTjp8bf4EEFVXS<501e{0&M;D>J!AjBs=E@&6$t)Zjvtz>Y%No;wBtx)_ z3o+xvt20W*87DpvnjA3?F z-I2D-FAkL(-8Dqj{X8P3lI*aSCgNW{Cw8-kV&YdUu=8?J(*?NW4->$2ADAa?ku{Y( z4nO7O;W?=QXI3UrwxQN2RHmas}M z{{6W*0jF$S$&XhpC5oJ$*SqHvBMP{?+xX0Eb!%|yO!#7GPF?6@yB6-+9u`~T6*^CM zHL?S^$6DY>Lk|-qS2}Hj@z__sp4>oXs=G@P<`!H_zwpdp)XaOIve%~F>k57BGDkZe zzDrd=q8GrjJ~A-Dh8)rxU*1&OQ@u4(77qFAiarr_{HBA(u^fUC_7LVLC^Lk>K!d*MY@ zXT8Bzrj+1Be;`Z>Z&e3VYh2j@CkAv$Fr0|B|8dfsLiOCQ=94NZESxnaEAkb&Fh%~! zAsi8<3mxPU`bn-?D&Q8if%pP)5eG9{@k`yXPoi6+ry{y_`?d4QMpZjPR(0HNfH3^l`I}&y5;v;~F8U z)O=dS0DM!YlC|Cu0NubSVbnAdz8N7sz>qb2%ed~7<4nn!X`}0kNjnQO3^q)C0dt3u z|Ew}pDVC3_3;1VGA#w&>mFob<5vkYcwq{<-%K04Oh*v0NT+Z|G;bzN0*xI0sW8$EA zJM2g_=!g`F)JsB->S!e^?|m^O{{i`e&ud2dSkc+zMUMf)B){Lo-!Dk?&D||L(_}~_ zgBdb1?%Vef%QJzkW;bz;n%T`Heo8#44+Hqe&+#M+;u!^n{GFmP%O8zNWjumX zOjsF@!=8~siQ|RyY&^aTl95clKIztmJJy*c{Kgva;sCmT;lB`mShuP=F9Y5@Bgh@H zInEb-ZvBwRV}5P_Ugqs_;z0J;^J}{#OvBDuVrBwd1u?2PuSszVsT4=zG~Md%Rb(oU zw=~DT%d4O1$6t1moEXGLj%xFAiR`n_8l@1F-A}MPK7%N91~W`42Pu^+H4_ALwcB9^ zV29-ldSx&;$f(42S@~Z-SX%n#I=HQRMkZ9%@Qb$%XV$b?N23$l${*$cev=k0Qh4g? zp|jmFR>J3K!Tx+>31}1B5qB*LT|oi$-mBPf0w8+AT5Td}fJyYAv;T_XHy0dXl*OWj zhZeC{OG+1n|1+e9?U#kl5Dxy*csOjnRmF1ije2gBs)3sM>nIQ>H@`y-GY_qwY-27J z5?w4-#80V&9%bZ68krfEDO%bWD|Jk&y7Z(>BACH7E3?lCVYGPe;v+~!s7r!3^0s{80V1h>1 z)mquVW>Hi1Z$hCrZ6j-rixngU`C`eUr+$22(h-RNvGi2|QFU$Gk}@iZt-Q6jngmi--EiL~0dB1<3bKu_2z1Dpt0w|o6-8=s6zg6F&?LBl% znKx)5&xihVq$}B`n^nr=7xbM5Qir7C~A~Sig zDeDE-6X)LQzKp z(1v?$7JgQkd}>JiQXte3vz*MQ^=Z1!dEM(vipRRM(c&~k6&06a;wov^lNwL^^F@uk z?@}IZdk!QP$?NwNDgn!LW|fqXwN!u#$q!k0_&4Q{aDmAYP{T#kvubO}{niN_lbgBX zfLe0$Rwim=U^a}>tMvQvTSiWFu1lZgBCf7Pvy2Wp^g9ci~P1za_( zF8}=b!8Kj5%nayDU+x226Ht%YZ3;Q7*U^NY`ZE(gFqWf*!@>cLlG`9tBV9IOQwv3d zs&tPg-k2yS(Qp}~`LjWi4m&m4LC&|Emk@&}KnmH+1Oj3;pB{^sX{e2T+wWWj?$r8Q z@sU2umuxP&**9*7CAhxVR5TV0ahavzm&i-#i-pKN*)V`+a#EgvBUE*WlD&`*lOykY zX5$O{0o%d`qru?t6-_kGX|`r|?Z##=t(H?35ennN@y3@pr`{L{T9D6ex-|2Ru5|k| zbv@3=>UXFMBj}sP*v)gky3}Ae%N_f-LWtZ5qtyw>H__m$a+uO^a6=q}&sn2fWOYoU zJS-DY8goAC^M`xdAJebnGF=o~h=;;5BQfDyDLQvj`lf#d+6YC$i|Ii`LT(#`?qAgg zFc8+QwY0Q8Yfa|7pNZD1(2(9J9)Yh$WZ6w+(>6-*!1DN>N4Dy(@9whP5p5L;x1Zkm^CDu*F11sL@K*X$IVopPMJ9ek;gLY@Ybl^;GC>UTPcp%aX8?b?+- z1u}+zR2;y7ok9`$vW)q$`tAvdc>QuNk@$3twjWXKAi@}0G-y&C2{f-Dhz+*kl3My) z{cDtDqUsNMlk>y8)d@7XoG)(ezmq_fPBw-=Kn*($C@w$m;}TvTP)6p1B9$_Dj?~qP z0w0sFpB9iaglh6u_Y-+rrAtm?X^BFW7&7Ds8>%=t0!hh68hWMU0x8^#ot)TDPftJN z1Of2JY=tI06g!@ni_@;@0}^>cN{XNUq2+~t2_bFB%d^(89StuO;oEENPe4QUJ5xa* zNL7ky2wysejwDD954;LCrJ)WS*GSPsCN#WAeoMOg$3jgx@l`` zPvr}1gWYY3ce}^zDh6a}RJ%4S0!Q6%GkdnUO1M{5IqT?fIo=Fyg~$$S*bR4wJ(_AB zS2PL}x~i&=WN8mv^qv*o+i7y=BcA`zBy&=gF03D~!x=y52~vc^h@7Ln<_?1)(zke< z>ePl|DUyLWor2RNiE)4TpDJDNGA)DD;1CsxAKJR-XLDb^7_0vk`C&b0M)Xv`-;w1s zDF=o-?|ImsVczwGew0U_^v0%~H9hlX40@LudB_0(*8`&5WRrIymZu(I6gCxN$o4Fr zk-n*S)IOo!eQrDbbjC`uS2M1dYvv<;?{Kp|E&ogM z-!r`xk)JqVUA|Vg90-2*RVQ^R6hY8stq+LkWO`aaPyORRLo}t40gr5o^z(c2-)?C} z;oRs*CO`1q+CHh)Mj!5E`YC|7F*;z;IzYX69c0lsrbzTS;9r8%l@fyicE5Yn@e}U@ zA%hoyZ2nCvo`4)FM~b5Iz&CRuqfhSND>U7_yyV>6xZ!Zqko~K;KB#cDhLX07soGpL zMmX#P-MRO$dV*!Ds~3dGc%tflq13711KCyLLhp_Jid{ueR&MIInZKq4UY1VAPpt?k(E#KP@ z=LMFr7#&}i-u-#}3T(qh9*iKCFNVh4IDJh&7PBi-RI^*=DLfaAEM`zW^Tiz^2<*5% zw|}7Tn>VtR{Mj{%eHPAcZ{oFxDyki#Vqldp7j9oS-vB|I6y87;bpAUuees-*#p>nt z1Binj7C$4c5>l~w0p_703JFAoAyG+$D-aAu{h~Ej{!ci<;MnnHnAogXDh#nf(k>lq z>m`#}QzqGFLU|@%oK;bmQN56T$%1{2s_{bO_!*GITB8ED3cIP)-U(%CG#6T35Gdk>ig5bL+z&a zm_?DW3&(@TVzq@An&eIf$iD@zSyM>o!b-0dEYoOERI>hk+Z2JO zED68^=Y&!+nQ+ey$=HhIT3fPURH^dw4AuncqSar9SAHS(q4XHKX?|nbj`HSrL&SrP zFS6m&&M7Iiih(oRIKmR`dX}!?)|N1_oL!pQl|k6&(BSPZA>k*bEIg}O%AstC#+Sn> z?hIeqEl!h2V`F3MI+k^5A*3l{wsl`L)aC{-*Se}uG@mB~b+y$s>txr@r3lgwGl)Iu z@H|YWkiEm3VQ$2+s|F=oJkA5M7|j-7LOi85V!5vxwH&#;x3XBwFOiLz=pBAF=#wa) z8mULJ-T%riBDxh(p^0c-A&|HTBL?G&CKL48VZj^OqR{tth$aChCscLywyVhF($7T7 zCb#j@>x&U!3RJGu_YC<~AQ}V2%t9;K8yiTg`cb#9lhU{=^=dP}IWIDNpt!0jzNpgh zmj+*{tByDvb`rfha6fJ`iwMN$KMFbJ{BVK!qj)$TcklQps*KU1;XW6ku@()^{VoE3 z=1b04L}Vf4Ge%I#yL$2$ZIcb^kfkcRn=C%JdeQjbCuN^pew)p3P|EBV^`-?e zq74^>jj1ih)LZ@aBH++cWE>$CG`7RC->6{hI5vqV`~KI&R$YXN;F(DEJ-r4lc!V^O zdc|bXC>Fab{)kyQXD9uyprZ8F^}4 z&27J&G%}G>-C$;SvVPY%5OBk?wR}10wP$(a9&jI{H5U}jmztl$gvULVevLwJ!hXXI zi%cNqn(%q_U*ISQ-tl~O8DwhV&qPHKBItsGw7 zMDsw4fyN==dNELC&}f|>+O<984@JeM6oK4jxM(?rP{5@^yovUA{35uyB(>db<$YF? z2%vuQ(vB-LHhmQ#-n<>_CZsDw%e@Ld%9yHQe+EuL(k(KrpB{Q*d0FA57r(;4iKy39{Y-(32rdtYKRRsRj7iYI7%`l${l(KlY_j2k}h)g6i^ z#4inIX(an?jHYYY(nnFBUhI;)Q?rNxbOsvn4KhcBEM8td;fql7|^ib#{cs$ zX`!=~;ziRgP{WP12&chJzntW-Q5l@U$G*U@fX!Ufqvg-p`HusB$4~CNBfK`&SXs z%$Q=dSPZrIDylo=UvPGI_89w{pr_jmE9w}C8}+g7&?TzHHvevXFY-4&lY+!2O^*Pd zPkKZa(tY#wA{gO3(%fT-!Ck_*@5ur}h(9?+LTKkTQzMrUbh1BeF05bL`0)Ab&r-sO zkQoWh*NC+-iPG-Q|u?dzeTgVNvrDLvG-^{0Vz5c0N$4une__C|&R*Buu;gy4TRfIoG0 zmh*SJLDtYvBrw9#Z7-gQ$fb2~FC$atk_mzP^V@DVkt_%slyM|ZB2CZ0az#xvzx_CF z*6q>6&ht`IcD6Lez@eg#FFTzJ(~y9I!jI>~{M1D*s(=gJ@j@Z&feh^GnHg>mU<)a; zYQyF|58W++P%9F*sVr;|8r!c1#xQr>^$6hGIVDmSui`!Z1wbYJM%$#oV;-U~=M-6R zGL<;%DA{E1zn1`Ymmotvk|J0$*PZG%p&SgS&fa!@j+3EZ^Ma;o~Xlm+yG6F`lKC@Wg??e69Hp5h)URg#$1&gIG) zBztX+j-_*TDI3yt1wT;?r#fblaV47A^UQiH&|F3@)1||1B2#~tNS=A#GMt~Z2uDsd zvkh}s=HT)9a&K%KJzhD@6Ocxzds;u-k8xIahbe13pP0lPF^n!ruEnqRAvG0jg1SdA)ipN5x z3PkmS?QZFM#_TV*B=AvpE`%Wc8q1UVLTMxHw5Z(8j8MGRoMuGg!h%3tnw;02bs*V-4jZ$WGmM$NEj4=1PK)Qk@@v)CXBtS{+jEynf<%ebNcVDW zw?5i;L=00z@eu0-y#z>?2$IW&m+XDS)DfY_tKW((9}oGs#H&Gxl-yHg+R}{BHvd0gy=qGsLhD~Q z$;{e}{mixHHF3N8D4$?mO!3!L(D^51HHWQe$Q-D>B}E7Yl!H?@mfKa7W-<`a$Vvj* zhyaNq;`G1#%a1v0=D&}AzRSW3b$Ofe*g%J{{U*sQG93Wux0ma!=I3Y2)v2K>4Ywt; zzhv|{qVzBgr9*K*aN!Pd?uKa-5ALrV++C&SDMmJei}DvjxagI_;mm7XZ^Ckt4j%h9 zhNyWa;x8s5bTuWcj?~77K@C z4at_(EJ7j(!zbHJ9>4J};N7LlMP#CZ2~ef3ZD!n`KR44mqAxv#amfyDroFpI+*A3u zeWzy@lrc`=iVLTX(!-@i=({cLyXWkiCFK=I#|Y{Qa#C9bbePa z6JoF!gM?GSnA@B{GLfnj^9_5)x_-@lai}9Yge+QejLLrLhvO&c@G@Z8Q*2c?R3}AN zb!@owgr!NIcU>n37Q*$NR$Eld77se3zqzACWnruEQD+ql{AG>*a&P1mNq4uW+s#A+ z+DXc(8D|(E9^7TUI*wJJ9}pBmXgN#eZp0aOHc~=Up_Da0!2|WqB10H@NiY!cz~FG; zNLNxuJON+GkC7mB;LiPs4I9Rkv)v-=)wT~VCoPN&jTeI=Xdlg?_ZcwVN&>heP?q*m zCEjOlHFJNxFM}wv8H_rtiJ6IRNF$VNW&f*ezkhR6(u&lsV%g>5-{GUaeL`_1{`zO3 z%#+WAJDR4y5=+hT1S0a@sSQji1VVW~rIjQ}|4cm8AWb&4;0}G5aNp_^DKHlC$f6;k zWiVvh#ebfWAGlMIW3({3m?jx|ViJF$m)$OXRpct_dzL=INw~hYWqTu3OqlacRU+Vl zs(6x#4?-%*=U@{Al-}D|Vh~C0c?JSJC9iCH13z;Yg(4DIJkn#j*u2DAxpJVAdCeD9 zHQ?BLiD}RZhU`wW01n-IBa&4D<4-LBibVTrU+ zaK+;4eVK6cs}l&Y0w%BIIoG`}V!JK4j&7&H;)!D|f8-lRT5DHKwr4Ci%IJ@mvA`U7 zOC1!6fP^|0DjwcUYmZlw0EXg4eBq=0j$LPnx>m?aGUZS6E11QL&=j59Q#oGag5_@E zO1_t0RYXgBo)@%AAnYeD)%#o&@?yZqvV|Ns5YfLZU_1CplvTUBK}Mn)D|+G^YKcv> z{Dyv7%pz(>zK2JwoXRlpujb}P6*?nJ^UKo&O_AnPemWuGVDt*v@imlh&^<3r;oT+V zr4>dO+I%gvfF$&tVELKDT#KYlOn(!A!nbXr>+|sJ#xnBecEdQoh-Iq0@81J^88NtX z>x>8z5qVp!lP?w*ztflAjtXr&&qq@yJb%>F8tRc8US5351p4WAiELAk$~v?1Ak%$7 zrz#|Dguesl7YbUXzJ4QWNdXO|d4;cs?ol^Sk*1q_{ELB`lGTx(y%jdel&5dia6ipD z2_e8}UEax!;?gp6s~T{O8{#fmsYSWfdChr4cz8{k-Jy*VaVLQksgKydV|poua{3jK z!nx|axVZSDA-rBKpMcr4t4$@cIPj>5N@!e08L4s6{VK}cNd+d&#({4PW4~zI26AU^t7epB%I4wyMz!o#X}vImMKu}& z{r6$VmJd;)2#U(@!cx8=CmV$Yg&?ekyi)VV?QK0K7DKVuepR_IqUGie^W%#TIIzdx zx+MsT>D~VfK{cRIhM)PaKy_aNLY^=kMGdBGs~Ir>$yHlRvGJ3VN@78v}KOv2Rn}Q z+FM5^0>+@ywyFUiUHHOJdWM}$7s@KgXuOBP)5L`5%&bEEU%=!eDf>f2Y4P_2tB=3s zpt^2L+SAufSKZdMleu%r(XyW>naX(1@I)9S*4ZY)+L>%77uB2|dM7{bL_@z9KFg?u)J ztV=|wB?pU^3{CqTnkSQU*Unrs9CLH5oQSxV{7UaF;%FX{C@UhPWsqJpQvXgah;WH` zToF=R+i5=Q4d`ez%T!2nv^3)2Ys_6Q$4}q9Mz|uousi=pS0UlLS; zMQtDOkkG}BP;?3fws4UY)X3AZ8>r5*9PJQbkrH{*Q>8#4hJ8PN(L`g~rgKj4~JJI%jFy+ z*yo&C|L8sggdhFXb(f!3+@&xK*{Bt5YV#KaE!=dRc;Nj+6RwJhLs;5dr_`N2hID9% zeSTnx_brMs66|0n#fW%~JD@s=!iolzXaJZC5S4v}`c}B)oTR&}3PDF?1Wv56{Sfo@ zw|!#n%Ix1>JKYm3KhV1sXU?FzDZyeIO(MOJW@6Rhar8-He}9BIZz4Dy+0>q7bIJ&21hr3$Qi7p9W|Q z38kK{LQ(3#!y(Nxk(;t$c+}PAp)UXXeDthMPsA)pg7*}m+?KjD_~Q*EY-l@F9UPuE zUyQf=HhXyMw~2IJ!u;4idONX7^~|ehK)7jVZ9e7fCkDrLjQ_09=fs^8*}BOYPbZJQ zJ#hJBx%&gr2B#@9bBKUq2x8VKZSbc2#D3DGlk%&1&B}+^{kk z2QH%nd?y2Ol;92GE$bF_Fl?SdGJ5!jvy5;n7}7_D6m|9QOmvvrL51sHh$<%LH=YW? z-cAJg+#Dcz=Fwre-mgzGRmad`f8DzVK_WN5r(y|Nq|*>b+oUPn$)uk%v?ct3o#son zpcmebx*g2MCrWg8@JAd-WJboeSvyJYM_q=~9Hj>;PZ`@qwFA>Gpg@4Argfw_?N-s9 z0LbQw=-f%sv6EchydZ@18^X{%Rq(8Y3;U<>k=seQB<{>^q?`u+A~_Ql;^rdL%*!A8 z$lZNvrbfIT{w-|)iA3(xC&Dmbj{IzC<1P)!jc=Vsr zmKlyi$DrgU2IRgkdE0+H~yn>ted6fG(K)GSaa2)t+ zq?Q24$AN8) z9!uguF{{~pr3=qP_um>G{=?HC&W5b<){m((g`I%hBRK_it;cX<^^^1$IW|9Zlp^O+sZH8D}HA(>iQl-E9Vt zmr&XLc`BfUD#o;Xl9bgpXXr*k**NOo=P|gLx`kKf;fEBW zL!ouzDS!`1jJ92 zLN@yjCGh^GQ>(ql&8ZXuvcoyC-GhQHpKI5=$>)vm#GJ4{1sA+;z%hxLZ_oeOJQ()- z8ge|A#Z?QT4eD%#wVz17)FHzMU4AXnJ}Nwz4ES!Q%LhxN0aj23pLvwbaERCWvRnt9 z`&1j=x?Jb$#p>)(lRVpx0NT0#MvsFO!}&X9dvYY~L`Jx&uf98BzvHUsfl-%}ias_r zY#ET2g5k|VesjcDnfhbJyW^GBZU>+7!4!e8v$GT(LCMcTPc%?im>h=S=Vd?I3?gJ= zGkXfLRO|hCdu89DkPLILPSin{Dmr6xC)^7szLS6L~k(3jN&6h3Ok_wz7j?4e2K z+5}p))W~uvuu|4D8VUwM-&W7Rk6Od7dsF`ga%8>xM`h}JxS+GkI|mP{%y%{{D|QoY z8&VyOE(Y-&oK$8J4K@-PlP-muvD9pV482QFCxZ1as~y3XK~22Rf9!19tK1{VESbdJ z4(Exc3VCrEF-FIHz<73(5dbON)#SqeSl;231=In);#hopb@_YAX0Kr)W8x%}^j44@ zzC{hE!6;2K8$opF5Q=iwX@-?dmnNf(GMe|Hfdwl%ao|w;aFZD|(A0*?F<0=Vd zLE&{|MOiO$oK{n^Ia<_{=afGeV;C`@6$XFp+7pR@ME6xy1r74ijNVP*&U6k1V|F8s)&o zQRh%D5U8>aJ3FY>em*wb1I8;8wVXj4-NXon=AvnAvs;WVZn&2;jUwG=H29X)8Otu| zNKKov7}0rML`)H@**|tzekSH7#7uwKpr3B$Id0g%nBM(M`0Th>pm3M-fB(e@s6aT% z;A}eYUSBZwJ!GfOmF_f2E7$fd_})2Jm7Pd*h+eITOrVoVaG_DlwRl#Rwsw~A>hN0% zA`1GLrQ4rhKV&<7e*dTYGr$OE2{uUSr%}f*35N*1E>-+^!5c>hswFPW?M$uQm;UT< zstM>XuTXz2bTepuIf3^&HXUoak-9;Gk{P^jV&pVL19-*5-+A-z#iQN`GPx2y& zcoK=%6Z|yy-n>p{R-Ad@kWMM%S)cjV)zZA^jU}ZE9KxNa^*){D@z$PUG{f)QnPRh5 zZiMIi7n64_Hn2!!<$EEDQPIcj7#B*i<3+Ap0iGLFHEP2ePtge(KoC5@?UIwDv&U51 zG;kKAd-29wfwdY5g64vt=!p}$xK2$bXohSk)W7f@om0>W&+>ON0gg1)&=voj9vo$k zYkmf!T>Yn2zpbNzc@m3`ooij#CmIO4I*!A!gPW?yu*T;0{mLJ~bl(R}#SiJef2GIN zQ&G%#xv?r=g)SHeGx+bE;6~4lC}ibmKe7UP%H>S7~3g&R=`P zez3XKd7F-T~0pbRW5XfJwHXqRKZ|6V~m zF5MIO05PC(D6X+&cY3;-;?CcC?+8f3Z%oA4E3W;?i z8U|zLe^p5|ivrFvX71PBEY(PRgga+6qq?`WPP;TQ+bhrNHX-`s@2&T?b z>!`c6_YxxBW|hAm_3-gbA@|rU5~~4lFH4H?BzS-Qg%#gX%YY zxMX$-f9-#qx=I_Lp3#C1uq}XGCmlGz1XQ^n=i|do?4@a-v+ZQ1ACPI@0ctV?38KZi z-eANxOI&obizpz@9z~z4Ocu=rYeO9{h8aZUVQYwmIMKv~vyxZc%n>QT7kCLoinukS z8&W9c=a-UCGGkVHXs&46cjdIM=q$%NL~Jsd=pT!DwORf<<~Xy1>@z$Od&Ee>i;|{LO>;aECCMLo>~$$w&ZEN}+0zxvGc;XH zi_*lUA6F6G`5^L=ymlIMy#T1l{_@6h2n`T}OA+m`X?;dRCuEHbN5PmFSN=u)U9bWD zBtoZB7YooWjRFEl_vd@}d=YP@#dz~+G;}C(qqAlrNKW547hu#$0PKzv8rzLjt+7Y? z>bd{vJU*|4Mup2cV?ty-;y&OZd`=OJY<^d^a%H=P*`4+fEQe|uR;$q$gYzpi`p*k4 z+fDe7qypacV?DFPb?NWtekt#}%Qw|DID-n6vS(&3%E&ERgNj;@ypiL@XIb<_F{oZq0!StsnV4axgJ z)=@=pC;ER+t!TE=9Z>l#vFC(67sIG96Tn1fDSd`Zp0E?*hJo_x|6CI@*IIzzRLzS6y>E&?|QThJq)aS2poe3sX3b-0;zm$17Y@&&a z2+B7qbhL0u{==-y-e+EUH1(y=;!ZTOU_?qJmE_hO3r|{H4(daAlVt)k{~bm^t`0Hu zkCl{#A=vzzOO<+N5dK5jc2_XZokHh6Yi}=&8O#pp$?dJQ~uc0E0HFcSC7Mc3x zX<2k2b)-0lIDLX9WckyL1-9!Eh*aRu`L7xT55G4Rq@R4F(xrksTQ4Q&d|E?vEnx)& zBY7?IbBmwEwdPHWUYGxOKN$EF!GnHqaKJnJ0od!0em1lHY;{d(u$kP6z>$Dm0R)h; zvYgsp7*RayyGqOPUmO<+@{B0!9CAo0+^GB~66>hDhpJbxYfnOIt=9z*QIasN^%35XaC(;*enISuPl3Gc*hdMFkt^~!P#52bj zLZnu$_RET16O$qB)JZ<4niP;R=b&t-0TE6ALpk8)NcjEx+LF7!UZz9 z#SbAb?7k2T8JN>9c2Wd!6V;UIUWC3hLG~H{4_OLz(x$3O7Xs_;Xu{Ph9v5Mxl6EyB zgT|q&0-uXpy|77jBx#oZqrD$UAF42Ru=41%u9ng3St0}93}1fvn^keFMcf_uXxa9U zIq`-%Vq+snbF~>;!t)vN;WQpfmnzOPM%RL(IrD|ic*o?A(SJu8aHwGh1Oz~-s;aAP zGR(lSfFXr=Gw>R#=jG8c@OH(HS16Y#R`B{@IFd90H`^@>qT%j~t~i86PVFS_h8i=o z5#H{oShlZwwk_m7jQ*s)*@tCB-3A>f*1qRhA8@~4qjZw=!_JysAL;<_k=$`}IEm4_ z-Q51EIDs)5f{~OR9%UrM&951C6WoCOa8sV505aYWWC$IgoaDgh7f^k6Hvx{ za_YsLddnH_;E{b7$W;j?>orr&uMhZ>Euytcr}S~cxqa# zbGip}%t`9f&HsO?e*8j0LStT)DuX}EUCmB~J%~0mk4Y!m^&oZrC(_RspExPSN>yWA z>b$VhCVIMEDLU>b6XP?!7OOtF++QAkZL>c>cjY_boox$i8!)P z6x|Iwel#(vQseQOYro;+@aD*b>eCiGS{#os_t=RXSf4lo2#q^gT58>s=*ov|mu3V& zWwhdnVz>gVrnEc_R8-5PO>Ck(%PC<~QhS(SDaSDC`E{?T?{Ji`**b-A2_$%pLiqg+ zCLrqJ8mVncMYi~;e>jw3D-j&{KF7f0OhcYMr13iq?)4hpwihSc3#(!p;X<~!0k42kzQ7Vff zY|12(Ag5c>{4qU9*f^o%BXu=^Ed%JXb*vI%KB9NLut}(kDDqa(2!}?myE60@U{B!6 zcvKBg8(?52J@Neo?wT@bL0EF?+P`D0{@E|i5Aw{a8NHKyp&QNg`-jwCWpIM+hOIJC zxU0u)ePgXs*iDl0^$2})v1S$5c1D`CMtNk1I^*l+pdR&i=+74Cgfi8}v%l?sQ;61hDV^V)LWA^}UJ=e-rzR8mg}mvu zP~?i{XPWWzmg5-g+{TYhzTSUPW*K7fwZ!HfcEy12qMZ`>ojZNvqR#LS57XNWx}$vK{1e&vO=Qea=n9okaszOaRDpyZ?_@uOl}U2iy&}<` zAnp6Rjpa-t>2;(5`3yv`vqH!yOSEZ*tBh6!#9_#-*e9_g>cL^#?@s5T39vPGK z1*6pY&CkG*>XK|6n-R)OXD~=faGEY$sA|LiRhe2CFJSm)P~Kg0W^!5&QRpCeh_6SA zPR`gR<9EiwXywjtw_rk(=GLF^4jZ>6V9~{_UiFSD)z74a*4?3Jod+r_E8pMWkFDGf z4ao^xjQZYGBLAXbQp<;lD7y^*#}pJhr5~=*-5G)_I7S#DR^NoVOep%%l~wTMEant2 zoR*zqk~*v1PrAgLR1$ujML^z~8H=2R^NmvC*>EU^clI_hwi%(AUlJKj8MsCnIR0@_ z@Shy2ucV4q_ZMEUzbZ2W3P=MhbgK06wzsz%g2v$t4b=c2?B?T>>CB(eFos!;BW!WD z*6u;UCvE+DqX|$njRc%`Gd*FE@hg^<(Ay_Ff3HZzJl)}CDz7?JdJP5VBSY)PQV=uG z&Fc=iA!UpcHZ0*Jr)cX*(x$TJq;@&dXzW(iJl{PbKGCNNGVjAr^YU{ew@v9UGc%Tn zmFuTZ<(xjHSJQy-CgA`3F?98Nebmu16SS^Q+afLG97i-s2>dwz-`vnBmXbE+^~{)f z*4s>f+w=TT?=p(F=8g#yppv*h!l%I9qm$x(Wk~;Q84BeiKG~=qWK#5{uUUBKblA}h zr0kr>w4xJ^{QMnbL*8nmf~%-t|Ka{30>_%$Cyi0n+8Bio`9CKlpvq5Aiy~0FpKI(h z|KscJoncU-$hfn!b9M>V@Z<6Gf)^9&@NMWaPn8)E94(n3OBc^8Do=*4YKl42wS2gRpM9{1iJkRXCrcI8J=n;5(*k1?bqFyZ9^HjF?>~4MYNk` zWF8J0H8h{%Rj%5wdR=7By767tOihE-?tQE8#)r3HRN$@QlvobOTkMV`H6-E0rZK5c zoCl&^(AHaxVv|svlrEFQC>ju?jZ1M! z?0K+e6iI`pu1lJxm+Io1=eh7GxCN-RgY0&QJKqc8d#f!mWB*-=^}L@h6R~}qR#5Wi z)mqTT`w}9Xcp7h{hXHjYfSS8eDgY6Gtm9SKlMnih?w#3wC^4j3@cU60=AuyEF2Dee0(_t1xrAo zuHWiXFy%1y`#=#47MtzUgdsN~-Y=Y~_aodKAYKTS_jAk<5`W@UzN15;VuYCm zKggGX_eG=9lxy)D=t21_ckzm{mA59Em1;<9O?O#~2s0PNY_}~C5D>4P1__G~j001T zWurLo#3C=cNQP4fe@%>H6Qt)h#eHoEWOSH>!|M*0385nB#4N0vwNGlsouhCm9?zS> zc*4n2!G7Rs7qihFpo}5u9r3f-aq^CD^V|y>d{W)YdlUHtVxomIX4AC*xxsLaoAXWT z{}gjWmV0BVT1+%0LVyt`Q5h2Er=K0;sPWHO2>l#wzIBayj>X|9P#oO`Cwj|gxstF# zb?J-$pI;V$J+NW^F5&`UDvRfHS=FSJVqsgxG&=)x=>CO1k}bK(a)3{B3W?jXAcF?| zU>@F=UXiA2VD;;L!O1K#DqVuQED>8Q~IrL=R(uR;Ka;rNSir*@OQrbBpBLjydBQ6-yZe0UIs42A)-NbzI8 zGu*z)xLwPTkwoqSU$;duZl~mW@42rB9~qZ9lS618 zH+-2FV&(}~*4Ir~FT`-Zlm6!(cLt*A6IFo$m*!x$+T=iQg>hVV1E^#t&=(UWAvxxM z(OjN_2V}f;!bzzg>d>G{gY2t9fpu?>0&v3qZb0wP+?c=SqopD5zC*N<6XH?c5%9Y( zH>lw1l;cGeqk?LT;f=G9W&cD0RDc1A2DY^8F%S^0btuYGHscR{FXS+kg{sd1b8Y(v z`prv3J~aqCL*2$JiD$2E2|s$M?H9eH1LU(Djg+bU7b9i!A(uPyE51NP$`!(o?f_p0 zHK*mek}B456$Rl`fGe{BhF zzJ&R~g{ZG~v3F7k*-Cl91kd@I$%W>Ggv|+SIldB4PhNXwXj*iRJT1s}hwu6u18Cf{ zl13>f3m+nl!ucN~B8D?75T#@T%#1zJ2qLFN(k=Ge={&(d-KNeYX~@KvP=?^c)C6XO zN6+uiojoePKYJXh`5k{y`FIiM@~Za`6W4AqT7mEXW|{p9)st&bNmDu*EHBECIV?n) zt^Q`t|9(*cRk(KD9Zh-BDJH^|;;3%8>aAJE6Sf}vdZV|XKb)Wf1OmZTE=<(VUjb$G zYdq0maVkJX2oS0sl@^&#^Ou|4AKC3 z7f$3Xp+i1wj*TS5Jo+K8+Wo47_*sU2$gra6VCu~D&FrSe3)DgQ^gO=y`^ zi#uaX7!>)hENSg*o4XZh+mKJ$+fv>_Z570Yuk+Qh)?tu-bG1Duq@~Rg+1C$wI2o_u zk$yP*O%BOoOV_(;qBicrGH*=*gwWsTcj{L1Z-i9~Qs=m)8~%Hyhu!zz1tQBMZSqT2 z=X}kjv-rdGo!N;VIc4UI&L^GIc^RM@naIE%5oL#^uCD%t{Yuf9LuQAM?rl5+jY>WS-jOO7Nz-p16@}*c1XsK(r~Ue@OZhh@IZb*p@3s&; z`El0uB0FFt$mk`2LfADrgse3kVfGK%o58;APYh7~*1u@^Ln4iyg8qfxU`58GGnKz# z_>q8t2=<8ZC;qLwpb|dU)t?$1Mo^+jwgY*LtOS%dxa9H6#alYiceA5|LDE3A&sya? zXjZJyaFVegz#dB)(tncWpnqKBjBAKkNFXMGnu4M9^88Th9k!RSNM*$gznFqWFArG@ z`A6-Xs&Y9A|B6I0a^~?)DpAa3D68Tb{phRZM;b4FFH661UgoQPHyimT+E9*mFrY3;u%DkbEJe?j$g4YEozaLwCxc!*o%c8;y;Dk1;H0X47Y2O1_fEcg5czBj-%9#dL1D z=@8kL%OKBTtp&->f$s3o%;`(*#*N*THvHT%v4<`|sGxZtEZxy}on@ zKS#$O235~M72YBg*qS)!m;I3(P5u{;*85H))+88g@6)z|0g@TMH!m`9kMTak^bcY_#`$ z$g4}PA1>>wzECsALWRe9oq|>p2Dh1D@UQe>sKdDaX)D2pngdzrLMLrVNJfy!!fxWbToICJuMj#J*tWAX$9sxXjpw<&3p3{6XUM4Vj<$)HG9a-VHg*7m@c zM78H%E(USn$lUjN0+m8xdFvf*T#2ts$nUj*WDmvi_)6u$(1X4u#i6KAc)kkQIY|Uw z#VkoRvZ$y^vsiI|Fvd)i`Xw->Hh_^D{VtSI7L_XPj}`#;+XdU9^AjoT$`;6lg{cRE z-17KG-9>NgAO_}U$rjCrryd-3jG(C0Lx0*pk4|ADd~= z6W`un>NCfmqRH%H;JKv)3Bmh7%KmEHW$?#1X(6bRI%dR~6)<4R)QdQGg|K+-W+-wH zdjNB_2V!PXzEi0ks!2DC%|!ZmtgxTu-1*Z&7^86(nZS%v?7k!9Ilw7$ge~1np1|}& z8wPx~PYD)T84f{Xkofka6?XpK{l=Rjv?3i%XjpVx;YL-Oacvv-Fm!)o(!#|WIu;56 z5nP&$_!dCQ8Lhq)fi~LU)TL+XgFiF~+p7M9YO4B9g>>HQ*Ii3hA#q?L$drK)FDEa#4a}rvGcPfvh=qv| z_aGJ0_DRJ{$V~6B2I7{hWb|A{T*xqLKN#|6b4R-<-Tl9w&N3jX@B8{nw@4$>LrQm- z#0)JUlF}uObayx6fRu!w(j_6?-O`Onr*!@get*w1FXrV8bMEEbv-e)>v-*i-5Tv+{ zPoM~eBYm^z=AN7-+=t(zR_yUi1GZGZ(0@JrP7?&t134<+V<7~MN7SA|Ctc6u1TV&O zaQYUm9=ouReNNUF*5JO!*G2r%tL?A2Jp3F&L-$;%SchhKy@AoC@OjDq4)=;JhGzM_ zra#DZK2V}WC*-zuD0-b7ot+PtsBZUTg7FoF*3NQG(n?Py9j*fT8g*?eV!8B5p*PDe znD9&INbVpbCS!LVMTd~KLa8icfRENI5)!a4knT9F**he;%7RqxK7>3|JGeVbDUb9D zYRIwsQcn_topqlM+?WTma~HqOwLjF}kuKJ<%(69w-0L$(dcTRJExk|*SrX$@kmsdU zGh{KnS7`W{)02Fju%Y(&Lp>8zDx3!Wp^Y<)pqQ#eX`qx2p`%$VlePQ+G#K^GP?ElW_mHye5&(`KBP@Ojn zTY1glMw%q`?;9|nO8DPZb*>Z`zDW_heGB#o$zBwvmXi~={rO&Ltmq?>${ zgj$UH+a^3k*TVH~?;`ue;c|Q8FHBmCsKd{i(?uUc%K~rrL@%=2-nch+2m?d^dZQ2E zB|IvoLXxzzz@_d%8zbj5k@9Oew{v_jSV7ANedm4FN1;-xRYcu`Yb(^#zRGA3O5L15 z)e8Aaw&Nj*!lsLs=#WjdSeudz`3(t)KTvyh<5c)e)3Y?>eL5zM8Vzm|Ud-kPMkIQa zHu2skt+3XM+cV0Z_jjx}2Mr`{Z{ve^`WB!MT?VTlkKYgKp47cq&46y?N}AiVIeRWz ztiKa47#m_A$SkjmVK^yti4jH>3j4;<8*+H7tWW>A;JDm(NT_OS|iHu&nYQ9q$ULL)6u90ZAe}wqq^kV1CSud_60?@1kXCn#&$z=SrlZT~J`|TD|Yfaqrx+IgSlH$>>UbLxl2AM~CVW zb+mtUKAG~9Tu*B@$-I`j@vQtrg&J~*0&A;7V0?c|Hjm>c`rVruoKPBAHWz-@ur#Fd zdM)M#9+|Vy$kfo~M-F`JiVbWl#6sC**QtV<`{yC=86A22!YD1fJX$YOTEJ zKtlVGNp&^8C&#!8QjA;Mv|m2m_mKofa!YPY+5;-y?bsW(DO!Si&B43+Fi4(kntb77 z{p7qCuX2)|P2Mf++CPXy=WS!Dl&a{E42Y%IMH0W2nUSW^i(%Vtapve3^_MBXyapvK z>OH|k3b{Xjy=RU;71~nl(Mv~3qQ%Xv7~|FV2Yc>Fo%w6szjO$QI}Dc{Sieoa}^1*p2aqA38q7pf!)4DgPJN$=$nM2L|dFV|wqbl6j`at{mQ1-3Cdt@t#! z+$}^ec?=Orbm}Irq#bU*xzKnx5@<`D(-7qu^U+Ri`RV(dhtJJxfEQQecwC$R7uiM#`u8_jjpZ_93b)6!VDeXdVneudO$OYm5y0u%zO+R!hHNGH z!yf$;|M!rCe!mED-)r=mA^S01{(Tj^Dl3UB)|H;|Ui_@MlXhOX^wxI@ykePKnbw}A zZq$&g%g+P5S_E0if#iH{>}LNAzZ-h z@@Q@8$qSDm)2BZ&UIYgQa@=fiXb@LEG-E#>c6?5I`nUCba}<%2lk;k_0CDo*5^y0+ zTUGv4z3FK*q^Cl;tU#M7{CK-y}J7Eh$Ksm7s(kHW6ZFGw?*{_m3tP+j{lE;1@XP!Cs z>o?yt-r{{gO$bgVUR5n>WCqQE>8?Xo$N9i)i(Ene7&3LR`%jbPv$jT(>S2T$T+uwId}H|pxRO9Ule z_tGV#BuhK4pJ57G_-KuYXyADyWuIRN@UxqyBM_cuNF`XvLQE3l2|V##-o2JdLh)-` z-R866z1Z$o8#C)Yo3!&j+Y0-t3GnW_aS`Os?cw)Az2&-7%1Lv^k4_r4f{p58sOgL=rZ2ZD#8`{3(mry+4wpJISNBNNn-Hv(u$XD9GTix5bcy?l%7U#vR-1 zz9DYV@M{h<`cDNKV*nngXH zXC4DX`zsdr&BQCWZPfB#yj0SxXr@QlPo*RmQ-W`cdi)?jCDT}=yeX(pE zZ}qzKYD@FCnV*IZ_q(YJnX`N8-S!}8pn)m#v!ZB;OyxT*pVLJaOh+b(m>6IL#i!Gc zjl@1|ecJDZC|Bb7WhEv7mQ#w7?I>UAJsJM1!?^IA@1$ZUPLpVFw5hm+ z;$Ox@-$5+|W+=mer0eA1mri0B_3a$(tM~{U+iivQ^dkh``wMN~fOH8pK?xsIlR{T~K#{g~2u)=`{02FiHOqo~QpP=97 zDU6Bxy-sTOaiY z$HYjwBq0fem}=ha1+U(ZBIxtwmR&ffKHh3?RgBw?Ku66zX>>*-?*F+ro-=^QT5`=u zs1AO-B(+-*sX6FyQXxhvDJo`uOMB%?!f%)E4A0*gEmX5nTItOOs{ld&fTRGy#B0$g z;4Qq5ejNCGAXNx0V~$s93Ma;KZ@@G4kMl-os?%keBzc+LB@$1G!ArbQzZfol?LYh93iWiM{Dx|vyF-u79{c3lgP zKnGiyT#s#sP2>9oHoy-DGu|tp;s4GyOPaM*t@#AO1~PrQ#%4c!A2>q9N5RD4b^hue zZY~Pj9}Lt~S0zrA31~OfGAwCDt<#m?MmE=QW%tqL7Be$1O!Fw8BOZJVjKd$!dR3ry zG$V!0=4DP#_HXaRsKbtpjpb^`ippQtAU)GIS?!3Zn4SePr!U7}53gPV#E%jliUtQ(&rK<*aZKFN5X1u9>6;%4s%Ax`bY6Qi3Lo#8S?o5coC3l6Y8Bo$1x6j5nG3 zSsT>zXD{u=EfSZXe+qp@#?!;ijQmnLfA#wtp^ppl?ZTDSjC}e*xyoByh>4uavyhw~ zzsEh|JV-ug9V>}Hi7-n69NQVsS35T#D?vu;;r5cG<_`$T0>VFN0!@nG`+P;eE1FiV zaKLhYi5CK}BRjf;Z{L7s4-KsC1(RS>u&J$(Y`QWaVh1uRjx4v=t-eoJrZ-TOk>75_ z@Y8w5ajAuMjD;bY8L;B+=EQOwt0K`F0ZZ{-StR)Wzw3w0@=3EqPf(*+hY*lvuO^8x z?d|O+zQ4uHA4dj{0tP5N!9ET`U70E}gE2bZl^Rhf zbJbip)O>;#QI86F;slA}y2w^l?cR#`8l*;V!K=d+RW%=44q%nhHPtBiY)Hhece(ZI zoncE?PA!9jN{PXx&39LI<~nNbO@{GTADGl3FdqsgHv?G_nMB+K1!!B44~>5Nm}4g` z`D4x}rJJZS1uK!d%+HZW9~wqLMO6GbZ4?YgRcjrL6jL}P&G|iN9JVKz9!wlx4Vs&1 zFiM$na&k<|M1~#hI>%|zX8|*fWIip!>Gl~c5q|eX^S{q=p9N>jcnkq+8i&a{vrZu6 z`XW3508W3^4hq`VT8D?9fJx;`plCj2t$U&p2g$PzLD(zAM2O||5v6M-g~p8ExBj=Z z7%KSX<>e$|-U59en1aCYeOuSyZ9HTA$iz2wR4_ZZnuI@rMZn7H%)&6k*w!XJ)RC@3 zH{p!GgM)*SJbYEs1+wiuoCLJ%m-~*aYkjxrZ{`i|3^xnp@WP$3)M zxJXxdOm0*U%ZnURfL%{&yVRs#Pny7(SJ3?IV9umFc4b)nzT?ke+9@mhlN?&(GW%af ziu*#Cjqk9fb`FGd)qgFL%s3%vD(j*CoMY!Jwy~3tcf!Ca6%W`F-N^P^$i6u5w%a^> z{{+ic2t`3QwdJPbOVdVwSb+jw_=yIUQhZ3&hbaY1rdLS+?qSK5RR4YQ2N{(-oDKWM zi%Atv-j)<}3k!=n;81J}Y~%d4rpUuIM?~(o6(Wd8Y`js?sKskWj>7*eCe%!4tvNK; z!M)wlecG$FbppMq#CPOfmc7i*6uE+2k@Osf{%DS#m*k|m@lH<0w1CBnP+rQYKL#~& zaodPx7M+q8wHGjjpwbIPWp6=dFL9rL1J8fwUo{0m<8YWiv;JDjJne&UP+ol(uw}NO3_nUEdE@4Kdh3Ep+ zf75+gF;oJ#7B+G&;d`7Fok+DhKdE`B6!SHeA(-7#i~&i+D4j( zRHH4$D8l&3!|2yL@%RAJlY4q@&IB;yHkSgwKzEI{Kel;ABjItLpbdPdGJI0=@Sqzz>dmnuuKo}!#PrJ~}H@rYS z+*;&|IkLOsI(A&C-Ba{K`%T!Tmo2R@aP^E{kSi@xmmoet?fhFD8AuT z+wJCS6WrOG5g8pc^K*U1WIq~uArzl?II=&p>T?y6o67j# zBi2QN#91tV&y-Z%*lYal(v>%`)k=~T)1tAkeAFio(4!7~PPZ|H_djV920O7U+hQ2s z`p?fkU-ktmj*vppWL$pbVMB=^l7*uf$A^@&-b=^~Zs{`N248&%9Cqf(6*qbG&#(`% zKmxV-M;&f6+IIx^sMk}^U;W;rb~xR{KKU+&dn#rZ%lg;V_$ZvvQa?LM!NZR=-?n_Cnl3nQ0z+PP;+6!A+t6 zMG6v@y%@z{lMQNAq2@@QrK9s-*tg{*$cX;e6^3F*%fIE(hNdWqrO$uj4E9weJ?^hU zsJ<>Qj8@{PCH4?-&oRizCi&$WxO?AlL;`8046jRySuKJKyqCUuD`m6%YZp!i^He;u z6|AWPlR{%T+hNHSGI17)aP-7h5b=P4{B^N=mU)L!+{BV>pmZH&AddnX4N%-|v z4DC*w0vtm5Mm#mM(s725WZsm=#$KqUa|rwPo-~y7_75$KfL_TPYOevZDkN05#*ei`B{IJOK!JmFUDU0JW+w6ss{{f>B*h zuo}nZ-5Z(32K$^2FZ=#=U{WFgZeWQIOJ=|~N+=UW&^9dY8+E+&A^BG9_BSGr?IFXW z$4`<^d_@t{{x6KZh@5`j4KZ+nG@XhiU3gR9@`-mSol<@UoI5BO9`?!McD6l0Hu$_FgIhf7nEGjt|xZyWXL`_1{)ELSC(k`kM zg#Qu!5a*Xfu2OTsSfFgYXU(B=x})+-cPXKhNrwOKs}P1qgyzMXQ9V6Spum&J9i`#~ zcypCiHw^XCKfm$ss7&AN#>_><#)cLa76Mm(dOC|A_a5E#d%+qjmaYs7_^`f5=x?ud;2^ zHEHCBl*P$+LA1gC>h9TZXwDnfjf`%w!19RTb|aY~ASl?HDNqM;v#G+536KolZf+ zWv=U2Bu}``mY6Jc&J3`=3MoNYm zt{H5is55IJ4+D9f?QKvAI@ zd%JP)4(&()6Xa%$Rw{5y(zBVxBl*f+=zUm&(Zn9;((J2xfnXNdv_CF~f70wD&vG6d zeF71&t5|cZXPtvzHIB#{j=IP$a&|2oc;*uWVEszIH%M_48(r;=yW2ST_w`rR_^LC+G{ zwiXz<7K%-F2(%+PHZv6$KrHp@r>pJ5aTT0vIsp~sH&@eNz|qO0JUyfu<>2N0bwBnr z5m4}QfDDLHJs->SWHSJKLXJ-LJD}EhuAC`?&d$zWM|t^oL#{Ot`RMk1X$?d}$`mMP z7~0yhCXWD)S`NTLJ{%f%lh_T!z(iE0(8q`29qSiZY&>O`g*o2)2cF&W)cj;@UI8zXUHy0p-l$q%N|9^Eu5- zll_jKhuTBVm8!Y04S>Pj*=Zd-VaX7F4)|}lWJfK47|RWn^6zJIL=`ez&*2QuB})Rd z$)jeP?q^0UX7s8^@0x7Z?;x#A5w>n&YqV$3E4PGnRk|9pBRaVD7%5)1H z*F)NWj@G&rkkjSwfQF6X+Gepqo6S!P0)ZF;*-#3nN$AbsDW$ltUP+i!8V9C?A1RUVK~^TJ8BU(8ns7IQrRC;afP4tq4&Q zg)eDv%Wfj_)eWy!7v1!tC+Ca{8pYB~rA7bRHAo`f(RBfCzAT9D#?V$|b3%0E{oQp% z%!U!@pC4Zw3r$yAJh?jCdlAQqj)-~gm;og7$LT*TM`^mh&s6CFax5aH??tvqSQiyK zEo@V*TIB~M;BjAhLixo;JCT;4PloT7zQ8BJe)ssRRgtpzj8i3H$kq$CYxGtA`a#Hz zehxa@^QJ^cR&jO_K^ShN+Y$w9!ucRS@+ph>e+pm^8F;72mulH z^~j435~|moH`@67;H>ywGW}vuOlq(R4uNPrju3)VoAky)+~+9UQIcRUl{yD2_hb&C4TSi z1sMlbqax z=^BYGzw}_7Js`zzeD7{NtVEX;`D-k!teB#D9DOw2Y43zDyIKWM!5*-4Hn}-q=X!Oy z;2(T}l&V>ISHNoYeyxN?{O7$C2x8}N(C=uH>*`5C@`@TreUC=0!0&vKjY@=N5ZL60 zEt!UqQSNxA3E3QXtvGf9XMf}MJQt@w*D|2i#zw{WpsCUqotl$;|ATSuF2Sls~0SxtH#=GgB{ZqEE$Q*D1+ z*L!5N!%1!4bQb-Qzr&J=EF7%ce>;*uLoV0+RQ_0E>aMZ^OQ%TzPjNH{W?NUJ*6E~1 z${!yrFYX%zdS5G322F5B)G9_OoT_eU7bc|>ovK$LIVQ8T3k`%*p*wTd`xxOr5J%av zcs;0aY_h`euRj{^eEQ~ouuwyp35ky&`({dL=k-5J9?9Ohl=hT;8!puWYD6V`2 zm(E^(O^JM@{_-$;@=go+cyMOacO0 zzGG$~w*o-}sT%<4OkALn^--Q;htGLu5*q}gbUY-gfcm>X)vvyI1VdsNo%vc1RbRiB zf;nD0+5N$k8%q}rL+C{%oC2{^@;SJaes9o+VrQqAd zFyxo;UI{ZYbzQKyC7SyeHB$U~fen%z7#6|8(nq7 zLTVNmnpM?yYEp9|I>z~BhlS^=VNkU5D3cc)UB13mlEZs;`xiCmmJKuUv-&8Gqdk@O-ux z>k4(o?rb!X1e#yMj=_R!s9Gg_)vYNxNy-xU^XO#HOvk``_{mtQyc&~v!>xs>rLYuu0X z`+8y2oJH{dUj&JquC?n$uY)aJ=We*2+0Ac9?&siKK(Pc)1bCQ4Eb)%ikSOPIu2y2qJ^3?wB>VV#7*LcIk}sPi_0LxlQYr z%in6ru*s$+nOB#5)QO&o5p%JiTaB7Cs&AWo+XdWco#??`yAFS&yjje>+3o!+#+ig* z9=V&Ta|7?gyPs06>};%$0dc+$%$Uynk1%W4ZMxuQ3VmV=Lc9U=v(JSmhutj`CE(vG zLjRAZ|HP5Wn-T+O-V}0y};|$ z*Tf`ryiO;rznguN&quCmFR-gb{uKXPe{e@iux*rU_MH8#ssG~%MC zvt*k@#MQ(6we$Gk)JEdZ``3e+KY2HLVN!{_`+75x56BSIxvM{H98tK5wkt+%8b5qa zR=fZlUMVpkNcc}bA8D;3SHfpnf>flE6kZR1o{e9ne}$#yk*yWV1CnNZ^~CB>h*i)8 zn^0PFM5%v=>QN@&VRmc6xLRA*e<^gy7s~2wDxGXC;vr-F?w+h3@2K}FUkHcP&OIkG zdA@AgBpw9m;k&~Q6hW%XkPPzlqms!ghtC5lBosUq+ z&#cT0-?ug`T2}BNl}DrwqUZZsG4jzO^YUBc5)|mg{_p-k(KrK*X8NUd4dyM)59DCq zj(Eh4KdKaBClfl21gzP&;Jgk?Y?<**e6Rsq(4cFuU8dQ=uFntPw(0BnApy4YvcTyA zx%|A^dz(|yTN|E!_!VNLQzLkB9Yn7s;4Royl@WZ_eA2SrOJ=58^ms~nF}`xf&u^M! zuhZm|0VVbh6p3Z}vf)zEn? zNWezIzKN@=YlXdQk363HTd=jiU`765-+W619DE;0%iM8DwT5iecU3Bu_xDQ-#`Sk~ zDkzKl-9ApHL}XZecAWXa74%`<%(`5opnoP!(9k(8;_F`{d>+ObjP_4)oo&?pqm@?@ z&liiL3c(kUT!@t1YHX>=DvbPls>^>D{Wwg?WU;GVxI7VH1_BlF?`&T`G4Q@4B^oBB zFGdSsE8wt`cJqkuk8~|06Ru=jlIl38y6<;lxpus{%q{#J!IvB?&3xbq!}!VcvG~I+ zl<|)eWi)QHf`HdF(q4yq7o{a$ak){aG>_dOg9x%!1X?5OLaac(Lv|pSlJ(bV3dB#O z-h*d!@59!Pzg2Ttt1s=dn$kn?B7XY4KZ6);{UT@(i{bWmH8mqsiC%I*RkZxpR+nl5(dnh zeRs67(LD}6Mz}VOo11*k^y@R^HBcOPv6RyIDI zodWv=LT_EaIa1Uq@}N^}-6IbI+ZWEQo0Cap)^)K1NS;X9wwEOWZWlQ>AYs+nNFLu0 zRLx$aCs98@>pV^CAPQYCUrHIyvbF{-BSM~YFY(YruYXU90B2j2agL68$ri81KnfVu1M3KF5@GK!yjRN<$fNck!1z9;`$_Ux| zvn9NFp`fC;%SGGHb<}X^3`LM(9Nmab`K;Sb4^$_j4-<9F4;k~#8f*&q@=b%q3n>Ei z3_>;5jNCMw@d9F-I7|(cV6)2lf{r%qYpx$ldR=1*(H>h*4#pS0XOR(MK8LJw!?VpyH{QJE;9C-9wC znaskrJ^atw7vw;XtF4@gVjXdJp+)Q~pQp=?<}rcnO&3y$Ld-`H6%n!Yy#gr}r|$Q? zKOWc@IdV05MazObT4{_BiqPHHg01m1$vbpUdvUJcJeTUX7h0W&{mF?5+O(w2MEZUQ zVW8yAR$pIC;dfwaNcZ{egxB}pi!G;6UbuGM`H* z@qZ=TTDIAX{9>Y+J$K;4nLWMG*b;hS%W#^Fes_bX3f{~ZyE$7xE z?~gqK$-}E`k2;*`#$5^7z2J=qFoMLA>h~{qS*}h32$C`u9wfS)mXF`|R3RWk;k+*;$A{oJ~mW(948(@MN}hI$Ysn<+-!!_A05sT7BAf~V^JuEJSK z?SGFdcijDSe)yJhs*2y~Z)~;JbJw$gtTgfGgBgxCe_MV@o*a4LTs z;`0BP>S34BNQ+s`B=YQL{?<%U?sizm_-^^jv4E%GWsmVYJ68&UpBpsvm|RnTu7^_Y zm()@W2GTtu*24HHsp<37dvW^il_^g$-DGe|%gx_tW~_qVSNvQg~B)*}u- zi4^P$sMp1=#MX%EGZb5vtc_o}DvR9dB?ZgySA`Ge5B=)}gEuMHvU+pc4+G_vTKUN* zew33oMH)KSQ>39kmfTv49%Gso1!^W~GohbhCcF+x^ZkX>U6V0d!z_XW?a2=}nYs8o zX?d>h2Bz4WI@p$t*&*i8wb8PABU)DeE<;`rP!2tg#gIDZTIPgG84}&!%rgomA<>A@ zkxvF})vjdV4EKb+InkF=cU<5mjHR}MPBI7nS0{cxv~;NX+fvs^YqoeEBoorH+O^(aFj|_G_JXilgwro999?Jr`&E>C+oyRx}h5 zA=pT8RCAAf0i#8s#0CmA%)LC}?tc;l3_)rpPW0>mqY4D>(aOC`WxA^h&9oM zbpa2YF1M6OS8>;*4MW+3%R}eG$vEF5V5E=#ga)bKNj4jwXQ;~Q5Js4V2dzH0~bl!-C^igjRAxr2(l?jC#auz - - - - - image/svg+xml - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - diff --git a/open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg b/open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg deleted file mode 100644 index 8542c6e3d..000000000 --- a/open-machine-learning-jupyter-book/llm/image/scaled_dot_product_attn.svg +++ /dev/null @@ -1,351 +0,0 @@ - - - - - - image/svg+xml - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - diff --git a/open-machine-learning-jupyter-book/llm/image/scaling-laws.png b/open-machine-learning-jupyter-book/llm/image/scaling-laws.png deleted file mode 100644 index c501579591b5f507dac8a7b9ae3cab964b9d48fc..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 86847 zcmZ6z2UJth)-_6z9(sV#MU*DJgCX?Zr5AyK2uK%^-g_5OI)oxkkt$8;9YF#jO^^zFAMgKt@7?zpgKPb^Bj&T!cIb95MXhwwS)A$r9!>}aFcSas z^S?utx^r>u|NCnuT{$JPkpDhom)dgM^?%RCo?(LSKF03n8I9Y(fDE*4wtH}`me`I8GL8? z7_%I97xJ;JY-Ep9J$1M%f40_E{=(|Nul+LtI}}?R8^P5p^)xmKIPUe$jrZAKKTc`s zIk5VF|7D7VZ#ih0J?uInP5yT7`iJD$U-RfcHIWwca zL!@9V_;9H8#qG~^lmvV_aKT~LLj)eU_9o>^t-9_j9rUU7c!-6EQ?-K(Y|rX&bVkN| zP15?sH`{XGR{6b?4Hh>yPni)}dW4;4LomCWz^qYUzDMgS;raP_HtgxN{>ztqC8ecl zWwIt}P$)-9NeLArBX5-1zrZx6NK&!&`6(gF)^sv9n^B1QSRqz$9^7`h`&q zmY#3%bjH;Cp4Ud@3Ef(LBePn>f18;b`F7ZR9|MSWGT5j;coP`AVqg$0d-mM_o{C+I zA5VKEXT5Wt>=s+t!-rfPCI#xzvXnJ@v@|K_(JIvo2T5UO-=K}(y8^nb=?Ub03$F97 z<~Qjgy?g=d`83lv^9TkCGbB}i|09+>cmiUhq?s%sQMvlh9VvrnwJ|L`JuJ;)Gj?;_ z=4FB1z0d33R_jdCX$ZGAd55*OAR={W{WrD>{D@a_?X}*!AP-3K+&VKCqR#>4S;*O< zZD?d2Tw)#f9p(Z#8#s&FJH9OaE#ZGJ_t2u+GHYSDc^B;8R0* z-4(UFggysn)^}|7pE-7gwX*ar$>#fy4V$maB{qC7EM%HdS>K4Fgb@4r`Js(I1r`&W zCvX>bD^tonOuu)!BKCT&4*GNjgNDlw1tB})7;OA;nWa3$Ym5IaU#mJbJKmLr=F9gqnhahmdGnQtzdHYbuNQ>J^Mgn5} zpURp!GKj8+Wa_ldRWXh)W$`{9;%lLGM)^_kWBsPfD$lnTPW@R8zbEMlDQIZ1k!q&} zO&%wW*1rk^Y_+M~mz#Z_uQ4pBhWIB{BpPE$seSaV#}aGqMC);iO%a{YV=Vb|#Q z;NA4AUQq(@UHsSJwoFVrO*Mon$>il)z@z634|JYcg^o~L#9eK4`h&Io^%P%1(p)uN zE)0D=kEfdh48k609XLce;n|BGZ<>!Tme$*ci3hJWP=6>)6Seu8(^{=GCGla7LVv51 zQq9R;J)8y-Fq9%*wOe@aGkU}KfLuFD-I$A1>+xaK(ALEkF!zvQwf;O5l^dWV2p_Hm z{poE?`pN&^b**EwBViO=&+4EDtqQn1Rd#6jxrbW*@%Bt3gyO+;#}j))Go`7zfQ@wZ zIzOZh`txqgfjYCq%ZD>|iJ?Ig_kRp$n0z+XB!Az~AbO3N(zlnRASM>BHKDEo!7_O$&6xy*SlktCrZon^NeT9VO{Tr36YL(n7f8^MN!Cn_@FLU9oE^}59s6+WzI8z#^`lJmhZV|rxM^gkK|eG*mpwf< ziQhN>%Wig%($lmAHPXPYbN=UiM?uCn2*m83B_i3 z;+@8nl^7H_sb~o*W&G~$-j(}lI=fzb$gq3{^tHWSZ_xL@`|%^~M}YdLAdk*`+7%h& zrc}k;3$YhSpAr}U-|r{ZKJ^1gPr*? zAHH5Ep|habXG2rzn2*@SwBx3y*80*tRq)rplK%T1xz=`z=Q7*hZ@yw(pICyFU=IXD zzAHqVx6^#jEIDv?5!4ExA4__-SD17}x-Ls(kY9v}$QHbbjJn9}Gj1UO> ze5MX9_lb|!eV1+qkoTns{%KU?ecyn@E%8>TDH)L!*h(B)Z($xLOh2-lpAc{J$eGuT zB<7OzOoRUo8G{IO={;Ej-ZR?FSLP~DxNAIjA}YKXBEHPj=j+onfsJJHQtoQQT znWb3qehW=nNP{c|QrpIG(}s{qB=WbnKT@3ODXo&cyga*5=XuF*-eNPDgccv|WwV!$ z4{fo&P?vzcy}imt;{mUc9nba2|IB!>N5}Q?=(|u0tI7YFb+zqA10dR3PEPKp58qJX z1ov-IE0Fc$iDy_*4J$h0sxX-Up$PGOGa;ae|o^tsBkD#3Kq}J zrMko?qEPv+zc8}g5qKgiP`upc<0NAii?Di^l@*sG=43zomc#ki$As4s*gBEByY{7J zsdsH4>1XXr_J4piH@B#VID|g+$M7&!M6jqa7luUjiQJ!4rJre;ENVx^lRST>>3GBl zo7CT*3-nv2fmC=oUXmpmB^n=Q+4yL0kCfl9Aj)GQ^D{&A_>xm=)w_|(z)kHt48RBq zB8P{;1t5f8DUK%*!FDeGWcGJ49a zXC7u3IYM^Gl%L#DT*ducS@@80ie95lHr(kop=3}_`(><1I3exm-l~Yp+oGdrHN2X> zN`N*#K>}P@)xnr+JiwM`1)X>hbW754%eAhh%9&UWefn4FZlx#cmrL|Cnk43y)o;8_{sbBQ`Sa(1aqsNM%nS_NquH~y|B!o?-tOlKMC|klbk1zQB+@XpXoh+Y$UkRK;OZ<>(96b zL6KKMCnrcSLepMBxcACM+|C?4MaMJtMQ(GEot$*W=B3f|NSaSCgK7?0Ga|_2><5Iq z(Rn*8$pyRSYM~vD=V2u@Ei{ClP~*Dj*r=a2-qKWsTo3XMtkGbX`zw%Vc&*5NQv3W# zf>-49#b>}ftB|!-wp>0Xd0P^%L|+l{VJ0pB<&(a%f%uZc0;rdWaDSWcdDkQH`zX2M z)sGEG?ko`{#RqI-vz0Eb#uXDzsUf480!KfkmhIq$ax|fLI~6c3?JnQFnbqzH{4I$C z>@q|3Q99gQo@~i9L$@J5&Q1TirD@F&;HCFVowGAOp=;@BI&{ABZ~u~2m1_|`eh9=S zdD89X6EQ9rVP zSRL-=hSsMwbSLZM^tiZRK~-j6ZU%cWk*!{=gR5;g%Bzy_dfZ|B3Bt(u7pTWNWNV+6 z^jV_Cl*o@w@C<8mx^+lmAgNtOxR8RsC?|k|Q<77E%OdEc%t~;3;vB8F+v(Xr{uzE6buDu+)JBd#8Xf_HUIDAQdI}#cg#>=#J>~6A zF#G}=^}k2DB=bah;7T7UMTR!4tk?jKbO9&C?`MNiL-Gb|x@K3aZI(*45CpRcYDd$F z6xZi0$3uANkuew@S0YMPlWAeY&3TU}`yi$V%~WthTd!U4nd8%i5P@td1t_#j2yDXI zW#8E)z=?-s|9l}v8cvM|MDj0Y75afCx}SqaKDb{*jV7oO*ko)5OwIN9>abm{qXk5I z@5tAaP3KdED=6s-#ufcz8B7RUF}uy zsr>|a4(3V`ql2jD5~}KwgqZfS%LRTlUhdlc^mziyRqQXe8D+*qM{B6wjCZ}xoCPp^ zKbf&AOQCygEq`~7cAhLD@c4cp319=8&(v?}#Dr>E$mIX_3;i*J(@Mz2s@+qSHW8#M zM31y^bF0So&D`jIq>8zDcCh-s;c|57OFdjm@`dVqw=cn^{kQ8{+hLM#zB5Ugup~{Z z+ppW}w+UCPO}$<5nfylF5J7+WM%tv<_0N9QQpJO2NI(i@D*Kn~ybv z8RBFrd#VwFE^hOY40*iJtEU)!k~Nd7lDb7XyeCob^eVfBVkw&ROKZX<<}bKvZwn`Y zBWc_~loaQ(hho>EPN=kqgfvGThU`9=*&vnmFU5IBN?j)pE-wJrp+5vi z{?@_`wijqcuv8q`&D7fK7|3cG)dFI&(UthQk$IQqxzS=w8T|uD1q+e=E9#x!lyfLP zs(NDeA0NVvKcy4Qf!2N+gqY@3cx}2ixvm;Km*sjwsL?9p<2fbScfT6PltLG4n^4a+IyGBO zF}oEx3#fX9`>r6jqeP>UKfe#*kNK2$@%B;r;p)#V2`F2rU%=5#;(Q=4Hh9zk#@}5$ znTkoGm6e>JeHl2iCTDi`OW|_svwK@AE+=XBS(i2`P3A~wYiilhQbRLt$3c1i9j(tv zTUb}C+XY}Q>MWO#s9eCDp7U$$Al_1#4QDAowr=FoTu%SYA-)MdVRbDdY|}rLX8ZQ% zwu_agVfWmST|-|#%gZE1&4GQZ{KbnGxwa-BA*3Pi$_Iham;Bu#Tag9D&PxvDON^Za zha4kY61)F!rXJmHT?(_57Sn?Q%FHSV8G~RywyRBp%imdNet2)3{6mg4yL0%#;I*Dr zxs>JF!GS>2+doaNabK)jet>Uh#9qByz+waPXWw2|p%!1geyxHfJ0^>jiTEuV`oU)^ zUnk_r$5pnilx_x5H^X18sa>l*^n2*@-zOaq;Y~Cl4PIXzx?lgjmE=GC1v;Ww=7bzm zMZQC)mEpRbX#<78aocEl?vkzzzMVN2uxF4#1S}=0@MAkrd^H}lNt?$(jSp7uZOn|veKg2;W7i!t=u_GgSs2i{)yXf1ekx=8 z48IiU{JTg*QEBhzXadN_naW_a+KSS$H!kKP)D_G4|`a-8kR}AngGpAM0ZL&30^~wp5@!%BkGmm>cz3cAQYKwVNse;s%;qU7yU68~}pk<4P z0Gy3&nWmY~W2s%H^L&jAgIs3U%@vOb3-8*P)-RBya~Mh;X!Txyc;bDIT5b5*?7r|Z zZ*nY0tUmbSSPSN4xeX(9(%xn^pkk&~bx62`)GTZ~l&Q@$x=Zs=YU^{2m;VY<)C5$S zt;uqzx9-Jb*0PlfO0w26hl;lmk~1IHdClDeP8ccR z>pmei@|@Hx(-%&>=?C=rk#vWKuk?sYe+@WzY%CpXJE`!29{gn@eX~SnLsWV%{wt61 zf=VTNqpi^v(s$0t0qy1?O7yp3NTH5K!fkwo7>^VvT{=r+iRQW(>Rwxm_u=KCB{7{O&F}3+_yX(Cg?vmH4gpnuQC3*V zpn^=!Y-V8h{AR@|(N}W(g`wDChH6nYQB$Vy&BF%|)}mQl$DIO6&!7s$y*`L<`TR){ ze@bMdAxJuyKA6&BEx`CqN+#--kdW|46CJ~Xe#;XpC0O()k=3mAg30PVzGU-*nBdPx zX3r`F$wETeu$;Qp!N5IULS7*G`Lhz7XH73nVvwL!r;H_!(Pe08nl$`;{L#aQ!;pDj zr_=rU2M<3-F4Hw{lo{M0yY5z2SG_H}elEr?hY1M@$%KW%+2x@Q zv9iKJs?7F$1$P1J>Q6_IpWAZhGaDHHv+r*{jH^bUUG@;MgEQBh734;|^Qq-f2QABN zv>3u*+g1VC)JDhIQpVZ)bZryWbC=>UrhNBePqNTrih#ggEV1_3tfsfX?{5lo?Kdht zW%hx|xs#&mGMutEj@nP?S4&$8BHu>7S9(9aHspTFASB*_!~`4te)VJcF%9*1DRET* zF0dm)%ixftxcna-fN+1`GOl@2GO;{dQ`F6rAgx~QWE4;=4IiKNG zOq8ScI?{goX^Lerbzvp$sSm@uOB^Dy|77qtJw<1Us-k8!)-Hu)`*Zd9a4;HR6S(u} z=wS+r%&+zh%<3(&4Z(H+X>nhmmxEvoDB>u4cu9ldbi5m!%;?VNZ^ zt({mc0?kAxlcxlGUUg*GH9+(m?e!=$f6KD=_;jyf!nWQ%M*0XmZQk{*Na3UUe8Uxa z7sK6I*d6y!p{QeGKuyAO=#_3~OQwxS;^p_lZ`zQdtJ8zN0>35RB}e+Xj*G3l&7hUt z-Iq33Dh-W|lb=i;yNqhz^dqkZ7;N2I@bw+t=IRt{5I-K*ipwZmok#4YMurr|+{tLbjX3^EXuc&LrW-BJRF{!mD@q?Iy9+YA44Y4_U z_xmu9=mxQ`ABr!4<-kqgM*z%;TAGPck}Xlr^mVGq7e)c9OnaIZAHP`5c=?zoDa62L zOrlQ^eOsx9B;Pr+m(0SHMQlH9sg_*Fc{JM_HGo09*k3ymA|!6YTDK*}+R0?85&R@(8V+-kIpJ5Bre)vkEK-q-jdQifGLGbr)Po{m zHM4lthi_U@5tn+FxLmeZ^Ih+db=B#>xFf}4-a-`gLPYN{&VDKLC|v2L5e}i9?6I~; zv{j1GPNRq61C`PsmzDmA?7mCvM?ZWCWU;oHzcRuG>|ksKEYuR5u{T>AhYe}a1FQ%) z<~sLcr90xE>beP`4cy!=AR{GxH#>vPKKk1RIeRS0+(9)1RyStd1MB?{;J)eK_%TEP zP&l(ktD@{5V+OfEOt(XtoGHib;_@4UG7-DruK>N?!z!6(A0ep zmVT*41#8dt=Q(GuuqbTU6(37u?+ICc)hLuoHWx0k>pU7{u7ADr6wNa1JB3p0 zD;N2Tl}u^K6`X}vIsrVGtXgvq>{!kaeU$RzU6J|R1g%)&B32PXNdk)lQ^2T`$BWoD z=SKWH%}Zl$UAzRNAr{|5X$8C1v=;sM3bs*yDZLm^SbmtwMcE@@28E3At!}6O@T!%+ zsb)&P#je@Jj)^anprebg>Bl*NEx;eZJj^#{ybnvxRlT(;4U>(9>3F~|7&S2!SjP#Y zrKc!WHh>)XG1NPUh7MayuZEZnl4G+w&N7!mM=yX!W6zckv+fU-Q$v$v(;S01%W7~` z+Z?~k6*BMPYY4U%EQb^deJ7~PE?jMLDi#b};61$%Dy0L_K`nM}v;FnP()IA8ha3X` zJz|De)1oO|2$O`r?L2O{DtPw`iio?wv=y{^aoEP>Vf-%aR;vN3kLxQ!-6Ic%vG@?*-qzfb)nt58gWHeiko z?k~{PVeAd1y^6*Y_NE%2lOTzR-d}f`_ntTt+o_WWj-@CTmu7IXgQN?a2lZrlsRB5`J}wfTt~1I<$l-yMJPB{Gie4^FS2iKlobO}oY??aWq*T*?!nZ7^O#tpR?(2yiAtxxcv&#p!2J{_NqE+>q4s$}k>WVp{sdOP zo;9W_NscUzTZY5oV8rjD)e@|6VW@YF9JJXdGRX}Kwl8#ugnI_|2#~W8QL+JxJ!+p2 z5eUQLGEtTSDNNS_ zQwboEfYR$%-~7WYhEoil{Q5q=7K{y67|Zpq$AX2=9{~o8JSiwE%q2{c;55|CIqx2~ zT>!gTM7*p@b}q|GW;4^4HYnjvEUp5MvWH-(YdTIX8T*;sACU9V@Q!cS zm)&=0m6RLjU+Gr_bx2@BVE{0DVDNd~%@BK7Hlh%W2^G1{Ky;(KZZG%rsg_voH@u}) zHsOurJ>KbAYqPW`4$)47K&1s!7yt>5N8I=nxiUr1hd<$>EoHnMdA@D-M^s@wAMs~c zoqTof{+sc985;YSBq=!U^nr-vSs)e5+gHnrg+s*(D!WR=^ZI4b z$MS_vecy<0e3*-yl|}^iF~s5Zl=4E?sYl=4UDrHqot-NK4%SguzDog!0#xuiB8d%3 zs$7V_utLMm&VJEp5qmLJq)92k8AjZd5v+%597BRSFglJ^I{czcac~8u2N$eoCnq0} z(lFP|-i7dG@}RYaaJVXD>PrkHB})9hIkVJxE?k9EX#^zYb=~g-Oi2m22(Z7mDO#@E zsK5rh$gJzcK{BTKh`RmxsyS1^<79F}FVy@NAyG3P(JruXjp!Oyqy!ZdzMvQfKaH_L%iJ4vLgRVoZwmy8c?{UNfd56E$W?}3@t zX>6}!KQXH!K=kA8JBm)jvgn2R@34#8WTv7ba$QMZf|X`WvsK_oj=r7)&KY!PbqfV9 zbiex@Ap9+S&wDwUuc#uo>7H}7*VEbVahxN18`ngqz$NA1kb{^nKHcz;y2EhClTbB5 zRJOQIT@eR-UQ}pV+IET3VHwPZ#iKtkdTQfz8sb*m$*c zJV@GCgzDnzK?9dV&PEgtjf-1^ARIP?T`X@!7H2HCeH6la2pEh}zwe%G*)nPn-eaMh zZ05$upfCj?aSyUfHzL6MmytDkE--zk3H>?bB!;*9NhMdi7uPf?Om7mQ#2 zXJX1Zdgj-Z`s#Qk9#(oGaA*}K7i~YAA*gc(IWM+)$5`Ue7{^UGb&9DzfhR>4R1pQ+ z2K+od$5J+`ZXM@qNi-oUP_a_4bJOXedRGd*CSGWM0U6^0mkHKQ+GEJCN^!HI`UdWIOpbQ!wv-%3{My83+G}tPb zQU_cNE}y8gBO~>K#JSz1GSAUoUZ+0YW*#KNhYa}#bJD~wD>t{7PHdiK$6Co!++L&g z+ql_-Q4t_kfsXCyCBH+&b582>F-K!sq)3{R+t69kRk2}@|4%2e*;2#qa5bF54Btt` zPReV!^~Pi*^_0kd@qbWFOy{HE^*dw)(PrPTja25kjpf?(%=Fj?WvaBL6ddA$Bo{9n z7-P`iqC}qR!Uixrfjh;LB=BJBy7ISI{4PW4`cee4S!FDOz%)74hSi-+rB^RyvJ8P! zM&mE9fuq1|UY1J8|zPox5trXd2y`(IWiUWEb22!VgVPKRkbgzfT}6|2}0^zBIJ z^aazO95(+#N>^gxd>Ahda6gmHFT>3Cr@mOsmY8Kv=2>AZ)l%9V;7aw?2H5Cl$mR4A zU(^moyV@QTN$zDF{Uz7A^56q0;1x^s!jBuDBmM5j_>wD*5(lWJ0=Vfy)u<9Ippbv} zx#tA5yC*%})x%S?mqlO|9IJ4P)uFazxXgJHyD~5}=6#CdS@nsHir9Xv3?>4-P+5EZ`H;hiyvIziKUtHVcgx)+VHoWz$rY5!&*Q!%u-71c}T_}0ZvdwVLp z`yzTRyHH9Ss++Fh?aImlrKI@i;w;jB5)4n-dm;FV8gC3;#QjdVH>wReQ$>g|J-T=i zyp(d5j~F?NE({IA20r5A$YK=EUMy_qEF|x93ZZ5Bh(79VHNvo=ks>hfPX97~;Ov5ElT~XUp zu45Ef2;2u&0_Unf!s|Ru%PUsOq#c`r=!e!tLpOM29o|cqGvt;`3rT86-_j$cpVI-g z^<+BM_eb@iT0&h-#8?l3fAjbg`5Af0cZJkGykOw(YTj2w}==bSS-m_ zrIqPcOZ1S_dz^(ci`Zra4+s1k0bseJKGGgGYez!Pv?ygn5BTW3(gSKs!|aa+IMII+ zDi0ut@jnh=pO$TBQ&U2j&lI1~{46?DcZ|7cO7j^UL2TxJef^%^g z%k7HDMhY~Qf4I-7?S!(z@3rjGUC`sg-8NG`%&dePNeIO~Y3Zt+s!zDFkAm5^r>m=- zGjQcTT4${;Ue5;FXO}oWa=;Ib1^w~Zq2|CGPT)H}ZvPPFGm0xk)B^lYMxT2M57J$~ z7b8wKJ-I%B`>82wjH!@JS=i}k?xQG~L|g2K8TBmDf9s2>WMv~arp-2xNI=exrFE** z6VSa=cD0q~`4EuBrV#k^{B}}9A&n}ACrSSwcR_cmtS?^+i$Z3^M0t`D25w#yNVXeZ zFI2MVI^dP^Z?k^%TM~w#IuI)vZ0>57AoK;KejK+i)nX;hzC$$1(3d#G3UzA{q$p6V z4ujQV+4o#Tc5ib)7>EFxv`g`~<;CbN75Lha()#YX>+2mUJQ8k>*q3C@(U4^_rW1d- z>&vyH{Y{O9$Q_Am!G*ncI}(^LMi~9w-gV6{5$_L^+8nCQ@nUh^CrN2K-eAoqkyj`9 zCz1e$OE^Yep&*s0E4p7mEja_nn-O01%Yn04O*sERz#J#7%e^P(93S(=4H7S9{V0q% zjfNO~Ip_bXBlX}~?}+6=z_}TwIV0r`dp1A7X`k4qoeZrRqodf2mh;N)*h@)1(WSu;rx-I z!<^(rq7JF1>8}coaQm#Lh-kFcxgGFS|v7g(Wc6 zn`~LTXDsmWs~0u$3x|weXz3<)8P-sP%#;fFfKaG5=fIf zD|nkDgx-NlDebwO*R^|;3d4NOa9tnfDjke|3+s}IHB?A7$}v$es+d%|Qvs3c=;Md& zyR>hnbv;VwjXT2|of05`8;u|H=}NW&w@cz)rN)P+ z5KhriF9T@;z=jbx=caA?YO|x}(L?7i5XInq&E0t<`Y*8)ndzvK#!koQk?;hVw;In1X|Ni2- zxg6PfRnzM?5I(34!A6AkIu$aCzUJ-GlTVMBy_7!&qd=&YUtzIgUlKp{^ zQjhS~#@xpBSy2yJh-Llx>I)7+{4EcV35~lUb4;gV>gCZwor~Dv+v^*(Rs~(%8rSFw zSSx#JxlnIm9bjycIhbm_LdH^^4HR&FJpR2t+~LnTXu z%Dx`5$9FWtZ`rO9YJ2cZUfIY3l(v4a@UdJni&LFD%ktyBI)hM2Uc#MX(R;N5def!) za8@$A0unGgDNUcVp;;wGF?^Uvn?nWSc2f*82K3=C+d}*Rr8M%jV~)+@UKi(=!&kbc zF#_W;6WrB%n4$MSTUR%cN845hA|a<>mhmd*hhN-(s}}21j}0@aaZOUGaKIA?Ta6M( zD@mDtf>Mt5$9>Qg@a4|hKDeBPBdB!M0g{X#O(XYWBWguCX){@zRbU^S= z4V?A)0lYUlP9o7A)RORKZ(u>Y8vWgKt%)idp=H-4F(y<1<%RXySqj`^-U>?{_@(Ou7sbg%UtQPfg>HeYvY6zF$G0Wa!w=4D$B$9 zsMn|U`W@PPht3Roy{0=yogK-$&^O`R`6RX))30^L!~d5yqV2m<{VUt|UyH5EZp&u# z=?-}hbE3NR6KvDsUym)aJy6{S+3OeU!KDf0d@iK|2eqIO`ms`IcyG?qd6ePK1HJh- zaG;!lpi%4}qNHWRufMeQs6C`-dA4~2=L=Iyasn>$KD1U8hbt{g%LiUEmV5Ybo`YI;;L|+bxG?fJN=63<{KdLBA4mT}%g|({bFa-0hY6X9rC{*X1ieC- z)T`0!DL*d6<0n4DhjE%^Hzb%**c}0TjP5w@P}#u*35x1;V_aV$f8EU4;wpJ;RI)?X zcU(%3G9%+wdov#-#g1`|QYOwJWu1f?KE=PQBA=tVWHB&w{=K21-{KewHL2ra9!T*u zNLpu9*y+eQcXwX)r>5_}IV3TSY?TkSy7zgyf_w z8kXu5pgI(=;|0KH^Ws?6IouA)-yHA2R`rO*XTr+VMBZI_+X|8MR2bXxa5P8bJF7{d zI$kG81SG~Oq`7`l+ zgY7wOYB@6{n6SC`pBLx*+vxA(aGV%%S3eLfOzs0+ zHkh^2v(f@UJvPfP=|lo!=06=W&5cEsLo@M~WFN&~$+^*s{7okqoR!H?fP9C4X2d=y zb^_t^Vf98QWDRn`TH-yX!7fdAzW;nxPtEzr;zvTdL9&}Lfc!dZ(3vf6$&W2-QpIWa zCjy-V$4sRtNt=l~7sDyJ!o+j%2bIKvTaP*1cmB4hKBKy%;~i>nOq%Xq54JAu%r}bl zRSEto_2Ro4bV5r==M8E9n=TdVtEm}61*rJ6F%z?E^uef|aJh>_F? zQ2Q?@CA1sHqit zJdvMu^}#7jZn(}nAx?$zlGG|8apYSZ;FD9BAl1kU3FH=2bKl!9jk{q(|7DLi|7%Fp036XCmdM z5N7;q*Nrj>^Z^~ZD;#OnQW3dd%SnwVEOTysv<%FVR%ieyKLSF>LL`o{*kQwN1O=LG z?fyzw_}B$**0z;=qCw7vLHje`*lzQ~(D zdRW*A$unooQ{$S!UG17V%Uai(iLQf4lBcrR9yg9)(thM?BCw(DcBb3k*rhY@MJOcY z4Qp86mM*jYm5)^cHhaPmX;A*M*}tzEQcq=!c*b~Zup%T^?M3p}N#%nJA>||9H@teY?4l+!44&VD9uB%; zXw$nM3tI>z4JgeD^S)QFmv?K*0^!r|lV*fW*a@3{fq%L~zlpWqZN6V^a5c9^CsgP) zUWyAXY~z;aRwGCGi}_7xP;n^PSl(2flob@vth@d`%F$ZE8ihH|GGlJuHRK3+b_ug{ zlFq$(JL|-l=;^<-)8wRuf95XmjdwplpMQ$VKNRd#PT&P0$8|Guw%~t@G6xJHbfg%qnH<#PgU9Zxaq?;9*m51>1^EUK8CCQ8Euk=h~HGE)2)tOkId*BkL#c1 zaUWLCfq&c=n(5$lVOKeR9=y3k0I;{$Xr>XS+0b8v;R7^p(H3qqu)MrD#}0*dH<;MX3PDM14g)P}-+o^w30VJEk8A}!^V=4!yqO%uHQ1hVxt zV2*~6=nVH%a7wPsm_9@8sV}JB(?&68TD~cX2^kMLA#YJTDfRuCyu)uF<3qM`g^hm(c8}x@)KPC;SbOOx-2+ZE+DZ|F6_#W{ukzl5mdjN{ z`=0TN_r3G!m!^96p3~+u!3s~MsF6r+0BMSpmcD$*bCl0z6R5{O=Y&~ANQ#4mpBJ*7 zbVu_aHG86x>H6>pq+D@@{R)S+j#`D70IdgyZ=cI#Quo4r*4ie%kI%emF1n93V8?el zR)TBOXMbRs-R;P}m?3gPK!Q}@9PdCHK0c1B>=ca-BKvVV z9^qmIknbz$K})P51rU`v2{@&~kFu2N`@|Mm?txYMsuYFzl|6G@M;9U5@Z1bPa3OLp ze3sC?pY24aD&^lwT{fflvWH5KF6{x1KE8$IwYIj#3ri4IiBF(g0SDK&SIDvxwiGf< zwH0Y#;C!~(kGJ~T+x45L$aZup@SrRrlA|KihClR8TiB?zx%M`0DN%Oe;#7p{fq8KO zqK64@@TQR;XUHM>#feo4mm}LhKoZY3`vUYpZ@nN;#3aC56=Z{KiA^xHYa8fV5?Hwx~k4(wA~4 zGIF!PQ2-hd9NRnv>UPFq!jojPzo$?Q!gDr=H$ynbuR91J8{CtKYScwMF*8z~y#xdM zOm8`dHbl%};nb9wC3y%Acl_yVQFM5E@kNc+{e2SM^tXX+w%%iDwbCp08@MSr)kXRa zr0~hs7x`Qbo8l^U8>?y0?y|}xRGZt2|V!=e5@SC8hc|{UH_d_`JF&t zH)V9$+CPh>8`9St>D_Mc?_OLJdpP?r8CW;_ZHx`G-+1{o>nvgVGkJ3U)E@x+$gO~z zlhE8Yya}+&g!OgFNx#W%767IB7RC@Q-I23~9Y#J%VwiWYFwOY4;o4Xai$qx#>kaS) zt}{U)hI^kUq>5IS5$%NXZ>|ASORpQ_Z%1o72#{e{aT^;*fL=y^K%&WMpVKg;;}VF# zSWoTgVu`>p`oMnY&+12-hMJz>zANSTt!6$a&If5MgiY#c!9U&^K6 zi;Egeh_b*6&joC}u^Z+FbMZyYL+)FuNWyz1GK<}7yoH1M@$iqhm99`lu(q?{id@NTAt6mH%14gL z6woTPL^fK9Yn9!8Cqry>mN)y9QHOfv!h%^q+?N46)0BxErn`>g{2gC(HbtvXpu1O? zk!^SX>Z_MyJ~izxw6SeP=(O;0Y7>ZWG^#n4Ew8bF-D(yZ`dHb-eRF5#4OXhTlD^<`FD2UI)zYQX0{&8 zC&8Cn87TkSzpOvf+M@t9Ez*6a@f6wvNt93+W<%L`*hs2lU-&uETj?M#otU3~ z*k|h&&`O~SMxd{Wttdplg`Xu*lh!i=F?8e%{m#!cloP9PV7vmoG)(*(xzi$iR|1%#KQ@>fUb0g+&9s5eGCFdX|xdbw@8+38={+WO49^dueict@-pWwfJ zXH`rBn9oFNkkDJ0*ex>e$oXWNeY;}Zw=a-rV;1GJpn&}^0OhZk+9#ZRpEZBA>trU| zua76UIx;qOtHs)4Ic{*)GMfhXD4#vBzkBz%wr^qiv^zY9xxQZlWOe=Y!<_zGa8Tem zZ}JYoa@jrMr0x0@wmwWw1^gm^o80!}z4KPx+Xwx+H#->*8NwR<+T)#?hf8^0R0gKf zv;)5{5}m0YX;F{;I0Am#Pl2;D#kSKj(ea#y*o&48EW2e<`3gc&24P~0%fdk6*RX}M ztvQAk;Nl_!Jg15{+T>)%#pxm5Az zzN@3bs98q)9w`_Z_WX#EsTEZNkkUQ5Hy1fco!o5qGS(2TVC07-nZ}?4&{L7}IhZiC zxBk@W{}6T7K~aD2+ov0mZcq>bX=#C_1e6dFr6rX`Qbd}iOHxEaV(Ahjq>)aQSh^cr zI(Nw>euvNZnR))kapryY#2wf5y2YxFOv<&V_l^v!*rlyG7waG2td(umU>Rii7JYV4 zJFdri628aAOu+h~8ii*1!y+Z`o|Hh2?s-hfkl4q2O_CD%7NoHDSy{cjH)BS}MkRhb zN6!Mjd}NsyrK$9-Cv|`A*G%O5tX8a^BI9C~-lgWuQu{I81vQVUR>cg>YB5|K{2`xhC*>6jjl4r6ugLnRDKwg*%hF0p> zHK}Z03y-~$yetNqJ<-U5G@`b@tr28WG7Djzz*$#4^^ow{de4}`#Q$Hwf;ve63*sB*J9=Bfoyk?(VK({!4@X@iOsn2yZ+f?SPY*fFAO}w7P~db zJdlDV>K1HXHZ7!!&j8VuGA}Sk)=#2hL>uAVkplAvJ+fOD?ZUws31?G$8E{yyn@{I@ ztdo6*l?W70fJ5RNdhAl~itpPl>o?C0j}KlfZmE~oAkokYxy842r_}_;7WRX%I53XH zL~`lg_}Hi;SlD0B@02UBlO72eRS``g;q@-*%KIKCV>TyyZjbNj&yhAaJkcJxIs7C7 zXuD7Wc!h`Sptabwt6=j6gberms4ege6_=Q3ZJOhi6_7aaXnZt#@y71=CWXPGqo#0b zY6|Jp`w7$t!p?=z{a?Rh@9M-aM$L$lJ4v?eCWl=Q+c|a?@BCV_hTtdvLY#ZUz_ znY-6{95ZHXi)dzf3qN?>DcH0__5&K?gX{$wnXn8T1EVpnf8jQ@Xlv1oG*f5ftZ_|? z)jFis%Fm;{9OktQtV08Wt^BSmIbR^kYta{;B%WS`_T8Ku4_;Z?YE}yl{dR}Mqf`lpFLXZJ7Hljm_fU)$owgTF6f-y! zM^Q5L&l`nJUgZo;h9Tlh?BnAGz;%ic^c7*WksN1?bu|t^cxx+M`(BE8jmi%%5AXBq zamy0S^!%e=-2xu@pUK}J2p0oS8B^4tB#)5l)Bt;&^d99Y!Y(Kd*RN|QdQalh6>2~& zzQ?qkJ_u>P&SVxf#S*Nu1wCMhlok5kulPo zr)fHn@_(%XDdvncll=6}k+Ggj6+8axOc zI0qG(ydxe=R^_#9*ZWuH&`ThWA1w2eXEb0A*|pb64TdZ&irl$_1>8&&=O=ryJ2SOK z-~LTfe|Z2qHrrnFcYquqtL<*l#pIntu!+v|-T2CA3%-7(Ig6QOUbW5gJAV2{u#+8i zyfn0Hw|~&8!XRvqM?kS1TWvR?PEG^pHJBSi8Tqqyz8JdPZ1f-zO_>j!T5V(O!;Q5%AOv{7bSRu(*!pi~$ zN=)P&N3JjSC;RP1{Y#$@uMc5h-617%g(;bP*p3tn7z!6`un1l?X z7lvJ)S+Kt=W&d0dEVJaxGw^K<;gTN^pqmo?2)kq)0msO4zpE`6uZ^xs>%F`AaKmtp40JX_M}pEYf~Bm-xoI5M)JilIqmVD1UXD;|XkE1{FB93%uV$GN`7&=g&6ZB!{#RX+uR6-G|Zq@`!R1`V=^3QIZFo_Kz6X6qozXZ19 zbar(;^RO^j`>zhYsCE=;@#lcGhHPx< zjn-$f?vlV1$*p^r-8uY}jcEj7nneTE552b}pjAGXS|&gxxBRcAXan~&cTEay?DmA% zS3AmJE$E+jPPw^$a&WVtexP3;fqazd9`~}}FSBf#(V5c6GEEqoNkQyS>i$3HOt6&_ z{1>v24pIz!IH{<2SmuQ?oLJ-DoxeYj=$Yb z^(*j;%12}G3F4+T{=S~eVFd+B?PT>FIOmiFWBtUrKXE*keo-e z7>n@V%kcFM9`8$fU%C2RPAH+xmO}6ybzT%YZNP~r&qdwvB76K! zZH4%CQR}YI3}3pn9EN^{W!fgShzV%kPyhsrcQ}59WrOVpe^(JtQRjqVdt;M z_%D|)xnF495ya$jV2M@7f+O+o-#_juYepG2{?cE&k0{q__74xgKwBF+)&(WpcWbqML74~znA+q3v+?%QQk{J_6i~Db~R`tR?Q6JYW9V7 zvZ~9Tr;QlU-|DGi6dfvaVh6ULg1{kzoS`(Cn{`&EKTl-o*H&%omf)G6W951B*)u?~ z?Q}cx_~D@ZbztQFy2f6n@&d6z6*o!otxof?W*$m(-voR4#U0fW6kQzRGYq*adJpb^ z8oNV@W=~uI!dl3m*b#rSbE)xfbh5#|akC7OJYQ2s6tL#{iHF!^R@#j}spBJG@Ey#0Ri5!&brPpJ2V^A0qU;>Uoa^P?lZP zrH=)gw$#2VKvnFK7rN$^uIzk=BQl6jf#dt3qJ(f$w;b0Dan(-$9PRcB$W`!K+t;8U z^9PpCGxcZv3*Nr05)vkb>ivU~;EZ6|$eN4DGf~(xGTg&ac+H7Ak7n`PLgRR07-`MR zy0Mn#L>jLXH}GrBDg5T`c!do&b^_ja+!S%UP&IZKjjyjSl8d|j?>?|1 zs)pTS1sgZyth8>KWY%LP`ic5OHC*0X>=1rR93a5|Y=l+)dY$^#b5Dd}@_t!MXxki2 z=)S{o&x&8EO4z|bxJd2QDA>moAVj=1PR-OZq%Frhu{csOV6e6v@J{!BmQx1ApxSnf zc*qhgIr4u?ZtRRE`fHi~x#dcERnUtpq;xLNgPm89x2-KGZ2yq2EPT=F=-=MD^cePt zd+<|;T+>4y)xOpAyM5ozRq_&za$(5$4Zm zw(bX*@Z%W0!0x+yNwb+hc=`BXQxrOHvjVk!>TWqBgR3`RkXg{?KmW1afp>K3!Fz3+ zpR41QK!Je|aXI=H>4a{An#!q2k~7HqnWp|Axrn;A=*EBAB@u|8H(r~r%x;b&dHyY2 z(v{&iU5COe1_4r6i{^$K6E@Sh!D2-F1@IlvB4oRM$w}Bx-ci9ka^H(A$QNu z&$FCY?sFWM)tx^ggEwJE&RLaz@QXr^v;X|7OM%E6VUN;G zDtDt>T6w67FI<(*ksA0dWRv0WnTVFQT+=3cwX#6|+O+t}TCq7@aXTm@fUJ@p9&%F0 zP6FM+@eIy7?fX|i2wua`a%&3+(>;STKflK*Gy|ehgq<{RKXXZAf^~@Fcs-XtK?Sf8 z>mN@ah?w)&2H!=~oDOZM?-#p7s>=Zy#wF>iI?7}3P={K%eTD~mjd38m1Q z%gPUK=Hf4L2MX&&pw5s$RFJz&8#-sa+A|S&-(uxR=+)t>dy?fTBNAXvM8n6hxUK2K0Y|-=q+x`>%-!*F5R-`u(-%ts}MB~ z2vdObA{o)4#$)@4wp7v!uBv z?Q*&iozDOzYFKiGs$+HWk>m5@{hmK3TUv9Q4HbUmKX31SRvkrU4uOo2*l1=Z!Bw&K zz>goGiWR!g+>NUw(7}Tn0c9d3(j{0ADaZ<_#^yPF*I52aIQGXLLyZ%IP?Qah#=R5y zfC-BQ(u@Qr+ghlnz-k@8mk5TmpfsVD!6SdZ z#Kqm+dE~(ONT8@pDdXN-+RWbVCXW}NE^9+&mj4KnzF)w+9?b91PL^@M^(9_wMGait zmmpgt1SP+iLl62@0WUI7yZOiOM!m$>3Ykum%zqDBe4r4Ly7m!kZeeVQP6^@?Y@Vb| z3uT-MIshHXKXVy5co4p$GW|ffWrklR78|}e{=;ht&uv|4X!P|$bH1R^=kbA>)BMSiDkzvMtU}y|?S{R> zu|Mj0aH47o$yy2^2p*kXA;1If$ib6v&G@6}V{g^)mH-BU(Fz`P^dp?{wfpr)=yv_( zF=&>;IHCDp|F@k10t{S>1p|^Q^fbkyFKY*(Y+NpDRx6hMF8)I~YLRUNLw|HkU9fuD z$>@qiPl32C*F3$g^Aq*KVh{y~#PO*wKTwe_+~MNFue9!O&&0kI-d_0Q;Xe34cB&WK z#SKsYY;MFxOuS?2_PL{@V;wLIh#Y3bA*GKG4vrK-Zn_Wt%Vl5$?EoIIn7Lo>sY&oB zFuL=5GA#@lF;jw`7{sw*EcdlZNP$euw6uM4$vim?B8$#DOPd-zqw zNim?~*YDRCQW3}KwxLE3!^e?sv8J%c{O=n!j=LGey=f)R^DK@6q$n?FPWqOhY_iI` z`g^g76Tfh@F88D>Uc-abN~&6OpfXY-##iFgM?*ZVjTi4`0`K9N$EQg4a^dyPa#on` z#P{;-ehYI79?|U-<?-aCA9gDmy_qO; z`_rkWRe#H0=k2>4Yk#$Rk&80&xRv;v{&||Hc)&c2j=8VKrW&fy5&9z#mY8OLCD%o? zNBE1)z>R8f)bzXYIw2n3lj;jP0i#7=UFZbQWias8P>DhZ0AdhQKuK8#Y?^QVZH0jW z(hC1f$~kjuWyQ+Wi{nUfomy2y>d_<1^~Hd784s!X4RPjJSF<3$aj)XN!b>kEa6YW1 zbhFG1UWV)UxrPytF_^>cn7uB*P3B8^YG`QqWfL{$zOXbtOyikzqS9e%+&)Uq_%y8?x`w@=Xzn^6+zqZJVRBt|NkAD=$x}pJZ^ijvnsJ1e!4O& z0)OJp_(yADVWJHu>z_~v84CA#)E!@pH$kZ2$2Ol@2L35(E=yRYd)jK7OZGZQD%Y(8 zq@V3(6>+5Sy5)Qo&QD5Px-@QFt{CGsRXSt*Dc{9QWG9zhVe;uh=T`IZ-`H1I->$XW zv~sXwEB=VyBFuT%nfh$WQPUpLgK~a#{b8Y`8-Ww>wL_L2o@!>2Z{Ks_gR6^?N6xU9 zZ6iEOSX%>qXb#dmqS}Wemd?m(C6CpDunQjYX-Ia-tN6UlI#a!pi^h3%MmweG|BgIy zBD1Jz^l=Qtc+2^@`Vc!6)d!`z2juSC(S7?qxX1la25WQkJMHc54}uS7vYUxuxbwx= z-7$(~^RR>CvYf$^-iK(T)vRXBBZ^l!(n)DWg9Vh8pvCHOanQr3WyT@q2rM!o20!0? z^1PL+0^)d65_)=iR9eKSdzQAfBWb&pUrD#soc@ZlxcFWL!FB10n%B66{dBqMF7~gA z;Ly;NS`g4fK|sF)dZX{&SFya+VSWxu;^>A(A~jO2gtOkScWf|2YCVk@{kiv~rWbY8 zS7{ZrAcfJWfE>3iuzukwp0fL!lPlSnngj=Tf0;k)ld=A;9cE)O9 zn+*AF6M($(RQd&}Sy?&8N7nd;?n`Wv3d?2$Lgv&Ht^1_thLS0b|Y%S@46HCX}Z@5QCpNI)F}y4UQ75f8$DsETwx&CoQiclV;Qd6xsqgpVsEk0Ii+zMV=S?Cj zB`v+71Y^@1&NIO+6&I3#+$Ej1(9_7c>oxm)+ zOJQqWEM#4krU4BWJA#3GUPoaZkfsPDQhMJitNvLI!zG5I7nr>k zWM0qG6{Fl30115D819m-wNNN%mrc6uDbWn^pf(A5@i{)>QwS+txJXEm<(Hx;Znkx`WbAVIX~r_ z&|*rtEP`TcWN-dBkNi+VzI(qYF8nZH-orGie3X zjW~nZtVu4vE1kT%%^>NRM#Hf)RapQk)|3Fb)IvS{4WJ!3(`CR_f`0*A$jEx~$jdBV zkG(R1FD>%@ybLN_d6Qc;%t3?{?fXj|&Wo+sH_xlI#{aHx#4WYMc%5y?ZO=${wX5mc z=XCW$jd(l)0$r_bLCyt4ylH?Y|g}2vwyh}u9PeVxNHI*v zPZ4O%pHjwNN#X8Y?eHwT1`(fVTFfA{KRRD4)pEYkbN}dwR_Ce}mtqkFDT%jzyhDzp z^xni={Y~KOcfkTk;55Ihvt$UY0Z^=&l-7OU-YYzqC=;FAH&3Enf7HYJuD~T<^OG5I%zMww>qhxTae^5d+d&I&om$L^$fuDV)L#0=5Z=i zw@x+de4WD}g&&}paRh@Nicie;U(pMV9+ z1lTU)iI~@rD(=-{u-p~8g?LgE+l zJn4>z+JeA6l_=cV`}cU&4p(@8Gk(yHtn@aBKyOKU9lAgwcSKI$Y_OWWoUE*>i@MEx zMSm1onM@Os=6qJZ>lcs#ZDn#VP=)Tg7#a*)eBybc*Hejn(|Fz&vT|S}To_@=L_wuB zM4B@FiSFE$RP}h9>`8Yam&_1s@k!%;n`GUOSM*ia?$jK*M<<#&MUDfx3>gQ%`UWox z;9lxV`vObJUlNX^MOFbu_dT#GJ|X7_DlFuoyuJje3`7xt-Stc%sbeL>9NS%*CYH{c zw+`MQTx|}v-dxPP+#w)m@$XNS%qe^_K{agiER5hTS}lg<6Tl47)y{hrYf8?2jHl81 z6cZy+=F9=KI-F{~I_tBH^`z{I2lKvCU(6~_)fFu$3CP(ZxZxfqc-Lq2rh(Q!9xYB7 zIwt>=V*!veF8*|-#E{dd&t6It64BT+b?B%{TM$?cxwO1@m$Kd^YID2BBVL1^`Xy5x zX%bGdrg6YOkbH>v=v8+n>3!nDzWoou;iJk9qg&5uGe4Urno4xVA&biP>6?%gCO)<2 zLTh0P;sQ6YfXyh1&fEgR7SxT$gl|CKO67G_pB9IcK$!{`3dQfmSjDba$<}9ZSC-}@ zK?`^H#S4>Qa%PFloeBAs#H8l#-Z|@{wWmKli$KnP3$U%I0YCT8LS)7%YmL8_WaMmnY&CAVf=FMeJ6z!)drf-A~Nu2c}JkvAu zo{?6)IBVI|Vg(OOLX9rHN<2GwG_GTv=RWn9JVqTH-@+N9ojtF<`qDtJh8P~;HiUPUp1fpSi_o~S{9w1 za3r#pB6wLi0zQ9$4f+7>zBexb_n{lG(uO@FCMNy~7(=j!rB%S!hAH`Z4XRfwNoKxr zN1PlSeHz-}YLD0n}u9; zz09;}`eJhxlx7Lc7cSSMInXZj63XueaZrH~k5zgWQO9Q4Pe<}OPT3vLhzWiiwlB-Y z8T{l{YDCd=()?zTOybk&FF!TsYL0gy>15U}gA}hatDujITNq0-><7vUgbPk)ewMTG z>a}NJtZfy1o!{YW%PG&l<2TQ00dPKAyy~K4vqT1FvZ&j>s!YPH<^T-PkMmlSe9_s% z(V3qiIT@x@3gJV!u;w(h2;9psdriHk@=E3Go{xM$*43kuI*IHM-^)9ka^x%A+G>@O z&8=DjGC6B~JDc@+(RLs#Y4`Af-DhzkCl8_ ztTj^~{J3@2d_#<@1$EPeA+YJ#$*wO?*d-jMgWrwhKW(Gc^C=^{DW@WL>De!s)gI#h zI~w}J!XhAFA#CEK#9`EY6&iktbn*+Z;G?k2cj4&<<=9Q!p%-^zEmO*wl(Y9duT{4oxJ-PYH#q2fTvw1;t4V%`X18P((e9h@ z&Vz~6gZ;r)lau*lC5Dsd|JC2Go}21L2pF>-c>Q`-cRVGuZ;>HQ^m0MxCPoE&U^g!1 zwg1#ikZ^Y!A)Bpzjg3BoDO0&=BZhP0oZ}HM^)q8hjo{t&p`1L8KydQY2sqKHV($WX z=YI8eRC}V39sq|G7Z+z?x*`=56Jt);5jlCDJhNB0|5YafX;8B( z&$nwq9iI|Y20Ny{??l=qR&b}!nK&WaD>MwQBsI=owGj;sXxdTV7JcJ4*Tx`qe*XzW zbB=Z((M9$!C)F)&9ZjctE_8|uYd+B+j{=2^@)2l{#Py>C0EiW!;D0E^c*CXjGYGJh zqeKP#XAY(K3D#L`9wis|NsjUuvpa*1UREY?q_o zawsk?9!+Kv`ZyMSNxsCDru# zo6cvs@gH4F1k1Z%i1}88++$z%#46gWswl#fUlCo`*}wV6ep1N&mOi}V&Rsq;*>kbm zliyo`GWRU`!4bRD1k^luuLr6pn8M(y&=L6dO@N%{kK>@(nc}E;jX{UP;?*YzrmkO3eK9vU&gIRIN@O!oQoCEoeN*D8yu{#-@{{VG4OVxI(5{~)d~zndIxKqK zE>I_C3=;N*G*jat=VB=lquyBvA()BDEV0=Y0)ipaN|jL2OWc`#0(2^JeRDz|E~!0x z%@jNv$^bj;%a<1)&o`hZK7Y8>_s#qfGAC^jmla?-hQbTIEG8a$^?8`VhnV2sKy>QV z81Bb)Jf-Frso&^vVMr4ja>kRwTgqF~k$_c8vu*F)BMb8y-(Zvs8Lym#zf%|}40!9! z@usDO!kl;uFFdX_4T=c#MVccAXxJ| zHG~v$vzi}{*?o}W7ftq(w|FYd+*ZR^b9L*O9CmoV7>>r!9`gR)m@Cltb29kuam1*E zQ2Iz#34QGUw#%pey&DM(`OD?HFLljM4_R=!uo>@9yBn8(+b>f1;^&9Zt2Omk-kcfp zM7?WnG>gm>U7xbrm}-8lc&R#FdQ_*26MxH>@EKRd(qNElwfiY}0VSo|&9Mk1Ako@U zAiWG#1T&1T;8KK=@+iub<|$$g){}2ZfRnGQNP2lr&MFyW>U{b>my_+}{h55#dK2g;_yM-|^^)osFaEQ|F@j${ML^=Pr4=1?G5B z+6l9>CI_>AV9w5Di=X^0)txMjEv0SNn z?SY=#d%vTbk+~P#^_#C`TA^UbCI-X+9d|UMv+mm;c933<=YaOesq3~n85;ygnSi`A zl$;su6rJA&Jb1otRfI+mpJFQ8B=b^@;1^tr(|Y5ZnPuh zr$F&Gs&1V2^c`8IaDk#fraG*M$ILD7?1DhkJM+L_m8!p94QwVNTG!sX zEF!9!VfVq`jSHi!_P$&CsEo4o*P4p`^j1ih)m5vf?HOTNo>hC|EJOBjI zl>Fi3T4A@{nrCu$H+eEv`$?-q?LrJnsFCOg1j9B`Rx4_oihzo1l6R8#>gk(J3t*26sPwZ8 zvB9#E>uo;iD=v^i3|aAi(E$2ysHMCl2=!eS=W=&bv2&MV&1jXH4R?LH)ruL2%DDLes__x4SDQ2 zsmYGRXyT4fLBzyuPbHF0=UmCL~1bnUHgxM8W&r2>eDzqt|QDCn2Us9qIV?N|W zz|>=&+E?CY$N%X3PMSphZ-vp0)zI!fwbumq2&QbHi)gDLEFB|lVvs-HA8&)RFsXTv zZ>(GrPP$nOV(z*B*k{b?9T~6l67(3Vig=2Zr^?$TU5ekxr3mb-OzJ`Ametv<3 zNXPcO+$mvKxZ5v$otuVJ9n1eq?&uqz)87h&X!V+I_CiBB8|jfqXUorQpU}8=y(M`& zo6qK(L4tqXcY#uWg_$WRaRJ@+HnuUW?CO-%nfMQq3lQ=_8Y^z9vvKNt&L%tTDh^eKtt9?BGg&Qklh+(`X)#IpTYnad&7QaN7 zav!*QGGJQ(<{a)1RcwwntpHA!n>NzYI`s{Lu!|Frqa(C7d;IG!zjiD4_(WD`EtEvvWfPpC&c-RkON~hj zmP@y~o@8-K;l!ikCepk)%$_d4vJkzG4x3Z-dpp?Nf4Uuwe7ILR@1mJHiUl^1TntiG zW8tzrPy3GB(w4{K0N<($>YOmceCnAWOL*UxeYd@2P5W1YaEtTeW0kfgb28)}Nrw_w zAAkw8@jb%j8>7Hgk&HYUVkl1vx%e`vCa-q?80n4{sb0(RScC9hd}F$uoJ=03EW*&# zMqxK5vR+gW7qIN*{yrkyS^2l_p6;Tkq`YVu9_a4<)ZQ|^2=2y-a0CW)fJjQagXZlv zjP(x<`}<&vIS$|cK%Z8Vw(!A;*-bybOk`2td()PF_Udn~&1^R7{eHuU>}+bgPc~xY zj9^zKcIN)eo-?4AoqsE`ds$BC=+z*Q(tN~u>|F_~RMqt>qU}0upH0jYv@2_G%ICUGu8DhcjUQ) zR444sRF&8o%~$o0*jr?i4`@6}ZYglJr(xd0um5;#zB4huhpS8Qz)V*+9jr}-b&~eI z6u)u83E6|YzT46)6&21m=S>xOJjCbL>NOCTQ17x1WwU72m86!=jHYwT+AYDaiF#~H zg+rsSY(oYM(C=N6gT|d&0w(4?*s1Ac+Gw)O0(D-?cSJh*h5jIHfAdT$3!E{e%`=+k zUxSAab58g6OzR9jUH)RX3Ek+Gyem8H z>UC02LD#afV3I+AYpevndo3Gk{Ql?5<$~L_xykJ&Q6%*}WW+U}A6#y-l*|xL2Trta z$$xVd9|JNZCgaC>tBWJeBky0KV)v&D*L(OUlSWRz%mP=(%H$b1Ib?W5NP(_K#X!Tfk!oc;WH;igM?rba z6KVc3-LL-EY;sLkG$Xyos!vGeYidL@4WM<9+MEjiashU}4SKp%uT#;SZp3Z&YqyDapb5Um9q#0siF96l9kqfV z)$R_=pOUR`(+mO*>E%$N`BUp=UXzULFC8j*mQT!CvAnM30L2~4f>lK! z>I$RDn`y9ClQ`IG_NrJ8E?JBr{Jvw@S@y8b(cteAb*ZAR3z#XVt*O}=fJ^x1@1H*+ zh#x6nGHI^_=44Op!_Q0GeNd)m*Kj9T4v0NpOoUEn+bvjx$2@cgNjTFT;K2JH80gzp zDl9t#l`~y`NZoSG6xfD~vD%rg{@q*L_^LuxSHF-;spWQLH(DWnX0L1cb=*U!dfBm* z_NRw`ru0Wf_O1h01F8Aq^npx#Lh>(Viyxa~!|sP!w;lXsv4#S41%jWaYku%$iOiD< z_JDdB<%8HDp{u!d?s{~|AR^J5|L5Aa(`K7?o5r4m@!DHjnhr@p-4FZZdp@=?MgFWP zO#G;q{Y353b4?K9ML09zG4Prj#JY@~)6A3YK(}PQj;4$kn7o$i1>}xZcUxRb=C>tiC*&}UkoA31x82zg&VLOd zU-Os1*j(hBo3=t?Q0?NzZtwog=T8h$nL_uUrTK@46QP)xwC3{kh%PVft~+SgbOy~r zRDN0M(EOr5G4mE-N_Uqwg<+IsC$q!&8))d#p|+d$+6%4n1oEx*N0NKh3V@=X_g^#w`ko|uv}_RpF3x%%Gmm@mNS zhB>3~7JVX4A9A(+Hxp;zIpu07OF>Pn&eljV!UlVwU{Zp%xtDgSS%wFHX6%6Xx4<8v zhaq!4#XAfZcjB-;X}ve0Ye~uNCu1!Mac@myuk}J7KPa?1WO6ky?$@1DlIQ0qJDK+u znw^^i(Np^+XbqH(DkEHMXX`}VcBbFLt;cHokK)r_2WNCUbViq5g>x@# z_%l|n-Q||)ZG4&YjsaZi(zF+r4+pJBC1mGZ1RA1WN5pZeng3pFd++9UQ!Q~#F}i~$ zJ?P=q=38>5^ksac(S$QgM?ru0Sm{@@dFIEI1-ew% z$w7>J8IRVKQ!!r1F5vR8B7YaC=9;cOG}O8ycr@JcN1D$2WRMZFmuX1CLVwsRM?+;~>+*B{A`VD|}*_QF8fmP76}{pQH1 z)=h>I)Aj4*b&wsK2O)9&{j7)iy@VWSH6i&sWjah@?z{B_wGvK|Rp0Hx(rJNJa)IBl zfee85`5l{%=$=xZ9bbl3SOd7IGewWY^_!AedM>$Rxg62l_4Igc95?*?PYvW|)>I1C zz*M47RDb)5C|=@ttYk@n6gXThBM@%t7rH3=CIkN}arcu{mb+Zi$emW$kj{XFo9dAK zaVgMGHXZ#a-a?6!6P1)YgCNR^*mqW zt=q}Fb-ImNsDPzXt8X`EfXEWS2>Ul$$2%{RJQjg1X~^?j@FR;DEy8z_QvTyk5Ks%{R4IzfY9ywTL%Q1@bKAxDgQ=9MgAbRZ_oh|u*mEri zcsS(q^4>mq65bK^L#qiqql>a)P1lN=Miu0wb1R*sEp4KD!4S+pr=yrK{g3$<{rg^9 z5vMe*{F6OOp6F~{*^w2qY`iMt_Yu=(bw+TVEF5`GEFLtucFW5S{pB?21N30!Gb~+# z5@pm=0?b~qi$<>iRRzKZ%Pt6OO%yaPo4J7PA}%!K`{G&3hilh)C#@H{N*OyLkKDGy zLHGZ!h?SL9zs57bp7eEM^WGNm!R*w6PXD9n8VA0t3xLBEEqfc-?mPDV70_L<+?4x4 zkL31s9ZMBs4P9;X8K0>GM!>o+xoCdvCF(`=f0IFwAG2_w!`6Ocdlhur$0lK6e#sFt zrW@gI3vS)h;>CHVCE#U|I#If|5%l%i!WL^d^p!g6>gaL4U)q#O>F9fbdoM)>Pg;oS zE^N^3v@qk=+(eq|;UTKR@#vw?aT_0pNs8PVOEbGHe#iLfVg`AD!8Y9|xvV0p_7)-j zSHXf!N2=^T8vG$LUD9$po+^3a&w3Liya6hQud2$b%bVmQr%;CQ#|3#oW8%FRe^3j{ ztU?>u3!zUnrZkXVU3m1wewP!6XlD1V}Px|`x z4zR#6OW3yp)NKR+di{1hy8Qh4_F1#+#ePT61DX-cb)P8WcTFOkVt1mqmvn3V2Q_CP z-MP5N=91-ww385%pBRyLBZ~Z-f{8@ljD074c+_|xbg&>ZRoDUt=vl1by3!B`6R*9{|5&i`xoydH_A9tmj6BRNu_J|9B$L<6!3Q`c$zb0r@qdU;j+RLF(nwkMR*3 zThZz}YU~C_du0#J8N(sQmw-$dK(V8f-RC{IRiha0zE(R;OBL-Ai@BLRSeXESm?Ck?YJ-_HZ}k@&;I4=ji1$HkJS-L#hGw!z%k{CR=59J5)9C&jm4`6L}XCnx)omn+hd3T(#VUITb?$!t9tydnt=uLe+!;J2b z*2lU9{@DPc&2^=cm z^2D7maUp-9t4&5Rp>#dFvYHyMe_rxXH2UgS-JWN7JeJ5A4>qQ8ju7Y??IngP$^ zv@8%27obtwMU80L&?)$AUGXbD;Hkj|TVAlYI=;@cJ&zQs!6>Y7b@33zN~j$E5fBIP z{_eS)`tE_)S&s(0ex_0?lWVv}I~|ww^40RFWO|EH*?0K4CC&4m3PJwDA_x#Q3z;>u z05g?X%fqGhb)Hbc4!*8^%viBLC(pXakz38;3f4s7PlUu{p~n3{nuAJOy3!EeojYP1 z!p)~ngjZ#kPi9ZN2*qIkz>u0P4=JyyvOk$aqg+6afja9%=K%3$4k#C{j}<3*esY+s zSoOQcWK>y-ns<)eh>I}gH=CMWRWbCwwN(aF*<0kdY{^{9B>KxS3%j|5;b|KRZeWSIbkV7EpYK1{1Kf z>c-0R1g;HhT!)pN$0*0n@VDD&a8+s3G_RV9UR+x|eLDD39>h*lgxA9wLMp&ZRSsr_M6t`8Mj$Kd z;5r^WQc?!Zgo#<_2gC^Q!ve@<@ocL*F5t@VSZ_!WX$%0 zw^C|VQHXPn`%i1*hDTvwi7A*NWEp-}^$e03K8H`yyK^&*of7;H|E_U&H7UBe)jXCk z^g4TL_-*5)9*Tg2_)IZEy%$JleTi56z9-?%@6{4fG@__WUm*~Cm>5MH%`YWMZgsaS zY|~yQtULzdrJK%8)JezQ&O)23{)IL{eD}wxG0AMLo_QjzpL0p|4L1qy^Up~^uTZKh zUr3pk{PaztDH>u__qui&8oJq0uRAK9Yk|@OS1dd*^)4DcTe$$iY*lJ4=w)~mjcbcE zoBWWO;(hKw&9!yUnpT}bR(U_?>t}f|Syvp30lT%v`=|ENw78TZQZ4r-&E!Soya(u3 z6rHoD!pmxY@U-$?@uJ@uloLi`4o+p+nvN6NcyNwKCzdec%C4lBhn)Xjm^j@GUpwU1 zO2MP2Pa^qobSf!#&OZ))YHgkW7_Xd&mT(-n27%`d1phk$xP#9|Ookc@aN$=j_`41> z^tEy)BIgvW0foPbP(~n2TK)JR+bS=fNJU2XEJf1EG;eTk{0FmO4r_| zMv;=5h{BhU5a}4*DIy)CyJIl6-7|mpzV1J;ozFSXdCn8>=l%Zu=j}LWw?XWo;veH} z7pvYYUXG?<5i_G5NEv{#X6{%)* zZainjeOBgdNcQ>Kxe%PX$PwhE{i}RF%h{3nB%hn;^5!TeN8HK(>R;csCO%%h7W9O@ z^T5Dkphh3#Nnw4n9%11hwthG#Y_b3QX?&LF%0YRGveI^6>#Ewz2VSs9nm=;ala6b+ z{{sB3BQQz_hladI8hB;9*;dd+7kUW&DDb_~pO z865t(oQ<^abLL%1S3aZtyL+JW$bz`1NBbjOmeIdTq&b>yhNl;!>nZ8{giz8fYo|@g z^w~LMKJX3LUEkavYkI_5?Lt9be9#KIK0;p6pi>T}3$F%*tx7zDgK_2k4T{3E>ow}* zj2SUC+FS0{jWDyUDK=<a#Daqc*ubH)~Hqx1u?ao%nVDl%0hh`qM)~b6!y{l_zJ4ezpeuVxUXfAawr$FPN zaI9hdHccIJJvV*tkAZ&oT)K?2bcqT53JU5qVRLOG<=gk~k?7r#ym$bg1xX0PQQ>M# z-R#*F8OR8;cf;s)I;v0M&RkDH91e@6@nHf>Vt z=hu58R{B6Q#2}M(ZtnL_KH~-E_sk7r^Y8V;=X0E4&Bz>0o}oMDsyX@F1oUxeK$I^9 z1qD4;*YK+mE+<~Or6ZR8E~T$deM5t5cYBKES=921Ty6nYXsF7nl^l#HfSexE-F5dt z`bX2@rA40V{m)+2C+l8lw;`b=7DMH%-){<{-9?V#nbj9#JB;Q}R+Tk#l{zg=6lZHr zUrZmUv48vfxF8ywbC2MiI~^-OOkGbKP8WkTV#U@ldHbvN=4(Geg;aq|n_by&`Li<; zBBU32?h-S{@y@rdrH!a502`}aukS|F#sNq<4w#OQ+9+;k@35rUoy-bo2`ZwpmT!?h&a zM%Y-fIe*NYU`E}`PTGKjZUIVa>Q7@ul{g&5?jjg!QcB)A7#)M6fWCPpCX)ctSzyfM z8Ggcvvxux}T#qc|#$QF?#Tgn=MWczr!ynl6^jvq8&Ot-`aPLa>!t5A*=F2)FW4ko& ztlBdM$-r!-dk$Lqgj&sY5)u;E#aNk#g7EU>?LBp5Va zc|+WllR6>tsu&Y>M9q2b29|zdJ5L~U)@P*~fVqt}Z}!x3eOw>+>H5X3K~yRT)jHUJ z?gP*l|G_-`ihZ_n`cW8~kt!xkyUnsSS;QlmioNnnV^$Gxt>0GSG|$LKh~K1Iir?2e z21U5s-0mLHjv>QYpCv%+bo40;z0x@+=6xb`lmBypwu5BFXNha)AO~v&b14GWSP&s4 z-npQs&*e=r$a|C&b?$$%^>OSbkNAT&NR{18t<-G5p&t6|>~l&MzzrX>YdkD_zyROG zq5av4;RV7wzKaTq!*=&(xf-v<%~tZIN~M+S33m;Pwtv{QV97qhbn(kRc+PRVErxh5 z&Fw3k%Qnk|-~r`K>gxmd&e@%NY!j7EMUiNdX|FlHHxLhE4X4Y^yj5Cut+o+ZxXk%O z5ine#kXVh_WxG}44Js?L%IbdhkFu9mM+BS3Lbcm=ujnOdV;x3eZ8yG8_d53`3LB;C~v>M7Cm8kP-up9nE${mEPhP?VEU|vPd!JD{ZEv2Kyw{4 z>MveOC5Ku>y;pk9av{doxcs%$Yke#o;+G#6%k6y+l%aPs>VA=L^hmhXvtKzlD?hXD2-?Qkoy_tdUaVY6bhkC1?`udACi<2<|a*!Z~bG$6tdZA84cvY1(n z@Byoz^RsQBaV@G;rF9u^_?Jtxpu_T7rZ|A{eoLkdd>Yq8`e^NMzEYO|`9n#Ezcl(vQzTID9a5u`@|VX8_s=FB zO0Mm<+ZQAeitf4(DUX}Hw=G(iQ!Ii23nLX=*CKrXkyYv%YY=$-#4OJq+*SBmN0de1p4T7CS==L*~4c23Ov<7Z}2rT=6SlP+}E~s*T z3lZWS=G+{(jek&hg0vn2)oUX}|A8Cla8AAw*sgp%u>knr${QZ(n{f|KgNM9x`QG|2 zgB-7nQ$s)Pea4p1VH((t@61%##YCoS3of*(VR|IrzE$5ZK&&if;pW945bl!3^8!WV z62Kf$^rE5%q$PyQO#3}m!n>F?rNlArF>JJcr0-f9aL?faMOLLAPGo0$&> z1Eg!kE5}f;tBcZS(WTHrHKoFBeIpe(&~AUe9p9pT>=&mK@oY1%%Ql2aKm-aGe6EH& zR~5H!cw<)s=0syjix#*)+FzbHjqPP3B|6+2aob+qD3>onFzo44{72)55jhIrw9O-=g&wu+}~rsW>GMrQ8ESJ0Ka9ETYOwrfM#Z{`}66qS@*_uE+< z!0(8&K7x3RyG#cx9`mAitFgEGUMOdNCpxl_!h=)eU(j;^WVpWbs+aZ@QLi`)IadL1lOLY4{#?$m_@)lLf@+MtJkC!AlD0APYC(7uYhWTOF zzvqeB+Q689m!IaisrB26-)F`Z-l0Hf5HnNtN#@Z5JU=)LdvA{|_QhHULpJ*0P5Ha*>+)3b0L-$4i_VQ(Le zxJQ3XG#9i+-lNxdR(9VjWH>i_+p|6>H8`9h3{s{)&uK%xXZQc-R9(`_S`YLr*G8(i zy=rB3F1f|7m=PZA;wHcDoafbxC(q4Y77d(SIyW<2=DbJ5Ue-yc*$-sM^BmdIqMx=6 z*FoZB2IYKZ$LC4mEg@6~lT%ZsA&m_j*t5gc=u>>M>Dy)?+1=+3o?z7nR>Xd%>%p7F zNaB=?G0N?!hG?leY}u^BCHRlycA$m~NT{MANke$M6|N(T03vKMS7HF{vj6in?yAp0 z%4~OMlE%ShC}Kb#?HH_$v?6C9#QrJ<^9-psvoU7gGYEZ_bsc7 zqH$;H%E_ElggMcjU=>6@(Z~E6FX{`sM}V6}s^asnR(00WK1Mp6B!zy!fk5N#^fl_0 zL3KaBTHh;RGM#WoQY_1sJL^h1RH1Qg+ro`a$AiCGtAcIw`TfMc%SmiB3~|f3ee+p0 zcxv$N+R#9zf-5SN#d>GD=3p{w>_PaBLjfbVvuOR$*U*4s1PsS7WYI`e=O5H}jP_P}0dfUslmVBhs*3vA@X;rKPJSA{e)Sh_gC&ORm<~h^eGY&% z0iaLQYSrr5;v1wqB)Bl{oLpyi%x`^F$ZnuZtYws znV?PGV0|Z^9xU#WloEkip9&Hmy{+YTB;ZKJGzbaBs~yXzD^s5z8~YIP^#>u8&IsfQ zjTba$h_97NLGaB${%QD1oF)C~RAQI+(Zi-&gRMW)GoY15^_W!_ZPo z;>*<3Ps|+aTvEYKro9u+&wZ4htNe+j+tw_XTBO72R4GWmj;6KgS{CaMGI`Y#6GOQ3 zn<$^LlsZiSn3G-Ep<&#`?~PjgPID%(<~2dyx)a@!OKSu0W#L#dT`J2moR+Z8G7nH&pD2_&l@eE z#s7SzA9GuLpx)naWkM72HLy~L^41+IXUYKP-SjF6CIFb@4}t&jgG3Uc{eM&yNr0+O z@?BRX&XPvjzw_ZuB0i#Nm8WeI`l2v14dp1Ec&+2Ws6d?Q=Lp+z<*~vdC&bDhg0~-= zcXx_5Qk>=ZNp%{i%&{9VaUDQ4g2OwBO>&2y2>`3zApKNHhC)GE7K%@67taNz=1;~N z!xt^M30*b&fykjb$v`nK|1~Sc;d;3=BXOnUrB83%Ha%^Hj-sQYhQLHQ)tqe?46u0( z|1g!=M^6ztnsTTJhWjXYlp>DQ40g6Gi*JeEN6{(mr!cIj-i(?;<*z)@UQ7 zV@`_}a*#~UV9bX2&l`2NChA-Ht=5e=lz7V%B2jM+;vVYKnThNDiA{^P_+ziRmw$3^ z90AgUEFb34ocUSKjF{N)g>OBLDGZxJEuFA4rcJZt4@sigClhDZDx zoQ8hacE(;TJk3wjUwq*5H9v`~_2f2gebiR)x|O(cluxm3i5P-u*I+ZGApe({|IPq8 z;RNqVGCi7P3t5>-UdYxzUpU-_w||Rdyn5}LjfHBCTphvC>9&sa+X$)zhg=Q>#zn=0 zL&xdA2b$AI&)5SlHXno1(6^2NSF~G(#3XJpgev{w$y0SSj+;bxcv4x12BVbG2WNk zF9Wro2}YjKqg1*&39;*O%KJ}U5pr0X;W<>wjk6QtjKnp4$?&=IJZ05mJ0e+Y(DqUh z>ub*Yp#OYxBFGBkINbSy_~Z*Wpb$6a6r~nl_mvkor6Fu6RI9766G2To-^^``>ejDC z<;~%mK=k?{Dc@Af?q8(0!3f&R_pT2S&7+pXSNrm>jGla7&@ltn83lkj5MR~9-s z!zyzkgs6~l;&Z;S{CCLgTNtOv;~@NkPDxYmw*1TKLr!X|}Mb*#C7a>mPZ=QACGuj=?AjGB{j5NaLrFS_7Ps~CoAepDVTy4~#yMEQ0z*p&lesjf=c@iZi_6_k_kW}b zQWYP6O6nJ7DKMNu_CENa6;3!Khtq?^dil)`PE0}(R%ZjmkuFDiP1aSE>)!`e0(l<3 zE@?~?M4yJ{J#j<|0n#&0f+4;(^R7d|@pWW+ZyT9NDK3JB3r`o2KeAJ6g()Qx zU4Q@vZfBlJ$Up^F=`ILf$D~&>3AreMkpA}t zUSYKC_B~TZ2AeY6?FpWD!G@DaREApu(fSDQDF2mxQI9K~fHnBDX4!zWZ@>vla&olG z9PbbG@9*k5Z=YOWkT@)pk-S7PS|8F!^6I~4S$79)(zxv!q;RL)oO^`q4aa{pE9Zx) zp@ytvk|%_FfUwGotDvynbAwY%pRPX+K#!KN?w`)>?e9yMWDjQmAq59E-tXVPkFVF0 zX4ajwVVwhlF^wFdaL#}ubRqHo+j_6d)Y0<`3qdJRd!8Wtwm?bi^k?|X{j4`Y_lO^R zU^975=ehbNi*fFeS9{=`u3EG{T=SHxq13h2 z59>ReagVU>dmrghta2MrMc@{zw=;LpZ|fPYC63&G@jL~CLcS)q0U_mfSPF-do0{8+ z{kOehn9zX^*haaL^&4lw5*Kb(|%+?63?9z1HLfmvb(0*)?5i%iRgh%QTr&dc{ zEzeLbwTCm&s(nL!V8Bjc8N}g6sn0fS>y-V@CN$z_{ZC7u;l63H~9;hLe55z+zvQ{L*|@xlN+Te;ekr zS<|`iw0qrtJM97C(_W5ljJ|O*GS~+|iuQCD3poyUC6#%m54WrlHRw;7B;v1gARW$B zO+aIfe~yyZn?y!y8HmCZ8dS%cZQO zQieciOp^i>7sM5K{xUL{5ugE&CDaV~ z!7)jG_}gKT6~1@K9q6hVJ-7!)Txr(LBT7u(w0)zB%AgAL+t=ysNgNPWn>!iI9=@XA zEOT2w@J=s(XIrY^R?f5$YBHJ)^_k8;J2wbk0o;IkkZP;TsnTiGxHD`_eO$aoSX%Z+ z=a%`C@wKFPfT>%(OVo`z1$q7FZa&$CGXD`6qp$`7d;92eAB4_*UN4GrQ8_Oir*0LO~g2#F}5TJNK%UJf{*o8pDcufvBwg-2(`Qw)R@W6 zo0CPFY0SxB^-1L4H|IER)^a~2$Ok(Jr5wr5?k=rX_@`VHDjBhuI|fs$%8%>nx~u+^ ztz7-q>UhlfpFea1ABRL++hCloyI8vUVNLF4y)lb84HkE-Lg}U1V!aa!F5+$hAh`TPR9j%XO-l9!7JlxKE^+ER7#1tej zg=vd^t4yfx(rb-=`0G4-+q4OWTmJ+hAGMT1)x5o03an zEudf?V8htwep$>JN+`$Y_r7}dI@@Shx%w>GbQ$A5ViW5!h&@HMA}GH0{t;%a{7VNI zP%S%7Ea_KauAj9NS9$sK9!KJy-CmFh^uOwLiJwPaN&n!8r8-RK0VM3)pvm2Q1zktZ zA3gl8{bKFRmWk-_L@)5Hd>N`iT*?BO+7jP|lxX3BeO3D{|sss?3i}CI|&H;s?pEf$vo@3^qwZ|#P;pt$jC{$6!oV?MrTi@iqJpUTLnm?jx>^7+Y?AHzq%TyJJQsnSk8n!G=G zTQ%xLa3Mw|fEhtZ5DinMmMH5Yv&ySZ7WH(o#q|_9dHsYsUD>PdgvC#Ok42$1OkO$1#GWZE8hj2yHS|+@_@z+m{&ydPH z!N@GklD{uyE3)aJ;UlGuLY0?p!W64VT=|K_pQiX-Mx?+GYjtO(gNK< zvC|UJ$^mF$=|N%@Rn=t)>9^Ce|GI`Ha-3!tjC;Bz6+ZPCw(B31fBVwzz*ddwF}?zQ zpB3m#K7U@3l44^dY}(1hJXg>y?xVIJg8=VU;@S2xGk3l_Yjdb*%09+;+r~o2fcw8o z?dUjb-Lb#b+aFKqKik!@&eZ^pF1x+fBfAGWoJn3fBB~i zq@cRY3KaIfWpXtf&kkRA*^79e!?1DG5uPW;4Kclpc3p4_TOO$!EC50I#KQ30{PnvQ z0lTR*c8i+LyG6IE=3lA(7bvD=2{%qV8XH@-d=&RTV(hN|ektTVar-}l(=M%7BI2u{ z+lr#1B|)7KK|w8$GEbEEy8{4rbouA&y<7St=znJb%&8a{RLQg*5&~ zZU0<6sL+E6wCkB7yn%o-TgWZ?g9~me;1f@r;VGSZt_~&S3Xx`Cp!h@(cpxheayk)N z{vvdbdOmUlhV36hbBVG#3tNg(eJ2~bpnpPm=*BUD<|7BTV?&bD*tl-@rO0=Z?HGr-P$my&F zoY^ERWbHPxdRLxbsZsV?H{N zz>`rKx8n(~_?9g7lb;9C^aX7Sh5Lu$ytk)rPL3Z$9nPpE*vy0!kt8N@t3R{Pin!3N z;2g7wQ&zeF)oE?w`+B2<84w+vPc8u~Ge9L^5tN9cd!g~lUb$9pDsF1-DdGn+;>+rXVJ&U#d!nO~&p@F3<`J#@n^}%} zdhaUss1DJoUMIFEA!PNR6A~C`xD=V_l#awiYUfEL<#!dQ@muL=o3Oj*B}eV0qZY5Z*!sIaWf(@hLvwH(P(L^&=4-2i?}jUhIj7V0 z9KHPY2#k*7>YL{xQjsfhD=t(6Di@8{s!xJ6suxZ=#pkTpZ=K*DCBuKeO!kf50KGvy z)i2-vyUImzvXFT?o3cB;t+(6SIfGLv0pwdN9AQa2A5#I1WGEbT>;P*qqd`Vc>9o>i zTJ0rK*XXMbF1i|xlzuBSDTzsZ$1d4R>m!BGB8w?DXNi07HeJm1mI)HlLPOpYHYrwJ zA9w%>0tLl*f0|Ue$qy*_Z_k7*)1U|Ll?#Mf_&*7*CaEyqI!%+k_Q)65JfEiJ8!n_I!-@juMKA==+n551IH<#UM(mZMbc zkJGsjcNM;OV6Pnd!rcpCxRHva(28p6Sdlt*_C%K(g&N+>!~6YLM^tMSXXyH!Ysv7Z z+L}v=y+!u_wuD!SkI@{Czix!v4KRbPtE2Z0MTcy@lHQUHwaRk%7IF|jvlh=&s3fiy zi3N(Mr33-rsdUiJjrI!HF;!aGDX)d8^;qryEkAEGb-4^GRP9sFfZp-a)r?m!svZlE zjOW#cDpQod?2H3G@xc>V`!%C4P$XpcY*AO)o8cgUcO zB>~7*Ze{$1`@{!2~0aQFpo~q{2U?-7ZLU7B*r4nXf0GdJh0=en>{|BH!UZqf3)YjlVBNA5WP=X}> z(YO<~^J&cL{N%B&`{kDEegH`LAp%I)6|0{%2^dHhX2c`QjzyADzXXLwr zN`9JRmC5`+i`yXi{Q2{mj2rLUfPyPKfvYS+(|-4#-qY!g*E-70J#hyBmG9l?wU6m8 z+pW7i-yc)qGc4=>9BE6|HphtAbjbF-ludd51&Bv)WJ4}hllI5Ul>6H zuOR|Yu22553udwKAucp3kuD)_>*!bj#Hj56X0a?qQBF!hn`q1Lzek|v@BxiX{;IjG z4?qvgV8n+!Y_VY%6XA_)H^P?Lz$4ynP7ukOXCu>)UdN!^K2O~cQqS)XvS%5YDI?Km< zrT0`L$_RnN_VvnCO!X#w4RBoSRD?g@hcG3HY-1#^<>zx?V`6CNv97MJyh}fb>~YdPTTVWsJ&!J~P?bq98v$<6)dK>JZS_ z(h=q{!04V^59a)2%Gq^L>2y5sFX4(igoJ1=H0=5@Es!^$_fW2o^w zLrj9(c5FdBAWyb^(f|a2*nxRT;nr~B{b_tBU&8Sn?9OdKxuoAcj**#}TvJo?m(vvU zOs%KH;OLtTa);?4SMPM?>f?!c*VV$g`NonCtG_2#51;BeeFNZ=m1XBN2%B&gj3i!Z zpMzY_>^;NTws3PP_grhRens1FCySr2!7y;6#!Fu0m#e+ZgX-lNw&;0hKX1m1D$CpC z#V6|V?criD_jkW~a%>3%k@Y9dkrN5X_UC>~hc%(!7rIyF&)O>~bRB!E6lB#^&~$~z zQk}qOs9nJ0eGS*e_ZKF6xFVH=F?q+^4?yIE9`2r;xhz zlxL4%utugm3@V)4F-%fZUEK-*;1V{zfpD|9IjDE=$g7ETE~fQ|sArQ@|I~c6#p1`T zOa}{9gDEh_I!87oZn)LuLFOCo9(TM}{E!gl+(}6P9MWixWG;ixWg6uyV}1;~q7`Y( zT(7D=$5;Z)ZqY1v?DEdX1oy$AiA@n=X1d4L`0(qEPyB>Hg(Z92zSx~onWN^)F71ohDj4g{>)+3FR&K>O$@57H(@}WIHrYYHBwR=Vup?W4q*yqnx1_NJ zJR?ETgU%rutC{h6VERNvqVgrsC69Tr8y#s%`pUXjX=P4=2_k!2lT7(=PsXp~p}4YX zBBxx?USey6R(BFNQ!Ev@0BPm=KE{ING+6fds`Ja<^>ge_xLGn={7v(#zqnj#RS>gY zeh0!R8158_Z*FaERa{jQ1Vtcx9s!}xbHsaJL!6I=l1fuzr$T_NHgZZVXBdl zh2lQvxokfYfl!;qOT@+2tF4*sAS0;1d1FZm)St0&O_^jdv*r6aE{2omca-xF&0ir0 z5*%{|dH{I1Vs_yZa%Ufr1mz93k1M3s>1iL~PX8#{@g8vN_%%5Z=mc{8WCLU=qEm#L zJE8a(%pCSoKP)X2rbEXgEj(BsmMh{`tAg*BnN`E%u>B>vVhCJr`dbxk?Wp>b)6>of zdcL@Zf&PAVfIR$e5Ls1K^=$t1=qQodzpt|j%x9#XtIe6jRi7nsyYxyp4IecC)MM2p zm;JH8QwLH&yd!YttXgg=Y7~UBRy@|e8!&$G&DO{jw?FcD8?Wd@F}x#Hg$Y` zrq3Y2dq^67N@R`af+F8JjCvo%DLA(ljc$s0ISAsW7;a(#-x9^k^;yaym-xgJzoOrO z)1&dVCyaR``Z?)F*DtgE1Dz?Q^3yp};gOp2TLfD!hzD~^Tkdnwvw@h!Bz+ht(p?10 zggJIRt)gT~S+f}S{0JzT1J93|fxt=_;Fhuk*c{$@Ftdic{6hl}EJDRyoaY+NEz09@ zp)3{=xAfA!0J_D4-AwP<+Zb1O_fMWh|7y&s|8)e=;TxIorw`$`|9CG_L#-F?O=*be z^*^N!fyrrj!i>1mGVw2N@SH2Go_4mgNVXUM$6?ayIU#6Pl#Imncd=2Uz}uN4XW}+g zc&K^6X)&I2J-H9V-tgL~8Dtr(xgvaF|Hf0ZX zvSj-f_W0|=H9(V12&d}*1cZfi%K1HuI=?o&3AMj-pg9PE%8>or7R2~CD-aF64#kdN z(uWOm)8T6=A&=%rk*PEwo2YGBzZ}Rs^s?}A9G+RkECTS|^9T#~*Z)y_{ybSRJ~~=2 zyZdMhfS-HSoEyc<&KnzSGyfTbqaW8~?8N)d+LC5f&Q2Y5oGAjss@gfI{x)l143d)J zpAAij_)BLVUx?ooZsU^fMst??EqsWOij~U)JZU zk%i|q5hW!3+lH_7#nlJ*e)-_3`&?){pPs(yo1_l{%x`5A610H-+ZZm!EKy0(`%5Jf zk44!%eFw+|^9Q&y-(Ly0{(g$1MXITN`@M+Z^C>W{rK?B{aG7#T6oM`GqApMQ+kTCwd>5a(36CI0QA+ZbZr;GIiy%{#W z{$^|YtZ-VZKV9_b@1~;l#~4|~Nhf~gGj9s7|K1lBb3VoVd(Wya;Qp(`%)p-;9Y<{*Bhe4HBKc3qe?OqVs>W1$W4w-6y*;QwW~(saSo{hKdYwv-K-C6emy+?y#?Hu~Qa9>#D17(P~*uB4e`9dW5@x=ox&vsi0 zsT&>;YI3-zdIPzB>!#K_<8pXAnd0v~8+2%|3jUy&|IPCfF6B+T&Y@JpEmF?bb(d265L+Idw56^r`0msvzx zvxB(xO?dh*QQ_U>42R2S0@>VG(*XyX>LPbsjj73+gzmBG=vaz~NemV93VP5{zV@?k zcHzWp-2>U!2r`ksMxU&}-%AC@QVcG}&UK)=0XtZ{_@`m-G*R+Fr6461z9gm8f^Qm3 ziwcXIQ35v~3?0;nxR`7Q`##5KySCbv`5apKh`8L(vo+NU%9b;e$uzKda$GuOo9fSp zdQL;5iM>11dpxvM*0HAKNb?$<6^a=a^Wmfpn~OF6m;-?_oZmdL`-PAv5_|F=)*cO4WDJ2vC%Ka%pcx6qrV3E&J=;-=N^L^j(C2HNuw2(;Y z6V^Aok;S>`YGsWTn!)P*S94dIRS;J8wC|rpqOD^1UNnx8ZYyIbN^lCKP!lIrOWm}e z6b6QhSF^cc+cbau{91RnuGsX{a|l(iN9Ib3R^3D!{(uFJ&c%NjLx1kd@6e(H`j771 zf+B6W?X0z<1BnF-9MIVm^IU%3H^m)uO*2S+xPdQWe$M5NE@c2)1BBT{=h-5XE0ebk z_uLL!(E8rkSMYH};H2~5@*AF`{3pe*XKpjhasNK#!KKM!i-ss5gk4&eMW!I*eW4U6 zFe^R%&gziXwXBhik~%Of-((zP)>KGpyI4~~zPKKb>xf3CTlG@OC8X!eb43h6t)w2y zVj(nM_a8!iTtzM_i#z5Zut*kp<>$`XJ=1t4xo}|(&NqYAck$2l0 z!RQuy1?78lWMkV#AJX^*SYEHJ9BSc7!$Or?{|HUfU7F|2iCrr~k`LmoBa$=Wo$X)=4qBXQ!D8IgByxECu_E{(FBkt)31ckFG zokt)(Q1Z=VZhh}HqHy&>lshk3o@#;`m$F{7cNav1GO(bOJ3S%R)NbN;t#MHz7eitl z{GKe=Y72v_@r$ZsGqOsZ1w;WFaYgU`mjtAK4W|3EuU22zwZo&alLZs-Pr0w}w|_g? zougxSaOm^!3R(S2%38G~I4gbhVD7WORn?A=-{5#Ou?R&e*0K8FPHyeL3r3yB>pK~C zh<2KMd-Tz9nJC@HXXBHy&w4K|b+?g9uE!&(54+ra8H>*-aIy&C1!>dz`uf+eJKL7v zdK#)WE(zB4|6-B;vc4F*-%f|LT%Xn1Q>5n!QI7qK_@svUzIT&na=FF7`zzmk(H}u>eA%N&G`lgkb?wxh?Y2B``2>I8UGZXk%38mviG8IY$ynF)sMk@vRv;+HC0Moc0C=HSVUS*>}CVl@{v9g z4|V&ijvdKp&7-f8WL2sQ*hYT9hQx5t@9c`o{oUXsJA(T?YCPz!UZ=I_hZGfErh|9w(L@c1{$H3gvug~jlNid}N6<&MCv zJUMhKriGN@JUG;SZlOC4eV^eEKs2J9Ojo_) zM#SCy==NSYEiAs}^fm(j)aRbcHz!o&XB9Le7WZEu?a)n*e6uNJ_^Gf>6f8AbPl?hPIR0(xL!80dLFXGu|5I<8pQ)jx zYh*Um?xKp1v*CH$c@Mn3zksqL<&`Uj$?rh7hLE1#x^=53?SQcUjb_`$TLFho-t|X6 z6?buU+yOyJ3USseaqKSR9#m5U>Nk#_lLswm_1OY@dqdx$AC#qm)+cZ8%{jG?)i|-^ zG+Z3)xqJWpQq0(E!eaWvE-o9%$arjuv+f>cz2iD*Lej^V@x^?ev8u8sz2~nOptJ=zY%%<-;(Yt zodHR`3YO2Q{A6$yzwMHx9>P0F-Cpg2Q{>2f6xy);%5kEMz2k0Nq}*w2GPkB1xlwX4 z?LTX|C8>BB8UITK>^v$I(UE;Per?VAXQZt}jVjlFL-;3CM}CG~3H&-@uez?*14&)6 zd-oU`cGHBwJiRpF#s=lGGTT8e8{t;5(a6U4#B7j*|4JFTJ^U{BrpUE=j(PfUP>1hM zP&&$~;=^CYLdB#rYj5A#AL01>EKy(G%YrA!`8%I-Y$#0ceW>zb=~Wh%@b>5w%LOv` zBI6o{PvN{cw^keIO)U8lH0Q-yoR);O1dKGeyI#Y-VC`Sr2=EFG9IhQ5GO|+XLnOjs&glPD(nx z^M>q@f12P*MQ&dHi>Qw@XWZS%&EC^0zTrW&bSp z(IZ-bg1%Jv@RL~U1l4ou!2NwP=*2xxfdTT)yZ+*a#DoVNYZk5BJSIK0kK}P5-cRRw zgCB)moEXzn!B?N`z_R>5qB;o z2{uTVb#W0O^rD^>8iuum@pJ zyK!)M+ZGW5i9IqHU@_b&`7v=$u+kH^Z7^xBAlLl%r<9PbaHlOZ`~LS)=`F=hp8pIl z5!TF}K@-tB{=IYm{YMsyvsPUO3_QoY>H(|lDEp;Qv!x?1uwn!&W0*wV;ElD{j!6o$ zOU6%Am5+-G4}sDzb+`mSzq*ORzoT-$k-f87M+NcRyeUtG-=5*r3G}kuO)_?rXB1hP zK2>BETA71-%07O!459a+_7~Me4OOs;Z(H1VypzB`jNq-Pe<7`Tm$0aroBRSVj|&V- zD>EoRbq}CB`bfD`G1l$G#a+;A1Q$}P>Zr}!SUDR;*C&no(JV2HJB06TjA6^HdTE)M zm%xckf`|esPen_s0jPA!94CGzd<^Vqa0c?;9WmGWk}LiNk6hxYvEl9V_>ghu12kW% zr4%tmv6@zk@{5=)Q3&h1(eU8XW1%0>3x?fGQP?kuR6w9y6Pqw@F>iBX8$qI zj-4`gq?wAh4Z*g&Zy-Pv3Ad*x?ub}Fqt(W9(^ZG|O{Yyr^Z^gDLHpoAbLyrvMeger zH0;l_dO5EvG|w~Cto(c&;#LURAJ4t*63F305UHVJ5V=y_2F3$M6|zxGLM zk1F5ME)&74lxNptlkOTlJlzoBO~f~hO*XlA5TWO)BSx(3#yxMdVT#ojU#o=u*9I59 zqpn+(^6^vrw-~bwni!TOO#YKjqS!ryNARD&ob`eUXQ}j`@gCr;n-}xEjN>Q@A3N7; z*EgtK^mrWflfu1147ZQuIAw;{q@s5o!8I&S*@~5 zNC`_*xIy|>uKV)wri>ct!#(yBKhCLu>C%-Z8^?yL;ekdkUH4z}J|dq{dG<8FsjsV_ z1<-1NuvU9g1COAI>iqF1jYYYaaGWAfo&Z6SH$1rO(zGQws~CN~`qFK)Z$Fc2*ujDH zwW_~(&Xrp-9%PH;cn5r|iYOyXfb(MXj+X-J=k~qJ56>VWr9)%?{F)MwenE}zFQS)5 z2HV$gjPZX|!&l`O%DHhZJ?6G$+X0DJ&!721Zt?5%l_Q|WAu$;w!)wHkZwL##TKaO| zuOy9}aH*#r`=R2tr^K`eDKlYgaWdfp?Z2_af=f{s-ULtND=QZ$MFyn4lg zqFVaZY-$@M$>zQ$$7R&3EOXl50%o&~RUgEH%Zi~zoID60JV%YfXpL*|wLg#l|3BZ_`u+CFf4kjz3YKf7c+16FlFj%_ z3-90=U1LbL+f{czc3Vd&v7z#YaYq~j!zBd6@7sQ;8_}8Yy?3wMnRIw(YM52dR$*bp zJ?V$8%|A~ss>u)8B|Yr-c{xtvfwmWX;eq2s#{E>aM7PQm45~mW{<(1p@7aJ~&iG$l zGw^R#NS*^!{q$0Vg z(Ef}TnB$>K`T8qC@>8pRURMy zRRCmJH~fz`LPZzI9>f@LRI1BLEgCjgQ~YPVs1-X1aU9pdY`hN z6^evo9iXJCeowFQX()FSfwuf2OlwYSR4I0i;NDpaYMy=jBey@{N#=YZ%OpIvNo>%` zuchrPn|sgZ|E{~UD^6p_Rp*tz*Mo&x?ovEje5;x%SRy;}DPhVJ%P26ylV~Gq-339myG{NQtw5%=6_da#e zMLYz&=MVbJpra}f!~etFTZTm$b#0(3N=dhLN{WDlbPwGiDguInq)1AmbccW-QbP@m zAe~auFtl_E!_W;wojtzq_kHKj`G2m9Kj7ueexAMeiu=CTS|lXmA9r}3Om1ZTa z?+@c-Mzyucd4muRG1pL2f!W>DLmYPts8b9l5`w@t%(9I%)p>N8G#+2cR9^oOq5b+^ z;l1zYinw94zv7$j?j7u`?A+WR>uTD+q`WHccIFPpwVk(9JH1bV5<{P5*Z5g~a|Kq5 z6_@8+(R6dsfAIZQRzHvbaTdkw3iz`4I1CkdM4R(>HR9$l2kRD&H^~K4jX?K8sas(K zl_}ttuou{w%G45+AdoD8RR5@{F`Vc=+Mb~tlD=s~V8hYNn58ki#Mr-*ArR^fv$b$<)F<`ta0qH2K6M&A|ECuF2pBs*-dhkDs`+g+B_}h_cN8{!21c6G8P~hWT3O{+f5Gy^eeX3fPHb{=e(`N5XXmB$bu2s^0^j<7?L>p|CX7t<-Pb!ov7lfKXzmI zLD;5m$9a2^BNNYUVwD%z5ktQQ0`YPz{)VYu5{4rIuGY=D;M*RIZ!VZYPgJi^2S&v| zFPD44A}@Cn#-1*tuD&5CJT*1dmaZ#y3mme#kKOd5WqXPO9kOEMM)APMH!^E1<{iM_~W zWo6ATieLOp5E9VuC<6e!+^lUI^a9V>Z(sBSec7cTrk39y4A4+m!IL`VfE2y8b)=fc z_uDB0c!rnf7T`;pW(Yw$rsj(IXqleVqJjTV1?+FHQQN-7^3>(1rB1K|^*fF5D8M8j8_vYZggKIv|O7 zDf;es#-XM2q+`apOLmGW;cM@J0@h<1ipSZ96 z^hNAhn>LPk;d)2k<215m=Q(oyh4w9)=A%xIoXq4Rr|NPiDK*4DyU3IV)4MwJ)n7%Z zCf@aPQi;$UBnd*bu`V$rr37qU%Dr8qSoD48W>5t;ZE$sDf9anVu8uVh#XgGM#Pq9d3r~)P3hIgsv#@9_oLeLK{q0?tQmFNfjF9D$k`iY*t#{k9J{jzX6qtMJWThORO-xon|?hnglD`GAYZsbdqBRA?dNdTOJP@blWh@l zNz4ti!X`S+3^4<4XN}#vAXI7BzxS>TbLpa$!u_(ubg%t~%F}g*JiCH#|MQfBoC&FK zmgp(T8OQx3N4%Srf|cHNY(EL{TXY3k97X448|jm{L_4^Er0sl3Q%Prq=wwdT1kr+Ix?8dP_`_xB&%Gw*Ho`gavh zBfc5&E*YJB4?dz{^U%WVs{hE@srE^hd1ogW-b42_7CeLHJp>)ap?V5$Q z_ICLxPp|IktK&)2v%PjgFboe`{D=^J^~A|2$<0;iVTeeBn-dH-Fi*KA~{979u&)zrC2~f+|y{PjV9$HC>?l%fltZ1H+kjV%%4m+1-hhB9x?b+4%?M+K-{xJ13 znR{g7D!5kChLx96M0LCk(FolRt0J;T97s-$U|+KXzT4 z0ml1BlmWwqeGKkZArBrFOGJLh8Q;ACD{kEm^mTF)XlrXD0zAmCUo!7!Yv+b2$U@<8 zZNOs9>V$d?$cvq{(Z;4~Mb>>j+m?4cyu^GJq0%*uK*tTxd=gj}+TmQ_y9(-lmRk1I z7l1rEYX6S1-T#~afU3aZ&buyafWLZ|RJNX~A>!xQhMPSOEiIzlUKtzjgBinNEd|*M zN${A^P$FUsFcmih@fP<&7mJ1E3nH5rVQB}R&b@i6A=ANhLXOIm39v-)DZ{)3r?Pb60i|tdP z2c(TcfZfO|9wrzUVx;l;+e=<Shq$=7h|t6+#CQ#Y;bIG_yiz&3Tp;rgYTjUN zyt=yOvc-?nqBR#i0CV{2Qu&Mxm?Z5uI4lm5;ud6!pJ?naCPHM4k4;3tmD{? zYXV*2o3Pl386w)G@U|JeWb~cOtFLu=LflSqwiGp7wBK`!8;|A*{r&YJ((K`?ATM&% zYAPX}7@0IUtU=?4oa0+SkJla=$I@{OW&=~uVSvsQ`Kz-T@^9bfOoU<+%mLITBJj-U zdA$p#CjoVoi0@<<4a-+^gtv4mcCA6KjviQMrW#^2iI|oo^61FL0>7Ql;2wD(F32d0 z>LP7%E&hIuy`(*BR*StAMx$-7fustXW4Nd4nt}s9h=>o8MH#L{-_6Zp6b#zW6kS>r*;jSPjWsB+B_Dam=wJE)>yHfy~>xNY_d?O8mCUg2o9?FA|APs8saP^a$v z^)0@0FH6iegu;MPU_Idc*bK(tvay^C>S{5Y2$9~NdK@k|di<)8!^QL;F5?cG6t{w& zuvFsn*wWD{{vVGg7Qgq1^I2Spq5*I*nu|r_>|?4XM7QKn>wgi8wN|~Ppk}c;%DLE` zYhEWUapf#&*%wA=;@FIyY|7)_b6-ZHjBjOKYLYyAdJDuK+|obCE;3rgq;$>U`hd>p z{De|bBcq{GIG5=0j35K_4uiwmyrQUu+JGFqqau84On*t_dcTXo3d}rsR{Z>P1m$Bc zqrQJw&ii~8$I-kfG&B81oqZ(Q|%`L#No{d)48kdz;CA_3vM^rM4hu zrD0&>L2_m8@c9ZUo1?yhxQ;j^9KDNoO2{z^gd~NaPd&H64NfSrK!;&? z)~bbn`P+MR)cejm_pA{%jRXu@e}G&;Jb1JG#;LO6bguQE)7~Xqf1FBrC$^MEPpBWX z_5kYLRH9kdUU9dr=emMn5Zbk&jK|l_q@~6U!GOs{0Q>^G?I+1oZn*aMYnc zz-WrwIK&^Bbb-vDj|>u{oy6E(E7j)Qi#vO*XXEbIA^c}`yB~N=%kT84g!R%31$aIB zZ8(mOXCU{;-A$Wp*rYCeh{U{~kQo2w?YgjRKI^sNmRsI=q!A8w7g+&^nQVgcq_Fw* zLIndjWYUU{yj;`;L4+PZ3|d5EI~E}M%W$oZz72kF+r%cJ^E%xj&+b)%b7sfKymLk;X`RTej;pA_$N_w*K=sM4t`Lo6@ z%s&v77OsFyh3?6dyGYkIG;pg4yqJ0+_X)fZJ)Y5z=MzQ_VVkmkEf+ft%i_;6zJ@ig zpaBOnGc$7!Fpz>-LYI;-9&X6Qb-2qA)W{u?D8zcS@D^;UdeqX$z*{_*DfYZ@&x;E` zGaJzd*M?~m+g$%BRo;=|`l$xn==mG-$#N1wfD>$S2 zHvz>LcR;uJ+cY#Z7AnDc_{iz1qBHMtrzBouB3fD=F?oc0%jJPu`TMB~3v-5g*u47+ z{=P`L-;FQR<$c_$`{yA#2-41bQzKDd*spnUp9hF!c+qxgrBd{v;TAl-kKKdZ9a&8x zaElHJ7lbfT%Z)g0vr#p(R*!{W?7skCn1qT9&(=#RlE&dKi2PMC*>?fcC6*5WcDX!I z;7T1;s>&f9)z%|aUQt1kA?-H%oT+}}Ru^sem8+nSqCV=4f;fY zjV%Krk%j@2k;Ilh{?&w`8Cx7aB-ws9m`dC?#wiQEK+@sfZDNR; z9_80jy-k;ayx!m2BU*^yrT2ttk*zOpic-|mN)FbFF+8fHrhE`s>>U7cT|Y+vvQmE z&ta6$#fkORk`8~ZW|~+caX#w=x*!=%!=KWIJHfOK)Ntcq#Bc5M?}`WgS|AW3Okqg6 z%VUg%6ol#rdVM|^tHSZAmsP+Kp?!h zxR}W^M~`}c1BBvX`k1Q6=o*+#Wyx`)Ga&F+6;nqzgB^Js9m6cdq53Ja_eSK-`_PNP zEVg`%AQLrLB7~TpK8_-*cKsa#>TwVwJmCG^1@?PURRI?@Q>OTR2y1D3=jQbWm)3@F zS&B=qeHBEgGk+sT5mU}m;dR2yOy^m?4^M&Jd z&3ezSK`MEfk5?&Ac*IZA4yP%5PIDs*rq8N!+D1E#569wC zTQ+ivlggQeH;mS+Ne43kxk(f{#SBxkPh6~$l{6PDEG`nUtM0SoS8X=`i)#_}9#OCT z?q1jeutRVD{4jedu95#F+wbO@x$MR*=(-Ze2lXAYTyQ9BjQElXo4!-n=tZ)aa3It2 zqSp`Bdz#WEpM9vGP*wq7AHMKD8dl0G$m_~p4suCrzB(}~S6@;f+tKP$NmD8m4cPG1 z6T9t-3wH?ArPbG7$EM@XCDmqER4kz8IkEu60G^&v=I~X}Oz!$yLT8C{U&2Vfaz*8P zot8y+{&_@lj*)N3$T(ksHk0b~0z|TzU~U&z?D|2b@kTZ8!%lNAU68K}uz1PB2QhaYx8Y%!Mb08y@3p%fy$5wr|p=;-+lsYkI2HcaqK#`bAN1K9SA*#%ZSgw0DM0$JDZ+@{x3&A+_p@Rm#m%ogFLXPGnZ!RWD;1b9C&cm7WY`yvHqkR@Q56_ztuOV?Ig`AD!b(l%;X8 z15jrow$db7wu?GY$`Oq&&+Rn12JHad{sQxyY}09U_+K5tAWQPg@XUM+FE1~Pf|Flz zrf4N4UsYUjqg?+>FRG!B3`~$yjqPj9!m{rP$My z>nOP~YwEgp89vpovN_jW40_l4VnuzKEHcGO7W3 zkaKs7!n{8qCA9yxgQCQjb}SrzL2|30T{EXb910=yw-TPpSeXVL+T?LW;Tv1 ziMFS_1?aQU3FqM7C+{%c z>>WN;67*UB=sjgf7^3xTcFC{oe;=iu>xgqKO#vg!nd_m_x@^?-1YQA)qvCRMTd}JU zuF53&j;8CaG3R;foGVb}KvZ{uJRFagnXj}aMaQgPIcGCEJ{1)eb??K74~H6jqKhU= zam!UXZ<96q?k&pl$A~YjuCj*T=ny{*f0l{dO~{jhS2_wlZWLoZ_%3M8?9|Pv`YHNz zDMnKnQY^MMZ}6T$;q5eTpS=1^SUrZIsl-689}MoQ=Y!=EnXs-oX>`?)=?tyMEEL`5 z)cq_bn|~M--5>zJ(9{29L)|!i7ud>w7urq0`EaSyiAK!jf^u7WC(Qc3MBF#ymsTgf zXs1Gj2@bIiWsp?Z^OT>dkR$8xpP=r{pfZ|HJY4k8t_W+dQY00Wr)O)v$q4%mZZ7Q7 zjilQJ&qX}}S0FwKQ|{Fbe>fi|b4AAC^#8kH^r!3RQ4WvnDbN&OWTLG_S9$>9ga~+de6@J=-M1 zvLGKqKz@!w!Udw{_168Ho24v=(jP&78<RwJx&_P) zp{D=AQUU4byfLGO_a%tunOXkG*t6VCnV`S=lb_peIAQ6fq_T7@m4rA>{?ReGw{)50 z3zfC(K_Hn}dRc~}h^-xkL9%_+Qd9G>X8}37F}$vu>*<86e6+CfQE@V8ZJ{{o2R!Lt zm}AT=$&Dpky5oM(o(`C^&(CY$iyYi{FvY8>_`g$3U2i%UOU2IQ$FBXnWi7PXX5Nhi zMYdcjnx*C3Q3OFVQ3rFJAsK;Z3VmVbSMAuYZrbdjT;!S6AVd3i>s{k~x(1ipTYhJ5 zh_@^;{G7s$j5MsC{YQP80iGi;==v()Ze6v%Pdy{Cmb2oB_x|0e4@F>p`2}3K*8;|Q zng5(lmSta%ODmS65~OU$o0FXf?fqbcr2AbRB$7GMzJUlB-Yd{^peoZhence(jFUB|bRx zKj#8XoCg4G^xyjEDJm)i_#M7sncUPrPDgnY%~OdH?tXjUc9BQpc$V_2VcqTj;|rtC zO~m0RE6sqTV{k!=+}oupZof1(Oe`TsFX-P1biwiQ&ofd}{kr}Vpi#(7f+~O&l2OdR z@U=8X3w`(qY1Z4n><=O=+}CE7+}Adt?zX8|!fCjA)zAK|jtayy1eIXD`W)_yX1brd8Z~r1`9x~aw*F&z}Z|ZKFCVQAy zoW6Mbca7VK``^Zc4^Go(MYoy+a!GawwZm*}YZOflm(|!e3Oc$gc+46ppfcs<<@rIj zeu>&F%klvrIhCZRpt-rrdb{qset=5sft)TAxMsF(Ff^?VNERx69nfE0N6K0flf25K zz-pPbN02v*hJW$eaN=3a(*knquL;a^3hgB&q@NHPGaUh zi~XUcsrhW8TZnw#@5aY&x>Dso+QW8zv`z+KOQ6-(70OlO>Sz=(S;pYDJCFZZ>+$8T z-wi3W0I(UoCV&53T8$-HvXv(yB366wzI5_yQF~SDKhl)9O(Z@26>mM$2NQ4a}Z!O;E6&0+``^$=g0FQ1XeTPn(G@zi1 z!rvLa(yO${wT!2Za*-2y2a3}ZQG55d@Y!(Kc5*|X1PIqRC ziaF7Yh?x!UIOoy&!1oZKR|x$^xD&Y49`a_jJFfSO+f$eS*A8$TaRweW>^94nnKV^C ziwDi1gtoS~74O-QTu6GIbbu*LFW8M@=q2H3VgL4w;bn<^M*RrFD~F+>+TOKf*y9>( z>&u$)T*&$vpEu0!>X5`N`(VszuPp35oZ&A?|RofRr)^&!Z zDx0!4Rl2gz=irB*(@85!@}#*Dvqh*nACZNhmJM^KnMK34Twc5SmZsAt!{@W_8%pg7 z4@4Cf{30kG{dhko)_E`{VP|hY43!U=nI9rF%3kPiUoIL?IN>|*63R806k@-StNJ=S zJ4@E*>;M|AXf@^Q^MtRpPGXYBadB`EF=wpBnN-_70ELCm*=NGZx%c`Ww=70HZYJ`x zoq_RI=Bg!yKcGZF>)rRwOD@KOAO@o<`I5(DdOym`TY;#$a>tO8I*mqD)umMMj>dYa zeM>zYAu`kK>bhFm+Vdgpp`Hw4r7}y^D z^pDQ4oZ#o@C!SQBkvg45h^Q<)n4+_v;T}bJB@0@UfDRNn$6oMp#HBJe7*#QM@)2_J z1shiI*j4frA!(Hp((G*C!Z{7(`gWJ#6JM6WKp>W^s(087og6;=n%-b=EIC^`%&*gB zt^w+ZxNEKm7^)!)-x3qar>krYIlp&r&(q9h=)2dHUUUf+3=6@m!_bK2qm#I1!xVvY zU+onTAu$xzY9*dIbhVSzYiTaz>trY7*7pHeKw;3)I7F<##A{pV?Bask7Rmdz*2Lp) zqH5IBNw-2VmyGaLAuIST(9)l){^Kt1S{J*S65$PyLq+)HV|>z0SH3Kg*F)-%-7=Uh z!!jay~L4S76sbYsMFZR?uQNvjCn(chpWy2Hh*de37nVTbu*oT-E?)oROsUdJ&6BMw6Qf^bsr!NI$sMhsp*)EydK9gRDc;f76k{fZ<1rR zn}2H1kU^D)g~?@u!Z*Yx6(NvdBID`q!9fZ-`sqyb^l{>#jY6FAa4!}J}5Db?WGYj)RdwRi&P4C9V{b#)H`=kW6O$5A*z^3ox2_vvvv-mtm zMGO?Eal-po7kEKuy3y##LB!P5;b8tIf4CfzZ#`q2Uqkv{{jC9+4-Jje^|zuvouN=% zYOa({GS)6%oQgKZSS2>neW_^gP&>xN#29|1%0od1GXc`WsfMS=f7Ga|NZ#OM6fY;wnMu2jt=N%F z_F=JY8pbN^urhXxKObTE-rk2-wI6N+$)WB9U!+^N=J2QcKHL60;GZas3i1L|ylg;}Gh(Dun3r;si}Z=W4^KR4Eg z#PYCePAV6)XlR0O_-lRz6_BF|PcqX)~1$pG(A{+?WUS7_!D=~r#njNKG zgN_>JUonx3`4Po#rcqi_=L%#WQ2?n?fIcR-N4DfSHw8BR?2axi>5KuTZk zwTCcAeJ=3OW)dmfeEs$<>-9p5UrR-hS+5*RmHvmQa9jf5u=ImV?FMeIEd@hl3;4Km zlL4I();(K~EE^-Uq7Vr8b^Gvvh|?zl;(oIA0{vjx{I9sY)#!N81Nz{}LP69t^;+bB zHkX~WVk?0S|B&=Loh?6u$?5MlUzB2(uqhMiwzgmmtEGMU!t5zt0DDC_)+%cdZ(WJf zjM?vwW64SiW0aNk&+uF=qH!Y2CvjI?{!WX9yO}qyL|xvjmp>W|$)^`Qy2zTfLzz-C zL7Dzkc_ktG>qkF`)CU7$KdIlv`FYsoHP*0^nq=~rGc6fe4se1&NCq(%S;P($cP0$?!01Jv+gjt`UWZcCS?2dhqS9CYR^~pl>VGx z^&*xq7r$^r;n-byOpm<;b)`C7QmC&zq5^2*^})m$naL3Eao~S5a8C^MKR+WC zkYL?1SR!{}(YCpDWts+$QXix+=zO@6k1vpV^eH!lE6fGSS#QLzCfy8NB$wJamTO0( z2vrQa;w{d~UfY3%{7x2UUO$LtTZIdyyxFO!s9taVDQ9j9$f6JsC~8>=7JDLvLa+$h zj49g^a)o(vB~Rq4QSu9n04?v*+l38`*b~^M>Zqg+tGuiH9skmgSE%F81eWPc%{)S> zGnsxIRmn7$mU)dtmtOho$q6!$$=yM6W|nDqXuDV1y`}3U?p2G5TY6_kH?+EEU6i#)hq%E^wRRqERe#i`pS!dXFdenqGL~HoY>J! zR<|H(4!-!{iWiKzFg`GX68nTp6d8Qq{7*Wj}C{VBOpxc(}7Yr_fc_0Yr%O8`ztZhCd|e8levJXl6##m z@>1!LoD@W%QG61se4=7vNI`RhdpL5`1KIPnUG!e+MR3@UzhLTg_^`A;gwT4>4mr!% z9joh3FaRs^1w_z^0-!g0gC^oa=d10%SLjQ|BA}WNNOD#=V)sS^=^v7%XmpkO2YLcI ziz+i+zH`8!+Lc6sPI9vi7k|B`i}6OQ+{|wP8BZ8enwXd@OCZFgQHbo9LCvQx9VGTU zNZH2!6gU^4C!)uMXJRpG1VNy2$aR%8i81a{f8Uy;Upbp9P`PoOt5RRo20I>&FnaiT zPatTJ5ihiCsRy`%NFyy_AXZ+GIg?R zYHI3SSd|{eridY#u(t~Ql74uk?zI}HR7vqUZ<~y|)H6EnMnfvfCNntCIVVo{;80mx z=iL{sl&_qGIBu$rj2Yi-609LQu~#sH*|=;~EcfomOeHeIEl6$lesF5CT#*KOWYo8E zTdAmnV6BE42=UB!j})KhGaAcdlobwYBRLqQui#7aXeEy#|9N9&5)--ez zwQcT%G2yl_>Ezo>*;0Ro4iX&4zU%GF%L**ixdOCaL@N97A#1ieq17?mQA~t{{Uf!M zPg(5K`vh_FY8#l$>oTMyNY<3GPC9Zt)^l-4s$A3)6KmKafYU(#jV$whw)@{n;1YN>Nuu%iLAwlr`)Q z(w!M9)AT(>8+Fh1b-_nBZ6r>VOyLLtVo7Tu1LdKc?TXCGeu(39vw6+T8f_OQ4AtuT zQ=rE_Yz8HulMeHA1)0-F4AH@(hU^6Z^ zXFhhZWwe_)Z%uwd==_07kIM1|Y(rQGs>wn^h^!8!ec@?R~el8s$-b3sADGpN>~>}brZpGvV1+y&2z64g__(RM8hZ_P9*HU%rq_s8P1 ztOS4?BEciWm1Ai@#)h(%>+6&@2zT$ZKrb1G*^V}9^AF`Di@z(d=JHtKJ~W1(sz^E& z?k3n!$F3YOLTEYqb-u%4-?AJ<{_)RJ3G9|WYpw{*Men*-kooF)II@7# zKEV&U6^Ud60dwPiL_fVi?P1WN~qHDvZiz%}b*+^aRx`GUF&v z#_)^yc&#PK^g(T5N;v2gK%kq3fBZ_KUPNW<%WyFdYcS_QFobtQtbk@33vD1>rSp@L zEE}|y$iuQ#6``R(2ahZE_?Un=xB#y{P|B5d*y@ht${4g!GrU5F94~8~o_x`$&MCLn z)PB1+#N~-A7FdZlPs(8Of&Vx(1^?|z_p@P#@U~;lg8;GNjgRe1zX%Dz2jLIG9TfCx z^$-HNWx07d;fQ`|uS-rF1^6%WaqZ>7?U`&fgq|Ri>#yNVe_ZOjObaiYenwxa7#|x^ zZ6g3S8l*acw=;F{ zaUDLoIzPf`awkInZzq*3?Vs-SPBOpgu0akro3hht#SdiXse1LS5Q*hC6!Y zvU3Fm6j^=|R#57JUC_4gbsohl)5Y;Y5*Ong(EvK63EL5t?lOQS>#7q6XNUa&{Vo3n zO8q@5vQgK^BpGzc&@I;>VP$=9bEk$BRXF>Myt0`0r~o5C3`2iCm)?AUFHshJa2zf2 zax@Z?<+IihnL(|TBqhQ@^(Je*0ELj&JB70ysr-oFRgg*&C*oacm|g=y!%0WMecWjlIT~srY@O)4C2o3O2SLBd}?MdlLTjvQ_3&S_X%4K_7iPkljh(Ai!5)-6326$M#+$;fdibGClq{`|#(Q?E7|D*${soQF$aMuV2j0y3~7nFa=DdF!_J|0nd4+|?IPHsS_a{5+~)UZa;Hl~3A zuc5}MVtyV+QPg-G*m&G^<`4s)6pobe$c_q2LMQuAp09dt<@-5s<##2@ebScCt2FbTIT$6{Tj%+DJqAmTbeZgMp9L#-j6q6%vTmYmU(rm&ga2 z?NC%O9W{(>=`(XMWauhdg+KF8rxfg605D>Czl67TsFbUm&3BU+r((`!WpV=bmPcH0 zm^~7*w)8nGfP*VWSJ>)Pn68i23!j|y5iE3yk6D6EU758o5}+PdGN9cpc}^f`DxO<8 z;u4f1gr5%~9U#jq+?H?doY`$Ow(R&;AW^vR1i!ctaI_Aom`lpp&tJRwc&+ z?fHo0<<#wCu`?D}R+ApY^MoKc}t~Ako*+3g6InnQm)2~bW13D=UZB4 zVly9@8pQJDVoQbR=c=x(MgUBj+vuBkkfi_s+cWlxx(V}Ma;NC64xh?NVduUZz7afw z73Q|iimqnp4!=7vd3q~>QU>8Uy&GgE5KX}_ zHhVosxFQTW=h9lRu4c?QiK*|*JA#o+U#wpo@l}Y0ieoHEfiLV07FLMB;nN8U4%Q5a z^TuWV@Is2&_TA%KqaQUPQDRs&Mq_tyL{X1M_D4OmcNttE))Z@uSP>9aM2)SL)$`v= zh4{eB4nD0H*^t;U$<^c-7{nEGK>2{#ugdB*O4)Y^&FkAQa5U8NP*H`=Z|Xk^dChXB zMl>50mdg&Sq^|eQYIQ%;4JR76V`2}=3eC#ON+=O<$r$ODQmh}odO4#TkbHsVT0P>R zaCMq>?v+VkJq;P+LH<~~1I3#nIxyIHp?OVprJ>!I$d3(#Fon zPkL7+I3U`?DSW}7+z?_%7v_e-*UN6Hnvz!a%aS!55v=0G|M@$32as}rvHADo4AfO< z1;BrPlEnbq{6D`?1sBNw|NkHJAiDa=F&(v3fgvF*63n;-4@5b5p9u?~Nf@C{&=N|c zo=uvijU_fArF94!{cJjbmeb4TP}cfY&o+qIADu+NkvM>$Av~s*3ZI-mC%Oub>|6_= z)Yi_~y42QN@jvKPa%jGi+0NXXvu$yI-*Vz&hUif&F_R7=$o#{ z*5{J)JZ=l2u?W_`}7pLB3 zC|olx5JkH0LkpV^2hPaL+ZJJ^sbKY&CB-Fmcj=B5Lh_D5CmQi$F`~@kCj7ddWQi?9 zLEX}Ym4TOw4>leQLw&?0CX{_4WAW+40S|d(zNR>PSxkCMXwD#|!xBUM6RLoA_iePN zy6l_3CT00m-YB7`CG$m%;Y?qR@JS}~+HIwoSffG$>CwG{g#7#esGaT4MYr)uPG_<3 z&?Yf49#9+{-j7X;yW6r&f0lQ6KitgWt`F?EmzzPj;`N@9p(9{xLRwy=HOZGaY+B$qgAz_leXsFx?1NI!7$;dWC0-L6=BAa01 zq5I}}^?~HVk!USLZE53|7Xr|hT3-9~8vQ%Mx;5DmGx=vfjcnXsXMuRR!C%S_f_VLTjI2yp>n9~$XO0OM zLYR5aMn4?MX{%g&xh#70r-o1{fL6* z*`r_XP5e}_cmCwIH7sJB##0SsEswbQ&HwNvi}%YrxxK(@;S;)}$Nu0=2Z!X=(+4~g z+eaVTllj|+&pTmnG3NJ;BWYn+Z@xbObzqbiM4dtBwVrFYXdiwHd$ouoJh<0D^vmNV zo?cV}_EoS}+trVq5BuctSbZ1cws+Wf<9lC{E7p2k-Ff!d=`;NljQ@?--8 z5qL2A!zB+Nf_H^AROPMSIC)d=8&dB1hTo!;-D!O1qOn*%RiYMS)Sf-O(XeU%_w%it zfM3|#9-EFL4!^p5_#=OBhFz7w`hT~^C&iVtD^srhUA!z$`%%UB#cqh|xmH@h=wzFA zj^+GFR%zSs*;_Z4<8(11nTNh)3&&@afIx3gNqd*rR2fSVsU_Z1y;~K7C2#xez2+U- zN5{i32U99d7sYU=)N+yNR=X{$NdDA7XK0wYu{T`Bnr_NU5@Bt9#KTwaqM)E!V9(|> z692=Yk2*Pufjupo-jer&Qs&opz&M`<*TZXG`f9%@lJ#Xq`CV81xNpC3U84TZT2>Qt zkai0Hofo%R%kPZ`=B>R|dxQ~sC$T+R1OeN3PO! zQ+k}Jw;R2ZCiF7AzeCxbH$9C1OPTQe+dJNFw2Ge1(e;lM6|w8RX`kdh55LlQ_~%X8 z+ZgBZ)P{(QNm)}ryk?IZS4Rzc+iP0PNGc4;m9Sc(l;SP6`1tflA0tUsyQ#?YB4X{q zujH>+4qi}+y?shi7T*%EsHS8l(t0cSt20s3BR{vPcYRcd7?Ccnp3^9Gj@=7^aUdjE z`@ACRn#`xXN>o;-asAd)j%+YhDo77D$ShVq_KD20>rLL2w#3x3&Uw?=%bze4|HQtR ztNAaMiCPTY6{b&~7e({&)5n}>BwlLf3Rz9lM;E!R{1T&dvAc5rW`9gw{mWgcj09+p zT(PFLORWPQCO@6KC*8yzBUVsDQ0D9JidD){7VG#E!B$;$w(t>Kjo0?iVb;*%_=!fx ziAo&b*}C2o)Gz1Q_K!CzuVU%s z)l`+t;IME=SW1--j+s|WEb2|@IF5zNTOYv@>{$mhLV5t+Rdv!x{HlFM*$*s->uB^9N zZ)Jp=SqEzQZQWlNbzWMwi*RV@ZAuY*X}MFcIK_Z1UXQ7vhUezva-D^hnlt%y=G)To zyO-g|8ZWD5zkk&j-i~Sg@Lil?l>K|wsPo&#O{nAdkB&U7291UlH}54L$#qd@z8HOI zHG+JB_arsY<8a+9>-u%21Jg$`L5GD0UXMG&CyDMqJR;vTYB`r|)9WRuS?@}fhY?i7 zOL}ygmG6wrM}hjM1bV!ou;ZcT>s*)q_sK17G%82h9O|DP%Z64@y-!>bFVtJTH>e&W z@F7^aKbJ#f-MV=$bU@kk?L=}^cX(?VL-(BA?+JQ!AYg0_bshvYqb3t zo4E1ASMU=@QGM5=O4pI}dw-hWv!&AU>`%u#tuJCNTwzd(7#7dhYPtQX$b#=anWB&y zSyee=u)Z@{cep5KOYOV)kW$8{a@**si!uE`o$qV%bnif^hTBr8WY?z-xaDVb*1W$yHE1@aX? zXP$yI`3>v4C=1Kplz;u;nie1l;~E@asI!t&aV(`6a`6vpf0f!y_NhNLyyz!t_eG%P zQ9WdP(L>+OO^>~o)8ZU46b(X2Z2VP=sYv&5l*>8=(ei;l-;$1j$%J&&pDmi->W zTwkB~Fe%dr`rVs(tPpY9O(6X_DZ^X6KIHq-Ppt)|(XN-7 zlYAKcDUw*ZA&Q*#T|-6ad#;0P~0WK-Q8N;-L-ge5AN>nR@{OI zx0CPRGS1!3$X&+BN>*mpn)1%)O}ZNnHtnkJe^wSb%6*RMw^?eLeHF%p@q1Sh8XQyt zEcj84ay)(!@{v$1$iJJH(`R)&K^{*Q>Us{XYqopPVsEg%`E1~IK`5WQ{e5NU_~DHr z`O+>TUC4az9>c_V^=PXfWTxbn^dd*UZK%ja9G$wxje9DbWTRf^a-N5aYJ9GT3pr(4 za!isTlz zZsIAGd6A9-ha16?|JHlU{x(cA%l33_b>iwXA{q`|23XYSA!>Qu5;IpG}D|2Eb6io_(9jItn53qaLTWqP9^{^wZ; z;ahi%5rS$xlIKvS^OC{KV`u0tC#xt0$kg55H+Df1RLLj>$Z)7`l>pA$| zKh(d%${ctJpyz0GBJ`WP?Q#J7%oydb|%QXrI zEoXUcdQ4|G3LQb&HG)*0fg*~hgg8<@{V1K}qOUOpe%Nc7T`$;nl%ur`?lh2kz@@k4 z!t3nu?JUayT#M%I(#)i^yo`v8OM}NZNvR=nUNWq{WP_zCuQdx^Nhn~4#+@7jM=RQO z>ug97_eJPBVY}XoRd3?}a6cm_tso;QuJsKSyoqqJ?|j)~%@J)N*nT_KoRCCW*7|;C zy;^=Q^^}V-^8qQfLe$8F9Anpb+y4s7&E|GMZH~-C)WwS{ATU%9eqpMQBzCrT$(!gE z0(N-w!egs*lH-HwK(FHyD;W+4{@lY&MNScOSmGzBCZbpJ?9Q+{N{6+CbP30q%Rl;p z^{%`EzHg7I-|GS%J7pECOTH31;mhe|v;3X9@Hk#6nQo5P+YQ#K*C1vcbR$zqR{NOG zCi=eE_-iiC7{c<6l^$k~j5bc-jE)SI zn<+OLyn(}`M8NpyP}h-GBaJh7BKqEudmFF3NGv7`L1EBl!M!oOGz_GG5k*M zX1f*fI5#of$s7E-)aUP(bB#0~^cGuSHhij92)ldpK9ct$9FRfzKt%P>DNLWho-p+J;r5iktkz_t=GKe5u z*HP|DO41QnYE!1ov5KQN>%XzY%R^Nw`IpwI=KVs#u}&D}b|g1HFukNXJU)%Nr81u2 z`VPV|22J}D-GKB|7DxLHH$y zbSeo0DLq;WbbV7U#(o73STGnpQ`7n4;N9tr78aue&7q@YkzU(9ATN43IOnCB{$kCu zn%YUZ5w4M0GdHRBS1zYCKf6Rq)6^TA{AU7BG^(VJFl+M;Syvn=TRYxyNNm@oy}Ify zB)Al)AUCR_Gw*~eRVq+(A@ugU$if&$PlO$`QE2=dA=9<)Ql?i}<_Gc4f!q*U9U-rI1o=;G&RzoIo z`eto$lNj0Fym9G>NpEMGuP98Xlo|%99!p;*aOwXi8h_$GLFw|6P~_<}RLj~T_{y;= zd?Vm@bcMFAVJj)Cs7?Bv42l-dh#Al%?T8#ACG#$^6h1ZX-SGu(B~1HTzDmQLsdAUl z1Q~%ESzFzoKTMGLi&aRc&fbohrjHE>liN%-pppmL*6yqraST3)Hh`k+bJQhvXj0~R{u`b;2mYP?`2iY0zo11}dl2En3pTIe^_d|uS$=Qi zEDNB!>H4(jPAidS`l_hu%6-e<2xt{zznO~EtFCb*Ww;vLlxs`YW%w#-v3V`y@(STc zCSv=9NqMz(;#g-0y=IG>IbmtE@ctP>+nBCWK;p2XfJ$FAta(k%AV1^3I(#PWEF7;k?8-B_Sk1qYGBob??_-%UVaV|kCHcKX z4N*8e!#JHEiRg$TS@bBlfqJZ83TtH$?hLp&egDOk_g=m5-7_1JGjB^3 z9V)?!pH77Y zLUcwz@>Gzm2S=lF)B9Idzs4@>3Z=ZVBnU(Nh^=khh{$# za@$nv^ml&3*t+b;(Nf?*z_lW|qfv6@OSjbvhMQsX&EX-vXxD_U@Hg_tiraBy{us@i z1oN+dqoCPad&s?E!a%P-$hweoV4pN>6tlLuS=wqx2^v6Q&;Uv#H2Z#6O#;c{j&K!` z(kMt~+j(=Hy5hmb3ttydKPoVR@fu-wP*Xg@?FID50)6|6r@HOly!xrl|J3{h@g9Hl zmgW1t5ZsAahf8+_O`B|lv(Dr%LsX1P0XMic`#wa71xnldpt?%7OsU6rf=Lfl-d!t@ z)Qjt(sDGc0FYByY*A+Xq!IT2~ZGY|0wAbEj%>lj*(JpvvB1w8+KKfg|L9vEyYZ5*G z;Yg!4i8sONOF)rahT#E`v(t_2WaMN#AkAA!|48?gcCZ9R)0j8A{;fa>&|AS7rYKKe z>)x_0b(Q~DCSU)EaB72Yv9x~Dvte2CZO7vDu^B%#D&SL5J?lW%{C(M{Jf-@wmwEN7 zrR%Zf71}tP($+Jwt*RPnA?_1+@Lhe` zkKmBef%=TjOR6dwTHYtjXS&l?vTvO>xDrQKt{&H{Pgzd{qQXeK{@zOMRi|iQtOg!b zSg&!A#ny)OW4;nmMb=_)o)n~^X!u>^6L3a5WyJfou5cozoo8j37yDFhSoYX zfnBH!xs+X?73XC~5sJ}LS{Mqy6nfdN*Kz}yLu!OqdxK<&hi|u*Cx|MSTkhHYPfT60 zQlP;JQuj&CTFYoz@&r&tP9PM~(hJDpcH;Ol6^Q2jr?N)j)~F>Z>p~m?CJ)W@yN%VC zaqL!Ms~ubrsS`A=buqRSO!`Ify=@(l9!)K7^^b#BU_FMD4q5?GdvfExHkakER0&eUh` z$q+hgXdR0Stz##TZ%5-{UxiIbkNvk$8Pg5aBqH901j@y~DO4m)Gl%qDo&$BC-Ti|l zE0y<#^t$<{Mk$`e%0!Rka28taxofL-pM4t(^>F|fs}gwhk8X1Z&q>Q$R1r|0=HLtT zd<`Y_c|!C(URgJZ%V5bQC8WUIY|66Llq0h7#}0rXSaLitL*I^jD}Zyq0}^;)vjI%&|`gx({)U zi#$_bU-@#oSdE0ubQUEhZBF7i+-cCmr)9#A{g$P*?Dzr=9^3=L{bTrCzgO!AeH1ZK zV8U{pe{iD;+^=HE6-~+U+_g?TCGG;EvXGiI&Ca0=C=E*vIwyc>iXMuGI-jMb;X!dQ z+fPNRS^{%lvJKUG<*0Ox*{lPvW&e#Ar(674AFE#)bCWE_8{(SiV^6Yy!0T-D z1%ZrXD=CO8kJN1tx0)Wd*}&sMGwTvfO4KLoAx144G2JXICx#*I{#6gWc2YFuaH(U7 zZHr}UoHHN{)ZIOJ?|Uo9F3Dv@R^>fcJUZ-ix?I}v^C?qx*Hal&Wf)ZmC@@f}jb053 zP9~hmwceDov3`=*B+@AUhK*$gJl;beWmVxYRfqTe_p;bxL8c;18aOk4YigwoOLE^+ zK3a>WTn<;`b^WdOJRXO+&v*-#Qu>aQI~QzavdJ}IcXi&ac zs=5E3ZS5B~E)i zF`iH^Um}_RT@B@QYj)j*)RrQ=o!%4X*pTz@E39@zZ=}ldv&{A>{uyI;GI+(b<=@E# z>>62B#8z+l!<^-MeVND@8C;Xx{n#{mCtRCD5L^fHcvO8DlF_7O{#cM)dVeDP^tto7 zLs(v3fz_uw*Wx~isO+w%+KWI{4Ehscdn~7JmY@)F_EN+OsLZMjB>okgZD2$uTKZ8b zEx=|xxs7R>vpb|C7`hoK**!mx_xcWw!HG@jyu0FJ22>2#^uwclN7{W)y6{)`z8O;P z)i~GZEA7YkecWKCe75*Z5y<)g&Q6s>WL7?f%uH!D?qlUwo%g(B9o7!0GUs3XQQo-; zeYwHS@vckPdLN~E{aP$9mywDUA!InJ3=^FJ>Z$JdBJ-RHbjZz}IbmISb0xJL%j-QIL77thgFD7{xEYdklO5m*r! z;oHMg^f^A1ot`?umCYJ>Uhp&MoraG8W%rP&b(gd=Eg_|ZPd&bz^qz33J>9 z00l^d{od?kEchDD-|p$MQf$NwWLbnfB<{4Im1eP?$Q~*Z!KrVitwabXHqu+2Y=I(9U`aN+Oj zQRIbROiNCioMg@D&oqRo6^$cG1aCI0vZV$4>913+Pw+^KBn*k9SdLyW1}D=vmARhS zBycdvjcm2JP1}u!t9c7^%)mz`-XT?q{C9))N1%6uPZ`C*vkL2vWax5g+eb)Ab?>J& zWse=@1^6y-zZ)2W-;OqXb6#G*<^09KVl6+aeq-C;u*ZE3$LD$b)ME4G*y$$LY%tWQ zZnMjybeG*?tra`R|3XSHHgUn*?OH*eU;)WCV+6IjjMJ}z^NqkRJnl17STD;$k0dTd zgKYuUpv@BM+59SWe>S*j^PQ+5s>0zi>#pXuoc8C9u(2})0*Rg|#!je328g^m#OpR~`H=VU+rqC_oc*$gcm+&Z5ruxdK zS7q$v*u*d7eMrTjM&Bz{d1X9tsUC~YLrS?spd3%I8+-Ns5`_~x)@!gpV!enO2(^jI zHUN32@U(j3`#rQrx(McokEK-R2^VmbiQv8sHRWwF8~!m(lo~T2EamiTpc8tk!UYl0 zSvSF;w(KTzG0vQ{G}Ruv1Zi2EKj?3D67$6_2-ey#$(`q-g)`K;i_ zma3-}FP#|Cp*a<~yi( zx8L5dWPyVKf_vyzR$DK-zaS6R58yN0OY!m&qc**@ahaCOG(c2|4z2SUePyl5T1fgz zpT)Ei)#>J=2HLS@{KM@W_jHA)T$HoQ!J=k7I@3y=`!pmR2Ty1~ANpq?juzVL2rZt- zF#cCqG%A}OP{G5Ki8e6DxZgvjZoT|~;BrwAN*_-*n!`5GGNQ$?0B_&*mJcDa3It7c zg;$!BCs+O$e(ZR-DQDWZ(H?WF?lC*(bi0cG*MFYOro`IYG41 zF8u-fIXv!+pJ81na7}loahQ4dCc6jqhvF;8wdJvJQ&(xfSEz|9fBeKu+LzB;Pr1f} zblP+#*3Z~NA}ee2Rb$e<2YHFA84r(Y5xoUphR4-}dtJ{=2DFSHnd1Gg-36I1Vl#3I z{#G~;8%ZztOuGS!v=EkGepzx00HETP~t(^WHUgZo9!@%a~&X?pm{WlL9@Qohbk2uX(&HP0v|CaahBC z*%1&zm)$GZXNh4b+vm7es|)zCcq$92>*jaw68U@7OOq2270Rvt9$m@*m9&vyL6`gwNMu?; zNJ?~~hDvhSs{cpe28D7$0S)UEbRXfa?#76xD@^(2br_VkhW?K@?&i;b5UTsIthwCq zsPS>x{f+-*W&iV3g91yT?12A$m}1ob*y8{9m?vNVA1_*eY~j_0+R#bJ%F4E+WG9w@ z)nyG}pGyiUDFi8y21YyG#r*8JpzY)&s^Vg0sJ<;KN*-%tEY$}N>)?=Fu%4}}6lh{@ zMov@xLz_WV!|V&xYH`mLD)gZp*fE&c2NwJ%u!vp<%a6oPX{MAB)a@pqrBpu{VgFQ% zqiF~jjb#`f6QO>7ZN$KN;>;np(dNbMlZ*|^@2j#S z^ZKNc*K#v@JbZZ=^&-LE(;6k9>R%fS_UoH!ENB7Ezl>`doPN0I+`=DC&*=i}*wK!g zRG$X4wf0ORU7xGN2J**wc-C}wOnxS4nL&vLbiKj5a1Q5^Np;X~R1C~{6}REX&#y3o zp272I@`v%$1hihjUSb$krpz2u{ z);e19K&*AkAqF2lexvXDz`L?CjKc`GqHWpi;zY^-kurW;g`uJ-)79kZu?pd5yPm&i zwry#7OvdwFm~h;ZNaOM4M0zEV1>aIY&Bvnt&Y$4>A_>(Ze8H)Ychk+w4R2AU?lqp0 zsMShJ3_!^_huVlSr9iQXve&hv12}R`(9P_NU%go4Ahy9`W06rrXjmL<(gvhPq8wE5 z=QC)sO5i&*W`3ydNKL|S4Q~UHdWGDDsRzAH$VvTcyp1cAPFwu~Jw6)3D_vTrN2H=7 ziH8rSWK3iKgc@&@e?M0jk`mX6<~;bau^%5C#6$j>L=&nS&ygw(TqEo0=77^WA{hS#KM^A4j#p+bg9<(7dHLd}6pgXyXR+$*yj7(x9 zZn9$~&h$W?sl7_W=$ofHe_3G;ZBN{gRz+gg%sVbu-?QEzsWaURv}`gpX;90RaFg}H zA#Ph0!QLM`OiIvj>u7-8bfz}m^b)Qu2QT)g{|J=~rd=%7~KkX+A8(00?g<(`T;(prq zVca($*e_ZNsZS*T*J8s&oslW?jNj8gPgIZ740WcEvzd~K4!9_;3{#xM88aQ08|utn zL&r!JLAqRPAwdOc?D$4SDX?s~;)?<=eEPYW1;3Q1Fa`Tte0HleL^zefG_o-Ipo&Ca zvE@nBLV(MDJ%~S^z5fC8s%5aI&b5CtQIm>YPxfDWhgJuGn>>KYc$>)lxb^4vATk?b zeMeRvjg`~8py%A*ne~EE(M3~iY`?R2ZHU^4ifyQzY9^dv6}RArN;?Qy2L@I6?ygvB zYkDS)mwv{j{wAK~RnV9mF$rLF%e~YzppQMu)R7m_vLOO_Fbz^7IpY!F_pZ<42=T{? z)T2?*eg80aIQgrzloc7J$hpb6`;H=Y^wqB}fqc8x1eH21Gvd{zSRV5kHft5J4}r-_ zvk5;Ja=_& za%bc*3w9}WNKJ0bhd1e!WpIHPZsv&ak!fsoQS>MMPE&v|q zP9IaB^(~uLM$Aw@9ea&aBiy6`Gs<=>Qb&}8GZDZ_oDLva)8N((>N}LXpvZ=PtdEDU z|90EpMpmvjC0dS`>t~3$QVv*HI09ONsc>5DB}I#6{~Q8J?w!EG-Y~eYZbqwEBZ_zg z5$5CV$%J1G>vH8?rePVf&fuNdF-JD>fdP#jChknBwG&Votw9yZ#+O(|ICQ^VqJx=w z!>dZLIYW2j7yh+js^m?YTfwB#i{_| zduwTOohTi-(Sb;lU(|$L!);bJ6|6_s>Y`K#u}Qs1$_0JViSYxHZ_H(KO#`kH47q$K zoy3dw$f=J64&)9^%z%^~p?Q0ARSApbpKBQ)xORLYe+f%i)@BT@y-l{300-oHNRgU% zsPNxhS2l1mdEwP=_y(Gfb8^-dYU$pda_@foa?_j^q|GKxg}iH4DO}p)M09+#SFxx zq_Ih{aj}Na`n5>6xMI@lfsF*6IqQ!9P|-LR7on%v)046h_}+d%RI% zd^E4x5|8o7b*RYU$EN10l5X$I|HM1eQ^Q7tr0XqJ0a*#w9XZbbyDibhh{x{p8AQ~U z2c`>uDGvp3D5BAGpNWZDQ9ISCI-U=6rJbDiINxiVGXGnzTA1eM2rjrGA`agc6jp>XwJ;& z>g4vnhIf0GeAen&bnR}`_Q62}_9j1iZVH{`gyMyMS4t{c#3Zf{dUq*lOh^sY^e4( z0UaA^G8SpQx}L*WX8qlM0jx1!hRHYW`$Ufjm44lfxA3^p;jPTxDul;&Uu$dN+4CZW zkHVB_c?~mUnT3x`%6-Mcj%{Bj8WG`vSxQ$-1Dc}p<4#f-P$0^qqdP=P7v3$b=Wg{G~6*S0)T)Amec^cmhH73knnI zWk}9%1fp!cnu~dL6B$2guTjPd!<{hYFCh5T&ktuKp%>TqEH!B%&fLA&x^6JeR9k(L z*S74*2Ge0<&j2%U= zwm*Gp$SX3<(K0RBS#3vM!2F)D$J)HZ(vd^wjgfgJ!sxht`yT~%=3~_H%I$x{l)T2v zyp^~_3*hqgSj-}emn2L2e92m;el((9b~|c!5r8JIWRFI=f>fmAA2F#&<<)5rU@f(T z`o3t);gGYi8C;(_p*zo!Mw%%G*6*s}rSYZg9T~3H6Qr*g?Pr`%DB`ISOvEPRRDbsk07 z4$h7840q$j?z2;DHB-h^l5$ZmTHIJ_b;FPcF2P!Ys%&N{T8)D+*6SPFRaWJifQi;K zX_e~rl4PKb0UOFGP{LU!*v1?S6_ip6~KyaosL*txNEHV zD)P&I9K3WT37eGm-?N;^l?UI{cOTQT@jhiol--(H{fehe@p+0Z>m5X?u4r|V* zTY{e|@Wr(3I5Aw94&*-q3mR)QWxyIRlO!zBrjEC9c!ug-g0QJQvE3% z@Y2z($OU=ty>92LPKOgLQ7O}}J2nj2LLZr+r_PmfNKY7gud3v6PqdNn!p}3*iwi#^ zxnTiZ%YPP6<;$O0LIt$InnSa7_2DMtxI;4Ld2*es!XjhUX$>$XA7Kp4@3Kch8Nk}b7 zd?m?-8b^uCbn{5Rz`>n|TOQ9)jO$v!jN0!%bu zUNUB=Y7nY_TrBHvCzx0MA(S>n_NY}&1PRG0g#pfvdC9}#vpVGPp!PwE$rTa&JL?GD<(BEhrH=?CuJu*;zR1x^FGC zDGy85gQCOfV*cXS14q*2vrWQPdnh4@HLIQ&6=qeEP8uuc!R_sOd$=jsu1yp)+8A5A zw~j|n+RBQ0>m3AUb17lFO2~ZQqs}MP7>&CVKDst?_Kn&e_V0KeyHdK1p{(P|imq)U zZ^Dfdcq!BYnR@z{2}ES}Rq+rpV2mv70txetDdCX7QLOfrX1`NDHOnn+E}YiXRpjHH z9<0iQyftV=APGdo2TG;1EEeqnS0Le9t+x!mv|xsD$Gwmg#iSQ;o;9tVU(J5n4z;WU zW%`Hvwgg~(c0w<+`l61)(9PtI*}&y5+m`$O#-63ztN;kSN2=NeGd<*$i{aPGJk>r4 z(Xc{DJtZ3mYNO4FWi%j9$Z#u2XHPHoQQ%OkXH_B`nz>3w3smH=!7L)vkX`-&+E#C+dh8CO0WlRyONc$->ew9dFtRP{8zF;&`3Z86=iIOtS$s6PF5 z@mDg0<}c_ed}+p;d{jC=qLXTzS=buIrQ3;C3Ecc>B%?WcM+nO-Xzo z5cqw>q>&Xe$MB>Z&^on3*`<^Fxk>-oPUfw&HXb{l?H8fcVsF+9g=r>VUI|xUuU-7t zCFN9iv8m}V=*RoM2kY}JT2TbQ8{tMY88Nac>^9-{qY1KBLC6JzD`bi-A*MloZph%M z+uB?4N7Te_<4hwCPt?C?+`M~1g)r0<@};E3ie)CJw~{fvdDg2hY6nnCZ`U9NlCQ6` zcST~5ceq6=lumZ)c)845Xp-6yPAj!FY1`3E=9(C~+s19YU#6bFYjj(dtkre)E6t6W z{)9YAo|^1u{gT@D<7RMHe5s}^| zc}n))XV4qb8E%SyH!WD_?9(AtorrLkb+XH*Tzq*;b;xuqQp;zwW3MV4Vz%wle2zrP;BRVU1h4Fu4cT`rfbQ0N9VCj2pMQLRtY zVC77D z)pNOhxlB(irpQn+G2sAiy~|lTorSSz4`zTc&o#PgRSCLuej{c|oY*c^^I@cc_~IDV zm1{n3RZ=npO$#$3yzaQ{L+4>qZ_`h(HQ5tT|47xCd>aP0EH%la(45sz6y@5})2CDq zwtC3kxEZpXNL7gwHCDTd3{Iedf4>83eIgHN6+X7F+Zb8xepXVtRkU>XkRD^VnE>NTha~=N8^rb(LCX^pAd0ShRdS==46@CvQ8nZrd@;+<}Y7%I=u zOIe?lz4oZ&>*@w3Omqy9FL=2e3>q~z|BCpj{Dp56_GibY;b!jE!nF2C!oPGLB0y7O zaH$;{c3f=wT-H6XyQI$wc6X; z6>@@wE|(=`j@rUvB7cqyao43syD(N+0Nk)_q8qy4oc#gNM5>g>#gH(WelRtb9nQVi z`%I^0&Zv?b1ipfC-7E^t4X$lLv9k)=E&szp4~0d%J>olR9iRMCrWopVJK-4DZ$=i% z1a0?k3)ZBW=(FW3Q3hrGO6T2|nxRMfC$xro-ZmDo*be5<6Y#{d)?nyS)0K-N;3Qxg zQ?&?lR`3p46itF2#vB=1bPt=+ey-wyvWCg~R!KqE_bqhc|A>lQ`PCT>DT@ z=bWefUKOOerqOFx%zqtG$1*keK}?bmF!$RTzQg6~t|Vdil%a2-AnAEMHDH#58)}T4 zmnC?$QiBc`V;FaSA;6F+DQY+uWBH%04tl=WORep$PV=No(98oXOBM_EVIa?5BH-kc z=uw+TlK)*flCMnK?AF={+O(*M1|>L3-jP}Wg0qBSU(%`UP#b9o`K*aI`FyJ$oX2WZ{fMHcN8@ zH2JQ$ptmX!tu=orh(2=^Bv+QV%qM3i`<#-J_^+B8QYz(^*GE|)=t3M^3aF_EN&REw zH3~2OF*3@)uYjoVh=WauKXzqIb=c5AYhVgh&e7bs-as_)J-kEW+Oy1FCuaDU`PDS z?fjI#aQQVqcyqEmmX+d2YRnt%UvgGjv7QdD!~CXMN%7>fP%2d)%xSuN)Vd0HPKl9T zRVmMrYUVvLaO@LqW=2(han{rHtPaRMMFP^;EYP27HPMXP=@xdT-G;D?Pn~FZ7>j8u zY?`(3OINmmM4L5VjiXD*^ZPw;ddDVkZaNyL5EN8Tu(9H<#s_9CttLSP8KS~8^q4J= zCs0NiwnESpPzp~mBie*G9v<3tB7v3}O<9y%6ljsfy1)g+GrFK-^h2XU;g2G8o|nSy zASWmpx~`hDNiU-?s>XEGWiI}n0aD4jG9k73_v1n<8#`NHv@S1Cqvyv8Gf#TmQPIw( z6A!zG%Fy3MaxJaY;*J^u#TiF*omO0WXAG3W^*(}BJA?7Jc=w|)WSAu{j&R0c*>K^Y=JgIYbw3PGJ9O52 zsZ@k4_~LOc7Rnz9Z8PJ?j!e=!_{8ECPw5&uDQY*_b$bO+F@>h$mOL*tcUw*Bc#5!D ztR&oMnkjlRLT?(0+T#j{z2Y!pM2oaFQhSuLYlEkNd)}5mV;zXEz2nmITwDp8)h`wt z`0wGEw&=9i?dP`Q?;6vNQ4b};i(ooZ0^KBR3d9i`eubg9$VXL%1}xe5yRonyj#C+` zV7SsR-SZNiI~R3(Mk_APQJl^n2gpqZ^?qORz+@pZ>Hdsu#&!|qtNN|w{zIsmALnTY zfR?q|N%VxH7&gVwX0FojiHE!m^g%92hE_x5d0idz%1pyyL6zZTZdpuQ;hJeARM#xr zPp~p__~cUlH;#=->FsK)yAhR{?lSt(&d??m-wFyrWA^7ZK|NOMeWv4TQJJGHJ8?se z%S;@FK0QX>LQC{ifQ3=O>CEcf7?c+wxUO+TU+SgS;1SS=K}OEW3$kR5vN@0?L2$VJ zb1#Sp6(!|xvNSp_SdHCzv;vA%A}LQug5SKj36W>&MH*6CB~#HndMtJrl#h7sRzKq5 z=}-t1wO@Ge|I^XpWx_4}j6aiVMEPh`A~uRPKLAV8clicWYd|;HwC0P$U|z9^@lhK- z7gm+Urh*dEh_zcDHI2I@bdN(c4yko&JBZQ^UZ#7z)r+e3`mdm;!&1K`KL68uxo8?w zljCb&kg@c_Wi-x%q=&3t9-xyJoo#@`;jzy3ZS*mr@)3$ z%mbZCbbMn(oTJ7ELhvZSZGSYt)hxu$XYwq*;r94s*}2$oxPTjW@PD5l*UF#@3-AZb%Bd1G`90P8{vHLQe*7B7jU#O^s7w z@?0w%VJ;X3Q-*BTb4$60Fmh-QC}tFuJ%T+PF*G<5&Y<+$DkaN4TD+N_;sZ(hbJ zwKZnTrjc?w!xS>*!_S}`aW(LTC~)QoFbf0BfV`GY^yy@U5}(&6A6xAucKNx$NJeuF ztgr!UV3(WRNRTV_;tHeE2K0oZ919?h{{p;dB*+xculpZrDF8OdUs@um4GS}-U1in0 zN6l!Yj5-UjLteNMPw(mHPGv;>Ta$TX=tbX$pFC?EXA(+i5PWIqDQwig@8g*D13{Je z)V$-N{#wXv?Cf71w&Ami*Njh0BAV3hzDDJg6OB=ftHAX@%fQG|h<{QL25?R!79@T1V@u1kge zfp$cp+gdP0=mp%5T5Jn?a+T=XM}!1QwD>b!Uc-bGw+vJw)tevnf>il6`}&=z6TY8c zKv&stegu0|nvcQejD17!H9uatC?CIzPsI8}XzW|D?9?Ft@ksi;fQ5h^M>l5d@fq5O z=l8&Qd+; zrS%eLTdr<`%%BrX%XEBDU1Be*{S)PO;3H9z?^sgU&8#aSoT2coNEuM??k0HTVFu5f zr^7`2{aOfZdE9^z?3IVd!q)#0`dML+E6r1bP7Q@SVT~o_hMI2SG3MCfGQI}dZps;5 za1b9nN%D+SJgKApPvOqRoco)%++AJg=;F>La4uhAKl&j1XI7-|X-fC>G%`|Zbx3L| zoL&}qTfiRW&i1x@9>4CJTxwK&WvLMLDjy)a1XqJe|AX$@F^}U2X}LaipTjKjh@~pp z7TjI)R6qQ#rSvp>TNIig8vTgJ@-n}1V<^pHbM>z?aaEb-77Fd0U)u1m+KW#$F3OS* zu5lFj$CitW&EJYcz9ZfGDY|LvC!`uOgg?1PnoAF^Vnn94_01Vmff**K$arzL5sHTV z9J{!STU)J(XAHTl*|;8)Cnu^Sd%=_y8SL00)G1nZ2@#u>y&P*YX%>e8nc-#a^EMS} zmXRn@RrF)uBWf&5+1s)6^ki8@>a->os&)a(E;<&j`s6#+i$c2Im5x&*8zbZ;iLvedv+zeI z#k8H~A5U3~HFY_WYWrRXMG}Dc0u+XdVqh{v(5piGH{GoJc8e*qhIF-d`S-z8SEQ<0 z^K1keoH0tCES&O#RCO`0<*97D=nZo@=bO5@of^M|Px2hxzklbUSJ-4bzF5J>*cpV- zWfe@Z=(Knqvs6q(vuI4=>Aqk+tHk}TEZ?BaH2DoHrzyi_giT~a-3x>^#s+D3x>!(< zO0m56{oU}^pzOE`_;@6DTz*D3 zWa;Q;`0^1|xzoNqbE7k$c@%d;cdu!J@AY;6y7rurja9JmQo~^iw4dF^(_%|Sg!^dW zc~@eTQ^4s)r@Df!ycBC(XVPbreb7BOhZ%X;)@{*_F3?1(2}V*FYe9FbPZ8VaO8NHK zSgt>EXMvh;K?9y81B`*Q)j>8@He;X^KaGZWbSDW$0 zI5-;d2px3^EpB%Ljpy^g!(RNFjGfQ!Xv`G`EqItwD6Hm1%=0uOvJuUQ-blvsb*fNb zE<8+yC|2_p`*i0Ch3H<<>z<0SCnE#hd%&ZORD~+i(!Fpeu~|q+8n1or-jXL^rd>PD zMwrmguZ`;6QBo$R^F1CVp-PLUv!EBuGhWx&tIP z#wh!4y#yRZnI+Hww>j%wxmJUJNOhH>skQiPaGfXOk=`zG^c%6aq8ln=V-lDdI;>Y^ zPtN=Mfg6SiIM>=Z+fp?&!@d-!PtC11bozPVzfHQ%Vc`xzC;|n(#S4dUQ7(7C12t4h zU(_t-i8Eo7Idroce259zIM^`rwKGEu5bKp-b{VbAT(0-6hx~x_Rt3_;C*;kvk^cc| zF4ef!qX~4h*FI0nP!^3eVsgktq4rc&5*EEK_9*e4N41mhnNm^k@Z^uf8z>t<(Fpt@ zuCsxC<~OLHC5O?1mcv;3spxKIWw(^8Tyn}T{(duT?bPu<8~}E`h(10Y?3<*5Q|Vu` zR8K{7Oj-u#-cF-+cQDMiwS zs{4D%=eg-ba9ywb|EPF(2%Es-M$N5av}JU2d7tez=W?a`3%H|v6!6Q=M9)ZN}*djouk=j=~Ff~~{yoxufE zAo*+4(Pg%gCv2Q=G&HGovSU9$QH$2Ce@vHfDVWHyYGu-AMGkx4dl4QgZhLidCz@R- z^4G1y+bcM-_(5I$Qk@>V#Tvp*w9H7FQ^eRp%OMl|=vbu>M%#OV$&%F=rA_2D_9fMO zG+w*iGxgFVOmV0*#3CYqVf9Vj&f-z`!UsFpRUyr5`}$)qX@Ax1%8GIMV}XbwDf!3{}ZF^3$kWe!VC&DJHY0wQIR@m0d9MG&KG0`bI9X%nD^~lJaaZ|$BY;v=Pl^o zyBd;ZlV~kq;`_38Mz86@WSb&A?dZghA3vpK`~+_<i0wG-NhC5d#jM)qoQZgcPL)u>K)k`jdJL-(E>TQd>u<*UObunI3uflDK|OBSV$r3|ER!N z(0=U*WC^yTOsbw40c}20lxnZ>0-v{?_2f!3{>Q-Ztgso6cST!TNl8k>?U9K0e@Vs; zdPG~E@$otGbHm$Z7wduNX#Wqv*#A#zHu>R&)HYx0TOlFb=MT__jD(_im8em`{|ANw Bm6iYi diff --git a/open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg b/open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg deleted file mode 100644 index cc73d30b3..000000000 --- a/open-machine-learning-jupyter-book/llm/image/transformer_architecture.svg +++ /dev/null @@ -1,118 +0,0 @@ - - - - - - - image/svg+xml - - - - - - - - - - - - - - - - - Encoder - Decoder - - diff --git a/open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg b/open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg deleted file mode 100644 index d8d59e5df..000000000 --- a/open-machine-learning-jupyter-book/llm/image/warmup_loss_plot.svg +++ /dev/null @@ -1,1579 +0,0 @@ - - - - - - image/svg+xml - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - diff --git a/open-machine-learning-jupyter-book/llm/introduction.ipynb b/open-machine-learning-jupyter-book/llm/introduction.ipynb index a53688e89..01464a46f 100644 --- a/open-machine-learning-jupyter-book/llm/introduction.ipynb +++ b/open-machine-learning-jupyter-book/llm/introduction.ipynb @@ -54,7 +54,7 @@ "\n", "In the narrow sense, a large language model is described as a probabilistic model that assigns a probability to every finite sequence, whether it's grammatical or not. This perspective emphasizes the probabilistic nature of language models, indicating that they can assign a likelihood to any sequence of tokens, regardless of whether it conforms to grammatical rules or not. This perspective highlights the fundamental nature of language models as probabilistic models that capture the statistical regularities of natural language.\n", "\n", - ":::{figure} image/implicit-order.png\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/implicit-order.png\n", "Implicit Order\n", ":::\n", "\n", @@ -108,7 +108,7 @@ "\n", "6. **Decoding and Output Generation**: Architectural design also encompasses decoding mechanisms for generating output text. Techniques for output generation, such as beam search or nucleus sampling, influence the overall architecture and performance of the LLM.\n", "\n", - ":::{figure} image/scaling-laws.png\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/scaling-laws.png\n", "Scaling Laws of LLMs\n", ":::" ] From e33cd46aff47a5248429803d9029f20fbaf22a40 Mon Sep 17 00:00:00 2001 From: peng_windows <2686728826@qq.com> Date: Fri, 29 Mar 2024 06:39:06 +0800 Subject: [PATCH 4/6] update attention --- open-machine-learning-jupyter-book/llm/basic/attention.ipynb | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb index 792958402..33bdae25e 100644 --- a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb +++ b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb @@ -86,7 +86,6 @@ "id": "a55aa49c-36c2-48da-b1d9-70f416e46a6a", "metadata": {}, "source": [ - "- No code in this section\n", "- Translating a text word by word isn't feasible due to the differences in grammatical structures between the source and target languages:" ] }, @@ -128,7 +127,6 @@ "id": "b6fde64c-6034-421d-81d9-8244932086ea", "metadata": {}, "source": [ - "- No code in this section\n", "- Through an attention mechanism, the text-generating decoder segment of the network is capable of selectively accessing all input tokens, implying that certain input tokens hold more significance than others in the generation of a specific output token:" ] }, @@ -179,7 +177,7 @@ "source": [ "- This section explains a very simplified variant of self-attention, which does not contain any trainable weights\n", "- This is purely for illustration purposes and NOT the attention mechanism that is used in transformers\n", - "- The next section, section 3.3.2, will extend this simple attention mechanism to implement the real self-attention mechanism\n", + "- The next section, we will extend this simple attention mechanism to implement the real self-attention mechanism\n", "- Suppose we are given an input sequence $x^{(1)}$ to $x^{(T)}$\n", " - The input is a text (for example, a sentence like \"Your journey starts with one step\") that has already been converted into token embeddings as described in chapter 2\n", " - For instance, $x^{(1)}$ is a d-dimensional vector representing the word \"Your\", and so forth\n", From da5bb23c0ab351065e6a8acf0615a8134edb5d15 Mon Sep 17 00:00:00 2001 From: peng_windows <2686728826@qq.com> Date: Mon, 8 Apr 2024 14:56:56 +0800 Subject: [PATCH 5/6] update llm --- .../llm/basic/transformer-architecture.ipynb | 85 ++--- .../llm/basic/attention.ipynb | 311 +++++++++--------- 2 files changed, 184 insertions(+), 212 deletions(-) diff --git a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb index 208716078..407963ddf 100644 --- a/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb +++ b/open-machine-learning-jupyter-book/assignments/llm/basic/transformer-architecture.ipynb @@ -47,7 +47,7 @@ "In addition, the decoder includes an additional attention mechanism that focuses on the encoder's output to incorporate context information during sequence generation.\n", "Overall, the encoder-decoder architecture based on the Transformer structure allows for effective semantic abstraction by leveraging attention mechanisms, position-wise feedforward layers, residual connections, and layer normalization. This architecture enables the model to capture complex dependencies between words in the input sequence and generate meaningful outputs for various sequence-to-sequence tasks.\n", "\n", - ":::{figure} https://media.geeksforgeeks.org/wp-content/uploads/20230531140926/Transformer-python-(1).png\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/Transformer-python-%281%29.png\n", "Transformer-based encoder and decoder Architecture\n", ":::\n", "\n", @@ -62,7 +62,13 @@ "\n", "The Embedding Layer in the Transformer model is responsible for converting discrete token indices into continuous vector representations. Each token index is mapped to a high-dimensional vector, which is learned during the training process. These embeddings capture semantic and syntactic information about the tokens.\n", "\n", - "Implementation in PyTorch:" + "Implementation in PyTorch:\n", + "\n", + "We define a PositionalEncoder class that inherits from nn.Module.\n", + "The constructor initializes the positional encoding matrix (pe) based on the given d_model (dimension of the model) and max_seq_len (maximum sequence length).\n", + "The forward method scales the input embeddings (x) by the square root of the model dimension and adds the positional encoding matrix (pe) to the input embeddings.\n", + "Note that we're using PyTorch's Variable and autograd to ensure that the positional encoding is compatible with the autograd mechanism for backpropagation.\n", + "Finally, the PositionalEncoder class can be used within a larger PyTorch model to incorporate positional information into word embeddings." ] }, { @@ -148,19 +154,6 @@ " assert math.isclose(output[0, 0, 0].item(), expected_first_element.item(), rel_tol=1e-6)\n" ] }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In this code:\n", - "\n", - "We define a PositionalEncoder class that inherits from nn.Module.\n", - "The constructor initializes the positional encoding matrix (pe) based on the given d_model (dimension of the model) and max_seq_len (maximum sequence length).\n", - "The forward method scales the input embeddings (x) by the square root of the model dimension and adds the positional encoding matrix (pe) to the input embeddings.\n", - "Note that we're using PyTorch's Variable and autograd to ensure that the positional encoding is compatible with the autograd mechanism for backpropagation.\n", - "Finally, the PositionalEncoder class can be used within a larger PyTorch model to incorporate positional information into word embeddings." - ] - }, { "cell_type": "markdown", "metadata": {}, @@ -168,7 +161,10 @@ "## Attention Layer\n", "The Attention Layer in the Transformer model enables the model to focus on different parts of the input sequence when processing each token. It computes attention scores between each pair of tokens in the input sequence and generates a context vector for each token based on the importance of other tokens. This mechanism allows the model to capture long-range dependencies in the input sequence effectively.\n", "\n", - "Implementation in PyTorch:" + "Implementation in PyTorch:\n", + "\n", + "The MultiHeadAttention class defines a multi-head self-attention layer.\n", + "The forward method performs linear operations to divide inputs into multiple heads, computes attention scores, and aggregates the outputs of multiple heads." ] }, { @@ -264,16 +260,6 @@ " self.assertEqual(output.shape, (batch_size, seq_length, d_model))\n" ] }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In this implementation:\n", - "\n", - "The MultiHeadAttention class defines a multi-head self-attention layer.\n", - "The forward method performs linear operations to divide inputs into multiple heads, computes attention scores, and aggregates the outputs of multiple heads." - ] - }, { "cell_type": "markdown", "metadata": {}, @@ -282,7 +268,10 @@ "\n", "The Position-wise Feedforward Layer in the Transformer model applies a simple feedforward neural network independently to each position in the sequence. It consists of two linear transformations with a non-linear activation function (commonly ReLU) applied in between. This layer helps capture complex interactions between different dimensions of the input embeddings.\n", "\n", - "Implementation in PyTorch:" + "Implementation in PyTorch:\n", + "\n", + "The FeedForward class defines a feedforward layer.\n", + "The forward method applies ReLU activation to the output of the first linear transformation, followed by dropout, and then performs the second linear transformation to produce the final output." ] }, { @@ -347,16 +336,6 @@ " self.assertEqual(output.shape, (batch_size, seq_length, d_model))" ] }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In this implementation:\n", - "\n", - "The FeedForward class defines a feedforward layer.\n", - "The forward method applies ReLU activation to the output of the first linear transformation, followed by dropout, and then performs the second linear transformation to produce the final output." - ] - }, { "cell_type": "markdown", "metadata": {}, @@ -369,7 +348,10 @@ "Layer Normalization:\n", "Layer Normalization is a technique used to stabilize the training of deep neural networks by normalizing the activations of each layer. In the Transformer model, layer normalization is applied after each sub-layer (such as attention and feedforward layers) and before the residual connection. It normalizes the activations along the feature dimension, allowing the model to learn more robust representations and accelerate convergence during training.\n", "\n", - "Implementation in PyTorch:" + "Implementation in PyTorch:\n", + "\n", + "The NormLayer class defines a layer normalization layer.\n", + "The forward method computes the layer normalization using the given input tensor x." ] }, { @@ -434,16 +416,6 @@ " self.assertEqual(output.shape, (batch_size, seq_length, d_model))" ] }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In this implementation:\n", - "\n", - "The NormLayer class defines a layer normalization layer.\n", - "The forward method computes the layer normalization using the given input tensor x." - ] - }, { "cell_type": "markdown", "metadata": {}, @@ -459,7 +431,11 @@ "Multi-Head Attention sub-layer that attends to the encoder's output.\n", "FeedForward sub-layer. Again, each sub-layer is followed by Residual Connection and Layer Normalization.\n", "\n", - "Below are the Python implementations for the Encoder and Decoder structures:" + "Below are the Python implementations for the Encoder and Decoder structures:\n", + "\n", + "The EncoderLayer and DecoderLayer classes define encoder and decoder layers, respectively.\n", + "The Encoder and Decoder classes define encoder and decoder modules, respectively, composed of multiple layers of encoder or decoder layers.\n", + "These classes follow the architecture described in the text, including the use of multi-head attention, feedforward layers, residual connections, and layer normalization." ] }, { @@ -579,17 +555,6 @@ " return self.norm(x)\n" ] }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "In these implementations:\n", - "\n", - "The EncoderLayer and DecoderLayer classes define encoder and decoder layers, respectively.\n", - "The Encoder and Decoder classes define encoder and decoder modules, respectively, composed of multiple layers of encoder or decoder layers.\n", - "These classes follow the architecture described in the text, including the use of multi-head attention, feedforward layers, residual connections, and layer normalization." - ] - }, { "cell_type": "markdown", "metadata": {}, diff --git a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb index 33bdae25e..9efd1058a 100644 --- a/open-machine-learning-jupyter-book/llm/basic/attention.ipynb +++ b/open-machine-learning-jupyter-book/llm/basic/attention.ipynb @@ -54,7 +54,7 @@ "id": "a2a4474d-7c68-4846-8702-37906cf08197", "metadata": {}, "source": [ - "- This section covers attention mechanisms, the engine of LLMs:" + "This section covers attention mechanisms, the engine of LLMs:" ] }, { @@ -62,7 +62,7 @@ "id": "02a11208-d9d3-44b1-8e0d-0c8414110b93", "metadata": {}, "source": [ - "" + "" ] }, { @@ -70,7 +70,7 @@ "id": "50e020fd-9690-4343-80df-da96678bef5e", "metadata": {}, "source": [ - "" + "" ] }, { @@ -86,7 +86,7 @@ "id": "a55aa49c-36c2-48da-b1d9-70f416e46a6a", "metadata": {}, "source": [ - "- Translating a text word by word isn't feasible due to the differences in grammatical structures between the source and target languages:" + "Translating a text word by word isn't feasible due to the differences in grammatical structures between the source and target languages:" ] }, { @@ -94,7 +94,7 @@ "id": "55c0c433-aa4b-491e-848a-54905ebb05ad", "metadata": {}, "source": [ - "" + "" ] }, { @@ -102,8 +102,8 @@ "id": "db03c48a-3429-48ea-9d4a-2e53b0e516b1", "metadata": {}, "source": [ - "- Prior to the introduction of transformer models, encoder-decoder RNNs were commonly used for machine translation tasks\n", - "- In this setup, the encoder processes a sequence of tokens from the source language, using a hidden state—a kind of intermediate layer within the neural network—to generate a condensed representation of the entire input sequence:" + "Prior to the introduction of transformer models, encoder-decoder RNNs were commonly used for machine translation tasks\n", + "In this setup, the encoder processes a sequence of tokens from the source language, using a hidden state—a kind of intermediate layer within the neural network—to generate a condensed representation of the entire input sequence:" ] }, { @@ -111,7 +111,7 @@ "id": "03d8df2c-c1c2-4df0-9977-ade9713088b2", "metadata": {}, "source": [ - "" + "" ] }, { @@ -127,7 +127,7 @@ "id": "b6fde64c-6034-421d-81d9-8244932086ea", "metadata": {}, "source": [ - "- Through an attention mechanism, the text-generating decoder segment of the network is capable of selectively accessing all input tokens, implying that certain input tokens hold more significance than others in the generation of a specific output token:" + "Through an attention mechanism, the text-generating decoder segment of the network is capable of selectively accessing all input tokens, implying that certain input tokens hold more significance than others in the generation of a specific output token:" ] }, { @@ -135,7 +135,7 @@ "id": "bc4f6293-8ab5-4aeb-a04c-50ee158485b1", "metadata": {}, "source": [ - "" + "" ] }, { @@ -143,7 +143,7 @@ "id": "8044be1f-e6a2-4a1f-a6dd-e325d3bad05e", "metadata": {}, "source": [ - "- Self-attention in transformers is a technique designed to enhance input representations by enabling each position in a sequence to engage with and determine the relevance of every other position within the same sequence" + "Self-attention in transformers is a technique designed to enhance input representations by enabling each position in a sequence to engage with and determine the relevance of every other position within the same sequence" ] }, { @@ -151,7 +151,7 @@ "id": "6565dc9f-b1be-4c78-b503-42ccc743296c", "metadata": {}, "source": [ - "" + "" ] }, { @@ -175,20 +175,20 @@ "id": "d269e9f1-df11-4644-b575-df338cf46cdf", "metadata": {}, "source": [ - "- This section explains a very simplified variant of self-attention, which does not contain any trainable weights\n", - "- This is purely for illustration purposes and NOT the attention mechanism that is used in transformers\n", - "- The next section, we will extend this simple attention mechanism to implement the real self-attention mechanism\n", - "- Suppose we are given an input sequence $x^{(1)}$ to $x^{(T)}$\n", - " - The input is a text (for example, a sentence like \"Your journey starts with one step\") that has already been converted into token embeddings as described in chapter 2\n", - " - For instance, $x^{(1)}$ is a d-dimensional vector representing the word \"Your\", and so forth\n", - "- **Goal:** compute context vectors $z^{(i)}$ for each input sequence element $x^{(i)}$ in $x^{(1)}$ to $x^{(T)}$ (where $z$ and $x$ have the same dimension)\n", - " - A context vector $z^{(i)}$ is a weighted sum over the inputs $x^{(1)}$ to $x^{(T)}$\n", - " - The context vector is \"context\"-specific to a certain input\n", - " - Instead of $x^{(i)}$ as a placeholder for an arbitrary input token, let's consider the second input, $x^{(2)}$\n", - " - And to continue with a concrete example, instead of the placeholder $z^{(i)}$, we consider the second output context vector, $z^{(2)}$\n", - " - The second context vector, $z^{(2)}$, is a weighted sum over all inputs $x^{(1)}$ to $x^{(T)}$ weighted with respect to the second input element, $x^{(2)}$\n", - " - The attention weights are the weights that determine how much each of the input elements contributes to the weighted sum when computing $z^{(2)}$\n", - " - In short, think of $z^{(2)}$ as a modified version of $x^{(2)}$ that also incorporates information about all other input elements that are relevant to a given task at hand" + "This section explains a very simplified variant of self-attention, which does not contain any trainable weights\n", + "This is purely for illustration purposes and NOT the attention mechanism that is used in transformers\n", + "The next section, we will extend this simple attention mechanism to implement the real self-attention mechanism\n", + "Suppose we are given an input sequence $x^{(1)}$ to $x^{(T)}$\n", + " The input is a text (for example, a sentence like \"Your journey starts with one step\") that has already been converted into token embeddings as described in chapter 2\n", + " For instance, $x^{(1)}$ is a d-dimensional vector representing the word \"Your\", and so forth\n", + "**Goal:** compute context vectors $z^{(i)}$ for each input sequence element $x^{(i)}$ in $x^{(1)}$ to $x^{(T)}$ (where $z$ and $x$ have the same dimension)\n", + " A context vector $z^{(i)}$ is a weighted sum over the inputs $x^{(1)}$ to $x^{(T)}$\n", + " The context vector is \"context\"-specific to a certain input\n", + " Instead of $x^{(i)}$ as a placeholder for an arbitrary input token, let's consider the second input, $x^{(2)}$\n", + " And to continue with a concrete example, instead of the placeholder $z^{(i)}$, we consider the second output context vector, $z^{(2)}$\n", + " The second context vector, $z^{(2)}$, is a weighted sum over all inputs $x^{(1)}$ to $x^{(T)}$ weighted with respect to the second input element, $x^{(2)}$\n", + " The attention weights are the weights that determine how much each of the input elements contributes to the weighted sum when computing $z^{(2)}$\n", + " In short, think of $z^{(2)}$ as a modified version of $x^{(2)}$ that also incorporates information about all other input elements that are relevant to a given task at hand" ] }, { @@ -196,7 +196,7 @@ "id": "fcc7c7a2-b6ab-478f-ae37-faa8eaa8049a", "metadata": {}, "source": [ - "" + "" ] }, { @@ -204,9 +204,9 @@ "id": "ff856c58-8382-44c7-827f-798040e6e697", "metadata": {}, "source": [ - "- By convention, the unnormalized attention weights are referred to as **\"attention scores\"** whereas the normalized attention scores, which sum to 1, are referred to as **\"attention weights\"**\n", + "By convention, the unnormalized attention weights are referred to as **\"attention scores\"** whereas the normalized attention scores, which sum to 1, are referred to as **\"attention weights\"**\n", "\n", - "- The attention weights and context vector calculation are summarized in the figure below:" + "The attention weights and context vector calculation are summarized in the figure below:" ] }, { @@ -214,19 +214,17 @@ "id": "01b10344-128d-462a-823f-2178dff5fd58", "metadata": {}, "source": [ - "- The code below walks through the figure above step by step\n", + "The code below walks through the figure above step by step\n", "\n", - "
    \n", - "\n", - "- **Step 1:** compute unnormalized attention scores $\\omega$\n", - "- Suppose we use the second input token as the query, that is, $q^{(2)} = x^{(2)}$, we compute the unnormalized attention scores via dot products:\n", - " - $\\omega_{21} = x^{(1)} q^{(2)\\top}$\n", - " - $\\omega_{22} = x^{(2)} q^{(2)\\top}$\n", - " - $\\omega_{23} = x^{(3)} q^{(2)\\top}$\n", - " - ...\n", - " - $\\omega_{2T} = x^{(T)} q^{(2)\\top}$\n", - "- Above, $\\omega$ is the Greek letter \"omega\" used to symbolize the unnormalized attention scores\n", - " - The subscript \"21\" in $\\omega_{21}$ means that input sequence element 2 was used as a query against input sequence element 1" + "**Step 1:** compute unnormalized attention scores $\\omega$\n", + "Suppose we use the second input token as the query, that is, $q^{(2)} = x^{(2)}$, we compute the unnormalized attention scores via dot products:\n", + " $\\omega_{21} = x^{(1)} q^{(2)\\top}$\n", + " $\\omega_{22} = x^{(2)} q^{(2)\\top}$\n", + " $\\omega_{23} = x^{(3)} q^{(2)\\top}$\n", + " ...\n", + " $\\omega_{2T} = x^{(T)} q^{(2)\\top}$\n", + "Above, $\\omega$ is the Greek letter \"omega\" used to symbolize the unnormalized attention scores\n", + " The subscript \"21\" in $\\omega_{21}$ means that input sequence element 2 was used as a query against input sequence element 1" ] }, { @@ -234,7 +232,7 @@ "id": "35e55f7a-f2d0-4f24-858b-228e4fe88fb3", "metadata": {}, "source": [ - "- Suppose we have the following input sentence that is already embedded in 3-dimensional vectors as described in chapter 3 (we use a very small embedding dimension here for illustration purposes, so that it fits onto the page without line breaks):" + "Suppose we have the following input sentence that is already embedded in 3-dimensional vectors as described in chapter 3 (we use a very small embedding dimension here for illustration purposes, so that it fits onto the page without line breaks):" ] }, { @@ -261,10 +259,10 @@ "id": "299baef3-b1a8-49ba-bad4-f62c8a416d83", "metadata": {}, "source": [ - "- The primary objective of this section is to demonstrate how the context vector $z^{(2)}$\n", + "The primary objective of this section is to demonstrate how the context vector $z^{(2)}$\n", " is calculated using the second input sequence, $x^{(2)}$, as a query\n", "\n", - "- The figure depicts the initial step in this process, which involves calculating the attention scores ω between $x^{(2)}$\n", + "The figure depicts the initial step in this process, which involves calculating the attention scores ω between $x^{(2)}$\n", " and all other input elements through a dot product operation." ] }, @@ -273,7 +271,7 @@ "id": "5cb3453a-58fa-42c4-b225-86850bc856f8", "metadata": {}, "source": [ - "" + "" ] }, { @@ -281,8 +279,8 @@ "id": "77be52fb-82fd-4886-a4c8-f24a9c87af22", "metadata": {}, "source": [ - "- We use input sequence element 2, $x^{(2)}$, as an example to compute context vector $z^{(2)}$; later in this section, we will generalize this to compute all context vectors.\n", - "- The first step is to compute the unnormalized attention scores by computing the dot product between the query $x^{(2)}$ and all other input tokens:" + "We use input sequence element 2, $x^{(2)}$, as an example to compute context vector $z^{(2)}$; later in this section, we will generalize this to compute all context vectors.\n", + "The first step is to compute the unnormalized attention scores by computing the dot product between the query $x^{(2)}$ and all other input tokens:" ] }, { @@ -306,7 +304,7 @@ "id": "8df09ae0-199f-4b6f-81a0-2f70546684b8", "metadata": {}, "source": [ - "- Side note: a dot product is essentially a shorthand for multiplying two vectors elements-wise and summing the resulting products:" + "Side note: a dot product is essentially a shorthand for multiplying two vectors elements-wise and summing the resulting products:" ] }, { @@ -330,8 +328,8 @@ "id": "7d444d76-e19e-4e9a-a268-f315d966609b", "metadata": {}, "source": [ - "- **Step 2:** normalize the unnormalized attention scores (\"omegas\", $\\omega$) so that they sum up to 1\n", - "- Here is a simple way to normalize the unnormalized attention scores to sum up to 1 (a convention, useful for interpretation, and important for training stability):" + "**Step 2:** normalize the unnormalized attention scores (\"omegas\", $\\omega$) so that they sum up to 1\n", + "Here is a simple way to normalize the unnormalized attention scores to sum up to 1 (a convention, useful for interpretation, and important for training stability):" ] }, { @@ -339,7 +337,7 @@ "id": "dfd965d6-980c-476a-93d8-9efe603b1b3b", "metadata": {}, "source": [ - "" + "" ] }, { @@ -360,8 +358,8 @@ "id": "75dc0a57-f53e-41bf-8793-daa77a819431", "metadata": {}, "source": [ - "- However, in practice, using the softmax function for normalization, which is better at handling extreme values and has more desirable gradient properties during training, is common and recommended.\n", - "- Here's a naive implementation of a softmax function for scaling, which also normalizes the vector elements such that they sum up to 1:" + "However, in practice, using the softmax function for normalization, which is better at handling extreme values and has more desirable gradient properties during training, is common and recommended.\n", + "Here's a naive implementation of a softmax function for scaling, which also normalizes the vector elements such that they sum up to 1:" ] }, { @@ -385,8 +383,8 @@ "id": "f0a1cbbb-4744-41cb-8910-f5c1355555fb", "metadata": {}, "source": [ - "- The naive implementation above can suffer from numerical instability issues for large or small input values due to overflow and underflow issues\n", - "- Hence, in practice, it's recommended to use the PyTorch implementation of softmax instead, which has been highly optimized for performance:" + "The naive implementation above can suffer from numerical instability issues for large or small input values due to overflow and underflow issues\n", + "Hence, in practice, it's recommended to use the PyTorch implementation of softmax instead, which has been highly optimized for performance:" ] }, { @@ -407,7 +405,7 @@ "id": "e43e36c7-90b2-427f-94f6-bb9d31b2ab3f", "metadata": {}, "source": [ - "- **Step 3**: compute the context vector $z^{(2)}$ by multiplying the embedded input tokens, $x^{(i)}$ with the attention weights and sum the resulting vectors:" + "**Step 3**: compute the context vector $z^{(2)}$ by multiplying the embedded input tokens, $x^{(i)}$ with the attention weights and sum the resulting vectors:" ] }, { @@ -415,7 +413,7 @@ "id": "f1c9f5ac-8d3d-4847-94e3-fd783b7d4d3d", "metadata": {}, "source": [ - "" + "" ] }, { @@ -449,8 +447,8 @@ "source": [ "#### Generalize to all input sequence tokens:\n", "\n", - "- Above, we computed the attention weights and context vector for input 2 (as illustrated in the highlighted row in the figure below)\n", - "- Next, we are generalizing this computation to compute all attention weights and context vectors" + "Above, we computed the attention weights and context vector for input 2 (as illustrated in the highlighted row in the figure below)\n", + "Next, we are generalizing this computation to compute all attention weights and context vectors" ] }, { @@ -458,7 +456,7 @@ "id": "11c0fb55-394f-42f4-ba07-d01ae5c98ab4", "metadata": {}, "source": [ - "" + "" ] }, { @@ -466,8 +464,8 @@ "id": "b789b990-fb51-4beb-9212-bf58876b5983", "metadata": {}, "source": [ - "- In self-attention, the process starts with the calculation of attention scores, which are subsequently normalized to derive attention weights that total 1\n", - "- These attention weights are then utilized to generate the context vectors through a weighted summation of the inputs" + "In self-attention, the process starts with the calculation of attention scores, which are subsequently normalized to derive attention weights that total 1\n", + "These attention weights are then utilized to generate the context vectors through a weighted summation of the inputs" ] }, { @@ -475,7 +473,7 @@ "id": "d9bffe4b-56fe-4c37-9762-24bd924b7d3c", "metadata": {}, "source": [ - "" + "" ] }, { @@ -483,7 +481,7 @@ "id": "aa652506-f2c8-473c-a905-85c389c842cc", "metadata": {}, "source": [ - "- Apply previous **step 1** to all pairwise elements to compute the unnormalized attention score matrix:" + "Apply previous **step 1** to all pairwise elements to compute the unnormalized attention score matrix:" ] }, { @@ -507,7 +505,7 @@ "id": "1539187f-1ece-47b7-bc9b-65a97115f1d4", "metadata": {}, "source": [ - "- We can achieve the same as above more efficiently via matrix multiplication:" + "We can achieve the same as above more efficiently via matrix multiplication:" ] }, { @@ -526,7 +524,7 @@ "id": "02c4bac4-acfd-427f-9b11-c436ac71748d", "metadata": {}, "source": [ - "- Similar to **step 2** previously, we normalize each row so that the values in each row sum to 1:" + "Similar to **step 2** previously, we normalize each row so that the values in each row sum to 1:" ] }, { @@ -545,7 +543,7 @@ "id": "3fa6d02b-7f15-4eb4-83a7-0b8a819e7a0c", "metadata": {}, "source": [ - "- Quick verification that the values in each row indeed sum to 1:" + "Quick verification that the values in each row indeed sum to 1:" ] }, { @@ -566,7 +564,7 @@ "id": "138b0b5c-d813-44c7-b373-fde9540ddfd1", "metadata": {}, "source": [ - "- Apply previous **step 3** to compute all context vectors:" + "Apply previous **step 3** to compute all context vectors:" ] }, { @@ -585,7 +583,7 @@ "id": "25b245b8-7732-4fab-aa1c-e3d333195605", "metadata": {}, "source": [ - "- As a sanity check, the previously computed context vector $z^{(2)} = [0.4419, 0.6515, 0.5683]$ can be found in the 2nd row in above: " + "As a sanity check, the previously computed context vector $z^{(2)} = [0.4419, 0.6515, 0.5683]$ can be found in the 2nd row in above: " ] }, { @@ -611,7 +609,7 @@ "id": "88363117-93d8-41fb-8240-f7cfe08b14a3", "metadata": {}, "source": [ - "- A conceptual framework illustrating how the self-attention mechanism developed in this section integrates into the overall narrative and structure of this book and chapter" + "A conceptual framework illustrating how the self-attention mechanism developed in this section integrates into the overall narrative and structure of this book and chapter" ] }, { @@ -619,7 +617,7 @@ "id": "ac9492ba-6f66-4f65-bd1d-87cf16d59928", "metadata": {}, "source": [ - "" + "" ] }, { @@ -635,14 +633,14 @@ "id": "46e95a46-1f67-4b71-9e84-8e2db84ab036", "metadata": {}, "source": [ - "- In this section, we are implementing the self-attention mechanism that is used in the original transformer architecture, the GPT models, and most other popular LLMs\n", - "- This self-attention mechanism is also called \"scaled dot-product attention\"\n", - "- The overall idea is similar to before:\n", - " - We want to compute context vectors as weighted sums over the input vectors specific to a certain input element\n", - " - For the above, we need attention weights\n", - "- As you will see, there are only slight differences compared to the basic attention mechanism introduced earlier:\n", - " - The most notable difference is the introduction of weight matrices that are updated during model training\n", - " - These trainable weight matrices are crucial so that the model (specifically, the attention module inside the model) can learn to produce \"good\" context vectors" + "In this section, we are implementing the self-attention mechanism that is used in the original transformer architecture, the GPT models, and most other popular LLMs\n", + "This self-attention mechanism is also called \"scaled dot-product attention\"\n", + "The overall idea is similar to before:\n", + " We want to compute context vectors as weighted sums over the input vectors specific to a certain input element\n", + " For the above, we need attention weights\n", + "As you will see, there are only slight differences compared to the basic attention mechanism introduced earlier:\n", + " The most notable difference is the introduction of weight matrices that are updated during model training\n", + " These trainable weight matrices are crucial so that the model (specifically, the attention module inside the model) can learn to produce \"good\" context vectors" ] }, { @@ -650,7 +648,7 @@ "id": "59db4093-93e8-4bee-be8f-c8fac8a08cdd", "metadata": {}, "source": [ - "" + "" ] }, { @@ -658,12 +656,12 @@ "id": "4d996671-87aa-45c9-b2e0-07a7bcc9060a", "metadata": {}, "source": [ - "- Implementing the self-attention mechanism step by step, we will start by introducing the three training weight matrices $W_q$, $W_k$, and $W_v$\n", - "- These three matrices are used to project the embedded input tokens, $x^{(i)}$, into query, key, and value vectors via matrix multiplication:\n", + "Implementing the self-attention mechanism step by step, we will start by introducing the three training weight matrices $W_q$, $W_k$, and $W_v$\n", + "These three matrices are used to project the embedded input tokens, $x^{(i)}$, into query, key, and value vectors via matrix multiplication:\n", "\n", - " - Query vector: $q^{(i)} = W_q \\,x^{(i)}$\n", - " - Key vector: $k^{(i)} = W_k \\,x^{(i)}$\n", - " - Value vector: $v^{(i)} = W_v \\,x^{(i)}$\n" + " Query vector: $q^{(i)} = W_q \\,x^{(i)}$\n", + " Key vector: $k^{(i)} = W_k \\,x^{(i)}$\n", + " Value vector: $v^{(i)} = W_v \\,x^{(i)}$\n" ] }, { @@ -671,8 +669,8 @@ "id": "9f334313-5fd0-477b-8728-04080a427049", "metadata": {}, "source": [ - "- The embedding dimensions of the input $x$ and the query vector $q$ can be the same or different, depending on the model's design and specific implementation\n", - "- In GPT models, the input and output dimensions are usually the same, but for illustration purposes, to better follow the computation, we choose different input and output dimensions here:" + "The embedding dimensions of the input $x$ and the query vector $q$ can be the same or different, depending on the model's design and specific implementation\n", + "In GPT models, the input and output dimensions are usually the same, but for illustration purposes, to better follow the computation, we choose different input and output dimensions here:" ] }, { @@ -692,7 +690,7 @@ "id": "f528cfb3-e226-47dd-b363-cc2caaeba4bf", "metadata": {}, "source": [ - "- Below, we initialize the three weight matrices; note that we are setting `requires_grad=False` to reduce clutter in the outputs for illustration purposes, but if we were to use the weight matrices for model training, we would set `requires_grad=True` to update these matrices during model training" + "Below, we initialize the three weight matrices; note that we are setting `requires_grad=False` to reduce clutter in the outputs for illustration purposes, but if we were to use the weight matrices for model training, we would set `requires_grad=True` to update these matrices during model training" ] }, { @@ -714,7 +712,7 @@ "id": "abfd0b50-7701-4adb-821c-e5433622d9c4", "metadata": {}, "source": [ - "- Next we compute the query, key, and value vectors:" + "Next we compute the query, key, and value vectors:" ] }, { @@ -736,7 +734,7 @@ "id": "9be308b3-aca3-421b-b182-19c3a03b71c7", "metadata": {}, "source": [ - "- As we can see below, we successfully projected the 6 input tokens from a 3D onto a 2D embedding space:" + "As we can see below, we successfully projected the 6 input tokens from a 3D onto a 2D embedding space:" ] }, { @@ -758,7 +756,7 @@ "id": "bac5dfd6-ade8-4e7b-b0c1-bed40aa24481", "metadata": {}, "source": [ - "- In the next step, **step 2**, we compute the unnormalized attention scores by computing the dot product between the query and each key vector:" + "In the next step, **step 2**, we compute the unnormalized attention scores by computing the dot product between the query and each key vector:" ] }, { @@ -766,7 +764,7 @@ "id": "8ed0a2b7-5c50-4ede-90cf-7ad74412b3aa", "metadata": {}, "source": [ - "" + "" ] }, { @@ -786,7 +784,7 @@ "id": "9e9d15c0-c24e-4e6f-a160-6349b418f935", "metadata": {}, "source": [ - "- Since we have 6 inputs, we have 6 attention scores for the given query vector:" + "Since we have 6 inputs, we have 6 attention scores for the given query vector:" ] }, { @@ -805,7 +803,7 @@ "id": "8622cf39-155f-4eb5-a0c0-82a03ce9b999", "metadata": {}, "source": [ - "" + "" ] }, { @@ -813,8 +811,8 @@ "id": "e1609edb-f089-461a-8de2-c20c1bb29836", "metadata": {}, "source": [ - "- Next, in **step 3**, we compute the attention weights (normalized attention scores that sum up to 1) using the softmax function we used earlier\n", - "- The difference to earlier is that we now scale the attention scores by dividing them by the square root of the embedding dimension, $\\sqrt{d_k}$ (i.e., `d_k**0.5`):" + "Next, in **step 3**, we compute the attention weights (normalized attention scores that sum up to 1) using the softmax function we used earlier\n", + "The difference to earlier is that we now scale the attention scores by dividing them by the square root of the embedding dimension, $\\sqrt{d_k}$ (i.e., `d_k**0.5`):" ] }, { @@ -834,7 +832,7 @@ "id": "b8f61a28-b103-434a-aee1-ae7cbd821126", "metadata": {}, "source": [ - "" + "" ] }, { @@ -842,7 +840,7 @@ "id": "1890e3f9-db86-4ab8-9f3b-53113504a61f", "metadata": {}, "source": [ - "- In **step 4**, we now compute the context vector for input query vector 2:" + "In **step 4**, we now compute the context vector for input query vector 2:" ] }, { @@ -869,7 +867,7 @@ "id": "04313410-3155-4d90-a7a3-2f3386e73677", "metadata": {}, "source": [ - "- Putting it all together, we can implement the self-attention mechanism as follows:" + "Putting it all together, we can implement the self-attention mechanism as follows:" ] }, { @@ -911,7 +909,7 @@ "id": "7ee1a024-84a5-425a-9567-54ab4e4ed445", "metadata": {}, "source": [ - "" + "" ] }, { @@ -919,8 +917,8 @@ "id": "048e0c16-d911-4ec8-b0bc-45ceec75c081", "metadata": {}, "source": [ - "- We can streamline the implementation above using PyTorch's Linear layers, which are equivalent to a matrix multiplication if we disable the bias units\n", - "- Another big advantage of using `nn.Linear` over our manual `nn.Parameter(torch.rand(...)` approach is that `nn.Linear` has a preferred weight initialization scheme, which leads to more stable model training" + "We can streamline the implementation above using PyTorch's Linear layers, which are equivalent to a matrix multiplication if we disable the bias units\n", + "Another big advantage of using `nn.Linear` over our manual `nn.Parameter(torch.rand(...)` approach is that `nn.Linear` has a preferred weight initialization scheme, which leads to more stable model training" ] }, { @@ -960,7 +958,7 @@ "id": "915cd8a5-a895-42c9-8b8e-06b5ae19ffce", "metadata": {}, "source": [ - "- Note that `SelfAttention_v1` and `SelfAttention_v2` give different outputs because they use different initial weights for the weight matrices" + "Note that `SelfAttention_v1` and `SelfAttention_v2` give different outputs because they use different initial weights for the weight matrices" ] }, { @@ -984,7 +982,7 @@ "id": "71e91bb5-5aae-4f05-8a95-973b3f988a35", "metadata": {}, "source": [ - "" + "" ] }, { @@ -1000,10 +998,10 @@ "id": "014f28d0-8218-48e4-8b9c-bdc5ce489218", "metadata": {}, "source": [ - "- In this section, we are converting the previous self-attention mechanism into a causal self-attention mechanism\n", - "- Causal self-attention ensures that the model's prediction for a certain position in a sequence is only dependent on the known outputs at previous positions, not on future positions\n", - "- In simpler words, this ensures that each next word prediction should only depend on the preceding words\n", - "- To achieve this, for each given token, we mask out the future tokens (the ones that come after the current token in the input text):" + "In this section, we are converting the previous self-attention mechanism into a causal self-attention mechanism\n", + "Causal self-attention ensures that the model's prediction for a certain position in a sequence is only dependent on the known outputs at previous positions, not on future positions\n", + "In simpler words, this ensures that each next word prediction should only depend on the preceding words\n", + "To achieve this, for each given token, we mask out the future tokens (the ones that come after the current token in the input text):" ] }, { @@ -1011,7 +1009,7 @@ "id": "57f99af3-32bc-48f5-8eb4-63504670ca0a", "metadata": {}, "source": [ - "" + "" ] }, { @@ -1019,7 +1017,7 @@ "id": "cbfaec7a-68f2-4157-a4b5-2aeceed199d9", "metadata": {}, "source": [ - "- To illustrate and implement causal self-attention, let's work with the attention scores and weights from the previous section: " + "To illustrate and implement causal self-attention, let's work with the attention scores and weights from the previous section: " ] }, { @@ -1044,7 +1042,7 @@ "id": "89020a96-b34d-41f8-9349-98c3e23fd5d6", "metadata": {}, "source": [ - "- The simplest way to mask out future attention weights is by creating a mask via PyTorch's tril function with elements below the main diagonal (including the diagonal itself) set to 1 and above the main diagonal set to 0:" + "The simplest way to mask out future attention weights is by creating a mask via PyTorch's tril function with elements below the main diagonal (including the diagonal itself) set to 1 and above the main diagonal set to 0:" ] }, { @@ -1064,7 +1062,7 @@ "id": "efce2b08-3583-44da-b3fc-cabdd38761f6", "metadata": {}, "source": [ - "- Then, we can multiply the attention weights with this mask to zero out the attention scores above the diagonal:" + "Then, we can multiply the attention weights with this mask to zero out the attention scores above the diagonal:" ] }, { @@ -1083,9 +1081,9 @@ "id": "3eb35787-cf12-4024-b66d-e7215e175500", "metadata": {}, "source": [ - "- However, if the mask were applied after softmax, like above, it would disrupt the probability distribution created by softmax\n", - "- Softmax ensures that all output values sum to 1\n", - "- Masking after softmax would require re-normalizing the outputs to sum to 1 again, which complicates the process and might lead to unintended effects" + "However, if the mask were applied after softmax, like above, it would disrupt the probability distribution created by softmax\n", + "Softmax ensures that all output values sum to 1\n", + "Masking after softmax would require re-normalizing the outputs to sum to 1 again, which complicates the process and might lead to unintended effects" ] }, { @@ -1093,7 +1091,7 @@ "id": "94db92d7-c397-4e42-bd8a-6a2b3e237e0f", "metadata": {}, "source": [ - "- To make sure that the rows sum to 1, we can normalize the attention weights as follows:" + "To make sure that the rows sum to 1, we can normalize the attention weights as follows:" ] }, { @@ -1113,8 +1111,8 @@ "id": "512e7cf4-dc0e-4cec-948e-c7a3c4eb6877", "metadata": {}, "source": [ - "- While we are technically done with coding the causal attention mechanism now, let's briefly look at a more efficient approach to achieve the same as above\n", - "- So, instead of zeroing out attention weights above the diagonal and renormalizing the results, we can mask the unnormalized attention scores above the diagonal with negative infinity before they enter the softmax function:" + "While we are technically done with coding the causal attention mechanism now, let's briefly look at a more efficient approach to achieve the same as above\n", + "So, instead of zeroing out attention weights above the diagonal and renormalizing the results, we can mask the unnormalized attention scores above the diagonal with negative infinity before they enter the softmax function:" ] }, { @@ -1122,7 +1120,7 @@ "id": "eb682900-8df2-4767-946c-a82bee260188", "metadata": {}, "source": [ - "" + "" ] }, { @@ -1142,7 +1140,7 @@ "id": "91d5f803-d735-4543-b9da-00ac10fb9c50", "metadata": {}, "source": [ - "- As we can see below, now the attention weights in each row correctly sum to 1 again:" + "As we can see below, now the attention weights in each row correctly sum to 1 again:" ] }, { @@ -1169,13 +1167,13 @@ "id": "ec3dc7ee-6539-4fab-804a-8f31a890c85a", "metadata": {}, "source": [ - "- In addition, we also apply dropout to reduce overfitting during training\n", - "- Dropout can be applied in several places:\n", - " - for example, after computing the attention weights;\n", - " - or after multiplying the attention weights with the value vectors\n", - "- Here, we will apply the dropout mask after computing the attention weights because it's more common\n", + "In addition, we also apply dropout to reduce overfitting during training\n", + "Dropout can be applied in several places:\n", + " for example, after computing the attention weights;\n", + " or after multiplying the attention weights with the value vectors\n", + "Here, we will apply the dropout mask after computing the attention weights because it's more common\n", "\n", - "- Furthermore, in this specific example, we use a dropout rate of 50%, which means randomly masking out half of the attention weights. (When we train the GPT model later, we will use a lower dropout rate, such as 0.1 or 0.2" + "Furthermore, in this specific example, we use a dropout rate of 50%, which means randomly masking out half of the attention weights. (When we train the GPT model later, we will use a lower dropout rate, such as 0.1 or 0.2" ] }, { @@ -1183,7 +1181,7 @@ "id": "ee799cf6-6175-45f2-827e-c174afedb722", "metadata": {}, "source": [ - "" + "" ] }, { @@ -1191,7 +1189,7 @@ "id": "5a575458-a6da-4e54-8688-83e155f2de06", "metadata": {}, "source": [ - "- If we apply a dropout rate of 0.5 (50%), the non-dropped values will be scaled accordingly by a factor of 1/0.5 = 2." + "If we apply a dropout rate of 0.5 (50%), the non-dropped values will be scaled accordingly by a factor of 1/0.5 = 2." ] }, { @@ -1224,7 +1222,7 @@ "id": "269df5c8-3e25-49d0-95d3-bb232287404f", "metadata": {}, "source": [ - "- Note that the resulting dropout outputs may look different depending on your operating system; you can read more about this inconsistency [here on the PyTorch issue tracker](https://github.com/pytorch/pytorch/issues/121595)" + "Note that the resulting dropout outputs may look different depending on your operating system; you can read more about this inconsistency [here on the PyTorch issue tracker](https://github.com/pytorch/pytorch/issues/121595)" ] }, { @@ -1240,9 +1238,9 @@ "id": "09c41d29-1933-43dc-ada6-2dbb56287204", "metadata": {}, "source": [ - "- Now, we are ready to implement a working implementation of self-attention, including the causal and dropout masks\n", - "- One more thing is to implement the code to handle batches consisting of more than one input so that our `CausalAttention` class supports the batch outputs produced by the data loader we implemented in chapter 2\n", - "- For simplicity, to simulate such batch input, we duplicate the input text example:" + "Now, we are ready to implement a working implementation of self-attention, including the causal and dropout masks\n", + "One more thing is to implement the code to handle batches consisting of more than one input so that our `CausalAttention` class supports the batch outputs produced by the data loader we implemented in chapter 2\n", + "For simplicity, to simulate such batch input, we duplicate the input text example:" ] }, { @@ -1305,7 +1303,7 @@ "id": "c4333d12-17e4-4bb5-9d83-54b3a32618cd", "metadata": {}, "source": [ - "- Note that dropout is only applied during training, not during inference" + "Note that dropout is only applied during training, not during inference" ] }, { @@ -1313,7 +1311,7 @@ "id": "a554cf47-558c-4f45-84cd-bf9b839a8d50", "metadata": {}, "source": [ - "" + "" ] }, { @@ -1337,17 +1335,17 @@ "id": "70766faf-cd53-41d9-8a17-f1b229756a5a", "metadata": {}, "source": [ - "- Below is a summary of the self-attention implemented previously (causal and dropout masks not shown for simplicity)\n", + "Below is a summary of the self-attention implemented previously (causal and dropout masks not shown for simplicity)\n", "\n", - "- This is also called single-head attention:\n", + "This is also called single-head attention:\n", "\n", - "\n", + "\n", "\n", - "- We simply stack multiple single-head attention modules to obtain a multi-head attention module:\n", + "We simply stack multiple single-head attention modules to obtain a multi-head attention module:\n", "\n", - "\n", + "\n", "\n", - "- The main idea behind multi-head attention is to run the attention mechanism multiple times (in parallel) with different, learned linear projections. This allows the model to jointly attend to information from different representation subspaces at different positions." + "The main idea behind multi-head attention is to run the attention mechanism multiple times (in parallel) with different, learned linear projections. This allows the model to jointly attend to information from different representation subspaces at different positions." ] }, { @@ -1387,7 +1385,7 @@ "id": "193d3d2b-2578-40ba-b791-ea2d49328e48", "metadata": {}, "source": [ - "- In the implementation above, the embedding dimension is 4, because we `d_out=2` as the embedding dimension for the key, query, and value vectors as well as the context vector. And since we have 2 attention heads, we have the output embedding dimension 2*2=4" + "In the implementation above, the embedding dimension is 4, because we `d_out=2` as the embedding dimension for the key, query, and value vectors as well as the context vector. And since we have 2 attention heads, we have the output embedding dimension 2*2=4" ] }, { @@ -1403,10 +1401,10 @@ "id": "f4b48d0d-71ba-4fa0-b714-ca80cabcb6f7", "metadata": {}, "source": [ - "- While the above is an intuitive and fully functional implementation of multi-head attention (wrapping the single-head attention `CausalAttention` implementation from earlier), we can write a stand-alone class called `MultiHeadAttention` to achieve the same\n", + "While the above is an intuitive and fully functional implementation of multi-head attention (wrapping the single-head attention `CausalAttention` implementation from earlier), we can write a stand-alone class called `MultiHeadAttention` to achieve the same\n", "\n", - "- We don't concatenate single attention heads for this stand-alone `MultiHeadAttention` class\n", - "- Instead, we create single W_query, W_key, and W_value weight matrices and then split those into individual matrices for each attention head:" + "We don't concatenate single attention heads for this stand-alone `MultiHeadAttention` class\n", + "Instead, we create single W_query, W_key, and W_value weight matrices and then split those into individual matrices for each attention head:" ] }, { @@ -1488,9 +1486,9 @@ "id": "d334dfb5-2b6c-4c33-82d5-b4e9db5867bb", "metadata": {}, "source": [ - "- Note that the above is essentially a rewritten version of `MultiHeadAttentionWrapper` that is more efficient\n", - "- The resulting output looks a bit different since the random weight initializations differ, but both are fully functional implementations that can be used in the GPT class we will implement in the upcoming chapters\n", - "- Note that in addition, we added a linear projection layer (`self.out_proj `) to the `MultiHeadAttention` class above. This is simply a linear transformation that doesn't change the dimensions. It's a standard convention to use such a projection layer in LLM implementation, but it's not strictly necessary (recent research has shown that it can be removed without affecting the modeling performance; see the further reading section at the end of this chapter)\n" + "Note that the above is essentially a rewritten version of `MultiHeadAttentionWrapper` that is more efficient\n", + "The resulting output looks a bit different since the random weight initializations differ, but both are fully functional implementations that can be used in the GPT class we will implement in the upcoming chapters\n", + "Note that in addition, we added a linear projection layer (`self.out_proj `) to the `MultiHeadAttention` class above. This is simply a linear transformation that doesn't change the dimensions. It's a standard convention to use such a projection layer in LLM implementation, but it's not strictly necessary (recent research has shown that it can be removed without affecting the modeling performance; see the further reading section at the end of this chapter)\n" ] }, { @@ -1498,7 +1496,7 @@ "id": "dbe5d396-c990-45dc-9908-2c621461f851", "metadata": {}, "source": [ - "" + "" ] }, { @@ -1506,7 +1504,7 @@ "id": "8b0ed78c-e8ac-4f8f-a479-a98242ae8f65", "metadata": {}, "source": [ - "- Note that if you are interested in a compact and efficient implementation of the above, you can also consider the [`torch.nn.MultiheadAttention`](https://pytorch.org/docs/stable/generated/torch.nn.MultiheadAttention.html) class in PyTorch" + "Note that if you are interested in a compact and efficient implementation of the above, you can also consider the [`torch.nn.MultiheadAttention`](https://pytorch.org/docs/stable/generated/torch.nn.MultiheadAttention.html) class in PyTorch" ] }, { @@ -1514,7 +1512,7 @@ "id": "363701ad-2022-46c8-9972-390d2a2b9911", "metadata": {}, "source": [ - "- Since the above implementation may look a bit complex at first glance, let's look at what happens when executing `attn_scores = queries @ keys.transpose(2, 3)`:" + "Since the above implementation may look a bit complex at first glance, let's look at what happens when executing `attn_scores = queries @ keys.transpose(2, 3)`:" ] }, { @@ -1541,9 +1539,9 @@ "id": "0587b946-c8f2-4888-adbf-5a5032fbfd7b", "metadata": {}, "source": [ - "- In this case, the matrix multiplication implementation in PyTorch will handle the 4-dimensional input tensor so that the matrix multiplication is carried out between the 2 last dimensions (num_tokens, head_dim) and then repeated for the individual heads \n", + "In this case, the matrix multiplication implementation in PyTorch will handle the 4-dimensional input tensor so that the matrix multiplication is carried out between the 2 last dimensions (num_tokens, head_dim) and then repeated for the individual heads \n", "\n", - "- For instance, the following becomes a more compact way to compute the matrix multiplication for each head separately:" + "For instance, the following becomes a more compact way to compute the matrix multiplication for each head separately:" ] }, { @@ -1562,6 +1560,15 @@ "print(\"\\nSecond head:\\n\", second_res)" ] }, + { + "cell_type": "markdown", + "id": "f24450ba", + "metadata": {}, + "source": [ + "## Your turn! 🚀\n", + "tbd" + ] + }, { "cell_type": "markdown", "id": "3a819062", From 975e013a284d1c30959c4c7ee1a814146b0b762d Mon Sep 17 00:00:00 2001 From: peng_windows <2686728826@qq.com> Date: Tue, 9 Apr 2024 11:24:30 +0800 Subject: [PATCH 6/6] fix --- open-machine-learning-jupyter-book/llm/introduction.ipynb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/open-machine-learning-jupyter-book/llm/introduction.ipynb b/open-machine-learning-jupyter-book/llm/introduction.ipynb index 01464a46f..6e458c7d6 100644 --- a/open-machine-learning-jupyter-book/llm/introduction.ipynb +++ b/open-machine-learning-jupyter-book/llm/introduction.ipynb @@ -70,7 +70,7 @@ "\n", "These models have demonstrated impressive capabilities in understanding and generating natural language, leading to their widespread adoption across various industries and applications.\n", "\n", - ":::{figure} https://raw.githubusercontent.com/Mooler0410/LLMsPracticalGuide/main/imgs/tree.jpg\n", + ":::{figure} https://static-1300131294.cos.ap-shanghai.myqcloud.com/images/llm/clipboard_20240409_112330.png\n", "An Evolutionary Tree of Modern LLMs\n", ":::" ]