From 243bd18dfcb407fe805da2ec7be7db7d2308585f Mon Sep 17 00:00:00 2001
From: Jing Zhang <jing3jour@gmail.com>
Date: Sun, 8 Sep 2024 17:07:18 +0200
Subject: [PATCH] update

---
 deep_learning_from_scratch.ipynb  | 12761 +++++++++++++++++++++++++++-
 figures/dlscratch_cnnlayer.png    |   Bin 0 -> 17642 bytes
 figures/dlscratch_img2col.png     |   Bin 0 -> 47906 bytes
 figures/dlscratch_pool.png        |   Bin 0 -> 82599 bytes
 figures/dlscratch_poolflatten.png |   Bin 0 -> 72379 bytes
 figures/lena_gray.png             |   Bin 0 -> 42588 bytes
 6 files changed, 12751 insertions(+), 10 deletions(-)
 create mode 100644 figures/dlscratch_cnnlayer.png
 create mode 100644 figures/dlscratch_img2col.png
 create mode 100644 figures/dlscratch_pool.png
 create mode 100644 figures/dlscratch_poolflatten.png
 create mode 100644 figures/lena_gray.png

diff --git a/deep_learning_from_scratch.ipynb b/deep_learning_from_scratch.ipynb
index 1bf2975..12e1790 100644
--- a/deep_learning_from_scratch.ipynb
+++ b/deep_learning_from_scratch.ipynb
@@ -6441,43 +6441,12784 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "### Implemention of convolutional and pooling layer"
+    "### Implemention of convolutional and pooling layer\n",
+    "convolutional layer are computed by img2col, which is to transform the image into a matrix, and then do matrix multiplication with the filter matrix.    \n",
+    "![img2col](./figures/dlscratch_img2col.png)    \n",
+    "The process of pooling flatten into a vector, and then do matrix multiplication with the pooling matrix.    \n",
+    "![img2colpool](./figures/dlscratch_poolflatten.png)    "
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": 1,
    "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(9, 75)\n",
+      "(90, 75)\n"
+     ]
+    }
+   ],
    "source": [
-    "### Implemention of CNN"
+    "# example of im2col\n",
+    "import numpy as np\n",
+    "\n",
+    "def im2col(input_data, filter_h, filter_w, stride=1, pad=0):\n",
+    "  \n",
+    "    N, C, H, W = input_data.shape\n",
+    "    out_h = (H + 2*pad - filter_h)//stride + 1\n",
+    "    out_w = (W + 2*pad - filter_w)//stride + 1\n",
+    "\n",
+    "    img = np.pad(input_data, [(0,0), (0,0), (pad, pad), (pad, pad)], 'constant')\n",
+    "    col = np.zeros((N, C, filter_h, filter_w, out_h, out_w))\n",
+    "\n",
+    "    for y in range(filter_h):\n",
+    "        y_max = y + stride*out_h\n",
+    "        for x in range(filter_w):\n",
+    "            x_max = x + stride*out_w\n",
+    "            col[:, :, y, x, :, :] = img[:, :, y:y_max:stride, x:x_max:stride]\n",
+    "\n",
+    "    col = col.transpose(0, 4, 5, 1, 2, 3).reshape(N*out_h*out_w, -1)\n",
+    "    return col\n",
+    "\n",
+    "\n",
+    "def col2im(col, input_shape, filter_h, filter_w, stride=1, pad=0):\n",
+    "\n",
+    "    N, C, H, W = input_shape\n",
+    "    out_h = (H + 2*pad - filter_h)//stride + 1\n",
+    "    out_w = (W + 2*pad - filter_w)//stride + 1\n",
+    "    col = col.reshape(N, out_h, out_w, C, filter_h, filter_w).transpose(0, 3, 4, 5, 1, 2)\n",
+    "\n",
+    "    img = np.zeros((N, C, H + 2*pad + stride - 1, W + 2*pad + stride - 1))\n",
+    "    for y in range(filter_h):\n",
+    "        y_max = y + stride*out_h\n",
+    "        for x in range(filter_w):\n",
+    "            x_max = x + stride*out_w\n",
+    "            img[:, :, y:y_max:stride, x:x_max:stride] += col[:, :, y, x, :, :]\n",
+    "\n",
+    "    return img[:, :, pad:H + pad, pad:W + pad]\n",
+    "\n",
+    "x1 = np.random.rand(1, 3, 7, 7)\n",
+    "col1 = im2col(x1, 5, 5, stride=1, pad=0)\n",
+    "print(col1.shape) # (9, 75)\n",
+    "x2 = np.random.rand(10, 3, 7, 7) \n",
+    "col2 = im2col(x2, 5, 5, stride=1, pad=0)\n",
+    "print(col2.shape) # (90, 75)"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "### Visualization of CNN"
+    "### Implemention of CNN\n",
+    "![cnnlayer](./figures/dlscratch_cnnlayer.png)"
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": 12,
    "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "train loss:2.2995362058887374\n",
+      "=== epoch:1, train acc:0.311, test acc:0.308 ===\n",
+      "train loss:2.2963005738557154\n",
+      "train loss:2.293603547766879\n",
+      "train loss:2.2884076469310015\n",
+      "train loss:2.278925014254402\n",
+      "train loss:2.265921866541648\n",
+      "train loss:2.2535849542742965\n",
+      "train loss:2.245003457603123\n",
+      "train loss:2.218813938881403\n",
+      "train loss:2.169671803356559\n",
+      "train loss:2.135996574969285\n",
+      "train loss:2.1447259119723703\n",
+      "train loss:2.0744724481989927\n",
+      "train loss:2.0291252127726422\n",
+      "train loss:1.9901688480893676\n",
+      "train loss:1.8961561773426754\n",
+      "train loss:1.8389760220099478\n",
+      "train loss:1.752274682254193\n",
+      "train loss:1.7087302557051505\n",
+      "train loss:1.5541750335157132\n",
+      "train loss:1.4806052643952965\n",
+      "train loss:1.4886417840448996\n",
+      "train loss:1.3014568356043916\n",
+      "train loss:1.2597812449422419\n",
+      "train loss:1.2497878303182606\n",
+      "train loss:1.1185777776643178\n",
+      "train loss:1.1015115736246783\n",
+      "train loss:1.03241412687908\n",
+      "train loss:0.9350416991616189\n",
+      "train loss:0.8285826180979486\n",
+      "train loss:0.8966355721801236\n",
+      "train loss:0.866788286601\n",
+      "train loss:0.746887836745562\n",
+      "train loss:0.8661508356191806\n",
+      "train loss:0.7667373830675487\n",
+      "train loss:0.6368185448183228\n",
+      "train loss:0.7847853674483365\n",
+      "train loss:0.7472912635713104\n",
+      "train loss:0.6122450509386127\n",
+      "train loss:0.6888880707196741\n",
+      "train loss:0.7731712120990576\n",
+      "train loss:0.5547860085175733\n",
+      "train loss:0.831467733576241\n",
+      "train loss:0.6858566291946968\n",
+      "train loss:0.5881971807862099\n",
+      "train loss:0.5377546177909613\n",
+      "train loss:0.6085414554123344\n",
+      "train loss:0.6352170605684923\n",
+      "train loss:0.518837993390407\n",
+      "train loss:0.5188881606480904\n",
+      "train loss:0.6262163694315077\n",
+      "train loss:0.48653930211725777\n",
+      "train loss:0.5996035402248621\n",
+      "train loss:0.5958299933823917\n",
+      "train loss:0.47785559328177796\n",
+      "train loss:0.6442040469925483\n",
+      "train loss:0.5294109516385664\n",
+      "train loss:0.6108989047725126\n",
+      "train loss:0.49579772890691926\n",
+      "train loss:0.5155124467964299\n",
+      "train loss:0.4425233369938194\n",
+      "train loss:0.5723815732448505\n",
+      "train loss:0.4444025741017111\n",
+      "train loss:0.5682502396333912\n",
+      "train loss:0.4812394544517227\n",
+      "train loss:0.5733820018653012\n",
+      "train loss:0.5976324097802066\n",
+      "train loss:0.42834611851298554\n",
+      "train loss:0.33299373106962193\n",
+      "train loss:0.4824415889920817\n",
+      "train loss:0.46172672532810366\n",
+      "train loss:0.39976769965368997\n",
+      "train loss:0.356991520423221\n",
+      "train loss:0.3971668972051343\n",
+      "train loss:0.3077737100712962\n",
+      "train loss:0.3688735462768081\n",
+      "train loss:0.39958060822977315\n",
+      "train loss:0.5228967063738835\n",
+      "train loss:0.4051070789088767\n",
+      "train loss:0.4087904735541978\n",
+      "train loss:0.4699611248842445\n",
+      "train loss:0.49664353328580413\n",
+      "train loss:0.5394036471918038\n",
+      "train loss:0.4386570870688064\n",
+      "train loss:0.542043869320365\n",
+      "train loss:0.461831085349672\n",
+      "train loss:0.33590317449397183\n",
+      "train loss:0.3641995285252705\n",
+      "train loss:0.46013235184048185\n",
+      "train loss:0.4629640846375168\n",
+      "train loss:0.40744457251974603\n",
+      "train loss:0.3204745620465407\n",
+      "train loss:0.28314928145092516\n",
+      "train loss:0.27531900677480964\n",
+      "train loss:0.2884335623712579\n",
+      "train loss:0.3461165795192081\n",
+      "train loss:0.26766298417593565\n",
+      "train loss:0.4132386147539512\n",
+      "train loss:0.45452523323166216\n",
+      "train loss:0.32100168104459925\n",
+      "train loss:0.33963758411907435\n",
+      "train loss:0.38065300436167054\n",
+      "train loss:0.36010924425797874\n",
+      "train loss:0.4772028751507527\n",
+      "train loss:0.3183786024897384\n",
+      "train loss:0.24063273539834373\n",
+      "train loss:0.33785816463573815\n",
+      "train loss:0.4096383795978882\n",
+      "train loss:0.3048112892795876\n",
+      "train loss:0.34850397319185866\n",
+      "train loss:0.4135990596240788\n",
+      "train loss:0.33631199253844507\n",
+      "train loss:0.36752554798070736\n",
+      "train loss:0.256289882515971\n",
+      "train loss:0.3434391522806523\n",
+      "train loss:0.5232171725615155\n",
+      "train loss:0.19122013617288178\n",
+      "train loss:0.24729049348021387\n",
+      "train loss:0.3788099794336589\n",
+      "train loss:0.22372122871779498\n",
+      "train loss:0.35003048898501987\n",
+      "train loss:0.3937635495265932\n",
+      "train loss:0.35450644272616266\n",
+      "train loss:0.38630379716477053\n",
+      "train loss:0.218081377276597\n",
+      "train loss:0.24068221540129456\n",
+      "train loss:0.37324452864502583\n",
+      "train loss:0.3980932619476245\n",
+      "train loss:0.33279776979069686\n",
+      "train loss:0.3126887242701013\n",
+      "train loss:0.311902560710905\n",
+      "train loss:0.3120501240043874\n",
+      "train loss:0.443248093420433\n",
+      "train loss:0.44772141309588626\n",
+      "train loss:0.3586080908819963\n",
+      "train loss:0.1985248679078512\n",
+      "train loss:0.3918575781138622\n",
+      "train loss:0.39030192273756137\n",
+      "train loss:0.19934570139414529\n",
+      "train loss:0.2699178555789283\n",
+      "train loss:0.5132984854142603\n",
+      "train loss:0.4162387821752907\n",
+      "train loss:0.2778727447209852\n",
+      "train loss:0.3026132236208509\n",
+      "train loss:0.23681846983354904\n",
+      "train loss:0.38732628396856683\n",
+      "train loss:0.30786760492468646\n",
+      "train loss:0.23318151214949712\n",
+      "train loss:0.28280193880545157\n",
+      "train loss:0.28744316984206875\n",
+      "train loss:0.314981055595686\n",
+      "train loss:0.19083085124681667\n",
+      "train loss:0.2468391300175253\n",
+      "train loss:0.5022710810056712\n",
+      "train loss:0.35210132284195766\n",
+      "train loss:0.4941417020864269\n",
+      "train loss:0.36300258296633986\n",
+      "train loss:0.2676306086995317\n",
+      "train loss:0.3956934523216956\n",
+      "train loss:0.49948617647803195\n",
+      "train loss:0.3398118281461233\n",
+      "train loss:0.23759418222895584\n",
+      "train loss:0.23565343639075323\n",
+      "train loss:0.26843572722452697\n",
+      "train loss:0.21510891732013285\n",
+      "train loss:0.32265419030998205\n",
+      "train loss:0.2985231765435053\n",
+      "train loss:0.26796717955155425\n",
+      "train loss:0.13599667430793513\n",
+      "train loss:0.34081862862193046\n",
+      "train loss:0.2403844382247804\n",
+      "train loss:0.32277883745308705\n",
+      "train loss:0.316461166608473\n",
+      "train loss:0.28049930012175817\n",
+      "train loss:0.2526955343690795\n",
+      "train loss:0.31225783151124564\n",
+      "train loss:0.241816725687005\n",
+      "train loss:0.26389694405728603\n",
+      "train loss:0.31704352907423977\n",
+      "train loss:0.48549474895706035\n",
+      "train loss:0.24423741761031376\n",
+      "train loss:0.22575933191828781\n",
+      "train loss:0.23379593692065895\n",
+      "train loss:0.43392939490175325\n",
+      "train loss:0.18763946088912448\n",
+      "train loss:0.3208163733495436\n",
+      "train loss:0.625560786031706\n",
+      "train loss:0.329178574853311\n",
+      "train loss:0.305694113117804\n",
+      "train loss:0.24654293474144828\n",
+      "train loss:0.2988656427193461\n",
+      "train loss:0.31576310016019665\n",
+      "train loss:0.34275087788715597\n",
+      "train loss:0.16360756268246418\n",
+      "train loss:0.19034393778847528\n",
+      "train loss:0.20139547839516606\n",
+      "train loss:0.35010435428873005\n",
+      "train loss:0.1585764920725295\n",
+      "train loss:0.18727900924059035\n",
+      "train loss:0.165339390268121\n",
+      "train loss:0.22369499079273247\n",
+      "train loss:0.24250080043458813\n",
+      "train loss:0.1632615950092321\n",
+      "train loss:0.2734568335302263\n",
+      "train loss:0.3141910145466435\n",
+      "train loss:0.24037194163794828\n",
+      "train loss:0.2051913327986346\n",
+      "train loss:0.32774778398544285\n",
+      "train loss:0.24699771884523855\n",
+      "train loss:0.2394694922447969\n",
+      "train loss:0.4169391437431564\n",
+      "train loss:0.2762179762629919\n",
+      "train loss:0.26164059586413874\n",
+      "train loss:0.16685245494574485\n",
+      "train loss:0.2103067986930876\n",
+      "train loss:0.21084263660581104\n",
+      "train loss:0.21102815261202865\n",
+      "train loss:0.3097679498100641\n",
+      "train loss:0.25315938853630127\n",
+      "train loss:0.26383700690836326\n",
+      "train loss:0.280468002599833\n",
+      "train loss:0.23808897982497895\n",
+      "train loss:0.20662748413570045\n",
+      "train loss:0.175746309591344\n",
+      "train loss:0.5696416699441348\n",
+      "train loss:0.13381465055359118\n",
+      "train loss:0.19067264894584926\n",
+      "train loss:0.16904414048923053\n",
+      "train loss:0.1822128726006831\n",
+      "train loss:0.2455060327676183\n",
+      "train loss:0.37535637564889035\n",
+      "train loss:0.30200796181204514\n",
+      "train loss:0.1388047093918803\n",
+      "train loss:0.21289344515928185\n",
+      "train loss:0.20380716469732513\n",
+      "train loss:0.2101381553210851\n",
+      "train loss:0.16829230402085674\n",
+      "train loss:0.250032923585419\n",
+      "train loss:0.2886474852305738\n",
+      "train loss:0.19874232094638555\n",
+      "train loss:0.2095868392663101\n",
+      "train loss:0.1839072516471734\n",
+      "train loss:0.21268794827322535\n",
+      "train loss:0.14048787981378572\n",
+      "train loss:0.11868821513153958\n",
+      "train loss:0.1350360226429506\n",
+      "train loss:0.23519816355548628\n",
+      "train loss:0.2568848076933926\n",
+      "train loss:0.2746125129456926\n",
+      "train loss:0.16470336822015927\n",
+      "train loss:0.18285312368802792\n",
+      "train loss:0.22133049525665663\n",
+      "train loss:0.20680602756401856\n",
+      "train loss:0.29588096790317386\n",
+      "train loss:0.2374773845710119\n",
+      "train loss:0.3123640189654112\n",
+      "train loss:0.34295290562362496\n",
+      "train loss:0.3425199579878354\n",
+      "train loss:0.27863041706294644\n",
+      "train loss:0.23782096699187033\n",
+      "train loss:0.28248441842950434\n",
+      "train loss:0.18766664070837064\n",
+      "train loss:0.24190456312306136\n",
+      "train loss:0.16606736607268846\n",
+      "train loss:0.21183019599123448\n",
+      "train loss:0.20964940557268347\n",
+      "train loss:0.2605947313513408\n",
+      "train loss:0.12494620144378356\n",
+      "train loss:0.17850585411827105\n",
+      "train loss:0.2438759175504828\n",
+      "train loss:0.1869960853763924\n",
+      "train loss:0.1744526186980288\n",
+      "train loss:0.2285773652563308\n",
+      "train loss:0.2642578364839423\n",
+      "train loss:0.17462434197416962\n",
+      "train loss:0.19833523317712712\n",
+      "train loss:0.3164126681478402\n",
+      "train loss:0.2839711135557014\n",
+      "train loss:0.20242114829877303\n",
+      "train loss:0.23557962271573565\n",
+      "train loss:0.1862007034093785\n",
+      "train loss:0.3368870283198877\n",
+      "train loss:0.21399323223647465\n",
+      "train loss:0.3393876531508587\n",
+      "train loss:0.1969068383064066\n",
+      "train loss:0.19589774626301287\n",
+      "train loss:0.22985644997292698\n",
+      "train loss:0.1792416589477409\n",
+      "train loss:0.2751420518127082\n",
+      "train loss:0.27091263844084634\n",
+      "train loss:0.1445833535957494\n",
+      "train loss:0.11727344378112566\n",
+      "train loss:0.17994335282343557\n",
+      "train loss:0.16285182276730897\n",
+      "train loss:0.20906913163704677\n",
+      "train loss:0.14057559137109815\n",
+      "train loss:0.20754404455636613\n",
+      "train loss:0.23670296667084476\n",
+      "train loss:0.15389187496043136\n",
+      "train loss:0.14486240610153567\n",
+      "train loss:0.13581088766821622\n",
+      "train loss:0.20782290809050147\n",
+      "train loss:0.12728044385638207\n",
+      "train loss:0.23610729007209247\n",
+      "train loss:0.28035096460006265\n",
+      "train loss:0.38397575431010655\n",
+      "train loss:0.15490292346718088\n",
+      "train loss:0.2843760640417501\n",
+      "train loss:0.3943114836334488\n",
+      "train loss:0.2900687091983363\n",
+      "train loss:0.20844222475020707\n",
+      "train loss:0.26486001716905877\n",
+      "train loss:0.1196667685978009\n",
+      "train loss:0.19836248766039472\n",
+      "train loss:0.38178973185355175\n",
+      "train loss:0.4218883695869735\n",
+      "train loss:0.2291154794970348\n",
+      "train loss:0.35943387117571246\n",
+      "train loss:0.2501316835106568\n",
+      "train loss:0.18511712690420246\n",
+      "train loss:0.19252567128704423\n",
+      "train loss:0.12660403944830181\n",
+      "train loss:0.13874234389370468\n",
+      "train loss:0.19403264050923405\n",
+      "train loss:0.11546229290528254\n",
+      "train loss:0.13655376105428005\n",
+      "train loss:0.19251669486149153\n",
+      "train loss:0.11342051444007563\n",
+      "train loss:0.1416544233730254\n",
+      "train loss:0.14633851524265198\n",
+      "train loss:0.24367762630662207\n",
+      "train loss:0.09087026690320454\n",
+      "train loss:0.2295413594137116\n",
+      "train loss:0.23866056266903604\n",
+      "train loss:0.2545641100569238\n",
+      "train loss:0.12784069258444938\n",
+      "train loss:0.16113552964506428\n",
+      "train loss:0.17403864978800918\n",
+      "train loss:0.17586664628416768\n",
+      "train loss:0.09805519807544234\n",
+      "train loss:0.1616212275997714\n",
+      "train loss:0.10594300046896105\n",
+      "train loss:0.24117889255151478\n",
+      "train loss:0.1483258334837773\n",
+      "train loss:0.30338077582411016\n",
+      "train loss:0.12938384425620245\n",
+      "train loss:0.20016031775671356\n",
+      "train loss:0.20518066966401677\n",
+      "train loss:0.2782778505961009\n",
+      "train loss:0.30250002066951237\n",
+      "train loss:0.1294634927398526\n",
+      "train loss:0.12799820766377196\n",
+      "train loss:0.0994178289788845\n",
+      "train loss:0.23549048015300086\n",
+      "train loss:0.2123750862830353\n",
+      "train loss:0.21734263750329955\n",
+      "train loss:0.20000440136154782\n",
+      "train loss:0.24858626109130819\n",
+      "train loss:0.1769848343946891\n",
+      "train loss:0.16396227422081605\n",
+      "train loss:0.12319936885728369\n",
+      "train loss:0.28064240184650374\n",
+      "train loss:0.20187288611237952\n",
+      "train loss:0.15817218086752713\n",
+      "train loss:0.11195060140416249\n",
+      "train loss:0.13166810842012708\n",
+      "train loss:0.32022762193508514\n",
+      "train loss:0.2651203077602952\n",
+      "train loss:0.08238882417565833\n",
+      "train loss:0.23917079819999185\n",
+      "train loss:0.2672918944342174\n",
+      "train loss:0.16726702162422474\n",
+      "train loss:0.26275383378967826\n",
+      "train loss:0.15316873431995812\n",
+      "train loss:0.28588653753697546\n",
+      "train loss:0.15991443471586195\n",
+      "train loss:0.2493342429747267\n",
+      "train loss:0.3354113405116221\n",
+      "train loss:0.14047989451275325\n",
+      "train loss:0.22334883915212062\n",
+      "train loss:0.2629987505317322\n",
+      "train loss:0.13083901773979728\n",
+      "train loss:0.1295064967899669\n",
+      "train loss:0.1821648448159373\n",
+      "train loss:0.23012745321839098\n",
+      "train loss:0.19945757435631908\n",
+      "train loss:0.1264889970765247\n",
+      "train loss:0.14571031896755982\n",
+      "train loss:0.27501513063312377\n",
+      "train loss:0.23566538390957847\n",
+      "train loss:0.380803831948321\n",
+      "train loss:0.22609911624690773\n",
+      "train loss:0.17862681746741463\n",
+      "train loss:0.17774901809571145\n",
+      "train loss:0.160711283960583\n",
+      "train loss:0.2498393840250065\n",
+      "train loss:0.2540339531763633\n",
+      "train loss:0.10447327831238047\n",
+      "train loss:0.09081545987660697\n",
+      "train loss:0.210966709936758\n",
+      "train loss:0.27702957258034727\n",
+      "train loss:0.14240522004925207\n",
+      "train loss:0.2247305220802777\n",
+      "train loss:0.2297815295241078\n",
+      "train loss:0.1603681518889312\n",
+      "train loss:0.13170611200797633\n",
+      "train loss:0.10022615460590219\n",
+      "train loss:0.1307846923879241\n",
+      "train loss:0.3573686574490344\n",
+      "train loss:0.1282467505300392\n",
+      "train loss:0.12773233797615688\n",
+      "train loss:0.0920185329077778\n",
+      "train loss:0.20783130599905242\n",
+      "train loss:0.10725366047900531\n",
+      "train loss:0.20065925496205025\n",
+      "train loss:0.331419450917982\n",
+      "train loss:0.15886719879197714\n",
+      "train loss:0.15676280268618784\n",
+      "train loss:0.15290716706401322\n",
+      "train loss:0.2308953781253323\n",
+      "train loss:0.1816390968953019\n",
+      "train loss:0.2591972473897047\n",
+      "train loss:0.15560271941656098\n",
+      "train loss:0.1617264316274939\n",
+      "train loss:0.07370684442370674\n",
+      "train loss:0.17999474808316956\n",
+      "train loss:0.09006538309884592\n",
+      "train loss:0.1281307272179306\n",
+      "train loss:0.14477718910101983\n",
+      "train loss:0.23322955289181863\n",
+      "train loss:0.2516932649382623\n",
+      "train loss:0.11960468833803979\n",
+      "train loss:0.13922155877355322\n",
+      "train loss:0.09764880728855806\n",
+      "train loss:0.18905814076621616\n",
+      "train loss:0.15631976670585154\n",
+      "train loss:0.26533567284021775\n",
+      "train loss:0.27741170590515263\n",
+      "train loss:0.1674910997915362\n",
+      "train loss:0.12343744626663934\n",
+      "train loss:0.10277734500012216\n",
+      "train loss:0.13016390829790725\n",
+      "train loss:0.13973411685428194\n",
+      "train loss:0.08909993000763959\n",
+      "train loss:0.18329198508849592\n",
+      "train loss:0.15599246412831164\n",
+      "train loss:0.17663948274467106\n",
+      "train loss:0.07973494937826837\n",
+      "train loss:0.1374780895555915\n",
+      "train loss:0.1622684625816559\n",
+      "train loss:0.0656944427659519\n",
+      "train loss:0.12422665872080944\n",
+      "train loss:0.1594186199698269\n",
+      "train loss:0.1468969464826564\n",
+      "train loss:0.07618377867042189\n",
+      "train loss:0.21959065456352372\n",
+      "train loss:0.15556954170990436\n",
+      "train loss:0.16395356715339518\n",
+      "train loss:0.22270154702496675\n",
+      "train loss:0.1483091307682986\n",
+      "train loss:0.14985848078776187\n",
+      "train loss:0.10741860008623799\n",
+      "train loss:0.08748394847746865\n",
+      "train loss:0.09830819077128457\n",
+      "train loss:0.1351282809127371\n",
+      "train loss:0.10990186382171162\n",
+      "train loss:0.1825326487228333\n",
+      "train loss:0.13790293943983012\n",
+      "train loss:0.18014682345440744\n",
+      "train loss:0.26999183081352096\n",
+      "train loss:0.13025666233326333\n",
+      "train loss:0.16723555511435317\n",
+      "train loss:0.13320384930897045\n",
+      "train loss:0.17889677421248995\n",
+      "train loss:0.14308810663597568\n",
+      "train loss:0.15119282201189974\n",
+      "train loss:0.10243301525986336\n",
+      "train loss:0.2039734000502691\n",
+      "train loss:0.07343025958025606\n",
+      "train loss:0.11540528196609127\n",
+      "train loss:0.16540301777221736\n",
+      "train loss:0.09720459922924451\n",
+      "train loss:0.15034350640033547\n",
+      "train loss:0.1749782593217526\n",
+      "train loss:0.12757041344468198\n",
+      "train loss:0.05042917593515421\n",
+      "train loss:0.1123980000023787\n",
+      "train loss:0.047670269631326326\n",
+      "train loss:0.09218928558276217\n",
+      "train loss:0.19163188301545972\n",
+      "train loss:0.28823547455254195\n",
+      "train loss:0.06960942354382678\n",
+      "train loss:0.13865111888593612\n",
+      "train loss:0.060488119131594076\n",
+      "train loss:0.21071457083153708\n",
+      "train loss:0.16077154755738257\n",
+      "train loss:0.13414236430072832\n",
+      "train loss:0.1785579469810826\n",
+      "train loss:0.13496182801903367\n",
+      "train loss:0.10270292484434272\n",
+      "train loss:0.09200172018306216\n",
+      "train loss:0.0886432461963844\n",
+      "train loss:0.18820088996241388\n",
+      "train loss:0.1748564306800397\n",
+      "train loss:0.12638995970012118\n",
+      "train loss:0.08355907445898642\n",
+      "train loss:0.13639431725061582\n",
+      "train loss:0.07245666989169693\n",
+      "train loss:0.1304797847128204\n",
+      "train loss:0.13483055301676541\n",
+      "train loss:0.16188278735288664\n",
+      "train loss:0.043502174108246354\n",
+      "train loss:0.16263503284865016\n",
+      "train loss:0.06917987986703342\n",
+      "train loss:0.20469652701795962\n",
+      "train loss:0.1313729580246157\n",
+      "train loss:0.11716312011893361\n",
+      "train loss:0.04467443566822679\n",
+      "train loss:0.20261041629211365\n",
+      "train loss:0.0971251074604594\n",
+      "train loss:0.27132987912088063\n",
+      "train loss:0.10596555175983921\n",
+      "train loss:0.2209541783731231\n",
+      "train loss:0.15664645858374052\n",
+      "train loss:0.13195055069812134\n",
+      "train loss:0.19611417576408774\n",
+      "train loss:0.21370580591526347\n",
+      "train loss:0.1346351951112406\n",
+      "train loss:0.0989933362671312\n",
+      "train loss:0.18255602344500513\n",
+      "train loss:0.12640442356232642\n",
+      "train loss:0.25426509663860863\n",
+      "train loss:0.18245032172708966\n",
+      "train loss:0.06547328590635973\n",
+      "train loss:0.10883557996299366\n",
+      "train loss:0.09374651867202438\n",
+      "train loss:0.13771163851044668\n",
+      "train loss:0.1302785109734555\n",
+      "train loss:0.11426886342144589\n",
+      "train loss:0.07117475188325571\n",
+      "train loss:0.17958827894701992\n",
+      "train loss:0.11480015066446772\n",
+      "train loss:0.09456015353872241\n",
+      "train loss:0.1905275513078982\n",
+      "train loss:0.08435356663597045\n",
+      "train loss:0.12593613284280347\n",
+      "train loss:0.22359727954427025\n",
+      "train loss:0.24320546762278528\n",
+      "train loss:0.18016065535191006\n",
+      "train loss:0.10223428875171563\n",
+      "train loss:0.0919863118333515\n",
+      "train loss:0.05083351251644275\n",
+      "train loss:0.21485631827949367\n",
+      "train loss:0.1489787019014207\n",
+      "train loss:0.08091794804291669\n",
+      "train loss:0.049867321822047346\n",
+      "train loss:0.15895584240943816\n",
+      "train loss:0.08672026955791115\n",
+      "train loss:0.26670377152682534\n",
+      "train loss:0.10806776133649505\n",
+      "train loss:0.19211519384959427\n",
+      "train loss:0.07308070263657843\n",
+      "train loss:0.08509968630805068\n",
+      "train loss:0.14474903761791494\n",
+      "train loss:0.15674616695437435\n",
+      "train loss:0.13326687111581093\n",
+      "train loss:0.19255676015814818\n",
+      "train loss:0.16304882103761198\n",
+      "train loss:0.09745226053736179\n",
+      "train loss:0.18150891113683304\n",
+      "train loss:0.1629305961499786\n",
+      "train loss:0.19943268625600083\n",
+      "train loss:0.14928665672336985\n",
+      "train loss:0.12086343324710626\n",
+      "train loss:0.06963367185968021\n",
+      "train loss:0.13656334045232466\n",
+      "train loss:0.09693043504618029\n",
+      "train loss:0.23742407591566356\n",
+      "train loss:0.08411429808332399\n",
+      "train loss:0.10070259763859603\n",
+      "train loss:0.12086839681659688\n",
+      "train loss:0.10511760456944516\n",
+      "train loss:0.10882019475720106\n",
+      "train loss:0.1391582016931682\n",
+      "train loss:0.11715988245112802\n",
+      "train loss:0.0836485117589737\n",
+      "train loss:0.1118329170955436\n",
+      "train loss:0.08954948487345937\n",
+      "train loss:0.12853334996903706\n",
+      "train loss:0.06771352340200942\n",
+      "train loss:0.08569362894119292\n",
+      "train loss:0.10880250906507559\n",
+      "train loss:0.1314239509203147\n",
+      "train loss:0.036067871127909115\n",
+      "train loss:0.16522557048165748\n",
+      "train loss:0.040317875586856164\n",
+      "train loss:0.09223834308740025\n",
+      "train loss:0.07087163041549135\n",
+      "train loss:0.16868302844959424\n",
+      "train loss:0.09473146115964597\n",
+      "train loss:0.08242012998779036\n",
+      "=== epoch:2, train acc:0.969, test acc:0.962 ===\n",
+      "train loss:0.10567043043106424\n",
+      "train loss:0.035399697698602815\n",
+      "train loss:0.11152512309116867\n",
+      "train loss:0.22522079410357965\n",
+      "train loss:0.17565566229939275\n",
+      "train loss:0.16686835160712168\n",
+      "train loss:0.12504658053774004\n",
+      "train loss:0.07544892173013264\n",
+      "train loss:0.07730940363843308\n",
+      "train loss:0.12115757637939055\n",
+      "train loss:0.16926484181389026\n",
+      "train loss:0.09731834037929434\n",
+      "train loss:0.04438945911530027\n",
+      "train loss:0.2212419111194723\n",
+      "train loss:0.0955204796700148\n",
+      "train loss:0.13965347127342548\n",
+      "train loss:0.059651909485076524\n",
+      "train loss:0.13471990991256516\n",
+      "train loss:0.114284870015461\n",
+      "train loss:0.33269356861132215\n",
+      "train loss:0.21967782932074115\n",
+      "train loss:0.07690377080910016\n",
+      "train loss:0.1665090344985183\n",
+      "train loss:0.10516886823499982\n",
+      "train loss:0.10044788182472017\n",
+      "train loss:0.07874527062530028\n",
+      "train loss:0.21635838283750344\n",
+      "train loss:0.1141392048563779\n",
+      "train loss:0.08566652170293802\n",
+      "train loss:0.051639984867280324\n",
+      "train loss:0.13022440925648887\n",
+      "train loss:0.09672690073560948\n",
+      "train loss:0.039806183029892146\n",
+      "train loss:0.21078978880224344\n",
+      "train loss:0.12203217912232683\n",
+      "train loss:0.20806694173595633\n",
+      "train loss:0.11704811312470179\n",
+      "train loss:0.12733400007846984\n",
+      "train loss:0.11863574365834824\n",
+      "train loss:0.12585967226627706\n",
+      "train loss:0.159797001786201\n",
+      "train loss:0.0902805242357857\n",
+      "train loss:0.05534506277893532\n",
+      "train loss:0.07303956722276761\n",
+      "train loss:0.07980315318989345\n",
+      "train loss:0.11244356798226067\n",
+      "train loss:0.08480528692152035\n",
+      "train loss:0.18460431824933654\n",
+      "train loss:0.19037966840285203\n",
+      "train loss:0.14395569521957766\n",
+      "train loss:0.183172574334989\n",
+      "train loss:0.10664419097036555\n",
+      "train loss:0.12350555266422594\n",
+      "train loss:0.19570554085689001\n",
+      "train loss:0.11463679674441368\n",
+      "train loss:0.17006189068798835\n",
+      "train loss:0.06674272301569104\n",
+      "train loss:0.18916921845904133\n",
+      "train loss:0.16276138753571556\n",
+      "train loss:0.1732295449055278\n",
+      "train loss:0.11196674105772664\n",
+      "train loss:0.12072916065971949\n",
+      "train loss:0.13416449500939737\n",
+      "train loss:0.14876760875508857\n",
+      "train loss:0.09384663699495087\n",
+      "train loss:0.09437538013541884\n",
+      "train loss:0.12420987841771262\n",
+      "train loss:0.06910195684832668\n",
+      "train loss:0.05594701099179878\n",
+      "train loss:0.09438126083503068\n",
+      "train loss:0.10201008138962063\n",
+      "train loss:0.22942996061028953\n",
+      "train loss:0.15970469979809243\n",
+      "train loss:0.10090229510283973\n",
+      "train loss:0.07619102931084981\n",
+      "train loss:0.0689976252241214\n",
+      "train loss:0.04844562929664903\n",
+      "train loss:0.06865666770400287\n",
+      "train loss:0.14483806418611775\n",
+      "train loss:0.07359983078002569\n",
+      "train loss:0.08417421028902092\n",
+      "train loss:0.10588816624316649\n",
+      "train loss:0.07430964562437946\n",
+      "train loss:0.08317411993536156\n",
+      "train loss:0.08932786694375507\n",
+      "train loss:0.05971271530462312\n",
+      "train loss:0.043101704214222186\n",
+      "train loss:0.2203357599326123\n",
+      "train loss:0.14041965901118014\n",
+      "train loss:0.07855022864202388\n",
+      "train loss:0.054395030792659255\n",
+      "train loss:0.042988678455292965\n",
+      "train loss:0.13470230956793494\n",
+      "train loss:0.18915949087544576\n",
+      "train loss:0.06121982536757053\n",
+      "train loss:0.07946035497995449\n",
+      "train loss:0.08970583814725859\n",
+      "train loss:0.0984300580473948\n",
+      "train loss:0.1308950474698835\n",
+      "train loss:0.06242005521896999\n",
+      "train loss:0.11095052513855341\n",
+      "train loss:0.18114988712409744\n",
+      "train loss:0.07738739719335216\n",
+      "train loss:0.1396554482310675\n",
+      "train loss:0.09666214199957825\n",
+      "train loss:0.06627109654775251\n",
+      "train loss:0.19092279197098352\n",
+      "train loss:0.12799422310033073\n",
+      "train loss:0.23336198938808814\n",
+      "train loss:0.06162106906606076\n",
+      "train loss:0.12745142028592676\n",
+      "train loss:0.06888614012108893\n",
+      "train loss:0.194780986189548\n",
+      "train loss:0.08803220093291524\n",
+      "train loss:0.08219593540614654\n",
+      "train loss:0.17479472288545422\n",
+      "train loss:0.15323085313751997\n",
+      "train loss:0.12783162665112152\n",
+      "train loss:0.061999376941728634\n",
+      "train loss:0.09673345174575075\n",
+      "train loss:0.1648861817571646\n",
+      "train loss:0.059887319771234744\n",
+      "train loss:0.2178341440627502\n",
+      "train loss:0.07133974368489147\n",
+      "train loss:0.08997449453131512\n",
+      "train loss:0.05099545412635589\n",
+      "train loss:0.1353487855044161\n",
+      "train loss:0.034321318242953536\n",
+      "train loss:0.04367717740557611\n",
+      "train loss:0.086047607254425\n",
+      "train loss:0.07896686492896618\n",
+      "train loss:0.12190509670550012\n",
+      "train loss:0.06468059628255962\n",
+      "train loss:0.05034544913190926\n",
+      "train loss:0.11027908048087655\n",
+      "train loss:0.048471134983695645\n",
+      "train loss:0.06753482886596457\n",
+      "train loss:0.03157269374218568\n",
+      "train loss:0.09658623257999728\n",
+      "train loss:0.04689056026127876\n",
+      "train loss:0.14921366437521175\n",
+      "train loss:0.07204319310204765\n",
+      "train loss:0.0847644999050876\n",
+      "train loss:0.05901513583540208\n",
+      "train loss:0.08033973004452317\n",
+      "train loss:0.10348031600496979\n",
+      "train loss:0.13859463846931036\n",
+      "train loss:0.14512712010379358\n",
+      "train loss:0.13285707475103625\n",
+      "train loss:0.031104206217529577\n",
+      "train loss:0.06731293922371154\n",
+      "train loss:0.07852183416716152\n",
+      "train loss:0.06455812841648487\n",
+      "train loss:0.17712387410000718\n",
+      "train loss:0.08711250226450243\n",
+      "train loss:0.08501438569949901\n",
+      "train loss:0.25948089150972786\n",
+      "train loss:0.03452027701304878\n",
+      "train loss:0.13038314822122415\n",
+      "train loss:0.11566604020674635\n",
+      "train loss:0.1523593335719743\n",
+      "train loss:0.06487576150619388\n",
+      "train loss:0.04475056132328741\n",
+      "train loss:0.13905737576816876\n",
+      "train loss:0.08858116195754424\n",
+      "train loss:0.14212505401320438\n",
+      "train loss:0.0795320427906903\n",
+      "train loss:0.05049673962107381\n",
+      "train loss:0.1269703042172585\n",
+      "train loss:0.10371272019522959\n",
+      "train loss:0.10413108413251425\n",
+      "train loss:0.09622257683628523\n",
+      "train loss:0.09322690616419421\n",
+      "train loss:0.2368108202049671\n",
+      "train loss:0.10673072456821371\n",
+      "train loss:0.09271529421486685\n",
+      "train loss:0.1224719467711636\n",
+      "train loss:0.0823793542366674\n",
+      "train loss:0.04961645251031983\n",
+      "train loss:0.09426933907171424\n",
+      "train loss:0.05588659889514631\n",
+      "train loss:0.0973187787170392\n",
+      "train loss:0.0966925045643045\n",
+      "train loss:0.09931285163124212\n",
+      "train loss:0.06265927990653312\n",
+      "train loss:0.06861753883508982\n",
+      "train loss:0.06462351831185914\n",
+      "train loss:0.10355242082208362\n",
+      "train loss:0.06943806721818933\n",
+      "train loss:0.0898022623422363\n",
+      "train loss:0.12331520537686927\n",
+      "train loss:0.05231490364368601\n",
+      "train loss:0.11083252379548238\n",
+      "train loss:0.0599724489172472\n",
+      "train loss:0.0970717230810238\n",
+      "train loss:0.047726887694604654\n",
+      "train loss:0.14029351536831508\n",
+      "train loss:0.12898386243553558\n",
+      "train loss:0.1418105125041675\n",
+      "train loss:0.07364641067847515\n",
+      "train loss:0.17503981517441847\n",
+      "train loss:0.23052378810791638\n",
+      "train loss:0.18570381759329913\n",
+      "train loss:0.07374078634620936\n",
+      "train loss:0.06457210990112981\n",
+      "train loss:0.06375487259287681\n",
+      "train loss:0.04606384519295615\n",
+      "train loss:0.0409130008101336\n",
+      "train loss:0.08696261907062225\n",
+      "train loss:0.05188033948649776\n",
+      "train loss:0.11801959209505812\n",
+      "train loss:0.04589113376692187\n",
+      "train loss:0.10698109005979278\n",
+      "train loss:0.04871636887893382\n",
+      "train loss:0.0767899809639784\n",
+      "train loss:0.08284503285309236\n",
+      "train loss:0.06245776826203446\n",
+      "train loss:0.10839617279181396\n",
+      "train loss:0.13264727889970498\n",
+      "train loss:0.062311315341553264\n",
+      "train loss:0.08949314430151362\n",
+      "train loss:0.05238533563066998\n",
+      "train loss:0.06831668059532754\n",
+      "train loss:0.09904675784059583\n",
+      "train loss:0.12650175925017854\n",
+      "train loss:0.06642130884774369\n",
+      "train loss:0.030078601047421934\n",
+      "train loss:0.037193462886189105\n",
+      "train loss:0.13619583224045215\n",
+      "train loss:0.062349899718328675\n",
+      "train loss:0.09254664270336958\n",
+      "train loss:0.047749255980321835\n",
+      "train loss:0.04303718858472296\n",
+      "train loss:0.11278504686867517\n",
+      "train loss:0.16524791028525712\n",
+      "train loss:0.07894923868931025\n",
+      "train loss:0.026563240868207164\n",
+      "train loss:0.10936951361077427\n",
+      "train loss:0.054577715219357836\n",
+      "train loss:0.04205495474069336\n",
+      "train loss:0.09141671144864721\n",
+      "train loss:0.1446929257959578\n",
+      "train loss:0.0538858695404806\n",
+      "train loss:0.06822448601884991\n",
+      "train loss:0.01898471018030265\n",
+      "train loss:0.04725850268704303\n",
+      "train loss:0.14677347427482415\n",
+      "train loss:0.07786559974533698\n",
+      "train loss:0.08588899048774078\n",
+      "train loss:0.06878630622934111\n",
+      "train loss:0.24524331092092758\n",
+      "train loss:0.1666480237111597\n",
+      "train loss:0.054403580301291825\n",
+      "train loss:0.1028338415557665\n",
+      "train loss:0.12079787166723524\n",
+      "train loss:0.08193679405274387\n",
+      "train loss:0.04386201459256925\n",
+      "train loss:0.11665999890234247\n",
+      "train loss:0.3370661102564865\n",
+      "train loss:0.2116417255817848\n",
+      "train loss:0.05480243362912765\n",
+      "train loss:0.09969446800166523\n",
+      "train loss:0.028055679325947155\n",
+      "train loss:0.10913824344960991\n",
+      "train loss:0.20447586311254123\n",
+      "train loss:0.1110581783197072\n",
+      "train loss:0.13930435499654237\n",
+      "train loss:0.06713709516947043\n",
+      "train loss:0.06553083042180276\n",
+      "train loss:0.0661417432898066\n",
+      "train loss:0.14268645581735848\n",
+      "train loss:0.029597637322657938\n",
+      "train loss:0.07904852934240664\n",
+      "train loss:0.063577594512256\n",
+      "train loss:0.058248288981219885\n",
+      "train loss:0.055520745245602666\n",
+      "train loss:0.0651517117210649\n",
+      "train loss:0.05166443483825071\n",
+      "train loss:0.04741291012433283\n",
+      "train loss:0.10205397304880877\n",
+      "train loss:0.09672990903093102\n",
+      "train loss:0.029369363938135863\n",
+      "train loss:0.05364785561438499\n",
+      "train loss:0.09737386714683324\n",
+      "train loss:0.09066658347127653\n",
+      "train loss:0.04638468024980118\n",
+      "train loss:0.05295750261571801\n",
+      "train loss:0.17190463192866154\n",
+      "train loss:0.06809278868724009\n",
+      "train loss:0.07585884150146109\n",
+      "train loss:0.12110473006046267\n",
+      "train loss:0.022291029231205427\n",
+      "train loss:0.1066730732768966\n",
+      "train loss:0.03223280249216093\n",
+      "train loss:0.0737317892616462\n",
+      "train loss:0.06714858561351597\n",
+      "train loss:0.12530267538947035\n",
+      "train loss:0.07546031847773312\n",
+      "train loss:0.05248283818335482\n",
+      "train loss:0.06122084816094251\n",
+      "train loss:0.13928421304609623\n",
+      "train loss:0.06410417982237022\n",
+      "train loss:0.09348648820585069\n",
+      "train loss:0.04508553631595855\n",
+      "train loss:0.07669850467292291\n",
+      "train loss:0.10718140889161941\n",
+      "train loss:0.07948029128127194\n",
+      "train loss:0.07620248290788671\n",
+      "train loss:0.026639286252294836\n",
+      "train loss:0.12101035045056652\n",
+      "train loss:0.04775075852969752\n",
+      "train loss:0.06944726048327288\n",
+      "train loss:0.039893118197333724\n",
+      "train loss:0.038712669214135985\n",
+      "train loss:0.07832294667549765\n",
+      "train loss:0.06250321944635791\n",
+      "train loss:0.04118610228079974\n",
+      "train loss:0.11690306111259136\n",
+      "train loss:0.07310400302604705\n",
+      "train loss:0.043716912591450836\n",
+      "train loss:0.06791165239419569\n",
+      "train loss:0.08397090235263983\n",
+      "train loss:0.15988557841479908\n",
+      "train loss:0.07560066860746878\n",
+      "train loss:0.021160317727112177\n",
+      "train loss:0.11802349140465226\n",
+      "train loss:0.051118506423152875\n",
+      "train loss:0.0833850317757701\n",
+      "train loss:0.036177093005702654\n",
+      "train loss:0.10476365786993515\n",
+      "train loss:0.06261347815796663\n",
+      "train loss:0.06904723714620886\n",
+      "train loss:0.05318889570402117\n",
+      "train loss:0.1260898537777717\n",
+      "train loss:0.16203821785371977\n",
+      "train loss:0.11255430455397807\n",
+      "train loss:0.09052549111092008\n",
+      "train loss:0.11600263492710478\n",
+      "train loss:0.06324590272247753\n",
+      "train loss:0.09362440999049385\n",
+      "train loss:0.09802479640629613\n",
+      "train loss:0.11526055804587867\n",
+      "train loss:0.12533292699233267\n",
+      "train loss:0.04888704858379206\n",
+      "train loss:0.034963175658116866\n",
+      "train loss:0.17544105802991208\n",
+      "train loss:0.023824999631380243\n",
+      "train loss:0.10968365964201231\n",
+      "train loss:0.08740812759081004\n",
+      "train loss:0.0967769012428632\n",
+      "train loss:0.1302879782621083\n",
+      "train loss:0.09962723662950534\n",
+      "train loss:0.06948409495582995\n",
+      "train loss:0.10809804895686578\n",
+      "train loss:0.07830639284122566\n",
+      "train loss:0.11658973654676273\n",
+      "train loss:0.05492148136611799\n",
+      "train loss:0.07017474723774766\n",
+      "train loss:0.06961883159753328\n",
+      "train loss:0.06881924315289757\n",
+      "train loss:0.0893056556191383\n",
+      "train loss:0.05956716325464935\n",
+      "train loss:0.09635751179145113\n",
+      "train loss:0.04382046924124374\n",
+      "train loss:0.018436684818234166\n",
+      "train loss:0.08020770825367032\n",
+      "train loss:0.028296904837955657\n",
+      "train loss:0.019445283073082065\n",
+      "train loss:0.046084299638108284\n",
+      "train loss:0.027839804655184538\n",
+      "train loss:0.11004770181041637\n",
+      "train loss:0.01839937326901776\n",
+      "train loss:0.04988207021209443\n",
+      "train loss:0.02847991212459914\n",
+      "train loss:0.06840662375411892\n",
+      "train loss:0.08146512358092017\n",
+      "train loss:0.10353138638930146\n",
+      "train loss:0.08387668966769163\n",
+      "train loss:0.06180773446580299\n",
+      "train loss:0.11251988917596072\n",
+      "train loss:0.07048866944889406\n",
+      "train loss:0.02675241362019006\n",
+      "train loss:0.050615473253064085\n",
+      "train loss:0.06856222412734671\n",
+      "train loss:0.03220625607226808\n",
+      "train loss:0.06609099504851779\n",
+      "train loss:0.08698393269519963\n",
+      "train loss:0.08498948382366103\n",
+      "train loss:0.0550358195597354\n",
+      "train loss:0.121911096297613\n",
+      "train loss:0.13243540354608485\n",
+      "train loss:0.12961864240005655\n",
+      "train loss:0.05807198418415394\n",
+      "train loss:0.0743378450890999\n",
+      "train loss:0.0760931896765355\n",
+      "train loss:0.12908564043684256\n",
+      "train loss:0.10153507934404876\n",
+      "train loss:0.06858174382352743\n",
+      "train loss:0.16998919317120187\n",
+      "train loss:0.021861133729344978\n",
+      "train loss:0.0290599636501473\n",
+      "train loss:0.07737220411739508\n",
+      "train loss:0.08063137640648076\n",
+      "train loss:0.21245879397604356\n",
+      "train loss:0.04926000770342153\n",
+      "train loss:0.05119492407820527\n",
+      "train loss:0.043605017418368426\n",
+      "train loss:0.15626952399303598\n",
+      "train loss:0.050664151108208634\n",
+      "train loss:0.12571852742744866\n",
+      "train loss:0.03892711821291215\n",
+      "train loss:0.13099508570305843\n",
+      "train loss:0.056580525091383255\n",
+      "train loss:0.06399774096392047\n",
+      "train loss:0.057075609287914233\n",
+      "train loss:0.10713367574181952\n",
+      "train loss:0.04100365085024252\n",
+      "train loss:0.029130846571031564\n",
+      "train loss:0.08246515985715595\n",
+      "train loss:0.07215751989074208\n",
+      "train loss:0.12102008969778026\n",
+      "train loss:0.030423297541165753\n",
+      "train loss:0.09405087507058905\n",
+      "train loss:0.14268822243533305\n",
+      "train loss:0.07528622864151309\n",
+      "train loss:0.15255920681158155\n",
+      "train loss:0.08655478746821826\n",
+      "train loss:0.11302895335302461\n",
+      "train loss:0.13251712732353632\n",
+      "train loss:0.05037181564015339\n",
+      "train loss:0.07602653832022684\n",
+      "train loss:0.06721249420071154\n",
+      "train loss:0.08725542905431175\n",
+      "train loss:0.08556324196173305\n",
+      "train loss:0.10043013073994951\n",
+      "train loss:0.07983142318314669\n",
+      "train loss:0.03464302380868023\n",
+      "train loss:0.1361382423966104\n",
+      "train loss:0.08206120907114134\n",
+      "train loss:0.1054774508292664\n",
+      "train loss:0.061864074083619505\n",
+      "train loss:0.1763497373072806\n",
+      "train loss:0.06478702085387705\n",
+      "train loss:0.06847050217497\n",
+      "train loss:0.0303255946484074\n",
+      "train loss:0.04059786938131213\n",
+      "train loss:0.17238627276551838\n",
+      "train loss:0.018196324449259376\n",
+      "train loss:0.053781023288106536\n",
+      "train loss:0.05622757122167351\n",
+      "train loss:0.05392405380640633\n",
+      "train loss:0.03337944842644434\n",
+      "train loss:0.07110553207556519\n",
+      "train loss:0.06581931327765746\n",
+      "train loss:0.028585140342569693\n",
+      "train loss:0.03335611316596218\n",
+      "train loss:0.08074529538147894\n",
+      "train loss:0.07092254439623413\n",
+      "train loss:0.030294194897738255\n",
+      "train loss:0.11631373522117949\n",
+      "train loss:0.05103560394942226\n",
+      "train loss:0.03634208034314995\n",
+      "train loss:0.08315385831037697\n",
+      "train loss:0.05528364673717614\n",
+      "train loss:0.04613037442281104\n",
+      "train loss:0.08811888278940191\n",
+      "train loss:0.1302100638347669\n",
+      "train loss:0.1132747730893744\n",
+      "train loss:0.0701264676844081\n",
+      "train loss:0.07179090514003762\n",
+      "train loss:0.09898884486031921\n",
+      "train loss:0.03782863745164786\n",
+      "train loss:0.07875041426571165\n",
+      "train loss:0.06301438272875508\n",
+      "train loss:0.08138679689254108\n",
+      "train loss:0.03876057219299272\n",
+      "train loss:0.04226980330128525\n",
+      "train loss:0.07444360318684874\n",
+      "train loss:0.04768913834980068\n",
+      "train loss:0.038212219578009264\n",
+      "train loss:0.053394095200782685\n",
+      "train loss:0.18300948801485314\n",
+      "train loss:0.06149548905275332\n",
+      "train loss:0.05992859145542497\n",
+      "train loss:0.057324541494924504\n",
+      "train loss:0.031339143471004756\n",
+      "train loss:0.14717468151319013\n",
+      "train loss:0.08667550447091826\n",
+      "train loss:0.08802621594746149\n",
+      "train loss:0.03869056047686819\n",
+      "train loss:0.11103731152310997\n",
+      "train loss:0.059539908009508195\n",
+      "train loss:0.019024196819031464\n",
+      "train loss:0.06302375212954725\n",
+      "train loss:0.07385682596569612\n",
+      "train loss:0.03259450006676674\n",
+      "train loss:0.08226824724727125\n",
+      "train loss:0.04281561695092809\n",
+      "train loss:0.10420082959050532\n",
+      "train loss:0.0653242950468115\n",
+      "train loss:0.038098026031583555\n",
+      "train loss:0.05609086333007669\n",
+      "train loss:0.04532468289910656\n",
+      "train loss:0.052212915896017974\n",
+      "train loss:0.07452347968303392\n",
+      "train loss:0.0772443685011458\n",
+      "train loss:0.05901067768215151\n",
+      "train loss:0.11083519647400081\n",
+      "train loss:0.0453190178417231\n",
+      "train loss:0.02504706173820131\n",
+      "train loss:0.056875064333495165\n",
+      "train loss:0.2757886502517094\n",
+      "train loss:0.020938664838901812\n",
+      "train loss:0.08490206759232967\n",
+      "train loss:0.05749890555356075\n",
+      "train loss:0.03179310886665485\n",
+      "train loss:0.06723229770621432\n",
+      "train loss:0.0594968290053043\n",
+      "train loss:0.08265415391594566\n",
+      "train loss:0.09294684143339588\n",
+      "train loss:0.09558107721139905\n",
+      "train loss:0.051818884548047256\n",
+      "train loss:0.047180000945854325\n",
+      "train loss:0.04912769543795161\n",
+      "train loss:0.031136353525095445\n",
+      "train loss:0.0516147007645152\n",
+      "train loss:0.19505201231154817\n",
+      "train loss:0.026848974193845507\n",
+      "train loss:0.10790285374523938\n",
+      "train loss:0.04311816118248252\n",
+      "train loss:0.05485669346593894\n",
+      "train loss:0.04335968743193425\n",
+      "train loss:0.05416934927373299\n",
+      "train loss:0.030181811833786057\n",
+      "train loss:0.0709774066286797\n",
+      "train loss:0.050575645797922134\n",
+      "train loss:0.04650632801455921\n",
+      "train loss:0.03839049172691952\n",
+      "train loss:0.037685473648000525\n",
+      "train loss:0.04286790771520143\n",
+      "train loss:0.10298703444529797\n",
+      "train loss:0.08673044592547002\n",
+      "train loss:0.07947521743440315\n",
+      "train loss:0.04724135353446634\n",
+      "train loss:0.058472198288141736\n",
+      "train loss:0.06346009578744889\n",
+      "train loss:0.038329039414776646\n",
+      "train loss:0.09221701895916225\n",
+      "train loss:0.08253686302431146\n",
+      "train loss:0.03345851597922944\n",
+      "train loss:0.06370407712053304\n",
+      "train loss:0.11301121760565593\n",
+      "train loss:0.037391341384401665\n",
+      "train loss:0.017723663050308373\n",
+      "train loss:0.12834567395315577\n",
+      "train loss:0.05186441349857985\n",
+      "train loss:0.0345809968653173\n",
+      "train loss:0.04434494410202424\n",
+      "train loss:0.0738166323941895\n",
+      "train loss:0.12797159084051454\n",
+      "train loss:0.03734017522171578\n",
+      "train loss:0.08276208077532277\n",
+      "train loss:0.031213911404879387\n",
+      "train loss:0.056113832147608046\n",
+      "train loss:0.037708690771947266\n",
+      "train loss:0.0635818499297054\n",
+      "train loss:0.06412671763430182\n",
+      "train loss:0.03383076714761229\n",
+      "train loss:0.03414605565417669\n",
+      "train loss:0.03228400374194713\n",
+      "train loss:0.08029521591362455\n",
+      "train loss:0.019029156603179675\n",
+      "train loss:0.05641111393119988\n",
+      "train loss:0.11603071440777125\n",
+      "train loss:0.2527216664182237\n",
+      "train loss:0.03969768032639173\n",
+      "train loss:0.08321041119148895\n",
+      "train loss:0.09302160961161868\n",
+      "train loss:0.15455033937005397\n",
+      "train loss:0.024508135054221922\n",
+      "train loss:0.08782642347469354\n",
+      "train loss:0.0498697397191813\n",
+      "train loss:0.02094518142195306\n",
+      "train loss:0.10038052181559451\n",
+      "train loss:0.0516536470960026\n",
+      "train loss:0.014529917189405944\n",
+      "train loss:0.12609930336097197\n",
+      "train loss:0.12780216022740876\n",
+      "train loss:0.0659022082461845\n",
+      "train loss:0.03780998384634144\n",
+      "train loss:0.0550284330304601\n",
+      "train loss:0.015045600381989077\n",
+      "train loss:0.10004237347723315\n",
+      "train loss:0.03168023609882437\n",
+      "train loss:0.05245399510737341\n",
+      "train loss:0.09636330860268824\n",
+      "train loss:0.055961168090278506\n",
+      "train loss:0.08323279927525336\n",
+      "train loss:0.08127454709280973\n",
+      "train loss:0.07469569185174493\n",
+      "=== epoch:3, train acc:0.977, test acc:0.98 ===\n",
+      "train loss:0.09644599996739901\n",
+      "train loss:0.06880614117143509\n",
+      "train loss:0.06995086431990388\n",
+      "train loss:0.036011799532240804\n",
+      "train loss:0.07781468541483569\n",
+      "train loss:0.0754843267439297\n",
+      "train loss:0.09535368168919486\n",
+      "train loss:0.03204607803554125\n",
+      "train loss:0.03779920315465613\n",
+      "train loss:0.07094830013932654\n",
+      "train loss:0.0385067579697714\n",
+      "train loss:0.07413380912778778\n",
+      "train loss:0.0656520436085621\n",
+      "train loss:0.04601103104532475\n",
+      "train loss:0.03499440335740928\n",
+      "train loss:0.06179822501819055\n",
+      "train loss:0.04618439925431669\n",
+      "train loss:0.07556498936980932\n",
+      "train loss:0.07894440024902967\n",
+      "train loss:0.0292007515568275\n",
+      "train loss:0.2541871297640099\n",
+      "train loss:0.07559122977479076\n",
+      "train loss:0.04656788884734832\n",
+      "train loss:0.04472526082638418\n",
+      "train loss:0.10805508589114382\n",
+      "train loss:0.03171288629629992\n",
+      "train loss:0.039466981124525666\n",
+      "train loss:0.030403885808200838\n",
+      "train loss:0.046535932874830606\n",
+      "train loss:0.051397919668906694\n",
+      "train loss:0.06754558029108516\n",
+      "train loss:0.020031660792796935\n",
+      "train loss:0.016793857497404214\n",
+      "train loss:0.058024612917034375\n",
+      "train loss:0.05227009159530275\n",
+      "train loss:0.06109141324461631\n",
+      "train loss:0.05381443541529062\n",
+      "train loss:0.041950982778488646\n",
+      "train loss:0.029425234008739934\n",
+      "train loss:0.044834410705859466\n",
+      "train loss:0.058153646287722224\n",
+      "train loss:0.041774015913894874\n",
+      "train loss:0.06132700299826116\n",
+      "train loss:0.04207687279350931\n",
+      "train loss:0.06585424714573476\n",
+      "train loss:0.025571112243363454\n",
+      "train loss:0.05290769874623081\n",
+      "train loss:0.02775146485474824\n",
+      "train loss:0.023072680132992664\n",
+      "train loss:0.070648196404331\n",
+      "train loss:0.0637874073412999\n",
+      "train loss:0.11635170433626671\n",
+      "train loss:0.08088923445544685\n",
+      "train loss:0.033867799730607016\n",
+      "train loss:0.04098975146164546\n",
+      "train loss:0.020153203911476406\n",
+      "train loss:0.09813681362397121\n",
+      "train loss:0.08966461242715992\n",
+      "train loss:0.03556487181409524\n",
+      "train loss:0.031135086140383152\n",
+      "train loss:0.075427380494993\n",
+      "train loss:0.09088747835488568\n",
+      "train loss:0.04843102468633722\n",
+      "train loss:0.08194145561027975\n",
+      "train loss:0.03970010871056074\n",
+      "train loss:0.052467235039644636\n",
+      "train loss:0.0571875775796845\n",
+      "train loss:0.0740930399843841\n",
+      "train loss:0.04182180698148054\n",
+      "train loss:0.02273953447718724\n",
+      "train loss:0.012148913470861979\n",
+      "train loss:0.04004431038695304\n",
+      "train loss:0.03937913100996165\n",
+      "train loss:0.011124047466397122\n",
+      "train loss:0.1864440217354493\n",
+      "train loss:0.021218789312918616\n",
+      "train loss:0.1281210977738702\n",
+      "train loss:0.04261822120339897\n",
+      "train loss:0.023775204894124463\n",
+      "train loss:0.05638743025372759\n",
+      "train loss:0.06025422920033969\n",
+      "train loss:0.08753601655888837\n",
+      "train loss:0.06445368864174017\n",
+      "train loss:0.0253870137970624\n",
+      "train loss:0.02468804362205401\n",
+      "train loss:0.08075696224945124\n",
+      "train loss:0.0968082653075744\n",
+      "train loss:0.05345256760037982\n",
+      "train loss:0.078759401482837\n",
+      "train loss:0.07234009827137024\n",
+      "train loss:0.03358055192448519\n",
+      "train loss:0.06078215902653412\n",
+      "train loss:0.047120935240140556\n",
+      "train loss:0.04528274467866034\n",
+      "train loss:0.05939296983316699\n",
+      "train loss:0.06755601096443561\n",
+      "train loss:0.016771278458672143\n",
+      "train loss:0.034556505922324655\n",
+      "train loss:0.08320646613515487\n",
+      "train loss:0.045332788036973035\n",
+      "train loss:0.019023980197406924\n",
+      "train loss:0.019613289388616215\n",
+      "train loss:0.05887715707903335\n",
+      "train loss:0.014184956666854608\n",
+      "train loss:0.022304138707454957\n",
+      "train loss:0.11618846485353884\n",
+      "train loss:0.07836022142662351\n",
+      "train loss:0.13861602002032533\n",
+      "train loss:0.04643687097017618\n",
+      "train loss:0.062212569614399256\n",
+      "train loss:0.05787083273226525\n",
+      "train loss:0.04442595478467554\n",
+      "train loss:0.08956289234542206\n",
+      "train loss:0.027904148846585593\n",
+      "train loss:0.07016416719489764\n",
+      "train loss:0.12389306242940609\n",
+      "train loss:0.07840964440410897\n",
+      "train loss:0.11866560122382408\n",
+      "train loss:0.029206390721896635\n",
+      "train loss:0.05841501073869744\n",
+      "train loss:0.08519910426449724\n",
+      "train loss:0.056586588061278906\n",
+      "train loss:0.10822636324815565\n",
+      "train loss:0.12501744871777493\n",
+      "train loss:0.020370411689426316\n",
+      "train loss:0.0534594890943022\n",
+      "train loss:0.14211433573668286\n",
+      "train loss:0.05386526062969512\n",
+      "train loss:0.0797436608429531\n",
+      "train loss:0.013665988866859274\n",
+      "train loss:0.027718717944378954\n",
+      "train loss:0.035017636676992\n",
+      "train loss:0.163156948590686\n",
+      "train loss:0.03385964203858918\n",
+      "train loss:0.08691420328495729\n",
+      "train loss:0.07413312079963362\n",
+      "train loss:0.07168917968530943\n",
+      "train loss:0.024949626201607022\n",
+      "train loss:0.03518473588172928\n",
+      "train loss:0.04561504567923984\n",
+      "train loss:0.0899015954612505\n",
+      "train loss:0.14358503168317763\n",
+      "train loss:0.07125518681443482\n",
+      "train loss:0.04273213704605311\n",
+      "train loss:0.06793164448862883\n",
+      "train loss:0.04361133204421479\n",
+      "train loss:0.015162521074986541\n",
+      "train loss:0.02435229341682359\n",
+      "train loss:0.13557899604611084\n",
+      "train loss:0.041077006556161075\n",
+      "train loss:0.15233386119732908\n",
+      "train loss:0.09852214910234552\n",
+      "train loss:0.04291227161181383\n",
+      "train loss:0.10990766844782789\n",
+      "train loss:0.048066107633045314\n",
+      "train loss:0.04177986501365535\n",
+      "train loss:0.02907025811999128\n",
+      "train loss:0.08048198178793292\n",
+      "train loss:0.06452472063418595\n",
+      "train loss:0.04307973648711402\n",
+      "train loss:0.08331951677387364\n",
+      "train loss:0.07095875936942436\n",
+      "train loss:0.15692205593581574\n",
+      "train loss:0.024773810739545873\n",
+      "train loss:0.11091920319039879\n",
+      "train loss:0.13478959386688885\n",
+      "train loss:0.06805854505592704\n",
+      "train loss:0.08484217577611844\n",
+      "train loss:0.08367456603385132\n",
+      "train loss:0.034796016975805394\n",
+      "train loss:0.14165712896609212\n",
+      "train loss:0.08592563395742533\n",
+      "train loss:0.03464520935072061\n",
+      "train loss:0.04693789581668048\n",
+      "train loss:0.036895050194126276\n",
+      "train loss:0.05196067901208619\n",
+      "train loss:0.11537999105655027\n",
+      "train loss:0.05398950218461482\n",
+      "train loss:0.09005745330577584\n",
+      "train loss:0.06820025752624558\n",
+      "train loss:0.03146646256271454\n",
+      "train loss:0.059408047491974585\n",
+      "train loss:0.03477074232091945\n",
+      "train loss:0.08367881753866849\n",
+      "train loss:0.0181358935057658\n",
+      "train loss:0.05785705745130243\n",
+      "train loss:0.03789074751646783\n",
+      "train loss:0.07395876667885583\n",
+      "train loss:0.0887071249714981\n",
+      "train loss:0.04915353717109381\n",
+      "train loss:0.055155166573771994\n",
+      "train loss:0.039925615366869796\n",
+      "train loss:0.027644079279778397\n",
+      "train loss:0.03957921771035004\n",
+      "train loss:0.061840582087598836\n",
+      "train loss:0.058639158071129883\n",
+      "train loss:0.04616279763753968\n",
+      "train loss:0.06671967473926607\n",
+      "train loss:0.13338520673929907\n",
+      "train loss:0.016511088417614377\n",
+      "train loss:0.017658645694663747\n",
+      "train loss:0.02344280774627607\n",
+      "train loss:0.027803178533163234\n",
+      "train loss:0.03777855554116214\n",
+      "train loss:0.02016551528158237\n",
+      "train loss:0.04694604877468267\n",
+      "train loss:0.01832044425981719\n",
+      "train loss:0.030933294960170672\n",
+      "train loss:0.06879934473589998\n",
+      "train loss:0.037054556929143036\n",
+      "train loss:0.060276208314745026\n",
+      "train loss:0.01648777013763984\n",
+      "train loss:0.06735632034618763\n",
+      "train loss:0.09292905329777\n",
+      "train loss:0.046917628638080675\n",
+      "train loss:0.013584727789716176\n",
+      "train loss:0.029939814083395126\n",
+      "train loss:0.02889783531493901\n",
+      "train loss:0.09488982765430805\n",
+      "train loss:0.05124831294898352\n",
+      "train loss:0.1046068524244204\n",
+      "train loss:0.023274559913365685\n",
+      "train loss:0.03561344669091676\n",
+      "train loss:0.03551582379272158\n",
+      "train loss:0.08592576252484023\n",
+      "train loss:0.10148083550599044\n",
+      "train loss:0.041283853199644366\n",
+      "train loss:0.09224789539322728\n",
+      "train loss:0.020118832392777612\n",
+      "train loss:0.039921425988345166\n",
+      "train loss:0.08200291085006482\n",
+      "train loss:0.03749553762444185\n",
+      "train loss:0.05874880974917371\n",
+      "train loss:0.017203002771614687\n",
+      "train loss:0.10200232206642522\n",
+      "train loss:0.04423750812464749\n",
+      "train loss:0.06853598751501458\n",
+      "train loss:0.04790119250473787\n",
+      "train loss:0.0584312998662858\n",
+      "train loss:0.03827917197428352\n",
+      "train loss:0.06025062718909317\n",
+      "train loss:0.06535156468227954\n",
+      "train loss:0.03075143472873197\n",
+      "train loss:0.13399686006093708\n",
+      "train loss:0.025658113018213424\n",
+      "train loss:0.01798137807048421\n",
+      "train loss:0.10264668644669035\n",
+      "train loss:0.018005680602772328\n",
+      "train loss:0.047474608914802305\n",
+      "train loss:0.13315109983469822\n",
+      "train loss:0.08295821336168398\n",
+      "train loss:0.06763206479602331\n",
+      "train loss:0.09728608163024885\n",
+      "train loss:0.04246729023750269\n",
+      "train loss:0.12138181187883362\n",
+      "train loss:0.01449411630083654\n",
+      "train loss:0.020933913276438166\n",
+      "train loss:0.06380191000950926\n",
+      "train loss:0.03600769444132849\n",
+      "train loss:0.04038000254223142\n",
+      "train loss:0.010462192885601563\n",
+      "train loss:0.10390840986745749\n",
+      "train loss:0.14594570469217305\n",
+      "train loss:0.04552737465945402\n",
+      "train loss:0.028582729175989828\n",
+      "train loss:0.05579709649755385\n",
+      "train loss:0.03511207094053389\n",
+      "train loss:0.12314829585807416\n",
+      "train loss:0.0665834116577946\n",
+      "train loss:0.026423862361989237\n",
+      "train loss:0.04937396100946267\n",
+      "train loss:0.016622110324222658\n",
+      "train loss:0.13974429003091465\n",
+      "train loss:0.06997842741034965\n",
+      "train loss:0.0558716321832959\n",
+      "train loss:0.10436785645574359\n",
+      "train loss:0.06625181089670124\n",
+      "train loss:0.12088152434764238\n",
+      "train loss:0.0819074425968852\n",
+      "train loss:0.07601713737755074\n",
+      "train loss:0.11186133967794376\n",
+      "train loss:0.04471513038891496\n",
+      "train loss:0.05338247019111271\n",
+      "train loss:0.08613593605482879\n",
+      "train loss:0.03745666244169576\n",
+      "train loss:0.1323708253102051\n",
+      "train loss:0.037899382754781656\n",
+      "train loss:0.02820300107036538\n",
+      "train loss:0.06741788748350974\n",
+      "train loss:0.013332818841924523\n",
+      "train loss:0.0591343193439284\n",
+      "train loss:0.10986201721380075\n",
+      "train loss:0.03468858558445678\n",
+      "train loss:0.020304802755598678\n",
+      "train loss:0.025605515047301453\n",
+      "train loss:0.02019504763787722\n",
+      "train loss:0.010762056315447228\n",
+      "train loss:0.07693978101486597\n",
+      "train loss:0.11038825257495224\n",
+      "train loss:0.04296557846938932\n",
+      "train loss:0.03870224774036005\n",
+      "train loss:0.04219371534302416\n",
+      "train loss:0.02719706589370883\n",
+      "train loss:0.012759846112978876\n",
+      "train loss:0.15569723904231147\n",
+      "train loss:0.02993185102759269\n",
+      "train loss:0.0434685615220247\n",
+      "train loss:0.05591850582373268\n",
+      "train loss:0.019935645009641623\n",
+      "train loss:0.08645718432281858\n",
+      "train loss:0.039935034640305095\n",
+      "train loss:0.03644574149684492\n",
+      "train loss:0.08636615296847852\n",
+      "train loss:0.03707970155674871\n",
+      "train loss:0.03448267878312523\n",
+      "train loss:0.06445830262356464\n",
+      "train loss:0.012189903394685004\n",
+      "train loss:0.059864701377279665\n",
+      "train loss:0.1001217856575255\n",
+      "train loss:0.07597622763285879\n",
+      "train loss:0.06118351672766566\n",
+      "train loss:0.12243445240512404\n",
+      "train loss:0.03494259393509023\n",
+      "train loss:0.01764094451975399\n",
+      "train loss:0.09644334543779444\n",
+      "train loss:0.026866905395789467\n",
+      "train loss:0.026450158291073333\n",
+      "train loss:0.0684841318635148\n",
+      "train loss:0.0840548821445397\n",
+      "train loss:0.024879763726130162\n",
+      "train loss:0.023057001510374393\n",
+      "train loss:0.07929728824010879\n",
+      "train loss:0.05525044000591223\n",
+      "train loss:0.011350267544838373\n",
+      "train loss:0.04182994015066807\n",
+      "train loss:0.03051335844321477\n",
+      "train loss:0.06632416650926293\n",
+      "train loss:0.0428476361892348\n",
+      "train loss:0.0062603034953816125\n",
+      "train loss:0.020409297696778433\n",
+      "train loss:0.049199368404914796\n",
+      "train loss:0.04491280513800481\n",
+      "train loss:0.008269873865740594\n",
+      "train loss:0.04091457655232547\n",
+      "train loss:0.019124674606559033\n",
+      "train loss:0.06030009271944887\n",
+      "train loss:0.012557463566821579\n",
+      "train loss:0.06612574733274201\n",
+      "train loss:0.035029500852252815\n",
+      "train loss:0.02540040174271506\n",
+      "train loss:0.04060703201870816\n",
+      "train loss:0.03693000008283849\n",
+      "train loss:0.07925116617785798\n",
+      "train loss:0.14524027514174923\n",
+      "train loss:0.04721806205710992\n",
+      "train loss:0.013680909329108076\n",
+      "train loss:0.12602836815080543\n",
+      "train loss:0.006341494890883343\n",
+      "train loss:0.04146889103285086\n",
+      "train loss:0.017767499253415957\n",
+      "train loss:0.015163068340752062\n",
+      "train loss:0.035787022968951976\n",
+      "train loss:0.034416572004085544\n",
+      "train loss:0.0374212946189244\n",
+      "train loss:0.03419911833232119\n",
+      "train loss:0.03655889866802134\n",
+      "train loss:0.025073138221303894\n",
+      "train loss:0.014459049578271438\n",
+      "train loss:0.013110176451010914\n",
+      "train loss:0.052460802711657976\n",
+      "train loss:0.07763140325579747\n",
+      "train loss:0.04450076455546097\n",
+      "train loss:0.10156092573793181\n",
+      "train loss:0.040594369930659464\n",
+      "train loss:0.022372253861977845\n",
+      "train loss:0.027999243608560064\n",
+      "train loss:0.01014064122400368\n",
+      "train loss:0.047410783326050664\n",
+      "train loss:0.0320671189435469\n",
+      "train loss:0.02523131759069993\n",
+      "train loss:0.019840278772253556\n",
+      "train loss:0.028514862942452774\n",
+      "train loss:0.20683046824662998\n",
+      "train loss:0.034146365512526776\n",
+      "train loss:0.03701439958346612\n",
+      "train loss:0.04628542140998229\n",
+      "train loss:0.08848321658220361\n",
+      "train loss:0.07872767687648001\n",
+      "train loss:0.07481723189954154\n",
+      "train loss:0.03340562529592199\n",
+      "train loss:0.14537114772477655\n",
+      "train loss:0.03498823555346578\n",
+      "train loss:0.039822431714362276\n",
+      "train loss:0.0666708398124754\n",
+      "train loss:0.032336989690469185\n",
+      "train loss:0.03861541078809987\n",
+      "train loss:0.02214309982175696\n",
+      "train loss:0.09246402160283287\n",
+      "train loss:0.03790001534825258\n",
+      "train loss:0.04200129761131716\n",
+      "train loss:0.01153175410935548\n",
+      "train loss:0.034210134111359064\n",
+      "train loss:0.01971493001755007\n",
+      "train loss:0.03775839159723452\n",
+      "train loss:0.08919615609810368\n",
+      "train loss:0.06619279218403491\n",
+      "train loss:0.03034763150938963\n",
+      "train loss:0.06952023448928055\n",
+      "train loss:0.057396625670308676\n",
+      "train loss:0.03412360085584292\n",
+      "train loss:0.027903991204281095\n",
+      "train loss:0.14269074200131018\n",
+      "train loss:0.034549718997527464\n",
+      "train loss:0.007388844087012293\n",
+      "train loss:0.026473723001364958\n",
+      "train loss:0.02469271302280803\n",
+      "train loss:0.0632687970531018\n",
+      "train loss:0.05443851498313784\n",
+      "train loss:0.010658633620339912\n",
+      "train loss:0.034033485030832415\n",
+      "train loss:0.06571879556778808\n",
+      "train loss:0.08398287453469168\n",
+      "train loss:0.02832330099154502\n",
+      "train loss:0.0581684308915794\n",
+      "train loss:0.017627295656855203\n",
+      "train loss:0.02167258286667233\n",
+      "train loss:0.06556225541080392\n",
+      "train loss:0.03081741678981517\n",
+      "train loss:0.04738781823782401\n",
+      "train loss:0.04539840767874075\n",
+      "train loss:0.11813078950042279\n",
+      "train loss:0.03681741918873715\n",
+      "train loss:0.030646445073929045\n",
+      "train loss:0.07222061069171093\n",
+      "train loss:0.07696272144272488\n",
+      "train loss:0.04968239020055098\n",
+      "train loss:0.14694227786497108\n",
+      "train loss:0.07791647973048833\n",
+      "train loss:0.015889941512742454\n",
+      "train loss:0.027496627522794168\n",
+      "train loss:0.12336158476925753\n",
+      "train loss:0.07547502933295765\n",
+      "train loss:0.04467441830042589\n",
+      "train loss:0.06465006149658781\n",
+      "train loss:0.03670704491576578\n",
+      "train loss:0.033748157374301486\n",
+      "train loss:0.012540955144292196\n",
+      "train loss:0.029091648284429814\n",
+      "train loss:0.0586576636861099\n",
+      "train loss:0.03677636586040965\n",
+      "train loss:0.03608801091367918\n",
+      "train loss:0.06405947365943786\n",
+      "train loss:0.0429188776370901\n",
+      "train loss:0.028865073259815147\n",
+      "train loss:0.025315575730919096\n",
+      "train loss:0.025502505615783398\n",
+      "train loss:0.07380855750188742\n",
+      "train loss:0.07434889373462236\n",
+      "train loss:0.043104055526087655\n",
+      "train loss:0.1571156797150133\n",
+      "train loss:0.027836103286795266\n",
+      "train loss:0.008919947626485233\n",
+      "train loss:0.0063160584969858624\n",
+      "train loss:0.016603542964793726\n",
+      "train loss:0.032285859117608645\n",
+      "train loss:0.020014028035350916\n",
+      "train loss:0.05079584476956402\n",
+      "train loss:0.09799989274584535\n",
+      "train loss:0.020372067267270703\n",
+      "train loss:0.053322095645597783\n",
+      "train loss:0.03726550893602791\n",
+      "train loss:0.029631068059853938\n",
+      "train loss:0.04631101637495773\n",
+      "train loss:0.018690240478966075\n",
+      "train loss:0.012938090293524196\n",
+      "train loss:0.0754230348153271\n",
+      "train loss:0.04009132321775696\n",
+      "train loss:0.06780921691444922\n",
+      "train loss:0.05305550909330241\n",
+      "train loss:0.040329411586161754\n",
+      "train loss:0.09838564584257745\n",
+      "train loss:0.05712309680730482\n",
+      "train loss:0.0539205732248583\n",
+      "train loss:0.018848984787034232\n",
+      "train loss:0.042045979112386445\n",
+      "train loss:0.04664366911011471\n",
+      "train loss:0.07322112659830272\n",
+      "train loss:0.16154747083629079\n",
+      "train loss:0.023919962775624658\n",
+      "train loss:0.042415621921936444\n",
+      "train loss:0.015103835956812774\n",
+      "train loss:0.06187670922605529\n",
+      "train loss:0.07589368749479879\n",
+      "train loss:0.05827025920301518\n",
+      "train loss:0.027206133912337737\n",
+      "train loss:0.014725775357362612\n",
+      "train loss:0.07703998411982747\n",
+      "train loss:0.053274080209912734\n",
+      "train loss:0.052057757842974735\n",
+      "train loss:0.012625730686558449\n",
+      "train loss:0.058330927850976017\n",
+      "train loss:0.04786823602262728\n",
+      "train loss:0.02968244421845053\n",
+      "train loss:0.02374835955755449\n",
+      "train loss:0.032924633655006065\n",
+      "train loss:0.03628790621262892\n",
+      "train loss:0.04531492796274841\n",
+      "train loss:0.05970151611983907\n",
+      "train loss:0.1572246362382861\n",
+      "train loss:0.016530341090360723\n",
+      "train loss:0.02638194530556753\n",
+      "train loss:0.054518865640588166\n",
+      "train loss:0.049908195642481\n",
+      "train loss:0.03073413692292412\n",
+      "train loss:0.030928819437204744\n",
+      "train loss:0.021250093467341306\n",
+      "train loss:0.020110087098154535\n",
+      "train loss:0.04107049420751659\n",
+      "train loss:0.050407611744619434\n",
+      "train loss:0.04846335658197085\n",
+      "train loss:0.017757523535559438\n",
+      "train loss:0.02151966377912161\n",
+      "train loss:0.08673726767846392\n",
+      "train loss:0.06082214421398044\n",
+      "train loss:0.00996800830014802\n",
+      "train loss:0.0754932471821692\n",
+      "train loss:0.025153582515002153\n",
+      "train loss:0.023979931219336108\n",
+      "train loss:0.022624558538657212\n",
+      "train loss:0.06281443129964444\n",
+      "train loss:0.027755100479489724\n",
+      "train loss:0.029117905496834382\n",
+      "train loss:0.06585365661755327\n",
+      "train loss:0.049151965812813087\n",
+      "train loss:0.10787822162593332\n",
+      "train loss:0.019753697987385483\n",
+      "train loss:0.06746414680130265\n",
+      "train loss:0.05682163802291438\n",
+      "train loss:0.009628217595259243\n",
+      "train loss:0.014089792635322219\n",
+      "train loss:0.021227961059692033\n",
+      "train loss:0.021191828170037764\n",
+      "train loss:0.11097406753328073\n",
+      "train loss:0.035260125925662515\n",
+      "train loss:0.029313728179830082\n",
+      "train loss:0.03679121380850001\n",
+      "train loss:0.10510869339411516\n",
+      "train loss:0.0991333543204398\n",
+      "train loss:0.09467237857310475\n",
+      "train loss:0.032310159367797076\n",
+      "train loss:0.08099686966889293\n",
+      "train loss:0.0418222903956388\n",
+      "train loss:0.015122739870620319\n",
+      "train loss:0.04340836777961496\n",
+      "train loss:0.06043217496512021\n",
+      "train loss:0.019869508542164557\n",
+      "train loss:0.036161688613140515\n",
+      "train loss:0.03981779081543062\n",
+      "train loss:0.06247267170562238\n",
+      "train loss:0.012827533070744006\n",
+      "train loss:0.025984426224935277\n",
+      "train loss:0.0303426062345681\n",
+      "train loss:0.02172733164151032\n",
+      "train loss:0.03245006623507111\n",
+      "train loss:0.021560962152687654\n",
+      "train loss:0.07200249316098933\n",
+      "train loss:0.019406123171017642\n",
+      "train loss:0.03685233363030365\n",
+      "train loss:0.012799415669313148\n",
+      "train loss:0.037633778083260835\n",
+      "train loss:0.06055858083866\n",
+      "train loss:0.017344151544921627\n",
+      "train loss:0.03477420402643999\n",
+      "train loss:0.05125901930270706\n",
+      "train loss:0.059686527768224196\n",
+      "train loss:0.08685356261413933\n",
+      "train loss:0.018253926670954945\n",
+      "train loss:0.05446114312125552\n",
+      "train loss:0.03126957345537809\n",
+      "train loss:0.01931847973518191\n",
+      "train loss:0.037412657753143515\n",
+      "train loss:0.04101066738580608\n",
+      "train loss:0.03784379689132337\n",
+      "train loss:0.037846507165230275\n",
+      "train loss:0.02510898034516871\n",
+      "train loss:0.034141962776980374\n",
+      "train loss:0.03619222317868714\n",
+      "train loss:0.07028937503492216\n",
+      "train loss:0.033840945974039235\n",
+      "train loss:0.02591138860716278\n",
+      "train loss:0.07669984609593772\n",
+      "train loss:0.017170038397862936\n",
+      "train loss:0.058454133157994505\n",
+      "train loss:0.13239693230494456\n",
+      "train loss:0.06806573064724426\n",
+      "train loss:0.034872198291480354\n",
+      "train loss:0.0223273106670127\n",
+      "train loss:0.035186964842929894\n",
+      "train loss:0.043527170895616335\n",
+      "train loss:0.054822788147290163\n",
+      "=== epoch:4, train acc:0.981, test acc:0.979 ===\n",
+      "train loss:0.13614181322749602\n",
+      "train loss:0.13988922369496723\n",
+      "train loss:0.10833527650011698\n",
+      "train loss:0.04614268007161565\n",
+      "train loss:0.03548917010485433\n",
+      "train loss:0.014844252156379929\n",
+      "train loss:0.05070356246657693\n",
+      "train loss:0.024684528715046814\n",
+      "train loss:0.030690620429120588\n",
+      "train loss:0.013350234267475353\n",
+      "train loss:0.0193041440418538\n",
+      "train loss:0.02798069005292076\n",
+      "train loss:0.04609925231934917\n",
+      "train loss:0.017648117787967786\n",
+      "train loss:0.03066912960131231\n",
+      "train loss:0.0578470445635473\n",
+      "train loss:0.056872692023074746\n",
+      "train loss:0.018514735901557104\n",
+      "train loss:0.1457609864335909\n",
+      "train loss:0.022103762410206583\n",
+      "train loss:0.058811314445195925\n",
+      "train loss:0.08607562522264049\n",
+      "train loss:0.05594877061206502\n",
+      "train loss:0.030986711188357705\n",
+      "train loss:0.024577855671668373\n",
+      "train loss:0.029892083738013376\n",
+      "train loss:0.049217267129203275\n",
+      "train loss:0.05122156233649663\n",
+      "train loss:0.03615900806616803\n",
+      "train loss:0.04741256042729694\n",
+      "train loss:0.080978581034061\n",
+      "train loss:0.026239135852536833\n",
+      "train loss:0.014514747078586356\n",
+      "train loss:0.013089905375661855\n",
+      "train loss:0.0377265544030144\n",
+      "train loss:0.04135118246690835\n",
+      "train loss:0.026425884445421023\n",
+      "train loss:0.008149795456225337\n",
+      "train loss:0.07369640630705962\n",
+      "train loss:0.0418173654594102\n",
+      "train loss:0.0243371607541084\n",
+      "train loss:0.034007723379589884\n",
+      "train loss:0.04141732997430591\n",
+      "train loss:0.08038848612538399\n",
+      "train loss:0.12427336987030257\n",
+      "train loss:0.023217787578193615\n",
+      "train loss:0.028303635423850216\n",
+      "train loss:0.02818329856163563\n",
+      "train loss:0.02503510814893104\n",
+      "train loss:0.040257589156875204\n",
+      "train loss:0.04061375801310813\n",
+      "train loss:0.030480978957246732\n",
+      "train loss:0.011772519700736582\n",
+      "train loss:0.0852505522402847\n",
+      "train loss:0.03635929669776829\n",
+      "train loss:0.013171396449634687\n",
+      "train loss:0.014986990574250963\n",
+      "train loss:0.022189748512153062\n",
+      "train loss:0.013096154220236265\n",
+      "train loss:0.017355069339191717\n",
+      "train loss:0.013122592302009649\n",
+      "train loss:0.05869184450663156\n",
+      "train loss:0.011945513869821877\n",
+      "train loss:0.061540439598404204\n",
+      "train loss:0.012099227990322401\n",
+      "train loss:0.10655304880450003\n",
+      "train loss:0.05654435356111708\n",
+      "train loss:0.0945842314261529\n",
+      "train loss:0.05494214004938207\n",
+      "train loss:0.03603414557725487\n",
+      "train loss:0.03213498870855318\n",
+      "train loss:0.16986339195232206\n",
+      "train loss:0.09999790023397653\n",
+      "train loss:0.025875532124061275\n",
+      "train loss:0.012427239408498068\n",
+      "train loss:0.06439533477268883\n",
+      "train loss:0.020918833660888193\n",
+      "train loss:0.050238209179649926\n",
+      "train loss:0.053467919729468925\n",
+      "train loss:0.03144491392916473\n",
+      "train loss:0.05600358198580415\n",
+      "train loss:0.03361222531933963\n",
+      "train loss:0.005173874747623465\n",
+      "train loss:0.012125799646614395\n",
+      "train loss:0.0531731105997076\n",
+      "train loss:0.04261977025460561\n",
+      "train loss:0.030717972037259723\n",
+      "train loss:0.0746676548201474\n",
+      "train loss:0.016992752358439358\n",
+      "train loss:0.0321810788499671\n",
+      "train loss:0.06722245522839128\n",
+      "train loss:0.04811452337746643\n",
+      "train loss:0.04795031293688761\n",
+      "train loss:0.05201633492714608\n",
+      "train loss:0.045174081244077596\n",
+      "train loss:0.1457956801486585\n",
+      "train loss:0.040527620049791185\n",
+      "train loss:0.023969645744845995\n",
+      "train loss:0.015042605415383974\n",
+      "train loss:0.05478779612028682\n",
+      "train loss:0.06474002082414977\n",
+      "train loss:0.043338564750956377\n",
+      "train loss:0.06300198324759813\n",
+      "train loss:0.04101543143126984\n",
+      "train loss:0.02298778270533329\n",
+      "train loss:0.020558657581426304\n",
+      "train loss:0.0168500767856568\n",
+      "train loss:0.017488532175260432\n",
+      "train loss:0.0768625115131354\n",
+      "train loss:0.04848984606156069\n",
+      "train loss:0.019925570364894066\n",
+      "train loss:0.0339660006545067\n",
+      "train loss:0.019782233209377397\n",
+      "train loss:0.012366452091527168\n",
+      "train loss:0.09866589737559263\n",
+      "train loss:0.01959650634240173\n",
+      "train loss:0.06851518964572198\n",
+      "train loss:0.03607800030560808\n",
+      "train loss:0.04421764945881456\n",
+      "train loss:0.0383251970727238\n",
+      "train loss:0.0669849538790062\n",
+      "train loss:0.043291819997729235\n",
+      "train loss:0.019242727898324517\n",
+      "train loss:0.015607788583513393\n",
+      "train loss:0.04480030543206758\n",
+      "train loss:0.07757633916200636\n",
+      "train loss:0.055160495656003346\n",
+      "train loss:0.03137098420872068\n",
+      "train loss:0.021988127803924572\n",
+      "train loss:0.11321140669285713\n",
+      "train loss:0.0341418561154588\n",
+      "train loss:0.04991012628962985\n",
+      "train loss:0.051399771352983556\n",
+      "train loss:0.09319080574600147\n",
+      "train loss:0.025733525331491754\n",
+      "train loss:0.022476467061143205\n",
+      "train loss:0.05195431467387638\n",
+      "train loss:0.03149885131071233\n",
+      "train loss:0.0180124024287705\n",
+      "train loss:0.026295293376507652\n",
+      "train loss:0.04383365598896611\n",
+      "train loss:0.019402770377773364\n",
+      "train loss:0.11170154487525515\n",
+      "train loss:0.02420600379968837\n",
+      "train loss:0.012364180207116651\n",
+      "train loss:0.04503303343179889\n",
+      "train loss:0.02652412921710466\n",
+      "train loss:0.030509419693277556\n",
+      "train loss:0.020312951382274576\n",
+      "train loss:0.02957773909519101\n",
+      "train loss:0.06030876644544739\n",
+      "train loss:0.11139254030082706\n",
+      "train loss:0.050761652256118915\n",
+      "train loss:0.02527325494665129\n",
+      "train loss:0.03706595602508819\n",
+      "train loss:0.02257444550890048\n",
+      "train loss:0.12374047202787468\n",
+      "train loss:0.036956159663021404\n",
+      "train loss:0.028121639818326098\n",
+      "train loss:0.0343223872405414\n",
+      "train loss:0.04633836052693708\n",
+      "train loss:0.08744537063305988\n",
+      "train loss:0.07447677096638253\n",
+      "train loss:0.0892020315963311\n",
+      "train loss:0.04600991489260875\n",
+      "train loss:0.029526899716785814\n",
+      "train loss:0.0373125045471922\n",
+      "train loss:0.03327292346011716\n",
+      "train loss:0.07573235609709578\n",
+      "train loss:0.024631044343714933\n",
+      "train loss:0.02825575443806058\n",
+      "train loss:0.011177742656522964\n",
+      "train loss:0.021062023638705504\n",
+      "train loss:0.053312163467452336\n",
+      "train loss:0.041548690888449015\n",
+      "train loss:0.06068656935268893\n",
+      "train loss:0.06337523512097165\n",
+      "train loss:0.027460329385856314\n",
+      "train loss:0.01524232788681095\n",
+      "train loss:0.024241803825575028\n",
+      "train loss:0.022773957554952543\n",
+      "train loss:0.03732688595429572\n",
+      "train loss:0.04228147749026478\n",
+      "train loss:0.031925600943702184\n",
+      "train loss:0.03452537323278512\n",
+      "train loss:0.05057595295649787\n",
+      "train loss:0.019434567803208794\n",
+      "train loss:0.044453671387025044\n",
+      "train loss:0.03935040771205054\n",
+      "train loss:0.012330569087323339\n",
+      "train loss:0.03076161031196225\n",
+      "train loss:0.04976808579195152\n",
+      "train loss:0.032808591766979804\n",
+      "train loss:0.017318085288330985\n",
+      "train loss:0.013147684752772329\n",
+      "train loss:0.02385837530183242\n",
+      "train loss:0.026112693169644326\n",
+      "train loss:0.01066816972810686\n",
+      "train loss:0.018150539589939438\n",
+      "train loss:0.030592217190000163\n",
+      "train loss:0.007552487850224502\n",
+      "train loss:0.01173743083018775\n",
+      "train loss:0.016166575198821306\n",
+      "train loss:0.043832648503591615\n",
+      "train loss:0.013126329301869308\n",
+      "train loss:0.029196214795349\n",
+      "train loss:0.02576775448934007\n",
+      "train loss:0.013049579677994522\n",
+      "train loss:0.05018195429669202\n",
+      "train loss:0.04192734293479797\n",
+      "train loss:0.06764333561800685\n",
+      "train loss:0.06881088077948604\n",
+      "train loss:0.016078681400486853\n",
+      "train loss:0.08395393778188165\n",
+      "train loss:0.031954376197870914\n",
+      "train loss:0.057234260802195236\n",
+      "train loss:0.07422659560241088\n",
+      "train loss:0.02458792154878174\n",
+      "train loss:0.028493010480675585\n",
+      "train loss:0.07242743535461287\n",
+      "train loss:0.0059309635260938656\n",
+      "train loss:0.009955432537704114\n",
+      "train loss:0.01732257987281039\n",
+      "train loss:0.05637214692616008\n",
+      "train loss:0.07817721211121549\n",
+      "train loss:0.018943668759999616\n",
+      "train loss:0.09344260618592294\n",
+      "train loss:0.03694138625285668\n",
+      "train loss:0.051721259250152106\n",
+      "train loss:0.028287082365712188\n",
+      "train loss:0.0375418093351322\n",
+      "train loss:0.08466879678978372\n",
+      "train loss:0.0300913453920537\n",
+      "train loss:0.04266290359572149\n",
+      "train loss:0.025378505952194416\n",
+      "train loss:0.03507205127451695\n",
+      "train loss:0.026188171570517546\n",
+      "train loss:0.012450306703371905\n",
+      "train loss:0.045488861980187555\n",
+      "train loss:0.01823832900364532\n",
+      "train loss:0.02997984712156706\n",
+      "train loss:0.012182226717740035\n",
+      "train loss:0.016778684019864487\n",
+      "train loss:0.07640362399788073\n",
+      "train loss:0.08225123856270761\n",
+      "train loss:0.05048290604660766\n",
+      "train loss:0.03079470041049888\n",
+      "train loss:0.048023981043107726\n",
+      "train loss:0.016122466709355507\n",
+      "train loss:0.11733706089363125\n",
+      "train loss:0.039691245288628677\n",
+      "train loss:0.03827217399157111\n",
+      "train loss:0.07645618057726254\n",
+      "train loss:0.03669631297868014\n",
+      "train loss:0.016743389122431288\n",
+      "train loss:0.12174758080567687\n",
+      "train loss:0.015708861442442955\n",
+      "train loss:0.05168473413496686\n",
+      "train loss:0.04473490030277154\n",
+      "train loss:0.010077728132718297\n",
+      "train loss:0.013113243553138866\n",
+      "train loss:0.09303271567376963\n",
+      "train loss:0.01606014257184567\n",
+      "train loss:0.043336408888947714\n",
+      "train loss:0.017159083939251463\n",
+      "train loss:0.07082160106971931\n",
+      "train loss:0.02495488893317877\n",
+      "train loss:0.02471899049763301\n",
+      "train loss:0.03871811824079107\n",
+      "train loss:0.027856940662276584\n",
+      "train loss:0.007406736612645449\n",
+      "train loss:0.03904276716073268\n",
+      "train loss:0.04455946621140451\n",
+      "train loss:0.10024383569711581\n",
+      "train loss:0.08508956987266879\n",
+      "train loss:0.09636460218322071\n",
+      "train loss:0.026017708813580183\n",
+      "train loss:0.04062609586773261\n",
+      "train loss:0.06686450946909747\n",
+      "train loss:0.013116701720844108\n",
+      "train loss:0.1393477116726916\n",
+      "train loss:0.027732203012992757\n",
+      "train loss:0.023412825488201054\n",
+      "train loss:0.06792154358011498\n",
+      "train loss:0.04708397857621029\n",
+      "train loss:0.053520405840233966\n",
+      "train loss:0.043912351253682376\n",
+      "train loss:0.07126102205104794\n",
+      "train loss:0.013742772450315506\n",
+      "train loss:0.018909004444973686\n",
+      "train loss:0.044293909467504625\n",
+      "train loss:0.008599236545454607\n",
+      "train loss:0.01571129419938356\n",
+      "train loss:0.11603417261642139\n",
+      "train loss:0.03747668339651636\n",
+      "train loss:0.03921350539860833\n",
+      "train loss:0.08466330882974234\n",
+      "train loss:0.04528024336316496\n",
+      "train loss:0.01593461450951934\n",
+      "train loss:0.029976656355002552\n",
+      "train loss:0.04233776947948484\n",
+      "train loss:0.04417828201971103\n",
+      "train loss:0.04936324009012122\n",
+      "train loss:0.06789237549348354\n",
+      "train loss:0.029213837349509518\n",
+      "train loss:0.0170344732299033\n",
+      "train loss:0.047588903123551074\n",
+      "train loss:0.019338901308146517\n",
+      "train loss:0.023019584936735758\n",
+      "train loss:0.02613053117253797\n",
+      "train loss:0.07356399629397561\n",
+      "train loss:0.02116579236876749\n",
+      "train loss:0.049966853875974014\n",
+      "train loss:0.019829482952088105\n",
+      "train loss:0.011352076331485637\n",
+      "train loss:0.010948602577690359\n",
+      "train loss:0.01898372439769727\n",
+      "train loss:0.0546438527757372\n",
+      "train loss:0.04285483526572184\n",
+      "train loss:0.07527357786051685\n",
+      "train loss:0.03725984628086679\n",
+      "train loss:0.0051647813776405415\n",
+      "train loss:0.03317176552564607\n",
+      "train loss:0.02492494632923693\n",
+      "train loss:0.007676715793209681\n",
+      "train loss:0.04122969400291077\n",
+      "train loss:0.059184607540392425\n",
+      "train loss:0.0807650930507699\n",
+      "train loss:0.03560427085286386\n",
+      "train loss:0.04384423699748794\n",
+      "train loss:0.02417893075882544\n",
+      "train loss:0.030891048619423867\n",
+      "train loss:0.033023619171181\n",
+      "train loss:0.01535828311007754\n",
+      "train loss:0.03359834462945468\n",
+      "train loss:0.018746821210858806\n",
+      "train loss:0.024384826683422466\n",
+      "train loss:0.013296305784594538\n",
+      "train loss:0.012122081493323727\n",
+      "train loss:0.018765061796412037\n",
+      "train loss:0.02307578480040122\n",
+      "train loss:0.009436453005065692\n",
+      "train loss:0.07123311217047058\n",
+      "train loss:0.08051962376378871\n",
+      "train loss:0.05417536417696562\n",
+      "train loss:0.006663246574152439\n",
+      "train loss:0.04956888702335279\n",
+      "train loss:0.010561559346069237\n",
+      "train loss:0.06943960723111972\n",
+      "train loss:0.016526853869173134\n",
+      "train loss:0.062146212376883864\n",
+      "train loss:0.002450905667569948\n",
+      "train loss:0.00862744804113943\n",
+      "train loss:0.09136282923651277\n",
+      "train loss:0.040070715201866725\n",
+      "train loss:0.06826940448483618\n",
+      "train loss:0.05575837921489395\n",
+      "train loss:0.03968476250010683\n",
+      "train loss:0.03173308895838103\n",
+      "train loss:0.021160972769975486\n",
+      "train loss:0.032893812106105313\n",
+      "train loss:0.11832605196546851\n",
+      "train loss:0.028562107400116648\n",
+      "train loss:0.014636658347795024\n",
+      "train loss:0.011869305971502842\n",
+      "train loss:0.01350486409911798\n",
+      "train loss:0.039417617114313803\n",
+      "train loss:0.042802945949964526\n",
+      "train loss:0.029851365625293124\n",
+      "train loss:0.07807229867798837\n",
+      "train loss:0.0595826861014658\n",
+      "train loss:0.009528144026908032\n",
+      "train loss:0.04007820682016095\n",
+      "train loss:0.027478603783065117\n",
+      "train loss:0.012493545450547925\n",
+      "train loss:0.08715134785081526\n",
+      "train loss:0.009116865416806548\n",
+      "train loss:0.029686936192487606\n",
+      "train loss:0.024115112136389425\n",
+      "train loss:0.02083956320178967\n",
+      "train loss:0.008701456767494363\n",
+      "train loss:0.046238745808900104\n",
+      "train loss:0.051415693586310374\n",
+      "train loss:0.08940215798964898\n",
+      "train loss:0.04199110373376825\n",
+      "train loss:0.010637692802635192\n",
+      "train loss:0.018788374759953732\n",
+      "train loss:0.009456496895886388\n",
+      "train loss:0.044643190567696944\n",
+      "train loss:0.07387939421969465\n",
+      "train loss:0.042139229966386464\n",
+      "train loss:0.018619599781026216\n",
+      "train loss:0.05709549602543564\n",
+      "train loss:0.023286786847429786\n",
+      "train loss:0.005982751150704235\n",
+      "train loss:0.01588317396364299\n",
+      "train loss:0.04098587694924442\n",
+      "train loss:0.08545765966993402\n",
+      "train loss:0.12435609478719986\n",
+      "train loss:0.04635327894169161\n",
+      "train loss:0.04813189532127796\n",
+      "train loss:0.009310414346959981\n",
+      "train loss:0.024836236539361137\n",
+      "train loss:0.03976353907650805\n",
+      "train loss:0.05782597304727284\n",
+      "train loss:0.03915233990320147\n",
+      "train loss:0.10341440249939808\n",
+      "train loss:0.03965337120006645\n",
+      "train loss:0.05937576055328322\n",
+      "train loss:0.02544844467914937\n",
+      "train loss:0.09659530959315242\n",
+      "train loss:0.006026892382481983\n",
+      "train loss:0.053048217688267105\n",
+      "train loss:0.013645598823160434\n",
+      "train loss:0.03249592247270706\n",
+      "train loss:0.04956235038579755\n",
+      "train loss:0.07099020812109508\n",
+      "train loss:0.027814754682288428\n",
+      "train loss:0.018373542383196332\n",
+      "train loss:0.03534704662048139\n",
+      "train loss:0.056705300700339585\n",
+      "train loss:0.022705465265217488\n",
+      "train loss:0.0906036655851131\n",
+      "train loss:0.01901031883327721\n",
+      "train loss:0.08937071804587067\n",
+      "train loss:0.03518337626716592\n",
+      "train loss:0.055552532785293145\n",
+      "train loss:0.021090213543935102\n",
+      "train loss:0.02465155949271249\n",
+      "train loss:0.02167844208879302\n",
+      "train loss:0.04094683512076733\n",
+      "train loss:0.010875545364966481\n",
+      "train loss:0.024334988613554195\n",
+      "train loss:0.018045639025233096\n",
+      "train loss:0.061033005051545006\n",
+      "train loss:0.03040594144530039\n",
+      "train loss:0.013155669403693986\n",
+      "train loss:0.020193891015651076\n",
+      "train loss:0.023786280248956026\n",
+      "train loss:0.030108644983596732\n",
+      "train loss:0.017605745121888117\n",
+      "train loss:0.03774153135152125\n",
+      "train loss:0.006163256093386115\n",
+      "train loss:0.07667879741373806\n",
+      "train loss:0.03968374397668613\n",
+      "train loss:0.009774105327088434\n",
+      "train loss:0.09089734905764754\n",
+      "train loss:0.036662499700769194\n",
+      "train loss:0.013951313263617099\n",
+      "train loss:0.015348354014151269\n",
+      "train loss:0.019297980717608403\n",
+      "train loss:0.013361179001670886\n",
+      "train loss:0.05344038251645204\n",
+      "train loss:0.011989950955495593\n",
+      "train loss:0.023522149927243673\n",
+      "train loss:0.026596584510884824\n",
+      "train loss:0.017365738299739215\n",
+      "train loss:0.019513381667836956\n",
+      "train loss:0.04075210228642066\n",
+      "train loss:0.023027861512222015\n",
+      "train loss:0.022090945227686773\n",
+      "train loss:0.028394138566821808\n",
+      "train loss:0.03136113283543265\n",
+      "train loss:0.01781654644261372\n",
+      "train loss:0.05586342790950075\n",
+      "train loss:0.047473396697932595\n",
+      "train loss:0.02972750457642014\n",
+      "train loss:0.014812613434478072\n",
+      "train loss:0.05516837713140136\n",
+      "train loss:0.014521621659025399\n",
+      "train loss:0.010619562161853635\n",
+      "train loss:0.008038083265295023\n",
+      "train loss:0.03345570705178975\n",
+      "train loss:0.04678590762854684\n",
+      "train loss:0.02319321089295287\n",
+      "train loss:0.007363265778514325\n",
+      "train loss:0.009489595303459497\n",
+      "train loss:0.022111319532910133\n",
+      "train loss:0.04804881597477237\n",
+      "train loss:0.03182343505952094\n",
+      "train loss:0.05242816048255374\n",
+      "train loss:0.0092717930724479\n",
+      "train loss:0.053738200315132996\n",
+      "train loss:0.013935322758099897\n",
+      "train loss:0.030318736396207635\n",
+      "train loss:0.0069054228846270015\n",
+      "train loss:0.008155510267331108\n",
+      "train loss:0.02146826009620953\n",
+      "train loss:0.04563870141138913\n",
+      "train loss:0.059097909985536205\n",
+      "train loss:0.008641008368648872\n",
+      "train loss:0.012859331386589816\n",
+      "train loss:0.040027992114809834\n",
+      "train loss:0.04580808349416996\n",
+      "train loss:0.07742667535311282\n",
+      "train loss:0.04445178723785455\n",
+      "train loss:0.018143853547539442\n",
+      "train loss:0.08155413701756407\n",
+      "train loss:0.021905288353233323\n",
+      "train loss:0.018143467788453244\n",
+      "train loss:0.027720284531972682\n",
+      "train loss:0.10984156100482935\n",
+      "train loss:0.014868023100759716\n",
+      "train loss:0.02342783077132851\n",
+      "train loss:0.11819936522106705\n",
+      "train loss:0.01770724305453528\n",
+      "train loss:0.03354682174683446\n",
+      "train loss:0.05685137085887233\n",
+      "train loss:0.16833899750180262\n",
+      "train loss:0.0611629548366938\n",
+      "train loss:0.015089227146599923\n",
+      "train loss:0.059412165573144736\n",
+      "train loss:0.05307108365762954\n",
+      "train loss:0.04717764487426053\n",
+      "train loss:0.0495931004339012\n",
+      "train loss:0.04328323396791903\n",
+      "train loss:0.010930856503913218\n",
+      "train loss:0.05933171731881362\n",
+      "train loss:0.016177524904400897\n",
+      "train loss:0.04065936867002505\n",
+      "train loss:0.02508232658381683\n",
+      "train loss:0.015861414174150608\n",
+      "train loss:0.09535315745519339\n",
+      "train loss:0.014516866890058069\n",
+      "train loss:0.04002762201522025\n",
+      "train loss:0.03135848303713348\n",
+      "train loss:0.009481261243858764\n",
+      "train loss:0.03374038800696127\n",
+      "train loss:0.0063729145447188485\n",
+      "train loss:0.02418746522824089\n",
+      "train loss:0.006510320967766207\n",
+      "train loss:0.021977936407616278\n",
+      "train loss:0.030681519597150483\n",
+      "train loss:0.014501082753709055\n",
+      "train loss:0.014048866725674503\n",
+      "train loss:0.009278231538369475\n",
+      "train loss:0.010909558564343128\n",
+      "train loss:0.026321698859848378\n",
+      "train loss:0.015120527890346035\n",
+      "train loss:0.04342484435232639\n",
+      "train loss:0.01362966640584521\n",
+      "train loss:0.017507649462094273\n",
+      "train loss:0.020479663141357227\n",
+      "train loss:0.016552491071714973\n",
+      "train loss:0.03582646443826552\n",
+      "train loss:0.03505739267780685\n",
+      "train loss:0.05381391474573313\n",
+      "train loss:0.03250499209327579\n",
+      "train loss:0.035389424338323365\n",
+      "train loss:0.02616377851353549\n",
+      "train loss:0.01166299966650289\n",
+      "train loss:0.01062888239504295\n",
+      "train loss:0.0469160333308244\n",
+      "train loss:0.04435325908649602\n",
+      "train loss:0.039798541874441894\n",
+      "train loss:0.0211458157444851\n",
+      "train loss:0.035032664334021094\n",
+      "train loss:0.007571534179136649\n",
+      "train loss:0.007960370753440521\n",
+      "train loss:0.061153089025977075\n",
+      "train loss:0.08216068197656357\n",
+      "train loss:0.03526056946680147\n",
+      "train loss:0.02291616427617228\n",
+      "train loss:0.02093099629225151\n",
+      "train loss:0.026124481797438804\n",
+      "train loss:0.09003327801055433\n",
+      "train loss:0.044252715522238445\n",
+      "train loss:0.03466958849448294\n",
+      "train loss:0.012929143842856712\n",
+      "train loss:0.026722481529615646\n",
+      "train loss:0.03713423012126638\n",
+      "train loss:0.008147078548536713\n",
+      "train loss:0.04605904631925984\n",
+      "train loss:0.07348430327126232\n",
+      "train loss:0.03737583482471938\n",
+      "train loss:0.016824940978696506\n",
+      "train loss:0.031555638865627234\n",
+      "train loss:0.026669423761374423\n",
+      "train loss:0.005998927927007138\n",
+      "train loss:0.03294494754260786\n",
+      "train loss:0.028530425001498563\n",
+      "train loss:0.00485546521429006\n",
+      "train loss:0.04385815811660348\n",
+      "train loss:0.01628513557303745\n",
+      "train loss:0.0439994380580084\n",
+      "train loss:0.009289756987924218\n",
+      "train loss:0.0033686796861901206\n",
+      "train loss:0.02186438838076439\n",
+      "train loss:0.022278984859949655\n",
+      "train loss:0.04743962216546358\n",
+      "train loss:0.038008261548763085\n",
+      "train loss:0.00965300335449213\n",
+      "train loss:0.0062467420473565185\n",
+      "train loss:0.018770777644317633\n",
+      "train loss:0.020399836188197757\n",
+      "train loss:0.00910006617308737\n",
+      "train loss:0.042684605039799886\n",
+      "train loss:0.021595431130570185\n",
+      "train loss:0.026297433336369785\n",
+      "train loss:0.0023942636947489977\n",
+      "=== epoch:5, train acc:0.983, test acc:0.981 ===\n",
+      "train loss:0.017095988987452706\n",
+      "train loss:0.008098281415507128\n",
+      "train loss:0.04306889575835673\n",
+      "train loss:0.028317602949993165\n",
+      "train loss:0.00937382256338845\n",
+      "train loss:0.027557259371400463\n",
+      "train loss:0.01705453597833812\n",
+      "train loss:0.03999394014052395\n",
+      "train loss:0.02665557740904657\n",
+      "train loss:0.022218244569406263\n",
+      "train loss:0.06882688328730285\n",
+      "train loss:0.025012184002542417\n",
+      "train loss:0.01308808975980115\n",
+      "train loss:0.017159241706776494\n",
+      "train loss:0.007477865457946025\n",
+      "train loss:0.01962234249222301\n",
+      "train loss:0.004300529847186558\n",
+      "train loss:0.020419742755230702\n",
+      "train loss:0.03409915168097206\n",
+      "train loss:0.006033702132192825\n",
+      "train loss:0.05429985938335906\n",
+      "train loss:0.01865160460912085\n",
+      "train loss:0.015838293075357737\n",
+      "train loss:0.021304366116336886\n",
+      "train loss:0.061829335347260796\n",
+      "train loss:0.021613217853968737\n",
+      "train loss:0.02853534938286408\n",
+      "train loss:0.00799105729619299\n",
+      "train loss:0.0016651603683949117\n",
+      "train loss:0.03843321465643174\n",
+      "train loss:0.0222760378584557\n",
+      "train loss:0.07151709928548054\n",
+      "train loss:0.041783423556628556\n",
+      "train loss:0.018049289879206953\n",
+      "train loss:0.00481654413188136\n",
+      "train loss:0.08304808186011474\n",
+      "train loss:0.01610027953967244\n",
+      "train loss:0.09017292735933742\n",
+      "train loss:0.012382116113549741\n",
+      "train loss:0.0729317069902299\n",
+      "train loss:0.015323954664926\n",
+      "train loss:0.05940094033874435\n",
+      "train loss:0.01020742408235392\n",
+      "train loss:0.018963831421489167\n",
+      "train loss:0.025172102930687824\n",
+      "train loss:0.05384569269090242\n",
+      "train loss:0.02910512401382535\n",
+      "train loss:0.023079583085780297\n",
+      "train loss:0.02674886351486489\n",
+      "train loss:0.006390561354750049\n",
+      "train loss:0.035921644369447545\n",
+      "train loss:0.022533394648289032\n",
+      "train loss:0.16836191122704655\n",
+      "train loss:0.058010693620808504\n",
+      "train loss:0.02072510392871736\n",
+      "train loss:0.011649438123690061\n",
+      "train loss:0.008948014501926251\n",
+      "train loss:0.04289416209365823\n",
+      "train loss:0.01803896499084558\n",
+      "train loss:0.03721593931141251\n",
+      "train loss:0.018517144958109678\n",
+      "train loss:0.012555075157200278\n",
+      "train loss:0.014029963232926377\n",
+      "train loss:0.06748832918308603\n",
+      "train loss:0.011884430188332698\n",
+      "train loss:0.01638280110326394\n",
+      "train loss:0.029685314685623267\n",
+      "train loss:0.01305635686137228\n",
+      "train loss:0.022498778929189875\n",
+      "train loss:0.00785562572436136\n",
+      "train loss:0.012264734626060496\n",
+      "train loss:0.02266401099694447\n",
+      "train loss:0.038810230406983386\n",
+      "train loss:0.09811677519047485\n",
+      "train loss:0.02870588872604664\n",
+      "train loss:0.053925247201902504\n",
+      "train loss:0.03469421713774433\n",
+      "train loss:0.01608252761733957\n",
+      "train loss:0.011757270443044816\n",
+      "train loss:0.02808974314771624\n",
+      "train loss:0.02214408177845175\n",
+      "train loss:0.0039863750416261464\n",
+      "train loss:0.030374870240101822\n",
+      "train loss:0.005877384865981735\n",
+      "train loss:0.025331193026673807\n",
+      "train loss:0.04708701962082054\n",
+      "train loss:0.09614894005471167\n",
+      "train loss:0.04240250505554264\n",
+      "train loss:0.025621210424236165\n",
+      "train loss:0.023069718497241324\n",
+      "train loss:0.046528284122229036\n",
+      "train loss:0.023032394289578702\n",
+      "train loss:0.03454247907516298\n",
+      "train loss:0.05773781884926804\n",
+      "train loss:0.04885844019295748\n",
+      "train loss:0.012901871547161768\n",
+      "train loss:0.03298469330032046\n",
+      "train loss:0.029661992310190528\n",
+      "train loss:0.03513862272926802\n",
+      "train loss:0.06302991265622816\n",
+      "train loss:0.06962719935258148\n",
+      "train loss:0.029270731367635507\n",
+      "train loss:0.0307688208983392\n",
+      "train loss:0.010764033595985753\n",
+      "train loss:0.03510727913190384\n",
+      "train loss:0.04279451473766488\n",
+      "train loss:0.07720294130257317\n",
+      "train loss:0.012372459397296453\n",
+      "train loss:0.05384296923771666\n",
+      "train loss:0.009159706383174777\n",
+      "train loss:0.011952992580924318\n",
+      "train loss:0.02710975344585748\n",
+      "train loss:0.04654706901272812\n",
+      "train loss:0.04498699941557835\n",
+      "train loss:0.01942016057355337\n",
+      "train loss:0.04180582564496378\n",
+      "train loss:0.02127033520589143\n",
+      "train loss:0.05449973087477402\n",
+      "train loss:0.03374613456224227\n",
+      "train loss:0.053643876594847234\n",
+      "train loss:0.017566297293179572\n",
+      "train loss:0.11751550031552269\n",
+      "train loss:0.013195752775974503\n",
+      "train loss:0.12393382793680552\n",
+      "train loss:0.006774201636979418\n",
+      "train loss:0.015334598223525201\n",
+      "train loss:0.05556436155916192\n",
+      "train loss:0.05127552781682119\n",
+      "train loss:0.037854500872628614\n",
+      "train loss:0.0161583012331388\n",
+      "train loss:0.0129327603928571\n",
+      "train loss:0.045662113199230754\n",
+      "train loss:0.02888910056081264\n",
+      "train loss:0.030065145344262002\n",
+      "train loss:0.04204068134743808\n",
+      "train loss:0.045438433540505245\n",
+      "train loss:0.050868814932999405\n",
+      "train loss:0.03154620458337553\n",
+      "train loss:0.02072307709172329\n",
+      "train loss:0.024964639509414976\n",
+      "train loss:0.01943283544776725\n",
+      "train loss:0.01790409727519633\n",
+      "train loss:0.056586551943306806\n",
+      "train loss:0.008870480714152904\n",
+      "train loss:0.01859376239174915\n",
+      "train loss:0.025234562802935002\n",
+      "train loss:0.061527296650516865\n",
+      "train loss:0.021648801450997786\n",
+      "train loss:0.010172288981383974\n",
+      "train loss:0.014489122251949784\n",
+      "train loss:0.024503193446432946\n",
+      "train loss:0.012760297450635038\n",
+      "train loss:0.014975640079107516\n",
+      "train loss:0.1448136982268607\n",
+      "train loss:0.026009507596949342\n",
+      "train loss:0.03767127968733072\n",
+      "train loss:0.012173960039960267\n",
+      "train loss:0.01637364267102756\n",
+      "train loss:0.010486078613163965\n",
+      "train loss:0.015090144292346239\n",
+      "train loss:0.0064093811464283135\n",
+      "train loss:0.021250038970978342\n",
+      "train loss:0.03891241625527539\n",
+      "train loss:0.028962433728947655\n",
+      "train loss:0.018822670177077346\n",
+      "train loss:0.02772893278000992\n",
+      "train loss:0.07347831057278531\n",
+      "train loss:0.014030013831936052\n",
+      "train loss:0.08913841608837009\n",
+      "train loss:0.02789576654238652\n",
+      "train loss:0.005386522222775727\n",
+      "train loss:0.004754696557030509\n",
+      "train loss:0.032725803332140854\n",
+      "train loss:0.009806252099714426\n",
+      "train loss:0.03184446609558407\n",
+      "train loss:0.022850642574245402\n",
+      "train loss:0.01664000449769053\n",
+      "train loss:0.018296393233202937\n",
+      "train loss:0.0064061108095743915\n",
+      "train loss:0.089008828990426\n",
+      "train loss:0.00980040751796044\n",
+      "train loss:0.11385523031403968\n",
+      "train loss:0.011862929794217249\n",
+      "train loss:0.004572570850792531\n",
+      "train loss:0.06909936065286443\n",
+      "train loss:0.008137406479356353\n",
+      "train loss:0.023238134817015844\n",
+      "train loss:0.022424844839110992\n",
+      "train loss:0.06668334730321585\n",
+      "train loss:0.02298736159296905\n",
+      "train loss:0.06272639399150924\n",
+      "train loss:0.011395905773684923\n",
+      "train loss:0.01079863228024286\n",
+      "train loss:0.03629598611746465\n",
+      "train loss:0.016305845289096604\n",
+      "train loss:0.03895256029724598\n",
+      "train loss:0.01993635585810087\n",
+      "train loss:0.007107632740970197\n",
+      "train loss:0.04542777457230232\n",
+      "train loss:0.014336957331369984\n",
+      "train loss:0.03825131755175121\n",
+      "train loss:0.021694414343555058\n",
+      "train loss:0.020941797814202405\n",
+      "train loss:0.005016189977472932\n",
+      "train loss:0.03905684260868672\n",
+      "train loss:0.005464580929197298\n",
+      "train loss:0.006591308430556645\n",
+      "train loss:0.02393483927320732\n",
+      "train loss:0.03126033271641227\n",
+      "train loss:0.013551417448319052\n",
+      "train loss:0.08536793776518188\n",
+      "train loss:0.02212013023492543\n",
+      "train loss:0.036806035149784595\n",
+      "train loss:0.015383631706004312\n",
+      "train loss:0.016093053493769903\n",
+      "train loss:0.005804047158269695\n",
+      "train loss:0.020333508203021187\n",
+      "train loss:0.01788165587134432\n",
+      "train loss:0.04879351550097906\n",
+      "train loss:0.04063473615767996\n",
+      "train loss:0.061873280563324834\n",
+      "train loss:0.007703442045782726\n",
+      "train loss:0.06116896217329868\n",
+      "train loss:0.006496112923947152\n",
+      "train loss:0.012853510319436587\n",
+      "train loss:0.04460851625194964\n",
+      "train loss:0.014219449765468146\n",
+      "train loss:0.0030760911777916893\n",
+      "train loss:0.007770963446567568\n",
+      "train loss:0.03188989107248431\n",
+      "train loss:0.01623113406473351\n",
+      "train loss:0.019688525787595407\n",
+      "train loss:0.007521447509891916\n",
+      "train loss:0.0720711908399066\n",
+      "train loss:0.03661930602595638\n",
+      "train loss:0.006773563670897645\n",
+      "train loss:0.01234308589355974\n",
+      "train loss:0.01430141785204694\n",
+      "train loss:0.01492218101860366\n",
+      "train loss:0.014053188098878528\n",
+      "train loss:0.03359341767573572\n",
+      "train loss:0.03040206910168641\n",
+      "train loss:0.05401146788339865\n",
+      "train loss:0.021186259445379985\n",
+      "train loss:0.04689376305416235\n",
+      "train loss:0.05613863130592807\n",
+      "train loss:0.006147500233734499\n",
+      "train loss:0.04305127967700785\n",
+      "train loss:0.011057515480120064\n",
+      "train loss:0.012337991748814319\n",
+      "train loss:0.018057558014124203\n",
+      "train loss:0.045581986098472214\n",
+      "train loss:0.10349266732232291\n",
+      "train loss:0.02816349918694025\n",
+      "train loss:0.050292511442864514\n",
+      "train loss:0.008517507221011314\n",
+      "train loss:0.022798546355951372\n",
+      "train loss:0.020420112071190127\n",
+      "train loss:0.04680249359292835\n",
+      "train loss:0.012497157763642176\n",
+      "train loss:0.036195415066314934\n",
+      "train loss:0.020093400835891995\n",
+      "train loss:0.00998850825061128\n",
+      "train loss:0.06345006515353682\n",
+      "train loss:0.04905246947408582\n",
+      "train loss:0.008456969287800696\n",
+      "train loss:0.014182984330892521\n",
+      "train loss:0.023258504719240757\n",
+      "train loss:0.00468004805080663\n",
+      "train loss:0.04195768558757444\n",
+      "train loss:0.04272194093445303\n",
+      "train loss:0.008157321803013406\n",
+      "train loss:0.049625654304874034\n",
+      "train loss:0.025645784919620954\n",
+      "train loss:0.021430100814364215\n",
+      "train loss:0.016689168547674958\n",
+      "train loss:0.04193998696118767\n",
+      "train loss:0.0655205753272174\n",
+      "train loss:0.008383686602650419\n",
+      "train loss:0.04569143901206751\n",
+      "train loss:0.014420874243673219\n",
+      "train loss:0.013661103984437244\n",
+      "train loss:0.011226673082849813\n",
+      "train loss:0.0124075176116647\n",
+      "train loss:0.05558228454207223\n",
+      "train loss:0.07663250684585753\n",
+      "train loss:0.009387446190345757\n",
+      "train loss:0.037449004613392234\n",
+      "train loss:0.021621382918480925\n",
+      "train loss:0.016951133012239543\n",
+      "train loss:0.033870878616162114\n",
+      "train loss:0.02464125159618489\n",
+      "train loss:0.038508642296487185\n",
+      "train loss:0.04881890745206028\n",
+      "train loss:0.061722623100490745\n",
+      "train loss:0.023298774373800374\n",
+      "train loss:0.010013536645651522\n",
+      "train loss:0.06280389848855523\n",
+      "train loss:0.0833669121733403\n",
+      "train loss:0.011435196432317052\n",
+      "train loss:0.024497712919379383\n",
+      "train loss:0.056964434447518045\n",
+      "train loss:0.012575120598953975\n",
+      "train loss:0.01539858587825398\n",
+      "train loss:0.0038066159853119923\n",
+      "train loss:0.01611530606473001\n",
+      "train loss:0.04417326429086628\n",
+      "train loss:0.011513054645964478\n",
+      "train loss:0.048289754580512476\n",
+      "train loss:0.0345798692078198\n",
+      "train loss:0.011736645040028503\n",
+      "train loss:0.009087162977754717\n",
+      "train loss:0.013543503192964374\n",
+      "train loss:0.01354998344792842\n",
+      "train loss:0.029236979528191234\n",
+      "train loss:0.03270522033566634\n",
+      "train loss:0.07163084598570321\n",
+      "train loss:0.08084695777054579\n",
+      "train loss:0.010327067607679027\n",
+      "train loss:0.02279503702346028\n",
+      "train loss:0.011784488359460584\n",
+      "train loss:0.029478907002665734\n",
+      "train loss:0.025860389420885886\n",
+      "train loss:0.013507830221919542\n",
+      "train loss:0.009633279781020717\n",
+      "train loss:0.04909748264579222\n",
+      "train loss:0.0074625753235532475\n",
+      "train loss:0.0159649444150298\n",
+      "train loss:0.05569460050641434\n",
+      "train loss:0.020383448991916173\n",
+      "train loss:0.02687752261369398\n",
+      "train loss:0.031907889655207515\n",
+      "train loss:0.028861521720654097\n",
+      "train loss:0.006542892666866477\n",
+      "train loss:0.008769094665312054\n",
+      "train loss:0.11726929354652901\n",
+      "train loss:0.025975498563194255\n",
+      "train loss:0.015681741411239597\n",
+      "train loss:0.03261597650015279\n",
+      "train loss:0.039731704585880684\n",
+      "train loss:0.015476040448249241\n",
+      "train loss:0.043614952250886685\n",
+      "train loss:0.007998035131048398\n",
+      "train loss:0.08027948498930887\n",
+      "train loss:0.01091418954431316\n",
+      "train loss:0.014918230951769619\n",
+      "train loss:0.02456804917697037\n",
+      "train loss:0.03149043417293889\n",
+      "train loss:0.036412742566632156\n",
+      "train loss:0.011923507571295655\n",
+      "train loss:0.09846704597262629\n",
+      "train loss:0.01974264913315048\n",
+      "train loss:0.06337742352258471\n",
+      "train loss:0.035702344419647966\n",
+      "train loss:0.03832199457767713\n",
+      "train loss:0.008318689651338555\n",
+      "train loss:0.008793883272857943\n",
+      "train loss:0.0126658297242591\n",
+      "train loss:0.014784907243580272\n",
+      "train loss:0.016122636396755853\n",
+      "train loss:0.04079913989355026\n",
+      "train loss:0.015019593442369724\n",
+      "train loss:0.029791335118415314\n",
+      "train loss:0.035136413589467994\n",
+      "train loss:0.03151087615768727\n",
+      "train loss:0.037321616717437216\n",
+      "train loss:0.06366299894790477\n",
+      "train loss:0.014775998201792118\n",
+      "train loss:0.006400090786908084\n",
+      "train loss:0.019791539788059455\n",
+      "train loss:0.024688016046513103\n",
+      "train loss:0.017525644497893247\n",
+      "train loss:0.009138228183983793\n",
+      "train loss:0.0134124382109524\n",
+      "train loss:0.006885662286145861\n",
+      "train loss:0.010577501201775039\n",
+      "train loss:0.04452602942670873\n",
+      "train loss:0.011515543977677543\n",
+      "train loss:0.026485702800371384\n",
+      "train loss:0.05805049754433881\n",
+      "train loss:0.06642304371376652\n",
+      "train loss:0.008990556361734781\n",
+      "train loss:0.036445459420712624\n",
+      "train loss:0.0069789591333029545\n",
+      "train loss:0.03018429256414766\n",
+      "train loss:0.02697318052252881\n",
+      "train loss:0.031482737107046126\n",
+      "train loss:0.006723096681147552\n",
+      "train loss:0.06102262577476979\n",
+      "train loss:0.018887648004186467\n",
+      "train loss:0.04559589916411719\n",
+      "train loss:0.003900071114842368\n",
+      "train loss:0.02772481960683396\n",
+      "train loss:0.02025993960452975\n",
+      "train loss:0.011913896687286523\n",
+      "train loss:0.004094115400812024\n",
+      "train loss:0.024571704621338623\n",
+      "train loss:0.03116340252546991\n",
+      "train loss:0.10159270673220643\n",
+      "train loss:0.007997219120402221\n",
+      "train loss:0.04203617447948042\n",
+      "train loss:0.009928237835694137\n",
+      "train loss:0.05332546576237871\n",
+      "train loss:0.007883874327112796\n",
+      "train loss:0.007029309007842932\n",
+      "train loss:0.0346168048578776\n",
+      "train loss:0.005773420750233992\n",
+      "train loss:0.015934418974424883\n",
+      "train loss:0.030720839013835315\n",
+      "train loss:0.05513804738626352\n",
+      "train loss:0.009156928326512466\n",
+      "train loss:0.07426135213270352\n",
+      "train loss:0.041739443426189744\n",
+      "train loss:0.013709781181498156\n",
+      "train loss:0.01229276535818119\n",
+      "train loss:0.0014734484550771039\n",
+      "train loss:0.013004005655734419\n",
+      "train loss:0.006639542999354017\n",
+      "train loss:0.018974989383488607\n",
+      "train loss:0.012327784565542009\n",
+      "train loss:0.07023792893283048\n",
+      "train loss:0.00963118137959133\n",
+      "train loss:0.009639805466372962\n",
+      "train loss:0.008514268869261104\n",
+      "train loss:0.0030576560713397897\n",
+      "train loss:0.061815967363901284\n",
+      "train loss:0.02700871737161879\n",
+      "train loss:0.004858537184330976\n",
+      "train loss:0.03664470149417212\n",
+      "train loss:0.006607701034542137\n",
+      "train loss:0.029721607232473204\n",
+      "train loss:0.05081868427638586\n",
+      "train loss:0.009889742519923474\n",
+      "train loss:0.04045630739255112\n",
+      "train loss:0.023560100227002133\n",
+      "train loss:0.010340926514865876\n",
+      "train loss:0.05100847991511154\n",
+      "train loss:0.024236908057501387\n",
+      "train loss:0.04771958059749332\n",
+      "train loss:0.04618754540070847\n",
+      "train loss:0.021450436145198468\n",
+      "train loss:0.020370694390459153\n",
+      "train loss:0.01732312081614011\n",
+      "train loss:0.0693120098647547\n",
+      "train loss:0.028670878105961986\n",
+      "train loss:0.04600716689987336\n",
+      "train loss:0.03385108193229938\n",
+      "train loss:0.07236311109704852\n",
+      "train loss:0.01906336893061132\n",
+      "train loss:0.06378749729679921\n",
+      "train loss:0.013664759552495244\n",
+      "train loss:0.007827885583399093\n",
+      "train loss:0.003011138599932629\n",
+      "train loss:0.006940396482276011\n",
+      "train loss:0.02833473145992676\n",
+      "train loss:0.023436785330792947\n",
+      "train loss:0.02295279369985635\n",
+      "train loss:0.03354088773862542\n",
+      "train loss:0.026588207023590233\n",
+      "train loss:0.04596759635351553\n",
+      "train loss:0.002236712822829867\n",
+      "train loss:0.010303357022935597\n",
+      "train loss:0.008915725581576682\n",
+      "train loss:0.017929635811053755\n",
+      "train loss:0.011038224353062375\n",
+      "train loss:0.00537131045556152\n",
+      "train loss:0.04896921082286302\n",
+      "train loss:0.02036726025694237\n",
+      "train loss:0.009360417594650608\n",
+      "train loss:0.011874679823326994\n",
+      "train loss:0.008123551146858428\n",
+      "train loss:0.026214200988157264\n",
+      "train loss:0.009817204087209912\n",
+      "train loss:0.015093179064345388\n",
+      "train loss:0.00790333570551424\n",
+      "train loss:0.01106979781065495\n",
+      "train loss:0.026511433123808815\n",
+      "train loss:0.025969495113651087\n",
+      "train loss:0.01100931621407524\n",
+      "train loss:0.00975215252027202\n",
+      "train loss:0.0023875299004491573\n",
+      "train loss:0.01311614759520336\n",
+      "train loss:0.02113032283504347\n",
+      "train loss:0.00797566786840766\n",
+      "train loss:0.04735253915943206\n",
+      "train loss:0.016051035749898402\n",
+      "train loss:0.021875886105756588\n",
+      "train loss:0.039673667291969\n",
+      "train loss:0.007984697161008908\n",
+      "train loss:0.018587526250840602\n",
+      "train loss:0.005727857255764231\n",
+      "train loss:0.023096294195402534\n",
+      "train loss:0.010827017450730576\n",
+      "train loss:0.007873242062472254\n",
+      "train loss:0.07614649454891623\n",
+      "train loss:0.02014169681198589\n",
+      "train loss:0.044775177769304675\n",
+      "train loss:0.04863137682991883\n",
+      "train loss:0.009775834561340102\n",
+      "train loss:0.002674665451160624\n",
+      "train loss:0.015584260207895807\n",
+      "train loss:0.014698819199822239\n",
+      "train loss:0.02665008422967243\n",
+      "train loss:0.06313605805694891\n",
+      "train loss:0.026219967658215437\n",
+      "train loss:0.01663325585991789\n",
+      "train loss:0.025292518295665648\n",
+      "train loss:0.004012249707736077\n",
+      "train loss:0.024334416855112635\n",
+      "train loss:0.008863768837474537\n",
+      "train loss:0.06890654760831781\n",
+      "train loss:0.012173576736101185\n",
+      "train loss:0.012378656578605396\n",
+      "train loss:0.00518348547272684\n",
+      "train loss:0.005706497269763731\n",
+      "train loss:0.017773730051364525\n",
+      "train loss:0.11304016302431688\n",
+      "train loss:0.005178145210933291\n",
+      "train loss:0.017166041777282412\n",
+      "train loss:0.01734442653737476\n",
+      "train loss:0.010115168846707802\n",
+      "train loss:0.005696258297025692\n",
+      "train loss:0.021783231922563026\n",
+      "train loss:0.020518879684734953\n",
+      "train loss:0.009176450421854167\n",
+      "train loss:0.03029658825515895\n",
+      "train loss:0.01387781266354121\n",
+      "train loss:0.05983839882920775\n",
+      "train loss:0.006616606000482425\n",
+      "train loss:0.009873506411610958\n",
+      "train loss:0.0032954707659283693\n",
+      "train loss:0.05355597172820582\n",
+      "train loss:0.05029590541444521\n",
+      "train loss:0.03009206275593539\n",
+      "train loss:0.05087661919595957\n",
+      "train loss:0.028442390988002293\n",
+      "train loss:0.024273844965507867\n",
+      "train loss:0.025597095373926085\n",
+      "train loss:0.011240999987105151\n",
+      "train loss:0.02254707817723738\n",
+      "train loss:0.009502651467413104\n",
+      "train loss:0.023873318303491722\n",
+      "train loss:0.0925762116034028\n",
+      "train loss:0.04709916423545788\n",
+      "train loss:0.011341409498885685\n",
+      "train loss:0.07188429605758506\n",
+      "train loss:0.01578768306490346\n",
+      "train loss:0.032580558298501594\n",
+      "train loss:0.01989053769262266\n",
+      "train loss:0.018902864322287792\n",
+      "train loss:0.04040934053626476\n",
+      "train loss:0.012381283504163178\n",
+      "train loss:0.01278501098838325\n",
+      "train loss:0.010447434634080993\n",
+      "train loss:0.0540720111354568\n",
+      "train loss:0.006087310338142847\n",
+      "train loss:0.02442577945068002\n",
+      "train loss:0.01140376282353437\n",
+      "train loss:0.007483060343334653\n",
+      "train loss:0.030272211079750783\n",
+      "train loss:0.01884655763454915\n",
+      "train loss:0.01521956998489757\n",
+      "train loss:0.06080215374154393\n",
+      "train loss:0.03130249854550443\n",
+      "train loss:0.02544420718317743\n",
+      "train loss:0.014654788580333564\n",
+      "train loss:0.014230762080392647\n",
+      "train loss:0.026942258796862358\n",
+      "train loss:0.023942650697099343\n",
+      "train loss:0.034089960502718814\n",
+      "train loss:0.014501198021823107\n",
+      "train loss:0.008442944808991494\n",
+      "train loss:0.023501015812579268\n",
+      "train loss:0.08215813645737448\n",
+      "train loss:0.030699545280513166\n",
+      "train loss:0.01356134527142648\n",
+      "train loss:0.021627120703581516\n",
+      "train loss:0.042796588684527966\n",
+      "train loss:0.00188643938644199\n",
+      "train loss:0.01699597556673412\n",
+      "train loss:0.0025558796230022223\n",
+      "train loss:0.13981434407258836\n",
+      "train loss:0.023812390044832564\n",
+      "train loss:0.010875143978309323\n",
+      "train loss:0.05965584018564838\n",
+      "train loss:0.019028090968068564\n",
+      "train loss:0.036251695945578\n",
+      "train loss:0.01688129926996423\n",
+      "train loss:0.0406498638185965\n",
+      "train loss:0.005683759888435546\n",
+      "train loss:0.007697411077721178\n",
+      "train loss:0.02997717260223812\n",
+      "train loss:0.014012592959914192\n",
+      "train loss:0.07405760366864467\n",
+      "train loss:0.010167422732725431\n",
+      "train loss:0.016053342517104642\n",
+      "train loss:0.0353307256561621\n",
+      "train loss:0.09287015778253141\n",
+      "train loss:0.05450366833601098\n",
+      "train loss:0.02378319607048566\n",
+      "=== epoch:6, train acc:0.989, test acc:0.98 ===\n",
+      "train loss:0.020809248777164877\n",
+      "train loss:0.04280698175257031\n",
+      "train loss:0.017105015125009216\n",
+      "train loss:0.02480643181652417\n",
+      "train loss:0.004335607420940324\n",
+      "train loss:0.0066707024321253534\n",
+      "train loss:0.015888614789769034\n",
+      "train loss:0.026481337962065504\n",
+      "train loss:0.004617463381011827\n",
+      "train loss:0.008850178144600969\n",
+      "train loss:0.02021403694571872\n",
+      "train loss:0.03449938466182247\n",
+      "train loss:0.004987728987375381\n",
+      "train loss:0.009986087452975403\n",
+      "train loss:0.007129557704164784\n",
+      "train loss:0.04612682177004778\n",
+      "train loss:0.004539793384710632\n",
+      "train loss:0.011070844772968437\n",
+      "train loss:0.03726291393468726\n",
+      "train loss:0.042042759056089334\n",
+      "train loss:0.023020628004249824\n",
+      "train loss:0.02338505400935517\n",
+      "train loss:0.013702409424299877\n",
+      "train loss:0.011571584104264388\n",
+      "train loss:0.007781663960610931\n",
+      "train loss:0.08555600489780323\n",
+      "train loss:0.020221805791106155\n",
+      "train loss:0.016398284038152567\n",
+      "train loss:0.017715909346725317\n",
+      "train loss:0.022693110671105254\n",
+      "train loss:0.023355153591307357\n",
+      "train loss:0.07508612429666342\n",
+      "train loss:0.011188550551325012\n",
+      "train loss:0.02257848564753707\n",
+      "train loss:0.03882399363296872\n",
+      "train loss:0.011891487438093339\n",
+      "train loss:0.021111339134118286\n",
+      "train loss:0.053538516976313726\n",
+      "train loss:0.13704307004187113\n",
+      "train loss:0.01363197185565789\n",
+      "train loss:0.014717143480828398\n",
+      "train loss:0.017249178214105387\n",
+      "train loss:0.015911285906090637\n",
+      "train loss:0.044813092838316894\n",
+      "train loss:0.008342447643476029\n",
+      "train loss:0.032588644829345544\n",
+      "train loss:0.007508165954416296\n",
+      "train loss:0.07931734144922352\n",
+      "train loss:0.03543794432522584\n",
+      "train loss:0.016442647426072765\n",
+      "train loss:0.009903655663909695\n",
+      "train loss:0.028365729910649642\n",
+      "train loss:0.010630553336799181\n",
+      "train loss:0.00864170714345996\n",
+      "train loss:0.007712137468477478\n",
+      "train loss:0.0050083897903676015\n",
+      "train loss:0.017651377345426652\n",
+      "train loss:0.012815898071447857\n",
+      "train loss:0.010914556049725094\n",
+      "train loss:0.0291804926442202\n",
+      "train loss:0.01617441537623946\n",
+      "train loss:0.0017531476885635146\n",
+      "train loss:0.006503723039688103\n",
+      "train loss:0.029059484323284927\n",
+      "train loss:0.009178402928734973\n",
+      "train loss:0.00688284281267498\n",
+      "train loss:0.02839486387559476\n",
+      "train loss:0.0032965387406802167\n",
+      "train loss:0.018697185005644653\n",
+      "train loss:0.08311024440658882\n",
+      "train loss:0.030859987849921836\n",
+      "train loss:0.008758973210287591\n",
+      "train loss:0.008675839181584292\n",
+      "train loss:0.012010352255444728\n",
+      "train loss:0.02221867624183986\n",
+      "train loss:0.00959431661722213\n",
+      "train loss:0.007325754770276197\n",
+      "train loss:0.014143295288370572\n",
+      "train loss:0.010791072766207783\n",
+      "train loss:0.0306841807090863\n",
+      "train loss:0.05550896505468139\n",
+      "train loss:0.07295187011947447\n",
+      "train loss:0.04075188398537205\n",
+      "train loss:0.013545378335510543\n",
+      "train loss:0.007912904307139439\n",
+      "train loss:0.012665277473073541\n",
+      "train loss:0.022260847694021865\n",
+      "train loss:0.012177260441226528\n",
+      "train loss:0.004811047183380167\n",
+      "train loss:0.011052420578776992\n",
+      "train loss:0.007320963423369932\n",
+      "train loss:0.03322347851413692\n",
+      "train loss:0.052999147843802455\n",
+      "train loss:0.026652177674814138\n",
+      "train loss:0.0034866583499630565\n",
+      "train loss:0.017479355979730434\n",
+      "train loss:0.010388929979624666\n",
+      "train loss:0.017454088436366973\n",
+      "train loss:0.06829203291404755\n",
+      "train loss:0.090404587899744\n",
+      "train loss:0.013939262256811165\n",
+      "train loss:0.018373310569176026\n",
+      "train loss:0.08236765334820077\n",
+      "train loss:0.02968142171111124\n",
+      "train loss:0.004040345957426434\n",
+      "train loss:0.0103926982510638\n",
+      "train loss:0.026504593036992992\n",
+      "train loss:0.04963288057987163\n",
+      "train loss:0.028227288641843064\n",
+      "train loss:0.019718280103822438\n",
+      "train loss:0.007316882577192631\n",
+      "train loss:0.03255116396121127\n",
+      "train loss:0.015594388733193376\n",
+      "train loss:0.02338481954196949\n",
+      "train loss:0.004604076518188741\n",
+      "train loss:0.020686477267039406\n",
+      "train loss:0.020174174671110295\n",
+      "train loss:0.01332586988831263\n",
+      "train loss:0.0375977097757921\n",
+      "train loss:0.01695982794590465\n",
+      "train loss:0.016483160469200692\n",
+      "train loss:0.003932326104475061\n",
+      "train loss:0.00794813884657515\n",
+      "train loss:0.019995672413912768\n",
+      "train loss:0.008277479731439922\n",
+      "train loss:0.013398558649394855\n",
+      "train loss:0.0365928278432557\n",
+      "train loss:0.008770371166473202\n",
+      "train loss:0.03713701522660624\n",
+      "train loss:0.007627181672050879\n",
+      "train loss:0.07282916447850438\n",
+      "train loss:0.0395614718058807\n",
+      "train loss:0.006402829326788623\n",
+      "train loss:0.030514612717519846\n",
+      "train loss:0.009762341810865825\n",
+      "train loss:0.0356118128398574\n",
+      "train loss:0.011794326742686876\n",
+      "train loss:0.005787927547976388\n",
+      "train loss:0.00636566955014915\n",
+      "train loss:0.014835765502923601\n",
+      "train loss:0.004808219909978044\n",
+      "train loss:0.01343151477325226\n",
+      "train loss:0.00827235684972218\n",
+      "train loss:0.027298212501995937\n",
+      "train loss:0.004225733640621753\n",
+      "train loss:0.00419484189172584\n",
+      "train loss:0.013756640556676197\n",
+      "train loss:0.013297527587317209\n",
+      "train loss:0.003370542018613678\n",
+      "train loss:0.031200207631927726\n",
+      "train loss:0.0181918627707107\n",
+      "train loss:0.008927528011449765\n",
+      "train loss:0.0485700409815767\n",
+      "train loss:0.014291742027193298\n",
+      "train loss:0.05748806484747143\n",
+      "train loss:0.06309089118716399\n",
+      "train loss:0.009966624492865015\n",
+      "train loss:0.10204753717030586\n",
+      "train loss:0.023737723579144675\n",
+      "train loss:0.016535873345276876\n",
+      "train loss:0.008029569555983586\n",
+      "train loss:0.06330236607431491\n",
+      "train loss:0.02133109822014214\n",
+      "train loss:0.007208229037177019\n",
+      "train loss:0.0061294552717994725\n",
+      "train loss:0.05819418846312024\n",
+      "train loss:0.016620021320978912\n",
+      "train loss:0.006390185412670376\n",
+      "train loss:0.019739165107361466\n",
+      "train loss:0.04241637983492183\n",
+      "train loss:0.04134732607630026\n",
+      "train loss:0.022852496088015425\n",
+      "train loss:0.048362392433810575\n",
+      "train loss:0.007046400899976949\n",
+      "train loss:0.013783171250045911\n",
+      "train loss:0.012312347583093117\n",
+      "train loss:0.017827247133830406\n",
+      "train loss:0.015346978905298174\n",
+      "train loss:0.0070869532087255585\n",
+      "train loss:0.016079784971554777\n",
+      "train loss:0.0031406955758659615\n",
+      "train loss:0.020580962071854142\n",
+      "train loss:0.015485631934724195\n",
+      "train loss:0.018534733918832067\n",
+      "train loss:0.06579958152296225\n",
+      "train loss:0.008409560794497672\n",
+      "train loss:0.0066497772206427115\n",
+      "train loss:0.01770137127132873\n",
+      "train loss:0.11056042088244396\n",
+      "train loss:0.027622815775052423\n",
+      "train loss:0.00952313127326749\n",
+      "train loss:0.04230439518534692\n",
+      "train loss:0.01698358872400787\n",
+      "train loss:0.013193334176323415\n",
+      "train loss:0.024487914234480376\n",
+      "train loss:0.009521120310506568\n",
+      "train loss:0.009586749646341453\n",
+      "train loss:0.02053839755548124\n",
+      "train loss:0.0018680961820362954\n",
+      "train loss:0.003521511662360626\n",
+      "train loss:0.009097020123734322\n",
+      "train loss:0.027322081480953567\n",
+      "train loss:0.008001863295541341\n",
+      "train loss:0.010433397225368648\n",
+      "train loss:0.012435983905054286\n",
+      "train loss:0.028241333710143906\n",
+      "train loss:0.010132116888925475\n",
+      "train loss:0.0055021906522837085\n",
+      "train loss:0.03839104305953717\n",
+      "train loss:0.02248541226764173\n",
+      "train loss:0.07061904910937153\n",
+      "train loss:0.00724093298883528\n",
+      "train loss:0.011663888443150732\n",
+      "train loss:0.010033318025087619\n",
+      "train loss:0.007270615132072935\n",
+      "train loss:0.01337949896567193\n",
+      "train loss:0.0018178027441787545\n",
+      "train loss:0.01881740482226054\n",
+      "train loss:0.024195709622437726\n",
+      "train loss:0.014417367426050551\n",
+      "train loss:0.005868396534104707\n",
+      "train loss:0.015800635352561477\n",
+      "train loss:0.03394670831082867\n",
+      "train loss:0.029960360576263844\n",
+      "train loss:0.00947264029031772\n",
+      "train loss:0.023058400149015154\n",
+      "train loss:0.011660883603437852\n",
+      "train loss:0.031687907130747164\n",
+      "train loss:0.07179361501887215\n",
+      "train loss:0.0032143372588013253\n",
+      "train loss:0.03395803980616228\n",
+      "train loss:0.005243012675934622\n",
+      "train loss:0.010924672845739867\n",
+      "train loss:0.012418562859866583\n",
+      "train loss:0.06737098313607133\n",
+      "train loss:0.006620871992735097\n",
+      "train loss:0.013187435742194942\n",
+      "train loss:0.012692665917296309\n",
+      "train loss:0.012582268126119305\n",
+      "train loss:0.017638209988133518\n",
+      "train loss:0.011585321961192685\n",
+      "train loss:0.004775141845797196\n",
+      "train loss:0.007765355102380515\n",
+      "train loss:0.013413790410501036\n",
+      "train loss:0.03449292291724224\n",
+      "train loss:0.042268189564898834\n",
+      "train loss:0.008596861583998134\n",
+      "train loss:0.013040475619343333\n",
+      "train loss:0.016815119209621848\n",
+      "train loss:0.01954748835542172\n",
+      "train loss:0.011815020596097086\n",
+      "train loss:0.0045935401017322235\n",
+      "train loss:0.008578487457604568\n",
+      "train loss:0.03272032217171873\n",
+      "train loss:0.014769505128623659\n",
+      "train loss:0.025126460712147122\n",
+      "train loss:0.016005973807077283\n",
+      "train loss:0.05205634947309013\n",
+      "train loss:0.006670259699277926\n",
+      "train loss:0.01020395123604235\n",
+      "train loss:0.000963726298429902\n",
+      "train loss:0.0014061441426578899\n",
+      "train loss:0.006406628192806986\n",
+      "train loss:0.07805146706324514\n",
+      "train loss:0.017276147708167176\n",
+      "train loss:0.00945453907317787\n",
+      "train loss:0.001824946182836841\n",
+      "train loss:0.018477602011928763\n",
+      "train loss:0.0496541783553689\n",
+      "train loss:0.0155697796699337\n",
+      "train loss:0.013423364891226586\n",
+      "train loss:0.008976504714780119\n",
+      "train loss:0.002362965249241658\n",
+      "train loss:0.023871342964506114\n",
+      "train loss:0.006813483988632424\n",
+      "train loss:0.03536700397754598\n",
+      "train loss:0.008244565671430514\n",
+      "train loss:0.007542800108929964\n",
+      "train loss:0.012281122088792356\n",
+      "train loss:0.0021101015131190647\n",
+      "train loss:0.09382630933134532\n",
+      "train loss:0.009596858154450362\n",
+      "train loss:0.03282508864367207\n",
+      "train loss:0.005535628894779722\n",
+      "train loss:0.04242968742890021\n",
+      "train loss:0.0030988353432914325\n",
+      "train loss:0.007547963556639127\n",
+      "train loss:0.0048338991205182135\n",
+      "train loss:0.020166337373904688\n",
+      "train loss:0.034963244083627255\n",
+      "train loss:0.007416021969552986\n",
+      "train loss:0.02500987911456712\n",
+      "train loss:0.004474938928318284\n",
+      "train loss:0.0176622105949477\n",
+      "train loss:0.035477215329794524\n",
+      "train loss:0.033573083841638415\n",
+      "train loss:0.00888968474856313\n",
+      "train loss:0.05245175152114598\n",
+      "train loss:0.020228257088735338\n",
+      "train loss:0.0015132314248476567\n",
+      "train loss:0.03515202084663895\n",
+      "train loss:0.04601995718271242\n",
+      "train loss:0.009882265255643577\n",
+      "train loss:0.015798593874407307\n",
+      "train loss:0.03562433516997582\n",
+      "train loss:0.011082431952416769\n",
+      "train loss:0.01846445969182892\n",
+      "train loss:0.006666818631239749\n",
+      "train loss:0.00508080930230332\n",
+      "train loss:0.05511118372453045\n",
+      "train loss:0.024219163948958358\n",
+      "train loss:0.009827913417693568\n",
+      "train loss:0.011167715089300033\n",
+      "train loss:0.010866441833718255\n",
+      "train loss:0.03876497503112247\n",
+      "train loss:0.006720832828052224\n",
+      "train loss:0.014625509010896744\n",
+      "train loss:0.009791070505243818\n",
+      "train loss:0.06489966361341203\n",
+      "train loss:0.0020924118714172844\n",
+      "train loss:0.027728298249511226\n",
+      "train loss:0.02575405568788292\n",
+      "train loss:0.01458250052695387\n",
+      "train loss:0.014604386406094478\n",
+      "train loss:0.04352435070686962\n",
+      "train loss:0.09421235654432694\n",
+      "train loss:0.01193188385359771\n",
+      "train loss:0.023771528884129558\n",
+      "train loss:0.008537026032471674\n",
+      "train loss:0.042209363550771456\n",
+      "train loss:0.00857605591599234\n",
+      "train loss:0.04334155753907901\n",
+      "train loss:0.0036502978131723748\n",
+      "train loss:0.014688374408946304\n",
+      "train loss:0.09604820958658035\n",
+      "train loss:0.04547479052305297\n",
+      "train loss:0.05762434855648964\n",
+      "train loss:0.044509508417865655\n",
+      "train loss:0.015572171923829084\n",
+      "train loss:0.02013043099953425\n",
+      "train loss:0.00913032639806441\n",
+      "train loss:0.013106500122036829\n",
+      "train loss:0.012797703742500168\n",
+      "train loss:0.05748894826467701\n",
+      "train loss:0.0128157081504665\n",
+      "train loss:0.006474387865598263\n",
+      "train loss:0.007611726465849926\n",
+      "train loss:0.0094723432905301\n",
+      "train loss:0.025128763286226798\n",
+      "train loss:0.014201280146699434\n",
+      "train loss:0.006679451633874179\n",
+      "train loss:0.00946767158450158\n",
+      "train loss:0.02629641212877858\n",
+      "train loss:0.028221081752611954\n",
+      "train loss:0.021127732086048033\n",
+      "train loss:0.046554268452795394\n",
+      "train loss:0.02300109916244846\n",
+      "train loss:0.03000244535618763\n",
+      "train loss:0.004553741526860236\n",
+      "train loss:0.004223980636905411\n",
+      "train loss:0.06037396922711741\n",
+      "train loss:0.08647350050406283\n",
+      "train loss:0.015635447391812524\n",
+      "train loss:0.009281619381977418\n",
+      "train loss:0.006955378025979687\n",
+      "train loss:0.08583653305074501\n",
+      "train loss:0.005995203918140393\n",
+      "train loss:0.013606224032355698\n",
+      "train loss:0.0037605139788986913\n",
+      "train loss:0.06056741984961788\n",
+      "train loss:0.008795500444621365\n",
+      "train loss:0.02243822122181579\n",
+      "train loss:0.013597047737893352\n",
+      "train loss:0.014575443760547232\n",
+      "train loss:0.021529233500228586\n",
+      "train loss:0.010887616303205288\n",
+      "train loss:0.026033443694321742\n",
+      "train loss:0.015014614474419976\n",
+      "train loss:0.03157149724011072\n",
+      "train loss:0.01748235718533156\n",
+      "train loss:0.0659784342826122\n",
+      "train loss:0.01045424986890562\n",
+      "train loss:0.010778221115326584\n",
+      "train loss:0.04762640307022289\n",
+      "train loss:0.0351749023831557\n",
+      "train loss:0.006818524397292334\n",
+      "train loss:0.009663595485421707\n",
+      "train loss:0.015305341907841946\n",
+      "train loss:0.013731277959440246\n",
+      "train loss:0.030446067264604523\n",
+      "train loss:0.010583714858779306\n",
+      "train loss:0.008488690135433422\n",
+      "train loss:0.040323098304507844\n",
+      "train loss:0.09008042188949603\n",
+      "train loss:0.060739413204336436\n",
+      "train loss:0.029482785463964923\n",
+      "train loss:0.02287122268258054\n",
+      "train loss:0.004581488923680552\n",
+      "train loss:0.05552950065450073\n",
+      "train loss:0.005660653052420261\n",
+      "train loss:0.009871270694412435\n",
+      "train loss:0.024992273084008837\n",
+      "train loss:0.087800080806733\n",
+      "train loss:0.023512781330453997\n",
+      "train loss:0.013832496156696319\n",
+      "train loss:0.01884421689547292\n",
+      "train loss:0.013468720280050204\n",
+      "train loss:0.04331903432659285\n",
+      "train loss:0.04206645302852786\n",
+      "train loss:0.016255073497263853\n",
+      "train loss:0.006729655932782086\n",
+      "train loss:0.007932791190439605\n",
+      "train loss:0.018923371226395657\n",
+      "train loss:0.016635688821256843\n",
+      "train loss:0.03442371748349411\n",
+      "train loss:0.003809161880378331\n",
+      "train loss:0.014782499889974337\n",
+      "train loss:0.11931809083427497\n",
+      "train loss:0.013729190082938653\n",
+      "train loss:0.02491886629830096\n",
+      "train loss:0.00861392913424082\n",
+      "train loss:0.009577094004522451\n",
+      "train loss:0.03425863465460088\n",
+      "train loss:0.06162996251306315\n",
+      "train loss:0.003175046852826711\n",
+      "train loss:0.01944478884329686\n",
+      "train loss:0.024359434305519016\n",
+      "train loss:0.0769988054903299\n",
+      "train loss:0.01798669165426874\n",
+      "train loss:0.013286320173288395\n",
+      "train loss:0.04180297193878093\n",
+      "train loss:0.02916838278115042\n",
+      "train loss:0.016312773777493465\n",
+      "train loss:0.001757112669200138\n",
+      "train loss:0.013231937928831948\n",
+      "train loss:0.004028227780415562\n",
+      "train loss:0.019123590964783322\n",
+      "train loss:0.020042015896531783\n",
+      "train loss:0.028172625926464395\n",
+      "train loss:0.004978880667782554\n",
+      "train loss:0.014892803907064907\n",
+      "train loss:0.004845275340388292\n",
+      "train loss:0.008070869650068323\n",
+      "train loss:0.00213941730511085\n",
+      "train loss:0.022372723927750506\n",
+      "train loss:0.04139379189260781\n",
+      "train loss:0.008583796967161881\n",
+      "train loss:0.01939548438543638\n",
+      "train loss:0.012157173404279385\n",
+      "train loss:0.09187485188003638\n",
+      "train loss:0.016892263971905686\n",
+      "train loss:0.007055641865045124\n",
+      "train loss:0.018715324468625217\n",
+      "train loss:0.005107659303215895\n",
+      "train loss:0.006314653963376458\n",
+      "train loss:0.020918695440553265\n",
+      "train loss:0.015301526925996624\n",
+      "train loss:0.04924879956626792\n",
+      "train loss:0.007560923725372831\n",
+      "train loss:0.01734750175897683\n",
+      "train loss:0.016664405623518074\n",
+      "train loss:0.014619609323934553\n",
+      "train loss:0.005180201742792677\n",
+      "train loss:0.019198321853120183\n",
+      "train loss:0.012808655931170653\n",
+      "train loss:0.0553536056575699\n",
+      "train loss:0.008208572938657233\n",
+      "train loss:0.07439308202567248\n",
+      "train loss:0.026584619141343458\n",
+      "train loss:0.014378758444315737\n",
+      "train loss:0.01790251722272139\n",
+      "train loss:0.0167106853795964\n",
+      "train loss:0.01022915958982189\n",
+      "train loss:0.026129249189144686\n",
+      "train loss:0.01453828792646734\n",
+      "train loss:0.029444484186945265\n",
+      "train loss:0.016466315150008445\n",
+      "train loss:0.024808405469490206\n",
+      "train loss:0.00681718129510558\n",
+      "train loss:0.022525561931176833\n",
+      "train loss:0.10888930553988722\n",
+      "train loss:0.004414452703650684\n",
+      "train loss:0.023993379372217002\n",
+      "train loss:0.03461587819976671\n",
+      "train loss:0.009733105875241987\n",
+      "train loss:0.008337443998287222\n",
+      "train loss:0.03674844934300464\n",
+      "train loss:0.008975536240320575\n",
+      "train loss:0.013796523910091452\n",
+      "train loss:0.02481327421055235\n",
+      "train loss:0.03525529791958709\n",
+      "train loss:0.0062176611036669095\n",
+      "train loss:0.04700332452112433\n",
+      "train loss:0.014211643596859598\n",
+      "train loss:0.0014481593593348007\n",
+      "train loss:0.001476257599833817\n",
+      "train loss:0.02884078735023594\n",
+      "train loss:0.015460063245078077\n",
+      "train loss:0.03726499656596056\n",
+      "train loss:0.03310023421438287\n",
+      "train loss:0.016772415026359497\n",
+      "train loss:0.006426788777046181\n",
+      "train loss:0.02404109757440666\n",
+      "train loss:0.003843115753492566\n",
+      "train loss:0.17042807733733983\n",
+      "train loss:0.0028106675140630666\n",
+      "train loss:0.03839473777233248\n",
+      "train loss:0.005198466921000345\n",
+      "train loss:0.0068515156051725855\n",
+      "train loss:0.0070158557471584814\n",
+      "train loss:0.014001461366159412\n",
+      "train loss:0.024622266788174463\n",
+      "train loss:0.0060924982993270415\n",
+      "train loss:0.007047428139295148\n",
+      "train loss:0.03928759429278078\n",
+      "train loss:0.058711560658202516\n",
+      "train loss:0.016049865713206112\n",
+      "train loss:0.01159218975309086\n",
+      "train loss:0.0025535382261230423\n",
+      "train loss:0.01199029609550222\n",
+      "train loss:0.006528752888731282\n",
+      "train loss:0.006602675591130948\n",
+      "train loss:0.00644016764576374\n",
+      "train loss:0.018638713918810744\n",
+      "train loss:0.006403925965340916\n",
+      "train loss:0.009753049564898328\n",
+      "train loss:0.006009829552255877\n",
+      "train loss:0.011916095453845475\n",
+      "train loss:0.007762599685915638\n",
+      "train loss:0.002278832418287037\n",
+      "train loss:0.024976695477014154\n",
+      "train loss:0.005630587745236326\n",
+      "train loss:0.005716754527749716\n",
+      "train loss:0.05606594371997517\n",
+      "train loss:0.009753323580953455\n",
+      "train loss:0.006864570047499329\n",
+      "train loss:0.004394556600792008\n",
+      "train loss:0.031854836929854025\n",
+      "train loss:0.08713623539533412\n",
+      "train loss:0.007617256651817057\n",
+      "train loss:0.0064244164160508896\n",
+      "train loss:0.0094187076717414\n",
+      "train loss:0.025009123439645443\n",
+      "train loss:0.009369160535721193\n",
+      "train loss:0.02179726368129152\n",
+      "train loss:0.006921784434882303\n",
+      "train loss:0.009914411826559258\n",
+      "train loss:0.007745551151296837\n",
+      "train loss:0.017325211121772936\n",
+      "train loss:0.008910167379908594\n",
+      "train loss:0.17275655999161074\n",
+      "train loss:0.059960336701004514\n",
+      "train loss:0.011283323617067274\n",
+      "train loss:0.05822223309671057\n",
+      "train loss:0.006264379769804215\n",
+      "train loss:0.010462987789936295\n",
+      "train loss:0.01772587230367596\n",
+      "train loss:0.006394170444722862\n",
+      "train loss:0.015381989843846484\n",
+      "train loss:0.012783933149471285\n",
+      "train loss:0.005074295169901638\n",
+      "train loss:0.00895331960890494\n",
+      "train loss:0.015279933063073712\n",
+      "train loss:0.019940291476679167\n",
+      "train loss:0.026197801343484577\n",
+      "train loss:0.012717396925290841\n",
+      "train loss:0.014562950116072552\n",
+      "train loss:0.011479106332857605\n",
+      "train loss:0.0028764223284449887\n",
+      "train loss:0.016716245108294062\n",
+      "train loss:0.0033800538139871256\n",
+      "train loss:0.007849184720110347\n",
+      "train loss:0.05538361777910181\n",
+      "train loss:0.004790928962094059\n",
+      "train loss:0.003648742086366079\n",
+      "train loss:0.023045236845168765\n",
+      "train loss:0.015479244570466637\n",
+      "train loss:0.007249725515925362\n",
+      "train loss:0.002423734183757652\n",
+      "train loss:0.0033663786363238575\n",
+      "train loss:0.03196931514242186\n",
+      "train loss:0.04815781478369884\n",
+      "train loss:0.04087771032113952\n",
+      "train loss:0.018112580668646083\n",
+      "train loss:0.018207163193823032\n",
+      "train loss:0.04027555505264094\n",
+      "train loss:0.01864160435013278\n",
+      "train loss:0.012038654539406396\n",
+      "train loss:0.0065278446483718324\n",
+      "train loss:0.003176104004295105\n",
+      "train loss:0.027249023854944446\n",
+      "train loss:0.013597704769899013\n",
+      "train loss:0.005234227194714566\n",
+      "train loss:0.029929844808225466\n",
+      "train loss:0.015075793903695234\n",
+      "train loss:0.007420803878997204\n",
+      "train loss:0.03492714893684354\n",
+      "train loss:0.0010460917381492074\n",
+      "train loss:0.008766901115125552\n",
+      "train loss:0.008394361552983889\n",
+      "=== epoch:7, train acc:0.992, test acc:0.984 ===\n",
+      "train loss:0.014363152128975291\n",
+      "train loss:0.007033571856582509\n",
+      "train loss:0.00207400986837361\n",
+      "train loss:0.021342008518471563\n",
+      "train loss:0.026526496406864754\n",
+      "train loss:0.008965823927418175\n",
+      "train loss:0.006329275010948764\n",
+      "train loss:0.013414149621443156\n",
+      "train loss:0.005274858741432123\n",
+      "train loss:0.03608319718655305\n",
+      "train loss:0.013796150662729609\n",
+      "train loss:0.027935838530122444\n",
+      "train loss:0.014246459652915284\n",
+      "train loss:0.002576427493742615\n",
+      "train loss:0.021562228300364333\n",
+      "train loss:0.006823621051468793\n",
+      "train loss:0.006251712513224987\n",
+      "train loss:0.015581864805458806\n",
+      "train loss:0.04296794850731553\n",
+      "train loss:0.004041320889850882\n",
+      "train loss:0.003411330574374225\n",
+      "train loss:0.011241020163566783\n",
+      "train loss:0.010655350566330848\n",
+      "train loss:0.014678064547694747\n",
+      "train loss:0.008884638029416277\n",
+      "train loss:0.02419531262617672\n",
+      "train loss:0.003733305753473124\n",
+      "train loss:0.040042168568008504\n",
+      "train loss:0.009986210271175211\n",
+      "train loss:0.017770856720439585\n",
+      "train loss:0.02656050791691103\n",
+      "train loss:0.012055411708328625\n",
+      "train loss:0.0816090260400478\n",
+      "train loss:0.0053247059497629\n",
+      "train loss:0.08529995226475455\n",
+      "train loss:0.011312500746456068\n",
+      "train loss:0.021146097247947963\n",
+      "train loss:0.015295137892420773\n",
+      "train loss:0.07509325305210517\n",
+      "train loss:0.003350348613527308\n",
+      "train loss:0.0025223577425391797\n",
+      "train loss:0.004715878018087583\n",
+      "train loss:0.02518813292472962\n",
+      "train loss:0.011763721378790946\n",
+      "train loss:0.01056085595857297\n",
+      "train loss:0.004847316526981664\n",
+      "train loss:0.028382207263317447\n",
+      "train loss:0.012449762822862302\n",
+      "train loss:0.002931660850223981\n",
+      "train loss:0.03405963270183723\n",
+      "train loss:0.009592873770136708\n",
+      "train loss:0.020162923304524093\n",
+      "train loss:0.012252188513421969\n",
+      "train loss:0.02810013889851174\n",
+      "train loss:0.05374381701932334\n",
+      "train loss:0.007556332458723394\n",
+      "train loss:0.022563195919169773\n",
+      "train loss:0.047539202947133846\n",
+      "train loss:0.01032140674668398\n",
+      "train loss:0.008579042123116571\n",
+      "train loss:0.010687201792094838\n",
+      "train loss:0.007835774474332771\n",
+      "train loss:0.0430494199309749\n",
+      "train loss:0.008644640142217459\n",
+      "train loss:0.012418644273915612\n",
+      "train loss:0.03417668873803087\n",
+      "train loss:0.012950387161486176\n",
+      "train loss:0.004636599817612791\n",
+      "train loss:0.015489634860942559\n",
+      "train loss:0.025711003730775495\n",
+      "train loss:0.00776655514189615\n",
+      "train loss:0.027870541781678974\n",
+      "train loss:0.008480991681513268\n",
+      "train loss:0.014379170996818817\n",
+      "train loss:0.014603185413184916\n",
+      "train loss:0.043524455113276315\n",
+      "train loss:0.003931136502361108\n",
+      "train loss:0.010104617488532239\n",
+      "train loss:0.010399669409552819\n",
+      "train loss:0.006149821481402505\n",
+      "train loss:0.008840725204710598\n",
+      "train loss:0.003107019269070701\n",
+      "train loss:0.022486994571328914\n",
+      "train loss:0.03192835605176301\n",
+      "train loss:0.012652827159555933\n",
+      "train loss:0.043860271624739246\n",
+      "train loss:0.0043416024547938145\n",
+      "train loss:0.0038607310767981005\n",
+      "train loss:0.00986150033097764\n",
+      "train loss:0.008533167692103234\n",
+      "train loss:0.005992597513263568\n",
+      "train loss:0.012659494935829954\n",
+      "train loss:0.047858932483589106\n",
+      "train loss:0.015057222156716068\n",
+      "train loss:0.02586517822098546\n",
+      "train loss:0.014861708746690669\n",
+      "train loss:0.009851921913698847\n",
+      "train loss:0.0021482008666622106\n",
+      "train loss:0.02165878048885429\n",
+      "train loss:0.008311261942350951\n",
+      "train loss:0.002163258515290414\n",
+      "train loss:0.01814440206335869\n",
+      "train loss:0.028363030676533944\n",
+      "train loss:0.009341430181495866\n",
+      "train loss:0.0009920270989337678\n",
+      "train loss:0.06247628539930543\n",
+      "train loss:0.0036645202408386248\n",
+      "train loss:0.007946195169319879\n",
+      "train loss:0.018042394597269126\n",
+      "train loss:0.0036832216509757572\n",
+      "train loss:0.019221854794545922\n",
+      "train loss:0.032884351372743474\n",
+      "train loss:0.004913280772990965\n",
+      "train loss:0.010708391246496831\n",
+      "train loss:0.040321614176530934\n",
+      "train loss:0.021824585148909934\n",
+      "train loss:0.01072804057609878\n",
+      "train loss:0.011101843941788848\n",
+      "train loss:0.04833970357100647\n",
+      "train loss:0.011699028636338999\n",
+      "train loss:0.038523296794532945\n",
+      "train loss:0.002368993146874696\n",
+      "train loss:0.005459042645553226\n",
+      "train loss:0.03285749287812087\n",
+      "train loss:0.011662558349676038\n",
+      "train loss:0.02883893807335081\n",
+      "train loss:0.005314211213439711\n",
+      "train loss:0.00769611829094939\n",
+      "train loss:0.06178752487310951\n",
+      "train loss:0.012383609766296728\n",
+      "train loss:0.015545084493242456\n",
+      "train loss:0.009831084420418565\n",
+      "train loss:0.01649733090060461\n",
+      "train loss:0.05560059563919494\n",
+      "train loss:0.01133156372527748\n",
+      "train loss:0.010719384817566712\n",
+      "train loss:0.010654250263565175\n",
+      "train loss:0.020503679174513515\n",
+      "train loss:0.03151693436327652\n",
+      "train loss:0.011083856542041257\n",
+      "train loss:0.004527602848336248\n",
+      "train loss:0.03264321160835543\n",
+      "train loss:0.031195452956825336\n",
+      "train loss:0.028121098593957332\n",
+      "train loss:0.0027629460419038103\n",
+      "train loss:0.019525868872497017\n",
+      "train loss:0.03891598516178136\n",
+      "train loss:0.02663446716353741\n",
+      "train loss:0.020629498434949368\n",
+      "train loss:0.01570608413297307\n",
+      "train loss:0.022456750603344674\n",
+      "train loss:0.011619676624447595\n",
+      "train loss:0.004043010781027087\n",
+      "train loss:0.02988201341911239\n",
+      "train loss:0.002212225534314845\n",
+      "train loss:0.00621161171310503\n",
+      "train loss:0.016974232901566293\n",
+      "train loss:0.006308452982926863\n",
+      "train loss:0.024569746358214448\n",
+      "train loss:0.02031279147065749\n",
+      "train loss:0.040659484214987114\n",
+      "train loss:0.037435371219900794\n",
+      "train loss:0.01340061094574412\n",
+      "train loss:0.009193936685592189\n",
+      "train loss:0.014206831456849374\n",
+      "train loss:0.007689300198808823\n",
+      "train loss:0.048131408862488165\n",
+      "train loss:0.0059429844283658785\n",
+      "train loss:0.01470783507550877\n",
+      "train loss:0.023932739810270786\n",
+      "train loss:0.0060946840422595495\n",
+      "train loss:0.023073930856874564\n",
+      "train loss:0.013526778177852835\n",
+      "train loss:0.008583918452476209\n",
+      "train loss:0.01327449803742816\n",
+      "train loss:0.02351905443662581\n",
+      "train loss:0.02863079732038587\n",
+      "train loss:0.01826028327589507\n",
+      "train loss:0.00666902799333393\n",
+      "train loss:0.11166022168876749\n",
+      "train loss:0.020817956600958042\n",
+      "train loss:0.007685404098795664\n",
+      "train loss:0.015349065096869598\n",
+      "train loss:0.009954958593046419\n",
+      "train loss:0.018029230286965705\n",
+      "train loss:0.014655237554577387\n",
+      "train loss:0.01710249593209224\n",
+      "train loss:0.046123056645654845\n",
+      "train loss:0.004382201803201445\n",
+      "train loss:0.0036846418342378778\n",
+      "train loss:0.012533632349470731\n",
+      "train loss:0.014104488427985262\n",
+      "train loss:0.0024453364020094763\n",
+      "train loss:0.00646673064122037\n",
+      "train loss:0.006621904930166631\n",
+      "train loss:0.01721134708970607\n",
+      "train loss:0.007798966058060198\n",
+      "train loss:0.15940305075065822\n",
+      "train loss:0.023864539893656656\n",
+      "train loss:0.006192842495972677\n",
+      "train loss:0.004196187767460145\n",
+      "train loss:0.013183759238438668\n",
+      "train loss:0.019317353313337577\n",
+      "train loss:0.013771731266497101\n",
+      "train loss:0.007201579902966346\n",
+      "train loss:0.03728649756450392\n",
+      "train loss:0.021453172101904557\n",
+      "train loss:0.052217893748569916\n",
+      "train loss:0.030656459819851635\n",
+      "train loss:0.025115302515607448\n",
+      "train loss:0.013954450059839509\n",
+      "train loss:0.030631855260316906\n",
+      "train loss:0.024892148939642967\n",
+      "train loss:0.005322851323437121\n",
+      "train loss:0.019077598711773672\n",
+      "train loss:0.017767554640900768\n",
+      "train loss:0.0053154111360156\n",
+      "train loss:0.006536347554404853\n",
+      "train loss:0.0219608371618859\n",
+      "train loss:0.04428057680141047\n",
+      "train loss:0.0402166119532345\n",
+      "train loss:0.0059780776010570245\n",
+      "train loss:0.009021707210029872\n",
+      "train loss:0.012341925440219234\n",
+      "train loss:0.005900041008139614\n",
+      "train loss:0.02406410412074043\n",
+      "train loss:0.006421765485764834\n",
+      "train loss:0.011101518218551414\n",
+      "train loss:0.009258883635582667\n",
+      "train loss:0.002515876209831508\n",
+      "train loss:0.030381721782078427\n",
+      "train loss:0.017898353406574266\n",
+      "train loss:0.013711717441749477\n",
+      "train loss:0.03969650142333577\n",
+      "train loss:0.007117225995278333\n",
+      "train loss:0.025524273090788068\n",
+      "train loss:0.02177278312606385\n",
+      "train loss:0.058234123226577054\n",
+      "train loss:0.028671340915029583\n",
+      "train loss:0.005603496464231311\n",
+      "train loss:0.022158018428129457\n",
+      "train loss:0.008337314536157273\n",
+      "train loss:0.006884801935540347\n",
+      "train loss:0.007856757678701027\n",
+      "train loss:0.04864678765233454\n",
+      "train loss:0.00715003859888284\n",
+      "train loss:0.01330522893308704\n",
+      "train loss:0.017471331318194352\n",
+      "train loss:0.0024226722412337533\n",
+      "train loss:0.0058350060130743465\n",
+      "train loss:0.022433290536393598\n",
+      "train loss:0.0021547910421011924\n",
+      "train loss:0.017350735079439548\n",
+      "train loss:0.009916483902562475\n",
+      "train loss:0.03233025776120434\n",
+      "train loss:0.0378808141269913\n",
+      "train loss:0.02206430232851824\n",
+      "train loss:0.009400569493048521\n",
+      "train loss:0.01290913621675159\n",
+      "train loss:0.005088646863681676\n",
+      "train loss:0.014754463063714474\n",
+      "train loss:0.009480875997029127\n",
+      "train loss:0.019470763736186635\n",
+      "train loss:0.025190169549207133\n",
+      "train loss:0.008035343361491946\n",
+      "train loss:0.007386943633280055\n",
+      "train loss:0.009618307273291347\n",
+      "train loss:0.011070836386088944\n",
+      "train loss:0.027271724963280265\n",
+      "train loss:0.004646972846692528\n",
+      "train loss:0.003643521645253227\n",
+      "train loss:0.006917292476410163\n",
+      "train loss:0.04386017340914776\n",
+      "train loss:0.016535452855640224\n",
+      "train loss:0.0101545261378036\n",
+      "train loss:0.005747958451925652\n",
+      "train loss:0.003390419386604301\n",
+      "train loss:0.007943567269061919\n",
+      "train loss:0.00610782806077898\n",
+      "train loss:0.029633757672674288\n",
+      "train loss:0.014854622974257903\n",
+      "train loss:0.02432686501394976\n",
+      "train loss:0.017683333387315064\n",
+      "train loss:0.027976597088941547\n",
+      "train loss:0.012336642869798638\n",
+      "train loss:0.0075193808220231076\n",
+      "train loss:0.014439623823788782\n",
+      "train loss:0.024689507898593886\n",
+      "train loss:0.0073097132066460725\n",
+      "train loss:0.01776995320896947\n",
+      "train loss:0.0021541356345245984\n",
+      "train loss:0.0897373246590249\n",
+      "train loss:0.025657760503645\n",
+      "train loss:0.04003103119372525\n",
+      "train loss:0.008969370927062612\n",
+      "train loss:0.00799312473944502\n",
+      "train loss:0.00628174619196412\n",
+      "train loss:0.014288259074925843\n",
+      "train loss:0.00963024367807126\n",
+      "train loss:0.0076484364416707605\n",
+      "train loss:0.00752217527026754\n",
+      "train loss:0.003560848529980571\n",
+      "train loss:0.04124146141483469\n",
+      "train loss:0.004618601817126495\n",
+      "train loss:0.00368024215717066\n",
+      "train loss:0.016217907868054427\n",
+      "train loss:0.010943147074148245\n",
+      "train loss:0.024328260585589675\n",
+      "train loss:0.00945326415269631\n",
+      "train loss:0.0229466077663924\n",
+      "train loss:0.02728666787741791\n",
+      "train loss:0.02145675702284539\n",
+      "train loss:0.0033196942344786307\n",
+      "train loss:0.0051663436679038124\n",
+      "train loss:0.008812735109810737\n",
+      "train loss:0.010446019306689118\n",
+      "train loss:0.001254150567656198\n",
+      "train loss:0.06264211109514997\n",
+      "train loss:0.043633433170441265\n",
+      "train loss:0.022623229893667586\n",
+      "train loss:0.05558981320344354\n",
+      "train loss:0.017063276592825333\n",
+      "train loss:0.0231707259146441\n",
+      "train loss:0.004118752492904156\n",
+      "train loss:0.010163786140910007\n",
+      "train loss:0.014117815276155321\n",
+      "train loss:0.0317017723446522\n",
+      "train loss:0.010065792622957028\n",
+      "train loss:0.009860071948840674\n",
+      "train loss:0.022121662970992637\n",
+      "train loss:0.052454222811119465\n",
+      "train loss:0.01312443177526744\n",
+      "train loss:0.01266901972727701\n",
+      "train loss:0.007825183558742793\n",
+      "train loss:0.009671488267010463\n",
+      "train loss:0.07328856714564126\n",
+      "train loss:0.005572795406987286\n",
+      "train loss:0.07371606718809837\n",
+      "train loss:0.02119768298536345\n",
+      "train loss:0.06260563836734909\n",
+      "train loss:0.0030866455046565083\n",
+      "train loss:0.005922206548110341\n",
+      "train loss:0.01270328808865814\n",
+      "train loss:0.00785176751940913\n",
+      "train loss:0.027307905727675844\n",
+      "train loss:0.005502989722169633\n",
+      "train loss:0.004995125634151989\n",
+      "train loss:0.012597066268042278\n",
+      "train loss:0.03007996134421315\n",
+      "train loss:0.012294691384766031\n",
+      "train loss:0.004695693189457263\n",
+      "train loss:0.030599744969299455\n",
+      "train loss:0.0644139319828548\n",
+      "train loss:0.012476506262290118\n",
+      "train loss:0.005178610998507178\n",
+      "train loss:0.01821324849370761\n",
+      "train loss:0.040298871508969256\n",
+      "train loss:0.010290388011271967\n",
+      "train loss:0.030130984794933945\n",
+      "train loss:0.01921773685236579\n",
+      "train loss:0.011127433600632806\n",
+      "train loss:0.009735098285788078\n",
+      "train loss:0.007190130549762142\n",
+      "train loss:0.03603240692084499\n",
+      "train loss:0.009068636765998485\n",
+      "train loss:0.017215803498406998\n",
+      "train loss:0.0077486260732619085\n",
+      "train loss:0.011593460956297888\n",
+      "train loss:0.01696817670529409\n",
+      "train loss:0.0035439050588639766\n",
+      "train loss:0.004801727434136162\n",
+      "train loss:0.01723353960890429\n",
+      "train loss:0.009705226897251973\n",
+      "train loss:0.015326823276635442\n",
+      "train loss:0.024420954549795412\n",
+      "train loss:0.005955983313453221\n",
+      "train loss:0.03308454176183849\n",
+      "train loss:0.012631694593257272\n",
+      "train loss:0.006062934840941384\n",
+      "train loss:0.04570629158793685\n",
+      "train loss:0.019852783607094908\n",
+      "train loss:0.009735469759912374\n",
+      "train loss:0.007926842781988932\n",
+      "train loss:0.007230916851472053\n",
+      "train loss:0.004864160477983795\n",
+      "train loss:0.002332295318183278\n",
+      "train loss:0.1063405696480648\n",
+      "train loss:0.006527483332482776\n",
+      "train loss:0.014585383217740909\n",
+      "train loss:0.023073884398483666\n",
+      "train loss:0.003781017411351894\n",
+      "train loss:0.0031843643715775993\n",
+      "train loss:0.022211919423445768\n",
+      "train loss:0.002136537029752058\n",
+      "train loss:0.004131961555302059\n",
+      "train loss:0.004601664513029659\n",
+      "train loss:0.009712801848585785\n",
+      "train loss:0.03974725383817571\n",
+      "train loss:0.1663924812823563\n",
+      "train loss:0.00708416656528604\n",
+      "train loss:0.0020625837776648763\n",
+      "train loss:0.017585136906216423\n",
+      "train loss:0.02407724626572578\n",
+      "train loss:0.01749412490380966\n",
+      "train loss:0.004038499865478538\n",
+      "train loss:0.012333787401712017\n",
+      "train loss:0.0025478915833981918\n",
+      "train loss:0.011230201575166066\n",
+      "train loss:0.00576784267547954\n",
+      "train loss:0.02286531318816793\n",
+      "train loss:0.02119561155070296\n",
+      "train loss:0.006077234347561947\n",
+      "train loss:0.011842499595788432\n",
+      "train loss:0.0042998511720604045\n",
+      "train loss:0.00719357706814513\n",
+      "train loss:0.04165019940820077\n",
+      "train loss:0.0030839571683261295\n",
+      "train loss:0.08185212293743707\n",
+      "train loss:0.021181538289021598\n",
+      "train loss:0.009429015655982243\n",
+      "train loss:0.013804640483235072\n",
+      "train loss:0.01711994629973957\n",
+      "train loss:0.01609049200044616\n",
+      "train loss:0.03396107125755674\n",
+      "train loss:0.0023360200660687055\n",
+      "train loss:0.0065796978640807425\n",
+      "train loss:0.03240199632893357\n",
+      "train loss:0.030434352942398078\n",
+      "train loss:0.009886626811411107\n",
+      "train loss:0.019410740189931035\n",
+      "train loss:0.022462868897990648\n",
+      "train loss:0.009666583080274634\n",
+      "train loss:0.008636079974488862\n",
+      "train loss:0.015131176037914694\n",
+      "train loss:0.015029995820075416\n",
+      "train loss:0.04527779850282123\n",
+      "train loss:0.007639782480623933\n",
+      "train loss:0.01577361815656739\n",
+      "train loss:0.0204563701528341\n",
+      "train loss:0.020191438740370483\n",
+      "train loss:0.0025414807849445716\n",
+      "train loss:0.07886268292127162\n",
+      "train loss:0.004015873111450965\n",
+      "train loss:0.005558770026089423\n",
+      "train loss:0.0029117849513192328\n",
+      "train loss:0.013559245188811946\n",
+      "train loss:0.023034526197556082\n",
+      "train loss:0.00698040840185576\n",
+      "train loss:0.04707286779088357\n",
+      "train loss:0.0020386559574113046\n",
+      "train loss:0.06783117294214014\n",
+      "train loss:0.014909139659848329\n",
+      "train loss:0.007048599011483463\n",
+      "train loss:0.02523067044373471\n",
+      "train loss:0.018031181927503435\n",
+      "train loss:0.005862193678891379\n",
+      "train loss:0.018992040485632973\n",
+      "train loss:0.02255649360772989\n",
+      "train loss:0.058734127690382054\n",
+      "train loss:0.006090374211113886\n",
+      "train loss:0.025425463783895663\n",
+      "train loss:0.07694182786300266\n",
+      "train loss:0.00960502916868856\n",
+      "train loss:0.0013352768816037566\n",
+      "train loss:0.012369892033534755\n",
+      "train loss:0.024136181234736882\n",
+      "train loss:0.012660552382455992\n",
+      "train loss:0.03735774129006351\n",
+      "train loss:0.012331078923173372\n",
+      "train loss:0.004150231492646033\n",
+      "train loss:0.0035187963938592788\n",
+      "train loss:0.009330733142246337\n",
+      "train loss:0.004293828723471361\n",
+      "train loss:0.0031809202856163898\n",
+      "train loss:0.011213016407908068\n",
+      "train loss:0.0044832766354900855\n",
+      "train loss:0.008365248319843744\n",
+      "train loss:0.03483174563318356\n",
+      "train loss:0.005723816203313538\n",
+      "train loss:0.008183860981425546\n",
+      "train loss:0.01288570705462485\n",
+      "train loss:0.010466264110265928\n",
+      "train loss:0.00922315214734169\n",
+      "train loss:0.017080562533579396\n",
+      "train loss:0.013830754247914544\n",
+      "train loss:0.002588578252469427\n",
+      "train loss:0.002359247575686825\n",
+      "train loss:0.016162101734307342\n",
+      "train loss:0.005075084313396704\n",
+      "train loss:0.0032079218419668854\n",
+      "train loss:0.027926539592094447\n",
+      "train loss:0.02142640135431657\n",
+      "train loss:0.04877488850101355\n",
+      "train loss:0.012394591317917345\n",
+      "train loss:0.010728689203011434\n",
+      "train loss:0.14569719162453457\n",
+      "train loss:0.04841627533799971\n",
+      "train loss:0.014361576885948819\n",
+      "train loss:0.006172124020638247\n",
+      "train loss:0.008089555276782248\n",
+      "train loss:0.05746295625815996\n",
+      "train loss:0.0038345826046908916\n",
+      "train loss:0.008902853076649534\n",
+      "train loss:0.004645527908415149\n",
+      "train loss:0.011128551721278373\n",
+      "train loss:0.017938099674691857\n",
+      "train loss:0.013707771300109965\n",
+      "train loss:0.0072043673652052356\n",
+      "train loss:0.023091222172878078\n",
+      "train loss:0.07377486297611247\n",
+      "train loss:0.016062807798677387\n",
+      "train loss:0.045640311039927745\n",
+      "train loss:0.014356970250411705\n",
+      "train loss:0.005646443773492038\n",
+      "train loss:0.012110067862024995\n",
+      "train loss:0.004810121038577918\n",
+      "train loss:0.003268293906416672\n",
+      "train loss:0.0033686946610144443\n",
+      "train loss:0.005736047400881381\n",
+      "train loss:0.06429421301984206\n",
+      "train loss:0.018375345433429526\n",
+      "train loss:0.003669635944194465\n",
+      "train loss:0.014497197391153481\n",
+      "train loss:0.00325389202371195\n",
+      "train loss:0.010850390073940739\n",
+      "train loss:0.0365870416416772\n",
+      "train loss:0.014290351039244113\n",
+      "train loss:0.0028335354840447204\n",
+      "train loss:0.015307730010777015\n",
+      "train loss:0.007404959239719277\n",
+      "train loss:0.11928628908763883\n",
+      "train loss:0.007827484057883282\n",
+      "train loss:0.02258216970514461\n",
+      "train loss:0.0696743315409074\n",
+      "train loss:0.02098224022802101\n",
+      "train loss:0.0019155324389726117\n",
+      "train loss:0.04760008703517712\n",
+      "train loss:0.003447484877698972\n",
+      "train loss:0.005534484721110139\n",
+      "train loss:0.0014192055852549068\n",
+      "train loss:0.00484296873300258\n",
+      "train loss:0.004462903194064829\n",
+      "train loss:0.007473824152032996\n",
+      "train loss:0.025723245644247016\n",
+      "train loss:0.003750161396883716\n",
+      "train loss:0.0019801565129629536\n",
+      "train loss:0.003564030234293655\n",
+      "train loss:0.0022077374051524606\n",
+      "train loss:0.0008252602922706741\n",
+      "train loss:0.06500577851965735\n",
+      "train loss:0.022751735866429857\n",
+      "train loss:0.027450737537127335\n",
+      "train loss:0.004034057556957521\n",
+      "train loss:0.007363149630942103\n",
+      "train loss:0.0010485294999615432\n",
+      "train loss:0.0029913040891653584\n",
+      "train loss:0.007992596818559088\n",
+      "train loss:0.0500020894450527\n",
+      "train loss:0.001599259438154814\n",
+      "train loss:0.005117667300297052\n",
+      "train loss:0.01083012504162666\n",
+      "train loss:0.008089864502459207\n",
+      "train loss:0.007998571184647302\n",
+      "train loss:0.00926079531610022\n",
+      "train loss:0.05785912557353757\n",
+      "train loss:0.007664318886145549\n",
+      "train loss:0.0027520271022457915\n",
+      "train loss:0.022451484217707485\n",
+      "train loss:0.008128244190857397\n",
+      "train loss:0.002890399913314712\n",
+      "train loss:0.003991359591986023\n",
+      "train loss:0.009656237726383497\n",
+      "train loss:0.010307953918459439\n",
+      "train loss:0.02640888065144468\n",
+      "train loss:0.028516761313693934\n",
+      "train loss:0.004952183359174758\n",
+      "train loss:0.025759838876403034\n",
+      "train loss:0.0679095614813479\n",
+      "train loss:0.010597303796939248\n",
+      "train loss:0.006543098027010562\n",
+      "train loss:0.00798790317061746\n",
+      "train loss:0.06016511009613695\n",
+      "train loss:0.0050921484501256185\n",
+      "train loss:0.005435889882715286\n",
+      "train loss:0.008130017096084652\n",
+      "train loss:0.007417673764682549\n",
+      "train loss:0.017344277153819948\n",
+      "train loss:0.14564857219407398\n",
+      "train loss:0.03642807194950174\n",
+      "train loss:0.013068973171863972\n",
+      "train loss:0.003650276265667539\n",
+      "train loss:0.006082913371010344\n",
+      "train loss:0.011434319259457341\n",
+      "train loss:0.009100604681033736\n",
+      "train loss:0.004624425393029769\n",
+      "train loss:0.003560646082188661\n",
+      "train loss:0.022680458133122926\n",
+      "train loss:0.006578504507110082\n",
+      "train loss:0.025732308551664244\n",
+      "train loss:0.030437403576003684\n",
+      "=== epoch:8, train acc:0.99, test acc:0.983 ===\n",
+      "train loss:0.01188645972180352\n",
+      "train loss:0.01347452655006645\n",
+      "train loss:0.04248575282812874\n",
+      "train loss:0.0016060374469501716\n",
+      "train loss:0.008901727739654546\n",
+      "train loss:0.01558199782543342\n",
+      "train loss:0.0026786057300957405\n",
+      "train loss:0.01959615468349815\n",
+      "train loss:0.0066834576136000955\n",
+      "train loss:0.013394337111380179\n",
+      "train loss:0.006940790996846247\n",
+      "train loss:0.028308553286179493\n",
+      "train loss:0.00782048045228374\n",
+      "train loss:0.003144961272156263\n",
+      "train loss:0.007212374756157522\n",
+      "train loss:0.003051797980478538\n",
+      "train loss:0.028086431402743015\n",
+      "train loss:0.030612327464535256\n",
+      "train loss:0.010879725384382501\n",
+      "train loss:0.00980748787471033\n",
+      "train loss:0.003094421507982909\n",
+      "train loss:0.004623118688442995\n",
+      "train loss:0.022969949514269717\n",
+      "train loss:0.0040186501308792835\n",
+      "train loss:0.003347007862327877\n",
+      "train loss:0.018660531191297687\n",
+      "train loss:0.0078511268522323\n",
+      "train loss:0.0457431464300496\n",
+      "train loss:0.02005238563790976\n",
+      "train loss:0.0229714806344032\n",
+      "train loss:0.0115041100809642\n",
+      "train loss:0.0030581941122588513\n",
+      "train loss:0.008999781071457984\n",
+      "train loss:0.009921572653356326\n",
+      "train loss:0.004297463510619493\n",
+      "train loss:0.0027573025710736377\n",
+      "train loss:0.0651170337101901\n",
+      "train loss:0.09634101393682122\n",
+      "train loss:0.02928831765559476\n",
+      "train loss:0.02294224371777588\n",
+      "train loss:0.0381387568845075\n",
+      "train loss:0.059744409677777346\n",
+      "train loss:0.004982958466312586\n",
+      "train loss:0.023711837034025292\n",
+      "train loss:0.015304400169897902\n",
+      "train loss:0.0032437397121691503\n",
+      "train loss:0.006180525038749473\n",
+      "train loss:0.015724135494813333\n",
+      "train loss:0.03168723323105611\n",
+      "train loss:0.022278150069065923\n",
+      "train loss:0.014076616813144068\n",
+      "train loss:0.02415908139419603\n",
+      "train loss:0.0023866147690719606\n",
+      "train loss:0.009688774355915036\n",
+      "train loss:0.0060855057202895135\n",
+      "train loss:0.014487607841959127\n",
+      "train loss:0.08320000571566279\n",
+      "train loss:0.09468966023391431\n",
+      "train loss:0.032045791104950566\n",
+      "train loss:0.004100706471474388\n",
+      "train loss:0.042596535454179095\n",
+      "train loss:0.005046187880616705\n",
+      "train loss:0.010846818084326798\n",
+      "train loss:0.009806363640480936\n",
+      "train loss:0.0182952215097739\n",
+      "train loss:0.007655526798355072\n",
+      "train loss:0.015892004765347603\n",
+      "train loss:0.008527672489079857\n",
+      "train loss:0.003508106916934715\n",
+      "train loss:0.03301369264401838\n",
+      "train loss:0.005961774306705517\n",
+      "train loss:0.03201686208932682\n",
+      "train loss:0.02669082814445824\n",
+      "train loss:0.004465180571294049\n",
+      "train loss:0.003274701557115177\n",
+      "train loss:0.007614774602060963\n",
+      "train loss:0.06735202810005798\n",
+      "train loss:0.016341250667475046\n",
+      "train loss:0.007178265524246971\n",
+      "train loss:0.03130090869288684\n",
+      "train loss:0.005757632541884597\n",
+      "train loss:0.06234479511693348\n",
+      "train loss:0.01906863558821351\n",
+      "train loss:0.03267457048771052\n",
+      "train loss:0.04401692589385079\n",
+      "train loss:0.02074431834502839\n",
+      "train loss:0.030232786389654813\n",
+      "train loss:0.019249105655263773\n",
+      "train loss:0.007115803642129995\n",
+      "train loss:0.03333698072196401\n",
+      "train loss:0.026985046573926075\n",
+      "train loss:0.005362241573491471\n",
+      "train loss:0.03427487011035835\n",
+      "train loss:0.01038508992046969\n",
+      "train loss:0.006650949007899766\n",
+      "train loss:0.01179728182679768\n",
+      "train loss:0.0025626058358159697\n",
+      "train loss:0.002836329072612403\n",
+      "train loss:0.009093057282619098\n",
+      "train loss:0.011608286858477819\n",
+      "train loss:0.001156295708690354\n",
+      "train loss:0.013463646634477754\n",
+      "train loss:0.0021416828867336147\n",
+      "train loss:0.021943989256392377\n",
+      "train loss:0.011693177933653551\n",
+      "train loss:0.009314554413611729\n",
+      "train loss:0.009791738610366555\n",
+      "train loss:0.0038679503244219288\n",
+      "train loss:0.05185267024423574\n",
+      "train loss:0.004002609592291216\n",
+      "train loss:0.005092421114471343\n",
+      "train loss:0.0314028820490648\n",
+      "train loss:0.009651407567091291\n",
+      "train loss:0.008085005999132854\n",
+      "train loss:0.013907328717070986\n",
+      "train loss:0.009554201381000334\n",
+      "train loss:0.0032730726988273906\n",
+      "train loss:0.012186709789584131\n",
+      "train loss:0.005252531734610512\n",
+      "train loss:0.06050996052190524\n",
+      "train loss:0.008745444420834409\n",
+      "train loss:0.010249977145256247\n",
+      "train loss:0.010340635377310533\n",
+      "train loss:0.004624946563267051\n",
+      "train loss:0.023341781026796724\n",
+      "train loss:0.005820757227417732\n",
+      "train loss:0.008211604919197665\n",
+      "train loss:0.004368557223014141\n",
+      "train loss:0.0349620548867639\n",
+      "train loss:0.004029069162513296\n",
+      "train loss:0.01888133010269523\n",
+      "train loss:0.0015506828479391862\n",
+      "train loss:0.014213104358257178\n",
+      "train loss:0.05182241064459344\n",
+      "train loss:0.04904021428021897\n",
+      "train loss:0.01085532383034541\n",
+      "train loss:0.006513851547900541\n",
+      "train loss:0.00522801408736298\n",
+      "train loss:0.005717037784148692\n",
+      "train loss:0.008688246915917998\n",
+      "train loss:0.008318538150177156\n",
+      "train loss:0.013307259075996813\n",
+      "train loss:0.006644319767108346\n",
+      "train loss:0.0035852246376515485\n",
+      "train loss:0.03795446205619293\n",
+      "train loss:0.003899023303637252\n",
+      "train loss:0.07398356599127526\n",
+      "train loss:0.008001462451980327\n",
+      "train loss:0.01237272938331252\n",
+      "train loss:0.01118696141354973\n",
+      "train loss:0.006381371738044368\n",
+      "train loss:0.03038287869691704\n",
+      "train loss:0.004521972031883374\n",
+      "train loss:0.021666239758140086\n",
+      "train loss:0.0033995613459186485\n",
+      "train loss:0.02122836207117763\n",
+      "train loss:0.005634974328044437\n",
+      "train loss:0.01273859654836532\n",
+      "train loss:0.0026634764993632883\n",
+      "train loss:0.008326238908588269\n",
+      "train loss:0.013565781000296291\n",
+      "train loss:0.009485571219994092\n",
+      "train loss:0.008901877049471105\n",
+      "train loss:0.02082067169744113\n",
+      "train loss:0.0056404072260545675\n",
+      "train loss:0.03800192253880512\n",
+      "train loss:0.01173786541472641\n",
+      "train loss:0.018669483273384847\n",
+      "train loss:0.0027246933919989735\n",
+      "train loss:0.016348839598239955\n",
+      "train loss:0.005259467484018818\n",
+      "train loss:0.006087969909208065\n",
+      "train loss:0.014316954586354227\n",
+      "train loss:0.006217565190233156\n",
+      "train loss:0.003557860105312824\n",
+      "train loss:0.00775873121356181\n",
+      "train loss:0.006213479980274709\n",
+      "train loss:0.011553464421270906\n",
+      "train loss:0.023965999700791697\n",
+      "train loss:0.0057602633232144125\n",
+      "train loss:0.01811609929845818\n",
+      "train loss:0.010247867107438961\n",
+      "train loss:0.002104561878353091\n",
+      "train loss:0.01770289983602788\n",
+      "train loss:0.009749841378088682\n",
+      "train loss:0.004221404152212377\n",
+      "train loss:0.006120982724085282\n",
+      "train loss:0.013350807137679707\n",
+      "train loss:0.001974336290107773\n",
+      "train loss:0.006786367862466073\n",
+      "train loss:0.015616532234007434\n",
+      "train loss:0.014557975514230498\n",
+      "train loss:0.0022092976961144433\n",
+      "train loss:0.004154426073973889\n",
+      "train loss:0.004490254250695261\n",
+      "train loss:0.017499003913252664\n",
+      "train loss:0.006812229171780541\n",
+      "train loss:0.021311288876107736\n",
+      "train loss:0.0051867776818280585\n",
+      "train loss:0.00327201485994957\n",
+      "train loss:0.007932621290456719\n",
+      "train loss:0.009819530011068082\n",
+      "train loss:0.03924410162816367\n",
+      "train loss:0.01687638412883939\n",
+      "train loss:0.011895123081463588\n",
+      "train loss:0.006950786534614403\n",
+      "train loss:0.011848594404016264\n",
+      "train loss:0.005198428906763736\n",
+      "train loss:0.002936860399956231\n",
+      "train loss:0.012642333221095332\n",
+      "train loss:0.06527898683040942\n",
+      "train loss:0.0024368203396035615\n",
+      "train loss:0.015209366402564895\n",
+      "train loss:0.010011344406757107\n",
+      "train loss:0.0037058636069679613\n",
+      "train loss:0.015375448254040308\n",
+      "train loss:0.011609840637147104\n",
+      "train loss:0.0039744399511777405\n",
+      "train loss:0.010012095880963934\n",
+      "train loss:0.01148947569933388\n",
+      "train loss:0.006691104226345422\n",
+      "train loss:0.04558723280851702\n",
+      "train loss:0.00183566466230106\n",
+      "train loss:0.0028494885633654327\n",
+      "train loss:0.007180455136601741\n",
+      "train loss:0.003477636227681663\n",
+      "train loss:0.013449647474403288\n",
+      "train loss:0.008954759670105599\n",
+      "train loss:0.011360437881260304\n",
+      "train loss:0.01059332525773159\n",
+      "train loss:0.009288543741334199\n",
+      "train loss:0.007046679525691749\n",
+      "train loss:0.07378733107716776\n",
+      "train loss:0.01638978707330192\n",
+      "train loss:0.0205847203253252\n",
+      "train loss:0.015400770953401095\n",
+      "train loss:0.010379555912196197\n",
+      "train loss:0.010025548400772508\n",
+      "train loss:0.0023581496184625623\n",
+      "train loss:0.004782399662199058\n",
+      "train loss:0.008954775839072003\n",
+      "train loss:0.001439813142399382\n",
+      "train loss:0.01657285669595656\n",
+      "train loss:0.0066709699820333805\n",
+      "train loss:0.00608889395393872\n",
+      "train loss:0.002714983539455479\n",
+      "train loss:0.01951979308843669\n",
+      "train loss:0.009426425232287032\n",
+      "train loss:0.009053505976198382\n",
+      "train loss:0.0007713661651063201\n",
+      "train loss:0.005626061562231111\n",
+      "train loss:0.002592750406620087\n",
+      "train loss:0.010460812739454555\n",
+      "train loss:0.001828909722123872\n",
+      "train loss:0.04866588891922677\n",
+      "train loss:0.013429057538032328\n",
+      "train loss:0.0458994616014322\n",
+      "train loss:0.002132899094760416\n",
+      "train loss:0.010319817280682764\n",
+      "train loss:0.006863351801558361\n",
+      "train loss:0.06814596820646708\n",
+      "train loss:0.007881954994295705\n",
+      "train loss:0.0022531051531756814\n",
+      "train loss:0.018085994433377616\n",
+      "train loss:0.0057794897222867465\n",
+      "train loss:0.0012038426232806087\n",
+      "train loss:0.03679092838901802\n",
+      "train loss:0.003232372412415805\n",
+      "train loss:0.011677826702300971\n",
+      "train loss:0.007002816852535582\n",
+      "train loss:0.0038701478802197105\n",
+      "train loss:0.006745331642789882\n",
+      "train loss:0.013526590900039217\n",
+      "train loss:0.025324002706376265\n",
+      "train loss:0.0038719432116892283\n",
+      "train loss:0.0018776780822869702\n",
+      "train loss:0.006076485922501228\n",
+      "train loss:0.029167190483985053\n",
+      "train loss:0.007926853498602776\n",
+      "train loss:0.010803514379302407\n",
+      "train loss:0.02985995587697032\n",
+      "train loss:0.011906145421854794\n",
+      "train loss:0.004933145505537178\n",
+      "train loss:0.01874031779767815\n",
+      "train loss:0.0028926092867523935\n",
+      "train loss:0.005170134504092804\n",
+      "train loss:0.013803421058106257\n",
+      "train loss:0.016463198382924295\n",
+      "train loss:0.012132845831679395\n",
+      "train loss:0.010441443853142138\n",
+      "train loss:0.033423322628579395\n",
+      "train loss:0.007597781587609028\n",
+      "train loss:0.003636224716419623\n",
+      "train loss:0.019015025064262826\n",
+      "train loss:0.0007843113418255831\n",
+      "train loss:0.010689116587784575\n",
+      "train loss:0.0015281571975359306\n",
+      "train loss:0.050388105014372755\n",
+      "train loss:0.0025539120213839113\n",
+      "train loss:0.019985284658253583\n",
+      "train loss:0.006881020073896864\n",
+      "train loss:0.004856136136719777\n",
+      "train loss:0.003929692442508258\n",
+      "train loss:0.004488226691264045\n",
+      "train loss:0.0038723060090335134\n",
+      "train loss:0.02402053859657151\n",
+      "train loss:0.004066800943585667\n",
+      "train loss:0.008504960709299847\n",
+      "train loss:0.018377165662271794\n",
+      "train loss:0.005697933190767811\n",
+      "train loss:0.004450271954907922\n",
+      "train loss:0.016051197998615457\n",
+      "train loss:0.004441264267739158\n",
+      "train loss:0.01108946980555657\n",
+      "train loss:0.021534120310245822\n",
+      "train loss:0.011663112369223247\n",
+      "train loss:0.0022808938299453585\n",
+      "train loss:0.0015033016740917105\n",
+      "train loss:0.005792145499013739\n",
+      "train loss:0.0021963123295298407\n",
+      "train loss:0.04081845647789455\n",
+      "train loss:0.006908755683803727\n",
+      "train loss:0.001866298128586715\n",
+      "train loss:0.03203688708062852\n",
+      "train loss:0.0034513305051780446\n",
+      "train loss:0.011540734572686718\n",
+      "train loss:0.0020069647639058465\n",
+      "train loss:0.012071211009226835\n",
+      "train loss:0.031142150437501015\n",
+      "train loss:0.005277467395256327\n",
+      "train loss:0.015267199809204282\n",
+      "train loss:0.005694012456930386\n",
+      "train loss:0.0046893816609647684\n",
+      "train loss:0.12421859096240874\n",
+      "train loss:0.002350590106039283\n",
+      "train loss:0.02833357990756934\n",
+      "train loss:0.003348953270960347\n",
+      "train loss:0.007066944547441587\n",
+      "train loss:0.010522158860359778\n",
+      "train loss:0.0027411606014723262\n",
+      "train loss:0.007618434197540122\n",
+      "train loss:0.010281987015286098\n",
+      "train loss:0.009539989806755283\n",
+      "train loss:0.013859851036182867\n",
+      "train loss:0.0175474774720536\n",
+      "train loss:0.01228800492032329\n",
+      "train loss:0.024243051080895604\n",
+      "train loss:0.007096427845363864\n",
+      "train loss:0.009961518109129685\n",
+      "train loss:0.006607007582799302\n",
+      "train loss:0.002849801378085461\n",
+      "train loss:0.011463910074502424\n",
+      "train loss:0.0021198514591075504\n",
+      "train loss:0.032438809110148\n",
+      "train loss:0.005961853513155234\n",
+      "train loss:0.0005086461757048791\n",
+      "train loss:0.012800347324339512\n",
+      "train loss:0.028360133840527508\n",
+      "train loss:0.014017640784938452\n",
+      "train loss:0.002691719869111752\n",
+      "train loss:0.0007117672821970615\n",
+      "train loss:0.015213493064569887\n",
+      "train loss:0.011673113507523264\n",
+      "train loss:0.009403308001102754\n",
+      "train loss:0.00588947230818109\n",
+      "train loss:0.012114613812044863\n",
+      "train loss:0.0022372727035973134\n",
+      "train loss:0.014539961535217355\n",
+      "train loss:0.015880641068999727\n",
+      "train loss:0.0022311402095992815\n",
+      "train loss:0.009029941651493482\n",
+      "train loss:0.026350675610771757\n",
+      "train loss:0.03142956109354359\n",
+      "train loss:0.0053142995778498615\n",
+      "train loss:0.01676872485127268\n",
+      "train loss:0.005383326533427571\n",
+      "train loss:0.03260990162172762\n",
+      "train loss:0.004516186691577224\n",
+      "train loss:0.005375536287686175\n",
+      "train loss:0.003583352900342869\n",
+      "train loss:0.0054276941650441175\n",
+      "train loss:0.002166826016784943\n",
+      "train loss:0.01230312049928878\n",
+      "train loss:0.007067168076430231\n",
+      "train loss:0.010004792868103029\n",
+      "train loss:0.006749522290049951\n",
+      "train loss:0.012738689939791246\n",
+      "train loss:0.05166471800342041\n",
+      "train loss:0.019350911913702962\n",
+      "train loss:0.013651231106996509\n",
+      "train loss:0.02030375065296912\n",
+      "train loss:0.008098429363953863\n",
+      "train loss:0.0012600911566379958\n",
+      "train loss:0.0022841212758855016\n",
+      "train loss:0.01619609254795412\n",
+      "train loss:0.002543281862410633\n",
+      "train loss:0.010112693048818823\n",
+      "train loss:0.008369010168843276\n",
+      "train loss:0.026364970524326778\n",
+      "train loss:0.0024178883505823995\n",
+      "train loss:0.021220467508401217\n",
+      "train loss:0.00937849012688132\n",
+      "train loss:0.0017022346044754655\n",
+      "train loss:0.01998655866096037\n",
+      "train loss:0.0034198050430395764\n",
+      "train loss:0.009966629651654846\n",
+      "train loss:0.006278492963823856\n",
+      "train loss:0.07888137861878994\n",
+      "train loss:0.010180482274050854\n",
+      "train loss:0.01880590381415822\n",
+      "train loss:0.027178950212849794\n",
+      "train loss:0.01440825485236078\n",
+      "train loss:0.006022731706388761\n",
+      "train loss:0.08369338840758647\n",
+      "train loss:0.008893505206268484\n",
+      "train loss:0.0019598096655467803\n",
+      "train loss:0.05156270037505229\n",
+      "train loss:0.009403007992477286\n",
+      "train loss:0.0032978537394533596\n",
+      "train loss:0.0015194731699299802\n",
+      "train loss:0.0057140560653281035\n",
+      "train loss:0.012949902238713111\n",
+      "train loss:0.008817011090677217\n",
+      "train loss:0.011421102634194026\n",
+      "train loss:0.011975489533202996\n",
+      "train loss:0.00396768245107089\n",
+      "train loss:0.007626971203996864\n",
+      "train loss:0.003502322187319944\n",
+      "train loss:0.014567862108018853\n",
+      "train loss:0.025331039087304753\n",
+      "train loss:0.003897423417427569\n",
+      "train loss:0.004750876813728569\n",
+      "train loss:0.0022565907372734846\n",
+      "train loss:0.0017966380945177705\n",
+      "train loss:0.004661101189721768\n",
+      "train loss:0.02101477437435566\n",
+      "train loss:0.013139424122578115\n",
+      "train loss:0.00188843568259534\n",
+      "train loss:0.002871361527971193\n",
+      "train loss:0.011492076814370127\n",
+      "train loss:0.006359129004417513\n",
+      "train loss:0.0007241961424577985\n",
+      "train loss:0.017554096040105934\n",
+      "train loss:0.011198416581458584\n",
+      "train loss:0.044779334391667876\n",
+      "train loss:0.017364631312232995\n",
+      "train loss:0.03482991403726323\n",
+      "train loss:0.022780185086519977\n",
+      "train loss:0.0023671079016541873\n",
+      "train loss:0.011677848489784965\n",
+      "train loss:0.010528306864332294\n",
+      "train loss:0.018244991788817083\n",
+      "train loss:0.03429103921217494\n",
+      "train loss:0.011292813997836451\n",
+      "train loss:0.009788714181120428\n",
+      "train loss:0.010330580377614902\n",
+      "train loss:0.0020868095741626213\n",
+      "train loss:0.0038374218856361705\n",
+      "train loss:0.0032343348434089414\n",
+      "train loss:0.004512741748892224\n",
+      "train loss:0.002091701980209125\n",
+      "train loss:0.014871035882130727\n",
+      "train loss:0.003607681800888429\n",
+      "train loss:0.03633192545203326\n",
+      "train loss:0.0021050121400772253\n",
+      "train loss:0.004100603557297052\n",
+      "train loss:0.0014892363272516515\n",
+      "train loss:0.06306085610909577\n",
+      "train loss:0.01126795096329818\n",
+      "train loss:0.014647113541695738\n",
+      "train loss:0.00601882009498558\n",
+      "train loss:0.00329618865399214\n",
+      "train loss:0.09199856752609953\n",
+      "train loss:0.026867628396684787\n",
+      "train loss:0.0036786377025077323\n",
+      "train loss:0.0008435906333870923\n",
+      "train loss:0.006507349071414156\n",
+      "train loss:0.006925537806601493\n",
+      "train loss:0.0021492853140364487\n",
+      "train loss:0.003752799143263048\n",
+      "train loss:0.00119171175586391\n",
+      "train loss:0.0037323416363914037\n",
+      "train loss:0.01086048719903042\n",
+      "train loss:0.00741668294588685\n",
+      "train loss:0.004367808808175767\n",
+      "train loss:0.006646783650608956\n",
+      "train loss:0.018726677690648195\n",
+      "train loss:0.003801594810469404\n",
+      "train loss:0.006467321455152267\n",
+      "train loss:0.007644928199788499\n",
+      "train loss:0.004192429016363005\n",
+      "train loss:0.0012488100741586473\n",
+      "train loss:0.002247712873984968\n",
+      "train loss:0.0035132123016988386\n",
+      "train loss:0.0017231032309434108\n",
+      "train loss:0.014224016562168429\n",
+      "train loss:0.006271859300590658\n",
+      "train loss:0.00584487954944428\n",
+      "train loss:0.01484803979919639\n",
+      "train loss:0.002452257390939185\n",
+      "train loss:0.020092875887085\n",
+      "train loss:0.009096854837366743\n",
+      "train loss:0.0197180156357149\n",
+      "train loss:0.007645163677371756\n",
+      "train loss:0.0057338506334987695\n",
+      "train loss:0.004754201347148742\n",
+      "train loss:0.009979122871644755\n",
+      "train loss:0.011361555731771214\n",
+      "train loss:0.009658025629743813\n",
+      "train loss:0.0013948185668876372\n",
+      "train loss:0.008538098543480019\n",
+      "train loss:0.0033131487971165426\n",
+      "train loss:0.011272486389170465\n",
+      "train loss:0.0010743213851221874\n",
+      "train loss:0.02062673089709153\n",
+      "train loss:0.0064142104728211625\n",
+      "train loss:0.016788539525288568\n",
+      "train loss:0.005382124492993978\n",
+      "train loss:0.009329927060384739\n",
+      "train loss:0.011915638663185058\n",
+      "train loss:0.006946171317188282\n",
+      "train loss:0.005827442284992733\n",
+      "train loss:0.0009541310391392595\n",
+      "train loss:0.04332754573919809\n",
+      "train loss:0.005588052573073536\n",
+      "train loss:0.007215800529639732\n",
+      "train loss:0.006091229277171631\n",
+      "train loss:0.05544698397645512\n",
+      "train loss:0.0025434118856303718\n",
+      "train loss:0.0035134930310313796\n",
+      "train loss:0.0030248676198584195\n",
+      "train loss:0.0030268196524509228\n",
+      "train loss:0.009455195522847303\n",
+      "train loss:0.005855313950076432\n",
+      "train loss:0.0062942158312882645\n",
+      "train loss:0.0011076400956146102\n",
+      "train loss:0.002695153719464172\n",
+      "train loss:0.008651419572438762\n",
+      "train loss:0.008826617143081172\n",
+      "train loss:0.009478315585886028\n",
+      "train loss:0.0020952947460629954\n",
+      "train loss:0.0165635546996748\n",
+      "train loss:0.01032422501024336\n",
+      "train loss:0.0023519573953229935\n",
+      "train loss:0.019195453556237866\n",
+      "train loss:0.018441646566587058\n",
+      "train loss:0.0076321634240315815\n",
+      "train loss:0.0035703160040015712\n",
+      "train loss:0.005889974921781707\n",
+      "train loss:0.004635668652526133\n",
+      "train loss:0.031817142288231\n",
+      "train loss:0.011353095389948291\n",
+      "train loss:0.004930120949502645\n",
+      "train loss:0.013450070304227364\n",
+      "train loss:0.011930473958122969\n",
+      "train loss:0.018119833495224398\n",
+      "train loss:0.009673446252959284\n",
+      "train loss:0.0014855569817751975\n",
+      "train loss:0.04991440024872602\n",
+      "train loss:0.008743817126093125\n",
+      "train loss:0.0022478672736883777\n",
+      "train loss:0.0042389771831586286\n",
+      "train loss:0.021100406957308557\n",
+      "train loss:0.011453258654759239\n",
+      "train loss:0.010886956090358627\n",
+      "train loss:0.008722723720850851\n",
+      "train loss:0.013471399350401807\n",
+      "train loss:0.005470577167896272\n",
+      "train loss:0.003721672645612328\n",
+      "train loss:0.0020798586375573914\n",
+      "train loss:0.001455946289021374\n",
+      "train loss:0.004316455203439658\n",
+      "train loss:0.006958267695104405\n",
+      "train loss:0.008525661586388426\n",
+      "train loss:0.01454890161968215\n",
+      "train loss:0.006412060527552165\n",
+      "train loss:0.007230027784604293\n",
+      "train loss:0.01595394939952177\n",
+      "train loss:0.008256646768880108\n",
+      "train loss:0.013710615165162929\n",
+      "train loss:0.020824835934689166\n",
+      "train loss:0.010007718242732338\n",
+      "train loss:0.0032964682238191645\n",
+      "train loss:0.004177912191359412\n",
+      "train loss:0.001461890734950237\n",
+      "train loss:0.005716855379229319\n",
+      "train loss:0.011822617936903916\n",
+      "train loss:0.014739570738252515\n",
+      "train loss:0.01233921445464905\n",
+      "train loss:0.0027323447381901677\n",
+      "train loss:0.0013244960814736897\n",
+      "train loss:0.024782247043748196\n",
+      "train loss:0.01830563181393928\n",
+      "train loss:0.009862272444581039\n",
+      "train loss:0.011683569899145456\n",
+      "train loss:0.002024165191626997\n",
+      "train loss:0.006872892407130789\n",
+      "train loss:0.05155107831891297\n",
+      "train loss:0.00663013029350974\n",
+      "train loss:0.011374784649598674\n",
+      "=== epoch:9, train acc:0.991, test acc:0.989 ===\n",
+      "train loss:0.0005902993607911365\n",
+      "train loss:0.03403512966181197\n",
+      "train loss:0.026544332774083704\n",
+      "train loss:0.013036511509574598\n",
+      "train loss:0.0019217732034296261\n",
+      "train loss:0.0012686060037369416\n",
+      "train loss:0.0009057361853687425\n",
+      "train loss:0.020432517644094584\n",
+      "train loss:0.01564549525313301\n",
+      "train loss:0.005623166718221372\n",
+      "train loss:0.005712436507029545\n",
+      "train loss:0.011163753007628568\n",
+      "train loss:0.013286080430677513\n",
+      "train loss:0.009839381198889816\n",
+      "train loss:0.10152736219572958\n",
+      "train loss:0.010209279985344592\n",
+      "train loss:0.0005326091710334813\n",
+      "train loss:0.010781252150029254\n",
+      "train loss:0.036420721032177965\n",
+      "train loss:0.017177161195032327\n",
+      "train loss:0.011556795459317277\n",
+      "train loss:0.0023433587057652967\n",
+      "train loss:0.026238138055669255\n",
+      "train loss:0.005701832185415089\n",
+      "train loss:0.05104000424792982\n",
+      "train loss:0.003066027106009272\n",
+      "train loss:0.013841499513069136\n",
+      "train loss:0.026470773856233837\n",
+      "train loss:0.004222574889305926\n",
+      "train loss:0.012571567434697973\n",
+      "train loss:0.014947943378221515\n",
+      "train loss:0.007675010322966721\n",
+      "train loss:0.011959749630695644\n",
+      "train loss:0.003996396750123907\n",
+      "train loss:0.023058509817378513\n",
+      "train loss:0.01350536526138589\n",
+      "train loss:0.026480338680420018\n",
+      "train loss:0.001146880469402616\n",
+      "train loss:0.04132340175886836\n",
+      "train loss:0.007259465973540716\n",
+      "train loss:0.04353523198371065\n",
+      "train loss:0.014010720980208408\n",
+      "train loss:0.0019426495113676952\n",
+      "train loss:0.007203932723874186\n",
+      "train loss:0.010750840169231135\n",
+      "train loss:0.005489021171356081\n",
+      "train loss:0.022535534316454074\n",
+      "train loss:0.04457916399463162\n",
+      "train loss:0.004095385251084725\n",
+      "train loss:0.011119962512038388\n",
+      "train loss:0.011794289008149566\n",
+      "train loss:0.008351451731878567\n",
+      "train loss:0.043627661230301575\n",
+      "train loss:0.004162489317232776\n",
+      "train loss:0.019559922957494487\n",
+      "train loss:0.006067225736598134\n",
+      "train loss:0.008570523369869276\n",
+      "train loss:0.002344016437972436\n",
+      "train loss:0.0622038066140401\n",
+      "train loss:0.0011615705821892855\n",
+      "train loss:0.015050606796601237\n",
+      "train loss:0.00556392491106866\n",
+      "train loss:0.025884614692118686\n",
+      "train loss:0.006856439951766701\n",
+      "train loss:0.0030213227108914\n",
+      "train loss:0.028317904019841175\n",
+      "train loss:0.07112627600425583\n",
+      "train loss:0.004212936770942807\n",
+      "train loss:0.006808805468519172\n",
+      "train loss:0.05196891205457971\n",
+      "train loss:0.003992050846006745\n",
+      "train loss:0.004384894332145231\n",
+      "train loss:0.017962582458863553\n",
+      "train loss:0.010672062450666032\n",
+      "train loss:0.009673488624559304\n",
+      "train loss:0.009485600088167286\n",
+      "train loss:0.00514241962089908\n",
+      "train loss:0.004066337133382261\n",
+      "train loss:0.00510896314111249\n",
+      "train loss:0.003281497108463901\n",
+      "train loss:0.010011725499467996\n",
+      "train loss:0.01452088148443098\n",
+      "train loss:0.0038043006444267507\n",
+      "train loss:0.00840060861174079\n",
+      "train loss:0.00975696045627341\n",
+      "train loss:0.005597039425722226\n",
+      "train loss:0.014733008867345583\n",
+      "train loss:0.028846477754435176\n",
+      "train loss:0.007546125543991176\n",
+      "train loss:0.01743399356045389\n",
+      "train loss:0.01596444087647493\n",
+      "train loss:0.0027075250429061675\n",
+      "train loss:0.004624512234316232\n",
+      "train loss:0.007085771039021298\n",
+      "train loss:0.015760895063063746\n",
+      "train loss:0.002025988098439069\n",
+      "train loss:0.0047175524449249925\n",
+      "train loss:0.02231818470952067\n",
+      "train loss:0.0028048797344295585\n",
+      "train loss:0.008283411792129528\n",
+      "train loss:0.004157387756368873\n",
+      "train loss:0.01168123950722627\n",
+      "train loss:0.016160722398123523\n",
+      "train loss:0.037838606679977085\n",
+      "train loss:0.0026874763023748615\n",
+      "train loss:0.0032704394955008864\n",
+      "train loss:0.004005635232598578\n",
+      "train loss:0.02191137042705782\n",
+      "train loss:0.027966635258106082\n",
+      "train loss:0.0036981055831814373\n",
+      "train loss:0.0011086652414176955\n",
+      "train loss:0.015602502044201747\n",
+      "train loss:0.04571940682320122\n",
+      "train loss:0.011712758320676088\n",
+      "train loss:0.007862615088358869\n",
+      "train loss:0.005102260996691183\n",
+      "train loss:0.00659240568513351\n",
+      "train loss:0.004299533285240232\n",
+      "train loss:0.004175933497511417\n",
+      "train loss:0.01749881233389215\n",
+      "train loss:0.019398532146184623\n",
+      "train loss:0.005812620994765082\n",
+      "train loss:0.007053217112005361\n",
+      "train loss:0.02407215862676717\n",
+      "train loss:0.001887483002108158\n",
+      "train loss:0.0019168609862626495\n",
+      "train loss:0.004050068048868868\n",
+      "train loss:0.013772585422361231\n",
+      "train loss:0.01317388419177774\n",
+      "train loss:0.0043089992301008\n",
+      "train loss:0.0025518215472686455\n",
+      "train loss:0.013014883024879234\n",
+      "train loss:0.052143341765724605\n",
+      "train loss:0.002764742604672677\n",
+      "train loss:0.004110525191834567\n",
+      "train loss:0.025794103930550864\n",
+      "train loss:0.0066822288574269785\n",
+      "train loss:0.0005976650044655312\n",
+      "train loss:0.005074258005027747\n",
+      "train loss:0.0016998144626370155\n",
+      "train loss:0.01060516596531905\n",
+      "train loss:0.002443142312782575\n",
+      "train loss:0.0019788972913053608\n",
+      "train loss:0.014288045777104939\n",
+      "train loss:0.01611790996461081\n",
+      "train loss:0.008351811508369687\n",
+      "train loss:0.003462855374829302\n",
+      "train loss:0.003557971528503705\n",
+      "train loss:0.008945047823296056\n",
+      "train loss:0.0014259963004796658\n",
+      "train loss:0.003798413995036376\n",
+      "train loss:0.001740792031896421\n",
+      "train loss:0.0019318304347466996\n",
+      "train loss:0.026231299214744076\n",
+      "train loss:0.010282454472483611\n",
+      "train loss:0.010187319748385035\n",
+      "train loss:0.008212474435990111\n",
+      "train loss:0.0037317884540967222\n",
+      "train loss:0.03724534146870414\n",
+      "train loss:0.009203226737962861\n",
+      "train loss:0.005873172619022399\n",
+      "train loss:0.000673737165640369\n",
+      "train loss:0.006394030391033884\n",
+      "train loss:0.0008633193097103726\n",
+      "train loss:0.004521438114789038\n",
+      "train loss:0.002919658409529095\n",
+      "train loss:0.007584144870345617\n",
+      "train loss:0.011095217400342085\n",
+      "train loss:0.0009658971397914397\n",
+      "train loss:0.018071846747045095\n",
+      "train loss:0.002426360204260963\n",
+      "train loss:0.014747498607059983\n",
+      "train loss:0.003522504445179024\n",
+      "train loss:0.005755402791431958\n",
+      "train loss:0.00971618640429724\n",
+      "train loss:0.012479665812840912\n",
+      "train loss:0.02936436446667801\n",
+      "train loss:0.0012585531312225382\n",
+      "train loss:0.0028463698494033845\n",
+      "train loss:0.0038931113447005657\n",
+      "train loss:0.0027786675016566063\n",
+      "train loss:0.001984385996626566\n",
+      "train loss:0.008482555883225985\n",
+      "train loss:0.005076005097460446\n",
+      "train loss:0.007972620072669474\n",
+      "train loss:0.011559662569838728\n",
+      "train loss:0.01289263234315735\n",
+      "train loss:0.008263111405322134\n",
+      "train loss:0.04922014486518313\n",
+      "train loss:0.0024810554332523977\n",
+      "train loss:0.002840736558369323\n",
+      "train loss:0.013258698867391991\n",
+      "train loss:0.015159907074835331\n",
+      "train loss:0.02857862433808464\n",
+      "train loss:0.0032133205246973077\n",
+      "train loss:0.013609574669890823\n",
+      "train loss:0.0007954631521968809\n",
+      "train loss:0.0067635363980689325\n",
+      "train loss:0.005804694000757257\n",
+      "train loss:0.011070614348253122\n",
+      "train loss:0.03467582852156345\n",
+      "train loss:0.002631740563207594\n",
+      "train loss:0.026228667770651613\n",
+      "train loss:0.004499764566064619\n",
+      "train loss:0.0024957076374909214\n",
+      "train loss:0.010808978092271031\n",
+      "train loss:0.05846781249065958\n",
+      "train loss:0.025380263678885667\n",
+      "train loss:0.014102585034342014\n",
+      "train loss:0.012906671429253107\n",
+      "train loss:0.015208402184522223\n",
+      "train loss:0.004694239649607978\n",
+      "train loss:0.005473208095488871\n",
+      "train loss:0.021743390148122682\n",
+      "train loss:0.0032255877148841677\n",
+      "train loss:0.006967549110816409\n",
+      "train loss:0.011621884625017192\n",
+      "train loss:0.024291477455977244\n",
+      "train loss:0.03492017607326153\n",
+      "train loss:0.002843313695822205\n",
+      "train loss:0.051686368717381345\n",
+      "train loss:0.0054119952353788815\n",
+      "train loss:0.0036881152450715326\n",
+      "train loss:0.0027317291563933766\n",
+      "train loss:0.004281091375243681\n",
+      "train loss:0.010913890654216461\n",
+      "train loss:0.0033970130193789892\n",
+      "train loss:0.009014667342532097\n",
+      "train loss:0.0009422580121984322\n",
+      "train loss:0.08561098205595805\n",
+      "train loss:0.01446078062724565\n",
+      "train loss:0.027670057325861183\n",
+      "train loss:0.004816356397470394\n",
+      "train loss:0.004462722432782995\n",
+      "train loss:0.004433115808016966\n",
+      "train loss:0.024158608532180853\n",
+      "train loss:0.007087376479529551\n",
+      "train loss:0.003930470643242003\n",
+      "train loss:0.007306881868514758\n",
+      "train loss:0.008018683178154908\n",
+      "train loss:0.01167685888020285\n",
+      "train loss:0.003710924847422856\n",
+      "train loss:0.002151311386103539\n",
+      "train loss:0.0032209928859439187\n",
+      "train loss:0.00574971839807132\n",
+      "train loss:0.011018603697619294\n",
+      "train loss:0.013986763567898252\n",
+      "train loss:0.009165166992611497\n",
+      "train loss:0.04379250090883091\n",
+      "train loss:0.005823338702910391\n",
+      "train loss:0.008685921750822395\n",
+      "train loss:0.00127850496606733\n",
+      "train loss:0.009579144223450487\n",
+      "train loss:0.004559377265024207\n",
+      "train loss:0.0019825488259478294\n",
+      "train loss:0.004435916230710923\n",
+      "train loss:0.021930266189404343\n",
+      "train loss:0.016108998984859072\n",
+      "train loss:0.0025341634876324733\n",
+      "train loss:0.03692776693626827\n",
+      "train loss:0.006153356826988565\n",
+      "train loss:0.004369950851858173\n",
+      "train loss:0.0030608601247814633\n",
+      "train loss:0.0022712412975201624\n",
+      "train loss:0.03595452158880824\n",
+      "train loss:0.004160967847048388\n",
+      "train loss:0.02176604051529482\n",
+      "train loss:0.006468030502166083\n",
+      "train loss:0.003733227048713024\n",
+      "train loss:0.008473977257721641\n",
+      "train loss:0.04307541330594716\n",
+      "train loss:0.004500356894464957\n",
+      "train loss:0.01294613947124066\n",
+      "train loss:0.004725818377212832\n",
+      "train loss:0.01726763582139779\n",
+      "train loss:0.0023397650069373575\n",
+      "train loss:0.006935735822036072\n",
+      "train loss:0.009267559333772182\n",
+      "train loss:0.02289851949074081\n",
+      "train loss:0.014668346283440527\n",
+      "train loss:0.04343490941255535\n",
+      "train loss:0.005392486758494316\n",
+      "train loss:0.009258763043241955\n",
+      "train loss:0.013681839394000634\n",
+      "train loss:0.031322100451792304\n",
+      "train loss:0.0018873017203015741\n",
+      "train loss:0.004473871762060277\n",
+      "train loss:0.0012550211374336138\n",
+      "train loss:0.002599533827864814\n",
+      "train loss:0.07498426658744929\n",
+      "train loss:0.0032558776681716538\n",
+      "train loss:0.028890592235265796\n",
+      "train loss:0.00571472249770388\n",
+      "train loss:0.011693113001846011\n",
+      "train loss:0.0095527149347024\n",
+      "train loss:0.011798272082914856\n",
+      "train loss:0.007065457522452855\n",
+      "train loss:0.003475585903096091\n",
+      "train loss:0.01493441788570096\n",
+      "train loss:0.005265355431852451\n",
+      "train loss:0.026143378966956943\n",
+      "train loss:0.00491744933759224\n",
+      "train loss:0.009139593651029449\n",
+      "train loss:0.0015026661444194434\n",
+      "train loss:0.03240064544444028\n",
+      "train loss:0.014455733480721884\n",
+      "train loss:0.009003356656630908\n",
+      "train loss:0.005923712631090092\n",
+      "train loss:0.005292778282371534\n",
+      "train loss:0.009395740792082276\n",
+      "train loss:0.016599876973693216\n",
+      "train loss:0.019115806610656653\n",
+      "train loss:0.008196028942502619\n",
+      "train loss:0.004247783478219096\n",
+      "train loss:0.003353410943224375\n",
+      "train loss:0.004884246955476793\n",
+      "train loss:0.0009807511753146636\n",
+      "train loss:0.02856832213484933\n",
+      "train loss:0.0033067316195050656\n",
+      "train loss:0.004078704580442304\n",
+      "train loss:0.007325851876259669\n",
+      "train loss:0.019383274833242873\n",
+      "train loss:0.020456623834272717\n",
+      "train loss:0.0007812974351538493\n",
+      "train loss:0.010843857314912695\n",
+      "train loss:0.004374667658293031\n",
+      "train loss:0.0241626129337257\n",
+      "train loss:0.007859955065776712\n",
+      "train loss:0.039929271634942504\n",
+      "train loss:0.00708102940051759\n",
+      "train loss:0.002624410636700337\n",
+      "train loss:0.00650547230371779\n",
+      "train loss:0.010568597329989065\n",
+      "train loss:0.008234310313618346\n",
+      "train loss:0.004725266822917307\n",
+      "train loss:0.004753673996025746\n",
+      "train loss:0.007698820845181547\n",
+      "train loss:0.009896022916885955\n",
+      "train loss:0.009365994011561601\n",
+      "train loss:0.007576837032226577\n",
+      "train loss:0.0344796782165176\n",
+      "train loss:0.003909452744092582\n",
+      "train loss:0.005380109141842455\n",
+      "train loss:0.005993575018270446\n",
+      "train loss:0.01097936458937783\n",
+      "train loss:0.01757947685048142\n",
+      "train loss:0.008758991604156737\n",
+      "train loss:0.005185181542705261\n",
+      "train loss:0.015785332399428354\n",
+      "train loss:0.0067532064157028926\n",
+      "train loss:0.0034247557514774695\n",
+      "train loss:0.004716121845813802\n",
+      "train loss:0.001155578829662111\n",
+      "train loss:0.006303270081247864\n",
+      "train loss:0.008560927814341457\n",
+      "train loss:0.03023966141859234\n",
+      "train loss:0.006041512768135148\n",
+      "train loss:0.0066959446266778985\n",
+      "train loss:0.004147336417380455\n",
+      "train loss:0.005078271105181157\n",
+      "train loss:0.018578621732352224\n",
+      "train loss:0.034121246516879755\n",
+      "train loss:0.0056593755391443415\n",
+      "train loss:0.00596388220405205\n",
+      "train loss:0.05147466660415796\n",
+      "train loss:0.0027005323502088344\n",
+      "train loss:0.0033507742051761013\n",
+      "train loss:0.007075705192100596\n",
+      "train loss:0.009565265695464793\n",
+      "train loss:0.03134279510666721\n",
+      "train loss:0.004745172566859552\n",
+      "train loss:0.012181324638229659\n",
+      "train loss:0.0011054559326565874\n",
+      "train loss:0.005382196864583967\n",
+      "train loss:0.009121434129752932\n",
+      "train loss:0.011631778190075652\n",
+      "train loss:0.004555014918595227\n",
+      "train loss:0.004889650157748363\n",
+      "train loss:0.009096710086587034\n",
+      "train loss:0.00924553269244445\n",
+      "train loss:0.007944747461751377\n",
+      "train loss:0.003459500841613428\n",
+      "train loss:0.01566008527427497\n",
+      "train loss:0.0008911464800127508\n",
+      "train loss:0.001091698205476946\n",
+      "train loss:0.0009505770695874688\n",
+      "train loss:0.007867986287712734\n",
+      "train loss:0.004968000324636349\n",
+      "train loss:0.0056121285885257525\n",
+      "train loss:0.007904635754945705\n",
+      "train loss:0.005197557588620477\n",
+      "train loss:0.005312977587160819\n",
+      "train loss:0.0020087423437651327\n",
+      "train loss:0.005294712491600181\n",
+      "train loss:0.0012405246341708844\n",
+      "train loss:0.0031783980822181774\n",
+      "train loss:0.1039724174242753\n",
+      "train loss:0.0037169780116194113\n",
+      "train loss:0.009759011343440406\n",
+      "train loss:0.0048018812051014645\n",
+      "train loss:0.006350426595224866\n",
+      "train loss:0.0008512095176504292\n",
+      "train loss:0.024253785368810593\n",
+      "train loss:0.0025916427487574062\n",
+      "train loss:0.0033269182947714727\n",
+      "train loss:0.09459760155500772\n",
+      "train loss:0.0047064993301111566\n",
+      "train loss:0.009045834540739643\n",
+      "train loss:0.0092606440169875\n",
+      "train loss:0.0075359863657588665\n",
+      "train loss:0.0183934461734446\n",
+      "train loss:0.01849892897252604\n",
+      "train loss:0.050874723831401104\n",
+      "train loss:0.007198767667317597\n",
+      "train loss:0.018661939001193967\n",
+      "train loss:0.0022028017841392174\n",
+      "train loss:0.018174116082275663\n",
+      "train loss:0.0017078903514663622\n",
+      "train loss:0.1459966672693368\n",
+      "train loss:0.006329843863072363\n",
+      "train loss:0.014378688194342711\n",
+      "train loss:0.01018378928810117\n",
+      "train loss:0.0009835564430315331\n",
+      "train loss:0.0056429997290290725\n",
+      "train loss:0.0010823893738730554\n",
+      "train loss:0.0043434085569439545\n",
+      "train loss:0.0030214509992223726\n",
+      "train loss:0.009926291548281655\n",
+      "train loss:0.015583007579840477\n",
+      "train loss:0.04296343915525511\n",
+      "train loss:0.005700470032062838\n",
+      "train loss:0.006183856552480422\n",
+      "train loss:0.0103129350615071\n",
+      "train loss:0.005623782881403926\n",
+      "train loss:0.004771483108086103\n",
+      "train loss:0.0022846827441920757\n",
+      "train loss:0.0070627647061852605\n",
+      "train loss:0.011700000064499427\n",
+      "train loss:0.006878545896291811\n",
+      "train loss:0.022344043115474315\n",
+      "train loss:0.013311365446273444\n",
+      "train loss:0.018434942623669858\n",
+      "train loss:0.01775084619643194\n",
+      "train loss:0.013375670648267272\n",
+      "train loss:0.002056485497154178\n",
+      "train loss:0.0018363345092885655\n",
+      "train loss:0.0011910873054445333\n",
+      "train loss:0.0021574398454583185\n",
+      "train loss:0.006961013560578591\n",
+      "train loss:0.04977116139251549\n",
+      "train loss:0.024193106720847082\n",
+      "train loss:0.019680667616881774\n",
+      "train loss:0.0052599166607132705\n",
+      "train loss:0.015396218206327075\n",
+      "train loss:0.0025889084914699186\n",
+      "train loss:0.008656963674933916\n",
+      "train loss:0.011366341339881484\n",
+      "train loss:0.0022448520707168173\n",
+      "train loss:0.0003427579085717119\n",
+      "train loss:0.014017099789055083\n",
+      "train loss:0.005286911899970182\n",
+      "train loss:0.002328978683078542\n",
+      "train loss:0.004394594403320315\n",
+      "train loss:0.003106831655873691\n",
+      "train loss:0.005016104437630236\n",
+      "train loss:0.0431066040586051\n",
+      "train loss:0.0076347954210563476\n",
+      "train loss:0.004873603683503737\n",
+      "train loss:0.01043082406237778\n",
+      "train loss:0.006092876008172955\n",
+      "train loss:0.005255821672366996\n",
+      "train loss:0.005628588583878875\n",
+      "train loss:0.0018846267040014606\n",
+      "train loss:0.0034342418193344998\n",
+      "train loss:0.0031498871117091242\n",
+      "train loss:0.016940578448135055\n",
+      "train loss:0.003260483259769068\n",
+      "train loss:0.01609160404934448\n",
+      "train loss:0.02206764472597681\n",
+      "train loss:0.008207123560034134\n",
+      "train loss:0.004358644595891921\n",
+      "train loss:0.007164831073969598\n",
+      "train loss:0.02634636059960875\n",
+      "train loss:0.017935455134055398\n",
+      "train loss:0.013872612345251953\n",
+      "train loss:0.014928728286260637\n",
+      "train loss:0.0033717031192576656\n",
+      "train loss:0.00822355061135789\n",
+      "train loss:0.005488733158157799\n",
+      "train loss:0.02550137055464332\n",
+      "train loss:0.011211660941091495\n",
+      "train loss:0.020907006192941047\n",
+      "train loss:0.012536649616985808\n",
+      "train loss:0.0031854216375890552\n",
+      "train loss:0.005385548680637622\n",
+      "train loss:0.024517075339633574\n",
+      "train loss:0.0074035240331445365\n",
+      "train loss:0.004254139067435342\n",
+      "train loss:0.0058829869904971\n",
+      "train loss:0.00853292432222234\n",
+      "train loss:0.02421496105187329\n",
+      "train loss:0.030974674281876852\n",
+      "train loss:0.013558521101576858\n",
+      "train loss:0.001858841869825652\n",
+      "train loss:0.02249357178600816\n",
+      "train loss:0.000500688914726695\n",
+      "train loss:0.012754598342220121\n",
+      "train loss:0.0023250307927846618\n",
+      "train loss:0.015588251430763464\n",
+      "train loss:0.021620929276049094\n",
+      "train loss:0.0032378818279185924\n",
+      "train loss:0.009692469524749886\n",
+      "train loss:0.08153647493011407\n",
+      "train loss:0.0056021477028105685\n",
+      "train loss:0.0008558945811552965\n",
+      "train loss:0.0024013208819945843\n",
+      "train loss:0.002022813904675698\n",
+      "train loss:0.0037117716681564776\n",
+      "train loss:0.00839416765208489\n",
+      "train loss:0.0022782768690073757\n",
+      "train loss:0.002042345942153006\n",
+      "train loss:0.037942080621939274\n",
+      "train loss:0.022064514240763406\n",
+      "train loss:0.0024210047231849226\n",
+      "train loss:0.0022182282787880678\n",
+      "train loss:0.002960918835554736\n",
+      "train loss:0.01687819454455562\n",
+      "train loss:0.032778960805304874\n",
+      "train loss:0.007558917867781768\n",
+      "train loss:0.006276258094573075\n",
+      "train loss:0.006903485956991742\n",
+      "train loss:0.01329753019022129\n",
+      "train loss:0.006100418517242749\n",
+      "train loss:0.0016190198364459498\n",
+      "train loss:0.006357330621681482\n",
+      "train loss:0.014051048830323026\n",
+      "train loss:0.0019165278873068514\n",
+      "train loss:0.012283409253111272\n",
+      "train loss:0.0004767032382887312\n",
+      "train loss:0.019315072446860048\n",
+      "train loss:0.013422040127603563\n",
+      "train loss:0.0006387369676177679\n",
+      "train loss:0.044423117100681545\n",
+      "train loss:0.009307951672255227\n",
+      "train loss:0.001652964685332176\n",
+      "train loss:0.0065048787566231135\n",
+      "train loss:0.021696148099957905\n",
+      "train loss:0.0009728583363724333\n",
+      "train loss:0.0011444721589118868\n",
+      "train loss:0.0015734742321270748\n",
+      "train loss:0.02143178689379717\n",
+      "train loss:0.007039078397827944\n",
+      "train loss:0.0009912872874003715\n",
+      "train loss:0.0034309013445205966\n",
+      "train loss:0.02275822068859825\n",
+      "train loss:0.003594258553329393\n",
+      "train loss:0.032178619615598064\n",
+      "train loss:0.013172302543865875\n",
+      "train loss:0.027110886972945013\n",
+      "train loss:0.006881847360590838\n",
+      "train loss:0.013791915707289467\n",
+      "train loss:0.012229731573256577\n",
+      "train loss:0.00996833478747986\n",
+      "train loss:0.0024573178923401173\n",
+      "train loss:0.003561402526371457\n",
+      "train loss:0.013769882671671042\n",
+      "train loss:0.00409994060061491\n",
+      "train loss:0.0033968952817230265\n",
+      "train loss:0.01747878703594751\n",
+      "train loss:0.00522876012038518\n",
+      "train loss:0.021446393141952794\n",
+      "train loss:0.008928426123211796\n",
+      "train loss:0.02051899569395377\n",
+      "train loss:0.011205647668535773\n",
+      "train loss:0.006976013388955122\n",
+      "train loss:0.008869331220677415\n",
+      "train loss:0.008539244512300198\n",
+      "train loss:0.0036322542240998412\n",
+      "train loss:0.04234968919276943\n",
+      "train loss:0.00334966715203166\n",
+      "train loss:0.009029382628082144\n",
+      "train loss:0.02172250058242092\n",
+      "train loss:0.015452175988101098\n",
+      "train loss:0.0027217219598769642\n",
+      "train loss:0.007235037686440381\n",
+      "train loss:0.01051027117794479\n",
+      "train loss:0.003938026690403711\n",
+      "train loss:0.003802565763352752\n",
+      "train loss:0.004761418443436841\n",
+      "train loss:0.013461807305842228\n",
+      "train loss:0.002036456168702846\n",
+      "train loss:0.003032060781626539\n",
+      "train loss:0.01591748573551803\n",
+      "train loss:0.006706472419507407\n",
+      "train loss:0.0034934718405741404\n",
+      "train loss:0.00937394216336106\n",
+      "train loss:0.00396047530166028\n",
+      "train loss:0.01915445294472339\n",
+      "train loss:0.010915591020089887\n",
+      "train loss:0.0034371283521484846\n",
+      "=== epoch:10, train acc:0.989, test acc:0.986 ===\n",
+      "train loss:0.004152176802283203\n",
+      "train loss:0.011444205147666206\n",
+      "train loss:0.0030225899060887028\n",
+      "train loss:0.0032246510361145764\n",
+      "train loss:0.010826100063612305\n",
+      "train loss:0.01925036409598172\n",
+      "train loss:0.03748163254670555\n",
+      "train loss:0.006804604279780044\n",
+      "train loss:0.0015559248017458487\n",
+      "train loss:0.011588371973761667\n",
+      "train loss:0.0011244018593414996\n",
+      "train loss:0.005755780073222744\n",
+      "train loss:0.001335918719565163\n",
+      "train loss:0.008787724078582478\n",
+      "train loss:0.00445628450573777\n",
+      "train loss:0.008226852223695907\n",
+      "train loss:0.00866795057454833\n",
+      "train loss:0.050613944977183155\n",
+      "train loss:0.043935737654510164\n",
+      "train loss:0.0004296089270960264\n",
+      "train loss:0.003922235465478686\n",
+      "train loss:0.008777787015849805\n",
+      "train loss:0.0015961356420109904\n",
+      "train loss:0.0019552588034362868\n",
+      "train loss:0.005105671738137553\n",
+      "train loss:0.0016221446635984735\n",
+      "train loss:0.0009908816073145917\n",
+      "train loss:0.012369650943206403\n",
+      "train loss:0.0037909478719594415\n",
+      "train loss:0.0019424340382218042\n",
+      "train loss:0.010982227484106459\n",
+      "train loss:0.009838119056063263\n",
+      "train loss:0.018283744681136062\n",
+      "train loss:0.005254002601885219\n",
+      "train loss:0.003976950807736353\n",
+      "train loss:0.01436884271221458\n",
+      "train loss:0.0065748140868349\n",
+      "train loss:0.009784215468705733\n",
+      "train loss:0.009065781362288914\n",
+      "train loss:0.00010137585387863971\n",
+      "train loss:0.008197679727035882\n",
+      "train loss:0.023851416361961073\n",
+      "train loss:0.006409149879717166\n",
+      "train loss:0.012286175915197048\n",
+      "train loss:0.0005813449133810354\n",
+      "train loss:0.00887810792283175\n",
+      "train loss:0.036218745003282475\n",
+      "train loss:0.0013044255918975697\n",
+      "train loss:0.004941251583070005\n",
+      "train loss:0.0039889271424887705\n",
+      "train loss:0.0059408621521440585\n",
+      "train loss:0.004704955631276692\n",
+      "train loss:0.003533674453316602\n",
+      "train loss:0.02296450142280602\n",
+      "train loss:0.0013853727530979076\n",
+      "train loss:0.011181854269216569\n",
+      "train loss:0.0052373813177011355\n",
+      "train loss:0.007009587040047091\n",
+      "train loss:0.016271642438821066\n",
+      "train loss:0.018392050337313066\n",
+      "train loss:0.003102470684099587\n",
+      "train loss:0.008863592255879254\n",
+      "train loss:0.006629543464891228\n",
+      "train loss:0.004526557809150268\n",
+      "train loss:0.006377352673897548\n",
+      "train loss:0.003976700677642091\n",
+      "train loss:0.0124848273485682\n",
+      "train loss:0.0022984895374087485\n",
+      "train loss:0.004144639287414399\n",
+      "train loss:0.008308529154235515\n",
+      "train loss:0.0015845167795295737\n",
+      "train loss:0.015645885479871596\n",
+      "train loss:0.004743162896889404\n",
+      "train loss:0.0018403363497701219\n",
+      "train loss:0.004996461947039934\n",
+      "train loss:0.0036374079437583844\n",
+      "train loss:0.007232472291638347\n",
+      "train loss:0.0016694687015468485\n",
+      "train loss:0.0030586325196168755\n",
+      "train loss:0.0018718630895809421\n",
+      "train loss:0.011632124940242404\n",
+      "train loss:0.007338786194562198\n",
+      "train loss:0.014984224186938582\n",
+      "train loss:0.0039652338125618376\n",
+      "train loss:0.00280052742728254\n",
+      "train loss:0.003036844016987424\n",
+      "train loss:0.007955794362476374\n",
+      "train loss:0.001988605167315156\n",
+      "train loss:0.00794910421496666\n",
+      "train loss:0.0011481258337704614\n",
+      "train loss:0.005662495582648102\n",
+      "train loss:0.007415225524966663\n",
+      "train loss:0.0015038768900600625\n",
+      "train loss:0.00984864938862302\n",
+      "train loss:0.006938702746062476\n",
+      "train loss:0.018039497244816197\n",
+      "train loss:0.003025990341469284\n",
+      "train loss:0.00562091273474571\n",
+      "train loss:0.0065531564300997935\n",
+      "train loss:0.006755176382765522\n",
+      "train loss:0.024288310092201594\n",
+      "train loss:0.002506539743258724\n",
+      "train loss:0.002304213266988145\n",
+      "train loss:0.0031478540437329646\n",
+      "train loss:0.009582751781976777\n",
+      "train loss:0.004033307919316155\n",
+      "train loss:0.004218567261496712\n",
+      "train loss:0.014702040811399213\n",
+      "train loss:0.0067563482393045125\n",
+      "train loss:0.023191206883774983\n",
+      "train loss:0.005646810998936548\n",
+      "train loss:0.09237152758798681\n",
+      "train loss:0.003892881996505428\n",
+      "train loss:0.0018961048079536877\n",
+      "train loss:0.006085788404784319\n",
+      "train loss:0.01041089233763039\n",
+      "train loss:0.006309472880187589\n",
+      "train loss:0.007559100464773488\n",
+      "train loss:0.016832195231354763\n",
+      "train loss:0.06050888288980862\n",
+      "train loss:0.017723686363937986\n",
+      "train loss:0.0053843628834639445\n",
+      "train loss:0.008171828780703922\n",
+      "train loss:0.018670230332516958\n",
+      "train loss:0.00587239759795825\n",
+      "train loss:0.004058870698659836\n",
+      "train loss:0.002293619182250444\n",
+      "train loss:0.027587863356592468\n",
+      "train loss:0.004877951808091856\n",
+      "train loss:0.035544737605570045\n",
+      "train loss:0.008851568508111167\n",
+      "train loss:0.03060677191262822\n",
+      "train loss:0.004274815906810381\n",
+      "train loss:0.0012645923572440032\n",
+      "train loss:0.00895642186452837\n",
+      "train loss:0.0019503432033776485\n",
+      "train loss:0.011977206119950474\n",
+      "train loss:0.0064871738449147455\n",
+      "train loss:0.008791048347450396\n",
+      "train loss:0.03941645974285413\n",
+      "train loss:0.0031139785254499512\n",
+      "train loss:0.008804812225670198\n",
+      "train loss:0.015446399163162518\n",
+      "train loss:0.01178995413121598\n",
+      "train loss:0.012705173513700605\n",
+      "train loss:0.002776205881560229\n",
+      "train loss:0.0033605426617531347\n",
+      "train loss:0.015559460055161982\n",
+      "train loss:0.19253853323839526\n",
+      "train loss:0.0014801544180012053\n",
+      "train loss:0.0010453993262331813\n",
+      "train loss:0.016880021050123865\n",
+      "train loss:0.0016743053392563908\n",
+      "train loss:0.02552353378583486\n",
+      "train loss:0.02667154541932183\n",
+      "train loss:0.01636829730417767\n",
+      "train loss:0.010398691566562696\n",
+      "train loss:0.002952536639568324\n",
+      "train loss:0.0022809507835539313\n",
+      "train loss:0.03177860560299269\n",
+      "train loss:0.0024455194081306852\n",
+      "train loss:0.04250535281196976\n",
+      "train loss:0.0018801631652352016\n",
+      "train loss:0.010116311668957413\n",
+      "train loss:0.0026409221025671587\n",
+      "train loss:0.004528142863142443\n",
+      "train loss:0.039915295636342\n",
+      "train loss:0.0007970028629885858\n",
+      "train loss:0.007907190568488152\n",
+      "train loss:0.003619876160318143\n",
+      "train loss:0.006463545515821144\n",
+      "train loss:0.0030356602720395865\n",
+      "train loss:0.009037781312917973\n",
+      "train loss:0.017996336650215314\n",
+      "train loss:0.0032715921455107287\n",
+      "train loss:0.0022573405763575633\n",
+      "train loss:0.004957120138560164\n",
+      "train loss:0.0014058721564725399\n",
+      "train loss:0.000508258734177101\n",
+      "train loss:0.02629597096937675\n",
+      "train loss:0.0020708313090385664\n",
+      "train loss:0.010570595215563711\n",
+      "train loss:0.014442033299753122\n",
+      "train loss:0.03163789662125663\n",
+      "train loss:0.011214360851325593\n",
+      "train loss:0.0016539218637576616\n",
+      "train loss:0.015065405649349614\n",
+      "train loss:0.0042481774234032045\n",
+      "train loss:0.0365635818937145\n",
+      "train loss:0.0187607010391974\n",
+      "train loss:0.016236444759842424\n",
+      "train loss:0.010210423897323408\n",
+      "train loss:0.011535402305981315\n",
+      "train loss:0.0031863022810900005\n",
+      "train loss:0.0077927554284317295\n",
+      "train loss:0.03162864656650884\n",
+      "train loss:0.00913570046842477\n",
+      "train loss:0.006321493403483717\n",
+      "train loss:0.004688769559091957\n",
+      "train loss:0.01720723521895977\n",
+      "train loss:0.017491022681434033\n",
+      "train loss:0.02022690844253853\n",
+      "train loss:0.014531308201273897\n",
+      "train loss:0.011142833573037539\n",
+      "train loss:0.04209690975951949\n",
+      "train loss:0.007515261623397651\n",
+      "train loss:0.0016203399270297264\n",
+      "train loss:0.003551369828198259\n",
+      "train loss:0.0038023169971498206\n",
+      "train loss:0.005365255833491056\n",
+      "train loss:0.00940407506604825\n",
+      "train loss:0.007076022876153748\n",
+      "train loss:0.026968088595980433\n",
+      "train loss:0.023553248840064694\n",
+      "train loss:0.012660836317151179\n",
+      "train loss:0.023381899167499314\n",
+      "train loss:0.004440872321325066\n",
+      "train loss:0.0022297048031715656\n",
+      "train loss:0.0036135693744899517\n",
+      "train loss:0.006062412900495341\n",
+      "train loss:0.00763313630780239\n",
+      "train loss:0.005312463037937148\n",
+      "train loss:0.012403768945458775\n",
+      "train loss:0.016727559698249763\n",
+      "train loss:0.007407309776003328\n",
+      "train loss:0.0030617948111259265\n",
+      "train loss:0.004050634106630292\n",
+      "train loss:0.01631553155875064\n",
+      "train loss:0.006314531440893537\n",
+      "train loss:0.011844594576885487\n",
+      "train loss:0.004479272722390536\n",
+      "train loss:0.014012735044091389\n",
+      "train loss:0.0020072968060891707\n",
+      "train loss:0.0019800034861049186\n",
+      "train loss:0.009935626827723233\n",
+      "train loss:0.009022507922718817\n",
+      "train loss:0.005707372971983091\n",
+      "train loss:0.006792622433184225\n",
+      "train loss:0.006062630121579763\n",
+      "train loss:0.004544462390145606\n",
+      "train loss:0.022192662565561416\n",
+      "train loss:0.0012970715676425546\n",
+      "train loss:0.00448858824184376\n",
+      "train loss:0.0038395385266455895\n",
+      "train loss:0.0048711400312821465\n",
+      "train loss:0.008309863808701808\n",
+      "train loss:0.0020361374737851434\n",
+      "train loss:0.006884572311579962\n",
+      "train loss:0.01185198496009391\n",
+      "train loss:0.012054044074346787\n",
+      "train loss:0.005285779416891919\n",
+      "train loss:0.0020569351920501616\n",
+      "train loss:0.0030297854828291474\n",
+      "train loss:0.002498707360927585\n",
+      "train loss:0.0036363989884754078\n",
+      "train loss:0.005337760399657166\n",
+      "train loss:0.003403019697501354\n",
+      "train loss:0.013257334164382779\n",
+      "train loss:0.0017378598209657383\n",
+      "train loss:0.0016579008210053962\n",
+      "train loss:0.008346985201645759\n",
+      "train loss:0.0061259862445442945\n",
+      "train loss:0.0045629837272112474\n",
+      "train loss:0.0001983821698395381\n",
+      "train loss:0.006355842387703583\n",
+      "train loss:0.004232285647050108\n",
+      "train loss:0.028713156911074504\n",
+      "train loss:0.02871413283706186\n",
+      "train loss:0.004311365173556812\n",
+      "train loss:0.011279562259503075\n",
+      "train loss:0.00035561967257554365\n",
+      "train loss:0.004302829084402483\n",
+      "train loss:0.002431436222318472\n",
+      "train loss:0.005694393072288307\n",
+      "train loss:0.006387481247450998\n",
+      "train loss:0.0040429165825466165\n",
+      "train loss:0.005574183775178038\n",
+      "train loss:0.017985103883166483\n",
+      "train loss:0.001374959562611247\n",
+      "train loss:0.005701622335423664\n",
+      "train loss:0.003809419616025812\n",
+      "train loss:0.0016931908411246049\n",
+      "train loss:0.004758264490107222\n",
+      "train loss:0.011711880174206959\n",
+      "train loss:0.0015842980897652964\n",
+      "train loss:0.002384628454336574\n",
+      "train loss:0.003682258117314366\n",
+      "train loss:0.002315576213999536\n",
+      "train loss:0.005122052109774559\n",
+      "train loss:0.005407160028527019\n",
+      "train loss:0.00514888027637812\n",
+      "train loss:0.005212652568258155\n",
+      "train loss:0.0027991643082745017\n",
+      "train loss:0.012892082505825499\n",
+      "train loss:0.0003681072234235739\n",
+      "train loss:0.004865555307634981\n",
+      "train loss:0.005695629210203078\n",
+      "train loss:0.0020208502631254223\n",
+      "train loss:0.000651128735427313\n",
+      "train loss:0.00181868095114618\n",
+      "train loss:0.022031493880561786\n",
+      "train loss:0.012881247562812155\n",
+      "train loss:0.001844323761778206\n",
+      "train loss:0.028795428161077163\n",
+      "train loss:0.0015846145404458778\n",
+      "train loss:0.008839389546673489\n",
+      "train loss:0.0030965027902207435\n",
+      "train loss:0.011549420196784298\n",
+      "train loss:0.017043338295091037\n",
+      "train loss:0.007023413321391934\n",
+      "train loss:0.00887168483199522\n",
+      "train loss:0.012174658259342995\n",
+      "train loss:0.007792827800663774\n",
+      "train loss:0.0026899757270953746\n",
+      "train loss:0.025075986017332123\n",
+      "train loss:0.005902624836237986\n",
+      "train loss:0.00457414448368691\n",
+      "train loss:0.002908225517702709\n",
+      "train loss:0.003425037249992881\n",
+      "train loss:0.004905477688690763\n",
+      "train loss:0.0028410894719949164\n",
+      "train loss:0.0020146865772631526\n",
+      "train loss:0.006100498241430243\n",
+      "train loss:0.0076703592051360955\n",
+      "train loss:0.001792293992331043\n",
+      "train loss:0.0037053007186089687\n",
+      "train loss:0.0033137442692250502\n",
+      "train loss:0.008214007799259416\n",
+      "train loss:0.01542315713600698\n",
+      "train loss:0.023903368614027764\n",
+      "train loss:0.004982187815756821\n",
+      "train loss:0.0003014732299713153\n",
+      "train loss:0.002187395432970617\n",
+      "train loss:0.003799415270831005\n",
+      "train loss:0.00706557449140978\n",
+      "train loss:0.011181486548667897\n",
+      "train loss:0.00089989862115641\n",
+      "train loss:0.013690704950368069\n",
+      "train loss:0.00482062036968136\n",
+      "train loss:0.0012661481840969796\n",
+      "train loss:0.02417832496007756\n",
+      "train loss:0.004163536522437518\n",
+      "train loss:0.011083192148831135\n",
+      "train loss:0.000559622598117533\n",
+      "train loss:0.0043615348229493036\n",
+      "train loss:0.007026372287966698\n",
+      "train loss:0.0005844082797375357\n",
+      "train loss:0.0026095645571620712\n",
+      "train loss:0.013850987393363353\n",
+      "train loss:0.005509457367032912\n",
+      "train loss:0.0013625886615260118\n",
+      "train loss:0.006679453855470479\n",
+      "train loss:0.001012464890558809\n",
+      "train loss:0.0020700354474246264\n",
+      "train loss:0.003899697094818998\n",
+      "train loss:0.0011171999219902063\n",
+      "train loss:0.0017787370271516864\n",
+      "train loss:0.0015948969997811211\n",
+      "train loss:0.0033144073081751175\n",
+      "train loss:0.003400788306595921\n",
+      "train loss:0.012866487458998878\n",
+      "train loss:0.014934543337768074\n",
+      "train loss:0.0024092666364162775\n",
+      "train loss:0.004509725451379536\n",
+      "train loss:0.006087250501243405\n",
+      "train loss:0.003412877020397752\n",
+      "train loss:0.03549528885845218\n",
+      "train loss:0.01731130227395986\n",
+      "train loss:0.001963949070901295\n",
+      "train loss:0.012289496842646533\n",
+      "train loss:0.010885711008645084\n",
+      "train loss:0.009048469292206174\n",
+      "train loss:0.02494724220522563\n",
+      "train loss:0.00730088706480893\n",
+      "train loss:0.002202662849544785\n",
+      "train loss:0.00546929617721405\n",
+      "train loss:0.002146131981499185\n",
+      "train loss:0.004173250010805002\n",
+      "train loss:0.0058655372247093255\n",
+      "train loss:0.0034196185980422665\n",
+      "train loss:0.007359830253231411\n",
+      "train loss:0.0005625306240807986\n",
+      "train loss:0.00250165555656888\n",
+      "train loss:0.0014903902292149563\n",
+      "train loss:0.01050926809532355\n",
+      "train loss:0.00195855215372101\n",
+      "train loss:0.011277552165263428\n",
+      "train loss:0.0008932906974572232\n",
+      "train loss:0.008678430198588293\n",
+      "train loss:0.0034394295486614357\n",
+      "train loss:0.014722541522140295\n",
+      "train loss:0.009254407211968126\n",
+      "train loss:0.016478383425742864\n",
+      "train loss:0.0018557601946513063\n",
+      "train loss:0.0871654388006691\n",
+      "train loss:0.0020563337853506725\n",
+      "train loss:0.008130153859024747\n",
+      "train loss:0.009006549155444507\n",
+      "train loss:0.008658453115534401\n",
+      "train loss:0.040885234437240736\n",
+      "train loss:0.012704277380765296\n",
+      "train loss:0.006412736227737689\n",
+      "train loss:0.01620393712192926\n",
+      "train loss:0.0012255161257290405\n",
+      "train loss:0.004944923273327346\n",
+      "train loss:0.004387143974960709\n",
+      "train loss:0.0006583260201638161\n",
+      "train loss:0.03294001025408699\n",
+      "train loss:0.0045972662028786766\n",
+      "train loss:0.009423016962468988\n",
+      "train loss:0.018466078065348734\n",
+      "train loss:0.009473973994762684\n",
+      "train loss:0.004803933909670826\n",
+      "train loss:0.010253255422843452\n",
+      "train loss:0.020636076702257443\n",
+      "train loss:0.005895734754781571\n",
+      "train loss:0.008560995298061527\n",
+      "train loss:0.004687866093769502\n",
+      "train loss:0.0007045204347177945\n",
+      "train loss:0.004793279131762746\n",
+      "train loss:0.025294653797476946\n",
+      "train loss:0.001347653879263792\n",
+      "train loss:0.04198975789722918\n",
+      "train loss:0.019157339273533826\n",
+      "train loss:0.006460850551584963\n",
+      "train loss:0.0016024836189168984\n",
+      "train loss:0.011548659978805088\n",
+      "train loss:0.007959428459098264\n",
+      "train loss:0.011130536083085134\n",
+      "train loss:0.0070220949853000226\n",
+      "train loss:0.008931085417561239\n",
+      "train loss:0.002867603419298801\n",
+      "train loss:0.007200632069659862\n",
+      "train loss:0.0019428967756518938\n",
+      "train loss:0.01735106648634741\n",
+      "train loss:0.03123150758891803\n",
+      "train loss:0.003264810820726164\n",
+      "train loss:0.011570399391342433\n",
+      "train loss:0.005250982654359495\n",
+      "train loss:0.006200933216706049\n",
+      "train loss:0.0033276458533423934\n",
+      "train loss:0.01634724938216908\n",
+      "train loss:0.0010329655746597233\n",
+      "train loss:0.0033890800047324835\n",
+      "train loss:0.004385547719369044\n",
+      "train loss:0.0074912768119097615\n",
+      "train loss:0.0028394554257394396\n",
+      "train loss:0.040512575737471225\n",
+      "train loss:0.033099269300293686\n",
+      "train loss:0.004399687116174101\n",
+      "train loss:0.020637662457113612\n",
+      "train loss:0.006700726987157735\n",
+      "train loss:0.003375010208945153\n",
+      "train loss:0.010251718479151068\n",
+      "train loss:0.001612451318164467\n",
+      "train loss:0.0010587946604568123\n",
+      "train loss:0.0051255233133494105\n",
+      "train loss:0.0028857640796250473\n",
+      "train loss:0.009301080572349234\n",
+      "train loss:0.002861213706256039\n",
+      "train loss:0.01582008909028616\n",
+      "train loss:0.0028376316141985798\n",
+      "train loss:0.0008556938289184822\n",
+      "train loss:0.007671839270960295\n",
+      "train loss:0.004225138823378669\n",
+      "train loss:0.025046317996923514\n",
+      "train loss:0.004392524039315647\n",
+      "train loss:0.001460340885398165\n",
+      "train loss:0.017345740290986015\n",
+      "train loss:0.00042582717328334\n",
+      "train loss:0.005895356631702772\n",
+      "train loss:0.002010271754766567\n",
+      "train loss:0.008133768900899507\n",
+      "train loss:0.004839356646756464\n",
+      "train loss:0.0003346819470559107\n",
+      "train loss:0.01601454837629541\n",
+      "train loss:0.001059611319036028\n",
+      "train loss:0.00821891258959908\n",
+      "train loss:0.005278390799694309\n",
+      "train loss:0.05144385589176356\n",
+      "train loss:0.0446838107420751\n",
+      "train loss:0.0022184476931855694\n",
+      "train loss:0.0043880040233680835\n",
+      "train loss:0.00035135967949394866\n",
+      "train loss:0.004176994001430222\n",
+      "train loss:0.00407940805583126\n",
+      "train loss:0.00460755439721878\n",
+      "train loss:0.0019918239846153886\n",
+      "train loss:0.004124948402163198\n",
+      "train loss:0.03496529276619796\n",
+      "train loss:0.004360507561666146\n",
+      "train loss:0.0023606379163853518\n",
+      "train loss:0.03050022619299674\n",
+      "train loss:0.016695070083234988\n",
+      "train loss:0.0057462448305650145\n",
+      "train loss:0.022633087348306317\n",
+      "train loss:0.0076068234918503985\n",
+      "train loss:0.002640002767394285\n",
+      "train loss:0.002212288814500479\n",
+      "train loss:0.06170205698477452\n",
+      "train loss:0.0053634253667708106\n",
+      "train loss:0.0021556532068859155\n",
+      "train loss:0.0024424308560409046\n",
+      "train loss:0.0015856334333924442\n",
+      "train loss:0.0020509425956977603\n",
+      "train loss:0.002311091361551937\n",
+      "train loss:0.006375821186815756\n",
+      "train loss:0.01327338080286732\n",
+      "train loss:0.004214986969628925\n",
+      "train loss:0.008766110435489112\n",
+      "train loss:0.013034686070446215\n",
+      "train loss:0.006659086692630605\n",
+      "train loss:0.010018540377748344\n",
+      "train loss:0.007917543456941838\n",
+      "train loss:0.01748703939097911\n",
+      "train loss:0.0032995605383161065\n",
+      "train loss:0.0006794536762746111\n",
+      "train loss:0.0035242809570686685\n",
+      "train loss:0.011107767653358509\n",
+      "train loss:0.1632281403389209\n",
+      "train loss:0.058854250019591506\n",
+      "train loss:0.021345068207908\n",
+      "train loss:0.007124172699745627\n",
+      "train loss:0.043099357273770435\n",
+      "train loss:0.004009789756743425\n",
+      "train loss:0.005289111646842782\n",
+      "train loss:0.007865039647149805\n",
+      "train loss:0.011149513684191316\n",
+      "train loss:0.00967526020072876\n",
+      "train loss:0.008723442618054559\n",
+      "train loss:0.015312079653140274\n",
+      "train loss:0.0020735847637150568\n",
+      "train loss:0.004785437195525382\n",
+      "train loss:0.011125161818402838\n",
+      "train loss:0.03995094178378101\n",
+      "train loss:0.011394234986826084\n",
+      "train loss:0.005271050114117268\n",
+      "train loss:0.0012102201741897977\n",
+      "train loss:0.011462184385073855\n",
+      "train loss:0.005897526837784661\n",
+      "train loss:0.0006637649192019796\n",
+      "train loss:0.023315687872842328\n",
+      "train loss:0.0014734861250271917\n",
+      "train loss:0.0035057163305640907\n",
+      "train loss:0.003071592461783681\n",
+      "train loss:0.0030116382528132223\n",
+      "train loss:0.014601909607848051\n",
+      "train loss:0.011775535791360606\n",
+      "train loss:0.033980496909716704\n",
+      "train loss:0.005553313289321587\n",
+      "train loss:0.008525879879512366\n",
+      "train loss:0.006046704337049359\n",
+      "train loss:0.002792824937024913\n",
+      "train loss:0.020220408316787074\n",
+      "train loss:0.0007643390140597054\n",
+      "train loss:0.009490095343874919\n",
+      "train loss:0.020379235412240244\n",
+      "train loss:0.0029084238461013715\n",
+      "train loss:0.001765403109198668\n",
+      "train loss:0.0010202268477971781\n",
+      "train loss:0.012777063716856331\n",
+      "train loss:0.004869947458613834\n",
+      "train loss:0.004992299044964749\n",
+      "train loss:0.007199493166733641\n",
+      "train loss:0.006340438055080669\n",
+      "train loss:0.0017292664871103023\n",
+      "train loss:0.007319864058387008\n",
+      "train loss:0.001729700599264023\n",
+      "train loss:0.010348993635794728\n",
+      "train loss:0.004312398459853025\n",
+      "train loss:0.030368879879162935\n",
+      "train loss:0.0024497006119297096\n",
+      "train loss:0.0016662532553674185\n",
+      "train loss:0.01022428975966313\n",
+      "train loss:0.011020861716290157\n",
+      "train loss:0.007965307314351853\n",
+      "train loss:0.01672853047633309\n",
+      "train loss:0.009729522764196906\n",
+      "train loss:0.006314165267587124\n",
+      "train loss:0.0016331708177720647\n",
+      "train loss:0.01650790452903283\n",
+      "train loss:0.0018466096247191318\n",
+      "train loss:0.002823484735125333\n",
+      "train loss:0.004568096642175817\n",
+      "train loss:0.009614866810638757\n",
+      "train loss:0.0018029136311278988\n",
+      "train loss:0.0031469256911874637\n",
+      "train loss:0.0013833586903249108\n",
+      "train loss:0.016071206173172564\n",
+      "train loss:0.0024205915202175667\n",
+      "train loss:0.004569414386879623\n",
+      "train loss:0.0030047736697423793\n",
+      "train loss:0.0036970984381252997\n",
+      "train loss:0.029323190220021077\n",
+      "train loss:0.00483714128581959\n",
+      "train loss:0.005673180224060389\n",
+      "train loss:0.0009612323619111001\n",
+      "train loss:0.0022567460623184234\n",
+      "train loss:0.00888058270899279\n",
+      "train loss:0.03626939592835377\n",
+      "=== epoch:11, train acc:0.996, test acc:0.986 ===\n",
+      "train loss:0.0046927828200519165\n",
+      "train loss:0.001730212249779993\n",
+      "train loss:0.0037038832562686034\n",
+      "train loss:0.019515350940279803\n",
+      "train loss:0.0018721190960816214\n",
+      "train loss:0.003458201268484668\n",
+      "train loss:0.0008072177802511082\n",
+      "train loss:0.016843495157538736\n",
+      "train loss:0.002336712820631938\n",
+      "train loss:0.0025426882435072686\n",
+      "train loss:0.04007504770004573\n",
+      "train loss:0.0033943375631721967\n",
+      "train loss:0.005085230959254645\n",
+      "train loss:0.01047729762445885\n",
+      "train loss:0.01720206163889212\n",
+      "train loss:0.0027365914074844516\n",
+      "train loss:0.0013238513129382084\n",
+      "train loss:0.004813491762837606\n",
+      "train loss:0.0007577562813795548\n",
+      "train loss:0.002861503306049375\n",
+      "train loss:0.0051313556926099145\n",
+      "train loss:0.006987420473733599\n",
+      "train loss:0.0022100118930466215\n",
+      "train loss:0.027434583851913587\n",
+      "train loss:0.0023460527354198708\n",
+      "train loss:0.001617734838627232\n",
+      "train loss:0.013696642424561511\n",
+      "train loss:0.02998865829894999\n",
+      "train loss:0.0014236881514425733\n",
+      "train loss:0.008901156371783913\n",
+      "train loss:0.0018488512506746129\n",
+      "train loss:0.010227026837583069\n",
+      "train loss:0.004895195307048737\n",
+      "train loss:0.001296435914134556\n",
+      "train loss:0.0039383545155283655\n",
+      "train loss:0.0024625221065896633\n",
+      "train loss:0.0008095093857667415\n",
+      "train loss:0.0005045870610615032\n",
+      "train loss:0.003386565223136765\n",
+      "train loss:0.007943735044616379\n",
+      "train loss:0.002579314751532921\n",
+      "train loss:0.0023912150825315215\n",
+      "train loss:0.02939266954232469\n",
+      "train loss:0.004577588427072399\n",
+      "train loss:0.014838678090189818\n",
+      "train loss:0.12065589434246075\n",
+      "train loss:0.004769946494141588\n",
+      "train loss:0.0035403509719385915\n",
+      "train loss:0.006785569427551568\n",
+      "train loss:0.0176581599144324\n",
+      "train loss:0.007920229416770624\n",
+      "train loss:0.003083426002354803\n",
+      "train loss:0.003269725048909108\n",
+      "train loss:0.026363832185844913\n",
+      "train loss:0.0020513398877254514\n",
+      "train loss:0.0015993076319247303\n",
+      "train loss:0.002328489915252304\n",
+      "train loss:0.00396779470315162\n",
+      "train loss:0.010829262197618338\n",
+      "train loss:0.008246331218984364\n",
+      "train loss:0.017241462005732146\n",
+      "train loss:0.011819494893977605\n",
+      "train loss:0.0007689009951712352\n",
+      "train loss:0.005146464979721995\n",
+      "train loss:0.0012902640547254366\n",
+      "train loss:0.0010469500614460894\n",
+      "train loss:0.0006430037931543271\n",
+      "train loss:0.0005930521501728216\n",
+      "train loss:0.004399640957576296\n",
+      "train loss:0.0077712208146303455\n",
+      "train loss:0.0021598372365207704\n",
+      "train loss:0.0034220980247771575\n",
+      "train loss:0.007464862912667244\n",
+      "train loss:0.0006975155484991125\n",
+      "train loss:0.026882180255929793\n",
+      "train loss:0.006906415481317387\n",
+      "train loss:0.007905562599134119\n",
+      "train loss:0.0011315925833574247\n",
+      "train loss:0.00971874766056834\n",
+      "train loss:0.0029440983982673796\n",
+      "train loss:0.004563240174735952\n",
+      "train loss:0.006487367968885054\n",
+      "train loss:0.0442093471758195\n",
+      "train loss:0.01032242055750972\n",
+      "train loss:0.0013141951589451402\n",
+      "train loss:0.004215838459698886\n",
+      "train loss:0.001112230168249428\n",
+      "train loss:0.007776273897079846\n",
+      "train loss:0.0070586418214086365\n",
+      "train loss:0.006450645384517646\n",
+      "train loss:0.006561514128385527\n",
+      "train loss:0.0023831567408051117\n",
+      "train loss:0.0023025642046404274\n",
+      "train loss:0.0038216951336722993\n",
+      "train loss:0.0042412862726552425\n",
+      "train loss:0.0036921822172067565\n",
+      "train loss:0.003433694935875195\n",
+      "train loss:0.022181333140680794\n",
+      "train loss:0.006474966165706463\n",
+      "train loss:0.002594071418871366\n",
+      "train loss:0.0040617200579514436\n",
+      "train loss:0.007672779849836477\n",
+      "train loss:0.016863787630687466\n",
+      "train loss:0.0023875315996254964\n",
+      "train loss:0.0075080161439440955\n",
+      "train loss:0.03973253592087797\n",
+      "train loss:0.005027767259151228\n",
+      "train loss:0.014300717886048342\n",
+      "train loss:0.00590830732197374\n",
+      "train loss:0.003222310027061669\n",
+      "train loss:0.007066358355885813\n",
+      "train loss:0.0142540343160634\n",
+      "train loss:0.03549295288765461\n",
+      "train loss:0.002766220052530057\n",
+      "train loss:0.012812503508192392\n",
+      "train loss:0.00447994224033231\n",
+      "train loss:0.005366614535928535\n",
+      "train loss:0.0007002034249442603\n",
+      "train loss:0.003641524035408453\n",
+      "train loss:0.0006361098907971715\n",
+      "train loss:0.0024471430274702125\n",
+      "train loss:0.008564242015069432\n",
+      "train loss:0.008113371180690124\n",
+      "train loss:0.035217700693233794\n",
+      "train loss:0.0014676672995602542\n",
+      "train loss:0.0040154673887946996\n",
+      "train loss:0.06258507734234285\n",
+      "train loss:0.002121824783745121\n",
+      "train loss:0.0039547568378801475\n",
+      "train loss:0.003312538081326497\n",
+      "train loss:0.010669975326557564\n",
+      "train loss:0.006393669541296728\n",
+      "train loss:0.03569270221376572\n",
+      "train loss:0.013334603176983992\n",
+      "train loss:0.0033345303055432748\n",
+      "train loss:0.012168524850784474\n",
+      "train loss:0.001120952571883308\n",
+      "train loss:0.0053748545160571485\n",
+      "train loss:0.025984957789206717\n",
+      "train loss:0.027497441357579578\n",
+      "train loss:0.006677097351775015\n",
+      "train loss:0.010550976502824794\n",
+      "train loss:0.0015978898510777897\n",
+      "train loss:0.029309778847610154\n",
+      "train loss:0.0016635233499721317\n",
+      "train loss:0.0047437136385169104\n",
+      "train loss:0.00405463930016966\n",
+      "train loss:0.0034086386801385864\n",
+      "train loss:0.001491692394294545\n",
+      "train loss:0.005806096525999055\n",
+      "train loss:0.008868594696476934\n",
+      "train loss:0.0006600914938366249\n",
+      "train loss:0.0010519751512899714\n",
+      "train loss:0.0034366397982870858\n",
+      "train loss:0.003161914693274874\n",
+      "train loss:0.05815235909039389\n",
+      "train loss:0.00399682107136351\n",
+      "train loss:0.004538817156584775\n",
+      "train loss:0.004819702862643653\n",
+      "train loss:0.0044183587336733\n",
+      "train loss:0.003405883374724896\n",
+      "train loss:0.03690196807984236\n",
+      "train loss:0.0034790735700450247\n",
+      "train loss:0.003058741124781665\n",
+      "train loss:0.020106920270880936\n",
+      "train loss:0.012283233554968735\n",
+      "train loss:0.003848952443052774\n",
+      "train loss:0.01212661671748817\n",
+      "train loss:0.0011816647839549745\n",
+      "train loss:0.0019407714657557968\n",
+      "train loss:0.0015481848147886803\n",
+      "train loss:0.002440855932960091\n",
+      "train loss:0.020126682825228247\n",
+      "train loss:0.0014184747938272158\n",
+      "train loss:0.007977916035122145\n",
+      "train loss:0.030531333521751573\n",
+      "train loss:0.0081545615298041\n",
+      "train loss:0.025827861842246662\n",
+      "train loss:0.010173791354276422\n",
+      "train loss:0.008234847584883624\n",
+      "train loss:0.001962144898353665\n",
+      "train loss:0.007246603795722218\n",
+      "train loss:0.005181124227843119\n",
+      "train loss:0.016131903199234847\n",
+      "train loss:0.0012935957125459227\n",
+      "train loss:0.027539533535814526\n",
+      "train loss:0.003955899637661323\n",
+      "train loss:0.01491466948908333\n",
+      "train loss:0.00299190576330765\n",
+      "train loss:0.015280849525923134\n",
+      "train loss:0.013724502225004029\n",
+      "train loss:0.0009055741799527606\n",
+      "train loss:0.002640668903950562\n",
+      "train loss:0.0055162753905496605\n",
+      "train loss:0.013091351560480502\n",
+      "train loss:0.004857414965574689\n",
+      "train loss:0.006538897984851252\n",
+      "train loss:0.002432246138139135\n",
+      "train loss:0.015474245436760457\n",
+      "train loss:0.001824972086511069\n",
+      "train loss:0.0007515317685806835\n",
+      "train loss:0.0032644898164920296\n",
+      "train loss:0.003118674250849102\n",
+      "train loss:0.0020841949679050067\n",
+      "train loss:0.018902032671798533\n",
+      "train loss:0.0052745008617129585\n",
+      "train loss:0.013106959571672317\n",
+      "train loss:0.0028423342648072145\n",
+      "train loss:0.00038734539763264804\n",
+      "train loss:0.034273249069100224\n",
+      "train loss:0.0031608051290538835\n",
+      "train loss:0.0024426116809007988\n",
+      "train loss:0.0016573656428878992\n",
+      "train loss:0.002940588339870061\n",
+      "train loss:0.007152398218207527\n",
+      "train loss:0.0033259723728623542\n",
+      "train loss:0.024054019087551847\n",
+      "train loss:0.011359113687234688\n",
+      "train loss:0.002988149311127139\n",
+      "train loss:0.0019113812898063638\n",
+      "train loss:0.003061231606654664\n",
+      "train loss:0.002628351732898558\n",
+      "train loss:0.0076276682016799225\n",
+      "train loss:0.012034151092819675\n",
+      "train loss:0.0046543381259469925\n",
+      "train loss:0.0015455705826686317\n",
+      "train loss:0.004983751461141312\n",
+      "train loss:0.003500484891428882\n",
+      "train loss:0.0011841784307994182\n",
+      "train loss:0.005756025229879796\n",
+      "train loss:0.005048022304982861\n",
+      "train loss:0.003579769274101924\n",
+      "train loss:0.030260839465735795\n",
+      "train loss:0.00970299521773383\n",
+      "train loss:0.0009451822095723276\n",
+      "train loss:0.00908209362225059\n",
+      "train loss:0.006289070190290707\n",
+      "train loss:0.004171542314382857\n",
+      "train loss:0.003160250180964411\n",
+      "train loss:0.004497888423072236\n",
+      "train loss:0.0011329083187253369\n",
+      "train loss:0.004257242188313983\n",
+      "train loss:0.013651142823967687\n",
+      "train loss:0.0016768691499877359\n",
+      "train loss:0.0023810075933780855\n",
+      "train loss:0.004287408782700802\n",
+      "train loss:0.0012100985244105765\n",
+      "train loss:0.026298014464333525\n",
+      "train loss:0.0012674278922622554\n",
+      "train loss:0.0013599600745321672\n",
+      "train loss:0.0013005445914666682\n",
+      "train loss:0.007644598396100083\n",
+      "train loss:0.0004785242026665329\n",
+      "train loss:0.0007537480231809156\n",
+      "train loss:0.0015890400068608087\n",
+      "train loss:0.002920538168779892\n",
+      "train loss:0.006348234189236395\n",
+      "train loss:0.013966442683729836\n",
+      "train loss:0.013916052298748528\n",
+      "train loss:0.015174687132905453\n",
+      "train loss:0.0013613811512541872\n",
+      "train loss:0.02008631734513622\n",
+      "train loss:0.005378136831602338\n",
+      "train loss:0.006117318909533397\n",
+      "train loss:0.0029178166944324423\n",
+      "train loss:0.013035588845115482\n",
+      "train loss:0.0008923083923972186\n",
+      "train loss:0.0032947912631343845\n",
+      "train loss:0.00991973802884954\n",
+      "train loss:0.01593967768210907\n",
+      "train loss:0.004086199004396033\n",
+      "train loss:0.0014249951154592813\n",
+      "train loss:0.005210117901374343\n",
+      "train loss:0.00604788204910442\n",
+      "train loss:0.0003476524976375582\n",
+      "train loss:0.004391570931912198\n",
+      "train loss:0.0059204628986581735\n",
+      "train loss:0.009926677966892835\n",
+      "train loss:0.002694886511838026\n",
+      "train loss:0.006392083742442129\n",
+      "train loss:0.016452952647465043\n",
+      "train loss:0.0064311980148070635\n",
+      "train loss:0.008719404872458593\n",
+      "train loss:0.0018813200050569712\n",
+      "train loss:0.0012661380121365548\n",
+      "train loss:0.005337610002339225\n",
+      "train loss:0.003585427463647646\n",
+      "train loss:0.005464982340742948\n",
+      "train loss:0.0036800417590197574\n",
+      "train loss:0.013658298238483059\n",
+      "train loss:0.001007851403164006\n",
+      "train loss:0.0037882393422829165\n",
+      "train loss:0.007744054315897952\n",
+      "train loss:0.002446706813561284\n",
+      "train loss:0.005152821288180609\n",
+      "train loss:0.004207717162813095\n",
+      "train loss:0.004541628757494783\n",
+      "train loss:0.0026831696430072156\n",
+      "train loss:0.04529225116007257\n",
+      "train loss:0.009647496876176718\n",
+      "train loss:0.0012107653738688716\n",
+      "train loss:0.0038979083661867947\n",
+      "train loss:0.0275097471889487\n",
+      "train loss:0.0012810698977255802\n",
+      "train loss:0.0018042819558379477\n",
+      "train loss:0.0017454779071847178\n",
+      "train loss:0.004674938504878227\n",
+      "train loss:0.001309598562866691\n",
+      "train loss:0.006959468492690947\n",
+      "train loss:0.004553519470910895\n",
+      "train loss:0.003175400982700339\n",
+      "train loss:0.004352458278604343\n",
+      "train loss:0.0018650293591514684\n",
+      "train loss:0.001405364844883363\n",
+      "train loss:0.004799674579418097\n",
+      "train loss:0.007657451976596521\n",
+      "train loss:0.0016068992180850116\n",
+      "train loss:0.006353476514990165\n",
+      "train loss:0.006201903065164542\n",
+      "train loss:0.005405713623799008\n",
+      "train loss:0.0007671164058568569\n",
+      "train loss:0.0006114743337208116\n",
+      "train loss:0.00423977735686354\n",
+      "train loss:0.005881697773831995\n",
+      "train loss:0.0043698681094241024\n",
+      "train loss:0.004938915110957582\n",
+      "train loss:0.05593968634866563\n",
+      "train loss:0.001494218099118648\n",
+      "train loss:0.002771596976408812\n",
+      "train loss:0.0053168498519110315\n",
+      "train loss:0.0010808065503424532\n",
+      "train loss:0.002887677251654555\n",
+      "train loss:0.007199460806565205\n",
+      "train loss:0.0013474810754026381\n",
+      "train loss:0.0054280827225032845\n",
+      "train loss:0.03124603584712779\n",
+      "train loss:0.009269406179321768\n",
+      "train loss:0.018433917272668583\n",
+      "train loss:0.004364003383621181\n",
+      "train loss:0.006107042412959865\n",
+      "train loss:0.02197162353281057\n",
+      "train loss:0.006474086223626685\n",
+      "train loss:0.0020837867187792425\n",
+      "train loss:0.026671680544403782\n",
+      "train loss:0.005956732292704765\n",
+      "train loss:0.0005855430151112346\n",
+      "train loss:0.0006673898919602773\n",
+      "train loss:0.00579833295428112\n",
+      "train loss:0.003687521558553689\n",
+      "train loss:0.005606858311242525\n",
+      "train loss:0.00232589383527603\n",
+      "train loss:0.013460722781594603\n",
+      "train loss:0.019467645669661242\n",
+      "train loss:0.0009657394264172703\n",
+      "train loss:0.0031034017465484415\n",
+      "train loss:0.001997255687524797\n",
+      "train loss:0.02214965268312331\n",
+      "train loss:0.003094384148932179\n",
+      "train loss:0.0015443842666298592\n",
+      "train loss:0.0002443833054373029\n",
+      "train loss:0.0018143263838136378\n",
+      "train loss:0.015631036772162214\n",
+      "train loss:0.0013189482152528708\n",
+      "train loss:0.01481815315031537\n",
+      "train loss:0.008845353744519805\n",
+      "train loss:0.00462728185283378\n",
+      "train loss:0.00068871730083089\n",
+      "train loss:0.009106421029643286\n",
+      "train loss:0.005636749099059859\n",
+      "train loss:0.005284107602563009\n",
+      "train loss:0.01466704199644938\n",
+      "train loss:0.0012912868495247101\n",
+      "train loss:0.008086944015843329\n",
+      "train loss:0.0009629208369823317\n",
+      "train loss:0.0018989861102775014\n",
+      "train loss:0.004599882985562592\n",
+      "train loss:0.004630853470340065\n",
+      "train loss:0.004784097185970278\n",
+      "train loss:0.0026294830561276773\n",
+      "train loss:0.00046067092997396317\n",
+      "train loss:0.002740230322042791\n",
+      "train loss:0.016527578369135945\n",
+      "train loss:0.003836272151366965\n",
+      "train loss:0.018388471778640173\n",
+      "train loss:0.0046988538852795\n",
+      "train loss:0.003047083408877404\n",
+      "train loss:0.004263209047082225\n",
+      "train loss:0.00902242995029762\n",
+      "train loss:0.007096568260656116\n",
+      "train loss:0.020139281295632996\n",
+      "train loss:0.020486036660674247\n",
+      "train loss:0.008424060441130234\n",
+      "train loss:0.015345925943318966\n",
+      "train loss:0.0031863520704660036\n",
+      "train loss:0.039756590707627576\n",
+      "train loss:0.011359522829618779\n",
+      "train loss:0.0016451089733192639\n",
+      "train loss:0.0017939290178995357\n",
+      "train loss:0.005109220412197244\n",
+      "train loss:0.006497205341437712\n",
+      "train loss:0.002047240372497496\n",
+      "train loss:0.0005341233543427288\n",
+      "train loss:0.0029712487704214104\n",
+      "train loss:0.0008220519832915025\n",
+      "train loss:0.002575997228128108\n",
+      "train loss:0.003708478095728605\n",
+      "train loss:0.009395067537604924\n",
+      "train loss:0.0019798135248328564\n",
+      "train loss:0.0040437296485576315\n",
+      "train loss:0.0006400824478434121\n",
+      "train loss:0.0017209755835213283\n",
+      "train loss:0.003404553562498887\n",
+      "train loss:0.00967528691546122\n",
+      "train loss:0.01855429694772093\n",
+      "train loss:0.004668419876897564\n",
+      "train loss:0.007786669175781472\n",
+      "train loss:0.001936512969865132\n",
+      "train loss:0.012370449189652674\n",
+      "train loss:0.014036014693979115\n",
+      "train loss:0.028915146370910562\n",
+      "train loss:0.0015066157045339971\n",
+      "train loss:0.0017265788410152328\n",
+      "train loss:0.0012200876184709645\n",
+      "train loss:0.0011504737262906148\n",
+      "train loss:0.004138885240381396\n",
+      "train loss:0.0031572642716184702\n",
+      "train loss:0.0045150472016305095\n",
+      "train loss:0.005316804633704501\n",
+      "train loss:0.010419742618161925\n",
+      "train loss:0.009084888678481446\n",
+      "train loss:0.0032142996234478423\n",
+      "train loss:0.01302345853456228\n",
+      "train loss:0.030421429519878886\n",
+      "train loss:0.0006320825795780187\n",
+      "train loss:0.0013807668340670334\n",
+      "train loss:0.05051126946413922\n",
+      "train loss:0.0005632508354900169\n",
+      "train loss:0.0019673642587872435\n",
+      "train loss:0.005549318870990336\n",
+      "train loss:0.0030809783263114877\n",
+      "train loss:0.0024283792543110575\n",
+      "train loss:0.008220068150142135\n",
+      "train loss:0.007796677681340878\n",
+      "train loss:0.002559415561751806\n",
+      "train loss:0.027085095152595495\n",
+      "train loss:0.0034307406622066995\n",
+      "train loss:0.0036104246278462894\n",
+      "train loss:0.0034185312939593844\n",
+      "train loss:0.0017432276283604914\n",
+      "train loss:0.0011618670789699127\n",
+      "train loss:0.0033601543711467828\n",
+      "train loss:0.005149028214399423\n",
+      "train loss:0.01872521891847144\n",
+      "train loss:0.0009315826833341512\n",
+      "train loss:0.0064741841601437524\n",
+      "train loss:0.0024724127131606887\n",
+      "train loss:0.0026058663910869095\n",
+      "train loss:0.0017379622853237534\n",
+      "train loss:0.005463855056595776\n",
+      "train loss:0.005596691518155601\n",
+      "train loss:0.02115199231470499\n",
+      "train loss:0.011305702707041869\n",
+      "train loss:0.015211638682375135\n",
+      "train loss:0.005604549119141204\n",
+      "train loss:0.007238419155723588\n",
+      "train loss:0.0022260123743476877\n",
+      "train loss:0.002534090659958281\n",
+      "train loss:0.0014734203542925625\n",
+      "train loss:0.013055353719873203\n",
+      "train loss:0.0060784959529301685\n",
+      "train loss:0.011979224234194228\n",
+      "train loss:0.0005008776680418938\n",
+      "train loss:0.001761081579189952\n",
+      "train loss:0.02159847536047025\n",
+      "train loss:0.031034240550220183\n",
+      "train loss:0.002056718935998179\n",
+      "train loss:0.0030048634731285567\n",
+      "train loss:0.004018533499846466\n",
+      "train loss:0.0004722218289149653\n",
+      "train loss:0.023521607285671552\n",
+      "train loss:0.028340335401444715\n",
+      "train loss:0.00512354396507975\n",
+      "train loss:0.0012081246116970674\n",
+      "train loss:0.007090472511258866\n",
+      "train loss:0.002993237055117547\n",
+      "train loss:0.007243078914977255\n",
+      "train loss:0.003344522463923466\n",
+      "train loss:0.0007106535183807822\n",
+      "train loss:0.0007164081870774143\n",
+      "train loss:0.0012213114459520922\n",
+      "train loss:0.06857259278937236\n",
+      "train loss:0.002931536500413545\n",
+      "train loss:0.0050468787531352255\n",
+      "train loss:0.0017593210860889614\n",
+      "train loss:0.003560564577366736\n",
+      "train loss:0.0005057291046460092\n",
+      "train loss:0.004214673092185986\n",
+      "train loss:0.002034803397956064\n",
+      "train loss:0.004277443689927452\n",
+      "train loss:0.00514903160122575\n",
+      "train loss:0.01688510010542266\n",
+      "train loss:0.002122765171078783\n",
+      "train loss:0.001968785881633218\n",
+      "train loss:0.004869563315143399\n",
+      "train loss:0.002402785203590315\n",
+      "train loss:0.0034932736843803097\n",
+      "train loss:0.0010438315015232672\n",
+      "train loss:0.0013045781149015763\n",
+      "train loss:0.0034878159785441424\n",
+      "train loss:0.011753035768574296\n",
+      "train loss:0.004910840306639627\n",
+      "train loss:0.006707203544908785\n",
+      "train loss:0.0019541366169274973\n",
+      "train loss:0.0021396235646928396\n",
+      "train loss:0.002219846552573296\n",
+      "train loss:0.007698390231681566\n",
+      "train loss:0.002203845466457724\n",
+      "train loss:0.006141778515834048\n",
+      "train loss:0.005532210886529931\n",
+      "train loss:0.0012542018531291957\n",
+      "train loss:0.0022282099537523055\n",
+      "train loss:0.022535302539686754\n",
+      "train loss:0.0009463850789706245\n",
+      "train loss:0.0066679947636763435\n",
+      "train loss:0.0002576172125695573\n",
+      "train loss:0.01645135920137427\n",
+      "train loss:0.0018018391330213604\n",
+      "train loss:0.004382186941664965\n",
+      "train loss:0.001744404860762974\n",
+      "train loss:0.0019663498688237556\n",
+      "train loss:0.011592980792216117\n",
+      "train loss:0.009086693421076496\n",
+      "train loss:0.0034423216743163275\n",
+      "train loss:0.002140924999129697\n",
+      "train loss:0.0017724046208369892\n",
+      "train loss:0.01542639856384149\n",
+      "train loss:0.015000562159127113\n",
+      "train loss:0.0024531965453031767\n",
+      "train loss:0.005620999112970138\n",
+      "train loss:0.0027051783261324604\n",
+      "train loss:0.008193066519604504\n",
+      "train loss:0.033303177200467764\n",
+      "train loss:0.0019301355283259417\n",
+      "train loss:0.008570544938008382\n",
+      "train loss:0.0016933871051129246\n",
+      "train loss:0.00571853635170145\n",
+      "train loss:0.007457142679365796\n",
+      "train loss:0.002283539969392863\n",
+      "train loss:0.000502254858005948\n",
+      "train loss:0.00448337645611542\n",
+      "train loss:0.0016344315129658011\n",
+      "train loss:0.004057337219080597\n",
+      "train loss:0.008571570508457861\n",
+      "train loss:0.004360808978755069\n",
+      "train loss:0.010315799078391865\n",
+      "train loss:0.010052960953152983\n",
+      "train loss:0.0005258944530297333\n",
+      "train loss:0.0012863372732286238\n",
+      "train loss:0.012617057778269864\n",
+      "train loss:0.005759501510382749\n",
+      "train loss:0.006067058610917969\n",
+      "train loss:0.00030275936060001604\n",
+      "train loss:0.001871223744490972\n",
+      "train loss:0.01391872910832837\n",
+      "train loss:0.002001216743124955\n",
+      "train loss:0.018439735448744686\n",
+      "train loss:0.003486259844172216\n",
+      "train loss:0.004417167937594703\n",
+      "train loss:0.011788320904798704\n",
+      "train loss:0.0067117458455261516\n",
+      "train loss:0.00234715401997214\n",
+      "train loss:0.02819058171418581\n",
+      "train loss:0.00587817783587941\n",
+      "train loss:0.01755761228696425\n",
+      "train loss:0.0039047837005430587\n",
+      "train loss:0.001907583812000747\n",
+      "train loss:0.027494574258075714\n",
+      "train loss:0.003298522356168853\n",
+      "train loss:0.008109134810672736\n",
+      "train loss:0.003069528520124767\n",
+      "train loss:0.023530056778730723\n",
+      "train loss:0.002793855771785195\n",
+      "train loss:0.005190388209196679\n",
+      "train loss:0.0159163277201782\n",
+      "train loss:0.006894944554746678\n",
+      "train loss:0.00484638511290693\n",
+      "train loss:0.0029346594168625493\n",
+      "train loss:0.0035716929204164588\n",
+      "train loss:0.007840391628773041\n",
+      "train loss:0.017462753520586634\n",
+      "train loss:0.0039681476494076075\n",
+      "train loss:0.007602483302035061\n",
+      "train loss:0.010483623519570374\n",
+      "train loss:0.008414050204746072\n",
+      "train loss:0.003353078657267244\n",
+      "train loss:0.00874888558710851\n",
+      "train loss:0.005722623226888764\n",
+      "train loss:0.0008869756399206468\n",
+      "train loss:0.004243580619460445\n",
+      "train loss:0.005687148872526684\n",
+      "=== epoch:12, train acc:0.995, test acc:0.988 ===\n",
+      "train loss:0.004094429892582126\n",
+      "train loss:0.005261023900179001\n",
+      "train loss:0.002442075123339899\n",
+      "train loss:0.002570397318945974\n",
+      "train loss:0.0016753265427764251\n",
+      "train loss:0.008654657467746147\n",
+      "train loss:0.004972109331414275\n",
+      "train loss:0.0021578788181848864\n",
+      "train loss:0.0009335678904998342\n",
+      "train loss:0.007278414584871868\n",
+      "train loss:0.002485439199660215\n",
+      "train loss:0.035996788986532906\n",
+      "train loss:0.0005887988008215807\n",
+      "train loss:0.0031090904325107834\n",
+      "train loss:0.0007611435934915932\n",
+      "train loss:0.007117786936205045\n",
+      "train loss:0.002408970880265077\n",
+      "train loss:0.010357466466219276\n",
+      "train loss:0.001259026096399502\n",
+      "train loss:0.005207031193806337\n",
+      "train loss:0.0017905141407115067\n",
+      "train loss:0.003831617481977786\n",
+      "train loss:0.002075452630107041\n",
+      "train loss:0.006501533337721585\n",
+      "train loss:0.0008532695342503204\n",
+      "train loss:0.0047250897981212\n",
+      "train loss:0.006883958117081258\n",
+      "train loss:0.0033211261330697685\n",
+      "train loss:0.00906949382295159\n",
+      "train loss:0.0025203984614259566\n",
+      "train loss:0.020966887693823107\n",
+      "train loss:0.0006485888399137528\n",
+      "train loss:0.005505968540776095\n",
+      "train loss:0.004852117539490966\n",
+      "train loss:6.818113774147882e-05\n",
+      "train loss:0.016944513664241555\n",
+      "train loss:0.0012712620682178788\n",
+      "train loss:0.002204030683067075\n",
+      "train loss:0.004589155526205788\n",
+      "train loss:0.002219377527500937\n",
+      "train loss:0.0022411722445911876\n",
+      "train loss:0.01007179195720815\n",
+      "train loss:0.0005284420600613869\n",
+      "train loss:0.0005104469837465504\n",
+      "train loss:0.01199541915922653\n",
+      "train loss:0.002924604124938354\n",
+      "train loss:0.005542961397510816\n",
+      "train loss:0.004312080483593168\n",
+      "train loss:0.0038271647514522495\n",
+      "train loss:0.005530810048353581\n",
+      "train loss:0.013520442896102219\n",
+      "train loss:0.002477671625878594\n",
+      "train loss:0.00031266044674827664\n",
+      "train loss:0.0015117100371385141\n",
+      "train loss:0.0003327165433079442\n",
+      "train loss:0.0189988510920475\n",
+      "train loss:0.0039423573981734895\n",
+      "train loss:0.002377114519963809\n",
+      "train loss:0.0008095288076929991\n",
+      "train loss:0.007327408119086363\n",
+      "train loss:0.007076981661788998\n",
+      "train loss:0.003768934859405198\n",
+      "train loss:0.001886211543246276\n",
+      "train loss:0.01426752398106976\n",
+      "train loss:0.0004467375864682516\n",
+      "train loss:0.005884243416296528\n",
+      "train loss:0.0031085312645764444\n",
+      "train loss:0.0004051028681619269\n",
+      "train loss:0.00294795920436732\n",
+      "train loss:0.011537916156927328\n",
+      "train loss:0.006514410481750343\n",
+      "train loss:0.003551495926155399\n",
+      "train loss:0.0126742156713549\n",
+      "train loss:0.002246212540816481\n",
+      "train loss:0.0007094130580226577\n",
+      "train loss:0.005737558483530429\n",
+      "train loss:0.0021653849267061504\n",
+      "train loss:0.001914917778447058\n",
+      "train loss:0.018131321585331598\n",
+      "train loss:0.0034972974986618084\n",
+      "train loss:0.0014174518734359585\n",
+      "train loss:0.005709002656854071\n",
+      "train loss:0.02161328610838316\n",
+      "train loss:0.002595211095367918\n",
+      "train loss:0.01704159745833076\n",
+      "train loss:0.008540488974347971\n",
+      "train loss:0.003783914669046126\n",
+      "train loss:0.0443610049439096\n",
+      "train loss:0.03143523308988851\n",
+      "train loss:0.0004413802931677694\n",
+      "train loss:0.031456378521494105\n",
+      "train loss:0.0026139296406128077\n",
+      "train loss:0.004014788133690297\n",
+      "train loss:0.0018750742967454104\n",
+      "train loss:0.0061306779165481206\n",
+      "train loss:0.011979563814763465\n",
+      "train loss:0.0215053227333843\n",
+      "train loss:0.023897939469313224\n",
+      "train loss:0.029011375449119154\n",
+      "train loss:0.019949480245319114\n",
+      "train loss:0.004336373469016014\n",
+      "train loss:0.0074270410678799435\n",
+      "train loss:0.00027591892888389883\n",
+      "train loss:0.011153117063309766\n",
+      "train loss:0.002750589491995596\n",
+      "train loss:0.0021421411139703676\n",
+      "train loss:0.005639655256625705\n",
+      "train loss:0.004951664790846788\n",
+      "train loss:0.0019385218130225986\n",
+      "train loss:0.010427194838391772\n",
+      "train loss:0.010477766081575895\n",
+      "train loss:0.003716811775118586\n",
+      "train loss:0.0018570063228076042\n",
+      "train loss:0.015714665333348685\n",
+      "train loss:0.015713720211733822\n",
+      "train loss:0.0010550290436682412\n",
+      "train loss:0.013321647162836734\n",
+      "train loss:0.003415331580839552\n",
+      "train loss:0.0016379931342395604\n",
+      "train loss:0.00859815393362425\n",
+      "train loss:0.0011813358810563229\n",
+      "train loss:0.0018631119026836665\n",
+      "train loss:0.003495272176030166\n",
+      "train loss:0.007466766859292273\n",
+      "train loss:0.013595898014310685\n",
+      "train loss:0.0012706765058522302\n",
+      "train loss:0.0019297641771617341\n",
+      "train loss:0.0020352340052596936\n",
+      "train loss:0.0126055146845348\n",
+      "train loss:0.004996199664738648\n",
+      "train loss:0.037511669973296434\n",
+      "train loss:0.0016947680045532936\n",
+      "train loss:0.0008286578988047913\n",
+      "train loss:0.0013005255152212308\n",
+      "train loss:0.0027003157170060977\n",
+      "train loss:0.0065732993585604325\n",
+      "train loss:0.009131113626360477\n",
+      "train loss:0.0057834182928424285\n",
+      "train loss:0.007771730783929681\n",
+      "train loss:0.004183976518678645\n",
+      "train loss:0.006399124074382644\n",
+      "train loss:0.002677650067686255\n",
+      "train loss:0.0014211201696027204\n",
+      "train loss:0.0030117448297853062\n",
+      "train loss:0.0008756795326712762\n",
+      "train loss:0.007903176613367107\n",
+      "train loss:0.0034654944526455837\n",
+      "train loss:0.0027782454597969674\n",
+      "train loss:0.0029481519916215175\n",
+      "train loss:0.00023473191517917342\n",
+      "train loss:0.006824198873002016\n",
+      "train loss:0.0045450964262967915\n",
+      "train loss:0.010283850159509248\n",
+      "train loss:0.0006769874284677244\n",
+      "train loss:0.0014835011544019052\n",
+      "train loss:0.0030690860687156416\n",
+      "train loss:0.02271488742404836\n",
+      "train loss:0.005637393514371191\n",
+      "train loss:0.0005838671361768081\n",
+      "train loss:0.0003487897730195141\n",
+      "train loss:0.018585123521203156\n",
+      "train loss:0.003346903412718141\n",
+      "train loss:0.002911389674950056\n",
+      "train loss:0.000963612832122165\n",
+      "train loss:0.023947332835812584\n",
+      "train loss:0.002803712858735942\n",
+      "train loss:0.003544586611446233\n",
+      "train loss:0.015337998594742355\n",
+      "train loss:0.003451746747652927\n",
+      "train loss:0.0043128354253818115\n",
+      "train loss:0.0014754606813472421\n",
+      "train loss:0.006083511399781744\n",
+      "train loss:0.007970218876408612\n",
+      "train loss:0.002482309719050997\n",
+      "train loss:0.004393027226990098\n",
+      "train loss:0.0030432932494030827\n",
+      "train loss:0.0015661616873127193\n",
+      "train loss:0.001925161832131529\n",
+      "train loss:0.0197381892637316\n",
+      "train loss:0.0017221843735364664\n",
+      "train loss:0.005712573039540945\n",
+      "train loss:0.003181010309056696\n",
+      "train loss:0.011573189032453485\n",
+      "train loss:0.0006435480546380368\n",
+      "train loss:0.0011290706335443748\n",
+      "train loss:0.007185777061343549\n",
+      "train loss:0.007455754978234337\n",
+      "train loss:0.001342667313368562\n",
+      "train loss:0.0024253368665405634\n",
+      "train loss:0.006604448414682826\n",
+      "train loss:0.0012370502415029528\n",
+      "train loss:0.009856602929702014\n",
+      "train loss:0.0001873114065085817\n",
+      "train loss:0.008372479428338847\n",
+      "train loss:0.0008254556111387645\n",
+      "train loss:0.004107270454112264\n",
+      "train loss:0.01609700706610278\n",
+      "train loss:0.001522228826513818\n",
+      "train loss:0.001901458123945687\n",
+      "train loss:0.0030296678864267608\n",
+      "train loss:0.0022892014675150833\n",
+      "train loss:0.0006323492752544485\n",
+      "train loss:0.01250070688514484\n",
+      "train loss:0.008682442341959746\n",
+      "train loss:0.004966913936866093\n",
+      "train loss:0.003813809211533217\n",
+      "train loss:0.0015014831258485673\n",
+      "train loss:0.008208041702731608\n",
+      "train loss:0.007912039808507819\n",
+      "train loss:0.0002787897297504664\n",
+      "train loss:0.006643033234316395\n",
+      "train loss:0.0019248327481250133\n",
+      "train loss:0.006095837600687498\n",
+      "train loss:0.019249732187045055\n",
+      "train loss:0.0016800659756624342\n",
+      "train loss:0.007327603663875567\n",
+      "train loss:0.0056317843265237645\n",
+      "train loss:0.028382859158018494\n",
+      "train loss:0.002010869708424784\n",
+      "train loss:0.0034433506513790917\n",
+      "train loss:0.003730155783581758\n",
+      "train loss:0.005807646494719416\n",
+      "train loss:0.03871629050959134\n",
+      "train loss:0.0067885548946422765\n",
+      "train loss:0.0015329760372672016\n",
+      "train loss:0.0012532653978851258\n",
+      "train loss:0.0037465758241997053\n",
+      "train loss:0.004540699391442121\n",
+      "train loss:0.0031359254768144936\n",
+      "train loss:0.004908984033022808\n",
+      "train loss:0.0011516710488026147\n",
+      "train loss:0.0020194844421794985\n",
+      "train loss:0.0017999333445597374\n",
+      "train loss:0.0036371911713064943\n",
+      "train loss:0.00239212797681574\n",
+      "train loss:0.002254080776409289\n",
+      "train loss:0.004293534334719512\n",
+      "train loss:0.003277762599446064\n",
+      "train loss:0.000647873217879364\n",
+      "train loss:0.007958955215066923\n",
+      "train loss:0.0023031873005442704\n",
+      "train loss:0.0025801158214393692\n",
+      "train loss:0.005263930817486173\n",
+      "train loss:0.016052723924724566\n",
+      "train loss:0.0011053312965768447\n",
+      "train loss:0.0029918025095532806\n",
+      "train loss:0.0016269887916132968\n",
+      "train loss:0.0017402289807634112\n",
+      "train loss:0.0008997077382256072\n",
+      "train loss:0.0018838675878568194\n",
+      "train loss:0.004001273665799431\n",
+      "train loss:0.0056408482378319325\n",
+      "train loss:0.006134037674732977\n",
+      "train loss:0.0029595287563134705\n",
+      "train loss:0.013402604317151257\n",
+      "train loss:0.004606525462564992\n",
+      "train loss:0.022693852626915246\n",
+      "train loss:0.0005504854254887462\n",
+      "train loss:0.005996985856697226\n",
+      "train loss:0.00256974510477006\n",
+      "train loss:0.000987490624315418\n",
+      "train loss:0.0055626909244653\n",
+      "train loss:0.0025216158179410366\n",
+      "train loss:0.0025660825848031073\n",
+      "train loss:0.002866505411405679\n",
+      "train loss:0.009708717249940477\n",
+      "train loss:0.0020439652670832626\n",
+      "train loss:0.001033406149940218\n",
+      "train loss:0.0028239749873169334\n",
+      "train loss:0.006084337083605341\n",
+      "train loss:0.001286421044475476\n",
+      "train loss:0.000853860353409048\n",
+      "train loss:0.005481936896972648\n",
+      "train loss:0.0015953243911413766\n",
+      "train loss:0.008138132926437622\n",
+      "train loss:0.0032099507033182713\n",
+      "train loss:0.0017957109576445763\n",
+      "train loss:0.0044677222170941815\n",
+      "train loss:0.004928356172966479\n",
+      "train loss:0.001763549747297903\n",
+      "train loss:0.011859096881819797\n",
+      "train loss:0.006023013103380961\n",
+      "train loss:0.0017793859767953268\n",
+      "train loss:0.0020738636492460493\n",
+      "train loss:0.0013649035197595464\n",
+      "train loss:0.012310112540614225\n",
+      "train loss:0.0007401425653863886\n",
+      "train loss:0.002719788559580498\n",
+      "train loss:0.002013909483632039\n",
+      "train loss:0.004238358077023333\n",
+      "train loss:0.004213545916998429\n",
+      "train loss:0.006420119426494143\n",
+      "train loss:0.004694929956971996\n",
+      "train loss:0.028011478292341324\n",
+      "train loss:0.0002628164238628929\n",
+      "train loss:0.00011443104287655019\n",
+      "train loss:0.000987382974815261\n",
+      "train loss:0.003798300517627326\n",
+      "train loss:0.0014567362049434185\n",
+      "train loss:0.0025790861936244424\n",
+      "train loss:0.0005815291730886966\n",
+      "train loss:0.0024484268468585703\n",
+      "train loss:0.0016802969420941422\n",
+      "train loss:0.000639317679905143\n",
+      "train loss:0.004969210824840858\n",
+      "train loss:0.00023802622243479212\n",
+      "train loss:0.027042804354513582\n",
+      "train loss:0.0027181127843891144\n",
+      "train loss:0.0013631095327825016\n",
+      "train loss:0.0010226887023397282\n",
+      "train loss:0.0026883851907170387\n",
+      "train loss:0.0025035654371895516\n",
+      "train loss:0.05285079767757364\n",
+      "train loss:0.002061810094237324\n",
+      "train loss:0.0027766206982007074\n",
+      "train loss:0.05407389129518723\n",
+      "train loss:0.0010760371677665775\n",
+      "train loss:0.005928762686892528\n",
+      "train loss:0.0007384230327994526\n",
+      "train loss:0.008335913865086142\n",
+      "train loss:0.0047510873861969\n",
+      "train loss:0.0012316453042330309\n",
+      "train loss:0.00853788716890394\n",
+      "train loss:0.006716602548600324\n",
+      "train loss:0.004977106544944775\n",
+      "train loss:0.007788780420050421\n",
+      "train loss:0.02266777419571417\n",
+      "train loss:0.0005695919861709947\n",
+      "train loss:0.013100543671633632\n",
+      "train loss:0.007674537312930979\n",
+      "train loss:0.011584087863735533\n",
+      "train loss:0.0023884132218787353\n",
+      "train loss:0.0007805224434444267\n",
+      "train loss:0.0022953004679366145\n",
+      "train loss:0.004300569281914284\n",
+      "train loss:0.0339003488574762\n",
+      "train loss:0.003518524767235532\n",
+      "train loss:0.006373226164552453\n",
+      "train loss:0.001491204097497734\n",
+      "train loss:0.0026293288984931113\n",
+      "train loss:0.002105115815833644\n",
+      "train loss:0.0010453090358878542\n",
+      "train loss:0.001576360523723156\n",
+      "train loss:0.0001615365227681291\n",
+      "train loss:0.003945598440181246\n",
+      "train loss:0.009738841921126371\n",
+      "train loss:0.0015600068332627052\n",
+      "train loss:0.0016878615088376553\n",
+      "train loss:0.004536409194045265\n",
+      "train loss:0.01187180964348855\n",
+      "train loss:0.03606811288438572\n",
+      "train loss:0.0068486076468404974\n",
+      "train loss:0.0020829780374755124\n",
+      "train loss:0.0016547960598501834\n",
+      "train loss:0.005600173553942629\n",
+      "train loss:0.013852534989523234\n",
+      "train loss:0.013548995687352168\n",
+      "train loss:0.0028062850488407187\n",
+      "train loss:0.017505938636219506\n",
+      "train loss:0.0018407026722212984\n",
+      "train loss:0.013831835898189726\n",
+      "train loss:0.0034327848573748976\n",
+      "train loss:0.003596372380048969\n",
+      "train loss:0.005479162158884852\n",
+      "train loss:0.0053091738114356265\n",
+      "train loss:0.003078371348702978\n",
+      "train loss:0.006415015289789527\n",
+      "train loss:0.005594335832719983\n",
+      "train loss:0.0012699211677269584\n",
+      "train loss:0.004387700306291921\n",
+      "train loss:0.0010627723147289033\n",
+      "train loss:0.018743939907629646\n",
+      "train loss:0.004906369719739015\n",
+      "train loss:0.006269109186818774\n",
+      "train loss:0.005695515585201152\n",
+      "train loss:0.00032792853524655813\n",
+      "train loss:0.0013033257391169947\n",
+      "train loss:0.003829966593879124\n",
+      "train loss:0.010382397235927694\n",
+      "train loss:0.0010918436008721217\n",
+      "train loss:0.0018282549686691552\n",
+      "train loss:0.0010470275061909783\n",
+      "train loss:0.0106067260442727\n",
+      "train loss:0.011683687026345453\n",
+      "train loss:0.006150170996352805\n",
+      "train loss:0.006059338360985832\n",
+      "train loss:0.0017433443579596124\n",
+      "train loss:0.0002048332761208539\n",
+      "train loss:0.005120740819584852\n",
+      "train loss:0.006318008058379863\n",
+      "train loss:0.018903914312952713\n",
+      "train loss:0.0002855242150560647\n",
+      "train loss:0.0037018638682376026\n",
+      "train loss:0.0008926635248822024\n",
+      "train loss:0.008068834119723452\n",
+      "train loss:0.011898369984415333\n",
+      "train loss:0.00786797821955785\n",
+      "train loss:0.002163099340807175\n",
+      "train loss:0.002818552762811723\n",
+      "train loss:0.0015065487115952679\n",
+      "train loss:0.00037886337171555846\n",
+      "train loss:0.0013136532619781873\n",
+      "train loss:0.007600737891300277\n",
+      "train loss:0.011941565300335492\n",
+      "train loss:0.002170905864514182\n",
+      "train loss:0.01134790592905465\n",
+      "train loss:0.0012065482005252023\n",
+      "train loss:0.006109378016160617\n",
+      "train loss:0.003023475548111945\n",
+      "train loss:0.0023241141665401245\n",
+      "train loss:0.0008365702394313982\n",
+      "train loss:0.005946456235549738\n",
+      "train loss:0.004388980321368512\n",
+      "train loss:0.005767150449392555\n",
+      "train loss:0.00033341638981024285\n",
+      "train loss:0.006363929407458609\n",
+      "train loss:0.001523275215380275\n",
+      "train loss:0.012291981262199807\n",
+      "train loss:0.002421364386741466\n",
+      "train loss:0.0025321457758580197\n",
+      "train loss:0.011138719771991578\n",
+      "train loss:0.004948356657985535\n",
+      "train loss:0.003423727700619913\n",
+      "train loss:0.001631827045780375\n",
+      "train loss:0.00011341114031615918\n",
+      "train loss:0.013756115767149281\n",
+      "train loss:0.007776301715222148\n",
+      "train loss:0.002127706092996907\n",
+      "train loss:0.004086796431485622\n",
+      "train loss:0.0027904333834956087\n",
+      "train loss:0.018066432937042357\n",
+      "train loss:0.008138083967876597\n",
+      "train loss:0.0013332814395681134\n",
+      "train loss:0.002515947565672989\n",
+      "train loss:0.00679188511208345\n",
+      "train loss:0.005954146029960372\n",
+      "train loss:0.004904215279604245\n",
+      "train loss:0.008410986636990293\n",
+      "train loss:0.0022660349819904222\n",
+      "train loss:0.00039129042460613815\n",
+      "train loss:0.01966595422628345\n",
+      "train loss:0.001041587052239422\n",
+      "train loss:0.0027159613856128717\n",
+      "train loss:0.03549218008472292\n",
+      "train loss:0.0002945175394593685\n",
+      "train loss:0.003985874730598481\n",
+      "train loss:0.006976536790590262\n",
+      "train loss:0.002478313771207038\n",
+      "train loss:0.002659726725917822\n",
+      "train loss:0.0012912812681594753\n",
+      "train loss:0.005109876612525168\n",
+      "train loss:0.00934457006099987\n",
+      "train loss:0.0015527581420736248\n",
+      "train loss:0.005344527911591731\n",
+      "train loss:0.0041971740402040325\n",
+      "train loss:0.005191528302220754\n",
+      "train loss:0.001079205630087957\n",
+      "train loss:0.0016884564870287694\n",
+      "train loss:0.0021324911888739046\n",
+      "train loss:0.01428919898639252\n",
+      "train loss:0.00027663580294374666\n",
+      "train loss:0.005867896946383212\n",
+      "train loss:0.0024869935658353463\n",
+      "train loss:0.0015901396590000602\n",
+      "train loss:0.0038523664807850093\n",
+      "train loss:0.0010387755292129272\n",
+      "train loss:0.008790648438524739\n",
+      "train loss:0.001464357088488329\n",
+      "train loss:0.0006874227904424616\n",
+      "train loss:0.00114426823160731\n",
+      "train loss:0.0026579591943164255\n",
+      "train loss:0.008739827725796573\n",
+      "train loss:0.0052771289993086565\n",
+      "train loss:0.0025780860778286587\n",
+      "train loss:0.0005074560097408915\n",
+      "train loss:0.0035924646699626245\n",
+      "train loss:0.002046532334763607\n",
+      "train loss:0.01091222311493465\n",
+      "train loss:0.007425848078356793\n",
+      "train loss:0.0017138494858984497\n",
+      "train loss:0.01454771773051254\n",
+      "train loss:0.0014256508641386785\n",
+      "train loss:0.00215340845979578\n",
+      "train loss:0.00017953437491480625\n",
+      "train loss:0.008140256600469832\n",
+      "train loss:0.00655145287040255\n",
+      "train loss:0.004066356390528975\n",
+      "train loss:0.001352075255451126\n",
+      "train loss:0.001670135477544643\n",
+      "train loss:0.0034908546197978613\n",
+      "train loss:0.011749427558782298\n",
+      "train loss:0.004138891578753358\n",
+      "train loss:0.0014819007956448278\n",
+      "train loss:0.007165965869502637\n",
+      "train loss:0.003917694396665725\n",
+      "train loss:0.000352560268100198\n",
+      "train loss:0.006158965745964545\n",
+      "train loss:0.004299918410896758\n",
+      "train loss:0.0003098513095227214\n",
+      "train loss:0.010874769951073155\n",
+      "train loss:0.008660973402609745\n",
+      "train loss:0.002148437746404626\n",
+      "train loss:0.01468025262540083\n",
+      "train loss:0.0018698172279477498\n",
+      "train loss:0.01053450258166777\n",
+      "train loss:0.00040466531367389937\n",
+      "train loss:0.004244390113344554\n",
+      "train loss:0.013966640899150612\n",
+      "train loss:0.002390705819273206\n",
+      "train loss:0.00718085918413309\n",
+      "train loss:0.0011016521981142403\n",
+      "train loss:0.0026767080128633747\n",
+      "train loss:0.0008117096498206577\n",
+      "train loss:0.004270054076115087\n",
+      "train loss:0.0012784886054911645\n",
+      "train loss:0.012400696163660845\n",
+      "train loss:0.0022810612187913245\n",
+      "train loss:0.02251424479546145\n",
+      "train loss:0.0012797948267965751\n",
+      "train loss:0.000586629328042763\n",
+      "train loss:0.0010693814222483022\n",
+      "train loss:0.003030547151569848\n",
+      "train loss:0.0007024844884213796\n",
+      "train loss:0.008395365365368397\n",
+      "train loss:0.0015625615124113365\n",
+      "train loss:0.00039072827281221267\n",
+      "train loss:0.02270506441935419\n",
+      "train loss:0.0019564328406903174\n",
+      "train loss:0.010138725249243353\n",
+      "train loss:0.0017505481793671446\n",
+      "train loss:0.0014568283042530677\n",
+      "train loss:0.0018321635255226637\n",
+      "train loss:0.0012163148368121105\n",
+      "train loss:0.014612052673312857\n",
+      "train loss:0.0015061801333467798\n",
+      "train loss:0.0019507670384796738\n",
+      "train loss:0.007277066558128904\n",
+      "train loss:0.0012766566327453044\n",
+      "train loss:0.0029082961280741367\n",
+      "train loss:0.006984997957769928\n",
+      "train loss:0.0024557685254453197\n",
+      "train loss:0.0013475673949574548\n",
+      "train loss:0.0020853171633411875\n",
+      "train loss:0.0022257139684603356\n",
+      "train loss:0.0028931876834721038\n",
+      "train loss:0.0011515621512384457\n",
+      "train loss:0.0007297760193586417\n",
+      "train loss:0.03705749742985768\n",
+      "train loss:0.004981375724416929\n",
+      "train loss:0.005198425771267189\n",
+      "train loss:0.0015661416503779937\n",
+      "train loss:0.001306352236311534\n",
+      "train loss:0.005705302589186708\n",
+      "train loss:0.0027016080787002116\n",
+      "train loss:0.003548514121322543\n",
+      "train loss:0.001015124596165243\n",
+      "train loss:0.00601002175225526\n",
+      "train loss:0.006573766912721588\n",
+      "train loss:0.005153073594761292\n",
+      "train loss:0.0037416221943822954\n",
+      "train loss:0.001802750016133855\n",
+      "train loss:0.022689209089221018\n",
+      "train loss:0.0036025555536511266\n",
+      "train loss:0.00895222149735643\n",
+      "train loss:0.0013900448581270033\n",
+      "train loss:0.031162542041477073\n",
+      "train loss:0.00534014330371717\n",
+      "train loss:0.0007257308403450262\n",
+      "train loss:0.0018361194646969656\n",
+      "train loss:0.0025560129626985345\n",
+      "train loss:0.012664963580119933\n",
+      "train loss:0.002264703327379244\n",
+      "train loss:0.0012845116210233759\n",
+      "train loss:0.0008115836144130206\n",
+      "train loss:0.00188425789684086\n",
+      "train loss:0.014894305304221114\n",
+      "train loss:0.004011545112277644\n",
+      "train loss:0.006692884737827951\n",
+      "train loss:0.005698281530288978\n",
+      "train loss:0.009905791129393082\n",
+      "train loss:0.002941893980186788\n",
+      "train loss:0.0014480538345863281\n",
+      "train loss:0.003193118010152324\n",
+      "train loss:0.005387307573357491\n",
+      "train loss:0.00552009420571138\n",
+      "train loss:0.000889586446859261\n",
+      "train loss:0.0011965430820883225\n",
+      "train loss:0.004970566756863797\n",
+      "train loss:0.0007967258907968532\n",
+      "train loss:0.002120958511880381\n",
+      "train loss:0.002410246036272641\n",
+      "train loss:0.02080263756806534\n",
+      "train loss:0.01134499450447776\n",
+      "train loss:0.0005384351623773685\n",
+      "train loss:0.00103703871758848\n",
+      "train loss:0.0033492889029920958\n",
+      "train loss:0.0017553780008795511\n",
+      "train loss:0.00010776168059167683\n",
+      "train loss:0.015680023159968724\n",
+      "train loss:0.0072647205929702915\n",
+      "=== epoch:13, train acc:0.997, test acc:0.984 ===\n",
+      "train loss:0.000980668072447895\n",
+      "train loss:0.0015432538609776003\n",
+      "train loss:0.0005201830777332154\n",
+      "train loss:0.0031476187627350275\n",
+      "train loss:0.0020622109968045517\n",
+      "train loss:0.0027788306912345326\n",
+      "train loss:0.008613347397108184\n",
+      "train loss:0.00676742989606274\n",
+      "train loss:0.0008554894835635363\n",
+      "train loss:0.0009245723051217335\n",
+      "train loss:0.010886145742768739\n",
+      "train loss:0.0017077055367909764\n",
+      "train loss:0.007292974229416069\n",
+      "train loss:0.0016168421686295444\n",
+      "train loss:0.0028553418689121523\n",
+      "train loss:0.004828701333753466\n",
+      "train loss:0.0008761066644170507\n",
+      "train loss:0.012611733723350775\n",
+      "train loss:0.005089762364580273\n",
+      "train loss:0.006268022414885586\n",
+      "train loss:0.009905485327366716\n",
+      "train loss:0.002111631665113651\n",
+      "train loss:0.022242743611348627\n",
+      "train loss:0.002420634980242378\n",
+      "train loss:0.0011846956503228153\n",
+      "train loss:0.0006493139345946934\n",
+      "train loss:0.009972796792547286\n",
+      "train loss:0.0011854193875466261\n",
+      "train loss:0.002954451505769779\n",
+      "train loss:0.0048282562624363005\n",
+      "train loss:0.001755654075197868\n",
+      "train loss:0.0009045593875364645\n",
+      "train loss:0.006910630487456672\n",
+      "train loss:0.0029022913786163618\n",
+      "train loss:0.004585298771518221\n",
+      "train loss:0.002867807754047888\n",
+      "train loss:0.002154982693198649\n",
+      "train loss:0.00045410702859921976\n",
+      "train loss:0.010724555612526132\n",
+      "train loss:0.015160660543346433\n",
+      "train loss:0.0007714833634092051\n",
+      "train loss:0.010137643338900834\n",
+      "train loss:0.007214950436486362\n",
+      "train loss:0.003443215794175308\n",
+      "train loss:0.002223031543877282\n",
+      "train loss:0.0070222994000706095\n",
+      "train loss:0.0036546818906493116\n",
+      "train loss:0.002474263783022612\n",
+      "train loss:0.0016619902457291802\n",
+      "train loss:0.007857027864025696\n",
+      "train loss:0.0029534635176238044\n",
+      "train loss:0.0015324484832123489\n",
+      "train loss:0.0038443894395616468\n",
+      "train loss:0.00044183054681590206\n",
+      "train loss:0.0006687086722789275\n",
+      "train loss:0.0008774800200070951\n",
+      "train loss:0.005215848500309316\n",
+      "train loss:0.004707991388276897\n",
+      "train loss:0.010738765780879306\n",
+      "train loss:0.000980650868293323\n",
+      "train loss:0.0021518997492228037\n",
+      "train loss:0.002707003698174949\n",
+      "train loss:0.0034471098035178103\n",
+      "train loss:0.00035543543003473094\n",
+      "train loss:0.000783164987859141\n",
+      "train loss:0.002324516077600614\n",
+      "train loss:0.005912627870150892\n",
+      "train loss:0.004246714248770638\n",
+      "train loss:0.0006132466731658942\n",
+      "train loss:0.007353858446205177\n",
+      "train loss:0.0007497978000144598\n",
+      "train loss:0.001055765259743001\n",
+      "train loss:0.001251542771701336\n",
+      "train loss:0.002042254153623084\n",
+      "train loss:0.0021752227256362547\n",
+      "train loss:0.0035226701847818763\n",
+      "train loss:0.004034823801458041\n",
+      "train loss:0.0012257940128841006\n",
+      "train loss:0.005232432833527712\n",
+      "train loss:0.002388923235501122\n",
+      "train loss:0.002371958390665963\n",
+      "train loss:0.0064591210168298725\n",
+      "train loss:0.0005730102061754457\n",
+      "train loss:0.0041733246604465195\n",
+      "train loss:0.0016887221896860483\n",
+      "train loss:0.004126403785526469\n",
+      "train loss:0.00039158699948507465\n",
+      "train loss:8.472231218806764e-05\n",
+      "train loss:0.0018725037726049617\n",
+      "train loss:0.002268088593969249\n",
+      "train loss:0.0015829995236284974\n",
+      "train loss:0.0009029996367551136\n",
+      "train loss:0.00048261050683183764\n",
+      "train loss:0.0008841144966251853\n",
+      "train loss:0.009285307753569453\n",
+      "train loss:0.0026816196991960163\n",
+      "train loss:0.00435900804337183\n",
+      "train loss:0.0005350649199246127\n",
+      "train loss:0.0009906978174369111\n",
+      "train loss:0.0009982070885888\n",
+      "train loss:0.0010818384439299652\n",
+      "train loss:0.003174642415543906\n",
+      "train loss:0.021182003720132833\n",
+      "train loss:0.0037962737058721426\n",
+      "train loss:0.00236639989423074\n",
+      "train loss:0.003047330497507742\n",
+      "train loss:0.000489095979143658\n",
+      "train loss:0.0014152900021525445\n",
+      "train loss:0.0018927734408379952\n",
+      "train loss:0.0026242028165384176\n",
+      "train loss:0.0010141733149466759\n",
+      "train loss:0.0025441443809702926\n",
+      "train loss:0.0005094621466544245\n",
+      "train loss:0.004986836228997798\n",
+      "train loss:0.0010275270983111434\n",
+      "train loss:0.00023175223674225164\n",
+      "train loss:0.0004724828399550227\n",
+      "train loss:0.000763921678837212\n",
+      "train loss:0.004700030610819931\n",
+      "train loss:0.006466739136898638\n",
+      "train loss:0.0004583258800215384\n",
+      "train loss:0.0029478317696742095\n",
+      "train loss:0.0007780696200553182\n",
+      "train loss:0.0026863677302213433\n",
+      "train loss:0.002172490878845281\n",
+      "train loss:0.0028723144358083004\n",
+      "train loss:0.0007094948154580354\n",
+      "train loss:0.00010715287790214882\n",
+      "train loss:0.0012005191715878402\n",
+      "train loss:0.0011713687565917924\n",
+      "train loss:0.0014355064738887338\n",
+      "train loss:0.005007260482291793\n",
+      "train loss:0.0018803190676037502\n",
+      "train loss:0.006587490539051941\n",
+      "train loss:0.008408762345635921\n",
+      "train loss:0.004943034231964991\n",
+      "train loss:0.0015457869280185584\n",
+      "train loss:0.0016808657532264035\n",
+      "train loss:0.007311245131317828\n",
+      "train loss:0.0011552888629769082\n",
+      "train loss:0.009101303261093657\n",
+      "train loss:0.0005451538680743358\n",
+      "train loss:0.005567132527197656\n",
+      "train loss:0.000673229734522774\n",
+      "train loss:0.0016969046363655\n",
+      "train loss:0.0025459228287764662\n",
+      "train loss:0.00745137584198925\n",
+      "train loss:0.0069943253224648595\n",
+      "train loss:0.003249382551600024\n",
+      "train loss:0.003132268068275482\n",
+      "train loss:0.0008507935866862579\n",
+      "train loss:0.0009330914461558899\n",
+      "train loss:0.0004001390110897881\n",
+      "train loss:0.0005530529265943364\n",
+      "train loss:0.0015519350279821753\n",
+      "train loss:0.004212457215434972\n",
+      "train loss:0.001160164212626418\n",
+      "train loss:0.0003805032469230436\n",
+      "train loss:0.001157679065055928\n",
+      "train loss:0.006675472785883787\n",
+      "train loss:0.0002901662839974176\n",
+      "train loss:0.0023228266998774937\n",
+      "train loss:0.0012363674207705349\n",
+      "train loss:0.027311676372915462\n",
+      "train loss:0.001068942705173454\n",
+      "train loss:5.7275555781357985e-05\n",
+      "train loss:0.003797316877085049\n",
+      "train loss:0.0019301142491706667\n",
+      "train loss:0.002655374270464354\n",
+      "train loss:0.0019511779978294466\n",
+      "train loss:0.00013538356743723117\n",
+      "train loss:0.017452999175294196\n",
+      "train loss:0.0006366939620685423\n",
+      "train loss:0.014393565356312945\n",
+      "train loss:0.001640512681976393\n",
+      "train loss:0.01332417531984375\n",
+      "train loss:0.0018668090141364558\n",
+      "train loss:0.0012767039600208585\n",
+      "train loss:0.0028842233958043086\n",
+      "train loss:0.007402035453819658\n",
+      "train loss:0.013526714271058086\n",
+      "train loss:0.0029160549242786345\n",
+      "train loss:0.003568650393293946\n",
+      "train loss:0.002607099158573739\n",
+      "train loss:0.006084577015365454\n",
+      "train loss:0.005342310057704545\n",
+      "train loss:0.0012619487368318466\n",
+      "train loss:0.012544377246525573\n",
+      "train loss:0.0015065025157776023\n",
+      "train loss:0.0013178293852693281\n",
+      "train loss:0.00021679782083885312\n",
+      "train loss:0.0015570081707072328\n",
+      "train loss:0.0011475262592003017\n",
+      "train loss:0.004046838131271529\n",
+      "train loss:0.0005676868884128659\n",
+      "train loss:0.0017565422713915714\n",
+      "train loss:0.0011301576248258268\n",
+      "train loss:0.004386644524196685\n",
+      "train loss:0.0019863657036474206\n",
+      "train loss:0.005372699376725722\n",
+      "train loss:0.0020074523144830003\n",
+      "train loss:0.0006245117182635809\n",
+      "train loss:0.00411300016804857\n",
+      "train loss:0.0065190566857470855\n",
+      "train loss:0.006812957921110194\n",
+      "train loss:0.00761992411819673\n",
+      "train loss:0.00025534508311141264\n",
+      "train loss:0.0008283664599707638\n",
+      "train loss:0.007177900175308255\n",
+      "train loss:0.0067438566165233214\n",
+      "train loss:0.008256582319053738\n",
+      "train loss:0.016102049846659694\n",
+      "train loss:0.0009235247761864112\n",
+      "train loss:0.003161205035641859\n",
+      "train loss:0.03165879871539782\n",
+      "train loss:0.0015449837083693033\n",
+      "train loss:0.0055597243152195365\n",
+      "train loss:0.0022592557380489505\n",
+      "train loss:0.0006293309084596808\n",
+      "train loss:0.0038667674321636674\n",
+      "train loss:0.00030650124936510385\n",
+      "train loss:0.0025285130730809385\n",
+      "train loss:0.0035946975023219197\n",
+      "train loss:0.0012137969698083116\n",
+      "train loss:0.0004006926626556222\n",
+      "train loss:0.0016533912812223559\n",
+      "train loss:0.0007627912137696069\n",
+      "train loss:0.0034644578321732676\n",
+      "train loss:0.023852216838927277\n",
+      "train loss:0.005648590784303079\n",
+      "train loss:0.0039401284159619986\n",
+      "train loss:0.00041850924749792977\n",
+      "train loss:0.0014996913908347973\n",
+      "train loss:0.0038469839375678255\n",
+      "train loss:0.0175267692240069\n",
+      "train loss:0.0013544028378410713\n",
+      "train loss:0.006718595841614023\n",
+      "train loss:0.0030066476220108867\n",
+      "train loss:0.0015297825467442854\n",
+      "train loss:0.004319981298495254\n",
+      "train loss:0.000436211979895326\n",
+      "train loss:0.0023476885037634746\n",
+      "train loss:0.00297239100956119\n",
+      "train loss:0.0017358222545477446\n",
+      "train loss:0.0005187503930540805\n",
+      "train loss:0.0012769905109887847\n",
+      "train loss:0.00325716900857411\n",
+      "train loss:0.002127234325062577\n",
+      "train loss:0.003027527458028627\n",
+      "train loss:0.0010245158287118996\n",
+      "train loss:0.002562314185464313\n",
+      "train loss:0.005073208004756414\n",
+      "train loss:0.0028742621559129738\n",
+      "train loss:0.0016150669404078441\n",
+      "train loss:0.0006698051921834597\n",
+      "train loss:0.00020193318269897973\n",
+      "train loss:0.0021959175237791467\n",
+      "train loss:0.0007933014463044669\n",
+      "train loss:0.0025081529608133373\n",
+      "train loss:0.001504317954722333\n",
+      "train loss:0.002360919494353802\n",
+      "train loss:0.012106700788936142\n",
+      "train loss:0.0010559590549735585\n",
+      "train loss:0.0012893107092639092\n",
+      "train loss:0.0329394899338803\n",
+      "train loss:0.0040369477959840624\n",
+      "train loss:0.0003010962497052507\n",
+      "train loss:0.0033616782874424374\n",
+      "train loss:0.001688517213677269\n",
+      "train loss:0.0010708526199544642\n",
+      "train loss:0.00256143099195854\n",
+      "train loss:0.001134485429802284\n",
+      "train loss:0.00048464511180498976\n",
+      "train loss:0.0007366679140580021\n",
+      "train loss:0.008203166456671582\n",
+      "train loss:0.008756226367292983\n",
+      "train loss:0.004808162107982949\n",
+      "train loss:0.0031003696138903465\n",
+      "train loss:0.004612596279301306\n",
+      "train loss:0.008307206481499548\n",
+      "train loss:0.0018999568705054131\n",
+      "train loss:0.00023154064412019094\n",
+      "train loss:0.001008528351943782\n",
+      "train loss:0.0009720529370422292\n",
+      "train loss:0.014467998497894811\n",
+      "train loss:0.00853353270203431\n",
+      "train loss:0.003979350161644905\n",
+      "train loss:0.0015404090068141987\n",
+      "train loss:0.0013839926000097482\n",
+      "train loss:0.0015746726269552445\n",
+      "train loss:0.0011583940510126243\n",
+      "train loss:0.005608896421542985\n",
+      "train loss:0.003264496963404299\n",
+      "train loss:0.004751482729195355\n",
+      "train loss:0.025968835774801025\n",
+      "train loss:0.008270862720304372\n",
+      "train loss:0.00020158718930432426\n",
+      "train loss:0.0016845818665882672\n",
+      "train loss:0.00042330492645128217\n",
+      "train loss:0.0001693042702946995\n",
+      "train loss:0.0009472676522040097\n",
+      "train loss:0.0024743498742307653\n",
+      "train loss:0.000995589288710081\n",
+      "train loss:0.025327410489038228\n",
+      "train loss:0.006412260913348185\n",
+      "train loss:0.010588773777735363\n",
+      "train loss:0.003569496826866954\n",
+      "train loss:0.0016581641699233154\n",
+      "train loss:0.0017631030840985528\n",
+      "train loss:0.0014849554661379395\n",
+      "train loss:0.002275593719527926\n",
+      "train loss:0.002981615262158272\n",
+      "train loss:0.007560639395962347\n",
+      "train loss:0.008757175649968963\n",
+      "train loss:0.008319072669035095\n",
+      "train loss:0.0007056095025464093\n",
+      "train loss:0.0031205667667879773\n",
+      "train loss:0.0011935247182611366\n",
+      "train loss:0.042225725315768375\n",
+      "train loss:0.004408935070088288\n",
+      "train loss:0.001251162363352742\n",
+      "train loss:0.0033178169606852877\n",
+      "train loss:0.0003186671931123913\n",
+      "train loss:0.0036666381602437153\n",
+      "train loss:0.003564393004347224\n",
+      "train loss:0.0022798223389058443\n",
+      "train loss:0.010504947583792421\n",
+      "train loss:0.0009929485488638582\n",
+      "train loss:0.006160295071518862\n",
+      "train loss:0.0007200785782932561\n",
+      "train loss:0.02099320059390609\n",
+      "train loss:0.01341027386885773\n",
+      "train loss:0.0021056379379569556\n",
+      "train loss:0.0028320532198188564\n",
+      "train loss:0.002712160809455242\n",
+      "train loss:0.0005990039935271961\n",
+      "train loss:0.0004193679847358557\n",
+      "train loss:0.004766935435486138\n",
+      "train loss:0.001160445587161103\n",
+      "train loss:0.0047493938556952865\n",
+      "train loss:0.0006327386639613067\n",
+      "train loss:0.0008922757338410751\n",
+      "train loss:0.0009486592409041375\n",
+      "train loss:0.0027342554169872547\n",
+      "train loss:0.01104584296727668\n",
+      "train loss:0.02404182597934729\n",
+      "train loss:0.001234126090172919\n",
+      "train loss:0.009592808186319655\n",
+      "train loss:0.0005732648707064673\n",
+      "train loss:0.005524726192616184\n",
+      "train loss:0.005398479036733589\n",
+      "train loss:0.0014263042006174092\n",
+      "train loss:0.0006188532963490176\n",
+      "train loss:0.00048054834784529296\n",
+      "train loss:0.0018803222614616805\n",
+      "train loss:0.005244628951732215\n",
+      "train loss:0.0008720063268086631\n",
+      "train loss:0.0031560983202541678\n",
+      "train loss:0.0008435703223133742\n",
+      "train loss:0.002467039427321831\n",
+      "train loss:0.0068154777875574475\n",
+      "train loss:0.0007955635161654929\n",
+      "train loss:0.0022119896221050863\n",
+      "train loss:0.018015394214017346\n",
+      "train loss:0.01100531027445606\n",
+      "train loss:0.002098548047094125\n",
+      "train loss:0.00020037276893455057\n",
+      "train loss:0.0022857340362817556\n",
+      "train loss:0.007258434602689784\n",
+      "train loss:0.0013356310133656654\n",
+      "train loss:0.0007060259559785695\n",
+      "train loss:0.005106808793091457\n",
+      "train loss:0.0032810738268256173\n",
+      "train loss:0.03856768497404055\n",
+      "train loss:0.00224957505931029\n",
+      "train loss:0.006464331113854499\n",
+      "train loss:0.001966446402250324\n",
+      "train loss:0.0035001007286926313\n",
+      "train loss:0.0017748748087855354\n",
+      "train loss:0.0018033298178182452\n",
+      "train loss:0.0032884060728626163\n",
+      "train loss:0.0022171766278698143\n",
+      "train loss:0.0006770517224071608\n",
+      "train loss:0.0034788344584237864\n",
+      "train loss:0.003426254842991783\n",
+      "train loss:0.0011626811075730085\n",
+      "train loss:0.0041702554115991735\n",
+      "train loss:0.0017207169870201366\n",
+      "train loss:0.0010488824073274091\n",
+      "train loss:0.005255300231250112\n",
+      "train loss:0.059742526469592966\n",
+      "train loss:0.0032027944410141125\n",
+      "train loss:0.00417029753645637\n",
+      "train loss:0.015050016524128806\n",
+      "train loss:0.0060077260183389455\n",
+      "train loss:0.0013252229741955346\n",
+      "train loss:0.0022382887269756114\n",
+      "train loss:0.0015529348198744253\n",
+      "train loss:0.00474658101609288\n",
+      "train loss:0.032264744225805886\n",
+      "train loss:0.012716857724440915\n",
+      "train loss:0.0015706179621849936\n",
+      "train loss:0.0005769256703656267\n",
+      "train loss:0.001757121487091467\n",
+      "train loss:0.0008327433175335781\n",
+      "train loss:0.004165362798534168\n",
+      "train loss:0.003955574181041224\n",
+      "train loss:0.003708463144301455\n",
+      "train loss:0.007189679746460662\n",
+      "train loss:0.007031635873873027\n",
+      "train loss:0.0025733143025340618\n",
+      "train loss:0.021364157011995302\n",
+      "train loss:0.005648566820512993\n",
+      "train loss:0.003321639958605599\n",
+      "train loss:0.007328148048632891\n",
+      "train loss:0.0016689095170632244\n",
+      "train loss:0.01024313729237624\n",
+      "train loss:0.0012033352226983346\n",
+      "train loss:0.0045885425775883445\n",
+      "train loss:0.005998457905053255\n",
+      "train loss:0.017380184913765286\n",
+      "train loss:0.002223296814928134\n",
+      "train loss:0.0010821320632397068\n",
+      "train loss:0.008600006711929793\n",
+      "train loss:0.0014134175902650046\n",
+      "train loss:0.00034742226169540277\n",
+      "train loss:0.004610359896328102\n",
+      "train loss:0.0012683625865149905\n",
+      "train loss:0.0017607870624595375\n",
+      "train loss:0.0005882156804865923\n",
+      "train loss:0.0011507987618279006\n",
+      "train loss:0.0022497300421240075\n",
+      "train loss:0.000583620758203903\n",
+      "train loss:0.0007761610147914002\n",
+      "train loss:0.0007988389876838654\n",
+      "train loss:0.004416783391264384\n",
+      "train loss:0.005597954488222269\n",
+      "train loss:0.0018731729748587178\n",
+      "train loss:0.00044038305369695837\n",
+      "train loss:0.0005844018977962599\n",
+      "train loss:0.0003393554279994995\n",
+      "train loss:0.0025978129227494555\n",
+      "train loss:0.0002938196781235047\n",
+      "train loss:0.0036037561779609815\n",
+      "train loss:0.013399063958278007\n",
+      "train loss:0.001178320355090888\n",
+      "train loss:0.0011028720642612456\n",
+      "train loss:0.0002843618151837493\n",
+      "train loss:0.004664626102212818\n",
+      "train loss:0.005114304486654702\n",
+      "train loss:0.0044936679112715155\n",
+      "train loss:0.0008757603036049739\n",
+      "train loss:0.008077986734360334\n",
+      "train loss:0.0002329334354398851\n",
+      "train loss:0.0012385522015986006\n",
+      "train loss:0.02289845785406187\n",
+      "train loss:0.004219893525304872\n",
+      "train loss:0.0028379985933622864\n",
+      "train loss:0.007851032543450997\n",
+      "train loss:0.0023217646483923906\n",
+      "train loss:0.0004100474931123326\n",
+      "train loss:0.0016328254631333386\n",
+      "train loss:0.010389166102209127\n",
+      "train loss:0.006244115399184721\n",
+      "train loss:0.024821871482486543\n",
+      "train loss:0.002430283543511875\n",
+      "train loss:0.0019520116577506326\n",
+      "train loss:0.023757333784694953\n",
+      "train loss:0.0007113042947605895\n",
+      "train loss:0.07076153829777017\n",
+      "train loss:0.002162929977508678\n",
+      "train loss:0.001047881665153765\n",
+      "train loss:0.0028015219830542027\n",
+      "train loss:0.004549004598871516\n",
+      "train loss:0.0002163336445051652\n",
+      "train loss:0.006557147766523313\n",
+      "train loss:0.00457384912191359\n",
+      "train loss:0.009334182115436537\n",
+      "train loss:0.0018168728101970776\n",
+      "train loss:0.002829023170696401\n",
+      "train loss:0.02359185652610664\n",
+      "train loss:0.005509291987585246\n",
+      "train loss:0.015134367342872858\n",
+      "train loss:0.0008258910060766267\n",
+      "train loss:0.0001808500971009399\n",
+      "train loss:0.0098847583179516\n",
+      "train loss:0.001202796314007398\n",
+      "train loss:0.00015798975577867237\n",
+      "train loss:0.0024576727319578777\n",
+      "train loss:0.0021079003962618404\n",
+      "train loss:0.0021520122758429366\n",
+      "train loss:0.0029713351110176056\n",
+      "train loss:0.0003027864587905426\n",
+      "train loss:0.0006316089123021947\n",
+      "train loss:0.0013089389301159113\n",
+      "train loss:0.0028938114729244392\n",
+      "train loss:0.001378143341715452\n",
+      "train loss:0.001394016050725536\n",
+      "train loss:0.0033323854835843607\n",
+      "train loss:0.0020071163034560326\n",
+      "train loss:0.00577124977680556\n",
+      "train loss:0.0017406946333486397\n",
+      "train loss:0.013205347760155917\n",
+      "train loss:0.002735965704244604\n",
+      "train loss:0.00038658307803170596\n",
+      "train loss:0.002827174707394869\n",
+      "train loss:0.0015018848427369698\n",
+      "train loss:0.0005661628689639265\n",
+      "train loss:0.0014821030249327705\n",
+      "train loss:0.0016310947704609605\n",
+      "train loss:0.004490286848350929\n",
+      "train loss:0.0010602842020651882\n",
+      "train loss:0.0017189798375913214\n",
+      "train loss:6.682572946258413e-05\n",
+      "train loss:0.005594981346902611\n",
+      "train loss:0.0014098759941325472\n",
+      "train loss:0.006627956855770774\n",
+      "train loss:0.006514132253014645\n",
+      "train loss:0.0007187025817873412\n",
+      "train loss:0.0002922382538274821\n",
+      "train loss:0.003310164715441524\n",
+      "train loss:0.001623035261319925\n",
+      "train loss:0.0028028114445569673\n",
+      "train loss:0.003489410859217571\n",
+      "train loss:0.007700114290489971\n",
+      "train loss:0.004919177845810439\n",
+      "train loss:0.004985969574656606\n",
+      "train loss:0.0013459799577581464\n",
+      "train loss:0.0036422137245669993\n",
+      "train loss:0.005195748964075567\n",
+      "train loss:0.0019342949232717304\n",
+      "train loss:0.0033345970044177577\n",
+      "train loss:0.00017348297395281974\n",
+      "train loss:0.0023555092258015563\n",
+      "train loss:0.0001719720726937964\n",
+      "train loss:0.008356002107081066\n",
+      "train loss:0.0010142150765612827\n",
+      "train loss:0.008137497368966039\n",
+      "train loss:0.0028068269400412845\n",
+      "train loss:0.003193351803188742\n",
+      "train loss:0.002070555771619084\n",
+      "train loss:0.0036681938083026793\n",
+      "train loss:0.0037665758129091804\n",
+      "train loss:0.0034102548785513464\n",
+      "train loss:0.0007219195916634071\n",
+      "train loss:0.010289095626331854\n",
+      "train loss:0.0008907048354391617\n",
+      "train loss:0.001997999912919165\n",
+      "train loss:0.0022429211580492826\n",
+      "train loss:0.005213591121860802\n",
+      "train loss:0.001573606283671545\n",
+      "train loss:0.003087390456435163\n",
+      "train loss:0.002325881272462075\n",
+      "train loss:0.011434747430345166\n",
+      "train loss:0.006476163731538749\n",
+      "train loss:0.001850200041814675\n",
+      "train loss:0.0026373341089862543\n",
+      "train loss:0.0004614524689075312\n",
+      "train loss:0.00024062601906566384\n",
+      "train loss:0.0009604511246443558\n",
+      "train loss:0.0018793738162398578\n",
+      "train loss:0.00513931206599355\n",
+      "train loss:0.003271651655542817\n",
+      "train loss:0.002969579999184354\n",
+      "train loss:0.0073085338553784684\n",
+      "train loss:0.008888366176764613\n",
+      "train loss:0.00030456309255674335\n",
+      "train loss:0.00043746205746862557\n",
+      "train loss:0.0034033910504434302\n",
+      "train loss:0.000636838400561045\n",
+      "train loss:0.0183535633953773\n",
+      "train loss:0.017768688116939334\n",
+      "train loss:0.0035928398411242644\n",
+      "train loss:0.00037646167537956214\n",
+      "train loss:0.008659120747075312\n",
+      "train loss:0.004853393640164436\n",
+      "train loss:0.0007245711281156901\n",
+      "train loss:0.0004871459777855455\n",
+      "train loss:0.022858000047800787\n",
+      "train loss:0.00035596742071022155\n",
+      "train loss:0.003380844441707825\n",
+      "train loss:0.004553488204581912\n",
+      "train loss:0.0021612565585467564\n",
+      "train loss:0.006001048816359579\n",
+      "train loss:0.0033485953370154735\n",
+      "train loss:0.001265833180462722\n",
+      "train loss:0.007544067414729626\n",
+      "train loss:0.013825175758773447\n",
+      "train loss:0.0010081118758594487\n",
+      "train loss:0.002743070681078623\n",
+      "train loss:0.005560423122102793\n",
+      "train loss:0.006103886285491279\n",
+      "train loss:0.0013546702269394515\n",
+      "train loss:0.0015777353969843805\n",
+      "train loss:0.0025796785485401453\n",
+      "train loss:0.010996065798204504\n",
+      "train loss:0.0006754303645314655\n",
+      "train loss:0.0008152014525114865\n",
+      "train loss:0.0002975216352187709\n",
+      "train loss:0.010760501388532469\n",
+      "=== epoch:14, train acc:0.991, test acc:0.977 ===\n",
+      "train loss:0.005295270403425826\n",
+      "train loss:0.0009016994158020441\n",
+      "train loss:0.008377256693058036\n",
+      "train loss:0.0025434938910941485\n",
+      "train loss:0.0026597592125962\n",
+      "train loss:0.005600410521822117\n",
+      "train loss:0.0011510643565423016\n",
+      "train loss:0.002183799054431248\n",
+      "train loss:0.0005550154891463305\n",
+      "train loss:0.0006380932512473784\n",
+      "train loss:0.0028791169403043354\n",
+      "train loss:0.0016368540652772464\n",
+      "train loss:0.00389588859207202\n",
+      "train loss:0.013893006752842083\n",
+      "train loss:0.01910477074880143\n",
+      "train loss:0.004193054613751184\n",
+      "train loss:0.003398833619257342\n",
+      "train loss:0.0004146519918612077\n",
+      "train loss:0.001353249538482104\n",
+      "train loss:0.0010027933440386707\n",
+      "train loss:0.0053800009883736765\n",
+      "train loss:0.000999202218662566\n",
+      "train loss:0.002628593080921437\n",
+      "train loss:0.005932016917975061\n",
+      "train loss:0.008126203086023697\n",
+      "train loss:0.0028581284038844\n",
+      "train loss:0.0011621744001397024\n",
+      "train loss:0.0005873674622335719\n",
+      "train loss:0.0027394765778073382\n",
+      "train loss:0.004673347852978257\n",
+      "train loss:0.004163114079683724\n",
+      "train loss:0.002009891414556306\n",
+      "train loss:0.0016520981287653303\n",
+      "train loss:0.0008123659811509641\n",
+      "train loss:0.0077758013995116145\n",
+      "train loss:0.001274329749346156\n",
+      "train loss:0.0012132763290489878\n",
+      "train loss:0.00624251896138706\n",
+      "train loss:0.008956894187374933\n",
+      "train loss:0.002345006393847668\n",
+      "train loss:0.003878452983325473\n",
+      "train loss:0.0011840863188568784\n",
+      "train loss:0.00040339996785864374\n",
+      "train loss:0.004687298367086714\n",
+      "train loss:0.004410354276640962\n",
+      "train loss:0.0004663478588940193\n",
+      "train loss:0.002137884181806555\n",
+      "train loss:0.0023435809154577288\n",
+      "train loss:0.0003790600216805019\n",
+      "train loss:0.009422092880505977\n",
+      "train loss:0.002769107007052847\n",
+      "train loss:0.006104838058413301\n",
+      "train loss:0.0013028427534384453\n",
+      "train loss:0.00027493626508471637\n",
+      "train loss:0.005258593334315731\n",
+      "train loss:0.005856522298127904\n",
+      "train loss:0.004268903894288115\n",
+      "train loss:0.004186847464425142\n",
+      "train loss:0.01450209216430671\n",
+      "train loss:0.001714720339850108\n",
+      "train loss:0.0022542943813428176\n",
+      "train loss:0.0021191340285284088\n",
+      "train loss:0.006745346791565904\n",
+      "train loss:0.0019143988621188514\n",
+      "train loss:0.001895224665622225\n",
+      "train loss:0.002151468864933937\n",
+      "train loss:0.0017425588280571733\n",
+      "train loss:0.0039923249552838505\n",
+      "train loss:0.023495707884893927\n",
+      "train loss:0.001308249062897483\n",
+      "train loss:0.00367203342003999\n",
+      "train loss:0.0007613757515449569\n",
+      "train loss:0.002211843109189702\n",
+      "train loss:0.004212281169350167\n",
+      "train loss:0.006839549780163406\n",
+      "train loss:0.00045122636918649387\n",
+      "train loss:0.004816761989418714\n",
+      "train loss:0.001309474507708261\n",
+      "train loss:0.003757728441332613\n",
+      "train loss:0.0006752661434596735\n",
+      "train loss:0.0016480983992613228\n",
+      "train loss:0.0004944723050303645\n",
+      "train loss:0.009822392451004348\n",
+      "train loss:0.0013742457436961585\n",
+      "train loss:0.0006786483471198741\n",
+      "train loss:0.020119806070334076\n",
+      "train loss:0.00119780979989019\n",
+      "train loss:0.0006112457239462336\n",
+      "train loss:0.0013046660053105319\n",
+      "train loss:0.011907026197084647\n",
+      "train loss:0.0032605094866240756\n",
+      "train loss:0.002327150125799972\n",
+      "train loss:0.0017665243851931022\n",
+      "train loss:0.0022582925113228303\n",
+      "train loss:0.0008735932355047705\n",
+      "train loss:0.0025341671163456717\n",
+      "train loss:0.0024190162533889066\n",
+      "train loss:0.0011636072038105842\n",
+      "train loss:0.002928231909505935\n",
+      "train loss:0.018943857663514266\n",
+      "train loss:0.00042603084615104575\n",
+      "train loss:0.0004421243750486418\n",
+      "train loss:0.003978068581838413\n",
+      "train loss:0.0009386717323534568\n",
+      "train loss:0.000384631070759704\n",
+      "train loss:0.0018233829390971463\n",
+      "train loss:0.0012335518645635816\n",
+      "train loss:0.0034697705693486334\n",
+      "train loss:0.0017227252064376008\n",
+      "train loss:0.0010539351299156834\n",
+      "train loss:0.0020546905892726507\n",
+      "train loss:0.0006902243093285555\n",
+      "train loss:0.003093984424457935\n",
+      "train loss:0.00028197431058381723\n",
+      "train loss:0.001450266119767213\n",
+      "train loss:0.006663542701237738\n",
+      "train loss:0.0002474193237737261\n",
+      "train loss:0.0008190955031565375\n",
+      "train loss:0.0008477441552077391\n",
+      "train loss:0.0007695880521801536\n",
+      "train loss:0.01758957536319639\n",
+      "train loss:0.0024534617132133977\n",
+      "train loss:0.006905887682215818\n",
+      "train loss:0.006773919735535046\n",
+      "train loss:0.0020946020784671678\n",
+      "train loss:0.010214873345229738\n",
+      "train loss:0.003645982965875368\n",
+      "train loss:0.0014039067281831604\n",
+      "train loss:0.007656302087295508\n",
+      "train loss:0.003022455199086889\n",
+      "train loss:0.005296377380301188\n",
+      "train loss:0.0012879707364068205\n",
+      "train loss:0.00044965682251885803\n",
+      "train loss:0.00423269120830359\n",
+      "train loss:0.0012566727517725545\n",
+      "train loss:0.0012693850137359824\n",
+      "train loss:0.006128951310367522\n",
+      "train loss:0.0006633357251162586\n",
+      "train loss:0.016313860043710337\n",
+      "train loss:0.0021554844774169125\n",
+      "train loss:0.0005051536911424201\n",
+      "train loss:0.001915333455064641\n",
+      "train loss:0.01490089145941999\n",
+      "train loss:0.0005010184780964155\n",
+      "train loss:0.0005032118125573862\n",
+      "train loss:0.0003440743879134113\n",
+      "train loss:0.011571781586124333\n",
+      "train loss:0.044835121871518974\n",
+      "train loss:0.0014255572330289568\n",
+      "train loss:0.0009441797938504254\n",
+      "train loss:0.004285054433881924\n",
+      "train loss:0.0017047201321519268\n",
+      "train loss:0.0007116104535080472\n",
+      "train loss:0.002069688648173477\n",
+      "train loss:0.0005273152512905864\n",
+      "train loss:0.008778004356789042\n",
+      "train loss:0.0032290060348331966\n",
+      "train loss:0.001414021680461504\n",
+      "train loss:0.0037018084132868583\n",
+      "train loss:0.0002642670252822706\n",
+      "train loss:0.001985919418251274\n",
+      "train loss:0.0014379533091877037\n",
+      "train loss:0.00027253528499663\n",
+      "train loss:0.0016946649924092603\n",
+      "train loss:0.0021678147478861338\n",
+      "train loss:0.001486582932489179\n",
+      "train loss:0.0010798301091223294\n",
+      "train loss:0.0027166900228923043\n",
+      "train loss:0.0027067558145289837\n",
+      "train loss:0.0042993515203736425\n",
+      "train loss:0.0015557347030699304\n",
+      "train loss:0.0002373105381134959\n",
+      "train loss:0.022396474402055288\n",
+      "train loss:0.04149736869530262\n",
+      "train loss:0.001647692932809478\n",
+      "train loss:0.0007610701527916564\n",
+      "train loss:0.001637441590529612\n",
+      "train loss:0.00021065757561142994\n",
+      "train loss:0.004823585614408479\n",
+      "train loss:0.0011534283790117451\n",
+      "train loss:0.00010609290139425947\n",
+      "train loss:0.0084370558546632\n",
+      "train loss:0.0006829000186604053\n",
+      "train loss:0.004137345437426623\n",
+      "train loss:0.0027259615870761037\n",
+      "train loss:0.00031835144605969924\n",
+      "train loss:0.004094067204095635\n",
+      "train loss:0.0006970606950709847\n",
+      "train loss:0.007474376999226596\n",
+      "train loss:0.0027340868344878254\n",
+      "train loss:0.005501682600097268\n",
+      "train loss:0.0033761588649837294\n",
+      "train loss:0.0038739178878960325\n",
+      "train loss:0.0046515929225271575\n",
+      "train loss:0.0033917147901166207\n",
+      "train loss:9.756509911405723e-05\n",
+      "train loss:0.007987572097191024\n",
+      "train loss:0.002311809736636817\n",
+      "train loss:0.001092557246131464\n",
+      "train loss:0.0008057708192203026\n",
+      "train loss:0.00010619404086839926\n",
+      "train loss:0.014931067706158865\n",
+      "train loss:0.021629400128066165\n",
+      "train loss:0.004437701273032665\n",
+      "train loss:0.0046696476788253664\n",
+      "train loss:0.0003850687872450969\n",
+      "train loss:0.0044413293678195525\n",
+      "train loss:0.023481510286756605\n",
+      "train loss:0.0009253826204552676\n",
+      "train loss:0.0005577063429007017\n",
+      "train loss:0.006927621756317987\n",
+      "train loss:0.004983127409619198\n",
+      "train loss:0.0045512537238416835\n",
+      "train loss:0.019908646608544564\n",
+      "train loss:0.0027041540342746294\n",
+      "train loss:0.007140321644194111\n",
+      "train loss:0.006937762327692814\n",
+      "train loss:0.023324466966848006\n",
+      "train loss:0.005346113122400673\n",
+      "train loss:0.0037637097445030352\n",
+      "train loss:0.0018537498095123138\n",
+      "train loss:0.0002169276455103134\n",
+      "train loss:0.0016814294913972997\n",
+      "train loss:0.00271810019559709\n",
+      "train loss:0.0013223772816811981\n",
+      "train loss:0.0017449076736427063\n",
+      "train loss:0.0007646486811920777\n",
+      "train loss:0.003004763974136052\n",
+      "train loss:0.030312853731730205\n",
+      "train loss:0.001368487271025066\n",
+      "train loss:0.0013231699218793627\n",
+      "train loss:0.0032467545637530825\n",
+      "train loss:0.0059868640987815955\n",
+      "train loss:0.001597773788875782\n",
+      "train loss:0.009793774250031216\n",
+      "train loss:0.0042012983404335355\n",
+      "train loss:0.001419168941325607\n",
+      "train loss:0.0011632958811076297\n",
+      "train loss:0.0008610740168699911\n",
+      "train loss:0.0007909678426802353\n",
+      "train loss:0.002762188479815874\n",
+      "train loss:0.002660914264464105\n",
+      "train loss:0.0044515289411432265\n",
+      "train loss:0.0047048351177110285\n",
+      "train loss:0.00011577262100506802\n",
+      "train loss:0.0039027958509113775\n",
+      "train loss:0.002147904620915923\n",
+      "train loss:0.0043540002929159435\n",
+      "train loss:0.00026639256102267147\n",
+      "train loss:0.0021792951222329424\n",
+      "train loss:0.002860662239519878\n",
+      "train loss:0.0018911330823098964\n",
+      "train loss:0.00345606200062497\n",
+      "train loss:0.006549529601118459\n",
+      "train loss:0.00027515749827920267\n",
+      "train loss:0.004596058232185497\n",
+      "train loss:0.10164556545108608\n",
+      "train loss:0.00041244430832474023\n",
+      "train loss:0.000302642815742701\n",
+      "train loss:0.0007277629302950238\n",
+      "train loss:0.006641113915252958\n",
+      "train loss:0.0010557187565536608\n",
+      "train loss:0.0011632238186912915\n",
+      "train loss:0.0007504288912262434\n",
+      "train loss:0.009829555292140562\n",
+      "train loss:0.0009589919423726222\n",
+      "train loss:0.0007074166652100716\n",
+      "train loss:0.00594913786850516\n",
+      "train loss:0.006186022257334658\n",
+      "train loss:0.0025332128198461267\n",
+      "train loss:0.0008527082205710668\n",
+      "train loss:0.004707146367645973\n",
+      "train loss:0.002210579767203551\n",
+      "train loss:0.0023510828574347235\n",
+      "train loss:0.00622616852892391\n",
+      "train loss:0.004754849948059761\n",
+      "train loss:0.00193038491027798\n",
+      "train loss:0.0018503942234789155\n",
+      "train loss:0.004138626175906713\n",
+      "train loss:0.0008430632284775575\n",
+      "train loss:0.0030188276155454414\n",
+      "train loss:0.007683604449000745\n",
+      "train loss:0.0038517002730255267\n",
+      "train loss:0.0004667917675343634\n",
+      "train loss:0.002585348780828203\n",
+      "train loss:0.004694113148658705\n",
+      "train loss:0.0020330769506814207\n",
+      "train loss:0.00033575114988010623\n",
+      "train loss:0.002947986823490995\n",
+      "train loss:0.001146768646991972\n",
+      "train loss:0.003442124982484906\n",
+      "train loss:0.0006467801594245155\n",
+      "train loss:0.0068345483121328545\n",
+      "train loss:0.008744750541190902\n",
+      "train loss:0.003777934447868087\n",
+      "train loss:0.0012448106290885918\n",
+      "train loss:0.0004921747547351362\n",
+      "train loss:0.001461615561052338\n",
+      "train loss:0.0044129689870813895\n",
+      "train loss:0.0033723157248168978\n",
+      "train loss:0.0005485283009707259\n",
+      "train loss:0.0026158556398334525\n",
+      "train loss:0.0036887873040982976\n",
+      "train loss:0.09278486515059933\n",
+      "train loss:0.006399613303713628\n",
+      "train loss:0.002613038977166701\n",
+      "train loss:0.0023094120943109755\n",
+      "train loss:0.00018684895802040732\n",
+      "train loss:0.003962481768532571\n",
+      "train loss:0.0034504700276628687\n",
+      "train loss:0.015921997676875584\n",
+      "train loss:0.007044169730785117\n",
+      "train loss:0.00014503689902963876\n",
+      "train loss:0.003974551201397947\n",
+      "train loss:0.0024023901593713446\n",
+      "train loss:0.0020231501100488308\n",
+      "train loss:0.0006761750650266156\n",
+      "train loss:0.14825166674035808\n",
+      "train loss:0.005958010226328141\n",
+      "train loss:0.020538703830776765\n",
+      "train loss:0.019453872849888542\n",
+      "train loss:0.025402728363051966\n",
+      "train loss:0.0036284397848130134\n",
+      "train loss:0.004119299573643871\n",
+      "train loss:0.029977111230611843\n",
+      "train loss:0.009417375213142556\n",
+      "train loss:0.030842074481914478\n",
+      "train loss:0.0030253368992522543\n",
+      "train loss:0.002491226350267361\n",
+      "train loss:0.012302465810263428\n",
+      "train loss:0.005122417465076297\n",
+      "train loss:0.01762803843062925\n",
+      "train loss:0.0274779274286751\n",
+      "train loss:0.0031764848555530367\n",
+      "train loss:0.0053623005915702515\n",
+      "train loss:0.011199371316039491\n",
+      "train loss:0.0004365067774132935\n",
+      "train loss:0.002785231263522044\n",
+      "train loss:0.005279937065996755\n",
+      "train loss:0.002676754923927258\n",
+      "train loss:0.0010056018009253654\n",
+      "train loss:0.006692722879549861\n",
+      "train loss:0.001821115758679287\n",
+      "train loss:0.009460394465027512\n",
+      "train loss:0.008456490824182636\n",
+      "train loss:0.000984084964341651\n",
+      "train loss:0.0002708560140200246\n",
+      "train loss:0.0016680958329615297\n",
+      "train loss:0.000953564342937778\n",
+      "train loss:0.007593314116798739\n",
+      "train loss:0.007683968394221517\n",
+      "train loss:0.031120076708903616\n",
+      "train loss:0.02351785144779952\n",
+      "train loss:0.002028508111939616\n",
+      "train loss:0.002565153629385423\n",
+      "train loss:0.0015632793806596562\n",
+      "train loss:0.01915251134001885\n",
+      "train loss:0.004564349839931063\n",
+      "train loss:0.015187290530801653\n",
+      "train loss:0.009001774859273035\n",
+      "train loss:0.007697634615506072\n",
+      "train loss:0.0022696318549205578\n",
+      "train loss:0.0041365609363603175\n",
+      "train loss:0.004743440651555918\n",
+      "train loss:0.00371953842973227\n",
+      "train loss:0.01200273192069925\n",
+      "train loss:0.0014437174356983174\n",
+      "train loss:0.0013442961696821323\n",
+      "train loss:0.0014862585697633555\n",
+      "train loss:0.0009125089250097933\n",
+      "train loss:0.0002910363237812587\n",
+      "train loss:0.003421945453117676\n",
+      "train loss:0.0017676003882358563\n",
+      "train loss:0.0009741229067713399\n",
+      "train loss:0.007969400966146666\n",
+      "train loss:0.008496014012745873\n",
+      "train loss:0.0012501043540160816\n",
+      "train loss:0.00025584407109670076\n",
+      "train loss:0.003191037764156762\n",
+      "train loss:0.00680831210552467\n",
+      "train loss:0.009910637977341997\n",
+      "train loss:0.025404732630495176\n",
+      "train loss:0.0008265953307054353\n",
+      "train loss:0.0010274258547687991\n",
+      "train loss:0.0727200455903091\n",
+      "train loss:0.003696956549532219\n",
+      "train loss:0.0010492985165098145\n",
+      "train loss:0.0003508032582074749\n",
+      "train loss:0.00013258510976312936\n",
+      "train loss:0.001579193139245233\n",
+      "train loss:0.019965419986858014\n",
+      "train loss:0.002259991124392093\n",
+      "train loss:0.0011200692902090346\n",
+      "train loss:0.018297405679516317\n",
+      "train loss:0.00020758091774652332\n",
+      "train loss:0.00045291626106597286\n",
+      "train loss:0.0004769253744904104\n",
+      "train loss:0.002683249967815818\n",
+      "train loss:0.0003203212101466171\n",
+      "train loss:0.00671309963542286\n",
+      "train loss:0.001491814319926144\n",
+      "train loss:0.005961534498334491\n",
+      "train loss:0.0015177959934522239\n",
+      "train loss:0.0037627052697626006\n",
+      "train loss:0.00022700431321451958\n",
+      "train loss:0.0011650455323228047\n",
+      "train loss:0.0012606384051683094\n",
+      "train loss:0.0006073459240101972\n",
+      "train loss:0.003448683855149862\n",
+      "train loss:0.009367474199760197\n",
+      "train loss:0.001872400343193579\n",
+      "train loss:0.009817626604092666\n",
+      "train loss:0.0015605338184137764\n",
+      "train loss:3.725291052603859e-05\n",
+      "train loss:0.0024703853687030324\n",
+      "train loss:0.00377142356042767\n",
+      "train loss:0.005012706896446153\n",
+      "train loss:0.0006100523915618955\n",
+      "train loss:0.001366858307039313\n",
+      "train loss:0.004733284463066456\n",
+      "train loss:0.0022521990101559766\n",
+      "train loss:0.0007543565585523486\n",
+      "train loss:0.0068544203875989815\n",
+      "train loss:0.003758184178599038\n",
+      "train loss:0.003635323909197101\n",
+      "train loss:0.0032109136407516784\n",
+      "train loss:0.005073238043119057\n",
+      "train loss:0.0026476180509226134\n",
+      "train loss:0.010610124365194215\n",
+      "train loss:0.0012235546550912126\n",
+      "train loss:0.0014932392911823245\n",
+      "train loss:0.0030118108390294134\n",
+      "train loss:0.009385746140452412\n",
+      "train loss:0.0024290807375695373\n",
+      "train loss:0.005398242621236124\n",
+      "train loss:0.0012112220942305468\n",
+      "train loss:0.0003982828546845108\n",
+      "train loss:0.001718503306137875\n",
+      "train loss:0.008378595587171626\n",
+      "train loss:0.002087848682639145\n",
+      "train loss:0.005832858248170924\n",
+      "train loss:0.003092772954727774\n",
+      "train loss:0.0012748884855209028\n",
+      "train loss:0.007836016108785192\n",
+      "train loss:0.001008010971931415\n",
+      "train loss:0.00028761808005848313\n",
+      "train loss:0.017011319990439464\n",
+      "train loss:0.005923273807906105\n",
+      "train loss:0.0005451944099529787\n",
+      "train loss:0.0007826394863352573\n",
+      "train loss:0.0004897864442553995\n",
+      "train loss:0.001547738144555632\n",
+      "train loss:0.00043144944651310094\n",
+      "train loss:0.00040224496576511345\n",
+      "train loss:0.004951594504466757\n",
+      "train loss:0.0044900213057510645\n",
+      "train loss:0.0023977109992617857\n",
+      "train loss:0.0008871785924948626\n",
+      "train loss:0.011133855851584584\n",
+      "train loss:0.006083340764537045\n",
+      "train loss:0.0006572114398635622\n",
+      "train loss:0.0033635719337789865\n",
+      "train loss:0.005225509317706005\n",
+      "train loss:0.0018432571176962285\n",
+      "train loss:0.0005508160379496532\n",
+      "train loss:0.005817780038385451\n",
+      "train loss:0.0003174029900763067\n",
+      "train loss:0.0013565931488920774\n",
+      "train loss:0.012263204063853521\n",
+      "train loss:0.0015911154852730532\n",
+      "train loss:0.0010181584197499185\n",
+      "train loss:0.000173205745514178\n",
+      "train loss:0.0031253824481213394\n",
+      "train loss:0.0006736399087181361\n",
+      "train loss:0.003170712193491199\n",
+      "train loss:0.0037837985896096243\n",
+      "train loss:0.010018063276938232\n",
+      "train loss:0.0038999117848921433\n",
+      "train loss:0.007169157874517758\n",
+      "train loss:0.0008637039429257628\n",
+      "train loss:0.007122391500486835\n",
+      "train loss:0.000785821030949814\n",
+      "train loss:0.0008517214471944465\n",
+      "train loss:0.008535350017355446\n",
+      "train loss:0.003436114387201204\n",
+      "train loss:0.00736464258666075\n",
+      "train loss:0.0026657434234492066\n",
+      "train loss:0.00207170635719532\n",
+      "train loss:0.002591063236258171\n",
+      "train loss:0.004095769477109733\n",
+      "train loss:0.000818727987521407\n",
+      "train loss:0.016985582257783884\n",
+      "train loss:0.0010256372589947144\n",
+      "train loss:0.0022402691963525696\n",
+      "train loss:0.0018757607962420668\n",
+      "train loss:0.0012080124945980863\n",
+      "train loss:0.06646869612519865\n",
+      "train loss:0.01222395203739153\n",
+      "train loss:0.0023661091380658985\n",
+      "train loss:0.001192737325265361\n",
+      "train loss:0.0006885542536914347\n",
+      "train loss:0.059840588568021734\n",
+      "train loss:0.0024865834679340714\n",
+      "train loss:0.0024916250211850554\n",
+      "train loss:0.012730089244137195\n",
+      "train loss:0.0010415863376581306\n",
+      "train loss:0.004171950079650017\n",
+      "train loss:0.0021791265664371164\n",
+      "train loss:0.0021452380771580593\n",
+      "train loss:0.00045974254526157186\n",
+      "train loss:0.004687667507961629\n",
+      "train loss:0.004834500239361998\n",
+      "train loss:0.0007475785053549125\n",
+      "train loss:0.001587127327922911\n",
+      "train loss:0.0006263383707443327\n",
+      "train loss:0.0006914215735001404\n",
+      "train loss:0.000197586313798846\n",
+      "train loss:0.0006032875710297676\n",
+      "train loss:0.00045859451316247366\n",
+      "train loss:0.004774478223985788\n",
+      "train loss:0.0008630754630920948\n",
+      "train loss:0.008467928089551699\n",
+      "train loss:0.017108735534966907\n",
+      "train loss:0.0019874130003276953\n",
+      "train loss:0.0029359946416088125\n",
+      "train loss:0.0015849793010635571\n",
+      "train loss:0.0016655218105547564\n",
+      "train loss:0.018746778280269306\n",
+      "train loss:0.003591349485536093\n",
+      "train loss:0.004343503675626161\n",
+      "train loss:0.002202429015487362\n",
+      "train loss:0.0023916209157628086\n",
+      "train loss:0.0064786628813652955\n",
+      "train loss:0.0007635979279356949\n",
+      "train loss:0.00925522600680715\n",
+      "train loss:0.00028246039563725994\n",
+      "train loss:0.00012579601809845816\n",
+      "train loss:0.006015135454102479\n",
+      "train loss:0.0014145649231878746\n",
+      "train loss:0.0010939203186159363\n",
+      "train loss:0.006273886101262189\n",
+      "train loss:0.04313876670696634\n",
+      "train loss:0.002048486998104939\n",
+      "train loss:0.0013979076927302402\n",
+      "train loss:0.0028519054998319636\n",
+      "train loss:0.003823824113783883\n",
+      "train loss:0.0007699429319831344\n",
+      "train loss:0.0019815498481655525\n",
+      "train loss:0.0005325204893234533\n",
+      "train loss:0.004850710904972487\n",
+      "train loss:0.004545846069755554\n",
+      "train loss:0.007542589573681051\n",
+      "train loss:0.00012624809424397295\n",
+      "train loss:0.0003629340974318049\n",
+      "train loss:0.0012680528472458858\n",
+      "train loss:0.0004063951994502649\n",
+      "train loss:0.0016471535386753224\n",
+      "train loss:0.0002819172117012553\n",
+      "train loss:0.015938710514329803\n",
+      "train loss:0.002410100176688494\n",
+      "train loss:0.0023690618263767214\n",
+      "train loss:0.006894518552206745\n",
+      "train loss:0.0023248022481352844\n",
+      "train loss:0.001040359424038802\n",
+      "train loss:0.005294501854885249\n",
+      "train loss:0.006897279506605935\n",
+      "train loss:0.01212924912660337\n",
+      "train loss:0.0020910759225395994\n",
+      "train loss:0.00257216615662608\n",
+      "train loss:0.0012556570577405579\n",
+      "train loss:0.004650474155626983\n",
+      "train loss:0.019369879095983387\n",
+      "train loss:0.0018105978814562745\n",
+      "train loss:0.007849593273065987\n",
+      "train loss:0.0018804230752172122\n",
+      "train loss:0.008429172350673309\n",
+      "train loss:0.004173224711949093\n",
+      "train loss:0.004143046797648709\n",
+      "train loss:0.0066190700328991\n",
+      "train loss:0.011410392501168503\n",
+      "train loss:0.007649361039506569\n",
+      "train loss:0.0019876782517357482\n",
+      "train loss:0.002565774388868544\n",
+      "train loss:0.00025557370279504876\n",
+      "train loss:0.007192642644363046\n",
+      "train loss:0.0077797864090682145\n",
+      "train loss:0.0058033126977306585\n",
+      "train loss:0.002753305308674553\n",
+      "train loss:0.004238501541120249\n",
+      "train loss:0.03277275073801611\n",
+      "train loss:0.0005455713575318933\n",
+      "train loss:0.0022867074416387146\n",
+      "train loss:0.013309435337598147\n",
+      "train loss:0.001265433152991833\n",
+      "train loss:0.02312427736716475\n",
+      "train loss:0.0008595360472787434\n",
+      "train loss:0.005251352067168814\n",
+      "train loss:0.01459603082500627\n",
+      "train loss:0.002791523773322519\n",
+      "train loss:0.0031293107606583\n",
+      "=== epoch:15, train acc:0.994, test acc:0.984 ===\n",
+      "train loss:0.0032397652727585496\n",
+      "train loss:0.004069880482810998\n",
+      "train loss:0.00014774091642985176\n",
+      "train loss:0.0017019388553772858\n",
+      "train loss:0.0028487869778942966\n",
+      "train loss:0.0015184864892426352\n",
+      "train loss:0.018206211583550194\n",
+      "train loss:0.012137071681856671\n",
+      "train loss:0.0012335677587895684\n",
+      "train loss:0.002687848956572496\n",
+      "train loss:0.004865027539890431\n",
+      "train loss:0.017784591958044042\n",
+      "train loss:0.0001243463295919291\n",
+      "train loss:0.0011792907002461139\n",
+      "train loss:0.0034865637123615827\n",
+      "train loss:0.0007697953937801986\n",
+      "train loss:0.00017149653308470093\n",
+      "train loss:0.0022642096300299216\n",
+      "train loss:0.10478395475678187\n",
+      "train loss:0.00850718482977179\n",
+      "train loss:0.001332977442096335\n",
+      "train loss:0.0011070878584249274\n",
+      "train loss:0.0025487015845437134\n",
+      "train loss:0.0007211963087939484\n",
+      "train loss:0.0009627977744896289\n",
+      "train loss:0.0014303065465913509\n",
+      "train loss:0.001177517279781651\n",
+      "train loss:0.00891139328039403\n",
+      "train loss:0.0014580424544659771\n",
+      "train loss:0.0026434751311648326\n",
+      "train loss:0.0021242008186173177\n",
+      "train loss:0.0019403826694527671\n",
+      "train loss:0.00037803186129402213\n",
+      "train loss:0.002654958381697955\n",
+      "train loss:0.0362198603911933\n",
+      "train loss:0.0022755891645288525\n",
+      "train loss:0.0019221574283154739\n",
+      "train loss:0.0012893195092505438\n",
+      "train loss:0.0049034455987005034\n",
+      "train loss:0.0006051615747182185\n",
+      "train loss:0.009959404008937297\n",
+      "train loss:0.00018091808681250517\n",
+      "train loss:0.006411838042917883\n",
+      "train loss:0.0006358009419141036\n",
+      "train loss:0.0005743403666752179\n",
+      "train loss:0.01570882943205831\n",
+      "train loss:0.0006457342698241471\n",
+      "train loss:0.012825001958553135\n",
+      "train loss:0.021907852242024344\n",
+      "train loss:0.009776476962943264\n",
+      "train loss:0.0029773894240806504\n",
+      "train loss:0.002155859833138858\n",
+      "train loss:0.0008236499099058647\n",
+      "train loss:0.00041102779241130477\n",
+      "train loss:0.005110277888505067\n",
+      "train loss:0.007044593662540717\n",
+      "train loss:0.000593569318563984\n",
+      "train loss:0.004999790470062392\n",
+      "train loss:0.0020686147730927373\n",
+      "train loss:0.0017184488267177613\n",
+      "train loss:0.004954746432105282\n",
+      "train loss:0.006040980165965547\n",
+      "train loss:0.005610698891911911\n",
+      "train loss:0.002462481518760018\n",
+      "train loss:0.001844245138256634\n",
+      "train loss:0.0005562130403474589\n",
+      "train loss:0.0033925186321522266\n",
+      "train loss:0.00041453062954254876\n",
+      "train loss:0.0028027498362907443\n",
+      "train loss:0.0007037803461619874\n",
+      "train loss:0.009689200531634568\n",
+      "train loss:0.0004892690087086578\n",
+      "train loss:0.022393429963207177\n",
+      "train loss:0.002378723308020015\n",
+      "train loss:0.0005989655802556794\n",
+      "train loss:0.001984930940481826\n",
+      "train loss:0.01626867890651928\n",
+      "train loss:0.0015232208764589868\n",
+      "train loss:0.0007515004849794966\n",
+      "train loss:0.020076773556386533\n",
+      "train loss:0.003941445651600895\n",
+      "train loss:0.0018511873623193087\n",
+      "train loss:0.0012576800629674064\n",
+      "train loss:0.0014264613838432703\n",
+      "train loss:0.0058322136155685225\n",
+      "train loss:0.00018133917798791592\n",
+      "train loss:0.0008946526049682865\n",
+      "train loss:0.0012911467384732337\n",
+      "train loss:0.0008475629152159916\n",
+      "train loss:0.0023157356258634836\n",
+      "train loss:0.004837009235218201\n",
+      "train loss:0.0013811876698848608\n",
+      "train loss:0.0006228320635094974\n",
+      "train loss:0.006042413148381878\n",
+      "train loss:0.0019092754028198983\n",
+      "train loss:0.0023470179198962255\n",
+      "train loss:0.0010827980826463165\n",
+      "train loss:0.0014612148252563177\n",
+      "train loss:0.0019480499960490977\n",
+      "train loss:0.0006423088699981845\n",
+      "train loss:0.0017620518322407936\n",
+      "train loss:0.0007477288284175503\n",
+      "train loss:0.0008761152098121679\n",
+      "train loss:0.0008967295456207468\n",
+      "train loss:0.0030121914745853983\n",
+      "train loss:0.0015308332056334453\n",
+      "train loss:0.007227591618777419\n",
+      "train loss:0.012934263883231092\n",
+      "train loss:0.000259598110728978\n",
+      "train loss:0.0004399037801446692\n",
+      "train loss:0.0004101721570428549\n",
+      "train loss:0.0014927183439499348\n",
+      "train loss:0.0008159525487921079\n",
+      "train loss:0.0030760225554139565\n",
+      "train loss:0.0024018555584340226\n",
+      "train loss:0.0005324671512791994\n",
+      "train loss:0.008285937998975753\n",
+      "train loss:0.0010416067354028407\n",
+      "train loss:0.00620132059858686\n",
+      "train loss:0.00039840775547268987\n",
+      "train loss:0.001518087431911344\n",
+      "train loss:0.001374277060433228\n",
+      "train loss:0.0015605860744442085\n",
+      "train loss:0.0036772836652044076\n",
+      "train loss:0.013576082599494634\n",
+      "train loss:0.0010649452160089309\n",
+      "train loss:0.0008412167106612845\n",
+      "train loss:0.004531711055187284\n",
+      "train loss:0.0015539702965054849\n",
+      "train loss:0.0021715976977655835\n",
+      "train loss:0.0020115545256626506\n",
+      "train loss:0.0005329583569394397\n",
+      "train loss:0.002863148611530494\n",
+      "train loss:0.000963897235137716\n",
+      "train loss:0.0012205811007282691\n",
+      "train loss:0.003963101640933106\n",
+      "train loss:0.0031266905585268655\n",
+      "train loss:0.004176583033816756\n",
+      "train loss:6.11482283419242e-05\n",
+      "train loss:0.001615874212020473\n",
+      "train loss:0.006036249005501351\n",
+      "train loss:0.0005749931854839054\n",
+      "train loss:0.000337385224665875\n",
+      "train loss:0.004289313053094713\n",
+      "train loss:0.00025201085991862767\n",
+      "train loss:0.017930301528398324\n",
+      "train loss:0.002203353056297415\n",
+      "train loss:0.012913011859252452\n",
+      "train loss:0.002444936405183138\n",
+      "train loss:0.00267173531880079\n",
+      "train loss:0.001149138068612488\n",
+      "train loss:0.003509722297435544\n",
+      "train loss:0.00047818563540041734\n",
+      "train loss:0.002644322885023561\n",
+      "train loss:0.0021834677057977494\n",
+      "train loss:0.00537676759773726\n",
+      "train loss:0.0017194839288876206\n",
+      "train loss:0.0009477471771531515\n",
+      "train loss:0.00431082847485732\n",
+      "train loss:0.00013657014800289344\n",
+      "train loss:0.001013239160801463\n",
+      "train loss:0.0004132568818514193\n",
+      "train loss:0.001105283225804852\n",
+      "train loss:0.05279060582055919\n",
+      "train loss:0.00018840654507833063\n",
+      "train loss:0.0009011806921334514\n",
+      "train loss:0.003634458530591142\n",
+      "train loss:0.0031766672327607137\n",
+      "train loss:0.0008335979389555302\n",
+      "train loss:0.0005890419395142602\n",
+      "train loss:0.0021724055602451633\n",
+      "train loss:0.012631438742093236\n",
+      "train loss:0.0006569043256087882\n",
+      "train loss:0.001887215628862682\n",
+      "train loss:0.0005848918753411037\n",
+      "train loss:0.0033840696618567608\n",
+      "train loss:0.031607517088349135\n",
+      "train loss:0.0023361545598157164\n",
+      "train loss:0.004523382104104986\n",
+      "train loss:0.0006649841731433784\n",
+      "train loss:0.00876175639998079\n",
+      "train loss:0.0016568994312511093\n",
+      "train loss:0.0015310565227964138\n",
+      "train loss:0.0010777861559464934\n",
+      "train loss:0.013214804740295391\n",
+      "train loss:0.010138958265182374\n",
+      "train loss:0.0011723146412911832\n",
+      "train loss:0.00017118263436239977\n",
+      "train loss:0.0014800170874957428\n",
+      "train loss:0.00029873415869749806\n",
+      "train loss:0.0007411424943288875\n",
+      "train loss:0.0007696603752259678\n",
+      "train loss:0.0021253467568931924\n",
+      "train loss:0.0032628952838687055\n",
+      "train loss:0.0022483547877185563\n",
+      "train loss:0.00014016657014793155\n",
+      "train loss:0.000592034074989929\n",
+      "train loss:0.0016537919347121965\n",
+      "train loss:0.0008265552788093161\n",
+      "train loss:0.00027832685010669757\n",
+      "train loss:0.0025427614397676725\n",
+      "train loss:0.001357081338112833\n",
+      "train loss:9.896415116583715e-05\n",
+      "train loss:0.003522650986705012\n",
+      "train loss:0.004192343324999707\n",
+      "train loss:0.0010417546740079673\n",
+      "train loss:0.0006128683387921562\n",
+      "train loss:0.00031883712666903646\n",
+      "train loss:0.004271401799395457\n",
+      "train loss:0.0032493351208288984\n",
+      "train loss:0.0018749473045618\n",
+      "train loss:0.0053868921007219515\n",
+      "train loss:0.0006194863463102138\n",
+      "train loss:0.002095569897398598\n",
+      "train loss:0.0011066386252688546\n",
+      "train loss:0.0003848392576072463\n",
+      "train loss:0.0023226686383132477\n",
+      "train loss:0.0006082524726005494\n",
+      "train loss:0.0036564569027248174\n",
+      "train loss:0.0001509829338184141\n",
+      "train loss:5.910504826861739e-05\n",
+      "train loss:0.013270800522210807\n",
+      "train loss:0.023354275427322935\n",
+      "train loss:0.0018698385644993806\n",
+      "train loss:0.002054251460893282\n",
+      "train loss:8.55644217332018e-05\n",
+      "train loss:0.001999307951774275\n",
+      "train loss:0.0020957704810561167\n",
+      "train loss:0.0028993007264821254\n",
+      "train loss:0.002033529541365454\n",
+      "train loss:0.05516769204651508\n",
+      "train loss:0.0016669193915937015\n",
+      "train loss:0.004074906083439264\n",
+      "train loss:0.007937393597695462\n",
+      "train loss:0.0009644260337239501\n",
+      "train loss:0.004961963854878944\n",
+      "train loss:0.014006375052523558\n",
+      "train loss:0.00045552930435093056\n",
+      "train loss:0.006960717145860786\n",
+      "train loss:0.007619347918250787\n",
+      "train loss:0.00044508779776097923\n",
+      "train loss:0.0005567659839391684\n",
+      "train loss:0.000745837099187966\n",
+      "train loss:0.007102999803457109\n",
+      "train loss:0.005119335175397117\n",
+      "train loss:0.0006464202347690386\n",
+      "train loss:0.0004660007313578854\n",
+      "train loss:0.0026422873026192744\n",
+      "train loss:0.0007658895221148679\n",
+      "train loss:0.0017283495657235343\n",
+      "train loss:0.00046403620482921113\n",
+      "train loss:0.0064210605671641725\n",
+      "train loss:0.006382575574174404\n",
+      "train loss:0.024985067998522962\n",
+      "train loss:0.007328595057183256\n",
+      "train loss:0.0005395877556882035\n",
+      "train loss:0.004586463071017983\n",
+      "train loss:0.0008678194517385441\n",
+      "train loss:0.016336167974458013\n",
+      "train loss:0.040273559816740905\n",
+      "train loss:0.00952078160019387\n",
+      "train loss:0.0007435976740001336\n",
+      "train loss:0.008462716775750346\n",
+      "train loss:0.0013838108192954256\n",
+      "train loss:9.662045739809445e-05\n",
+      "train loss:0.0057015540753835095\n",
+      "train loss:0.003489350894164405\n",
+      "train loss:0.003776877531654635\n",
+      "train loss:0.002872192107525548\n",
+      "train loss:0.0016731321560852345\n",
+      "train loss:0.0039789844901829425\n",
+      "train loss:0.001026162395805878\n",
+      "train loss:0.014834593024079109\n",
+      "train loss:0.0022117359524764903\n",
+      "train loss:0.0010057126763719758\n",
+      "train loss:0.005250667218186027\n",
+      "train loss:0.006966343663974253\n",
+      "train loss:0.0020935583531454163\n",
+      "train loss:0.004297882580876642\n",
+      "train loss:0.0017565258100410298\n",
+      "train loss:0.0043123849910263095\n",
+      "train loss:0.0017941747011057138\n",
+      "train loss:0.050251976095517285\n",
+      "train loss:0.0004952535824295501\n",
+      "train loss:0.0032239779795580296\n",
+      "train loss:0.0003496065032716525\n",
+      "train loss:0.005109187307947188\n",
+      "train loss:0.004520531670992093\n",
+      "train loss:0.001374841434778139\n",
+      "train loss:0.00012962151835203532\n",
+      "train loss:0.001108847843357278\n",
+      "train loss:0.004199221653685792\n",
+      "train loss:0.0020452193241171485\n",
+      "train loss:0.00044871880019535585\n",
+      "train loss:0.0009326387046016597\n",
+      "train loss:0.0026849142228444196\n",
+      "train loss:0.014175576104812897\n",
+      "train loss:0.01318682984224991\n",
+      "train loss:0.002383760509084649\n",
+      "train loss:0.0032743360794909625\n",
+      "train loss:0.0007697163102196153\n",
+      "train loss:0.0041253147492214025\n",
+      "train loss:0.0073884988761018944\n",
+      "train loss:0.0025818364964714994\n",
+      "train loss:0.0011502478470660482\n",
+      "train loss:0.00031104601302914665\n",
+      "train loss:0.001147393152140801\n",
+      "train loss:0.00016371566647514066\n",
+      "train loss:0.0004860096154661885\n",
+      "train loss:0.0034035626216197155\n",
+      "train loss:0.0012772843683616828\n",
+      "train loss:0.014502519836647\n",
+      "train loss:0.00018674286308961158\n",
+      "train loss:0.03110734140864995\n",
+      "train loss:0.003509666690566313\n",
+      "train loss:0.002911356954559855\n",
+      "train loss:0.002926287028928096\n",
+      "train loss:0.00325186855751043\n",
+      "train loss:0.0037750757806304124\n",
+      "train loss:0.007771394608666566\n",
+      "train loss:0.000920974963061045\n",
+      "train loss:0.007959062346072278\n",
+      "train loss:0.0016831348631913038\n",
+      "train loss:0.0026424477461654983\n",
+      "train loss:0.004620387919983737\n",
+      "train loss:0.0002750054243689916\n",
+      "train loss:0.00473176093339547\n",
+      "train loss:0.0025692844926373227\n",
+      "train loss:0.010087482147970504\n",
+      "train loss:0.00257051167244999\n",
+      "train loss:0.002078865814366453\n",
+      "train loss:0.0022056652450955006\n",
+      "train loss:0.006720680180564067\n",
+      "train loss:5.031576150143758e-05\n",
+      "train loss:0.0007032780026483589\n",
+      "train loss:0.0016747756439347951\n",
+      "train loss:0.003426818616476097\n",
+      "train loss:0.0011437602577595908\n",
+      "train loss:0.00246796743101373\n",
+      "train loss:0.0012658223265285063\n",
+      "train loss:0.001965093327538609\n",
+      "train loss:0.0016895495354845478\n",
+      "train loss:0.012066923045736835\n",
+      "train loss:0.003518572646002807\n",
+      "train loss:0.0029549654590746866\n",
+      "train loss:0.0014314379129631615\n",
+      "train loss:0.007154751176150652\n",
+      "train loss:0.027475039636954225\n",
+      "train loss:0.00023632599603151365\n",
+      "train loss:0.0008203178224716727\n",
+      "train loss:0.013880161279119674\n",
+      "train loss:0.0028527484744491688\n",
+      "train loss:0.0017973643775943302\n",
+      "train loss:0.001924820927642422\n",
+      "train loss:0.0003352054506370575\n",
+      "train loss:0.0009453839435825974\n",
+      "train loss:0.00016854848476129098\n",
+      "train loss:0.004050578936581221\n",
+      "train loss:0.00190650339327381\n",
+      "train loss:0.0008354348444733786\n",
+      "train loss:0.0006027773718789842\n",
+      "train loss:0.00560517579491598\n",
+      "train loss:0.001463112451685027\n",
+      "train loss:0.001302919205931852\n",
+      "train loss:0.0010174484345350765\n",
+      "train loss:0.00024785999547995364\n",
+      "train loss:0.00177469308840798\n",
+      "train loss:0.0007782406396266442\n",
+      "train loss:8.298801367254108e-05\n",
+      "train loss:0.01360051665402478\n",
+      "train loss:0.0011569466108887992\n",
+      "train loss:0.00040091448288755724\n",
+      "train loss:0.001646497658573734\n",
+      "train loss:0.0012416519202798494\n",
+      "train loss:0.002614638926450202\n",
+      "train loss:0.0001654947419235437\n",
+      "train loss:0.03333953643148731\n",
+      "train loss:0.0002890731403324747\n",
+      "train loss:0.007709238848084872\n",
+      "train loss:0.00012455753913503126\n",
+      "train loss:0.0008690512086441622\n",
+      "train loss:0.0020776777092315446\n",
+      "train loss:0.0008435733869728983\n",
+      "train loss:0.00033385604921736846\n",
+      "train loss:0.001389125653537837\n",
+      "train loss:0.00014651379490027223\n",
+      "train loss:0.0006649912187250414\n",
+      "train loss:0.01199631873712488\n",
+      "train loss:0.01116894209974776\n",
+      "train loss:0.0015984270824816271\n",
+      "train loss:0.0014910522363014724\n",
+      "train loss:0.002302526044541847\n",
+      "train loss:0.012897047756706937\n",
+      "train loss:0.0006417821675530823\n",
+      "train loss:0.00027926860513719144\n",
+      "train loss:0.007916396105981043\n",
+      "train loss:0.004328335968992134\n",
+      "train loss:0.000793344300577205\n",
+      "train loss:0.005093071565315374\n",
+      "train loss:0.00060635409944068\n",
+      "train loss:0.006889395403171576\n",
+      "train loss:0.0035649825530311785\n",
+      "train loss:0.0009713347595906991\n",
+      "train loss:0.006764570084752596\n",
+      "train loss:0.005430291439584025\n",
+      "train loss:0.005091207490075364\n",
+      "train loss:0.004927444193801179\n",
+      "train loss:0.0023639134479317187\n",
+      "train loss:0.0047124383246257735\n",
+      "train loss:0.0033614137098941055\n",
+      "train loss:0.005975870138350484\n",
+      "train loss:0.0023914690451406497\n",
+      "train loss:0.0012966057964916159\n",
+      "train loss:0.002000862948660208\n",
+      "train loss:0.0015131322967830158\n",
+      "train loss:0.0037789041288543964\n",
+      "train loss:7.20482470945194e-05\n",
+      "train loss:0.003532597191426754\n",
+      "train loss:0.0025696096851964226\n",
+      "train loss:0.0025455162942231136\n",
+      "train loss:0.0012257947652909598\n",
+      "train loss:0.006377058780187567\n",
+      "train loss:0.002916644027055661\n",
+      "train loss:0.00013931507964608288\n",
+      "train loss:0.0014638295092363007\n",
+      "train loss:0.006780650672110914\n",
+      "train loss:0.0009405411350653637\n",
+      "train loss:0.00022367274227091269\n",
+      "train loss:0.0015879136414453752\n",
+      "train loss:0.003351955688505456\n",
+      "train loss:0.00027179524223305135\n",
+      "train loss:0.001748739766062308\n",
+      "train loss:0.00519303376435561\n",
+      "train loss:0.0002986729042965361\n",
+      "train loss:0.0011755886357787957\n",
+      "train loss:0.0037426693872784794\n",
+      "train loss:0.0006652476149964201\n",
+      "train loss:0.006567251930869841\n",
+      "train loss:0.0029440988232235714\n",
+      "train loss:0.0036316924567891483\n",
+      "train loss:0.02408921852277033\n",
+      "train loss:0.00033835439203977176\n",
+      "train loss:0.0008187807609979285\n",
+      "train loss:0.0045644048945391955\n",
+      "train loss:0.00036497940835896804\n",
+      "train loss:0.0035277807865424053\n",
+      "train loss:0.002816254641285686\n",
+      "train loss:0.00045756589798155343\n",
+      "train loss:0.001309266820019097\n",
+      "train loss:0.0003283247456655993\n",
+      "train loss:0.035803793074377616\n",
+      "train loss:0.000935570039279045\n",
+      "train loss:0.001985220627864091\n",
+      "train loss:0.0032293865481290065\n",
+      "train loss:0.0008438759222931399\n",
+      "train loss:0.0069832839185171\n",
+      "train loss:0.004770340721169265\n",
+      "train loss:0.002149324494372427\n",
+      "train loss:0.008847241543523496\n",
+      "train loss:0.0006860084424948254\n",
+      "train loss:0.0014727939191501104\n",
+      "train loss:0.005229369586590409\n",
+      "train loss:0.001888504217129601\n",
+      "train loss:0.005034672204300972\n",
+      "train loss:0.002119287559883147\n",
+      "train loss:0.001460439437157552\n",
+      "train loss:0.004983467548793089\n",
+      "train loss:0.0016992342891195373\n",
+      "train loss:0.0029445673737929627\n",
+      "train loss:0.0008191143962910349\n",
+      "train loss:0.0004221466167698104\n",
+      "train loss:0.004398736621388166\n",
+      "train loss:0.004766923309860863\n",
+      "train loss:0.001669845292853565\n",
+      "train loss:0.0022741805437126355\n",
+      "train loss:0.00011405659733277757\n",
+      "train loss:0.0030974511524923675\n",
+      "train loss:0.0007385930614406047\n",
+      "train loss:0.003078320824618205\n",
+      "train loss:0.0009121267488487423\n",
+      "train loss:0.0008530157776647728\n",
+      "train loss:0.0014046889466541284\n",
+      "train loss:0.0028145323591638773\n",
+      "train loss:0.0020682521507103388\n",
+      "train loss:0.0021494709800742007\n",
+      "train loss:0.003965245363734608\n",
+      "train loss:0.0012100132597256087\n",
+      "train loss:0.008521144173531607\n",
+      "train loss:0.004782120175471135\n",
+      "train loss:0.00015674340757336347\n",
+      "train loss:0.0010591915316991323\n",
+      "train loss:0.0030948654981382954\n",
+      "train loss:0.00023976735918534398\n",
+      "train loss:0.00116060441445797\n",
+      "train loss:0.0004800178777158333\n",
+      "train loss:0.004102238766418144\n",
+      "train loss:0.0017765155341650338\n",
+      "train loss:0.000554845808678729\n",
+      "train loss:0.004411708048245148\n",
+      "train loss:0.0010534810290256488\n",
+      "train loss:0.000573282461861122\n",
+      "train loss:0.0018156051621244956\n",
+      "train loss:0.003488266852952931\n",
+      "train loss:9.629773473966421e-05\n",
+      "train loss:0.002016283207163754\n",
+      "train loss:0.006417496329146019\n",
+      "train loss:0.0003137740042194626\n",
+      "train loss:0.0009290925004129215\n",
+      "train loss:0.0007580319412089535\n",
+      "train loss:0.003521604147388419\n",
+      "train loss:0.0007373570089742484\n",
+      "train loss:0.007485059348026787\n",
+      "train loss:0.0022309274066350223\n",
+      "train loss:0.001193813726025917\n",
+      "train loss:0.008420775603332625\n",
+      "train loss:0.00016920103957028957\n",
+      "train loss:0.0005596640261361732\n",
+      "train loss:0.001066896131141084\n",
+      "train loss:5.2807764374587664e-05\n",
+      "train loss:0.0011746401534522266\n",
+      "train loss:0.003129579805164258\n",
+      "train loss:0.036389189733620986\n",
+      "train loss:0.0021795984952361364\n",
+      "train loss:0.007551976785314718\n",
+      "train loss:0.0017977449087137431\n",
+      "train loss:0.0012209020508019551\n",
+      "train loss:0.0009989640111976387\n",
+      "train loss:0.012586910462519568\n",
+      "train loss:0.0004132185932137461\n",
+      "train loss:0.0006075475533815895\n",
+      "train loss:9.033301292843979e-05\n",
+      "train loss:0.012080749083982925\n",
+      "train loss:0.0011423884652849608\n",
+      "train loss:0.0006193408588896841\n",
+      "train loss:0.0019703890971923712\n",
+      "train loss:0.003164754735209926\n",
+      "train loss:0.0003175871446030883\n",
+      "train loss:0.003949343997281708\n",
+      "train loss:0.002312819110263616\n",
+      "train loss:0.00033162498333510175\n",
+      "train loss:0.0027202222158457497\n",
+      "train loss:0.0008037153846963345\n",
+      "train loss:0.000411709853243364\n",
+      "train loss:0.0007693684653926572\n",
+      "train loss:0.002349695047982958\n",
+      "train loss:0.0037796330517299425\n",
+      "train loss:0.0005110720732132507\n",
+      "train loss:0.003275777827769946\n",
+      "train loss:0.0003108867934166341\n",
+      "train loss:2.824535882840313e-05\n",
+      "train loss:0.003912413841903415\n",
+      "train loss:0.00038715470807641604\n",
+      "train loss:0.001170421606942369\n",
+      "train loss:0.00018587257158546484\n",
+      "train loss:0.00018985566617107073\n",
+      "train loss:0.001145045488801332\n",
+      "train loss:0.00017396231963100154\n",
+      "train loss:0.0004350217506199496\n",
+      "train loss:0.0028842541715450665\n",
+      "train loss:0.0038934214681696474\n",
+      "train loss:0.0006174568329045785\n",
+      "train loss:0.0008887331702994999\n",
+      "train loss:0.012318657312488457\n",
+      "train loss:0.011990224711382969\n",
+      "train loss:9.615529367545412e-05\n",
+      "train loss:0.002165072904639889\n",
+      "train loss:0.0002745489861930194\n",
+      "train loss:1.335949049202723e-05\n",
+      "train loss:0.005495751490049925\n",
+      "train loss:0.012304213231563213\n",
+      "train loss:0.0006244919749324867\n",
+      "train loss:0.0008759182103681884\n",
+      "train loss:0.0017470344629200816\n",
+      "train loss:0.0017766012302765151\n",
+      "train loss:0.0003321117842415633\n",
+      "train loss:0.0007609236884456637\n",
+      "train loss:0.005168600445731498\n",
+      "train loss:0.0015683671866570418\n",
+      "train loss:0.0005129055882059894\n",
+      "train loss:0.001093723995719312\n",
+      "train loss:0.00020641160132522918\n",
+      "train loss:0.0007050944149724417\n",
+      "train loss:0.0020255939970736567\n",
+      "train loss:0.003476789827492332\n",
+      "train loss:0.00138531481028189\n",
+      "train loss:0.000548936103882458\n",
+      "train loss:0.0001046651814870439\n",
+      "train loss:0.0015665962158093636\n",
+      "train loss:0.0006460389182245152\n",
+      "train loss:0.00019757142643653412\n",
+      "train loss:0.0011534211514478909\n",
+      "train loss:0.0031882672696765374\n",
+      "train loss:0.0019862522384414137\n",
+      "train loss:0.004427588329344357\n",
+      "train loss:0.0014514753534261172\n",
+      "train loss:0.000533157184069062\n",
+      "train loss:0.04693709558486421\n",
+      "train loss:0.0012170704085740195\n",
+      "train loss:0.003567263641136259\n",
+      "train loss:0.0014561013129257385\n",
+      "=== epoch:16, train acc:0.999, test acc:0.988 ===\n",
+      "train loss:0.0017582999220133804\n",
+      "train loss:0.003944548977933845\n",
+      "train loss:0.00034179019529944883\n",
+      "train loss:0.0016876584190863436\n",
+      "train loss:0.0005742751450549771\n",
+      "train loss:0.0022509897718930847\n",
+      "train loss:0.0002184306013722232\n",
+      "train loss:0.0009251452251222573\n",
+      "train loss:0.0002999587244051553\n",
+      "train loss:0.0045073437342237925\n",
+      "train loss:0.00022103040295068282\n",
+      "train loss:0.0009300713541218064\n",
+      "train loss:0.00018834029684574322\n",
+      "train loss:0.0029040068030556114\n",
+      "train loss:0.00592657949217719\n",
+      "train loss:0.0008357148312299384\n",
+      "train loss:0.001190954476748208\n",
+      "train loss:0.0012680078374061124\n",
+      "train loss:0.0012478929491602292\n",
+      "train loss:0.001249855591380049\n",
+      "train loss:0.0050981104492628234\n",
+      "train loss:0.007271989268273644\n",
+      "train loss:8.956981166969263e-05\n",
+      "train loss:0.0005593242115863546\n",
+      "train loss:0.002524970184955055\n",
+      "train loss:7.769092783030565e-05\n",
+      "train loss:0.0016768321616219904\n",
+      "train loss:0.0006034253460861631\n",
+      "train loss:0.004875581300699698\n",
+      "train loss:0.0020154992573016618\n",
+      "train loss:0.055170889722022484\n",
+      "train loss:0.00254652197728088\n",
+      "train loss:0.0009938312623715235\n",
+      "train loss:0.0008233282106526456\n",
+      "train loss:0.002140740039961363\n",
+      "train loss:0.00033452334066135105\n",
+      "train loss:0.0024403992144879567\n",
+      "train loss:0.0017089366647020444\n",
+      "train loss:0.0015275352063325643\n",
+      "train loss:0.0030498851890066207\n",
+      "train loss:0.0006679394060945586\n",
+      "train loss:0.0008070849592829454\n",
+      "train loss:0.0012855822623570802\n",
+      "train loss:0.00027473057323224265\n",
+      "train loss:0.000780956459852467\n",
+      "train loss:0.007329304896561825\n",
+      "train loss:0.002201373562891596\n",
+      "train loss:0.00011370235230840763\n",
+      "train loss:0.0008538547443069684\n",
+      "train loss:0.0037016821368557372\n",
+      "train loss:0.0031228926610491264\n",
+      "train loss:0.0005693598919912659\n",
+      "train loss:0.003878199691314846\n",
+      "train loss:0.0017468256821229135\n",
+      "train loss:0.0036823655130602\n",
+      "train loss:0.002301840243764097\n",
+      "train loss:0.0004524233079043383\n",
+      "train loss:0.0025090371984933047\n",
+      "train loss:0.0014709533684511664\n",
+      "train loss:0.0006604960918876195\n",
+      "train loss:0.00013068810750061475\n",
+      "train loss:0.0008174938758322008\n",
+      "train loss:0.0034808976864879756\n",
+      "train loss:0.001110555640347111\n",
+      "train loss:0.0011971313405386636\n",
+      "train loss:0.0011352246876902906\n",
+      "train loss:0.0013745899691443594\n",
+      "train loss:0.0008721693111758903\n",
+      "train loss:0.00011737519755098124\n",
+      "train loss:3.2071323175403185e-05\n",
+      "train loss:0.001865514864363246\n",
+      "train loss:0.0025040547002632123\n",
+      "train loss:0.0012465855224214903\n",
+      "train loss:0.0013833009445397984\n",
+      "train loss:0.0007551077129626952\n",
+      "train loss:0.0003948994611025018\n",
+      "train loss:0.0036115686905245793\n",
+      "train loss:0.007363465502324944\n",
+      "train loss:0.0009883786410563452\n",
+      "train loss:0.0004847866299539164\n",
+      "train loss:0.000337273244939767\n",
+      "train loss:0.00031927411490918\n",
+      "train loss:0.0003935408964765294\n",
+      "train loss:0.003034038176177944\n",
+      "train loss:0.000272937824403089\n",
+      "train loss:0.0004793279179409166\n",
+      "train loss:0.00022721823334452326\n",
+      "train loss:0.0015269798516946975\n",
+      "train loss:0.0009197403730864173\n",
+      "train loss:0.0009706448458426243\n",
+      "train loss:0.0006715135887770518\n",
+      "train loss:0.0001664895462948491\n",
+      "train loss:0.000882211149955597\n",
+      "train loss:0.0017723650428553119\n",
+      "train loss:0.00039165801010620046\n",
+      "train loss:0.0014286367568731196\n",
+      "train loss:0.0006118534639169781\n",
+      "train loss:0.0015447965149898621\n",
+      "train loss:0.00031711105658042494\n",
+      "train loss:0.0027479721655195815\n",
+      "train loss:0.0032422317508318467\n",
+      "train loss:0.0014573764117730612\n",
+      "train loss:0.015676339475675583\n",
+      "train loss:0.002474835805924034\n",
+      "train loss:0.0021244738228844854\n",
+      "train loss:0.0011609053891119547\n",
+      "train loss:0.0026127637348979747\n",
+      "train loss:0.00275008774503845\n",
+      "train loss:0.0006772195836109518\n",
+      "train loss:0.0004592502365021609\n",
+      "train loss:0.0008958528316442516\n",
+      "train loss:0.00041123800850808017\n",
+      "train loss:0.0001285821629435828\n",
+      "train loss:0.002783639448866432\n",
+      "train loss:0.0005944740386506741\n",
+      "train loss:0.00023207309886815784\n",
+      "train loss:0.0012413649515487508\n",
+      "train loss:0.0014188300160429437\n",
+      "train loss:0.00036916182329300924\n",
+      "train loss:0.0009811682800667306\n",
+      "train loss:0.000502448168108315\n",
+      "train loss:0.0006138272765193181\n",
+      "train loss:0.00026875212251062223\n",
+      "train loss:0.0006831757182430222\n",
+      "train loss:0.0002543936523420126\n",
+      "train loss:0.0023158394438185055\n",
+      "train loss:0.0036251313037945765\n",
+      "train loss:0.0012230609793988235\n",
+      "train loss:0.0013822492799123694\n",
+      "train loss:0.0011278754518439403\n",
+      "train loss:0.004072033366990291\n",
+      "train loss:2.1831930802937543e-05\n",
+      "train loss:0.0025226509158849013\n",
+      "train loss:0.004537667803858763\n",
+      "train loss:0.001515207826721542\n",
+      "train loss:0.0006514215432117766\n",
+      "train loss:0.00019520649341886923\n",
+      "train loss:0.0012872587120796705\n",
+      "train loss:0.0025081301699919727\n",
+      "train loss:0.0012479760255684857\n",
+      "train loss:0.0001272634255515174\n",
+      "train loss:0.001552707358276282\n",
+      "train loss:0.0004263846965507406\n",
+      "train loss:0.0017716168412963665\n",
+      "train loss:0.00047165329192560526\n",
+      "train loss:0.0006879467824708915\n",
+      "train loss:0.0005624284375149542\n",
+      "train loss:0.00017727535449484243\n",
+      "train loss:0.003031854664581723\n",
+      "train loss:0.061255431162352485\n",
+      "train loss:0.00021056744081690586\n",
+      "train loss:0.003928800060561956\n",
+      "train loss:0.00040067958461272133\n",
+      "train loss:0.0002730280397015682\n",
+      "train loss:0.002552533253897774\n",
+      "train loss:0.01987888715213287\n",
+      "train loss:0.0012922013729276373\n",
+      "train loss:0.0014978312156634175\n",
+      "train loss:0.0007450877058078069\n",
+      "train loss:0.0007864617401071732\n",
+      "train loss:0.0014469963536163167\n",
+      "train loss:0.0004676853779463658\n",
+      "train loss:0.003923863386657001\n",
+      "train loss:0.0016561550249064682\n",
+      "train loss:0.0005063724679687715\n",
+      "train loss:6.61365942289523e-05\n",
+      "train loss:0.0011223890025478744\n",
+      "train loss:0.0023038356351388735\n",
+      "train loss:0.0007379328532028323\n",
+      "train loss:0.0003351773841015356\n",
+      "train loss:0.0011313128966597209\n",
+      "train loss:0.003939320120263297\n",
+      "train loss:0.0006131821168653635\n",
+      "train loss:0.0009075508456291033\n",
+      "train loss:0.001943409172646157\n",
+      "train loss:0.021096221628920553\n",
+      "train loss:0.0012439735321103824\n",
+      "train loss:0.00043275675031407857\n",
+      "train loss:0.00015337931479154172\n",
+      "train loss:0.0007024813691949569\n",
+      "train loss:0.0014505683059522418\n",
+      "train loss:0.00012362675359533373\n",
+      "train loss:0.0009559119847648871\n",
+      "train loss:0.0005526640335101947\n",
+      "train loss:0.001751672218620095\n",
+      "train loss:0.00482312479874069\n",
+      "train loss:0.0002485399021659373\n",
+      "train loss:0.0002191697858375134\n",
+      "train loss:4.455350453267918e-05\n",
+      "train loss:0.004389614726515932\n",
+      "train loss:4.19401199020098e-05\n",
+      "train loss:0.00044401238165783026\n",
+      "train loss:0.0023912466478523105\n",
+      "train loss:0.00024291264990751636\n",
+      "train loss:0.00024324282725467423\n",
+      "train loss:0.00012425127457682785\n",
+      "train loss:0.0006831142555904809\n",
+      "train loss:0.0019426883681161237\n",
+      "train loss:0.002616235767491727\n",
+      "train loss:0.0012204471425146782\n",
+      "train loss:4.584917229063972e-05\n",
+      "train loss:0.012483097155400509\n",
+      "train loss:0.00016506299485325156\n",
+      "train loss:0.0007643547281598171\n",
+      "train loss:0.003773931725416669\n",
+      "train loss:0.00336399729348996\n",
+      "train loss:0.014174176788706094\n",
+      "train loss:0.004087650656845692\n",
+      "train loss:0.000537076321613296\n",
+      "train loss:0.06350927042597371\n",
+      "train loss:0.0017329405625156\n",
+      "train loss:0.0011644472398342634\n",
+      "train loss:0.00022153055068402852\n",
+      "train loss:0.0016927881870073584\n",
+      "train loss:0.0021618932113039667\n",
+      "train loss:0.0019815455463885936\n",
+      "train loss:0.006274249343055515\n",
+      "train loss:0.0006253235014806748\n",
+      "train loss:0.0004767946095994162\n",
+      "train loss:0.0016558060877493732\n",
+      "train loss:0.0013554022572372052\n",
+      "train loss:0.035710302521773055\n",
+      "train loss:0.00029996396583844507\n",
+      "train loss:0.00216233564216746\n",
+      "train loss:0.0023372800735024356\n",
+      "train loss:0.006600637595971631\n",
+      "train loss:0.0027969793885827695\n",
+      "train loss:0.002789240558016929\n",
+      "train loss:0.0023820194212027627\n",
+      "train loss:0.0010426417419444791\n",
+      "train loss:0.0026562942837508275\n",
+      "train loss:0.005383883797517855\n",
+      "train loss:0.00019928791010886005\n",
+      "train loss:0.0011568016328131186\n",
+      "train loss:0.00027176473306993366\n",
+      "train loss:0.00047763874220072423\n",
+      "train loss:0.0001213861660222856\n",
+      "train loss:0.0032301310282456057\n",
+      "train loss:0.007583148804599434\n",
+      "train loss:0.00944161126764183\n",
+      "train loss:0.006212352345848835\n",
+      "train loss:0.005297201928278837\n",
+      "train loss:0.001071263752054232\n",
+      "train loss:0.0008909944588275808\n",
+      "train loss:0.007222779393455072\n",
+      "train loss:0.00027520197212081417\n",
+      "train loss:0.0004519530492009234\n",
+      "train loss:0.010435274497562947\n",
+      "train loss:0.0007289016326722863\n",
+      "train loss:0.00029756419643480283\n",
+      "train loss:0.00206458736425594\n",
+      "train loss:0.0016945165639595267\n",
+      "train loss:0.0038557358798051496\n",
+      "train loss:0.0027468260929683257\n",
+      "train loss:0.00019143651386650424\n",
+      "train loss:0.0013974119834164712\n",
+      "train loss:0.0004464645225133489\n",
+      "train loss:0.0003085973429321018\n",
+      "train loss:0.000931702533309999\n",
+      "train loss:0.00032576030928680637\n",
+      "train loss:0.07319686344676715\n",
+      "train loss:0.00024540474686449767\n",
+      "train loss:8.098881574371157e-05\n",
+      "train loss:9.352924767779424e-05\n",
+      "train loss:3.680682256151892e-05\n",
+      "train loss:0.001448340570141027\n",
+      "train loss:0.00011854185038222188\n",
+      "train loss:0.0018398603273018135\n",
+      "train loss:0.007902946857863523\n",
+      "train loss:0.0022038419220709402\n",
+      "train loss:0.0020269162815382716\n",
+      "train loss:0.003703780793406374\n",
+      "train loss:0.00020341587595987316\n",
+      "train loss:0.0008028350289876886\n",
+      "train loss:0.0011438503976701443\n",
+      "train loss:0.0011487433342483606\n",
+      "train loss:0.001392566955659929\n",
+      "train loss:0.0006437651033768913\n",
+      "train loss:0.004455851504589434\n",
+      "train loss:0.0030265208660600657\n",
+      "train loss:0.0023186956698132964\n",
+      "train loss:0.00820949326670181\n",
+      "train loss:0.0011490565132238207\n",
+      "train loss:0.00017306605021089543\n",
+      "train loss:0.0003643891860144878\n",
+      "train loss:0.0013317833254217423\n",
+      "train loss:0.0010783135073208535\n",
+      "train loss:0.002235271414700939\n",
+      "train loss:0.0019593382896514497\n",
+      "train loss:0.002355138621859486\n",
+      "train loss:0.00136726042525598\n",
+      "train loss:0.0016678367852610975\n",
+      "train loss:0.002005513791933336\n",
+      "train loss:0.00012643409998382878\n",
+      "train loss:0.0011214300962012623\n",
+      "train loss:0.0007801047901982609\n",
+      "train loss:0.0005555890498999648\n",
+      "train loss:0.0403200374984329\n",
+      "train loss:0.0010605101454818236\n",
+      "train loss:0.0015840737992831352\n",
+      "train loss:0.001521918121775368\n",
+      "train loss:0.0021373974244186834\n",
+      "train loss:0.00035060281265820856\n",
+      "train loss:0.0014643268810309899\n",
+      "train loss:0.0010817949931633559\n",
+      "train loss:0.0014327174851359048\n",
+      "train loss:0.006554073727600064\n",
+      "train loss:0.0016517591746031896\n",
+      "train loss:0.0010048775095969082\n",
+      "train loss:0.006124615763730027\n",
+      "train loss:0.0012201811206435592\n",
+      "train loss:0.0010386388759554656\n",
+      "train loss:0.00026995791527448547\n",
+      "train loss:0.002144817963608315\n",
+      "train loss:0.0010424738526554102\n",
+      "train loss:0.0021611388907992165\n",
+      "train loss:0.00594742463098841\n",
+      "train loss:0.0008076219218444136\n",
+      "train loss:0.0025112813211926703\n",
+      "train loss:0.001191853413855036\n",
+      "train loss:0.0007552671722838691\n",
+      "train loss:0.003588400792711545\n",
+      "train loss:0.0012412780358402199\n",
+      "train loss:0.0003822533892199237\n",
+      "train loss:0.0008954707996986843\n",
+      "train loss:0.001415679597605034\n",
+      "train loss:9.104985088594444e-05\n",
+      "train loss:0.0003341283275084864\n",
+      "train loss:0.001615917833134424\n",
+      "train loss:0.0021480519246466208\n",
+      "train loss:0.0008381005062647755\n",
+      "train loss:0.0036495747787094056\n",
+      "train loss:0.0003509194866758769\n",
+      "train loss:8.902442208400323e-05\n",
+      "train loss:0.0012644854821466118\n",
+      "train loss:0.0004345703040956606\n",
+      "train loss:0.00012248898505825184\n",
+      "train loss:0.0012043742413828577\n",
+      "train loss:0.001315735491831011\n",
+      "train loss:0.0005720866272562107\n",
+      "train loss:0.0037461314711911605\n",
+      "train loss:0.0011807038329645708\n",
+      "train loss:0.000385052352963265\n",
+      "train loss:0.00135236341095682\n",
+      "train loss:0.001449374237237369\n",
+      "train loss:0.004661943511308059\n",
+      "train loss:0.0019307791323953478\n",
+      "train loss:0.007911689514380177\n",
+      "train loss:0.0019898436849784804\n",
+      "train loss:0.0005580046578003969\n",
+      "train loss:0.00016195299207915847\n",
+      "train loss:0.0011699544537123446\n",
+      "train loss:0.0007769225451467362\n",
+      "train loss:0.0005228174714244528\n",
+      "train loss:0.00021469150737042593\n",
+      "train loss:0.00025020483408718893\n",
+      "train loss:0.0032593762377886716\n",
+      "train loss:0.002524099044474866\n",
+      "train loss:0.0008740733312194932\n",
+      "train loss:0.0026472650242537537\n",
+      "train loss:0.002632275721962421\n",
+      "train loss:0.0006342675544290968\n",
+      "train loss:0.012605060072130444\n",
+      "train loss:0.0020793622590757374\n",
+      "train loss:0.00013591669968274073\n",
+      "train loss:0.0002223366268004627\n",
+      "train loss:0.003148389391604954\n",
+      "train loss:0.00018286149396523895\n",
+      "train loss:0.000379086128451454\n",
+      "train loss:0.00048140013684133003\n",
+      "train loss:0.00463466670727198\n",
+      "train loss:0.0013343163351319295\n",
+      "train loss:0.007202627840193934\n",
+      "train loss:0.0003939686470984353\n",
+      "train loss:0.0004164089485033917\n",
+      "train loss:0.005667870401323527\n",
+      "train loss:0.0006164837044832433\n",
+      "train loss:0.004529651114173213\n",
+      "train loss:0.000986414030715124\n",
+      "train loss:0.0006226938615132399\n",
+      "train loss:0.00044583850292117643\n",
+      "train loss:0.000622411932940038\n",
+      "train loss:0.0046916270687337195\n",
+      "train loss:0.009445373852958848\n",
+      "train loss:0.0007539809709894968\n",
+      "train loss:0.0022218848573152894\n",
+      "train loss:0.00022802834136561112\n",
+      "train loss:0.0007948941628484166\n",
+      "train loss:0.00097598171222426\n",
+      "train loss:0.0020384004282934186\n",
+      "train loss:0.003543544986689066\n",
+      "train loss:0.0007765314238654635\n",
+      "train loss:0.0036223047398643566\n",
+      "train loss:0.007248752351887319\n",
+      "train loss:0.0003309838668733135\n",
+      "train loss:0.0006081494992998603\n",
+      "train loss:0.004277078837526386\n",
+      "train loss:0.0018438844309843373\n",
+      "train loss:0.0013508679173348985\n",
+      "train loss:0.001519516046701864\n",
+      "train loss:0.0009383829417283039\n",
+      "train loss:0.003842185464599045\n",
+      "train loss:0.0019409028815521897\n",
+      "train loss:0.00017364067740495198\n",
+      "train loss:0.0032094946532938494\n",
+      "train loss:0.0004987258454212905\n",
+      "train loss:0.0028038641430959445\n",
+      "train loss:0.00025983517450421623\n",
+      "train loss:0.0005508225397572331\n",
+      "train loss:0.0016402454790494557\n",
+      "train loss:0.0020926015258990894\n",
+      "train loss:0.00012346196193192236\n",
+      "train loss:0.002638923901013614\n",
+      "train loss:0.01092015617594077\n",
+      "train loss:0.0005138920848244728\n",
+      "train loss:0.001004596368392022\n",
+      "train loss:0.0020148559585911594\n",
+      "train loss:0.0008294028862457377\n",
+      "train loss:0.0017914876905864732\n",
+      "train loss:0.0010791161171768125\n",
+      "train loss:0.0006448848899704588\n",
+      "train loss:0.0011571228444496448\n",
+      "train loss:0.00045987729944376186\n",
+      "train loss:0.0008745913708055598\n",
+      "train loss:0.0023388984857699406\n",
+      "train loss:6.414151130503845e-05\n",
+      "train loss:0.0012526510568311444\n",
+      "train loss:0.003702159895876804\n",
+      "train loss:0.003970850172700187\n",
+      "train loss:0.002462979829922056\n",
+      "train loss:0.001360226156721705\n",
+      "train loss:0.002180966850512325\n",
+      "train loss:0.0019253585199627613\n",
+      "train loss:0.005149236508767427\n",
+      "train loss:0.0009272849935166892\n",
+      "train loss:0.001417026266130281\n",
+      "train loss:0.0009048417201857632\n",
+      "train loss:0.00902034775390861\n",
+      "train loss:0.0003191439658276205\n",
+      "train loss:0.0028369394408274207\n",
+      "train loss:0.0001219125291607011\n",
+      "train loss:0.0007254296653479749\n",
+      "train loss:0.0008232405133718706\n",
+      "train loss:0.0004566701320990643\n",
+      "train loss:0.0007359848411621382\n",
+      "train loss:0.0007911369468864725\n",
+      "train loss:9.56339964413809e-05\n",
+      "train loss:0.0007978944780463708\n",
+      "train loss:0.0033511595432524316\n",
+      "train loss:0.0024595527220975305\n",
+      "train loss:0.0001095465885116912\n",
+      "train loss:0.0027763820516224563\n",
+      "train loss:0.0070266597608720106\n",
+      "train loss:0.00761517307920341\n",
+      "train loss:0.0006149936865260174\n",
+      "train loss:5.485260378127934e-05\n",
+      "train loss:0.0005261053263718411\n",
+      "train loss:0.0006554361152405473\n",
+      "train loss:0.00248779497917738\n",
+      "train loss:0.0007039607475032004\n",
+      "train loss:3.895158285666154e-05\n",
+      "train loss:0.00033218308103715274\n",
+      "train loss:0.0011870293322198732\n",
+      "train loss:0.00025238615413001204\n",
+      "train loss:0.00037411855881902523\n",
+      "train loss:0.00023496722767244352\n",
+      "train loss:0.00024531430358304537\n",
+      "train loss:0.0067689093019934\n",
+      "train loss:0.0005421662921089496\n",
+      "train loss:0.005735617971912612\n",
+      "train loss:0.008200865406309477\n",
+      "train loss:0.0008057890512898001\n",
+      "train loss:0.00025996784110731905\n",
+      "train loss:6.16685355694556e-05\n",
+      "train loss:0.002028896794203463\n",
+      "train loss:0.00017232538380483476\n",
+      "train loss:0.0011561937621661312\n",
+      "train loss:0.0001778317144859885\n",
+      "train loss:0.0007695393644188145\n",
+      "train loss:0.002589859279148077\n",
+      "train loss:0.0028461516974140556\n",
+      "train loss:0.0003098916039123784\n",
+      "train loss:0.00014442566033244853\n",
+      "train loss:0.0006998797111374103\n",
+      "train loss:0.0005449941811876729\n",
+      "train loss:0.00032396899992527957\n",
+      "train loss:0.0022842905522077782\n",
+      "train loss:0.0007086104664920716\n",
+      "train loss:0.00043835632584729813\n",
+      "train loss:0.00015530607680400523\n",
+      "train loss:0.0008761075686014884\n",
+      "train loss:0.00017952900958541248\n",
+      "train loss:0.004074757808123502\n",
+      "train loss:0.00763900733211882\n",
+      "train loss:0.000895789956442668\n",
+      "train loss:0.00010377074777250966\n",
+      "train loss:8.55031028969495e-05\n",
+      "train loss:0.004851317330535822\n",
+      "train loss:8.549138854217955e-05\n",
+      "train loss:0.0023128991333120647\n",
+      "train loss:0.0037435604939249894\n",
+      "train loss:0.0009086771688484081\n",
+      "train loss:0.0008965621763603088\n",
+      "train loss:0.0005630121925396551\n",
+      "train loss:0.0049361734768176265\n",
+      "train loss:0.012345000359231768\n",
+      "train loss:0.013373583871813775\n",
+      "train loss:0.004347552319363865\n",
+      "train loss:0.00020791967455644853\n",
+      "train loss:0.00032071180203704816\n",
+      "train loss:0.0004299759323748823\n",
+      "train loss:0.0007727212535427336\n",
+      "train loss:0.005195136462642085\n",
+      "train loss:0.019947166341662968\n",
+      "train loss:0.00043744529713949694\n",
+      "train loss:0.000735049809869701\n",
+      "train loss:0.0013731677486533683\n",
+      "train loss:0.001370100762937856\n",
+      "train loss:0.0011118993573063516\n",
+      "train loss:0.00045616385833194446\n",
+      "train loss:0.00015354851459214511\n",
+      "train loss:0.001885386511651041\n",
+      "train loss:0.0001038829120347471\n",
+      "train loss:0.0011024588978280944\n",
+      "train loss:0.002298775457240504\n",
+      "train loss:0.0016509754365802846\n",
+      "train loss:0.0004201499731036245\n",
+      "train loss:0.000624898856829724\n",
+      "train loss:0.0008159464052417419\n",
+      "train loss:0.0009443179687623171\n",
+      "train loss:0.0037763164755276234\n",
+      "train loss:0.0016849874755833455\n",
+      "train loss:0.003902871201326864\n",
+      "train loss:0.0003539673713846549\n",
+      "train loss:0.0008431043415010042\n",
+      "train loss:0.0017871345968767463\n",
+      "train loss:0.0007081709589874595\n",
+      "train loss:0.001497292489590464\n",
+      "train loss:5.471459060158774e-05\n",
+      "train loss:0.003448536882600371\n",
+      "train loss:0.0024903642717709912\n",
+      "train loss:0.0009141849806722335\n",
+      "train loss:0.00016044892774401274\n",
+      "train loss:2.9459345853118683e-05\n",
+      "train loss:0.0007478451846848329\n",
+      "train loss:0.0009737734208622255\n",
+      "train loss:0.01082258014259804\n",
+      "train loss:0.0014754817967312602\n",
+      "train loss:0.00035940477123813685\n",
+      "train loss:0.0026771630370909434\n",
+      "train loss:0.0014336192548771393\n",
+      "train loss:0.004394326626642422\n",
+      "train loss:0.002884741594947247\n",
+      "train loss:0.0020071710958638037\n",
+      "train loss:0.0002644655937781062\n",
+      "train loss:0.00047330306811532946\n",
+      "train loss:0.0035762019419390066\n",
+      "train loss:0.010648172057648815\n",
+      "train loss:0.0013567466378221918\n",
+      "train loss:0.00197736535261263\n",
+      "train loss:0.0005690127446134509\n",
+      "train loss:0.0004872038065229816\n",
+      "train loss:0.00241274547997243\n",
+      "train loss:0.0042739832454321775\n",
+      "train loss:0.001277490814280513\n",
+      "train loss:0.00026011169066435426\n",
+      "train loss:0.00025053295261414114\n",
+      "train loss:0.0001584094886334484\n",
+      "train loss:0.0030828702750959995\n",
+      "train loss:0.0010323950211807006\n",
+      "train loss:0.0010590720883513597\n",
+      "train loss:0.002141435570790481\n",
+      "train loss:0.012499660562732596\n",
+      "train loss:0.0006326094148617608\n",
+      "train loss:0.0006827613703861647\n",
+      "train loss:0.0024199862722254157\n",
+      "train loss:0.005311772782314405\n",
+      "train loss:0.0013312231529390372\n",
+      "train loss:0.00250131068811138\n",
+      "train loss:5.767400479787071e-05\n",
+      "train loss:0.007199553458399769\n",
+      "train loss:0.002664116884428415\n",
+      "train loss:0.0009486463024943243\n",
+      "train loss:0.00061363997131429\n",
+      "train loss:0.0008320213751247113\n",
+      "train loss:0.00019691769477491245\n",
+      "train loss:0.000699639664088761\n",
+      "train loss:0.0023415886539033762\n",
+      "train loss:0.00043646943468432216\n",
+      "train loss:0.006732637955792805\n",
+      "train loss:0.00017613118945424507\n",
+      "train loss:1.4935578284004895e-05\n",
+      "train loss:0.0036386054756933463\n",
+      "train loss:0.0013919914586803367\n",
+      "train loss:0.0014557987979008536\n",
+      "train loss:0.00037830185345877386\n",
+      "train loss:0.0008183242749497364\n",
+      "train loss:0.005321204018786584\n",
+      "train loss:0.0009295894764540354\n",
+      "train loss:0.0005313527759971655\n",
+      "=== epoch:17, train acc:0.999, test acc:0.99 ===\n",
+      "train loss:0.00529649864772141\n",
+      "train loss:0.0001097765405847573\n",
+      "train loss:0.00042904505884104035\n",
+      "train loss:0.0010310293874101408\n",
+      "train loss:0.000202921596944486\n",
+      "train loss:0.0012311216096129824\n",
+      "train loss:0.01725094191246529\n",
+      "train loss:0.0004796165305408107\n",
+      "train loss:0.004794433103009817\n",
+      "train loss:0.0005445124850991227\n",
+      "train loss:0.000920603655381986\n",
+      "train loss:0.002587376495246138\n",
+      "train loss:0.0017394747791539664\n",
+      "train loss:0.0009714731833170471\n",
+      "train loss:0.0009884829359190523\n",
+      "train loss:0.0018086085624830195\n",
+      "train loss:0.00037608438213456835\n",
+      "train loss:0.0009180952789731627\n",
+      "train loss:0.0001447130733116748\n",
+      "train loss:0.0042167454899873355\n",
+      "train loss:0.0005416034013433393\n",
+      "train loss:0.00023462452538529452\n",
+      "train loss:0.002680307737113034\n",
+      "train loss:0.0014463987480443057\n",
+      "train loss:0.002336077388447556\n",
+      "train loss:0.002827402832632056\n",
+      "train loss:5.7016040417657384e-05\n",
+      "train loss:0.00015715656566688333\n",
+      "train loss:0.0030588399679984447\n",
+      "train loss:0.021732545317148078\n",
+      "train loss:0.002560277637310323\n",
+      "train loss:0.00018018117326336336\n",
+      "train loss:0.0010495725667855013\n",
+      "train loss:0.013847893912006129\n",
+      "train loss:0.002752083961990941\n",
+      "train loss:0.0004685669249428051\n",
+      "train loss:0.000900206715623575\n",
+      "train loss:0.0008130694082962149\n",
+      "train loss:0.0012302443736693172\n",
+      "train loss:0.0006340742643394868\n",
+      "train loss:0.00032389361823883996\n",
+      "train loss:0.004663810225957319\n",
+      "train loss:0.03051358773045923\n",
+      "train loss:0.0020837795338623545\n",
+      "train loss:0.00016604865307943254\n",
+      "train loss:0.0003443931658531924\n",
+      "train loss:0.00923411223440943\n",
+      "train loss:0.0029353827630885713\n",
+      "train loss:0.001164048017334726\n",
+      "train loss:0.00258529247206968\n",
+      "train loss:0.001871668531414577\n",
+      "train loss:0.007743598499648823\n",
+      "train loss:0.003259379855724673\n",
+      "train loss:7.081811615718645e-05\n",
+      "train loss:0.0020013360877645225\n",
+      "train loss:0.00017238200084099614\n",
+      "train loss:0.00439545689904593\n",
+      "train loss:0.0006522958520236374\n",
+      "train loss:0.00695724561671771\n",
+      "train loss:0.0008872782810554055\n",
+      "train loss:0.0012444601658876968\n",
+      "train loss:0.0003373753580669044\n",
+      "train loss:0.0007252098163324905\n",
+      "train loss:8.937152247470596e-05\n",
+      "train loss:0.005489927908891653\n",
+      "train loss:0.002855621227130749\n",
+      "train loss:0.0014711976973171854\n",
+      "train loss:0.0001827024556359811\n",
+      "train loss:0.001930274221869306\n",
+      "train loss:0.0014683286828642403\n",
+      "train loss:0.0007769477080613379\n",
+      "train loss:0.003732774171918446\n",
+      "train loss:0.001582203927434128\n",
+      "train loss:0.0005030027292922421\n",
+      "train loss:0.002788004437283454\n",
+      "train loss:0.0005049119783531814\n",
+      "train loss:0.0004139670364005366\n",
+      "train loss:0.0010588696806388782\n",
+      "train loss:0.00037725036417694097\n",
+      "train loss:0.0005868857635129165\n",
+      "train loss:0.0007338044629295788\n",
+      "train loss:0.0011680194137243852\n",
+      "train loss:0.01634660078471837\n",
+      "train loss:0.0004420799797684152\n",
+      "train loss:0.00046238071711829555\n",
+      "train loss:0.0012445844146625467\n",
+      "train loss:0.0006804018144437268\n",
+      "train loss:0.0018048689985092584\n",
+      "train loss:0.0060859138415561885\n",
+      "train loss:0.0025336134845123652\n",
+      "train loss:0.011657110089773643\n",
+      "train loss:0.0009283134742467576\n",
+      "train loss:0.0010610474256033228\n",
+      "train loss:0.0003465091571427151\n",
+      "train loss:0.00024340620064580053\n",
+      "train loss:0.0006403287495306371\n",
+      "train loss:0.0002372981616191561\n",
+      "train loss:0.002025252848295288\n",
+      "train loss:0.0002981215956311731\n",
+      "train loss:0.005127655370436059\n",
+      "train loss:0.007119394090195594\n",
+      "train loss:0.00840858375007657\n",
+      "train loss:0.0014605062291600216\n",
+      "train loss:0.002771925564075537\n",
+      "train loss:0.005892833561739045\n",
+      "train loss:0.006063468975144817\n",
+      "train loss:0.0003585351672199074\n",
+      "train loss:0.002558184050471816\n",
+      "train loss:0.00044101951477292607\n",
+      "train loss:0.0009508998189699475\n",
+      "train loss:0.0006992504128289289\n",
+      "train loss:0.0020988529214155996\n",
+      "train loss:0.000558640088744938\n",
+      "train loss:0.0027561998142733566\n",
+      "train loss:0.00020731398414150905\n",
+      "train loss:0.0013865852944756143\n",
+      "train loss:0.00033397526060971764\n",
+      "train loss:0.0014872962232796564\n",
+      "train loss:0.004614595005981675\n",
+      "train loss:0.003724034500701946\n",
+      "train loss:0.0007023031271799792\n",
+      "train loss:0.00029431620606398085\n",
+      "train loss:0.00031915554375935005\n",
+      "train loss:0.0008174165117316816\n",
+      "train loss:0.00015061614799012093\n",
+      "train loss:0.0016387099422519782\n",
+      "train loss:0.00021890439484164133\n",
+      "train loss:0.0014424932071069945\n",
+      "train loss:0.0008883832116733204\n",
+      "train loss:0.002366098800094411\n",
+      "train loss:0.0016503849767743944\n",
+      "train loss:0.00084580562971115\n",
+      "train loss:3.646951749679609e-05\n",
+      "train loss:0.0007900031049246607\n",
+      "train loss:0.0002384322427017002\n",
+      "train loss:0.00022979577414854958\n",
+      "train loss:0.0643451302389254\n",
+      "train loss:8.655410461324705e-05\n",
+      "train loss:0.0037148191030258364\n",
+      "train loss:0.00036500971184058844\n",
+      "train loss:0.0009649506002156902\n",
+      "train loss:0.0018804930598810376\n",
+      "train loss:0.0004280001724386485\n",
+      "train loss:0.0012725763942842755\n",
+      "train loss:0.0008969288191262372\n",
+      "train loss:0.0007280334199655722\n",
+      "train loss:0.004275857525710959\n",
+      "train loss:0.00045867702345161776\n",
+      "train loss:0.0006763577934823329\n",
+      "train loss:0.0007554605141315508\n",
+      "train loss:0.0008469968368895988\n",
+      "train loss:0.0002648859635869607\n",
+      "train loss:0.0012451024983111607\n",
+      "train loss:0.006110052817472857\n",
+      "train loss:0.0045871655996569645\n",
+      "train loss:0.0016936033645279808\n",
+      "train loss:0.0014757108424909684\n",
+      "train loss:0.0016888895939969073\n",
+      "train loss:0.0023697640585755068\n",
+      "train loss:0.00012482768614454522\n",
+      "train loss:0.0013599870628325033\n",
+      "train loss:0.00016729487555038132\n",
+      "train loss:0.0011293267461833048\n",
+      "train loss:0.00037095533744942974\n",
+      "train loss:0.003383877201490386\n",
+      "train loss:0.003630454123603612\n",
+      "train loss:0.0016743891615497914\n",
+      "train loss:0.0004753794245676441\n",
+      "train loss:0.0023362477646342926\n",
+      "train loss:0.0014441042615017511\n",
+      "train loss:0.00404471650896511\n",
+      "train loss:0.0050394702515069245\n",
+      "train loss:8.392546122535087e-05\n",
+      "train loss:7.391232340691089e-05\n",
+      "train loss:0.002028894542235323\n",
+      "train loss:0.0013942277027702752\n",
+      "train loss:0.0006083629576918953\n",
+      "train loss:2.5562977747224178e-05\n",
+      "train loss:0.0005159329685472955\n",
+      "train loss:0.00012886350714174353\n",
+      "train loss:0.005290505015687369\n",
+      "train loss:4.604183182329553e-05\n",
+      "train loss:0.0006977993475397238\n",
+      "train loss:0.004029178392100167\n",
+      "train loss:0.00460729641296833\n",
+      "train loss:0.00019184633137692944\n",
+      "train loss:0.00011230109231463941\n",
+      "train loss:0.005021935181170005\n",
+      "train loss:0.0032763748776558025\n",
+      "train loss:0.0029669887507167182\n",
+      "train loss:7.174541522035673e-05\n",
+      "train loss:0.0019673076113933445\n",
+      "train loss:0.0007388354267730789\n",
+      "train loss:0.0009937324319938266\n",
+      "train loss:0.0029535655952059804\n",
+      "train loss:0.00012844508980318082\n",
+      "train loss:0.0013955205501840048\n",
+      "train loss:6.796570838976155e-05\n",
+      "train loss:0.006557580167227835\n",
+      "train loss:9.111491102852014e-05\n",
+      "train loss:0.0010263957608615379\n",
+      "train loss:9.39525308647669e-06\n",
+      "train loss:0.0008462007472992361\n",
+      "train loss:0.0032139077379827485\n",
+      "train loss:0.0022391123200034948\n",
+      "train loss:0.0006066236976196857\n",
+      "train loss:0.0007886818033087695\n",
+      "train loss:0.00012010812259044723\n",
+      "train loss:0.0009104769961104057\n",
+      "train loss:0.000431076349183074\n",
+      "train loss:0.001984299148761432\n",
+      "train loss:0.0004926801726578616\n",
+      "train loss:0.0001477322554454662\n",
+      "train loss:0.0006675174395694276\n",
+      "train loss:0.001556657150548559\n",
+      "train loss:0.0005196423073096565\n",
+      "train loss:0.0006825287967760129\n",
+      "train loss:0.0013378431628399275\n",
+      "train loss:0.0011789985349805931\n",
+      "train loss:5.7827740779626734e-05\n",
+      "train loss:0.02152404821099009\n",
+      "train loss:0.001463356690650307\n",
+      "train loss:0.0014232268909550316\n",
+      "train loss:0.003176921293044826\n",
+      "train loss:0.0006501023614957094\n",
+      "train loss:0.0007387468641411796\n",
+      "train loss:0.0003034276955768719\n",
+      "train loss:1.7781690189335298e-05\n",
+      "train loss:0.001347022536483\n",
+      "train loss:0.0007072688064629082\n",
+      "train loss:0.004692762281826185\n",
+      "train loss:3.389472560697742e-05\n",
+      "train loss:0.00031863122864467697\n",
+      "train loss:0.004001233031977154\n",
+      "train loss:0.00036832739245094506\n",
+      "train loss:0.00045373964842838953\n",
+      "train loss:0.0005843013634278267\n",
+      "train loss:0.0021675222836265917\n",
+      "train loss:6.852136618978316e-05\n",
+      "train loss:0.002355998166997441\n",
+      "train loss:0.0007763808972916236\n",
+      "train loss:0.004308744497357297\n",
+      "train loss:0.002444205864798243\n",
+      "train loss:0.013938145272974071\n",
+      "train loss:0.00014926047994325763\n",
+      "train loss:0.004195225484348869\n",
+      "train loss:0.00033504464615698274\n",
+      "train loss:0.0003422977634455339\n",
+      "train loss:0.0011674354173673988\n",
+      "train loss:0.0007513937975321225\n",
+      "train loss:0.0016505186053672022\n",
+      "train loss:0.0012108241270740742\n",
+      "train loss:0.005814492075767575\n",
+      "train loss:0.0008109654109163173\n",
+      "train loss:0.008047810538426532\n",
+      "train loss:0.007240761916409203\n",
+      "train loss:0.002409519044820267\n",
+      "train loss:0.000871061615811555\n",
+      "train loss:0.0016671128346103174\n",
+      "train loss:0.00010276666118547067\n",
+      "train loss:0.0014153986319893506\n",
+      "train loss:0.016020179002271272\n",
+      "train loss:0.00038893321961855755\n",
+      "train loss:0.02255704072642771\n",
+      "train loss:0.0006003821382434193\n",
+      "train loss:6.843245394266243e-05\n",
+      "train loss:0.0018111661025129403\n",
+      "train loss:0.0007019495079789593\n",
+      "train loss:0.002024869262253315\n",
+      "train loss:0.0021389083453553777\n",
+      "train loss:0.0033803571042992603\n",
+      "train loss:6.12252423299937e-05\n",
+      "train loss:0.0019771066549105465\n",
+      "train loss:0.002788558194858798\n",
+      "train loss:0.0023983561971086744\n",
+      "train loss:0.0016781029269870012\n",
+      "train loss:0.002335544268099719\n",
+      "train loss:0.0006169879837792052\n",
+      "train loss:0.0009127381181399206\n",
+      "train loss:0.0011660891156280537\n",
+      "train loss:0.0034591059062294206\n",
+      "train loss:0.00021469424681733982\n",
+      "train loss:0.0023275887105934823\n",
+      "train loss:0.006123258597123613\n",
+      "train loss:0.0036543468397113293\n",
+      "train loss:0.00020157779613943884\n",
+      "train loss:0.007933127475387205\n",
+      "train loss:0.0011363482917057949\n",
+      "train loss:0.0018922710002551568\n",
+      "train loss:0.00018708573864764213\n",
+      "train loss:6.652511133067519e-05\n",
+      "train loss:0.0024678174094301137\n",
+      "train loss:0.001435095765844832\n",
+      "train loss:0.001408232401171331\n",
+      "train loss:0.002184752313083066\n",
+      "train loss:0.0004595262605245968\n",
+      "train loss:0.005595537453845285\n",
+      "train loss:0.0020770957356415296\n",
+      "train loss:0.002179168141329399\n",
+      "train loss:0.0014868139443709246\n",
+      "train loss:0.001931653801543139\n",
+      "train loss:0.0010988112297232574\n",
+      "train loss:0.0017482647523980497\n",
+      "train loss:0.0020940754020850054\n",
+      "train loss:0.00425993828521751\n",
+      "train loss:0.0006324896899552867\n",
+      "train loss:0.0014680101049076217\n",
+      "train loss:0.0014986447794986572\n",
+      "train loss:0.0045221975715619925\n",
+      "train loss:0.017040392581907232\n",
+      "train loss:0.0007478043473100659\n",
+      "train loss:0.006098366691764756\n",
+      "train loss:0.0004861475181422591\n",
+      "train loss:0.0012565573673904993\n",
+      "train loss:0.0007851844905622668\n",
+      "train loss:0.005612300728829123\n",
+      "train loss:0.0023265910537685953\n",
+      "train loss:0.00012644030156757797\n",
+      "train loss:0.00015336591023567146\n",
+      "train loss:0.03649275334661023\n",
+      "train loss:0.0006605232812024677\n",
+      "train loss:0.0058285778735078465\n",
+      "train loss:9.594313917651073e-05\n",
+      "train loss:0.001182743906780062\n",
+      "train loss:0.004346030297176277\n",
+      "train loss:0.0007761183791788447\n",
+      "train loss:0.000639193827144597\n",
+      "train loss:0.0004275847172871096\n",
+      "train loss:0.0035688566559540363\n",
+      "train loss:0.0016100424278322617\n",
+      "train loss:0.0009150562084851904\n",
+      "train loss:0.002015851360512777\n",
+      "train loss:0.00146447003371947\n",
+      "train loss:0.002595818685101439\n",
+      "train loss:0.04012533215897179\n",
+      "train loss:0.001525271248522345\n",
+      "train loss:0.004132304888464021\n",
+      "train loss:0.003407991153031144\n",
+      "train loss:0.0015099397858086394\n",
+      "train loss:0.00032196047833579105\n",
+      "train loss:0.003666364110999884\n",
+      "train loss:0.001270529661358778\n",
+      "train loss:0.0007210301099317175\n",
+      "train loss:0.034431946675365095\n",
+      "train loss:0.000551624277643462\n",
+      "train loss:0.0023574112327611196\n",
+      "train loss:0.0012894234666728554\n",
+      "train loss:0.0013370164275203558\n",
+      "train loss:0.004410576835032474\n",
+      "train loss:0.0005011797923489956\n",
+      "train loss:0.004195711098441512\n",
+      "train loss:0.003274611584527036\n",
+      "train loss:0.011973313518711115\n",
+      "train loss:0.0034661040822764376\n",
+      "train loss:0.004798981881863656\n",
+      "train loss:0.0015620353157617492\n",
+      "train loss:0.03906689152012058\n",
+      "train loss:0.01852848102367532\n",
+      "train loss:0.033319057310786104\n",
+      "train loss:0.0005944913441515504\n",
+      "train loss:6.771114589043644e-05\n",
+      "train loss:0.001178410746141117\n",
+      "train loss:0.00034616197788552635\n",
+      "train loss:0.002546465772634222\n",
+      "train loss:0.01447605880648827\n",
+      "train loss:0.0007610289766299267\n",
+      "train loss:0.002109892203466103\n",
+      "train loss:0.0025405792637557116\n",
+      "train loss:0.0016761886270264662\n",
+      "train loss:0.04997489853730242\n",
+      "train loss:0.006502567792305221\n",
+      "train loss:0.002154295359493234\n",
+      "train loss:0.003996042204973332\n",
+      "train loss:0.00042902979264061286\n",
+      "train loss:0.0015363784006917876\n",
+      "train loss:0.0007417239159784271\n",
+      "train loss:0.0024570818403156676\n",
+      "train loss:0.004034362000243835\n",
+      "train loss:0.00015948653077888428\n",
+      "train loss:0.0017644035246602615\n",
+      "train loss:0.0005619892604328463\n",
+      "train loss:0.004955182406135777\n",
+      "train loss:0.0028784295435152524\n",
+      "train loss:0.004221453942225503\n",
+      "train loss:0.0005877742040565022\n",
+      "train loss:0.0025807599621656767\n",
+      "train loss:0.0037415952599350664\n",
+      "train loss:0.00031034893700078903\n",
+      "train loss:0.0024698458551015413\n",
+      "train loss:1.9904745044775516e-05\n",
+      "train loss:0.002869137448744854\n",
+      "train loss:0.0006586271040289759\n",
+      "train loss:0.002405602852356083\n",
+      "train loss:0.00019139975597888824\n",
+      "train loss:0.00328175721915474\n",
+      "train loss:0.001846915949006077\n",
+      "train loss:0.0019701504473415648\n",
+      "train loss:0.0031764370485824573\n",
+      "train loss:0.0033861783947613266\n",
+      "train loss:0.0012974540863892364\n",
+      "train loss:0.0013713363100312267\n",
+      "train loss:0.0016297596214103808\n",
+      "train loss:0.0013629915534552912\n",
+      "train loss:0.0055014172081052715\n",
+      "train loss:0.007028983895893876\n",
+      "train loss:0.00017220379557243866\n",
+      "train loss:0.002751538126911322\n",
+      "train loss:0.0017809036900671466\n",
+      "train loss:0.00014124911890886998\n",
+      "train loss:0.0033998365913597006\n",
+      "train loss:0.0011472739975341084\n",
+      "train loss:0.0005524598263505328\n",
+      "train loss:0.0002659977239774073\n",
+      "train loss:0.0021331045926117702\n",
+      "train loss:0.0012082950511210085\n",
+      "train loss:0.0007236521914422061\n",
+      "train loss:0.0038296023037695226\n",
+      "train loss:0.00885357930338106\n",
+      "train loss:0.006358785912966015\n",
+      "train loss:0.0001440149205018985\n",
+      "train loss:0.005839731300337263\n",
+      "train loss:0.0005909434818854861\n",
+      "train loss:0.00471408527559813\n",
+      "train loss:0.006695507877142503\n",
+      "train loss:0.0015011890914436285\n",
+      "train loss:0.0023770575000124174\n",
+      "train loss:0.001474321184277374\n",
+      "train loss:0.0010297757130871576\n",
+      "train loss:0.00038005313826746116\n",
+      "train loss:0.0033187418435313104\n",
+      "train loss:0.00025150494564459816\n",
+      "train loss:0.007653718303423646\n",
+      "train loss:0.002110198264945662\n",
+      "train loss:0.000709478518941115\n",
+      "train loss:0.0036541961845346126\n",
+      "train loss:0.013322026885576353\n",
+      "train loss:0.005649779122572358\n",
+      "train loss:0.002179235084149041\n",
+      "train loss:0.004666637924565756\n",
+      "train loss:0.004748225047324075\n",
+      "train loss:0.00514076057989443\n",
+      "train loss:8.161751122384047e-05\n",
+      "train loss:0.011968657137435864\n",
+      "train loss:0.005353937403301759\n",
+      "train loss:0.00021323615786826295\n",
+      "train loss:2.9550891632007586e-05\n",
+      "train loss:0.0014005656578891393\n",
+      "train loss:0.0021958080675131097\n",
+      "train loss:0.002800454447290498\n",
+      "train loss:0.0018410147088665765\n",
+      "train loss:0.002055661127301154\n",
+      "train loss:0.001445141383563693\n",
+      "train loss:0.0024634644283683398\n",
+      "train loss:0.0011264223922074963\n",
+      "train loss:0.00037985840365164093\n",
+      "train loss:0.0009626491577940012\n",
+      "train loss:0.0019467347939642289\n",
+      "train loss:0.0025526085098778665\n",
+      "train loss:0.007025002682075475\n",
+      "train loss:0.0013579024567708011\n",
+      "train loss:0.0022766573809829666\n",
+      "train loss:0.0037695538229546776\n",
+      "train loss:0.002648675187226906\n",
+      "train loss:0.0017921239125916755\n",
+      "train loss:0.0017808835562491605\n",
+      "train loss:0.0024069914475563074\n",
+      "train loss:0.0006533496388786457\n",
+      "train loss:0.0018247361711204871\n",
+      "train loss:0.0002698170049616059\n",
+      "train loss:8.584573886239455e-05\n",
+      "train loss:0.0008994579895110726\n",
+      "train loss:0.0005161192195520264\n",
+      "train loss:6.732344162632796e-05\n",
+      "train loss:0.0030201307544730537\n",
+      "train loss:0.0013778513227848178\n",
+      "train loss:0.0003460019086268109\n",
+      "train loss:0.0015798600784083208\n",
+      "train loss:0.00032687466170776264\n",
+      "train loss:3.1411700052953365e-05\n",
+      "train loss:0.00032675569391359675\n",
+      "train loss:0.00044500726308035354\n",
+      "train loss:0.0024564005613118734\n",
+      "train loss:0.003845773877872179\n",
+      "train loss:0.007329069047341241\n",
+      "train loss:0.0010186291475231524\n",
+      "train loss:4.9049287516226726e-05\n",
+      "train loss:0.00013468549970073367\n",
+      "train loss:0.0003422081306653164\n",
+      "train loss:0.00029640023894942567\n",
+      "train loss:0.0005291686715019206\n",
+      "train loss:0.00033194335026242525\n",
+      "train loss:0.00015236185319093758\n",
+      "train loss:0.004385977075739143\n",
+      "train loss:9.992706846957834e-06\n",
+      "train loss:0.001728121550564739\n",
+      "train loss:0.0005244171500857503\n",
+      "train loss:1.6251196003631075e-05\n",
+      "train loss:0.005309471655753374\n",
+      "train loss:0.0015182650850205082\n",
+      "train loss:0.0009406198592451559\n",
+      "train loss:0.0002219172248988946\n",
+      "train loss:0.0010497423623777725\n",
+      "train loss:0.0005457212319961857\n",
+      "train loss:0.002752873061901551\n",
+      "train loss:0.00012204088246259863\n",
+      "train loss:0.0012477597979581407\n",
+      "train loss:0.0030064625422022654\n",
+      "train loss:0.0027937565119787013\n",
+      "train loss:0.00791434824773723\n",
+      "train loss:0.013567393875163185\n",
+      "train loss:0.00041410611556675655\n",
+      "train loss:0.00027544279957211624\n",
+      "train loss:0.0007310198514626715\n",
+      "train loss:0.000771704026290016\n",
+      "train loss:0.0008146395709173615\n",
+      "train loss:0.005521616838307406\n",
+      "train loss:0.001545665665432561\n",
+      "train loss:0.00018791566793775971\n",
+      "train loss:0.004027185015612562\n",
+      "train loss:0.00031473849837371\n",
+      "train loss:0.0037482189282808964\n",
+      "train loss:0.00019228337398749312\n",
+      "train loss:0.0004221817962483752\n",
+      "train loss:0.0005183327929659253\n",
+      "train loss:0.001902986301556909\n",
+      "train loss:3.6433541591672756e-05\n",
+      "train loss:0.02691381044062284\n",
+      "train loss:0.001914498839687024\n",
+      "train loss:0.0003521378097555091\n",
+      "train loss:0.0006624786267925388\n",
+      "train loss:0.003670546245066785\n",
+      "train loss:0.0069543882786777\n",
+      "train loss:0.0013877728781186366\n",
+      "train loss:0.0013966908062666976\n",
+      "train loss:0.0002324539940020275\n",
+      "train loss:0.00043860794193189195\n",
+      "train loss:0.0007186367962900586\n",
+      "train loss:0.0006977561562482264\n",
+      "train loss:0.001161768024164053\n",
+      "train loss:0.00032369906358923397\n",
+      "train loss:0.002624947369998617\n",
+      "train loss:0.0003126261521376604\n",
+      "train loss:0.0005264371908646862\n",
+      "train loss:0.010490611662014008\n",
+      "train loss:0.0005579149575897724\n",
+      "train loss:0.0012065037654207752\n",
+      "train loss:0.0006081516988340478\n",
+      "train loss:0.00037261909507785095\n",
+      "train loss:0.002807806964777113\n",
+      "train loss:0.0004975792607500242\n",
+      "train loss:0.0007112740219018675\n",
+      "train loss:0.00018815114732522551\n",
+      "train loss:0.0019099230727896363\n",
+      "train loss:0.0023308900184184733\n",
+      "train loss:0.0026387959336414786\n",
+      "train loss:0.002944699237391962\n",
+      "train loss:0.000763879221733956\n",
+      "train loss:9.051481057027023e-05\n",
+      "train loss:0.0009228309000253507\n",
+      "train loss:0.001356005853511857\n",
+      "train loss:0.000489555461301137\n",
+      "train loss:0.0006928999157352945\n",
+      "train loss:0.001268742199065022\n",
+      "train loss:0.00031466752730147215\n",
+      "train loss:0.00023613471293143263\n",
+      "train loss:0.0013809121543318036\n",
+      "train loss:0.002229588853232352\n",
+      "train loss:0.0009173689966870639\n",
+      "train loss:0.01574791104378402\n",
+      "train loss:0.005127866341548992\n",
+      "train loss:0.003954235003912256\n",
+      "train loss:0.004724779460277238\n",
+      "train loss:0.0026142435821161682\n",
+      "train loss:0.004305200417141725\n",
+      "train loss:0.004000013032497437\n",
+      "train loss:0.0009034027383910067\n",
+      "train loss:0.002153631915674115\n",
+      "train loss:0.008956510737844465\n",
+      "train loss:0.0014826292230084345\n",
+      "train loss:0.0009024662118808286\n",
+      "train loss:0.002627368836384354\n",
+      "train loss:0.00917789168493926\n",
+      "train loss:0.008704647082839755\n",
+      "train loss:0.00022330872884579215\n",
+      "train loss:0.0016108525757353707\n",
+      "train loss:0.0005069470295942664\n",
+      "train loss:0.0005647709056483123\n",
+      "train loss:0.0006218659277886706\n",
+      "train loss:0.0011242712853077633\n",
+      "train loss:0.0006762790105253373\n",
+      "train loss:0.0032260332017637365\n",
+      "train loss:0.0032696217587132826\n",
+      "train loss:0.008611964315565967\n",
+      "train loss:0.00029563607998101915\n",
+      "train loss:0.0012300995504321772\n",
+      "train loss:0.005071761012063751\n",
+      "train loss:0.004135591075278628\n",
+      "train loss:0.0030809434516318674\n",
+      "train loss:0.0004201180539801171\n",
+      "train loss:0.002210568635324676\n",
+      "=== epoch:18, train acc:0.999, test acc:0.985 ===\n",
+      "train loss:0.00016840831251569593\n",
+      "train loss:4.1071741229043185e-05\n",
+      "train loss:0.0005643631920274524\n",
+      "train loss:0.0007637748253937854\n",
+      "train loss:0.0002298048578922272\n",
+      "train loss:0.0003839245690326395\n",
+      "train loss:0.00552881723267125\n",
+      "train loss:0.000619904306622759\n",
+      "train loss:0.0029005105621006484\n",
+      "train loss:0.0005473479600941299\n",
+      "train loss:0.020546337562806105\n",
+      "train loss:0.0006899381778719435\n",
+      "train loss:0.0012728007103062977\n",
+      "train loss:0.0003403884143865073\n",
+      "train loss:0.002575202818947197\n",
+      "train loss:0.0003664695668521761\n",
+      "train loss:0.002992490227743116\n",
+      "train loss:0.0015347601585755132\n",
+      "train loss:0.0015961306093469627\n",
+      "train loss:0.001099423682617452\n",
+      "train loss:0.004843741381510463\n",
+      "train loss:0.00019351874335478403\n",
+      "train loss:0.0014566825443104573\n",
+      "train loss:0.0010016852442777796\n",
+      "train loss:0.000876636988592642\n",
+      "train loss:0.0031581619410498\n",
+      "train loss:0.012159176665713084\n",
+      "train loss:0.0031861820397066426\n",
+      "train loss:0.0011176884881274748\n",
+      "train loss:0.0008144726250607358\n",
+      "train loss:0.0007736734988533906\n",
+      "train loss:0.0009825788843025826\n",
+      "train loss:0.00022538422108690846\n",
+      "train loss:0.0010780477322263815\n",
+      "train loss:0.0007792940871927584\n",
+      "train loss:0.0004936500617477524\n",
+      "train loss:0.0020858981052807907\n",
+      "train loss:0.00022293333252730837\n",
+      "train loss:0.002415571115106081\n",
+      "train loss:0.009953436584301206\n",
+      "train loss:0.0017152600908219614\n",
+      "train loss:8.39755595517393e-05\n",
+      "train loss:0.004861218779360548\n",
+      "train loss:0.0011774461635483656\n",
+      "train loss:0.0008091531000296523\n",
+      "train loss:0.0002755467832089066\n",
+      "train loss:0.0016104371778319882\n",
+      "train loss:0.0065459747731361675\n",
+      "train loss:0.004146924131530835\n",
+      "train loss:0.00497426355948364\n",
+      "train loss:0.0009186311774399264\n",
+      "train loss:0.00021584760369511394\n",
+      "train loss:0.002531271434233856\n",
+      "train loss:0.007984732478039309\n",
+      "train loss:0.00010756369733896733\n",
+      "train loss:0.0017644127527547673\n",
+      "train loss:0.002409580690537212\n",
+      "train loss:0.0004891138852378835\n",
+      "train loss:0.007127534557947256\n",
+      "train loss:0.004622650671673781\n",
+      "train loss:0.0027263515781480456\n",
+      "train loss:0.00019083049721215254\n",
+      "train loss:0.0017056439637824215\n",
+      "train loss:0.002453974080813034\n",
+      "train loss:0.000921763539543025\n",
+      "train loss:0.0009667662227825333\n",
+      "train loss:0.0002955885016870814\n",
+      "train loss:0.001559506235087088\n",
+      "train loss:0.002572224164912656\n",
+      "train loss:0.006834111438972402\n",
+      "train loss:3.207015790079667e-05\n",
+      "train loss:0.014957194344282356\n",
+      "train loss:0.0009652719670654542\n",
+      "train loss:0.0012904011617460866\n",
+      "train loss:0.0037872833775626845\n",
+      "train loss:0.00078302901702519\n",
+      "train loss:0.008462792992345685\n",
+      "train loss:0.0008457738586929979\n",
+      "train loss:0.0005631520900663918\n",
+      "train loss:0.009380562090032973\n",
+      "train loss:0.0003066985793498184\n",
+      "train loss:0.014155742967010266\n",
+      "train loss:0.002461519884677363\n",
+      "train loss:0.0002702289205036588\n",
+      "train loss:0.002418796125621207\n",
+      "train loss:0.00023381514311824635\n",
+      "train loss:6.9675715378606e-05\n",
+      "train loss:0.002302476210755608\n",
+      "train loss:0.0012030333671876087\n",
+      "train loss:0.002081445057188151\n",
+      "train loss:0.0005524369154808967\n",
+      "train loss:5.335685290464253e-05\n",
+      "train loss:0.0008370165853677686\n",
+      "train loss:0.0011851271283971742\n",
+      "train loss:0.0008810494160311066\n",
+      "train loss:0.00390083174090908\n",
+      "train loss:0.002790050731805678\n",
+      "train loss:0.0038634994519183108\n",
+      "train loss:0.0010872879289343482\n",
+      "train loss:0.0015443576283492937\n",
+      "train loss:0.00038872349818980294\n",
+      "train loss:0.0023866695336813865\n",
+      "train loss:0.0006722596354904076\n",
+      "train loss:0.002295311807169194\n",
+      "train loss:1.2863519777696246e-05\n",
+      "train loss:0.0005851754335052893\n",
+      "train loss:0.002671170576560264\n",
+      "train loss:0.0012362059023774525\n",
+      "train loss:3.775195191735021e-05\n",
+      "train loss:0.0005708897707111369\n",
+      "train loss:0.001744772453935698\n",
+      "train loss:0.0017863446927246537\n",
+      "train loss:0.0006939616256294465\n",
+      "train loss:0.004639590804832937\n",
+      "train loss:0.0020088318737707545\n",
+      "train loss:0.00017786718520270865\n",
+      "train loss:8.238200136495712e-05\n",
+      "train loss:0.000109584525139189\n",
+      "train loss:0.0011038741316866723\n",
+      "train loss:0.01045284899067305\n",
+      "train loss:0.0009007249662435638\n",
+      "train loss:0.0025261059192472944\n",
+      "train loss:0.0001759349579520167\n",
+      "train loss:0.0004213058225703748\n",
+      "train loss:0.00029012583771788136\n",
+      "train loss:0.002341099441447979\n",
+      "train loss:0.0015183642817550335\n",
+      "train loss:0.0014969958412152625\n",
+      "train loss:0.003876602429727461\n",
+      "train loss:0.0006440857457347121\n",
+      "train loss:0.001078845285110365\n",
+      "train loss:0.0005366817114698897\n",
+      "train loss:0.00040238053541626117\n",
+      "train loss:0.0017991541988556419\n",
+      "train loss:7.544660738606756e-05\n",
+      "train loss:0.0033064259830746182\n",
+      "train loss:0.0008336786852002477\n",
+      "train loss:0.0014250738022530366\n",
+      "train loss:0.0034230516992965632\n",
+      "train loss:0.0006004452609250548\n",
+      "train loss:0.0008489428801966539\n",
+      "train loss:0.0005960468852849369\n",
+      "train loss:0.0011428121303736667\n",
+      "train loss:0.0026516162376356996\n",
+      "train loss:0.0005266678159729692\n",
+      "train loss:0.00029619312762901965\n",
+      "train loss:9.009486528573279e-05\n",
+      "train loss:0.0006846119990295067\n",
+      "train loss:0.0006400240800428562\n",
+      "train loss:0.002449041387399308\n",
+      "train loss:0.00014728676512667586\n",
+      "train loss:0.0011335256255392224\n",
+      "train loss:0.004080632227191448\n",
+      "train loss:0.0030102384281644363\n",
+      "train loss:3.276813385375164e-05\n",
+      "train loss:0.019226420153038416\n",
+      "train loss:0.00018874201196637748\n",
+      "train loss:0.005305589129869519\n",
+      "train loss:0.0003942262840872435\n",
+      "train loss:0.002288046374445681\n",
+      "train loss:0.00024432361609751004\n",
+      "train loss:0.00010703669980770448\n",
+      "train loss:0.00022344609472182755\n",
+      "train loss:0.0014883454600656304\n",
+      "train loss:0.008426477380671598\n",
+      "train loss:0.0006642178210869426\n",
+      "train loss:0.00016181503514773145\n",
+      "train loss:0.0026931192496604402\n",
+      "train loss:0.0038176978368305516\n",
+      "train loss:0.005076306470131666\n",
+      "train loss:0.00047092823601683977\n",
+      "train loss:0.004356228059487743\n",
+      "train loss:0.004035782482753133\n",
+      "train loss:0.0017662930743445042\n",
+      "train loss:2.823285500423691e-05\n",
+      "train loss:0.005896164459326369\n",
+      "train loss:0.0005832120120060905\n",
+      "train loss:0.0027767856609951097\n",
+      "train loss:0.0027254930195017646\n",
+      "train loss:0.003307558193734884\n",
+      "train loss:0.00033012333122866834\n",
+      "train loss:0.002877435247387753\n",
+      "train loss:0.0001486900943522399\n",
+      "train loss:0.002287850247827273\n",
+      "train loss:0.0016116223927377107\n",
+      "train loss:0.0020381445103247737\n",
+      "train loss:0.0005145031594167966\n",
+      "train loss:0.004907892436550152\n",
+      "train loss:0.006251167924955783\n",
+      "train loss:0.006435920598959835\n",
+      "train loss:0.003254316533627123\n",
+      "train loss:7.056349661720925e-05\n",
+      "train loss:0.0015864367473632283\n",
+      "train loss:0.0010966343228395305\n",
+      "train loss:0.0018818468124963646\n",
+      "train loss:0.0008819629974027953\n",
+      "train loss:0.002845752513267142\n",
+      "train loss:0.0037981400594095105\n",
+      "train loss:0.003621195724620806\n",
+      "train loss:0.0021103377926253875\n",
+      "train loss:0.0018686255137018565\n",
+      "train loss:0.0005252527500045485\n",
+      "train loss:0.0013196741923936107\n",
+      "train loss:0.00030610142796171097\n",
+      "train loss:0.003075082464957058\n",
+      "train loss:0.0050741194806232705\n",
+      "train loss:0.0005868011883736\n",
+      "train loss:0.002702987496898475\n",
+      "train loss:0.001303897716176039\n",
+      "train loss:0.006880054942259062\n",
+      "train loss:0.0004033616720290942\n",
+      "train loss:0.0033230390801661075\n",
+      "train loss:0.004514214401451409\n",
+      "train loss:0.0016139955304796088\n",
+      "train loss:0.004573711442186004\n",
+      "train loss:0.0014174711428730305\n",
+      "train loss:0.00021456236476774228\n",
+      "train loss:0.003423279763155534\n",
+      "train loss:0.0036402556871435493\n",
+      "train loss:0.002708718654088865\n",
+      "train loss:0.0012840324493853578\n",
+      "train loss:0.0009793335104303054\n",
+      "train loss:0.0001733175551187286\n",
+      "train loss:0.00021629520216741394\n",
+      "train loss:2.1208561870349778e-05\n",
+      "train loss:0.00010497619529036662\n",
+      "train loss:0.0025788586825350457\n",
+      "train loss:0.0052703167968618046\n",
+      "train loss:0.0019355668024999186\n",
+      "train loss:0.0009775558691898759\n",
+      "train loss:0.0005895703945958313\n",
+      "train loss:0.017030324704734593\n",
+      "train loss:0.0002816628199310151\n",
+      "train loss:0.0030340078344198176\n",
+      "train loss:0.0011347194446495064\n",
+      "train loss:0.002653067743800612\n",
+      "train loss:0.0007351903517450058\n",
+      "train loss:0.0005701652694622011\n",
+      "train loss:8.97063213837596e-05\n",
+      "train loss:0.004130009731190847\n",
+      "train loss:0.0009738721719509086\n",
+      "train loss:0.005608245312422045\n",
+      "train loss:0.0015590282574210784\n",
+      "train loss:0.011769405033116232\n",
+      "train loss:0.0012034828339201854\n",
+      "train loss:0.0009729366491537503\n",
+      "train loss:0.00946465839153601\n",
+      "train loss:0.00280722479363725\n",
+      "train loss:0.0012103404145414791\n",
+      "train loss:0.0009583768507617639\n",
+      "train loss:0.009420411885644176\n",
+      "train loss:0.0007604507367709968\n",
+      "train loss:0.0017398387379950148\n",
+      "train loss:0.003229751227714644\n",
+      "train loss:0.0009124697801469545\n",
+      "train loss:0.00247002721363657\n",
+      "train loss:0.0019935962263981535\n",
+      "train loss:0.0013714831425241387\n",
+      "train loss:0.002144313893129755\n",
+      "train loss:0.001422542277564043\n",
+      "train loss:0.00023390836309305566\n",
+      "train loss:0.005489825041252023\n",
+      "train loss:0.0007626666899526655\n",
+      "train loss:0.0004397677188406917\n",
+      "train loss:0.005614354824341699\n",
+      "train loss:0.001699827001082667\n",
+      "train loss:0.0006098435211210642\n",
+      "train loss:0.0006625402640822868\n",
+      "train loss:9.012009510178238e-05\n",
+      "train loss:6.889381203336346e-05\n",
+      "train loss:0.004205401113089762\n",
+      "train loss:0.003094303506179704\n",
+      "train loss:0.000514747420878415\n",
+      "train loss:0.000452773148827875\n",
+      "train loss:0.0018898639496166554\n",
+      "train loss:0.0013769319810043629\n",
+      "train loss:0.0006398198901289243\n",
+      "train loss:0.0013320768972174346\n",
+      "train loss:0.0012504068470467113\n",
+      "train loss:0.00020193593543771\n",
+      "train loss:0.0009859947753640473\n",
+      "train loss:0.0027071650456737106\n",
+      "train loss:0.0012992293896041346\n",
+      "train loss:0.0040930413989842375\n",
+      "train loss:7.192073504898249e-05\n",
+      "train loss:0.003212221835210305\n",
+      "train loss:0.0006459854293568376\n",
+      "train loss:0.0009363112026956916\n",
+      "train loss:0.00021154767447796263\n",
+      "train loss:7.352427438811808e-05\n",
+      "train loss:0.013425885642534796\n",
+      "train loss:0.00017466661525618207\n",
+      "train loss:0.001705546179628122\n",
+      "train loss:0.00013371327691937134\n",
+      "train loss:7.275312521792998e-05\n",
+      "train loss:0.0007737650475175353\n",
+      "train loss:0.0024128029407094486\n",
+      "train loss:0.0008876551578842132\n",
+      "train loss:0.002538626961127899\n",
+      "train loss:0.0005907627335806046\n",
+      "train loss:0.0001793305479577088\n",
+      "train loss:0.0006572449149783118\n",
+      "train loss:0.025164948706480318\n",
+      "train loss:0.0012785270828813743\n",
+      "train loss:0.0005099768607833759\n",
+      "train loss:0.00037708341795953006\n",
+      "train loss:0.0005796475948781334\n",
+      "train loss:0.00016126029504651956\n",
+      "train loss:0.000816651058661133\n",
+      "train loss:0.0040539729503590765\n",
+      "train loss:0.0004829779104338531\n",
+      "train loss:0.0004417927798935907\n",
+      "train loss:0.02781394617911632\n",
+      "train loss:0.0009752012063242543\n",
+      "train loss:0.0002492187006446086\n",
+      "train loss:0.015175826826954477\n",
+      "train loss:0.0028668642126364846\n",
+      "train loss:0.0002958229175690656\n",
+      "train loss:0.0006418800890834758\n",
+      "train loss:0.000962616788839666\n",
+      "train loss:0.0030469076346624326\n",
+      "train loss:0.00013440684153114728\n",
+      "train loss:0.0026431432385521643\n",
+      "train loss:0.0049187519472778315\n",
+      "train loss:9.433841923192559e-05\n",
+      "train loss:0.0002921448596109379\n",
+      "train loss:8.747295455157559e-05\n",
+      "train loss:6.947678151365999e-05\n",
+      "train loss:0.0004944410541975282\n",
+      "train loss:0.00029394317452978994\n",
+      "train loss:0.00045388307003656913\n",
+      "train loss:0.0009376984243407742\n",
+      "train loss:0.0008163468756432875\n",
+      "train loss:0.002620816718094788\n",
+      "train loss:0.00335273633659938\n",
+      "train loss:0.0013869906531214409\n",
+      "train loss:0.00017436038882310398\n",
+      "train loss:0.00037366096179663105\n",
+      "train loss:0.002265695628841397\n",
+      "train loss:0.01674072173911842\n",
+      "train loss:0.0009123045917322325\n",
+      "train loss:0.00029795137749632307\n",
+      "train loss:0.0002013665926880532\n",
+      "train loss:0.004194035345352971\n",
+      "train loss:0.0008826632506348494\n",
+      "train loss:0.03613660841227778\n",
+      "train loss:0.0017884517360702138\n",
+      "train loss:0.00038195420664301015\n",
+      "train loss:0.0015080223659659356\n",
+      "train loss:0.0015453830799379042\n",
+      "train loss:0.0020788650782020235\n",
+      "train loss:0.0018830981807793968\n",
+      "train loss:0.00011697768424666997\n",
+      "train loss:0.00012336073532560573\n",
+      "train loss:0.005582890929594762\n",
+      "train loss:6.872526809705579e-05\n",
+      "train loss:0.0002618733703579749\n",
+      "train loss:0.0005762799957185216\n",
+      "train loss:0.000546141594634754\n",
+      "train loss:0.0002201422208018571\n",
+      "train loss:0.004497808768261689\n",
+      "train loss:0.0009378028633397717\n",
+      "train loss:0.0002523641171200336\n",
+      "train loss:0.003468084369932091\n",
+      "train loss:0.00026840330994599206\n",
+      "train loss:0.004497018990671465\n",
+      "train loss:0.0018100618542051768\n",
+      "train loss:0.00032286027028648916\n",
+      "train loss:0.0009692617458704028\n",
+      "train loss:0.002368877514095066\n",
+      "train loss:0.009749467959630923\n",
+      "train loss:0.00012533950729814243\n",
+      "train loss:0.0018806868503674506\n",
+      "train loss:0.011258208844306046\n",
+      "train loss:0.0004465117028885686\n",
+      "train loss:0.005985279032788368\n",
+      "train loss:0.009758274791543738\n",
+      "train loss:0.0011771707900479968\n",
+      "train loss:0.0005301360009068153\n",
+      "train loss:0.005912150898063945\n",
+      "train loss:0.002964094964161705\n",
+      "train loss:0.00019452570104060317\n",
+      "train loss:0.0061494622398261065\n",
+      "train loss:0.00023840444379090502\n",
+      "train loss:0.0013783922602819936\n",
+      "train loss:0.004308765378178042\n",
+      "train loss:0.001062495787377402\n",
+      "train loss:0.0038448679792790134\n",
+      "train loss:0.0019261390654462\n",
+      "train loss:0.01720236056359547\n",
+      "train loss:0.062010267426241016\n",
+      "train loss:0.0006878643261074564\n",
+      "train loss:0.001158513159701854\n",
+      "train loss:0.0011860672483521547\n",
+      "train loss:0.00034163043617877544\n",
+      "train loss:0.004224084902306446\n",
+      "train loss:0.0002617065027786931\n",
+      "train loss:0.00018197915835386754\n",
+      "train loss:0.0043128634775789765\n",
+      "train loss:0.0018918559117215678\n",
+      "train loss:0.0001611282681826458\n",
+      "train loss:0.0008978927704649799\n",
+      "train loss:0.00013365785988568156\n",
+      "train loss:0.0016738205366216695\n",
+      "train loss:0.0027475095536915566\n",
+      "train loss:0.0015446298731605543\n",
+      "train loss:0.0012033512050042738\n",
+      "train loss:0.0007267537878815909\n",
+      "train loss:0.0010810970718424095\n",
+      "train loss:0.0028570176733199344\n",
+      "train loss:0.001332154086677099\n",
+      "train loss:0.0027045852251518596\n",
+      "train loss:2.1688621528012426e-05\n",
+      "train loss:0.0010085571172022514\n",
+      "train loss:0.000520076402236073\n",
+      "train loss:0.00029675421591343875\n",
+      "train loss:0.00010293978544484612\n",
+      "train loss:0.0041056077760521364\n",
+      "train loss:0.0001557746425503504\n",
+      "train loss:0.00025680415699049614\n",
+      "train loss:0.0007616399536475006\n",
+      "train loss:0.0008820736738195897\n",
+      "train loss:0.002237827271876456\n",
+      "train loss:0.0009080937634230054\n",
+      "train loss:0.0016465775764032005\n",
+      "train loss:0.0007963785516535689\n",
+      "train loss:0.000271710250984186\n",
+      "train loss:0.00017355478694370738\n",
+      "train loss:0.00560473626202711\n",
+      "train loss:0.0015124306980218495\n",
+      "train loss:0.001978191574265688\n",
+      "train loss:0.0003478815264682403\n",
+      "train loss:2.7974285299428715e-05\n",
+      "train loss:6.220662930715582e-05\n",
+      "train loss:7.99521570097432e-05\n",
+      "train loss:0.0007339429619166269\n",
+      "train loss:0.00023395245030130238\n",
+      "train loss:0.00016478426873946892\n",
+      "train loss:0.00022775574316507115\n",
+      "train loss:0.0002445100657903331\n",
+      "train loss:0.030563736206458265\n",
+      "train loss:0.0002518714034776671\n",
+      "train loss:0.0004705164056985883\n",
+      "train loss:0.001527731247727847\n",
+      "train loss:0.0007569505969198174\n",
+      "train loss:0.0020163838948786296\n",
+      "train loss:0.0029328796731797374\n",
+      "train loss:0.002751841902672361\n",
+      "train loss:0.0015270480052140057\n",
+      "train loss:0.0015143796258999158\n",
+      "train loss:0.002135579364599659\n",
+      "train loss:0.0023330262683956\n",
+      "train loss:0.0009704519216279936\n",
+      "train loss:0.0052110346555601225\n",
+      "train loss:0.002637216341281742\n",
+      "train loss:0.002462333760759106\n",
+      "train loss:0.00011211284446285658\n",
+      "train loss:0.0017860449764777356\n",
+      "train loss:0.002196145306722232\n",
+      "train loss:0.0014010692058219498\n",
+      "train loss:0.00024761138687479355\n",
+      "train loss:0.0012369741862283535\n",
+      "train loss:0.0003554670513797348\n",
+      "train loss:0.000859494076843954\n",
+      "train loss:0.002008749415089745\n",
+      "train loss:0.0015742043508043377\n",
+      "train loss:0.0012636608934753196\n",
+      "train loss:0.001246901881847769\n",
+      "train loss:0.00040473681461361885\n",
+      "train loss:0.0018665302212957505\n",
+      "train loss:0.0003673891618510809\n",
+      "train loss:0.0001485670877711396\n",
+      "train loss:0.004179959465077497\n",
+      "train loss:0.0008251864378661804\n",
+      "train loss:0.0004518331688993005\n",
+      "train loss:0.0009063094403157946\n",
+      "train loss:0.0007762611401283715\n",
+      "train loss:0.0002628118165095828\n",
+      "train loss:7.879281595655134e-05\n",
+      "train loss:0.001110261621046111\n",
+      "train loss:0.002650604421750157\n",
+      "train loss:0.007530584938024058\n",
+      "train loss:0.0016430168598005237\n",
+      "train loss:0.0005556047932087575\n",
+      "train loss:0.0003735470214032005\n",
+      "train loss:0.003085634974252188\n",
+      "train loss:0.0026379643856063485\n",
+      "train loss:0.0016422973268852025\n",
+      "train loss:0.0007638096615836891\n",
+      "train loss:0.0005338454681564288\n",
+      "train loss:0.00588645369389348\n",
+      "train loss:0.002831659156804419\n",
+      "train loss:0.0025677493462717125\n",
+      "train loss:0.005312494688948707\n",
+      "train loss:0.0022474632374207843\n",
+      "train loss:0.00024178158703388732\n",
+      "train loss:0.0010042607118616084\n",
+      "train loss:0.010737062230379554\n",
+      "train loss:6.63094088816681e-05\n",
+      "train loss:0.0001230676282991285\n",
+      "train loss:0.006589613706810169\n",
+      "train loss:0.0009748302324524031\n",
+      "train loss:0.0002797846722038302\n",
+      "train loss:6.75741857903316e-05\n",
+      "train loss:0.0017572203885612766\n",
+      "train loss:0.0004645553434585878\n",
+      "train loss:0.0007541944374860085\n",
+      "train loss:0.0015943354516082453\n",
+      "train loss:0.00031750978396717053\n",
+      "train loss:0.0009835341077524672\n",
+      "train loss:0.0024045403430884684\n",
+      "train loss:0.0018328770361182014\n",
+      "train loss:0.011503684657119937\n",
+      "train loss:0.004992638983759658\n",
+      "train loss:0.0006408134144312585\n",
+      "train loss:0.0006742879960081817\n",
+      "train loss:0.0001645644980367372\n",
+      "train loss:0.00034127705116577747\n",
+      "train loss:0.006019717357773232\n",
+      "train loss:0.0037163845011086135\n",
+      "train loss:0.0006228625350967763\n",
+      "train loss:0.0002734720783972108\n",
+      "train loss:0.00036872204011992424\n",
+      "train loss:0.00041676827405960707\n",
+      "train loss:0.00010721847888572645\n",
+      "train loss:0.0012860623137444177\n",
+      "train loss:0.0014288798454205735\n",
+      "train loss:0.0024601778780107835\n",
+      "train loss:0.0012378482273575273\n",
+      "train loss:0.008837024921957643\n",
+      "train loss:0.00016941407681388314\n",
+      "train loss:0.0008814235232502776\n",
+      "train loss:0.0008745417151102341\n",
+      "train loss:0.002240826536427365\n",
+      "train loss:0.0010216739950254768\n",
+      "train loss:0.0018914169335149738\n",
+      "train loss:0.009086594658955317\n",
+      "train loss:0.00229491201148888\n",
+      "train loss:0.002793688957706313\n",
+      "train loss:0.0003074973122737487\n",
+      "train loss:0.0011513220279602793\n",
+      "train loss:0.002468602281486436\n",
+      "train loss:0.000174510399375162\n",
+      "train loss:0.0004983697786096146\n",
+      "train loss:0.0006909485983277561\n",
+      "train loss:0.0011415354739220368\n",
+      "train loss:0.0036549655465641095\n",
+      "train loss:0.006195615781961495\n",
+      "train loss:0.00034423958183848016\n",
+      "train loss:0.006009027632475704\n",
+      "train loss:0.0013151792716944192\n",
+      "train loss:0.0006052120390411204\n",
+      "train loss:0.00019822440391172911\n",
+      "train loss:0.0004405988325793321\n",
+      "train loss:0.00012427902304533938\n",
+      "train loss:0.0008373376717784106\n",
+      "train loss:0.0005071477971753355\n",
+      "train loss:0.0019048912916934243\n",
+      "train loss:0.0017537214763588005\n",
+      "train loss:0.0021759752341941837\n",
+      "train loss:0.00040983752204014535\n",
+      "train loss:0.0009434300887789341\n",
+      "train loss:0.0013819541046872837\n",
+      "train loss:0.00048046821997064967\n",
+      "train loss:0.0007732995788941822\n",
+      "train loss:0.0013766791758781825\n",
+      "train loss:0.0007327294107213661\n",
+      "train loss:0.0005031146182334187\n",
+      "train loss:0.003431852593105187\n",
+      "train loss:0.0004776137326494528\n",
+      "train loss:0.0031918606539401945\n",
+      "train loss:0.002304364832040249\n",
+      "train loss:0.0004946141636865046\n",
+      "train loss:0.00013894234800012364\n",
+      "train loss:0.002674187643900661\n",
+      "train loss:0.0005985396948756983\n",
+      "train loss:0.002649453293392576\n",
+      "train loss:0.00026157586641198784\n",
+      "train loss:0.002227584555782466\n",
+      "train loss:0.0005190507824767979\n",
+      "train loss:0.0009781406854965976\n",
+      "train loss:0.0008812345198438854\n",
+      "train loss:0.0011814678110238653\n",
+      "train loss:0.0002437733850049643\n",
+      "train loss:0.0002556045662914611\n",
+      "train loss:0.0007577022689024528\n",
+      "train loss:0.0011509224773051468\n",
+      "train loss:0.0006778654979645811\n",
+      "train loss:0.00046662936256197446\n",
+      "train loss:0.0022671670245488834\n",
+      "train loss:0.003237351184217897\n",
+      "train loss:0.0013649602149454795\n",
+      "train loss:0.0008836354299455479\n",
+      "train loss:0.0006572610119760508\n",
+      "train loss:0.0008252898428169997\n",
+      "train loss:0.002107379552377114\n",
+      "train loss:0.00022065402345832028\n",
+      "train loss:0.022594065687966864\n",
+      "train loss:0.0005101759017936313\n",
+      "train loss:0.00010235677316545006\n",
+      "=== epoch:19, train acc:0.997, test acc:0.988 ===\n",
+      "train loss:0.00201562673727647\n",
+      "train loss:0.0005738821420634085\n",
+      "train loss:8.584274181179314e-05\n",
+      "train loss:0.000646741624033347\n",
+      "train loss:4.221455373470476e-05\n",
+      "train loss:0.0017331758638621708\n",
+      "train loss:0.0003555173176857092\n",
+      "train loss:0.0028662709547707392\n",
+      "train loss:0.0016401104696126803\n",
+      "train loss:0.0006290523269648842\n",
+      "train loss:0.004197247401317439\n",
+      "train loss:0.0003746467566445499\n",
+      "train loss:0.0003205023808302073\n",
+      "train loss:0.002698333408777993\n",
+      "train loss:0.0001722335426406409\n",
+      "train loss:0.003758922222542738\n",
+      "train loss:0.000571995878227311\n",
+      "train loss:0.0007160325461869798\n",
+      "train loss:0.00030654212997040497\n",
+      "train loss:0.00805401716426235\n",
+      "train loss:0.0010954957270582664\n",
+      "train loss:0.0013285688011125953\n",
+      "train loss:0.002432302451395929\n",
+      "train loss:0.0002126198619939483\n",
+      "train loss:0.0013277385640800552\n",
+      "train loss:3.6084589209132347e-05\n",
+      "train loss:0.00044517014622645704\n",
+      "train loss:0.0024223288946936135\n",
+      "train loss:0.007032399394834072\n",
+      "train loss:0.00421317161054626\n",
+      "train loss:0.0036575036170661403\n",
+      "train loss:0.0002901515788292286\n",
+      "train loss:0.004902332430008246\n",
+      "train loss:0.0005722351666270119\n",
+      "train loss:0.002208123155720256\n",
+      "train loss:0.0008660925771962748\n",
+      "train loss:0.0001645306644798551\n",
+      "train loss:0.0004413248429086699\n",
+      "train loss:0.004867894211282838\n",
+      "train loss:0.0010179494499526025\n",
+      "train loss:4.799572228552165e-05\n",
+      "train loss:8.950325016450355e-05\n",
+      "train loss:0.0013484538469949222\n",
+      "train loss:0.0016283393507306467\n",
+      "train loss:0.007471204943485437\n",
+      "train loss:0.00047778113743786245\n",
+      "train loss:0.006568268410619131\n",
+      "train loss:0.00154705190056182\n",
+      "train loss:5.990102657734085e-05\n",
+      "train loss:0.0011521274295942271\n",
+      "train loss:8.552913255871227e-05\n",
+      "train loss:0.0009174531283128963\n",
+      "train loss:0.0002425593112382108\n",
+      "train loss:0.0012163704795576766\n",
+      "train loss:0.0008225546920514215\n",
+      "train loss:0.0008744487109544416\n",
+      "train loss:0.0003378695874359519\n",
+      "train loss:0.0031679123248877393\n",
+      "train loss:0.0013734358864490783\n",
+      "train loss:0.0007683123532072091\n",
+      "train loss:0.0003684332423496734\n",
+      "train loss:8.905028207044287e-05\n",
+      "train loss:0.00023969121890444356\n",
+      "train loss:0.002312663565784003\n",
+      "train loss:0.005535709902674533\n",
+      "train loss:0.0021170988426417008\n",
+      "train loss:0.002393593272924301\n",
+      "train loss:0.0027635547237329335\n",
+      "train loss:0.002183398693004779\n",
+      "train loss:7.639680299867609e-05\n",
+      "train loss:0.0009357011258897599\n",
+      "train loss:0.00015994512183376553\n",
+      "train loss:0.001510084185387153\n",
+      "train loss:0.0004220938230620774\n",
+      "train loss:0.017172443918326123\n",
+      "train loss:0.0005551756777641205\n",
+      "train loss:0.002539242901107027\n",
+      "train loss:0.0006618786341030771\n",
+      "train loss:0.00011130550459460006\n",
+      "train loss:0.0003011537896396056\n",
+      "train loss:0.0027367423739486653\n",
+      "train loss:0.0008749496484705701\n",
+      "train loss:0.001354951445011406\n",
+      "train loss:9.431455925977238e-05\n",
+      "train loss:0.0002837738108929504\n",
+      "train loss:0.005298487047217685\n",
+      "train loss:0.0002456988232937063\n",
+      "train loss:0.001639404144778367\n",
+      "train loss:0.0008315946107248422\n",
+      "train loss:0.0029510799752572262\n",
+      "train loss:0.00033906366624857466\n",
+      "train loss:0.0017953643289535179\n",
+      "train loss:0.0008329891755657161\n",
+      "train loss:0.0009341825431847478\n",
+      "train loss:0.023924736479076603\n",
+      "train loss:0.0023254252558215878\n",
+      "train loss:0.004636674077047982\n",
+      "train loss:0.00014629320521406309\n",
+      "train loss:0.0012638818592574064\n",
+      "train loss:0.0011543306961161446\n",
+      "train loss:0.0010645127222807243\n",
+      "train loss:0.0002525618146565728\n",
+      "train loss:0.00010981706007977568\n",
+      "train loss:0.0011418238251197305\n",
+      "train loss:0.00026907573338869105\n",
+      "train loss:0.00835970570301537\n",
+      "train loss:0.00033525125777401075\n",
+      "train loss:0.0010487987016567935\n",
+      "train loss:0.004498035147237731\n",
+      "train loss:0.002444924378569104\n",
+      "train loss:0.008515039118977934\n",
+      "train loss:0.005121440307490713\n",
+      "train loss:0.00045516233746311263\n",
+      "train loss:0.0008701329662212882\n",
+      "train loss:0.0042407031248265364\n",
+      "train loss:0.00027504946476849217\n",
+      "train loss:0.0006765014528193099\n",
+      "train loss:0.041168012895678056\n",
+      "train loss:3.3288900870106754e-05\n",
+      "train loss:0.00015135656137786542\n",
+      "train loss:0.0003316138537802417\n",
+      "train loss:0.002384722201644558\n",
+      "train loss:0.0028987535862759657\n",
+      "train loss:0.001237956475727634\n",
+      "train loss:0.00041273366754099685\n",
+      "train loss:0.0007773064988105473\n",
+      "train loss:0.001891754282486741\n",
+      "train loss:0.0007823124137727366\n",
+      "train loss:0.018537969867314617\n",
+      "train loss:0.0015121544324406442\n",
+      "train loss:0.0004551413263382978\n",
+      "train loss:0.00346233680515184\n",
+      "train loss:0.002096378963786115\n",
+      "train loss:6.644812987597719e-05\n",
+      "train loss:0.00014842889974354918\n",
+      "train loss:0.003776390599851215\n",
+      "train loss:0.0031762863707388966\n",
+      "train loss:0.001712372333681476\n",
+      "train loss:0.0036252991015460743\n",
+      "train loss:0.00020819483279546186\n",
+      "train loss:0.00029498666249786295\n",
+      "train loss:0.00026604286607457135\n",
+      "train loss:0.0006674656087718843\n",
+      "train loss:0.011046733077596897\n",
+      "train loss:0.002844491881091754\n",
+      "train loss:0.001384760592289852\n",
+      "train loss:0.005681464087263197\n",
+      "train loss:0.00016518026857191301\n",
+      "train loss:0.0002485362417457427\n",
+      "train loss:0.000438651852280454\n",
+      "train loss:0.0007351154784804641\n",
+      "train loss:0.00023762850161507357\n",
+      "train loss:0.002168542409509366\n",
+      "train loss:0.0005854041609653163\n",
+      "train loss:0.0034368776536537653\n",
+      "train loss:0.004807251017783537\n",
+      "train loss:0.00016694203462981204\n",
+      "train loss:0.0008518171161053169\n",
+      "train loss:0.0015078026046227962\n",
+      "train loss:0.001597216904822876\n",
+      "train loss:0.005097177531994812\n",
+      "train loss:4.173643915215239e-05\n",
+      "train loss:0.006367570915876319\n",
+      "train loss:0.005653150876744977\n",
+      "train loss:0.00036103745884132836\n",
+      "train loss:0.00030469585599897245\n",
+      "train loss:0.0011146344702771289\n",
+      "train loss:0.00041729532618309397\n",
+      "train loss:0.027733572729749346\n",
+      "train loss:0.0005826072693332742\n",
+      "train loss:0.00611048399149567\n",
+      "train loss:0.00033191972410400903\n",
+      "train loss:0.009608710556373848\n",
+      "train loss:0.007135993859316434\n",
+      "train loss:0.012746025346633437\n",
+      "train loss:5.357021138749502e-05\n",
+      "train loss:0.0009724700244976611\n",
+      "train loss:0.0022421895860598967\n",
+      "train loss:0.00031043206379086004\n",
+      "train loss:0.0005611573045984767\n",
+      "train loss:0.00474430851230446\n",
+      "train loss:0.003210654240023339\n",
+      "train loss:0.002673339774837046\n",
+      "train loss:0.0030281717954051624\n",
+      "train loss:0.0006523479157631639\n",
+      "train loss:0.0016658141150751357\n",
+      "train loss:0.0035826560475634185\n",
+      "train loss:0.0006290968093489898\n",
+      "train loss:0.003315382188386398\n",
+      "train loss:0.002486259932165057\n",
+      "train loss:0.000579756054411207\n",
+      "train loss:8.121829959833062e-05\n",
+      "train loss:0.0015778563489285554\n",
+      "train loss:0.0004048463157683449\n",
+      "train loss:0.0003115987376515391\n",
+      "train loss:0.0016948048634879591\n",
+      "train loss:0.0005730551719601684\n",
+      "train loss:0.0006116651254999414\n",
+      "train loss:0.0019298740288611381\n",
+      "train loss:0.016392215387112406\n",
+      "train loss:0.000878923135796477\n",
+      "train loss:0.005330877709567838\n",
+      "train loss:0.0029896491411537233\n",
+      "train loss:0.003977474836956842\n",
+      "train loss:0.000982498737211159\n",
+      "train loss:0.002503608122322425\n",
+      "train loss:0.0022634240493438593\n",
+      "train loss:9.272619840380558e-05\n",
+      "train loss:0.00108174784443588\n",
+      "train loss:0.00017742496341430436\n",
+      "train loss:0.002747535938566787\n",
+      "train loss:0.0016798066593863787\n",
+      "train loss:0.003448902397315172\n",
+      "train loss:9.942222987120088e-05\n",
+      "train loss:0.0004317665145998878\n",
+      "train loss:0.0017721797204072525\n",
+      "train loss:0.002633748807825329\n",
+      "train loss:0.005823373309490812\n",
+      "train loss:0.0002552907480015143\n",
+      "train loss:0.0008304314866740185\n",
+      "train loss:0.00101618060161584\n",
+      "train loss:0.0002742195191514226\n",
+      "train loss:0.0029673447096534513\n",
+      "train loss:0.0010541931143333435\n",
+      "train loss:0.0008730192611471706\n",
+      "train loss:0.00048278691728056365\n",
+      "train loss:0.0007358073845739019\n",
+      "train loss:0.024766515959101114\n",
+      "train loss:0.000598266198801763\n",
+      "train loss:0.0003611643935058745\n",
+      "train loss:0.0002586400316807489\n",
+      "train loss:0.0017631355178854633\n",
+      "train loss:0.03512889254555709\n",
+      "train loss:0.004892176587574364\n",
+      "train loss:0.0023444210668304177\n",
+      "train loss:0.00464996004910188\n",
+      "train loss:0.0004998254249681981\n",
+      "train loss:0.027198669015190774\n",
+      "train loss:0.0003838884798553287\n",
+      "train loss:0.006235031318717982\n",
+      "train loss:0.006032732293399396\n",
+      "train loss:0.0026841630833612014\n",
+      "train loss:9.107957944663506e-05\n",
+      "train loss:0.0478707443539909\n",
+      "train loss:0.007439507334840413\n",
+      "train loss:5.85698031220205e-05\n",
+      "train loss:0.007414038413266884\n",
+      "train loss:0.0014315534505867588\n",
+      "train loss:0.0034082641435285504\n",
+      "train loss:0.0018395949559591477\n",
+      "train loss:0.0028274813872695274\n",
+      "train loss:0.00210866735814968\n",
+      "train loss:0.0022813818764869124\n",
+      "train loss:0.004216093912009445\n",
+      "train loss:0.004303640528944524\n",
+      "train loss:0.0010084138835416622\n",
+      "train loss:0.001013628595579011\n",
+      "train loss:0.00017886377236425315\n",
+      "train loss:0.0036720283523055735\n",
+      "train loss:0.0029889427702976183\n",
+      "train loss:0.006170841262436226\n",
+      "train loss:0.0056168085122332845\n",
+      "train loss:0.0014440209028591133\n",
+      "train loss:0.00017456258653666758\n",
+      "train loss:3.367394803116703e-05\n",
+      "train loss:0.0007139370905795557\n",
+      "train loss:0.0018612458174460064\n",
+      "train loss:0.0015725946628407093\n",
+      "train loss:0.0011340487084645448\n",
+      "train loss:0.008366606101848697\n",
+      "train loss:0.0050497570391574865\n",
+      "train loss:0.0006173029238928476\n",
+      "train loss:0.004140077788615883\n",
+      "train loss:0.0006205559213562187\n",
+      "train loss:0.0005478490688415936\n",
+      "train loss:0.0010041926868561916\n",
+      "train loss:0.0028541551383136393\n",
+      "train loss:0.005919807566736969\n",
+      "train loss:0.001143849916802972\n",
+      "train loss:0.0013219934650442302\n",
+      "train loss:0.0007171925795887306\n",
+      "train loss:0.00021240595048886018\n",
+      "train loss:9.079450799494716e-05\n",
+      "train loss:0.0008555387678533657\n",
+      "train loss:0.0006535275827521129\n",
+      "train loss:0.0028735281139254387\n",
+      "train loss:0.0023076875194869874\n",
+      "train loss:0.003122469393470128\n",
+      "train loss:0.0006195102275365334\n",
+      "train loss:0.00019871291327192106\n",
+      "train loss:0.0015783756336767171\n",
+      "train loss:0.0019700521011579305\n",
+      "train loss:0.00020755383613426967\n",
+      "train loss:0.0008998388768699287\n",
+      "train loss:0.0009878041807769272\n",
+      "train loss:0.0004566369643070693\n",
+      "train loss:0.0010107434104172663\n",
+      "train loss:0.0015771535242870116\n",
+      "train loss:7.145147628787198e-05\n",
+      "train loss:0.0001477578521052238\n",
+      "train loss:0.0014538028419469628\n",
+      "train loss:0.00028519323090133857\n",
+      "train loss:0.0014461469528576094\n",
+      "train loss:0.0012001775394762245\n",
+      "train loss:0.00020923390341947098\n",
+      "train loss:0.004775955554646396\n",
+      "train loss:0.009968608803728596\n",
+      "train loss:0.0011461611835624544\n",
+      "train loss:7.782287775413652e-05\n",
+      "train loss:0.0008099063554755522\n",
+      "train loss:0.0036101670116682714\n",
+      "train loss:0.0010554468811131935\n",
+      "train loss:0.0002780517107854672\n",
+      "train loss:0.0016167932249672313\n",
+      "train loss:0.0003202299202396304\n",
+      "train loss:0.000869576363022842\n",
+      "train loss:0.02284751155813229\n",
+      "train loss:0.0025063513084733076\n",
+      "train loss:0.008675362673555289\n",
+      "train loss:0.0021909973541238046\n",
+      "train loss:0.001939820902244141\n",
+      "train loss:0.00028808559823077963\n",
+      "train loss:0.0002912078662617112\n",
+      "train loss:0.0005154776855868839\n",
+      "train loss:0.002332781900759128\n",
+      "train loss:9.086141929328353e-05\n",
+      "train loss:0.0026398756605815464\n",
+      "train loss:0.007072997043472512\n",
+      "train loss:0.0005828139590274104\n",
+      "train loss:0.0004141946655607246\n",
+      "train loss:0.002807064091364975\n",
+      "train loss:0.0014425620804020534\n",
+      "train loss:0.001969390687009602\n",
+      "train loss:0.0006980814470788889\n",
+      "train loss:0.0003870200649133003\n",
+      "train loss:8.941715814052527e-05\n",
+      "train loss:0.0036943851329821133\n",
+      "train loss:0.0841729254000139\n",
+      "train loss:0.0006543640372564392\n",
+      "train loss:0.004645010709393847\n",
+      "train loss:0.005773031932812265\n",
+      "train loss:0.00011594964066509634\n",
+      "train loss:0.0004847760609787515\n",
+      "train loss:0.000977065659430292\n",
+      "train loss:0.0005370695717950736\n",
+      "train loss:0.0003042438031744199\n",
+      "train loss:0.001160091822054909\n",
+      "train loss:5.6208544613496785e-05\n",
+      "train loss:8.576191336236843e-05\n",
+      "train loss:0.0007949063856598358\n",
+      "train loss:0.00025953187552335566\n",
+      "train loss:0.0009394369105703679\n",
+      "train loss:0.0075852417609582525\n",
+      "train loss:7.651914645177443e-05\n",
+      "train loss:0.00047532101899023913\n",
+      "train loss:0.0011598855151222496\n",
+      "train loss:0.00021808518400729952\n",
+      "train loss:0.0003209493673684025\n",
+      "train loss:0.002306948483491966\n",
+      "train loss:0.030996265025029227\n",
+      "train loss:0.0012993583701120285\n",
+      "train loss:0.0005679186342692227\n",
+      "train loss:0.001659753305221893\n",
+      "train loss:0.007541509441909535\n",
+      "train loss:0.0013550614663562872\n",
+      "train loss:0.000514837484089126\n",
+      "train loss:0.0016579514969157377\n",
+      "train loss:0.0026496588014587397\n",
+      "train loss:0.0008271099954844008\n",
+      "train loss:0.002508382946058963\n",
+      "train loss:0.003543976173548667\n",
+      "train loss:0.00022772169345163968\n",
+      "train loss:0.0006886773754489764\n",
+      "train loss:0.00046552146095566995\n",
+      "train loss:0.0013247505044254007\n",
+      "train loss:0.00014165743498813032\n",
+      "train loss:0.0011338607897472494\n",
+      "train loss:0.0003182577273590469\n",
+      "train loss:0.0016046450326717193\n",
+      "train loss:0.0003422058743844433\n",
+      "train loss:0.033463511398700116\n",
+      "train loss:0.000991904317115498\n",
+      "train loss:0.004249333006745761\n",
+      "train loss:0.0005864282055251304\n",
+      "train loss:0.0027191155151357668\n",
+      "train loss:0.0032271863429720825\n",
+      "train loss:0.0005664587463699685\n",
+      "train loss:0.0012369243813821868\n",
+      "train loss:0.0002434108388913198\n",
+      "train loss:8.483089078941899e-05\n",
+      "train loss:0.0002259032033877728\n",
+      "train loss:0.0012361167925284018\n",
+      "train loss:0.0038905226605616657\n",
+      "train loss:0.00013210851155146712\n",
+      "train loss:0.0001366292703322705\n",
+      "train loss:9.144706460096856e-05\n",
+      "train loss:0.0001348209889548926\n",
+      "train loss:0.0006089205226553399\n",
+      "train loss:0.00038976270713085315\n",
+      "train loss:0.0024821740177786654\n",
+      "train loss:0.0008800831244150058\n",
+      "train loss:0.0002845254076083477\n",
+      "train loss:5.502842107027729e-05\n",
+      "train loss:0.0034608518367675034\n",
+      "train loss:0.00045882714207694735\n",
+      "train loss:5.6265931654745603e-05\n",
+      "train loss:0.0005243654561597552\n",
+      "train loss:0.0024458822133993125\n",
+      "train loss:0.00013899935000366405\n",
+      "train loss:0.00011356573694417582\n",
+      "train loss:0.00018716744124576003\n",
+      "train loss:0.0002253614479048799\n",
+      "train loss:0.0007682784076920035\n",
+      "train loss:0.0005933607117697006\n",
+      "train loss:0.00020636707936150285\n",
+      "train loss:0.0016473009054688847\n",
+      "train loss:0.0009648305643258276\n",
+      "train loss:0.0007602042587731963\n",
+      "train loss:0.00191305305151286\n",
+      "train loss:0.0004906616589757719\n",
+      "train loss:9.289524407918928e-05\n",
+      "train loss:0.0006769777813715402\n",
+      "train loss:0.0013396614504627025\n",
+      "train loss:0.00014178154597280058\n",
+      "train loss:0.00022133925231092587\n",
+      "train loss:0.001024759005523674\n",
+      "train loss:0.008665840571297831\n",
+      "train loss:0.0003155442785148567\n",
+      "train loss:4.2369365314337985e-05\n",
+      "train loss:0.0015541317828533562\n",
+      "train loss:3.157734954192132e-05\n",
+      "train loss:0.001621785782490444\n",
+      "train loss:0.005024154367888245\n",
+      "train loss:0.0001344958356597919\n",
+      "train loss:0.00020723697078415352\n",
+      "train loss:0.009011128439449763\n",
+      "train loss:0.000979951857468748\n",
+      "train loss:0.0010614486244896338\n",
+      "train loss:7.65491058154957e-05\n",
+      "train loss:0.0016451286483667405\n",
+      "train loss:0.0005048393302178722\n",
+      "train loss:0.0005503543493214078\n",
+      "train loss:0.00578654471850681\n",
+      "train loss:0.007472172606871271\n",
+      "train loss:0.0016590542764813763\n",
+      "train loss:0.00015031941047846827\n",
+      "train loss:0.00019627523845281282\n",
+      "train loss:0.0002049852914258531\n",
+      "train loss:0.0017579260418005603\n",
+      "train loss:0.0003421723866009297\n",
+      "train loss:0.0008199465926888073\n",
+      "train loss:0.0016231027917538955\n",
+      "train loss:0.004036178538447944\n",
+      "train loss:0.000958693245927841\n",
+      "train loss:0.0008204892121547831\n",
+      "train loss:0.0019381273045792295\n",
+      "train loss:0.000945267672571838\n",
+      "train loss:0.000709042431886426\n",
+      "train loss:0.0003013031015133319\n",
+      "train loss:0.0004241414351309533\n",
+      "train loss:0.006384110608368639\n",
+      "train loss:0.00012920187321641973\n",
+      "train loss:0.0012066475460768542\n",
+      "train loss:0.0075370546384813004\n",
+      "train loss:0.006112779063997176\n",
+      "train loss:0.00026868426821866477\n",
+      "train loss:0.0006371404276123883\n",
+      "train loss:0.0002149244205567416\n",
+      "train loss:4.863376760757092e-05\n",
+      "train loss:0.003951825300239647\n",
+      "train loss:0.0010578021645361502\n",
+      "train loss:0.000797873801506392\n",
+      "train loss:0.0033573005806040923\n",
+      "train loss:0.0026607042151810647\n",
+      "train loss:0.0039401141924132735\n",
+      "train loss:0.002469701526503155\n",
+      "train loss:0.002240215165863572\n",
+      "train loss:0.003130014094916654\n",
+      "train loss:0.000706193058471721\n",
+      "train loss:0.0028189488776458062\n",
+      "train loss:0.01732086182853873\n",
+      "train loss:0.001429764705252115\n",
+      "train loss:0.0008190795260651664\n",
+      "train loss:0.009298371021012225\n",
+      "train loss:0.00040176678809288003\n",
+      "train loss:0.0003754179156305222\n",
+      "train loss:0.0005859743705516128\n",
+      "train loss:0.0038411070289117465\n",
+      "train loss:0.00020916988875515976\n",
+      "train loss:0.007259732681302364\n",
+      "train loss:0.005728377278139529\n",
+      "train loss:0.0020712612837809905\n",
+      "train loss:6.956316547755206e-05\n",
+      "train loss:0.0002278824650866555\n",
+      "train loss:0.003067141189208879\n",
+      "train loss:0.0011341443883366852\n",
+      "train loss:0.0016955849325181582\n",
+      "train loss:0.00048463488570750344\n",
+      "train loss:0.0008152306792455408\n",
+      "train loss:0.0001313403917158109\n",
+      "train loss:0.00014830677024366503\n",
+      "train loss:0.007980224560968958\n",
+      "train loss:0.00025035843042210665\n",
+      "train loss:0.0036522074015055135\n",
+      "train loss:0.0006900979711607576\n",
+      "train loss:0.05889396021090247\n",
+      "train loss:0.03234272275410196\n",
+      "train loss:0.0025189445568412306\n",
+      "train loss:0.0018024223029741646\n",
+      "train loss:0.00021806282892738955\n",
+      "train loss:0.0005225352429549934\n",
+      "train loss:0.0001492795494826197\n",
+      "train loss:0.00018721075322119404\n",
+      "train loss:0.0002568299945009554\n",
+      "train loss:0.0028006525456081154\n",
+      "train loss:0.004817970816075981\n",
+      "train loss:0.00031133708945982135\n",
+      "train loss:0.010466009359128566\n",
+      "train loss:0.001217897553092976\n",
+      "train loss:0.000513472423418304\n",
+      "train loss:0.00294928042273202\n",
+      "train loss:0.0005576294929338445\n",
+      "train loss:0.0011793106223474893\n",
+      "train loss:0.0026778636371399965\n",
+      "train loss:0.0016010436044605622\n",
+      "train loss:0.0008360925993324805\n",
+      "train loss:0.04330548665778476\n",
+      "train loss:0.0034121579697360053\n",
+      "train loss:0.006154588281945171\n",
+      "train loss:0.0002321805921926395\n",
+      "train loss:0.0009756304338040253\n",
+      "train loss:0.0012720136792768969\n",
+      "train loss:0.0008111579826422985\n",
+      "train loss:0.00020225085570226025\n",
+      "train loss:0.0014845095336292846\n",
+      "train loss:0.0006903652483634244\n",
+      "train loss:0.003033536037125984\n",
+      "train loss:0.0009740468048759786\n",
+      "train loss:0.003214252629359056\n",
+      "train loss:0.0018524852840957073\n",
+      "train loss:0.0033013364852061246\n",
+      "train loss:0.000943464271813532\n",
+      "train loss:0.0009146952856603506\n",
+      "train loss:0.003914981444115151\n",
+      "train loss:0.05550105814897471\n",
+      "train loss:0.0003096186089576023\n",
+      "train loss:0.03722798976347605\n",
+      "train loss:0.0037890241475087376\n",
+      "train loss:0.0003372445838245382\n",
+      "train loss:0.0013659475203158095\n",
+      "train loss:0.021017577641938158\n",
+      "train loss:0.01060645322549989\n",
+      "train loss:3.789638324708787e-05\n",
+      "train loss:0.0004575531923380112\n",
+      "train loss:0.0004078292153083727\n",
+      "train loss:0.0016315281103408794\n",
+      "train loss:0.00011531067541335383\n",
+      "train loss:0.0003769209781876958\n",
+      "train loss:0.003335348569822409\n",
+      "train loss:0.003105881457863847\n",
+      "train loss:0.0007226024632956497\n",
+      "train loss:9.562816303370371e-05\n",
+      "train loss:0.004096060887382989\n",
+      "train loss:0.0001804082282854047\n",
+      "train loss:0.0014500579914384372\n",
+      "train loss:0.00018680917687264765\n",
+      "train loss:0.0010637874935106725\n",
+      "train loss:0.0034769577851601636\n",
+      "train loss:0.0403003740122033\n",
+      "train loss:0.0011443698961990704\n",
+      "train loss:0.0013550813376324298\n",
+      "train loss:0.005036035180184865\n",
+      "train loss:0.001137757382815503\n",
+      "train loss:0.006842024663262129\n",
+      "train loss:0.0010361746246836908\n",
+      "train loss:0.0018744734122034753\n",
+      "train loss:0.00021488869827814155\n",
+      "train loss:0.004690074235887949\n",
+      "train loss:0.003259761927713636\n",
+      "train loss:0.001445923885069762\n",
+      "train loss:0.0018321784638876911\n",
+      "train loss:0.00041066486803164324\n",
+      "train loss:0.002595856217709082\n",
+      "train loss:0.002292120977327015\n",
+      "train loss:0.003270356384322628\n",
+      "train loss:0.0002526832185564584\n",
+      "train loss:0.004436755774814239\n",
+      "train loss:0.0007699347384054879\n",
+      "train loss:0.0011590679741262985\n",
+      "train loss:0.0009052868668999485\n",
+      "train loss:0.0011120933406526471\n",
+      "train loss:0.0005047541683990603\n",
+      "train loss:0.0002747686743207911\n",
+      "train loss:0.0006356508417106651\n",
+      "train loss:0.00034948971010808293\n",
+      "train loss:0.0005160475152418321\n",
+      "train loss:0.0012043508712382205\n",
+      "train loss:8.597500915004078e-05\n",
+      "train loss:0.0021310802520638603\n",
+      "train loss:0.0012926688269895923\n",
+      "=== epoch:20, train acc:0.998, test acc:0.986 ===\n",
+      "train loss:0.0009425861847802156\n",
+      "train loss:0.0012388363971579009\n",
+      "train loss:9.101555403906664e-05\n",
+      "train loss:0.004728422802568407\n",
+      "train loss:0.00023838271249729205\n",
+      "train loss:0.00041537156864013497\n",
+      "train loss:0.0075691746601021645\n",
+      "train loss:0.000647486959083769\n",
+      "train loss:0.0002895595479668609\n",
+      "train loss:0.0029580294070317864\n",
+      "train loss:0.0004927343516117552\n",
+      "train loss:0.0003518224345357958\n",
+      "train loss:0.008083003383835663\n",
+      "train loss:0.0013034784509826673\n",
+      "train loss:0.004910388911753011\n",
+      "train loss:0.0012538773133524198\n",
+      "train loss:0.00014605684340317907\n",
+      "train loss:0.003601883869544861\n",
+      "train loss:0.0003522992197519906\n",
+      "train loss:0.0015467262595895647\n",
+      "train loss:0.0012919672774440944\n",
+      "train loss:0.0031827785314142913\n",
+      "train loss:0.002255624129629947\n",
+      "train loss:0.0018224670070908422\n",
+      "train loss:0.0065667043291677985\n",
+      "train loss:0.00021349643823793583\n",
+      "train loss:0.0007646257933801858\n",
+      "train loss:0.00011087012988483565\n",
+      "train loss:0.004203865229227702\n",
+      "train loss:0.0009517126023603621\n",
+      "train loss:0.0003421244643227871\n",
+      "train loss:0.0008739422467270372\n",
+      "train loss:0.000890651783206102\n",
+      "train loss:0.0003434185165136161\n",
+      "train loss:0.00937542751428539\n",
+      "train loss:0.0005081690247327377\n",
+      "train loss:0.0019505682320912043\n",
+      "train loss:0.0002537208435086196\n",
+      "train loss:0.01267354884184957\n",
+      "train loss:0.0008333877247637054\n",
+      "train loss:0.001100077346866\n",
+      "train loss:0.00047709160376524284\n",
+      "train loss:0.0018677662601679895\n",
+      "train loss:0.003728712862791359\n",
+      "train loss:0.0020796971992892263\n",
+      "train loss:0.0014110030499124473\n",
+      "train loss:0.00016603797273870947\n",
+      "train loss:0.00023591173801858673\n",
+      "train loss:0.0001496385343514474\n",
+      "train loss:0.00031028622249515604\n",
+      "train loss:0.012559667058977654\n",
+      "train loss:0.0013072865568489098\n",
+      "train loss:0.0016414125901341855\n",
+      "train loss:0.00012245526809393778\n",
+      "train loss:0.007964424427532597\n",
+      "train loss:0.000789095033537243\n",
+      "train loss:0.00231515994261443\n",
+      "train loss:0.007090388467682131\n",
+      "train loss:0.0004823213043491772\n",
+      "train loss:0.0009401341246989569\n",
+      "train loss:0.00397880956994372\n",
+      "train loss:0.0014521595276336256\n",
+      "train loss:0.0015376197715854024\n",
+      "train loss:0.002840369131081458\n",
+      "train loss:0.001903359297839217\n",
+      "train loss:0.0013096914726940011\n",
+      "train loss:0.0008669870959536692\n",
+      "train loss:0.00042782191641501787\n",
+      "train loss:0.010935201233907224\n",
+      "train loss:0.0052576146498982505\n",
+      "train loss:0.00354826530355679\n",
+      "train loss:0.003091662090025083\n",
+      "train loss:0.00024098287628655817\n",
+      "train loss:0.0017609086261463202\n",
+      "train loss:0.0011708998077138112\n",
+      "train loss:0.005457382123405183\n",
+      "train loss:0.0006475488200460526\n",
+      "train loss:0.001102485700035777\n",
+      "train loss:0.00031584104980218996\n",
+      "train loss:0.0005227499943363776\n",
+      "train loss:0.00015927457821359812\n",
+      "train loss:0.0002993828642307263\n",
+      "train loss:0.008854881519005355\n",
+      "train loss:0.00023548980955284844\n",
+      "train loss:0.0023262034235469586\n",
+      "train loss:0.00042037920062980793\n",
+      "train loss:0.001763647657674306\n",
+      "train loss:0.0007220522435549287\n",
+      "train loss:0.00024069305000101672\n",
+      "train loss:0.0003251168370930077\n",
+      "train loss:0.0004552240265334595\n",
+      "train loss:0.0019033005103227657\n",
+      "train loss:0.001733801696223089\n",
+      "train loss:0.0022210213565775428\n",
+      "train loss:0.0022578703289856815\n",
+      "train loss:0.00014959081019652495\n",
+      "train loss:0.00893629861174099\n",
+      "train loss:0.00030062477377721563\n",
+      "train loss:0.0014283274424947098\n",
+      "train loss:0.008053185636835733\n",
+      "train loss:0.00016945357366119834\n",
+      "train loss:0.00014829339008212044\n",
+      "train loss:0.0015677653684597675\n",
+      "train loss:0.005616514744878929\n",
+      "train loss:0.0006464268689205742\n",
+      "train loss:0.0012122076912671354\n",
+      "train loss:0.0004833958915189033\n",
+      "train loss:3.502771549600554e-05\n",
+      "train loss:0.0005713796249051744\n",
+      "train loss:0.0036116738501899585\n",
+      "train loss:0.003173443903287778\n",
+      "train loss:8.444042003322916e-05\n",
+      "train loss:0.00037797267230871957\n",
+      "train loss:8.129496695173586e-05\n",
+      "train loss:0.0002766763545309833\n",
+      "train loss:0.00022361016491818928\n",
+      "train loss:0.002090536836920973\n",
+      "train loss:0.0005859137088015143\n",
+      "train loss:0.0002476141588577824\n",
+      "train loss:0.0007226559314355051\n",
+      "train loss:0.00015085053959417978\n",
+      "train loss:0.0023841236000502436\n",
+      "train loss:0.0006499789937835087\n",
+      "train loss:0.00039305286491898796\n",
+      "train loss:0.0006957451384045457\n",
+      "train loss:0.00027135745357328853\n",
+      "train loss:0.002012052038573128\n",
+      "train loss:0.001207424023239694\n",
+      "train loss:0.0002264568815983594\n",
+      "train loss:0.005071936635134354\n",
+      "train loss:0.0014838145989027544\n",
+      "train loss:0.000160393054232794\n",
+      "train loss:0.0005939029148699644\n",
+      "train loss:0.005073864462708926\n",
+      "train loss:0.0018857263963152854\n",
+      "train loss:0.004981493386087465\n",
+      "train loss:0.001146199007322667\n",
+      "train loss:0.004179061784502698\n",
+      "train loss:0.00013004066934308453\n",
+      "train loss:0.0002368817517524879\n",
+      "train loss:0.00047235813189443654\n",
+      "train loss:0.0010458552472576335\n",
+      "train loss:0.0021140566199371238\n",
+      "train loss:0.0020068707779158867\n",
+      "train loss:9.20833474125542e-05\n",
+      "train loss:0.001037585381317053\n",
+      "train loss:0.00027462842758606086\n",
+      "train loss:0.000561351606707536\n",
+      "train loss:0.017014664304499008\n",
+      "train loss:0.0017174375700743265\n",
+      "train loss:0.0016031245481580429\n",
+      "train loss:0.0004239245806847206\n",
+      "train loss:0.00011524981545295205\n",
+      "train loss:0.002427140700510552\n",
+      "train loss:0.0008079705567012836\n",
+      "train loss:0.000256616387449517\n",
+      "train loss:0.003203452418382935\n",
+      "train loss:0.00030443271278705333\n",
+      "train loss:0.00037821171605104635\n",
+      "train loss:0.0007139961914450623\n",
+      "train loss:0.0006759546284906755\n",
+      "train loss:0.00045796253773051633\n",
+      "train loss:0.0008742250551499956\n",
+      "train loss:0.002487365383698131\n",
+      "train loss:0.005765443114840585\n",
+      "train loss:0.002233866289703528\n",
+      "train loss:0.0004954872729608948\n",
+      "train loss:0.001966617346697114\n",
+      "train loss:3.13622050687855e-05\n",
+      "train loss:4.53245503358474e-05\n",
+      "train loss:0.00014279642342877858\n",
+      "train loss:3.6862509689700825e-05\n",
+      "train loss:0.00019730196228709776\n",
+      "train loss:0.000748150142329769\n",
+      "train loss:0.00030703806851618186\n",
+      "train loss:0.003409282330285271\n",
+      "train loss:0.0014972721331696765\n",
+      "train loss:0.003393905763582497\n",
+      "train loss:0.00014907415154635957\n",
+      "train loss:0.0016093228475829167\n",
+      "train loss:0.0012530271978802088\n",
+      "train loss:0.00468442111268226\n",
+      "train loss:0.0012119511152496511\n",
+      "train loss:0.0019748367202385934\n",
+      "train loss:0.0001272322588914395\n",
+      "train loss:7.487761177828373e-06\n",
+      "train loss:0.0004079513715561998\n",
+      "train loss:0.00020810570709084206\n",
+      "train loss:0.0010232006915663005\n",
+      "train loss:2.5645424956172078e-05\n",
+      "train loss:0.002114165030172764\n",
+      "train loss:0.0016065528103116544\n",
+      "train loss:0.0024585292386469794\n",
+      "train loss:0.00037404360900281016\n",
+      "train loss:0.002326003123258971\n",
+      "train loss:0.00029705358264903924\n",
+      "train loss:0.0002481026696519242\n",
+      "train loss:0.0005359791485046928\n",
+      "train loss:0.009335142473862744\n",
+      "train loss:0.000343467037724635\n",
+      "train loss:0.001930253466526114\n",
+      "train loss:0.006991514824123902\n",
+      "train loss:0.0003632586065887616\n",
+      "train loss:0.002846721262749492\n",
+      "train loss:0.0002655342610402095\n",
+      "train loss:0.0008598675611555981\n",
+      "train loss:0.0025812205612426725\n",
+      "train loss:0.00013960764281782274\n",
+      "train loss:0.013390082275221083\n",
+      "train loss:0.007860822828198588\n",
+      "train loss:0.000565547014769682\n",
+      "train loss:0.0003562673731853779\n",
+      "train loss:8.560054121321916e-05\n",
+      "train loss:0.0005814400336847422\n",
+      "train loss:0.00028061368149923284\n",
+      "train loss:0.0016857260102265037\n",
+      "train loss:0.012693351770905336\n",
+      "train loss:0.00032612040107412893\n",
+      "train loss:0.0007966389289124369\n",
+      "train loss:0.00027469027683662693\n",
+      "train loss:4.3775809756334204e-05\n",
+      "train loss:0.001956635579243276\n",
+      "train loss:8.76552773640285e-05\n",
+      "train loss:0.00041835575408297437\n",
+      "train loss:6.526030292048548e-05\n",
+      "train loss:0.0014689498155823024\n",
+      "train loss:0.0030637728696215278\n",
+      "train loss:0.0006777306230672887\n",
+      "train loss:0.00015190570605915464\n",
+      "train loss:0.002095369389451946\n",
+      "train loss:0.0015710918981686574\n",
+      "train loss:0.0020646493654179593\n",
+      "train loss:0.0002612243751442766\n",
+      "train loss:0.0022135394324263864\n",
+      "train loss:0.02211458908860858\n",
+      "train loss:0.0004888938885244338\n",
+      "train loss:0.0008544214237611805\n",
+      "train loss:0.00024353056743609512\n",
+      "train loss:0.00020952349302581555\n",
+      "train loss:0.000764681945270984\n",
+      "train loss:0.0005116845633989662\n",
+      "train loss:0.0004345605608874708\n",
+      "train loss:0.00015537186883517683\n",
+      "train loss:0.00444215376771068\n",
+      "train loss:0.0027547638211427084\n",
+      "train loss:0.0014594997126293652\n",
+      "train loss:0.00014093963671174594\n",
+      "train loss:0.003389646870211599\n",
+      "train loss:0.004727999547109214\n",
+      "train loss:0.0008863494392802324\n",
+      "train loss:0.00025011560399679824\n",
+      "train loss:9.488386529665633e-05\n",
+      "train loss:0.00019391833554872212\n",
+      "train loss:0.0006041402057030052\n",
+      "train loss:0.0010045366104686896\n",
+      "train loss:0.0020727327829625384\n",
+      "train loss:0.00021825303258031715\n",
+      "train loss:0.0008084639309995906\n",
+      "train loss:0.0020882465727042913\n",
+      "train loss:0.002683579713531566\n",
+      "train loss:0.00038924262084869037\n",
+      "train loss:0.0017367259575286822\n",
+      "train loss:0.00013491174486179769\n",
+      "train loss:0.00019315241095744172\n",
+      "train loss:0.0002507826911201591\n",
+      "train loss:0.0004846486620904259\n",
+      "train loss:0.00024924696981304964\n",
+      "train loss:4.791431846384136e-05\n",
+      "train loss:0.0011547323505677477\n",
+      "train loss:0.0003429478766962117\n",
+      "train loss:0.002119932073379297\n",
+      "train loss:8.377224214868108e-05\n",
+      "train loss:0.0005114312977761742\n",
+      "train loss:0.0002671412920282315\n",
+      "train loss:5.008843826708595e-05\n",
+      "train loss:0.0014906129790537204\n",
+      "train loss:0.002595710896341383\n",
+      "train loss:0.0009368810309852091\n",
+      "train loss:0.0003648482437580534\n",
+      "train loss:0.000333050132961956\n",
+      "train loss:0.00035877012487198754\n",
+      "train loss:0.001971957671426692\n",
+      "train loss:0.0014041749397305462\n",
+      "train loss:0.00010105124261527269\n",
+      "train loss:0.001524815234359918\n",
+      "train loss:0.002391429915599316\n",
+      "train loss:0.0002102312124738284\n",
+      "train loss:0.00014850143377009668\n",
+      "train loss:0.00041462461948765\n",
+      "train loss:0.000357332442624609\n",
+      "train loss:0.003468977893794626\n",
+      "train loss:0.0010485145565369343\n",
+      "train loss:0.00026931025542976434\n",
+      "train loss:0.0005296147277379873\n",
+      "train loss:0.0005903896222498539\n",
+      "train loss:0.00048249610611469164\n",
+      "train loss:0.00017362218347893507\n",
+      "train loss:7.908355831710902e-05\n",
+      "train loss:0.0010751678919641073\n",
+      "train loss:0.0007560364283239174\n",
+      "train loss:0.00010365637583925882\n",
+      "train loss:0.00023221183944872162\n",
+      "train loss:0.0020047202178928216\n",
+      "train loss:0.0008201433678076563\n",
+      "train loss:6.603841057659462e-05\n",
+      "train loss:0.0018726082259801372\n",
+      "train loss:0.000427298614537771\n",
+      "train loss:0.0006512422715800708\n",
+      "train loss:0.0003570567123374701\n",
+      "train loss:0.01504500378413068\n",
+      "train loss:0.0006352561995676522\n",
+      "train loss:0.0007086743091663452\n",
+      "train loss:0.0012775544018596193\n",
+      "train loss:3.1077320830054995e-05\n",
+      "train loss:0.0007613096951401357\n",
+      "train loss:0.011405662999800588\n",
+      "train loss:0.002654078832693878\n",
+      "train loss:0.0012766922294892133\n",
+      "train loss:0.0007289622243255747\n",
+      "train loss:0.0008893560059015253\n",
+      "train loss:0.002910471589144093\n",
+      "train loss:0.0004848187217231621\n",
+      "train loss:0.0008316566587841973\n",
+      "train loss:0.02857819490304024\n",
+      "train loss:0.0004949785125775109\n",
+      "train loss:0.004651712362509775\n",
+      "train loss:8.623962137258477e-05\n",
+      "train loss:0.0007570575411507611\n",
+      "train loss:0.0008602871249397819\n",
+      "train loss:0.0016466805831438728\n",
+      "train loss:0.0006324202675624813\n",
+      "train loss:9.845203219919959e-05\n",
+      "train loss:0.025076208545402066\n",
+      "train loss:0.00025051675268889564\n",
+      "train loss:0.0011270128046603689\n",
+      "train loss:0.0009126996814187498\n",
+      "train loss:0.0002470057160646998\n",
+      "train loss:0.0004596824322795424\n",
+      "train loss:0.0023667299489506344\n",
+      "train loss:9.054018496489466e-05\n",
+      "train loss:0.0028867802492717532\n",
+      "train loss:0.00034749743003443303\n",
+      "train loss:0.0001659326073922809\n",
+      "train loss:0.0011458760842771867\n",
+      "train loss:0.00040581812868240524\n",
+      "train loss:0.0007268107723639776\n",
+      "train loss:0.0007815921429208093\n",
+      "train loss:4.726653396168877e-05\n",
+      "train loss:0.0012895903289368157\n",
+      "train loss:0.001336429694547479\n",
+      "train loss:0.002246395423447744\n",
+      "train loss:0.00033630266861504746\n",
+      "train loss:0.003834508153360563\n",
+      "train loss:0.0006894325048725972\n",
+      "train loss:0.0006612270840826455\n",
+      "train loss:0.0009997056854919844\n",
+      "train loss:0.00029278139964484086\n",
+      "train loss:0.0016687564785534883\n",
+      "train loss:1.6691474643913894e-05\n",
+      "train loss:0.00032293635876864415\n",
+      "train loss:0.00031050612572582466\n",
+      "train loss:0.001861239707948453\n",
+      "train loss:0.0008627022525421672\n",
+      "train loss:0.003750777704089753\n",
+      "train loss:0.00020954819436083685\n",
+      "train loss:0.0020122954793865304\n",
+      "train loss:0.001508332858636077\n",
+      "train loss:0.00032108575942297387\n",
+      "train loss:0.00024322969273723183\n",
+      "train loss:0.0015627097682159772\n",
+      "train loss:0.0015391672459574123\n",
+      "train loss:0.00114134023655816\n",
+      "train loss:0.0016682191785412648\n",
+      "train loss:0.00021238025711776097\n",
+      "train loss:9.804488063335371e-05\n",
+      "train loss:0.00022379109196076706\n",
+      "train loss:0.0004104436404655618\n",
+      "train loss:0.0010404273034931237\n",
+      "train loss:0.0005223671184075929\n",
+      "train loss:0.0027655563945111995\n",
+      "train loss:0.0019843848187221036\n",
+      "train loss:0.00037010783579957733\n",
+      "train loss:5.432449179459114e-05\n",
+      "train loss:8.093771493979583e-05\n",
+      "train loss:0.0004989762215389655\n",
+      "train loss:0.0010931575275523236\n",
+      "train loss:4.075714234806638e-05\n",
+      "train loss:0.0009382236766928662\n",
+      "train loss:6.299269680026603e-05\n",
+      "train loss:0.000159304562071122\n",
+      "train loss:3.3449180697312816e-05\n",
+      "train loss:0.0003903193227378153\n",
+      "train loss:0.00246885174514562\n",
+      "train loss:0.0004511198198763038\n",
+      "train loss:0.0013570610539983224\n",
+      "train loss:0.0003583157313954908\n",
+      "train loss:8.28041059038483e-05\n",
+      "train loss:0.0005798687664197886\n",
+      "train loss:0.0022749996687549605\n",
+      "train loss:0.00021920874792091577\n",
+      "train loss:0.00018199439607921434\n",
+      "train loss:0.00014656890700848839\n",
+      "train loss:0.004715619534438283\n",
+      "train loss:0.0009163064736904939\n",
+      "train loss:1.1430661731761482e-05\n",
+      "train loss:0.0019417988142812924\n",
+      "train loss:0.0014202623005032237\n",
+      "train loss:1.1257841843530857e-05\n",
+      "train loss:0.0005314283504522823\n",
+      "train loss:0.00011695861589034135\n",
+      "train loss:0.00029891699637397506\n",
+      "train loss:0.0013037986611950036\n",
+      "train loss:0.0005119191333642605\n",
+      "train loss:0.01716401218539764\n",
+      "train loss:0.003517752838209627\n",
+      "train loss:0.00014838949000064983\n",
+      "train loss:0.002314176837163169\n",
+      "train loss:0.0014775497879213025\n",
+      "train loss:0.00031392067246595226\n",
+      "train loss:0.01933280489539096\n",
+      "train loss:4.67416382514252e-05\n",
+      "train loss:2.733441900064944e-05\n",
+      "train loss:0.0018614484710271246\n",
+      "train loss:0.00011388598438773198\n",
+      "train loss:9.531057465130018e-05\n",
+      "train loss:0.000286888120332867\n",
+      "train loss:0.001559253817948824\n",
+      "train loss:0.0002405721730903334\n",
+      "train loss:0.0029190211853752633\n",
+      "train loss:0.000753697107923011\n",
+      "train loss:0.002083793516321092\n",
+      "train loss:0.0013303413410174158\n",
+      "train loss:0.00025875165913693854\n",
+      "train loss:0.0001907645721331581\n",
+      "train loss:0.00027645163407492146\n",
+      "train loss:0.03320815299071024\n",
+      "train loss:0.0007329960169586183\n",
+      "train loss:0.00013231148514934838\n",
+      "train loss:8.523595249005568e-05\n",
+      "train loss:0.00021018724486788606\n",
+      "train loss:0.00012478491967100828\n",
+      "train loss:0.0005017121680978088\n",
+      "train loss:0.011441801949601515\n",
+      "train loss:0.005008138912130445\n",
+      "train loss:0.0009392064690660033\n",
+      "train loss:0.005522185548574482\n",
+      "train loss:0.00035100422880633995\n",
+      "train loss:0.0005699932163528782\n",
+      "train loss:0.0011206856779456163\n",
+      "train loss:5.8723750890966444e-05\n",
+      "train loss:0.0062115622494227155\n",
+      "train loss:0.0011379659367674773\n",
+      "train loss:9.814787996068519e-05\n",
+      "train loss:0.000130587855576745\n",
+      "train loss:0.0032267834785380824\n",
+      "train loss:0.0002192184244833853\n",
+      "train loss:5.5471510090964474e-05\n",
+      "train loss:0.00020448857447463\n",
+      "train loss:0.000189885057365809\n",
+      "train loss:0.0003704838080724721\n",
+      "train loss:0.0006223572606423542\n",
+      "train loss:0.00012831959370162668\n",
+      "train loss:0.00038160155159826166\n",
+      "train loss:0.001867945123577388\n",
+      "train loss:0.0021615490075254547\n",
+      "train loss:0.0001237324388116061\n",
+      "train loss:0.0006246921850746903\n",
+      "train loss:0.000586986523191294\n",
+      "train loss:0.0007737760937049842\n",
+      "train loss:4.282840176555636e-05\n",
+      "train loss:0.00023271520613639445\n",
+      "train loss:0.0011224884830807686\n",
+      "train loss:0.0011285311620885868\n",
+      "train loss:0.00023849419744926046\n",
+      "train loss:0.00016628779170828098\n",
+      "train loss:0.0005634738667915095\n",
+      "train loss:0.0009161788999471213\n",
+      "train loss:0.00018933893659517896\n",
+      "train loss:0.005514325025695049\n",
+      "train loss:9.742388842647574e-05\n",
+      "train loss:0.0009538385104339952\n",
+      "train loss:0.00035924525941105303\n",
+      "train loss:0.0014549212219407844\n",
+      "train loss:0.0012689100345773458\n",
+      "train loss:8.746804836110639e-05\n",
+      "train loss:0.00019122700881084168\n",
+      "train loss:7.329672708828998e-05\n",
+      "train loss:0.002415757871069771\n",
+      "train loss:0.0007278017791386408\n",
+      "train loss:0.001156680081443522\n",
+      "train loss:0.0028757877400457875\n",
+      "train loss:6.594369918200518e-05\n",
+      "train loss:0.0011657253004596392\n",
+      "train loss:0.00014895012446948015\n",
+      "train loss:6.428937022642236e-05\n",
+      "train loss:0.0011542588158305837\n",
+      "train loss:0.0015791571473676568\n",
+      "train loss:0.0001386459203918016\n",
+      "train loss:0.006425342547014735\n",
+      "train loss:8.934492625214053e-05\n",
+      "train loss:0.00015913082775983935\n",
+      "train loss:0.00034708001990634533\n",
+      "train loss:0.001109693445594383\n",
+      "train loss:0.000626800875762785\n",
+      "train loss:0.002237792035993713\n",
+      "train loss:0.0013153034331304329\n",
+      "train loss:0.0001755514546323963\n",
+      "train loss:0.0015133919276047575\n",
+      "train loss:0.0004947925410573802\n",
+      "train loss:0.0033944569744796776\n",
+      "train loss:0.005981536318583529\n",
+      "train loss:1.3838619134940197e-05\n",
+      "train loss:0.00029628767882246726\n",
+      "train loss:0.001074951667069093\n",
+      "train loss:7.622959681189665e-05\n",
+      "train loss:0.00010471018992325475\n",
+      "train loss:0.002990537507859546\n",
+      "train loss:0.0010192045838691047\n",
+      "train loss:0.001118448686870212\n",
+      "train loss:0.001605177557888191\n",
+      "train loss:0.00018834889415526572\n",
+      "train loss:0.0023826029396472955\n",
+      "train loss:0.0011515474117715471\n",
+      "train loss:0.0005494424543653156\n",
+      "train loss:0.0011406445352204503\n",
+      "train loss:0.00022719561542847798\n",
+      "train loss:0.001918255291291415\n",
+      "train loss:0.00037986704151997993\n",
+      "train loss:0.0001260070010940137\n",
+      "train loss:0.001934416511196194\n",
+      "train loss:0.013666301895285867\n",
+      "train loss:0.0014237240514040537\n",
+      "train loss:8.650932613670146e-05\n",
+      "train loss:0.00018333988240184316\n",
+      "train loss:0.0016084894343109462\n",
+      "train loss:0.010507380167872553\n",
+      "train loss:0.00027171275024165056\n",
+      "train loss:0.0004235431855046625\n",
+      "train loss:0.0008248793947197382\n",
+      "train loss:0.0009978094605020023\n",
+      "train loss:0.003470162421089079\n",
+      "train loss:0.00033223467295880445\n",
+      "train loss:5.902889460174268e-05\n",
+      "train loss:0.00017828316546947718\n",
+      "train loss:0.00042649773528573894\n",
+      "train loss:0.0006428779074085575\n",
+      "train loss:0.00010192116631651286\n",
+      "train loss:0.01570528480504136\n",
+      "train loss:0.0006643369417929383\n",
+      "train loss:0.0006881083049066118\n",
+      "train loss:0.00020101217887106952\n",
+      "train loss:0.00040427887220698594\n",
+      "train loss:0.00016870127999343336\n",
+      "train loss:0.0001170223805050098\n",
+      "train loss:5.40364776229251e-05\n",
+      "train loss:9.566062418562013e-05\n",
+      "train loss:0.0006399778116263151\n",
+      "train loss:1.7265005976848826e-05\n",
+      "train loss:0.0010508218748398318\n",
+      "train loss:0.00335709435895161\n",
+      "train loss:0.0007029123245917468\n",
+      "train loss:0.0017683701271607375\n",
+      "train loss:0.0004833753840440376\n",
+      "train loss:0.000538878813969704\n",
+      "train loss:8.816664367615469e-05\n",
+      "train loss:0.0011446740918959488\n",
+      "train loss:0.0003876886110401492\n",
+      "train loss:0.00016198636372687526\n",
+      "train loss:0.0013754244711397293\n",
+      "train loss:8.386619100245291e-05\n",
+      "train loss:5.45988646408109e-05\n",
+      "train loss:0.0003068779327934376\n",
+      "train loss:0.0025399836962673255\n",
+      "train loss:0.00012375890812875887\n",
+      "train loss:0.001011480451211636\n",
+      "train loss:0.002185332658291104\n",
+      "train loss:0.0005710736631396594\n",
+      "train loss:0.0006964381467156402\n",
+      "train loss:8.628368463067649e-05\n",
+      "train loss:0.004495527049952172\n",
+      "train loss:0.013089196746745963\n",
+      "train loss:0.0005102174787462591\n",
+      "train loss:0.0036629680550218625\n",
+      "train loss:0.00018905311553597739\n",
+      "train loss:0.0008466780891603237\n",
+      "train loss:0.00023206871022192266\n",
+      "train loss:0.00022453889171138684\n",
+      "train loss:0.0007484123594856247\n",
+      "train loss:0.00012627246448164135\n",
+      "train loss:0.0014397780567760982\n",
+      "train loss:0.0014710353081448249\n",
+      "train loss:0.0019050016134575742\n",
+      "train loss:0.00033739980958520833\n",
+      "train loss:0.0048669945150845596\n",
+      "train loss:0.00032157915705620474\n",
+      "train loss:0.0013140608942161813\n",
+      "train loss:0.0005311059299363021\n",
+      "train loss:5.411515254639891e-05\n",
+      "train loss:0.0015661317263335572\n",
+      "=============== Final Test Accuracy ===============\n",
+      "test acc:0.9879\n",
+      "Saved Network Parameters!\n"
+     ]
+    },
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
    "source": [
-    "### Classice CNNs"
+    "import pickle\n",
+    "import numpy as np\n",
+    "from collections import OrderedDict\n",
+    "import matplotlib.pyplot as plt\n",
+    "from dataset.mnist import load_mnist\n",
+    "\n",
+    "class SGD:\n",
+    "\n",
+    "\n",
+    "    def __init__(self, lr=0.01):\n",
+    "        self.lr = lr\n",
+    "        \n",
+    "    def update(self, params, grads):\n",
+    "        for key in params.keys():\n",
+    "            params[key] -= self.lr * grads[key] \n",
+    "\n",
+    "class Momentum:\n",
+    "\n",
+    "\n",
+    "    def __init__(self, lr=0.01, momentum=0.9):\n",
+    "        self.lr = lr\n",
+    "        self.momentum = momentum\n",
+    "        self.v = None\n",
+    "        \n",
+    "    def update(self, params, grads):\n",
+    "        if self.v is None:\n",
+    "            self.v = {}\n",
+    "            for key, val in params.items():                                \n",
+    "                self.v[key] = np.zeros_like(val)\n",
+    "                \n",
+    "        for key in params.keys():\n",
+    "            self.v[key] = self.momentum*self.v[key] - self.lr*grads[key] \n",
+    "            params[key] += self.v[key]\n",
+    "\n",
+    "class Nesterov:\n",
+    "\n",
+    "    \"\"\"Nesterov's Accelerated Gradient (http://arxiv.org/abs/1212.0901)\"\"\"\n",
+    "\n",
+    "    def __init__(self, lr=0.01, momentum=0.9):\n",
+    "        self.lr = lr\n",
+    "        self.momentum = momentum\n",
+    "        self.v = None\n",
+    "        \n",
+    "    def update(self, params, grads):\n",
+    "        if self.v is None:\n",
+    "            self.v = {}\n",
+    "            for key, val in params.items():\n",
+    "                self.v[key] = np.zeros_like(val)\n",
+    "            \n",
+    "        for key in params.keys():\n",
+    "            params[key] += self.momentum * self.momentum * self.v[key]\n",
+    "            params[key] -= (1 + self.momentum) * self.lr * grads[key]\n",
+    "            self.v[key] *= self.momentum\n",
+    "            self.v[key] -= self.lr * grads[key]\n",
+    "            \n",
+    "class AdaGrad:\n",
+    "\n",
+    "    def __init__(self, lr=0.01):\n",
+    "        self.lr = lr\n",
+    "        self.h = None\n",
+    "        \n",
+    "    def update(self, params, grads):\n",
+    "        if self.h is None:\n",
+    "            self.h = {}\n",
+    "            for key, val in params.items():\n",
+    "                self.h[key] = np.zeros_like(val)\n",
+    "            \n",
+    "        for key in params.keys():\n",
+    "            self.h[key] += grads[key] * grads[key]\n",
+    "            params[key] -= self.lr * grads[key] / (np.sqrt(self.h[key]) + 1e-7)\n",
+    "\n",
+    "class RMSprop:\n",
+    "\n",
+    "    \"\"\"RMSprop\"\"\"\n",
+    "\n",
+    "    def __init__(self, lr=0.01, decay_rate = 0.99):\n",
+    "        self.lr = lr\n",
+    "        self.decay_rate = decay_rate\n",
+    "        self.h = None\n",
+    "        \n",
+    "    def update(self, params, grads):\n",
+    "        if self.h is None:\n",
+    "            self.h = {}\n",
+    "            for key, val in params.items():\n",
+    "                self.h[key] = np.zeros_like(val)\n",
+    "            \n",
+    "        for key in params.keys():\n",
+    "            self.h[key] *= self.decay_rate\n",
+    "            self.h[key] += (1 - self.decay_rate) * grads[key] * grads[key]\n",
+    "            params[key] -= self.lr * grads[key] / (np.sqrt(self.h[key]) + 1e-7)\n",
+    "\n",
+    "\n",
+    "class Adam:\n",
+    "\n",
+    "    \"\"\"Adam (http://arxiv.org/abs/1412.6980v8)\"\"\"\n",
+    "\n",
+    "    def __init__(self, lr=0.001, beta1=0.9, beta2=0.999):\n",
+    "        self.lr = lr\n",
+    "        self.beta1 = beta1\n",
+    "        self.beta2 = beta2\n",
+    "        self.iter = 0\n",
+    "        self.m = None\n",
+    "        self.v = None\n",
+    "        \n",
+    "    def update(self, params, grads):\n",
+    "        if self.m is None:\n",
+    "            self.m, self.v = {}, {}\n",
+    "            for key, val in params.items():\n",
+    "                self.m[key] = np.zeros_like(val)\n",
+    "                self.v[key] = np.zeros_like(val)\n",
+    "        \n",
+    "        self.iter += 1\n",
+    "        lr_t  = self.lr * np.sqrt(1.0 - self.beta2**self.iter) / (1.0 - self.beta1**self.iter)         \n",
+    "        \n",
+    "        for key in params.keys():\n",
+    "            self.m[key] += (1 - self.beta1) * (grads[key] - self.m[key])\n",
+    "            self.v[key] += (1 - self.beta2) * (grads[key]**2 - self.v[key])\n",
+    "            params[key] -= lr_t * self.m[key] / (np.sqrt(self.v[key]) + 1e-7)\n",
+    "            \n",
+    "def softmax(x):\n",
+    "    x = x - np.max(x, axis=-1, keepdims=True)\n",
+    "    return np.exp(x) / np.sum(np.exp(x), axis=-1, keepdims=True)\n",
+    "\n",
+    "def cross_entropy_error(y, t):\n",
+    "    if y.ndim == 1:\n",
+    "        t = t.reshape(1, t.size)\n",
+    "        y = y.reshape(1, y.size)\n",
+    "        \n",
+    "    if t.size == y.size:\n",
+    "        t = t.argmax(axis=1)\n",
+    "             \n",
+    "    batch_size = y.shape[0]\n",
+    "    return -np.sum(np.log(y[np.arange(batch_size), t] + 1e-7)) / batch_size\n",
+    "\n",
+    "\n",
+    "class Convolution:\n",
+    "    def __init__(self, W, b, stride=1, pad=0):\n",
+    "        self.W = W\n",
+    "        self.b = b\n",
+    "        self.stride = stride\n",
+    "        self.pad = pad\n",
+    "        \n",
+    "        self.x = None   \n",
+    "        self.col = None\n",
+    "        self.col_W = None\n",
+    "        \n",
+    "        self.dW = None\n",
+    "        self.db = None\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        FN, C, FH, FW = self.W.shape\n",
+    "        N, C, H, W = x.shape\n",
+    "        out_h = 1 + int((H + 2*self.pad - FH) / self.stride)\n",
+    "        out_w = 1 + int((W + 2*self.pad - FW) / self.stride)\n",
+    "\n",
+    "        col = im2col(x, FH, FW, self.stride, self.pad)\n",
+    "        col_W = self.W.reshape(FN, -1).T\n",
+    "\n",
+    "        out = np.dot(col, col_W) + self.b\n",
+    "        out = out.reshape(N, out_h, out_w, -1).transpose(0, 3, 1, 2)\n",
+    "\n",
+    "        self.x = x\n",
+    "        self.col = col\n",
+    "        self.col_W = col_W\n",
+    "\n",
+    "        return out\n",
+    "\n",
+    "    def backward(self, dout):\n",
+    "        FN, C, FH, FW = self.W.shape\n",
+    "        dout = dout.transpose(0,2,3,1).reshape(-1, FN)\n",
+    "\n",
+    "        self.db = np.sum(dout, axis=0)\n",
+    "        self.dW = np.dot(self.col.T, dout)\n",
+    "        self.dW = self.dW.transpose(1, 0).reshape(FN, C, FH, FW)\n",
+    "\n",
+    "        dcol = np.dot(dout, self.col_W.T)\n",
+    "        dx = col2im(dcol, self.x.shape, FH, FW, self.stride, self.pad)\n",
+    "\n",
+    "        return dx\n",
+    "\n",
+    "\n",
+    "class Pooling:\n",
+    "    def __init__(self, pool_h, pool_w, stride=2, pad=0):\n",
+    "        self.pool_h = pool_h\n",
+    "        self.pool_w = pool_w\n",
+    "        self.stride = stride\n",
+    "        self.pad = pad\n",
+    "        \n",
+    "        self.x = None\n",
+    "        self.arg_max = None\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        N, C, H, W = x.shape\n",
+    "        out_h = int(1 + (H - self.pool_h) / self.stride)\n",
+    "        out_w = int(1 + (W - self.pool_w) / self.stride)\n",
+    "\n",
+    "        col = im2col(x, self.pool_h, self.pool_w, self.stride, self.pad)\n",
+    "        col = col.reshape(-1, self.pool_h*self.pool_w)\n",
+    "\n",
+    "        arg_max = np.argmax(col, axis=1)\n",
+    "        out = np.max(col, axis=1)\n",
+    "        out = out.reshape(N, out_h, out_w, C).transpose(0, 3, 1, 2)\n",
+    "\n",
+    "        self.x = x\n",
+    "        self.arg_max = arg_max\n",
+    "\n",
+    "        return out\n",
+    "\n",
+    "    def backward(self, dout):\n",
+    "        dout = dout.transpose(0, 2, 3, 1)\n",
+    "        \n",
+    "        pool_size = self.pool_h * self.pool_w\n",
+    "        dmax = np.zeros((dout.size, pool_size))\n",
+    "        dmax[np.arange(self.arg_max.size), self.arg_max.flatten()] = dout.flatten()\n",
+    "        dmax = dmax.reshape(dout.shape + (pool_size,)) \n",
+    "        \n",
+    "        dcol = dmax.reshape(dmax.shape[0] * dmax.shape[1] * dmax.shape[2], -1)\n",
+    "        dx = col2im(dcol, self.x.shape, self.pool_h, self.pool_w, self.stride, self.pad)\n",
+    "        \n",
+    "        return dx\n",
+    "\n",
+    "\n",
+    "class Affine:\n",
+    "    def __init__(self, W, b):\n",
+    "        self.W =W\n",
+    "        self.b = b\n",
+    "        \n",
+    "        self.x = None\n",
+    "        self.original_x_shape = None\n",
+    "        self.dW = None\n",
+    "        self.db = None\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        self.original_x_shape = x.shape\n",
+    "        x = x.reshape(x.shape[0], -1)\n",
+    "        self.x = x\n",
+    "\n",
+    "        out = np.dot(self.x, self.W) + self.b\n",
+    "\n",
+    "        return out\n",
+    "\n",
+    "    def backward(self, dout):\n",
+    "        dx = np.dot(dout, self.W.T)\n",
+    "        self.dW = np.dot(self.x.T, dout)\n",
+    "        self.db = np.sum(dout, axis=0)\n",
+    "        \n",
+    "        dx = dx.reshape(*self.original_x_shape)\n",
+    "        return dx\n",
+    "\n",
+    "class Relu:\n",
+    "    def __init__(self):\n",
+    "        self.mask = None\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        self.mask = (x <= 0)\n",
+    "        out = x.copy()\n",
+    "        out[self.mask] = 0\n",
+    "\n",
+    "        return out\n",
+    "\n",
+    "    def backward(self, dout):\n",
+    "        dout[self.mask] = 0\n",
+    "        dx = dout\n",
+    "\n",
+    "        return dx\n",
+    "    \n",
+    "class SoftmaxWithLoss:\n",
+    "    def __init__(self):\n",
+    "        self.loss = None\n",
+    "        self.y = None \n",
+    "        self.t = None \n",
+    "\n",
+    "    def forward(self, x, t):\n",
+    "        self.t = t\n",
+    "        self.y = softmax(x)\n",
+    "        self.loss = cross_entropy_error(self.y, self.t)\n",
+    "        \n",
+    "        return self.loss\n",
+    "\n",
+    "    def backward(self, dout=1):\n",
+    "        batch_size = self.t.shape[0]\n",
+    "        if self.t.size == self.y.size: \n",
+    "            dx = (self.y - self.t) / batch_size\n",
+    "        else:\n",
+    "            dx = self.y.copy()\n",
+    "            dx[np.arange(batch_size), self.t] -= 1\n",
+    "            dx = dx / batch_size\n",
+    "        \n",
+    "        return dx\n",
+    "\n",
+    "class SimpleConvNet:\n",
+    "   \n",
+    "    def __init__(self, input_dim=(1, 28, 28), \n",
+    "                 conv_param={'filter_num':30, 'filter_size':5, 'pad':0, 'stride':1},\n",
+    "                 hidden_size=100, output_size=10, weight_init_std=0.01):\n",
+    "        filter_num = conv_param['filter_num']\n",
+    "        filter_size = conv_param['filter_size']\n",
+    "        filter_pad = conv_param['pad']\n",
+    "        filter_stride = conv_param['stride']\n",
+    "        input_size = input_dim[1]\n",
+    "        conv_output_size = (input_size - filter_size + 2*filter_pad) / filter_stride + 1\n",
+    "        pool_output_size = int(filter_num * (conv_output_size/2) * (conv_output_size/2))\n",
+    "\n",
+    "        self.params = {}\n",
+    "        self.params['W1'] = weight_init_std * \\\n",
+    "                            np.random.randn(filter_num, input_dim[0], filter_size, filter_size)\n",
+    "        self.params['b1'] = np.zeros(filter_num)\n",
+    "        self.params['W2'] = weight_init_std * \\\n",
+    "                            np.random.randn(pool_output_size, hidden_size)\n",
+    "        self.params['b2'] = np.zeros(hidden_size)\n",
+    "        self.params['W3'] = weight_init_std * \\\n",
+    "                            np.random.randn(hidden_size, output_size)\n",
+    "        self.params['b3'] = np.zeros(output_size)\n",
+    "\n",
+    "        self.layers = OrderedDict()\n",
+    "        self.layers['Conv1'] = Convolution(self.params['W1'], self.params['b1'],\n",
+    "                                           conv_param['stride'], conv_param['pad'])\n",
+    "        self.layers['Relu1'] = Relu()\n",
+    "        self.layers['Pool1'] = Pooling(pool_h=2, pool_w=2, stride=2)\n",
+    "        self.layers['Affine1'] = Affine(self.params['W2'], self.params['b2'])\n",
+    "        self.layers['Relu2'] = Relu()\n",
+    "        self.layers['Affine2'] = Affine(self.params['W3'], self.params['b3'])\n",
+    "\n",
+    "        self.last_layer = SoftmaxWithLoss()\n",
+    "\n",
+    "    def predict(self, x):\n",
+    "        for layer in self.layers.values():\n",
+    "            x = layer.forward(x)\n",
+    "\n",
+    "        return x\n",
+    "\n",
+    "    def loss(self, x, t):\n",
+    "     \n",
+    "        y = self.predict(x)\n",
+    "        return self.last_layer.forward(y, t)\n",
+    "\n",
+    "    def accuracy(self, x, t, batch_size=100):\n",
+    "        if t.ndim != 1 : t = np.argmax(t, axis=1)\n",
+    "        \n",
+    "        acc = 0.0\n",
+    "        \n",
+    "        for i in range(int(x.shape[0] / batch_size)):\n",
+    "            tx = x[i*batch_size:(i+1)*batch_size]\n",
+    "            tt = t[i*batch_size:(i+1)*batch_size]\n",
+    "            y = self.predict(tx)\n",
+    "            y = np.argmax(y, axis=1)\n",
+    "            acc += np.sum(y == tt) \n",
+    "        \n",
+    "        return acc / x.shape[0]\n",
+    "\n",
+    "    def gradient(self, x, t):\n",
+    "     \n",
+    "        # forward\n",
+    "        self.loss(x, t)\n",
+    "\n",
+    "        # backward\n",
+    "        dout = 1\n",
+    "        dout = self.last_layer.backward(dout)\n",
+    "\n",
+    "        layers = list(self.layers.values())\n",
+    "        layers.reverse()\n",
+    "        for layer in layers:\n",
+    "            dout = layer.backward(dout)\n",
+    "\n",
+    "        grads = {}\n",
+    "        grads['W1'], grads['b1'] = self.layers['Conv1'].dW, self.layers['Conv1'].db\n",
+    "        grads['W2'], grads['b2'] = self.layers['Affine1'].dW, self.layers['Affine1'].db\n",
+    "        grads['W3'], grads['b3'] = self.layers['Affine2'].dW, self.layers['Affine2'].db\n",
+    "\n",
+    "        return grads\n",
+    "        \n",
+    "    def save_params(self, file_name=\"params.pkl\"):\n",
+    "        params = {}\n",
+    "        for key, val in self.params.items():\n",
+    "            params[key] = val\n",
+    "        with open(file_name, 'wb') as f:\n",
+    "            pickle.dump(params, f)\n",
+    "\n",
+    "    def load_params(self, file_name=\"params.pkl\"):\n",
+    "        with open(file_name, 'rb') as f:\n",
+    "            params = pickle.load(f)\n",
+    "        for key, val in params.items():\n",
+    "            self.params[key] = val\n",
+    "\n",
+    "        for i, key in enumerate(['Conv1', 'Affine1', 'Affine2']):\n",
+    "            self.layers[key].W = self.params['W' + str(i+1)]\n",
+    "            self.layers[key].b = self.params['b' + str(i+1)]\n",
+    "\n",
+    "\n",
+    "class Trainer:\n",
+    "  \n",
+    "    def __init__(self, network, x_train, t_train, x_test, t_test,\n",
+    "                 epochs=20, mini_batch_size=100,\n",
+    "                 optimizer='SGD', optimizer_param={'lr':0.01}, \n",
+    "                 evaluate_sample_num_per_epoch=None, verbose=True):\n",
+    "        self.network = network\n",
+    "        self.verbose = verbose\n",
+    "        self.x_train = x_train\n",
+    "        self.t_train = t_train\n",
+    "        self.x_test = x_test\n",
+    "        self.t_test = t_test\n",
+    "        self.epochs = epochs\n",
+    "        self.batch_size = mini_batch_size\n",
+    "        self.evaluate_sample_num_per_epoch = evaluate_sample_num_per_epoch\n",
+    "\n",
+    "        # optimizer\n",
+    "        optimizer_class_dict = {'sgd':SGD, 'momentum':Momentum, 'nesterov':Nesterov,\n",
+    "                                'adagrad':AdaGrad, 'rmsprop':RMSprop, 'adam':Adam}\n",
+    "        self.optimizer = optimizer_class_dict[optimizer.lower()](**optimizer_param)\n",
+    "        \n",
+    "        self.train_size = x_train.shape[0]\n",
+    "        self.iter_per_epoch = max(self.train_size / mini_batch_size, 1)\n",
+    "        self.max_iter = int(epochs * self.iter_per_epoch)\n",
+    "        self.current_iter = 0\n",
+    "        self.current_epoch = 0\n",
+    "        \n",
+    "        self.train_loss_list = []\n",
+    "        self.train_acc_list = []\n",
+    "        self.test_acc_list = []\n",
+    "\n",
+    "    def train_step(self):\n",
+    "        batch_mask = np.random.choice(self.train_size, self.batch_size)\n",
+    "        x_batch = self.x_train[batch_mask]\n",
+    "        t_batch = self.t_train[batch_mask]\n",
+    "        \n",
+    "        grads = self.network.gradient(x_batch, t_batch)\n",
+    "        self.optimizer.update(self.network.params, grads)\n",
+    "        \n",
+    "        loss = self.network.loss(x_batch, t_batch)\n",
+    "        self.train_loss_list.append(loss)\n",
+    "        if self.verbose: print(\"train loss:\" + str(loss))\n",
+    "        \n",
+    "        if self.current_iter % self.iter_per_epoch == 0:\n",
+    "            self.current_epoch += 1\n",
+    "            \n",
+    "            x_train_sample, t_train_sample = self.x_train, self.t_train\n",
+    "            x_test_sample, t_test_sample = self.x_test, self.t_test\n",
+    "            if not self.evaluate_sample_num_per_epoch is None:\n",
+    "                t = self.evaluate_sample_num_per_epoch\n",
+    "                x_train_sample, t_train_sample = self.x_train[:t], self.t_train[:t]\n",
+    "                x_test_sample, t_test_sample = self.x_test[:t], self.t_test[:t]\n",
+    "                \n",
+    "            train_acc = self.network.accuracy(x_train_sample, t_train_sample)\n",
+    "            test_acc = self.network.accuracy(x_test_sample, t_test_sample)\n",
+    "            self.train_acc_list.append(train_acc)\n",
+    "            self.test_acc_list.append(test_acc)\n",
+    "\n",
+    "            if self.verbose: print(\"=== epoch:\" + str(self.current_epoch) + \", train acc:\" + str(train_acc) + \", test acc:\" + str(test_acc) + \" ===\")\n",
+    "        self.current_iter += 1\n",
+    "\n",
+    "    def train(self):\n",
+    "        for i in range(self.max_iter):\n",
+    "            self.train_step()\n",
+    "\n",
+    "        test_acc = self.network.accuracy(self.x_test, self.t_test)\n",
+    "\n",
+    "        if self.verbose:\n",
+    "            print(\"=============== Final Test Accuracy ===============\")\n",
+    "            print(\"test acc:\" + str(test_acc))\n",
+    "\n",
+    "\n",
+    "(x_train, t_train), (x_test, t_test) = load_mnist(flatten=False)\n",
+    "\n",
+    "\n",
+    "max_epochs = 20\n",
+    "\n",
+    "network = SimpleConvNet(input_dim=(1,28,28), \n",
+    "                        conv_param = {'filter_num': 30, 'filter_size': 5, 'pad': 0, 'stride': 1},\n",
+    "                        hidden_size=100, output_size=10, weight_init_std=0.01)\n",
+    "                        \n",
+    "trainer = Trainer(network, x_train, t_train, x_test, t_test,\n",
+    "                  epochs=max_epochs, mini_batch_size=100,\n",
+    "                  optimizer='Adam', optimizer_param={'lr': 0.001},\n",
+    "                  evaluate_sample_num_per_epoch=1000)\n",
+    "trainer.train()\n",
+    "\n",
+    "network.save_params(\"params.pkl\")\n",
+    "print(\"Saved Network Parameters!\")\n",
+    "\n",
+    "markers = {'train': 'o', 'test': 's'}\n",
+    "x = np.arange(max_epochs)\n",
+    "plt.plot(x, trainer.train_acc_list, marker='o', label='train', markevery=2)\n",
+    "plt.plot(x, trainer.test_acc_list, marker='s', label='test', markevery=2)\n",
+    "plt.xlabel(\"epochs\")\n",
+    "plt.ylabel(\"accuracy\")\n",
+    "plt.ylim(0, 1.0)\n",
+    "plt.legend(loc='lower right')\n",
+    "plt.show()\n"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Deep learning"
+    "### Visualization of CNN"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 14,
    "metadata": {},
-   "outputs": [],
-   "source": []
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "before training\n"
+     ]
+    },
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 30 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "after training\n"
+     ]
+    },
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 30 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import numpy as np\n",
+    "import matplotlib.pyplot as plt\n",
+    "\n",
+    "def filter_show(filters, nx=8, margin=3, scale=10):\n",
+    "    \"\"\"\n",
+    "    c.f. https://gist.github.com/aidiary/07d530d5e08011832b12#file-draw_weight-py\n",
+    "    \"\"\"\n",
+    "    FN, C, FH, FW = filters.shape\n",
+    "    ny = int(np.ceil(FN / nx))\n",
+    "\n",
+    "    fig = plt.figure()\n",
+    "    fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)\n",
+    "\n",
+    "    for i in range(FN):\n",
+    "        ax = fig.add_subplot(ny, nx, i+1, xticks=[], yticks=[])\n",
+    "        ax.imshow(filters[i, 0], cmap=plt.cm.gray_r, interpolation='nearest')\n",
+    "    plt.show()\n",
+    "\n",
+    "\n",
+    "network = SimpleConvNet()\n",
+    "print(f'before training')\n",
+    "filter_show(network.params['W1'])\n",
+    "print(f'after training')\n",
+    "network.load_params(\"./params.pkl\")\n",
+    "filter_show(network.params['W1'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 16 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAegAAAGFCAYAAAA2OmCIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOz9d5Dd53Uejj+391629wYsOgiCBZRYREqUREmJpEhusRPLmbGtxPakOONkxk7xOJ7YM5lxEjtxLCluk2jUJVIiJVEUWMACEAQBAthdbO+7d/f2Xn9/7O85e+4FZAGOZQv73TODAbB7y+fzed/3lOc85xxDo9FoYF/2ZV/2ZV/2ZV9+rMT4d30B+7Iv+7Iv+7Iv+3Kz7BvofdmXfdmXfdmXH0PZN9D7si/7si/7si8/hrJvoPdlX/ZlX/ZlX34MZd9A78u+7Mu+7Mu+/BjKvoHel33Zl33Zl335MZR9A70v+7Iv+7Iv+/JjKObbeVG9Xsfq6io8Hg8MBsOP+pr2nDQaDWQyGXR2dsJovDt9ov098P8u+/tgX/b3wL4At78PbstAr66uoqen52/s4v6/KktLS+ju7v67voy/luzvgb852d8H+7K/B/YF+OH74LYMtMfjAQD8p//0n2C322EwGNBoNGAymWC1WmE0GlGpVFAsFhGLxVCtVmGxWGC1WlGtVuX3VqsVVqsVjUYDRqNRPqPRaMBgMKBSqcBsNsNsNqNWq8FoNMp3FYtFWCwWmEwm1Ot1mEwmlEolmM3Nt1Cr1WAwGFCr1QAAFosFAFCtVtFoNNBoNOQ1LpcLJpNJroHXA+x4OPozjUYjHA4HnE4nGo0GSqUSVlZW5HVmsxmlUglOp1O+32g0wmw2I5VK4Xd+53fkOd6Nwmv/hV/4BVgsFhgMBlSrVVkXm80Gv98Pv9+PWq2Gcrks62o0GuVn9XodwM7zrdfr8vy4H7hueu35O/3aarUKq9WKWq0mz1q/p1UMBgNMJhMMBgMsFgtKpRLy+TyKxaK8p16vw2AwyB9+F9+fzWaRyWRQLpcxMjICs9kMq9Uq9wPs7j8Acq9GoxEdHR3o7+/HRz7ykT2xD37u535OdAHPdq1Wk2ditVqRSqVQqVQA7JzDer0ua8ezVq1WYbPZ5DPq9bo8N76POoTPWK8TADlvBoNBdIcWfl69Xhd9wX0JAJubm3JfDodDPovfaTAYZK/W63W43W5cuXIFPT09cDgceNe73oVvf/vb8nlms1nulVKtVuF2u1GpVPCHf/iHe2IP/P7v/748L4fDAQByzmOxGJLJJNra2mCxWJrOqclkkvNKXc5nrIW6nmtQq9VEP1P/8HXce8Du/tC6hD/nd1arVdTrddRqNXg8Hvh8PtTrdVQqFdRqtZt0C6+NusLn86G9vR3pdBovvfQSTCYTurq65JqAnX1QrVblmnkvV65cgclkwp/+6Z/+0H1wWwaaB8Fms8Fms8FkMqFcLosxAgCr1QqLxYJsNotarSYHkj+n4TWbzXLQTCaTHCj+m0qW7+Xi8PPq9TocDgcqlYpsCm1geT2VSkUWzmazoVwui0IolUowGAxwOBzyvbxPLio/hwrEYDDAarWiUqkgGAyiWCyiq6sLW1tb8rl2u12eBa/VaDTC5/M1febdKLx2k8kEi8UiitblcsFoNIpSdrlc8qxpXLkfuB4AROnyAPC1WqHTYdNKEtjZ6FyTWq3WdMgByGfxMPHw0kjTiGiDwc/VDpr+TAAIBoNyL/F4HENDQ7J3qRS477j+JpMJ/f39OHz4MK5cudL0LO9G4bXb7XY5f3y2drtdHCe73Y5KpYJ8Pt/kRPP5UkHbbDbUajVxfKm8abD5eVxTitFoRLlchs1mk2fNtTCZTKIfKpWK7B0aXf6f3+VyuQBA/qZypmKmwdUO/tjYGJaXl9HR0YFyuYyBgQGsrKyIs8G/qaTD4TAMBgOWl5ebnuPdKLx2l8slDiqfNc+qy+WSAMpiscDhcIjTTl2gn2/rGQd2DTSw6wBrQ819YjabZS8AEB2i9U2j0ZB9WKvVxEmsVqs3ORg8w1xr6iltX8rlMqxWK4LBIE6dOoUXXngBvb298jz4nVarVc5HsVjEhQsXEAqFmhyNv0ruKAnCzU/jSeVGb4aHpFKpoFqtyoWVSqWdL/v/Hw4dCfOg0Hjrm+fD1Qtot9vFSaBipWfLh1kul+FwOEQhA5CoT0d/vE5+BhUy/9AYMQrXRtdiscDlcqFWq6FYLIqB4sEkgmA2mzE1NXUnj/nHWrihGGVoB4eHqdVj1A4U/+joREe8RB20t6z3AQ+82Wy+SdnzwGmHi7/XEZM+FK3X1Bo5t9673+9HvV5HOp1GIpGQe+D79F6xWCwYHBzEkSNHMDMzc8vI/m4WfX7pULUaskqlgnK53OQ4UzHzGfO92sFi1AVAXq8dKG1AaXy1TtARELCrrHUUzn3mdDrl2vg7Hd3piJx7xOv1olgswmQy4dy5c3jXu96FRqOBQqEAi8UiBqFSqcDlcsFsNmNjY+OujpxvJRq1oOHjGnFNdbTL/3PNgN0z2Cp87nTCdXDHvQU0O/s8x63rxu/RyKxGYHTkq1GeVj2l0bGVlRW4XC74/X4MDAxgYWFB7p1C3VcqlfDmm2/C7XbD6XQ2GfK/Su7IQHPj0lDzpnkzxWJRvCgA4jHxAGvooXWBSqWSPBgdkRgMBpTLZZRKJTQaDZTLZeTzeQC7njuVMj0oq9UqrwcghpjXTGOqoSzCHrweHnxG3PV6Xb43l8vB6XSiXq/D4/EIrMPFtVgs2N7eRjabxeTk5E2LdjcLD2K5XBZjqtMGfI2GhPj3rf4AzQZWR978N71l/Xl0yIi68Pv0Iebn8PDqQ6idNRp+nW7hd3Av8POsVivcbjdsNhsWFhZQLpeRTqebIDsamHvuuQcdHR2YmJhocgb3iuh11OkMrhXRq1bUjGeEeoMOrVbm/Dwq2lYnma/XhprGX68hjTbfw71KNIyfQf0EQBwKbUhaHdNcLodTp05heXkZZrMZX//61/HUU0+hWq1KKsdoNMLr9SKdTmN1dRUul0sitb0gXG8aaOrYVqQDwE1Ouk5Z6L+1w9UqGm3TKUntdPG6uCe14ddOYasjrtNr+vc6cm914DOZDNbX1+FyuXD8+HFYrVYsLS3J+7Sj8sorr6C3txd+v/+WzsMPkjsy0K3eBKMcfTGED/m3zi3qCIvQN6NWDWFZrdamw08IlQtIY0jnQN+wziHoh22xWFAul1EoFOQ7CW8Q8m71quhB22w2uUfm2vP5PDweD9ra2iTJT4NuNpvh9XqxtLQk6YC9JDwoVJo6N6w3st78rR6yVqA0xnzeOuekD4Q+LFxTGmgNk+tr0ga2NYrj9+sDzf/zmvSeIFoQDofFUVlbW5PcIr/DZDKhr68Pbrcbq6ur4v3frazdHyRcp1KpJGulFSN5GXo/MPKh7jAYDAJT633C83mrM8nP0RAogJuUL5045sopfF9rhM7P5ffweur1uqCA2pCTk1AqlVAoFJDNZtHb2wuDwYBCoSB7FNiN6PeSLuAZYVCmz3Erv4TOkD7HAJrWr9Wxb12j1sBOR763ipaBXTSP19ea5+Y18LO1g6DvodWp4M/W19fh9XphtVpx+PBhrK2t3YTovPrqq+jv74fVapWf/UgMNB9KK5GKD8xms4m3ypvWZC3taZXLZdmwPLA8dDwMPBw0xB6PB4FAAN3d3eju7sbRo0cxMjKCUCgkUTgPWLFYbMpLcwEIgevoi4uhF1sTfHSkTWJUPB6HwbCTxyacTvgVACYmJmAwGCQ62CvCZ8H8PdfnVpD1DzoQAG76uYZHW585DyIdrEwmg1gshtXVVcRiMaRSKTQaDdhstqbv5F5r9ZZ17lvnmVvvQe8f5t35WuZZ0+k0MpkMgN0UUHd3N44cOYLl5eWm6LuV0Hg3C50Z3h/zjVxD5pYJVevoms6zjph5tlrXRH8PnXFyHOr1Ovx+P4LBILq6ujA6OopQKIRSqSToXavRpS7QipbfpdMmfA9fRyeduoH/PnLkCFZWVmAwGHD27Fk8/PDDqFarCAaDsNvtWFhYkM8GsKd0gclkakJNdVoSQNPz02kvbWB1mgLYfU78P8+dNtw8Z0Qo3G63EH7L5TKKxWLTmjNQvJUR1yhsqzOvr18bbNor3kc+n4fD4UAwGERfXx/m5+fl+86fP49IJAK73d5kz2433fXXMtDVahX5fL6JXam95lbICNg5wDoy5gMiHMgo2Gg0wm63C0zkcrmQy+WwsbGBZDKJ9fV1JJNJZDIZTE5O4tVXX8XLL7+MgwcP4vDhw02Qu/bqtVekr5kPTDsevDbCkozsNDOZkDfh+46ODnk+MzMzTfnrvRQ5NRoNOByOJkIG90QrNMUDoT1qHem2Ric6etYHp1qtIpvNIplMIpfLYWVlRRjYxWIRyWQS09PT2NzcFNLID4rmdeTM+9H/b4XgWw0GXxMKheSelpaWUKvVkEwm0dnZiSNHjuDGjRtSzaDRhr0kmpClzxijBK4xiVL8GY15q6Ki0qVi1A49iTzhcBiNxk6u12w2I5lMIh6PY2NjAxMTE0ilUjh06BAOHDggsDcNcmv6QkfSrdevoW1GgABkfwG7uXKj0YhsNguDwYC3334b999/PxwOB1ZWViTnzLPAvb4XpF6vo1gsolwuC0FWn5PWqPVWaBj/r//N/7fyDsxmswQGNpsNbW1tsNlscLlc8Pl8cLlcaGtrg9vtFkMNQIIxSmtUzmvTUXprClYHityf/P3CwgLsdjusViuOHDmCYrGIUqmEV199FX6/H263W763lWz2w+SOLAcNDvO7GmIAdhUsX8OL4ENgpMGD2gpfNRoNOJ1O+P1+mM1m5PN5bGxswOv1Ynh4GNFoFB0dHYhGo6jX6wgEAjhx4gTe9773YXV1Fc899xwOHDiAQCBwUzTH77VarSgWi5Jb1t45/2jSGj+jVCo1sZctFgsymYywvNvb29FoNDA5OYlCoSAepd5ge0HMZjM8Hg+cTudNjGyup4atNWdBK0kNiem/y+WyPFceiGKxCLvdDrvdjlqthmg0KogM0JzCWFxcRDwev8nQ69wYfwbsKmp9YHg4CUnpyJp/HA4H/H4/bDYb8vk8YrEYBgcHcezYMczMzMi5oGEuFos/EE24G0XngDW/g+dJlzvp3KQuu9QkII1yNRoNZLNZ2O12+P1+OJ1O2QdLS0twu90wm83IZDJS5sNrMhqNmJiYwNzcHB5//HGpoLiVk9j63bx+KnT+rbkpmv1PeP/EiROYm5uDwWDAtWvXcPr0aczPzwuzmXlallrtFcnlcigUCkLqbQ2+dGkr0GwEyQcAmtE0vS7kL3CPud3uJmSTehbYOV9EQEKhEHp6esSBa3UEaXB1cKadBM09YaCno2+ty4CdVMfW1hb8fj88Hg+OHj2Kz3/+8/D7/fB6vRKssrLgRxZB02jlcjnJLTFaaTQaAlcXCgXZ3FR+OufEHDUXjYxql8uFqakppFIpeDweuFwuBAIBuN1uGI1GbG9vo1qtYmNjQ9i0XEy3240HHngAc3NzSKfTOHHihETBhE8JOdtsNtkgPLCa2KK9KL6fjEx+Vj6fR71eRzweh8/nw8LCAubm5sRge71eIabsJa/Z7XYjGAxKGYVWVq3GF2g2hnzOOlLSyAW9XovFIk6A0WgUlq3NZpNSHr6vXC6LAmVao1wuY319XRRqazStIyMdXWnorDWi5r/1gXW5XE355XvvvRfT09OCJmgOhjYke0FMJhPsdrsoQN4nI03C3IQ/6fwCkLIYzcLn341GA6lUCgMDA9ja2sLq6iqy2awoa5vNhlwuJ+zrTCYjzhn3BiP4Z555Bt3d3Th58qSgX1wbOgs0uLxmOns01OVyWQy/3ivAbkSYy+UwODiI1dVVDA0N4dlnnxUSKSMrXheDlL0gxWJRasg1Kqlzza1MbD4zIqRaB1NqtZ3aZOpcsuUrlQpyuVwTh4iGlHuD32MymdDb24toNIp4PI5EIgGguWqjNecMNEPZGm3la7STCezm4dkTw+FwoFAoIJ1Ow+fzNfEueG86zfPD5I6x10wmg0wm03STJpMJDocDdrtdHiq9y3K5LLlZ7TmXy2VhAvOweb1e9PT0SAkLc9rpdFoiWHru6XRalGA+n5dcV1dXF3p6enD27FkcPHhQDGRrfplMTR5Y/fDoMFB0NEDjDuzAVYFAALOzsxgeHhYYv16vI5fLyf3e7mLcDeJwOBAIBMR71TkkIg86yuB6E+bV+elb5a3Hx8fh9/ulkQiNqs4NEsLiutErBXaitWQyCZPJhLW1NSSTSTgcDtl/GnLje8hJ4HUxl93q4WuPm46m2+1GNBrFBz7wAbS1tTXlPgnfm0wm5HK524a17gbR7GgATYiB5g3QEdfs5Xq9LsqUSo96olqtwufzSStJ7ifuIWDXoPM91EO5XE6cYoNhpxTqxo0beOedd/CJT3xCAgVN6OPe0jC3RgLpiGvnEri5kVFHRwcikQimp6dx5coV2UelUgnpdLqpwdNekVwuh3Q6jUAg0OT4tkagRLI0N0lH2hrBYCC3sbEhNsJutzetUSvPiXpFl3JxXwaDQQwPD4s+0M6VTmcwpaIDSqZWWtE/bcy1rKysYHNzE3/2Z3+GJ5988qbAhLbI7XYjmUze1jO+4zKrRCIhZSW8KXrE3PzcyLoEB0ATYQfYja5qtRqWl5cxNTUlnmYul5NotzUfwEjKZDIhm82iWq2iUCiIQjAajRgfH8fi4iJGR0fxoQ99qKkWszWfyANNj59RWWvuhAvP33FBOzs78aUvfUmcBbPZLLAc2ed7RUKhEGw2G1KpVFO0rPNy2lvUELGudeVruYeIoKytrYnTxgOiUynaoeO6MZLnc2bDHDJrr1+/Lp43D6AmrWhn02q1SgMe/kxDnNoRKJfL8Pv9OHPmDMbHx/HKK680PRPNVwB289x7QdispTV/qCMP/XNGQYweiG5ow84SSiImpVJJmgxRCVNhk6tSLBal9phkTe30GQw7jOrPfOYzOH36ND75yU8K+kdErNVw1ut1OJ3Om3Qa967OUfP1tVoNfX19sje5b81ms6Rm9hrEvbm5iWg0epOz0ppf1iQwYLcUsjVKBXbOmN/vFySWNoLoJ8+/Tk0weNNOgk6vWiwWjI2NIRKJoFQqYWNjQ4IuvpYImo7+aVt0qaDeK/qMG41GxGIx/K//9b/wyCOPSCkvnQw2anG5XMjn8wgEArf1jO/IQOfzeXloumMSc7rc7Bo2Yscn7X0wcgZ2D8Pg4KB41fSaCZnzQNJBIDGkUCjAaDQiHA7LQXY6nQK9BYNBJBIJnD17FseOHRN4GthtSqA3ABeMOUz+XitzGiODwYBwOIzl5WV84QtfEEND6J05lHw+v6cgbp/PJ93TtDeqD6COMhj58PnoiIuGjGtM7kEul5McM0khdKQ03MgDBOzWn+vIdXNzEwsLC/L8W+tt9UHjIWN6wul0NhlzDePzMAPA2NgYTp06hTfeeEMcBTbS4T1ns1lUKhW88cYbfzuL9Lcg2Wy2CW0AdrkINKT6GetIV0fLrTXSwWBQ+g1QdwAQY0pjVyqVpB8BDb4m5aXTaVSrVXi9XjgcDrS3t+PFF1/E1tYWjhw5Io60jt60Q1apVBAKhbC5uSkwPl+jiUW8vnA4jOnpaSGH8pq0A2q1WlEoFP6WVuhvR1wulzw3AE1nhPKDmPE6guW+sVgsWFxchMPhaApuWnPAunxP6wXuA/KkSGg1m80YGxvD2NgYJiYmpNxNQ9etUHexWEQikRDnrzVgo/D93/rWt9Df3w+XywWLxSJrbbFYkMvlYLPZ4PF4UCwW8cQTT9zW870jA51KpUQB6e4xmmTDyJKQDuFszX4jA1g/mHg83nRQarWalLHwMxuNRlN7USrSlZUVmEwmpFIp+Z7e3l7Y7XZEIhGMjIxga2sLtVpNWIfaiFABc8Hz+TxSqZTAWzQkeqO1t7djfX0d09PT0hiFxAUqrM3NTVit1j3VnCCXyzWRfrjG2oDRUdPPmL9jpK1hI4vFAr/fj0Qi0fReGnS2j2VETSeOzpLb7YbFYkE6ncba2poYilAoBIfDgUwmg42NDVEErSVTGpKr1+vo6OiQrnDc2/pwMqIfHR3Fgw8+iEuXLklETSeT+2FjYwPpdBoXLlxANpv9u1m0H4EwatFQJqXV+dFtevkzppr0XqJS0++t1+soFAri9HPNHA4HfD5fE0mNuUD2hXc4HEin03K9Pp8Pzz77LIaHhwEAhUKhqYmRRsXq9Tra29sRDoelrE+XjGkot7e3F5cuXRLyIo2DLjMrFovI5XJ7qtSuu7u7KR1IHcq11Qgq7YZOHfL5ab5HKpUSljYAOUvAbv8MYLdHu2bpE/0Edhwm3QOc62y1WvGJT3wCr732WhMqq/UB/7ZYLJiYmBAdxbOthQHpM888g87OTvj9fuGelEolaX3McjCDwYAPfOADuHTp0m094zsy0Ey0UyFTeenEO3MFPBhkwuqclIaYGo2G0NJ1MwcSgHhoqbjJ5m00GkIWy+VysNvtGBgYEGyfLPB4PC6w2PHjx9HX1ycePpWGLqXignEoAnOhmuTi8XgQi8WwvLwsDVdYTsH7JKvQ6XQKZLcXhOUkekO3MiC1sWaUovO4wG60BQBOp7PJeDFyYtpCoy0k3gAQYojVasXk5CQsFgva2toA7BpSOpCbm5tIJBJYWlpqImnoQ8k97HA44PV6m/JTmpFaq9UQCARw8uRJTExMiCJyOBySTyVEWiqVsLq6imKxiEgk8iNbl78L4brqSJLCNdMpD6C5Dpj/1pwUrgu5IAZDc898ohR2u72JMBSLxVAsFjEwMCBkQRpERllUmJ/97GfxK7/yK+jo6BB+io7U+O9isYhoNCpOAJ1vHTH6/X7EYjF4PB6k02lR5AwCGEURrufe3QvSCm0DzUNIaPj0+dXNjTSSpZ0yjT7wM3ie6EDRQNMWlUollEoltLe3w+12C5FQk8ZI+vX7/XjwwQexvLwsSI9OP2m9VigUBDHU3BStN2ZnZ1EoFBAIBGTti8UinE6npF9qtRqCwSCGhoawsrLyoyGJMe+rSyZ0fo24O1+Tz+ebcr+abJXP5+H3+5HP52EwGIQJWKlUkMlkEAgEYLVaBaJiLoCHL5VK4cCBA7h+/TpGRkYQj8flcNDjohddKpXg8Xjw/e9/v4lkpMuoADQ1OiGE2gpxMk9Zr9eRTCYl6qPR14fX5XLB5XIhGAzeyWO+K0TnWTUZhIdRR5oawtKHmkpMHxLCgST8aFjMaNwdbECEJpVKYWJiAgMDAwB2B5W0ljkZDAbMzc1he3tbemjr69S55Xq9LhNuSHzUjkaj0cChQ4cEBiPcazQakc/nJbVRqVSwurqKVCqFYDAoTuteECIgmrilc3lM8VAYUbVONGMe2u/3N7XYJAmNn5PL5QSu5HPnd25sbGBkZARut1sQMmC3exevyeVywWAwoKenB7/5m7+JX/qlXxKIFmiGLqlU6/W6QN2aMMZ7dTqdQkqk/mIgwgg6k8nIYIW9tAe0U3Mr8pw+L62v18+QOjWRSIiu1+kBpjtp/AAISkGH2GAwoLu7G9lsVlKuTHkAu1A2UywHDhxAf38/Lly4cMvrpR0xGo1YXV2VagHtYBLB+973vofx8fGmOQ/A7h6i3urr6xN7cbtchDsy0HyY9GKZJ9QwsC574SJQwek/hJFZfqIL3nXEmc1mxWOq1+vIZDISJaXTafj9fomsDYbdwRxUBHQcarUaHnvsMYTDYek8pvNkJpNJIAhGwOyzTHY2sAOrr66u4p133pHpXozoWRvLRivMq+4lYkgrAYPKlM+Nz7L1UPJ9fC8PAL1lOm50djRBkF4z90GlUkGpVMLU1BRMJhOi0WhTrbX+DEJrPDzxeBwmkwmJRKIJWqPoDlh0sFjuxdf5/X6MjY1henparo9RtL7OhYUFrK2tIRwOi/O2V4TKrJVoQ6NqNpulUUQra17XpFerVSwsLMDn88HtdktEqxEzRp9sn8mqkc3NTdRqNRw7dgyxWEwiXA218lqBXUNCZf5P/sk/wa/92q8J0qHTNZo81tnZCZ/Ph0QigVgsJoSfQCCA69evNzmj3Afcb2SCt7W1IZlMNjktd7vo9dSBCZ9/q7HWiJb+jEqlImgG38/38jnq9CKNL4OsaDQKj8cj401JJtNoDIV7tl6vY2xsDKFQCDMzM03loAwa9HeurKyIXiBUbzKZsLS0hNHRUek0R5SPqCCdihMnTmBtbQ2VSgU+n++2eUl3bKB5WHij+oDWarWmxD6VHRV3qVQS5TowMACHwyGEIF3nVq/v1BnT8NEJSKfT6OzsxNraGkKhEFwul3jSqVQKpVJJkvlkjvLB8fp5TQcOHJBFYNTGaJ7vr1Z3xsR1dXXh8uXLaGtrw8mTJ2WxNCWfeQ725W1ra5PxY3up/66uXdURc6tXzEOrc7gasqKyDQaDsgbAzkGm48SoiUaW672wsCCDzunMkThCMg6NpGZu8/Nv3LgBj8fTVCpG0XCnx+OBxWLB2toarl69iuvXr2Nubg79/f2YnJyUQ8b7Y0kV2cgTExPo6+sTo7OXlDPQTPLRJVNcZ64D0Ny1if+mkzw0NIStrS2JUjiqEoAYPf3ZhUIBS0tL6OvrQyQSwcLCAvx+v7SdbGXXksgK7DbAMBqN6OnpwVtvvYXDhw9LUKFJS7zHXC4nUPaZM2dgMBjQ1dWFD3/4w6KjdBMjEgMdDofURhPq3ktMfhpkGjygGbpuTX/q9KVOeYXDYVy8eFE4BXydRq4AyBwF2ohgMIiBgQFkMhkxzmTRUy/z+oDds019X6vV8NRTT+Hy5ctN/fmBXf1GPZZOpxGLxXD69GlEIhEZs/ulL30JHR0d8pmMqguFApxOJ3K5HHp7e5FKpZBKpaREVTf4+qvkjodlkEXJQ0aPgkqWMBZvkJEtiVJGoxGJRALt7e2i2HSNqI5GYrGYEIAIa6VSKencFYvFAOwQl3w+X5MXz842yWRSlAE3RVtbm0TuOo/Bem0ASCaTGBkZwV/8xV+gq6sLjz32mNTYUtnWajVR8sxvFItFGAwGGaiQzWb3XJtH3i+VnzbEjG6AZs+10WjIXjCbd0dwknSmSyjIb+A4T2CHPZ5Op3Hp0iWYzWa0t7cLu58REA0CSRpkCjOSJuLDA02DSa9Xe+0WiwXLy8t466238Cd/8icIh8OSAmlraxNGOBUzc1Q02m+++aaQaNgVay+JJmYBzZOCqNSIRPHnPMc61bG5uYmDBw8KsZRGVjd2qVar2N7ebqor7u/vBwCsrq4iEolgbW1NeANaEdPpZk8Fk8kkxtpiseC///f/jn/4D/+h6AGdr2Qkfe3aNQwNDeHJJ5/E1atXUavVhACqURtgl+BqsViEAcyITzPW94Lo860hbj4DXTXDs81AjK+tVqt45JFHMDk52YRkEZHQiCaDnVqthqGhIQSDQWxtbTUNQCKsrHUy4W1NTOU12u12nDp1CtevX29C4TT83mg0MDo6io997GOYnp6G0WhEV1cXAoGAkFJ57brjJNOpp0+fxtraGiKRiDTxut198NdqEs0oWeeJmPdLp9PiofJh8lDQM6ay5kMn5OX3+8UTp0LmoV5bW0N3d7f0WiWRi2VNRqMRyWRSro3NVJi34ng5ei/VahUnT55sivDpBCSTSTz22GP48pe/jM997nOYnp6G3W6Hx+MR+FrDu1xos9mMWCwGn88nkZ3X672JQHM3i2ZNAje3zSNSwmeqoybtuGUyGWHTtjK6SQIiTOTxeLC5uYkrV66gp6cHbW1tQsJLpVJNhlmTiZiD4v7ingWAa9euSfkDxWAwCI/i6tWr+NrXvoY33ngDP/mTP4nV1VX4/X4AkGhIH3w9RW12dhbr6+vCIN9LKQ4Kn7Fu8ahZ1kS9NBuXzhkdaABSecF10GNgHQ6H6A2PxyNI1NDQkJTYcV8FAgFBXggrF4tFGaKgm95oGHp0dBTPPvssDh8+3FRFwpbAFy9exM///M9jamoKhUJBCEiVSgVf/OIXxREAmqdo2Ww2JBIJdHR0NPWq1uWId7uQ1Mt70tEuAIliyfbX0LXOR9Oh5/tpwO12O7LZrKSRgB3u0okTJ1Aul7GxsSGdB3XkrrvG0Smn3eF389/lchmPP/44Zmdnb1mNUi6X8dhjj+Hxxx/HSy+9JJymfD4Pl8sFr9crn6nLOQmDnzlzBktLS/B4PMJP4LO5HbljA00oUd8sD4o2WEz88+HQqNntdthsNsnt1ut1RCIRxONxlEolgScSiYQMRZ+bm5NZmrlcDrlcTpiy1WpVjD3LsnTJFMlcrKnV5I65uTn09fUJHApAYNSVlRX803/6T/H2228jGo3KxrHb7cII1bR/EoQikQjC4bDkIXgNe0U0KVArZG2gNckD2N2MugOQng5EWIo9uJ1OJ/L5vCAjtVoNS0tLGBwchN1uR6FQkM+hA8faQ+am6CDxcJMJqr3sra0tYX+yzMdut+PixYv4/d//fUSjUQwMDCCbzSIcDgvHgV45970mIhWLRUxMTKC9vR3pdFr2HNAccewFobHlWmvYUp8/Gjw6w8BuTTprxGloddqI1RmMqGOxGKxWq5RSUZ8AEEXNf7NvM6Mvt9sthpq8FmCnXOuFF17AyMgIuru7xZDk83nRAc888wwGBwel7zRRIpJm+d3UfU6nE0tLS/B6vU3R2F5y1IHdJiAarm1NfWliMLBb1qoduZWVFXHo2MyFa2AwGKTxSzwex8jICDY2NmRQER1ECrtX8rOB3fGyNN68Xn5fo9HA448/joWFhSbCbz6fx6c//WlEIhG88cYbCIVC0m6azobudMd1ttlsyGaz6O7uht/vx/r6uqS+7hRBuWOIm9AhlSv/r2EsYLf1H6EdLkoymZRQnw92Y2NDiDTpdFrIX36/H8vLy+js7JSc9NbWFrxer+D5zDXH4/Gb8s31+k59dXt7O+LxOLLZrGwQl8uFwcFB6dnMkoovfOEL+PSnP4329nZR6sx7kThE6IJRkzZOZJ3r8oK9pJi5CSlaOetcU6tSYuTCHBLbb2poOxAISGRmsVikWci1a9cwODgopBAW/ROJ4fPm8ASde+b/STYCIHD02tqaMHsNhp369+XlZXz2s5/FT/3UTyEej0vphC6dCwaD4oDq3JrFYsE777wjUYEm0O01IRTMSJROinbSaQR1/Tp/R0epu7sbqVQKbrdbolZGLy6XS6KObDYrsDb3ACOVVCol51Knn9gYhNeUTqdleIGuQGlvb8e3vvUt9PX1wWq1Yn19HZFIBA6HAy+//DL6+/sRj8cBNLP+x8bGmqpCuA+y2SzMZjNCoZAY9FZndS+ITklpQp4OSvisGGnrio96vY6enh5cvHgRnZ2dogtI9nM4HJLPZ9+BTCYjCJZ2wvh9DJy4F0jmYl6YukenQyuVigRWsVgMlUoFHo8HDz30kPRWcDqd0sCKpNNKpYLl5eUm/gWfRa1Ww0MPPYSJiQnZH5Q7Cdju2HLwYGrGI39OiFiPZCOrjouSzWYxNjYmjQMAiIE0GHbG+On66NHRURSLRWQyGSGJ0SPP5XJSWuH3+6Vemgw+RvB6+pb2olhO4/V6sbCwAIfDgU9+8pMyNrBWq4lHpst1OOaQm8PlciEejwsxjJurtfZ3LwgPmK5n1k5K6+FsdU4sFovkn1dXV5s2dalUkmiHKEYikcDo6KhEP7rekDlFOlA80MCOd8+mAbp7Ew262WyGz+eT6Gd9fR2XLl0S7kEqlZK9RsVCx0NDciQG1Wo1vP3221hYWEBvb6/0h9+rYjQahfmsO3hRtCPHNdF5SJNpp9/y2NiYnEueW0ZGfPYej0dqihk52e12qeqIRqPY3NyEy+Vqim4ajYY48VS6hJt1ftJqtcLv9+P73/8+/H4/Ojo6sLq6Kk4l8+h6b9G4EJ4nW9dqtSKdTqO7u1uY4JqrsZdy0MDuGddrrImBwO4AIgr/XyqVcPr0aUFEaDCJzrESx2q1or29XfLNNJCZTEZ0DdENls3p1FqpVEIqlZKmRrRR7FhHHpTH44Hb7UZ/f79E7Vx/XaOvg0A9T0JXDT3yyCOYnZ0VW6gdtNYg56+SOzbQmlTFm9OlDcz3aqNED4Y5akaxuu8uH+7y8jL6+vqQz+cRCoVw48YNpNNp9PT0AIDAoIQlFxcXhaTGZidmsxlra2vCpGQ9s2bvpVIpdHd3I5fLIRqN4r777sPy8rJEzkQA+H4qhomJCRgMBrjdbsk7Z7NZhEIhgcI1rLPXDmRrFzgKjReFykhH0NwHm5ubks/N5/OiiJmuYJMY1o6Wy2Wsra3B5/MJW7fRaMi/NVzJAxAMBmXqGqsOCEsCELKjxWLB6uoqbDYbOjs7kUgkpKEIFQm7ABmNO21ldRkgjXc2m8Xq6qrkHHUk0Zp32wtCQ0hnmvu9lY+gS9laDfji4qKsBxUko2eLxSJ8DtaWb21twel0wufzSWObQCCAhYUF9Pf3IxaLyeeRF8DaZYPBIMqdTSTI6OZIW6/Xi7W1NXEAqGP4fo/Hg42NDVHcJAzpe2Sum8+E19Gan90rohnZrY6I5mi0pvpo1Hp6eiTlqCuAdG47Eolga2tLKjQajYbwkxiosT8CDSVJaiRw0Sax6yP3LSuG+J0c1ENCGt/PQIDRdKVSweLiIg4ePCh7n4Y4EokI658NVrTcKnD5QXLHwzIASHjP/B8PJW+EQg9be440piR00dBVq1Xk83mMjY0hmUwKdh8MBgWW4gB0jvTSeW5CouxKxjywvq7t7e0m1vC1a9cQCASwsbEh3jY/l552PB5HNBrF2toaent7ceHCBfGc6HQUCgUZK0e4A2gebbZXhFAmlTPQ3DWKSrn1Z9qDJHkvnU5LG0edr+cksHA4DKPRiHg8DrvdjkAggEwmI5PPaOjZjJ7KkgdIj5okhEpv32w2Ix6Py/i/dDotkZQme7DRjNvtRiAQwGOPPYZIJNJ0+Ov1Oqanp5HL5UQB8LDq56CjiLtdeM41ix1o3gOaiQvspr3oAPN1yWRSGP58bTKZRFdXl/T+j8fjCIVCEv34/X40Gg3E43F0dXVhY2MDHo+nqfSNDWOMRqOQDZkG4bUQlUskEnC73U3NTnS3J9Zdj4yMYHp6GqFQCN/61rckhUI9YDAY0N7eLmka3q/OUe4VoX6ljm0NyloddP5N9JXnw2aziXNNljP3VTAYxPb2tjhVfNYMyJgOpTHl5/Jv3ToWgBhn8knIeXG73YKGslSW98S9zE6aq6urcLvd+NznPic9tYncer1eHDhwAAsLC4LoaERRVwncjtxxHbQ+dNor1IvBxL1WhlpBs1SFjF0q9EgkgkQigXA4LB6Sz+eTwec0jDSW7MpSq9Ukd2Cz2ZBMJoVNTtibcFYwGMTGxgY6OzsF/ibTWpMJaPg9Ho+0cjMad+qt/X6/kBjK5TI6OjoQjUYB4Ca25l7OQbYSPvSm01EjFThfx2dEJ69Wq4ljxfInQpOLi4tCANOQYSKRQGdnJ6rVapNx5sFm9MPDp2HRSqWClZUVaYxht9vh8/nkQHOPA5BI3+fzYX5+HtFoFC+99JLkx9ktbHt7G+3t7QKZa2hTE2b2iugaU61sW0VHl/r+OfmtUqkIE5Y8BI5x5f5g+ogIS6FQkO5cFotFqkFYzlapVESZ0uHiiD8iJ1tbW+js7BTkhlUj5Bnoa6XTZ7PZEIvF0NPTg/vvv1/ae1IPEgrXuk5Hla0o090urZyk1t8Buwir1gd0VJgnJtmOr+fvSMrK5XLwer3CPaCQDJrJZG4KFInw0FFgSlL3WHC73TJXgutCRI+kUa5ho9HAysqKGHC3241Lly7JTHgGZ319fTJhjYZYp1d5j7e7D+6YJEbyhWav6oiKE6j4Gl3CBECgKdYiMs/g9XpRLpfhdruxvb3dBEkbDDsdoBi5xGIxXLhwQVi8hMo3NzdFwTocDukaRbY4SUFsMdrd3S1dXejp0IFwuVzI5XJYW1vDyy+/jJMnT+Ly5ctC7WfEXy6XEYlEBHohrb8157BXhIeAaQ4NZ1E0vKVfQ2IQISpGy4SY2GiGZUmLi4vSrpUpEdbdt7e3CxHH5XIJ9M4pRezwpjsRVSoVzM7OivJtNBrShpXld1pB+P1+tLe3IxKJYHt7G4ODg2hvbxensFbb6SG9tLQk+Wg6La1e8l7aA8BuGZUePKKNNO9d8zeoNIHdPsfcI0RjOKGKvIFUKoWuri45w0SrAEjkRGYtmf8WiwXhcFj+7/V6kUwmpUQqnU4jGAwK0ZRVIVTWFO5bu92O7e1tuFwuXLt2DZVKBRMTEygUChK1k3Xu9/ubap41WW4vcVEANK13q57TiIpGEKhfyQWi7m80dsvV6Fgnk0mUy2VpGESHjvXwXq8XsVhMnO50Oi2RNCt7gN20LNepVCohGAwKmstUCG0byax0LkwmE+bm5hAMBrG+vo5AIIAbN27g9OnTsu8ZAAwNDWF5eblpXCmfg3ZOfiQ5aE0GYJTcOlqPohW0Lk7n54RCIXi93iZFZzabsb293US+IcwcCoVQrVYxODiI0dFRnD59GiaTCQsLC4jH4+js7EQmk5FFPXv2LOx2OyYnJxEOh3Ht2jUcOnQIGxsb0upza2tLjAPp8mTpra6u4vLly+ju7sbg4CAACIOcHjbZfzrHSSdF51X3EqwF3DzXW3vHOlLmvtAQGMk+hKkozDMyv7u5uYlQKCTrQm/ZbDajp6cHTqcT/f39iEajqNd3OkV1dnaiu7sb7e3tGB0dlWlT7e3tqNfrWFxchMfjkRK+RqMhbODW6/R6vRgfH4fNZkN3dzfm5uYwPj4u76NhSiQSiMfjkivnGWhlse41A022tIbzed51zo7MbGBXf/DZsC8Bna5sNttUOkm0iiNbdcRMo9nW1obh4WHhABw6dEgca7bWnJ2dhdPpxH/5L/8F0Wi0ae8RQQGaIVs6W8FgEI1GA6urq5iamsLAwAA++MEP4vz58zI60GQySSqMkGorUaqVj7AXRBOe9H1pCBvALfc/f9/aYtNoNEq5bLVaFYIgz6nL5UImk4HH48H8/Lyw7peXl2G32xEKhbC+vi7IaTKZRCwWw8rKipTVck1Zq0wnisED74HXlEwmsbi4iPn5efh8PjzwwAP49V//ddx///2C8pRKJQwNDcnER+2g69TGnaY87xjiZp0poQLm4nT0AKApH0AIAIDcDLBL6gmHw7Bardje3pbuTPx8ksXK5TIKhQLOnj2L9vZ21Go7E7Py+Tyi0SgmJydx8uRJjI6O4l3vehf6+/uxubmJxx57DJOTkxgfH0exWITX6xX4k3+0UTl37hzS6TSOHz8Or9cr7RqvXLmCd955R4wuy7tCoZAoHiptwiM6/7hXhMpLd/nR0KYut9GQk/4ZNy8Z21arFYFAQIxzLBaTHusAhA9gs9nQ1taGubk5+P1+KbHJ5XI4efIkvF4vOjs70dvbi/vvvx/d3d1IJBJ47bXXhA9hNpvFCFCoaEgSGR0dxRNPPIHjx4+jWCziq1/9KkKhED7+8Y/jlVdeaepQtbi4KJ/Hkg6NKunIYi/tA0Y0zAlq2FKThujM8DyTMNNoNKTGPZ1OIxQKIR6PS3veeDwOr9cr0ZHOGxMds1gsSKVSKBQKOHnyJHw+H6amppBIJOD3++H1eiU9cvHiRfzBH/wBJiYm0N/f33SdrVUH1WoVkUgE6XQa3//+97G2tiYlYSdPnsTTTz8ta059RUIbdZ3e89o47SVHjc/rVqkNnvXWHLQWGmeijtvb24KmkMBHB69erzcNTVldXUV7ezs2NzcljWAymZDNZrG9vS1DcdbX1zEzM4NSqYQDBw5IhMw2nK0kP62jMpkM3nrrLVy5cgV2ux2HDx/G/fffj8997nMYHh6WplVECA4fPoy5uTlBg1vh/Fvl5H+Y3DFJzGq1Ss5AU9mBXeYkc7mEF3Wf03A4LO3RSHvPZrMyJpKGnYMrKpUKstmsNDZxOp34whe+gGq1imAwiL6+PlQqFXzta1/DxYsX8d3vfhef+cxn8MYbb6BarWJ1dVUajOhWjPSieSDpoX34wx/G5OQkXn31VanP5tALQh8kkYTDYfh8PoFUCe9TEevawL0i3MyMFFtrPHXEQIeNv+e/o9GolCpZrVbpo2s0GrG1tdWUU+b7zGYzXC4XEokEhoeHpaHB0NAQOjo6YDQasby8LK08V1ZWcOnSJXz2s5/F8PAwZmdnEQwGpXeyhlqpJMLhMI4dO4ZTp05hcnIS3/rWt/D9738fkUgEv/d7v4epqakmpCiVSmFjYwOBQEDyoK2HsZUgt1eE8Harw6aVcqOxO7BEdx/kGXrwwQdx5coV4R+0tbXB6/Uim80KCYxkK36O0+nE9va2lDjFYjGkUim88sorCAQCiEajcLvdCIVCMg42mUziJ37iJ3Du3DkMDw9jfX29KVrWzGN2Hpybm4PNZkN/f7+UTx45cgTf+MY34PF4mmr6G40Gurq6/ko4c6+lugA03a/+m/faSorTqRAawY2NDalhJ2ydyWSkVwVha0LGdrtdyKErKyvSbSwej0v0yvTU8vIyQqEQTCYTnnjiCTH2JPRpRxpoDj5mZ2fx0ksvwWw2Y2xsDEeOHMHw8DD+4A/+ABcvXsTDDz8sNdeNRgPd3d2SC+c50M6LDtbuhCx4RwaaNWG8GUae/DK28NPJeS5KNpuVkpfNzU2BxehFBwIBiTpJ3mLeye12o1AooFgs4t5770U6ncaNGzfwxhtv4MqVK1hfX8eZM2dgNpvR2dkpHYoefvhhzM7O4uDBg039mAnB1et1uFwufPvb38Zjjz2Gt956C+fPn8fKygrW19fx1FNPIZVK4e2338bFixebnkMymUQ0GhVUgc+DfzSJbi9FTreCs7QHrY2y9ha5aUulEjo7OyVvyK5OVKSEuQmTEdIkeejAgQMYHByUfTY7OyuQtM1mkzGR29vb+NznPod3vetduH79OgYHB28q++B9jI6O4siRI3jyyScRi8Xw3e9+FxcvXsTk5CROnz6N3/iN38DLL7+M+fl5cUxrtRrm5+cl1UEl33ou9ppSpuj0hIaEtQLiswbQ9Gz4u7GxMVy6dEn2Bttj0kgSPtf1px6PB8FgEFevXpUUCFNivb29OHfuHEZHRxGLxeByuTA3N4djx45hfn4efX19UpLTakzMZjMOHTqE559/HrFYDJlMBr29vejq6pJ+DF/4whfQ39/fVA9NqJX5cKa7tJPC77rd0pq7RRi5agcHQNO96+5xOg9Np/vNN99EW1ubtFHmeQ+FQkgmk6jX6xKpEoJmm08OoCBP4PDhw3j11VfR2dmJra0tGarxyU9+EolEAj6fDxsbG4J2ETnRyO/Fixdx5coVuFwuHDt2DMPDw/D7/fB4PPiLv/gLbGxs4Pjx4011z41GAydPnsTMzExT6SefDb/jr4Ok3dGOsVqtsgG5ILptms41a8IVDxH/XS6XcfnyZTgcDmkawT96nmutVpNOLWRefuELX8DRo0exsbEBn8+HaDSKy5cvS6QM7JCLPvjBD2JlZQXj4+MCi+jrq1QqGB4eFgLKf/7P/xkmkwkejwdnzpyBx+OB0+nE9evXpYsZ78Fg2GmoQniW7GAN7fHvvVj/qNmo3Ac8jABuaQiphMvlMnp6eoS5ncvlhLFrs9kQCoWkjzrhLo/Hg46ODhw8eBDJZBKrq6toNBpIJpNwuVw4fPiwTMN5++238f3vfx+vvvoqenp6sLm5iZ6eHinT0FGe1WrFE088gUceeQTxeBwXLlyQIe6dnZ04fvw4/u2//bd4/vnnpTsU90A+n8fGxgba2tpEeeg9ryH+vaaYgd26dhrm1hQXpRU+pqEGgMnJSRw5ckQiWipWi8WCZDIpTHk2AgkEArh27RpyuRwGBwcxMTGBBx98ELlcDh0dHdje3sbw8DBmZmZw4sQJBAIBJBIJQU10WSBLJHm9oVAIL7/8MsbHxxGJRFAqlXDjxg0pvfva174mbYE1bEkGOFm7uvYZuLkfwF4STRK+FUJAfa+Fr2dq4NKlS+jt7YXdbkdbWxsuXbqEkZERbG9vi9OmS1qnpqaaWrVGIhGsrKxgYGAATz/9NEZGRgTR5FjYZDIJh8MhNe7UPZorsbW1hf/xP/4HgsEgxsbGkM/n0dPTIzros5/9LHK5HE6fPt2EkplMJqne0EiS5uO0PpsfmYGmh6hzdto71NEUu23p4QoejwdWqxXRaBTt7e3IZrPIZDKIxWJNeWlCHF1dXchkMnjttdfgdDrR19eHRqOB9fV1gcCy2SwefPBBiWqLxSKOHDmCcDjcFMXTGBAaNZvNePXVV/Hcc89heHgYgUAAbW1tsNlseP311/HYY4/hD//wD5sGPBCWZ/TMaJnEBmCXIEHlvNdIYjpnxP/r2m/tIbcSRfSB5qEkg3Ztbe2m4RLlchn9/f3o7u7GxsYGNjY2JDricIS+vj4sLy+jq6tLGNwmkwnT09MYHBwUJU9HiRAYCY4bGxt46aWXEIlEhEcQDodx6tQp/PIv/7KkUzRBEtipheRcYN6jNtLaidVe9F4RKqRWx1SjZvwZzwGAJiNZrVbx8Y9/HF/84helNKWrq0tqkOmEud1uOBwOLC0twefzoaenR+plp6am5D3lclki2e9973swm80YHByUPgetLH1gd006Ojpw4MAB+Hw+XLp0CaOjozhw4AAuXLiA2dlZtLW1Nek3kttisZi0KWYkTyPdehb2Uh08hevZWk7auvdvZax43s6ePYuxsTE4HA6EQiHhFbCkyWTa6WPNfgiEr6vVKqLRqDSYYTdJRtPME3OymA4kaBvIIzCZTNI5kN/TaDSwtLSEv/iLv0BfXx/6+/tl/+je7wMDA5iamkI0Gm1ibrdyD3Sw8iOBuEny0CO8qBB5w1SGvGkOu2e5U7W6M0nq2rVrcrHd3d1wuVxCqQ8EAkgmk5icnMSJEyeEDMKmEoQPLBYL2trakMvl8O53vxuHDx9GOBwWWIIPk1NxWO+cSqWEEOR0OhEOhxGPx7G+vo7jx49jdHQUf/zHfwyz2SyGl8xjthX1eDxNB45wvL4+7UXtJblVboWRoob3NEkM2B0hVyqVMDg4iPn5eWkk0dXVJYaNkBYdKTaxCAaD2NzclBIMMjo5PpRrsLCwIH2VWWfNw0SIjEMTWNNIuNTv9+PRRx/FyMgIvve97zUZXIPBIG1j2UxHOwV0zFoj5r2YgyY5jPfaWqYI7O4NGi7+DezmL2/cuIGjR4/CaNxpj5tKpWT92WOgUCggmUyir68Pa2tr0p+/p6dH1pJOwZUrVzAwMIBardY0UY6dpVwu1029nHkd4XAYxWIRY2NjmJycRKPRwMGDB8VhpGJnTr1er4vjR2QQ2OVfaIOwF1EU4OZmJfo+NaLQujf4zCwWCyYmJjA+Pi77o1AowOv1yrAUOtVLS0uoVCrSs99ut2N5eRnBYFDyv2R/37hxA729vQB2mt6wpE+vje7D7/F44Pf7MTExIeWe5XIZ3/zmNzE6OgqPx3NTHbveA60lx3w2vFc+Cx3U3o7cMUmM3jBLowAIpg/sMr21UmQts745Qo7lchmJREJKKlZXV1GtVtHb24v+/n4pBmebUL/fj1QqBZ/PB5vNhpmZGRiNRqyvr+O1116T6yCpjP9mkwG2hzOZTAiFQgiHw9ja2pJ86MbGBlZXV4VF2Gg0ZHwmy31CoZDkx0iK4+aiEeIG1AZqL0ir93erJvCEEzVZgr/j74GdPeBwONDX1yeGm8zsYDAo0TPTCHQMzWZz0zS0fD4vLV9fe+01MZjs1w3sdr9LpVLiNJXLZSEk0YkkCsO538BuSRGJj7lcrqlBB/d5a/0z0z6tz2cvCBWR7lHfmnvkM+V50PtAv9bv9yObzeLatWtC4CFxaH19HQMDA9je3sba2hoGBgbQaDTEkD/99NM4ePAg2trakEgkMDQ0hIWFBZkPz+iL55mKHICgY1wzm82GkZERrK+vY21tDf39/fJ6BhNEfZiOI0GQToCGODW7mHphL4nWBdoJB3Y5CpohDezmYjWJimuQTqdFr2cyGWFDp9NpXLt2Db29vdIJjOVyiURCuEDZbBZer1d6FzAabzQa0rGQZ5YVAsBu//BgMIh0Oo1YLIZcLodgMIhsNtu0x9kDnI5/NBrF8vIyotGosNA17H8rY30nztpfy63TTRm4CIR2eOG6NpIeRigUkgcUDAZRLBaxvr4usEUikUBbWxsCgQDm5uYwNTUFv98v+cl8Po94PI7h4WEZtOFwOPDd734X8/PzeOCBBzA0NCRdoYxGIzY2NiR/wQia0AvzEKTeJ5NJgVzpZBARYC9uMg3ZCINtKlluRgeBm3AvdY8Cdsk+FB4wHTVzzbk/iCbQkeGzD4fDGBsbw/z8fBNk2tfXh97eXni9XqysrEj5Er+v0WggFothZGQEfr8fc3NzuHr1Ki5cuIBIJIJAICBzgI1Go3SUy2azTUxrGmzWwlLJsiZXw5JUyI1GQ+BtwvE8uDrlwWvVDlprfvZuFg1htzqg1AVkzeracLZC1ROlHnroIVy6dAkAJFplG2C3243JyUmBHwHIui0tLeETn/gE0uk0rl69KpBzIBCQBiR07Px+v4yd1SWQ1EdsmmGz2eDz+YRzQgSNpVPMW7P9KNsUW61W2XOswdXGmve+10SjS8AuEQzYjaABNBlj/o660ul0Sp0xhwwFAgEsLy8LSz8ajWJubk7qnDVCEg6Hsby8DJvNhieeeAIGgwFbW1vIZrMy2pj7QHeuo41gDTSbFrE/RjQabeqPoWF7OmqsuzebzTKw46/iHWgy3e3IX3vcJDcqPQLdlAJAU0RN71oPNnC73VJHaTAYpFHFm2++KVOMgN2Zo6urqwgGg3A6nXjnnXdw9epVzM7O4u2338YHP/hB3HPPPSiVStje3kYmk5GWcJwpzeukgudBZa9lTjniQbTb7bKZWMPJzcY8POEtwpxsuMD7J5y/lyInPkeuG7BrgPlvXcuoPWuiLySLdXV1SXtF7o9gMAiTaWdMJAk+rI8vFApIp9OoVCoIh8PY3t7G1atX8dJLLyGXy0lHN41usCMcv0MTfICdBincl3a7HS6XSwYs8Lp1ZFyr1aT2kvX0vG8A4hCwKxUN9p3kne4GIZpBJ0ZHmfq8O51OeR2VF3sRUAwGAy5evChn1WKxyPo3GjuVHsvLyzKwxGAwoK2tDW63G11dXVhaWkKhUEAwGERXV1eTc8hgYnt7G4FAQNZeQ6000kyhdXR0IBQKSXcpzaeg7mC6jl0TdZMaDn4AIH0jdN31XpHW869LLKkHuBc0YZLwMvus22w2nD17FsPDw3Jmy+UyUqmUNJ7y+XxS8UM9TpJxOp1GLpfDJz/5SSwuLkoVAA00O4hVKhVp+9sKyRsMBgn6SqWSlM8R9eC9MligvvD5fAiFQtIKWvOO9Ozp1tzzjyQHrXOurTkFYLdelXXSzNkxkqQipAfMAvRkMikL+Z73vAcrKysCf3OABUcCWq1WPPzwwxJ9/+t//a8lfxiNRuVwu1wuJJNJNBoNgbRpUPkgee3lcllqMePxuCwisJuncjgcyOfz6OjoEMiTz4Hv14eRjd2Zs98roqG71mYfutZPR9Y6N6s9Ua6t0bjT15x5ILbWrNfr0qXL5XLB4/HA4/EAgDTRf/rpp9HX14eOjg4AuzCz2+2GxWLB1taWOJBU1vw/I6JisSgICUllVKZEg5hvjcfjUjOvp7ZpbgadEDZa0M7MXhE9CKIV1gd2IW8qKz53PaiAaYdYLIaf+7mfk8YR5JwcOnRI+CLkCvD/Kysr2N7ehtPplNKaQCCAVCqFdDqN9vZ2qZtnFKSrSHg2gV1IPpVKoVarCaOYxCMaZw7jAXaHbPCa6YDwvoms6TIx5tX3irTC21o38Jnw5+Tj8DVsz8zPoBFlOWUsFpPOckTM2traJI3A1AlJff/4H/9jzMzMyM+BnT26tLQklUIciEE71EoUpME1mXY6w21ubsLr9TYhIawIKhQKGB8fx8LCAtxud1M3MpaDaWSmtYHP7eqDOyaJ6To/rbz40HR9GcuyuDDc6LlcDkePHsUzzzyDZ599FuPj43A4HFheXsa5c+fQ0dGBYrGIdDqNyclJ8YiZU1xYWEBHRwf++T//51haWpJ2nSaTCW+88QYmJibw9ttvI5lMNnnKrYxaDXVyQ6XTaWnZSQXM3zMvQYiMyoc1uQDkWRDuoMOwl+RWpCftIWsvV3vWhLeNxp0WmfV6HUNDQygWi/B4POjr6wMAGekXDAblcPAAEyr0eDx47rnnMD4+Ls5cPp9HIBDA2NiYRDE8iLxWGktGtxT2aOfQBWA3v86IyWTaaS1LBjmNVGuuGYAoEl6zRlf2gmiHi6KRAj4P6gddr8yxjzxvo6OjCAQCmJ+fRygUAgDMzMwgmUyKoezp6UEsFhMDQB7Kt7/9bbzvfe+TFBWHWlgsFiwsLGB9fR1zc3OSiyQCyPPNxkJMY9XrdWxvb0u/baYuyuVyEwGWilqXUVIhc68S3qfipsHeK6Lh2laoW58J/ly/ns/cYDAgEAggFouJ7szlcpiZmcHw8LA0K3E4HHA6nVheXkaj0cDGxoYM1zl27JggKAzKfD4frly5gkwmI+lIHTTyXLbaBHYrI+HM5XIJQkJHi3t6aGhIUnD6HDBI4Gu1gdZR9e3IHUfQzMPRS9QQgCYHMZok+1l72sFgEJVKBU899RTGxsbw/PPPCx3e5/NhYmICuVwOvb290mmM04voYZfLZbz++uvSXo9ezNDQEE6cOIFPfepTOHDgAN555x3pVKV7SDcaDVy6dAmrq6tS+8j8h8VikSha59s5T5jGRzdW4R/ms/WCaTjvbhcN9QHNzRd01Mm/da6SCrJarSKVSuHUqVMYHR1FoVBAd3c33G63jJNkP3SPxyNOHyNdm82G8+fPS6s/fp/f78fRo0extbWFWq2G3t5ejI+Po729XRANYLe8o6urC0eOHEF/f78wwoPBIGKxmMDsJJsR7ioUCuJw6WicULaGzKhweO97KYrW66kdXA1zUzfokhYNLfKMk/cRCASwvb2NQqGAQ4cOyVS6RmOnEYXH44HL5cL8/DyOHz8uM4AXFhZQKpUQCoWQSCQQCATw0ksvYXx8HPfeey8eeOABNBoNPPfcc4jH402RvN1uR3t7O9bX1/Hyyy8jFosJc/zKlSsIh8MAIKkS3ivTG7rsFIAocSptfc/A3ppophEy/h/YRdN0HbRGWTRRjINvnn32WRkrubGxgYGBAVy7dg0jIyPIZrNwu92IxWKIRqMwGAzSdCadTmNsbAy9vb3IZDIyNvTGjRuC5t64cQNzc3PSWhhonppVrVaxvLyMq1evwmazIRKJoF7f6d3PNtMkpjIFyioezqNu/TydSqWN4B5gFH9bz/hOFkQ/bGB3IpUmxpA0oY0YPUun04loNAq73Y6JiQn8+q//OvL5vDQryOVyKJVKGBkZwdLSEur1nek1Bw8eRDgcliHYCwsLsFgsGBoagsGwU35TKBSwsrKCYDCIRCKBL33pS/jMZz4jnanoxTQaO/XYk5OTmJ+fx/b2NlZXV6WlXDqdlgEYtVpNpuNks1lpsk7FwM0GQPJrzFHoCHIv1T8SLSCrWedYeGC5GXX+lQc5nU5L56WpqSl873vfwz333AO/34/r169LlyeWOXCGb7FYRK1Ww/DwMKanp7G2toaNjQ3ZY8FgEO9973vRaDTw4IMP4tChQ1KfSvY+m+MbjUYcOXIEH/jAB9De3i59tZnTIvmHcCcHNaytraFSqQiprDWHpUXn3u+UGHI3CKF8rjONEBELttflujHiZqRaqVQwODiIZDKJTCaDtbU13LhxQ9Zia2sLbW1tmJ2dhclkEqN79OhRHDp0CFeuXMHk5CRqtZ3JcyzHOX78OKampmR+/MWLF/HCCy9gZmYGP/VTP4VGoyEOs91ux8jICGZmZgDsNL2YmZnB6uoqAEgZDg0NlfDm5qZ0GGtlKWuClHZYibbtJV0ANKc99Vmn0BFipErDzDPr9/sxMzODJ598UiJnn88Hj8eDw4cP4wtf+AJOnjyJarWKTCaD7u5uFAoFQUR+6qd+CkajEclkEpVKBW1tbfjzP/9zqQpJJBIyXKNcLiOZTDbxjIrFIp5//nlcvXpVppOxSqdYLMqscZvNJkTWYrGI48ePY3p6GpFIpKnXBe+T+pBBCp/NncDbwF/DQNMb1tGIZrHyIDIapQKNxWJYW1sDsDNbd2BgAJ/73OfQ09ODz372s1haWpIcII0kAHR3d+Nb3/oWLBYLXnnlFbjdbtxzzz3o6+uD2+2G1+vFyZMnceXKFeRyOZw9exbf+MY3cM899+DjH/847HY7tra2JJIfGhoST4g5h3feeUdgr2w2i2KxKAQDAIIERCIR2WBa6RL+0mxlGoK9FjlxrXUbUw1jA83sTSptq9WKlZUVzMzMoLe3F06nE52dnXjqqadQLpdx9epVOJ1OrK6uwmAwCIFEEzICgQDeeustvPLKK9LP3eFwIBgM4p/9s3+GdDoNo9GIWCyGc+fO4dq1a6jX6+jp6ZF1Y53rBz/4QZlSk8vlsLq6KtfNvJHOm9VqNTmojJ5albI2Vq3PinD7XhFNDNWRsx41S9Y+mz7wfRMTE8Lej0QiyGQycLlceOCBB1AsFrG9vY1wOAyz2YyRkREZG/v+978f3/jGNwQuPn78OA4dOgSj0SgkouvXr8NkMmFwcFBKM//rf/2vGB0dxcLCgqxxqVTC+Pg4vv/972N9fR0A8NM//dNwu91ixDc3N5tKanjPxWJRHEl9vvksdNdEOrFaWe8VIdlOI5NaN2r9QH3AoOvxxx/H9PS0ELImJiawsbEhKatKpSKpq3Q6LcMt2EM7kUjgPe95DxKJBDY3N+F2u1EsFvH1r39doOxsNovh4WEMDg7i6tWrCIfDyGazTanHz3/+8ygUChgbG0NHR4dwVsjiZxqDxpk6nd0JW/PsujpA8500mqJnU/wwueNOYowgGNbzcNKLYB6a+TtGlmazGSdPnkS5XIbL5cJbb72F1157DXNzc3j88celbq2/v1+o8KxJvffeexEMBhEIBDAwMCAEEJa4vPjii5Jvisfj+Nmf/VlcvHgRU1NT6O/vb2pUUi6Xce3aNUxOTuKjH/2o1DmynSdJXq3sX6Nxpw5al+m0sheZd+HrGEnsJdEHURODuA8YWfHg8ncsiThx4gSWlpbQ0dEBr9crrVzZJIKlMcz7cXYrsOMMzs3NYWxsDENDQzCZTOjv78eDDz6I559/Hjdu3MDm5iaeffZZXLhwATabDfPz8035RoPBgEOHDuG5557Dq6++CoPBgPb2dvk+3gMbkgC7iiiVSgmEeasIWeffCaPTMJtMJuFK7BXRThr3ArupsUMT9wL5I06nE6VSCd/5zndw7NgxnDhxAvPz82KE2QUqmUwiEomg0WjggQcewLVr15BKpTAwMICtrS0Ui0VMT09jZWVF2N3nzp1DX1+fdBq0WCw4ePAgvvKVr2BoaEgG3rCb4PXr13HmzBlpRMHcOPOErCQJBoOyJ6xWq+SqtTImSYiGiYqc0SKw2x9gr4iu3mDkTMdVpzKIpPH5bG9vI5vN4vTp08hkMrhy5QqMRiNOnTqFqakp4SWwdhmAlEnF43FxyukEUlefO3cOBw8ehNPpFKd8bGwMBw4cwPb2NpaWlgDsrBcnYjkcDnR1dcHr9cLtdgv/hY6lx+OB1+uVUq16vQ6/3y96ig4DcOu2rtqJoXEmifV25I4MNB8wN5nuosPf0Tuh58AWepw4c/78eRSLRZw4cQLnz5+XQ7y6uopIJIJ4PC7er8fjQSwWw9bWFt58803cuHEDs7Oz2Nraklpn9k8Oh8Nwu924cuUKLl26hM7OToyMjGBlZUWULTdNIBBArVbD//7f/1tqWdPpNADImDtCmCQN9ff3y/+5IWmk9ExRwuI85NyYe0laYRoN33JD8tmwHMVsNstM1lOnTuHd73431tbWMDU1JfWysVgMVqsVmUwG5XJZ2m+S8Tk1NYXZ2Vlcu3YN6+vrGBwcxIkTJ/Dcc89hdXUVS0tLSKfT2NzcxPHjx+F2uyWHSCcyGo3ixo0bAn2ZzWZpisEGJNlsFtlsVqIko9GIzc3Npj1E4f9pqBg1829+7tLSEhYWFv72FulHLDqPq/c+sEsAYokkAKysrMDn82F+fh6f+tSn8PDDD+Ppp5/Gv//3/15SDVevXoXBYMDc3BzC4TAWFhZw4MABLC4uShTOCNvj8WBkZAQ+nw/BYBDvvPMOenp6MDc3hw996ENYWVmBybTT8vX111+Hw+EQEpDZbMbGxgaGh4exuroqpL9vf/vbotjZgKS7u1v2cKPRQDqdRldXV1OXK+pB5p2phDVfg22PWcqzF0SXkNJZ0SkfXWbI/ZBKpXDmzBkph/3KV74ik8KWl5fR0dEhHCWn04lQKCSzvjs6OgRde+9734tIJCL9Cr773e9ibGwMfX196Ovrk26FiUQCa2truP/++5HNZiVgsNlsWFtbw+nTpxEKhRAMBvHGG2+gs7NTWloXCgUMDw9LhQ/3+cDAAFZWVgQ25xrrMwA0j94lysCW0rdroO+IsaCZeXpxeCFcGBrrRqOB2dlZLC4u4tSpUzh48CCuXLmC119/HbVaDR/72Mfw/PPPw+fzIRKJYGJiAkNDQ8LUY+4gn89jZGQEBw8elJm/q6uruHbtmng4hCb/xb/4FxgZGZGoiQtIg7G1tYVIJIK/9/f+HkwmE86dOyc9YNk/nCQvQnSZTEbm0WrPmOQVRguENcgE5Oi0vQRr0SvWjWj4PLTxogJnKc33vvc9vPe978X4+DgajQa++c1vwuPxoLOzE4lEAi6XC5FIRJ5hJBIRcp/FYsH29jampqakzK2/vx9nzpzBt771LclZssvcsWPHYLfbZYpVJpMRxZHNZoUNPDg4iJ6eHly8eBH9/f2ynqlUSmYJZzIZ1Ot1xGIxOZBsXsHnQdIiiYIApKlNOp2W8kD9vrtdqtWqGM1WCBPYddpaS/HuueceLC4uYnJyEoODg7jvvvvwzW9+E6dPn8bg4CC2t7cxNDSECxcuYGhoSJA47i8OyigUCigUChgdHcV3vvMdtLW1ScqDDS8qlQqOHDmCQqGATCYjDUUASCqElRqjo6PIZDKCniUSCeFZsF+D1WrFtWvXcOTIEalCIOStZ0AzT8nvIkTr9Xr3FIqi0xqtaR46bhr+dzqdGB8fx/T0NN566y1sbW3h6NGj0l8/mUxKyWwikZCzRcIxgyKOj+3p6YHH48Ff/uVf4t5774Xf70e1WkUgEJA1m5yclAloS0tLgqLWarUmkum1a9cQiUTEiWQf8O7ubmxtbSEcDksgMTAwgIsXLwqZTKMGvFeNMDLYDAQCQjC8XSTljiJoesaaGMFNqltdalLE4cOH8bM/+7N45JFH8JWvfAVLS0vSVOKhhx5CX1+fdH7a3NzE+973PikoByDlM6urq9jc3EQwGMTFixdx/vx5KeHid7JJxTvvvIPt7W3MzMxIFzJG0bXaziSimZkZXLt2TQhfpOfTsG5vb8PhcMBisQh71OPxSHTEzcO+vqTQ897ZaSoSicDv99/JY/6xFioqPQIQuDVBxGjcaRCxsbGBhx9+GG1tbZibm8P09DSy2Szm5+cxOjoqXeUAYH19HS6XSxwrOkUzMzPIZrPSgvHDH/4wvvOd7zSVcqTTacTjcckfGo1G4RRo2Jxs3Fwuh9nZ2ZvY1yQSsikKnTQShSjaO9YQHvPQKysrUofpcrkkmtwroglhusSKe4Lpnmq1inA4jNXVVSwvL2N0dBR9fX3SpOb++++XskRgB8Xq6emRWnY2EapWq1KfarPZcOjQIXzhC19AV1eXGEWOpQUg3Qc5wlBXV/C6DYadhiqJREJaz9JIFwoFOJ1OiX50GaHD4ZCf0fGi864RBPaUJ8S/lyo6DAZDU/9xvf7aMPNvckzW1tYkxzs6OipNSdhTnSWNrBhiFQXP1QMPPIC+vj5kMhn83//7fzE2Nib2wmg0wuVyIRQKwW63Sx8Lrhn5QYSq2auAffh1Tp09uVl2RTtHR56ExtZ7JYoI7JYjsi84CZK3G7TdkYHmjGadf9UQl46snE4nJiYmcPbsWfzhH/4h1tbW0N3djWq1ira2Nhw6dAh/8id/IjDYm2++CavVij/90z/F6uqqwKMulwtHjhzBmTNncOLECXz3u9/F3Nyc9NrlH3qoZOFpggohi3K5LKUzPp8PLpdLym+YmwJ2PR4OAne5XE11n6z51mxQEsWq1arUX3d0dAgTdK+JRk1aRaMMZO8zL+h2u2VCEUso2IrRbDZjYGAABoNBZi2T00CDe+LECbz3ve/FM888IyVv3INcA6IYVMwk6jDS1SV/jPppfHUVAkusqLx1tygqYUZM+t/AjsPHgRqt4w73gpAfwFw7oyUdUZBsxVKkaDSKxcVFnD9/XhizKysrAHbaK9LoZTIZ2O12fPnLX5Yxk1SqQ0NDcDgcGBgYwLlz59De3t5UZ089lM/nxVgQhuQ5dLvdqFQq0q2s9ewXi0UZV7i6uipGg6VgvCd+Jg2AzkNzn2iSGaPAvSK61A64dRtLcjJYlsTnsbKygt7eXoH+ScxMpVLSd2B2dhbvec97ZAAGU2WXLl2CzWbD9evXkUwmMTIyIo4C9x75DFxXEvt45vP5vOwDvoekYDoITFWQMEyS8cbGhpR/8hlQNIrI76aeY5CoA4If+ozvZEHIwGttcUivA4DkXu12O86cOYMjR46gt7cXN27cgNvtxszMDOLxOJxOJ7a2tgAA8/PzGB8fRzQaxdraGu69914ZP7a4uIhr164hHo/D5/Ph4MGDonQZQevyHrZ5o/LWMAxn+momNg+1ZpwTxmX03tbWJsqbZDE+aJNpZ14tI7lSqSSQPQ/mDzJkd6Nob7E1atZ/E1np6enB8PAw6vWdKVM8oDxYmUxGkA+/34/+/n6cP3++qcfx2toaisUiVldX8dBDD8mgDHq0bATBqIXGmagOWcX8Hh4+AELioaOmyy/y+TzK5TJWVlakWYHOtVER6H3GnPni4qI4HVTOe8lA87mTBAbsRlG8T64fdcTCwoLk/Pr6+mQqGetTTSYT2traEA6Hkc/nMTY2hmeffRZnzpxBo7EzIGNlZQV9fX3SXY5RKcsi6RS0ljPRUAM7pT+cbmSxWMTRIJ+EOeN8Po+5uTkcPnwYwI6x5ThLPUCBf5N3w/2Qz+dlLC51gEZg7nah/mxFlbSR0oxqlq95PB4h+JpMJnFmw+EwSqUSMpmMTKCKRCLo6uoSR+7hhx/GoUOHsLKygj/6oz/CE088IWgt9yLPIvWQHhkM7Eb+nIBHzhT3DteRTqjf74fJtNNjfXBwEPF4XJy61rXVUHetVkNbW5sECtQPjNhvR+7YQGvCEz1GGiJgd84rBxN8+ctfxtjYGGKxGF5++WXcd999OH78OC5evNgU6ZJl7XA4cPToUXzta18TZXjs2DF8+MMfxuXLl/Haa681QYXMRVER6q4tVL7aw+bC8T70w6OSYLMMKgC73Q6n0ymwWGtuhYqKnrpuF8mDu1eklaUKQHKvOnoyGo3SxefatWvo7+8Xo8k0wszMjMCK9G63t7elDSzLIlZXV5HP5/HRj34UDzzwAObm5ppgL2B3b2qnTTcQ0flhXb5FSAxonsyVTCal7V86nZZ6fnrZ+mDyfvlM1tfXJbLkvtBO7V4QPk9tnIFdncDUUjqdFmMLAG+//TYA4IUXXkAoFEJ3d7d0DzMYDFL6RqLQ7/7u7+J3f/d30dHRgY2NDVQqFWlswRxxKweGaBkjVh3pct24nrx+Gnm+xuFwyKQ7ll7FYrGmVAmw66jyu/k9bIxEVruG/feK0PGkQQMg6BOwa7Cq1aqM/2UHNjrLrNjhGF/+rF6vIxQKSYoDgCCqoVAIv/d7v4ePf/zjwhvQ7XjpZFFPE93Q+0GXcemySHIP6Gg7HA5hbTcaDakoYYviVuOsORlshpRKpQDspnx4jbcjd6QxWCqjoUHmcPgzo9EoQy0ajZ16sTfffFNylwAQi8XgcrnQ29sLk8kkjSHY2m95eRkPPvggCoUCOjs74ff7MTk5iXPnzsHpdDYxqDUpjQdBl/qwAb7FYkE+nxeYnpAJIVceSnaL6urqklGE7Byly4Y0Y5V/p9Npqc3WZRh7KffI3tJ85pTWjc6oiAxJHuSpqSmBC7UDs729jWQyie3tbeRyOTz44INoNHbqHdfX1xGNRvHzP//z+PznP39TFMLomJGzLnOiQdX9dKm8ef3RaFSgLL6fo+PYnYiKX6d2WuFKo3GnRzQhfN2jnSSivSQGg0HKCSna8eUZsNvtTeVQxWIRQ0NDwkNYX19Hf3+/RDNmsxmxWAwf+9jH8D//5/9EKBTC8vIyxsfHcerUKbz88svo6uoSZxjYjda5NnSQtQPB9bXZbHA4HMIsJ9Li8/mkvt5kMqGjowO1Wk0IpMxRUgdpXaNTfXTqOJyDzqOukd8rQj2rnU+djy4UCgiHw4jFYlIimc1mm/KyHC7DmQwkZJ04cQJTU1N47rnn4PP58NRTT2FlZQW/9Vu/he7ubiELageIwRlL4fi7anV3sAlfx5QK90itVsPKygoCgQA8Ho84X/39/VhaWsLBgwexsbEhZV3aOdNOCh20Q4cOIZFIyL7WKM/tOmp3nINmcl+TYhi6NxoNuN1urK+vS1cvMnPNZjOOHj0KYMdAc1E2Nzelm88bb7yBoaEhzM/P46233pLauP7+fjz33HMSpen8D7DbIICKt1QqieJnqUUikUAulxNYgkaGP9fkj62tLfH2Ojo6pJSCD1fDE3pzWiwWIYQRLt1rI+a0V6rzKfzDA8IhKRwawPabw8PDGB4elpKH9vZ2aQ7AevW///f/Pmq1Gq5cuYKvfvWrqNfr+Nmf/Vm8+OKLTfX2eo15LXpf6Gtm2oMHJRAISLMCj8eDZDIJAEIASafT8Hg8WF1dFQ9efw9fq5m79XpdjDMjB23EWmHXu1k0GYi13vpnTENFIhEZ0Vqv19HW1iZlbDTMQ0NDsNlsUmLZaDQwPz8vc9nJCTl48KCsIVMOPJsAmuayEx2hk2AymdDX14fp6WlRoPl8HplMpmn6GVMgADA8PIxkMon19XUhruoIUaNJrREUgKa9QUO9l0SjprfiH5hMJoRCIWxvb4vuTqVSyGazaDQaWF5eRjAYlPad5HjQ6SmVSnjhhRdw7NgxfOQjH0E6nZbc8/DwsEThmpynn7/mRdRqNeEKsZe6Lo/VRp1dBM1mM7LZLHw+HzKZDIaGhrC9vd30DFqrFqgb2d2OBDPurx/k3P8gueNhGSxFAtAURWmSzIkTJ+D3+7G2toZ8Po98Po9r165hbW1NeuaS9FOv16WLSygUwvPPPy+9eKvVnWklVqsV7e3t4g2xYQCVAw0DF5kMvGKxCJ/Ph1/91V/Fgw8+iGAwKLAYR5htb283sfZ0YwlCMzTmzFHTw6ZRIrGEAxq0EaHB2kvCFIbO4/JwctMPDAygVquJoWbN4uzsLJaXl2EymZBKpRCPx5FMJmWGr9lsRjAYxMTEBCYmJtDd3Y177rlH2kLqvJIWXd6kDyjJIGxEwFwz0xpdXV3iGDBCDofDcogJVwPNtZ63ck74TAi3aehtLxlnALK2rFjgM+JeYI0wlWoqlRJHqFqt4ujRozLXfWtrC5lMBvPz8xgZGYHD4UBbWxuuXr2KaDSK3t5ezM/PI5FI4OWXX5bIllPu+N10hPR5434hn+XYsWMyFY093wFI0xx2rLJYLOju7kYsFgMAScsAzdOZ+B1U8tRPNDQ8F9pZ2CtCdIRkK517ZjqI5Wg2mw1erxdmsxlbW1s4c+aMtHktFArS1rW7uxsTExNSasV89aFDh2AymfDlL38Zjz76qDxrtu/lmrQaTGBHN5DktbS0JGdSs/YNBgPeeecdIakyPTE9PQ2Px4N6vY5oNCoorP6OVseg0WggEolIlH2ryPl2eUl33KgEgBxKYDf/RoYdiRUGgwH9/f3o7OwEALznPe+R/rX0RJPJJDweD77+9a9Lx6jDhw8jlUrh3Llz+NCHPoT5+Xl87Wtfw9bWVlPEw8VgnlnDaZqsEo/H8aUvfQlra2uYnJxEOp0Wr5zEMXrYZrMZgUAAKysrcDgcaG9vv6mciFAtW4fyvkk64HXQYdER114RkmzopFB0HjidToviikQishYnT56UKCgej2Nrawvb29vSIMRms+H555/H22+/jUwmg+HhYdxzzz14+umnBcXgM+d3alakTnlQaWxsbODatWtYXFzE2tpaE9s7GAxKe1G+ntN1iMa01j3zPnW+Cdhl7AIQJa2fy17aByS7sPxM51gZsfDn1WoVJ0+exPLyMgDgvvvuw/e+9z3Mzc0hl8vh6tWriMfjePzxx/HMM8+IQWYjEp/P19Q4hHuLeoffS8dZM6qpsLe2ttDe3o6FhQW0tbWhp6cHPp+vqYaZZTgkKBJB2dzcFDhUG16df6TyJeNbk8gYUQLYU44a0SamBnWKi0aO+s9sNmN9fR2ZTAYDAwM4e/Ysenp6YLPZsLGxgc7OTtjtdhw5cgTz8/Pwer1IJBJ46KGHcOTIEXzlK1+B2WyW9rBEOjQpj3/0fABeD5Hf/v5++P1+BAIBLCwsiFPOs840GddzfX1dIG+W6lFPaN4Nv4d2ibXYtCutzsuPJILm5vL5fEIM42YF0KQoq9UqFhcXMT8/D7fbjaWlJenak0qlYLFYEI1GYbPZcPToUZRKJYyNjeHNN9/Egw8+iPe///3I5/Po6upCvV6XMY/8Ts3Y0zNmtSfHMipGNQ888IAoWKfTiaWlJWF88uGmUilUq1XEYjFRMGSiakhVz8Plc2E+jMqff+8l0SMWW50X3ZSCh4XTh2ZnZ/Hoo49KnXMkEpFyN5PJJJ7sysoKrl69ikgkgmPHjomCZqkSnUQiJfr560iXUBv3yMrKCtLpNAYHBwUl6ejoEPIKFTpLaRqNhhhu/b36MFJIOqNi1tCfhr/3Eoub9f961J5GslphvBs3bsBiseDYsWM4e/YshoaGYLHszHh3OByYnZ0V4s7Vq1cxNDSEWCyGnp4erK6u4sSJE1hdXZWyPbPZjO3tbfj9fnmumpyjnzd1RD6fF0OSzWbFyXI4HJibm2tidwM7qTgiQKyH1ux9Kmn9DOr1ujiveu25V/eSPqCDpHutU+go0WB5PB4YDAaEQiHE43Epr8pkMkKeOnbsGKampqRJ0Qc+8AGEQiF8/etfx9GjR/Ef/+N/xKOPPtpE7PP5fLIurUYQaIa5SfYdHBxEPp+XvtwAcPXqVfT29so60VFbWlpCe3s7jh8/juXl5SbWukYNW9MejNiJ6uh1v5N9cEcGmkYxn8/fNFaRX1ipVJBMJlEqlVAoFGRE2OrqqkStbNqwtLSEQCAg5Kq+vj784i/+IorFIkZGRmAwGPDWW2+hXC7LgSJJQ3swOhfE6JYHlLnNRCKB5eVlaVxhsViQTqdlUancvV4vNjc3YTAY0NPT08QS5kEntNdKiGF+Wuer99qhpEOmN52OEKiYeQA4MvDMmTO4ceMGgF1Hr6enRzpHNRoNvPHGG5iZmUFPTw9OnDgBAPjQhz6Ey5cvSxMYvj8QCMh3adY285Ma+mw0dhqPRKPRppGhnZ2dWF5eln3J+2J+s1AoSKMDHf1yv7U+AzqPdB50tHQnxJC7QagQufcZMbVCeLVaDdFoFMFgUMpnvF4vDIadEqf+/n709/djc3MTvb29WFhYgM/nw9jYGI4cOYKvfvWrOHr0KL797W9jbGxMIi5C0fxe3YuB+0DnFskJcTqdeOONN6RXQbFYFIfMarVKnwRCkyQ0Mi/Jva7Z2qwOAXY7yPEZ6bMB3Dz17G4WrmFrJ7FWY9TZ2YmlpSVEo1EAO9382traUKlUpGRye3sb3d3deOaZZ/DEE09gZGQE6XQar7zyCg4dOoTPfvazeP/73y8pB6apWDHQik4R3dR2gKlLVhjpVNna2pqURWrCH/uBR6NRmRPPe2zNufN7iRjoFKhOcdwuvA38NSJoGjxd18lIkXmhEydOIBQKwWaz4dFHH8Xrr78ufVTr9bpMlWEJxquvvopIJIKlpSVcvHgR6XQa3/nOd/D666+L8dT5X/7hgwAgA7VrtZ2RhgsLC+LlRyIRnDhxQqZg9fT0YHZ2VtrukcikITP92TzwvAbeL6FWlnnp8h0upPbq9oroyJVOC7CrkLa2tlAqlaS13cGDB6WpPHM/dKKy2SwikQiee+45KW+79957MTU1hTNnzuDZZ5+VxvjValVK2DQMxfVgVO/3+xEKheB0OuHz+WTuM/uz8/fRaFSMqS7RYimNjqZ43/y/hrmB3c5UVOLaedTEur0i2iDrnJ1WWLFYTMY/1mo1HDhwABMTE4JacE23t7dx6tQpfP3rX0d7ezvS6TQcDgfeeOMNdHd34+2338aJEycwPT2NarUq55atFlsjWDrgTqdT2nzSYUyn0xL1zs/Po6OjAxMTE7BYLNIS1mw2S2MNr9cre5f3yrWkMtdICZ8NgJuMBnD7uce7RTSEr8liwM5adHV14e2338bY2Jj0FOCI1+7ubsk1k+Q7ODgoOedLly5ha2sLnZ2dKBQK8Hq9cp70YCNt+DRrPx6PS2fBSqWCRCKBjY0N3LhxQ4JIXt/g4KA4W2xQks/npccFbR6wWxHCe9RkaQZp1EX6tfp1PzKSmG7J10p+ajQa6O3tlaiEXmswGBTGZHt7u8DLPp8PTqcTTz31FAqFAjo6OqRJxIEDB1CtVoV5zShH549Za8gyr1QqhYWFBSwtLcFgMGB7exvXr1/HpUuX8J3vfAef+cxn8N3vflemqVitVhQKBWn7yEPHqEB7QYyuNGubBodRJQ+wjrBay5HudiHBjs+B98pohTO9x8fHhWAViUSEb7C6uioOESHK1157Teqj77vvPiwvLzc1L6HCpUHQ8KJON/D6UqkU1tfXkc1mpWnI22+/jZmZGbzzzjuYnJzE8PAwFhcXAUBQEc6YZdojk8k0MZN1fgvYNVI0NNybwC7Er8vx9lIErXO8OoLm/6vVKkZGRnD27FmZQOXz+eBwOJBOp9Hf3w+Xy4XFxUWJZg4ePIhQKIRCoSDprRMnTsBoNEqLRKIn1D1Ugjyf1BUkenV0dKCnp0eQk/HxcfT09MhQBUaARPVIKOK5bWtrAwDhl2iDpCFsvp658FYOikb59pIw/ww0tzptNBoIBAJYX19HOBzGysoKotGoBGecvT4zM4P29nbUajUsLS3hkUceER3x+uuv4yMf+Qg++9nP4tChQ9jY2IDP54PRaJSpYnSkgeYJe0RDiGaazWa43W60t7dLR0MaeTpq5JtwrUulEvL5vDgGGq1tPc8aWtdlqHrPUDR/4YfJHe0Wbi428LBYLOI112o7k0rYaD4ej2Nqaqopr5PNZqXT19jYmAxjJ6OStPTHH38cN27ckBwnWXc0BOzwREadyWQSdm53dzfGx8cRiUQQiUQwPj6OM2fO4P7778fDDz+MoaEh5HI5GcTADUbSGw+sJjdotID3qiNs/Xy0gW7tsrMXRMN/2ltmNF0oFHDPPffg2rVrCIVC2NjYwNraGmq1Gubn56XEJh6PyxSxeDyO9vZ2dHd3w2KxYG1tDR/96EexsbEBAOIB6/yOx+MR1IJOlMlkEiIHfw/sKJGDBw+iv78f0WgUR48eFVY5SSJcU66r3gM/CJail9yaFwcgBBk+s702D5oQ463KnVjNQBJOtVrF1taWkII4gz2dTuPo0aOo1WqIx+MCWXNi1aFDh/Dqq68CgBBv6JSxfS+Ng2ZMk9THXuosralWq5ibm8PMzIyQgkhcI6ufKZJkMolcLic92YFdljh1kWb1A81EUqAZceFr9hKKwmiRPbM1gsBe1fF4HAcOHMDGxgaOHz+OxcVFtLW1SUAG7DaJ6enpQaPRQDKZxPnz5/GpT30Kv/u7v4vTp0+LruH6830sraXR49/VahUul0v2pO42tr29LbOcX3jhBYyPj4teIfJH20D9wCBAB2hcW51rp+HWuWftlNHJ/JEYaOafmfzWZTaseWP0Ojo6KsabDUH8fj/S6bSUN3k8HnR3d8vnm0wmjIyM4Itf/KJ8NutWaaTJjqVDwLIvelE04lxEDsZg57LZ2VnMzMzs3Lw6YOwWBEBKRDRkp1sBagKSPnD0nPiZvJa9dCiZmmDkSMPIZxQOh7G+vg673Y729nYkEgkcOHBAxr9xAAFRj2Qyic7OTok8isUifvEXfxEvv/xyU36ZDH46hTTCzCfrQ8DuXZVKRVo2crTl8PAwjh07hvn5eYHYWQbBXt3MbbGWt5V4AqDp4NFx0HAWALlug8EgiM9ekUajIbN52ZGJ98d60o6ODmxtbQmJ6MqVK1JqeenSJfT09GBqakoUKfeHz+fDsWPH8M1vfhOhUEiMIhEJwueMvNkyVPNEyMRmnSwbZvT29qKvrw9+vx/z8/MAIKWaRHXcbrdEXT6fr4lsBuzqjVZonXqHOWgNg9KR3UtOGiFiGjvNPyAHKRgM4urVq+LwRCIRADtnNBQKCWrK8Z1ra2vo7e2Fw+HAb/3Wb+G+++6Tvhd87poMXK/XkclkkMvlxFkHIGeOncg00sYBGS+//DKi0ag00KHjp/lN/EyNGOp0h46oAYh9YApOO+zcU3dS0XFHBpoMZ4/HI1ASL4weBGeexmIxGSHGxDyV9uTkJABgaGgI169fB7DTFMDj8eDpp5++iZVJ4g5LO+ih+P1+JJPJplwEYe9arQaPxwOr1Qqv14uBgYGm8gez2SyGmGQBHm5NgNMQJj1sbbRb4VYNaRCKb42072Yh9EPDqPeAZkzabDa4XC4Eg0HJ+zGn19XVJf1tWT5TLpcxPj6On/mZnxHuAQ0tnzX3GGFIGge2iOUho6FPp9OyP4xGIwKBgPR7LxQKYpSpVPlerdhbST4UnXNjNKkVM5UVlbjupLcXhNEqo0XtFNNhKhaLwtbf2NgQZ4W17uFwGHNzc0in0zh16hSmpqZw+PBhrKys4MKFC/D7/TIxiF3CWLdMg0hHr7e3F2tra8jlcrJHgZ11Yj+EQqEgrRzb2tqEDU4EjRPUyMJmsxruceYmNSqmnUPd6le3EiUqcyeR090gmiRM0VA+DR7L10gA5HPlLAbWl5NVnUql8MILL+DAgQOS52+thtCk5Hp9p95ek7ioJ8hF4B9eG3tzsL0zU2y1Wg1ut1vKgblniRbRQGuirLZ9dPR47mlfNPlMO2w/TO7IQNvtdvFGqIRpQDVhh4St7u5u6WdaKBSkBpkjxjjtB9gp3frWt74lG5leE2EMXf/KJiSFQgHRaBTFYhHr6+vS8UezLulJ8xoJt+roWROD6HHRkJPIBjTDUzoPxab4OtqiAtFlN3tByHbVEWNrBEtHij228/m8wI46h0cSRi6Xg9vtxuDgIJ5//nlhZvL7dMkMvVNGS06nU/rl+nw+6RhGpe5yueR6XS4XXC6XNMsgRKeHsTcaDdk/jNZZG6mjYx44viebzTahClxzdswLBoMYHx//21qmH7lwvnUulwMAMUjAbhRB4hyw2+Sos7MTxWIR8XgcVqsVp0+fxvz8PF555RUxxh/72MeayHV0clsbxlAP0dCSiJTJZKSUh7nreDwuTgOhdELknPNNXUBjTEIs9YN+L69JQ6usDOHoQ52n5P7y+Xx/q+v0oxSz2YxUKtXUflkbbOpcPUzI4/FIcxKLxYLl5WXU6zsd5hqNhrRbzefziEQiovOZd2YARvujz+r6+jpSqZQYTx116/p12qz+/v6mSg8KHXbqLxpVpkGA5iEprVyTSCQiZWTcU7Rf9Xod8XgcExMTt/WM75ixwOhCs9E0izGfz2NrawuJRAL5fB4ul0vmodbrdSwuLko3r/b2dmlI/2u/9ms4cOAA7HZ70wEn/ERPneUvhK3W19cxMDCAJ598Ek888QSGhobQ2dkpA7KtViv6+voQjUYRjUYxPj4uh47eOCNBnUfjPXm9XvldqVRCLpeT/BXzX+xABuw2Tef76/U6EonEnT7mH1sxmUxSeqQhff7ObDZjc3MTk5OTMjqOpXFc01gsJs4ND53NZsOVK1ek73Vr/oZGlgqfTiCNo65ZZXSnI1ySSubn50UZayNLZa9Z3VarFaFQCF6vF06ns6n7lPaALZad6Vws76Jzxlz5sWPH8NRTT+GjH/3o3+5i/QiFZ5RnHmhmqZZKJayvr+PjH/84RkZGZCAJz0c0GsXLL7+MQ4cOSXRCRfZHf/RH+Omf/ummEi7+YZTE3vrME5LhPTg4iJGRERw6dAihUEiiIIfDgXw+L4qaHcqMxt3hCtwPrLUmoS0YDDbVfTMNQhIhoUtC7yzb4h9gJ7hpa2vD66+//newWj8aIWLAs6gjQxqkd7/73cjlclLDTiemUqlIw5DLly+L8xyNRmE2m9HV1SVOFXUEU1DkiGhyKnPHwC6fQBtq3T+DTh1TGVqvOJ3OpsEpJKZms1npOgns5rRbbWCtVpNmOhSicqlUCi+99BLOnj2L9fX123rGd1wHzQiTbQ01LJhMJnH27Fl89KMfxaFDh6TucX19XWAjljtwCMbGxgZSqRR++Zd/WeBuYvTMN3ORaDw5C5qR1sLCAi5cuIDLly9jfn4eW1tbUlJhMOxMKJmdncUf/MEf4OzZs/gH/+Af4MEHH2waHwhACGGcmsJNyMlWjIaq1aoMhidZjUZLR1lOpxMbGxu3vRh3g3Cj61IZTaRKpVK4dOkSHn30UYlMOKKUrfXIriU7e2hoCC6XS5R+JpNp6sury3ioxKnQc7kctra2kMvlkM/nxTsmSsKyrFqths3NTbzwwgs4e/Ysbty40ZR+4IGm0WdUz0MKQIy0x+MRRjKdTfbspbE3mUxwuVw4evQoHnvsMdRqNXznO9/5u1y6v1EhC9flcsn66gja7XYjGo3iz//8z5FIJLC9vS2d+yKRCA4cOID19XVcvXoVDzzwACKRCLxeL775zW/i05/+NH7zN38TH/nIR8QRzOVysNlsAkkCu1ERHTzW3G9vb0sag84SHa98Po9isYjnnnsORqMRP/MzP4N7771Xco8aui+VSk2jBaPRqMz3Ji8mm82KLmCJp44m6YT6fD5MTk6iq6vr73LZ/kaFejCbzd7EbgeABx98EF/84hfR19cnCBafY6VSwdzcHA4dOiQ9LoiylMtlHDlyBMFgEMvLy+JcETqmaPSC0+/YupdwOIU5caZXurq68OCDD0p5HVOxwC4yw4CN+5aIEasMWNGknRKDwSDpEjqsjcbOIKXz589jcXERQ0NDGBoauq1nfEcGur29Xfqu6o5SvLF6vY5f/dVfxfLyMq5fvy5RMrATAUUiEYEfL168iMnJSVgsFgSDQczOzqLRaIinScVKeJHEILPZLI0HcrkcAoGA9MumF0ejW6/vlF6RjfmTP/mT+MQnPoFnn30Wf/qnf4qPfexjMuuV5RpkgdIBYR0lF9BoNIqhttvt0i6QcL9+JouLizAYDMIU3QvCWlEaWnqRNGSVSgUf/vCHsba2huHhYZTLZXR0dAjbsre3V5QkCUR04srlssxR5oB1Dj7hQQGa20mS0EVUhMI8NbDbPIepDqNxZ1rVyy+/jMnJySYokt4w92kqlRJlTXKK2WyGy+WC1+sVz5/lGfTYzWYzDh8+jPvuu08IaozC9oJEo1HpKEhDyJy+y+VCPB7H2toafuInfgJvv/22nGuDwYDV1VWEQiEpc0smk1LqxJGDTz31FH7lV34F9913n/ADqPh1e05GOySiMSepUxIa6iR3wmKxYHFxEZ/5zGcwPT2Nf/Nv/g36+vokyqOCJuyuy3lIHmOHO5b2eb1euSfN5g2Hwzh37hzcbrfA33tB6JBkMpmbUlgejwcvvvgiTp8+jeeff14M59bWlvTE+M53voNwOIxgMAiTyYSzZ89idXUVHR0dYuRCoZAM2mkta2J0qx15RrH8HfeE5hBxfUOhEI4fP45jx47BarVKaSaNviaWEQXWUDlnV9NgM2Agaqi//8UXX0Q+n8fJkyfhdruFI/HD5I7roDlGb3NzswmGpEGlMksmk5IHOnLkiERGW1tbKBaL6O3txebmptwQBxo0Gg3Mzs4K+YNeFZU6HzTz0lpp6xZ7ummEvn4AOHz4MH7hF34B3/jGN7CysoLTp08jl8sJSYRlA/X6Tns4q9UKq9UqJUGM8NbW1iQHo721YDCIZDKJra0tXLhwoamX890u7MDEdWFEzZwvN+Pw8DBu3LghKY/l5WXE43HpUZvNZjE4OAi/399UvtLW1oZEIoFoNCqbmCUy/C6uER0EPns6Cq2HldGTPpy8j62tLVy/fl3YwzSumgXMCJrefaFQQC6XQ7FYRCwWk5ppMpmNxp0RdY899hguX76M69evS85srwjJoO3t7U1s/lptZ4hGLBbDT/zET2Bqagperxd2ux3Xr1/Hvffei87OTszPz+PRRx+VaWFs8WowGLC5uYkXX3wR/+7f/Tu8+eabkvYiqgLs9mAgVwXYHdBAVI+GmQ6d/jew2wkrmUziX/2rf4UPfvCDeP/734/19XU0Gg34/X5kMhlkMhnRc4TDCe+63W5EIhHhO3DQClGU7u5uPP/88+jv72/iVuwFsdlsyOVyMvdd52Wz2SzGx8eRz+exvr4Oq9UqkwE5ipioV1dXFyYmJjA0NASj0YhwOIzr16/D6/Wit7dXInSy8/kZXGNWEZDvo+FnvRdIGtMEZ6NxZzzyAw88gKNHj2J2dhZra2vi4BEdIamVARptgubiMCXCvu383euvvw6jcWcyoy4HvR2542EZHCLBB8YQnt5EOp0WxnQoFMI999wDi8WCnp4eFItFvPLKK/I5Xq9XSB2hUEhu7IEHHpBOZFwUGnEqZaPRiLa2NolaqYhJTtINJXhgqEB5oE+fPo3e3l5MT08jEolIxyBuMCpqGgaS5MrlsjCTW2FtXuvVq1cxPz+PQCAgPbr3grCsjeugGY3VahWzs7NSLgNADtDo6CgOHz7c1Et5aGgI1WpV1jkYDKJUKmFrawuXLl1CIpGQtAOfNfPKJBkySqECpXNFY8H1JhKj2cbsJkW2eDqdlpI7AKKYSRohq5+fy5F1hUJBynyAnT4Bp0+fxssvvyxtY/P5/G2XVtwNsr6+Lo1kdJWDZrC/9NJL6O/vR7FYhN1ux8jICM6dO4eZmRlpWrG8vCzQ5OjoKBYWFmA2m/HzP//z+JM/+RMkEgk88cQT4mzp3KNOq3R3d8sMb6beNMNWE7p4vZrY09bWht/+7d/GI488gtHRURmGo0mwreWV3PNsB5nP5xGNRuW62Mmso6NDyuz2kpMWDodFfwO7JWQ8B/fffz/efPNNMbzJZBJ9fX3C0O/r65PpZJubmxLo0JF2u93I5/MIh8OSvgKaSxyZ/iRaqvlFXHsaZ6Jw5XK5iTjMzzt48CAeeughvP7664IGkdnNz9MpLM6gp8Gu1WoYHByU8bm1Wg2JRAKvvvoqHnroIZmIpm3QD5M70hgkyVBpauYmmXgf//jHMT09LWVQ586dkxxwOBzGkSNHAOyUbAWDQbjdbiH5OJ1ORCIRbG5uorOzU6Jx3ZmFhfFsvs5IulAoSDs27WlrKIyePg0KN9nw8DCMRiO8Xm8T0YWHXY+gpKMQCARkKldnZ6ccXnbH2d7eRjQalXzVXhHmXhiJ8jlqpXT48GEMDg6iVCohEonIXllZWZHcLQ0cIWqtPHO5HAYGBjA+Pn4Tc5cOEfM/LIMxGAxIp9Oy3ppRysNJEpB2qEwmE2ZmZoQgpssh+L1U8Pw8Ek34fhoVese9vb2w2+3Y3NyEzWZDoVCAy+XaUyU2mUxGcs08o3zmdEgGBwcxOTmJer0Ol8uFWCwGv98Pj8cDAEgkEvjUpz6FixcvolarSVpkdnYWX/va1/DpT39a0mLcd4xKgJ08ODsNdnR0wGg0Ip1OC6tYk3j4Hl2Cw/WiRCIR/NIv/RJ+7dd+DY1GQ2aE6xpsDZvz5/wepmWoC9i0iQ76rcqS7maJRCKSzmit5DCZTJibm5PSSpPJhLW1NWQyGSmfW11dxYEDB7C0tCQNonSKyO12w+FwYGhoCNPT04J4kIPCM8oxspyKx3SHRkp0DTMhb50v598kGJ4/fx7lclmqPnSFDteb90mH0WAwYHh4WBosVatVXL9+HYODg1LeR3IjOU4/TO7IQBeLRQSDQYRCIblpjdMDwDPPPIPl5WXJT9TrdczPz2NgYECK0Xm4NzY2ZBZwOp1GOBxGOBzG4uIivF6vEC6KxSKy2awQQmq1Gvx+v0AlhKMIs9Cz1b1T+V7tGVPZB4NBdHd3Y2RkRJqmt0ZbvEftRbOVXFtbm+TAy+WyTO1ityOtBO52IZTLDaZLihqNhrTti8fjACDPaGVlRaBKrl8mk5GccDgcRiaTwcbGBt71rnfhueeew9zcHIrFIlKpFABI/tdisUhbThJIcrlck3LUB1R72zS4+sABO9NsmHNcW1trIgxpJczPIXS2sbGByclJeDwemEw7XfHGxsawuLgIh8OBTCYjDVVuF9a6G4TpKxpjHWU2Gg08+eSTWFhYEKeJzY2AHcW+vLyMY8eOCfy7vLwsFR9Ep+bm5jA5OYm2tjZJpxBiZO0rOz4xauZ+AHYnKulcpW44QeH+Jarz8Y9/HP/hP/wHuFwu9Pb2NuWwqYiplBnBNxoN4dOQS6OfDffM7eYe7wbx+/1wuVwS/VL4fBKJhIwj5XCTWq0mDWpIIiWKRiO7traG9vZ2ZDIZvPXWW3j/+9+Pqakp6X2gDSWRN5vNJkQuBg5cH66d7tmgRaOgtVoN9913H8LhMC5evIjx8XGsra3J+7Q+oC4h5N7f39/UNCedTuPGjRs4ceKERNlLS0vw+/1Nuuevkjsy0GSsEZrW7dNI3kqn00ilUtLrtF6vY21tTQhYbMXY09MjryOUXKlUcO3aNRw8eFC6FNHbofEzmXbbejK3yO8heY0lNFSi9Oxv1aKPBqbRaKC/vx/5fF6gVg2n6fZ9RqMRW1tbqNfrQn4Adli+sVhMiAZ0IPbSoeSzamWtayOdz+fFi+RarK6uyhowVdHT0yORDIcTDA4OSi0pc1gWi0WIQq0GgdAx83uaWa6jJL3m+v/a+718+TKGhoawvb19U+TF1+j3Ly0tYXZ2VkpyjMadyT02m03II1Q8JNftFWk0Gk0Oh45EDAYD3n77bSH/kODZaOyM/eQko1KphGg0iu3tbakPJpxJ5mwgEMD3vvc9IeMRjeFZZ1TldrvFEaCjrh0GomHcFzqiBppbhXZ2dsqe45q3lhAxEgd2Zk3TMPN3brcb29vbonu0vtorQrTpVl0FAUgpGgApuQSAgwcPNqFLTE1wrzC1xCCMc+G5/iyxZZRN6FiXOlGva06KRtVoZLUjT2k0GvjoRz+KN998E21tbU1Mbf1+/dknT55EvV6XfgiMnoeGhoRwlk6npd/G7fbG+GvNg2YrP33BwI6BamtrQygUEo83kUigvb0dPp8PZrMZoVAIADAxMSGQORVyLBYTqOqP//iP0dbWJixpfRhIW9fQhe7ywz800LoxCdnGzEPpg+pwOCTXSEhUe70a1rLZbJKL56Zk/pTDINgEYy8pZmC3ro/PENj1mk0mkwwzYNP5RCKBbDaLnp4eeR6MlKrVqrRTZKqDI0rn5ubgcDgQDoclD8TonKRBAJJ3boWvW6FpzdTWf1Mxc4gKI3vmw6iI9H3WajUZ5s7I0OFwoKenB/F4XPYcIzq9B/eCsPZfKzetIEOhEBqNnXnghKEzmQxOnjyJq1evoru7W5pAMBK/cOECkskkxsfHYbPZMD8/j/e973349re/DY/HIwqVZ7BYLMroSpLNaLz1elEhagYusKsLNL+BqMx/+2//TZwJzQjXjjo/g84GnUnCr+Rq6D2zl9A0GhuguYWpTgXx7PL5d3V14cqVK1LvTBTEaDSKw8a1Xl9fx9GjR/HHf/zH6O7ullGTJAR7vV5Eo1HUajVks1npWUGHSkfPrcZZc5TozPF1RPhMpp321SS/UidQ19Ae2mw2+P3+Jt5NKpXC5uYmwuFwE3nS4/HIZ92O3JHGSCaTTUqQm5/Yui574UWWSiWcPn0aExMTyOVyiMViTQXuunwlEAigWq3i/PnzmJiYQDKZRCgUQjgcFkXn8XjQ3t4Or9crHhSwy/QlaYALw/ppelD0oDXhQHtSbGRC0T1f9c/YQpQQLksIaHSYK2eru70iulEInSA6MsCOE7exsSGICAcfPProo8J4pkJjtFMqlZBOp6VRQTKZxJNPPomlpSWYzWbp30zSRiKRkDpczezU68TromfPP615JINhdxY4AExPT2NtbU06kPHgMmLnewnXkaFvNpvR3t4ugxYYNdHB4/XuFdGMdl1eRslms/D7/ZIKYs309evX5bysrKyIMi4UCnjPe96Db3/72/D5fHLmP/OZzyAYDErrRZZt0jkzGo2C5rU2GQJ24UutUIFdA0unnkqUP79w4QLuvfde2Z+t0SE/V+8xTRLc2tqS9xLd4b/3ipDMBexGlHofUC8Q7XC73djc3JRaaA0XsywykUhgZGREUmQ8T4cPH0Y2m5WImc4fiXi5XA7BYLAJsaADxutrzTnz39pA6/V95JFHJMjT0prT9nq90qaURLSZmRmZM2EwGKRaic5F62f+ILljl56lTvSUq9Uq0um0jG1k9MT+tkeOHMHFixeFHMKhFG63Wx6o3W5HV1cXlpeXsbS0hIMHD+K3f/u38Y/+0T9CLpfD/Py8MGx196/19XV5UITBWZ/Gw0hvWsNTmmLP15H6zsb9fJC8Pk0GoHJPpVJCmGLbOx2xMdLeSxC3LmXQKAq9wnw+j1Ao1OSw9fb2SsMHQsFEKFjvuL29jdHRUQA7vXkXFhak+5vD4UAwGES5XJZey4zCNRuXDhqw6zFznbWBJiyqmcA8dDdu3EA4HG5q0K//plSrVZnaVK/vNDhpa2tDJpMRB5E9hunQ7aV9wNQNKyeA5lwezyvXJ5VKwe/3w+v1Sjc+DVWzguL+++/H1tYWlpaW8OSTT+Kpp55CMBiEx+PB7Oxs04QznsVAIICZmRmJUGl0teOoc4jct4S6dUTNvTA0NITjx483IQXaCeAepHNIwiPzrUTUgJ0zoyO8vSIMzlp7FFC4rqwVByCsb7fbLakBpjHz+TysVis2NzeRyWTQ1taGxcVFceqZpuCakSBqMpmwvb0tncTojNFIA7vr34qg0KFoNd71eh2PPfbYTa2h+fkalSNplQ5bLpeTtA2dv0wmA5/PJ87e7fblvyMDzTZ7ACRZT6PJG2C0A+wYw+XlZZjNZik2J06/srIixnRubg5vvPGGlFbl83m88cYb+PM//3Mkk0mMjY2JYaaRZ52qx+MRZcv8MxeB0IquT6VjQWhKwxWNRkNg91QqhWw2K/CcXkSr1Yre3l6BXckeZR01o2fC7rciJtzNQsPLNaHHyigjk8lILqZWq0mOlnkjGvN4PC4Hq6urC4lEQjzNVCqFRCIhip7PkS30SMjhMwfQZEw1pKXz41QqlFZYnMxdi8UijTj0QdalPt3d3ZI7JS+Ch5lOSDKZFEO9l4RQLwldvGdtCDOZDKLRKAqFAiKRiPTKt9vtSKfT6OvrAwBB2fr6+hCLxbCysoLHH38cv/Irv4JSqSRdCTnZip9BOFl3muKZZ59mXhMNKg0IjbPmUOi9wLpmlgCl0+km1IiIXaFQwNGjRzE/P980aY/OPyNIOnx8bntB6AD5fL6mclc+c7vdLikOr9cr3ALqRY/HI05zOBxGMplEKpWS55ZIJNDT0yM1z4Sy2fzFarVKNUcymZQeHTSaOn2hkRQdrGk9oVnfDCIvX74sfCMGYnwNsLNPstkspqenhXORy+XEOeWas/Miy3c1avdXyR0ZaHqTnLWrjSHJWhaLRVjQQ0NDTR4ysAP/ENun4bzvvvvwyCOP4Nq1a+jt7cXBgwdx6tQpicg4Wo41btVqFSsrK/KgGBERjqYnoxtU0AumktUKl9JoNDA6OopKpSLTT0wmE1KpFFKplGwwennz8/PCLCYpKJFIyHVz0fcSMYTPjakMTbQDdhnTPp8P1WoV/f39SKfTMu0pm83KHnA6nRgYGEAikUCpVEJ7ezt6e3tRLBYxNDQkhf3sKEajz97K5XIZKysrTc/XYDA0kftolHlt2mFqzQnzc9rb2zE4OChVBeFwWBirdL6Mxp0GB0tLS1LdwI5iAKTnMB045iP3imgCHslZGiLu6+vDysqKOPDJZLIpJcWJd8z337hxAy+//DLe+9734sCBAzh//jw+8pGPYHZ2FufPn5c1o7Gj8iXyxcYRXG8ay9ZIWnNTtA4jxK2Z//Pz85JWI+EpFothe3tb5snTABEdIPGpWq0im81Ke0o6BHup5JKGRndh1Agju7wxlZnP5xEIBKSVcjabFbSTQy9IFPb7/UilUrDZbJidnRUDCUDKr2gEM5kMenp6hNinna1W7giwGyFTbpUP5muvX7/e5KBpzg1roEulEra3t2U41Pr6ujiNwWBQRpiyXt5oNEplyg+TOzLQlUpFpjvp1pZa1tbWxEvheDGWG9ntdoGmcrkcFhYWsLS0hFKphMnJSQwMDODGjRuYmJjA9PS0KHLmulwuFwwGAwKBgEQmnOWpDxY3jYa3CFHpiJmKXJdBkHXLUWh0SHjtm5ubWFlZQV9fn9D+GSnRc6ICZ450L0XQRqNR+g8zagZ2R4OS4c+RlOzYxYYWPISMdJeWlmRWL+vHWTs+Pz8v35HNZhGPx8X77u3tFbIac7u6DIp/04ng53D9ibqQn6CNtdlslg5ChNIikQi6u7vR2dmJYDAoThhH53V3dwv6QieO8B7ZwHspkibUyIgGQJPSIzrGc0guxvb2tjQ1Yi+Dvr4+tLe3y/Cbzs5ODA0NySjSYDAoUQ0jFpLD3G63pNDomNPgAs3KV0dL1BkaVQOac5UvvfSStGbs6OiQXtw87zMzM3jnnXfQ09Mj553TmuLxuNTCM3oiurZXhHqRBhrYLUNl3XM4HBbnmmVVdLJ4rl0uF+r1OtxuN8bGxlCv1xGLxdDb24uLFy/KzGlWbLDFKp3+8+fP48yZMzItkc4WUx0AJCjT68z9QHi+NZgidM0znEwmpa88x9iyNJDOw/b2NlZWVlCpVODxeKT3PAcmsfkKkbofJndkoImp03NikxBucpvNhr6+PmGsUUHzIGQyGTGUFosFg4ODOHz4sBSwm0wmKWLnXFDm+kgICIfDuHHjBgqFApxOpyhRfdCYH6OweYkmDHAxtMEmdBsKhVAoFLC1tYWtrS2BXex2O4LBIPr7+7G9vS39mLe3txGLxeQamGfgNewlchCRELLdddQB7DAadW9yknv06wkFGww7PXv5WkbjbW1tSCaTEm0QSsxms7DZbOjo6EAoFMK5c+ekCxkjJx46Ijut+SUaYx4o/l/Dn+vr67K/SXSjcQgEAvB6vZifn5dD1tnZiYGBAYmaOXieOTFGB3tpHvTIyEhTsxit2EicCQQCWFhYQKVSQSQSkZpYAFJO43a7sby8jK2tLYRCITz33HOo1+t45ZVX0N3dDa/X25RvJnxdr+8MbJmenkZvb29TX3gKo2Q6g63NdQiD6nXReyaTyeDAgQPCe+B0so6ODvT29sLhcCAajQLYYTTTAG1sbMDj8Qh/hfwc6s69InTA2LlRM6eNRqM4sGxUQl4PzypJZoyuk8kk7HY7Dhw4IKhbIpGA1+sVnWEymYTAWSwWMTMzg7GxMbhcLqkGYOClK0u4prQPramNWyGqAHDixAlsbm5K3wc6l+vr61I62tvbK5D84uIitra2pC+CyWSSoIWMfvbtvh25IwPNiFbj+cBu4p2LtLm5KdEjaxLr9brUsYVCIZRKJYlUOzo6cODAAVy/fl1yO3wf38thBCsrKyiXy/D5fELS0bCKhl2phPXvSXACmokuNDLVahWBQEBmC7NUKh6PY2lpSWDt9vZ2BINBrK+vY2trC/F4XIwX25Yy/7qXyEGElrPZbNPAFBL1CoVC05B7wtM670IGNw+mw+HAxsYGBgYGJCLWjOtsNisdmfr7+zE6OopnnnlGytmImmg2PmH3Vha3VtQaptVQ2MLCgnj13IN08nK5HJaWlqQRxYEDB/DhD38YPp9POiLR2HDoCgBxJveKsIc6S0c0QQ+AOKh83rrnPok9W1tb0kmws7MT4XAYZ86cwaVLl2CxWHD58uUmJ43KlFOtXnvtNUSjUYG8ATSdc92OllEVu0wBu/lSHXUBu0hAZ2cnFhYWYDQaxRlh1Ga323HvvfdienoaRqMR3d3dWFlZweTkJLxer0Reeg8wR7pXZHV1FRaLBdFoFLlcTgwQAxWv14t4PC6pQqYOa7Wdzo9k+DNNYjabZRwsSy9pyOhshUIhQTDeeecdhMNhPPDAA1haWpJUgg7AdAqDeoF2Sjtjt0JQAODQoUOYmpqStSTbnFG1w+GQcr9XX30VL774Ig4fPixOq8/nw40bN2SwEjk0PxIDXa/vtkP0er2ywQkpW61WbGxsNOUk6vU6gsGgNDApFArSmm19fR1tbW1YWVnBpUuX8KEPfUgK29kg3WQyYWNjA+VyGadOnRLlSMiUh1E/cP6cioAHtJXh26qcuTCMAHloSRghTLq1tdU0Xm9mZkbyYLo5C5V7a67zbhZOsSKMq4V5ZZbXaMNmNBqlPIIwNp02l8uFtrY23LhxowkiowNVLBbh8/nQ1dWF/v5+fP7znxdPmqIPGP9Pp0GzMImacN25pvpPvb4zVJ17kHBtNptFKpUS6Kuvrw8f+MAHYDAYcO3aNdTrdSkxYw6No0iBm3Ped7NwfB+AJiecz59jKEmWJJOZkTRJg+FwWCLPYDCIlZUVrK+v493vfjd8Pp8ocgCSO56amkIikcCJEyea+A+ahc08NB0xonZU4jz7re+hUME///zzuOeee1Cv1+V+maIDINP9bty4IYxy3gvhdk0S20soCvthtFYo0MHRrZr5f+Zs/X4/yuUyotEoMpkMwuGwcFGAnbOyvb3d1HTG4XDAarU2Vfv09/djY2ND1otnnftR631NENPXqQlut5JgMCg5YyIG7IPBwOHFF1/E1atXcezYMWm+5Xa7hSvl9/uluyWHb9yO3HEEzaQ8c6/aA+GDZHN8hvLM9Y2MjKBWq0mHnfb2dok6nE4nnn76aYTDYQC79cdDQ0P44Ac/iGw2i7/8y79Ee3u7lDHovLIuowFw08Lo8ioa89bImZ9Tq9XQ398vZVb8THpB/f39CIVCiMViuHz5snTHstvtUpenSWZ7STiLW3dF0jl2r9crZSfMPbLFqybV8A8jUc5PZVOAarUq8DIbAUxPT+OrX/2qcA8odIK0965/Rmatdpi046SVM3++vr6O7u7uJqi7VCohlUrBbrfj8OHDePe7343NzU1cuXJFctIOhwMdHR2oVqtSC0vG6l6aZFSv16XkktwQ7fjU63Vp3EGYl6kMOvilUgnLy8uiF775zW/ixIkTOHToEC5dutREKmPU09bWhuPHj0tjCEa/XHOeZx196bQH11qfa8Ki1FNaia+uruLYsWOiXMlCNhh2Wlm63W5cvXoVgUBAUhrUi5wVr9nKP8gI3I1CRIJtNlvP38LCAgKBQBNLmsgi2/VGo1FxxPW8BDptREH4TNl98uDBg01NcLRz2Pq3JonqaFo7ldrJ0+sPAEeOHMH8/LygN1arVfR9KpXC5cuXUalUMDQ0BI/Hg1KphEAggNXVVSHEmc1mIcVRf96O3JGB5rBqdgzTnicvYGlpSVh5XAw2deDsZj5gNiHPZDIIBAIIh8PY2tqCyWSSMqqpqSm88847CAQCGBoakrmchER4WGjQ+ZB5cClaMevOQDr3yPfSw9PvZdmAx+NBLBbDuXPn5FppBCwWC9bX16UGjpDuXpoDTBYzG9Pr3C7XLZ1Oi9JkXSgVNl/PgScskbBYLAiFQrKZiYBkMhksLi5ienpaIhRGI4yEKHSwdJ5PH97WXLR25nRqxGQyYXV1VWBufjZzXMxFv/XWW1hdXW0qqaBx3tjYgMFggNPplDaQe4nFXSqVxAkLhUJNSBTb/pLUQzQtEolIDTSbG7366qt47LHH8Oyzz+K+++7DwsICFhcXZWQrIzDmitnUhtwTlk9qBjejZJ1XpHLm3mp1nPl6riP3bzgcFpheozpEZ9rb22VwTqPREISwUqlIy0qHwyGI0V7SBSaTSWrUWQ7Jc1UsFpHL5RAOh5tK0xhUkQCYy+UwPDwsXdva29ulNbR2qO12OyKRCNra2oQZ3tqCU6dctVOgSYOaOKydNv1apkv4/0OHDmFmZqYJcWXenR0Qx8bGxJn3er1YW1tDuVwWgiEDlVwuJ7yZ25HbYizwZlnzqGuESYCwWCwCRdPbZK6I8OD8/LwQRUi5Z6OK2dlZIVMNDg5iZmZGlJpu+MCHSoiJsBUAgVJpULW0GmQuvC5/YbSjNwVLe2jYeTAJ9ZMIwVo9/XmEZ3mtdzObm9fOdpxGo1FGp1HobDG1QLiKinRlZUXao25ubmJ4eBjXr19vYmVub29LmoQlCwCalKP2flsPqI6WGDXTM29lcNJ4kyCo91cymUQ+nxcly4iR+6hYLKJQKIiiIQzPWu719XV4PB6B4dPptDA598I+YG/pfD4vPA229+Wc7I6ODszOzkqUxQEbPD/1eh333nsv/uzP/gzRaBQDAwN4+umncerUKVy4cAFHjhxBKBTC4uKilC8xlcTZ70w3MHLTCBmAm/YH9yvvg/uCexTYdTCoS86fPw+XyyWkVRpwIgPk23C/r66uwufzibPAz6bTor//bhTt5DIQm5ubQ09Pj7S15ehHPlOuD0cxUsdub2+jvb0dr776Kp566ilMTExI34z29nbMz88L/4i5axpcnmdG1xoR0Uxu/ow5aO4JHVzR0SdXhtdNJyGZTArBzWq1ik6nM8i0mN1uRzweb2oJbTLtTLzb3t6WZ8fhIj9sHxgat7FTlpeX0dPTc7vrty8/QJaWlqT9290m+3vgb07298G+7O+BfQF++D64LQNdr9exurp6UznFvtyesGSDc6PvRtnfA//vsr8P9mV/D+wLcPv74LYM9L7sy77sy77sy7787crd6cLty77sy77sy77scdk30PuyL/uyL/uyLz+Gsm+g92Vf9mVf9mVffgxl30Dvy77sy77sy778GMq+gd6XfdmXfdmXffkxlH0DvS/7si/7si/78mMo+wZ6X/ZlX/ZlX/blx1D2DfS+7Mu+7Mu+7MuPoewb6H3Zl33Zl33Zlx9D2TfQ+7Iv+7Iv+7IvP4ayb6D3ZV/2ZV/2ZV9+DGXfQO/LvuzLvuzLvvwYym3Ng96fXvL/JvsTbPYF2N8H+7K/B/ZlR253H9yWgV5dXd2f//k3IHfzDNj9PfA3J/v7YF/298C+AD98H9yWgfZ4PACAf/kv/yVsNhsajQZMJhPq9TrMZjOq1SoAyO8sFgvMZjPy+by8zmQyAQBKpRKsViuKxSJsNhtqtRoMBgOMRiOq1SosFgvK5TJMJhNqtRpsNhtKpRIsFgtKpRKMRqP8jt/Nn1UqFVgsFvkZr7NSqcj15fN5uN1uGI1GlMtlVKtVVKtVmM07j6JWqzXdu8lkks93u91YW1vD5uYmenp64HA4UC6X0Wg0UC6XYbVaUa1WYbfbUSwWEQwGUavVkEwm8Tu/8zvyHO9G4bX/wi/8Amw2G+r1unjPnFhqsVhQq9UQj8dRq9VgsVjkd3w9vUU+b/7fYDDAYDDI6/TfZrNZPqdWq8FoNMr66vVvNBpoNBqyl4xGo3wOv5vrXqlUkMvl4Ha7YbfbZe/q6+B31Ot1uc/V1VVEIhEMDw9jaGgIzz77LEqlEgwGA8xmMwwGAxqNBsxmM4xGI2q1GhYXF1EqleB0OvHcc8/tiX3wy7/8y3A6najX67LnDQYDTCYTLBYL6vU6FhcXUalUYDAYYLfbZb2AnXW3WCwAIGvHvw0GA2q1mugMCn/HvcD9ZzKZUC6XAQBWqxW1Wk32AdeOf/MzuS/K5TKSyST8fj+cTqfsbaPR2HQd/ExeQywWQyQSweDgICKRCF544QWUSiWYTCaYTCa5Tl6H1WpFZ2cnpqam8LnPfW5P7IHf+I3fkHXX6wkAhUIB2WwWlUpFniXXl/aA55FnDoCcGf1/HaVr/Uzd0fp6rjWwq5sANOmrer0un2UwGOB0OmG1WpveRz0A7OwXriX/DofDcLlcWFtbw/T0NOx2O5xOZ9P38vX8vN7eXiSTSVy5cgX/5//8nx+6D27LQPPGXC6XKEsqoEqlIoePD8tms6FSqcDtdgMAqtWqLA6NMpW3xWKRA2C32+XzeJD0wlutVjmAxWIRZrNZPq9er8Pj8chnc/EBwOFwoFKpwGq1wmw2o1KpwOl0wmAwiDIBIMbcarWiXC7LvdIAuFwuOJ1OdHV14fr16/jgBz+IjY0NADubloa+VCqhp6cH2WwWhUIBPp+v6TnejcJr5xrwZ9zsfEa5XE6MqMFggNVqld9T+DP92frZ1Ot1WXOuUalUEqPLdeF+qlar8l0AxHnjYeXhoDidTpRKJTGodru96b28nlsZhPb2dmxtbaFcLqNer+Ohhx7Cd7/7XdjtdtnTNESNRgPLy8uo1WpwuVw3GZa7UXjtvF9gx/G1Wq1y73a7XYxSo9GQc0cjzM/h77lX9B7RzhKAmwy2Vpi8BmDnDPO7buXwAWj6Hl4n9xX1kVbsFG0IQqEQksmkOOYnT57E66+/DqvVKsanUqnAbDbDbDajv78f8/Pz2NzcbHqOd6Noe2C32+V8m81m0aFc62KxKAZZG1c+I73ut4J6W9/HM0sHH9jZK3wNf0+HjQEidQe/g/unWq2iVquJztD7UxtX/tH6hLaqra0N2WwWS0tLCAaD8vn8Ln5mT08PYrEYlpaWxJD/sH1wR0kQKlV6loxMAaBcLosHydcwOrVarU03zJvjYWKUxM/TEYzJZEKpVJLDwQfEjU9lzn9z8fgw6/U6yuVyk+ejI31urHq9LtfExaS3Xy6XUSgUsLGxgbGxMdjtdoyMjODLX/4yPB6PvJcbxe/3Y3V1FYVCAR6Pp+k57QVhhEDhv7mOVqtVlDfXRG/ySqXSZAipzG7l+TYaDRQKBVHoWqlrb5trCOx6rdwrrdehERZ+NxWJ9thbo3sAcLvdcDqd2NjYwPT0NEKhEE6ePIl8Pt8USQPA9PQ0jEYjIpEInE5nU4SxV6Rer4tR4hrq50xdoZ0eKlM6clq4bnrPAGj6HAo/R+8VnvnWz9XREF9LXaSRD/2dvD/9Porb7YbVakUsFsPCwgI6OzsxPj4ujhkjR6PRiL6+Prz66qtYWlq6qw3zrYTPijqXehDYCcwqlYoYwVs5YDTU+rxqQ0odTOEZKpfLKBaL8tnAbvRM1AaAIB8a1QB2g4v/H3v/HST3fd6H46/tve/VvTtcxR06QBAEi1jEYlKUZUqWIyuWLSu2Y1kzcpIZJ2OPk8nEf9jxJDOSrJHtyUS2E9mSItsS1UiJJkWKFQQJopc74HC97N1t77f198fl9dyzC/gnwF9LEXfumcEAuLL72Xd5yut5Pc9DxINOGe0BdQv3X+sSfi+fzyOVSsHj8WBoaAihUAjz8/NNdo3vPzg4iMuXL+Py5csoFAo3ILX/mNy2gb4ZRKkXgJEpvRFtVHlhTSaTQEn0rvizWolpb4U/ryES7ZFrz6hUKt0Ap/LwFItF+V1uJL1zXlQeED6TzWaTz7O+vo5arQar1YpwOIxUKgWXyyXRk9PphMlkgt1uF/jiZhf83Sx6P4EbI2AebK6JPuBcVx5gvZ8UGkl9ifk+FEawOmLle7caf+0c3iyiplSrVblUtVpN/s8zTIfT6/VieXkZhUIBc3Nz2L9/P7xeL6rVKorFIkqlEqLRKBwOB2w2mxj/VmXzbhYqK95LDSu3oiFms1nOBLC9v4xuKBqR4f/5Gvrf/L/+fa5vq3Hl6+iUhf4MWk/wbOhn1a/V+rm8Xi9WV1exubmJjY0NjI+PC8LHzz08PIy3334buVwOAJpg1He7cO3ooFGXauRJIxoU7r++axru1vqi1XHiHmu4WactgO2IVQcS+jX5LDqlwp9pTWXwuVqja37mbDYrqNHAwADsdjsymQyALWfCarUiEongwoULSCQSsNlsYiNuRW6bRkgYSH9IXhZ60YyeeaiZWwa2FChzgLxglUpFIGVuMv/NRS2VSrJYVqsVNpsNHR0d6OjogMPhaNogu90uCpzwuj5M9PL05vA59b+5gYSuG40GMpkM+vr6UK/X0dvbi/PnzyMQCKBeryMcDsNisSAajUo+lp50u0mpVJL8Oy8RDy0VsvZQtZEE0LT2wDY0xa9ZLJYbPF6LxYJgMIiBgQHcddddeOCBB/De974XExMTMBgMKBaLTTlnfX4oN4O9tYLm/7VxZ8TPs0CvPxqNIp1OI5PJ4K677pKvLy0toVAowOFwyOVuhXDf7VKv1wXKrlarkjKg3MwQakWov85/60gIaE43ANtRNF+b99hiscBisTSl4HSUDuAGhahhb81VoMHgc7ZG1JrrwJ+PxWJIpVIAgAMHDgi8GolEMDMzg2w2K/egnYSOdStaphFWGlS9d633rxVG1lF4a66YX6vX68jn84jFYpicnMTly5cxOTmJlZUVVKtVSXm0RuWtTpZ+Br6XdgLobFEHEHXl82xubiKdTsPlciEQCKCzsxOxWAyVSgWVSgWRSASpVAobGxvyWrSNtyK3lIOWH/6/ipd5ARK6+D39oUjgokIrl8tNRksTvEgKq9frQryiobfZbMhkMvLebrdbLuX09DQWFhbg9/vxvve9D9lsFhsbG7LAOtet4QlgO6dF50JfVHrbfC5uJrBNQuFn6ejowNtvv429e/eiVqshnU4jFArJZvOgtJPoiLUVgtQOjb6w3ANNxiN/QMNLzB/x63SqHA6H5KGLxSJmZ2fl93w+Hx566CHY7Xa88sorSCQSTb/L96OnzjSG9vC1Qufr8kJpwhA/czgcxvLyMsLhMBYXF3H06FFMT0/j4sWLqFarCAaDNygXrejf7cK1LRaLKBQKsNlsTYZUO2NU5DSK2pDr12M6TCtOrW/Ia+H5cLlcsNlsMBqNcDqd4jASamd0wz3UyFsrjN6qrHXOuxUd0src7/djY2MDPp8PS0tL2LVrF2ZnZxEMBrGwsIClpSUANzqc7SC8SzoNwD2m4dYGWt+z1jvPu8k7yXXSKFyjsUXGJbnTYrHA7XZj165d8jvr6+u4cuUKAGBwcBBer7fJLvE9Ka33W59dQtl8Xu3ca8OfyWTg8Xhgs9nQ29uLWCyG9fV1HDx4EPF4HFNTU5KSpV68VTTttlx6k8kkTGV9kRhxErunUuRFbFV4ZP5po9doNIT15/V6BarmAvl8PgwNDcFgMMDlcqFYLCIYDOLuu+/GgQMH8I1vfAOvvfYaJiYmmggq+lLwOXiRNaShYU165CSM8JBwwzKZDMbGxlAulxEMBrG2tobOzk5sbGyIg0GIl55XuwgNaLFYBLDt8NBD5boy/6QVJPef+1EqlUShMpdkMpng9/sRDodhNBqRz+exubmJVCqFRmOrdjCVSqFYLCKfz6PRaCCVSuH8+fOYnJzEe9/7XjzxxBNyKTc3N7G5uSnPSEWv0xv60t4MAufnanUuHA4HZmdnkc1mMTMzg0gkIpdVKyWz2SzRZrsI0QGy4QE0raMm2ADNkYzOzwLbBlArRk0INJvNUvVBHaOVHclIPG8+nw+dnZ3o7e29gYGtERw+l1bSrbnq1lxmq2PKdVhaWkI+n0cmk8HP/dzPYWlpCQsLC015VnJh2kX0XpMnBOAGh9xoNEpAQ6Eh1ogZ95WcIoPBIPedgZ7H40EkEoHFYkFnZyfC4TCcTiccDgdMJhN6e3tx//33Y+/evVhZWcH58+dlz3hmgG3Hj06YRmtoC1r1tobk9VliFO1wOODxeDAwMIBgMIhSqYRr167JmjASLxQKtxxB39ZpIcGKUGLrJeMmbW5uygZoD0jnjXTOt9FowOVyNUEAjFJsNhv6+vpQrVaxtrYGu92OYrEoZCQa9gMHDmBkZASf+cxn4PF40NXVdQPBg94R3+tmsBqVA59LX0gakXK5jJWVFfT09MBgMOCDH/wg/vt//+9wuVxSEsZIm2y9dhFNjtDRMQ84DzxL4jR5rl6vS2mb9pR5eB0OB6rVKhYXFzE7O4tSqQS73S5nKp1Oy6XQaRCS/tLpNN58800kEgn8yq/8Cvx+v+R/6ShRKZCUyM9AZwyA7L+G1LSRphLp6+tDPp9HMpnElStX8M4778DlcjXB2nQG6Ey0i5C5HYvFmtAO7qfmEWgHGNhG4ng3NaQJbDluVNgMCPh7zOWyWoJOGIMGkkoLhQICgQD6+voQCASwubnZlIcEmkllWg9olI3Cn9FwOF/H7/cjk8kgHo/DaDSip6cH8/PzALZTM3y/dhJNzuWduZmjTj2geUvacdFrCkCCqHK5jKGhISHm6nQhgziinLRJ1EM9PT24//77sWvXLpw9exbpdLqJi8BAQz97a6StdZQ+I8D2eahWqyiXy9jY2EAmk4Hdbsfo6CiOHz+OycnJJuIzURSfz3fLjtptQdxGoxEbGxsC3zGK0oSvWq0mOWANd3NDNLOThnpzcxO7du0CAORyOWxubgrpJp/Pw+v1imJkNGQymeBwOJDNZkWxA8CTTz6JV199FXa7HR/5yEcEctTGlt6Y9pS4eHwencfkhpEo0Gg0UCgU0NPTg97eXly8eBGPP/64PCONEJ+rnS4mo12d3+X+MqUBbJ0J7hEVaWuemZeJxpEG2Ol0ijNYLpcFPtVsUEZTvMhkjjcaDWxsbCCVSuHxxx/H4uIinn322SZm+c3yUNx78iFqtZpwG7Ti0OxSo9GIvr4+nD17FplMBl6vV9Ib5XIZBsMWaZKoTKlU+ont049bqOSWlpbQ3d3d9HWKVnZUzprfoXP+Gspm6U69XkcqlZJ7x73XkVqpVBJlDWzD6fxdu92Ozs5OuN1uzM7Oypljiq01haUDDuYKddTF72sjbzabpcRzfHwcn/vc5+Q1NI/BarUinU7/82/G/0NhHTkjWB05c8+oTzc3NyXlxzSEPhc8L0ajUVjya2trTakpjULQsKdSKVgsFtEbTINZLBYMDQ0hEAjgwoULWF5exsGDB2+AtbXjzjuu0zJ8X53O089MJzIej2N8fBx33HEH/vZv/1ZQZaIIuVwONpsN6XT6lit7bjuC1lETgKbLpD0Qt9stURSw7RVRGo2GKEyfz4eLFy/KogJANpuVzSgWi/I6fr8fdrtdfoZKXBPSxsfH0d3djf/23/4b7rjjDgwMDMBg2Gbe0ctvZYw7HA65qITXdU6F3hafi1Bno9EQSIMGJRQKoVwuiyffLsLSBl0/DkCiUXrFFJbe8ZBq+Jj1pzysfr8fwLby0xGLJgHxsBeLRVitVkmZaEJHJpPByy+/jO7ubvzpn/4pdu/eLdG0rpHUxplRYTabRTabvSlPQRsHKnqbzQaPxwOv1yvNcqgE6JzQYLeL2Gw2LCwsNCFMrRCxjjS1Y9MaOWmjmM/nEQgEUKlUxBkH0LSmwDbkzEhIR0D8GZPJhEKhgFwuh2AwiN7eXkQiEUmt6GhaR0fc00qlgmw224QW6lQIHU4ACAQCOHr0KL74xS9icXGxKZWXz+fFUaXuagcxGAzIZDKS4gCa89Ka0NdaucO906kDzV3K5/PizOj0CKPWbDYryIvH44HdbkehUEChUBDbQr3T0dGBhx9+GC6XC1/96lebYGftSLSmujTBWD+Hfn59humAf/nLX8aFCxea8uzlchlutxv1+lZZosvluqU1vq0IOhqNoqenpwnuY9kSLyejjkwmIwsEQKBfzYjmh3U4HPLwzNdxARuNhlwWu92OeDzelD/Ql7NQKMDr9SKRSKDRaOChhx7C5z73OTz44IPwer1i0LnJOuqi4g4EAkilUsIy18QArZRJUDh58iR8Ph9KpRKcTqd4jNxQs9mMc+fO3c4y/1QLvWCuPx0yfciB7ZwL95xrp88OHR8e7mKxKKQferat0BJhLpvNdkNdcalUkjwVI9e33noLFy9exIc+9CHMzMwIJKaJaNwzo9EIh8OBYrGI9fV1WCwWMRaaMKS97EQigVKpBJ/PJ2eptcmNwWBAMBiUu9EOUiqVEI/Hm4zUzUQbNx016NwvocRqtYpQKIRkMin7xPXWTjKRNJ4hdifkWjOK4p7W63Ukk0mEQiFMTEwgGo3eUFLHs0QFzXOYSCRgNBrh8/ma8o5Ac6517969OH/+PBYWFgBs1+tWq1XhzGxubiIUCv2z78X/K6lUKkgkEpJqpOj7qs+GNoD6fmhkolKpwOVywWAwCDOe6CaNv47QSdKiA9DaAIuG3ul04t5778X+/ftx5swZZDIZuN1uQcLohGuDTSSUDbcYbGhkiGfSaDSiq6sLs7OzuHLlitg7ptL4Nx3a6enpW1rj24qg2SVFM+6A5hpEfmAaws3NTWFw64XlptrtdlGG9IZ13pJko1KphGw2K9AHIypCnclkUvK+DodDIPKHHnoIJ0+ebCIq0cDS09cdwCqVCkKhkNSvAs35aX7m3t5eTE9Pw+PxNEHbzJuVSiWUSiWkUimEw+HbWeafeuE6aNKX9iYp2nhTNBNak7b4ujrPD2wpSu6xxWIR48tzUqlUhDQWiUTg8/lQLpeRSCQkX1ksFvHSSy/hd3/3dyWFwmegAtER/uDgINLpNBYXF5HNZoX0qHPtAJBIJJBOp9HX19cEn9Ow8FyFQiGEw+G2iqCnpqaE/EMjdbMUgFa8hDf1GjLKrtfrcDqdErVSWeqGD3xN4EZ2L1+P0RchT4vFIihXJpPB3NwcPvCBD6BQKIiR1pERpVaroaurC6urq9jY2EChUGjKlfKsVKtV7N+/H5OTk7h69aroRpfLJbn0fD4vhvr69es/sT36cQu7omnHSO/rP0aE0lA2fx/YTlsSAaOeZ5BHwijtTTabRT6fb/o5u90On8+H2dlZ5PN5eDwe9PX1IRgMwmAwwO/349ChQ1haWmpCevRZoi2r1+tYXl5GIpGAw+FoaiPL5+XPBYNBQe1KpZJ0z9Ovxd9LJBI4evToLa3xbRloegwUnU8kFg9sXRISfAA0dXniZWuFrzV7mq9BD0kTkHhRWN6wubmJZDKJnp4egZRJ1qpWq0Ig++IXv4iHH35YuprxcmoWn8ViQTKZlGYrVqtVLrLukrRnzx6cOHEC/f39criYh6CTwZwjD0y7SCu5QXvOJEfpKFp7mjoPzTOgFbHO++vcjb7sdPp4+IvFIkKhECKRCJLJpDAkealJIopGozh58iR++7d/W5iUms3PPzx3oVAIhUJBUi1AM6xZLBaRyWTkGQBI+oNnnPeBpUE6xfNuFzq6rUQaoDmy1EID21pLrI0rW/hqh48Mbr4+dQ3JZjwniUQC5XK5Cc3QTpXJZEIsFkOhUMCBAweEb8J90fB8o9GAw+GAxWJBPB5HJpNpivipcIPBIEwmEy5dutSEyrFvA7B1LwqFAjKZDCYmJn4i+/OTECJWFH0GWo2z5iFQqFOpA3h/6FgD2+2XC4XCDf0EHA6HVM0YDAa43W5Eo1FcuXIFo6OjglhR1zD91Nvbi4mJCbz44otNdgZodrBpt6LRKHK5XNN7a2TV7XbDbDZjdnYWPp9PyNG0ZbrktFwuY2xs7JbTnrdloDUsQWVGoZeqk/40eoQluYEkE/HykPlarVaRyWTkAhK6MpvNcDqdMJvNcLlcqNe3axwZUXEDWZtNGJbP8uEPfxgf+MAHcOjQIYEcaKgBCORJhdvV1dUEoxDSrVaruHbtGvr7+5HNZmUtcrkcHA6HeHSxWAz9/f1wuVyIRqO3s8w/1cKoSENOmjQBbPfP5XmgVwo091QmikEDx/PA3L7dbpe959ozp5vP56XlqsFgkA5vfH9gu284L/Lc3BwqlQr+4A/+QPKQPM80Jnwmq9WKXC6H1dVVcQp5RhuNLcb45uZmU7N7VhbwrBiNRkQiEQDbKZ52Ea5xK2mLQkVHac078jVIoHK73TJch8KzQJ4LI3DukSZ7rq+vw+fzNfEYeFa0GAwGXLt2Db29vXjwwQdvcL4BiKGt1bbqnKvVKmKxWNM5pS47cOAAXnzxxaaBPOQeEBGo1bb6IwwODr6rh2S0ioZ3W3PO+ixoFAXYRqtoKxgs8Xeou3nfGPgAEEOr4WySSE+ePAmn04nR0dGmahLuF+1LoVDA8PAwxsfH8fzzz4vuB5r3ViOsJP1Sn2k+BCN2DgkiL4r2gSS5ZDKJzs5OdHR03DJh9LYMNC8MvVwaTy4+jSMVOC8Q4QfmiSwWCwYGBpDP5yU/Q0XpdrullItwByNhk2m7tRoVeC6XEzhLl8m0MgYB4Ld/+7fxu7/7u7jjjjtu8IbIBGXuw2DYosNzcAY9+PHxccTjcTkghOvpRRFa56HJ5XJtdSk1Q1lHTjrdwXXkpdIEG146zT1gJETUQucQCVny9alMh4eH0dvbi6WlpSYyhyakET4nFF+v1/H2228jkUjgl37pl1AsFpuUM/OkhNJ7enqwsbEhbT15XorFImKxmBA9WomTVAYdHR3SuY4Kv11Ef2b9bx1Fa+Hea2iT0VMikUBnZ+cN66OJhyR1aqSFaa96vY7u7m5B7HSjI+DG1o+NRgPLy8sYGBjA8ePHBTblM+lUjc1mg9vtRjKZRDwel0CgVqshEAjg3LlzSCQSTWV65KMUCgVYrVYsLCxgYmICfr9fjHw7CHWfJuzRsPLf1BFAc005sB1VcxgOI0x+jzpBt8f0+/1SU01nf25uDktLSzhy5AhcLpekvohmcU+ZLuPe3n///bjvvvvw1ltvNTkMRIUY1edyOSwvL6NUKkmKlZ+JnJWlpSUJ6HiueR5tNhuSySQCgQB6e3tRLBZvufz2tgw0I09OENJeLCMcDXsZjUaZbsRLUKvVcPnyZUQiEflZ5qt5eHUTDL4+ST8swVpdXZUWm0BzAl+T0AhX04hMTEzglVdekcXk+7B/Khd+bW1NDMHi4iLeeecdSfR3dHQ05VGy2SwASLOExcVFdHd3I5lMNnVYagfR602ugE5zUBFrDoE22vy/0bjV/SmXyzUZd15UXf7C1EY2m0UgEMDY2JhETWazWZibvDjFYlGMs4ZS7XY7qtUq3njjDezdu1fq4fk+RHyYB3e73RgYGMD+/fvx8MMP45FHHsEHPvABdHd3C/dAMzV1BOb1euHxeCQPqeHTdpDWMimucauBbk1RUDdwj/P5PHp7e5FIJJqi69Yom2NneQdTqRSSySS6u7vR2dkpxv9mDXSo6HUe22Aw4Pz587jvvvvgdrubEBU+Jx0BOmLvvPMOjEYjPB4PPB4PHn74YZw9exYAmhQ/nXq73Y5EIoFIJIK+vj4hJrWLUKdqx4n3iI4QkTN9P/i7NIjsX0+UVJ+r1hQqnV2n04lSqYSzZ8+is7MTu3fvFuSDaREaf42E8r2ZgpiYmJARkDdLyTDAzOfzWFlZwZkzZ/Dcc8/hwoULePvtt5HNZnH27NkbHFW+FtO4ZrMZIyMjKBQKTdPzfpTcNsTdmmMgpKOT+dwkjo0kNFCvb42E9Pl8OHHihOQXmCNgfpgbUigUYDAYJJJqNBoIBALI5/MIBoOIx+Oi8IHtvBjhCJINNKFjbGwMp06dwvDwsGwYlQbzR0ajEdeuXcPq6iomJyexZ88eRCIRzM3N4dSpU03tIwmzsxYvHo9jaGhI2L309NpFSMTQU6WorOv1ugwkoSLUeSd6vLx4mjjFC0hnjXwCXRLV19eHnp4eRKNRcfz0OE+d/6Uy16QlnrVUKoW5uTk89dRTgq5ouJWd0jo7O/Hxj38cw8PDuHz5Mq5evYrJyUm8//3vR61Wk853zJ0yUne73fD7/U3zy7Uj0w6iYcmbsXZ1tKTTYnq/DQYDrly5ggcffFBQEk041JGMdsBXVlZQq9UwNjYGr9cr+05kh+eLz8Y/muMAbPVcyGQyePDBBwVJ0RwLPs/a2hqWl5fx4IMPorOzEwDQ1dWFoaGhpjQJn1M7arOzszh06BByuZxEX+0kdNRprLWTToIekSkiLDo65c9zyAQRWJbZ0ZEjH4CQ9ezsLJaWlnDnnXeit7dXYG+muoie8v6RaMygiummQCCAgYEBLC4uNhloBnqMuiuVCv7n//yf2NzcxMc+9jEcO3YMDzzwgJRscZ40f0fbnGQyib1794o+uB007bbroPkB6TFRwbFHqs4JkACmWbqFQgE/+7M/K9E2PQx6loQ52UrRYrGgUCgIc5eLm0wm4fP5hEykS7ioCDSsruulR0dH8cILLzT1x2XEVigUsLKygl27dkktc71eh8/nQ3d3N6LRqLD0eDDr9bqwNtk4hcMSgPaaZkW0RDtTPA/8nsvlku/rC6nz1jwLmoBFr5Z74fV6hcjDkY0sY+Heer1eKfUrlUqS+yN5hJPNeGF5XuikBQKBJm+30dgqk7rjjjvwyCOPyLADs9mMWCyGlZUV4RRoJ0UbALPZDIfDIf/WTQ7aSTSkp42vjiJ0/beGuflzuVxOyhR5PmhQHQ6HKDkq3dnZWYRCIWn7m8vlmnQMkTy+t9YJ2ojwec+ePYt9+/ZJoxE+Jx3LK1euYGNjAx/72MfQ2dkpeolcEwBNDggRHJvNhmvXrmHPnj1NyGM7Oeu6YZFGSbSTzPKyVjKg5iIx5Ul7QLSLYjabkUgkpGnV9evXYTabcdddd8FisQibm528+BrsFZ/JZGCxWORO8nyRuMl0WSwWa0J6aKTn5ubw9NNP4z//5/8s+zkwMIC+vj5x0ogU0CbQWWdLauo0j8fTlA77UXLbLG6d9NabA6DpchH6oILlz7Id6OrqKoxGo4zkI9TJ5Hkmk5EcYWdnJ3K5nEDIm5ub0vva5XI1RWtkBudyOVitVvj9fqTTaRkwz3zVmTNnZOFoQNj4n9FUMpmUBgOshT127JjA2mxE0Ghs9f9dXl7GkSNHEIvFBE5tJcu0g+hRj4QNtfHV50HzAjQZi8xMHmxeTqIldKjID+BAdLb+5PvRABQKBQCQ4SokepFdT7IXo6RarYa33noLv/qrvyoeMiPsJ554AkePHsWFCxfgcrmkNIsNSXh26YyyBpNKOhAINMGq+nO3i+j0lBZNIKQO0DldoHmYRjAYbGo0RENtt9uF6GWz2eByubC4uIje3l75HZ1n1k4inT0ATdEKHUPdfrFcLmN5eRlHjx6F0+lsSs+9+uqrCIVC+PjHP456vY5QKCT6TKMmrcih0+kUBG3//v1ChG03YWCkuzBy/zUawj3iHnOPGMUyPUBjTQTO7XaLLqfh3djYkLQTm5XU69tNYXTaiXrfbrejVCqJjggGg7LXRPUikQhWVlakhwE/F+Hsz372s8jlcuju7obD4UA6nZbaeuowOpI68i+VShgeHm4iO96OLvgnGWga4NZattZSGbK3+WBGoxFXrlxBtVpFOByGyWSS7mGhUEg8TzaISCaT8Hq9ssg0pPV6HZlMBpFIBPF4XBLubGbSaDTkPTOZDHw+n/RqpdL42Z/9WTz99NOSC4nH44hGo+jv729aRO1sENYkKY2wicViwfLysjC7NaGonaJnYJtEowkgjIz0z2hWZ6t3mc/nm0gfXFe+NvPKhK00v4G/u7m5KfWyuqNYsViEwbDVvYkRVLW6NV2KxprKYXFxES+++CJGRkZgNG41o7jjjjswMzODs2fPwmazIR6PC2JARaw75LFMh8/X2dkp8B2AJkXVbmeBRrY1z6wjaY2i6JQSAKRSKQwODmJpaUlyh4wyeVboPK2vr8Pr9cLlcknExH0BtnOadLLoQHBvuH/JZLLJgQSAlZUVuN1uDA0NwWjc6lK3tLSEoaEhPProo6hWq+js7BTDz9djSQ0/kyajXb58Gffcc4+k29pRNE+ExlcTwviH6U+NTGhbQWc9n8+jUqnA7/eLcWPZnNPpRCqVEmcnn8/L2WKgR71Uq9WEoEdeCZEzu92O+fl50R3All4aHBzEgw8+KAFkqVTCq6++ive///3SP2F4eFj0F88YHVB+TuasAWB1dRWRSKSpIgXY1gm3IrdloOlxsgRK53mokLk4NKaFQkGgS4Nhixn98ssvi8fDKDmXy0mXnWg0CpPJhGAwKMPQG42GeFQAZBCC0+mUTjIApPbYbrdLYxN6NnQWuEjHjh3D5OSkjA3s7OxEJpOB3+9v6h5FQ9NoNPDSSy8JLZ/eF4dHBAIBxGIxget1CVI7iT5cPKCaXEPRkCLQDH329PTI77Y2nTEYtnpY+/1+NBoNqTcmCsLJMUx/EEVhRMX9577pznSEvWnIE4kECoUCdu/eDbvdjpmZmSZPWKNAhKr5O1QI9LaDwSA8Hk8TWVKvh3Zi3u3S+ln0/1sddeDG8hXyPA4fPiwREhEwfWcqlQrS6TQ6Ozvh9/ul6YfD4RCYEthuO6lZ2BqdyWazyGQyTf3VdU6Sc4TJ1O3u7kYgEJCzxOY2OmJnVMx7zp+5fv06jh49iq6urqYhEe3moOn9JaGXzigdeBpRrR9oLwghb25uyhmoVqtIp9NN7VjT6TRWV1fR0dGB7u5ulMtlGX5B/c4yV3ah4xkgudRkMknU21r6S+TW7XYjHA7j6tWrsNlseOKJJxAIBOByueDz+SRoIDLHz09bppGVVCqFjo4ORCIR0QfagfmxQdwcAEG4WDNw6SUxH6w9V27kyMiIeDmxWEw8Z9YfFwoFdHZ2ipFn/pn5vnQ6DavVikwmI86By+VCoVCQmuhSqYRMJoOuri7E43GZIc2ol5GYxWLB4OCgRL4Wi0Uibz4v37dWq+Guu+6SMgtGgOzdPDExgdXVVYnm9dSmdso70THRIyLpnAHbOWrtSepI22QyiZNHo0tjWigU5JDzDOnGNMViUVIX9KadTqfsJRVyKpVCPp8Xo8EyPhplTdxzOp1Ip9OSVyZcrpVPPp/H2tqa9HXOZDJS96+9cN4Ffq2VsNROzppeQ2109WfVjlkrzG0ymbC+vo6enh65b+SLANujavP5vEyNI0Obff4Jf+uotjUfDECMPtMeRGSopJk6S6fTCAQCEnFpRj+NAdtPer1ezM7Oyv2moSevYnx8XEpFW53WdhEiZ9w3RqnaGOn1AZrz1rznnZ2dog8YmbpcLqneKZfL6OrqEoOazWbFoU+lUnA6nbDZbDKXW9skj8cjDafIEmcvbOpons9YLAan04m77rpLdBjr4Pn8LLszmUzwer2SitN77HK54HQ6MTAwIBE8CbRckx+LgQYgAwf0peOlYkTZSgYAtj3o/v5+gTJ0LTW9Y0YgTPCzrjiRSMButyMcDiOdTsPj8UgOgE3oGfW43W643W6sr68jHA6Ld0YSW622NTKRMCi/r2tmDQYDQqEQVldXsX//ftnUBx98ULwli8WCRCIBs9kscIzuGMVcSztFTrx4vJh0tnRKQDsk3HdNJtIGnk4O84J2ux0ej0dIhwAE1mTDEd17t1gsClrC92ckVi6XBenR+Wc2xvB6vbBYLMICTyaTsq9UPsxNA8Ds7CyKxSImJycFJaHyHxkZaSrb4me/2b/bRbTDoaHem+XdW3+W0RF5B7rWlfnjbDaLUCgkP8NyFb6u2bw9flJzPjSbmDwH7SzpWtt8Pi91qbVaTVIlWsnTuefrxuNxHDx4EP/7f/9vMeY0+BsbGzhw4ECTM0BpNy6CbvJE3an3mevdiiRy3zXTm4ZdpzHpxIXDYXH2Y7EYAoEAcrkcstksfD6fEMQ8Ho+UMTEV2zorgrwJImJmsxmZTAbz8/NwOp1IJpOo1WriNND5azQaTWkSOpaMmJn/rtfrWFxchMfjEbK0dtgpt+qs3zbE3doNqdUz1kZbLzo9CavVCo/HIw1J6H2USiUEg8Em9hsVNWHKaDSKSqWC3t5eUaSpVAput1vykKyBNRqN6OzsRDweFy+2Vqshm83C6XRK5yLCEySXaW9+enoaHR0duHr1KjY3N/FHf/RH2L17NwwGg0R3tVoNkUgEiUQCbre7KS/CdWm30grmE2lwNUKim4RQSWojTQiMJW10rnw+HxwOh3jMRCoILTH3wzSJ0WhEOp0WB0uT+ZxOZxPsxKk4dB56enrkcxDJYXkULzeVCx0Kt9uN4eFhIYMxB0oHQ+eage22hxrSaqfoSZNBNQdBrwO/djNoj18jlEn0inojnU6ju7tbjHChUIDP52uKsjQXQFeTMPfM96KjxShcG+56vS6jEglZk4jKs2sybbV+XVpagsvlwtraGiYmJvD22283OayxWEymZuluhRphaicURZ9ppoI0fK05CVwnnRbVwt4BTFVsbm4iFovJcCY6P+FwuInhH41GZXAR7QT3mEgr77/JZJLv82xcu3ZNfr9SqaC/v1/sCvUNhXu8ubmJEydOYHl5GefOnZMgkneCUDx1I8+HTgfcKnH4tgw0D5rOr9EYErrUm0CCEHMNurtLOp2WhujpdFqMNb0xRiPMBTNHwPmjCwsL0kginU7LptCbrdVqWF5eljpJelEkG3i9Xni93ia4QRuTYDAoEHkikcBv/dZvoVQqIZFIoF6vi2EIh8PSnpQHtXUz2qkHsyZEADfWvbYaIh3VtCopQlnMX1ksFmHrO51Oga/5ejx3HGSxuroq+Z6zZ89KV7JkMglgezQm55OzeQidRUbdbITfSmIi7Dk3N4fp6WmpoeZ515EclVMrvHq7rM13i7TWdut7r1NEjIRanTSuF1u28jUZtXR0dEiemEQh/p+NIziXnjqGCJmenKfvInUPp9UxvaL5MDyjOlIPhUJYX1+H3+/H4uIi/H4/5ubmAEBITrVaDclkEvv27RNEQOsTrkU7iTbENzvnjCq1o8a/eU60M8SI1uPxYHNzE319fbLniUQCHR0d8nPcd8LJ5KQ4nU709/djfX1dyrLMZrMYz3Q6DaNxqzFVNptFR0eHlHG6XC6kUilxGjWXgk6c1+tFd3c3Ojo6JHjQiIzRaERHR4cEMbqKQX/uWz0Lt91JTI9l04qJXyOEwJ/l12jU6FV0dHQA2IIVady5YRw9yXIb0vi7urqwa9cuuN1uPP7441heXsbZs2fR3d2NO++8E7Ozs/D7/bBarYjH47j//vuxuroql4heNptpaLieBw7Yyi9dunQJdrsdNpsN3d3d+O53v4vHH39cIHi3241cLoeuri4hLelxYq2eV7sILyENb2tulWvI/aPh4x+yrgGIQabny2k1RF5IytAMYL4+o6uzZ8/CbrfjiSeewNmzZ/Hmm29ieXkZJ0+exDe/+U38wz/8A37wgx8I8U8/m24P25oXSqfTuHLlCiYnJyUvBQALCwuCuJANTu+bEbX+vBpybUe52frp896KJOjzQkiUeoS5ulAoJAQfOlYGg0GMOec7s9dxtVrF+vo6lpeXxcDzLlI55/N5nDp1SmraWZdMshl/lo48HQm32y1ptJmZGZw6dQqf/vSn8Zd/+ZcSdLD3PkmCuoRIOzHtJlwrTcKkcJ81UU93FWTwRoIlu4MxXUjWPM8BjR4JX+yDb7fbMTc3h1QqBY/Hg+npabz55pvYv38/hoeH4XK5sGvXLhw9ehQHDx7EoUOHsLCwICMtnU4nXC6XENOoozQqaLPZ0NXVhY6ODsTjcXzjG9+QyhOeW11pEgqFxCbyHGqjfDv24LbmQbO/LEshtNese3DTO9VTh0h5J0GILf5SqZQsFhuWNBoNgS35euFwGEtLS/D7/ZKvZm4ikUggHA7jvvvuw/79+3Hy5ElEIhFMTk7i4YcfxsWLF6UnNhWpzo0DW8bC6/ViZWUF/f39MJvNWFtbg9/vx/Hjx/GZz3wGjz32mETj6XQakUgE0WhUNpORGbAdObQTQQzYhiY1SsLPCqCJMKKJWzofRfYsI6ZoNNqEOFDxARBiH+HqcrksREXmfc+cOYPl5WVUKhV0d3dLXnLXrl245557JD/Z1dWFVColKQheHM025ci4RCKBiYkJHDx4EAMDA7h48SKi0aigLkST6vWtMYlUyq0Omvac20lR62hA5/j4PWA7ouDf/Jr+Pe6F1+vFxsaGREjkk7Cyo1AoSK9/zmxnK1Xum36dVColpZmzs7MyxYr3k4ZUl4ryM1QqFYm82E7y8uXLiMVi+MhHPoJ4PI65uTmpr7Xb7VhYWMCTTz4pZYA6xdWOegDYXi86Whoh0egpkVVdZgRs80WYEybvhwEe0VWv1yvpSf4sYWU2Jrp+/TpmZmakH3exWMTq6io6OzuxsrKCyclJXLp0Cevr63jyySexsrKCwcHBJvIgP5MmKh4/fhzd3d144YUX8Oqrr2J+fh4PP/wwJiYm8KUvfUnSqsyjs0af0XMrstBqd36U3JaBrtfr0ozcZNpu08moQW8Ev04FxciZHq7P50M8HpcDzs5bVJy6c5jJZMLq6ipGR0cxPT0Nk8mEjo4O2Gw2DAwM4PXXX0cqlcLQ0BBeeeUVzMzMSD3zhQsXxGMG0PRMvDg2mw2rq6s4ePAgVlZWMDs7i1KphFwuh8ceewyrq6u45557AGyzdWOxGO6++25cvnwZXq+3iYDGjdYMvnYRrXj056Ki1Q3qecn4fSrnYrGIubk5DA4OikPG/so6tUFDzeYFjUYDLpcL8/PzAle53W7hBbCz2MrKCvx+PyKRCJxOJzY2NuB2u7G8vCwEIE3eYy6Uinz37t2IRCIYHByUKP3ll1/GxMSEPCsJa3RWeW4pvIBaYbVTFK0hfp1r1AQtnXtudVSIQi0sLGBoaEgiY977fD4vPbZZHsU8JRU1G08weiFxaG1tDXv27MHU1BRcLhfOnTuH0dFRjI+Pw2w2C8GvNbqnovZ6vfD7/Th58iRWVlakLOfjH/84zGYzvvCFLzSx+OkseDwe6evQmmtvR6EO5V63pva4bvwZvSb8vsFgQCAQkFKrxcVFgbnX1tYQCARkshz1ASt12KUrEAhg165diEajmJqagsFgkNbLa2trSKfTuHr1KqxWKx566CEUCgVMTExIWZzWaXz+np4ejI+P4+tf/zpmZ2exsrKCSqWCD3zgA3j/+9+PL3zhC0IQbDQa8n7BYFA4FTpipk5sXacfJbdloAkF8ULQc9YRFb+uCQP1el3yQsxDs3G8z+fDwsICent7pXyGvbsBSEvP/v5+6XO9vr6OK1euSIORvr4+OJ1OLC0t4YEHHkA8Hsfm5iZSqZTMCSaNnwsGQLw3v98Pg8GA5557DlarFUtLSwgEAvjwhz+Mb3zjG+jp6ZHab9beOhwOrK2tNbFOuQbaGdDlN+0gvJS6m5iOoDXk2QpL81ASiopEIujo6MDS0hIMBoN4zmTV6npkv9+P2dlZpNNpiapDoZCUtbEhzdzcHDweDwYGBsR4+nw+ibCYl+RzGo1GXLx4EZOTkwiHw3jiiScQiUSQz+eRy+UwNTWFkydP4uDBg3KGqDAMBoPkQbnnGtbjvWin/deiUw/AzaMCHS3r6BkABgcHcfHiRUQiEWHJOp1OzM7OCjmIjYF41lgyQxSLQcHs7CweffRRzM3NCUSeSqXw2muvYWxsDP39/dJEhk60LhMzmbZ6vcfjcUxNTaHRaIiOcrlceO9734tAIIC/+Iu/uCHqY09ojQCSKNnO0opAtYrmKLU6LLQNTD0aDAaBsg0GAzY2NuD3++F2u5FOp8WmkHXt9/uxZ88eXLx4Eel0GoVCQRpNhcNhgcmXl5exsLCAcDiM/v5+AFudxMju5nPwebu7u/Ge97wHr732Gj7/+c/j/PnzCAaD2Lt3L8bHx3H06FH80R/9kYxCZgDh9XqFuKbRM94PzUe4HX1wWwaai6uLvIm76/ZmzPFpQggAIW4NDAwgm83KjNS9e/eiWq1ibm5O2G/5fB7ZbBadnZ1Cf3e5XHjzzTexd+9eWdwrV64gEAjg1VdflR7QrK3m5pJlSwYfsJVnPnjwIL773e9KHmFgYACZTAbj4+PI5/N45pln0NPTI1AZiWiZTAaHDh2SSSpk9DIvq50X3VWqHUQfLp1X1p4zf45KlagK12V0dBQvv/wyDIatOc59fX2YmZmR12ATGt268c0338TAwACCwSAsFgsikQg2NjaQyWSauvocPXoUs7OzkruMxWLC6GxtN1qtVrG4uAiDwYC7774bXq8XFy5cQCKRAACcOHEC6XQa4+PjTQq3VqsJBMrUiYb5WpVVK/msHUSzUP8xw6xz0jq6ppMzMjKCN954A7lcThTu888/j+7ublgsFumtzCgsGAzKPF+SfCKRCM6dO4c777wTU1NTKJVKGBsbQyqVQiqVwoEDB2C1WjE8PCz5QI3ImM1bfdMPHz6Mr3zlKzIhK5VKoaenB263G0eOHMHc3By+/e1vC0GSCFmpVBInX1c0aMSA0k5cFOBGFIV3QDvurXnpVvRqZWUFgUAANpsNoVBI7jM5Squrq5KaJDFweHgY169fl1GzsVgMXq8X/f39cj5MJhP+4R/+AX6/X5rhjI2NNdW/U18DW6m0sbExPPvss/J7u3btwuDgIGq1Gn7mZ34GsVgMn/3sZyWY4O8SBe7v75fRxDxb2iDrv281gr7tYRk0VtoT1sZYQ0Y630sjaDAY8NBDD2FxcRGlUgmvvfYakskklpeX0dPTg1wuh2q1ikAgIBeYbDsAePTRR3H58mXpz33s2DEEAgH8+q//OoLBoHQkotKmYeAAbXajunbtGl588UWMjY3B5/PhypUr8Pv96OrqQr1ex/z8PPx+vywyvTxewlQq1WRIAAhCwP+zWUs7KWYeuFZCHKGb1mhZR01sl2owGDA8PIxvfvObcLlcwnolwZDkn3K5jEQigfX1dRw4cAAOh0Oa0PBsud1u9Pf3N5XghMNhAFuXm3kqHdXSG+/s7MRTTz0lTP+lpSWpfX/66acBAOPj4xLBscKAn5GN7zWpiO9xM7nV0op3g+gcso5AtKOmyWD8Hf37BoMBfX19OHnypNybUCgkvBJ2iSIRiV0FmTaZm5uD0+mE1+sVQqHZbMb58+fRaDQwODiIer0uAw10swxG/0zFxWIxSaGwxC8QCGDfvn149dVX8cMf/lAiON1JMRaL4cCBA9LkQjsireiJzm+2g5AcdTMSp85H634YQHOLWKNxq1HM4OCgsLg5PpZ9MfheZFy/9dZbwjEwGo2IRCKCyqbTaSmP2rdvHyKRCADgjjvukOCNeko71QsLC/jSl76ETCaDzs5O6bNhs9nwK7/yK3jrrbfw5S9/GcAWaUzXuJtMJiEJ89zrvdf6kXfiVoO22z4tJIfpvK4mWZD4w8PcWiNpMBjwne98B/fee6/UDJrNZvT09EhNs9PpxOLiIqLRKN73vvfBaDRienoa5XIZ0WgUx48fx8zMDCKRiNDry+Uydu3ahXA4LNGy9pCY86AD4Pf7pb/v7Ows7rrrLiSTSQSDQWxsbEibPy4kFVEul8Pu3buxuroKr9cri63hHt1aVJcZtIu0IiVcG3rGmgmr66L5s9VqFf39/QgGgxIlA1sHPZ/PSxchpjuIlNhsNnR2dqLR2O4KV6lUMDc3J2SSWCwmjgAA6cFOKEt3gctms8hmsxgcHMTMzAxSqZTkvY4cOSJNMphTYp27wWAQYiPXge+njXSrp9xOjhrQTITTClgThm6msPn/RqOB0dFRXLp0CUajUUZKMudMxIpGOBaLyZmoVLbmsl+/fh1dXV3SXapcLqOzsxNra2uwWCzYtWsXgO2xiES42PQE2J5zv3fvXkmbVatVHDx4EC+++CKuX78uOo9nm2gMAGHt6tyzRmp0aWo7ib7PwHaJKXWAPhOaNKZz0DabDQsLCzh37pzYDfIQyDtgSpH9EVirvLCwgGq1KpB3o7HVCKurq0uQLb/fL3XpRD1IQK1Wq9IMKRQKyTAU2hR2sPvsZz+Lt956C8A2G52T8xh8eL1e4aboz6gdNp0K/rHUQRMeaKWNa+VM4pjePN1pioe6Wq0iEolI7RkT/kzcHzlyBMPDw0in0wiFQnC73chms7BarXj77bcRiUSQzWbx+uuvw+v1Ip1OI5fLSX0ix9CxBIA9v6lUbDYbOjo6MDMzg+7ubmSzWRgMBpw5c0bITboDEQly2WxW8hc09tw4vSa6oUs75qJ4CXXUDDQ3yQe2c9Hay+ZBpcN06dIlWX/md+lFz83Nye+vra3JqM/Z2Vk8+OCDcLlcCIVC2L9/PyKRCPx+f9NAdK0YtfdMmIqd53bt2oVisdjUzY6kFO4fc2FGo7Gp/I9pE/25+e/WEqN2k3+MKNYKbbYiLPrnSQpdW1sTiJDtW9lU6PLlyzAatxqMsJyKY119Pp9MPFtfX4fJtNVlbnh4WIwACWWNRkO6z9Eg1OtbjZG6u7uFxBqNRoUTA2xP6tPk1Ww2i9HRUeRyOXR0dDTxTVpTQfozt5NQx/OPllY0QZfcaga9wWCQIRl9fX3iBLGRVCgUahpWw2mBLL1yuVy4evUqMpkM1tfX8cYbb6BQKEh7T64769yJ0rEPA5+T0TrP4ebmJtbW1qTPhQ48iZbU63Xp766rGXTQQqHzcjt56Nsy0NoDBraHWvPNtQfBCJsKmr2xuRmsbWSLP3qZQ0ND8Hq9ePbZZ6XParlclglGuVwO73nPe2A2mzE/P489e/bgrbfeQiKRkAHuhCR5gUnF14tHb4zN1NnTl8qbRpibqQ/a4uKiDGjweDwS5WmCFKHQ1jzUu1140TQq0EoSoSLm3mujHQgE5GeLxaK08uP+kHl54cIFFItFRCIRgT7Z3Kavrw/lchmnTp1CoVBAX18ffD6fICnZbLYJvmIags9Ah5HetN/vR2dnJywWi3QrYiqHjhYvNs8PG1GwOQqjcZaQ3azcqt0iKKC59SvQPCBF34ebRQ40zjRuc3NzUl/a0dEhnebI7CVqxeYWvJsdHR0oFou4cOECDh8+jPHxcQQCAeGhsOyO0TlhWT4bdZfX6xWSUjwelzSLhiZJCrXb7VhaWpKzyPI7BiY8Y61QdzsZae51awMinX/WOftWp50ICQDpJnju3DkZsGSz2aRhVCKRwOXLlxGPxwUVA7bawdbrdTz00EMIh8OYmpqSkaQDAwPo6OiQLoN0xtPpdNOELd5zBnOpVArxeFy4UOwYqXPq2uHzer3yPeoJp9Mp76HJ1P//SHU3k9vOQfNDaU+YhptepfYeaMAJHfF1Go0Grly5Ih/S7XZjcXER77zzDjweD44cOSIRKvujMlf09ttv4+tf/7pEwGR8Xr9+HSaTCQsLC1JrTciEB0grCvbXJSuTk1LIzNN5S5ZiHThwoOkSUiGzTaXH42liCGv0oB1Et7TTtfD6+3S8SMSgsWS7Pq5HX18flpaWmuBDOksjIyNN02MMBoN0EpqenpY0BaHy1nVuNBoShfv9/qbPoD15Pi/bz9br2yMFW4lufH2fz9fUQY99vdkqkkM9ONe2FXFqB9ElJFTS2ojp9FKrYmZnMCq98fFxTE5OSm7X6XTK0INkMomVlRXRM2Tmb25uol6vIxgMIpVK4cSJE7j//vsBQCahUcnrznEkDOrnASAtZzkAgR2n6JTR0GjWLydreb1e0Y3aAWy9K+0mbDJCh0ffEe3UEGGiIeSa65a75XJZnCJdnplIJHD69GlBtojgZjIZlEol+P1+OBwOXL16FWfOnEFvby/C4bDoCj2z22azYWVlRf6t9TSRFqaxOLiJhlt/TopGf/i58vm8BBK0f2xrrFHEW0173nZyVEOE/LcuoXC73U0Ueh5O1gpz06ampnD06FExwkzgm0wmXLlyRZS8x+NBKBRCJBLB0NAQOjs7cc8992BjYwMABJokFG6xbM2SpnFlYxQeBKB5pnG9XkcymZROMmw9yeckO5jdyRqN7SHzPHz0srLZLMrlsuQrGcm3UwStS5T0eurvk5jVSpLQrGxCmFrhlUolpNNpiZAikYgweQk58cL09vZicnJSFPH6+jqq1ar06mWjG7L9GUnXajVJd1QqFSmRczqdUjJls9nE2PJy6YYWNptNDDibp1itVikbJI+CRBf2Em+ncwDcmHvW0SL3WcN6/B02d+B6srUjuwiSFR+PxyUiSSQScLlcoleYm4xEInjzzTdx8OBBOBwOhEIhCQoCgYCMKuVAHO6l1llEvGq1GjweDzo6OuB0OiWlBkCcAyrtlZUVjI6OyhAXvb8857p7Vjsaak2CZV63db95X+m46PPCzl0cGcyWn5zX0NHRgfX1ddRqNayurkp1DjkkHo9HnGh2pxwcHMTo6CgcDgdyuZx0i8vlcsjn81LCy3QXG4wwHUoonWgqAEFR9P4RSaOTwolabMhFMiurDWgndUXTrchtGWi+aGv9GBeW0B4jHj3Ng55xsVjEysoKjhw5gr/6q7+SvM/KygqGh4el4UixWEQymcTS0hJyuRzm5+dRqVRw7NgxfO1rX8NHPvIRdHV1SX0rc8aLi4uoVqsyHhCAwNXA9jAHHh5uMJWDxWKRVoEcL0fIa3R0FGtra5JvBiAbToeAURujJ/aUbiehMtIwVitKwAupIywqbrPZLF4uPdpSqSQ5J3rF7Cy1urp6A9nOYNgqjarVticQVatVLC0t4cSJEzh37hyuX78uz6Bhd50jIoQajUaljCMej4tTxmduddj067GVpB7kTkTF6/VKeoa9v9tFtMPB+8WvaUXWmoOn/jAYttp30qkpl8twuVzw+/24ePGiRBx0rjo6OsSI8r4tLy8LK3tgYADA1n3ftWsXTKatBkfaQWK1ALDtYOiI1+VyNSFmNptNHDLqGH5vYmJCzrOGenU0zb/5nu1UcqkRktbKHg3p8v/65zW6lkwmkUwmpQc/DfHS0pI40ZxiuGvXLplCyJnQ2WwWc3Nzwr5mB8uuri5MTEzA7Xajo6ND7BLQ3M2MYjRu9dFmVQhnROhKJO4zz044HBb2Ng02nUxC5mSYp1IpGZ97qwOUbjuCpsek2bdUxvROSfQiqYzwEEkUu3fvRiwWw/vf/35Eo1H4/X6srKygWCzi6NGjOHXqFPx+v3gkAHDlyhVYrVa88cYb+Ff/6l8B2Br/pz3j1dVVWCwWPP744/j0pz+NoaEhMdiaac6RhqVSCV6vFz09PfB4PAK7hcNhWK3WpjFobN6+trYmtZTcMHppACTa40VkZNAu0pp/bs1B8m96i4SRCX/RYCaTSWHizs3NIRaLybxm9lwm5MTpZUajET09PQJhkrhBctni4iJqtRp++Zd/GR/84AdRLpfxl3/5l/jud78rjhMNQ61Ww/Xr1/Gd73wH3/ve92C1WqWsjiUcdDhpWLWSaSWc0cnTI0dpzB0OhzT1byehYqay0oZIO62an8H/AxAeyH333YelpSVp9LG4uIhgMIjp6Wn09/cjlUohHA6Lc8xa5XA4jDNnzqCjowOjo6PSApg56aWlJamjTiQS0g62lXVPZcraWoPBAL/fj4WFBUmPsHsV99lkMokeaWVu877zrGn0iNF9O0grr4J6oPXrNMTAdnMbTRx0uVzYu3ev9DUYHBxEsViUUb4mk0nywDMzM9LQKJlMSjA2MTEhVRe1Wg1dXV2wWq340z/9U+m5cO+99+KOO+5omtFNY+vz+XDPPffg7rvvRj6fRywWA7DN8CcEzjaumj/Vqgfo1NPJZzkYAOmSR9j/R8ltnxYdPdLw0YCRaKHLWqjQE4kEdu/ejY2NDcTjcTz99NNwuVwYHR3F8PAwnnzySXzve9/DXXfdhYGBAemnuri4iH379uFjH/sYXnrpJdx5551YWFiQUqh8Po9AIIBSqYT3vOc9KJVKuH79Or72ta9hc3MT999/v/ThBbZr2E6fPo0777xTmmSwoUUgEBBHgmQCp9OJ6elppNNp+P3+psttMGyPqtSQPzffZDLd8ma8G4TKiZEE/9YX0mAwCNuah5NnYXx8XGDsRqOBgwcP4utf/zqOHDmCQCAgrNz5+XlEIhEUCgXMzc1hfHwcs7OzOH/+PMbHx+V16blPTk5Ko4rvfOc7Ant95CMfgc/nw8rKCjwejwx6n5qawvz8PMzmrdGGJ06ckI50+XwePT09ErnTEWNkphsQ8DJqyJSKmQ4H0NxVqB2kFa5vLSOhUdLOm270UiqV4PF4MDo6isnJSQwODqJQKGByclIM8D333IOzZ88iEAjA7/fj2rVr0kyoUqnghRdewPHjxzE6Oiq55gceeABvvvmmTMGy2WyIxWLCyo9Go3C73TCZTPD7/RgYGMDi4iIuX74svQ04HSkajWJ8fBzT09Pw+/3C4l9cXMTevXuRzWbR1dUFoJmV3uoAANsIQzudAZ1X1QaKNgHYWhcSLjVEbDQasba2BofDgc7OTly8eBGpVApPPfUUzpw5g9XVVdEHDO7YWndxcRHj4+MYHBzE5OQkRkdHEQwGxWkeGRmBybTV2OZTn/oUTKatfuz/5//8H1SrVUxMTIjNAiDIzBtvvAGHw4Fdu3ahVCohn88LgsKSKt5trRe0jtPsdGDbqWc0DWyf/1uR2zLQhB648PQYefgNBoMMzubDMxJlf2T2Wf2N3/gNOBwOxGIx5HI5vP7669KY5MiRIzh16hQmJibwoQ99CJ/73Odw3333IZPJ4LXXXkMwGITNZgOwlYNmW7h4PI533nkHVqsVv/7rv46ZmRm54JTR0VHEYjH09vaiXC7Lc9Hbt9vtSCaT6OzslJrLeDyOAwcOyM+ydo4sUwrXgUaD69M6Q/vdLJpE1dqIhgQPAE1lCPzbbDZLpGI0brVxjEajOHbsGEqlEmZmZtDV1SUozOLiIiYmJrC+vo7r16+jr69Pom4yMxuNBqanp5HL5XDo0CEpjfrUpz6FRCKBz3zmMxgcHGwqhVhbW8OlS5cwPDyMD3zgA5icnMTly5cFKiXBiURBGiOPx9M0+5lCiAvYjpo0YsCLSyexHYR6gJ9PkwU1W5XrpCPrM2fO4KMf/SiWl5eFsFOv13HmzBn09/ejt7cXsVgMo6OjwpCPRCKSzujo6MCFCxcwPj4uZ6K7u1uMPUfRXrt2DR0dHdi7dy9eeeUVHD58WGqgrVYrjh49iosXL+LVV1+F1WpFT08PFhcXpV0nSWZ2u10+I6sMjh07Jndff0ad8tHpFL1m7SStuo6OiG6py/Umwsp5y3a7Hbt378bm5iaOHj2KYrGIr3zlK9izZw9GRkYEMaEjQMft0KFDmJ6elvvPfSG5NJ1OS6kVbUsoFMKHPvQh6fnOvHgkEkF/fz++8pWv4MSJE+jq6sLw8HCTDiiVSk0VIPybI1C1nmt1wHSprob4qY9+lNz2PGiDwdBUoK0hI16gcrksOVjWC/+bf/NvZNrIzMwMLl26hJMnT6JUKsFkMuHAgQPIZDJS9D84OIhz587h/Pnz+Pmf/3ksLi7i4YcfxuHDh6WZvc1mw5kzZ+D1ejE9PY0rV65ID+ZPf/rTSCaTAoWWSiUcOnQIX/ziFxGLxdDd3Y2RkRHJcROKdDgcyGazqFQq4gSk02mJrLUyZqkHc9kscNc5T814bwehwqLoUir+TZRFlxk0Gg2cOXMGV65cwWOPPYa1tTVJZXD6GfkIHo9HauLn5+fFMEajUVSrVUSjUalzX1xchNfrxS/8wi8IVNbT04PXXnsNFosFDz/8sERXwJZC6e3txcc//nHs2bMHV65cafKAeb5Z8wo0N+LRzHV+dt17XRPn9B969O0iOsfYSnhpJQ9SeRuNRhkoEY/H4fP5sHv3bqytreHcuXPYt28farUarl69CpfLJU4/dYzf78fa2poQfngu2HudE6dSqRRCoRCSySQCgQAeeeQRafUJbOWoWS8/OTkpKZU9e/YIo1jrMJ/Ph3K5LMgQ77yusaWh0OU2rQ4LOQ/tIqVSSRp2aGdEO6k01CSO0qG79957ceedd+L06dNwuVwIBoO4ePEixsbGkM/nsbCwIA4yU0xswZrJZBAOhxEKhXDPPfdgdHQUFosFQ0NDuHr1Kr70pS9hfX0d0WgUJpMJjzzyCI4cOYKrV68il8uhXq8LurK8vIz/8B/+AxYXF3HkyBEMDAwIukc+CfeTxpa2wePxNPVZ0JUt/LdG13gGOBjoVuSf1KhE51O010AymMm01U2sr69Pcnl/+7d/i8OHD+PUqVMol8vo6urCwMCAMDoTiQRGRkawvr4uOd2hoSHp7uT1evH888/jzJkzMJvNKBaLmJ+fRzAYxCuvvIJjx47h9ddfF9LW/v37kcvlpM2jw+HA9evX8ZGPfEQgrVdffRWbm5solUpS3lOtVqUzEdl+BoNBxmwyp8rPTPYnLyYZpqyx5Wu3i/Cz3qwTjmZw8mzQsRkcHMQv//IvI5VK4fd///dRrVbxi7/4i0L+MBqNTY0J2CifbHmDwSCQ6J49e2QOLJVrLBbDxMQElpaWEIvF8OKLL+KVV16By+US8l+5XIbH40EkEpHJR1euXMHMzIyQSAiThUKhpjIZu93eNFaS+03FreFb7bQajUaEw2Hs3r1bykraQXRE0JryoXJuJZGFw2FcuXJFmgzdfffdeOuttzA1NYVQKITFxUXU63Vh31erVTkLsVhM8r6XLl3C4OAghoeHxbm/cOECZmZmYDab0d/fj0qlgv7+foyMjGBqakqYvaxj7uzsxKlTp4S9f/z4cSwtLUl/A4fDgfX1dXg8HunT0Gg0ZGAPm2AAzY2a9OfVa8Je0u3ER2H0rDkmWge0svoNBgM6Oztx/fp1fOELX0AqlcIf//Efo16v44//+I9l3kK5XIbf75eIm4gl+6eXy2UZKbq4uIhMJoOxsTHE43HMzs4iEokIqrJr1y584hOfwH333SflmLqT2MrKCsbGxuBwOHD8+HH09fVhY2MDhUJB0lSNRqOJP6LTFtxzzeLnHdD1z3y/WCyGa9euIZ1O39Ia3xbETSiBNYhUQPqA6mhlcXERhUIBR44cwcbGBnp7e/Hiiy/igQcekLFgrFe7++678frrr2Pv3r3CnuOf8+fP4/jx4zh06BCuXr0Km82GpaUl6fwyNjaGd955B4899hicTieGh4fR39+PtbW1plxlIpFAqVTC6OgoAoEA5ufnZbwhKf0DAwN4++23EQwGpV1cIBCQtqB6MAY/K0lKLLdhzp0Et1uFM94NQi+Sjpg+B9qTZN49mUzioYcewtTUFL797W8jl8vhwIEDWF9fx9WrV7Fv3z7EYjHk83n4/X4sLi4C2PLO2eAAAFwuFzo6OqSZ/uLiIuLxuLB12bxgYGBAevvSUyWhg/XJ3Ber1YrDhw+jUqnI6zJ60hB+o7FVU80IWEdNbHBDUhSVEcuDuru74XA4kM/nkclkfqJ79eOWVoRIw/48D9pIX79+Xe5evV7Hd77zHZw9exaDg4Po7+9HIpGQXCOjTRpnku5IKiXsSSO/vLwMv9+Pzc1NDAwMIJ/PY2xsDBsbG8jn8+jq6pKpdkzFsQPZ0NAQotEoFhcX0dHRgXw+L4hNZ2cnLl26JCma9fV1HDx4sGnPgW1mskYTuC5sluRwONoKRaETznuqkSYaaq4378SLL76I3bt341d+5VdgMpnw+c9/HqVSCY899hguXryIhYUFgY5pGzhilGcpk8mgWCzi0KFDCAaDGBwcRKlUwqVLl+ByuYTwm8vlUC6X8fzzz8Pj8aC3txf5fF720mw2w+fzSdrq7bffRiqVkmZKJHmxsx0n2LHfBfPr2jlvLeMj4lqpVLC2tgYAGBoauuW0521F0GQ1s8aTm0FPQjelYCOJoaEh1OtbvUovXryIrq4uXL16FY8++ijuvvtumM1mRCIRfO1rX8PAwIDUMWtI6JFHHsHCwgLOnDmDI0eO4J133pHB3lwYRqtzc3OYnZ1FMpmUjlL0bjghJ5lMYnp6GsViUYbDp9NpyTVRYbtcLqyvr2NsbEyUKw0S4RdG05o8xdmlrbmHdhFG0K3KiHk6XWNuMBjwwgsvwGg04uGHH8b+/fsRDAbxyCOPIBwO4+TJk8jlcggEAohGoxgbGxOeAoeeAFtOjt1uRzgcxurqKubn56WLE5ESYNurZVkWHQqeV9Yt09lknSqVDI09c9lUsjQIbMCiFbIuvyAUBkDmGXMcXjvVwPKzEDFqLbNrPfecWNXd3Y1EIoFcLofr16+js7NTIlveYY1U0VEDtnQNx5QODw/jjjvuwNTUlETOdBZqtZqUupAYSKVMZZzJZMRxIys4GAxKowqidCQIcZ/ZRIPlWBQNaxPaBbbuRSqVQjAYFGetXUSXGfL/QHNnQd3q2W6344Mf/CB27dqFr371q/jBD36A2dlZnD17FrlcDvfff78Yvkqlgng8jkKhIKkGjUYMDQ1JlOzxePDyyy/D5XLB5/NJqpIpEN5Z/Yd6ijaNZyEUCskeEi2mw67PGMfhtqIHrRwUfm1lZQU+nw+jo6MIh8O3bKBvK4LmAzE6aDQaUv6iFTI968uXL8PhcAirLpFIYM+ePdi7dy9effVVVKtbDelTqRQ6OjpQKpUkj0w4sVQqIZFIYHx8HL29vfj85z+Pxx9/XKBlQkl+v1/qnY1Go1wu5hN0yRZbvtG7yWazMvhCd4nimEvdLEOXGTHnzgNFiLxQKMj4uXabB62nddGoaZKYhjfZ57Zer+Oll16S/ucWi6WpfhWAcAl4Lj7+8Y/jnXfekb3ggAQa5+7u7qYcn84ZApDf052L2KmKe6QhLCoR7iE/K9+DrRy1sdcKSX8WVjOQrakZze0iPO+aJAY09+bm2rEVJnvZBwIBXLlyRUpRAoGAkIEYlRIO37t3L0qlElKpFCqVCiYmJlAoFDA6OopQKIQ///M/x/j4eBNRB9jaExpmtvzVXcWIpGiWud1ulwEN1B0cTWkwbI1G7enpQalUQiAQ+EdLjDShaH19XXrEa5SpXYQOKYdbtKIIXNtcLid7cuTIEYTDYck/33333cKarlar6O7uBrDFrjabzfj2t7/dZDjZh31kZAS9vb34+7//e2kwpYnJ2kgyTclnpi7nM3Pghm4BrXVCJpMRY02UTPNrgOazr4mT6XQavb296OjoEOftVlMdtxVBM3zX8zS5CVyISqUieZ09e/ZgdHQU0WgU8/PzGBkZwbVr14R17fV6ZbhFX18fVlZW8Pjjj+PKlSvo6uqCzWZDNBrFhQsX0Gg08PTTT+Oee+4R75pGl00q2ImMX6eXQuYm4UuOONRt3HiwarWafD+VSqGzsxOJREI8O91gn79DZVWv12VcGTdJb1y7SLlclkhFGzM6I/yTSqWQy+WEzHHnnXfC5/MhHo8LwcTlcsHhcKC7u1vqnL1eryAX7MJjsVgQDAaRSCTEILbW3rO+kISUzc1NZDIZ5PN5ITYWi0U5F7pckM1pGGm5XC7xdMk+595rQpCOlvhv5rIZnRO9uVne/t0qVMoUnXvWhpLOvN1ul45vGxsbqNW2ugtSaS8vL0vKivv3zjvvoFKpYGRkBNVqFbOzs3C73Thw4ACeeOIJfO5zn7tpq0k+A5EU7gMjO7N5a6Qh01Kt5EbNOC8UCtLKc319XeYDE20DbmRx8/NzyALblvI52kVaUSsATfAxz4TBYEBXV5cMJfryl7+Ml156CdFoFOFwWIKh9fV1FAoFrK+vo1QqYWpqClevXsXIyEgTR4itnXt6evDCCy+IQeae04Dz6xxBzMErNN4acWGwQURFN17J5/NIJBKSC2cKppWxras7iMbxjuj0GA3/rchtk8R46BmhtD4kc7D5fB4rKytYWlrCyMgIgsEgVldXEQwGhf5OZUZI2OfzYWNjA0899RR++MMfIhqN4ld/9Vdxzz33oL+/X2AGdigrlUoSAREm43MyL0wPyWKxNDFDSRqy2WzSfxuAzIwOBoOoVqvo7e1tgrf1dKPWpgOFQkFaTfKiaw+9HYQGiwdPO2fANuzldDoRCoXQ19eHqakpxONxrK6uYmNjQ0heY2NjkmJoNLZGxXV3d2PPnj2Ix+PI5/OIRCLS2m9ubg5zc3NC5NFeKi8NHTEKHUntUDAFQ3iWnczoqLEPLxnmhMP5rLx8GsrTbG1ecCI8PHPtpJxp1LRzws+oy4qYvyd0zPwuSXis4kgkEnA4HPKajHi5fx6PB9FoFKurq3jqqafw1a9+FcB2H2ciFJq01GpA6VgaDAaJ6PTPOZ1O5HI5yYMzfaJTFbrEU+dctZHmM29sbKC/vx/ANtTbTrqAiCqAJl2nU51kTCeTSUxOTiISiaCvrw/d3d04cuQIZmdnsba2huXlZWxubkokvLKygpWVFeGkDA8PIxqNynscO3YMxWIR0Wi0CSVjMMYAjraKOob7pOFo3mXNp2KkTNtCuNxk2moxSj14s0oGnjE6msFgsKkO3GQy3XLTotuOoCk610wFyAtptVoRDodhMplw9OhR1Ot1rKyswOVyIZ1OY2NjAx0dHVIqRc/EbrdjeHgYX/ziF3H06FH4/X781V/9FUZGRvB3f/d36O7ubprDyQ/L99YwmdFobGJw8+AwmtIkL6/XK/1aOfC7UChgZWVFDH9rDpGbx0vHCJIkOr357SY6v0Tnh14zYctoNCpsSCISNptNCHckGnq9Xok4c7kcwuEwFhYWcOrUKbzxxhsAIBNorl27Bp/PJwZQ5z51+QoNLy8IX4OOhcHQ3DiAbTj5PXb7IYxJToS+7Pq9+Zn5b0KrRHEIwbcTksKzrZUegCYFSEe9s7MTfr8fk5OTCIVCaDQaUoHhcrng9XrR19cnFQ9sHnP06FEcOHAA169fx9tvv4177rkHn/zkJzE9PY3nn39eKjb0+wLb/Ql0R796vS53mHAn86I00F6vF0tLS9KWuFqtYnV1Fb29vdjY2EB3d7ecWZ1/5OflugDA0tISBgYG5Gf5Xu3USQxonlzFvC3Xhn+8Xi8GBwcRCoWkO1uxWMTMzAxGRkbg9/ul4QvJlCQCHj9+HA8//HDT2u3btw8mkwnf//73RQfrdJcmaGkSH5FX2gDaDRp0HXBq+JpcFb4en6X1rHMtGMRx6hknsBFNYo+IW5Hb7sXNh2kdzk3vw2w2Y3p6Wjybc+fOCePx7rvvFthwdXVVfj8YDGJqagoejwczMzN49NFH8dWvfhWpVApPPPEE/uAP/gB79+4VI0vvnZ4sFSC9JMJW9fpW7eTly5dlk9g+kq0dL1++LC37+BmtVqt0tMpkMnJZdbcoOiPcSDZTIJtYwz7tpJgBNKUGeBE0QczlcqG7u1tINSx1SqVSku/hz3JcJwCsrq7ie9/7nuxRJBLB3XffLU0M2FaR+6/LnfQ5BLZ7ZhNapSeezWaxsrKCQqEgl45lXnw9YGsmLI0x2ZyanarPALB9N+gpawcOuHEk57tdqJBa68L5PaZ+mNfj3WMqwuVyIZlMNhH8yFsAtpz1iYkJ/OAHP8C5c+fQ09ODcDiMffv24ZlnnmniwQDNzHpNXuL/2e2PEZE2IMB2j36mUFgHv7q6Co/Hg9XVVezatUv2WH/m1uiZ6RU6IzTOeo3aQaiPWwMSriUdNPYtsNlsiMfjmJ+fl6CIKQreR92Y5PHHHwcAXLx4EVeuXMHk5KQgpySe0mgTNeO+l8tlGY7D4TuFQgHlclmGKjFtxlG1fD3WKJMoyulqJK/q9E1rvhmApLdo2wjzs6LoZqTKf3SNb2dDmOvT4T+/zgfo7e2F2+0WRRkMBrGwsACTyYQvf/nLKBaLMh5sY2MDFy5cwJ49e1CpVNDZ2Ynl5WU4nU78zu/8jlzk9773vQJ36e4xzE3qD0tDmsvlBNL+zd/8TbhcLvT39zfBnHxerWgJgzF3mkwmm6IFRuyEO3RNNBUBN4Ab2G7SaDSEHdta70liTDKZlJzR5uYmotGooBE+nw/5fB6pVErmN+/fvx9WqxW7d++Gy+XC0aNHYTAYcOLECbz++utYXl4WmJQQOrBd6qE9aV4UNpMIhUIYGRkBsH15aDB5mZmDNBqNUldNEokupdAGgO9L5aBzr9phoGFvJ+WszzcRKp51ImtUwNlsVqZXBQIBXLt2DcPDwzKhLJFIIBQKYX19Xao+2J43Fothfn4eExMTeOCBB/CVr3yliTnPe0jREDuwrUSNRiP8fj8SiYT8DBsSVatVdHZ24vz5802Evs7OTqysrAjRp6ur6wayn47c+CebzUoESH5KK4mtnYQOidajwNbajIyMoKenR5xzVmKw5HVjY0OazTidTkFPyuUyXnzxRSSTSbz11lswGo34tV/7Nbzvfe+D0WiUunhgm8ypnTJtI2ikuRcLCwuoVCooFosyCZEpTo2U6jQanXldxcG91H/re85IW5eh8mzeaqrjthMiJFIQLgCap9MsLi421SMyJ/fAAw9g//79UkbhcDgEBvqzP/szHD9+HC+++CIMBgOuX7+OyclJfOITn8C3vvUt+P1+YeA5HI4boGZNFGE0ZjRuTUFaX1/Ht771LZw7d07a+LFhxZkzZyQq4yEjGSGTyWB0dFS8L14wDe2zFrhW25qwQ8eF7NBWg9EOQjifn11HD3Rg2O+6VqvB7/cjm83KWbh27RpMJpMMYT9w4IAMYWeu78iRI9i3bx8GBwdlTUkmAbaZo3SQqPh4Lmgs9OhPl8uFO++8E/fdd58gIoSduEckmFitVqRSKXFEdL0/oyQNbzO6JqICbENqPDM3I5W824W5RqYPgO1zQMeJk6ScTicqlQqi0Sj6+vqwsLAAg8GAcDgsqY21tTUMDg4K9+DMmTOwWq34pV/6JSQSCQwMDAjSBjTX22pHSEfO1FNms1nGDdrtdoyMjDSRnNh2mChgvV6Xkr5GoyGpFU0wvFlEzPw2OxjqM8Hvt4vUajUhVOqBEMA2wsIeB0QZnU6ndAQ8ePAgTCaTMKSj0SgASL+DQCCAH/zgB5ibm0NXV5es9csvvwy32w0AMlQHQBPRj8ZbkwUbjYaMKWZUzi6SvN/r6+tNUS71D50sVvsAzeRAYLuRF99TP1cr4vxjiaCB5hxbq7Kq1+sCJbA2bWNjA263G2+99RYsFguOHj0Kk8mEI0eOwO/3Y/fu3aKs+/r6kMlkMDQ0hFKphOeffx579uwR74yzNukRaQ9KK2sAUr7BMjCWT9HbCgaDCIfDqFarAmHybxqGeDwuMAehFw2r8HdIdmslgbSSUNpBqHCz2azsf2sEw4NttVqRyWTQ3d2NdDotTQA2NjZkstDly5dx4MABTE9Pw2AwSL6X9a35fB6Li4vS+IR7TQeAlwDYUgockkAkgzXI/JNIJJrIK5zxza8xZ86LRuKS9o65zzxvNPR0GrSHrf/dbhG0wWAQ3oUWKih+/lKpBIfDIX93dXVJ0wcASKVSWF1dxdjYmMzxfuedd9DV1YUPfehDWF5exoc//GH8xV/8hTi/JPaxHpkKVkdSfE6mYQyGrdGE7PfNuxsOh3H58mX5PavVKnXRGxsbWFtbw/j4uCB2POO6kgPYNr7lclkcEq5NaxqgHYQRL5GSVrifesFoNAobfmlpCSaTJYF2cgABAABJREFUCSMjI7hw4QKCwSB6e3tht9uxd+9eFItFhMNh+Hw+TE9PY2xsDL/4i7+Ivr4+eDwerKysyNx2lrhqB1pXZuhSKIvFAr/fj0AgAK/Xi/3792N8fBzAdoMRfU85mMNoNApC5vP5bkhbau6BTv2Sk6Md81bS4q3IbZPEWJ5Eb0VDvHzgRqMhnX3IvmPhOTsB9fT0iHdL5Tg+Po4nn3wSp0+fxi/8wi9IrSQVQaPRgN/vb4KXdd6RyXdGuEzIc3D7/Py8dAc7efKk1EKXy2VhkZbLZQSDQSnxoQHSEA4bWWgImwpI56e4ce3kNQM3dk3iBdEGiUxr5pxpVJ1OJ/r6+qQs4sSJE3Kon3zySYyOjmJ+fh7Ly8uw2WzSApbws9PpFAiM6Qh9SUwmk5RwFYtF1GpbXc2uX7+OmZkZzM3NIZfLSX6K+8fPwrI9XdevlSs/H7A9T5xKWHM0WmGvdjsHrHzQ0DYVDz8zsNX1qb+/XxqD9Pb24vr16xL5ML1w9epV3HfffXjjjTeQTqcxMTGB/fv3Y2FhAWNjYzIykE4R6+JpTLVy1dETUxiM9tlfQRvTjo4OzMzMANjup07nr1QqCXu7Fb5tdbyJ4hCCNxgM4rjr89QuwjNAW6Dhfv03U5WZTAYDAwNSBw9szfNeXV1FrVZDb28vAoEAkskkTpw4Aa/Xi0OHDmFoaEi4Cm+88QZKpZIYTpZO6bIoYBtlA7bLwGh4BwYGsHv3bmHkU5cvLy8LKgBsGXlOTtM8JQ1XawicDh6DCJ5D2gQdzN7qObgtSqGGtzWESENZq9WQyWSwb98+iSxKpRL8fr94qTR2y8vL+PjHP44PfvCD+N3f/V2srKwglUrhxRdfxKc//Wn8j//xP3DPPfdgYWEBgUBA6md5aVpzn+VyWbrIRKNRyR9arVZ4vV6MjY3BarXi/PnzGBkZEYY2L7lWwOwJzfIvQmSMlLk5vJClUklIBISA2+kitgpRB92BR6c8ODS9WCzC6XRKoxeSRHp6eqTs7v7778fU1BTuuusunD9/HoFAQAYpsDfv+vo6Go2GkA3ZlITvy30jDMWyDZZGpNNpyWtWKhVEIhF4vV4kk0m5nNxf1sFvbGw0NeFpVcg8d5oopMkx+iK2E4JCac03As2GmXnboaEhzM/PY2hoCH6/HzMzM1Lz2tvbK1PJWEHBwTQPPvgguru78fWvfx3ve9/78Nd//ddyhyuVCtxuNwqFwg298KmbaIjpGJC1bzKZsLy8LPvV1dWFyclJ0VfUY8wfsmzH4/HIWWBQwffjfusonhGePiPacWsH4WdkHl+znvlZWQbLaJbVHHSEiWCl02kpv3vjjTdw3333oaenB4ODg3jttdcwMjKCc+fOYWhoCJlMRvL7XFdgG+Lms1WrW73c2SfDZDIJQZj6olwuI5VKiV0iFM69ppNPpK41IAC2dYH+vi7Fbf2523HWb8tAkymtae3AdielTCaDQ4cO4c0338QjjzyCd955R6KR1dVV2Zz5+XkEAgH8/d//PX7zN38T5XIZg4OD+M53voP7778ff//3f4/+/n6pn6Q3RJIYqe6tSmJqakrKOiwWCy5cuIBarYZgMCg5jbGxMTz33HPo7e2VUovp6WkEg0FhejLC0q0/6ZwQDqGTQO+Yl1t785o00y5CBUiIV0O7hJISiYS0tsvlchgaGsLi4mKTkWbTh6tXr+LQoUPIZDK4du2akEqy2awMwiDsCEAcL0KndBQofr8fx48fl85ERqMRfX19MqectdgLCwsAIOhMuVxGOBwWI8OLpZtXcB9bS7wANF3qVi+ZiltzJ9pBNJrWaoAKhYLsExsOdXR0IBqNCrO+q6sLi4uLcLlc6Ovrw6lTp3Dw4EGsrKzgrrvuwuc+9zl89KMfxde+9jUhFnENNXFPN8ihaMiTRpwRP+Fxg8GAwcFBfOUrXxFWOZ+ZPfoZ4etRu/rzthIDSTZiSqz1LLSTLmCgwruvo0byesxmM4aHh7G6ugqbzYb19XVpFlIsFtHT0wO3242NjQ2EQiFcvnwZo6Oj2L9/PzweD1544QX09/djdnYWwHaKjfqWqS+XyyXrTYc6n89LZ0KSSldXV4VA3N3djXA4jF27duHChQvy3DabDYVCQRwvPReglWuh0TX+WyOK2mHQZ+BWy29vy0Cz/R09DyovYGvSSHd3N2ZmZnDXXXfh0qVLmJmZEQKA3++Hx+PBK6+8gn/37/4dnn32WXn4bDaLw4cPo7e3F+FwGMlkUowxa15J1OKFZwTPRSTM4fP5pKRh//798vV8Pi8KOxQKwePxyND43t5eITCwUQEvNL1gRs/6363kJCrtVuPcTpcS2I6eGDHzMNZqNWSzWaytrQl8xPpRu90uF7Onpwf9/f04f/487rrrLplW1tfXJ+1Vp6en0dHR0VQnD0BaLzKSZmREdnYsFpOzwsjsypUrEvVzkhLPDgljZBQzzVGv14XQyH3l5dfKmXtP504bKXrSdGTaKZJmhMK72NqowWKxoLe3F9///vfx8MMPY2pqSkb+eb1elEolRCIRrKyswGw2IxaLwe12Cw/lm9/8Jn7+538er732muyzJnVp9j4ZuoxsSEhieR/TU8CWA0fHeu/evTh16pTcc7YAZWkY0RBGUK1te6lotfJlAEM0RzsuJtN2Z6l2EeZddVqB+i8UCqGjowNXrlyRHhaVSkWm+5nNZiQSCWxsbKCvrw/T09Po7e2Fz+dDKpWSSVM2mw1TU1NNjbK4LwwK6CxQL9lsNvj9fkk9Mu12zz33IBAIwGg04tKlS4LY6bpoEt80fE/Oi47YW4mCPAutOWhgG+7XA4BuRW7LQNNrSaVS0mWJD2m320VxxmIx9PX14bHHHsPXv/51PPTQQ7h48SIOHDiASqWCV155BUajER0dHbDb7ZiamsKFCxcwODiI6elpmWZCT0l7HFSU9HIIOTOiZccnq9WK5eVlmSQUiURw8OBBnDt3Di6XC/l8Hl6vt6nVHzsI5XK5pvw0iSFUsjqa1mVXnLiioy0AbRU5EUHo7u6WKT2auMMJQZwWZDKZZIoLsJVz6u3tRTwex969e3Hx4kU4nU5sbGxgYGAAc3NzWFlZQSAQkLyT0WhsatHJiJr5wUAgILAkc1N0CDo6OoQgmEwmEY1Gkc1mm2r3mddkuoYwF+FvOlkaPeI55NnRBBVgO/9pMBik5rKdzoHBYMDGxgbGx8eFFMh10lEzB5psbGyII+T1erGxsSHK3e/3i2Ek0vLoo4/iG9/4BhKJBIBt0pFmUdNhrtW2ht2wcQkjOSpdRlT6ayMjI7h+/Trm5uYkd8he4RquJFpCwhCfpXUteOf5GTXpiEGEJrG2g1DvsluWLkFqNBpS5x4KhcSJGh0dldIqk8kkbX03NjYwNDQkrTaz2SzGxsaQSqVw5swZISYzGtb6lzlupiTonOt2o+VyGel0WnLfXV1diEQiKBQKWFxclLvKigSigqVSCcDWWdDOOrDd24COgU610IkEmh05lgnfqqN2WySxYrGIUCgkCpMKijk8Kim/34/5+XmcOXMGvb29yOVysNvtGBwcRE9PjzD6QqEQlpaWcOTIEezduxfxeBzBYBCbm5vi1RqNRoGa6/W6FLi7XC4MDAw0XQIaW+aFWULl8/mQTqfxzDPPNPVppYFhqQwXnMqmlWzSmmNzOBxCGgO2YTUqDV7Gdiqv4TozD0Qlye8xkiGBxGKxyGQqKrrp6WlcunQJwFYP9r6+PlitVrz88ssy+QdAU30tISqDwSAj4kKhEPL5PObm5pBMJlEqlZoMLGHQaDSKZDIpl4xOBrBd61ir1aREr9FoNCkBGmPN3qVoRUGSova0eZ7YzKFdRKMn+nzr9SHKYTab0dfXh2g0CqfTiXw+L+WOuVwOwHYLWbK8L1++LGU3fF2eBR0Y0AFivln3iecfvj4HdoRCIQwPD2N6elp+nqRGMsOpA9jsRuedgWaomgaJfSLoJGhDf7O1ercL0VQaYE2k1Y4RpVarYXV1VSotzGYz3G43xsbGxDliTfJ73vMelMtlnD17FkajUcZEUrewpIkNTvh7gUAATqezCe1k6snj8WBiYgITExPweDwoFAqYnp6WPaNuYFDIc6GNPY2udkS0webz0Sbw93QnMTZ5uhW57XGTOuKk8uGhY0lSIBCQWlgebLfbjTfffBMA8J73vAe5XA6vvPIKKpUKrl27hvX1dckpsHaV4yIzmYx4Z1TC+Xwe5XIZAwMDwhBkZyCXyyXlFzSyFotFfqdUKqFer8uUGt1YnbAII0N6xMC2F68VL+F0Kg3+HGEYKoZ2EUK/msGroX1GHC6XSy4Pc3rV6tYozoMHD8JsNuP8+fNSnzoxMYE9e/Y0jRbULUR5GWjoiGSMjY3BYrFgcXER8/PziMViSCQSSKfTWF5eFqWdyWRk5jOjGxpWRmaMdFjvSoXNOu3WfFNrDlrXPeq1IfT+3HPP/T/Zsx+H0GnK5/NNjmsr74KwLhtDEHljJMQyJsLenHZ27tw5cfp5rnRUQqVJB5uIVzKZxObmpugJDS1qqHltbU3uqXYy+bNEz+gk8lzz7LWWFFE4OpWICZ9fn612ETrBQLPzoWFurhsddD26EdjiDUWjUQwMDADYOi/r6+t46aWXcOnSJbmnelAT8/waQeGku7W1tabeB5wN3dvbi87OTulSGAwGxVHQz06HnU4W06zcQ9o1fc5bf5d6i1Ez9YwOHFji9aPkn5SD9ng8TQXZfCiO1fqv//W/4ud+7ufkYZaXlwEAu3btwve//31EIhFJzs/Pz+PixYt4/vnn8Yd/+Ic4ffq0lELpvB9ho2w2Kwb12rVr8Hg8OHjwoIwfZCTFQ8PIyGg04oEHHpA6bQBiNCqVilwsen+M+shSZAtPRtIkgVQqFYHkdWTGqJwj09pFGB3SaSIJkJcEAMbHx6XMaWpqSqIRYOtSW61WHDhwAGfOnJESvHK5jP7+fgwPD+Mb3/gGLBaLeLTANnNc9zzP5XJoNBo4duyYsGwrlYqURugSF6/XC6/Xi2AwiJdeeklIILrHLvdPl2dwfjibLZjNW5OtGC3riNHtdsvvEvHJZrM4e/Yszp07h0Ag8JPerh+r0AHZ3NyE3+8H0MxonZ6exuHDh8Uw6r7s7OLFGniz2SypJQ6ZuHLlCjY3N2G324UdTMIOnWY64CQU8tzxXALb5U/cq0KhgHQ6DafTKUQ3fhb+m05bvV6H3W4XHaGbIhFm53sAwMbGBo4dO9ZEpqSz2dnZidXV1Z/kFv1YhevD5j4a6qf+u3DhgjQh6u7ulnyxvk+Li4twOBwyECmZTCIej2NqakrSonwfBnx0/Gl/XC6XjCu+fPmyGOdwOCzBF520bDaLQqGA69evC/LFdBY/CxERps1YogVAJtsRzSFsrVMZPCf8N7BlP4eGhrB371688sort7TGt2U5Njc34fF4xFhpKRaLYhj/9b/+1zhz5ozUmK2urmJgYAAWiwVDQ0OYmZnBvffei5WVFRw7dgxra2v4uZ/7Ofzbf/tv8Yd/+IcCPfEy6qED7NVrMBgQiUSQyWRw/vx5uZwaXqIhTaVSSCQS+PKXv4yPfexj+Pmf/3mpvyNMzdePx+MYGRnB7OwsHA6HkM4YkddqNeRyOanDNRgM6OjokCb/9Kg4EevSpUttFUH7/X5pNp/NZmUMJwAxqqdPn5ZcI7Dt6bKV64kTJ3D8+HGk02np6Ga327G+vo5KpYKPfOQj+OY3vykEIBL6NGRG5IMRLh0vg8Eghp2wWC6Xk2h+dHQUf/Inf4KFhQV8/vOfl8YlZOCTse90OqUtYTgcRjqdljOeSqWkVMTpdAqspfPVzI1tbGzAYDDgve99L9xut4xXfbcLOSC6FEVDuouLi3jggQdw+fJlHD16FGfPnoXP5xMjHQgEcPnyZUQiEUxNTUnlRC6Xw8LCAsLhMEZHR3HmzBkEg0HE43GZGa8jVxpAlmB2dHQIxMn7yvQVkR9GMuPj4ygWi9jY2JDpejSs2tiwPIyOKM8l/61zjMViEYFAoIlYSOOcyWTw0ksv/b/ctn9WyWQyUi7F3C33plqtYn5+Ho899hicTicuXbqETCYjjhA5KhxBvL6+jmq1iiNHjogxtdlsOHXqFA4cONBk/ADcEBCxlWxvby/MZjM6OzubSmSZ2qRNYJDx5ptvwuPx4N5774Xf7xfUgzA0nYNarYa5uTnUalud89iRzul0CuLK0lK2LtW5ab7/ysoK/vqv/xrz8/O3tMa3BXGbTCZ5MN0Dl9DF8ePHcfnyZYyMjKCvrw8mkwmjo6OwWq1YW1tDV1cX8vk8nE4nnn76aVy+fBnLy8sYHh7GyZMn8Qd/8Af44z/+YwwMDKBW226fSe+Gm8HFMBq3mpEzP6mjVwAS+VqtVnR3d+NTn/oU7HY7PvzhD+OZZ57Bo48+KkaVERQnbLGRxdLSkswnZQ6ho6MDXV1d6OzslEHi2mExGAzo7u7G2toaOjo65FC1gwwMDKCzs1NIcSwp4B6tra1h165d2LNnj0DNnGJTrVaxf/9+nDlzBolEAqlUCouLi9izZw88Hg86OjqQSqXw0ksvYWRkRKBR5hJ1+z3mjKrVKrLZrDhImmXN5ih0sqrVKk6cOIHPfOYzOHv2LD7zmc/gP/2n/wRgKwIjClAsFuV8MZoKh8MYGBjArl270NPTA2Cr+oARNodx0Kvf3NzE6uoqrl+/jv7+fgQCAUEY2kG8Xq/0m04mk00DU0qlEkZGRjA1NYWOjg5cvXpV4OKVlRWYTCYcOHAAV65cEX1w5MgRmQtNJ6fRaKCvrw/ANqQNoCnS1Y0l6JwB2/2ZqQuoMwhpktDa39+PvXv3SlkPERuNCNFwMxJjmsto3OqZUCwWZcACIX/NY+nr68PGxgaee+65ttIF6XRaHBsaImB7nvc999yDyclJPPfccwgEArBardjc3EQ4HJaKGQBi6MbGxnD69GnMzc0hGo1icHAQTzzxBBKJhESsel6zZo/zfFCPkA8AbM9pT6fTiMVigtR88IMfxB/+4R/iscceww9/+EN897vflaiZaIvVapV0Jfc9k8lgdXUVCwsLWFhYkIZW5FB0dHQ05Z0Nhq3OlX6/H8888wzW19eFZ/Oj5LYMdCKRQDweRzablQVjtOH1evHd734X+/fvxxe+8AVhT3/ve9/D/fffj+XlZXzpS1/Cb/zGb2Bubg67d++GyWRCLBaD3W7HsWPH8Od//uf41Kc+hWvXrqFcLosC4GLz8APbk1TYVUaXwFBZULnSa+NB+PSnP43e3l4cO3YMg4OD8Hq9qNVqTYl95tMJj+dyOcRiMSwvL4vhyefzWFpaQm9vb1NHtYmJCZw7dw5Wq1UaarSLdHV1STmc7rhFhcZc0tzcnHjCHR0dePzxx+VCHz9+HKdOnWpiV5pMJoRCITG0vCDM/zAfydckzM2LSqgRgHjMrXkxlslsbm7i6tWr+OxnP4uNjY2mOlhGRaVSSWBwnfuyWq3o6urC2NgYhoeH0dXVhZGRESkl42XOZrOYnp7Gvn37bqihbAfp7OxEJBIRrogm0TkcDkGnDhw4gOXlZaRSKenEFAwG0dHRgXw+j3g8ju7ubokoSqUSBgcHYTKZZMAKhbwHRiR0hHSkpKNrbbwZGfH3WOMcj8elD8N73/te9PX1IZvNNuVNk8mkMPEJy9JY0zElMpBKpURHkRy3sLAgZaV6zOm7XVi5E4vFkE6nxVDTUE5OTqJYLOLw4cO4du2alLq++eabcLlcsFqtgqJQR3Z2dgLY6sfNkb+RSKQpdURiH9NTmqzM+82vMX1GLgiAJkjearXinnvuwX/+z/8Z99xzD55++mmpCKDO0FwYdi4j8ppKpTA9PY3JyUnMzs5icXERs7OzwoEAIMTmL37xi6hUKtIN81bktgx0KpVq6tyjazyr1SoOHz6MarWKwcFBGAwG3H///Xj44YeRTqdx4MABZDIZPPvss9i/fz8qlQrC4TDGx8dRqVRw7tw5/Mt/+S/xd3/3d1hZWRFviwaauQDmkKrVqswSJuTeWpdJr4cXSsMjbrcbv//7v49PfOITUu6l8+n0nPj5WDpiMGw1w89ms7BYLFheXkY+nxfF4Pf7ce3aNXR1dTWV5bSLJJNJhMNh6ZkNoAmSWl9fx549e6Tm2e/344UXXsCzzz6LVCqFWCyGjY0NqY1nFDw3N4dsNou9e/cKdMxzwotHB4zGTnvl/BnmnQll6nSHJvxxb958801885vfxJ/92Z9JjpPEPzp5fD06hOxgpQltjKoJcwLA6OgoAEi0305iNBrR1dWFcDjc5Awzorl+/Toee+wxLC0tyc8nEglMTExgY2MDGxsbOHr0KM6fP4/e3l6srKwIK9dut0t5ldlsxtDQkDjhjKz5f+45B+pw3xi9ct81IQzADXs7Pz8Pv9+Pp556CmazGcViUXKIhUKhqSMUX5evx/OayWTQ29sr0ZTT6cTAwAC+8Y1vSMREtK8dZHBwUMh5OnrmPXE6nejv78fc3Jz0jrj77rvxiU98Ar29vSgWi9i1axei0ajMh240GhgaGpLSObN5a4gGxxMzigW2kVIAkr5gagOA/J9OFB027p8ujTKbzfiZn/kZfPSjH8Ubb7wh07JIbGWAR+4R55izvwf3m4xz/pzdbke5XMb/+l//C263W7grt+qs35aBdrlc4nFq75QLdvDgQRw6dAi7du2C3W7H0tISkskkzp49C5PJhOPHj2NtbQ2dnZ1YWlpCqVTChQsXMDAwgL6+Ppw7dw5PPvlkU19rGkY6Acx5VatVbGxsiNLmZrFhBJUyf48KW9cyGgwGfOpTn8KnPvUp3H///ejq6hJ2pyaY6M/JRWcnHK4JD4Xb7caFCxfk/YPBoEA57SCEM5lf5kGnWCwWXLlyBfF4XJq+EN4JhULo6upCKBQSQhWJeiyzIAy6srIijHEAEvEYDIamPtsej0camdDrpXDfNdSk95+X9urVqzhz5gz+43/8jwAgeWl+n2dFs3cJ4xHJaZ33m06nm6IlssDbRZLJJJxOp4xrBbab8jQaDRlowAlRXV1dKBQKwoi32+2YmZlBb2+voC1cy0KhIPXRq6ur4uzSwWY/BoPBIIaUjTDI6tYdnwCI0aSRpePN/QW2UJ9gMIgnn3xSFDNTI5r8pqN0OgJmsxnxeBy7d+8WI9DV1YXp6WmEw2FZN5/P9xPYnZ+MkCGtnR7qWKNxa0BRKBRCf3+/RJzRaBRnz57F1atXAWyxuNm4pFAoIJ/PY21tDb29vUgmk/B6vZiYmBAOE7CthxkB2+32phJJkoq1DjeZTE3lTZpRrkv2jh07hgceeAAnTpyQfho8062NiEiEczgckoO32WzSCIWvmUqlxJkjUvhjMdCsI0wmk6JgdSJ9enoap06dEgM7NzfXVPKwubmJu+66C6+88opASW63G9FoFJubm3jve9+LmZkZFItFPPPMM7JAhJtYTzoxMYFyuSy9lcn4JuxOxUzFoaN8bqLezF//9V/H7/3e76GzsxPr6+uIRqOS5yAxhI4AYa5CoSDtCqkkfD4fTp8+Lb3I+Z63Cme8GySVSkluRxsdRjWJREKYlvRedSkTawQJibEnM8e6saE+8zQcTUoSF1tIMgeuORE05lq0t9xqIGmwTSYTTp8+jWKxiH/xL/4F8vm8sJL1RQO2LzYj8a6uLgwMDMisatbg62ifufR2MtCxWEyImszBAdtsZo/Hg7W1Nayvrwu3o9FoYHV1FYODg9L/GICUTXo8Hni9XkxPT8NoNGJiYgIvvfQSDhw40FReR0cJ2HaENJGQaw9sNwki/K1Z9/pn+JqXLl3CwMAAPvCBD2BmZkacA+6tjtyB7TIynsXR0VEYjVvtaIPBIF544QVR8j09PUgmkz+hHfrxi8PhQDgcFuOkjTPXKJ1OIx6PS3CVTqeF/Ev2v9vtRi6XQ1dXl9wrGml2I4zFYuKAcT/Irq9UKujo6ACwpZ/q9Trcbrc8B7Bdn38z9Fc767VaDcePH8d73vMenDt3ToihPHc6ADAajeIgWCwWpFIpmT1BvVKpVDA9PS0dLUlEu1VdcNskMYb5wHbdMx92amoKly9fRiaTgcPhgNPpFFYjG9Svra3hoYcekiL1QCCAdDqNQ4cOIRaLYXx8HH/0R3+Ew4cPi9fBPLPNZhMP2el0ymYSUqQhp2ejPR8enlb4i0p0bGwMCwsL6OjowPLysnjzGrbjghMW03kwPhM3plarSZ1oOxnofD4vrfN0UwKuZVdXV1P5A3NAhLs5ctLlcmF1dbXJcCUSCUSjURw8eFBgMm3oaIxZT0ruAN9Lw8j6wvFnmKei0m69dKdOncKePXsAbEdpWonrUh3WRLLjGb/H5vt0Uvi7t9qY4N0i7GPPFBTXlGK32+H1eiUaLhQKqNW2msbs3r1b6qBp+HiHHA6HDLhIJBLS2pV8EJ4DOkoul0t6J7SSBDWrnH9rZq022DpCPn36NB566CHE43EEAgEx7nTUdCUB2b2dnZ3o7+8XXkZnZycmJyflWUh4bSeioNFoFOPcmspjyevq6ioSiUST7uV9dDgcclcCgYAwsWu1Gnw+H1wulxjqgwcPSuvder3eFPiZTCYpgWMplS61pb7Xeom2QhN8dfOR0dFR9PT0wGg0NjVJoj2g/aCOY+OmkZERqTSyWq1NHfZYKcSzfktrfLsbwkjnZrnV4eFhgTyYqGfXIEahm5ubMqiAw9pDoRCmp6eRTCbx6quv4tVXX8Ubb7yB4eHhpjIJltIwh6m9KfZfpqLQF4+XkcZT12/z50ZHR/HOO+/A7/dLnpmGtbWcolKpyMB3RpN2ux2zs7MyZ9ZutyOfzwvbtF2Ea8d8vM7l1Go1qW+lQ2QwbHX+ajQaWFxclA4+1WoVwWAQLpcLV69exeHDh9HZ2Sn5xwMHDuDZZ5+F0+kUCIzK2ul0wufzCSLCjmHaEOvWjMB2i1gtrbA1n53dzyhEY/TlZI0zO5RpQgoVBddLG+p2kVwuJ+ean0sbLgBC+GOkk8/nMTQ0JEqZCAvvycrKCqrVqnAT1tbW8OEPfxjnz5+XCByA3K1CoQCfzwefzyfGutX51s/DfdZ3n6L3KZVKSbtaj8fT1D9ZO6RU9GwlzM9DyJUwLgMb9g5oFyGiqXlJwHavdELgTIdWKhUh0Y2OjqJQKEhvjWg0ilqtJgNVZmZmRN9fu3YNoVAINpsNLperCSWhI8/JddQB1Lu8kzpypi5ncKBtBLC1x4FAoImcRl1HndaKyvp8PkxMTIhxJlrIM02uQmsK5kfJbRloMmkJbRPG5MDujY0N2Gw2ZDIZGI1GxONx7N+/H8AWOcJkMsncVUYgBw8eRDQahdfrRX9/P7q7u/HJT34Sd9xxB6LRKLq7u2VhrVar5LZsNhsikUhTXkJHta1sS+39MjrWxBNgi+wUDodFOZPwwI3jc7DemjkHwrfMhbJWkwPq24kkRgalhieB7QEadrtdYBwAkifMZDLwer3i5NFpC4fDKBQK+MEPfiCK/E/+5E/wzDPP4O6778axY8dw3333wefzCWOXZ5AXkiQ1rWR1vawWXkhtsHV0tL6+joGBAckfa+eKyp2iPWgA0sxC12zSUW2Fx97tQmJmaySglV2lUpEGEtwnl8uFaDTaxKTVynN6ehp9fX2o1WpIpVLYtWsXzp49i+7ubhSLRXi9XrlbjUYD4XBYejfznrciX8C2E6EdMt1MRH8PAK5du4aOjg643e6mz6YRQ55tGnE6Z8FgEDMzM/IcHGtbqVTaiiSm+TnageUdLBQKUqVTrVZlelVPTw9mZmZEN2cyGWFEA8CePXvQ1dUFv9+PD33oQ3j44YfxrW99CysrK7h48aIgeMAWzB4KhQBs56ZbeQetneJanXV9BvT3vV4vYrFYU8vhVgQO2NKBDoejKS1KrhTnReje7JrI9qPktgw0FSIA8ZqY2K9UKujq6oLX6xVGY3d3N5aWlqTubXNzE4ODgwAgNYcvvPACHnjgAVgsFvzt3/4tgsEgfvu3fxsf/ehH4XA4MD8/j9XVVclz6b64ZFkD2638GNFq6KI156zhDlmI/2vU+/v7mwZkABACELB10b1eL+LxuJR9sVysq6sLwJYjw/w6gJvmRt+twrpPwps8dCw5IfMRgLDvnU6nOHDANvvSaDQilUrh8OHDCIfDSKVSyOfz+Pf//t9j7969eP311/E3f/M3+Iu/+AtxfjTZp1wuN3UkorLQKQ2KJvrQqLYiLAAwMzODPXv2NMFedMD484ygo9GoMPh5DmiUGTFq1KadoidyMHS0QjSLe+Pz+ST91Gg0BOZlgxuOd9QKlchHOp2G1+vF5OSk8BWo0MmUJTeB+W59BjWJtVUZtirn1lQNAExOTuLuu+9uymO2Cj+v7khos9ngdDoxPz8vzofdbpcBEe3kpAHbRCk6WeysxWCJZNJGY6v8LpFIIJ/PIxgMCvLJtIDL5cJbb72FaDQKv9+P5eVl/NVf/RVOnz6NRx99FE888QQ+8pGPCCOegSJ5QdoQa+NJIphGVDTCqs+JthWceqi7CzI40KlMi8WCUCgkhpeOWzKZlDOpq034uW9FbstAMy9HYS6Qh5ijw/h1PlA6nYbf70e9vtXVKZVKYWFhAcvLyzh8+DC+9rWv4dKlS3jqqafw/e9/Hz09PZiampLSC7aAIyxht9tx7do1+Hw+yX+QSUsPhwpZj5ojPK29KQ2HdXd34/z581LDTXq9LuOhITh48CDq9bocrkwmIyPxmB/XEV67CPP+ukkA14h/BwIBmWRlMpmQTCYF/mP5AY0e60wvXLggJXOvv/46Pvaxj+G+++5Df38/RkdHBZVhtMwoSjetYFTT6hED2yQhfVG1YdaR98DAgLQCZG10K+mNrUXPnz8vURWVFKNLpmZ4HtupSQWVMPOCVEz8Wzf+YVqCCpRdw+jQ1Go1dHd3w+fzSb7abN4aRcimQIQqk8mkTCPj/WdPbzJ56by3MveB5oE2OjXDc8Pv5XI5HDlyRJwK3YVMGwGn04lAIID5+XkUi0UZV6odwHQ6LcSldnLWNROe60qHjchmLpeTjn6bm5vo6emRMlWikbqBzO7du9HV1SVphjvvvBMnTpzAD3/4Q0xPT+Ptt98G0Ez+ow6ns66haDrHwHbDklZnvRXe5r6xn7fZbEZ3d3dT2kzfZ/7NGm7aHQZ6DGRZvaBR2x8lt52DZqcWKlid8w0EAvIAo6OjaDS2Jz0tLy8jHA6LJz02NoZGo4GXX34ZH/3oR7F3716cOHEC/+W//Bfp6uTxeGRAB2teq9UqPB4PFhYWUK/X5TIT5tLNCrhBNKyt3qtW1jo6yuVyGBwclEEc9Ayz2axAlq+//jp6enpEGdjtdqnhzGazCIVCwli+VULAu0F4+egw0Zuk8eVoQRKm2POY+XyTySTNbpLJJDweD0ZHR/Hoo4/i5Zdfls5gf/Inf4ILFy7IhbNYLJImoWKlQuTXuIe8nK3RD6M7/vzNYG5GZxy6MT09jQsXLmBychLLy8tIp9Pys52dnRgbGxOlnEqlZKA8zxK5Fw6Ho61y0DRYjJRaoWOz2SxkUTor8XhcEDCeI0YTTAdFIhGsrq4KZyEYDCKbzQqsmcvlpPkP2yyycoApFGA7SuLdu1k+upU8pj8DsMWp8Xq9SKfTSCaTKBaLQlhjqodE1+npaSm3ISkqHo9LLtvr9TbVC7eDsNSR6CXXWpOwwuGwDC8Kh8NNvfdNJhOCwaDwCKxWK2ZmZjA1NYVQKCT25Xd+53dw+PBhOQM06haLRUaMGo3bI2n5PCScAdsGXTvxfE59DrThrtfriEQiCIVCqNVqkpfu7+8XVJjvs7a2hpMnT0r6kzqRkTLPN8v3bpU4fFutjex2O9LpdBOMzA8GbJU8AFts3OvXr6NWq0mdZE9PDxYXF9HZ2YlYLCZzOH0+H1577TX09fVheHgYL730EqrVKsLhsCjMer0uoyiBrShufHwcDodDFGK9Xm+CXHkBqTxaqe30rPTftdrW5CxC1KFQCD6fTy4W21NubGyIw8BaOhbs+/1+LC0tiaPA3Hi7CBt2MLdOY6nZmezRDGy3X9VRS3d3N3K5HEwmE1ZWVlAqlfCzP/uzMlb0zJkzTbk7ffl1B59SqYT+/v6m128VbTS0cWb0xvOivWy3243jx49jdXUVxWIRiUQCsVgM8/PzKJfLCIVCePDBBzEwMIBYLAaj0YhkMonl5WU5c2S3Mv8KtNfYUZ2D5t3RsLLf78fCwgLMZrPUuNPZKpVKUg+cy+XgdDqFFb57925Zy71790qTC6JnPFterxc2mw19fX349re/je7ubnHgeeepO/h8jKBaWdm61EaXYxEF0rAkSbLAltLVyBAh0ZmZGXFSs9msENzYN6FdRDPlgWa+Bu+U5uMQNWHQQ0SEr+F0OjE6OoqpqSn88Ic/BADs27cP169fx9LSktxZlmnx7DkcDiSTSYGSuec6R66jZp0n10RR7dzpu/rGG29IlU5HRwdCoRD8fr/wIlZWVrC0tCRENsLZGl7XTgC5Nre0xrezIawto2dAT5gbtbKygpGRESmbII2eB9nn80nxudvtxpEjR9Df34/V1VVhYTudTnR2dsqF5wYEAgGUy2X09vbixIkT6O/vRywWE09Wkz5qtZow95gbIvxIg6lzjPrAJRIJuN1u6RO9sLCAfD4vk0jYPWl6elrWJB6Po1KpIBKJYGNjQ6JnRvvtVGLj9/tFYbJMhoaX8Fa5XMb6+rpAjprhyRwVYer+/n6cPn0azz//PNxuNwqFgkCLmoTkdDqbhmCUy2VEIhEAaGJ1ttYt69KJVudNw968wFSqu3btkjzq4OAgDh06hAMHDsgek5nOrlNsVkNIzGTaHs5Ax7GdHDWiBsyxthLhiBiQp1Gv15sGztCZYyRhNBplgE1vb680uVhaWhKHSg8z8fv9CIfDWFpaQiKRaOJ/tJZ9tdZoa+In/9ZwJc/B0tKSdDFjiR/PYy6Xw+LiIqLRKDo6OuRPvV7HzMwM0um0cCz0WrVTuostULU90EgF94DpzXK5DKfTKXpZT5jifPDe3l4cPXoUXV1dGB4exuzsLObm5iSQ4h+mmTioYmpqqqkJjH5/onw61aQjZ6J9rdUIACQaZkew+fl5nD59Gm+99RYuXLiAfD6PxcVFhEIhuFwu4Sutrq5KD4BsNivOHp/lVuW2DPT6+rqUumiFSzGZTNJ4oqenRzB4k8kkhBk2FWG/25WVFfzyL/8ycrkc+vr6sLq6KiQf/pzBYJBmFJcuXcIdd9yBdDot5KSbMUlJGiCTkrlPetj6sgLbERgvvdfrldKAeDyOy5cv4+rVq7K5ExMTcDqdwkDv7OwU48xcGo0Gy4TaQbq6uqTbE5nYdNoI95jNW03jc7lcE3OeuXzCk4SOHnzwQaTTaTz33HOoVCpYWFgQo8Y6RF7Qubk5ZDIZDA0NNUXwmoxF4dc0c1J3DtKlElTKBoMBy8vL0lc5k8k0MYaPHDmCI0eOYGVlBWazGQsLCzh58qQ09Acg41hLpZIYpXZrWMMyw0KhgM7OzibFwyi5s7NT+nSznIp5eZ4h8jTMZjO6urqwtLQkxDtd3sQ8nsPhQHd3N3p7exGNRvHaa6/hzjvvFPRFpzbosNEZA7YJijpq5v9bHf2FhQUMDw9LWovGiGz0XC4nJV6HDh1CKBTC008/jdXVVWmkMzg4KL3722mqHQBBF0gC1dAw15+QLtOVvEssryKhkCWOb7/9NsrlMg4fPoxgMIjx8fGm12R3N5/Ph/7+flgsFpw4cQJDQ0NNkLOOWHV+nM+kU1w6hw5sn41GoyERM1MZPp8PNpsNqVQKFy5cwBtvvIEDBw5IALd7925MT08LiprJZIRDQXSWadRbkdsy0CTMsLaNm0SlzCiXIwBJmqrVtqcEEQJkbubo0aP4/ve/LyMB2U2Gnpjf70c2m5VclNvtRqPRkM3lazNS4e8xgiGpg7+jCUU0onx+nYPm13ih7Xa7ENwACDGGeajV1VWpzc1kMkJaI4mmXYT92JkP4hrQ0BHWyuVycLlcgh6wN3cgEJDyKraB5FofOXJEDBkZwlT2s7OzuH79Onbt2oUjR44IcgKgSdFqCJt7yz0kvAVsO2Q6R6kju0ZjqycwiWg63cKhHpcvX8bS0hL8fr+U5HB6GecT00GgomoX4ex0kiI1wQrY3gf9memcsWtYOp2WMjuyeq1WK1KpFLq7u+V8AJAIh7XyV69exezsLPbs2SN3sZVzoHkFdBR1blqTwxhhtbK66/U6hoaGRMdoxT8wMIB8Po/x8XFYrVY8++yzWF9fF4JsJBKRtJluR9kuQuKUbqGp9SZRE2BrNCWbdhQKBSEKco1dLpdEocViEZOTk1hfX0cikZA7Spb80NAQDh8+jPPnz+O5557DyMiInMHWtGtr8Kb3WhMEufcaBQC2zpDdbpfe3Iz4OYKYz8XSsO9///uYnJxErVZDPB6XrmIMUJlXv1WbcFsGml1cqDx5yEmiIpOO1PJMJiO5WbfbLRHtzMyMzNT9/ve/L9j+6uqqsITZOo/TkU6fPg1gu9UiF5veOS8pDwfJK/wdHh5unIYcWyMvzgLmZjK5z6ld+/fvRzqdxsbGhnRLC4fD4lgQ3mLdYztBm2QjApDoULMS7Xa7lBcwBcL6VKNxq8NPIpEQQk1XVxcWFxclZ8lhA9wTrv+9996LRx55BKFQqKldooa96CjQaaPDwHOqPWntXGjiEM8zox89rIPnzWaz4eLFi4hGoxgbG0NnZ6d8n5N9DIatgSwaSdEX/90upVJJyitZ28tz0GhsEUZZVmcwGIRPwLvEvDSHDjCy4d3VBpT76HK5hLsCbJG42EKR0sqJ0JETyzK1Ueae8He0YSaawqEtZGDTsQuHw/D5fFhfX8fbb78tCrlQKEjDIp4Fv99/U6Lqu1l4RzUKBWzrZRo8BkrcGyKYhKgZTDF1FI1GhQBIQ06j5vF48MYbb+Av//IvkcvlcOjQIXg8nqbcMv+tCaZ6up123LRzz8+iyWI8uzogY7kfU6/sv/53f/d3mJ6elrTm4OCgBJjsqAdsdyW8FbktA61JV4SeCWtxOAYZzdwI5pYrlQrS6TT27duHUqmEQ4cO4erVq9i9e7eMZOvo6JCaQmArYj9z5gwAiCeiyVdaoerZnxQNZ1JRUzlzw+nJ6EjLZDJhbm5Ofm9zc7NpTN3169fF6DJ/QUPNlm8kjmhHoR2EU2o4alBfCqPRiGg0KheCqQ2WVLDHtdlsxsbGBkZHR7G0tCQlFqurq9jY2Ghqner3+9HZ2Sndhlj6pokgvITaCLZ60LqhjM5JA9vlFxruXF1dlYldLB/jBa5UKrjzzjuF5Uvl32g0xFMOBAIwmUxCMmMbw3aRXC6HeDyOWq0m/aq59pz2w5QYFRv5CUyPsARveHgY+Xy+aSQlB+IAEE4LnZ7Ozk4Eg0FR9BrS1vlDnQKhUHHr77fmo3VlyuLiosDcNEJ0MOiYnD59GgbDVse8UqmEjo4OmM1mxGIxJJNJIY+ZTKa2SncRmib/goYNgKwHe2MDaCpbcrvdKBaLsFqtMnrU7/ejVCohEolIdMp1Y13x2toauru7cccdd2BwcFBejwgpg0V9HvhvDWGTm6SheAA3dd5ZncTAkE12fD4fHA4HXnzxRTz33HPyGer1ujTvisfjSCaT4ogyXXerBvqWWNxcdLbtJBzNshoqzGKxCL/fj7W1NTFgfX19uHbtGo4cOYKlpSXEYjGMjo7ixIkTOHz4MDY2NjA9PY1Dhw7hW9/6Fp544glpBUiPh16vJhfR4+WwDipYfeEYNVGo1LU3pyFx5pqsVqt4cwAEpqeRTqVScsBKpRLcbjfm5uYEvuAz0ePm5r+bIyitxHK5HGZnZwWKJrOWSAhzboweOzo6kMvl4Pf7pY79+vXrOHDgAIrFIoaHh7GysoKuri7EYjEcP34c77zzjtRJ6ulF2jnTf+sLBmyzM4lq0DnTZ0KT13TZICMtRnQsrWEEqL1z3gVeXl5QMriZwyZhUK/lu1H47MlkUhj7dGR0u132WWf0Q0PKEhs6sHNzczh69CjeeustDA8PSwvR2dlZYfuyLEWTEnVNua5NbT0jNMZaD2j9wH/rHLVG4NiUh7lSOhr8HToSwWAQmUxG3o/12qVSSZy4eDyOWCzW9AzvRuGzr6+vo7u7u6l8ijqahCoiquyVzZRfMpkU0nBnZyd6enpQKBSQSCQk7bG5uSljQMkBoLEEtgMw2hrukclkkvuo2do8A5poxol0DKaYEtX11fwdIj403CQj80zze3yN2dlZcQaCwaA4+izb02v5j4mhcQsnZWlpScpZduSfLouLi+jr6/t//Rj/JNk5A/98snMOdmTnDOwI8KPPwS0Z6Hq9jpWVFcH6d+T2pPF/mdzMr74bZecM/H+XnXOwIztnYEeAWz8Ht2Sgd2RHdmRHdmRHduQnK+9OF25HdmRHdmRHdqTNZcdA78iO7MiO7MiO/BTKjoHekR3ZkR3ZkR35KZQdA70jO7IjO7IjO/JTKDsGekd2ZEd2ZEd25KdQdgz0juzIjuzIjuzIT6HsGOgd2ZEd2ZEd2ZGfQtkx0DuyIzuyIzuyIz+FsmOgd2RHdmRHdmRHfgplx0DvyI7syI7syI78FMqOgd6RHdmRHdmRHfkplB0DvSM7siM7siM78lMotzQPemd6yf832ZlgsyPAzjnYkZ0zsCNbcqvn4JYM9MrKys78z38GeTfPgN05A/98snMOdmTnDOwI8KPPwS0ZaI/HAwD4vd/7PbhcLjQaDTQaDZjNZlQqFQBApVKB0WiEz+dDrVZDpVKByWSCwWCA0WhErVYDJ1tWq1VYrVZUq1XxwGq1Gkwmk/xdr9fFszAYDPL7jUYDFosF9Xpdfs9sNjf9jMlkkp/l/+v1urx/OBxGJpNBvV5HPp+HxWIBgKbfNRqNaDQa8rq7du3C9PQ06vU6XnvtNXzyk5/E8vIyarUaqtUqzGYzTCYTyuWyvEalUoHb7UatVsPv/u7vyjq+G4XP/lu/9VtwOp3ydZPJhGq1CgDyt9lslnXhXlkslqb9rlarcDqdKJfLss/VahUmkwkAYDQaZY/1GTEYDDfsT+se8+vAlrfPM6jPgNfrRTablTPMM6Dfm+fWYDCgWq0iEolgdnYWTqcTly5dwoc//GFcunRJ3sdoNN5whuv1uqzHxsYGvvKVr7TFOfjwhz8sZz6bzcqaOxwO9PT0IBKJIJvNolwuy9pwHcvlsuiNWq3WtM4Gg0H2HcAN+633nPtkNptRrVbl3/y9VuHeu1wuAIDVakWhUMDS0pLsmcViEV3GZ+Yf/n6pVMLp06fh9Xpx//33Y2JiAleuXJGzrs8az1+j0YDT6UQkEsHv/M7vtMUZ+OQnPwm73Y5KpYJarSbfNxgMsNls8Pv9CAQC2NzclD3lmhaLRbkfWpfr1+D6aXtAXdFqD4xGI6rVqtgd2p5We6L/b7Va5cxmMhnk83kUi0V5f74395/722g04Ha7sb6+jtXVVWxsbOCDH/wgTCYTCoWCnGmeQX6OarUKt9uNkZERlEol/Nqv/dqPPAe3ZKD5Rg6HA1artemy2Gw2FItFeSCbzYZKpQKr1SoPRyXOD8oFplLkQtDYcqP0qGqr1QqLxYJSqSTvQaXPRafh5wJpI1+r1Zpe3+/3I5vNIhQKoVgswmazoVaryYXSirbRaCAWi+H48eM4deoU9u7di1dffRV33XUXkskkXC4XqtWqrIHJZEKlUoHX64XFYsH6+nrTOr4bRZ8Bm80m6wMANpsNm5ubYgi5T2azWdafipOXh+fH4XAA2Nofm83WZMxbR5VbLBZYrVaUSiVYLJYmp46/Vy6XYbfbmxxCPkO9XkelUoHNZoPBYIDH48Hm5iZ8Ph/K5bKcK/156fxZLBYkk0kcOnQIly5dwvDwMC5evIi+vj6sr6/LM1OJUAnw+ZLJJLxeb9NrvxuFz24ymWC322XdjUYjzGYzLBYLTCaT3CdtRAGIQ2Y2m9FoNEQv8K7SAGsHv1XR8jm4vzSq1Dv8ntlsvsHR4rPznAaDQSSTSRSLRfm6Poe1Wk10GV+XTkgikcDly5dx/PhxhEIhFAoFeUaz2SzOSa1WQygUwtjYmNyZdjgDvN82m02cczrYFosFZrMZdrtd9CFtQWvww7veukf8HZ4B3uNGoyF2aHNzs2m/rFarrH+j0YDL5UKlUhG93+rcOxwONBoNBINB+T3qLhpW3mnt/NXrdQwODiKZTMLtduOdd97BU089haWlJTmPvP+VSgXVahWBQADj4+Pi2Oq1/MfktpIgNEA6MuWFo3GmQuOit0Yj9EB0hASgyZjqS8iLyQXm62jFz3/rKKrV2BsMBlkYKmEadXpfPDy82PpZGo0G0uk0fD4fent7ce3aNXg8HjEY/DwWiwWNRgMej0c8M5vNdjvL/FMt2iukIaIHzUPPw0cFyn3Ul4QXrlwuy8/wNXlOeJG4R9rRoyMFoEmR84LyHLSeAR3Jca8rlcoNRkErAz4XAGSzWdjtdvj9fszNzSESicBoNKJYLDYZBgDy+qlUCm63+ye0Qz8ZYUTAdbRYLE1RB+8alVwrsqHPgY6W+Dd/X+fnGNnSCQC2zw2jX33m+G99lvTz83mdTmeT00iFzL3Ur8Nn3r17N8rlMlKpFF566SVMTEzI5wK20aR6vQ6v14s9e/aII9guovV1q4NltVrh8Xhu0K3ANhKmbYFGKnj/zGazrL825jT+dA74fwBNxo9BmT4TFH3H+bNGo7FJBzCY0HaPwjM3ODgIj8eDxcVFXL9+XQJW/XsMBvbu3QubzYZGo3HL5+C2DLQ2sPx/68LzwtKT1fAWH47KkJeaBlFfTl50/g4vJxUuF4zKVS+8hhd5afl6FL4PLzy9NVkYtbi8bEtLS9i9ezdqtRruvfde/Nmf/RkGBwebjEK9XofL5UK5XEY6nZbIv53kZgdbK01C/rw02sAxsuIh5f7qSJjvwXSB1WqVveal0VAizxwVLIAbImsAAmvx5zTkrp2F1s9K5WAwGLCxsYHdu3cjm81i3759+Ju/+RscPnxY3p+K32w2I5fLIR6Pw+VySdTeLkI0olQqiXHk1wGIwfvHPjN/nntIo64NtnbEeLb4c7y72lBzjzQCp50Avi7vM6McRmR8HT6Xdhwo3Ge73Y477rgD6XQaZ86cwfnz53Ho0KEb9I/P58P999+ParWKXC53Ayr0bpZyuYxyudyUntBOdalUkrtJ4bpou6HXnLqkXC6L7ubeWCwWsRk6NcYASQcGvIN8Xe0kaKefep+/b7PZRJ/TUOtn1+cnmUyis7MTJpMJnZ2d+Id/+AcYDAbY7XZZDyJ2u3fvhslkQjKZFMf2VuSfFEG3KkMdbVqt1qacBC+gho/r9TocDofA03rDNCzJ39MeF7/Hn+Ol5kbToOp8AS8NN5abSOPSekFbcweEU4xGI65fvy7Q9djYGJ555hmJpOk58tmtVmtTxN4uwn1g3k9/jU5buVyWQ8j14B+dk6Ox5gXS+0wlrqEp7Zy15rV0frJWq8Fut99gtAE0GWKtzHkx9efhGdIGZX5+XpTHrl27cPLkSXR0dDR57lT+vKzlcrmtDDT3RDseGsLk/7WC5l5oOJO/o++pzjXqqIyGT+ebqYj5fa1stWFodRa08taQt4bRgWb+Q6txDYfD2NzcRK1Ww1tvvQWj0QiHwwGDwYDNzU24XC7s378f6XQam5ubN3Ad3u2iUUOd3uP3tLHSqR/uv8456/sFNDt4NMo6uAOanUEATe/TGiEDW/f+ZnZDR9h8L3KktJHWASQlm81iZGQEZrMZDocDJ06cQEdHh9wLu92O3bt3w+l0IpfLibNgt9tvbY1vdTO0MIohts9Lw7wLoyJ+6NbLA2zDVfR4NZTcqoSZfM9ms4jFYojH41hbW0MsFgMAuRRUCK1wOAAx9lw4TU7QsLb+Hi+uNvSbm5s4cOAAqtUqent7MTU1Bb/fj1KpBJ/PB4vFIpEzsE1ia0fh+tJYVatVyUfx82vIWEdLvGitEQvQDCXyIhPlIEqjL3ggEIDX65Wzo+E2fQb0a7VC9TynWjnr86odxWq1in379iGdTiMQCODq1asYGhrC5uamrEUqlWriYfDf7SLMzfKuE+IHboxctYOkkTJge995//Tet5KParUayuUyisUiCoUC1tfXEY1Gsby8jGg0CoPBAKfTeQO8reVmThL3rNXZ1+eCZ7Y1fzgyMoJCoYBEIoEXX3wRe/fuRaVSQSgUwsTEhAQi1CX687zbhcbKZrPBbrc33Us6ZoyE9b3VawpsO7T89+bmZlPwxihX/0mlUshkMlhcXMTMzAzm5uYwNzeHbDZ7g77RKKi+51oXaeRW6yK+Fm0enREie7VaDeFwGEajEU6nE/Pz85ibm0O1WoXX68Xu3bthtVqRyWQAoCmddityWwZaw3d8YJ3LI+zACBbYZubywbSS0zkfLpImofF1HA6HeGqBQAA+nw/BYBAulwszMzN4++23kU6nEQwGJXLWBld75Pry3myhtIfVGhVyc5eXlzEwMACDwYCDBw/iS1/6EoaGhlCr1ZDNZoV4oPOm7SKt0Sj3lBHI5uZmk+Hm5dP5HZ4Bvfbc81aDbzAY4PV64XQ6ZV8dDgecTqdclmg0ipmZGRiNRuzataspr6z5DPqPVtx6f3SUxM+mzyIv7urqKjo6OlCpVDA2NoZvfetb2LNnDxqNBjKZjJDfKNp7bxdxOp3yObmGGjnj/umo+WbRjU6Z8f/UKcDWWdvc3ESpVILVakW5XEYikUChUEChUEAul8Pm5iaSySTS6TRKpVLT+mu4nHdSnwt+X+c9+az8ee1EaEdyaGhI0jXnz59HNBpFd3c3xsfH4XK5kM/nb2Ant4vwPpIIBmw7ZwzIiKYxNdGapqToM6Dz/oScgS3jxr0uFApSiQNAKgMuXLiA5557Dqurq7Db7cL/oYMPbOl27TxznzVXhfaN56DVMWtNy+7fvx9msxkejwevvvoqQqEQOjs7BT1j8EIH8lYd9tsy0NqbuBlrmcZVQ1e6jKU1YgW2c0oGgwGFQgHFYlG8WZfLJbh9MBiUkiWXyyUL2t3djf3796NQKOC73/0u3G433G53E3TFRdbwiz4U/BntQdGYaNienz8ej2NoaAiVSgUdHR1Nz64ZxBoZaBfRiIeGgFodM2D78HP9yHSmtMKZVHRWq1Vy91TOyWQSPp9PvHKeDWALQenv70e5XMZbb72F7u5uBAKBprXX0PrNcqO8jHQOeAao2OlsAFvnJZvNYmhoCKlUStjZZrNZ0JNWNKbdctBOpxNutxsOhwObm5sAtiFuAE2chFYIlCgXgKY90QaxUCg0VWgwYqfBI8yt01TxeByzs7NYXV1FLBZDd3e3oGvANpytyaB8bv6M5qW0oitap/BrTqcThw8fRjqdRqFQwAsvvIBHHnkELpdLSm40x6KdzoDJZILP9/9j7z2DJL2u8+Cn00zn3D057sxswu5iAxYZBEGQBEnRpEBKli2WinKZoiSXVaVQtkJZlq2yvlK5LJv6QZVIBZZEBUoMYpBEiKQIYAkicoENM7M7s5OnJ3XOufv7Md9z5rwNyNzVZ9JE19yqrd2d6el5+95zT3jOc87xwefzibGiM16tVsXZ0c4RAOEtVCoVg2OkHTfeoUQigWKxCKfTaeAjhcNh+Hw++P1++Hw+cRKGh4cxMzODra0t/Omf/inS6TT8fj+sViucTqfoKf5OTW7sRDc6UV+mMfke1FGlUgnBYBBerxcOhwOZTAY7OzuCKlCOGZS2Wi1UKpXb2uM7MtA0pp0Qoc47koSjc45U4ryYnbmDVquFUqmEiYkJ9Pb2ihIG9r0mj8eDVqsl718sFg15ynq9jmg0iosXL+If/uEfsLCwgL6+PkOkzGfRtbZ8TkL1VCCEKnkJOy9Wo9HA7Ows7r33XlSrVbzzne/E2tqagVygiSzddCkBY54eOCAH8t+UASpRzfSnrPB9gAOvO5PJYGJiQmBMzYjVZXAWy37tLQCD8fR4PBgbG8OVK1eQTCYxPT0tXrOGxzrzSFqG+UzlclkMi36d/vfS0hLOnj2LdDqNkZER3LhxQwyxVjj0orvJWXO73RgYGDCkpAAjX4AcDC7uHxVXp4HkH36P78FInE5To9GAy+USNEXLGOH2W7du4erVq+jt7UUwGDTIbGeOm3KlHTCdo2YURiOjq0gIcQYCAZTLZUxOTmJwcNAQOWvovJtkwOFwwO/3w+12izOryxJ5jtrIESamceM5awcN2DfibODB+91q7TPiA4GAwaGmTqCcWCwW+P1+nDp1Cs8++yyeeeYZ+P1+MeKa7c/34PPyDx0x/h6tLxi8MWpvNBpIp9NCIH788cfR19cH4ABB1hVONPy3s+44gmYkpEN8DSMTzuBhAAfKUdeuMuqmR9Tb24tcLifKrLe3V4w0FQA9Fz4LPTVN6Ln77rthsVjwqU99CuPj43A4HBIJa3hTe8iaAMIIWBOJdCMUCkO5XBYmbzKZlN9PI00kgA5At6zO/LyGhXjZNCFKX0D+nD5PbUC9Xq/UjPO1+vJRMVLuKHtkkvJ9BgcHYbFY8NWvfhWnT582OAz8eV46yqNWvjxD7VHrXBWfqV6vo1qtwu12I5VKIZ1Oy2ubzaZECvzs3UQQ8vv9CIfD0nBCQ9c8b51n1OSdznyvJgQxqvJ6vRIxcd9sNht8Pp9Exbz/nTlOOuJbW1t45ZVXsLm5ienpaTidztfB1IQ7SeqjwqYBoX7pLOHi76CMHDt2DKdOncITTzyBZ599FsABSY4yw2irWxZTHDrfT51qs9mkr4BOWQIwVO/o/hg6HUrklUZa81Xq9TocDofBieK9NZlMgug0Gg1MTU2h1Wrh4x//OKrVKoLBoDhc1E3aaWJKjnrB4/FI0MbF+92JCrTbbbznPe/BhQsXMD09LbJJQ065q9frhmZP/7t1xwZaQ7c8GCogHRHpy6bhZa3k6F3ZbDYpTWq399m9mgbP8gR9CBQIzZrm74hEIrjvvvvw67/+67hx4wZOnjxpKPvR7GI+T6lUkud0u92GRhma2KDJA6lUSppm+P1+ABDh0g04bvcw3gyr1WoJcxV4fW1xJwOWgt0ZVZJMwn+zXpyKlpERv1+r1QysXe1wkRGpFYXT6cT09DT+9E//FPl8Hm9961tFLnXUp40GYTkq004io47A+Jl2d3cln+R2uw0oARsomEz7JYa3C2u9GVZfX58oMn5m4EBH6EYPgLHuWKc29DnyfSYmJuT3cM+ZeybHge/DiFlzWvj72aNgfn4e3/jGN1AulxGJRAzPx5/TzTIYqZH30PnegLHrXaPRwPDwMD70oQ+hXC6jXq+jVquJEaGiLhaLXdUTgb24Ox1f4EDHaqeG+kGnJ7QDx3tM3lGxWBTGM9NErVZL0IlqtSr2g04WbQX1sMViQTgcxvT0ND7+8Y/ja1/7GoaHhwWBofOg0520XazlZvfMNyrF5c8y0BgfH8f4+LiQFhl8MoKmTVxcXLytPb7jOmjCTtx0Tfyg0NIAa4PdGUXygPiaYrEoXgZ/hsqYnboY8dAoaw+sp6cH5XJZEvIWiwU/+ZM/iWq1io9//OMYHh6WQ9OtPYGDXGgmk0EkEjFAcxqG1wdIgUyn03A4HIYIga9lq8+tra072eYf6KXhS/6/E/Kj8uusa+W5dMI7VqsVXq9X8o6dcCmNMOsuq9WqlG8AMFxGNgmgQ/bWt74VAPCZz3wGR48eFSMMHKAB2njkcjl4vV6JyrUS0RGCzlslEgmDE6YNeLu933mqXC6/qds7di6Hw4FsNitpAJ0m0J9fO23a4dWoGs+42dzv8JdOpw35Zd2cAoDcf7vdDofDgUqlIrrEZrPJWZAXUq/XcevWLVy6dAnHjx9HIBAAYMyN8zx1IDA2NgaTySTOO88TOCgx4rOfP38ewWBQlHinbiqVSqhWq9JKshsWCU90anRFDwBDWRnTVICxja+WDd45u90On89nqFPXjp/T6ZTUJ/eZup/savJCMpmMtJy9ePEitra28D/+x//A1NSU2BcGUxrWpi4ZGxtDKBQyBAeUEeAg3Wez2TA+Pg6bzYa1tTUAB+k+wuBMc21sbCCTydzWHt+RgaYS1J24uEF8WG4oDaHOSfEw+X8NiXJzqMh1Qp0fltCphtK8Xi8ymYx4py6XSyLrYrGIqakpeL1efPazn31dlyk+n859pFIpTExMwO12CxzFSE6T4whXkWWoozz93tvb26IQumHpqLbzXDuX3jOeGc+UP0uDrUkllAF93iTa8AJq2ejp6RGl7vF4DNBaqVRCKBRCb28vXn75ZUNTGg1na6JKuVzG8PAwzGYzCoWC4XNrBnCr1UIqlRJSImWLn5PvXalU4PV6DfnYN/sinA8cGF4tD7y/nakpjabpVFK7fdB9j/wD5gYBSFRKVEYjcoyg+LsTiYQB/dAw9m/+5m/i7NmzBm4Iz0yjP7VaDXfffTcikYhEbZpdrJ2MBx98EKFQCPF4XGBvzdJl06JkMolLly59T87j/8aiHuAd1nlkpjeYr7Xb7a8rpdPpHxp4og4k2HXm7xkhNxoNFItFuWcm0z5hb2BgAOl0GqVSCZlMRowv79/Q0BBsNht+93d/F263W+SItkhXGwH70PWRI0fg9/tFJwEHgRtfPzMz87q2ovw+7UU6ncbm5qY8323t8Z0ciPZktOKkQtJwhY6Omb/jw3Oz2T9XMyN5ATQ9nhfHbrcLXMSIOZPJiFLm5eVlpTI+efIkVlZWMDQ0BJ/PB8DYWpLPY7FYkEwm0Ww2xdhzgwGIcWaemvXPmhim8+qtVsuQK+mGpc8bMJapcPECsj83cBBVao4CsH8BWCup4Wez2fw6ZIKVAnwPKs5arYZQKCRyVyqVDCziZrOJkZERrK+v4+zZs4hGo4bcWGdqRg9/IESpEQI+N/ut6wiLxqjRaMjv0GzPblmdZD+d/gGMFRKahKMNI/cLOHD2aXw1aY9pJP072ISm1WqJA1YsFlEoFAzkMzrKNpsNdrsd5XIZv//7v4+zZ88KzKllmYsGY2JiwhAF8fUMRqampnDixAnE43EDUYr3n4p/b28PS0tLryu/ezMvIhdkNHdCxhpFo+Oiz6YTUeOdYRTOnwMgfAPC3DTmdIbsdjs8Hg+2t7cBQHQCo3vajEajgYGBASwvL2NrawuDg4MGEqCOjqnDLBaLlPXq7/MZ+/r6MDg4KKnRVuugQRHvCLuqpVKpO0JR7jgHrWsbNTTEDdTGiBeDob1mOdPo8mJy0wlzNBoNUWhWq1Xyk4yitra2YDbvT8/ShpGXm3/oDX3gAx/Az//8zwOA0O25NNTWbDaRSqUwPj5uyFlT0QCQjjD8TPSs+J6M3JjPzmazd7LNP9CrMx+rZYDyQY9W/5+eJ6NfKjoyK7l3WmEC+2xqngNlifJXKpXg9XrhdrslN873Bg7YlvTkH3jgAXziE59AOBwWCIxL/1yz2UQ2m0U0GhWSEGWACkWT3fj5dHRARcXaSKIt3bT059RRDv/W0Df3TJecUIER1uRkOd4r1j7X63VphKKjUyrdSqWC3d1d0SVMUWkiD5Wq3W7HysoK/uqv/grveMc7DHwInVqhYY9GoxgZGZG2pvpztdtt3HPPPaKLOmWPDsfe3h42NzcNaGO3LCJPnTl6HQBxXzVawnupG5pwEA8dYpKz6DBrUh5/nvdve3tbHHaWWPJ7vMOa93T33Xfjt37rt/DNb34TgUDA4GTqZwb29f/k5CT8fr8haODrQqEQPB4Pcrmc6CCd4qU9WF9fNzgxt7W/d3oYWjFpL5nJfG2oNGlAM6lJS49Go3A4HEIO4yExQqVXxryj0+lEPp9HPB7H8PAwbDablNtQyesckPaMLBYLfuqnfgo/+7M/iwsXLryuDpOva7fbKBQKKJfLGBoaEq+bSp412PrwKAhcNPYWiwXFYrGrvGamObRC0/v3RmQoDX9rb/nmzZsIhUIiN4xS2GCAilq/v9PpFPLI8PCwEMwYuejohfXUGlJ76KGH8Gu/9mt4xzveISgOHQxtXEqlElqtFiKRCBqNBnK5nDhjVBpc2hGgbPD57Xa7TGDTcvNmX5rVzs9FBU0EpBNp0/wM6gyr1So15Tw74AD+fiODbjLtTyFKp9PY3d2VCJuyCcDAIeDvZDTX09ODK1eu4JOf/CTe8pa3vM7B0Km4np4enDhxQshlpVJJ5GBqagpWq1UgecoZIzZGX/Pz87Jn3UQUpLNKxIh3sPO8dTDW2RSIr2eKkpEn31cbap36ZMDUbDaxvLwsVUCUAcqX7mSmUTGz2Yx3vvOdeOqpp1AqlaQyQKc9KQfUEUeOHBEElmt4eFiqTxiYdFaq2O12LC8vG7gSOp3yv1t3rDGYA9TGFzi4cLpGWpfJaDi52WwKDk+vWTd34KbS4NLYxmIxWCwW9PX1YWdnR/LNOqqz2+3iFZMxx99ptVrx0z/90/i5n/s5RCKR15EbgIOesOvr66hWq4jFYrDZbJifn8fq6ir8fr+hR7Rmq7JOe29vT3Jpt3sQb5ZFIacToj1FngMvCY0iZUCvYDAoxLxCoSAQpE4R8Fw05BSPx+F2uxEMBrG7uyuEEX4fgJTnNJtNFItFgzdstVrxoQ99CL/927+NqakpMSoaYmcExnOsVCoYHR1FPB5HPp/H2bNnDZG+RoQqlQqcTifK5TK8Xq84C505zjf7okMCGMmWOg2m00z6dQAEolxbW4PVahWUie9LuWKExjOl05XP57G5uSmtZT0ejyh14OAed6aXaDwtFgsuXboEi2W/ZpbGVT8jlTqrTF577TVsbm4ik8lgd3cXb33rW7G9vS1pN+pAnrXD4ZD2k06nE9lstqt4CDzrTsRC3zcNfXeS8hiF7u3tIZVKwefzIZ/PG+qjW62WoGhMe7FBzubmpqEdrNPplBp7IlqaXMgzBg6c6tHRUfzZn/0Zpqam5JnpSDBFV6vVkMvlUC6X8eyzz2J7extPP/00FhYWhCgHGJn9mjeztLQk6K+uRLiddccsbk0Io0BrL5qv0/CDzkG3222MjIxI8xF6mqS1U8FrSjqxe6/Xi8HBQezu7iIajSKTyRguIN9Pk9g65/yOjo7i2rVrOH78uIHQ0sngm5+fx/b2NsbGxtBqtTA0NASPx4PPfOYzMg+ZB6Bh23Q6jUAgIPnHblLKgLFXOmAcDaojal4AAAanTXvbbJdK6LpSqQh6wgiMk2royYZCIXi9Xmxvb8uFZnkGIyZCZjwTskkZvQSDQXznO9/Bgw8+KA6AZo/z92UyGVgsFhw9ehTlchmhUAhutxvf+ta3RK51dABAGLtut9sQRdwupPVmWVSMjBq0LmAUQfnvZHZz36xWK27duoXJyUlRhjwLGltGyxaLRfLYzB+Gw2EpXyEpVTuEukc/dRCfiQjHxz72MTz88MMi01pO+cyXLl3C5cuXBca8cuUK9vb2cPToUcNnpjEC9u9AKpXCK6+8gnA4bEjXdcvSbG3tVAEHaKv+P/Uz7zzvDhuPEHEiI5vnRaNMB81qtWJhYUFSDswDp9NpuW9ut1tqsQGIfGgUxmKxIBAI4ObNm5idnRWnTqNvDB6vXLmCz372s6jX64jH4+jr60Mmk8FXvvIVeX/KsJahRCKBTCYjuomoyvcE4ibUoBe9Fxpl5l41vEQyGJVlIpEw1LLqjaFybTQaUofYbrfR398Ps9mMbDYrbD5dr8zGJoRKSOzg+9ErarVa+Lmf+zn89E//NMLhsHwOHUm/8MILOHv2rMx75bSqaDSKy5cvG0q+NHRisVjEo6eh0Iy+bliMEvTlo7DRIBLS7fweL2i73cba2pogJHRytKInbKhrIDlUnRezUqnA5XKJUqezVy6XRakTnuKZ0Un8iZ/4Cfzqr/4qpqamDCkZGol4PI57770X+XwehUJBOBBerxdPPfWURIiUKZ4zZYB3gFFCN8kAAEOEQadNO2vaQGsURKNdJpMJ2WwWQ0NDouz5el3/yvezWCzY2NgQY8i8NRedfJ49DTzrdTuXw+HA3NwcstmsOIY0Cqxp/8pXvoLnn39eonaHwyHO3rVr1+RZ6OyTC9HT04O5uTkJTBh4dJvDru0Bo02eIfU0nS09BIPESnJ+AoGAfJ8lVpqJr9NEW1tbaDab0i+D7WZ1VK4bTjFy1ykzonwmkwmnTp3CX//1X0swxufl7/vc5z6HeDyO+fl56fXv8XgQDocF+WUKlo6G2bzfgS6ZTCKdThvQPeAA4flu644MNDu40LvRFHheQG6EjmR4Welxb2xsyNfYbYmDt8nU4+sLhQIcDgdKpZJEwbwQzO8yeqFhJiTBLjA0mlSofr8fTqcTKysrMlScubClpSU89NBD2NzcFMNBY9FsNjE8PCxlALruj5GThnp0vq1bFgWXhk4TZnSagPtCFINywtrwjY0NiUJ5ucjE15fDYrHIJWeLVwq/yWSSlpwADIhNpVKR2mM+By88L7fNZkM8Hkc0GpWLTbblQw89hJWVFcMF5GeNRCJCimTURjnXJBnA2Nqx26JonavTf/iZde+DThIR0REddRIl6Uw5MHUVj8fhcrkMZ0JSGNEXOno07OSwWK1W+Hw+UZAMDHp6evCXf/mXOHHihKFjWbPZxGc/+1ksLCwgHA4jlUohFAoZIEwGC3QOqSM8Hg/W1tZw69YtBINBlEolg/HolsXzZxctOlEa6uZ9Bg5SIBp5ohPu8/nEmaGR1M457/z29jaazSaCwaCBj8KAzWzeL71lqkkHUryr4XBY9AgduEqlgr/5m79BKBQSXQQAf/iHfwiHw4Fvf/vbuHDhAtLpNNxut3BmNJKmOQh2ux2pVAqpVEqCVM11uF2H/Y5Z3ADe8ALyIDRZiodBwTebzYhEItLggKQfzZZmRMSD4kQQeqC1Wk3gDo5y488yQjObzZIboGGnQif8+ra3vU0UsNVqxdramsCZ6+vr8Hq9hlwGP3c4HJYieR66ybRfe0nYnp+bq5u8ZioYOjw66tCRqt4zAIaIkrlfRpyMqqvVqjQh0bmlcDgsSAWJOppIyGcgS5TnTMdME4z4TCy/29jYgNfrlXx4NBpFX18fVldX4fV6hQGqjc/ExAQajYawx2kUGNEzX9pNyrhzUcFpWeeZaeeUZ0YZ0bpgd3cXLpcL6XRalKcufTKbzXC5XKhWq9jZ2YHVahXUiw4yz591tnwfNizhfWTkS7mlYrXb7Ugmk3j22WcxOjoKm82GWCyGubk5LC0twWKxIJfLiT6hPrBarRgfH5f3oiwD+wS1y5cvS+0s86ca/uyGxc9CHcjATBtf6nl9h3Qr1XK5jNnZWUE/vV6v2AWNztpsNqRSKQCQtAYROL/fL04bSZnAQXcy3YiE8Dmfi1H0mTNnUC6XMT8/L1H63NycoGlTU1OoVCoim7R9nGin+ynU63Vks1ns7e0ZKhP02d8uF+GONAhzhXzIzhyD7rKiLyI9LB198CBZRsGB5hwlpnvf8lLRs97d3YXdbpeaR2448wA6quEELJ0PowE5ffo0nn76aTSb+3WyOnrTNY304BqNBu677z5YrVb53Z0lAdpT6sbcI6FDOlUaReF+0Yuk96rZuRTUgYGBN/QiWYpisRy0XGUaAzhw+tLpNOx2O/L5vJwXL22z2ZQcI2WASI022AAwNDSEb33rW7Db7RgeHhY+BOWq07A0Gg08+OCDEkHx0jscDlEEVEid5RTdRBDiPmteChW0jpSIlHEfaXh7e3sxPz+Pvr4+JJNJw35pNK5YLErXJafTKWgLoxgGBdxnVpFQYfb09Eg9LNEWl8sl1SJMndTrdVy/fh0rKytotVrY29uTyJcKVue4gX0ug05v1Wo1lEolPPvss4I2EsoFjFOzumHxrvLcdCtMnqHWudwn/gxTCeSfEFXRM7+B/VLLRCIBk8kk+WrykyKRiJDwmN4gigFAUBa3223oxW6z2eB2u+FyuYQ5Pj4+jkwmg0wmIzMBBgcHYbVaJZoHIMFes9nE4OCgOG6a/7S4uIhkMimMf51zvxPn/Y5JYhou1otRCY2hjng0OQOAjP8qFAri+XJWZqlUQiAQkJmu6XTasOG1Wg1DQ0PY3d1FMBiUC8sLrpm/NAw6V9FsNuFwOJBKpWA2m3HixAmBNelxcxMBGKau2O3215WAWSwWEQ5dQtEJeXbLoiHUNY1cvKRaBrSc8CJzRCMJIRR41kGWy2W43W4Ui0VYLBYZwu71eoVoEY1GkUqlEAwG5XV0ooCD+cv00hmVs8EFCX1OpxNDQ0MiA4yyuKig6bBtbW1JraN2UOn5a6ejsylLNyEpOiIgzKs5BAAMjpX+OWBfyW1sbCAajaLVakmUSqfWarWiWCwim80KKsGUGXWG1WqVfed95Llog0CeQ61Wk5QW9RiNs8lkQqFQQLFYRD6fN7Sy5Z9AICC678KFC3C73aJjKBOJRALr6+vw+XyGNJjmOXTL6nRCNUJAI6QhXU38o8xQ39dqNezu7kp0DezLCPewVqshEokgnU5LrTPPOBgMYmdnBz09PUilUobgkPwDOk/s8c+UKlEaPmMwGES5XIbP55MUJ8+YOohp1aNHj2JwcBCt1kFDKiJpbFtMXUK55F51crn+qXXHddD6F/JQKHy8NFoxa6IMFSAhIQCGJD4JWTSApNezCTprkNPptLD2NM2dz0gPiaUuZNUyv723tweXy4VcLicGm4pZR77MK4fDYVitVszOzkr+GzjofMPn1MJJZdFtsBaFrJONys/KKEMzvXUOiqkIliJRgRKdKRQKUp7E6IzKlHklAMhms/D5fFKfzKhVw9jA/kXgbF424GfU6/F4JALne+iIANh3SIrFIlwuF4rFIhKJBLa3tw3QPD16t9tt2BNNDus2R033EdDMWBppwNhvn3vF1zKNZbPZJDrS8sPac0Y4hDh5PsFgUM5RE37oOPT09MgAGyI6dPIdDoeU5JBUSPkhxK6RgGaziVAoBKfTif7+fvT09OCRRx7BwsKCATnM5XJYWVmR+ciAETXptqoOzc4ncqG/pqFubSeItPE9WIWhc8/sp+Dz+VAulzE4OIhEIiGBXD6fF31QrVYRCoVQLpflXGlcmUsm/4DwOIM1u92O1dVVSW1oThQRAuohMsOHhoakXz+bolBmcrkc1tbWXte0S8um3ovvtu7IQGtoQhto/mJNEuL3dYcuu92ObDYrF4gGEIAYYJPJJDA3jSE3OR6Po7e3V8onPB6PKPlWqyX5ZvbRDoVCKJVKcLlcKJfLyGazovABCOTBvIR2QEhYsNls2N7ehslkwtDQkETK2lsvl8viQQNGUpj2Irth6UhJR1EaggYOBhEQHibC0tvbK6xGr9crjhH/JkzM3C/PmjA2p14x78XzoKEnkYwGlUQSOnzlclnyjwAkstZEF+Ag2mdUl8lk4Ha7cfHiRSGn6TxmqVQyRI9aGXWbDADGjmGdzgdhPs3i7jRMVMqcjKZ7L7P7FmvlTSaTnDuNNwlmPB+mN6jcNbmM0KfZvF+yQ9lg3pPPqs9MR3xWq1WQmpGREZRKJVy4cAELCwsSCbVaLdy6dUv6M7A8UDup3eak0dnWzrheDM50N0m9dDqL3Rl3d3fF2eHdjUajiMVicDqd0oKZZ8R753a7hRPCXDRlig4h4XHC1blcDgDESQuHw+KU6Xw6AxKHwwGfz4eNjQ3Y7XYcOXLE4HCWSiVks1lJjejApHPdbrndHUfQOuEPvB7K0GQgXdbCaELP+83n86hWq4buYySHkWBBT5t5pUAgIJ4xIfeBgQHJT9hsNmSzWXi9XuRyOVH8jKJ1hESh0RtID5BdwyYmJiTpPzw8LGxRDlHQ9ZU6/6LzTt3kNeuyMR0tcQ874W3uJVez2UQymZT9KhQKwmsADvK37XZbUhg8P0JfRFny+TyKxSKq1Sr6+vpgMpng8/kERvf7/dL4gA0C6PHyuSgPnd4usD9hjXBatVrFCy+8YEiTsAkKnU+NlmjSUDdyEQDI/mnnXC9N9tLpLirJdnu/nI55R0KCmUxGyKI9PT0oFArweDySXnI6nahUKggEAtIj2efzGepeqT8sFotEzmxsVCgU5Nk1X4ZGVJOT2u02wuEw/H4/Tp8+jWQyiZmZGUSjUZFxkklXVlbg9/sltaF1ImBEG7th8dw1hK9JYjx3piC0bQAgaEm5XJb8czabRS6XE4fb6XRie3tbGhJRppgWZStWljyFw2GEw2GxF3T4KR/8OTLwd3Z24Pf7JQ3CdBbtEAAJHPx+P44cOSJM8LGxMWQyGWHw1+t1pNNpsVfULZqn1RnQfLd1R9JCz0QPGtAXEIBhoAE9CCpDRkSEoH0+n+SaieEzP8DLQ6iBBd75fF4gj8HBQSwuLkqpjNvtRk9PD86dOwefz4czZ87gxRdflBwnPW3mJ/mZNBoA7DPzhoaGUK/XcePGDVHG2WxW3osda3iAnaQIKqNuI4YAMJCyOiNGXkIKOvdIoypk39OposepW7ty+AHzyIS8SM6ioh4cHJTuTozK0um0RDw+nw9f+9rXBH6mvOquP4BxlCQA4Rtks1ns7OzA4/Hg2LFjWFpaEmNOlICfWZOlOvkH3SYDmminSZGdRkkbp06+Alumdr6GpVSMnHX3Pka/ZO6GQiEcOXJEuv5xuIXL5UIwGMTU1BROnz6Nhx9+GK+++iqazf0Og/l8/nXGUjORgX1jMzAwgIsXL2JgYACxWAyvvvoqfvzHfxxXrlwRxWy1WrG8vCz5ceYrNXObiE83yYEOvDR8r2WfOlcTCEne1LqeHAMy7sntSSaTgpgRJWFagzo4Eong3LlzOHv2LBYXF5HNZjE6Oorjx49jeHgYTz75JM6fP4/7779f+EisFuEz8pnIadIpk8HBQdxzzz0YHBxEJpMRxM7n80n+2mQyIZPJGBruUC/Q0eOe3QlZ9I5Z3Gz6zwtKZaeJWdxwGmLdVWtkZAQWy/7UKHq2zP3RuJMMotm/Xq9X6o+z2Sxu3LiB4eFhPPDAA2i19ucyFwoFRKNRxONxVKtVfO5zn8OTTz6Jvb09IZZoco+OnJrNJvr7+zE1NYVyuYy9vT3s7OzA4XBgampKGH+aws/mKNpDosdIb/yN4L0382K+VY90BIyTYHgJqcAYwQL7Ajo4OAhgP4/MunpGIbw4lClCmjTUhNPq9Tr29vbg8/lw5MgRJJNJZDIZqVMsFovI5XJ46aWX8KEPfQipVMqAvLwRBN1o7LeMrFar2NjYECie82Dj8bg8DwB5dp1P00pfEw67LYLu5Jp0RsqdqAEdIn6df1cqFVQqFZkDzVwfSWI8M8pCtVpFMBiUvDQds2PHjqG/vx9+vx8jIyNwOp04deoUTp48iWAwiMXFRSEYFotFicb53jwrKtfJyUk8/PDDOHXqFGq1GpaWlnDz5k287W1vw4kTJ2Q4B2Vnfn4e0WhUIHMNk/N3dKMcaOOs2dz8/ExZ8nN35veZjkgkEggGgzLEhgx48oOYt2Y07nQ60W7vl1wmEgnMzs5iYGAADz30EEZGRmC1WjEyMoJjx45hYmICoVAIV65cwfT0NC5fviz9/4nKEe3TqJff78eP/MiP4MiRI4hGo0in07hy5QruvfdeTE1NIZPJGNrYMoUGHMg7z1wbZRrt21l3nINmRyVdfwoc9NDV7G2toMmkO3bsGOLxONrtNra2tgw1r8xVkszF92cXn97eXvT398Nut2NgYADf+c53UCwWpZQKAJaWllCv17G6uoozZ85gYWHhdSU9mm1qMu2XbAwODqJcLuPmzZtIJpPI5/MIhUIYGBiQnAMFioMSdI6iE+rV0XM3wVqMbnXemBeUnicNNYl2zOMRCbn33nuxvr6OSqWCXC4nnYMsFov0vqazRxkwmUxSUwhAzmd+fh4Wi0UY3q1WC7FYTODMRx55BGtra4hGozIAozOSAyCciFQqJcQzv9+PoaEhNJtNaWpD8gkJiBrKBA4MfrfC2p2r0/ns/NxU2oxGtC4YGxtDMplEqVSSfCOHETBNxa5dvL+cPBQIBBAKhTA5OSlNIUKhECKRCMLhMO6++26USiX09/fj2rVrmJubw/DwsMhZJ4kV2JeFSCSC8fFx3HPPPXA6nchkMlhdXcXi4iLe/va34/3vfz9eeOEFABDG7tbWliABzD3z+3ofum3xflPmeWfptGhnligCeT0M7Ox2O9bW1gR2ttlsyOVyKBaLhsZUemId88terxf33XefzIEmyWt4eBirq6vY3d3F1tYWbt26hT/6oz+Cz+fD3/zN32BsbEyenVC4Xg6HA29961sxMjICl8uFzc1NfO1rX8PNmzdx8eJFQW34Hul0GltbW9JqlM+tUSZtf1hlcDvrjhuVaChb/7/z34T8tGdFD2h3d1c8ZK/XK7k8rdyZPyDkyT64wMFsUCb9d3d3pYYtm80iHo8jGAwiHo/D6/VKq0YKFZW+xWLBhQsXUC6XUa/X8dJLLwHYRwo8Hg/Gx8extbUlpUAkjQQCAQCQmmvgwAvXBrvbjDMAOUPuhc4/aga8Zury/3TaGo0GFhYWBK7Uc7cJiWlGr3Z6MpmMfE/DSXt7exgeHhZYPJlMYnh4GIlEwiADmtzEyzMwMIDFxUU0Gg2srKwgFArB5/MJmWRtbU3YocD+ZXO73SgUCgb0RMNZbwRvd5vB1sqYi/v6RsQ4Rk68vz/0Qz+ElZUVNJtNafQD7JNBOd+bjrnD4UA+n0ckEkEoFILf74fX60V/fz+CwSCazSbuvvtuqT2lsXzmmWcwOzsrkCmdQN5/GhGHw4FHH30U586dg9Vqxfz8PDY2NpBMJhGLxfDggw/iXe96FzY2NkSX0SBdv34d4XBY9ESn00I92G36gHdPO+hcWo/zzDurexqNBsbGxqTMMZlMGkijbERDdNVsNiMcDsNkMmFmZgYejwfBYBDBYBA2m02CtfX1dZw7dw59fX2oVCr4h3/4B3g8Hrz00kvSIpa6pdNuve9978PMzAzsdjuuXr2Kp556CltbW9jd3cWjjz4quWzWy9dqNUSjUSGHdfItuCdvxOq/nXXH0qLxeUJWVL46L0kvQXtWFosFf/Znf4Zf/MVfRLVaxcjIiLTRZM0bAIlSCW3yQvv9fmxsbEhemo0ixsbGsLq6CgAypYqHrPu/MheuJ+M888wziEQiMJlMOHv2LBqNBoaGhjA2NiYD1nWfaAoXHQ+yf3X3LH0g3UYSo+LVtYb6b0I7VMa6uQMv6he+8AV89KMfRTwex+DgoESkTG1wP5mbJumr2dyvh89kMhgdHRUn0el0YmRkBNVqFadOnZLmAVQaun2gZo5ms1n09PTg5ZdfxuTkJFqtFgYGBoT13Wg0sLq6KjWXwEGEyFIwQv2d3IM3yjXe6eV8M6zO3HInogAcDE7odGIikQhsNhvS6TRyuRyGh4extLSEQCAgryU6Vq1W8cgjj8DlciEajYpTZ7VakUwmcezYMYm+LBYLYrEYnv7/Jg7lcjlBwADjtDOzeb+74Yc//GHUajXEYjFMT09jY2MDGxsb8Pv9eN/73of7778fN27cEBQA2EeTOIlJO4yUd8A4q6DbHLROgjDvFnDgrPPfuiyP97bVamFxcRHvfve7pdRuZ2dHZIiIKvU/e26Pjo6i2WzC5/NhYWFB0Nfh4WGZ9b6xsYH5+XnU63VcunRJ2PbssUF9zTx4MBjEXXfdBbfbLQjsuXPnUCgUEIlE8DM/8zOIRqNipzRKtrGxITZMl5sx1altAeX/drkId2SgmUfUh0Ivil+nUuz0rhhtNRoNRKNRrK6uIplMwufzYWBgQDpzWa1WaUrPaUWa2EOokt5wOp1GKBTC1NQUksmk5At5aDqa0ZEdoY+TJ08CgMwmPnXqFPL5PNbW1mSEnYYjCIcQqtEbz8hP50tuN9fwZll0SP6ppckidNqAg5pYOnCRSATXrl2TMrhwOGzo8c4yCwDY29uDx+MRaL3V2u9Axx7anMXabrfx3HPPYXh4WGAwRki8FJphStbl2bNnkc/nsbq6KvNdNzY2EIvF4Pf7DYxsykO5XH4dCfCNKgJ0vrWbHDWiIfr+A0YWL/D6kkPelXK5jEKhgA996EN4/vnnkU6nJe9HrgvbNlqtVkxOTsr7sAsU+zJbrVZpGepwOLC0tIRYLIa+vj4sLy9L1KMDB54Tm4yYzfu11f39/TJG8Pz583jiiSfg9XqRSqXk5ylTZrMZq6ur6OvrQy6XM0SHb0SK67alU1c6ctQBDHDQP4Pf13ehXC5jZGQEFy9elH7aOr3JFGhvby/8fj+CwSCWl5dl5KTX60UsFhOIOZlMYnR0FB6PB+l0Gjdu3MDExISh2yRRDj5rrVbD2toaXnvtNezs7KC/v19ajT788MMCazPnTL1POWI/cKZ0+d5ETDQPQTPDb2fdkYEmaUsnv2k8NbSryVJ8SF4Mm82Gv/zLv8RP/MRPyAfc3NyUbjK1Wg3BYBButxs7OztC+IjH49Kej3N57Xa7EEASiYQ0riD5y2LZ7x5E1jAVCQ393t4e8vk8/H4/nnjiCWxubuLq1atwOp1SrsOfIUOTsC03nAKovWYNcXTbomHrJAfx3/pvKil6xPxjNpvxzW9+E+973/vkku/s7MjFabVaQgpMpVLSgjOdTiMajWJ4eBizs7OCuNjtdsRiMZhMJgwODorSJvJB4hE9X/In/H4/tre3USqVEI1Gcd9992F9fV0iJxpnnidrb8kkZQ2uloNOb5mrU2F3wyJRUjsvWvlqB7lTSdF5ozPFemjgICVALorf7xfHe29vT6BFv9+PZDIp8LLH40E2m0W1WsX8/Dw2Nzfh9/ulcQWdK9Y/07nI5/O4ceMGjh49ilarJSV799xzj6H8U+dE+Rmz2ayhBEjLuXYIGV12kwzwM9FAA0ZdSCOtc/50wPVezc3Nob+/H7VaTTrDkYuQyWTg8/mkvHF0dBTj4+NYW1uTkiw6bA6HA+Pj49JIyOl0Cnri8XjkPRhksc00eTWlUgnVahXj4+Ow2WzScGR6ehrr6+sADnQ+DT4rAtrttjh6OqDj56eevNMUxx29+o3yavqXagXN6LkzrCd0RQ+FJVbxeFzaqyWTSTHShUIB/f39CIVCyGazKJfLMuSit7cXsVgMrVYLS0tLGB4eRq1WM7Dk6I3RUGh2ncfjQSKRgNlsxvLysoGAoAVJw1SEtYGDZL92TiicnXvSLUuT4XSuUX9eHWVQeesoljIQCoWwu7uLTCYjF4Tsyr29PXHaMpkMBgcHpUVroVDA6uoqPB6P1MuaTCYsLy8jEokYamoBSMRLWdXPxd9ns9mwuLiIXC6HiYkJgcQI0+vP3zl1i4pY55kYTXcSErtlEcrlnmr573TUACNhRiMPNHp+v9/g2FOB9/b2YmZmBrlcDl6vVzpGmc1m9PX14Tvf+Q7cbje8Xi+SySSy2aycP6MxTjSz2Wyw2WzI5/MGzkitVsPm5ibM5v3Wvw6HA7u7u+jv75d+B5Qn9o622WwywINcGv25/qkz70Z9wH2knux0xjpZy/wa7w37HASDQUSjUQBAKpWC3W6XTmFjY2MIhULipLNEl02oTKb9Npxra2vY2trCt771LTG4/xQniD04+G+LxYJ4PC7VRgMDA6jVatja2gJwUB5KHcAe4qyrByBInpZ1QvrcG96d21l3PA+6U/D0hWJ+RrM16VnW63W5GBaLBceOHcMzzzyDeDwu0AO7f01MTGBhYQGpVAr1+v78aNZGzs7O4iMf+Qh6e3uxsLAgB0WYy2TaZ31Xq1VDy8jO2mR6/+l0Gru7u3A4HLj77rsFwtAds+glNhoHjfnpfdlsNiGddJLh9D50y+qMDHR0RMPE82bzDqYbGHlSBi5cuIBnnnlGUhrMQft8PoyOjko+qlqtGnrs3rx5E//+3/97IYcRbvX5fPB4PIbccG9vr1w+Rj58fpvNJgM3SqUSfD4fpqam0GwezKHW0YBuV8lWpeQyMCqn8tE/p2squ2l15lnpeAHGATeaKGo2789050zvUqmE48ePAwCSySSazf12iZFIBFNTU4hEIkIgYw93AFKS127vt4bc2NjA7OysyBv1CXPVDARyuZwB9WOtfiKRkJpa6g2XyyWcCI0SUKctLi7C4XBIlzmiCTqC1ohiNxHEAKOjxaVz7bovAM9FBzm5XM5wR1ZWVgQZbbVaUjZ77Ngx7O7uSifJYrEo/IXNzU309/cjmUxifX0dsVgMly5dktey5z+JgNTrJPTRXlGn7+zsiK6ZnJxELBZDqVQSIjHPlQaZYydZadDT0wOv1ys8KebSKQO6s+btrDselqEvHb/GAyIeb7FYDJ1YAEinFxqw1157DQDgdrtlGonFYsH29jb29vZw5swZpNNpMYzlchnRaBQzMzNwuVz44he/KHM9eXkIW7Va+60e9/b2DDkHTVoAIMM40uk0vF6v1Dl3lowRmm82m1ID3G63pYsVPzMbahAG7caLqVnQ2thpchT3iLAl5cbn8xmafLz88svI5/PSrIQXJplMYnNzE8FgUFjYdJRGR0cxPT2NcDiMZ555BsA+MZAjSDUpg9ESqwE0xEjl0tvbi1wuh0KhAJ/PJ73bNRpCY8MzZjkInYdW66A0kKkVDXfz93WbHHSShHhftDLW0ZTOu9EZq9VquHjxIpaWliQi6e3tRSQSgdVqxeDgoIz9JBOfKFcul8PZs2cRj8cxNzcHi8WCubk5cQ4tFos47SaTSdrAdhIHm80mMpkMUqkUTKb9SWs9PT1SEcBzJCO53W4jn89jb29P0DQqfq0DGLEDxkES3bIYDHXKNXWf1WoVzgrvEAm8ujSWPa0ffPBBcbpbrZa0ct7b28PJkyeRSqXE+SVqwTnSV69exdLSEm7duoXJyUlJT9FZoyPOMk6NeurUG6Nuj8eDYrEoRp5Ls78LhYIYc90Rj8N92DecLaV1yu921x1H0Nog85dp6JLeD71mKkY2oiDF/datW/jIRz4itXBcExMT2NzcxNLSEsbGxoS5Z7FYsLOzg1Zrv0XokSNHhAFOI697sTIK0oegu9FYrVaJgmq1GnK5HAKBAOLxuMG71uVChFO5F9r7KhaLElUzcqPwdhOsRUeFivmN8u7aGNKwURlzb8xmMzY3N/GzP/uz0iGsUqnAZrNhZmZGZgAPDAwA2K97rtfrmJubQz6flzwxc1Umk0nykgAkH6QZ+PpyUKlQEdMLDoVCiMVi0iKQ0CihUJfLJYxdOhv8nFT6nahBJ5uzGxbPn7Kt88faYGt9QcPocrkM3ARdV2q32zE4OCjtXBnZbm9vIxQKCerh8/nQbDYxPDyMy5cvIx6PC48hnU7DbDZL203KFnDQKIOIBz9HrVYTxC4UCmF6elocbxppInNmsxkrKyviALJzoi4rpC602WyGyXfdZKT1naccMIVEOaCjpFMifK12YmKxmNQnWywWaeF67NgxxGIx7O3tYXx83DBJjtUWR48excWLF5HNZjE2Ngar1SrGkw4dZZFBHFOkdLr4p1AooFQqoVQq4ejRo1hfXxf9RSSVQVu73ZbGJLokrFqtIpfLyYwGOhsctvE9g7g1m1XXptKI8dD4EBR+Qgqcy7u7uwu/348TJ05IEj+dTgvph4YwEAggn8+jVqshkUhga2sLrVYL3/rWt+B2u8Ur7u3tlVFwfX19wgRlOzlS6QlJ6wtEWLRQKCCZTGJjY0Maaugoi4fCvAU9IkbXFDw2aNc9xjsL4d/MiwYLOGDlamNNw6dZs/So2ceczprH48HQ0JAY23K5jMXFRVFihLsLhYI4Qbu7uwCAhYUFnDlzBgBkqpnP5xM2cLO537y+UCiIbFKRMj+unTbW2W5ubmJxcVGcO8KZjO5pmHVNJ2BsfkOHgxEkHcLbbU7wZlkaUes0PPw/90pDvfx+q9XCysoKXC4XfvRHf1Rqz91uN7LZrLRbpaIn9Gw2m6VRzZUrVzA7O4udnR309fVJB6rx8XGcPn1aUDsNh1MmyR+gHLDfe6vVQn9/P1ZWVsTB6pwRvrS0hFAoJCkdjbYwQiMMyrPXQ1q6ZfGu6fIqzdrmvvPsufcM1qrVKtbW1gAA9957r3Rk8/l8aDQahm5ydrsdmUwG7fb+6E+XywW3242rV69iYWEBJpMJa2troptHR0dx4sQJjIyMiMPH83M4HGKrmP5yOp2IRqMSFOrhO3SwibgwUib0rZndTNdRz3H8JHUJZ1PfzrrjCFo/JIVXe8k8rE4mGy9CtVrFhQsXcP36dVy7dg3APszNAwGAWCwmCp5zQFOpFI4fPy4wYjAYFAHhXOBXX30VCwsLyOfzOHfuHB5++GGZVKKZ5mzfOTAwIGxxh8OBra0trK+vIxqNiuIGDjqoFQoF8aj5+Rkp0SMnGqAFVA+LeLMvHUHrRiU61cBcrC5DAw4YjbVaDZOTk1hdXcXW1pY4cB6PB6FQCDs7O9jZ2TFEaZxMdtddd8lcWJ6Hy+USb3lubk6QjKNHj+LIkSPY3NyUKWp8ZqvVCr/fj1KphEqlgkQiAbfbjVgshoWFBfT394tzRoeLeVCdW+ffjBJ0KRkNF+Xudksr3iyLd14jAxqpIJpAJ1477QBkYMGZM2ek5/m9994rEHWxWMTo6CgKhYLM6XU6nVhYWJBz+fa3vy3Gu93en4LEs9vb20MoFMLo6KiUcuo7SvbvmTNncO+998JsNgtJyOfz4ZVXXhGngJAs022UPc0x0U4YDTRlh7LChkndsBiVUv4p9zoNRvmnk6aja+qCvr4+bG9vSwRcLpclLUV4GICwsa1WK1ZWVqSPwc7ODmKxGBYXF9Hb24t4PI7z589jdXUVe3t7aDQaeOKJJ/Cv//W/lvnNfDZgf6rh2bNn8ba3vQ2Dg4My5rRWq0m7XwZ0dNw5GpdlYABEFthki1E5sI/oVSoVQWxvd92RS08vSOcd6H1SKTJK0RfVZDKhVCohmUxienoas7OzsNlsuHz5MrxeL/b29tBs7hd637p1S/JPDocDyWQSk5OTeOCBB7C8vAyn0wm/3y94P0swarUa3vKWt8DhcGBtbQ03b97Ea6+9hocfflhyFxSevr4+xGIx9PT04Pjx44jH49jd3YXb7cbk5CTW1tbEMFDYQqEQ1tbW4HK5XsdU1cy+TviXUVe3LDon9FIpB8zH0pskPKhRl1Zrf2zjsWPHsLKyAofDgRdffFGmRSUSCQwNDUm3sGazCa/Xixs3buD06dOo1+vC2g8EAjLEhCUWGxsbGBoaEkN78+ZNLC4u4rHHHsPu7q7BsQgGg7hx4wZcLhdGR0eRyWSws7MDABgbG5MRowAMKQ4O6iARUEcL3ActA0RedJlftyyz2WxguHNRIevvUQZsNhvi8bggHC6XCzdv3sT4+DiOHTsmStXpdGJ8fBzPP/88xsfH0d/fjy996Ut4xzvegYmJCZTLZTz77LMS1RKhe9e73iW/2+/3w2KxSB/2kydPYm1tTSLi/v5+PPTQQ9KdanV1FbFYDMPDw5I243kyd0m95Xa7pcRHOyoaWdTsdt6DbpIBpgsdDoeQt/QeABDboB00YL/++caNGzhx4gTW1tYQCoXwjW98A/39/VK2lEqlUCwWcfLkSSwvL2NmZgZbW1tot9uYmJhAq9XC5cuXMTc3JzXUFosF73nPezA+Pm4YMxuPx3Hp0iWEw2Eh9nES1dmzZzEyMoJ4PI5jx44JmZXNseLxuKRPOaij1dqfhMUZD4CxORXTKJqDRH2hS/a+27pj1oo2TvQKCRsSAtYMTs3wZKQTDAbRarVw9OhRTE5Oor+/X+DORqOBqakpxGIxpNNpvPe978Wzzz6LeDwu5A0eoNPphMlkkp6r6+vrmJubw3/7b/8NExMTeM973oPt7W1xFGq1GpxOpxDDnnrqKYRCIUnmM89AgQIOGpPoz8pLR+iSQqkhHJ1neCMl9mZdjD51mQSNFOEinYcDDiJqPczE7/ejUChgfHwc0WgUjUYDAwMDGBgYgNVqxbFjx9BsNlEoFPDwww9LN6Bjx47hxIkTEjHRo11ZWZFpMtVqFV/84hfx+OOP4y1veYtcan1+S0tLyOfzeO6553DXXXcJMlOpVJBOp6V0A4BAYGyQo2fb8vPrSgFdXqUj7W6CN3nftfLpjEw6UQRgP1p58cUXEQwGYbFYJE+7vb0tbROdTid2d3dhNu93+bp69SqazSZmZmaQyWTgcDiEbVsul5FMJtHf34/jx49jdHQUwWAQ4+PjyGazuHz5Ms6fP4+dnR20220hFVUqFTzyyCM4ceIEtra2sLa2htOnT0u+ksQ/k8kkJaHUabu7uxI5aU4D776u++W/GV13ky4g+YroJM+ed0BXMtBYAZCe3MB+W9e+vj40m0089NBDGBwcFM4HAOF9FItFXL9+HefOncNrr72GRCIhaCn7r9vtdpw5cwaBQAALCwvIZDJ45pln8Au/8AvY2dnBPffcI5A0HYr77rsPg4ODkkZlvwQGIDTMvN98No5N1tEwv89qH43aAvslv5xnfbtIyh0ZaNLMO0uWNBGMAshcHYX41KlTAPaV9fHjx6VLGD3X8fFx7OzsIBqNolqt4l3vehfm5uYwPz+PixcvolwuI5VKIZPJyPM0Gg3Mzs5ib28Pu7u7qFariEQi+Ff/6l/hd37nd7C1tWWoUyXLk4p2cHBQGlzojeRnsFgsAqtx7Jn2gPXYyk4iGGEc5jG7ZWnjq/eMZ03PkXAyPeh2u21oeD8xMQFgXx58Ph+q1SpmZmawvb2NaDSKer2OBx54AIuLi9jZ2cH4+DgAIJfLSQcp7u3c3JykMtgK8sknn8SnP/1pgTRpNPP5PFKpFMLhMKanp3HkyBHE43Hxgplf0peoXC5Lb3fd050ISycRUBsm/p/9hLtlWSwWQ5MS4ADy5n2iDNC4cZ67xWLBa6+9hrGxMbzzne+Ez+dDqVTCxMSEIBexWEyc+tHRUWxvb0taa3V1FS+++KLk9RwOByYmJoTsNz8/L3rlm9/8JnK5HMbGxiRfzMY0Pp8Pe3t7WFtbw8LCAoaHh0XBEopkhQnlo1QqIZPJyOejQdZsdW20+Rpd0tMti/qPUaHWCdQT1IHsLcHZ7bzPt27dwnvf+14cP34cy8vLWF1dhcvlgtlsls5w8Xgc99xzDxKJBNbW1nDvvffCZrNhaWkJ29vbMnxpamoKW1tbuHz5MpaXl3Hz5k2kUimcO3cOX/7ylxGLxRCJRITMxdbR8/PzmJ+fR6lUwsbGBiwWiyC0Xq9X+nGYzWYhfaVSKWl8QmeVOWedpyZBUPckb7VaSCaTt7XHd2SgeRBc2rAR9uNBmc1mbG9vY2BgADdu3EAul8Njjz2GRqOB3//938fY2JhEv3a7Hbu7uxgZGcHVq1fh9/sxOzuL06dPo91uI5vNIhwOC4mkt7dXIKjjx48jk8ng4sWLyGQy6OnpwcDAAEZGRgwTUTweD1555RWcOHECOzs7GBsbk2EY9HbYtYgQOi8hI0YqY36dxof7wmhaU/bz+bwwSLthada+NtTAAUFQcxKuXr2KqakpJBIJyftvb2/jj//4j3HmzBlkMhncunULJpMJiUQCgUAAt27dwtjYGK5fvy75RJY+mUz7k4wIO6bTaRw9ehTtdhsPPPAAyuWyzPrVda5skk+ojKMu2c+Xk6lcLpehi5gmOLF0TDtwPOc3kgEAIkccwNIti86JbtRDpaxz7rphjMPhwGuvvYZjx47hrrvuwtmzZ3Hz5k3E43GZGmQymZBMJuH1epFOp2WuM6FUv9+PRCIhZ2ez2aR/wfr6Oq5fv45QKIRXXnkFPT09OH/+PC5duiQOIqtMzGYzNjY2cP36dTidTpw7dw4rKyuGM06lUoZmRa1WC1tbW3A4HCiXy5LL1npQo0pU3JxBnE6nJfLqhqUNskaMKAOsfOAdbLfbOH36NF566SVsbGzgrrvuwlve8hZcv35dyFM8V7bvTafTqNfryGQyuHDhAur1OsrlsqTBCFPffffdUuo0OzsrjlipVMLQ0BCCwSC2t7elDr6npwcLCwvScCYUCgGAlNrpLoWcokcnkyNFAePEMjrvlAEGh8CB3tzb28PGxgZSqdTt7fGdHIgm/jBCJJRDmI8eAw9qaWkJ9913H6xWK5566in09PTgl37pl/Diiy8iFouhv78fVqsVw8PDuHnzphhHztZkr9y1tTUkEglh/2azWWlSMjIygrm5OfT29uL69euwWCyYmZmR0hw+bzQalctP2LO3txflchk+n0+GMDBKJzuTpDeW0WiCFAd2a4YwnQoyP3UZ2Zt96a44gLF7HCMmHVWMj48jFovh1KlTSKfTuHTpEgKBAH7yJ38Sly5dQrPZxMTEhBCv1tbWUC6XkcvlDGVrgUAAXq9XIlGeoc/nQ61WQzgclm5Q/Prk5KRUATCyCYVC4v2y3ze7FdHp6+/vR6VSkQb9LN1hPS4dEMoWYXbm4RlR8/ydTqd47d2yCG/r5i1aSWsyGADs7OxgfX0dLpcLd911F8LhsOwf4WcOu+jr65O8ITuM0QG6evUqrl+/jrW1NamDdrvd2NvbQ6lUkpIcRsBMq5GsxTRMLpcT5U9Ha3l52aBUc7mcoD6870tLS9KtTjcv4ftS9vW/U6kUNjc3sbe311V8FKYoWDapORvUC2ziYzabMT8/jy996UviEPX19eHixYvw+/24efMm2u22OGpsMMNugoxISea9evWqcFHuv/9+Oe9yuYxAIIC1tTUUi0WYzWZ4PB6Mjo5KqoJoWiAQkLnepVJJhiaxxtpms0mPdzpotEfUD5RNbQ+ILgGQtEapVMLa2hpSqZSUDd/O+md1EtP1jfoPcFATyqYSDocD3/72t5FMJuHxeGA2m/Hcc8/hzJkziEQiqFaryGazuHHjhng9hAmY7/N6vRgdHQUAg4LlJugL0dPTg2w2K8LBy2U2myVnzfrFUqkk32ODBH4OKh8qYl1GxsiBh8Y/jCZ46Vkz203EEF0yRaGkk6aVM71rQpq3bt1Cs9lENBpFq9XC7Ows7rrrLmlGwJm6gLFEy+PxoFqtYnt722D4FxcXhRSiI3juO5EcygY9eMLdnS1cOYiDykQbG5ZdADDUPutImpEEPWXKL8spWN7RLYuIQieqpqF/wMhZuXXrlvA7xsbGsLm5Kfl9nk82m0WhUJCeBLqkz2Qy4datW3LPH3roIYyPj0u7XirOeDyOVColz8A53gCk4iKXyyGXywkEv7e3J5BkT0+PYRQqn42RnJ7GBLy+7TGjbspdLBaTetnOVNibeTGq1AQwpjypj+mU0ln3eDzY2toSBv/S0pL0vGCzqEAgYOhIScJpT0+POO3FYhFbW1s4e/YsstmsdArjvaaTXygUcPPmTfk3IWrWX1erVTGeKysrKBaL0tOis0qD/AQSxYCD8bv8W080457U63XE43GJxLk/t7PuuBc3P5yGOnWTAgBSJlGtVhEOhwWGZlNybgIp6dPT03A6nZiYmEAoFMLKyopE0u12WyaVDA8PS8cpDTvqbi8UGMLKNJi81JyQQ8q8zp2azWYMDQ1JVyI6BxzAoWt6edH0FBMqEQqLjrK6ZbHUROf2dY4fOGDxkmVttVoRCoWwt7cHs3l/ahDLpGjcx8fH0W63ZbTjzs4OgsGgnFU0GjX0vmVkQ4XIs6Vc0OOmQ0nFwdd2lsnwvVl+RdiNHrl2woADGIswHg0R5UM3w9D937tlaSXE8hpNCtTOD/XAfffdh5GREWlMwd7ZXq8XJ06ckJ7aXq8XExMTQhTlbGiSQdfW1vDoo4/i7W9/O1555RWJ5nn3isWiICesKuAseYvFIjlFBgStVkv6F1B3eL1eMewkFZGrQHhbQ7qde0MdwT7OWl66ZdGI6nOnUwIc6GLev1KphOHhYfj9frRaLYmWi8WilEPabDb09/cjnU4Le/vKlSsADvr5b2xswOPx4KGHHkIgEBAiL404o1Y6UuQOkahGmaD9IspFTgVLpBqNhuglp9MpneVYEaLPUgeodDgZpCQSCcMExjvhJd1xBK3LqHgwut6VipnezuzsLOr1OpLJJIaHhzEzM4Pd3V14PB4higH7MOH169dx8uRJXL58GaFQCD09Pejr65OJN7FYTKJmwgtkCFIgGM1rhiEvIin3HJBA4aGgJRIJZLNZJBIJTExMIJvNCmGBzoImwgAHjRjoQQGQaEwbi25ZjD7JXtSpDo0o6O5bq6urwtr1+/0YHBzE9vY27HY7/H6/OHTtdhtra2vo7+/HK6+8giNHjsBisUjfXfa5ZZMBfYYUfCImfFZeShL+qHB1Jyn9WdgwZ2VlBadOnZL8Mp0NDWvyZ7RxZu6JfX/5Gl1u1g1Lo1bcCzonek9oxGksmbowmUyYnJwUwh7beI6PjyMUCsHv96PRaOCpp56S6VI0dsePH8fb3/52vPDCCwCMzXOImumWvVzUT2xaROSDZ6P7JzPSSiQSQlpcX1+XdIhucqLvAH/eYrFIn3/dT6GbIG7gILfKf2tHnftNmDkajWJ7exuRSAQ3btxAu92G3++X3DybGDUaDUxPT6NWq+HBBx+UYI/530wmg3Q6jYmJCUxOTorh1igq9QX/7XA4xBlntQYH9NAYU184nU5BB0hC7OvrQyqVkgie6ZJO1j4NP2WPgzQoW81mU3gpt7P+WRE0LyX/ZoTaaDTgdDqxurqK7e1tgaZ3d3dx8uRJbG1tYXFxEcPDw9jY2ACwb8wymQxeffVVBAIBLC0t4WMf+xh+5Vd+BTabDVtbW9jc3JTOQbrujIfPKVQaaqQx1VGdjsqpqKlQaVQmJibE2+Il46Xn+9Ij7IT2+XX9RxOGumVx/3Spkfae7XY7VldXxSkaGRnBa6+9ht7eXmxsbKBcLiMcDgvZwmKxIJvNYmFhAQMDA8jn8/jYxz6GX/qlX0IwGMS1a9dk6hURFMoe95fd3eggUB41u7bZbAqkzp+jDNBBsNvtmJ6eBgDxxre3t+HxeAxoUWfUxPPXqIl+DRV3tyyiaNxHKiT9mfn/SqWCgYEBeDwe7O7uwufzwefzSQepgYEBxONxGTXI0ql4PI4jR44gkUjAZDJJGeXjjz8uz0BHi5ClZs3ScWYnOC6TySTkTj4vqwn4++lgLi0tiRO5vb0tzF2+j35POiPsYri+vi5dpYgk3EmTih/0RT1Jx6Qz984KimKxiFgshmAwiEgkgnK5jJmZGSwsLMBqtWJ6ehqbm5tybgCwvr4On8+Hq1ev4ujRo/jCF74Aj8eDWCwGAHjggQdgNptx+fJlgywCkJ4FjIR5TjabTRj5TJ/pKVS0J06nUxA2h8OBGzduoNVqwel0ikGnPnkjHU9Ep9Xab0vNYJKo3p3YhH9W935dmK2jF0aqU1NT8Hq9uHnzpsDW29vb6Ovrk25MbrcbfX19QqWPRCLIZrN48skn8d//+3/Hhz70IblcFPJIJCKtQUlrpxdDIpcmrZnNB9OTrFYrIpEItra24Ha7DUJFyJa5MXYu43gzPoc2yJq5CkA+PxU1IY47yTe8GRYdD21wNMzFvA7zTdvb23A6nRJRBgIBIQG2Wi0Zdu/3++HxeBCPx/H+978fH/vYx/D444+j2WzixIkTQvgbGhpCqVQyDMCgZ0ynkYaUUQ09VovFgnA4jNXVVSEfUaGSDWo2m3HkyBFRJNFoVCI+XWIFHJBhdHqj01BRWQHoKuXMO8UzoFzwsxJmpFFjauv06dPo7e3F5uamKMJCoSDEq0wmg62tLcRiMVy8eBHT09NYXl7G3NwcisUi/u2//bd4+OGHcfnyZUNfd56vyWSSyVIaSSPsyP9TV3FMpM1mw9DQkLRlBPZZ58vLy1J3zRw3a6l1mSGNPc9/aWlJeAqaUNRNThqNM0vTdGkVg7Zmsynkq62tLeGkZLNZzMzMoL+/H7FYDNFoFCdOnJByWZPJhJWVFczMzKBSqeAjH/mITDg8efIkwuEwms0m0um0odxPlzPRaWNUTQNLiNtut0tpJs/R5/OJk0e0dXd3F2tra8L4pp7XLZ7ptGleQj6fl2Y3RJCoI26XMHpHBlrXeVHYNEGIsBILvSORCBYXF4VNxx7HFy5cgNVqRSaTQblcxuTkJCyW/RGUX/7yl3HmzBm4XC4899xzCIVCUoPMecBUdIQgNLWfH56Ke3h4WPrBUslyQwl9aGby4OCgPCt77WqCD5UxD4mCwf3QZCE+UzflHumUATAoHH52Cj8b1kSjUaTTaTmjzc1N9PT04IEHHsCpU6fQaDRQKBQwNDSEcrmMoaEhfPOb30RfXx9mZmaws7MDv98v6Qq73S4EE+0xUw75LMyD9vb2CjxVr9cNwxP4jCyDocxMTk6iXq9jaWlJCCuMuLgHgLEffbu930JWQ1c0HPwaHdtuWJTvTgY3ZaNYLEqlBclyLLchOYwDVNbX19Fu77fFDQQCiEQiCAQCOH78OGq1Gubm5vDCCy/A4XDgiSeewOzsrAGhomGmYqZTTAPC73OWtM/nk3tPR5NlnHqs6dDQEHZ2diR61xwHHTEDB86a2bzfJ5w183TiNdLWLYvRID8T90Pn8tlH32KxCELi9XoRDAaxvr4u3IC//uu/xtNPP41cLocjR47A4/HIBKpoNIpbt27hK1/5CgDg3LlzhrSVThu0222DI63JnuwgyDJNTsNiMBcMBiXdQnicyOzGxobUxRMOp2OqI2YabA7M0PZRP4/WV/+7dccQd7vdljm4wIHn2G634fV6hcrOhxgbG0O9Xsfo6Cii0ajkMPP5PK5fv46trS2Z97q5uYl2u41MJoMzZ87g6NGjyGQywsxlDocQAb0QDTlr0lK9XsfLL7+MwcFBAPveVTgcFiVNGINN2KmoKVjZbNYAlWrDS4GkkmDeuXNqUTdC3ABkaL0mxBAi7owuyDXw+Xw4ffq0OD6xWAzz8/MCM09NTWFxcRFbW1siOz6fD1euXEE+n4fX6zXARJ3tNjWbWyvOmzdvyiCF3t5eYZIzumYzCo6RtNvtkjtKJBIi052GGTD2o9bQF7+n00HdtMhmpTLTRCEABlIme2Czxpw9sJkCoeF+9tln0dfXB2A/GIhEIojH47hx44YgKQsLC4KO0GnqJOtoI8p8IaM5lm+Gw2GJ/MxmM0ZGRqRDIbCPBI2Pjwv7lv0aOlNYNNTaYSevhnJJB6abuCgAJFfPOd766yRWBQIBIf+xNzrbpjabTcRiMQQCAbn7e3t7ePXVV4UvBAC7u7uo1+vw+XyYmZmRagvmnnXJm76P+lnYH2Fvb09QUj1kiSiKLsPzer2SOjObzSgUCq/jHGiOEaN3yqc25G9kq25n3ZHWMJlM8Hg8Buo8H5S5G2282e+YUTUnWXGCzfLyMj760Y/iE5/4BKamprC0tCSlT8vLyxgZGUG73ZbLwwhKk4D4XNwoesR2u10O4+bNmygUCkJWYm6E0DmhGUZVvNgUBA1j6ty7JgFpJrf2rLuNJAZAGslzaSSFdcJUViMjI1haWpJ54AsLC1hYWMCtW7eQy+Wwvr6Of/kv/yU+9alPwe/3I5fLCYqxt7cntcpk8JN5zUvQeTF1E5Xe3l4kk0mMjY0J87tYLCIQCIhT4PP50NvbKyU2/AyFQkEUAA2Qzj/rHDgAgwHWqRY+W7fJAABpvkGlpu8hAEE96vU6gsGgNO6hYifpjzlmOrnhcBi7u7v43Oc+Jy0ah4eH8UM/9EPSoIKQoZ4sR4UJGJ2jarUqcCPrUAcGBqQ/ASMnTkpqtfbb0k5OTiKXy2Fzc1P6cPNz6miJCpuymMlkhNeiGf7dFD0DB3usEUbNpidCwjp01jRvbm5icHBQ2jzX63WcO3cOg4ODmJmZwde+9jWEw2FYLPs15C+//DJSqRQeeOABDAwMCPmOxD7m/LXjpAnAhKIZ/f7N3/wNnnvuOVy9elXuPJ0JOuImkwmpVAqRSAShUEgcChIMAWPTJqY1iR7o/2uUB4BBb3y39c9y6zXhQhPHdP7XYrHIJk5NTWF3dxeBQAChUAhWqxUul0s6/jz22GP45je/iampKZhMJkQiEbRaLSEa8X3p2XAcZWcNMi8CD8TlcqHVaiEUChmYme32/sB1Qllk9Pn9fiwtLWFgYABDQ0MCz1AYZdP+PyWk84ss1eLrOmuEu2V1Qvj668z5AQe56L29PQQCAYyNjWF5eRlDQ0MS+fr9fgQCAbz44os4ceIEZmdnMT09jUajIflA5oBp9Mzm/YlDJHjw/HX+j/JAWQAg04gKhYI8b7FYlDPWeeabN2+ip6dH5JAQZafy1146lbE+b+4JlVY3ISmU/1wuZ4gqNElPIxtW6/6MXpbU0KAFAgHkcjm88sorOH78uLT0/c53viOd5TKZDN761rfi1VdfNfBNisWioF56MAGVrjaahUIBmUwGy8vLQtaigR4ZGRGUTA/IiEajMJlM2NnZEXhWO4T8XLpig5EWyavA66O5blnUuUTNgAP9oG0B71A6nZbA69atW6IjOSCHzPp/82/+DRYXF+HxePDyyy9jbGwMQ0ND8Hq9wlUgmgHsG1dtmBnIUQ7Ih/B4PPJ8hULBQCLTlRfZbFaIhdlsFtPT07BYLILsAQfBKWWC+on/5llrY9yZEr6ddUfWQytfflA+oDbUHJLN+mbWHDYaDWxsbMDr9aK/v18miLCY3+fzoa+vD3t7e9Lv1u/3w2q1IpVKGd6buR0eCI0jafWsRXa5XBgbG0M2m5UB3SaTSYgBHJTBSUxU1ITU+Tt0nk0rYioLvbj5Ov/YLUvn3wDjKEmiCtybvr4+9Pf3I5VKoVAoCFRMUpDL5YLL5UK1WoXL5UI+n8fU1JShPjKXy0kbPj2Cks1MeBb0YpkTAiB191arFT6fD4uLi/K8JIABEJnSEReNuyacUfFoY0zZ053C9OXsJoWsF40Y2a/6/uvXEHVjg4YTJ05ge3tbiJk+nw8XLlzA5uYmZmZmsLy8jH/8x38Uxv3e3h5GR0dRKpXkjDgnvFqtinKm3FE2qAcYyfIc2Q2O6Q6WdpH8QweARsBsNiOTycjXeeaUMf6bd6LVaon8aNStM+3TTavzM/H//Jtd+iKRiKQwent7kc/nkUgkAOxPkItGoyiVSlhYWMArr7yCq1ev4uGHHxanqd1uC/+AXcX0aEcuOgscTEGklCm4sbExhMNh9PX1SQQ+NDQkuoPNkTjOstVqiXEnSqsdNco/l3ZM9dnroPJ21x1D3LwM/IU6um232zI/k00IQqGQTKAhO3plZQWrq6s4deoUFhcX0dPTg7vuugvZbBbb29sYGxsDAJw9exY7OzsCSdfrdckn86Jw44vFImq1GkKhkDQ45/zQarWKwcFB9Pf3o9lsitGlMqYyYFu3cDgsl1RHSjTGOhLixdRs7k5F1a2rE6rRDgx73zYaDWlCYLFYMDw8jEKhIGVT4XAYsVgM9XpdeqPv7u6K/ExOTgqzE9iP3AiBdz4LuwK5XC5DakJHRnTG2GkIgDhlHo8HHo8H+Xweg4ODkuagrBC664SwyRjthLE1pNWNipn7AxidUkYKnOsej8cNTX+YemD+OpVK4ezZs7h69Sq2t7exurqKc+fO4fz58zI0hZPm2JrX5XJJ/3TtJPKMmK8cGhpCIBBANBrFyMgIjh8/Lv2e2f+fkVculxMFSmce2O/P3Bk565I6fb6Uf93yVe9PN6Eo2hBRFjo5N2zXyhRBf38/crkcAoEAzOb99p9ra2syHGVpaQm3bt0SMmi1WsXU1JQQTovFIhwOh3SFZMqVMscAjAzs4eFhDA4OinNAmXjkkUdw9913w+12IxwOC4mR6ArLRVliyVy3dgY79bzmRrFKRafFdO75do30HdFKmfjXbQ91qQnLaOLxuJTTFItF+P1+1Go1MZA0aKlUCo899hj29vYwPj6OV155BZOTkyiVSqhWq9jZ2YHL5UK5XIbX65VSG0JLjF4Is+jORA6HQxqPhEIhZLNZGWvJTaQiByD1avV6HZOTk6JcNOkMOOiWw0vN3CUJZRrO6ITBu2FpGWCOR0eL9GpTqZT0Sq9UKggGgzCZ9ptTvPLKK+IohUIhYc+ytWs0GkU4HMbOzo5Mqmm1WohGo0I44rMQLaGjVC6XpTSOTh0jdj4Hy7symQw8Hg/a7bZAqVSunDdLzgMjRuAA3tUENdbz0zHQkG+nE9ctS3M+uEfcj729PUEcbDYbNjc3MT09LQMp6vW61LUSIVlbW5ORsBMTE1hfX8db3/pWzM3NYWBgAMVi0cBvYITM30HjqpUsa2vJwOV9pVI+d+6czJknusPmFHTkNewNHDiinQRB5ttZR60dtG5cjCR1EyeNoKVSKRw5ckT2t1qtCsS9ubkpOd3BwUFsbGxI6S3vkc/nE4Lx6OgoEokERkdHpdyNBGLWKxO54HNtb2/LMKRAIIBUKoWdnR2RmZ6eHkSjUUQiEYHFWV3EBjrkOCQSCbRaBw2YKBvAAReKss+qj06eDCNojcB8t3XHELd+SD5Uu90WyjrhjFqthoGBARQKBelJnUgkkEgkMD09LS09C4UCSqUSvvrVr2J0dFSMOmFtl8slwwZ0a0hNDCF0QYYgmXOjo6M4c+YM+vv7MTg4iHa7LVNS2JWmWCyir68PNpsN5XLZAIPpnJP2lqh0NZTSWX6ku8l00wWlodHOCxUWWdZk97KnMqMds9mMq1evolarYXBwEOFwGDdu3BDl/sILLwh5jxepVCrB7/ejp6dHckJ0jLj//P1W6/5EHHryJIFxhCm72+3u7kpem032OUmHRBK/3y8j8qic9efVkSJw0EFIR0o0JHTWuindQY4Bc/0awqNyHR4extbWFmZmZgRBYf6vUCggn8/D7/cLgWtlZQVOpxM+nw8bGxs4f/68KEY6N/pOOp1OeDwegbMZQXN2Nx101sGzVWOxWEQ6ncb4+LgYcn1/+fNU+Hp6EZ9BM/r5h/eeRNPONAdh/m5ZOqWgG5UQgWC1BoOiYDAojhubifBch4eHsby8LJU6sVhMkNC+vj6srq7KECM6AqxJr9fr0oCEqKrH45HyTOqlwcFBXLx4Effeey/uvvtu9PX1yaQr2gJynAAIYZl6otVqGRA1faepJ+g4aoibJZ/83vesFzdzb5yRrBVVPp+XaHVubk4o8vygzWYTqVQK999/P1599VWk02mcOXMGR44cwcjICOx2u8xaDYfDkh/igZNV5/f7kc1mZQgBD4QTQpjrbjQauHXrFlZXV6UMQEOQ9LodDge2t7cl1wDslwWxsTkhLH0BubSHpKFtzSjutqiJio4IBJmKZElWKhX09fVJp7j19XUkk0msrKygWq1iYWEBp0+fxo0bN7C3t4epqSlMTExgYGBAckU3btyA3+83EKx4qQk5sbGIHkOp80OM3pLJpMBdHo9H8uHkHxAm29vbE8/aYrGgr69PHMLOuu/OFp4kx+lpZ51GS5Mou2GxN7XX65W7o++Ibr3pcrlgMpng8/nQbrcxOTmJ2dlZmWB35MgRpFIpnDlzRmrmL168aGjXyalmVHIApLOgLt+k4qPDzdQX2/a63W44nU5cuHABg4ODMqmIqRHKFR03s9ks5VydzHwuGmEaXzoC/B5f36mD3uyLjjKbR2kmPxE26nA66MeOHUOtVsOZM2fknFOpFNLpNM6dOydk3VqthuHhYZw/f14IvawrJ6fEbN5vXkUZYO90jVbRkSwWi1haWsKNGzcwPz+PhYUFhMNhhMNhqTwiuhKPxyV65vhRjjilwdVEsE5kTTesAmAI7JgK+J704tZCyF/OQ+HFKRQKmJmZgclkwvPPPw+LxYJMJgO/3w+Hw4GjR4/C5XIhkUjAZrPhW9/6Fs6cOSNMSK/Xi2KxKN2iqHx5QQlD53I5Ydfp4R1UsIVCAX19fQJzcnM6YWh2osnn83C73QZolJBVZ5mEjpzI8KVS4r+5N92mmN/IIelk7xeLRYyOjsJms2F+fh4m0/58VXbpGR0dlX640WgUr7zyCqanpxGPx4Wh3cm05SKEWalUkM1mMT4+bmDb670vlUoIBoPo6emR3rmsf+Q5ORwOJJNJWK1W5HI5ccpIXiMaw0vHPeh0wlgDSYhbQ9udyrobFvePk306meqUfZJ4+vr6DGz8/v5+hMNhAEAsFsPg4CAqlQqOHDmCQCAgbTfpAJJpSyWtjbfb7cbg4KD8Tu55o9FAqVRCKpVCqVQSJndPTw+OHz8ubWdZxUGSKB0A3uk3csCpR4go8g7ozoUaXdL70i2LOlv3gNC60WKxGIh8dK6ZGiXvpN1uS3rjO9/5DkZHR3HffffB5XLJ2Eg6SzrFYTabhcnf09ODo0ePIhqNyvwE6uxGoyE8qHa7LSkIu90uwRhRWDqa7G7I+n06GrqHOzvp6cY41BfssaG5CzTeLP26nXVHloP5RW4ShZRRBRUQ870UUBamZzIZvPTSS7h48SK8Xi+eeeYZVKtVvPjii7j77rsl4tB5PF4y9lElfNTb24uFhQWMjo7C7XYbjCvhlUqlIm3brFar5KBbrZbU8nKjNGTJr+kWlTwcTT4iU5xOSme5Dz8L2cLdsOx2u6HuUCsc3b6OOWgqt2AwKGPhVlZWcPLkSVSrVXz7299GvV7HxsYG7r//ftlHeuF6OArbvlL2/H4/tra2hFQIQJCbVqslE2jI/K3X64hEIkL68Pl8Iluama3LJuiQkkhGeF8bWxoNnfrR+VjmyDwez/fuYP4vLLZK1LlZjao1Gg14PB6Zs53P54WLUK/XJerZ3t7G9va2RDXvfve7sb29LREn5Uq3bdS/w2rdn5Y2ODgoqQoNMet54ACEdwIcDNrg6+jkM/omdErdBxgrF7T8m81mJBIJkR8adf6fDOJuWaxjp/PBFKhO9TBdEI1GMTAwgGq1KulDt9uNGzdu4Pjx4zKzwe/3Y3Z2Vqo6NIpGlI62hhF8qVSSngoTExMYHBzE4OAgfD6fIDycG+9wOIRAbLVasbu7K/qGaTPd25uOvclkEsidaRTKHj8z5ZRRsnZYKSt2ux2jo6N48sknb2uP78hAUzl1dgvj4TSbTSmLyGQycniEGKanp3HlyhWk02nY7XacP38eQ0ND6OnpwR/8wR/g2LFjAocxr0CIUE88Yl/VnZ0dvPbaa9je3kYulzNEOXx9oVBAuVzG7u4u0uk0XC4XrFarJP8JlfOiWiwWISxVKhVh9rKuTuei6XHxZ/l1Hozf7xdIp1sWBZbEG70ojLu7u3j3u98tzQmoAJvNJiYnJ/Hyyy+L0hobG0MwGAQAfPazn8WZM2dk8kyr1RKDzIiaHiodsZ2dHezu7sqFZacqOpK9vb0oFApotVoya7y/v1+UMADx6hkN2O12kVFG1Zq9z8/NC9psNpHP56UeVDOamRdfXl7G6urq9+OIvm/L6/VKSklHroSI2W87GAzKNDBG3haLBTdv3sS5c+ekGxVzinNzczh27JggH1TONNj6rjES2t7elsg6m82KUWaKiizcUqkkVQHkjuj0FpUynbNms4lwOIxIJCJGmnCudtIYPXFak+Yg9Pb2or+/H+9+97tx//33/185q+/Vslgs0m9CQ7y8L88++yx6e3tx5coVxGIxcYjL5bJMOezp6cHg4CCi0SjOnj2LRqOBz372s2i19tuCMiLlPGYy7Bkcmc1mgaJffPFF3Lp1S8ryyF0xm81Sy8506erqqtx3k8kkfxcKBQPHJZ1Ow+12w+VyYWRkBOFwWGYHAAfTHNnQqFwui6EnikYy3Hvf+1685S1vkXKz77buuBc3laMmzpjNZpRKJYyMjCAajSKRSEgzel5celvHjh3D3/3d32FqagrJZBLHjx/HpUuX8Ad/8Af4lV/5FfT19cnF4OWkYebFAfYbD4yOjsqMTbaGo7IgjEmSksvlwt///d/j05/+tLB5dfs1XlIqi0ajIflLesEejwfBYFAOnnkERmdEFNg4obe3F3Nzc9Ico1sWnSQqQM1wp/B+9atfhc1mQzqdlsjU7Xajv78fgUAAzz33HO6//37E43H09/fj7/7u7/C//tf/wic/+UmMj48LGkEI1ePxyP5SObJPL6MYwuYayaCjx3r3ubk5vPTSSxgaGsLMzIycOZ0sQm+bm5sCoQEHzgcNAi86f18+nzcoeXrQ1WoVi4uLhrKdblisIw4Gg4Z8Glm9rVYLQ0NDyGazEl2Qq9Lb24uJiQksLy/DbDbD5/MhEAjg1KlTYqj/4i/+AkePHjU4WTxnRlCEmNmzeWVlRRj1NKQkCbGfQq1WQyKRwJe+9CWsrq6i3d4feQgcpEdo1Mm38Xg8KBQK8Pl88Hg8MiuaQzloAID9LmKEWBn5BYNBPPTQQzh9+nRXpbt47j6fT5AmoodsoXvfffdhYmIC58+fh9vths1mk+5sdJC+8Y1vIJ1OI5fLCWnvAx/4AH7v934P8/PzcDgcUk1Bx4sDd4ADBIuEsWKxiFQqJUEZ77bNZkMoFBKi4sc//nH80i/9Eq5fvy4RM42slrFisYhwOIx0Oi1d4lwuFwYGBjA6OirjUQlfA68nBHq9Xpw8eRIulwuVSgWvvvrqbe3xHUkLPVB+EC7mjre3t3Hz5k243W5ks1mMjY2hXC5jcHAQN2/exJUrVxCNRjE1NYXt7W1J3A8ODmJ9fR2/8Au/gP/n//l/MDY2Js0rCHUzgc9Smna7LQxOq9UqLF1GWDSazDc4nU488cQTePTRR/GNb3wDv/mbv4loNCpRPi8Uc+AsYM9kMjJ/NJ/PiwdIpl8kEjEUoNNb8vl8SCQSEpV1y+IsbeCgJhQ4MFyJRAKFQgEnTpzA+vo6JicnUalUhKV57do16QjEHHEqlUI0GkU2m8Vjjz2G//k//yeOHz8uuUHKAPODNHyMmHV5jS774SVjrtlisWB0dBTHjh3D7Ows/vIv/xJ33303hoeHxcGi88H+3Ha73TADVp8zFVEgEBAnjI4BjQGHJhCF6ZZF5jtnJXeWW9psNly7dg3Dw8MA9omibOHJdMbS0hJ2d3el7tzlciEej0tDky9/+csyR1x36uOdZlTN86BDr8lKRFO4KBdXrlzBU089hUuXLmFhYQEjIyOiC3SFCI0Qq02Yc3Y6nQiFQtI/AYCh3pfP6nK58Nhjj2FoaAgrKyuvY4S/mRcDmJ6eHnFySOZkmmFrawtzc3OGlJPf7xfn7MKFC0LSXF9fR7FYRCgUwsbGBi5cuCBIB1E1nXZiGko7x0xLUu/yTtMA53I5SVW++93vxr333ou/+Iu/wIc//GFJyzEIs1gs0qubfIt0Oo3t7W0kEgmB26nPvF4v+vr64Ha7JYhgOvaee+4R2xKPxw0kw//duiMDzbIF4ICZpstQdnZ28MEPfhCxWExKay5fvoxAICDzP4eHhxEMBoWJd/PmTfj9fty8eROzs7P4i7/4C/zZn/0ZvF6v5P0ASFlUu92G2+2WFoO63IvRD711TWLiITkcDjz88MN48skn8Wu/9mu4fv06ZmZmJFJj/TThbhpXEp3S6TTS6TTq9TpKpRL29vYEwuWz+nw+rK6uSm1dNzG56bU6HA4AB3LAva9Wq3j88cextbUlzOnFxUX4fD5MTEwgHo+Lx7mxsSFOmtfrxQsvvIBMJoNPfOIT+PznP49AICC1rwAE8gT2lSEdNZIXCSsSbuNF1oQuYP9iDwwM4MKFC/jzP/9zJJNJPPLII4ZmJru7uwLZUeGSQEaonCkYokqdDP6dnR2Bxu5kSPubYRGa7OvrE4MIQNCr+fl5eL1ezMzMSBXF9evXAUAIWWfOnEEqlRLSYDwel94HS0tLGB0dFWY/7zMVLaNpDtwgtE70jWkyXfZCB05DsAsLC/jiF7+Iz3zmM7jrrruk/p2Rr9lslq6GdPhZT83fSejV7XZLTTDl5aGHHoLf78etW7ekJ0C3rIWFBezt7Rka1tDBsVgsiMfjuOuuu6Rcsd1uIx6Po1wu48yZM8jlckilUhgbG8P4+DgmJyeltrnRaKCvrw/RaBSf//znJTjkPO5WqyVOsdVqRSwWex1xlY41dYQulWS063A4cO7cObzjHe/Ar/7qr+L3fu/30N/fLzrF7/e/Di0mXyWRSGBnZwd7e3tIJpNIpVJIJBLye+m0njx5EiaTCcViEVtbWzJM5XbWHeegd3Z2DB22qKAajQYuXryIv/qrv8Lk5KTUEH70ox/F/Pw8SqUSzp07h8uXLyOfz2NgYAA+nw/nzp1DLpdDuVzGW9/6Vvz4j/84QqEQTp8+LQfPfLFmRsZiMalV1dAXlbKGQvXF5oHabDb8zM/8DP7u7/4Ozz//PPr7+wHsE9xILAIOhsLTWNOz0lN80um0HCKVEy+3ptt3w6pUKtJ2lcoQOOiM4/P58OlPfxonTpwQos373/9+zM7OIplMYnx8HCsrK1hbW8Pw8DDcbjcmJiakach9992HX/7lX4bNZsN9990nhl8TuOhJb21tSRMaGgkaVc0kJZOTMsvLZ7FYZLbwzZs3pQlCLpeTmcAApM4WgDS+YArHbreLodYksa2tLfh8PsPedZOBZp98Kkw9UtFisSAWi+HcuXOipIhIEV0wmUw4efIkNjY2cOrUKZhM+z2ZqQuOHz+O2dlZ/OM//qMwfQEYnHASSFklQp2j771+PQlmbP9KvdFqtfDyyy/jt37rt/De974X4XBYXkuUBIDIEaMplmJxxvDa2ppwGcxmM6LRKO6++26srKzAZDIhk8kYiJRv9lWpVLC5uSnoE/eafw8PD+O1117DBz/4QXGkHnnkEWFMz8zMSH63p6cHw8PD6OvrE2eIJbuPP/44xsfHBbXi/vK+E9bmjHdde6xJnhptISubOsVsNuMd73gHXnzxRXz84x/HzMyM6Hm73S7oDBE8GmsiN6lUSnTj6uqqPFsgEACwfyfY1pSyezvrjgw0PUXWNWumotlsxubmJu666y5hy+XzeVy5cgVerxepVAp7e3solUr44R/+YXzxi19EoVDA3t4eTp48iXa7jS984Qv48z//c4yOjmJlZQUAxAumAeQH1rlmRr80wp3lLXw9cDCJiAf8wQ9+EL/927+Nvr4+uYCETDRLnV4Y35PPw0iOl52Qpy4F6Ka8kzaEmgFLo5dIJCQy4vcWFhYQCoVkqlAul8OHP/xhfO1rX5MSiNHRUcTjcXzxi1/EH/3RH+GBBx4QOKpzCAVLcFqtlni4mm2piXyaYaxlgBBmu93G+fPn8cd//Mc4ceKEGBpGvMxH8f34WTW7lEaKr2EDDjp3nWz3blgbGxvY3d2Fz+czwNvcHxpBh8OBWq0Gv98vDPpSqQS3241r167h/PnzmJ+fR6FQQC6Xk45vS0tL6OvrE+dXQ4K69jydTiORSGBkZAS1Ws3Q1Uoz6YEDtE+XhVG3mM379fe/8Ru/gfe+970wm/enqVHRa2YyFSwZyZVKBY1GQ7gG/B0nT56U52Xjk26SA8LL2vHgXrVaLWSzWVy8eBFbW1uCRK6trcHj8WBtbQ1bW1sIBAIIBoNYWVkRw8nOcltbWzh58iT+4A/+AJ///OflvrImutlsCh+EOWn2RqBuAiCcILKvO8tC+bfFYsH58+fx9NNP49KlS9JKVpfedfa6oNNNRLHRaCCXy8n7eb1ehEIhxONxkUU96OO77vGdHAijB9LfASPEfd999+HrX/+61CiSrenxeFCpVJBIJHDx4kV85StfwX333YdisSi9tqnUvvrVr2JiYgLJZNIg0BT83d1dmM1mDAwMYHd3V6LhcrlsMIa6iYguLNf1cbzIP/dzP4d3vetdePDBB3HixAkhiDBa5/voki2yijlujxEUxxrq0qtu6h5Eol6xWBT4CICwed/xjnfghRdeQCqVEkG0Wq3SYnN9fR133303vvSlL+HkyZPIZDKSUqCX+/d///cSqWhPkyzLnZ0d6cueyWQE1aAMaEiK50ciIGCcGcvvv+9978OP/diP4cd//Mdx9uxZaTvK1+tSEp4tlUChUJDSLcodf07XRHfT2traQjKZlFIrfk46uadOnZKe2wCk93IikUAwGEQul8PAwADK5bI48+FwGENDQ9je3sbOzg4eeughvPzyy0LILJfLoge49IhYRrp03rSDraNm6gzd1IYG/7XXXsNv//Zv4z3veY8M0KjX6/K31jH8uXQ6jaWlJWxvbyMcDsNk2p/INzU1hYWFBSENRiKRrkp3AQecC5J1uY/NZhPj4+P4yle+gkwmg2azKdwesqQ5RjaXy2FychI7OzvY3NxEf38/KpWKsOf/w3/4D0gmk4JqaUSSZX7BYBCJREJQVJ2aZLoDgME54/MDELTFZDLhvvvuw9/+7d+K3r558+brAjRd2cFOYxsbG7h165b8HMu8AGMbUPJqbmfdcR00N0hfEv7/ueeew/T0tBAFaBAbjYaUW2xubiIcDqPV2h8DyTInehsLCwsIBoOSx+bl5Oaw7zUAiVQZwfI1b2SoO6Ngnb+2WCz4yEc+gj/5kz8R6jzr23jRtVDocg/W4vHrJLWw1RvQXQ3yWf+nUwi6dObb3/42pqen5dz52Tk60O/3S7mbZmF7PB5poj83N4dGo4HXXnsNLpdLkBtgH1FhK1kABhIRL4x+NuDA69XpBv09ysxP/MRP4JlnnhHuABEAHR3wZwnDkQzDPaG3TSXFPeg2siD3hXpA3y9gPzfMBkDUBwCwt7eH4eFh5PN5uFwu9Pf3S3tPm80mxjsajSKTyWB8fBx9fX1CAq1Wq9Ixin2zqR+oiHWUQyOtoybNn+G/tS54+umnMTAwgGw2K6QwXdeqiWQAcOPGDQPb32q1CgOdETad1W4quaQh0r3qtVxsbW3h/vvvl4CONcScu93T04Pt7W3Y7XaZYkfkiajrSy+9JC2DGYgRJeHdY94XgJTYkSCqc+K8p9p+aSRFO9bBYBCf+MQnZEwtu8lpmeLPAZABPzoYC4VCCIVCwllhcHcnTtodt/q0WCzCZNYwrslkksYDhB2AA7ZnPp8XRhynGZnN+60Yk8mkFLHX63XMzs4aGttbrfuzZAlrcNOZB3Y6nYYCee09dypZfWF52VqtFiYnJ/HJT34SxWIRAwMDcgCacq/z2XQsGOVzX7TSAmAwFN2wuKeMVjUpAwDC4bBEurphPMsURkdHxVhmMhkZOUc4S+cqiaIw8i2VSnA6nQYmJ0v6KBu61Il7T4SjE6bkc/Bz9PX14Xd+53ewvb0tU9MAGBw0GnkyRNkTWL+G9dp8Bu5Nt+UfKev6s2qkgYqUd4XtNHO5nCGiYPXEa6+9JqzYYrGIv/3bv8Wjjz4qI2qJlLCsKpVKCWmU+kE/g3bWNZKl85id/ycZ9PLly0ilUtJYhRAnHREq51qtJjlHokWhUAhDQ0NSxcE90Kmwblntdlu4CMBBp7R2uw2Xy4VsNguHwyGVP7FYDMePH5e++LQBsVgMR44cgdPpxLVr1+B2u1Gv19Hb24vZ2VmcPXsW1WoV/f39hlaZm5ub0knQYrFIwyrmwhlk6WiX8kE9Axw4bvweezNUq1VEIhH5TNRFfF8ulnFRr7D7GHVCJ7J3uyWXd2SgqdCKxaLUwerohHAjG5rwUiWTSUQiEbjdbqkjs1qtSCaTePLJJ7GxsQGHw4GpqSl86EMfwurqqtRJbm5uyuZorymZTEqzdXonmgAEGHsl8/8UIk295/tbrVaMjo6Kp6UPjCUXjK7oUfM9mA/R9dE8kG6KoHnmxWJRogIN4RLG4j7pSUUOh0POnRFqKpXC448/jqWlJWHZ33vvvbhy5Qo8Hg+2trakgT0jJAo7m4nwuTpRD369M/fMi9pJHqSSPnnyJMxms6FJBnAAU1ER8N9UzryMdAiAg3aHlJ9uWYyKyYwGjGx5ngGZ77xLfr8f9XodwWBQyt+4L+yJMDQ0hImJCbTbbQwNDWF9fR2ZTAZra2typ0qlEpLJpMhCPp83EIcAGAwwz11HTPwcdDL0a//oj/5IdBjfRytcflaOLGXay2q1Ynx8XJxTfkbggHHeLUuT7Kh7+XU6OgywyKomJ4GVPLQVtVoN6+vrmJ6eBrCPlE1PT+OHfuiHUCwWMTs7i5WVFdy8eVO6gLFXtm5KxFJI7jOdd332erAKcNBgpNNRW1xcRH9/v+TGiYppfcFe3yw3I0rm8/kE6aNtoQHn89zOumOSGL1RHgQVIBUUo+dmsylF4twktnLjYSYSCbzyyis4deoUUqkUNjc3sbKygunpaVy4cAFnz57FyMiI0NJJXwf24fbd3V2DcdaeM38HlbIWHr5GQ3IA8GM/9mPSPJ3knjciCLHhPj8n94BtQdlxqHOedDcseqCEibRxarX2G3bo9qjM0dNoZ7NZgax5yUgU3NnZkb0+efIkpqamMDU1Ba/Xi0QiIVEq5c1ut2N1ddUw+g04gC41w1ynNPhvjYrw3z/yIz+Cqakpidw7ZYBGnReO4yn5efiZ6GyUSiVD/rJbFuFbyrh2hi2W/X7srCsm8dLhcCAQCAjsTfSp2WzKhDFOIIrH43jiiScA7Dt3jMbZDYqlW2azWebMaweM59HJ1NV/dD5SI2okjJ0+fdrA/NXOtibDETGiMR4cHJR8KdGBXC4nU/S6ZbndbrTb+93byNlptVpyH3TOlRGt0+mUCXUcOcrzyeVy2N7eRjAYhN/vRy6Xwxe/+EX09fXhAx/4AH70R38UY2NjMna4XC6LE+1yuYTRrVMJTEVpNIXnDsCAjFB2+ZqBgQEpC6OO16WalG2HwyEOBwlp7HRoNu9XA7jdbsP0t9tdd6QxLBaLkHl0sTjZsz09PUgkEnJwjLJTqZT04+7v75fLes899+DIkSPIZrNIp9N473vfiy984Qu4efMmHA4H9vb2pJMPD1o33O9UsrpBgN50bZx5qXU0DexfuL6+Pnzzm99EJBIxHKJW7I1GAzs7OxgfHzfUPZJAR6WtcxXdBHED+/ugS9EYkZrN+93CUqmUtOgzmUxSO0oZOHLkiCiwCxcuyFjSvb09PPLII/jUpz6Fa9euIRKJSHch7i27D7FGUffGpues97sz5/hGipby0G7vd5Wam5uTsh2dz2JE2Gg0EI/HMTIyIixPOiS8kKzX1p5yNxloRpMcPgHAcNdKpZI0oikUCtLcg04d++RTqU9MTODo0aPo6+tDs9nE9PQ0/uRP/gROpxN33303pqen0dfXJw0i9PQ6p9OJra0tierpUPEe6jScJjJRETOKBg4iauqZaDQqeow8AzptTPPcc889yOfzKJfLCAQC0mUNODhzpvooF92wNEeI6KHZfNDFsdVqYWtrS1AUdvpiXp7T73Z2dmCz2XDs2DEhl25vb+PkyZNoNBq4evUqTCYTZmdnYbFYMDg4KLq4Wq0KGqd7JlAONVkUME4V086ZXtQlHo8Hf/3Xf43h4WGEw2EDB4vp2HK5jFKphMnJSQwPD4tTQtnubE5EtPl7EkGzzlPnenQZwuDgIHK5HFqtljRnYEhPKIIt+1iW8OlPfxpjY2P4wAc+gK9//ev48Ic/jEQigZdeekmMPqM1KmqTyYSBgQFDhKovFb0nbWAJT3Z6yp0G9LnnngMARKNRBINBORg6IYQ15ufnpXUdo2q+Dw9BDxbplkXFRoOpHSTKQCqVMsBaVIwU8Hg8Lh722toaPve5z2F0dBQ//MM/jOeffx5PPPEETCaT5CR1aQy7AJVKJYRCIbloPGM99UY7YsBBmoJf0wRAzej+6le/KrlQOqWMlClbNptNCC6c5sP7wO5HJK7pSL5bFg1WLpeTzk38w3LIWCwmxpB3n8ZcdwIjsY6zo9nBLRQKYXd3F1tbW6IwOcmOJTXsacypdJ1kNY2i6cWoTf/h0tHx0aNHhcjGqUk8ZxLhSGLj1C42wGg292fLl0olYfl3kwywPpgBis73t1r7fbQ5j9vlcsk+0xawWZXb7Qawj5QsLi5ienoab3vb23D58mXce++9OH36NNLptOh59qog+YylWezQpQMzGnKto4ADgpuOmHVQBhwY+bGxMUxOTuLo0aOYnp7GwMCAdLFk69F4PC7IkMfjkVGotH86qNUVJd9t3ZGBZo9ZzsntzOmQnVur1YTV2NPTI4O2dRPy0dFRHD9+HA899BA+//nP4+WXX8bQ0BDK5TLGxsZkHifJBoVCwXAxdEkFDQXhdX1BNYuPhpNesiaK8LM888wzCAaD2NraEsXR19eH0dFRjI2Nwe/3Y2lpSRpVAJCevIzm2CydcEc3RU4U4nw+b2AsUsgLhcLrmsVQNlg3CuxfjOnpaUxNTeHs2bN4/vnnkUgkpLmHHvhOaIzeKSErttGkI6eJQNrx0o6TTlfo9IN2or7+9a9jYmJCZI595JnaqFQqWF1dhd/vl6iKZVkkvrCJCWWN8Fe3LDpfGtHiarX269MLhQIqlQqcTqewq5knJLxZq9WEILq5uYlarSYd5B555BGEQiEDIY/dp0ym/frzsbExXLlyRZ6B1RXAAeLFu62Vsc4rayKfdt5SqZS0boxGoxgeHsbo6CiGh4fl7JPJJAqFAgqFAkKhEGZmZgzOOpECACIT3bIo09ls1qALuIeEf+PxONrttvAEyuWyDC1iS11gX6b8fj++/OUv44UXXpAGNRxVyuoIq9UqTr7ZbJa0JO0F7zYDJH32fL43KrHjM2jnrVwuIxKJYH19HdVqVbrjXbx4EWfPnsXExARWV1dF1u12OwYHBzE8PAyLxSK2q1M33e66I8sRDAYF3mH+WZPEGFWTuBUKhSQfR4IIyUWbm5tYX1/HiRMnYLXuj/26fv06ms0mPB6PeGUk2fD38XLxEmpyjs4XaQXdybjTyrgT4vD7/QKTplIpbG1tYWNjA5lMBmazGYODg0J2Yv7bZrMhm80KrEWHQUfz3bII45BvAMDgqfLs0um0RFLFYlHQF9bE9/T0YG1tDfPz8xgbG0M6nUYymcT8/DxcLpeUN/DC6Dww5YGQKXBQTgccRMZclE9NFtNy25kKGRsbkx7rdDbZNIV7wGdk04pWqyWTrxgdMudExdFNZEFGBel0WpAR4OCusdaTe2YymSRSYnRaLBalXIrK8MUXX0S73cbOzg5GR0cNvfTJYWBjotHRUUFhqOT1SEhdiQEYS2roNHUSBvmsZrMZzz33HFwul3QrdLlcCIVCOHLkCI4fPw6z2YzZ2VlhGx89ehRTU1OSG7Va92eMU5ZZltcti2kOkiB1hEpU0e/3w+12SwUGHTOeDVsH12o1eL1ePPjggwCA7e1tLC8vi70g4tZqtWRsbbPZxMDAAOr1uoyN1ERMIru0ITrS59LyoR046oL19XVcvnwZrVYL29vb2Nrawvb2Nmq1GoaHh3HkyBEAkIqUSCSCI0eOyFlrFjkAQzrgdtYd9+LmZCj94WgM6TWzno0jIAl3VyoVUWJ7e3vw+/1YXV3Fj/7ojwpjcnV1VeAhKj9uqtfrxc7Ojgg54UoaZh6AzhXxgKjAdT5C99Dm3+94xzvwmc98Bl6vVxRtuVzG9vY2bt26JTk1q9WKwcFBKbNhWQD3RdfbdVPkRAFj/gk4KF0jWaNYLEpukMqXSoqkMIfDgWQyKRyEd73rXSiVSjKmkNUAPBemONh5jKxu7q0ur9KMfj6b/pp21jQEzUv54IMP4m//9m/hdrvlczHnWCqVZJwqiWS5XE4GwnPROBHO6rZUB4dk7O7uSqpBI1etVgvhcFiiFKIrRBWYvzOb95veFAoFHD16FFbrfl9l9ttn+oAwIoduzMzMwGq14umnnzb0SwYO8ox8Ft5Lomyaf0KEhCU32khzhCU74HFsLiPhvr4+Ka187LHH8K53vQvb29vSIzqbzaKvr09mYTPV1y0rFou9riGITnuSD+BwOKTiprPMrNXanwvN8ZOJRAInT55EJBIxtNjkWVWrVWQyGbjdbtx7771IJBJIpVJSwmaxWOS8GQlTFnRfAuAgzcLXMY2pkeGxsTF861vfgs/nE5S0WCxidXUVL774olQTWCwWPPDAA7h48aJ010smk+Jg6JkMDGBvZ90xSUzPadYQIT9osViUmayt1kHjBkJc9GZmZmbkwObm5qTWjRA5lavT6ZTayVgshkAgIPlmbqaGsPlcJC/wmekVUSHTYOsD4ueYnZ3FyMiIGHBGRWSbjo6OAgB2dnYAQGBPvrcu3dEt57phkSFJ6F6zo7mHhDBzuZxcDEI9vAhWqxUTExNS7766uorXXnsN73znO6WGFoAwXxmtcZ6zvoj83Zq9zUtN9jTRDn5dVx9oXgKV+jPPPIOjR49K9E4ZowcfCASQTqcRi8Vgs9kMxEi+nrKom1x0y2JqgUYWOEgrMA/n8XgMgwHYP4EsXlZ1pNNphEIhOBwOaZ1Lo8u9pPM2ODgok+Ju3LghLOA3SmeQRNTplAEQZUyZIIzOP3T6b926Jc2XqtUqSqUSstksNjc3Ja968uRJ3HPPPYjH48K7yWazkl9lKozP2i2LlSw8J94fGkrds4KlSNQdbKVbKpWws7OD3t5eGUbEErWZmRn4fD7hjtABO3bsGBwOB2KxmLwvdTjvWavVEodeI6z8W0Pa1AmsNqJM8Os2mw1DQ0MGxLBWq6FQKGBtbQ1DQ0M4cuSIlBKnUilkMhlBVpxOJ7LZrJSAcTTl7aw7bvXJQyBEqD2n9fV18Zr5UKxbZiTt8XgwPz+P/v5+XL9+HbOzszh27BgeffRRfOELXzD0WWW+eXNzEyaTSXow664umgSiSys0sUwrR+1QaLY1v0aD7HQ6DQac0Mfu7i6Ghoaws7Mj5AeTyWToiEaB0a0iu2VpZac9W+7h1taW5Jd8Ph9yuZzspdvthsfjgcPhwNzcHCYnJzE7OyslLffffz++8pWvSJkCcDBrGwD6+vrg9/tFWeocknaItGwAeJ3nriNZXaOr0RSPxyPNSnh+TGesra3B5XIhFovB4XDA4/EYmqUAkBQAoz8q/W5ZdIB1zlenGfr6+lAsFuVs2u22cFeYk0smk+JwkfHLPH84HBbnljLA98hkMlJ/TGOgHS8+izbaNCS645wuvSLKot/LbDbj6aefxtjYGEwmk+QZ+Qwmkwnnzp2Dz+dDOp0W7kkul4PP50MoFEKhUEA+nxcnv5ty0ISv6/W65KE7UwvhcFhSGSRRcXQnI2sOO2HHLTq7L7zwAiKRiJwRGeCcIrW8vCwIhuaX6L91a1h9j4m88WvayaO9olyvrq5ia2tL3rdWqwm5sdls4rHHHpNU3u7uLkqlkkD8RA6IJrBU+XaDtjt25wj38DJqj7Wzm4zH40E4HMba2hq8Xi/K5TJ8Ph+q1SoGBwextLSEH/7hH8bLL7+MXC6HY8eOGYhkhUJBcjmECDR8pZ0FGmcNs+nXae+eh6FZdfpgH3zwQVy+fBkWi0UiRXqGpVIJ8Xhc8tDtdlt6j1P4WEhPQ9VNizAT4UIt2DSEVJIkVUQiEWxubooTw8hrZmYG169fxzvf+U4888wzqNVqGB8fl2EljEJZtkBiEnDgaDG3TRnQ/ANtiDtJTJ2XlbLCz3Px4kUsLy8bJjDRUaB8si0hPyvfkz2jabQBo/HqhsV88Pb2tiHvS8UUCARkgAkAaU7BWnnu9bPPPotjx47JcJzBwUFRxFTK5XIZOzs7yGQyMr2KClaTwADIndTNJwAY4E0NcXMxjdZpoDc2NqRFJdMchUJBulkdPXpU0h0mkwn5fB4+n0/SfalUChbL/lxhlhl1yzKZTNJLX0fQ7fZ+Q6n+/n6Uy2XhLBFp2traQk9PD3Z3d6VXNY3YyZMnEY/HEYlEEI1GDWMkY7EYNjY2sLS0JARcbeiI1unIFzA2VNHQOx1nnQ+mUdfyEI1GcenSJQAQfhLtEomCuVxOouRyuSyVCMViUeSVLW3vhDh8W9aDio7wHclehDLr9bqM27rnnnuws7MjH3p+fl7gBqfTiY2NDbz//e/Hf/kv/wX3338/YrEY4vE43vve9+Kzn/0sLBYL7rrrLmSzWaGx0/PQeQZukC5h6YRc6TWzPo5GlDlBTeDRkKnT6cRTTz2FD3zgAwJLVKtVEYpisWggMVmtVqn3s9ls0sqQeU9G0m/mHCSf3el0olqtwufzGVjLrVZLZmWPjo5ibW0Nbrcbdrsdc3NzsFj2x6319fVhZ2cH73rXu/Bf/+t/xVve8hYZ03b+/Hk888wzAICzZ89iZ2dHzr3RaMDlcglpiCQs3blNExX5fzpinMCmZYAQlC6domHo6enBP/zDP+DkyZMy+5glFXa7HbVazZBXYjMOlhTx35Qdwnx6L9+MS7OTU6kUUqmU1KbTybFYLMhkMrBa97vGkUtCxca9Y1vXVCqFpaUlPPLII1heXsbMzAxeffVVnDx50mCcqVx5Vzuddf5bk1h1PfoboS2a6Kp5C3RC2fdAf50kNQ5+MJvNCAaD4li2Wi2Uy2UhmQ4ODsLlcqHVOphh3A0y0Gg0JFrc2NiQtECz2UShUEA2m8X4+Lh0FOMo1/7+fqTTaTidTsTjcQwPD8tdM5lM0rBma2sLsVgMFy5cwMbGhpQ68g6TYEZ7ALxx331tDzQCqFO0WnZo08iZIZGR0LwmQttsNpm4RtJqT0+P1HMz3dvT0yMBbLlcxvr6umEv/6llat+GpGxubmJkZOTOTvFwvW5tbGxgeHj4//Zj/LPWoQz8n1uHcnC4DmXgcAHfXQ5uy0AzP+TxeLoKpvt+rXZ7vwZwcHDwTUsSOZSB///rUA4O16EMHC7g9uXgtgz04Tpch+twHa7Ddbi+v+vN6cIdrsN1uA7X4TpcXb4ODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7gODfThOlyH63AdrsP1A7huax704fSS/3/rcILN4QIO5eBwHcrA4dpftysHt2Wgt7a2Dud//h9Yb+YZsIcy8H9uHcrB4TqUgcMFfHc5uC0D7fF4AAC/8Ru/AbvdDgBi9a1WK8rlMmq1GqrVKhwOBxqNBqxWK9rtNtrtNqxWK5rNJiwWi/xMvV6X17RaLZjNZrTbbdhsNtRqNVgsFjSbTZhMJrTbbZjNZlgsFlSrVfT29qJWq8Fms6HZbKLVasFisaBer6O3txf1el08O/7daDTk/ex2O+x2O+r1OiwWC2q1Gnp6etBsNuWZzWYzWq2W/LzZbMaxY8cwPz+P3d1dVCoVHDt2DKlUCjabTT4zP1e9XofL5QIArKys4Pd+7/dkH9+Mi8/+H//jf0Rvby+A/T1pNpsAIOfAczaZTOjp6UGtVpP9vB1vm/tuNpvRaDRgMpnk59rtNkwmk3xfyxSXxWJBo9EAAJELrmazKT/H56MM8nPwd/Hs+TP6+ZxOJ5rNJrLZrMgsn43PZLXuX61SqYTLly/jwQcfRCaTwe/+7u92hRz8+q//Oux2u+w3Py9Xs9lEsViU+2+1WuUOUj54x61WK1qtltwjfff4/tQPeo8pV1pPaL1BWaG8APvnV6/XAezLE5fdbhcZ1fqIv6PzGUwmEzweD2w2G+LxOHK5HFwuF5rNJmw2G8rlMmw2GywWC1qtFsrlMqrVKqampvDv/t2/6woZ+OVf/mX09vbKndH3oNFoiE3gefMM9PnoPQUgcqL1gL7/XK1WC8CBvtD/pmzw39Tr/H3UU/qsbTYbbDab4f31e2s7xd/V29sLh8OBcrmM3d1dAIDD4Xid7qCM1mo1eL1e+Hw+zM3N4ZOf/OR3lYPbMtDcIIfDYTDQNJi8qD09Pejp6UG9XhdB5erp6ZGfM5lMhsPk4ofXB0cFyE2y2Wxyca1WK8xmM2q1GgCI4eDP0zHgBdMX2uFwiGKgcqFzwddQIPgstVoNkUgEXq8Xf//3f49HH31UhNDtdsv7NptNBINBtNttrKysyCG8meEgLQM0bBR0XgQqIZfLJXuo5UUrPiquzqW/x0ujnSv9LFTK+m/+HN+LX+Pz8kzb7Tbsdrt8v91uo9FoiEHvfF99QXt6ekQWk8mkfEb9/M1mE41GA9euXcMjjzyCSqUCt9tteP434+KzO51Ouff1el0UEWVCG2zuq8VieR2c19vbK/JDvQBAnG3eWZvNJu9tt9vlHFutlrwH9UKr1ZJ7/EYG3OFwyHOZTCZxqvm7tU7g0rLabrclCLBarRgfH8fi4iJsNhvsdjuazSZ8Pp84EuVyGb29vTh69CiuXbtm2Mc34+q0B/qecP95ljx3bej4df1+vFs0fHSaKRd0oLUR5v+pgwCIjNEJAw5kjK/nMpvNqFarBgeD36esaKeA9kg/N+2I2WzG5uYmenp6DLqOf9dqNYRCIVitVmxvb9+2TbjjJAgvDADUajUx1BRaXgwqMR4Wv8aHotfNQ+VrtdfDS0Pvg5vKC99qtVCr1cQb4+ZRGetN4mXqfA+LxWIwAvoz8tnq9Tra7TZWV1cxOTmJRqOBt7/97fjDP/xDTE5Oys/zM7jdbuzu7mJjY0MOr1sWBReA4bz03msvUxtF7YDpC8v34ve1h8tomDLEs9aGtfP7+iJpBKfTQPN38dwoHzpq1wgOP0+hUBDnwe/3I5VKyfvoqG9ubg6PPfaYfDatNLplVatVcZr1fddRBJc+Ny7eQeBAAeroWuuDzrPvRFroZAEHssb/82vUGbz/1Fd60RngM+vIXSvzXC4nCNzw8DByuRwAGBDDYrGIoaEhOJ1OrK6uGvbjzb40OtmpE4B9J4t3+Y2iWC4aYf68Dtz487x/1Ps8606UlXZIOwD6d/H56EDrQLFT7+sgje/Br/Ez5fN5OBwOuFwu+Hw+JBIJCQ55B0qlEo4cOYJMJoOtrS2Uy+U3DE7eaN2R5eCDcyP4NT5Iu90WiFlfNH4wGrF2uy3RqlbyWilqz4gHpBUs/82f0xAKjb++ZMC+AqhWq/JvKkweFF+rlUCz2URvb68IyNzcHFwuF3p6enDmzBk8++yzCAQC4nz09PQIpNXb24ve3l6DknizLwpwpVKR/deXRitZnmvnheR5asVHZ01HOsCBIeXZ8nd1QtKA8XLzDx08LVsaQtMyxXPi3/q9Os+QF7W3t1egTX7GarWKF154AefPn0exWEStVpOorlsW90WnrTr3v/O1AAxREc9e6w+9tEOnf7ZWqxnSY1wajubP6UhYQ6ZMzWm0pTMFot+fP6PhdYvFglKphFarBYfDgUAggEKhIK+pVquo1WpIJBLIZDKo1WqC8nXD4lnTKef+AEb5oO7U0Wnn/af+1s49gNednz4jrXe0QdfIKYDXpcG0Q8Z0Kn9WO3udjjo/K51+oi/8XeFwGE6nE4VCQZ6j0WggEAhgcXER5XJZEKDbddT+WaEdhVznmgkTUQh1PlcbUg0XU9AbjYZBKerN1fBpuVxGOp1GNptFNptFuVyG2+2Gy+UyeFT0gDsvGxUqheKf8ur4DFarVbxsKo9SqYTp6WlUKhWMj4/j1q1bCAQCKJfL8Pv9sFgsWFlZEc+MyqRbltVqRalUAgBD3pCGWkNSmoegL4WOiIADCIoywP3SF89iscBut8Nms2F4eBj9/f3o6+uTlAOfh+fI9yDyAsDwPX32+ndoR09DthriNpvNqFQq8js8Hg/q9bo4ac8//zweffRRVKtVA2TfqSje7EvLNs9OO2R0wPV5ajSjk2fCe6vfV+sDvq/D4UCtVkMgEIDJZBJ9U6vVDLlf3nHKoYazG42GQKk6yNC8B/09bRx4poyiCWEODQ2JwS+Xy8jlcjh9+jRKpZLopW5y0gAYjJXeO61jOx0enb4EYLhffB/aBx3EaQPf09ODRqMBu92OQqGARqOBarVqiIy5OlGSTicLwOtk4J+KnDttGQDkcjn09vbCZrMhGo1K8FIqleB0OuFwOOTZ/inU5p9a/yyIm0sTcmw2m0ALjUYDPT09ckjcTH0w2nPW5AyN4WvCDV9Hxdzf3w+n04nr16/j+eefR7vdxsDAgGED+P7aAGuITRtwLh2ZUzj0hprNZmxsbKCvrw8mkwn33nsvPvvZz2J6ehq1Wg3r6+twOBzymUg+65ZFj5GXgwpPIyQA5ALw4nUaah0h6yiFF1orQq/XK6Qjl8uF9fV15HI5FAoFpFIpmM1muN1uyfECB+kXjYbQkdAOYGd0r717OnydHnqr1UK9XkepVBID4fF4UC6X8e1vf1tyznwdSYhazt7sSxMrGbVqXoKOmnh/TCaTGF8qQiJOdGYBiCHTKIzVakVPTw8qlQry+Ty8Xi+2t7flzF0uF7xeLzweDwqFghhOLXeUJ+aPO3WQJg5piL4z70gZZUqvUCiIHPv9fmSzWWQyGZw9exbXr1+XNEC3RdA8m8401hultPh3pzxoBxY4cPS4z5qHQLlJJpMSpWazWYTDYTgcDvT29qJYLEoAR7T0jdIq/L+G5AEY7IF2NLh01E/5qVQqqFarIqNOp1O4KV6vF2tra2IHARg4FN9t/bMiaJ0jJIxRr9clF0WPlZtaqVQMeVqyOnk43PhKpSI5ZavVKhFTsViE3+/HwMAAWq0WvF6veObT09N4y1vegkwmgy9/+cvo6+uDx+MxKAhuMD0f7R3pfAaFqzPPTdieQrW7u4uxsTGUy2UMDAwIjE2SAN+X0UA3KeZSqYRGoyGGWbNpdZ61E2rWhpLy05mn5IXk+dHIMtdHUpbb7Ua73UahUEBvb6/IjcPhgNPpFNnSvII3UsJ8Ti46jnTMgAOZeaPolwqq3W7D6XTihRdewAMPPIByuQwAshdut7vr5AA4SFWUSiUDIUbnhXlPec94royQdXrDbDZL5EOkq91uC3pmMpnQ398Ph8OBQqGAwcHB1zlbRFjC4TDy+bzhLmpoHTBC4vqMtUMJHBgY7eTzZ8nipdyEw2Ekk0nce++9uHbtmiHyp/PQLUufOdGiTmdXk7m0MeyMIslZ4j23WCyCrDWbTVQqFRQKBVitVoyOjsLhcMDj8SAajQpBube3F4ODgxgZGZFKH/ICuLTc6ehZkwq1fuL39Ws1V0mnS2igQ6GQkMIWFxcBwBC4lsvl742BpoGiV8kN1rkh5l50vlnnixjB8P348+VyGVNTUxJpUDm3220Eg0EAQDabRW9vL/L5vOFgq9UqxsfH8ba3vQ1f+cpXsLGxgZGREYP3yw2n56+NtM6LUdh0hKfhd/7/1VdfxZkzZ1AqlfDkk0/ihRdeEKSAQmY2m+/IW3qzrEwmA7/f/zpDS2XMSIiRiiYRaohLe9jtdhuBQADJZBI7OzuSvqAscA/p8FE5ahYxy56YauDv0pG9Vro6mtdKWOfC+DXKiM6pMYru7e3FjRs3pLwOgMEpJOx6u7DWm2HRaXU4HK9DQQhBshRSlz3qlJiOqnW04na7sb29jWq1ikgkAo/HI6kVljU6HA7s7u6KISfXgPLX09ODI0eOoF6vo1AoCLzNc9fIT2cqTBtg7VhqGdKoX6u1Txy0WCyYm5vDRz/6UVy5cgVWq1WgTbfbjXg8jkAg8H/tzP5PL6KdnVGzDoIYWWpHrJMfpFEN4MBRj8ViMJlMGBkZgdPplAiUctVqtZDJZOTfGnkJBoMYHByE3W5HKpUSnaSdMZ1m7XTaNeTNpdE0ykOr1UK1WkWhUBDyaKFQwOOPP45bt24ZUi1utxv5fB5ut/u20113pDH0huhIUUfBmsDBg+IDdsKLVNj1eh1+vx+JRAK1Wg2VSkUUQK1Wk83t6ekxeKCEjbip7XYbjz32GKxWKz71qU/hxIkTBjY5n0nDcdph4O9oNpviEQIHCpwHRmWRz+dx8uRJzM3NyUGyDo7wZ61W6yqIu1qtotVqwefzvS5y1rAlcADlEArUeaHOnL/FYsHCwgImJiYkn0PZoXGjQ8D3bTabKJVK8n3+DBGXYDBoYJICB7CVdhj1ZSMfQisQHXHprxF+u3HjBr785S/jAx/4gCFPRWPebDbhcrmQTqe/b+f0vV71eh12ux35fF7uhIYved7Mzet0EY0pF5UVUbj5+XlMTU3BZDIhkUjI+Wi9Uy6XEQqFkEwmYTKZUC6XDcxbMmynpqYQDAaRTCYN3AOdCwfwOiMNHDhZOv/IRUedaaxyuYxr167h2LFjmJubQ7vdlhLUQCCAXC4Hn89n4ES82RcDl2KxKLLOpSsi6LTSEdP8IP0+fI3FYkGhUMDY2BhqtRq2trbEqHb2RiCczN9FHUCnPBgMYmBgAJVKBalUSmwTcMCF4XtqGeh02HTA2ZkWof4plUrY3d3F29/+diwvL8tzWK1WeDweZLNZuFwu1Go1ZLPZ29rjOzLQJGJoqIKHojeZ3gxzAFSeOuHO17RaLWn8QKNIeIKL+YZ6vQ6n02k4TE0moTc0OjqKd77znfhP/+k/IR6P4/z583IgerN15FStVsUIdOaNOz0pHlo8HkcsFkMmkxHYlfkm/p+QfLesTCaD0dFRFItFA5FPMzk1zwCAAa6iUGtP22QyweVywe12Y3NzEy6XS4yq9nyBg0YGjUZD8k7ao+XvZQ7yyJEjGB8fR6VSAQCDodYRkvbk6e13klmAA0eQZ1upVHDp0iX88i//MpaWlsQolctlyZl6PB6kUilMTEx8387pe73YiCedTsvec994zrzvuuyFfRL0HdcGvVwu48SJE9je3jbIQbu9T7oh5M3/9/b2Sj22li0a4WQyiVAohOPHj6NSqWB3d9fgsOm0l4ZjGeHz/xre5tJGiPB2pVLB4uKiyFetVkMmkxE5ZUOLblg8d+pn4KA/RmfunkGcRiw6DS51fz6fR09PD/L5vDjBDABYt1woFOSuejwe9Pb2SvCg4WuimTMzM/B6vdjb28Pe3p48P8+J91mfeScMzp8BDpx1OgREymZmZvDMM88gmUwa0nz5fN7Q4OjkyZO3tcd3ZKCz2azg/XzozqQ+P1BnBxlC1noTtIfKJhf8QFSewD5LlgadipZGXzOIq9UqqtUqKpUKLBYLfv7nfx7VahV/8id/gsnJSfHOtBdFgTKZTEilUggGg4bfrZsX8HPSELVaLczOzhrgO0InjBacTieWl5fvZJt/oBcJW9rRoYEkUgIckEU0a1ZDylrZWSwWrK6uwufzIRAIoFgsSl6/kzEJHDiAOl/MaLrdbiMUCsHr9aJUKmFnZ0c4EITG9VlqwhA/F3OXGorTzhyf3WQy4R//8R/xIz/yI0gmk2JgTKb9LlOEs9jY5cSJE9+3c/per0gkIpwL4I0JQzp/z7PkWdBI09nnnWOu0eFwyJ5WKhVx4ugE0OmzWCyS8uI5MqXAFAlrT48cOYK+vj7kcjkDJ6YzbUeIPp/PGyBcrTf4c/zsW1tbmJmZwcbGhtwBq9UqnAjK2pEjR77fR/U9W4SPdTpQR686d0sDqEvVKBP8OZ4D9T3vIu8snXIacupuEgepl4EDhIcs7+3tbYRCIQwMDMBut4uN4OI5Ew3i14jm8gyZgu1Mh9BIx+NxpNNpgfbJy7Lb7Wg0GqhUKujv78dLL710W3t8RwaahpCQhfaWOllx3EDNyuXSsDhJYPSAqdTpDfFPo9EwvI6eERtFsO7Y5/NJC7ZSqYQzZ87A5/PhqaeekhyhvkCacdputxGLxTA4OCiMQO1ZacPS09OD+fl56Z7G3DVfR6GKx+NdFTlFIhFD7Z+GBAFjnTJhSRpGHX3q6COdTiMSiQgzlhCm1WoVJQxALgnljfKVTqfh8/mkxCGTyQhRi0jO0NCQlOMAMBheAIZzIwmOz69TNXwOm82Gp556Cu973/vEQDCd4fP5JE9PpX/q1Cl8/etf/z6c0PdnxeNxQy6W56vhRw1hapRCw+BaSRYKBUQiEUFnqAtYRkkFb7PZYLVakclkxFljVDUyMoJSqYRcLidoViAQENh0ZmYGDofDIJM6aNBkMrPZjEQi8TpDAsDweW/cuIEPfvCDeOGFF4R/A+zDr5lMRnSkz+dDLBb7fh7T93SlUikAry9P1IsyolELnV7qlBl+nY4c95x3V7OsW60W8vk8LJaDEsxqtYrR0VHEYjFxtP1+P8LhsFSEDA8PS726drio2+lAEKUhSbUzHcrPxdeNjIzg2rVrghxzbxg48lmWlpYwOjp6W3t8x6wVnTvQi5eHr9EQkc5Pa6JZo9EwMEDpoZIkQEPIjePFouIuFArY2dmBz+eTySqlUsmQX65UKrjnnntw48YNnDp1CoFAQA6l82LabDbs7e1JlG6328XoAgcUfL/fj/X1dYRCIUP9ZycZCgDC4TB8Pt+dbvMP7GJunQq282JSuDVU9EZNYLSTx6icaYyenh7Y7XYh2QAHcBrP32w2o1gsolAo4L777sPe3p6wunWUVS6XhcwVDAbhcrlQLpcNKEknF4GfiUaal5fPCwDLy8sYHR1FOp2WOuhGowGXy4VUKgWv14tqtQqv14u+vj7Mz88jGo1+H07o+7MI4Xcu3oFOwo1uCEJeCO86EZhQKCQksEqlApfLBafTKRE0oW1yU8iDYEQ8MzOD559/HtFoVCIdGu9isQhg37E4e/YsisWiOFZaSWt4lvefMgQckOD477W1NfyLf/Ev8Oyzz0rLVyriWCyGcDgsaAHls1sWo0SNJnAPGUUzuGLQxtXJASAiZ7fbkclkRL+Qj1IqlcTYaaTO4/FI8OZwOOB2u/Hcc89hZmZGZIxpFcqA1WrF2NgYUqmUwU5xaefLYrGIs6cdNB310/FbWVmB0+kUHedyuaRHO58jlUphaGgIkUjktvb4jlncuhyJixER2XVczCvRe6CxpZHnYTLJzhwDjR4Lvm02m+Qi+Nq1tTWYzft10SRpNJsHNddaSdRqNfzUT/0UfuZnfka6e/G5NcRGeCYej2NgYMDwOTR8sr6+Lu9LqJUCxLx4sViE1+tFNBrtKnKQNmr0cGk8O7kG+gy098xltVqRSCQMXjSVN9Miug5Sy1e5XEYkEsH09DQuX74sNdkkb/FMXS6XOIQkI9brdYPCBSCOAWWOHcD0a2h8zGYzrl27htOnT4vsUda1fLtcLoyOjmJxcVFQlm5ZRKM6c3I6LcGl7xlfw7OmM57L5VAul6VPtu7pTzRFM7UJk1erVZw+fRq3bt3CysoKZmZmAOw710w1UVlSD+3s7GBqagrFYtFABgUgOoROAA2HJooScQOAgYEBrK6uotVqSVqPDPP+/n7k83n4/X5pvqH1ypt9dfJKdMqSDlJnaojnoVFJTRysVqtwOp2G79Fx11Es00b8vSdPnsSLL76ISqWCyclJAAcyQJkhKkpHwGw2Y3l5WRxxAAZ9RiRNOw9ECIEDtHBkZASrq6vI5XLyWSwWCxKJhKTa3G43yuUynE4nIpGIoA/fbd0xxK3zdsBBVMQPoCPOTniY78HoORgMwuFwGJi2DodDSnhYRkNmt81mQzabxd7enkBV+XzeYLgZcfFA+L5msxn/+T//Z/ziL/4iHnjggdfBMprhmU6nkc/nEYlEJApjHjUYDMrv1BA5vWfCb1arVQ7nzTy5pnPpKLPT2PIiERHR0ZQmW/BM+vr65Gudl5t1zZ2pDirUu+66C+l0GrFYTAaV0PnS0CqNAZ+jXC4jGo0amhhoB1OX7ZDPoPOPFosFm5ubeOc734l8Pm8o4XK73YbLeNddd+GVV16Bw+GA1+sVD74bFtmrgLEVJ+/E/9ven0e3fV5nAvADgARBEDtAYuFOkRS125Yt24od747jNrGTVNmapDOnyzTTk3amW9p0pm1muk0mXWY6TTtJF58mcZs4ie3Udp3YseVFsmUtliiKEklxJ8EF+06QBPD9we+5vIDdRprva0+No/ccHdkkRQC/933v8tznPhfYRhvoiDUpB9hWeaMqH7Mj7gmDH5/Ph2w2C4fDIXXqUqkkxu7FF19EV1cXHA7HW0RR9NlhhwWzuoMHD8r5qkX/tHRvJpNBIpEQO8PPNzc3hwMHDiAcDgtyxGfAzJ9fX15eriIr1sOi7dM9xEA1x0SXhTTXgH/za0SbtLMEts+O0+mU1kv6hM3NTbGtr7zyCvbs2VOVBBKh0SUx1rArlQoGBgawa9cuhMNh2VtgG0FjArKxsYFUKoVcLie9zlyZTAYWiwWJREKSTdo7fh4AAqmHQiHhWFzJumoWN/F+HRXr2pImkPGwy4v9fw13JpNBPB4XiJIOTRNJtNgEiWnhcBgtLS3o6+vD0tJSVQ8mHYOufWsIhr/nF37hF/Cbv/mb6OjokM3TkAwN9OTkJLLZLFZWVmAymTA8PIxoNPqWCS2aAMUMb2lpCZ2dnQKl1lPmpJ2drgfxsvJ5MJDTJRGNbHi9Xpw7dw6BQAAA3nKx2O/My5pOp2EwGDAwMIBgMIiRkREJCsmUZsSrszX+LqImm5ub4tBJ6KKx0EGE5ib8+I//uKhUNTc349y5c/KZef5ZE3U4HIjH4xgaGsKZM2dgNpvhdruFYFkvi89No2k0nAxweC80sROorkm3t7fj9ddfl4yaBp9nxWKxSItSJpORIGpgYABjY2OIRqPo7e0Vx00RE43wMKGofV/sAshkMuIYdA1a9+0ajUYMDAxgamoKjY2NWF5ehsViwbFjx+Qz8ZxR+rFYLGLnzp04ffo09uzZg1wuV4UsvNMXbSefk0ZRgGphl1qOkk7y7HY7wuGwBEZM+HiXmSS5XC6RTW1qakJnZydOnjwJAOjt7RV0y2g0vm37H0tvutzqdrul7FULX9N2aN6Vz+fDsWPHEI1GMTc3B6vVKoI0tIc6USyVSmhtbUUikUBfXx/S6bTwlq5kXXUftM6gdA1XE6N0FMsags68+/v7YbFYxMFpDd3a+iEv/erqKgKBADo7O+W/V1dXZUN0HZvCFry07Kkul8sYGBjApUuXsGPHDrm0Gjbj55icnMTy8jKuu+46OBwOHDx4EC6XC9/85jeFuFRbGzebzYhEIiKQoKGdelkMToiK1NafarkHwLZ0I+FqAMJ41TUl9rUbjUZRjyL70Wq14uDBg1heXsb09LQccO4bnbl2zgy2qDSm3yO5CJrEwu9p2Owzn/kMfuVXfqWq5Wf37t2iR04HRCWjXC6H1tZWBINB5PN5tLW1IRKJVDGA62HRAXHva1ETErT4PZ4FYBu+BrbO06233op4PC7oiK5P0nFSsYsCFM899xwCgQDa29vl67lcTsiimh/B+8m9ZRklm81iz549sNvtVYQ3vi6DhH379gEAvvSlL+Huu+/GxsYGdu7cCbfbLaUTPfzHbrcjnU6jvb0dU1NT2L9/v2gn6NGk7/RFFT+geva65qAw6GI3BwMn3YGztraGwcFBkczk7+OdpM8xm83I5/NwOBxobGzEK6+8gv7+fgSDQRQKBZkmxXo37QADp3w+L76BCYHZbMbg4CCSyaTYMh108mv9/f14/vnn8eKLL+LIkSNwOp3YsWMHVlZW5Lzq8i5tktPpxNzcnMxvqEVhfti6KgfNupOOjHR2VC6XRThCQ0uMhuh0qQqjI28+FG6mnrG8ubmJ/v5+NDQ0IB6Pw2DYEjBgbYcfulgsSg8dLymjJ63q80u/9Ev4L//lv6Cjo0M+G4ON9fV1nD59Gvfffz9aW1uRzWbhcrnQ1NSEjo4OnDlzRg6WNuY0QC0tLeJUzGazGKx6WYyU+Zk0GYzfB7b7A3UkyqxUQ80Un2DfssPhkL7R1tZWpNNprK+v48Ybb8SpU6eQTCaFkdnS0iJoCw15qVSqypABCCzFy0YyF4M//b7L5S21uY997GMIBAL4jd/4Dbz//e/H9PR0FVGFmYPud6YB2r17N06fPo1gMFjVhlJPi6IivH8aTaGR03VJnhfd1VGpbIm8tLa2orGxUYRPAEiwY7fbYbfbsby8jJaWFgDA2bNnsXPnTng8HiQSCcmayT+hrSE8zZYbvjfeW/arU0RCZ3k04vfccw++8IUvoL29Hffeey8WFhbgcDjQ0NAgLZmsdTLg51mwWCyYnZ0V+U9KFNfL0hwcoHoKnA5yaC/pM/TPsTZvNpvFFlDghb6A7XLJZBJutxtLS0tYWlrCvn37YLfbpa7MZIw+h/aXzlcTuEqlUpWIFEuXDAz53pqamrBv3z78+q//On76p38aoVAIsVgMfr8fbrdbFCu1zyM/AoA8A/b9kzB2peuqHDQzGbYo1BIAgO1ag65H8cIyYpmcnJRN0tkrIyTCkRsbG9KCw+iHkRovMLF8wiN8EMlkEi0tLVIT5qWlRqrJZML8/LxcMrZhTE5O4uMf/zgikQjsdrv0ttK579mzByaTSQIRbqTVakU6nUYgEKii6QPv7OHstYvGVdcKaw0zf45Lw+LA1hlZXl6Wy1ksFhEKhZDL5eS5U2Qim83ipptuEpas1WqVyw9A6tLkLxiNRnGkDJo48o9IDrDFQHW5XFUkRhKDvvjFL+Ib3/gGzp8/j507d2JmZgZWq1Wy6OXlZTkDNAiEYjs7O4V8VFtvrLdzQGOk910javps6J/XZQcyaQld0omR75FOp6VcFQqFsLCwgL1790qmRfSCMovAFgyZSqUkENP8Bd5VAEIG3LFjh5RKuEwmE+666y786I/+KP7P//k/uHjxIux2u9iUdDqNpaUlCeiYQfM879ixAydOnMB1112HaDQKm81WV84ZQFVSpUtE/1TQTsdHlIK2mXvDoI9Ilslkgs1mk1Y1Jodmsxm7du0SGJtBFmczl8tlab8Fttrd6Cjpb6j3Tz8UCASqRoUaDAbpo/70pz+Nxx57DMePH0dnZ6e8RiqVkmfA907CMAD4/X7Mzc2hv79f0JNaUuUPfcb/Lxui2Zg6k+LF05eStSX+THNzs0iuARC4mJkJDzJhzu7u7ipjT3IIIyuytxkU8OFarVZUKhWZw8rDw8jpyJEjmJ2dlfc1Pz+PtbU1vPe978XCwgLa29tFZpSHkDVqGhItkrK4uIj29nb5Ob4PnW3Ww2KgpQ83L6g2yLVEDWC7lcLv92N2dlbqRuVyGfF4HFarFS6XC+vr64KE3HrrrThx4gRaW1uFFMLLTLIF4a1cLifGnXtDo9zR0SGkHSIqvOC8uK2trXjggQfwq7/6q9izZw8ymYzA7Fp6lIIozNiAbWWt3bt3Y3h4GB6Pp0qsp57OAICqlhndMsfzzmwJeCuqQliToiFOpxNGo1H6nxsaGkTHgKUB1nIHBwextLQEq9Uq+8hseW1tDW1tbQiHwzKSku+rWCwiGAzKmFrdWz8/P49du3ZJduPz+eB2u/E3f/M3eOSRR3D06FHs27dPapfktjArJ2RPm+bz+bCysiIOnTBvvSEpumyhdc5rzzqdN+0n/y0zcJa1gC0xLIfDIRl3NpuFzWaTUunKygqCwSASiUTVa7E1s1AowO12Ix6PCwGZdWgONwqHwyJswvdMqJtlFo/Hg/n5eWQyGfzhH/4hTp48iX379qFS2VKyY7cOy2iaHEmRnIWFBfj9fkkoNR/pX8RBE9LSTDV+SL7JSmV7BjINE4eak5LOS8psh9kG4SFCnJRGo+Fllr2ysiIZK2thjJ7z+bzUF7i5rJXp2vja2hpuuOEGIXns3LlTHKzT6XyLTCkvZ29vLxoaGqR5nT/H7K62HeBqNuOdsBh0aRKNrtdw0VFqh80ySHt7OzKZjPTAM1o1GAwSpHGq1fnz5+F0OgWSjkajMBgMwpDc2NiAx+OpGpLe0tIinAObzVZFOCPXge9tfX0dHo8H7e3tCIfDmJ6eRmdnJxYWFmC1WiUg0+IDuVyuiiDGNsBdu3bh1VdfldpkrRBDPZ0DBq3A9iQiOmbtjHSAqiFNGu3+/n4sLy9LGYJONZvNwmKxoKmpCaurq1haWpK7VyqVEI/HBRZdXFyEy+US20E2bVNTk/Spm0ymqp51OgXuUyQSQSAQQKFQwNzcnExG0qgYExIGXaFQSIwuoXO+58nJSezevbuKHKiDlnpZ2hawC6aWra8zZ5Y4mA0bjUaBqWlPOL5TcxfC4TBMJhN8Ph82Nzcl8KbKVyQSgdfrRSaTEefLRadfqVQQj8fhdDqryMtM7iqVCvx+P6anp5FIJIQVbjAY4Ha7JTHQ51snhTphJemMOuB8TjpxuZJ11RC3ngHLDeIbZR2QWRY3ipu1ubkJr9cLn88nTo6QEf/k83mEQiGsra3BYrFgdXVVKOskk4VCIYTDYXi9XuRyuaoozGKxIJ/PC3GDmTYhcEZt1BC+8cYbxWCzdqmNimalu91ucQy6ZtLc3IxAIFCVMTEo0FllPSxeqloSHC8mP29t3ygACXQY3FmtVmmPIwRVLpdl2EUkEoHRaBRFJvaUAlsXtr29XbJiYLuFjxKbREAoEkH0hRecLM5kMolisQi/3y/sXBqHUqkkAWGxWMTCwgIGBgaqWMH6jNFocP/5TIDqjOOdvhio0SjxrDPw1nVGngd+jVnv6uqqiIrwvNDYl8tltLS0IBqNisqcx+MRWDkQCCAWi6FQKGBoaAhjY2MIhUIS9FG8gnC37l/VwQDhaZPJhKWlJfT09MDj8SAcDsPn88k+lkolJJNJKbEsLS1haGioyhkxWJybm8PQ0JC0hmoSYj3B3LXOhvunu1z4c/QLOsCvVLaEh2w2m9hwYMvGkNTl8Xikp5xlKTpgl8uF1dVVCfovXbqEYDBYpRbJvvR4PC6DKoi4ANt+i/5lY2MDN910E4CtzhGfz1c1F6JYLAoKUCwW4fV6hQPFz8yWMKKw+rUYpF6pT7gqB80okQ++FsLixmhImBEGLw0zGlLlid2bzWYkk0mBMvn/7HMjBGo2m7G8vIxgMIjl5WVhYALb83yBbZKJw+EQoXJqPBMiY32rqalJyAn8PQBEDzgQCMBgMOD8+fNCAOEBY6sOa+E0yvVkjPXShCDtgPiZNVlIZ1HANomkWCzCbrdLuUFHpRaLBS0tLSIlSenHjY0NyXCy2Sz6+/sxOzsrDF3+bjoNQtyEJPl+yOju7e1FqbQ1Z9bj8SASiSCTyUi7Fp0EIVietddffx333nuvGBcGbr29vbh48SJcLhey2aycd9alNNpSL6u2dKFXbR1SM+b5b9gWBUB4BI2NW+P6urq6cPHiRbjdbnHCIyMjgp5Q4tHv92NsbAzd3d0CexLhY8DHEsb6+josFosEcUajEe3t7RgeHobP50OpVMLy8jKamppEAYxnPJ1OY2hoCOfPn5daOCfr8Uz5/X4hJOkZ4DyLwPadqKfFc6ADdL3XtZk0gCpHrnlKAITx3tXVhUuXLsHv98v/T09PC8+IrVcNDQ1YWlpCW1ubBGjkhmjxG54DEoiJgjqdTpw6dQo+nw/pdBrRaBRutxt2u114T+xMCAQCWFxcFIIxtRDoxK1WKxYWFtDc3Fwl2qSDGZ3d/7B11Rk0HyrwVok8Lq2zTEPHzIIXQxOwgC0YwuPxSME9kUhILYK/m1NISCRpaWmRuhVrg7z0bL1IJpOSVadSKXEOlUpFoK+1tbW3DFIwGo2Ix+OwWCwYGRlBqVTC4uKiwN9khbP2oevc3Ag6sXrKoHmRNEysW66At5+nygyJRo+Bmoa3DQaD9JGur6+js7MTyWRSHCFZt6wPEdIirM3yhtVqlQvMqJYZTltbG6xWK1ZXV+FyuZBKpZBKpWC32+V9aEgyGo3CbDZL33wikZDxdy0tLUJmmp+fl758nnktzkCnX2+LhpCZombu19bf+fkJc9JYUtzF6/XK/g4PD6O3txfxeBz79u3DG2+8IdAkgx0SCdva2pBKpQSRWVtbk2A+n8+Lo9bqYt3d3VhaWkI8Hkd3dzemp6fhdrtFaY6GHoBkRadPn8a9996LCxcuYM+ePXj22WcBQBxApVJBOBxGb28vcrlcVSugvhP1tIiG1d6b2u/x8/MMaDiYDtdqtcJutyObzcLr9WJ4eBh9fX3IZDLo7u7GxMQEXC4XisWilLGoj2C1WqXFjqUrm82Gzc3NqtGObA2rVCro6OjA5cuXUS6XEQqFMD8/j2AwCLvdjnw+XyVrzKBrdHQUg4ODGBkZQblcrhoYwwwZgAR4tAEaIdLByA9b/09Sn7X4OWswwDbMWAt18evLy8vyAMl205H05uYmMpkM/H6/tExoERLWIrPZLDY3t6abeDweIW6ZTFvaqVarFSsrK5IZp9NpeDweMZh0MBqi0n+ALWUop9MprTv9/f1S92aW5Ha7q/4N4TDtlOux9qidr0ZK+P9A9QQwGlY6MJJEGOjwACeTSWxsbEi/O0slFADxer2YmZnBxsYGFhYW5Iwkk0l4PB7pp+WZZNvOxsYGenp6sLq6KhAUh2yQmKgzQYPBgPn5efT19WF8fBypVArJZBK33XabvCbJJYRESVrkheRzuVpiyDthkQ1bO3QCqNYpfrsyiGa+cqQgIWSfz4fLly9LcLZ3716cPn0aHo9HSgp8TZa3GIjlcjnJkNm2B2wTG9PpNIAtxu758+cxODiISCSCpqYmBIPBKrRFG1MSvpxOJxKJBM6dO4cHHnhAnkU6nRaGLjM6/fkZjNQbD0G3sQLV+857q7t99Blh4MySFgNoIlpTU1Po6+tDKpVCKBTC5cuXZWyvhorJX2KLFaHvVColoiBM4EgsNZlMsNvtmJ6eFtErr9eLQCBQJf+q3zdbspqbm2E2m5FKpdDV1SW1Zp5Bg8EAn89XFZDokq8++1eyrlqoRI9849KGWqs7aQNdLBartJRbWlqEeWu326UOxQ1iPY8ZNGtFuVxOJoeYzWacPHkS8XgcnZ2d8hB27tyJcrmM/fv345VXXhHCGbAdrdE56wvDC01jkMvlMDw8LJHTysqKMD3Z68wsiksbZD6Telq1mTEdoVYQA1B1OemUaaSY1ZD9SiOWzWZhMBgQDAal/ryxsQGv1yv1yrGxMezfv79KY9fv92N9fR25XA65XA4rKysYGRnB2NgYlpaWEAqFEAwGMT09DZfLhZWVlSoISiuc8TNFIhHMz8/j/PnzuPPOO3H48GF8/vOfx6233ipQaaFQQEdHhygK6eehA1b9e+tlce+IGugAtbbERSMFVKtLMahj1sN+5+7ubmQyGezYsUPgZ5KB2JKTzWaxf/9+JBIJrK6uivDMyMiITLNLp9NIpVLiIPv7+5HNZrGwsACLxYJUKgW/349isSgyrDrZMBi2RtAGg0Fks1lcf/31OH/+PAqFgughUI3M7XZjbGwMHR0dVXPqtaOvNz4K7w9tYi3vRDtqBu98Drxv3H+r1Sr1Yp6BdDqN1tZWzM/Pw+VyCQ+kublZ+AC7d+8WERqbzYZIJILh4WFphyoWi8jn82hpaYHX68W+ffsQDofFP1UqFbjdbmQyGbE/Gg2oVCpIpVKwWCxobW1Ff38/nnvuOTQ0NMg8CPJpKPmpp9gxmNCfm8/uStZVWQxCxSRYMPOsbW3RtTeSaHhgg8Gg1JedTmdVHTOXy0k2Q3iBxDOn04lsNitG+Ny5cwiFQrj99ttRLBaxuLiIUqmEjo4OhMNhJJNJfPnLX8bHPvYxrKysyAXnw9E1EB4msnlTqZQweltbW3HDDTcgkUgIjEpj4/V6hYpsl6wAAGoLSURBVJymL7ZmBdYGAe/0RWdLZ1wrUMGDyeehFbwqle3JMrwAqVRKyDSsB7G+SJgqkUjA4/FgZmYGBw4cwMmTJ9HW1ias3jNnzojQBHsnOzo60NPTg5/4iZ9AOByG1WoVPWw6Fb5/GpGGhq0hJ0888QRisRgOHz6M++67Dx0dHfjsZz+LT3/605ifn6+CcdlSQxlCDYnpGm29ZU+82zz/RBRqSzr/1Odmm1JtH7rP50M8Hkdvby9GR0cRDAZhNBqRzWaF/FksFjE4OIjh4WG43W6EQiFcvHgRyWQSnZ2dcLlcGB8fRzAYFMj6jjvuwO/+7u/ixhtvRDQalaSAgaK2ZcAWqjc7O4tDhw6hvb0duVwOr732Gp566in8/d//PYaHh9HQ0CD916lUSpIEZpb8fTp7rqczwDvEEhRtgi4rAduqYLU1aWDrfGjicWNjI/x+vwRGi4uLglLG43E5CxyWEQ6H4fF4YDKZcPr0aTgcDvT09MDn8yESiWD//v3C8Pd6vfijP/ojHDp0CJOTk/B6vbIvup+fZ5jdGkeOHEF7ezvGxsbwyiuvYGVlBT/2Yz+GxcVFGI1GkXQm0VS3n+pnoMmCunX4n1tX5aDZRlAqleSh8iHrpnNuHNsv+OGpCDY/Pw+TyYRwOCwFe7ZVEIrgf/Pfa5ipoaEBHR0dOHfunBj4paUlpFIpvPnmm1heXsbo6CgefPBBqWPUMg31f1ssFrhcLuRyObzxxhuYmJiQ6PrAgQN45ZVXRM2GIgXMHmqhXb0p/P96ipqB7daCWifHz0kDpfdeMzmpnc1aks1mk55mwkUbGxvi9Agh79y5E8PDw2hqasLi4iLGxsbQ3NyMvXv3ijAJGfabm5v4xCc+ga9//evYt2+f8Ajerv2N5/b555/H+Pg4PvCBD+DQoUPo6ekBAPzH//gfcfPNN2NgYEAGp1QqFfT09GBxcVE+C4M93Q+pjfKVXsp3wtLa+TqD1udAkwl1/ZWGSqNkZOdGo1EpK5AgSgfe1taG9fV17Nq1C2+++aYgYUajEaFQSKDHM2fOIBgMYmlpCcViETfddBPe//734/Of/zyOHj0qqoTcG9oZ7t3ly5exc+dO+P1+XLhwAVNTU5idncXf//3f44033sDv//7vSxLS2NgIr9eL8fFx+P1+IaLSSbHlS5//elm0bdrxvh2CosnDGuKmLaBdbWlpwerqKqLRKPx+P6amptDS0iI8BaPRKGJERNXYotfX14edO3cim81iamoKTz/9NMrlMk6ePCltc7/927+NI0eOYGRkBENDQ1W+QJckKC39vve9D2tra3j88cfx+uuvY2RkBNFoFH/3d3+HRx999C3ZNtv4tH2kY9bIEZGGK3rGV7MhOmPixdDf0xkj+wb5ppg1sZbHr7W1tUlTOQ0roXBmVaS7r6ysYG1tDYVCQRx2Q0MDVldXsWvXLkQiEayvryMSieDQoUNYWVmB3+9HJpORS6LrQUajEXv27EE2m0U8HsczzzwjsHUoFMLhw4fxzDPPIBaLCTJASITtVqTca6P/djW4elncZ50V1AYh3NtaUgydYzablbFyzKJYwiALlmhFc3OzKL69/vrrMJvNMtXI4/Ggu7sb3/3ud3Hw4EEsLS1JHfOzn/0svvCFL+CDH/wgXn31Vfj9fmnJquUbfO9738PIyAg+9KEP4e6774bX60UikcDS0hK+8pWv4IEHHkAgEMDKygqMRqPUXgOBAObn52G32yWKrmW41/ZL1stixlxbS9NnAoDYAc301WUSqj9R+6C3txeTk5Nobm4W1GxzcxM+nw/hcBhOpxPPPvssDh06BKPRiEQiISz/sbExvPe97xWiYTqdxvXXX4///t//O/7rf/2vOHv2rMDnfG86cD958iS6u7vxwAMPIBwOIx6PY3x8HOVyGadOncLp06fxW7/1W6IiyOypoaEBXq9XkhSTaXu4AoNXGuV6CtIo4gS8NfHRZ4D2QvM8aEt573XXTmdnp9wr1v1pE2gfJiYmcNNNN8FkMiEWi2FiYgKLi4uIRCJ44IEHRD7WYNginr722mu47777BI1j941GOMvlMi5evIj77rsP+/btwxNPPIHR0VEMDw/D6XTCbDbjD//wD/Hrv/7rVcIsJDuS5Ky/zt+rA5lajtI/t646nNMQnnbYWqEJ2G654htkjfnRRx/FJz7xCWHpUqnJarUiFosJGSyVSskBMJm2lLqampowPj4OAAIrLS8vIxQK4eTJk8hkMmhpaREZN/a2MVqnQSFJqVwu49vf/rY4/dtvvx0OhwM7d+7Erl278Pjjj4vcGz8Xfwcb33WWz43Qh7beYC1gm/z1doGIjp6B7WfAiNlkMok+MrPecDiMHTt2IBKJiPIS2fY0wPF4HDfeeKMMISB55Dvf+Q4efvhhnDhxAl1dXZJ1v/jiixgYGMCpU6eqxkvSsG5ubs39PnbsGP7dv/t36O/vRzgcRjgcRjQahclkwm/8xm/gox/9qLCLtcHp6urCzMwMnE6nQPTaCWtYW6Mq9bJ43jUJBtjORLQ6FLNILp1FUHAiGAyKYIjFYkFXVxcWFhbQ1NQEp9OJ2dlZeY077rgDL730Enp6ejA9PY1CoYDZ2VncdttteOSRR3D//fcjGo3i+PHjcLlc6O7uRqVSqdpHJg+lUgmxWAyhUAgHDhzAqVOn8NprryEWi8Hr9aKzsxNvvPEGHn30UXz+858X/gkRxI6ODpw4cQJDQ0OIx+NVA3/4nGpJlfWytEYEsM3M5t2v7YXXvCV9H2KxmGhkEAVxOBxwu91SnrJYLJifn4fZbEYgEEAwGMRLL70Eh8OBqakpJBIJ2Gw2OBwOfO9735PJY0ePHsXBgwelE4MiVtp2U7jm7rvvhsvlwte+9jWcPn0aVqsV1113HQYGBmCxWPDTP/3T+OM//mNx+jzTNpsNq6urkgRo3g2w3cHC53I1XISrbrOqrTNpJ61JAZrBysyaRKHu7m4sLi4iFovBYDCgra0N0WgULS0tMBgMMqDC6XRifn4eiURCCAJDQ0PIZrOStWxubqK1tRV33HGHzIft6OioYswxayZ5gxHY8PAw9u/fj+bmZly+fBkejwe7d+9GNBrFCy+8UCVYwc9EeTfCszpQ0W1WmqVYT6uWVKfJQVw0RpoYw3/H50mt9KamJrhcLkxMTFQxg4m2AFttdX6/XzKgbDaLvr4+mce7vLyM/v5+eDwedHZ2Yvfu3bh06dJbGKUspxAlaWtrw549e3Dx4kW5qDabDWfPnsVXvvIVfPGLX0QkEpFsSMPbTqcTS0tL8Hg8VYQ1YDsz03900FJPS5cJ+DnZdqUzBd4DDftzNCtbJHft2iUtUjMzM9KKt7KyAofDAZfLBYvFgueffx5DQ0N46aWX8NBDD2F+fh79/f2YmZnBjh07cOzYMbS2tuI3f/M38ed//ue49dZbxWhSi5sBo81mg8fjwcTEhMwFZg/zyZMn4fP58GM/9mP4yle+UiXUxCCEYhUs+wHVbZa0G/r818vi/ddZo/7cGrXUvByd7LCOT70DqkcSGdUKgxSmisVimJ6eRigUwvT0NA4cOCBa7kQ1R0ZGYLFY8MlPfhI/+MEPcPDgQbHLTNL4Xjh059FHH0WpVILFYpERscPDw3jf+94Hs9mMxx9/XPaZn51iNy6XS3gZPFv/VKJ2NUjKVZ0Wsqzf7ut6cQNqiQEkhnz5y1/GQw89hJaWFlitViwvL4tiy9raGnw+HwwGg1DtvV6v1K3JAE6n03C73WhtbcXCwoJEXdxQPeKShpmrUqlIFkZ6/Hve8x5MTU3h5MmT0jutYQpCMXTMmqGu/653iBuonkykURLtlHk5NTFCQ0nAluMl5O1wOODz+RCNRhEKhVAsFlEoFLC8vAxgi7TT3t4Ok8kkDM5KZYtYyJYaynTOzc0Jy16rGlHpCdiua5ZKW7KRdAxsl/n4xz+OxcXFKoPDcZmVSkUgei0Jy0CQl5+Qfj1mUJoIpdvaGIjzudFR8x5ppM1gMMi9L5fLmJycrIICW1tbsbS0JLO4yTM5fPgw4vG4jJ0cGhpCNBoVhbDV1VV0dXXhhRdewKFDh0RGkueRgdba2profZNE6nQ6UalU4HK5cOeddyISieDNN99EpVIRYSV2qdhsNoTDYfT09CAajQobnUuXPDT6Ui9LlwjonN7u/jNw0Zkj7wV72mOxGBobGxGNRuW+UO6XIyS5N6VSCZ2dnaImef78eXR0dEjgTCLy4OAg5ubmsGvXrir5aRL5mAxyL9nZwTq32+3GPffcg+985zuYnp4WXhXPO9HftbU1EShiAKeDlrcjjV1poHZVDrqWXMMXqnVOwLaoSW39gZEF5d1Ys4lGo3A4HGhqakI0GkVbWxs8Ho+ouvh8PhlScOHCBZHaHBsbg8FgwPj4OHp7e6s0l7nJZILrQ2M0GuHz+URK9PLly5Jl62HfWmeYGQJFSbRRArYDEt1qVY8sbmAbJdCXlH+zXs9I8e1QF8I+bJEDIEZ3dnYWGxsbiMfjOHDgQBVExv0lrJzL5WRO+KFDh4TBz4yYHQesXScSCXl9tnhwjxn86RnVfP906Kw3zs3NCfKja6qaPEcnps9FvSztcLSmgM6iGhoaRIhIB2d8NsyyKBaza9cuEaqgYiA7JSqVCjo7O5FKpUQ8JpfLobe3F6lUCtlsFn6/H2+++SYGBgYk6Cdrf2NjQyRk2Xvf0tJSJSrjcrkQjUYRiURQLpcRiURkGI9GCPn/FMLQg0P0WWAAqwPUesqgge3WWu6pDmj5WdlXDGwzurXdbGjYGiNcLpfR29srz5Otk62traLq5Xa7kcvlRPWvVCqht7cX4XAYiUQCJpMJL7/8MsrlMlZWVmSEKbtCCoWCDDoqlUqCphgMBiElRiIRmVQ3NTVVNaec54B/dLlUs9f1560Nzmq5Of/cuuoMutbZ6D43DSOSyUgpT8JB/NnBwUH84Ac/wOrqqlDmmeGEQiG8/PLLmJubw8bGBiKRiAiZnDt3Dr/4i7+IpqYmXLp0SfrlbDYb/H6/kI7IFCb0wehW18Wam5uRyWQkS+vu7haCmm4boTCCnlFNOIVa3ED1RC8+Fz3isB4WD6Z2Nnyubxcd1w7U4LNqaGhAKpUSp8gaFJGSZDKJwcFBjI2NIRAIiEocURga3La2Ntx5552IxWI4evSoZD+EneggiH5oRR9muWQkU3c5EAhUfTa+d4pi+P1+MUx6ig2dDv9bB66aMFYPi/dIS//qcha5GsxMyUEBtp8jM21g68ywXY7tbix5tbe3480335TuCQb/Ho8Hy8vLEqDNzMzgAx/4gHSEsHeW56m9vR2JRELOBBfPL1XoAMg+17ZLcoIV73dfXx8ikYiQRjV5Vtfm9Vmot6U1JfQZAKr5R+QrMKirVfpzuVyYmppCoVCQu+7z+URDe3p6WlQGee/ogPv6+hAKhXDq1Cncc889cLlc8Hq9cLlcUopMJBIyo10jfsD2/nDvNjc3hRzMBJN3WncwkMDIITw8S+Qe6H2v5SddybrqcE7XY/nC2uml02lxttqI+3w+ecNmsxlnzpypmjZltVqRz+dx6dIlLCws4IYbbhComcY4GAyiv78fyWQS3/72t9HW1iZwNYUHmD23tLRgeXlZLp2GnZn9OhwO0Xa22+3w+/1SnyCZgIaV74NN9YQ41tbWRJ1G93ZyE+oN2gS2syAGI7XQNb9Gx6sNIDNYksOoke7xeETFZ2VlBT09PdjY2EB7e7vsI+GjaDQq4wgffvhhPPPMM8LezuVyAptR/pFBoy478GwywyaRiedIDzth0MazS/RF92PyM2k4VztsoL7arLivfB7ANoLGwJwjH6kGSMh7fX0d7e3tco4otaohUYPBIH3P4XAY119/PWKxWFXvdCKRQEdHByYnJ2G1WhEMBvH000/LRCkibul0Gl1dXZibmxPUhGeUr6mRD5ZOEomEIHw8M7RtDocDCwsLsNlsAKoHoWiWsg4I680OaGekA3H6BM0p0Wx2qr6xjVGjkqztt7S0IJPJIBaLiU5+b2+vKHdx76g8ODs7i4WFBRw4cAAjIyPy/bW1NZlC1d7ejpmZGbnnOuHkudOCWdRY59c034aQPm0/HTJJj9R/0OdHn4ErDdav2kHX1nL1plBRRUdQOgLhpTWbzTh37hx+5md+RjaRMFkwGMT4+LjUn3nJW1pakEql5CF6PB5Rd+Fm6sEZrBVpB6mjGr5XLRXJeiRrjTw0hLVYmyARgJB5uVwWhTMqppHpCaDKSNfD0odLOx06qNrgjWeGk6WIPrjdbiwsLMDtdiMajSIWiwEAurq6kM1mEYvFsLGxAafTKb/fZrMhGAzi8uXL+JEf+RH83//7f+FwOORZB4NBTE1NwePxCNOeZ48XS5O4DAaDEFMohsHAj7+TiFChUEBPTw/Gx8eltspImueKhpoGC9iGueoN3uTeM4vWCBWAKgfFbJJ3gdlwpVJBb28vZmdnZb9SqRQGBwcF2p6ZmZHWTAAyjrCnpwcrKyt417vehTfffBPj4+M4ePAgmpubpXbpcDjQ3NwsE8l02aGWXUvHzTPh9XqRzWarnE+ptDV61OfzIZFIYG1tTTgpGuoGILZF28N6ctL8XOT66NIGsM070loIuuslFotJGYATp/hvNjY2cODAAalNRyIR4R4RIW1oaEAwGITFYoHf7xclup07d6KlpUVQHNqPWCyG5uZmsf8AqrJeOmnai6amJjgcDuEp6L3TJVPuvX7v5FMxOHg7B38l66pr0LqNRGfP2ihrI8UPTsiqUCggHA4jGAyioaEBiURCel1pyClIwJpQLpfDzMwM5ufnYbFY8MQTT2Dfvn0yDKOxsVHmhLI2zYfDWiQlPjW7lBfZ4XBI3yTnjhKaZ6S0vr4u4+c0jEWCCQ00IfV8Pi8tHTT09bLocHStiZdP1+w19K2hvcbGRhkwQTRifX0dS0tL6O/vh9G41d/a1tYGAJibm0OlUsHq6qrUIB944AGcOHECN954o7yexWLBsWPHZC4wp5UB28YSQJXBZK07FotVlS842KW2TcZmswncDqDKODEQJYKiCWq12fQ7fWkGr85AgWq9dqBa7pL3iX87nU688cYbcLvd4ugymYwMnbhw4QL27t2L6elpQVXS6TQWFxcxMjKC+++/H3/+538Oo9GI66+/XjL7M2fOYNeuXWLgdZufDtZ4TqkDTaJYLpfD8vKyIHzAVpmkXC6jubkZS0tL6OzslAyKdoZGmo5KOyay/etlaU5RbRDGO6/LH7r1Vne5NDY24tKlSxgZGRHVsGw2i/HxcSl7dXd3Y3h4GHv27KlKqjgs5cKFC0gkEujp6ZEzMDc3J7wFDWVrB6l7lfleSV7O5XJVeuO0+fQfmUwGra2tVZoPRJVY4qVtoj/SnVBXsq7KQRPeBqovoc5SaYw0k5sbQvjirrvuwtmzZzE+Pi61YYPBIKPBKFRAdmdbWxtWVlbQ39+PTCaDvXv3IhQKiWHk9JMf/OAHGB0dRblcxr333ot3vetdiMViMpRDT7Xp6OgQUlo8HhcCEfvZCGvzcDGy4mfSn19D4ozQuPHMqOtlca/4mXjR9MHXNTg6LQ2D0+nNz89j9+7duHjxItbW1nDjjTfixIkTVfq6m5ub6OrqQrFYRFdXl4yWbGxsxK233oqVlRURuz927BhuueUWdHd3y3tMJpMiSqAj+3K5LNJ9LpcLN9xwAzY2NrC6uioX0GKxSJmFzjmbzUo5RjslTQzi//M50GHXk4PWRrj2b/392qWFW8xmM+bn53H//ffDarXiiSeewI4dO3DbbbdhdHQUS0tLGBwcFGZ/KpVCU1MTQqEQXC4XhoaG8KUvfQmf+tSncOjQIayurqKhoQGXLl3CZz7zGRw/fhyRSAQrKysol7cGrpDUw/e2ubmJxcVFpNNp7Nu3T+aDx+NxLC8vw+VyiS2i/aP8bFtb21t4OW9HCuXdIJpYL0uXczQZUpNqgW3nxz8M7lhDXl5exoMPPoje3l689tpr6OvrQ0dHBzKZDCKRiPQX+/1+LCwsCGxsNBoRCATw8ssvY8+ePdi/fz9yuRwMBgMSiQQ++MEP4rnnnkOxWITT6UQgEEBra6s4SJ1YUCzr3e9+N5xOZ1XbFrUZiJ7SCQOQbp9aAiCDEa0axrIfSahXsq5aSQzYhgW4+GaZqfJA6jaXfD6PWCyGvr4+nD17FplMBseOHUMsFsPc3BxcLhcMBgNGR0dlJCDryE6nEw8++CCWlpbg8/lEX9dsNsPpdKKjowP5fB733HMP7rjjDjQ2NuLFF1/E//pf/wtDQ0OSGfGw2Gw2nDhxQqYV2e12hMNhRCIRdHV1IRwOCwOVG+hwOGTqlobGyuWyBBJ00nToJpOpShK1HpYmR+l6kyY/MKLUpQ8GMrt370axWJRJRcePH8fa2hr27t2LeDyOO++8E08//TS6u7vFubNkwNFuDz/8MPL5PC5fvoxCoYD+/n58/vOfR3d3N958801Eo1Gsrq5KXVJHrNyjV199FQaDAT09Pcjn80in01hbW5P6JXux+ZnX19eFYe71eqsupXa8Okjh2dFGrF4Wz7UO2rkYDFGvv9aJmc1muFwuLCwsIBAIYHx8HG+++Sbe9773wWQy4W//9m9x5513IplMIpVKwWg0CvqWTqexsrIi2uwPP/wwDAYDFhYWMDQ0hBdeeAFerxdPPfUU9u/fD7/fL/LA586dk2yYdeKVlRUEAgFsbm7ixIkTkum5XC5h/rIOSUY3SUcABErV9qCWtc3gpd4Io9oBMajVPA+SqIhgMMskysq578FgEE8++SSOHz+OBx98EGtra3j99ddFsTGbzYoNIOGvubkZwWAQ58+fx+HDh2X0LCVjHQ4HFhcX8d73vhe7du1CIpHA5cuXcf78edkvngEy/bPZLC5evAibzVZVsuLPEvrW32PCxzPOBJUscQYCfFZM+q6ULHjVDpq4PrA971OTbMjK1QeW/+31eqWFyuVyid5xZ2cn1tfX0dbWhrW1NelljUaj+MhHPoJnn30WMzMzAjsUi0VR+GpubsbZs2cxMTGBixcv4vvf/z5+/dd/Hfv27cOHP/xhTE9PC+Fnc3Nrdu/09DRWV1fx0ksvCemEU7pojHmxeMhqB3vwc+n+Sn5uTQRgraJelj6EjGJ1Bq3rjvoQbmxsYO/evXjqqafgcDiQSqWwtLQEl8uFm2++GZOTkzJ5ZnNzU9S8yL5mueCuu+7CG2+8gdHRUYlqv/zlL+Mnf/In0dHRAb/fj5tvvhn33HNP1fQ0ntXm5mb83d/9Ha677jocOnQIBw4cEMIfWcM0Mpz3TQPk8/kAQPSDaYz1JeZn5wX9pwhj7/TF+nAtD0XXaqmtzvvE+2Wz2XDmzBkMDg4iFouhpaUFBw8exOnTp7G0tISHHnpIht6T6MOygsfjQUtLCwYHB9HU1IRTp06JfONLL72Ej33sY2hsbBRVr+npaZw5cwb5fB6hUEhg5sbGRrz00kv4yEc+ArPZLOWtiYkJOTOsLwOoGqhBaJOziBlIEt7mmeBd0e2m9bYYhOjgU2fRDIh1u2VDQwMymQzm5ubQ09MjKo4f+MAH8O1vfxvRaBS33HKLzFng70un0yIBy/KV2+1GJBIRJwtscVhyuRxGR0dx4cIFPPbYYxgZGYHL5YLb7Ratf6PRiNXVVbz3ve/F5uYmLl68iOPHj2NxcVGCKSI92tbT/tlstipJVx2YaKIcS8O6JPgvQhJjm5SOCvnirAvUEqwIhT744INCRe/o6BB25sbGhmSymUxGnPW9996Ly5cv4/Tp07jrrruwtraG1tZWaYPgxj///PMCYyWTSfT19eGXf/mX8Qd/8AdYWloSggDJAk8//bQECTt27JD6Jg9Qbf2QxKZMJiMDPOiAaaR0NqVrHdxMRu31sHgA9SXk59aHFEDVQQyHwxgfH8eRI0dQqVRw8uRJFAoF3HbbbZiamoLRaMQtt9wiLSsulwsAsLq6KmWD1tZWTE9Pw+12Y2BgALlcDkePHsV//s//WaL0hoYGuN1uXL58GblcDpFIRN53Y2MjZmdn8cADD8DtdiORSMje0HkSfiePgdO33G63vBf2YmvDpGEvniVNkmtqahIHX29LB6Sayc/AmPbBaDQiGo1iaWkJn/3sZxGJRJDNZtHW1gaHw4HDhw9jYWEBi4uLEky53W753WybbGtrw3e+8x05M5lMBrOzszhy5AiGh4eFfV0ul5FKpTAwMIBwOCwCE+VyGdPT0/jd3/1dfPWrX8XU1BQWFhbQ2dlZBcGTBKRbySwWiwSWhKzpuLVQDYMRIgn8fj0F6/y8tZKe/H/aT400VioVmVb2vve9D8vLy8jlcvB6vVhZWcH999+PbDaLiYkJsd28Q5RqbWhokAE6LS0tGBgYELTG6XTi/PnziMViSCaTcLlcokA3NzdXlUzl83ncdttt+JM/+RPRZ7fb7RJob2xsCLGQ/o17zjutS58a2tccLc6Y4CJx7krWVWfQenyghmsI8bJWwyhpYGAAFy5cwKVLl3DfffdhfX0dX/jCF9DW1ga73S762dFoFIODgzIV5uLFixgaGkKlUkE0GpU2LcJdFosFExMTVdBoPp9HR0cHWltbMTAwgI2NDak/2+12HD9+HO9617uQy+XQ1dUFj8eDhYUFMabNzc2w2Wxob2+XGoQ2vLV1JC3EwE3VDGFCp/XkoJmBaEcMbDN2dX2Hz2NxcRGf+cxnsLm5iR/84Af4oz/6I/zsz/4sbrrpJjz33HPw+/2wWq3CvvZ6vYjH49IXz5aVQ4cOCaFwdXUVzzzzDD75yU9idXVVMhu73Y5//Md/xNzcHPr6+qouDyPbHTt2wGjcmqTzjW98A319fXIGisUigsGg7Cvb7gKBABYXF2X+dy3ngsGYRpf4mu3t7Whra7viIe3vhEV0RGfSmjxIR0Ujtb6+Dq/Xi3e/+904e/Ysfud3fgc+nw/vf//7cfnyZXzzm99EOp2Wmh5VAVlXZgeF2WyGzWbDLbfcApfLhfn5ecRiMbzvfe/DiRMnYLfbpTzx3e9+F9dffz0OHTqEtrY2eT9suXnttdfgdrsxNTWFoaEhkRplRkzEji1gm5ub0mPPLFl3eTAw050MJK7SebydEuM7efF56v/XsLeuTRsMBgwODsLv9+NrX/savvrVr+JHf/RHZTDFmTNnkMvlRJmLhN9EIlFVOi2VSnLXmQVbLBYEAgFcvnxZOoIKhQJOnz6Nm266Ce9617tk+iBbgL1eL0ZGRuD1evH6669jYGAAO3fuRDwelx5tsvYJY/NcaAesuVi1UD+fD59JY2MjlpeXsbS0dEXP96octKaW00DVEm90jXZjYwNvvvkmHnzwQdjtdjz11FNYXl7G5z73OQwPD2N6eho+nw8bGxvo7e3FxYsX4fP5kMlkpGWpVCohkUggmUzK6zmdToyNjcHj8aBcLmPv3r0YHx+Hz+fD2NgYKpUKDh48iFwuJxfFaNwaUZZMJhEKhZBMJlEul4X93dbWhlgshoGBAWQyGfmMRqNRSEKa5MCMgBk9ezO5KRz24XA4RF6ynlYtxMnnQ+OmyYIf/ehH8b//9//G8PAwAOCnfuqnMDk5KRDVysqKOPdisYhoNAqn01lFyuvv78eJEydQqWwpSn3961/Hpz/9aSwtLcmlpc4zSV/U8mZrF5WDYrGYnI0Pf/jDyGazVcxeq9WKbDaLlpYWUStrbm6WKFqjBXrpzJlcBxLbYrHYWwQy3smLGQCzxVrIrpZ/0NrailgshieffBK33nor3vOe9yCdTuPZZ5/Fzp070dXVJV0a2WwWpdLW2MeDBw9ic3NTWuC6u7vxxBNPYHx8HC0tLTh37hw+9alP4ZVXXhHD3N7ejsbGRtxwww0AgFOnTokxZ+DA89rc3IxDhw4hEAjg+eefl/ojOzGoo2AymaQG3tnZKe0+hEorlQqsVmtVYEZkjq+3uLhYVzVo3a2j7YBeRKOMRiOcTieefPJJzM7O4hOf+AQ+9alP4ZlnnsHIyAgOHz4Mu92OxcVFWCwWefbxeFwEpEgcpeyrwWCQmQz9/f24dOmSyIN6PB4EAgHs3btXeqk9Ho/sLZExi8WCYDCIH/3RH0WhUMDZs2elNY8BN/umSfBiHzz3XjO89ax5XeYjDyscDsPhcFSJIf1z66ozaF0Er4V2dQSxsbGBnp4e9Pf34+TJk5ifn5eL/Oqrr+Lee+9Ff3+/sHLffPNNWCyWKmq+1WrF2toadu/ejf7+fhgMBnR3d+PUqVNoaWlBc3OzwAVNTU1iUDOZTNX743tiS4+e6cwLlMlkqiQfK5WK1KUZlACQz6jrEfwajREjLo/HIw68XpbWmQZQlTFoeB+AZBkvvPACbrrpJhw6dAj5fB7BYBB2ux0jIyPC0udFWFtbE+1lXbft7e1FMBjEysoK/vIv/xK/+Iu/iMuXL8slMpvNcokYdfP95fP5Kuer4SY6Z77XYrEorXPst6XmNomChNI1vKkzdb5uIBBAOByW33Oldad3yqrtedecAy1oxIDVYDDgoYceQiaTwdLSEiqVrdY4DqXgiNGmpiYkk0nYbDakUqkqpzc+Po6f/MmfRGdnJx5//HF85StfwWOPPSYGUosFcT8Jj7ONihkf/zYYDIhGo/B4PJLtAhBREwpWBAIBYXYD2+ebwTzbOjXMyy6RbDYrrWT1tt7OMWuiMPdvbW0N73nPe3D//ffj5MmTQtpLJBJIpVLYs2ePwMa8W06nU5IBIldTU1O4+eabYbPZMDk5iYceegjHjx8XUi4AGVdMrW06fO4V94ftdfl8Hvl8XsjKWt6TP2cwGGRCIpM2bfcBVCEkGmldWFhApVJBIBAQ0uKVrKvOoJkV8U3RQNNQAZCfSaVSaGxsxMDAABKJhNTyjEYj4vG4CMx3dXWhoaEB7e3t2NzcxMLCgtSbgK0+2M3NTQwNDeHYsWNwOBxyCBobG2E2m2VUGS8mFaU0SYVOnJKdNptNJlPxgVLogn3TFEGhYpSuq2qyFA8Pv+Z2u+uOwQ1slzY0KYT/z++RGONwOJDL5bCysoLR0VGZNrSwsABgm9VMQZJkMomzZ8/i0KFDwg3gOnr0KKxWK6anp7Fnzx6k02n5HomLdJ4sw2hyBrB9Lvl9XirCWGRk8nfRWIdCIVEz41nXASA/L59PpVJBf3+/1Lxqz009LHIzGADXtlXxPnD4QC6Xw9TUFOLxOPbv349CoSB98DTOHHhTLpdx6dIl/NRP/RTGx8clM21oaMDk5CSOHz+O0dFRfPvb38bv/u7vCvzI1ydBjdwFm80mGS/vKveNvJKmpibRM9jc3JRgrFTaErAxGAzCJOfvJwSqeQh6jyuVisysd7vdAt/Xy9L1dj5T4K2jVtn3vry8jDfeeAMGgwHvec97sLq6imKxiJ07d2J0dFSQRo5ynJ6exp133ilDVABIj/Hi4iKWl5fx6U9/Gt/61rfkDOifsdvtgpww+OMeMTgzmUyCmnBgBv2GxWLB+vq6lDqAbZIsS7ra7ml4Xydv9Ac+n0+SwCsljF51Bq0NHLAdRXJzeHhtNhsMBgPOnz+PSCSC5eVldHR0wOfzSeuEw+EQvL+lpQWvvfYaDh06hLGxMXR0dAhjr729HS6XCy+99JL8XtYGuaHANimJfcla9zmTyYjjZl8jYToa3lgshnQ6jWg0ivb2dsTjcal984HqC6ZZ3npqD6F3/UzqbWlEQZMCgW0Rm0QigWw2i9tuuw0ul0vE7RsaGjAxMSHtdNrI79ixQ/qaeYluueUW7Nu3D6Ojo3j66adx++23I5vNyl5oI6mds47gNSSt/y6VSpJZMQBlnQmA9FDqvmcaHv0cdBtfIBBANBoFAIHudWZWD0vXXTWKoOFdZrDA1vN573vfi4sXL2J0dBSzs7O4+eabsbCwgDvuuENm/RLR2L17N770pS9hcHBQjP/+/fvxEz/xE2hqasKtt96KP/7jP0alUhEpV54D3me2tTBborIg7YbZbJagjvKMDDqSyaR8DqvVKsxfDu/QWRNQTRDi11tbWzE+Pg6Hw1HlxOtlvZ2kKe+BZjKTS+Lz+XDrrbfi+eefx8mTJzE7O4vOzk5ps5yfn8fa2poIBHk8Hrz88svCJ6Iz3blzJywWC+6880488sgjVXtOH6X1v7UcK9ucAFQ5a23jKTKi7YIO4m0221sgfR0cal9htVqRTCZFJe9qfcJVK4nxoevDyQ9ITH9xcRGRSARGoxEdHR3IZrM4ePCgtDdR1o0XKJVK4dy5c3A6nZiensaf/dmf4bOf/axEv4QxNfmITloX8Rn98rIC1SIRTqdTDAahLkb6PAChUOgt0TSjHs1arIUs+R5Zr9IOup6UxOjEgG0xCkaTOpNqaGhAZ2cnjEYjnnnmGfT19SGdTguPIRQKyZg3OmJCgK2trUIUoYRqc3MzvvzlL+P3fu/3MDc3V0XO0OQ8vj7/X7d3MJDSxpRkMUJrnJKlfy9RGWZh3NfazJnBBycyEWlibbSeBGsAiHPTJQ4+Z+4BdfY3NjZw9uxZGRn4sY99DKOjo6hUthTicrmcSG2y1EG1Oc7afeaZZ/Diiy/i+PHjuPfee+Xe0xk3NDQIO5ddGDpb1kGlbv3R+0gbQ8SNY2YbGhoQiUTQ1taGXC5XRRwlmqJbLc1mM1ZWVuB2u6tq3vV0BnRgrlnrwPY4Uh04jY+PC+lufX0dR44cQSQSkTIQkUre13Q6DbPZjKWlJbS0tCCbzWJxcRHz8/OCajDT1XvLWjEREy1FqzNdDUFz37lP/LqW6eSoSZY+tR/Ubae0Z0bjliKiy+Wq+t3l8rba4A9b/081aEYLui5AY1coFLBz5044HA5MTExgx44dQurq7OyUVhOz2Yzu7m4kEgk0NjYiEAjAZDLhJ3/yJ/HLv/zL+NCHPiRKTqFQCOfOnZMBBfrBa5GUWui1UqlI/c9o3JqWwgEN2sFubm4KFBIIBGSkGWF29sfyoesHzc2gI29qaqqqU+spPvWwaqFt/cz5bHgh5ubmkE6ncf311yMej8Plcok4BAOd1dVVeL1eYWdfd911OHr0qIjff+xjH8OlS5fwC7/wC/iVX/kVUfrS7R00DrUELu6PHmDCOhTrWqVSCalUCk6ns4qB3tbWhnQ6je7ubiwsLIhetybBEd7jv6tUtvShFxYWJDonuUgbr3pY/Hx0bpooyO8ROiaqls1msXPnTrS1teF73/seTCYTuru7RewD2LpHFJxYXFzE0NAQNjc30dfXh8HBQRw/fhyPPPIIHn300bfAySxhsC5MZ6idKA2qbpHSe0pnzqyXY27b29tltKjT6ZTzpxFEXY83mUzCWwC2SZW1et3v9MUSp2bvMyji87BarbBarejo6MDU1BR27twJp9OJV155BU1NTejs7ITT6ZRgix00vD/XX3892tvbZaARANx///04fvx41esDeMsZ0I5Zd1jQOfPc0HYA22ebCJDdbkexWKyy7ZovpcmQ+gwQ5aWErS61XimSctUZtO5x0xeT0RQAJBIJgXpPnTolyi+c07t79240NzdjcXER6+vr2L17N0wmE3bs2IGvfvWr2Lt3rzSO79mzB88995yIiZAAxofC19TTs3QrEKFqfk8zLZkx8fKaTCa0trbCbDZjenpaJu7oqJcBCWsPNLoklenfzcNwpfWGd8LiBWDgoaFjGiu/34+VlRWBl9haQPGIwcFB2Gw2tLS0VMl3rq2tYWRkBNlsFrt27cKDDz6I48ePIxQKSesbI1Rt6HSmqyFMYCsrp/oUp4/F43EplVAGVJ9t7jmV6igfqLMrZmf8ee6/HjlIVjpr0PW0tHOshfVoXF0ul0wPam1thdvtxgsvvICFhQWkUil0d3cLpMxMaHNzUxQDKUa0b98+PPnkk9jc3MSP/diP4Zvf/GYVlKydLtE89sXSObtcLqysrIjdoDayHk8bDAaFK8MseH5+Hl6vVwy15llwj3V5gwTDWCwmsCZbT+sJSQNQlQXyDuhyB78/OjqKtbU1uN1uNDU14ezZs2JHyHrO5/NVQfzMzAxuuOEG6ebJ5XIYGRnB2toanE4njh07VlVO0X6Jd5CQNANxDujQGh1MqlgmYTlDE4EzmYy0e7JGrZ0/P69GEw2G7alcfH3ymLQj/2Hrqq0GYStuiiZMuVwutLa2SkTT0NCA7u5uGboNbOtaJ5NJLCwsYHl5Gel0WrKplZUVFItFfOITn5C+RJPJJG03lN/U9HbN9tWMSpPJhNOnT2PPnj3C8qURLZfLogREmJ0tHrVtXZqhywPBKIuGgvXu2oev2c31shoaGqrY0DSwfB5zc3MIhULCgmXr1AMPPICJiQlEo1Gp542PjwuEGAgEqhAYi8WCRCKBRx55BP/hP/wH2VNCTcB2HVlDlLz8fG3WNLn/7GUFgPn5eezatUuiWzL8SXBj3zOhLl1zrkVrSEbiGefv1N0E9bLIuq8td7GEwXF9e/bsQTQaFRjSarXi8OHD6OnpwezsLObn5+XnGaTdeOONeOSRR/DhD38Yly9fxhNPPIEf+ZEfwZkzZ3DDDTcgHA7LawPb9T8G6YQbme1wGlIoFBIYvLGxES6XS0bUsuRG482SBGfGJ5NJuN3uqpYanntdi6RhB7ZHDRL61JlbPSwGQRwpqkli5Cr5/X7s3r1bni+zT7vdDqfTKfMPDAYDPB6P+Jf+/n58/etfx4EDBxAOhzE8PCyCVf39/TIEBahWMyuVSlWQNiHvcrksCA5VITlZMZfLyc+tr6/LHhcKBSnRsM2Ldp6vx9fnGdAQOcukGq3Vjv1K1lU7aB1FcvHFC4WCZMkA4PF4MDU1VQUhc7iB2WzG+Pg4fvZnfxZ/+qd/ij179uDUqVOS/o+NjWH//v2SybJNg/A0sO08mdHwv/lz4XAY3d3dOH/+PFZXVzEzMwOPx1PFzNMqL4VCAblcTqBLRr36oeoavD6U7Jvj4dSQeD0tXfutrT3qZ0F4J5FIwGQy4bbbbsOXv/xlHDhwADabDblcDtlsFsDWrPDJyUkh1w0NDaG1tRXf+MY30NPTg7a2NkxPT8vrkaGv4S1G0QwM+b5oiI1GI4LBIGZmZuSs8D1ms1kRMeD3eDaoIEdURsO6fB7AW4V6tLHSta56W3SA2lgB25+btfj5+Xnk83nccccd+LVf+zVYrVY4nU5pp2PmYrFYkMvl8O///b/Hd7/7XayursLpdCKZTOK//bf/hscffxxtbW2CounyAuuXPBuaK8Dfz86NsbEx5PN56WXeu3cvJicnBYUxGo0SuLe1tUlvPj+zLu1oiJsBIpnIuoapz009LG2Ha88A7UQ0GhUFt3Q6jYaGBgwMDOD111+XtlifzyfcjtXVVUngjhw5gu9///s4ceIEenp6UCwW8ZGPfAQvv/wyXC4XyuVy1dQo3kFC3NoO8O+ZmRmUSiUsLCzg/Pnz8p7K5TI8Ho+MvAS2A79KZUumenNzs0pLWwcBtEU6WK3tduLSzv2HratusyJlXW+QhvR4UBsaGnD58mWsra3hlltuwblz56QpvFQqoa+vD4FAAGNjY7jzzjtx7Ngx3HTTTSLusL6+jtOnT2NtbU0+IEUjdD2Qr0+4na9NiM1gMKCzs1MgCj5EzvgslUrC6GZA0d7ejt7eXoE2uXSNgpdNG+NaiEdD4fW0iBLQOOrF58vWgubmZrhcLjz11FM4dOgQYrEY4vE4gC2GdG9vL5LJJNbX11EoFHDzzTcjlUrhb/7mb/DBD34Qv/qrv4oHHnhADBslW9kywWfNi0F4C9geMUnyETV26dSpIKWjXpPJhHQ6DafTKVC9dsY8X7rODaAqQCEprPZS1lOpQ8ObhG41/KsN08bGBoaGhhAIBPDFL34RX/jCF7C8vCxDDWKxGOx2O3K5HDweD5qbm3Hs2DG4XC44nU7cfPPNuHz5Mv7hH/4BgUBAWuz06FhN0OK+19ok1sGNRqPs++bmpjDImTmzTqiRIZJVWa6o7RzQ94Ea7rX17noL1hmEvB10y2CWf2ezWTidTjQ1NeG73/0u3ve+9yEajUp9ulwuS7nB7/ejq6sL//iP/win04m77rpLOj6effZZmdnA19SoBAM32noGRyaTSV6rUCjA7/djYGBAkC09QwLYRnpJEC4UCqIop+06fzfPoQ7Aas+n9gP/IixuXjpGO7XwIj8YD/HQ0BB6enqwtLSExsZGFAoFzMzMSOTc3t6O2dlZ5PN5ZLNZdHd347rrrsP09DQaGxuxuLhYJfdmMpnQ0tIil0fXg/nffG1GsTabDd3d3TJyjg9YM4rJ1iyXy8L+I9xSG/1UKhWBWjURQD8L7bCvBs54JyweSMK3XBrmM5lM6OzsxPnz52EymTAzM4MDBw5UwcgGg0Eg6Ndeew0f/ehH0draimKxiGeffRZHjhzBH/3RH+FLX/oSxsbG5Nmura1JVlYbCHGxhkwG9draGmZnZ7G5uYl4PC5a2lSq04xjGmkShXhmdN1JR8x8Pd6B2vpUrUOvl0XDRiNVW9bh52aQvLa2homJCfzUT/0Ujh8/LhlXLBaTu7u8vFw1+/mll17Cz/zMz+BXfuVX8Eu/9EuYn5+XkZPl8tYQDS27WPv6vL800sViUbS/eQYNBgOGhoZw+fLlqsyGtqBQKGBzc1MEM3R2xfetESTe+aamJsmemUDw99bLqlQqMszinwrWeR8cDodMjLvttttw7tw5uXPpdBqFQkG6bBwOB0ZGRrB37175+VdeeQVtbW1YWFiQbgqWKt6uhVGfATpOTWJsbW2tagd0u92IxWJynoGtPeZMaJb1dNbMz6jvurb5/P06UOH6F4O49cXUB48GK5FIwGq1ykM3m81YXV2Fx+ORAQUTExM4d+4cDh48iFdffRU+nw833XQTNjc38frrr0vEfPjwYayurqJSqYiCi9/vr2Lm8WEyAmptbZWRgc3NzVJLHBoaQm9vrzhxYLs1qlLZ6nVk+xcNs+6P07VNHZVzYwiJaziFG1FPGTQPsB6tqdssKpUK2tvb8eKLL+Kee+6RWr/L5UI+n4ff7xfHzIh5z549mJychM1mw9GjR9He3o6NjQ3s2bMHFy9erJLq1PW92nowI3XWMzOZDFZXV6UnOZvNorW1Ffv27cPFixdx3XXXiTwrYdBcLleV7eqWIQ1Vcn+JoOiyiIbdNVeinha7FvQ90WffaNxiwh89elSEiIiU5PN5GYpjMBik5PHQQw/h0UcfxcDAAIaHh/H5z38ev/3bv43nn38ef/mXf4mBgQGp67H1jfuhX9doNCKZTIpN4msaDAYEg0EpsXV2duLs2bMIhUIiRMFWSd33ns1mq9Ayfb9pkLnPzJp0YMC9rzXS7/TFevLGxsZb2sfooOx2O5aXl9HS0gKHw4FoNCrPxul0iiRysVjE2toaOjo68NJLL6Grqwuzs7P4wAc+gL/+67/Gz/3cz+H48ePo7++XKYZ8HWC7zKp9Aksa5AE0NTXB7/djaGgITqcTfX196OrqElidULwWNGKNnTac57w2MOOd13f97QJ3/o4rXVdVFCMMwTdOY8TNKhaL2LNnD0ZGRjAwMCDGkTq8+/fvx+Tk5NYLNzRgdnYWH/3oR1EsFrF79248//zz6OjoEJ3mcDgsxC32wZH0w42gcTCbzQiHw5ibmxOVsNnZWcRiMYRCIaTTaRlaMDExIcQu9u9ysMb6+jp27dqFSCRS1euonbSOiEgkYcuXDh40xFUvi8ZJ9z8C2/Cv2+3G66+/jsOHD+PVV1+F1+sFAKysrCAUCmFjYwNjY2Po7+9HoVDA8vIy7rzzTuEfzM/P43Of+xw++clP4ud+7uekJsSMa2NjA83Nzcjn81U9sNwXt9st748McqfTiUKhIDyByclJ6blMJBJwu92SdTNq3rlzp8hR0gBrciCwzRjlWeTz0FwI7aTrKXsCtvtgyVTVtUen04lHHnkEv/Vbv4UvfelLOHz4MPbt24fx8XH09fUhGo0iFovJyEmbzSZymgBw5MgRXLx4EV1dXfiLv/gLOBwO0c/XXRcUtSAJjQa7o6NDWvKMRqNwDBYWFlAqlVAoFKS8du7cOXG+Ho9HCKNkHhPaJuKm95yIIo06sG2Mmdlpw1xvZ4DlBH52DRE3NTXJSM9XX30Vra2t8Pl80hscj8eFyZ1IJASW7uvrg9vtxv79+/HKK6/g4MGD+O53vytdPCSB8c6Wy2W56/QHZE2fP39eIHgAyOVywtJua2sTBGV8fFzuqeYn6QRLO+daREw751oyce2+16Ky/9y6qrCesJDOXvhCLS0t6OjowOLiogyjIAOSbQ6FQgGRSAR79+6F2+3GzMyMEHJef/11BINBtLS0CKsuGo3CbreLFJvWVObB5x+r1YpQKASPxwObzYaGhgYMDQ3hvvvuw9DQEPbu3Yvm5uaq6Ti5XE4m21BdrFQqCXFFE6A0lKqjY5JTgOq6An9GQ971sOigeJk0atDY2IiZmRl0dnbCZrOhVCqhs7MTb775Jrq6unD58mUEAoGqKPP666/HzMwMACASieBzn/scHnroIfzGb/yGjCcFts/ZxsaGBEO106GYZevWu0wmg8uXL2N+fl6Y+SdPnsTtt9+OTCYDt9stQ03Y69jU1CQtXbWICLCNIulzqOuvGu4jlM5nVy+LgYsWdeHnK5fLGBkZwS/90i/hueeewz333CNlpra2NoyNjcHlcqG3t1cY8263GxcuXEBbWxtOnz6NQCCAr33ta7jhhhuk55T3ny122WxW1Ol4FnkmcrkcHA6HEJF4f71eLwYHByUopx0g0ka1Kdas2YbH4JvGVzsizRrXNXidZelyQL0s+gCiGvxsRuOW+E80GsWHPvQhnDx5EkNDQ4KUlMtlRCIRNDU1CTJKKc5kMins7tnZWSEhsg2KCBftEGd2E8lpbGyU9imTyYSuri54vV44nU54vV7s3bsX7373u/HQQw/BZrPB7XYjlUrJHtKuMeBjZq3Rs9qyTm0Wz6/VljX4tybU/bB1VaeFWQCzF70huVwOy8vL8Hg8OHv2LIrFIrq7u6WXlJf58OHDGB0dRTKZxK233oq+vj54PB7EYjEYjUaMjo5Ki40e8QUAqVRK+qm5EXw4vLysj29ubuLy5csYGxvD/Py8wBeEWEqlElpaWmA2mxGLxSRCJgmOm82LReIZDQCzBzptvUn8f035r5e1ubkJn88nRpGZE52Ww+GQyBXY0lEfHBwU8YnV1VWBNxsbGzE+Po719XW0tbWhra0NDz30EP7H//gfmJqakvIDgzNd2yWaop2ehpeZVXk8HrS3t4sQzsjICG6//XbR2tYtg5qtTfUrTT7U0JU++xq61MxhRvMclVhPSwdluocc2B4UMjY2hvX1dSwuLuI73/kObrjhBqysrKC7uxvj4+NobW2VTDgWi4kxve666/AXf/EX+NjHPoZTp05JEMw7SO6H0+lENpuVKWG1ATKdKYOucnmr1WZ0dBRLS0s4fPgwJiYm0NTUJGI17J1tbGxEPp+vGpCiERLNudCBiUZcAMhz0boB9bIYnGh0iJ+bUsuXLl2C1+tFOp3G3Nwcuru7YTKZ4PV6BZkAIOTfUCiE5uZmCbzY4cF9JK8E2HruvLMcUkH7zMxalxmKxSLm5uYwMjKC06dPo1zeYm5TgIbdPXT8REZ5t2uZ6rW2n/apNojj/5NwRhT4ip7x1WwIMxc6Kn0Q9ZCKoaEhFAoF/OAHPwAARKNREaYYGhoCAMTjcZjNZrz22mu46aabZNQY4QXKO2pWYLFYlAb0VCqF/v7+qtogo6CGhgasra2hp6dH5oqWSiVMTk4KPMWDEYvFqua+UhBFEzxohHVGTcNCJ0JIhe8V2HbW9QRxU9ec/w2gynnxIjALyeVyYuzo9OhYjcat1ieLxYJ4PI5HH30Uv/M7v4PZ2Vk5zJoZrfdY15z1tBuiPFSbW1tbE2i0oaFB5kxzL/leuJeakcoMXncr6PPIrzEY4BlhZs0WPnYi1BOSYjQaRdWvlsGrYV8G2qFQCOFwWDJS1nk9Ho88O042On36NH7u534Oo6OjMkmMvxPYVn8iactgMEg/M4M0XYbiPjc2NsLhcKC1tRVDQ0M4deqUwJ0MApmJl8tluN3uqvqizpzoiGv5KSQl0k4QYaLDqKdAjbZUa6HrgF0jiBsbG3C5XIjFYnIf3G63OGG32418Po9UKiVymsFgEIuLizJRjPvLfWC5g46QmbOGtHXgaDAYJGC32WzweDyYnp4WxI3KYRqh02Mla7lFtVwko9EoZ4cBAwMKInv0T1fqE67qtFBFRTOm+QB0RLC2tobW1lb5ObKkFxcXceLECdx8881wOp148cUXAQDHjx+XWiUF63nxisWitESxLsQHMTo6is7OTjHKNMJNTU1CxdcTrkjVL5VKAm3RkNNIM6uuVLaUjUgW0Aacl5Zjx/gaPKAaDiGTtV6W0WhEKpUSAXptmDVqQIk78gNItmFGCWxLsVKh5/bbb5expLUENGa2JIyRYMYsuFLZGmmYTqeFFKTnL9NJ79q1S1qxaFRr61Y0LMBWZE8Hwc+ro2YGDTabrUqByGAwVPXJr6ys4IUXXvjX2aR/hcXnoEmTul5HhIXjQzn+s6mpCYlEQoJmio5QRaxSqeD222/HCy+8UAVX0tbQgJJPAEBg7Hg8/pbRhHyfdrsdAIT5HQwGq/qm9V3nZ6L9MJlMYoM0GxjYhjV5DqjjoMthPBM87/W0dN+xds58RrzLVqsVmUxGgleWqjY2NpBIJKTHvKmpSRIAZs6akKgVwjRaCmz5oGw2W0X+ZS2cJDUAUrPmv9NOl3ac6G3tedBkwVrUgO+vFvLn67DUks/n5T78sHVVDprRIsXCazPojY0NrK6uwmg0Smba0LA1ZCCbzWJwcBBnzpxBMpmExWLBHXfcgc7OThgMBnzxi1/EwYMHpce6XC5LO1W5XBZFF26WxWLB/Pw83njjDWnVcDgcVVC01WoVKDSZTKKpqUlY3tRoZlbHgQmNjY1IJpNSC7darWIAuGF8f4z8KRdXC3nZbDb09/cLQ7Qe1sbGBlpaWgQB0fAOoa5PfvKTGB0dlX5nZicAJCtZXFyUCwpALh6dOR2nNpoaNs7n81hbW4PFYkFHRwdcLhd27NiBQCCAlpYWceD8eZvNhkAggJ07d8pF4fvW082IDGlSIrCN4GiCCr8OQKBQGiY+i3g8jueeew4XL17E7bff/q+3Uf/Ci05SK+7ROXP/Dh48iEwmg4sXLyIUCqFYLIpjpVGnoiDvKhEYtlWazWaZ5QtA7iFJf2tra3JOfD6fBElE27T9ICnUbDYjGo3KXdbOlQ6X+05iIrNEnTXr8hs/M+0UEQQ6iWw2i6mpKYRCoX/NbfoXXQzIdLurRhlZ3hweHsbk5CQCgYCMH9X+Ix6PI5FIIBgMYn19HYlEAgsLC6JBoBnVtWVMOnBgW/86l8tVcVGYOLCsyYSNQT+TCABSHmMATwSA/oUBHfdf8wu0g+c51R0mxWIRxWIR9957L+6+++4resZXDXHreiofMgARGaBMo8PhkLGSPKRWqxX79+/H9773PezevRvpdBo9PT04fvw4nnzySXzmM5+RyJZQKACBS6lrTOi0t7cXXq9X6hPMnlizKpVKAqE1NzfjG9/4Br75zW/C4/Ggo6NDNpCEkEqlgkwmg2w2i3w+L1N2eJCsVivcbrfAreyRZESoDywhlmeeeQajo6NX85j/zS9mrISHaZhNJhPuuOMOPPjgg3j/+98vbXZk8pvNZszNzQlho7e3F6FQSPoLvV6vRLmMWPXMYSq78dDzYrG2GI1GpW1jc3NTWu9YU6YRbW1tRSAQkBoVHYaW5qQMLOeM85IRviJ7lcZBK43RWGWzWYyNjaG7uxvvec97BHWoh0W2O2F8zVQ1GAzw+/34sz/7M9x+++3YuXOnSH96PB5RFnM4HOjq6kJ7ezvOnDkDv98Pn8+HhoYGeDwerK2tSebKPeJ90waaQSIZ3WazGS0tLbJfPDuaocszkc1mpa+Wr6WzoGw2C4fDUdWzy9/J88/3UyqVhCej69RMEO6++26ZhV4PSwvC0GHpGnwkEkE2m8UDDzyAnp4epFIp4ZIUi0XEYjG0trbC5XLB7/fjtddeg8fjwZ49e+D1ekXDn3ZftzLx9TXszDtMpJN+g4EVkyiiLZcuXcLy8jKSySQKhYKUxHTSyX2kf0mn0yJgRZtAlIW8JCaRuvRHnsaOHTvw+OOP4xvf+MYVPeOrLohovB7YZnNSs/jMmTMwm81YWFgQMlAgEMD09DTOnz8Pm82Gnp4eTE9PY2JiAqOjo+jt7cWFCxfwB3/wB/jiF7+IgYEB6WEDtok4VITiJpOlabFYhIHLn6UYPttpHA4Hfvqnfxof//jH8eSTT+J3fud3cODAAdEO5wOmIAGh8mQyiVgshmg0ikwmI0xPq9UKo9GI1tZW2RTCN42NjbDZbBgdHRU4vV4WHSI/l641NTY24u/+7u/wp3/6p/jGN76B66+/HsVisWoQOtspmpqasLKygqNHj4rcXkdHB86cOSPlB63GxDIGa740guvr61X6ubxUdL401mytofSgw+GA3++H1WqVepFuGTMYDHA4HPJ7SUhhBsagrKWlRRw2o2a+r+PHj6O3txeHDx9GPp+Xfux6WKwB0zATijYYDMjn81hYWMCP//iP4wtf+AKsVqs4cm1UKeM5MzODPXv2IB6PywCdeDyO9vZ2Cba18AzJaZVKRcpnwJYt0vPFa4ls5XJZHKzJZEJHRwcOHDiAjo4OhMPht5DMaGSZqWsCKDN3OgQKk/B88I5sbGxgamoKDz/8MM6ePXvF0OY7YTHoYqlD6xREo1Hs378fDocDjz32GIxGozhMQsZsrbNYLIIwxeNxxONxTExMoKurCzt27BCHB0DuNH8fv8b9sVqt8Hg8cj+5+DP5fF409m+++WbceuutaGhoQCqVQjgclnIlsI2SMslkFs6hO5lMRnxRQ0ODdA95PJ6q4JCaDHfddReOHz+ORCIhz+mHraty0JlMBul0umpsFx8YAExMTODIkSO4cOECbDYbrFYrXnjhBXR2dmLHjh04fvw4hoaG4PP50NTUBJ/Ph+npadHaffnll/E3f/M3eOyxx9Dc3PwWkQhGvpQF5PeAbSfOqE5nXYxiWGM+cuQIfv7nfx7/6T/9J8zOzmL//v1IpVIAthwBG9fpaAmzZLNZrK6uSmadTqcxPz9fJaBeqWz1Ur700ktSW62n9ho6QKIENIKVypaq0Lvf/W68+OKLcDqdWF1dFWnGQCCAy5cvY319HR6PBxsbGxgfH8fOnTuFTXnx4kW4XC40NzfD5/NJVkrSnoaqGNVqWU3ut+7P1+1ZjMBLpZIM7Ojs7ITVahWCItEZvi6hMhpd/n5qfPN8cVwlsHUvhoeH8a53vQu7d+/G7OxsXREFgS1NfnJCSPwiEcrhcKC/vx9f/epX8Zu/+ZsYGxsTCDqbzaJUKqG9vR2rq6toamrC6uoqOjo65J6Fw2H09PQIq5bGlsgasyjC4TwDWtNAt0UxiaAj16RSkpduv/12bGxsIBwOi/FMJpMSvPFr7Pmmk2A9NBqNwmg0igPmeZ2bm8PP/uzP4tFHH5VkoV4Wn3E8HpcSFu8m52a//PLL+MhHPiJDkdjqykSGU78ymYy0vxaLRen8icfj0kYFQBAJJmE8Gw0NDULqZOmMfBVdmqBP0W1we/bswQ033AC73Y7p6WmkUilxsBSp4b9pbGwUxI/aGQwqlpaWkMvlZHoiz0ChUMB73vMe/OVf/qWgOP8iDhoA0um0GD9+YGYwDzzwAJ577jns3LkT+Xwebrcbn/vc5zAzM4PNzU3s27cPR48eRT6fR1NTEywWC/bv349CoYDV1VU89NBD+PSnPw273Y4bb7xR4GrNriWUtry8jNbWVmmv4nvRTGpmXYzYtZG2WCz4/d//ffzjP/4jzp49i46ODgAQmJS1bMLn3JiWlhbJ4nlZmQGSWHbs2DGBaDRZoB6W/iyaQcnvkQTkcDjQ3Nwsfa4vvPCCcAxeeOEFeL1eeL1ecaYkEwUCARmkwvKCbnUgOZHkME1Ys9vtks3yMrGGSGOts6RKpYJoNCqtHel0uqqdgxe7sbHxbYlidOZdXV2IRqNVMGc6nUZvby+mp6dlvng9LYfDIYaGATsNHsVoPvWpT+HcuXOwWq0ybMJoNKKrqwuJRAIWiwWnTp3Cww8/jOHhYbmzXV1dwluhohdLSsxgGRyS4ctMjszbbDZb1a1BG1Dbm0oINJfL4e6774bNZhPJR6IrfN3aTJ7kNt7xQCAg54Dvt6enB4888giam5tFMKdeltPpFBtIZIR/1tfXMTU1hQ9+8IO4cOGC9DJ3dnbiwIEDCIVCmJqaQiAQwKVLl3DDDTfg4sWLKJVKIsVKrtLGxoZA0FyarKWTM91FQofNWjDtAs+T7qowGo3YtWsXhoaGsLy8jHw+L9k+kR/9u8xmM6xWaxVqS/SEWvF08m1tbfjrv/5ruN1uaQu70rngV+U5yLQjTZ4OkB8wHo/j5ptvRjKZhMPhwMrKCk6ePClN66lUCplMBvfddx++//3vS18aiTvf+c538Fd/9VfYsWMHxsbGAEDIO3QMvOQsurO1iw5c98EC2xA8afrcUBqTz3zmM/if//N/IhgMVrXlANskCL42Lzjh783NTSE28PDk83nJ4AmF19NilslnD1SrB1F9i0phhUIBlUoFfX19iEQiALbESVZWVqomkxUKBWxsbMDn8+HcuXM4dOiQQOjMdmrbebgvHLpAhiz3V0fPte0QmonLiUUkC9I48HMxK+B/A9tD34nK8MyUSiWsrKyIvjOHLGQymbo6C3R+/KPFZ/i9+fl5dHV1yR4sLCzA5/NheHhY5nEHg0GcPHlSDFYsFoPZbIbdbofVasXu3burhCJ0gKjPn91uRzKZhM/nkzKUPjvcdxpmlqS0o15eXsa73vUuGAxbffacrFXb402bouvPhNFdLpeUy4rFosw7ZwBYTzwE9vfzD20vV39/PzKZjEyrWltbQzgcRiwWE8Lc7OwsPB4P5ubm0NTUJMma3++X1yDyooMyYHsuACF2AJJ9M2DjHWXJgr+DSJwmfJbLZQQCAezevRtLS0tSt9btY7rWDkCCNE0SJo8G2NYE1x0ndrv9X6bNin2EnHkLbDutUqmEnp4ePPnkk0LuIj5vs9kECrjjjjvw+OOP46abbkI2m0UkEhFYYNeuXfj+97+PtrY2ZDKZKpKQxWKRupXD4UAoFMLS0hJMpq1JM7zEmkBCCF6TtmiYaWANBgN+7/d+D5/4xCdw6623ore3VyImXkAeBm4wMyoaANZWzGazZExNTU1VBJp6WXymuo6iiRqVSqVKnYvIhsPhwPLyMsxmM1KpFOx2OyKRiFwichXOnDmDjo4OeL1ekf9kxqIFEbjnRDcIRRGC597SkHKvtIHV7VLUeV9eXkYwGJRaJQkgetHxW61W9Pf3Sz0S2CJL5nI5dHZ2wu12AwAWFxcRDAblXtTDIoLEoFjXCMlJIXejqakJPT09ACCQMTki2olms1npez579iy6uroQi8WqHCEAER0qlbamEDmdTgl+otGoBA3kLOh/r0lNtatSqWBlZQX79+9HpVIRQ8qgTbPUNSmOXJi5uTn4/X5xzu3t7Th37hza2trkrNYTH8VoNIrOBJ2kbjHN5XJIJBLiB5xOp8DCrNmzvs/9os0nasF+ed5lZueaMKb3ii2WtBt0xvy3ul9bB/tcLLft3r0bKysrVU681g5ocpzNZkOlUkEymZQgrFQqwWazYXx8HHa7Hevr6xLAXaktuCrPQWIUoxIuZouvvvqqwEz8GokcHFzByInRpsvlQrFYlPab1dVV3HHHHcjlcnLxdG+j2WwWY0zdZcoy6r40/tHRr37Y2qEYjUb8/M//PP7hH/4BbW1tVRmYDkC4qXS6HKfIz8sGe2Z6AKqCg3pYhHHZt1p7aBsaGjA5OYl0Oi0OGIAYW7a7RCIR2T/CUgzSCCU+8MADwqTVJQzWPv1+v5BD+N600ps+E5phqj8Lf47cAYfDIcQjo9EoxoA/z2UymbBjxw6Ew2GJ7kulLa3nwcFBkQ9Mp9MiyFBPi/eBRB8+Gx0QAVvoBpmydOR+v19QqEQiAZvNBrPZjLa2Nqyvr2NmZgY33XQTUqkUhoeH8aEPfQj5fF5gRwBVXANC4yyB1Xab0GYxs9esc2286cgtFotk4iQ4sr6uMyjaCaoRkltTqVREhUwze51OZ12VOvjZiSYC23eKLH92VVADmzwA6k0QfWS5pLW1VUpbHo9HSHj33Xcf8vl8VXAGbAuBkIFPVFfL6+rWLM2qfjubwH/DEg5tDs+aZozzZ8noZ1BWmwxSxY7w+r8YxM2DzeiGD4hvuqOjA62trVWXNpvNwu12I5PJwOv1Yn19HbFYTEZQrqysIJlMwm63Ix6PY3V1FS+88AIuXrwov7epqUlgbF6oxsat8ZV0/nS8vHQacqFz1w9NZ7WlUgm7d+/GX/3VX6FUKiEYDAo7Gdie98o/AMTIkPFtMpmwvLxclXXVBjL1sDS8pCFnHvJCoYBdu3YJrGcymdDb24tvfetb6OnpEUEP9rtTMW737t0AgJGREXR1deGJJ57A2NiYSITqs+BwOJDP5xGJRIRFzUXYi73wAKrOBlA9GhPYhq3JymW3gD7n/HcApE7NPmmeNQos0PjwQrMeVjvx5528aJgYiOtshH3K+/btQzqdRnd3N8rlMvr6+vD444+js7MTJtPW3G2v14tkMolsNouNjQ309fXJ0BRyEb73ve8hl8vJLHmyqylkQrYtgKrxg/yj7zvfM8+GNqa838lkUoI/okG1Bpln3mw2Y35+Hm1tbXKmi8UiQqEQZmZm5PXtdnsVj6EeFoMPjSwwiKGz7OrqQjabFXlgp9OJS5cuSZsjS5gMaAFIuYJjiC9cuICxsTEYDAYZA6y7AtLptAjN0BHWqs5psqBOnmoDef4bBuAkoJGAqmve/N0kprW1tYl9rFQqaGlpwcrKigQsJMDRR13JuioHzUOuoWd+GIPBgGg0KoxbsqBJoHG73bDZbDLBpLGxEZlMRgZ3O51O7NmzB7/wC7+AZ599Ft3d3Th//jyi0ajUd4xGo/SlZrNZeDyet/Q5amKIJohpGJYwXK3QQFtbG/bv3181iEHXO8nKo5C6JipodRstZsL3US+L2Yp2fsA2atHS0iIiEySNnThxAocOHZLfsba2JpmPx+NBJBJBqVRCMplEW1ubzH7t6OgQUlEmk5Fnr8VCmpub5VKwnMH3xvegg0htTHQGza9RhIUZmjbuOtjjGeTrVCoVRCIRdHd3izoao3pC+PWEpJDZrgVB9N1zOBy4dOkS/H6/GNBXX30VDz/8sBgtj8cj94a/KxKJoLGxEe3t7RgeHsav/dqvIRqNorOzU8RpGPgUCgV4vV7Zb5Y3iPLUlrb4t86C2Cqm66c05MlkUgIQBtu1xEief9pDPpuWlhaZOU9boWcH1Mti6YnPj86RWev09DQ8Ho8EqNlsFtdddx3m5+eF0MeMmcptPEck2t555504c+YMTCaTdAHQ75hMJvh8PgnYAFSdAaA6qajlMfD96zIY/z2HpfBs89zVonRsFavV3NeyxrrcRrLZlayrctCsE/AhANv9kIQ9SYlnpkvKektLC5aXl+Hz+QTWjEajeOONN3D99dcjGo3i9OnTOHXqFA4fPoz3vve9uOOOO9De3o7l5WW5cMA2CSUajVbVeOlw+fo6s9PBhDbOOiv+8Ic/DL/fj1wuJ7CEFkPg72PdhXR/fk4tI8oN4gGol6Wdle6D5fOmGhjhIQBS27VarTKthhd6bW0NwWAQY2NjsNlscDgc2Ldvn5ypQqGA4eFheT2tDpZKpSRrAba14nUwprNnfk2fGX0hASAUCkk/I5cO5khSKZVKiMVi8jmYbWv4M5/Pw+v1IpfLSYRfL4vserbaaLRpc3MTqVQK7e3tYuAqlQq8Xq+cg/X1dWQyGXFsqVRKeoZJRvV6vThx4gT6+vokEKBaIEfD8jyQ5UsnqvvmgWpyH5euLfNnGYRZLBaZcqfhUJ4XwtsM1jS3wel0YmFhQZwVDXc6na4roiDtrIa4uRfsmPH7/UKcJcRPZMJgMMgZYM348uXLyGazcLlcCIfDCIfDiMfj+PCHP4zOzs6qAStETzTniHeM/639hvYF/JttesA2uqqdJxEZDeHz+0wKAVRB1kQXmajoVuRaNvoPW1edQfNN6ShJO+7V1VUZNUgntrq6Ki0LLpdLegX37t2LwcFBGQf48Y9/HI899hgWFhbQ0NCAy5cvI5FIwOPxyEHX2QszWj58wop82BrW0NEOH7S+aOVyGZ2dnTK3tJZMQONcLm/1PO7YsUMybf5+Hg5mcTq6rKfF6FI7PDppMtfj8TjcbjeWl5fF4LGFAoAIiExPT0s7RDabRTgcRrlcxs6dO6vELxobGwUx4cXU8BLRDIqg6LaKt4Pg9NnQjrVcLiMWiyEcDgvzn/+WP8PWolQqBYfDUeUUtNEIhUJIJBJVBr5eFo2M0bg9a5kIEveLJQO73S6znU0mk/APWlpaJAtm50VrayvK5TJWV1cRCoXQ0tKC+fl5mEymt4yb1MxpohaVyraKoCYx6kCMf3QNm0kDjTgH9lDSlnaCZ4VZFz+jy+WS7hYadLPZjEwmA6fTieXlZWH018vStpfoBYMS2oN0Oi16EOPj4+js7BS0jGeFSBQh8EqlgkKhgIaGBuzYsQPRaBRHjx5FIpHAyspK1XMke1rbfm2L+T74fmvtMfdc74v++WKxiEKhIP342h6Qp8Lfw7NL20O7R0a57uS40nLXVTlopupOp1MegHZioVBIIqfW1lZxorzMNptNYAy73Y6xsTH87d/+La677jocOXIE3/rWt/CLv/iLyGQyeOONN8T5k+3HbM1oNFa1W9Eg0NDzPfHh0QjToWu4i5vDB3/+/Hmsra3B6/XC5XKJtCchDmDLMLz22mtVQ8K1ohFhYLK966kObTAYpD2KmYeOKGmorVYrUqkU3G437Ha7iDSQGZnL5ZBMJnHddddJe5bL5YLNZkM2m8XLL7+M7u5uuXw2m03qjGTvtra2CjmDew5UD7TQJQ5gG+bShA8NdVcqFSwsLIhCWbFYlH5cCiMA20NdKMTCnk1yLQjVUmwDQF0NSmAmoMk4+rnyrtjtdqysrIi0L1nXq6urKJW25DYnJiawe/dumEwmYX37/X5cvnxZVJ+ovc36NssnxWIRfX19SKVSEswR2tYZMc8B/5+Gm2dFo2002oVCAR0dHVhbWxNkAIAQ3GgDnU4nlpaWqhTtyEyuVCoy195isYigRz0sOlEmaLWoFclWdN4DAwNIJBLC26Bj93q9yGaz8Pv9YqOpELmxsYEbb7xRYHLd9cFAYHNzU2r8DPQ0LE0+irZXvO9aAVIH0HS0drtdRtU2NzeLnju7R1gGy+fzmJ6ersq0WSPX6AlLM1e6rspBs3eRKl56mUwmUYhZX19HJBJBuVyWGoPZbEYymRTo0mw2o7u7G7fccgsee+wxnD59Gl1dXQiHwzJ9hFrX1M/lg9WqXxp+ZLuTftDM+nUdQJPI+Dlo0F988UU4HA7MzMwIO9zj8SAYDIqgxeTkpLCY+dn5XNhGxOi6npSDAMicVr3/OiNNp9PSh8hLlE6nxaDxudntdthsNszNzaGxsRG7du1CU1OTtKYMDAxgYWFBSBqEodiyMTIygsOHDyOVSkmfvO55pLHV6Ik2wjqK1u+fhoO64Dw/HIUXj8dRLBbh8XjgcrlQKBSErZrP5xEIBBAOh9Hf3y/SgRqSrZdFlTXt9GqzC529AMDy8rLAnA6HQwKg/v5+TE9Po6urC8FgELFYTCZT9fX1VQmREJJkeSWZTIoh1FwEXZbiewK2y2O6DY9L/6zRaBSIfn19XVTFmFnTAXV0dMhwDwAyW9rr9WJ2dhY9PT3C3NZ17npYTIb00raX/AJmxKVSSbTNabeJRFksFkxPTyMUCiGdTmN5eRmzs7PI5XKCSnHPWJpgqQvYVm6jw9Rohw7WdWuURtOAbTvA82M0GtHZ2YmpqSlEIhGRCWZCaLfbq4J0Ijz0S83NzYKy0SYRaaM9+2Hrqk5La2trVSaiI05uAuHtcrksWTQhB6/Xi0wmI1NwkskkbrzxRgBAIpHA9PQ0isWijGfUH5bN35QG1CMceQi4iTo6ZiTDDEvDHLpWycXhGoTZZmZmMDk5KZkfFbD4Gqw1srbGRnmdwdfTpSRKUcvc1RE0yYIkWTCj4KEnQZBsbpKpYrEYstksOjs7AWwFA3yO+lzNzMzgrrvukixMcwX0+eS/JeSk64Z8/9qZ8/2TcQpsj9Oz2WwCp5lMW+MJOWghHo8jl8shEAhIfY0lDgaHnHJUL4v9xkA1UxbYeo4+nw8XL16E3W6Hz+dDOBxGd3e3yKjymTK4aWjYHh+bSCTQ3NyMtrY2nD9/XsQgaKA5R3p+fh5HjhzB0aNHpabJjJgZPh0Bsyn+ASCJgh56op00e1cJ4+shLExGzpw5I10Gm5vb868nJibQ29uLS5cuwev1Sr9vPZU5ksmkPFv9DLk4EdBms4nugd/vl9Ypo9FYVY5wu92iwEc7UCwWMTExIagk1+bmJlwuF1KpFAYGBnDp0qWqHnMGc0ROifJx/3UrHn9Goz/0DSxTkHMRjUYRjUZF3tPhcODChQuiKqZ9ytLSEjweDxYWFmSKVm059Ietq/IckUgENpsNNptNjB2wfahbW1uRSCRgtVrh9Xolu2lpaRHhEeL5FBy5cOECfuInfgJGoxE+nw8rKyviIAllsLfY4XBgeHhY2jRqnTE3UbNKdVbFg8RDpWtJ3JwHH3wQ3/ve9+D1emUm6NraGubm5jA8PAy73S4GIhAIYGFhAdFoVFp/CMWzHsV2nXpZPOA8jBouYm2aU4XolBigED1pbm6uUmNrbGzEwsICHA5HFYOaMLLL5ZIs5vjx47j77rtht9sxPz9f1VZVW0tk0MCLzUxLnwcdZPDCsnVLZ9wGg0FIbCR+eTweHDt2DFNTU7jzzjvlrHJql9VqhcPhQCwWEyGDelkcr0l5Q82U5+K+ZbNZtLa2CgzN/mfuG+9sS0sLhoeHMTAwgJWVFQSDQXGSfB22NE5MTOBnfuZn8Nd//ddob28XfWRNWtMBOc8FsE1wa2hokM+gYU7+O86OZnBntVqF8NPU1ITZ2VkMDg5ibW0NiUQCu3btQqVSwcjIiMi8Op1OgTULhUKVytQ7fdF+6lGN+vlpEh+AKglWLTSkWdeZTEZQFbbocX/4ehSBSqfTuPPOO/H000+jvb1dSmAsLegA7e32l/eRZ0EHGDrpa21tFalSqhsyMD99+jQOHDggvKO9e/cin89jdnYWPp8PS0tLCIVCwqFgUHil5a6rrkFTIaoWImRWEYvFRBmsXC5XjQrjhpRKJezatUtkz06cOIFz585h9+7daGhoEMam0WgU4YhCoSARNuu9tbCFlvpkEzxr2DTa3Ai+f80A5+8aHR1FT0+PGHYSmhoaGoS8Ui6XcfHiRSn8p9PptzDYWSOrJ5ibrSe6fq+hQs2k5LPnJWTgRQfMQ2symeDxeKSnEdjOZgkPnzp1CqlUCj//8z+PlZUVLC0tVQVyQHXWrGvNvID8mnbKvMw68uX3EomEvF9GvwzQcrkcnnrqKTgcDtx3332CDBmNRiwsLKC5uRk2m01ajBht18viOEddz9UBSDablalklUpF7snm5qaw2tkrT3gyFouhra0N0WhUBunoIRkcSNDV1YUPfOAD+JM/+RN0dHQISY37rzNpkgeBbW6Ifs/MtDRTF9gW5GHATUPPz0wnHYlEsLKygrvuuguxWAwXL15EZ2cnFhYWJPMvl7emaPl8PsRisX/1vfqXWswKGaBpaJkoBu0wSXzUTqcd0TPf2RHE/me32435+fkqpjhLDLfeeis6Ozvxta99Db29vVWBmeYV1SaSvM/aXtHx66/TL1BghiUW1r11q9Tq6iosFgtuu+02jIyMYGpqCna7HcvLy+ITiBoQVbhSRbmr6vvQMIF+GFzT09Ow2+1C1ohEIlWsXX7w0dFR3HfffTh58iTC4TAefvhheDwePPXUU+jv75cPxIfETCkQCIgD1zUmnQ3zAvIy0klrAgMNCSM8bh4vuMfjqdpwHjrWU3fs2IHXX39dDqnBsDU3mNF6Op2uIi3osWfv9KUvJC8XsO3UMpmM9MID2wxvwsOa6LG2tiY1bYoVOJ3OKoPPWs4999yDjY0NnDlzRgwkDfvboSnaAOsAQkfO/De1jhmAcCd0eYIBV6VSwcTEBG644QYhuGxsbMDj8WBychI2m02ycEobOp1OCT7qYdXWd2szaKInyWRS7h9V1dbX16UmycAtkUigo6MD5fLWBCEiEcViUcojHR0d2NjYwNmzZzE5OYmenp6qSUfAtnIfg0PaokplW9hEE4V0LVTzFHStkk6WZ5q9/hx8cP/99+P06dNYWVmBx+MRSLS9vR2lUgmrq6sS2Hm93n/djfoXXAx2iapphILfJz+Ez1OXyLQCI0sAgUBA6v2U72R2SkGjqakpfP3rX0drayv6+vokc+a+AXiLXaCd0vaY54J+o1YZke+b741njYEatTiCwSCCwSBefvllrK6uSn+/zWYTchtJYhsbG0ImvaJnfDUbQvYsowl9IXXbBQ+6w+GQSIKzdYPBIMrlMnp7ezE8PIwPf/jDmJmZQTQarRJ5MBgMMm/TbrfLjM1aQgoPh2714fdrmb38m5C2zp60cT58+DBGR0eFIMRInBd8dHRUnA5btxgpxuNxYfvVGo96WGazGT6fTxj5et/Z26rnvvJAA9uDNkgkA7aj2lQqJZeH0anVaoXP50NDQwNWVlZk2Ab3j/vCZ1ybATHI4mXVdXJNGgSqJSqZKfM1NLGQkX5XVxc8Hg/S6TQKhQJcLhdmZ2el9u5yuURf+p/Tf36nroaGBuk51/V8ZqQej0eIUzTSzc3NouFN4mdHRwc2NzfhdrtRKBQQiUQQCoUAQDIulimi0Siy2SxCoVCV/raGs7UNIEuXyJrOqOgcavkhteQxo3FLb1rXKJn5A1vE2TfffBORSER4MalUSvp/K5WKzMMmklQvi3dKl434dba/cfQiAEHNWC7gvnZ2dqJQKMDj8Qi6Qua0LoEUCgVcunQJlUoFQ0ND8Hq9UkLjPeV7qCV90tED275AI28acdHJHJfdbq9ifJPh3dDQgIWFBRw/fhzxeFzkScvlsghzURucdeqr6eq5ogy6lglJw8Q63ebmJqLRKGZmZrB//36Ew2GR32SfKyE/Eju+8IUv4N577xVy2H333YdvfetbcDqdGBwclLousH1RtRgBM2MNsfKD640hiUt/Xw9U5wPXgy1aWlrw9NNP47777pP6MS9oU1MTMpmMZMgAYLFYMDc3Jw6dgQz/1r3X79SlA51IJIJCoSCtUzwHrBnrNhhGzRyQQTIJJ0b5/X6Ew2Ehi7ndbiGNWSwWkU8FqomJb7doGBh4kYXPTLuW3a8/F2vmOuOmYA0A6UJgIEYHQy3pyclJOBwOQVsKhQKi0Sjy+bwQnOx2e9VrvhMX33s2m0UikUB7e7uUshjAAMDMzAyArfPCAQLJZFJ633ft2oW5uTlpR2toaMDs7KxIf5K3QpiZQkC88wzKGWhxn5nNA9slj3w+X8U4ps0g01zXrPU5YHmlWCzKWeLgH+7p5uam8CYqlYqUwGKxmJT8GKCYzWYJWurhDMTjcWxsbEiwrvWuWQJk2YeiQmtra7BYLJifn4ff70c0GoXNZoPb7YbD4cD58+fR0dGBSCQitoJ3lSVRXS/WpVbef91SpZ0t0Vf97PV5IdKjg2kNmbOtTLfy0u7RVrDjwOv1ih/MZrMyCpeZdy0p8Z9ahsoVnJSFhQVh1l5b/+9rfn5e5k6/09a1M/D/v3XtHFxb187AtQX88HNwRQ66XC4jHA7DbrfXFUz3r7VIkgmFQu/YlqtrZ+D/93XtHFxb187AtQVc+Tm4Igd9bV1b19a1dW1dW9fWv+56Z4Zw19a1dW1dW9fWtVXn65qDvraurWvr2rq2rq1/g+uag762rq1r69q6tq6tf4PrmoO+tq6ta+vauraurX+D65qDvraurWvr2rq2rq1/g+uag762rq1r69q6tq6tf4PrmoO+tq6ta+vauraurX+D6/8DO488uGeA390AAAAASUVORK5CYII=",
+      "text/plain": [
+       "<Figure size 640x480 with 16 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import numpy as np\n",
+    "import matplotlib.pyplot as plt\n",
+    "from matplotlib.image import imread\n",
+    "\n",
+    "def filter_show(filters, nx=4, show_num=16):\n",
+    "    \"\"\"\n",
+    "    c.f. https://gist.github.com/aidiary/07d530d5e08011832b12#file-draw_weight-py\n",
+    "    \"\"\"\n",
+    "    FN, C, FH, FW = filters.shape\n",
+    "    ny = int(np.ceil(show_num / nx))\n",
+    "\n",
+    "    fig = plt.figure()\n",
+    "    fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)\n",
+    "\n",
+    "    for i in range(show_num):\n",
+    "        ax = fig.add_subplot(4, 4, i+1, xticks=[], yticks=[])\n",
+    "        ax.imshow(filters[i, 0], cmap=plt.cm.gray_r, interpolation='nearest')\n",
+    "\n",
+    "\n",
+    "network = SimpleConvNet(input_dim=(1,28,28), \n",
+    "                        conv_param = {'filter_num':30, 'filter_size':5, 'pad':0, 'stride':1},\n",
+    "                        hidden_size=100, output_size=10, weight_init_std=0.01)\n",
+    "\n",
+    "network.load_params(\"./params.pkl\")\n",
+    "\n",
+    "filter_show(network.params['W1'], 16)\n",
+    "\n",
+    "img = imread('./figures/lena_gray.png')\n",
+    "img = img.reshape(1, 1, *img.shape)\n",
+    "\n",
+    "fig = plt.figure()\n",
+    "\n",
+    "w_idx = 1\n",
+    "\n",
+    "for i in range(16):\n",
+    "    w = network.params['W1'][i]\n",
+    "    b = 0  # network.params['b1'][i]\n",
+    "\n",
+    "    w = w.reshape(1, *w.shape)\n",
+    "    #b = b.reshape(1, *b.shape)\n",
+    "    conv_layer = Convolution(w, b) \n",
+    "    out = conv_layer.forward(img)\n",
+    "    out = out.reshape(out.shape[2], out.shape[3])\n",
+    "    \n",
+    "    ax = fig.add_subplot(4, 4, i+1, xticks=[], yticks=[])\n",
+    "    ax.imshow(out, cmap=plt.cm.gray_r, interpolation='nearest')\n",
+    "\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Classic CNNs\n",
+    "* LeNet (1998), by Yann LeCun\n",
+    "* AlexNet (2012), by Alex Krizhevsky\n",
+    "* VGGNet (2014), by Karen Simonyan and Andrew Zisserman\n",
+    "* GoogLeNet (2014), by Christian Szegedy\n",
+    "* ResNet (2015), by Kaiming He\n",
+    "* DenseNet (2016), by Gao Huang\n",
+    "\n",
+    "The advantage of deepening the network by stacking small filters is that it can reduce the number of parameters and expand the receptive field (a local spatial area where changes are applied to neurons). In addition, by stacking layers, activation functions such as ReLU are sandwiched between convolutional layers, further improving the expressiveness of the network. This is because the \"nonlinear\" expressiveness based on activation functions is added to the network, and more complex things can be expressed by stacking nonlinear functions."
+   ]
   }
  ],
  "metadata": {
diff --git a/figures/dlscratch_cnnlayer.png b/figures/dlscratch_cnnlayer.png
new file mode 100644
index 0000000000000000000000000000000000000000..eb461e93af7da44f06783be15d8db3e52446acc2
GIT binary patch
literal 17642
zcmeHvXE@vK-?v`f?M1I@RZG!UQ53bRDB3DoYKy(9H6wP65ZY3TqP6#|5m{F3(N<A2
zRuY1sc7j+jLU_{ay6*eM^XhqC+{bah_@Bp-BguIrzjJ-h@A(~{&l#zwqrrNP>l_OU
z3+wY|PYqaDPWZC09D9A{6thN+v$2NxcFfyA;|U9@@A?w+%L&KF+K*XS%Hz%-ygbSL
ze%9lenKugyyU*{}F_HtHKMTv#kmpYy8~IzWOq}+=jQifcE_eIbA7P(9{qgk17MtDU
z!aL76Pa3(7pyy`HA20Ki+p1nxdnzHaed^YYk`F$xyfC)Blt=7Oo}4+xzOj<l^W@{P
zKVJJuuN|%TR5ny*Xpcw-_NDptWv+L7USxKFS*{m{LC^p5G2-UPvA;ig+<C|IZ`G~;
zk*b#KocqdC(rIaF#HPTSZR(^u`mz3*Gk+SHI+eT&4PB9KPKb}UOOlK-_SoAnzIOV=
zi4kC9Tm|jpFTHAJ$L<`S@tra>Gcz-<^D4X#BDtd93WY^P!1Y_hmCyS73J8jbi^ul&
zh4#dTWDzj{xUpDZjGtb0_5!>D%;oK2yMVINeD<u2kXsU#`lj4YO-=nN^ZD*nQWLQn
zlwmM5xGZ}|WV*<epFg_o*bSk3r=@&X$2+SRk2zR;jEyy38Z2CqB{^<uRf8r`v$L;x
z6nuVKpV!S@6^5KpT)Q@OF-Oild(e`n`s>$6xDx$V@&n@%y>G!Mbs_>a99zR#HFb4s
zeE+h*xGv$&$tWvZi<%%SRyAt7stpP?ORbUVw5!F~<sX7|3Va#&LGiFIH~?8%q@6*;
zLax|T%$o076n8d=aTO^&9=N9{zv^c9?AeDt5WWWv`mo2)O1?V~`|vH#oQ2QEYOOKM
zKutSC=4IHFWU4aGj!>PNnD`p~(&DaWKusGF;{pq2(6{-ilBd&pzd0CEuO1#Ais-*H
z(|UN*J;215#+Ok?yh6?%(kMHglP2k1WMF8>ap_V)dkohoaQdu+kmtAW?~Myb^Pquf
zqihG6h_kCNT1QYIiyYp!@<VBC3z>D^NEY;4xb*DO08b)v$G^cftNr6u=`7Dmv2<*a
zTb9DM^c|7F?d1WoEHH7AP72&)c{^H15A3@R&#@(69n!sbOe!jQiSMpi{V;JrYI;I|
z)24^bSh<KdTLF}g(M}dM={xs8{*2ky3#(?kwsZE5j)4r@_EG@6N?h4GI5@b}x=CSQ
zz}l9aZQq?@F)2aL&YwQYkB0KCrLT<D0B71`j9dM`?JbM}r-h67f4LwLv2{z0YHDhu
zSpV43YqD9n+bK4Eu`Lm4;r%DRtWyXUd0M^cvPzJ|$O;;-dkQik)%P`yjKk}1R;O6a
zCqY(RUpLPzJk_A&!dLEjJE!G#a2ISQHz*u948yKn^mop<U$=5KDs#`W_e|u9qOWOj
z<F)xM&~<yZk7*tg!);>Xjh}q6sv@<%tG%!e2a6XkG!29a3$FNpbor*fu#9<@Z*wa|
zvjW0YK+&osG<^%YkNE6CBC5zI3@C=_-d-Meo?s{Y$H~WoxWv3y%Zk(b^3wd;PWbt}
zY3xu+Z%wd&F-i}Y>%u#a7u#6*E)}{)TieFkW_)V<m@9d+Kq!sBc7Czx?25R{z;~ud
zOHG$7@u%vaUp3VKG&Y}>rBivS>kb>Wb~is7@mZRJM^+O2ldpYtO1;QmB|!4C-kdkK
zawi8F?#v8dT>dK5qpz|lQ1k~4RaSNn-a5aeld)f7gDV9fE&L<)WfBn&7sNR+6T-rH
zF?)&?>5hevWR}Vw>QuN%4GP>Kjm?zHwpAxzby!vMk87~@BW!;19UZe{KxhaH|7_S$
zZ;7N~N5+6m_B`9_gQh;ead1<^Y6tA*$Xne{Kk3Pe{_B@f@jmmnOdokpp<eZH$WAFa
zbaY&Qj$QNMp`sjn9&)eodUs`&r-&-^d``@Coseot@C%@9`kvN!{P;0mEGRV8L@m_z
zGz-fE1DHYKa^Tt(cOAD*r@K_UAAhhs;QYq+w~l<HdU5$5W%=Opf2#g76TfHRKQi&3
zt?>V(6taxUDjl)B0s;cU^74BK>mJ84QAz#4omJeo(9<irOx?pQUmFg*ESL9%WDZ(i
zLV!~&`xmpJE36pjA$5!kzKW1r)~=hH%d|DXtHYhv$K6vE-Lu&*UBZ>xYrPK-Cjz6b
zgMu23QRan$+UX?_jLWvUf9Rf@Y>ABx#sybFxGdI%_j_nntrH&~pKn&_Sjj{G>fUmM
zI>#Zhn14lVSyl~eg|!?-W;C55JyF`!T@Ai8mhT55Hkx^Qdh$}y^<(iLKAefeJ3o+Q
znhG5-)8d=*!p}#|Mz{I@?mY|j)`&bS#}ie+_nf%D{;t|R1^eNA=FFK-IXO9uGk;fo
zZ8*cUoqw0>QNL9#vwY2q0+(I7auG(?PAAt}Q1y-$tcP>+T4i4UI6}-}{8ai|b(P>n
zoR&^4Tz~&zYS{qsF6&cSlnbsof#}<0pyXb!CO7UI;a1Pku?n1^k$vZq$iuJ%1@Nlt
z1k<E@K2dK4%dkW{huB|`RnE~G{@eV0Q>j&c#`oa(A3mE((5?O%T{79aX}HNz7XzHM
z?(I`D!2k)yMum5nlUo|(QKot7RO#!^coCcC#_uBGFo)iB@;C;LqE1*qa{{bo$@t1i
zj(@t>V|@2rmMFT(TFe4tkRYf%T=r5`)~Wvyjf^)$reiMi@E|{Z`s-wpIH-kQlAC)2
zfHE3|fEaWduGaGvc#+r}eC6K<%@5gj^yCfAv}fSTwK#ETliy2e@PbAzT{UMKKc$O{
zB2iHvjYg&HXQ@>5a%)6VRu*W2*H_G8OXun9Fgv?axcZBaw9S|dnexehO#P{z`AS<M
zg`(s`>U3~@sm29hn2!Hb|5cc=fEKU(MAu!xetG}3T&8iH$dr@WiH4-~^irnz;j`F(
z^W>=$C9b0tc$oQ1RH5da2zi9zvj5xrVV2CgAogV=n6Z`II0SBIXNNF^y|`m2c;^|@
zI@=bFmCB1f>uKOy?9{Q%Lu>{$#Iq+<`BAOyc^}xujC)Ey*3(P7A?q~!)B5<|PxF}H
zcwRQt+|8kY-V4CP3?1xdz3jWq^K^7{Vr9=c9vw39XR?0$xDA>tSEUfpgs3D@8OOfK
zbg4PrHt{G7UO-S#L|S?%*hKi=Jw5w(T_sGD+a(}CqmDEj{M&0|VVUZgNw6(`Ug;18
zyI4?Ca_4=b1e#qy(=;#kT2`g|q%w$DW#&EKGmzL)<TT%tmaYt~59U#^=2y91^wPZv
z<+!i(CBS>Zg=tyhVP^ZAVTy;x4n+)$K9Yx#a4)9uhWWb{7DwKk6N>X9Cg0||%uMRN
z;)W%Q0cmdp#Xx90!m41OS_{B#LG>h*)=QMR4vK>J_k;ulU27&E)9FJ;OjTjFFTHJX
zNu)uybHke<P5o!Xcoc0uFEgz7aZb7m=nB5g)7535t{yHM*SF0xM%zbUT6T;TyK!Uc
zr=9vi?ZQH5S6_wBoK&$(VR4~3<)2>lJykyIi@4X$X4ms9W2<S2SZ@!%A{K>P%cHvB
zhlTOp9R@{~Wp2$BUNsF3QJK3&g<44p?911GKFlrDN_J|eUP5;KJ3bNS_?%(RnxmB(
z&d$#G>+)kY?u1ete2cCn;WD47f{~+*@Pm3dF!U`7n(5t$+<N#uUAN=oh^JyA_?fq3
zZx!D^&gkC)gs^VX1QJlV`d$Bh->l@eHsJt*&OS$nu>k!#MQUl9qZLr(n#Ru`GXCP{
zC8Hw5zglIgXD-2Zp)yW+xW*;~nG<}VKN&<zQs$y(H}6fJ*&m&O8wr?CPfwqN{bkpS
zn^8X5-TP47F!^5vl)1+g-R0*xgJFCvLw-5I`H=-7{X~xLWk(%^^X5ErhgF~u0H!E2
z-6lMstjAL6+WwBHl9H8$9H`8O_3w%JTVTBvL`+cm%+<eq`SO=|0WEWWpNfi#DhYm7
zSfD{tW+Ea?EMvvw9|_N_yMAeTxP<WNXyvF@1W@{gDWSsmZhSR&S0`V8;NM?gfnQ~A
zzB1`=<`@2Hp~(Ab*--gDt{#pJ0e1?Baa`koF;9d2YfBzFu~O2~80G;38f(_})Jh>@
z|5=^P(d!qNEe9(Kx$ZlaKxh<UsasD3jvtTffqSogJyy7^GeBph(3Uh>W#{4g+q(ZQ
z#Ap<>Y{n!%Oe8!4{^b3+bK)j_T*#Sb_GI~!qV~)-Cd;k5Q^bZ2Mx~Lof2APQj&C_p
zl+W&@yhE5yY*U|yZh9tDeT0Rb0GYOs{%gxE>$!$zl`MJ>=L5K{LN>+lVV#fd-x+=9
zg(W0qfvoTknF23rf*K~|Kud*NVk@)%nW_gCm%D9?M^B0kFECZsLi1@KZ8YV#gZgP>
z4mniDmOg|wFsK5J`GUr4^;@*B7SBb$O3F&IKa%;ocV9>IOiw9_nPZG!(60WL=UY-w
z=0Zb}TAAFZ*{<su*u^tL3tw<>6CiGb0}q+{!FI>9Q-Z!W00ciyh^=2i9W9bbO+I5o
z1u_yh4b3dtOQ@l*+D>jpH@Ft|1nELo>$M4(QX~<O4}QGB4Mq#=61=cD<Zym>D8%o(
zOW}f2NTuFJi(eucLxu(R1nw}bft4#MJe~-GO72bRODidefjFqUL5<*m#vxug)KNrW
zXS7l;-+?vw(GF?U5Cm7RQ%(>h<M;b+iL`w0v|j&QU)$A{XtL7KzLuqtH+x$Qmsl8i
z^g!&u)N3$>QtW6swl$;+$0FUG9PC!V6~978$nl`rq`(2^t!PsX>9v9wJ$r`k-ef~H
z(HfdTfl`f+sDxVcj|WbwpKOxjllu%T4PKJNT?^5>`q)WGTy|i2-0IZ_fJnT0AcD4*
zI9|_C2Uj*zYnH-+`{Wt<j4H;?G4jslm3@FvsML>6(d2KjXMMi&e+S&Gde@aC27mL0
zW&fIUh(nR@D$&^re^5EiC-RcwuzzjpftLa1S7kHVx0^r=+`FM|MUPNtxPMU(VuY%1
z!_kFk<=lPV5bYUhwlS4Azq)PmkbHw4CVQ+<L?y-ZWY*54ob30+V&I;CkB|1Xpgd#c
z2}3#7Oazaa&=C842-U-CdqfZFN(%f*E1k*NFD;0r(RXk);4nsfJy~`Z7{kMe>T8+h
zfJ#quvxa=ZJn_y^I<HoO<EH|D6szLUhP9OS!g{~8j({%rD8Yw@R)+MgVXIa<CHB=_
zxdRn#;(e1)V#FxWRo+5oJukU*Vp&pq-1DdP*{e(D+bf&(HB>)y-?bCm_I={ofh#Xr
zFEneolO~=y8%`3sUm1%w<kD2MhaXhSODJ0N$Y`qc9H^pmS~Ml9S^($+3=We4m0}8x
zBW=IGktzh)r`Myd@T(C2n)UXmYo-Pk^D=zv*og300U`TzQth7iNtJt#{P6O*Eh<4a
zD(G>WDY|4!NI`IKDXdL(1{3-w*U28$Uvi|37D#D=mA_Iqqy6-oR6+@QpCYFRtGMJa
zDiS;q6(?hhnfrx9G+gxV-52)V$w@6o0}l80*Iuq*N8GdC(e^a>L?mA}F#hy-xR05R
za0GV@QE-gbJjimCf=G&@RYG*mAGgdGQO9avMB~`8oNEUF)U<5<Q8vU}{y@J;Cb(Pp
zI#|pFwQ~2t)0-SmE?@U}J)`<oQ#hZ-O4m(c5xc$>-fR{iVwL}7ZPUcI(@wGXdQ?W}
zdb*^7vAL|7v@uUeM$f1MASX@%5xc@w8Z=v1zkS$TUrdwV5OsoPkWVfzw-z5f87}kV
zo6vVGS8mt@YmZo8|3i-WsZuqT4V9i(4maRXfrK0L$vssrL52juschAQ_6zF56?=gC
z{XCM3mG((A`r_-O>rwcZg$zaviR4}AwxS2n2)}j0XrOIphY-S@j4{dl_$mqXeCexY
z)4Gj1SAlIC%4F=Yi?`IWu!0%gWS$s5&-t?PDZ5$`gKQx6@$m^;gxSunXOqCA6H%ee
zaHX?JzLIOLY3D$m?BvqY;iPD#U&sVO<U3A@F4xi@4c-s6TC~lP)D-=d=|osz(@wiy
z{AE1J<6Mg#0HpW4%R1L)!3_+~X#9aLt#6jIpd>fKP)WyElGfhWZO)r@7xiXf{LtvJ
z{p%}n8AZ--F&ssvF<HDJ<NQmb4b^wV$7}?SAIzDcKBJ%;7p=zbJ&ibTK3M`o*=C3a
zYlaKgtCs<fT*8a)io`;>Gt&mETsJIF%D^l?{I!?MUgM_c8~NyD?RHYnF=R(}qdIGs
zJ!ft6%ZY4=W2ZxJAXO6}$bI+8HS-#$SSgW2fZiciwya2;5`bb%y}k~2)mqd$VDD8J
z&=b$ZL}{1J@coqwbZgeOY5*I)|8x!??keu=mik%Yw3KvGda^?=y3%#v2~a624!pje
zzCaF(%ej|#72F<UB*w)^ns+oYNzv)O?Az0h_>!MtlOMj#f8EBoJO}Ym!|b$nH)rme
zw+PuNHYF1oWirY;e5h*HX8QxT)MZRho{NeZ`&Awg+V?avGBPSUdZ=n?h&D!TvDm3`
zpIi+E$pAb<Tn9PODsbz8(Ws-K_$!WjpB$?)U3nCoZq1!hi%99TijW@cF{?y`Cd_}o
ztDI79kic0r1eX~9a;KMI(jsmc&bS-}XE_-Wu4UNaG(M=bM;9$L!gQXD`txp7Ydb0!
z^eYH@*!N7VhVXT-=g@08U(I+@kZHVH2a^9Ue^7sbDV@PCBPE}iv9+Q_?|gvd-4_Cy
zv-3LhD(UeR8j9?`PTvmpATw<X2wL;@c&;3BEQ&+ayyii<Xal&?O`GL4$_rm-C`2jm
zcr7tdhBt9qE7GEkSLuQSvj(>>o&0H85ZPj1svTD(vH(d<Xf{t!|8r=E`w6Z}{ked|
zWC=UyuwdV<?pacg-s18cz1GmAsd?iKkl()t$C87Vr|RI}ir6R00GrvYrS&9s-15DG
z9lU;A*cvYLG`|OiZ5z3k;{BR){dk>`8^9La(Zo8U$nA_l)3zUFX@^%2u7{1`mdeW*
zHDfxgg-=v<8Gk}Im9On@!@TCY&Uu55UZQro6giE5`?>WBn%e0y`ZO2ZGDb~I>#T|j
z0>7h;n`;$xeu^#KhNNWna_xsq;6mKES-CdDA-EpJ6MJviIb-qtH9{hl&W7)W#Bmja
z6SQGN6Zm8K9mkN&9dW&z);agoca-^?wp!qTPV2d|^9{J%U??+|*=_CZ)o4;}A+wDV
zC##erNwH89B@rG5PuD0a@&cF;ssg|tpJ_Il+*I{4$yro%79FQ7Poh&~heQ&x=mkdr
zI#n*i5=&FqkV}h9ZUFmRQ!Ig^4;9Q`rFW1SD%+)Adsg(o(iX^{!PauJf66Pc-vc(x
z#|EjY&Yy;>A^#LJ=l!zDby9O+4)b#D#QO^j9V|w9?|y5d=pmBni@J>DzGeJWm6~6S
zX)swLO9YC{#ZZR#m6k}7DvRsUA|nzW(w7$4`W>b1WTC6#faa1ya(I$StK$CR_kh;(
zrXgsqm^#JD`+m<vHO5Qf)|u1h1fvTdgLM}U8Sf7t{#9gaQXJg=jTgeea1XCNvzm*?
z7?!I9<=&jxhOM7`scBxOUWg8`jI$ZLgLUpu>j-n@5Hadk6EhzH>SQkw3RfoJKhFg6
zrUi}DJ~m*du63Y7y{h8|pkjRe-*u>q4Hr;`F<&3@z0Omn&J4SsRPZMLRorP&lOF6G
z2F#FQ&`W+kOA@!DKWpi%9LxV)@UhMvj*WtEjj{?uC>222Cl{6|&T(+Hv>=<=#5Asr
z-7~%Bk`~eTHA({_=;v4;nAE6@8f~7AxoZpMi2;Q~sAtSrSw$>V2fSZxoj=sv+_4vE
zCwJVWE)ja_6(sCx>{3p_J1QAI+NAqsA5wX&FU{w;h2Wr{Ml;VE{B>!uH(hB=WXBLb
z)RT5Bhf=+*>lFMRU%!2X8Yo0DMjbfs56A~loNF?TkPPq_<z2TXk^69z!E<00>M3x7
zF~#bl4;z;3sxGhJIWrt4Sy>f8`V(oaigDEk6aw0Dv5mRSDa#6;X4ZVAC#e?{bu8=z
zfxHEB;k|}+T!mF`U2L)tuQ#jimq0nz4#XNx+JO6ztfE`W0noBIMI}&+@E(i%y6U&I
zx!Z&SBXKL!ge`}+n~_-~qsBW|SF8Jwu4z&l^vMnvzPMu36N&;>91Gp<7xX6-z`tFJ
zjk#{P56nO+#N4<VSnTf?#ms&cC5f7lLu)37bqI`KZ)_K4)1oTVfR2!joDhi6h$K(m
zCf!Lhn+NkmzRFmIG-)G^_BU~-AwFBc=9ck|1nR&erF?zeDuWVS6i<IUdKD0HEYbzq
z0p;5Dmji&CcBcj}_yAI1WZAv*6GQ!M9Bg7k9ZG>N9=(3Dj3&3JvS1SK6E;Y@M!fln
zjgkin9>0yq|I@c;f7br*WSa3`Wk7;B-$_M>*FXNyJ|rJ!rHh)!HE(_apZhVurcu2u
ze(<Pf{Bkk3U{f(kaIJH+5QZg{bnGbSms9H_?&M5-i}mZ;HdRzcNm(AI@h)!H$L|M0
zwi=8_sIT){8umhYW<FI1hvX2q#rCi*!a~+IHj{u*wrXSQ(Zgbmz|<XWlkxqUMY5!o
z?CQ>u;c+-;r46M8@@yh}7{6fbc<`w@=;)2YpD_bJZFr55tfEyls{y5kdXCf+XME3H
zVQ0@bmj#w;CW&mP=lhxlIyM&y)z}p8pyLK^!rel&mo6~AsAM;e;E;zki1+T4UvR;t
zWY1T_J3-#5H%z`!&XeZ7?gtf;kf)Yx_cMI<;c+E_027LbM`Sp-PN!x3f%>IYP~w<p
z-*f%tPk}xi`M|BsBR7NIM=FF;VaW8xLe<WkyKJom?$+eANR(|yleB$mg*4-iVN0!F
zbI_IL*)PEE#>uUfeX6s|Fattq18$x;6n}r%n00Mg4}0Arr7!Eq^Hag101&SgifP3Z
zkl^HR<J)sMXvdk)F+Oc<&E|v<q8c+Phu(}maPBb8snk~?;Q_kIL9)6TCcDb6eWU;g
zwLPM$cIQq^Z5`?lD-gKL<~R;RRc^YsguA7ZR8{hh$Ir^nH?_636<IGRCAIO)EIYXD
z)iR-d9ldyX6pX-Wh(kFME)%Omui_>O#laJdeg<)kajD)<T`?x8xR_FTA85rGtfQ6?
z&ZDT?atxo(V2>SBdkejmwiSQ-y_j2;;3XQCuR%<bHWCdbkck&ojSaoK2y}NJ_Y@o4
z#=<lTH07k|AQ*Lckj*VqvUQ%$Xa!oe@T-byYz;LC@bz2N$SEO1&kn0}_>LB)@JeU)
zz)drwIC!twYbRgly5eF*91l|nzvh<345<Bfj7KZfq>Hpg@8_}U#K2^FFMj&9^2zsw
zzb;n0<q-k=u{^<rv*n(Zf2*v>4(SGzH^T+JWwKO(IwxVJP_t?8E3i@=|9PMkm6YBh
zh+P(c3&zcDcZ2=MZ}%#D(Gs9hHu1;|h|HFMNkWSAj~6E)rCNtPA*~4Socq%{IXlav
zGw+84w_SQ`<78j2jp3zs+l01$c!W$SJqRh8@$zqKZ#vAozePrhP89eb-12$8qtiuS
z7iBcBZbdw%#|v|}1|9BoBg&$8x_#3GgJgb)x72L48~QDVcW}b@PS0%he*UYdY_c<9
zZ}rL#-D3h;#GTDGQCP*?@ubiFrdjp1lZ5OJ=Q#tUr%eTY&7pU9TY|^M+ysZ<M+gAw
zjFnDy#1X!<?E{t?&+7^gOO7kn3RX$)V7J<QDp)Vwn6>B4z<xau%?ym=9$qD1ZNPZg
z#S^XASBMIh;6TQGvi_uvqJ~a?V^w}C12zTflPLKJtg)@}m8By3ium13H=9qmOpMEV
zg=k%H-vzj{l2E*Uz7<bSwMuLSSp}#H$8al~QKI(V^J)joBW(`p6bVW&q#mwMOK$}C
zQ@C~)FLA-fwj2)iI@*?U<a8zs2);`LhcAc*jOx5JafOcI)Aj)BAX{l+A>C{FR^eRp
zudU1e-`;%RetE%ZMuN{}gcw`TC?$5qcpPvLE8wV3u7ZmR_&|5@M#IaCrH&KDO)uyx
za+J#tFVRy!hZajblQd!L_Z!jx_Mg8{vLJx5%;~cRWv|uqHm(XVDg<iW(<-6yB>$jD
z@uoZ^!;JgzoSP@c?ZTlEHn{#-aqt9uCTUI`<l@m-ecK{%{-}t0v}>88GlsfOz@EfA
z!6}O9!KtHOyo!#&-HAs{rymwZA6-S%bhzua(<8Ys6m`~zy?UoRYHYv#0o+&mY$s_~
ztOmai*@8J<Hsn{&L#dOLYINF}LD_+6Fa>Y6(83llCjf+czEoeXa+@G3nEO{tD~7o_
z+}26zN;*o4TClgaS=MRmdQPa;K7ARWq<U$!=?UXP%NF(NLttLG3>B5LdZ3oTCvQ_2
zd)<CWC9cbi|9y7S22VgY4~c=0n!JksOdx!+X}{jI*P}BbRU2U<SSjix6E2jd&l|%n
zyP=or_mMwSKCsGXBm_M1;a<~PD}`*ev-mlDykJTKKKv<?zD2)kGGt%b1$^^P`4B+$
zl~#=GyI!8K6S(NCudnYiSiA0=xbpn^K%Zw*^}Wf?;Lqtsi&VGA8qKlmQ1NBp1vGv^
zsovRkHbcWYMNxA`2+SaJvbs8ohXj#d>}89!CQg8Hk+BsJU!z<tq77s^Qi&|R?9vcR
zDo)}*EPz{&=V!#~7YWJPEnFS5d<%MZxYDouWZ(2TYr5i#rqcSj)G_LwZEkpbGZJ;N
z_=+HJv>fGHtLr8+PPOyZDe@g9H*?FBwS{2}<=Xz(%mqmxC;HZ07TOHpJRo(<lh`s<
z$V1LOM}!ld5o|maY^Jqo{gUGV>TLlb3ySvCvV4RUp!bGS79=`+8<9LFSFRDz(!4DQ
zt+0JVOvdtcWzK%cN^suXAgFAY;1;x65U>IiL0&^aw~o5!-_M=C7c7@`5Xs*$&s#nt
zR;Za|Q0dUCv$U1+hyBXn;D9Ra#nETY#(fLM`gZ}%qdGK^lu{sf2oM^v-?lP=sx;;F
ziDIai)t57MEaD4PDIxu;HAne$ch&(_&i4(~-(G<<)QhSQ4nCq*8jNJ?eF^+%l4G2V
zf|8@f#H(73#I)cS1_trl_OpGZbIM{>@o+TgRLkXRwAEepWlLgSfo$VX8RDv5$4|)=
zxyhDE@>dY4cECFN*H71=qEJwHR9ad4xSWQ5bThwi`~jTuj@LnR)?aMWqmiE`el41|
z^hsW`lT^s+?K>)ZKE9BrTkc5uoA~XPE>74!;BoqTn-S6*G#U8H(b+HJJYZ?jfy&|F
z1U%%Q6{<|#xStejHr~UQ3QO$+k}R3q0fSE?1VF=*ZBEKA1!KgYALt)fp)Y(6U!FK?
z4p-H*JSYy{3;5VxpCWHi^HPFB53lpC%S@Gr{^Ec(euuiY=^mo6zn=D*N9M*01@$i7
zvwxKuE*X->;VZ8{im&X(FTV~%x}M+9+M4pK(bF6@DpocTbzM5wR-Cpzg6hMTCJ_k-
z>Q>NiunPhz+`n;3F1kbvaG`bkwnf9j0ZbkhyMEs9NMna&pM09uH#|`~<zm==c+o^g
zV^DYW<_f890^DGwC3xgjrMw@IIbRw|3O-VhJtKZ)qH&0whi3&>u_H=&YBJGp&>-~Z
zK*!J~UPx^>AY;gm9y8p<Ukth&cYoH^{>dzKf1Ag7p~u+-$xMquO)S!D#I{m1pY}4q
zvA(fT{Ub-5G^!9gmj0Pzmbh?a7$QCnYU$Ug*l{9=nD9Xeb_{uZ(?o{-TOV9EWV1b7
zVg327wu~ig>d<3d5;M9+)j%q3SXG~VTDk<ju$aePD&jn72kPk?j-|n^kX@dxk%Z*L
znda}R2bWDZ6<B!~wPe)p)3^ypH2onXxjE}4x#3-+G0x48oUP7X<zu6!mgpU<I0F>?
zDO1q(;%H*IAPwql$eGrR?cBmZ&Dl8$L|{kW66Ax;K9l$T=g##IJTl-f?U%dDblRe9
z8rpx0Tc({o*N^X+M+;UrU4nai>-$3z1R2dw8{*`Z#WYUOq&Rz55=$3t9)G6Am51hJ
z33%sy^upaTB3XkC4n$x$Su23Uv?-(9HQ~x(GO`aOGRK?cdDYve`}%;_X-I(I*MpTv
zEVK=ayC@x+cA?iAi3}h<J1WyGEqJ7sIc7Z(XpQ7OGDR7|x!yMg8`s}Qj-!9x^NYaR
za8T|Ws!DJ9nwTGB^^29d-zHk<{PFv;Y}@F17hVb0WWuqnxYS8?%R5o>U7j{f{&fPn
zzQxkCNu1%jhfE`JLyFT+S1S5S_2?Vs)CkEd%KE*_p$E$t@E@i)?rlJo{DaG+nt9P?
z4G;ZY+RvU8x!d*e9wOg?r+xvBwbY6;j*^!mepyWAvAV>vaN)hKgCq~(;6;{UgRvMz
zeyC2i2iO$s@fCM#A)B{G+naEE7KV*p+w2=DP1;ulZ{65Q!-YSHY{=T2d)%4%<7B%3
z+L4UwYf}BZA$0qY-zDFl@c@EiKMxIlf4X)I|I-`StkJwe!uYsbwZ0lUo4NY-nV(gB
zitsB#JP)x(#eg%(2fCn(9KYa%ttccq@$__dp!3tKlqZ9yP%7u*k%&-Kjs#!fTEdd(
z>}?S@l6uzR?LQoT%}Q)Nicsdt0M}*jZk=Zhq3KVI%#Vb)cYSZU91_3>jt*4Qg>spk
z2x%Sd{Mg>fEtoN&Su9<gsJqh)vGfb&l5s8ZF3m2O@|r_PBzf+xy%d$#)zYdI)XtU(
zZcOgUc%^hY^Sa7O{7PZUDVo6&dJ<n|7KvKbKUPt_h@~cg!71q?(|)nyOQV{HmA@{x
zP;=0Iqs=<e23$+TZPs&{GMhg%OAgc~e6+TtVi+!A?J12<7mk4bpmj&olUH0lzql7i
ztJ{H;)q)tL4ed&ODduv6e<*xElW=oomm$rln0~|9J+a*ZJe-^Z$YcA5*fC?2Skcxh
zx8=J?`9yb7Z`tO}_7$TzHcT{Q0B>hj{hp!2beb%#UNe__h3tbaYO{tM3D5$`O3?ei
z9)l6rmVd5oyQU=6J_}j2v-;~Zy3!`U-rU+d&&MM@IHLtV#S}wL7*QZ=_<KlQGNN1V
zmaHTE@T!6}cl7!XPQe>qai8ufnyy}CFCK!4nVV^vVgMQ)nwN14Sp^LEJka_D#~dRk
zTNw0VXtz15+h8z=psZ6oe-2B~ODw?|C}A()^E?f!zAB|;NHE2s&eObRCx<(rFSe6z
z>$;x(O32;Gl`dD*Cpl;RI?r=khOyNlX!3kM1@;4Z_8ip<Wp^+^>yQ$&N^!T&$-IrU
zs8yYo!x#=#2n|^LBq3BZ#0Cd0ozd0rbI|76suwhA&|pwq@;>ly9*tC_EVdR0D@oN@
z==_p_K4`9A4cUFEv<}e)aqXrnTg4w4Q1ARXMB=bdT^>^E>Ovhod$#vil9&Zvp=Mxm
za<ZNsf7tV)u&7X2Sn@e#W(EUUy0tI##!z1e3~4FFS|edLhx@st12A%Mb68n^2`Iac
zyikSMoFLovl5I34mFP=p_ocVi30iT@_kq!ZAyS&$kR8GVXHL6hr4;M%bkV7(BfGb^
z6lFbU*Kl^+v$gMy4OPA6-z+1E=}stWT4_M&DHl60FgRrXrR=U}VzW)Qi`*0WF5-HM
zx{i3UApXmHp-&iq;bhQrzM@mYcid+dOaK$v+$F%F=D`M=wz*L^Uy<#E8ky>EAwJr@
z<B`+RmB0QGm)(ZICjyuCydn<ipmaM<@g6HXf{1W8BjT&PRF&IzU827m2ykGc(op4s
zS^XkN%%5!U1H8KsUGp@?4YJWm=2G-R7=LL{T|Q1+@2E*<2j~v8ZPmVA>Y3-cInB_Q
z@0daNmtu@Y!4iRU#sX<^G-j-R)csw&?*8WemEIi4e%eJ%Q4{*&32OVb!SJcuJmzp4
zm9bI>aO)wIW$M`n=GmME62#CrhN_iInzavy9Vh1|?rV=JSm<ku?E6M+Jm@)d1{=}d
zAnU$(1G*M?Fo13U_|8n)I_KajQp3IJE18l~;#ru@GoqKE%$0>c+<lURpujj+oju<{
z-`N6ChH6BNhM-ONXp><>sj)Ic|EIi}Lh2Hvbx^<_dsf1}9XO46y2q#d1<}LM#P%@a
z=DtYaICVqdA4n?z*&BKndHC?TW4=y5hYG5kr~}7gYph3{R!k64mF>5n-fZRGU>~xb
z3vM7}HAF9lTdo>~b8B`I9G2{H%3@+WFup71V$7hoUxD%vY4}o4%rE{F@ZxK$&oje)
zoevw3o=Iq7?}XgI_cqSV14UoS;-vH34tnvkkn&K^dqdgv1qm8%_Rp+9LTmj>W0VFL
zpe0`41+u3m!q~ajbg*<CWVTb&9<&x(jq!gG%I93%<q)A97!w>cd$vhs-5oTFd7fe1
zA#z_)U9vlZyyO>Ryxh)eUU^GLYeuu=aff=T{}-HSVbsl~^z!`A^oQdH7Y~nIx|AHo
zszF^pf9hdM0UVcK+i-Lp?`br3V818*&F%fC?|l}ArlXo^d(Cv#T|E_s6yR*&wo~O^
zrXzNk9>q!i<zip>CaTn0Q4h$;K(&<?v_IySkwd1{SD{d!G<$KF%9f&*?b2q;ipP}s
z7sJ?1(bAm+6S$6^PV<hAN!L8i{2t9DX5StG+*5DO3DL>v8_Ca1Zlc*@PbnzKmgZ=N
z4FeiPxty@GezaedNoS%>%VM|wP~s>#SZSd9<DGIF^5gjij*Gl*tkkWrA&If4y<7)j
z*1YRkf(Pa5&hi#jcAf}_qdK*hAaOlR?z@JF_jH{r8{ko33*l8<OY!79Qh%d5#vJ2;
zsi3?Ewsm9Y_mt5_c1Gu|{B6IVwQlI5?MN6{Myiwh?`VEIrnoi7fXXhz(Oa!@s5#1K
z`fOI548*hAPF!AB@>_ARQTvcR<u((bkH<+xH^+7&ZFqxsJ)7>kpxBGF2|e21Tu96W
zhoj5Nj>IC&FL>}BNVv;fTDS7m0<gWv15<zdSvd-{pCnQ2v{DT=Z*q;i@?vt8h=K#V
zBXkb~q1_wlJ6nc()pdfU4;EOJ>@hbjs#=Ft&R>AL^>{UHhc)qt+o04uDS&)^QCWnk
zjy|Wa(di>+2=r@;c|902AIpb6`Lb!tA5A~E@hIz+V4%$79{GU*W}s{V<v6~H;Xqy@
zJU;dr02m5^+X?*;F81U(i8TLG6|Go`lYd$2Gmcd}3KXjA0_PPm=vr>v8xyCwcDEwZ
zD=QJ=jZ3_=-40lRqXzdcnbly!0rzMJizW1-Zeu<*S)71yK3{dSwhqn%OXZF?h4ysN
zNW1c==4TanY#RLi`32{MrG*m`gP$U-2MI$niDl5bWbq10E<J;$e~co*mQ%2>QT-}<
z`_6_94SSYeRwzn&=Z2hB0e!IMm9?=T^bVj4W-dcEL+h*E>TdBRxuBw4#dBwt4ojIK
zsL(%kkUCmJ<jy7X);7OOE!|sbXEB3Inx?CL;8q)cXsjdGvY!aSSF7+<N(>@m#KmYe
zB%ZZ8yP8^ZT#oVp!2(MdQsvVd>ngNNn(TSky&dmz&cGP}TOhip-#5zXBjw`GUm$M;
z*28qtXhI`=@{cVBcQviX3evU+LiGD?ITgNQgS%7$d`Yzl;M{i)s_a=zCV}6^pH_i&
z<OBc;Qe1J#x#w|BppTj;?qZXO0|cw65O6DO^yKp9WnM6LX|oeh&I=8GL&P*#a<qz;
zzDScg=LP<2JVDUW(Gl32j^R>y+4vjtghAE*Dm5+c>j`oTe;IgCYPx%RC&{75!$_CS
z&kwud7TcaajZ#iZ$Q!O=c^`J!PNwhL*C7**JI3}Fh!K+{e~&!zKuilnHo7XVolkEt
zeUDDrB+w3s+%!s!9K2oF$NyvHdep^^Ul_REv`z3n!;dn2%DcX(xR!Fwn;3?kTh;=Y
z!#2)!hCsewi-OGOj5CUUm5F6*kDIh@_=Mmll2Zd(ADR_U|0oI6nH2Vw(2Zl_V5CF_
z6}lXrWHox}wh?WXRe4z0rcLG%XTeii*UV`Y&h)W)E-UEmVM$HlMmM0<&Q!>-!@R2)
zZRvFbi%zo5=%-3SCJOOEitam}ojW|z`k$`B5_Z@Z!mnL9&GN=OQGeyWl@WOGsrKY&
zclK3g9^tB-PIn*}5M5%Z!ttBNbLRu|pVD!$X0ZaI-3=NtKs+36T{8hwTn*Hdn*dWk
z&$TO3@x41+<;l?!Ku0EW^3B1)<>6&F=!Tm&(tpKzVt*LUCAAA;4cJ$sHLXZ8u6krN
zLojd=?<qd!^VNEUzQxK<kWCHuQwJ1p-n>~~3ksxo8g0p!5~}JjGOoejRFc~lYV<<A
zu_hirBe6{gV|XkLvV+AlF->4@IUFNS33jABuA8OcXFTS+GMx9SgI`O7pg^fM6r$Cy
z3xbz}F$}l5*qtLLco4))|C?P<Iv_1@trLgH`_DpXD@Xa^L6T3flcUCxbG3Ss%uYpR
z>!PB-b2x&CU{SqnfZJm)wuKU%2?#j|;0JEb&kzTV8H2<dULc88dmV%Iy@hyCx0ZzA
zX0*8R>p7>n-6TC&ZiR@=%@?xdVVtbo1On(zUV-F@4p&dixqzKlvDWx)FT#GcHzQjC
zpH%l<G~6P=`@L})CjlxgS`&Cygy+lP<HyA8GQ&>AONf?A?R~c2#FI+DoIlVu9Osjz
zXG%>UWd8~Q+|3k0UG+>{Kt;X3<ar1v_jougeKcFg6hQw>Z@}Hl983i5Lc&EwyzJyg
zFMP%!W(4+EKAlE^nN>QtjFjbZ$vEjv4VJxJ1u<C=R%kSOQ@r-kih{2ZVRZsoXw{+f
zI|ob)A`#?dwD>ohX@CNCRy7k?{)eAu0EnwaYacw+Otr!=o@(`O1-JM)0K%FBD2W66
z6dSjr!n)$b$YYBCAfq&xz_-`S-TjA*eehy`UPv$I)w`?;`z}-Gth2`7MU9Qr2#j}E
zyz6?)`t<;%md_<6C0)i#%FQuu$ZldaCRt0o`6(+|P7mh-Kzg(@Ae$t7U@b{;7=sGh
z+xU$eOCOfMcjJ-30Ry?*ndx9J-r_cY;F;qEsF)pqQ#`o-!MUypFfoJ{)oufQVOg5m
zT0?J-@1c0T;$Qsr8dYZhM}`XyZpQD$CFh3t4Qvh$rT@lPJn2fQ(b<>({DLqUpPKlk
zH{(8Vg$Oop4tj5lu!J$O6~ID@FO$KOE*)6SaG7fsy7cmozsJnC{=BR*ctT<2*+|~(
zm)G{p94%tmbw4dc_nBp?zNDb7<U(MEz8n)E20*!*0f*zvn8082>A07q?vGzR<G#z!
zn)g6==v!Mpq8`LD&(L4VZO+@1m9l}`uDxlB8owc<wUV!@Z085$!^f%rFz%+#GvSuT
zcdZbt-1kkVW7DpRg0Z~6Q=&>R82+`H!|05;qu6}2SxT})Q~SqK;SXocdhDKXZKgx7
zUBg$%DKUNq7T1y_0(RVhMzLnUp|hfr>py3hNd%zG=WA)39bcY9sF)Pl*`ySigRE~{
zdxs6?<xX!IV_tP3LVu6?jl}Nlkw_biflZLq6MhxCUWj&D{@y0%9p9ITD{bNi%lQS}
z-$zWvEX(vfwPvcA6f276pMuMseR2}I)`CCG_*$%)DDWsS<KSR#6XP-cM9jL%Gz5ES
zp>ntuN`kBanv~ZBSRny#!285EoU%?+EUY{S$?A27D0gLUSX#)TTT4NeoUGcYnc2T^
z@i)J|F*Qs%c;-}ZHa+E3%fm-#R)1^05E)B8=z|wW+>*b3N*#n=kn#Dc$hE(DRg^{Q
zaqoO@22#s5ouz{-@|F7jmqKCaq%!3Y&xBdOUSek`ynNSuxS`dg^7;eK+9p_1^LTdf
zcRExh<OtawxA*)~`^Rdxj+jIJ8I7Ya!U=*d>jweEIuP6|_7d+uz{ks{mCyTb5UZUF
z=5;dcdTn00O?2nCIJyKZ2QdzMQ|O=rf3Z|Ktb%seT_6F0Nyfo@XE5e7^_W$(RY6;M
zIUU@ioBnXsu}v%$E^l}=JRCWLQS?k<aQ};|L^G4E@5@228V`5vKUxROKUdmZ<cH)>
zw&FL23m99o{7~8t{<<9-e?*IVh^!{8Az>xw#NW&JKw4gq37GX+%dK`NeTkvqO}poJ
z!T6PMSvYc^2@xJe|NFa2dGUXVoR7r+zqy_AtfQl&U7elf{~~hzIluKeKhx1M_BF2l
zZ}`5B$cVhhR4(*_x<{SY#?DU7fBiVchHZiU+v)#Z9EHDUkA+RG>>niQl-`MdG~a)!
z{xcK*k%|B8j^8uzzga1KHM?`;#@FV*c^@89?|62f?rqG0nZe@yi;ecbSuh|Ek1{Yt
z_#0|NmhJij+llXr+smV`CO5Bl;pH0js~^35ZErui<585CeT-${7Lzg?W?YiHyzI*4
zb&Xmd*X{1<F@iwqIZ6Y}0~8++U+h6}Oo}$b!f*H2+0}}PW@rQWJatWdR3S2Ocsg0k
zg2~8@aMtkh@-kwATdilV9eev0zmkptGMT;|s2@)$_kXs|QEpHF#vF!OlaewsT_-p1
z?!496d8@k{-hlEWFdoP<x$}_<V6thAs4=2UA!@Br-h08k1_Q)ZxGBMWmIh0WnSq_^
z?qrD?!uJGrcJ{<Ya$*Ks!0V@BP1TEO6-Y~1JO{f$HE9}I0cIb%CYPh(;vxe2z$wNg
zkJlresLK#dqt{=+-QSK;iDfTC!Q9Rvho8QG|IQp$EgN;D{8a?URK#z>+{4y`SSAq*
zuB&SCOL4WHWW}U2|Mi>B+?RLuYEMa4OOejpvk#}!!d=K!&c+2Cyu2tTQDX$_Z#+>C
zMxOF!a(|+%s(xPyaF2iT%3VvR2^DmJHwB--CF}BZ6cSm}jepKHjCPOV^^Cx;AsUYr
z6pvjT(zw?W8hV`cioBs3!3JaU^fc6eJT@+~^*!jV##S%#6k%P++X@)q81Nz7x(R|`
zVIoqQ)0?tqomNk}E+Ye0(aFl8L?^tOJl*KV1SXnu^0`mQ**NH88D7<rs#@B)uTu3E
z-fd!cTp~css}9m-1FL0)LYWjiocy`hsrH|LU&X-jDcuW1{1Qocr`Uzx3<6Z^RQN0z
zfJhZ>#BC-g{0S4hSC;cwUy{d7+CWHH_&JmL@*9$ALUb{#N#zk6icRy9UE$0+%y(o%
zq^*r4R>C?Y>UQnkWd&_i`(3<vk;zvS9?L&HJ9%ZS$_=D46Tq|{@oz61cF4Mp^q+e$
zZB~@29gyKA2iA$$UV#PbT>8cqBrYJJ_M6KU8+Ia)H7Y*Mu;XLw22Lk$=?s&L^_<)4
z{K7p>rJ*zG4PE$5(&9?2;6Iln2z3U&`5N}o=g;WmYi%!?m|-Tjz8`+t_C-g?dc1&f
z$?McC6uZDxS8JmlwYx9;U3Y&?#U-SAIFb(bG>8gpuP-AT$MToZ>7mz1q$|!^jY1(4
z-foo+Ti#JDziHkcGL=IZ(AfRL<<V9S&gf60b&^>tn)>?6JdzHAw28D&xlKz;cww$P
zT!ObnO{x)n@(b7Fu=ZX>_c+;}jnoaD;~BjYt>Za*MY}G2sAuO(5${=Ey{#8x?3Sp^
zd=Y@%>epLO|NA8R==l4(5N4zQ!xwYhUOGDV<yJ^qL)vQT@AjUn={!X}v3>V{0P#-4
A_y7O^

literal 0
HcmV?d00001

diff --git a/figures/dlscratch_img2col.png b/figures/dlscratch_img2col.png
new file mode 100644
index 0000000000000000000000000000000000000000..2dfcf7e818f719dad66c4cc9b9b6214487c8a8b6
GIT binary patch
literal 47906
zcmdqJi93{U_&#i_C?lnfvL-@^G1kVCC1g-a_9e?;vSf|Pl0?c<$TG&xNVe>vK@{1>
zSQ^>08*61O5&iDR=llD<@B1&j9Y;qp&vWnhb)DCFo!3Mf>feAJ<vL17M+ei<*0@ba
z$K*su#}LbW82m=+{NJnKA9}CbH?Gr_kS<Wcf0)o}C^b5|ve;vPZ4ZI}vv_EmdePCH
z^oIV@k2~@{qNA&;(a})5^T=v>?1+u|x1oQ(+OSG7_pY5hiGGdZv*nPb*A2OCrklms
zaXBw=qCtFY&#$j<quTu@mBH-jp`#8x<&1(+%dd#)_uEK6$JML<Z7DPizCJxv_Ays1
zP-T*D%+@q64>Ok0<F9`wUlWW0{9%{r-MsYQkD*~9(BSFl_`aT-Vxl{DY7fyjFmNS0
zI(n1KFY?8UxUR0=xs%{?<}(ajbWI^X-rm}G?^YFARNcO9bbF<0wU<ohi$+Czw%<?|
zkdl%*$&G3Q>q5tLgdRr6XR}kN-jZTj@2fMkzvtN{)u#RU@kb(pM%|lmA6t9z;;6H$
z>l-5j<lVmhK0#U8#B>>`r=$CKnCY0Gho4_rSt$?L;1Vf)`>H)&xLyoTT}Vh%^40b6
z@$oCosH;=__2}B1B6x+Jl}Uy!B*?2d1%aWw$RE2n-4<I@Q~Q(!?Ykngg%pu0z@#Y)
z8PtDZ0uP5i15+kTa@h<}zO|Rnc3$F_`s6dIXJi!rV5Fg&wOdzDyK>N%A5Tw5r~3W`
znT{?-R~=S7_<MhM?WUDgx;+{_Vm_KxQmtR@IwJJ&;lq=>^XZc&pyB5ipiwA0uf=tA
zh?-UfR29g@Ck!BDzKU1Qca@CQ3yO=QvU9epj?mGCkUtCGrBf}H!S^!M6<TtUoVrtp
z*sYT!&<x+qFXz*AKaw~|4@PTs2s}#P-XPJHb9#E-tdp$Z^XohrCiHYS7TzfsiL5m?
z>is_U+Y%97{kueg<NSF{DRj=Rp9QO*c?1Sti(<pzWc=gAS~y*x#+zKXXO7Zc_0~4R
zV(XVjKirgcA78GsMFsrp?C7v}|7E3Pa8nqXdPw0<>S;PMH>t>~LaRfp!(l2nvSd*q
z=Szs+>Zk|;fryL43*5SOYx4}$DmSwe9bdD;WEZXq^JV(iSE;rO)9*rB!r=A3xZU*R
zS561R8oku9V>9k-eS}Wq+}`O-I3n=9TN)=RAuY|-{b59PwXBGUh}4tYk4KvV{@%G`
z{FUb1pO?uuaqS4L0oshK>|{FCu)dbY!xx6V2xVElZz{OIQ79d6qi#KYC#SB;2u5wF
z3jR5v8-YN`BH9=47NUAmW@hZMFR@NzDWAsO!5OBb3uABr4KejSurkMB3T0|!Mlhgx
zI;LY_(KIf@oQ9D+MiytmZ&W?N_Ny}WOO-@<L(3W(`@fe3%=h3=Xw?57f5rY<QAsJU
zq(mQw!wHHic32$)ll|{XPkDR$YCGJ`TOO*v(Hg~hk%fijBsYhhJG4GpJkX`jc9J)`
zd-u|=uCD#^_p-ROG;y%4C(m<SymgC~3+<x^*he}6tymAoQF)OomlN>I4|Mc(;})EP
zE-`@P|7ct0DBW}BV%Or^#%7|kv$N99zwJ|G9>zs+%Na{cOA9J1rynf3+L!!WbgJJm
zZf=QjaVMS2B@oHI#wI3wIeq<?BqU<8P+&;RieSrCP5RT}c@BAfy}kS#tp@LxNcK9O
zp5>nn{UF=N9|~Gyuhx?{GII&P%8gwvyDd|u-5S-7(0syz1`n|7Lc49-&Ph3S{=Aq`
zzWy5)4x!}GCLJA}FgVnQ0(9@+=S)Vyi=6*%|BB1Va9`{;WIYEy%X(o@06yy-8WI6k
zs~o+SzG{o&B1e(Fz6uLtYUTuwgcV(0YI%?!82Fi5Q&LovpgucWXHyDOS>fjHTmAL@
z=4yV+p~nWE*H%{gD8nw2A1v-9=vC-1p5)-*WO(vEugyBv$KqYlySUS0#@fAM0!A`}
zuD+|ILxGJ-ib}a10oM*TqV^%--1|*@uSa+a4w}+p=t=TI&496m#g)!P$)v8XuJ_wN
ziZq3djEq+DBzEMF^IW^E6RESwwdJy2spI2QH8VS#(zfX2?EL!0@1r^f1~r4V_wH8t
zc-FXQJf6A8`&#P52ZH_~_88jtQ~LBM_eo%P`qJV5t&fiyi&Ldg)A-bV|25*4J!x5C
zuB9Tis-~uMe{a{}PV43ND*Z+7oq@&npa8Qv5nZaItE=m6V`HI*zaB~}+EwP}<)>_j
zcZ)j`mEHD^Ri#!{S9c5_r{LNGrXqzqfUUXs>+d&iTTwg%?sM*EtN&~2#I>$v_PYBm
zkoo=hqUG#YM?b{jXBs#d+_^{#r%E_;;JIV{5(*y_9Eltxr>i;i(Wz9sfi~#Cc8hn<
zp$YV;+GlVKr9_GtLep^Hp;4tBI!@{6>dnls*|y7NihPy4+OjiBsLeHT!MNl$WPM_S
z+Z}7uT}`&oxruTx;Q;Gw;qaM(YfpB#qqEcD+q-YzBn#?1_w(yXX>HXiC@Dx}6<2mf
zo)On8ppRzz)5|V;A%Ns0pQBQ7D@Q9OJ`Fd_#X;6+)3}N9iRMh7+q6|Z{pR#s=}%{8
z=ehaXe@#It@(;7S@KP<b%Y2z?@}3M_+|{+VEm|U_m*wER(d5!F{;bKDJi#>iyk{W@
zd^?FGy=G461&2OlBuZN~XX1J)II-g&{cTT>oQQE!AMEN37-TYa^mSw>^~RI04IVj5
zdM~ubBh5v2vgxG^3n?gA5}%D0*;_|04^_1N{fN)VIV_{8sVR+_(_`QY6GsYR-rvG-
z*HrSD=;%Za3=XF9zje!6L%}{(RFq$mROzV`Jn$`87*=)qJSnX6J%RDaacwt!x7c{P
z73V4Et+2cDIW4%47$&Oq5aM}Xz-hPOs9`V1&O~u!GpDBNQ|&C<sLk#ESA>#k66?nK
zIcpYmUF>h?XijcY?xEo{=gNZ{`$MaRhy3U)6I(tGZaQR&44|j>*Ln@=r(ARP{2mS$
zFBUJyZn=qExsnidi*wV=&+{k)+QE_gq4W)?hc<E70E>P*pNN;uoOx>&f&Uhz?oH%>
zya#O8562cR@&YOXqh_<#mK|GBS#;;V-Tl1G%qFWP{hf1Zp-l~wT~8Q~bSV}WPb;O>
zrh-9i$_>PX9DeP1E*KVdE}$Enil&wD*)AiM<t1x71im-4$;_*DWipC_@3ln@d$sPC
zE&QZ|q5E$4>)pf3^?PVVI6HT9u6eW$j1&|}tJmGPvjey*w*^-@ujf9=${r~*>VXBm
zk8afMD<c>1`PBD9D+(NJ6$ocr3vuQm`e$2qy+5vF%@?k6*7Q&gUW|g9$E#yW0~Cs`
zv2o&3gqyRPO|e3bisXSsqEo%h*%#8Pgn#Hls@>BtQ9qY9^ch<%<Ur_?P_d^(7<F?_
zfMFr1GE;8z?$tt|Z-6<zygDK?Y0O?*N7#App~ag|ay;cW8P$k`D-0==1V<?rMS2`S
zP4<rPE16PRn*RRZyD7QE8;@H!^XK#g!{BbE3OR?K0C=Gz4^4-T*;)N0*~uwOco&$c
z<Ml(hj}cA^lX|!B$1Pz<3wE@#q)PwcW>I|a5U#JUj~O!6z?!=n(nlXYLwbOcZ2bM=
z0(>?3d2+Fi!S+~D%@9roUr_XxHFgLmnq15RJ^h1+k<AH!u(>XZjQ?-@!L#v*ew{68
z^Y~3smaQRN<&EePzr6>jOP9(W#3~RFN}wCHQ>Bmgljm4H^rUJ&s((aq6pcChGPLRR
z>2kE>p(h8=tV;g~5gKhy89&Of#1agwh^;02qqG0_Lk<tsoGeWf5@q0~+K+*OJA)B$
z;%O$f8jtYMCa4Fxnngd=oF~tW*dIYQ!+reYdkN;EC*fz9*dm5-e@`>XZupAUFRmD}
zLmq}M<j3MYRp5MN^Z9<}w#Qop3p}n2ZAwW=0iD)XR~wg?mtSUvfFqqMv%HJyG&f1U
z1veTbpr`hOu_WW+&0GI9!!EINnrz7a-Ezl>{rgJn>;J|Q;&Jj@LA!>ecl^Qe1pE6P
zoX_cj(59!w3P#q}?$=-2aIFsET$WtAz4E`rKm8)Ye9Z!Z-sJr*`9Fq1M_)g2VIc(L
z%y>dQHe=Yh#KJj%&zH6@b}85Cec{^5v$Ga95-b5=;n*J#56je<8Ejey&7MxV=I{AE
z0uG#z<KfIt%)Y8Qv9~a!?QpnjdI{O_vLoiGuj+Id9AJx<O`KpR{f<GblKT~_D{tPs
zk;YOUU|knR8Uj-vjDIw3`2D671RmG;VPFkHN$#pS6$q+F#-*D{xcI%OTdO7avW3~r
zytH1<wzhp4!PGm+_Yd_BT_m4-J2&UpUwF}^GAW8}?8Qr@g_jmp@9vIWOeD|W_k>o0
zRzK#Flt`}?-1C|%4@dff-35}OHj|9%zlF6I$0^0}uI4rER5U(FqbaL1DCi5#S+Idn
zS7$~kGPQe=7d*w0@#+@kp*h9}3Nq;Ul4F>iyvwSJQHc_kqA(b&B%@tBh4ZC!MbQmR
z>CbHSS}k-`b<*J*jB%efn>nNEh81;mP-RJMg#`t#6J{<B9@hXN%xB@xM`nF)-M#A&
z;FFRudsQHlK1RQV6B*R7IgIh7n4)jqsZ?r)A!8QSYB=GJU&A++X5((2(Q$K&VWIk$
zf_rx`j>?)yeFV|I{Wko96Kx@jMKV+;LUwk3cH}1VbkeERg6fFSO^IIWho^b($~)GU
z^=41c5*kQ`WLt}2tk5}iyzk1e1A_nq7vCe=z3_1O5~VZ=bzZJRM>yl!?<d9k6fRYs
z%D`P;IN~4h-ge>G63YrLiH|=8L@s|#`utgAd<iTF^Sg`6(TYf?N;hy-+4lA_`S&rU
zjS?f(Zf_2u)15<MmTNL#(Vu1bG<5rA<cRyp5r0;6W7F1l96A%*KRrEy+J0B@J!Qqw
zYY+-{*&hTk;@FlLAKk!45CPz$()^x=!xalN*f<ndD*G=fikf9V-dA`$=*&K~+cUY$
z@w}yG^&fUoW!<u9m-u){8U&lSO-!8DhMqY=b}TD*aDQ~1H=UmDH7Wv;;z20a)mPe(
zfq#B0=184rY}g)YoAAv1H?YY)lGi=`Iy97lgYrlY+USI<KX~ZK!ocX=b;!f}4m+2K
zGF#lXxN_)^<z$qj>=-=?3Z8*Qg#wZ$Ntm>^NxXg@(XWR`VTIO@lJDpCe1wQtQ(T+f
zmzgMi_^5P!jPSSySa5NMYi^@bn{3Lv-o*>U&>+L;ps4+X%x90I4DLhansYqCkGFQ*
z0=~KF|D5dh8(MI%&ReRiiTQvUIkCHVw_ai7W6jdWY-ds@aANj<!ys($x)QLRu|GHl
znhRmS4#KMiHFdUpJ`V>l_<+TNu_YfTL%)ia6YD)6+~kCr6z*FGk1P|G^d4G|^~_`H
zw!9L%VJ80+MwUAvglcY{Ix|1dJ3d(NzXw_Q>nvcBnW^FDkj*JZ`iQK^InHtVh!%@M
z&H4>y^&(35(A-APB$2c>_$bhF&}EEVRlM|WqlXP`W^N9af*08UMQZH!5CnGvR-cQ6
z_xvs5n0A<xJB>Sn@dVa1b*SWDkq^n&mE=3W)YY&lI}zA5NEOAZpoc!Jj&-e=6y=o!
z*OhKekkXX$s*3OaUjKc3tD0I&b*1KgfKCpd3>2`iqN{~qXno~BGdxDle(Nbs1xnX`
z{mIYHuJiRx>>k{yN4;b*d$3=<`+L9P8{*sT+qWMyrvy=V3Ba(0Wb$RwhcM@>MZnGb
zBY1-@3ukzj11tPEb@SS<nsMv(u|evmDPl<tWiW^&OPo;hTYgoA8Pv6%G{M}p`3J4M
z;Yo+Urj=hJcVCJb-@Dftmv{4$q@;mXEDs2sOM&~d`d4mdar?>#g%6v3KV)PPx{i(n
zqRj7iqJ@GNFSb}I7JbaN0FYui?^8i(Y0~Fh4ILD!YA^~9ku$(<*gH55VZz(?8F%yE
zzSVZ@#*4JKw*xXKCh})qBLm<s${jjqR%#{^0TJW3^_ZGk5$jH#bPsgjW&1oi(l>p}
z;_=5nmb0Dn_foTBJF$DM(dTEUUP>l5hcf8s>CF(?phXbK<j-VbriZax$>5!xUEj=S
zVD+`k+e)((jibU47~{?v5u@e2afMG&>q{h-Oj&2u_1Ce`+SBm~{&YS?Cy*R|MqFIH
z@B4sYyH0ytTwEWf-^SQHxCZcH*4-1XG+!mDva+#HH?2|Xmx<o;OVH9ALPz$5Udu59
z^VCPzc&OJp*mD~~VZkGZPlj&SM_ak0>1vLHF2Y2?ReA2QGTidn1=l10MB5$p$DB8T
z(z82N+$M|V29K~Gg#tmXV1Z*3P18;v8DCFM?IB|*7sdw<I)bQ&t1^`f3(G$z6ze32
zEio52e+{o*Q7WoJZo*OS{SB5Aj&3cO)W$Vuw?V53^YyW|Kj|b{XnuE~X|nZ?Hd`A}
z^8CI({H$Epfi}0^^<A&%ejNSVY?GL_>SKlXA%cr94GrZ1lMM57A6+}^W7a!^K71$r
zk8tkeS|((J#a{)c&Fv!C)_lfC{uZy#)sH!Y=T+bGXL1TCGV0c1TwM|Uzl1Jb8p1Ri
z7<t8i{_-W|um)GKT6p7J=Y%YAe8f;-Z7*%E$giQb=8q@k&+1w2|GB8$>|_tZyE@&z
z{{9uj<c*e;;X1SS&d&Jc<a1cpN4*(#fK$rN%NxRkd>x>!_e@&;%d5%E!Pf8j1W@;U
zNV4eqmeIsPm9`VsOMA$bB}P|hV3)zb1cmXvKLliDDOTsLXi|!kdh#l*bvugf=ib<s
z(hJ10NPKKQabwUdLp3ic_{q<G#l8IOqTmi}njAWCu(3%)(`O7?MD`cVPKQ4!;aLXx
zr2>xTe$oV(a>#VZc0`GpIsBr)DNCAL5ED;5jhE;MJ>8LAbi4?uTvlCG1qHZQOzQ^l
z39&rL;c;mi&3tL=yxZ>9S`r9~mh;j#->k+jFF(K*{2MkK(IWy@v}^T;fTH5C`}3^W
zhl2X7w0T+4zyzYViKfG|TmoVE=3ahF;Ns$97lx0Y|5H(aqw4OloJzN_!`v;-J?Tim
zvC8C@>J%5`Y?)`sf!U8=|MTaM^vZX;O9~1@e=VMP<hA8@yc6|b9YQ<w&9n!_@@KjH
zc`+;PyKI6L@ffLbPttsOM#om|iJ*|sZ!5qksu~^=&!jFdBtQX`o_Can=5hC$_20ie
zIdZpiMx><~&Qr35X~p8j$n^6oPqM0R0$8|=u;k+6$^{s4YI<q~(Q@PIv&w;sOZfu>
z1Id~E?0|t0Vr}o}k`@-<xwo>j6Tm?_b>_@4)~9!Vx@n&=?o~1wdCgWnVMQ{2Zf=gh
za07r)<?4i)Q`qQ@oh=jLLV(0vhh<U_UPAP+=5%+}_;lZy-NKrX4>kPib&YQQguyh2
z2rn!#khr*l6`fx)c2b4A;XjNgjW_&@;Z{1}*x02Hn2Kv2KSEMqUw3C7)h<NtB}?sN
z$q9*xy86B`L~m~{E$=~mBy~~T$RTai$vu8MY@0{zRX7~+4G|X~-#;Ju)ldrQr1fV_
zZCbLH8hwU|(xijsb}l`+=YQAuF79R^ew;p<gJd5?dKI=^VkO(+&@*1bBSU=V`h>h&
zOy9i_)l%vn_;{z{8=p&Gt`4|)*d5%U-TvwrZ^zUZ_pdZ6T#vl${>~qrwzjL*GBO@I
z)KCoz_o^4WSMsX4%&vQS;W&~FyH@Wtum2wd8I}2%>l>eKZgWc3Oc<QY0>Ih^q6Z!X
zRR+--^6)z}selOhyM`>Qp64-_GnL{dyA1nwi}|Dc4CsaNI(k0@VXr$=rs}X1LCla$
z=}>$f+NgYAVJQtjp;1D{Mgz*d^8;~TW*&<hsaf+1#x({WSSYC}bT%p))wk;G=OaHo
zU_&*KpC-fyB29a*f@rUITxJ;ar3M|N>?RihPsxGT^X34;WoWSnp&YawkxifVu7786
zyw`1f0vVbT(=r)J%rVVR3Pf;(Y!|H$(yHs+4g6i#M-sdgm{%i7phtEVtt-f@!k_Nm
z75ey7j#W4TAuFq^ZugXv0%i@Iq|oDEY-pO4%ic*?763V^&73d^7Z*&|D_*oo{AO-%
z8+F7>%V1noO!N|5tp6^EkxN00Y<90rWUi=137>ut9dNVj=*?yEM}4Ca$GAzA<|r)w
z`@gzFC!<&eHK;v<w6fTB6a3uI8hwj_`V-*dPr7TIWHMskZhEq++JaIK0^#3r8t|LQ
zaD?;T(}aX{Vo2q=($(qewzouk@S!uI>PV!r6$p=?y{gKO^dsjAyYF^x-gGg0_xVd5
z+U4O2A_q^Wn-Xvn@huJ)r`yA+ApsLF3g)Ib{K-<ElP3@_R0&7(9FxiYHOK8KKZwON
z?8<)}+@M_8ulA8s843JjJ~6ltnZoz1Sk)Zf5T9zH8W3%n27Y|mrwBkoO9}`zIP94e
z&2PWoFlIODdHfiwNL|m`c>3R(ttjLm9Up^`c$y;eY;2&P<=h#{uH^(rX~W;Rk5${S
zUzOOpfWAsUZ+EY7euaNsKZRBPT?p8F9!T@8+?dXQ=K@Q*rn1z^HbF3e6GN8g-0Kf~
zQUdF4O2No6hTUUm&^q9yjGXINq;|sL9U{!jrE^jGIweLH^~G~h<9<gO826ofIAKzo
ztA7kB#b%fC(|zmwd*tc0^7q$S;P{Xh>m@ehqNRr20Qd7R=CHft-KmW$!(h@OhSEjo
z{7FOhZKn!^0@z{1!`%lc5D!!0Yt&~W1Ox8`Y&E!L#+HHb9fYypmox8o|DG*hu;Yyv
z%s-oH5Y$WkyM9#mV+-4aZ{^A`SW|`Wd$B7GoyO~{3OU_=^?numgFZRX@-m-G4AD~A
z4!@~pvlmi|yn)|QU<-^gYt)G7aauVexAB?$<NDLQk5c|)Ew}(N=Au|@Mp*F}l-^No
zsposoAqpa8`7Ra_8ghSbN={E+DX>+qg7lM=O7x7%ImFAhwpT~JTd1(R>0?f6ASv%}
zR_zbk4Hd0V){hOf9i*NBN0NUO#Ly<ds0YAroMI{|mJ;4KYxx{~=n0~hobg(Dyx}MV
z?=^4osa_vPq$gwWp(m;4I1y&nZSq#|Q=d`n&abwauz3Br@3pn7KbjRvsO5k1lQv}?
zN|v(wA2jT8))j8lf8v}Pm|pTuT}M!Y%|a|Ci4i&*nT2@5+0DVjefUh^Ea;dosRJJC
z$hZfXCeaCBP{lpGtD*8kSolLBpT|wi&8_!Bz$vx86;|ALAuYTYf@O}LB5i|*k?6%m
zunqDb)n9yZ|Hx0xkp@vh_9|IGQM67hv#bJbIbkViQmWLwo3Ry;;n;kpG!UQJT2!&G
z;GG7?CP<sFlv3TS4iK$}bS}DZ(uDo{!~+yyz?A4b7y)^Oj@){H+DYChFZF0e5mB`H
zw_9mLxR7xwwwd*8xrCUg=**xB(Ndy*>L|8w%&V2f@i8N(tG*$Jc10t_Uev1FkM~A!
zqrp~#@jCaucO_vL!Y*0%Ur8ooX6z;g3y8POTcJ(klTm#;ssW+>=4v)_;Sl}eSS^(K
zvO)L)H$`eSvW+KxU9#c(K>95}1a7^Mg%4;C*}mvs`f@M7N99JW&!+1IlwXKvWm<9l
z-179}=&g-?8FVn{i0=+-yJ=xKoXe%QnD2f10LC!u(=&Kffy!i5s{ZB%lSaheqX;-~
ztaHD%cm16uFV}>_&D7icqjqW$cv~<8^z$tKBA)3NFJI2C1@FDIcsSeE9Uz?iFvgv7
z&)YH}u-G%PesD>sbk3c)wNI2?Inh5+TDFC#-x(S2ISBaT)Gonx=<hMYU<#g|o)+H<
zHPG9=yTvUy;9OpQU2NB>?V`+dnE~fXK^EyeWrf-P1uWd5*V7zRmZzN_^-yL8I>11h
zRFH=k$OPY|y`YJS)i=+QuEeouY7%a|=*PD1%8yg`dU%XwZvTtf-VfSWpgr8k|Fa6^
zh{EFN^I<z`W5s}cj6H@d^CP4fbICX-?R%sY>!=@tCE~}`0;IFEo|UY*RtpvL&Ja}9
zgTp%jw+KT6O75pYETdwjbKB<1@89b^ooJOUg$iXM#nuZGctXIjt9TE)T3n0S@A|QN
z(|=1P$Z7feypj!_pkF{(8UoYlId%<Dcc*&Cud)FL(iqybGdbZ?fvBp~cC{=%HFMVc
z&{>K?B-oyJ;3_xe{0lHq_UrGK7LqI#ONgXUHqxc|!}rmai@!?i72uMlOA!?*Yv0Dd
z{dHaI?Lw|j6|H}>3obRbzK5=^7kLDMu1Dv|rbVe}OrM94iew{JK1#o_NO*_%4gYZL
z0jgm#N|W~{VE;}r9cIW`JHb6(xxBYEn>U{xe^au)i|N8n5H;O+<i5nlJ2#q({(IZ)
z2TCqR`;;w8LdGt6EWt;C?37nf)=Q&HlXGtn6cp61^eGa23;e3;`$La~mMDM`ZQ&G-
zdi%C$0e;ZqO8Uj-6bi+dV>qny0cuWi0_xe3fotiJUyt=aftvlEdkzUmnQu%o#$GW#
zTywq=d_l>XLgZleJp@U1N!k$P{maokJWyW2-q8V<9xoME2bPza{ubrPC{Kn53KqNQ
zVp{H95XFNu>+7!%5Ls~B@Y(3=8}=zL-7p)yZZIAnAK(36D1SbxK*28U)qvc9UulK%
zS}MQvj)~7d)fP_J_0XYdtq5NEXHgqc3wFKnCuaB7UDllB=<f_h>fHO<%NuY#7a8MO
z?zN1#h_R+BuMe(P4wPiK=~|cW$XCsG89syW9$!w%&K3a3lHkuxzIeoq(8d6>M??jI
z^wLa{%3i_Wy^MbeVV^Z3t79JS9J8@*SN2~omYYqsi;>z${fo*kuF1Mi-FxrfFVx=E
zvFK-*gG6txj)_edwJBicu881F1UV^daHQ`~WoVb*Z*Y-;;z75(=y-!!u%=({>W_fy
zrRu4pT2Ifc?G_`^GRfLdH0@nb^e*9hvWeGiWp-FJj4+6LPwgqp&TeKBHB%i--4yKG
zQ5$Cbl0tmfNb}uL5WRA_axkjMvfgiF;lu1HCZ0tc9AL|{JL%sDle(!J<BflImK=I#
zlRBGUN=Ub9x^nd`kC|U_C+;OiaVt?SoK3r`rJz<-Awm8TkV)HJ2wdw->`uL8=iOsx
z;SPXH!50J`u&Cu!wfNNde#Ga<EqSn)*Hue1bP;dKhUx}}7#Ei=?*XCrWr<@HK*Pf|
zI`jG7hl`md+};9iGYqcy#W!<O?`?0Jj+)Jn1^lbIFY_5mFMfL}EG<`{f)!zomBP4&
z;B33PadqJFeDXd;`FU+}u2M?Z942}iaO3)b;7WM&rhULJ_Du>Ljq?x^RvHO|Upy@j
z4l&&guK~Du@d&U6?(Q6Np3R)xb+da}7A>4Bh#B+zFRnEETwpiT6KYeG^#Q;@Dm)2m
zF{Kz!kokglm#k;K<Q`j1ZS1*o|Es)OQb4?K=@+i5*rqw}8&FT2d@%mkb<2(NFNWC@
z&)4o8@SqfANT=sWZ#etWUq5FH^m_^2ig;uin{kjy&dTnoFfX>w7T$c#PI9t5T@!35
zx8e0S3~q)M7;WZ+I{%{-r10d+QYldsB$PbohUSyg;e?cQxFyo*0%<>hdedEnY-i86
ztF*hg<!ctav564x?!Ik#_mbB<>CHnsu!3q0EK^hk;EqBJ2QV-kAj5F?%M}z9C=vdC
zTeqEDF^D8Ci$}qi2kCU59urqQe;T=ce5o7Nt+w<m=Tp=f5sJX41~wi=r{;p~Ftx*6
z6tO%)wU(!s=gR4eg&-rRkTPH+B=T_Eh5y#m!*26or$PSMky3$C`Ni^KWn?0bW3F4S
z6wN)H%b$N3!S!QxsN$Yg@U)M}<;z!}d|clc_g2ZXULq1z)Y^JCS6u5$)-Yp~V)OcA
zL%V&`lMQhv;YB)Fh<+!IL4+Hh$qxRDh>}wM;PFPc+g4U?gu%@{)e@phCFY;uH#=t+
zH^uuMtFQY;M@KPF>|jpeqGa+=DX{9xd9U*-N*5|8TP&ZagdBFukMLPxnRa$hnosxF
zN)P(Y@~0)^-~GeV&P^vd&W~`Dog{T&PFC1GM0-z<+{OnQCQxCC!Cz%TYH<Eu*Q?=b
z*VI4Uv%5$cyvSHh&nYHZ1iEMo{+~|9GJ_rk1UsB3XMFSW@`N&f<<X-aTl{?md*l4T
zMdnMw(=4DN!RKxta{~Gnlw?*`W{;_@?JLKJ!&Ntd@PXMWB!_JG1aQRS`}w7Hx^BCa
z8vF6TCFAfiAo?lfkNzg}mK-wIa(wISQfZ@*U6EZ-Fe7e1O!6s<)6RMo6r)#Grnkw0
z%$;Ag5E1ns+cDL}LD&2|oQsR#U0KsvVSv1pVrn`z8MTQAA?mV6t5Om9vPL6?MKj>o
z4SXOCD?dm~RWIGD?lDBn@5&cf6SUZP<jG49Q2(XY9RTmi?kH14<{VqyX`JZrksEhv
zJ_{4ZPc?PO75!vWtW;RK(6EHX2wf<D$~h6pJ-!4m;E0Dr6;RIT`k-J3S*@lj5NkK(
z;a^=>P<z%7c0nga)+-Tz)jd5$f28YsZa16rW5$=`46G9)-#{eQ?HuU3HCvC8B&zec
zBddo-d_c684p0L)7`2>$&kZ{Plc?P4n)tVKwjZm-c+{1vL;jN<pKX-;`YM0_z4h^)
z53cAm_w{|n3Cer}WyR$QZO_3$`eC17<D{N@>}k*rGXk2Kklyp&ArS-QuQ0s~L}C4A
z6jdQ7=>`k>Lc<`@+|JIfG4;#_r87J{e04Slv9^aF0krKj@_lfl2y@Yf<3URBhhx{&
z=PUa3wr7O-X!4Vh)j(fSzVLT_CaG~Ztt&M+n{(XW(d|VvGW&4!>VymT`qt`@*gt=&
zN`SiKj_Y3D4~_FAN#?W_9fbMXC&SDw6_TPaCM_j9lEVRLJt$@T{iDGLS<~7R^e}pq
zDmXUeoV7qSGnu$GD!wLRDjK}Kp3#lDhj#1p(8{N5vPsbTSe)2^z@qS8X(CX4;rP|l
zC@b526ENMEKf^<l7tZPE>PxqQt7E~L_S!aPmI+dsc!)7@G2lV~PndOMVi!o`ZkG_z
z67tqT8}$=@r;o%$Qdf~VNav4?TzS^{NIMWl=M`1zvAyA4<WZdIG?fE>&5QHFPQ!qm
zpQmoQpzx)XrT<B3Gs!T?(DP@~Uw_%KR^_bfwKnG0Sx+h3Js$k$CCh!_qNvE6mel3J
zGWX*gdv$Y(X8?mN*Y;Z0U}W+k$_gLi_;=f`%i%J)EtfyRYA4(P3^rthoy<;V0^&ku
zZqlTA3&#r5tNqWs&sKWxU@^t+VKe8v&+_^E3$6=;7{i=y1KED{+xN2r#l6|Pz}tKm
zJmy&F_`8JF*XFoMPT1d(Z4Ns{Lqhmlx4=QFklQ#(x@?(GG?eKA+_DV4o;7Ogc+~;=
zWd!sJvn}xu`}mTWsMxG~@P36w@N~2#{OVAao^`*=u4i*h4gz#R*YXrtev1yybQ&q{
z(ijmfbG_F_YpKUcTfG*Dkmza|!7;9l;^=ZJba#5yxVKU2IR=xvu!@{M4aw-}JdS{#
zJosVg;o{Am!ctvh-^8v|mB#w5DE)$!F)TC~)rE_I^k<(l7yJ$u5Oe-K*UXyVg!tf?
zN6wJrBI(gHXjkb*55}RbjvKVbqNFP-X2SlgcPqji^yJmoAazia54f6qVg3<UhOvbD
zeLZYF73oEtO-l9Q>iYb-6}JQb4DrDsyWn7iyi5*hbpcw5{!0q$%U*T){j$KO+=RMW
zy{-z^z&AdKfUj>i@A7j^?t6vRJIm)FoxJA=J!2kE0gve0J!&t<(f2(GdI5~fk$Hh7
z@b29=g2KWhnC4jJ){YKcPJ~MWQRrG#Wp#4=NbPrdER3CthID$tXzn_+yQZW0haX@A
zuR3fu7)hOAGUVn*BSRct;XQ;9h`WHYU#^<`!h-I*kq7Z7MA^R!?e0K?mz$q|O!FH1
z3ji_PgNf!6R^A1If<ioO=*rEgK!|z|+5M>##17ytOtSf|Wnh~BV<3Xr#Qn?>CL*d3
zga~#dJNpaTqa4@W&6RThmH<=lKeIwNysvZSi&(WByw6T%HOvizcXxN6<lyEwG1;}y
zo5MLA_G*@v?msgb0pP@#ASWj$(xvC>M@SpsA;*BDf9L)-_PGZkt4vJ`9aB9R#=GVr
z#IO9x>R8wr7<!-~aCc^QJ|q5^gvld~gWaNAZOz}IqejND{fUf?Jw5_-5OyH+sH~~E
z859)sJ^gQ$|F9k%NKtFbKYf*7AxET5y)6G3rx}MPN@8j0WvSU`q0HY8{acjwN3jae
zMuvxTLB=@oilwN(zdr%}$)lj4M9Q|!Bg1gPP>`GLM<EdD0YS(miIU(JGg`qra5KMz
z1Ri6+UgLqYUWUW_t@<?wdF4Htthkgx>SS*JlHdhSnT6?PcY3>dsl})}ih-qafv!%k
zUR%F%FQsN%8wI_!FdH!gZ1D7tPfrQYOE)dJt`K{M)wP2i$4gaKZsqh0mKhzux#W3%
z3r5k}rwztc+}FId&!h#Xk9z%Mye;!ETjb~kG<PHI+!;H|c^^nGU29He0NNBBE>%zr
zj`_}r+dHz^nGBMZeYRYqfLbEj#%N^_-Ytp(>6DTHADLsX<DG8BxKbuC6@gI!ClPqA
zRIXwrk)Wr^FFKBTHSVf|qu~FT*AC>a{Qm|Bn1IY|=}LeVo2hWj&URmac+HL8bRxEE
z?<edNLmt1Lt84KD4z4(=V|~_$p!eDojy_4|drRBPD;^8z-e1a!n0b@5e-T(}Yr)n4
zRO|Oiy&`TB*$cYI9yOVGe_2_X7!!H;sU1KQqa8*tb@Jc<MfJAukXTQzJmWac?ep1g
zz)EaqJCJu;I~=YK&bW8A5Tue(qAvA{t4M|&5f1w`Th!kPlwbZcSx07f&b!N$-ddYP
zfiCEpTb>8k4gR>klp1C(KOLAo!%d>)`{`dMduTDfh`NQ02vvxk*pgZd0K{wq0!=Gh
z>Vb=t1*?Aw$(mY|Ux19{zcIg}L$CeU4P*e1hHC7CwcUbRz9E-QAH9hek(8AC^vt}O
zv-ow+)}S1|mm?spy<M)>$;s&z5cF}Ve&C1ro)+A-6>xC6cFUZ4Ql3Vz;pkfB@CEOR
zq;12&yMb4BDt@IEx2Js3IVP}VH)((N6M1Awe*&|^6}a#u_yBxRs0}+o-Z+W+Up|Gz
zP~YWI5lCMceO?Zi^$ZM0Rq%XSh!%z<4<yYoI&b0pVSvM_F)7)sB%56x9{H?POXgsM
zDOGB{I3M0F4=FzEV@S)tOlZH}-1C0s+@P5NL^mZ%?vs-?KpB;)thENT%TtBOOj;Rm
zX&fZJ+=X4Ek@BsHB-GMiyXichLv}K1_@l#1s`FbUTOr29zJIQpAj}Jf!d{>yn9125
z+N3_|1j3Ag-OUTV@vLJ6{*VzKpl5xu2gAH8)6bQ8?0Hi@hd(Nv_GhUokC;=*?C}_{
z6rROjE>GVrCfO)BIQI7tM=nmamCZ5Xd<<eA$jVcu^CV%x!1#q|BbI@T@C+01`XC}3
z7urF%DcJT+D(~~;_^H`Uc(hVorxI;Z`rRos+QkS7g+oAc%I<=<(Ha4e`%h0rPsg8Q
zXBQ6}d9@B?l#Qp3uNl<zrsV(4Qk4TU5EyzN)Et`-+646DlcsX4W4DRjl$mS$<IO3_
z3&|G)PCXM9#!~^&vgd!B{ajWDi_`x4)JxCFY|Xf&r2H|5uHD<Qa5t7vYz}fzx@e0a
z2SWqSX!#)L{hQTXl|$mt6CWbm(yx2$CV&9W%zB0%BXg+pzBFPt;mwx=ymQO3O1~&~
z`5aAe4vv$t%-+{eA9^xryc6Ulv9bNTrtz<-1aaBZvUki&>nYCT(TaYmEQ+9?<`Rng
z1w@-(u}UNaKQxi67s4-PvgC(04IfkgBgQ=XeZc8@$oF$wr<pdDKlS6_r%#`bd;PjM
zhwpt*G^E0v$E5>OC47Xst8a%Jw8EkYJvi}xf`MG;4(=O44JA*WfGG?ONqf!daVq&1
zDab;8==FoIj5t7ej?Ov9CnTK0qRw&yF`T^&`MrM1aut7<V!annj=SVJ82Vjd;UzJ#
zrRV8zE+?n$8G<cp|0sjfB<rCkCpu`zd5cBNCRLax>88p~ZwWYr8~X!WZ#f1)7j(>$
zfLE<8;d{Sf5b=?A*Gns+`H`0kDY?1AjTwDqxF4N48D$=hpt87v1C~<E;%3{-Sy*gc
zqc2-(Es0JKtTtT>gX^o>QBL$h&DlD9W#E!K!&FjO7(b9OVAYZW=;3->llVR$KKumi
z!PQLgJ6*p629q*)agz0VO2ZodW<PeWU3dm~v@4OD1H+0JE{7jT@0c-=pw8prk0yG~
zH_9CdVK%cX_b@z8JxAU5SkR-M<)yKjj4vuHWx(1XS%ALoa-fyB>}f!djvsn*x&rYD
zr6+G!1>%2y>WBH6&|?x;mo|`#V+Sxh<fqI+NNY-Q`zePF(W~d*(FC=zq|a?_yeF;$
z6ah%bTw;oMHR12azdXX;kWORxk&4#5e{WNZ_+(VLV;6;~rhhBltDto3p0&TGzUh0_
z)|q+Cy_6E0)&^F&9iFkq;Kl;(Lts%f>eK?Kw$h8+!Hb(Qz3K4b>E={mKX6wjubLZy
z{3{hPbzWve!Q99pudtbuSi18FP>YjIVQ^p7^#%3g?WcL``|i|whH?f^Mg>KuzDY><
z3KWLzB+GjH*>`u#A5LpwAuf|W<?sb^?Zavz<io-HAI!8VKrGvC4ftYk#{e}l(B)?m
zAI>>^?Cl^9E^&dPp=J|yERfgNtaQm?t+Di`v=`+H2!38NE@4KohlL|jTW+i<RFqZ@
zSS{YR%y|-=Mudt{G}7r0$+)E!>uhk5DruHpttg;P`*{V74CtplTDPR?r0~7FdZ6v|
zftHw`+l!r&onMlyL?d-{EQ#plx$(=bDuKtBq$o)|v7osS?b8lv=V%MU6BLT$(XSGL
z^V~Y+`X_J487+%j(gd^X2*XVDZVDJDXE!M%P+q2#mrE;n{0yP;Pjg!&w>cq?aG5zh
z>b2$yV-j0%+Nw;=xaaFzO(eWjPV6?9P2HV6PoV0=DZ5QvZx8HMMC@1`etIU`cxMkQ
z#(eG6|3~$=KKiIpHY6)I!rFB>XXZziy62x~$7VkTs|fiYI^;M<Fma_wX_qc74U0e7
z?p>JvW{Yaz?*^*k;w_J!5D54_KLx~q7)sF_DZ5r}-Hyo+tz+dbzwQ?AW(l*pO<{w5
zss)e*gJdDa#h)y-BBvg>_VaK(0co=U1du*%_I>T=`?SlXf6N`1u1;Zz1q_pAsgD^$
z%A~fn=-s<_Z`-6@JKVB*+%L6>ET`9Ri`uE(C}aJ?a)lnI^`ar4hFAua#0{Px6XPq^
zLNyb=yw-8+XUSn4#cDY%c>F7^;0e8SnRW{(mm+Yhl)ihoidWkzC_R_py}{heFU_So
zV?STJuc7vG--S6xEgnkoZ!iLye?xIu?9@VF@6}6k!ZkOVGF4VVl1nc+3+cv?WnK75
z-&QS)=xDz(Of0++D}gNghEw7aV4Hm}#K6URHt@JVuerBrONwvo4P{?yp65tS<E8bv
z?j$Fr<lYdi7S2FexVh>yxehKu+3j-8QUQazI<Vt2#|d<}6a_M%i>cNRmzECmkiz0}
zp1}jA&N|6+ya3`J!w7iO^NR7L&Lna8Jv0gLAhE%m+m_`4bNEPoC3FzoX7U1wI_7AN
zN2^>dKqZ*oh4%z)X7popq36Xw{0oo6Zc~Aed$)3B)2w#iBw0&;dBSm$qaOVL{7^i%
zrwjop_J*CmGo{bafOQlQz;mWa4Tqg|BOuTMmq3!WB|d(~dh{H#=FWK0+qbWh(y{ua
z#vcq7`mvP_{Vw%;+t8K#u@F$KRuRz6Wi)-!T`yfkXvq=ErAYlbhVIx*N)f1Qs9=DF
zZOU*V11f>|sgma<rRmq+%lhyKifP+6Eh+j-8HZK>`2X3&z22BiKi+M~oK~!ZK5@Pz
z-q?j~Be;-tZ>i_nF0KFf07q<RTJTN<6~V!AUhJW9zbfLwOGYS_^s{s@#Ie0E&+)6l
zdqOmiE|($5D_nS(UHl|GT;>VT^rjN#gx2wgp6dTfOMdm;jv^j$n605g9+seAmY$l5
zk&M2QwxOx64$BH(&wUeexCM2*p0tdTv)WSluxOMKcjy>N=F5{b4=roI;R69g$X7C2
zH9jfn>;MnHdR}GafP3k2m{VoA&f~x20jXZwJD+fl^$?LobI$xx_`fQz-ua^USG6nT
zfe<rX7Ph5l$|Q46Vm!he=2Y!AB#=SeiT}Or)pIl4r|6qat5RmKhgM^&wRZnVKn<Bc
z<;$f@j?eG4zP%M^ZD@1!Ff{MiFW?4r=WMx3oF4v=J1dzeR=S^HZ*QMGiP{h7qMd23
zRlwgmFY5;6Cv3@9QL{h|9kdtV?CxGu1BKJCBS5sRm3&wR6&?NiZ$-~noqT+@^Gg;B
zy_X@js45>HLSjNGutmF2eHObvn7GY3|Kk|Xg}UR)CK$^TDRRLLEcSRDoJ1R*ywLZ#
z_4D_^fW9nC@U&_!(71-2$gM)lWcshA`ORKTvUH^Q9FaM}PP$yYGGiVBTqGWU0Sg6n
z&#wVi>1x#9XX1*N{x$Lj7?MohXDYq=<@Q?jrH27dzV{R}rFj|xCVFB(vPP!?iNem&
z4`Z}a&l~~S1wV*~kj{FmEfXX4<;lg&%_upToj0KgklP($zost%t<N|#Ukz|!>CW*B
z<x&@3oz1{nk%&JSENwwFy00gA2r6=<3xGy#OZXJ!Wol|FvevMdh!5DII&KQN5;mi%
zka6J$6!}6Vv~D3(q@87nWqm8Fd&hZrYnYUN4`(Q3Bl*gKbhwH6D$%Kk(2}a`zf$Fx
zrX@|-it1q(&Tm827QTBI$F1bok@edX3&oY}AO2g}2yuX@93S;BD^zrs_dnHA$7xsi
zEEx|~`d3_#_fUlj8w-s~t81WvzW#3rq;XmY@I3{`NC~&7IbDkR^Ln|bncYA=(C-&V
zu?081yx>SkXvW?Pb&(9!iQd50f86<f({u96*;F%``+)nod0M0ecgM~yC$Km_JpAzg
zd$*l`W)sn(qGAr~I~&p%624>@zDXu%Ul^AOpg94Wr!tAHv!}D7^Z?00rz!lk(S_}&
zRhaIdgs3RievFnKCwfyTUo$RJQ!@^Sb1A`?z@4WwbhNdv9%Kjw4;Fli)x3{K=AZxm
zhmuJY#Rp~=P^dKh;JDxL%eh(wubG5+gG7H}JfA-e=z2l1A;jR={@^Cv!40L3C$<|A
zl6}I*FhYF+r0o}xk!AmlCK?(|<kn9DwZbvgN`R|Pz6$;K8T>WR_k;C&aq#hBnc=4g
z=@T!>+NQ4YTqBH9KZc`~rkn%S39&mLYz(Mc?5KBdgwzu_Q;K$Y7D@p4s&bbN7t;#g
z;w$D27P`8+%i`b#ReJ%T=L1P5TkD%yKqoXcJ-w3mb6|X2f}5Lr;7e1c0ufCK{P|>N
zZr1O(d!G$BI^|fhhKpf^>#nb}o15$)k|!3_@ZB>rGwV7K6h<;Tg@*ujy{l_ndb(RF
ziwX!T9=!&onc&AXN+Xb<K->>!5CdyMI9w<b9PF*}z<bJ`)I7`cAD?nooJ*PP0_TjQ
zV_TGPKw=1-LuWk&NXf1>ie)5Td`}GIx<7pQV7q$`fd>p?5)g_7^w?-St)-A+*!(fS
z&P%U-=IB;jvk|)8!ucjW(b0NIrv-{(`EH*evtAbjI;k#Dta(ENR0!1^FiA~6T))3p
zV@rIa)B8En(R4z6SV^o8>vRcZN_~wJi367a<>c918^Hq_@c<+zV{X++0e(;e@-CsM
ztLQm1y05?4?>U!N47^eY+w)+!_LlN~0Y*@YKMV6H-r%Ku8xOYVf{I^Bl*-(>6sf?v
z3v%cfA_PI7>w(5>nmG-&k7M2ymc|3k$jP(3=u((uz{wR>i~#<VVc#Tbg|f8tEy*e_
zr2?JeV|LCYqUX!!R$qfMZYn<tlplf|o-w4~Jvc-HXX#NP-X=yskZJGm%)RuaIP+-V
z@O_529zYW@AGLo4E-ihI5|%YMB=QE>R={wjAhG)Q);K{zCyC<tU)fFrvlD}=+K-<<
zEZOB{g(z0XSYhaZmE)9>H_jD~DeQ(%IoArRpbOLldQ-W@E`JS_khFB`m5ufCFM)am
zuO5{~OddBW`BivGD|tFvzlI2`0#uZ%egnJ~qOGOcfBf}H5_szzSDG<)H(GAyyK{ei
zGbd)~V=1j}OkgZg+}wVo+V!^8Eq?6g>YFOw$2KwQphw6`D;$NP?`PXxW?uFTn7W~6
z!+LfFF>T>8QWjW$VboeEUI-<<lvYqxbp%#-wD{iO!1tS$*LY&@7~dU&C`{EE{!HrA
z1C<{0h=<V0qYN1N9Qsp-o=BaPc{fz&@Dg8c1F0c>pGh09BSYo0^aWiWJg6+)7$B2v
ztRWXBaGV|`iMK`NNNWM@Yb=lfScPG`A#oLa=?rMc;^TyhS+HaG;<xf2T@`i4CL@>G
zN8|MiiznZd{{49Ef34Nu=V3TF2v}H==;!^NJu1SWBowhk+ta-2l}qg>q}Ua8oCEr6
zy$lIZ9P4<1Wi^;&lm+tmu7YffdSU&3R}8nKfAi-^<O_z~hJFELB~&BR9xv2R%}*18
zqe(JE7cgFEy{9#g-2>%FyZ#X%;BD3J3vuk6>xLLZp;wT0purn*m~`6$nPn*+tsMI;
zxaPoWLQD*sBc}PeS3r+b{|>yZI}JlwD$`KMt!6?nj4nV;)egw$P~9lPs|U5|8k)@y
zK;4;EaAFrF%qD>P0qw9@*5)hz9m?i9B{RMru2bx<fF#n-y-)CxHLiDau5g2B$L5jq
zk8)7N;Y9?B@$8(O`qaiP%i4ysyGw@tA0S&r0Ws~Gq`469>X4l~bnFC-G>*uS^t2|e
zya-iQZ$g`zu09Nx^)Yb;AQoaJ9ElAR=Y|LBfhIk&ZMrN;V{>Ih82T!Lli+1`e*swt
zreG661%8B&yw}^mucEjRq5{pJvYqd(qLRbP;$3O~jyvEor-C3jyye}ycL8DW7(inU
z4T1(jzH+>#j|S<yA`xL>;Rgl~qX?x?j~+XENg^?{X)o`|@hUa;%lT)!-<=k-AEr^~
zYNu%FzuXqGjUMd6MblnEd8yZ8Qw&_ltihiG_yhejyr_V$v59tt<px(SD>&4!HQI-p
zDAda0A$r;;r7p3t8iW5aP?ln}jTX$;(y%ZlP<dFHRA(Q>bt&gYY<1FqgS!O{uGrP`
zlcszGe7Hg*zmILOZGjcTwHVr@IFnso4oV7c3j+4}l;5G4Rw~j6)b949yAwgZ%=)-p
zz)Az^oQl<4PpWXdkbc56$T|a1GHia(_9-cJCdh0r+?kx@Azj`xW%B_8MT4wo_eN3I
z6;SV)_&xcYeo3$7aI9m?&W2})bAF1$`n%r2m~S-1=cTVID>?vy4lcbC2F>xsY4`+e
z_0~AKCNBOb2BEs8Cnh?&Ka>>LE)ZS^vX?XD6{odV-{rRqI*Sg)BJ9t}`Hr3tehNyk
z=QO^qR7Qj<>vL0N5)J{JODHJNiQ)8%(!<N&l7`<%osY7TS$N5#j5zfzbkzFa;ys5+
z&9m*C=yVb2+a=DP7b`RFUjiTw{D~m|C=McaijE6`Q|lxfO|r5<SO*k$PB|-i3Z1Cw
zQ&lry4s`yRDg4@P3&8#Q2wvDBZqYmhz!{=l<@=W!ovy#OpSQpH*aFXjmM_z?#)q`_
zk4xnDD5<UP%|8V=@9W3oX(8c*kZd$$e*uwcTEEHI-q{1lm^;{G>=sJCtLOTf3?sr;
zCK@MbC4HfJ!{yR&s>0JCKu#XdGOu~?y2oly$KMqcQWc3vNR&W8J3M4gd-(h+IZOSv
z?O>qj*B<m+2m%Ojp|hTAD?(XzAQ8)fOY6CwZ{}pW&WIy^!OjjGLgO5S0Z^6iODr}A
z`eY)3FaXu)?1jS|-$HeO>;>{1nxB73FfVhH7M7L5;0uy4N4<)Yweb%(W73qziU#Fd
z!=#zuL|_$VK1Im|TD!zT#)I$nL)jP4rz|0g52W{LzN(puN{aUD1QY<GAQ+`(V_1Tm
zx<M839)zGFM6AII(janj_gYVqY?z&n<NPLqZnnm7V@gsz!Uv5qz?XS+?UL>V1vLah
zpcBH}%veR{FQ=H4-%o_eOt(hI>u?C;UI9{yagx=2@%=F1IxF<C_SC|!AJdC&S{VhV
z0%Z^N6X97x8~BQW3UrUGQQUt)NXTJZgZw_Ae4=T1fb^^M!zm4~>u(Q*ZiKz!aAwtV
z(0m3G#q~F0*{zIj<X7aO-jBzETAMg1Sf{-^ut6X{-Hh8z{v~G~-|uz!+~pNDyFG)=
z<)MphQM?mXx1Vy>J8P>OLcL`+%ByRB_jDm$4xVHc?m2e_=1eKF68DNx86U{FC6rtS
zmCMno2A?4FUGYB)DF|85!W=dq*P54CRCG&E>e&pBjCB8ONlZNRzv!FY7}Re{;0y&#
zQ&Pg<=cM&t+rYnU?rg+46B`oz9h|jWFw1WN$eY=7=(L|Nu`KueCOTC6Fz0_^2eTO9
zBGG7cN@y-{2!>+f;)o<o%2HS65mY1K>;pHw=HHeSKsNh-10Lq{K;IY>8=KE}HCz)^
zZ2?S}x}<;AE?i+qEmK4=>LMru#%&c!1}Yyo`7fqo&CR_)^P--IeH8yq`L9RK1r_)7
zl55v0+Z$+XE@Y#CfI8~zTqma}<W4jMNEVX;r<k<fUl<Km{eA3cmyZ=1R6oFu70-7o
z<+hVL;%I<T{J+tc2jc%ZJ7M7L^z{YC!5yQBtvuuQ);A`=F%ny!?~do}&rj{Ni$zlY
zH=L?9#$Yg@bXN$4ce>663LwKuM5<L}N|6x_Q`zTk54sP>JJJ?<3C|%ynVl*hA0p;_
z+3P3y6Hf&qniJl7iWzKB)N_(P{7ZGiT!R0XUv<#-T~%e0@!&qna4{Lr@QD<q{^!r;
z75!VcERi)`kk_GO3a>D&JqnoLqu!5<!14N!ZbRH;-K}cFu!B#VRB4=K`Ct)3@AWJY
z@&I6@Hv92R@xOL}gcTe<g_gi5m>yc@uK*Zs4{9wlBfBX6IS={-^aTf&Mi*0MUE`pL
z`Fc^neO9J;_+#+p77c~-dTLKtoGbT_OtNad;E~vP^mZi$ou2qj>|VQE!=CH<PJe(l
z;3RGqKh53!T?f&KWa}GDGKyi~O9<3-x8CSFn-&aw`}QlzhAA!lDk$WNd-LX!zuj*^
zJMVBoA;Dx0Tbo>YdK9SExwhK{hrVbbV}BWqVz+Bfsg;kcvWoxDzB2e|sOAXsiWe8(
zHvd1Iy$3X0?Yllaq>!Q{5<*@jx(K585+ynbNkp^|j84=r7!e_&M(@2uwCGU=sf^x5
zkLX5;!HhP1&z`*J{LcEmv)1=t|Gie$5@vRJ_S5d`y6)>Xvj36{`Z1>D+cf!U^#|^8
z;5-_D!3DW48j8L$vhdr#n@*)<9xaCmA$p!y{>cgJvcHe$kwDqF1C<ZBLj@L)-kL^3
zy{?|CU4ANC0l^{hRh1l0ebL8v$faOPihp|JYTq{-g=hGT&?<g!M<|LWAG$;m;reKQ
z?DXmQj6GLu*wYnCn-b4Ck5Tk!^A<(&?dKL=3UfnQv!W2i$a2sB6i9Y_*Y$_fktr(e
z`ETF)rDL)NQigTOXCXjX2R;Hjo6~Ds3rz#y#UFu?i$x*+;lssBo8QX)%|Zoj;e*g$
zp2CQKc^Ds$BT5CXG5q~940h)(Nb!2v*Rc91RnFdX)b&A}GUD$?$U(fn%$2^QW8|&!
z`X6EcrrT$rs3Ra7%045Ja$kPh3$`vjeMDtID+?g&6=2JvPzQV}q3obA2waqUW$|e;
zCr8_h4XJC>D5K-8C}cZ`#Pd+7Dbov2+52>%W^8qNvpN)b{*uj1vmbW19BBel|K<yU
z;&Fa)aop(c50Lh{P}f2z?aGullLM!ius>Xi5TB0DJeucPJQgtibPtLY`B06U&4*pj
znpX-^ExVEgaZkWgDLj~{>Xl<FHYkg6_C1`qeDx{^<Zc8kPbwf^Cq~B+mGgwF|8a-v
z&wOaTxcufI8MG&hh>C*(jYF!qbAo+2>QPzahSdJ%$+0uS@#G?V)_xaU?ioMwcssN6
zC&-5JfW6$Jv>Bg}ww!%U2Y9m};Y-*0-2dOkQrRf3>!}P2F;TwC7HgTr^t)S^uJ?(E
zilX0M<z>Hd<4&af@e5mf`&j%mLRU{;zk8Qa+B2!s5QI9W{0xC$03`c?Fp-^uLq*Kt
zCkr6{K{62AQo2_LP#dI9nASQZuiS9Mjlolyr0hn^Osv%gChOh0)j!AT%JFzjc_c#F
zt*#sOOC5n{E?9QN@hr0U8ra$vHSLZVEgmibKNOm8GbhJ*SyBcR!E45^?y&r4&-z1<
zoca5E*n)V=euN?N`&EN>+ro5fCcxO_Mm=0uT1q{P1u=C3$xhy;Y}9bZHgHBEi^%w;
zs-ckEA_geJIh}v4vUYg6DX;pR;EnK&IXQ|T+qL_-z?9S@%vR!Gm#l<01h?#9#*?EZ
zdF^{S=Gs80aRfSqs4P$m7_x$WWdfD_(Y@m(<Z;_ApauX0Fc8MB=oW}~$l@p2LPJA2
z`}V&J_?!DwFNgT;exuQCFGmtE8&)-T!#DTmm``lJP*c1BNn%~<Q}6HR=4E9mfaW7)
zTx;6yk#^ng7VlggF6ICNgNynH_?8nsplQG{S!^_p4D^3gch+oaqeAW}(I~7q+6Cm*
zmCqQ2h_B#>VtFatt1}|D2`AuA(NjPT1B|3_e5d{_&W};-QZzg*Q)Td@6nXyy<r9(Y
z<cpn0Z=8oqNm(sLrgJuV`Vf{kPs#p(2F+(>q!Tl=FCghsl|Q(@-(k>fp(DoxynC2l
zzQL9Bo2wg`c*GmKb&Fd6?CXf<)pou@uwU!9VX-*2NX_5}Ayyp7pF{6$&2^joLNNk;
zM39se`^xp1z2KsU+n`zQg9KLiX1~f;c9eYP-Z!o;N5ZN>ix860pRhZsZIdhOeg49Q
z?)5a}@xBne=byn-Y6!xl%U6m;dpXd^ov-OkZO(IT0+W<xE7-BaOzm*9XVQ;ihBv_o
zDiD+d2LRp-&nQW!`q2MrQFc}*1Gn)9+A~3JYs@s9TFXoy@=zMpwz>?A9rJxz%7}ak
z_cw7kwf+oVrG0~+zkgc{ib(WN3rB|WH@t7F^f>|ga&YZvkYoI~Ks(jiXVNBmC=mdw
z02&<C;&wY+uj_QIo3akJUO9Iyz*PB9XjAYth@cPpc6#)+@VLI~1kQ2sF+i??nnDq>
z;eZ^q2C2poKcv_CZYWNpOsdJkAe9O8amZUrlYxS|&2q5_rH}`<<;)b1tM~qlHseI~
zi|;nB<2{j-j8c}t)+XKy@zyqpBOUcS1Df4)Yq|m~FcYob<g#M(Zu;@OHrl%3BB9x?
zT1V^0w!UeI;AhZ$QVHZb0V6qE8$4B3owULwcP?1ruI8QAz)S$$%lhzdM(Pzjp7OrP
z)Bah_DEZ@0N59DAZ;RSTdXna&;MIRVj(q4J7nV}i2vJ;#=7tdp{RF<CPF~aWlo&RS
z3{|?)FpgAvZJN?3s;X}BLZNMTm~Qc@tj|9{4!E{jZWX8O-mg~$YKuE7gWYB}QL?zl
zG+7Sg)#%f_2ct%efq7NplGUv)PEKX@5!1m}R;UBjV>y!<Bzyw1QRcujg!UYcU!i2i
zD8$$bK;)YzAQKw*v#fd}B3Mo5W8Ua)kV=b=5(~<Nk#=}sz@qllvP#lu($)fF&nL&d
zTz$8@QC|NVCke_|{v2g{8_eO{v=zK>|4~Q1^d*2{TO1T4g?nWNPjmN3@r@LFz;~3I
z`R#w@R56#<=seEwS>>V0m~<c1F8EU(%<&}Jpw9lCDTS@O7)PmY?u>rn&`5@ul=6qX
zyl-99`G?E+VN-!=rNT6rihZZei<}wf?i5jLW<FBi9?7L7m}<yBV2U|;S87z(fnDg-
zHt?S^xIit9ve=Qs;K#w%RB{%CQ$kS~kR2|Daku;9ff-4jAk1^g+xpDImepsZ86Qnl
zI5rmPiUJ=mH{(9M<pcL>Wus9KLnDd4wqhlY8t#vBrs;2I_;!-&?D70=sB&}d1~ZV_
zDLHM|%E&J-4c^Zu-MC#68P;MpJKf9al}|$0ZCkUE^dqgxr)Xu*_0#aLSLCSp?4D{m
z%K)Ul*x_>j;kA?fUgWfI(x{(1GQD|R+PI&I=b&c%)m_AU8(6NW-HyG-a?F6OsRuJf
zph$IpQC)6ntc;mY3&^RPR-IA}yN0Rs*F}|aT>(N~aW1{-*{}>3i6|As51Z(`LmTQm
zS-YKMi5as~JSe_&3!@g(P2qry9oglIakpNXXJrCS%dFcO!*g$JM-bYHUuW6Pq8#Jn
zO=K~g+8Og7vb8;GJUG)<dOsEbHB+*jV}plQmw41W;!f%F<qDE+1)K6~Tv_{QU4M&U
zZjv|uFm%`;z*3c0Tkxpt6IR1bvE1BI;*1CY<G8`j10>#@Kf6GtApZxf@O>9UT?fxy
z2*|E`cIbJS!gsOKsT(62FkbJLn0(1;wL;&-tz_`9?0YYsb~Apb+WoIDpvT_n>o7(J
zcT8zrN6o0I{TCplq|=dr$Tnanl^7b!nQ|l3>FHaN%}Lu}n*#R2z6;QPq)Zvw-}J>I
z358wc_Y3PKuDE=mPcKr(xY=sostfw3M@!qM1!zOoYruzzB1MION;lUv`;9^X6Hp;H
z33zSp+}Gc0KGoAwQSBB8&p84U#-T2v{bInxefDdx19e2Q&9yB3@sr<h`FQD*9k_6f
zr-O0Yw=rSK`XaTDOvj?(Km`y%5wUpOs1Af);Q)Sn?<0KW9JoP+_mJ~~zbk19Fk|Uh
zlJlr(YOpMl%pQIEHG(8UF5^c5ry<I&rl=x5EzQCCIa#T+@Xtdn@JwLh`%KLR_H=I8
z4G|Nulg-u8-y(G`8|9Q&?VN0DM~VAW$P<&i@0)sTmlen$=VUCXui(u&E3(cM#q$y3
zD_K4fcal22$sam24xi8&1dEJ|SVCW=egO8AeY#TL)W6cB(&X%G=~pdDDW`1kkV76U
z;3ag|B-hMD?tmSjb(uX?9A&XLUOC$U+UN!-5VwyF7tS1JzXKB0+t;B`g!tQ5SeNYV
z=`oaBAwAYok-o8?s9<X)ejy;OS<s`V23M1Tf|Em34eT^TO^`l4f5ymhyn3Y)7>pnw
zZUJ_Y5{O@S^u0{zDQru^y04FT@l7bDE{hYlt6AWYp29eFQY`uLiSfx0Gz+_S5o(A4
z`rnpji;GK|z(mfAK<_9;%*y}wD)U3xlIF+0W3`UJ*Frk$KTrAc&D)s3Kt)>{8<8;1
zl>UI=XyQH>MtsIJ#m@&PlTCt^pet9cT{-ucS@_#LFiEAfA5obs_6DUQ<`Gy*TXG0A
zg`I}%{LZr8)Bt;z`CrEbkb>Y4ESCJVAomqf1f$SOe1Pgvtvh#K)|)<hV#lwoHs}GZ
zx-e4WZrwFlLPbTj(-xk=0o0_zpTm<aQm2Vx-LnDExwBm+B_!RKqfzWv&O`Tln+3ef
z+X_-$wqV0j`D~&#<8|OUqCe#6H&6o)$$mrI0d-PhqS=dP`&XmXT!BQ?_U}t}A!SHM
zt3MnaO45w51j4U6sFf9v;PG2wwlHBSVSU%_?ffDBD-H!?Bbq*t8wQLV242v`K-@d1
zNdc}T*2x)jgDAG8W&v;F8VtS{W7-3H!v524hN=?x21m!89H`0r%!s!NsGZc?{(&VX
zH610wFE3_8?oa_Rh)ZO-8qUZ|hEG)2_V6$$r#DeoqZ!y)dSGC{rLC=f`Kq=c9DMXN
z^WV#7p(`((NsNyN><0Hw`rko)gZ*PS;vdEXn?ojGE)))>thsB2qafUroD#Y*mQ5)U
zcse4(VXSHF*5E5nciaEnD{_#q4TZIVFQN9nqq~}$lWT>I;{n}3tg2__PHy)tnvtL&
zrrEAf&-EVP;|5x<n6x<d{*hj2B#?8;y(3kCHGQQ8zJ66@x|w3yrvAs?Ri^lr<>lq`
zb|s5pt=983wKW=swncG?iJibT%WB}8x{&AOV|06a^xobJaIfGo88g*Lo<@u0^~d_x
zAV$kJZs?pA%)awA8FHJna}xU;Ui;+>DWIzOL7a3EidYrYpm7V^%jlBs^zj}~h9}Po
z7m!6T_Pr1%d)e9!K=%+zCK1_Wcd4DS24TaPw6t4-L+RkfD3AmFeZZZ8SYO9&T`6_3
z5;JEAyjS}nQton=t&M(UXmP8(qK%1Hg(eIHY{vRml8Bu-I~Zyp^f?>rt~4HJ0aJ$&
zTkdD8z0^IYRovXRXU(+%rx0|G23URN<*^e<q9AtYJ&^i*DgO$m+z^QUu)4aDA>5;-
zb4`+=CoUo0lgj=k_cE|k6?_VsAV^5Ah_scOK0TRd5Z;?$0s<_T9~T{6`32W@ydhPZ
zECQc&`v?6<TzWhQU_1V9TWo!woC&y*C2NyY>bYuL*jb=`y)jtKu)FT@?cfmFQ)PgE
zn`-m|oJXWqx>OC~vA%@`t5OQM95ogio6$S-Gq%Fua#ovw#FWksrx|$qf@msgW!=Xh
z?UY7umynq`Bm1wRoD<}`qNMyPfJeWeJ%QqS-gq%3MR0Hb7y4kxv8|O4KrUoGOyIm)
zy2+80z{bfHLJ&N}Ct0=^dlP?tk`T7P=ipHCEx2#(jt_FT`O%|?P>7iaaQ%yCTw{O7
zSQM2Oy*mg#OVbt3_YZIVGgl5i(TBfFmeMnzS;2+oZ;NsdkARniLK2Es#<r4sO}GU6
z`+=d;aN0_`(hUhP%-2_{{xxT;Vy<5>2$Tys@T~<Vm>>$#mH<&92S94Ha-oUHfHmt}
z<$e`>Y?RWd-v5;!ML=p&UoOgI+OXkUkcbyyU%>iJ#L68SI=D~Yvg`ZPA`DJIb*ldS
zc|}}YoSL5>YURxQ*#jA1@}lnfs1|6{lao{aO@=uw0)ZGoR9rg}E`PF&{bv=u|I9?G
zn*zj%;&(Nf_8SVff!hW7`QL*?z#e5_e0(*9Nu*z}zu-qWC0wH>ML0r?m|ZpbHh{Ao
z<huOH4$-HOeKIjVpK5lPatmyG3|$tJFFTTnWJ`3Zrv_{Iw0P9=)b+u2GnM@0fI}-Z
zV}c|h?PxIF%zg#oZ*3>r$w6Y=#spWL?~100ht1jz9i1#{4O!XAWBUM8sENtRj?kB2
z!O`a~l!*BqxEo>@Evj_ps3`8F(c10EHP%B@+blFU2fv}YrflTcF;wL0a^k=5f_DOz
zD9a3Y^GG6U%-)|RVl{z8^}v${0K(!xzXnlY%5ZVxh%L(;5h40>2kK2)WK%W*DTBR_
zSpXcewkrBebnw*=CvX+mFCyEYBs^pbNV1MY%|taYQ+nN2g0>8>R%WOxDC$V$mb$Dx
zMt6560=w#@Z0#pu!KTEsA3?+X#v;J#x>jO3`C}J46Flh1Mr=QwBQD5*$Ak0+K_Xlk
z?s`!xwJJ*Vf0<o5Xk3u*(0<ZN5&EuGheC|<S4C~85PcE>nF|&Jl6Ahh+h0?2a<w%9
zmw18B<fi>-iJ&Z8YJ7q~`5bEkH9NH-bebfIJKZZ6Xxo2}nu=mX0tG#6N){PAB$RTX
z$NMW>;pEJ!7@wSrO6<NardFd<=NKZq?pg5^`oKMC0nth`6$AjbkRQ$gNRa)!C>SX0
zwAhGSm#5GZn*X|J;Hn4|bEe#zsO`w*Cxc-C-<_B3f>2<E`M*kxAmRJ{{gqEYlX~yJ
zUdpepA9EPI7konpx(H;XeLGxB;h72T6<Q4^(`Mj={rRHXWeWJREU}#-9PX=@5<_~o
zEtDVo)~d5xN;o$5tcZ0Hm*RDB1{N%|fZlcsmP&WWqOp4|ObN8*<~ltkI@8(V#8h=C
z!~~xJX{U}=*1ZXnzKo02{SIz_+Z&q3FFjRcU|e(`Tc{i(f;7Y2w8V81>SoDc?HacO
zFY)_QhHwqM!M@K%ZBF>_<5?10U_OXnM@<;f^N-}G7A-O7@gc^pzE{vofR^rXYOcKc
z)p6a<)p*`+sZyn9cKz-`3q%hiH1vwDf!=7GLd(bp2|?%=ElY1_ho6IkaS;UdB1roI
zZV%}CN3tcIHYPXBb%1!Zu1=qL{x>dA;{}T3fh01quW64;ADfRz%EFtVRtac_HaKkG
z=oj-)Ha*fOM`hrvKOSo%2JVGG_x8pZe3$xk)neDB(94E3c2=Wt6X3lQS3?#dqSnaG
zr~g*NnyX>6L;X3cBjSL12c<CGjTY`gA7rLMt^EzOc0mCrO}mW%L$D+?IEf;nKm|l!
zSBk8W>o(SEE||guWnHp+(GA)Sv+Boru(l>EP<Lm_b~bzd?AX1&W)J%kil2Ri=zK9g
zu`u1ZnJ}Khu>U);nZ5tNc)+lVGB12>#mXo`fU1tqsJ8vObCY@VraD(TJJ`1YPa^C6
zcHm0i-`|hLb_!@(SmdPn;l=<YXn}-wBmpmhLppdBo2@DWHFVvt=}g#lWv~K8Wn~)~
z-Rh?<A;<PXCI<q)5kGT`tE&XwJ@n#EY3D4N{P)cF4`)uCHtbDI(q#ZpfCcCoH0$oC
zd=I1R1jwH@51?fr)e-CfvJAib*z4Z2nlE1zlfg#6#H_B&{(~cU=~Qm3eDCSc0l5Sb
z!ldOl<_@Ffi;;9&Z145Q3D*Ol4S2luPL=~_^ue^lz18CK4uDPCet}*k>?PFASo)bn
z7>3fkKA*v2k)|TDHX>`w#dO_jKN~!XoAwt|E0HJuhV`!bEA;w6Wd^iFoWA`qG{~IH
zy=B&HEao(6Q)_>(%3(~X%5->Ws6B>5bAF{TGhrzcY3jN5^9IN=w4TUT3hBZ>Y4j{@
z!glat4>0Aagxw0vsLbw={-Qmgs=A2YHJ_k9$s+hQEhl&32q=9C`I`A38X9<$wpLYg
z?@GPzf2yK%IjZfMVluG(r)ipW>y<575p(u21FTNB!vlsUo21;sWe*c19%LVC3ZRA<
z_m=yLfSrF*TfTXpSMxSwJpnhN-I_a7oSpsoV!F?-mt>asjVaDF&`;y*p}2*`81zsD
zY6)-;Ik`Z<uc>3=QgF+kP_yCk<D88zqBuOL-fgLa9-d}?{$%w}K@{?6E(S<418I=5
zHUL0g05=uCGFYhj<VlJz<j%R>D?ek5oPVd<b*~8A(gUZat-JdG3~{y_Y1;c8wx}1P
zX1itXM+rfGshjyTOU;KHvEJK@Xl?N+OTaPd>%0SG|DiP0#b=>aCr~X!(pDfPOrTzQ
z`o+4a@Ng~Vb36Qb+cYt~v<z7dfpm;tzuyif4KrR%7CDmFpQT!kDY-{R`A;-t2o03(
z(yCT)cqe#8PKpc!sdb)QS(jSp87Hs{SdXL^hBkvBX#lKH=4JXRVzT`5N7xA0Z3=@<
zjw;>6O#%M8I+D*zK{pCB{2Rc;eMiiB(#EI%1(h2E+QpiC#X%S3N5MS?n1mrjt%!?;
zj*g2f<3$e0|4SvfOY-obWvDbRaAo>}A^!7O6q>GOPi22Q^W+w0w0yJ72%4CK*ho`c
zh78$KnGTpG@t=wpM#D6}Ju7FTPybyseY1R;aB9DUA~P|yOv0!Kh2_N-lg-4f4?E)I
z*myK9b7ctB9k*<Dz~MfYs;{MW-7jTyQFAqNaBz0CmDfLdRBCKnD=BMRT_2B2>;Iz?
zB4^tNAw3V9aW&;=2yU(iJsyIjr>nRabvz&>1H9sAgT;5@Q3o4dExg~qf5$<rjKUu9
zAym~)UD%(I<wQ{NB_K8t1|4_Tj>`09tjYnrwqM=;eI{mxu)gWqt6$zhQ<urHOErBf
ztn4_s>0|NdX|H)@4(s-$rgK6|dn;dH=Ym*)Xrz8AoeMvMh!PPOPc;=DfAvEoF-U}=
zVEHzCVOtdLZL&PF8AL9-*1$cl`wwTSVeX-%to0Gk&6_cpvPPYa1`PlH0GQsSy*5P7
z1c6C2z(qxYCNd@}vG-_3cYYkx%+qp!do5I3eCyjE`-vO6Cl*Cr2ge5UfzNpcb{*#|
zcfvatif0<`-Yq|@@?9U1@2tET%mZr{$pcEUTEO{s0!X_dL@aqRCR@jCu}ctT?f$_b
z{asOhzDgMi_XH}%0Z}xS?kjrU04H9^4q*}+bj9Od$Q}Jr^4?!a>=u~-LfBXTl->s(
zAI~<Sc|9mIiP>-sLmt?8a>R8wZZ1CKQ|#xMWx=tYL^GeZtrKf28R~gYXi_Tw`GOxd
zcKkIqA}sQq1i~@VMx8<W?(au3yv}7u3Y6L?W&TV_pP-GzT>U3<zWT?<^G?a03x)HO
zIc4K?3Lp=zK)U2(S<CTw@yH=w22oM$6(NQyb7Ws$(-=Qx{W<WAHFwOkJcTi*ZiYkP
zHO7X#*GP}On6sMWY>P?HO7P7!rkI{(!tg@Hvo^3f)X#5fG6sep_66e^x6K3pK^|=_
z;&#P$?A?D993A;?qeSQXlhYrSoAM!WYl1T;ABxL_P<98a9#9|uWEi)bJ;CU=^mXWH
ztZL0Zq|~_ObMqqSZsl>AZs%1mU+(SW!#ZqZc_!O_#WYvUaODwQX0-i=G1qg0W0UFT
z?aOHT{-6RSJGzs5J)dwM^+(z81)N*u`mVyC74yd-LbjKTDFn-!f9GO<y7hK0hhGF_
z%{vgLgoZ_kj;h|w>OahpHA~A@^I;aPzR&QUd#VhtnwWl=e<RYrbb1s-uT`dryY{<+
ztF9%xkJBdlJba`ZO3dc^2<e-JGiJ_V#K>_rT(|KWE0gVq(E}l?%GIc-c4@2TGcReZ
zc9P2O<*k2xrKf$1d?)L(3IJ=b08+k?UqYX@<@jJ$@$>pRXZtm@jLFx_o5}Gz$F^#}
zgZRfqnw}>xHt4|N1DbHu^dqp?8kX0s1PN<6o7DmFng+g~P_yFsHw2aT={r3ZjnA?j
zXOB8r)5U5Sz7k0+iL2S(h|Bj%V_UI)mbzx<T6ctd`A~1YD9*8*1HJ1!$u(;25KnhZ
zNQtu!kG=m8W3?Y5={nn;85>p*8N;pn+QDbv*4x7@=bu<T9{cGT-l9TZW4S|gEPJ^s
z6+Itylk&!XcdXv?4sEQwK~LNe6LQ(h=V#MW3z;_jRotzUD4FA*7ChS51@>L<;o5K2
z|E@CVNLH~`r3}G7K;t+B+6SHKG=I#i$LO`_dp?SYJ!4?X3fH3NtH3#h^uIcr-Myj$
z*;oKktg8QRRVMmtl&n)N5PWJt>LNb!o1o_e{83h*u~cH}a~IR>{(Q4&iV);WpsDLz
z?3MX_wa@)+VY6Nb2G?0RGkLR4)|h1mv9rzEIB3-`%B3E8&X~dK(66pNXC@*hQx<0y
zL%`XoHy@ps<WXIcUyZAE#CO7b5A)k~Gvdg~GB@<?3#if?7q)u&Vw_tzD}Q<QcFvUF
zl=`RPg~ojM38BGvyIT@leCQNgwcWV*WYyvLp6MY$sq$n%ptBGjzB=7f<)41EZQt?^
z#e7oXO9(s{W8-}L@MJXvJ1<y=E|(W?<6>dWx?|%;%HSw9*I#2`!cpgUXavm4eQ7#u
zB<5r<wl|JKU?mJcr&}@0dGr3=2z26eVe5_|c7#bCcc^xEOw*CN&@=0oHBI^%M@y#3
z_;Xxo_tk~`r`3gK`-HI}C(ZhvX?vHXLDv6e@!1!KxAX5jlG$(e{b^n<_d&<@2|nk@
z?RovTu6k3a`d`<|nFMz$S2kx71daqBIUTGcy+VW!MEMYj=nK6w0!Uw*;n^Ga9VZq-
z%coe89s6m%XkKvQP@4IpRJj|WCthK`a~12qrQ6SwdoX;n`IGZCfWLlC$eS=cTH+Nb
z+l98vTLBNF03x(d8;=;zkz^rW^r67k{mZFwtmpyO!$IBnZELSx)Cz4XHxJJitC1HP
zB*JovWKb={BbZ|%l9^*S2{=wk+iHD#MZ~-ef8EUS2W8}rKDK|Q2<J1*88Xt$e==ss
z_VD|Ho^v@bV!>&Dc$<|jO<wix;fexvyjS2YnW~4Tgs~~?_$3dQfWC#DH~pFvc0E!{
z|9CHe2&$mrN#m&da8?C^s5>W@KP_^}v0CfqwJA&(i@EDdONxW51MH2NN2Nz=>vQvD
zCkL;6f7ri|aXS2yE^g_8s09MrU`sr?@o%fiHvgMZ@_*jT9H(|#r8!}4z4K!7yMNQF
zNs9@?xcDNpBM!}_bxCq3=Ln&GPL}f5O%U;F7glA+A!=`#)oO_+-MF{TFz`6>ZGREe
zXfJMLeZFm%*zgeFesVO)W+um)R^k4rGjce!1a%kv^B?)w0ngm5aQpC12l$(ZgCt`$
z^$nPV`wu&7O~~@lrlga~uNX>2-PYl9yuGpncP}ai5KlJf5ToB95OQCP^(>m%iT2&5
z+r=)xPw?$7cRfbXpQ#S7e|jvVt!FS-K^OUh&7($p<b=fQxljM{LJ`o}rizy&J}iF3
z^f%R#TmU4<oK64D<7enkIAo~D^-Mr3YslB+o||JwJ+wvr+nw)4(CN|UCbsbnKSpw?
zR5emr+R}5X1Vq)9^qu>|R8y{3&uG<lrApr!Xm~_g1`3MMUN}DDGm8^9_3JJfeEJvJ
zCf5ZQvH%xCC>?L=4r9?^w8PzdJ1mQhc%zyh6mW=^2=^H$BUcpBaYBX_{>}nl-<3>E
zw5X44I#=MB+1k>DbuIFgYp;$TP`OLl=Udu`s&&aoA~urP5bH4h*`2)l<=<foR(!O^
zMsCUqgGOpfFNn()6$wS~132NV_wW9j_45;%ar8wX4>MA%=t-RhPWx7L+2xjs7`L8X
zY`4nXd~0qmTAGZo^L9msU31gh6xX)w$k7?6>u1vJZx`EiI3UEcwU-g(`L=M+#Cx&+
zY%+fR<Q;Dz`visr4|36rW`9<=0{jzwy&N@twGT*%-PErUH;81UDYu8mCb>OIabGT_
znJXLz(|WuuEV{!`YVYU~G4LC)v1FSNd=?XIbUnYQe#yNheHSyIX@*j71`0K_mHq`w
z*U#57#UAN2t*WC_qdoRs(}}B;HO=8&C(eZ}AZ8HvMjm==M@$)VHJ)r&JIKv3NOE!2
z+ME1#s6Q@~N7oGFJ9>}nAAM+0V!-$tl*WfRwdkg_VCnX!=pqw`n5N)TdHEIh#}A~_
z&R@HOdVNbrASDWkjm7?KIp*lr)(-hf|2LvF2eqsMdby5RQsfZ{e(N9OG2%Z*&t^T3
zpWGg9v_yySE==u`m0qoPT^lQGKe9go(x@2UG5i;S1+J*_7{_uH7svX@Vu_^dxVo^_
z=TiBE>u!l1QT{vc9mcU`C1WOV<hn0C5|_AubnZx~Yd#V7cAY!xjTqChO?-?_LMiAZ
zxin8=k>c&=&>oc=W!L9vyqX*xF`e~TLfuAeueN%1qGQIZ%8l1;^IHa^nr=PHJZg?H
zZ$B7IuBQ~w2htmkcwvgF>K}|W9mUBI#+^b_qXmj~ttty^)OrX%uU7sEB2||$-udC9
zam;CUGH723-@iQeWOijM$vH*5W15Pba*4XcJX{=(qJ)0{7Ad+W-uaWbQfG7%jc(k}
zpN~T$LIo3b?wD_ukCyMEWTBNHI|4w4!7^DS)M7h-N=i{E6DI%V8-^7Q$wxa&@*s;b
zEwt(s16SU20L*}6<166b$e%OR8h>=jZfbE4U7N1yJbk;lN>UP1OhdjRXJLwQXN7e-
zJ9-kL%-l+EnN-fd2>^}-*?0)4`6UOT8NapEo|%|d+Ps@YZps2fu6z?~t%|2`F-Xb`
z+kk+i8#<>LN`-|%{SOeICp4WLBSH{;%lV80dl+q~&;G4$79{yKUQ2+X*zdR?b>U?T
zK^u5kGy<<O?YYy!Jx~Y?n}iC>DW(Bvh*RuQr1bF%6SnK(4!42DN-$T{@U>Nl(fD*7
zz(1dq-5<o|v2%6qiqG6S*_%M+JO>!JyUpMY-O~`F62J=uvgF$oM%CiowuPDeMq&=p
zpnifUz-A%UEcl3vC;(+#NZ?7awQD7}>W?qfwO{C!9i0mM(FpC!T_7OR<&ysCpLWcs
z-{+zA!3tkyQZO1$(<3puqE4I;pz;Diw>J%wcNzv`IYGS4Z|q*wROKEzHo`bSG~)S7
z@A6wHG8|%8?aSUBZq}=U<oHHXhGGw06ogc_J_LAGt5uq*NNl0jy9;!DF-w`=KL8-z
zlVtH3N6ocqu7fgxvSg_lpbx5~x%{xDGk%c`y$r}4O!Qk3itvW`NJPIJ7@cci3II7&
zpp1U!minJv%Xy!}g%@hYr)Fk}PB02K0Gxv*_6K|y!T11nY4Xo&fp!A&(8xd_r6)sJ
zsy~+4K16>W9uSuC6sEBnj)B5OHfsTQ5&g@My8+CLcn%@pjFA@_nQ*<?#m!5A0eBV)
z&xaC$+JEn5m1)tkCTZD5a(8HuomUmn-(*XgjKkpo;s%{X5#qr<#d)daDy2IeeBleR
z?{{~9(>8aB>JHR4zs`gH_sdmB!3|qfsPRarwkaRn1cOMyr5@<pHy~)pxI`87sbo_T
z{Pf_%upiQS6$~hdIt=Jj0X@@PQ7I+X<8EQvKi-(`@1Oz_SJn+8%Bn~MUD-IwRHVP(
zHCLA}g)&o0B!~zOE<?D`?OA{!r~6h~S}nB$HCkh^q=%=cIC1yQA_8@;Yc7Ca3X2GR
zcfI+~nf;}i6VBn`VOhI=0G<$O)POz+DKrNQF9zv~MRRq2<I$fVgJ;OebfKAuJ(*<(
z3{!=q{;{-bu85xHomLcH!|Gu?uo+IeD*(}F|6XW|X;s9#<4QfqS#jMoJI;xJsZGVO
z84bN8g=Gk5y`4v*9X?#vxOLZi;X9orP~w@q4)pWs2OofuQ$R!bETHn>VukOQZGc-{
zG{&jH(EqU5EZI*aZHw{mz3hR`#d0y}cdMzpA>8e4SVX+N+%|nt2{+(e52EN7=WXp{
zJsCGF=_ojf?MJNS_a!b^1SAyz&t_J$<JHl5Xi$InP^?XrA{M18(?vXU5Rutd>bcG3
zCM)#>2x9(87P0yugY8qYPN!d0{`;VoP)l4B7M2>5wLM;{otO_*J?2OicM{1CuLWt2
zMej>Q4lfTt&+{4L3CufI?mV-0e;fJ^ibj!BIFJ276J#`y*Yfc4#WVYz9Af9^qCBWP
zSpVKD1h0MpjDxxn=<ddz`wJF)=w9T3{{A@9&v4oZTieuVS#<8N+gZRb==GBW$W}T}
z((^#X8xL$R0Sj(DQLu~2tE`Lx=K{?NO0&S;z!>@<b4IEbysY;@McZ0Uh~G!gQle)6
zZ>_~7H=iFRhA54M=K7z;c&I}HQWRnvf+W{vb7@U{l4Z6t6D|8y%Wrp1Px}%?tC+v0
zfzJf|dQ&MEzHRG!`r*Sz9RvBA>m2Wxk(-UUny>zQRnlwrtb_0_sHEqFChWG$bjTrq
zO?3TnbIj=D0X;-SJDaW?M{LPl1kExy8SEM;ug9i!D&<0t)S#(UgKCCPO2FCH=;xIC
z3|~Y*!sH<E&}-TM-V0o_LIhB2LT->&w6z^jXn`2HZQHTFC?6{CGq15nodV!`eX&f<
z>0;L)l~AhLiH6W!t9CqchYeAR1d?o==xIgdU52;czkO44cdxdk^tlCQQ-w=?r$3{B
zSP{AZH65q7S%%v1*<W?618He}`MJ4YjaO%l$}#zazTV}MPcDA6*msrO%8Om&IMKkZ
zSC#K4#>KG%@*FqNf2IKm>YrBE){bkc(k&k|WC75Ag7~8bv)go%;1tNXdJGylpdBDT
zr!k|YvA}Qghw17R^Jcw)j@-f(prWT7eEMT<OU>RW6&Qy;0+aDN-)B+avOiX76$Y&E
zUv8K^e$0IsdHJ&U1^1Jk9rvQr(wf9b{ysvy+~V`OAL<TT;pd=aQy2^gIRX_<o;3Z+
zqgl_NMjeby2uT6IDAE9apP!%*11OTWWeF0qfK?1vQBe_&$<GJ?$>(S-K>0CIzhY3R
zaDRJgLesAphyh<}vK(Tli;MIN$L$Xu0X;dWbx5*aj@{nwUg&q2sMQ84WzzAMEKmkz
z$@CPE%SvGOqas5EiPL0j=6aw)W1X4L_cqXG*4^ov5~s2Z#$OefAo{ASHQ;2DjT6hg
zWhNnoV_%2CeZxpoSR`k;ktWT{zoUMR!hWe0&_#H-&H!7N9;SHfUuQYfvQP1X&ou!D
zI&_ow_L}{1cbaHt^I4eeN+?Q9{}pEd=_hhV9f`p^YEYNH4#>zWS7=jgs}T@kL(KC7
zW5f$kr91t+<sgO7z50Mi-@#a*bKZAGS<<F?BoIdmpv*|Y4Xrw+JWdaUXa(>KO5!gV
z97}bf(*t_rZ#^H-lOWPFpi@~zCAo8X*(NkJQgC!u?m&O-Kpnz}!3FDF3BgQ*jsk|w
zU%>qmzkG&7n_R+qR?!XYvzIjV^kV8M|Nemt`h(Mkv72{~_fjfRh(o0K6kE62n4Liy
zQ<6D9v>5<|5x_f1q^4>qol}4xEU1129m<zUQ>eFgo&BMffeuz+U-*WbTSg|9hWP(C
zx4`5}Jczn?=qGB+`{}fYQn1!DJ0GV823RAswe|E(*uHao*SmL*<v-m7D6BOBrH(3a
zhRSgvk9`qE&bzz2pomOD08W69FaH0htH}2)`HxrI#v~{6sE3AvQkQI7TU)qC(7N~#
zq)h;U6)NcEKrc4si=~xS%EI#USAsQhBSLNiIO%)2ZTY|j*|vg#nNM&aCpV`d>g)Xx
z517&>@FCFywMhZdt}5GuIqX9)&sYsl0IT-h;mir<H=ie+Sj>#{^jJaCp(|FME7^*$
zw(@to@l_1CJ5o^Jl#-GHihL8(7h5XZpJn;^d!|Xl!#Oed*`O_&`Jc}UQ(BZUp_c-(
zK`jjbf(~@grP+1Sr#=oJN{iwLD)rp_fb`ee4`e6TLjtg0LXh{O!u!6tcfrCGz<c?{
zmIVX|163Y^_qNau{wXMS*Oexzbouh-u8OOzzifNkfB(L0yJ?d+bOlpIIos9h@fI-u
z8s6UZ=ioc_=>X{+n`i912V`?F*r`QZXK{00YAsMqY+z{8!N+|I4}WTNB{5FTLN6yT
zAM8Ue1~SD$wM__zcun^QGnX!c$sJ@#D9gKhR@Z52x`m<6l1GlzKC-oU6eXbu_$&=H
z1K#qsS$#!Q^0#k~=+`;V$mzRr%55y83W$=Y+uY*>Msm)mHa5ncG?L(;VBzrzHV(<x
zF4okbWDBtB<4g+;TQ~C)BXWwM&!vs1jMdvG@>qgecz0HZY|+K}IZZicvT%q4h7M&3
zZ~?&zd+ez%$s~ibEugUMeCx|o*4Fiqstr1*KjWAs6uOhMpN^*WubiA(k(`6x88}!J
zLFL5t^>;Cf-N~P?sddvs@+as}zE&z(^=wH8%#J|#_YZ>atg?O20|QMJm8YU`Uoc3a
zgW4QNOQ#7kwet${#_exY7c6S2hyDZ1`7d6&4s8pdgH;X4^*JB+nM;C1O$k3reEeE$
z!D0akl|2>;Ex@4=)EC$aWIs8JT`abr^7Uh=K#9w<SNd!;=BKL1-4F=r2-=~=Cd%Cb
z+a3&Y0%0`_GjS^b#(tFm6t1~OF54yoQWklGZ>8NQ0r|7&GB?t-3Y|9~fxjA*uHlWK
z{Z>Hs&$iT%CPs9HaZ;k(=nHgjtwsPu#j5TvY6HH-`O_2rWP)uy5>bVsn9WN{5mr*{
z<?(ntcyB0e>#Eh%3k7&dLnBSkGJ3P1J2f9LY}4n1XxD4S5hJ%kBO}GMK`p;z&DPQ#
ziOK(E)Po|fjjjtCpek%4i2M-RGdFp8t&WfOae6bW7z`Nmd?f2;PoCTadD1nbJEbKh
zT}Q)Q(S(N}L$F>Z#c`rm{mGLj3WsC|le)klOh{OmgGcbbg@pyXod2)*<CjkddtZU<
zjmlmhX6MOI_usyK%Pz||HGMw(B0=_ji|%4iI<Mi#*2Ab`-{}j=g)dfl^P8M9Z=vRB
z_i;3H3y02AT$!l=zEqXq=ACfvnV=sLk<gJ4P6D1kNA`t*V^#3f1rS5I6+tiMb}q`g
zdh`n!voA~NNQhR!g5vxBQ_*e~^vralgs|~*cCb3AiWQ0C84?Ob5|ZwFwh8Na69GD;
zF1gK@@L=H(N=A#vgDfl0^BX>l@1)UpeJr>^WdjoXi#=TXED4I!P2N=?Ygq$i53I`u
z^ks29ACK7F14;`(6o%GL-#aK^&3ZX;37DtCvI7IoW(NdVzo+gCN=-<3<R+K*G60wM
zxpie}GbM!qRI>bMXW!C{M7y=F#H(bGX*!RKT;cuT=!e}eHN*g!<tRBtm}e-Bz9bSy
z$-~1hh>8wAj3S)60Ng^5f?L(zYc&eX1EY2!jD5@YbWmUYl1yo@Cv~+N_p`Nid4CT0
zTjtg0d#l0iSGxm#Gs}(*Tl;W=a2eriOI$c%vpq$Uxj;<{KdTd>vHG##-Ff(fuBcnM
zUH|^mMN>N&wt$4E)xV$+cG_O{6ieLlr{?kTC+KeUyHB4a^f7mUQT43CDc5wJ=I>IJ
z7>qHdV5cr2bi`NCBl%$8)W*(E$m7w0(u^0V0S$H&Evl7GKc;ai_7|5&Pgk}KEEUQ_
zdAx2k?Z+?W@MAy%7#C+r$cYk_KehDp^rv_J{6nE^JANfxr$91{*yfF=>?@mh2vgBu
zKZ7QAm$ITegN8Sqk5>_{ZGKVcs*}C^qo-5K^@}BzV|&Yf*kx&FbgY*gk8~v*y)`$8
zI<B0)(QQ?{@SSWUwk=^Z{_VOmnk|Yyc6^OBIW2KuTi{Q+2X@vds#BiH@RpHgr)6*L
z!^*ueulzip^hso|lO!&1Md-s}5ms$s96p~duH4gboGZlZQ`~g)D;K+%AX~Y+t~&1=
zRTkY59xCMLkTTPP^%C2R$1@Q^jxf90dXJi(WVCEPH@u&I^BLDwH6MrGOx301y7?aF
z=^MxsJOHUOH=@a8F3i8JIGlgoxEXh7YRGYY+DDJ+g;VJe=M}a7z5dIa_%{01ZR#7<
zr_mX*i0(skP=6WBe~QQ!kh+wbB3vOzD4w`RPPip+-F3~)q)^W5;;M%x0~Eu<PF;!T
z-FDu~INGbdz+pAYw6ZM2Zoi3qLYd|fJL&6N*t9+V3Y%Ik%UygueY=d7$H+Bt5?e%;
zw-tC2v+1`#==#0pghT$IhhFa>ws@iQ<T2OoRH4moC1EzR)9iz8tN`+m7h7Lk7TsCC
zSM9r%nweneQM?erCpxf(cQ4a_qDk9%s&}~a2-9*)+yoJCyuGm5o@nhD$BaCh!1mSN
zMaT<ur!MrHwcH40u7B9PKiu2dR^D)@c|Xi({^!C>OcY`CRb^50QhV6@HE-8-aer>J
z77>Hm(is8K&do$KS$8#^z2CYEiSkDUzPlkUF=j~H$`ZmV2exR!ELJFLlaXTH`^`V!
zzjMUIr745;YM`Hv?V;Nb1NO6FQ1eieT5yOW!sm5UdD}pFn#vU(Z>u52`u8mNW32gO
z%BNh@jq#-_LTu`d(e4`CwOsRUX2-?Vz8kdlu67L)jG8%mQL)7>z6B+O<C6}WiAK`6
zo>$Ij=9W66g5IvW=CH)~50ZWR+val~v`p#UYMdWK#&;a)rm;&-$(qGj9yP^UDjqgs
zcci}$RJ~AWe1O8gKib$Jn}2EGrBQ`?WYLD+7dPXIH9aWuU1G)nVBPUsfrh7PwpNCu
zHa1@c7thRVfLxKGPh3vR_@TD?Evm!I|4?0WC(&pjLB1rqVWXAf<I)YixN%_vJvVC?
zP<~il9sP##zgM~&SzE72B?(>6IbGgpr1KaY|4PK3$jRwzB@7s{hHW-znik8Ll*u4t
zc4R!l--Rc0b?^u}cYG~(NPJ{6JisElZolx|K#Z|JmHp<k7?6B-O|ZWA(klCUiLlk9
zsEEHz%Sdhdgl4`ytRmNBl`3B2;o;r=-fhhMJv{zx%%^j0oq;^vm0M^pOI7>L>^ooL
zwuCr)@IBw!mj}zEdv&UskL$e9u{yslFK<(m;NlZBRJ>oWNGeGPyMVPF)Y!gB8}4%V
z@9M9>o%ERfnefplCq|}P$JF1mP;HuBk`vP&+xC79?RvEKVyv-7*{;NG=Y7-MLQI${
z`a`A(N8;4pv-+j%rTOlM6IMu6+qi@&|M6y+Rmn_%HAKZQwuV|Xg&+RZUDzUHif==@
z)rW64cmGrd)w1b$Bw4N=x9RWl`8Rc$y(m#Tyy@+>PC_Gl`amW~nBp?f<JhjAv9^Lk
zyEYtT${N$r(UJ997!vV=L-CHrTlRx1L+b-P+KuAv=hrXCytpU#0&t>>D}_F_aN5ov
ztB+RR+b%O$Ujc~u(fZ6h-KbW)-^qUL5f;>DjY!P60oRf@bJ|}ItDGw2@5=hN*s-?V
z8^?3MaUs3%<OU$g@$+OYd#@%>_O36=Q5ffhGsQZZBy@hLVb+*Z$OySzTleYbPKvsE
z!D81_o9<)_OuftlIV{*vsd(5uxB@E`n;akE1P)+&4^14c^ZJSk1MZ2~4RI$+J_}~&
z-~u)b2A4a+d0eDr9WZ8E45=wv;$1*|D{QZ4XZHc6Is;V>ww?})ST|GgA!vR%+TUK!
zZ(b=;H>!0wXM-fj3VEzP0GkicDqRkOFW3GPes*_tTp=zJg9X~LU0qtBhPy~x?r{(7
zOb{#K-ULVxXc+zPokc{ltYGC1KI>R6S=y`2(y+}ywyeRLmP@tmQT+*4bKY|wu$9Re
ztBy-cZy5C&_d;BM)OcymHJR;H>N=0-S>t^StDc8B%zS*5AcN0qYHU;&64<Yi*$L4_
z$0;w>+x@#=aE|g_CG!3v*Cb1x9s0R07|tZ5kIJ%YZc3>7G!4p<S?<bjMwicwaWkK+
zi_edr(YK5{tKBm<e!x@jwGg$iwv*x2&^sZLcDNJFp6b_gnf}i<wJCrcgsZ$)FDih&
zhXlbG<>fruWgxY^sU{PDusyITj=(uorLBufuSA~~@wsda)w3lDIx8DG$Q~DPb`}^w
z0mt^Mz^~!VyGwnK8`pf)<orfPE%)O#JM{(p)U6LQ5e-I5$6oUlj4Ier;qzMBsZJT^
z4e`AvT;l%D#l{DlZS#k439gV+6gI0?gHv_wSL?((OLb@1PGp^TW;ex`0}7rbocyM7
z?M~1CFIGCH2bLM)k_Rj=+kis;nuM&*^3o(0i@n4+GGw4YlkLq0_3uxrJD+Igzvv68
zc&;bJtWF1WiE5&;qIv_?>7*l$v8v1_YzxiE$f)iZ$<b-nQElGX;fYQ^Ys>S-@L^Z)
z`QEU6zYPxTyE)VsdZZm)XH;{`W4rb5xpq@*yk=uJW415V+kJlQHt|x9YF7C~q0X<3
za?mZ<#oA?uDubctbmL;Uz~3!DxyN(;g=aIzIb9CO`%Ev)8>`2PQ67&_TvSbZd_10v
zsgZ@u@pJ@jamTlKoj84XIQ`K%bHDm4i3oNl#WKsiJmK_1ar7Yu11g=?ILxq-MUv~w
zxGr}iJ5L#%pFeW;H|;^!>E@9AkqNp`DjAy+V;3xH{!40Hx{FxWordvAE-w+k0hO9%
zx(E9DZjPMfjSU)=M_p4-#h=;l{Av~#B;CT+;=7<EDI=&?_~OGzN6W@IC*05cWw@tn
zC~kBi3i;x`r#J#1Yn#|0THfBNBHmHB7*RRO1H?-rhKZ$mqd)DYa>OyXR0ieTjM|2~
zwp-)Z;*oESchR;3wL2z;i1ugYH7_VT-WuGr`s_%jt~JOsR@I_Dj!RS@dQQXMJ^lUC
z)Iqb%{@&U0l~(CI@69l5%gmEoZb;w4NvA3Q9F&zpuhO+XbM5~7CWV_|iz)$`3A1W<
zCsVIV&SjmY_(CgcLtXa%juHg}%Y%DBUeP)%Y9objNB%EN*tJytedfaqK(KB32VXSY
zY^)T;0-<ClCUqj;#iBq>rKWx7!u*G*j2kq1jdB=IT6vxsf@;I0k5X-k_rcXM27PXQ
zR~1UfDXxqax{fx1WH!ly5cT4?razkd=(%Fc3Hwd;&`5@k6t=>25q;XpgJ<=EUPVXC
zs!<&Ue48C9Q{&e;4I*U`&EfKgo!A!fUYfzguWt87P=DfN8cU3SbfufG-{)!6jG3>t
zFN=#6l8b{oon#!Xb?Zp(&Y2}RpNx*hZ0`J7DZ8^B-Guwj?md@WGY|vRQ5ugr$x4Oi
zfAb2w6D|m^&A5s@z{>=^(!*?rgP6E=(CbA%eF{iqQt?)4ktd@fdwZIV6_}9#8RUP3
z4=lIdo>qOIOLs2PVSPl<6PSDr*rX#ceVkvCUxvMnjHw+FD=I6ZJow1@_B1hC{l)4y
zqTUo7WjJ!q7k~s{TJAslGbVQOejTJ=1V;chC^iTTVrn{H*XL0*dANIi$|1W<gbiR%
zgLX|{Tzdl{8_*{&4H_N=9{maKMc0@erLR>dOl0iU8hfqpx?-k^wVs|*{&0ux9PWio
z!D9~Uw*cW_=aM$!z8vn76XN*xv_AVa@-tV?-Q~_wyZz;1?5?5nRKqT$)-3<5VEgXZ
zc*>KaqSA9Kte=&H!eTSHmqTw;YwXl~;F9n99ps{<0C4u8^P~(xq~ISfX-YQDeG-a6
z2evDM-C|#^{zY=gZJ>I-)Mdc;F4BmU!R~L+4?7JNFo6ytSV#Qn|Nd8o4+_6GPYVDY
z?ZAI8eNfIj7`9KH&nRkYXuQ<|u1zs9u^e*zJtbkGfHZxV11ymspTGRC1zO1#W|#?J
zgSBoS5&`_;Alo~Ra1mLDy6S4R+}*0uqOx~jiG*m8p`izesOD1RdOaX8<bZ1F4%UH!
z4Up^*HB<+xw*rJ%Q{gZ%ApsJ=0SK3ump8Vdp&<Zg07}m;Vr?MTM^IFOWSx20O&%c1
z=&r7~yl={GGH;`zfQ#qGV1NIO|D!Pc{{8#7N3EzGx1$CR42QV*sOhVhZ$W4K`_A4a
zE&s}(@_cdvEDfxH?~b*kG;gQzhdX2<2cMrrFo<>-lV&UGKHXKC>!KHCmCM>cg3Gh<
zP8(X3uV51ond7#Bf1)XN^A(j`G_QcEjetk}pGw~OC{U{e74($lf5he#;6-=$g}w*Q
zxXD0!ji6!?Synd&saoI2L^KeejeSSxae9f*IJFbnGQE4!xDRq$sQC?FG_Ot7e>EN4
z6+Q8tzmFlvjhMdrUrzg6QvsTqNS9p-m`||C^?ZJYrX$>FOKoDL`{*WehAr*>4B>MW
zG3n+ux3uTZQTVwmK)6!hC?j;iyZ*2{M@^o1d9M!}tml6P%&*zzkvyWPb2#(x=zfSY
z2`myS6_@{^a-ez_)OoWfVR1F_`>twryk>|+0%wV)tpm*B9yv&PdBZ&eqJ^q*M$a}~
zf6L&H64ks50{)-`){shwnBOLu{6~NVyxbem7VkI%Pv@CAI&gF))Ql#@#xNVTJ&@&r
z$|g-;fU))}oqW=5-j_EMY-@*TlY0C{qjq#@rs_*Ikm9RM&6u&1D!+T{4@P=u1>MFh
zpOf|HWNMeo+L1Ki&XX#O*M8db^@SNCTpWoe?%BqOU9N8RP+awF-f4xralk4aEl_!-
zupmCGbEU8x*26P7+V3i7gON?E5{ST`0Cl%;6FRkLRmA^G5Zl{x$k~>bn=^a3x^<2q
zQsP{&kt1tO`uxBUj^4i@5Vtb2Sq{-h_DVD=ftSv+7OA;pwZ*#ZtQt9YE}L@JlZ#(D
z$0XJD`!pE2+M7E;hjWQ$KHn~CKf_`{l?}<QoikWG*|GX&GBq?)(C|oHTH(2(tMPY6
zVC_v_%HRd;>Dw;(mAq)GN-|IiW%+g(O+a)$SBMl_tT=_eZ-^LSJ~ET02d40*kAWf+
zg=2t<i13JAWXVAheq6`4$hYJ+{`^S!n)C|^c<b5jcRACOi^Qe)>U4T3VC0OcBc;Mz
z98Yr1`+)^YgD5Zn#hyREkZ^L~ygw(4?}%C8oge7ruc{*|6qIkLC##2u?GLPc@_6C|
zQ?TiGBFG<&F=E#=1v<9{T5fS=y1oStQLL}rjvBWXQ@aJDr>Duel=NN|6cqe`ki60;
zWr~#P^$7!3c)CQdsZ{v5aCQWP$WT`3m7yog>L(ii*tJ>N^|4%0)}u&q>Cc=yyoCYg
z?BAK+C0ALaL0NWItQ(K0O)#ZO;7gWVwLkrSD+xyP3-HX!L>~)W0U)cvgzI%q#HIn)
zfLV~hd{Z{#J#*uDAH*gA!usKN#7O7ybHMuoUCP`HASajwg++H?gH{DVwSpP~3Q#~f
z6=)3X|9V9Q3J|oq^xM|79H(z~_)*E;;vIF4E}!;HO+!q3-(c>6wLV?>s4aUeiya!Y
z{u$$T#Flmk*h`YIMpHUAtdAJ2<?}!B`Y~$I;#;h~K>~Z_1$dL>&b3Lwg2Zszx}l;f
zByyB_0n#&!xh*EG!>BjXemA!eK+W?#x*mP2Q_TRWMo~aXHmu=i`MoBO^$|s3wC9@%
zk<#<x$b&FohpvP21EbtofCP3{@kFRZ_L)B}%q;15f&l8cVV}2xwr;-s8;teRcfU&&
zIm)Z6m~V4iNv@}wImiGzLB0*xU1q`R72Z0(siAM5j&cn^;fcE;QYbLwfYl+(y&kjR
zUCSVI!00IRm&)oJ8S&2jh>VS!w;XH&!RaMSkllj)>g+dVx|;Jc{H$=~rvJ$aNEJMs
zu4^_w{~cnIV5C<#z>0H|uba>H$4X&m98W9<;3`C%0oxGNm#r<ox8ViW7vEE0W~}fJ
zks=VMF5+TG$QD*x(Q*SorjW4sibFQ4yFdd>0OXA0)xXm@aC&qU79!DG3(oHc+3+VO
zrik5#L~k!xA6KcQz5y3l+=QMTTts(wSBUSaf-l0-`6s+=215%4qJcAsr2r#(=3b$+
zKb9?63c~0W<ZZg?xu80QdgVmWUm<BQTU*McwmfrkYg?{(2`@IZayUaMBiiM|Bm#B2
zlb=PU-vr5~xF6Fh9yl=)F9G1QR_Z+TZ)nK}s3L1d>bC$u*0Ne-FaADpCdu7|9eBbx
zH*W=_h8DiX>aO0u_DP`S_`87CDrTL=z>Ker(JKuV^sXRe9nkQgJX}o5DgzHcPeNcI
zsX8d_h+m!nE;1alvV`~Nnw1dR-I?+;rmLxu<x~IPMOAo8`P7)%%Hx^ipP)KjUVi?W
zYGhU~q1tw^tHc(tS&J|7T2x8uR+pEfL5k8C92F(*G|3GB_Lz(WC+9PFgO8bpX1JiT
zfwwSwAKHfBb7R)XT_g!gnEx!N15&qZL~Hc)jo1YHzar$`&vep{xc^vzKhEUcUgJ{*
zVb)#w$Cyw><NynRJ?$Ur`djxTUFrwccR&RQ(OCau^V9R@jEHA9eYyr>F+5pl_h-@N
z>#h4sAU@2!8xiwg3+MIRPoI)NUP;J*B18}+8k`nUg6EMkFpfL`B>?Suv{FO_e?{IJ
z`pdq+u5lMGYH5b0`H!!F*gTFTLI{j0Vkf)yhFef61vt+?2O}^=n6O^75%za~3A#VH
z8Aj_OQ8?riL2(LFOh5-U3*!Jt;|vz2;nKeNp5-Huw-SF3BeAh{d<W=0;dtDYQc2;{
zH1hxQLa?1z$(KHqJpw54{3ppab<^pjJNUGJ)ny>bP&*D(?bJ5-f13NsuqfMhT@}PZ
zawJ3=DMfmS0m&g0=~6&yL{b_-x*3!Z1!<6w98o|7L}HLsawtjZ?(W$4@V@VNd~2_7
z?_>R390z}HV0fPUxt{AjuR726#6$xZj^R;$p})c^KrESAIosBjC}=%74mc72x#oFQ
z9q_wg7E8a#G(aU)P+>AOeKw*nPc7W1+5}i&cHejHzmpr^_k@Im1|A1s0baXF(G(S*
zHU0EeR5)zu2?{=gub9MGCS2|C|28tC|GUQCv-<u#$#K@*t(7S5iUxI6b93|ALh66d
za{N_f|F4zPS$83N%t*l7?*BP<`#nfG?yUOT5>WsCRt5|=)dYCvi%q?~4^-_U8|1IJ
zN#&V5HZrnvj)HZ-<u;olipz<s|1)?4HumtZdaSHW5TFqX0s}3u=Rj0rK-f|!1&FuC
zsU!+|=yh_2vH@DmoT4J^C?15w^V%arLlv2zqW=!!VOW2~P~ip3iA2yx-tbDH{<H!d
zVt-iyX<-t~GQ}6QAO^eUrvrGq&zFm=9O!^?YHG4Hw<ofiHBG-Bxb;uz&i|Ic<p4T5
zI`iFv!DztA<v?@eZ<Au4OcalbjLczn)a(j|7McM6cm8QNAHBQpqFj<MRx%!OS<quM
zD_vz8H++u!x)}rHl!r&AUdN1#q@=eWEB~_|O#l<mJ}A&HVY_pu)tK&|vL^E+PFCB}
zqVqkE0)=ADztRgA$i<eEV2n}Cy(x)6NUNc-agZ*E6W#N~WG#Mo*Uj4QSxiooVm|tF
z9Iyr0Jve!oVk9j)4D{3BY^;3D@rRm1z>n_wvV8urUN+%AVJ$-~?#q`i<0#SBy2@kk
zJbD1=8ylLMaGe@pCU9%6vpoLrp@|9Ke@~-jeg4dw{{Ju2b<sx!YWkDXZ~@wJ(9g$x
zRIs-J=8UodB`}}?UHMe|6FU}kc=U92q$1j!!DQzfEf6U2xyYL<c^9aQ@&P{ZU0Pxt
zz$e)I``p3~j%SFU8GsjTdUkri>+%?IL!b`VmO}2S;csJQ#m^ZB+>Yp9RdeQa*<0Wn
zOdj44SCi3xY*{*FuvMuHl!`hNHN>Uqqd~vlZAxnDA+HM=vg(sSQ=8C=-0i<wWFG@I
zf+hV%!F-+-MH@Zbz1GkM{z2Fh0W@|*LFmYH9zN1u0!avRfR4bop|E~)KhSPX=p@oK
z5DJOk@m~<gsL0nJlCX2GaxK7iKFM%5zW>(r+$az#hPDL}zZgKn!U|g#7w9nhdQs$L
z%EB8_)s`&U`OcECPkN{)hyWSR2%PD5dd3ytn&)M~%R&ai)|0l}65<6hH~}A*_)xHM
z*KhN@Ofa#E`(jZHQvCEbGiDP@Fm`QnN=`gQ^B;HkZ@n7-VJYZ+qteGLZC%=o2bTAh
zl+-|CGnmT4=?M_W)Bz9|aj^t`5prElu-#2nHL{ut!XtiJ0>sTDI623EDv|^;-)O+z
zss|GNtB46E>yUI<oOXb*`^4o$;b(CNGLxv7s!gC31PIgG3DKhtbzcHl{yjYNca#MY
zR0EFkaI;zZR1?P*R=A<APVBc=Iri)BT9ujF??2-}gw7eTnzaH#1^}W4o?Wq=>qy=F
z<lxhqvQOG~P7_&sAH42}HZY;drX0up7IfTsMf%|mblhyN^G@F*Tn)ck<g0aOR9RUW
z%csYsZYu^Ji}|@v0g2))Uj=SM#}&`1wvH6ki-tbMFh{vYXB!!me)IryObfhZ!O`G+
z!SxD^q|m<#4_{(bqhOHSH`MYfD0p1nK@8)%efupKCHMs>*HqTi<xaumn4RZy+tDh0
zZ~(NuA)Gt1>7*es1%%iRsK9xjp1mlyb{{b4syd*dso8ZR>{$VLv_T;{>c-W~fM6++
z>OX)2J8>;{O?f3YQ&I=V!iS}*uFh7DPeLkyO6NUd%Rw;VTfsvjysb41$lk!n$?Y8L
zep8>*uc5Dv?FWi}Y(E54Cj4WSAeS)ey`{P(F+YV}i!qb#CzX_%cllt~{F;H|d3W<F
zaclRm57^-L6)`z^S~|K)ugeQ&F244M5=)-dOknz1I7*4>3{2}<GFJLJ2m>*Oe*A|)
z8HJ^wN75kaaobj2VazFMY*)V~0)fDc5gnHNMoOGy6sr30fFf;;+P3R3)8Y5ug@jN2
zvOCrN=DpO@l4W+uwk!JkKiAAo73^i564NYR5$ze19`e#0Yd$<ujrwuOcu6rD2ZKm}
z>S44k&Tq7UW^j9XnV_?=(drBO`u%?NXHGFYyf#>ccQUN<Nk6aBR5d6x-r?t9*11(K
z?{w`l&!&i}03gE7LRlrK*u-T*^KS+H(z_LB<qn3k<s0q&!~B_#JC7=lru>Ge2SP$h
zEPXHumTsM5ogN)7q?MLwb1pTQwZg7RGGV`>7YZ^Z&ND6B^-QPyTzWFP5M<-Y-Wb!x
zOUSF7*kB%fa-B+YTXLx)QrWAV)xmJ0>%1zUtszn)7#rMrY+w4UpSRSyhXWA?$SJeI
zgz-J*2A7HHcXJ}v-C4Z=)xmGnOufGQ1jmy}v3E4X=a&fP7*oXEmT8oo95110jl5P$
zH9hK1=jLRv&j*m93Q-Y{Oniim+$Tby5k-E%5}}NqRzd_AF#l?1a&R(NTC-9%ocTxz
zyed=UDRYiJs-le|9=2-luJakw%W8nJY1VX>feJ9s{UPl>u4Mr;e*;h{H$Yyy^V)`w
z+sw8lI?-J*Oqt$1`(-xG9C5T&HHaD7#k*}uAVeUfC1)!00afd-sAX2XWr^+%C7-y+
z=}Z4+cd_qIB)ym*37rVHT8&lUa3QF!+VtiLdGD|1F|xd4zQB?<53F2#Z^WessBiXw
zN%?ljbDH1sSOL>oe~9)_xdW42{A<bGfp70kLl>zSpEbOM=07gCcN#CY=|e+q!}S@&
zpIgz>Q?!wLkAN2_-(jp~A(%q#Wz3Jo0S;M;41zs#Km9yK7M%&PYnRA5pzg$s5^xe)
zp<8tWkL@ei+Y@1&sV*I2*2@FM8-@dXpIN_7n4Y}7$EZX^&3zwCHapY8EL5r#bvo7)
zJZ@bpB<u_bLPrck{2`a&>vJ9yLd9gZp&=((U0kwch0yX;_sLfqha28(O3`fjBQ)6i
zcQcgZc{LXh>vbv?-h#f{UH6)~sl7AR1`u$Bqny<bSMM)BvGjrDmuS(_`;_lZ1J!#7
zjrI0xzST{*q#YURq>>NxZs(K7-U((0o=gdptHPUEm@}_gJcB$w++LtKYR^^0my=+k
zCEv29q~U!&;?&$&AhJwWpPpyMLX7ltW>;Jrtu`B@`L3`Iy6%qm*_Mb*_7PPC1<fw-
z<MsNpB@a9G&x>L+@3gyFLg}yRO6X8Q8`i&<e46Yp$mv&%26sX$jz??FN&anmg3z0t
znHl@@B7VQ-WrPBx?^f`2z}+2wl@^6RGaks#_9)htL@okNrp=}3DBw6=>rRi-)YSh{
z%^!TkFVD4?i#);?UnYwc8y71|JMeI(BdSk_JQXx1+I>6h-hr=f__;<+wNGLgp(h_g
z-dgrP9Jh_JuuiwEKmJ(!ZJ&(Jwnv!5J|ScsprTpqoliu-<?@P4@_Lbvbw2aAy)w0W
z?A|8v4x&@TkN=8A4*#AoO9OJkw=v_|@c3ykzfSRbtRkjyiT~ZJqrS_-$jV`&qVorp
zvZ5o?leMm{wbC^mhpGjwZ&8cXD3O^5u@MB}6$aV7R#>lg;r<4VEF#`-%sd8PZ=r7J
zj`<4Me}Fo!UaUXmzeADH^*bp`OROF-Wy%Pfs(}9FF9<)$6%Y*2B#ngRCq#y^Xo=Zo
zB%!$IP~Ke>p>INW7FJ>e<il3FiQ%Dl3*yUvIC3IV=mo(ID`|-hZRn?n*hj?NRm*~V
zOyWna85z>H)M?Q9s=a9BR#UmI3NtNJnleTZ)P-GPbZ5@-%CRJpONB@P?8s0;xijCz
zihvxJ^BacRVSmP3=MoA#!%Qr6Z?a9C>U<%Wa%4gyt%8-AFJ6MuUh26OUmMJ#9n5jv
z#M?=HAuA@cf>&FkCqZ;C>b4@x<TvN`xQB)qPO@cw1tDiz)BSvZ)N3s=r(rwlqFh|U
z+r?j;X%RNeqZAJd^xoQkn?Mbat+uXIkw!@!TD4g<Zry}T?jXP08K>dYigqN(4%^nN
z60#S6Tw#h>_~BtJ!7EE088&Fb7=G3dI3}cw9X#jAd)Fmo)EUZ3N*B|bf@@p#Q_4E^
zQ?iW0_slJwwDLDzyupxkOoI1oMMH*yMRg)-<7my{KF=Kk`vnsgVYZ4yTT-#jUsgW&
zLI?iu@>B1+E>aG2-ZIV5=YtW?EWzJ$MEqR(Vv<sOt%2U9O+Zsq&J4MF?~Qyy-3f#p
z{xkc4-|9{4*Pe=8qT!kGpsN&&;mpB`)y+(P-t^2AaYYQCEgDemlf8GfLH;ulu#T62
z>aR7f@g%|SmCwvLhcqW+&<TFjFR9{AiO`wW_-2HEJ-2#F!WON|C7%Z9y@rc-IkWL)
zg7t+oW@~Jt$irOSc?dlq8CDW)Q@COKsKSE!J+Z3E;##SVv(~MtVmO@jzG)>B%KD7N
zir6X(=VeI(E=FcGWhzaOJxgu6?D(aLX)^EqZTkJ1>cf?<hbw*7D_+ltIEi9?Zf<gH
z(Xqz+#PoiS_N&=%VVb6Hf83xX$+>W}f&rb6$HlYqFLxI2q-XG5h3IZhxj5GDHTG*F
zn|z^f92w%pDhN1TkEM#$`^xLZM?#nMLzlM1$6rM5Y`P7mO|DE2hZo)vF>iUdU6OXf
zjkch9-QdYhuh+j#(}(x|ctr1the<e7=v?iIe*98J>eA*?z~o91Y5(=YnDXYW)<V4U
zjUio3WUFE4mQ~y4v)$35QtKUb-$Ln3p#pPVr>7miVZ>O};#uxvxzcMa`gY_mv1fU_
zOB4ng<`3y~44rQ3#qY?j>(dEkb2#R=9=iIh5IPc4?-k??Y?g4x-WjbF<M6It_fm0X
zWjXde5;=WiNJ8Xy<-n^hRN@NNs7ZJfluej4MO!1aQ-G3**oC-{rcW##xf7Z4oJx)K
zH}xrVp~nkV;=b{x$f+wKF}bR-R3eQIVSRkcm6xO~Rg-mnX^pzR+*~PrF+<M6rIu-s
zz7G4@i8Pb;DEvf^1ueQTwH&&oWJ9xt&hHlivt%zh4K|O3KmF|@neN(G<UAOHSOiC3
zU+K<Ve(w7R>Jl6B;!6c3nTwyw^4<ZgyY-~^V$7%B90QkM(d=Ch0=Uj3xH2+^GyGBN
z<p*C6Pe|x){+g6$vCuzA-;CK-O}-;d81I$~u}Tt(e1m#G;%j|?*S_aPsq03A)%gzi
zB)ZL_B;%T*n3(D#yp38y&o4>iUzZf)$v?$yM+PEC-9n5qJ?3=&p7aCI>#y8z^umaq
zqL2DwyVzU`#gDEmrI<20?3+g;o=cyA_<_DPIEA{nGU=B!Wa^pBr~{LDlqZ~R(e}wP
zizLz>RRKjp&kXdHVy@SP(#z~$snq)HJ@xb*x$;cCf0MtW%`IXp+`in8oxj{9&qAr`
zKN_)}Iqi-|{{d>aK4D7=pWnF8bpP<tW-Osf&B=;GE)clL=~0)K^t5tjYz$_y4L1&=
zr`tU3^pa|pwM`2oIVE+BkSo2~zoEBGhKU*311$Nkus8-Awd=<2NZaJFgb}k6n&Dg%
zj+4{<_cqBOL=ZB7P5Yy^Zj^QUx_Gy-v8(XcRgfIjnL!DhqyrgV|40~_0raQgHF3)@
z_qFkXZ!CE(F{aa^RqzcS?)EJ$cNTX6L&FE98n@-9RDG$%4xUcT#fruryDTv+Kd92|
z&J)5QKUR@P470f(=`mGFl_FzI^X=ixvlmjv%#!gx>eyc6)_quG=y$SXz0jR4pz?}8
zSV<Nl4d<Pzda3Pak!hY8?Om`HZgPC{4*=cJcLfHeKykz3*^Jxh6rufqNJcwYzRu(e
zUF2`&UMjQyBW~e0v-hUOTS}gdcw*q?5{N;lV-m5VKs>CirsX7YCw7H`x{nfz4GYk>
z86<#JUt269ujOau&LCm(<}t2<gN^o6=BwF%(&pG2YKyNeaYd2H@v)RWo5mzO7O5Vr
z8!SXT_v9?EXkx#hcTplIK=dy4=)NAjcDiv+VWvmsdb}3zG5wdX+{Y@{2By=r>u=zW
zwMZxxk4`xM9WL<bBkAIKzt3kvI3V{SdJrLqJ>=GYI^Tz@Ax9KtOs>6}nqu`QHQU*I
z*|-Du5|m3d0U6g4L!ky|0X0t%BkzqyC(N(|Mv_Z!Kx&wyh|^uxR|P@+8PLv{EP^b8
zP7g7IxDd$=!K|j1ScP!vY$m`4EwFXR$8vxSZ89KrL?|_V^JPYS`@vjVTFvQB;k{bH
z7po%!OcqfqT_5h@-<RkZv`k~@40loRT#C2><0!l9zCQjO1V#58QVcb<ABd#BIY$mE
z&=d_N#0A99I_xvljh%%aPDC=!&{~Uf?J`<W7lPJ<ZW3EG@thVVucKD9je0->?epd0
z<|ZsBmnOulEc9yYF^f3zYLxB?-t)i|Fh2ue*}-(ohs}hO@e2Ni)T2P3Y@637SvGTk
zd?gIcXdDi>?fE{>6}$5u$a&XTdU4}R>fl2=iq7yg#Ex&X<I&=_I=Y22mOYqECw%$>
zo*_qGp0RMr2b!Dm*tgJ2&^CS{u1D?5+;vokEm(Eb8;&dR;Jf(366zX>TPJIN(mN}|
zqFhLT7cZ+YlA2IlqCuIbddwy+eC<(F79Djr!-u7B=8CZNe72BbRf^8?l+BmOp5f)K
zbjepW_wP-B)=mHN3WMsAgN@N*`ox&0bDLBEqBLOdvumFrNFbEX=gVD_%lCvd@N~af
zItR2>^P^0mSdjN)PX+Ay8wR#K+)wWRR`=bEfjC+Ddjpo4dw57LCt3tNNW62BO^J(V
zv@Esp7XSn(e>?-y%rS59kQQd*Cu7uk`<ppSyLTPG4;HMKN9>J*fK2*Sgnj_6TUGeG
z`B$C}@4SFq=M6L>fi_u&K&U;tg0DxVCn{uAh!`(P(#vfTDDV`(K-2?wOBQ#v5xSb=
zX}-Tc(bdYKUUITtPu9G<FkI()@GW$dmp-ODCBcRPt#4I!4h@S*+kFzhB<n|i{x%tw
z7l<^0JS-D#-<|#y^Fe7Hr1~-|_;q3Ug4=Hw#xQ>07ttRlE4t+Oqm_AI`HF2<9Q-b?
zq3T0y|H-HjC?9if<*{yxxKV6UShe|}il7a@aSNF?*kK-TL|yVA)oZmvpjG>}zASF{
z8R`P=!aU0gtSYkUQ}G<t-WVCW?Q~}r@s9*LqS7#V*YapjnSBz-6*VLg^`n4COkSPJ
zLirHzU^CWxBy=Q^h0IUYtr1EZC;WLi9+LPP(hox;=erONi2&0m%)fsoZqbo)V=udt
z=m+Rb-Hf;DZe3%-gjaj*y<XW2pk%5QkfYQ#pToXyo|WTAQ9<Ku?dGZac<?182=?ai
z%IWOCu^<xYYc^AEe34I=CvWiWGK4?sonbNYE^3zA&^wNNqDKug5{RM9AVX4>IEzp|
z#NO9REDmhCN}HcCe)1A;S~w)g%U86&sxs$GphXOo(2!)3=rhJN_pQ<DPARWur6*2g
zqzm<d6TaZE3SY?1w`a>s{}QZ$y8Vfe8@`(PB-r(vH{k1P@)D7Jilc>6n3px(ti>a1
zK<LQ{lQi1^O8u8%H&9=`*sa2B2a0;Zt?e_g!I9PA2!;Xo56kyO9zM(IwkTsOYc9Sa
zOLmbk5_K<v2HLF9NRza)T+-R($mW}-Z{@F&AV1JhHV}HWTFIBQI;$6ECi>Y-TJbgg
z4Q)S)Dh<hcp&=b&&479*9+>gQUU+{t$mn#LCN8pC<Z-PQYhNW0iq^3pKF4^$6c-V+
z)TGOz)YN2rrIn7+gH`LsEn`B|qmQ;Ym%BqrkVK&OkSk=WdhrAao}hTu3l}<!T!H$}
z=qQpw(&onnav`)oegYa%vxI9IWP2BzVIFWQ_q93Sibd2E>#4}PSE3w!*{`V*CJ9Cs
z!tzJWQqW6IeqU|+Y+8M+sEK(gXo^+QqJxFU-)v!P>NQNAAwEBc%^8gGUVkLC*stkb
zV>-WD_hFHAf%Lbu*GA(vZ^mBY)b)6)gK;KH`dC;aPC~<9YZh<xQ#J`Wq@mj>lv}^p
zGd_eX-=_9GCAQ2m3R=7-^(ND~aje9s)&D%sWb1p`GS76j-z-p-Uq&@BBmn-$xt9ge
zH&!Nr<<CRxvccbLO6=$E2mDy)$$wtbu@G?>r8oJYl<YIq@;j&Q%<L5N4*Wi`6<dDW
zfF#B#k9go1x1xn}nq1mCCbpwbFDvSE9gjiVg@n`Jf}g~4gy4Bct+MAwtd~{$o*Zt2
z&?g(D^>Swl`P8q_JIV|JZ4>nGfij$qMzK9G!H=H@%gV=m@A@3}q<#MW)yEdYy9Dy+
z+zg)c8M^UhE2%F=mF_(HT+nm4hB@;UpHB}$gTDZNm9^A%@Qd0~|4!*#sn~igUe9|f
zstrGErb0OSF}qUqHWGKg=BRQzL?}oe<@9|VQ@b;203Fkg)jW=?DHVOOIt<*50a@%2
zz<KDJ;Tq7qlhbeP^=Ywtx6kziG{?YCmQ2pj;F?^G($Yi5Kb9?imvZm;L<(ot)zN7<
z3%VEycE|Lu|FQxyHd(Gq{kg(xpOrp7DtyfqU%K>LaX)(O_kqDNwZmGH(Z`(rq(W{g
zo;zBtVR6=DK6{a2$v(TIcUQW;m=ik*;`Z;_`Cfyq17oHwTuH7=-FF=@F{2<&^VR<3
z`xzPGRg=xV<c~JiNg~B7Kn=lqD`$Tw%f{vt#2PJFY@wRmh|RCw10PvARue1NhDIlV
z4^)cA$2~7OChv9GAw4~awFqV0PRLA9WNdujx{yPcM5)#2O|%unX4k6gNh^N|#F<T)
zwh(jb32F$k;)nAez|2u<H$dM#bs<FM#Gw+a-=?&{i1}2tfBia0vSi`79b@HmKhit@
zWXV@glgd|37SZj`8khw%a48)CqtBbME7^GeeQ)hwtb20uvIaHuq)cxcKCEvzyG{jb
zfBf}6Pm*ZVY^_)dct6}2<s=;{SIV9PN0b8)Ku(V8Qu&iNZ_GgrWTDmM)LDG;B`f;#
zr$JB=zL_Nk(%-DX2OTeJVnDtxK?+d*=YZf`Koa1Bg*zq(55%V;V3~pT0C-z>q=+qS
z_ozpi5)^C!uEJ~}`f-Z^1hrj65v@Qy^MM?SSP7^}senVnvfz%Ajl}s?Q)28Wx^2-{
zpXn_gWy$ZDxlEkpC{3K9&X^86D6hZ>gt|LE4DP4P26f|5(2x6emiZq^xTyVrP3PQQ
z{l=;a4r@>g{InQ%wsXAInF}^XB9c*3JTTMm>}YI3`s_fxO`^P0>Zqa3nSxU-Lh8vQ
zV-Isfzo|&koUxPD7fu);s?qEQG6v9$_CnvCk3#k%-+F<WL9FdDC%fD8AEI)Uc`Mi$
z01~Y6yRXICv%DCHUSn4>XHFX_ajY9FD3W(oL?Q<wl<ObwzmquiLnD$V&*T>B?GH9*
z1R6rGLUS@X$&nCs(v-9HQ0XjOYSd?%m1(fM<W*}y<(ThLw&<8Q>||CHql5wQ?*lrx
zC<!$J;A7MSO5ZCco(tIn9YlOp<a@dLqWv)LZP1~j4wUDWJ{gpTEqyx!sz}MBAOQNd
ziVNMyB)2tmFhxd)xG0}xvolydLq_XPzX{V(J3rg1GA17Y(tGSAm^E+nv9rT|OwQ_`
z%4R#;H2+r_>V^bJQteHgE`^%9^vUzXWlQPg-7&v2cM^!zw^W;qV*ux9uZ;Dv=fAxr
zweZ4PFFY#=+M@DaLJV;wkUH=!i0gQ{?m2f-qj;SdVsG6DAdck(!*$`KJiy6Bu%rp8
zoPk2hTW|%}CkTcCvmL?{*x-|aE(&!CXup%+CHmiK0O<hmvm4jUYK0>h?J;`shpHT=
z%xFUB;!dBlI#-M~JG;(fK>364B01&=U#f2mPcQ0f+6swgW<FdOjxVk=J3Wr}IojPA
zfy4rsAAasg+$wR|jD^``hu7m~Kv1eUf=OD+v&wtTYb-pB3at;h2e6pTTWBqLiX%fT
z2gtdpUtxHzu?XNdt(fyuagdudgU_RNKi>Qa2fU&^Kym1aJ2aLLX3_R?;$A_0nSN+<
zi@OTI{D3&zc8?-$UtXf42w8vAKYta-TfHQIWpka&(#9D$CIy;4P$%K(6CB>`it-h@
z#QOfq(SyrS9~(%<fP<3O)RS`{`{#9rstz^<ByhWA;?dVo8~nYYwZZ-<uK1M<1UuQz
zud(ZoJ`vOKGy4Eh_AtjO(3bRi@GT_t^c094J(@2Ow&ioc>>n-Wea={9CyQ-?JcI<<
ze_zOX4iulVKqV^+$g{vux9!wNF;$W^oAfjmv3GpIc7aG7_%ye>2%<XDOQDl8L9IF!
z8j)Jb7tFneK(`2JRY3D*&okk1X@q^WaHBqI*~()a=a*lo96Pa35`OkjXOf?SqW)bO
z^)}@PQIf*TeX<2<p{m7WjU-d-N|Aqgb+@~yi(RcDENYE3HS!l^^R%k{T3?&yX*pr*
ziIvhP9DHo0KjadhIMPJF6Chp!hg#&QD63+`OZ{|l*Trp<fZj5G`Y`lOt_X;gyAhhK
zIhswF@%+>sTFHBggp<R#p)LMsa@uicAhZ?CDAZ&Q*JA9JgKq$q-Lw96&P!x~S@O0c
zl7>nyP)0+JiKa}M?|PXkxqPhhC}RLL-UVAHa^rPFzn$(UjhVnnjyllmq8fgkPpv+b
zOda^JSMLQ&^HU~HM_x?Sg;=kT|IobLFjaD9o7Q~O6kH3hS_uyT!+&05gNi0`*~u0P
zYA)5;&a^Z&kl42BP=15h-)w_*%7dP?A76U_)#WSl4<;C2sHY={c+QUIdE^{_AXtrZ
znmKGUFeG~z0O8EgE8u0|yTqJR&5?jSS|(cJ)+=I`1ZD{x#}=YavaX?)_TSq<m#1%I
zV5pm#8+B4W_z89}I?A4aJy7;w3}sQ~7NWH9LBC9}@WKW5AtJKfcM7YbfqvNg1QBnb
zObD{AYGFGNs{v{I4@iOGf*P{utj)i|sz0yCLr@hfL#4yl3^f=K^Ug!oxpw(44?xWH
zp7ic=5yS){0=ZYS-^kbGQ|Y!sizL?N#}^h_G>1@3Ttzkmj)QLCzb7~H>R9k;@4DH7
zRMT7>i;fUE$At-BmBv4SXC#5E5^S>o{VV1Wq#}2))OHhc_EuoR6MVt8vId>u<?Fh<
z+k+Dx<Q+vT#}Xa|B6dR!Dr!2*8lUPlxPNDO2=>DsQ%XK_hU3+7<S=ulAjfR9BUzQW
zm{diEktzVVu_JbfSQMn#*vT^<WYhEs2GbPO+u>6ayJiNmg;F*Svv7MHZ#4Ia)@(#3
zeD@l!4R~Cp#IG$mFz!2&cJ!saiqE^#%qg+iQu<rl?z5}+K+UrK`Jj1xql%7PC~0JK
zTYYL+F-cjfLuV%sYt^Ld&mSsoeY%en=3N%D<Lz(kvVqes;tzmMI)cGaoevtf>$D_O
zF80s3LU<vf=uo!C;fW)9<Mmh#4hKw*X^7qI(HA@3>sQ#kX9UYjj8%s(__)ps*Yn>-
zJuCin{H>8__-ALyeg&XSdMN2#0R-QA0M4S06()8<`&P2$AO97PDTdD0_T;KZEhbJy
zN!K)uTRt)cHlJbT-?pA}2yC37ah<2`Ch$3C=RQX<6e+izF=(I=#->C9a)iJU0Mz+v
wFz7EloZoPccfjg{w!b0r|JtQ2twU+@djXOY5uAXO;lc$a1vUB42-Cp-0}Tb-MgRZ+

literal 0
HcmV?d00001

diff --git a/figures/dlscratch_pool.png b/figures/dlscratch_pool.png
new file mode 100644
index 0000000000000000000000000000000000000000..34896a1c6a3f90d4d853131ef9b6efef457b31b9
GIT binary patch
literal 82599
zcmb@u1z1(@w=PO|BhoAc5dmo_=|&L=k&rIwmhSFGx<ioe?nWBvUUYZYx)b5|-+TYh
z-OoAup1YpMU@_M>zZi3j@xJ37D?sLxI64Xu3KSF+`g;ixIVdPNSSTpi_eco9I}S5v
zdB8u=)^g%PP(_0zTfh%+`hrq|P*9~|sJB}1z|Y7Q63W(4P&nTo{y;B$1KC4CL9*YA
z2+G^5@6IDSKO3JxIHb<P5h6nN#&OIA<)(Q)qY0$96%iB>z`_>97U>ApOh-m}ErFYD
z54OT)gszo06pBtG7Zt*Mj|7b)!Pw_|GC1fx9}egFi*;ym(Q>HY1cs2k-a3V=vo{Y4
z`|%Yo2EP~K|M>dosrB@BL@yH>ImO=>3!yKx$m8o(JCpn4l|8xy2v|=8ptVCiUP%$9
zMIi8ar7jB^>_3D0{>Px4%6MOsAKrh4jEr1kf0R)r`peKzLQ2XfteQhrO|7zVz4HX@
zzd>ovW<25T=jT@?Du4bslizDor8=D7FH5RKMda=GW-Gl?Q&a!buSGFozg8GPAN>9S
z3}Um?0{v3G&Ku!uhw}AniCJTH9#E(~A4xKoHA2p()Ncd=dSriRouPWl<c@#BUf<r1
z4FVPWQrZ|zPMjy;5BX81V2V3K(vCfGk#PQPp&3>buU?7mxyU@D;^kGlDbf}DXR*Hn
za#{7qw2(w)M@L614nBPA;QG4_RuSCatkV0Lo7qWl_x(58v4{LJj}Px(^yTbN)kYmy
zzUof-|80-|FAh%rn{osxuJ4FlaL0^&P}Pug!4|W>5*zV~__?4#P=W!$90o5!aL1*&
zg4V?moUZW@B}E2t_g|aHeN*<6!uEm(&BX#sK_rLrkM=l5TmXEpxWcm6{@u}lEa>|t
z_jBt>h%fris|-5{9@lFNEJwQ<5NwW_ko+~?p0U6Grr6`$Y3T~KYM!~kVJF6d^JQHQ
zdvFglg&UL)oKzxgoZ-cI%dyPN``d#fct>jWT;wtHJn9DCEge;_YmP5Z4v$?utCHUK
z+HZFi(5*v;!9@<Bn9v66Q;tBUHx8HGo9yqW@9j^@;gyK7VOf(rJxe*zM?$p!+O7O{
zB=(U5$J+vB#KMEk(2M7Lu=^zL*mnsA-BPdVaQgU@5HB!)E*COXE@kXB{U#mf#%%32
zxA9>uvt%|QS-3OfAFa$3)0N_LcMEewQR_DGFpkB(bUdQykFUmBbm2f3p2^YCgNTpV
z%}M)zWH+RYYc%vn@%j6B$o@JJDaU`VbF<KVq<HG;JwP1xLV7Xe6|~umWqB#N)(uYo
z#~mL<p`CSXID;hTkqkvQo_#7}{L+mmE+zb^Y%-Rl0*L|wW+?MQ^T46?(9UvGqqg?<
z-0RC;b)<+o>%Td`&DyUW!<0O5m0uYml>@7m(s^1nZA*JKt!Lhqy6IB<jKgh{BOTG%
zrnheqd@2?-M8W>>7eZA?f90j^yd{D8xx2gw1CgJdLaO+Q;`K_pCZ=gKFLlj?8v&)S
zU03$_%Hc?v1i{<qBPGBc_&Xo{#qXHlsHvM1XcOWEbhYb`UDvsu?;tm6;KG4QM;wfO
zu0Yh87+ED;Tks7dN1Nxl{ZADn)ylicW0Cg3i%h&ht=UzosR+$gFwCx$@ToNOU(b|L
zM)d+g95*Y{o0XVaYjv`Dxy+HAu?Wd%7_Mn1MsLV6iA|oElX%0<SIa?_T_&3<0j}Qo
z=4AcZ9Gq$1>mN1amzMe@!Q4vjz!}`rJi0jfD*Nk<gm;)NQWAsSrM`U{Euh4z>AKrR
zWf0qOvVh=<9NyVJnhdS1QuP+&%b9Z=Y&EY(=k|ks!A@&9Sawr}5vp?E0aXLT7}0!h
zxUz`+%!-1gl78$=T#r%>{8L0*af-$?)1s`+7OaV)KSrI=@U$G3F4)$KMh|m4l9reQ
z4e^a_c$*OE&R?)MKOm_1^2UqL*wDZG81vepAXYNtGz}OF?_c5fW3~<$PvymCVmU>y
z?%>KUJUV6Bs~-+%v-h}K`P_CQ&EhI8f2#D5VuJLse~qIayc|sG8{QTX34OYjOHZ&Z
z^vC6X$y@t**_+N6_IZX>voC~u*x)jT<t)Hff^^JS5%EuxwCYWG7Fw&bvC<%C*R*5$
zk$8m6)i*);9z~8(ZOm6-E?CmPKKkc)fIrF82adZF+#9WPH1{LWIIEq}%k@viB^mYi
z$g#22FdxGPtv(A@{XJ>}vmr6LV2fwOWV~*;zrWR^ZHlb<!}lUS*XmE>lE6rIUy~GP
zMk1%ELw@wPxk)ADS4-LrN4vTm{A_rrIp>0s{EB;V*j%mie^5CfxC)-dIf)9?6jmKi
zb_h*Zro*ARELk4DVE%mFbplojonPFYl~(ar-rFtn7)&F1=Fx2w5&7t+%^{=~Y!P#w
zfP79OW<q84wG*<5ZHPgy3GSh<w_usiGmUi{yca|o*`^u656>DuiS<MsiLG#ix@}Oa
zc!wQ9`D9f;m7^}evm+;(r0j@MH&xz>u6>B##3^=H7Q<JKlFs0h^CkpY%~g2mnBfVl
zjG$Q}@_3}z#<6VjEaRZC*Fv5wOc@8xnSa?jX^$USM@aY>6Xw}OZab2>Do9X_h<8aD
zk)B(ja(sx2Vk@vPfw50i1oTlpe=zxLx}~1_K42P``+q~T-sf|}+uJP;|D7}Dg>5-h
zcw76SPaWy>iP(R#p5Iog`7X($7uL<gvn~1ux7Akcpm|8t`cI)wf1rZetD6rbU-=$M
zo<Y&|w4W)vtW(FmGH4nlB;`)kVo^ln!zF2Q^~_{08+LPYjKdQa>}VWg+Y`7$^48D{
z;!8t1Gg$6C$`KC7P;F5^@@8S}>#UjL;IU*~A2<95^u(!1M{-D~a&|`wr~jc1Dw8o-
zaBtJGKe>s78zV)a;5?~(BkIM48+vboze~9|I9s`1Ys7XwPGAgm6k&?kKT1R8)ZvVb
zlM;rGgG4N$3i37fXwuh)6Tz**wVts~er?r5B%1&6VD#MAT6nx|oyg3aPB8cg_zA6J
z_u$Q%L24khFCP%)`RmD0bAsCue;-|_?}vWMoLp%*uk^*OCS>JXknbTEcq`PHK=yTB
zf(Lul7>y69&;$2#&eAMDuH~L4G-@P7H4ynT<r3(ctc!mUMvMnnME-kH0ZQSghXg55
zr#lop3~o1AEF!(Vy{%|mujJ+-IXXHr7|DR&9xE!`V=Eqg2bx%gRXPH{ZYXbhrJl%+
z(#Y-GA|9jE|4CrAsx$7b^Q(;dkj_=i`pjR7z34HoD^DU){H+5FjhM8hBn8GILgH@L
z^{1XOx_omUu7D^tjMvj|-#AVAO4i7W3Aih^T1!9k0Fsc1@NjohZo6aN!PUn@SZiZy
z!ET2V&vxW1_~k!;gL;3_j6PMyh{1rD91)=aEhsgUA-qXhQ~U+pKKf7uI@^Y??Bb)~
ztSf)!baLzD9=ukgJ%P%dI5^fpnp|R5QI!Be#-a7E#k|J%g$5iC=kk?B2axA!LH)o_
zV*S5MDgQ6h9pE7Tn>_XZ)1a1Rw7^pjylBoLXZW9z-@$8uKs9x>hE4-?%*@!z%F4N`
zTgXEnegMb)H_L68H8P@=lasS>+L=H?K^Z-dp$F39sZwj|nQCK1etv#a#(4CnG}YG6
zkCKm1eU^fPw&6Mb*R)T2GcyX)ZA|WPPZwZ-_<}RLL(>KCt{eXR;w(xEiXM}f$3zhl
z&`hhEYygtqBWKP)eme~PmF3|oDPlM{I0L7?=f2P{pHAKCqv0IMJLfq3-$ac$BFasF
zlp`NMM2mR<LW7mCfD!yZP4oXhZQuV6JN!=v%_w`x4IPn)ueIF6hh!>?i*7vDv3GIN
znwU!8GUHE>rBb;OAHarJifvqaSRSzn*|oN}6wu&-n7CP|;!<R|0oDnerEI*W;PFvq
z+*Cf}&)qYToYqvG&L#tr5I*13oHNn+o@s(}3;pk)N&@U7!kI~eRJJRqd){rD@1*7M
zMQUP8JDovH`U^tN7xQn?3{yjY+<`<FE%{%tOlP>jV`I-FJvyT7!NA?AfjD+Y`n<+^
zC{{6^0QMIihZb@eTn^-ETaES(Gx7~%uSwk)2|d(Ze(KnHA`>+$M<_9CLoIw=e3m*u
z8l*ovR*{Sg{(njKx(+BbrNRU|@}>!jFG!Hh{K#O%v~M}`OK$98$|^S^N+|A@syFRs
z2?{vc4^U-aJO*xB(+PW%RA%B0r{>UwYI%>Ks@as!YZIGXtI(WuHhoQu%-?J|BTfC#
zXz?O~y1>cf8zZ+r?(JVs3D24jAs3y2y8OY(mWF4X&J}NPlxMJ4O3AN!J~$6I@;7m9
zdttbAYwuq#mT1k3oMHrYOM@_k4lpismd2#~KQw2!g!Jl(F7u~KwVv1XNenD0;bhD`
zg@>vR{*rG;3I$~6B<ck9Fn-6o$NRns)$=~zjCf&l^zZyN(4DLQB$S{oy4A`W!c=#&
z&I>K&?KQ_x6WKO_3N0{ZUoo~~2&drEm4z0!<mx|f+Y51sV=kk9S6vXcEyyNQt&GTX
zKS5IP^_D2%U25C;dzWFCDNO&jcOoc;oJ}S;U8bBlrns_Zj|s>J<@2MvspwzwZ^TQj
z&*9!oIV{+c$D;gj!Ok6zaKMU?-J0<o(OtP6WO@_Uhmcnx5~XQp>b8*NnV2SpmyJ_R
z#J?Z>IbjeX3NK&{H0X7M`1wd7_N&7wiKY12pzz_5iW;~esXj<+<#6BbMr%y5`wYNm
zXRJ+0HC-*dqW<?>Pv%qLsewiG9^&=`pVn!eO{V1|!%~>ktO<H8_fBTl*(X2AFJtwr
z0>4DG=6^Pa8_XJGJ%4GU;Tdn(c=*9!p2sgd_kRHtI-@^J&d?^L)q{RxR{pWYl2v~?
zfijdcvaTuRgEXhjKbtqb)b^!?&9)s8YjcFRh2vlwx`gq<TvLLNA70w}B#*MAP}7;<
zWBereW#00iaK-g*fP?k1xesEhk>+fVYLNLPso&K%%pS4Z>no)NXC`%n&tGQ6MI0VI
zg8WR9GZxM@?W_Ly(2MIz(QrxikR~*dJqYQD(L6jfZ{1SN@2-aE@)T|#uUJ!?#r4V;
z7q_VEeoO0YCgXjw{*4A{TxYDGv=4?5OHA!KzHhM-?x2zoEqnP^vrlfn=d9Uy&0wJN
zO-pJjknki=+dlO<lYzXL{1d_P15vsAs4om$<)|qgOgM+(3cvW8q52K}5QSVm{=)kt
zuGsW(bn^d-M1=a@A1?)_AxJIdn<}4^bStwK=IV!Z4-)dXK?YI`WJl7PNkl~d@h`EV
zxh1u6wJQ`@moFDIuKCCYtbg%R)_9Kf1%zKcGyjCbpvpcx=8)2q`*RT10ymO1LN7rS
z{*43*1cS9u8hoYTs`*UdyUU|R{(eJO79~xoJ>5;fVuII@LgMF>Pn{{v3WDXZ1+d7d
zXqJ(Rj~%jQ>}W+DwEd!wAgfrqa3j=M-3GlgR{?25rSRl$ysWV^KSA+G)%Ju6#FOT-
zr5{>*=(9S$xw%ii;WG;s-Z?V@=4Qa~^UC3@5mI}{%deM>Ea<G=XW}<Y9~yQHX~~e9
z4$z(JSD>n#A@ANPG8)8p$|t=m(Lj|`l__-jJodJ8+?54`Ud_9_F;$qmIZ2PpfAigw
zjnV8QVYnccuiv?s+>LNLUK3QA$N3i>X@HMn=9%oT9g`&3{|H#rX~lP!?D1n))me+5
zI#s~)n5rOr6TNPA(>flJ3Eouj5+sy;@fb8{`&1jS#tP`JrQ!kt!t}M<bB4ekIj=M&
z3p`%+W>RQAA)p3^x&J^wlzo96ZcUt_-M%(L<WW!=LQmF6Rx{Qq+>QXo(9fQ(K0s}F
zYr#$t89XK2g1n!iZZTekQjFuwJ6}=fpcK`{Y`5#2q_{n3*BD|*+0SIlhu$4K&~fgB
zKt5K2YlMRbdVGG@JaJWp-=#a@&*t64G)6Hy4ytE!>AoGzRn5*D&1zQW%a!efv+LMX
z#x1eif7Z_%3B3_htaToIZ?L6zT6&_OD}EPDs4}}8LcQH8Ix8u{5&9Uq9KiB;AqNG;
zrn*ZV1z)CIt^%t}N&XwaGBEbGUtsr|@vtk))Ey2QRZLw!QqZd;rS-a(c;vKFTxRjd
z3aq~w!eh1}_f0CSd&K3&^3?eqiu&^fQjba_V;j+93~x}K>*v}A%e=3M0~Kc{WjLRs
z*Z0qKDZV(Q{LX)|EzjD=1mwn;Z`$Gnt;0#)LHEUmvWR23n{*VXhOG@Crc1Ljnu4(L
zo<H42S8B?mH>|oF7IQ8L(QGH6w|-EA8mk1~FS9hZvd+VPLa;CNi|LtlfJB}82I7AL
zJ*70yC#(M#aM+xfYM9Pb`f9g>Bj}E~LebjdCsan_6M;3R6n;4?9oH<qvut*ScM>b!
zGEvZo4UL4sw3IBHV`MPC$B>)3K$&U5u=nmr;<sprdXEAvw87+_0vut}WDVYQDKXPg
zwz9+l2dXvyj>k{mR7wFI;HlrTw!8``%1yDt&>!kr|2#;+_u|V<hWyJEZl^OPRf<t_
z9!pA=H&33t3wf}}%$9YdK2t#o0<&-XbLkW;{8bHX!||@_bc8*)1HH2Dtb=u2)x8h<
z@+6KOj$B$nH-*RjP&a$~94cj>#N@DYoeF8nf9TpR&>HreR@dRD-2KQ&X=lO+6<cuD
z$k)c0KbNLg?+OS;@kmwX12L{*ChV7!*=FzyTWb)%YVg_~@Q}P+79tVxc?8Y5WAXxP
zD)mc8Mtr&Bub$0|540(InF)*TZL*D0l%8y4&iL|wZ61L&Z47bX+E|@@?xNm^IOyN<
z8{H38i4DwhKU0Fyz3SXBkFS<T<a8`xe9Y&}?MPP8zOHGt7<2fW@v0|OWU9OQMXy2B
zPHjck;hiNXk#*xS_fH-J6$e;*l+P3y#bmSmIE6L0Pyj3ea?p&?)m1wVUl%T@aoGKq
zOlcdT_S-rq6-Mw<Mtq!$$j?6~f`3~7Fz@+1M<8R}3kX-&ITSQNPl|g7zjz-X*R6{3
z{3WFJ8SYQ97XDM-V7G;yjrQmN9kzZ&$SWY?^JF6Owd5r$7S3E#_&dHAZIY=BEi<HU
zVBKB#m-=)A8piK%KHd6$!SzjFL^KZnl1-VCT=wi_v57i$AB?a(@~rXaV7va1YFovI
zICy73+~WOUKo0=r){B{!uNcE`gFmKSpb7DyIrEiCuWU#+3#t}}J){Hh#rdhp{=zEk
zlt16oUmDi~<cRgZkVYsex391tW~NKM?cvGEpYs^^-9Aoh3ryQRl<w3U3QP8D@~Zt$
zLhXa5BMp|Qw~KqG!=)~CV<+^}$}4t%Aq5%z*9w8;8Cyx>m}kbc+86hpkUMkOs@)(j
zNl|^>tS<sGQB|8**GT)uUtp=?VWTgvdicEPQ_)%_%WYS>{OT0NH<Q>dDJp!c)SN%u
zokirlE+9ID5-Z{@9@C3!!MA^ttJ{STJ#!=e;^oVin%#rRde;XMonen2ieU|`-SqOu
zK$rR!&B%$CR7X>>x>#5k2Ob3y7CrN6GaYk-%~3JCEW0{LL6=&i<7)L|c7pHaomL#*
zHmgZs?3Y*D={vI^Z6yOqIEC*ps<f2kE%{3eco3*|C|X9&kAI<MZb9H|Frg0yj0oje
zzTwX=m-L*?V3+UP_bdxec04~J#*7v>cbnImxpE26PO)0omv%=Bee)xRr;ZYC&=lo5
zLw{{fb&9nG+v#7>l@~U*LG$l#9zfCmw(#*k1uwvuxOGX<S>rvyk3kU|JKjGiDCoMz
z4dQCD&_Kw{%)HGWkNWp{c)0kNYAwrt;|AH-FEwJCU%&jP>N%*N^$ZS9q=<?=TCcx<
zk1h`n&v1q)Dvt%^&R`}BnWemJOh$9%rxd5{uF@|*nC824;`6;(zc*_KU*tkT{GkWU
zkS5jB=$@XQmg467z_Xo6N-i$tSryKI3LYOn_@ccv7>2s|zL)0<vkXugMz1ZD96m}*
zD_;DF=5;#P9Ltp@;B|b<ZZ_xN-~Vw>qBSxy@^Id6QEMQ9jf0fDrzAj{?@9z+_;+1J
z#oGe)Mk2|WH-g&Q8CBI5|5|wx8Oog7qKkpetf_^yO{Z+OY=fX-xFlt5jWD!fOR;LL
zoN2$^&IEcagPPEX4=4o+`RBo&YBd(neKGW(+u?#bGxq37w$NX=tq-`mypKGgXoOu|
zToW7mCy)2{e=QzD#5SBQiO=@z@a(Mf*Pk3~ArEhFxIJZ2$u|G`gC=5FM2rsJi#e|k
zKWGibNs{^9$m(tP+9nEBjwUtSe%bY`4PJuWZavENhn81+qo$f%xHB>`P)N4P`TMF&
zXU7gG+hME_9}wj>GtI_evbBLlUn)KsS&H@C{Z4k@J+RY7=M0?RH^Z;(A~vQVE?L!H
zw=Qd3h!`X?VFz=yxH&&gl5P%O+?*b~Xn$p9D<Pns#4m4P^I&~)J)K3W784j4Vx56_
zK=liclr&@v0RQ<4=uLclGwLUgbW4xK0f=Qpb8n^+n}8s9!6P&v<O^{n@&}!o$Q=EM
zd*@G|J{`(8<=n>O(aN`Y!lRC*pqS29`6MPLUXvIS^tlo?VF4wwrgMO>8uPvWDU~=D
zF3>bFK8^<s4ZUqFr@<vi&5Ln=k_uV!1G&l&pIKU3a$2n-2?+^VY>&|ts8nMp7OBb-
zUzRo6A8A*DUAsx`TKf7V@2(+t2X01j7VG_lCevl2rUiYA<BN+R@i4N^o#N)AK?ibS
z{8g^eWXKRdq~Bq%GYFqqw+G9nAF|dLJI-(1?u$NEZ^uZ)ZsJe-B`-JK-FcvT(KR1|
z<T{mVDE5HTO&pnZbtpXL86U?{{lDsL(Ts)`)wL&|D$FN8?Xi)g9|07Yww=9wfK(N=
z-MHJ;8h)kLLZg#i!|`uA`{Q0}*-E1?V+G2%VZ7%Ao?c!fKc$G#$an|}RjLP(v3Kuw
zG$3t&9fTl)oMrCF<lcSXO#76Z8@05um6O_4qSZdePx<D}N7LIr83bhf1a{NF`@5Tb
z4LY|NZ>(byLTz%pyKHKYx0irP1;%Q)Dknn(9El`!)a%YAg*AvSaNh8o4hwHzY|YnK
zU+eEKG^*TjcXV}WU!QIn39x+Ml&PF^z1cJ3#3Cg8#kZ0ASj__1tx2kMcQn%BSgL55
z7FXrA<(KG|h60l2P+PPIySu^L<N5i#P+9x?`ze3u1c-)@UpJ!f_D=rZy&28jANv*M
z7^|K#oA-t8tf+ZEfI3CyV7Lz}U^Y5cG4wqn^c8o{q0c*;?cAX#Eo8}9hA&RAJwxXF
zmg%Q2b2XZqt}CnOtb(pz+g+Ze!(6AT6sz-AGT7a$agCk`!N6`&k@K6MtPhMs2|2=y
z6{;xmzVeWe?FY9!(#rn+W^qk#(#cRZ`*1$ImQe-%*3ls98kK62PUXAYEgvcY=u@GA
zEJnN8>^aAsLW~f|T$UvRv{$%qjW-&LVEU~p^`|6(#U>ZaMVAXO;B3D5z3H*Ma+(Ue
zhy@-Y8<yzSdoHW~CJs-NKEOyK5Sr06!pRU=XflF#V=07*v#7nj2}E1pUrEl-jd?u~
zwg{Dw;^wh>0haq|wlK4w-zf2Own>@SmM2(K-#7Mq|E=g!#`6o-1uZVrv3YuWRy20K
z3em0QX2{1xs-;_8hTKb&%2|}_vIrVlV1oE~a=;Ap+iure)HVfofL#|10+~0ld9K>^
z8pE|DV`G-<e4=^N*6{GedqtNk*SqVp_VOV<rvYn)0;TlOi&!cYZoB=0{ix!ILvZaz
zN_!AvlYcy`{_@R1v!WmE>E?)VG6(o-BUQZo+W?cpv2Nvs$Gvkrz5uAt4RUvpf5P}N
z;qCk6)5D`~P&Z{5Pvpr-@1n-NN$m~ei}o=7dw<cO=f8xJbT_?tuTn-?1!4rF`OTL*
z0=BUN+8(VdIw;ud$9}Ph2#4f;QlFZ2W|Y{kV69j@L~n1Q5MqvdaY6SUIy5XCp7&D}
zl30VO{2Ix`OUZT%vcfckD~!TWVj)C7^a;r4ud`d}rIvslxJ_tYIctM5JnbWO+}dY6
z!RTRvx|*g6HkdBci|~a;#3Yli*82p5%xD!AK+AuBsq>gIi@vXO`Au4c@Y{#%;5$tB
z59^4?gurb<K_n!!v47R^k2VIuCaNSeRfd_N`W$)7DA3<)+o7&bbZuR>J}|oZe$R~U
zvg-~d9bZGzYiq*fyA&W*fo$g&=b@VuE}HY3yMG^vjlbi+JN<04+@#*<_-j$ck+=uW
zn`JjGjY)uLebM<u>eB(#6|1g|7h}^UOc>v_0WfDA$DU+2u5KLL9v*m>4#Aku4HpYe
z<77|OtH8KAzEg1fcK<YsR?PGkFIB%)&WYnA<;rq>BW~#j^=@mM5qI}}rt<!=BqikX
zXvIiJJ%%9qn@Ofv$XM^aE~oF?(`@EuqjB2G7xW=FK<L)^BP`%+*|X~07t5&F?AMQe
z*aC~0o9?+A+jZX)K{-BI9_}V>xN5gG3XjQuTQU<4;xJ3P*6*=EiYJkY!H!0TLPR4-
z8>H8SILC7fJj{0+C*+tU9#`}B@~GxyklWgoPdU_TFo|nB2?_}B^29eRG~h@o$uTWs
z7JW~W`i&H)G5*gA^x;3SS9eyO7Cs(6!ic}j|J0QA1MR=j9wEEPZ{-KY^2hbg@hp%B
zZumWiy)uyu(Vy#im-s}1iSbz9lHDkT^0_aQVSTP#3`0?O=NiV+p?xJVE-d`DPK+U@
zcRP)Zjn*+#(Hp;2;YT$mn^SM+S8IhFK@w_?Z+cQ`(na?-P=O<k6o}stNLVVPcl&uy
za`o=2oGNM9>21%6ZYms>P#fZvqeHX~i>i<-!n8iWGWbae5Dg?&;wHOXPl40m08}ds
zRa-y*`la_l2trFXe869EOOMCg-9ob)KfB3XE&?<Tj6}=G`XDzU9Kj<g{}7trBat`q
z+V0NvVzn&&M?T|%MBdIv07C%!Cx465M$%OOS5o!L4v9R9-?r|?%opkv!0|ucOyng}
z2`2PIl4rw)q?{4j|5SU%jzFGw*hn@)A~#-+y80*6WO<>8k6$SE?Wy$nkp2C0f9&}G
zfn?{O@hJY$0qMoyO*3l|xO7Vtgj5^ceC$aGobs<!x}26cbvx&3;7nsoajW8AxHW;E
z5bGeH&*KdP{5ioRa4VA_YN_AQ+Vt`&p?F6c|J=IDzOWbZ@IY2NeJ76IVvhwzgHhd^
z>*V=z{lQNyiJR{QyO7t{p*P48y|0iZ%W$!39k5@8?s^SBXB#OIlbUmN1ARuMoH!*A
zDM$!8y5{;{=<R<MVfB32NMkd?$hA*coZ0d)$sCqRP5e15Kfj{Gp%X)EQQ1jcF0a)&
zGD%E0v=|Moy-YLfZjGxwmbxq^2wb34SQ3eeP-%Pq5~L9n(0Hx=TV47)(Jhz4Zcx|A
zrp~3qSB<fdq!2AkJFyGdAZ+Z^beE73nvK=}W5ydDd+YIVe$?rTI4_{18&3%CaM!Ig
zir%<$BWj8?znZaR{UExUVQH^_=J3`no1YBjeE59U+SB;Xk6GJZU-c_528cdLB*wNw
zsh}rxKG-mGGGnG5l}V=~$^aL~097oB2W$9Ly@AJ+R^2E%v^h3dpqTGZDR=88FI7R$
zyC#^-6lJ7pj&-fBo|MhGGQK>-ae?`7q=%O()+I(q#s=HYD6`aOf)`r+<dG}y{YjpP
zHJcQJ^EPdh#T$<Mz=g09hZW}4t@@FdEmGWOd=uVkm-l@kC271VThCz>O(B&$BPAzy
z20qqP><-Ab1f|#}A>HSM<(5qsJDV<&cr^y~dt6`4LQ;gu+{@*vcW-cXzWGs3+xEnj
zQIy`cJ6J$8X@Qb8U6}Q9=ytN-(n@t5?Z!)D{Pr&J{J}Y;8utg0Clb^h<?ygDFwLk*
zvA#9&EBoL9TKjR0N!ka=x5&1W=NkhwHOMpavb2`@M=OJz1pD{q%OIt4dpZS7Ck)15
zH4kBj!LX%&D=W-$w|IK*Niwk(;;SPUChBu$SRIQ7sox#d#G0sfq)4x8ss%IOgksI3
zJ;o*ZF4g}m9x>fBRL#p+NQVAEg|fqb=OZ$!-rmF41WIK25xDl+b}+>qo_lB(V^uD_
zj1i{t9SbG}%kSKa)}fV)ag~u=aq73<={b4-jWb8o$0RR%s+#FZ{u|?-Z!Q4eO#U|V
z9zx3~f8L<}PTPWzy7XcLc8a}$xb8GQ_zd2cjL*Q{*t#&q+si3**|SPi*Y<J$azB>-
z_hfnj9s%u`6M>F?qt|}Abmjqd>QZOb{S3BvjJ5t@wSwPdp@BkmMwJ#5GBala`Hz=L
z)11hZ&G&dSeN>qLMV-@28!a;VBq;p-tgo}&e{9>w4o82G(8Q3J8szQW6o_p7bEVPM
z6niyNaq{xtD1U^;v5QktHXAbc#PB3!;bMXJ@<>n9N2~oyp1(NepoYh1NVFKdELgc*
z>$XtgEurp0GehLh{3N#Jf{!pj;Dd1nBeRoIeYjR)dpLoiMmM?htB5~1Ap#5Uxf|Yt
z+s#FjB{Gj?p!hON+wdrE8K5AQ)v{)`?iM9KeBaP~>cT^`n@-9~feE*GfzUK>38m8H
zMw*<DUoj&>`-ya_xH09C@zzE4m*~`he*kP5R=t6{tAG5J&Y6ADRN#xLyS2dEy_<jf
z5mnRWqW)dGBdY886d8Vppo#dn1h>LEQdEf2zcu`hm_pX{+_m^)d%xw~2wx43mpjSx
zm%B;<2N-=MIQzMG%UC?)6_I;b*pJLT1ug#^Hpw-{w}BgIt(#2C(^F=XMJ8;umFBh;
z`g*Oy(=*P9{;gvlHX0J^?!j({?0x6S**Nx$M#~LZ@(0z;#~!1~vXYXLIr$e9w632l
zE>1o3tG?L(SlZe8IWE=Gi7Gfym$B#_A9rd0irx8|gv{OurmBO;MAWCP?3L!h98H|l
zOV~krjrLu_yOGg8Wo`)(Pc8S>H5=yQSUSZbVB`|j>eaLRh<Yq+JW_r6BDpVyql<4q
zW4YRGMN14eN5dt8HNu7mU(a1UVi5+#L?V@)kaWpy@9^z)M%e{C-Qg*VS$oMNxR}wH
zUf-a8cZf!H4C?eW$<7U%p7vn?O7L&qyg{IMxEtJ|lB79q((1*-0`7#b4<Yw9IE&sy
zl^ET=l5F+al>EvI<+{M|b^c@0645B~d_7#Q$`dsoXQ~DtPfdQ+LJ>^(7O^)Sx5Rnt
zH3GEAm4zI`roE`Xz`3A%G~I{N*Z8L+BO}|cFh&yvy(e0J4;VpwVYPv?z9!8x3OA{O
z;#8a>6JTBlFi81=&-Z3phlZqQRYm~~;uWsF<1t_M{j;AU&K}*KA95n<|4kl3FGe%V
zTx4VSrlDXAl;pE_k|GekC&8CdV$=dlP<zg^#D-+;1_~Nn=(HQk?-=Svmu$T!<UQ)3
z23HN>aN=j|CSd|JJVO_8BcG^zb!ptKmfc%=pLA%wmAg#MM-hU4{5`kakq&{T%r$?$
zBffMS)EYAzl_&sIi5*T>u4)4RS&+R#o;L~}ZLWtq4~j3e?priik{|XuLwRra5qZ5M
zv^Lx)QnmK-?vZG6d5y1dHEisWQd58XT5rBFxmrh-XY8uZj^qPPoZ6FcTRqRk_C$at
zgD-d<9#YQV<clGEUueqFo%Vz7=o>M}wB9~CgsBCnl|#UpIo->rQK`#+wCpnLB)6Q|
znb82Pu9mxCFMs|sWazBK;_pWT{r#(qjigP;@1F9wd*_<k+Sci5LIwr~AYJkRSV9&e
zK#}sHsWdhqS5MgT;P6X{eY2;J(}M8L2l$2|)rH^-vx9)_Ug%9uQ?_+}wPK~%Njl&(
zAL<f~*E;7-k}Onw>R<Wr%`4D*Ha7Z55#Z8x6!}*YyS*y+mSwNCfLaRikuv0n^UsOo
z$NgCxb%&I+KL0r@tRmCdv?B4NMxnpz7%o_}kLYmG;x_*7K9c=dvLoTO;;Sw`?^&?%
zK>rA+WM|9)&R;38il6bna{j6Xf&#r|<VV%1Od6h0HYbYx8e|?KvlHk@0v7JE3LFgo
z!MQ)ff_q&?iG6-#E@8zsNawwos!5<}p<r6ZSZhyBL5soy@qZVPCg<-T81!G1R1BIF
z4Sko!?e9pu0Ji@xmv4A1K+RcSH%(u@8Tys&4xDbPMhBmR2?7Lk#q}eldPf-8uTB43
z&6=uKxI`yYOUr2r^z}BNeHInmBTRed;0BgprJ&_YLE2(N0C`We1pNAZxR)B>a4Jzo
zel?UG$HWG6pMIcqdqR(5@H$o;Pv&ZZ!M-5@)2I?z^g_1$a6D~RMwZ(#q(Wl%_A0ju
z95!KoDln^^!c!uzm)bRy>grXa4q5>eyfT_-(f?xA*ZQ#{fPIj0#3Hxb&$0kicX)UX
z{1GY8zllSt{QLtw*?6%n^>r*2XiBLuD4geE`oG%CaMzJp=P-7MdQoHL^KR^3=@Rg5
zUl;_cf8GF>>oaK0oMaxJty9)S7dBq#Qj=SqOcEU3;!C)GqkaC#1>r>}Cq_dB=hb|O
zmIj8vnLGF4=I7*LrsnCR!dMH7k&pd+k?AIvX<JKMap#^RZP?t?`Xy^o=>$!hT;DFt
zLSC`urE3Xh+v@OK4<6KHtCLSLyZ88Nxg+PMLov<-<od6AhuDURoDL~t^a~qQZyc?2
zA1rwtMKRk=t1q43jGVQ~WPcYpE8i)#_|55@M&fp`w!it-k$2Eo!&vu|`oPgxmYYtW
zC$+5Mg$bHl6Q(-JF#ZrfKa6!%?K1d^xKw-ePj;ENYOjT$W;giKs)TU@{lF)dPtCoO
z`@XGr3RNOTvm+tY8@E|&+*@n&qLoPZ%a*@>&YRBL>I0g}>C>QOHtd^kVQoMSA=9F|
z9!8e2M<pK@TM?J(_1$1H&(ELPFWA`de(>;e8RwD?W^6x7<|%JZA~G6M?P(9FQ}F28
zraR#W%lgx*I>;krqy45^B;}@+-WBaW=-^xk^)dEai`2lE)*3Gd#Xubs^Y0C&YUvX=
z%se3Vjp*;H%zYf26!i&TBU8EZ4Hk8zDR`B>pKY+DDyKzM#05NwqPd8OJZIHDR23lU
z`ayzkK4vrxFO5zVV#&I$J@`BMlKUHfTTFjz2c~-GNxJC=9;DZJg3?QgahxRd-{X~y
z`mAw&h@!!weNa15p=nU1&R9-chc@c`tp)QyC#_+%L$QXV+NkH_%3_xeJo_unTqFDk
z4gOm_0qv8Tj94<i!2C-JTAXupnr#xQhe70v9+L13A<NMWs51x}%tN~Jo5~mZTMIH=
zKyHl%8}#2)BLE1nM>xs;fe8h#$Peh9-Ch81(>DJXFz6*$zgM=yfQ<(m&noZ@Wq1Q^
z3C{EHL;;m9EtNtR==S-Y8L`(H@^jNvE~P&@SfxhnU15(vjH!my5se_=XtAJee+9W`
zB1PX!LATpm>y(tTlzl<97}LqQ@S#^k))GiovnlCtAXg?RRo!Z0M!w39OiJO-x>S6d
z3!p`kGeN&59cH^b^QjUu3P-<=l^jgObUVKrW{d4tBH0=vO<mp(f<#_SVGxNo-pE!@
zMO>*Lss`QP&1eApD2+t>kzN#&Sv!~MEZYAN)v(NqglHPSV-N)icC71N8p(rrjyt%g
zu*@fJ7M|u01VaI;{=nVb)T6Zyww}@RI3b0XF>aig`=F1VDr!bR@!jPIcb*-YXV~I4
zioWiO_7rEs&Ae-tv3Lj_A$F>raC?Hy@O9|Ez4-{=(jYX!NO6GeocHC|hZD)b0oczE
zG5<+)6XyYsJR7U|?JHvg&R34GuH>#>e0s;T?iKYiOKc~;crP1fv9OOkBfCDGQ%x7<
zw|I>pt&=XDo<DY(@7)t^=S-dkjs&!qA7t2<Cx^O^*S(H}67y#GSUJNM70pF6$&!bM
zE0>fflW$ePtvN(f>@1VZvo_sYyVzj7C9n#f;J7om6?qHqX$N{WfgR{dD8XZ0|I&HL
zM=v8gtB>+;xIz}*X7gOIT_w@@>?O&Z^)vhmdKTy=g$L`*^=3j#i`4qhQW)K8cU*0Y
zw}Q0qIm<0^#9Z&_81fO1m;bYR=A@`VpYLe`$bbn^wRmP79x(m6ewJdshWO_U;V%v_
zxd24_BG0J<IWYgZ+23v>v<~gkrQbj3dKAe1$bF7TNbvpegQrwy<AM3daCLQ^VGT+J
z@?8y#dC%FL5D_PSeLZRR!TU?1cW77Da&AERxv51*@aoeK1rr<`oDZAMS(MvutcMFz
z@SMtS<a|k-aLM`g#@$x+o_&e@0)7jN<X_lT+<YtZuPL+OwCf|eL6ekpXgnS{)CZrp
zzuCJ?(bX5@0gfy(5=D9_@WYXJc@Sz6oxA{iHWp!T(R{t-{jI~9&Hl3L@qwV8)K^EM
zva+%zFc_LcvHa7qBi>_U5@9?cP?Hk?XyK6p!_TJX({i>&H@R<ttT(p|7vbf~5{LQ9
zO`SP7>c+^R5vKn+!ZR_J^C5c-Y3Jj7IXl5hMMcd!6;rSww^9M1#-+UhMN-{-1-ABE
zxy=r{>ZjjsX&BfeXJZ&eAu;qS1}b_09fkuqvpf@TXuuaFg}qpuX!AzCpk&f_hH7_r
z*1HcVCbNf21r0(a)oY&CIoOvfQ|)OUX%j5v_Pa^M92x^nF7X2g%?GX1HRZVql|*y%
zCp+mnWW6-KKo=MV3@nUnX~wUg(1W;->x*U3Xv>z2#TsRT1IXkDVJsj(Wdc^mf{ma{
z+8UXTD!Vm*ai~<{SM`|*ttSy4Ik6kux}dSbGbihGNKs&aV0QUq_$BH?D!PZ+8|hXE
zxXSYB`U^~vZjnsO%&x=Zy6Ik?;FaC!avqTFBk5Yk4|?$ey|VG|Y8xCgwIh93`o3x3
z@WNtkW5LO|>YDigtq*h|+$~o{)y$?MH#n#;xz1o?p9j-^Xd#gYd`{>!A!}7tC$aRH
zaPQ(DdcOkXDQ-KIb+TS}>%K`QCo#8*)5zZE0ai~R3w$V4J5I1F@jFl^G@T=D*B{%i
zo$37x(Gc08ialrY;)y4Ia&vD95D4P>iPNH}KIy844}*gGZXI^qZ$NE91INUy>z>ot
zUn;JfCG~h}UjI6=<i&Qn7wUG^$ke+ayVP0R$)2d8quYQzY;(gs7~OVunUdryLDE4O
z5VYEad!z9#Q1Ea2-~3x(Em`+`&IGbK(krcgWTWRYf<QHs$#E5lXRZ;EFPSrd^-!*9
zweq;DJPI2!CjTaB29f?wwUY9|75|k>1di2|f|duXOvh!?(jvnxWN%PZ+|%$45954p
z3m8uwFd9>K+R5EfvYHh}ZwA`th8)vqHh|s^z!5$uZ}EPqHJfymc-&#nRDSl*I}2Q!
zQEbtI+;Uk1g_W3?7++@dP&qGmd<2Ys4kV@eZy^el<z;S#{R9ih<7&qVek&{#)w^xI
z-;16}KFo*%=t0E$mC0<$A-_9)09ncm1BS`w*soi#E+&Fo72ikjJDABfjUZfeEwo?b
z0O{MlY~#B}f1gwm^>DamK)BApc@<J0Q$6=p6N~&)3a7`?Ol2iKOD3^Pp;9IWWrL*)
zoaO1`?vVow5S79GTrI~~4X%?b1by6iY`Ch&#O$&%@?R<EUm*uDbdjHyJ0T~R;C<#O
zRY<+`f}1!!y0Lt{Sv<1?Mc$YEpDt|i@6*-kDv4J{>|DF)&tO_Z|8@#~&jbu1Ix4F0
zsQ7;I>vdFM@1>nfr~6#ga&Og$B5ru9ERn4bo>FC=0&5W}KEvwm%}#OH!N%Z3D9j(e
z)Mbmg++ytcfkuUYQZFDrs{w2@Eq2)l_a<%k0(Q+I*E8tMqF3QJ`c1N}t*u`NYTE1$
z2CpuLcFtb#zT6EXcIgaAbSix8g819ixq1D!!2y9Bk~B)!qFqYPN+&rV8<P;|-62B4
zV=puRJdEeIM1i#!QlwruZqEnz_^|(n=G`wkTHfAKK|YW=NlkT0H64cbwNH=4`Tia3
z2=j=+TrP7%Ljxe5w>jTcPQm1UZ0Vd%4GRzVomnXBytkfzK+uxk+ViW0!hLNGV~Eo+
zF|ptWaKa}qI8;hwCoRkIoE%91{`ZKOUPPX+*Qep~re|@R5*s^W1aKtc>iB1lwN`5%
z;@=$GMUg>oSB+KPZD)4cHWWb6sFVRefjZ=;LlbzS1*tq=ltuw|4O`Ywn;<;s&Zmy&
zw`jxLlJXxEVan6fJ%e8W$4&A<u>1b~2*eEj2Z`Kdkl6jbV?=8EuFpn$>7g@cjGGmo
zl`CXT_J-`-!@XwRF2)~46hc}zA_SOhD5t&v!}ZoMBiR`D?r(UP?-hmh<j_D{?CMoX
z`(uyn{r&wm=LbR+#&!oQh1(M1na?qngHW_|yNoI?b^GX>7Yp0&iwZGN{}>7EC=^zH
zH7uu~v?5#}azvB9-=Y70N?*HJF9*U+v3!10UC{k=9_)XqhZssdE>a#^II>$+Rdcgn
zKU`KN9l%ihoB8;mfA(Q`MgoM@j8u$D<6>{<-#SJg0<Zgo*pGj<o>=}3PGJpfrlW9`
zM&m;8`udeeZ?Y_b0>)<7R_e%cF22od)gQ3k?|U`})@FS@7PKC|P>?uUM>MJF;r`4~
zO|wxio10{DsU18`_arzUv$@^pocCyIzB*kHk3;umN|(o35R=Y5V~ux0&0V$~Q{I7E
zE?AxAKs0;yQmVc7u6a3DqiT6lcmuo(4(|U>4aaU8be{Y!Q!R+R#(3xrjUGK(Pw1>W
zGNVbi+G3+$(U%V_uO1lGHTZUS)zcu<N8aGzFK-Y;+T`%ir;?S5X?V+SFf0Hx5ua~l
zewD#eO-cr5yiK#dz<Ry2tFeIwo0~PQkvD}|+xx2zdQBFR9GL+LU2`>E*gR03Mui=<
zxaL1dKj_|x2=V)fi=dl&k9^QrgClt0=BYTN=oeJA_zHHzw!bT}c&ME@h}F5S$!}`~
zB+h%W?506q*=7EFQHfWN<R^)l=cE{VhgY>Bx|<|U#`!JcpGD=sEI;+`M6bGZ_3TI|
z4F+y40ir?Cf{u+$YMt)R8_C-X5kUL%8&$n=O|ZE{#t%v9O;xgp2qYYhtIWIyOBTG@
z%zf+>&6?`hpC1(;aCx#ZPGytb!^b-u>%e`w$bEB=ZXSz2&8)(7up`woQ6=@_hEqXW
z9i@7*8)!pCvXeJm#LV=o_{?BHUsGX7`-*`-7;P^Xtee`hT}sKMr6SzG=J1Sc(hCZ#
zO4w?5-bsDp+hR9RPrz==_9=0_Gt#54w7b-;TUzOIgz41D62?|_^^AKYM3W#KGv5kK
zj*n@DEhr-YY?qj76ph9_1UqB&CuBcY%iUSbI0O>h|M3<mUtg}pYK(DP8Igw<N-h2J
zv2ZMfG=6T@aJb~H=+fLNkUHFUtgmWEi|%r{bSq)HvplV@wj~pEA0(Cn9R%F)WV<4}
z4JdijTPkm3>%4Cd!28T>r~@N(ecr~)c1(t}Bv((2fZl-mo;XKqM!FlFj4WFdvYWCw
zb^Xbkj|VFXVvST&h>vilSuB!4+{r`^+V1q7la*nh02ma{hSb~;>}gFbf$yKkO4P6J
zBE*`!p|-ZRBqXIiJ3%LL+hM3wn{@)ppU{pD5u3QwrZ&I2elxxhNEs=T?n>K4Hj?ov
zPfBfJVAT*BgZ85DeC|b$T5h5nj^@fNjGD0V?6z`ZsdJy9DtDzWS1f^t+t6ast*YJ9
zNc|2TR;awleEmBx!~(i`;2zUe^6jxyVW7d7F@nNv13K}!Nn0CD=Ey}vmBG@+Pg3fP
z9WB6#sbb~gqt1}RVc?QyWLPl2fHpHTGm5x~ZRlc&)nh4=af`l!@qU|{74d~(zd-I+
z<fMe)n&T>$RV<)UIj&6t!m_uwh;Ekt(s6cQjRn|K*xv2>)bK&B=5}|6FZcQLxK2DF
zVU>))l_Q}TIe?;smd)N=zJP*FHG;u3jiu2Q(}Jv1UhcK0*gH;5PCbt*BJGjX8#7%{
z%FiLUT10+npwgw;>chKJy53+VwLSX`-RSw;luaI?=dAOO-&g3D4$=*#IWNAr*V{qm
zAvnffdIyMzwKAnaZLc90TL$DJDHYEtW%~8_pT564rgJF8M9q>|QG<c5)GEF}l8_XZ
zv%+CeYs#9~d+!ASj-{(Bg!t$pg7WuCWJF9t-ubKWj6pi`MQNH^NI}^OjO&dpy|acZ
zniTIx+uj%gzo^*kNY8R^eQ@T%Ik_@#IKHBM;VjmV%0jvu>6N7P{WaB&R#I>k1EhLF
zuySX>Z-c~rNkvbnM#*IUh=XDJv71<cp_UP*NC|y%B-sCh$7n}52LYRg?n?oArkf${
zUpsaudKm&X1_EE-$;o{ltG}nAp{e=473s*sh=X$r=smU%J^timj&1Z!zE2~!#*i4<
zz<I4A*n;;`{7tvePW$GFoY_gjM^SUQyc4K7?`La^3i(7r+sM1q1cj#0KAm?Vk-T5a
z!T6`?^atlnSG9U91-f;k1QNNZTZ)VmuDnd(aVm-A`%kq$6$R8^!cF0pB062{e{sU=
z2^UyP{3Y5{Plgmt*z30JckpKOrj7J)#5Vo|&^pBH%4Al&=9gw$yp(xgO$mYDDI%xe
z!m(OT``UC!3(Burz}mTbEn+lJLCcXBurovg_96=LG#Ott=%BK?r=$2y{NSYb1Fg8y
zG~xh0L3X4BV)8@WaCj%g&F0*#u*U+bu_p;ie=w<aM5h~zjI4-#^I~t&ZysE4x!SYH
z>9Fj5eP<@}@go|$$&@=F=gWom;KTH|+i=-(JI^6M?#A<XTc(-Y|9otm6^l-mJF=zs
zQLPfbF9)+tZ5)LNFU2>v8MB=|-<=g^A4xFWqsXFIHLyFRxy_o}s~n%D@?Q<V^1Hjj
zP>n~$_?1VDFWg5Dr4@)anM+IkgM&pb&6~9Vx#gwH5Lxz!n}}%-b>9@Eq9W;lTC_=A
zixkMXjVB6Pd!wkw$GwhK4AWMXBP@0Cz9l!{d+iq?z|Jo*@F`PcH=Q%8%wK&vttqB4
z`IAH>Sfq~W6XWTfg$6X2u0FPv;Gy0i)Hf9(K+R;nxrmL3NM}H{KU@;ry**RhohlVp
zgI`+x^|{7bIa!cF<Cs@fRZVt!R%P}xC(Drg8{rzqGoz%i>VZ1-Z4ZHS<bq^7qKifg
z{5&S)8tCV$*QWFeBvJK?X1OEzAR*%UU0qnc2)tjhxRvUd9F)jMvF7*r0X<WH`K1x5
zQuEynhEbVG(AgI#-2uqQZ0MAPNKfEfGD^%B`Ah(r-R;pvs()yx0`+-c(3{>Z0O^82
zAh<86LNYQ~fYNZdL`zuymz*f?y>!n9SpH&_91PHFHABw3KQP{Qn?6oP7Nr%$k++C3
z&i9NqG_5FwzTkF2y9Db{c#4d~7Qi8+tpHjL6lqMTQ*H>LS@{ZRj&9eAAXR&_<0(*p
zL~%xuD{ptrXZwR!v)O!|!CVb<0-Le#yULvwK!dozw{U22;r<}JRgK@t5%O`f<)YLQ
zl8`{!uiIfbJJ9%Arc<Tax<5g1t4TWL`i@qK;+!$w2lPodLnsRR^7}2T!PkJ;Dgb~4
z9>l@FHn`IP01j(k{>Ymfk~a5UldP<t%M!XTtITG9ewv9@gpw)M7L_mj`o;8?!*u4E
zO|&?K_KVCfIjS1Y`WDB|a~C~ZjbA%`_xFE%>zyI%1BokOKTub)-5j*^kle1iSxD=I
z@+=lh8nF>8a8!hx7k#S1siYVZScg}V28v5B0V!KF@D#yp;<Eu6rf#Mi`}%!w#Q`$;
z+N}@j6l79kn*=K7Q?gw5#%vT++Sf`WgW)vj?aBPG7Iy*p<AN*(Li<iDQPWe&6^$6|
z?-v$|#M^RN&<9_7Q5O~#4vTitH8eCT)d>JSC{U!k_4D-wWN!x$nUQwkJvA9e4(KvV
ztsOaUHsISiLTC+dUb`uVA78i~Zf#OYY+bW|b2%&X@Pb-9(<mL~-=tYaJ_sxVzAUB(
z8z2~%U;t#8**Bk7tvJk_OfsnmBNGcPPjt(2&aZu-4y;A>52`@jIU)i=p3WT=RPM3{
z{Jz--%1?%Fl9?JrtAm4s7B<#CiS>zJS8H~yZA<s;ENwjU^YiO8wnsZ5V#J3J`6{5F
z8V5s-qPB`Q6|0TFdf?b?*}YNU@^cO2KsYef7{bdUQTDJDZtm*z<Lem1bV~ab7C~H(
zx3mU2#8ePJ6bo;TcqRWbBjZ7FlM6?}@2Z-dW&)^fkx)_FiPnyq^A&X}<(fyVfCo^H
zPDBV3@ia|90ba%lR#!it7h0DM@JX1<%bh{nC*Rf8)pOPAtKKna_`>Ewy9=R1C7D9I
zqah9x-;PcV(fqEOtBoatc04N3-FdOJvIonr!Cx6G=(_6EBjXNOak^9@J5tdXRDeJS
z@9&%4+%qt6K?`i1qSUxAt}_ab!tTk^pEcB#3AwJdWxw4s9<{jgoG-oXW9nPBoVHXs
z8s9t!AJKt%)p++>=p&Q30DP*z*q?1fOJs=&pgbz7oz#Tm2Xb!U-1qiBUe9ulJ9$+@
zYLh{^)+V-Z)$^{4rMy)2?e0`9F7PcqpY<8p>j5G4BVIc_8L4=WTX|DFy|qC}L-XF0
zr#E=>n2Zt~TlQT@SfXOiC(^M3S>Zy(u`b1W8gyF2{)p18Sk&X!qH1I^R|V0NmjO}K
zmf$w{vl`y`Ee!>z@cp~dGY#ruCpILDyfnTKSU?YjOwg_RYf&;i6%BfFR@Tq+jA6%w
z^o!y#ucEb(09~j|YGzc4z$>z7GN|M0=#6*AyqK%h5eJ3$ywMpD6S+wcmHVGcv+=L_
zc(8qZ{f)(#&tDnMa<pIQIiEGR8psAW`|1y+h<^Y0mWacsz?VDw1-BXTx6nA>sqU_o
z#+*co18BVWoJTv@H5EoanitDXZuBE35k#Sem`(n8kucxy(dd5i{6Cz%bzGEN+x9=h
z(2YnVj3BKv(l7`DA_5{((jgs!bPb?%3J6M<fHX*_pn!;!bT>$M$8U{$Kl|SMzVG+*
zKJV|J{n!4;HFK?5>s;q~9N#18V<qEB(C6Cd9;5TO=*@dtiaLC_E>Ah9db4iVOIWLx
z_#8s;PnM!oeXzHRZbY8tzdc`$xj4bbvB=%H^UU=^!OsxEY08X`@{d$<7w9w)Hl5p%
zUaA+BdOmx3f0$FZZ?W;LZ?Tekn>G8cOkRVV<`a^-BZ~7aADR+VeOeuM5k|&BG(YVr
z-H*#z>-F!R&#EqP7U&;IT8~zfthvy`Rwu*d=-lul#r7ySi!^Fv)W7xi$XUFC?$&WM
zl?yuDWFWv!7wilT^_Ix8dMV!;uviH#yrRX2b=_~0iIQWpgruFPXf3)tItdeV+RbwK
zOp__mo>D)>gdQcjb#Do3Z+>1^tz&a+MetG_JxVTHz}704L!V=Ow<=0f@;fKKJjy@t
zU{V4<$`osDFh574^GGES&76a1BoL}POmo%J$f<A}zuM55pkCE{A=cfPYZ3l7;$1{_
z`_z{-kJX#g5nI=a_AhU2P&6Sw1mMVJl-jD%hfQbc;9pKtdRC>etdb2>R?&WGVM%G7
zFvnwmd4d0(MVCwLcp0=qfxvYUBM>K6_<t++vU{uK+j-w$7kch7@MJ~gIi5~kSo3fw
zg<*I7>*LKj3(9_p8yshnA&v9i_e{#}pW!-|GSj!7_F!c^8&)RK%fr(<IrR}DCJk(*
zF*7!&F!&KP%x$IYH@n;tB<6r#b=MMUn=EPw8cLwiMF84zLOIsswPfGpFYsCe3GVN?
zR}5}e$h?f3S{V*ip}A7f1sl2}j8=ns3ivt31*DKosZF)l4F;P(0`tTZnydcfw{1=6
zFo!_NoQ_KMg=+q>#m7F4t(B?^ZbhBLH4k4@Jz`~jNW3ael!euNaWownaYF>f8(8gy
z{t9Ni-za<KxWAPncsba6Q-c^`s@Qq+++*E3$z#b2{|+m*ie@I`j-3A(>+3^(9$cBw
zx@Y5@W6EuFo~KR{Kc0}<b4nEjp7)+^E_G^<qUub<JEQx~>(3Qw8psEybdgkyDKd#s
zi1A5V79w>11`00a#E^LE?D;Jw#@_s%m+R`&>ja$tq1n^$fK>k%$zES#{o#VZ9$uqo
z4TYZGN^ttxTcdkdMds8C+1+C63$u~6Htxpfu^uFsOI0I@&m$Zi=Q<2McP~XL@0Lj3
zt8?yTyYM({PM9q=0u4yb?N$eLEG#R1DH*qDI3k$>;r@edu|NCaNV!?02v)$wDx^`t
zUuUyTqxi9|g*nHTM5o863mwSnh-U?loz?l;IH!AbThowAeBJTtq-XX@1Ll71B`pr=
zE#fGKa4#(q#1d@#=qRv`|1$34P|E*OJZmM~Cjhc2<II2e$1&YbWYcSnF^*y`8<&*Z
z@h|p=lY^`@VU1&)#`e<T5s~w=5Hq+NJz<K%;NkL}wJwtLo60XX!UD$!Z%6C-f?gL7
zKI_ARoe>G+yWL0LHyZ5n-YjGNkEdUL#%A=0sjZ8!$v4=fv;2R7coO0?Fx44axX|3a
zG*_HfrD<LA@ss%B>EN}JGix%EL)p9RWi9%N?imsF*`Na>-R2Z5(d8;4!ezQ}<0dc5
zS>2UepFac#dJ8`a0za>swm&J8;CW}`1i={ua+&PAo!Mv6_^>Eh7o~Z9b#xVl?i=nF
zxoeTTv~1TPEB8;dojQ=}`<Qj-x(7#sK)`FME1WXy=yWJ)CMC%gCpB6+B7AWpOqV=>
zsg({9KU9@wEVbAj;c7USYOY-u;+|~6+E}<ffOjJn@2F{eZzxlK*jb+gC$P%4&Ek0G
zDnP>j@F~6vb5Jfe>Px$J)%^bJvr_jFM_0pjT-wg^?{zemQ_d9c^A7NnQVtls1?;yV
zRxO;KSBtWmy>vY<@uV(8vD`8Ym(TiemrHDRudQ=SN8(ta9d@5!(+7S{o8n;HxyR9D
z_0#7&Ndc84si^WTHkurJ?$y*Tm8ahI7k%w3eA`Sm-^n2zjiy(+{uV#YO5`BX3`fZ?
z8v1VEMDkx|cTD<FX`GI4HO-Ig%#ZQepjyS<c&elMWwnJ^kW*a`i}Ua7klmHuZH*B3
zWqphKe3!KfxiicLuf7Mml0}zV*J#asQZBs)y>`XzX3j=``W%*#lWVG8v(Cr(lH8WA
z@7>ns<xX5qMAKS1*ZZrya-QNzY|n$tnlR~5iI#->9xt3$ztNB&$DTfP>~epR7vvsE
z)VijdKi5^SrRltF$?7r)$|arP;-w3W)S}|~A9RnZ)tE@gf(#K<Z@fGu@0zlceSf^3
zBo4)MPN}-lk$ul}I)A<irF&$*@~rm!-gc5~s7V)3pk@)w{29Y1OBh2?Dv@1*$yUI7
zg34@rX9vuow0&_>?fk!<K<fwU1lX6I!k?wH-y0jQ7)*XP6Crz9S^K$x+a4})yuwHu
z9{eN}9a2{ObghgVhw|}grl6l=Ahfi!^w@D)B5(@QUtO$GMThngJy#mC{}eJ%cUm+S
z1<5WF96G4Gux!=O10}Cr?y_xxex#V(Rc#Vnif#A8@e)7#`o0&Q<^FWXbozp(z2Ugb
zzUvEq=0`~(7?K-U+RiPy$#v}6B-_Kvn%iH|6{-0k;u2FKR>Kag%iSBze!oQVsL+Gl
zv5VenaNUF16Mdei_m1czVlkYuqBoq4n6EC#LA;eo8Yv|<)=vx_2$Te6C%trok<O#^
zu8!(3SFhk7KjpUD-}=m+`Ox<f+P3y%Tz$2ZQyPz;xhuq2>!$MvPEm6ljhEi9`Dmtd
z=c{YcPwZFuTdF_$Tt*%0ho!WuYLaw=r3x{PV`^Wp$ey3W0zP$l_$JS(IFv`W8D4bA
zZSVF!Y4$HK@tPvV?vGfwgPV^jhK#oAX{{JO8WCQUi2X8=o6Sp;FMw&K?{&RZUFP$_
ztkwws8ID@dGT-!jPu!*ru~(n83;P<i3_j1_2}OEKC>RSs-`!GK!i<8X74-I?<Q8Sj
zN9{h{=Mc~~@9J#n%akLDvGQxSe45{);g<T7X7n0n#{{JRc&P3qKiq5a0Pf^;A~E{b
zXt=l1E^FExrgPc2t8?Oko40nh{%icSv_8d=v2)52Ejxd`i(LP1?B<F5e{Bw1hl_4)
z1Ss9{Q>d3iSR@mDztnOyIVOKKp?y&+SuU`{+Ts;xaN-pI{Cv6kJwO$$Q>g@~<sNaj
zb-QCtN~g+BD@%M1wb^S7wGwv0^s#bsoeX*I#9>K;8FMXI*D&W(c*{53@z?%rZ9i%X
z%qopXBOfdLz^Zb#3{FEUa!jsYo#{TKL*!h;MhnV)`Jh#!^I$|$Zq<?)E{;}K{^HlJ
zDXP6mD23!-H}yk+i{JnRQJIZ}Y>k4vYXH|k(&M~6n0)jUbnM1i{pQ3p>qU)aNt#!~
z#Zh0VQt|i}BX2-R?pOwgJN^Z6YEP2ygC~v66nr_elf!2-jjGsa3NenMpEb*OusnH8
z*_rin6yG1|pR1xtK_0ucaHrjVu`4CNlApkBCNi})OfJRgrrT@+x4ha4QeWf5Ug^WA
zQoFk(nRzh5wjw>}nK&$qEYyZbq6=0{mVQ}e{i4@ZIY__)%c%@7OdRviA{^t-bKcMa
zlPhX*cnKE+3q&R4HU;Zq%hym#Vgt$B&Ew-VHwmO?wECXdc#77IEMcmct-DkS93JqT
zt(HJ6w#NlVz%@SC+K5tR%H&2nmt?^Bs;@ZDb9>Qas(tP64m-CQD%@w2zr1s#(w;Qu
zoart0u{@B?)%cla%ov>8>o9$zM_CD;CiXEOL#HOAo2bNPOD~WKX6pK|@O{NVKtoo9
zgU0mWqUF4PDh1qZu2JV^bNHoKXU2O0%a4)$KjMLXB8U<*mg83tZer*=QBHOZbLF{<
z?5+7evG_4QKMwhZB<y#}J4;zJEvEiT1qmdtFA|MyoVa|BUD=!3hHfQIlwp&$*M|)2
zc|y4)#_=xXyaQfa9WvZ)*!~HqK3(+OKASpWURPcf`1E(Lg9)9V&9f@Y9yl*cgA0-x
z-Mj%t$kD`wMRcP-CH)SSCI{+{U5V+nxrkQ54QIht33&x^Csxsep{&Y$hLkFtn$gLg
z)Fj^}+M*2eXATyGZtI8(M+z*4FQ8C;G9Q6xH=fvq?NL6B;ywHbW^*SOeOC`XnPZAn
z1Fc;f*I9pyW20`*F0C?3{iZ$~Ee?Iz<@Pp%Zz)2QPQ5aEZ@JDWyfw?vBgJ=K)qX@?
z*Ls}xBId)&ba@~(VF!A;%yWZj?kXZQ6K}{b4ERmt#W{rcvPo$}WOW;2G4UN2x@lYK
z2F=QKgmw@^y(AzM@m5RF-q7ww6dTPH){5lifIT*(H{M330B{rDBkD=toWSDO(;Pp!
z9d&#@5fy!TF~@9L<4TxQQCqKMH1wnWeN;Tp-0TwecCSx1M|0+>F6H(~Jmp;3wJpg@
z+~=DQxQ`t|HJEgjRAET=r?eXLuG{H>++Le#o{}zi+v<fwd?8d3QzP*iKi5vZy)v5j
z!|=hA<hH_I#<4$D)d-6^f$SSX`$yst2(N+4gEpVdDaave?B97e0BZVH)K7Z)bAg*A
z+}ejfB1Ix#gN8Z7>RiLM-c%4jrIC*1O^xbDDy^SyUg+gB{*v=7&BM`i&vj1Wzz_m%
z2Z>^*z#LPW4@Bo^O}8k|1m^SjO#&KsnDr=yRK^_(yLD@GBl9(Gx36AZW(@^*ftOvP
z85JGv>veW0tazRE;ln|~A{(o&?<GiF=RH_;^#VRt6sS&Vkm24ap*bk9-Dc-rr#;v0
zvujaQ^b{kdk?<lU5)*(vGYS`ZKsOoH#bG{maESJn{AGat&+%S`k1s;Golb7*BvW^}
z0;emF=NEY!4!g2*tyEvGyaS$A{jaa_C<a8LNfunp^ipZy<3>{2?jo1Rd&+@ugnJ+0
zu@-qJC*~(oLpl_oo2nDBHf+~$S-&=FAjQw7utgdK{k(wQDWl96JlVV|1rY(wE72q<
zKuT@*H;)F%^Uh7`u~d7we?}?p6}Asv{#=zEqIJ2OuR0S&lJn82w@cabfzWpG4WUj#
zCi6p|9~IQZjUH9%jJ)I)u4^}u`<PMLu%ADF4$?s%Ja{lxZY_mSseo4qA}3DbanJ@$
zVHLy(T}~Bu@S8&0r6GqiRnfQ+f-!C!VE}Y7;y*wvly(c{f9FHioD_3@*RdR`n*2r4
z?Y+I%X@x}IFr%pc)dU(v?Y@+ZM+DV8dU|oFr?QF%Pephpx?@S1cKscrfe)P&V({bB
z#muimy)H!t&zKN2*9@9;((v(!tp`X+NT|Tl-|fg_r=Hf*AMa=m)&h{OvYoe)j)z3|
zg4C6R8<F|(#U~H`lBW+_+=|x#T%e`wK`;NuKYFyrt4Ua1EYSwO9SIE+%NPs_liCWg
zo92K&@k3smy%3IsBP_=9!J$DJMT*Yq<W`U-CSoCl*n}Xi9KmAF^)#YP>)lGht{Ewv
ze*55UMrJW*jarh)^SuhIA-GWwosF_L`)sgsOGPci3C+tx&KjRi;F=Z@Bd5r5Z|IYu
z8BebGL?l+kw|Dj$kJHud7hA=YnF01`fNSo-oVw~VW1tJg?MFka6wDce2fRe_LkZD*
zx4r2Ds)k}-&84RX0**y~B$Mg*P~pl@4Qa~SAnD`BkH@Cg#t*(fFO*fgTuIe9yuTua
z#3$&mi-6#uu?vm9C5%0RI)MM)L{oU|55>WVHIn2<+2z8(>Y`$Jm`p%3La!{j8&#_$
zA{?<H;ng|R+tYJ$>{_^V>?FHye?+JK=pL~Wl10M(N<2x(LDOcUShG=a+{4R8bX1)J
z&K|jInUpVu9JEYiL9n$)(l<RfM^t<q97us&Lti0<EeYI)AKmgjq{IgUi8woaSZ7cZ
z5xsi6$j@)Lmjyc;4bAy&wk&f7k+HGY3*S~bE_Yi9q?vwy_US#(<kgI{_75I(9zGnj
zms3*`+8d{~P4k3>>WwSnH-pw65W^c_Mdnh?(TsA$G&K1)DlnkFzP^>&u)(@ZZOSw(
zs5TdblSUr2aXrDG#nz&q)RQ7OpU_PkcUjLJ43m2s@^fo~PSjb=tE*?Uyb@2(&3m5t
z(Ln9v$<Ytt>i+J-#qewVzosr%j0>NONl&{zxoC(^E!EhX8k?#gRi-|kkxthxJdc`!
zyDMY!3VLZGJARR*5fBMEtqy>ef&_o_;Mb?m?jY94(kLk6lV%|(i%KW+uMGgQJg1KX
ztxU`5@UAfM7H)>?S#-!tt5XXwPSY0Ns2S2fJF*f;)5~ObJ&u;ExMNJc)5d!6Vz4uA
z)!2J~{&X)~k#jYtM>lPV!sjaaG$GaN>{5+>v0`_GR5{m)v9|7s*+8KN5f?sfPzV`^
zrAIN7qa)bIE4AaND|RVqe;m)@VM1d`(R0P2pWuiMlBl@X;D-KUcj3W7b*@_8H6S(v
zUt#}DJTchJ(ICG_qWHuRY(Rhkwxx`I7GFD0&vQxk;Najf7e+o~N|k^WD3agu7Hvl;
zzllVukkmNZ)Rno8F9AKH`Cx5@dzD6;-*@y-x9AH8ofMWHL4PcIcX@&+ToJDR(1pa}
zM%6Xa51vH<*s@Glo4jHMpo=q>V)bn+vg^mqg^7m!UE&J8ln(Sv_XMe1xA05oso@-$
zw3TZlx1<pEg$6=(BLUD3G3e?!MAhUO9@;w1x%DwANt`YaF1iZwf07UJ<l^NDfQHm>
zx{lbz4cw1LhomYcy}ByWeqFlNR2+d*Nq-F<j-=3EDV`#Ur9(GN^p<L`+Zu|XPMp|?
z3usg<NvNCU7%$V0_`cUAqsQgo^340vB31dDTZ6oygWE-Y@3OF3zTORRdAALTlDRsH
zC&O4+9%j*qB}P;(9?2~&9lsolU#;P=lX_2$sg&~N0TG+r<`mQ-FK?vEF11!!-oHuq
zfVxujs(*X<9wbm0EtVnG@zlikw_LgF^S&X{Tkh+npMquoWEsCaeEYK;{cLnQo%BL9
zV}hg^A@xx1hqqgey2zrViBJs5+wB?IlFcjgbB-|qDd_N09v(^w>s;!Dj!f(C-sKq3
zd(`Eva$#yd-aMvr=m!<MW4#(B%SY1#DU7??-->gBNxrSkG!RLs`S@gsHY7#L(R)G?
z?^+|aBzlaGF8x<__W2ge<bL#j4INCqc`GZrGO&yF$n5PaNIxUuEpaSQ8ZZ#8@;nAz
zYA0LBFS4{AQusrxRSxXjR`grMy?)4+-YY^-IH*Yov(YaFT=8>tMds?MnVRKOs8dl<
z$;-<l+_goNlEi$0+ozNx+{j5YLEV3>p=13I0pgpZD6pS?<f{(kw+;$PNm=z4%qC%W
zF|~_TL_cYL!(^$+DwQCiHe;@sbgDVApDIlqFJLvGax}T#b5BtDN-Rc>4LBH5p8<=%
zV0VgmJ{<)ZYxqS*Mv54wNoW};6+=!0e+FQVU_sZ(><6bBeBdI?)ZsJ&0rBzi01>)S
z@%2v)50TvAGUKhR^vd|$b)Nf__Tr~(QK9fNZba<@vyLLd-CaZ}MdTYdqO~mrg~<m(
ztPlqLs3QidIXHkP#ACHW_WHpHR1^IE0<dq0J1csSs}x#31wX#D%R2cNjxOaFq85fo
zvCdtv4tA@oj%<?vPbH;uP3Nrvzo=VbcGid7hL2hC!J~j#3N$#e(eKiy)C%;QB;E_y
zeB5AS_ZZ<rq|*<#LVg*s*;(c}8#e#S{fQ6b|HHDqKRuXVp*`CZcuv^4mEx6o%LSHQ
zFQ5sYv3L1f)B{Eg@O*E2I8KuK>C3@Q>txpy$Ux;hFlRP4!ol$G;}vs%P#g}-)Ep_H
zkCeELbVR>yy*6Z1o%B-cH-B~Pb46d7DV|Me?mF8pH@wM?<t+3#veppeqGpxkqS`?!
z>kz-U1w!S+W{AAy0rCd!)zxi+vlY8Mv6)NFDk*^q2;a{3UhvHZoj3C0S1W8Dcj996
z9{EQYB2QLQje&uTPWuX6{w%jhW@l&9FKJLNa_GJrbYWWF5AyDIZ=)Q<A9;Rpc(w(g
zK+>oKiTmsNTu*y`F<%Yg$MvULe<dF1)_URqERL$gk#ui0&+<unS-yNL<jjuA$?I9+
zIQ!~hIQz#YZK>(6A9YlIEo8_y{bZofT1|pF8zROlZKyC=hl+Ib2a#BSYm)yC6^n4T
zx1(a%6SymC{@D8}X`v{4NIqINbjqOAt~xeewo8%#X7Z*hUY`p03`d#@Ga>TQjWzW#
zYFR!-AsmK7hXit`5d`Q9)2)-O;%45EvtTqT^oayO^gscT*U556Z_pJr4uLEOUy%wc
zD~}(cf50^weN@~Yn;0T5s-p48Zk_>((0~XujzeSVPdQwx?2vqX@Q;1W(6HgctVgn0
z%C5{^#F!o1$c2SlP6sd#Fr1LP3X40jE&{`$K{vDrx~E5>fPvCF-J^d|6*UOEb-ja%
z8XR#^(}}x>qH5Z%JX=BG;;7Y)`gvaXcZM<$!3OHc6?EyIJW(A8l`D%DL{Ev#|NY}#
zDAWVfV<39f7JNc!4<=Mj!5zYT6guVj46Ts)?-zZy$5r=*ZGgNaEGP(jgLNSE2lb#;
zE@<YR4>F+-GGXXi&mzf7Mz5y0@mVRkmTr`mF8}Fz`TfbW43T;Z5wFL7NQh55@8s6A
zbxewv?;RcWSX@U}%gfjh;QHf-2t+n7R`)Fi)Tbn=U%$#d%Zknv6bYxpl=r_vZL9f1
zg%#o&KqpRg)}sz|B<ke6ynLlDTT@m`IH>6W3*Nvow!NcE)oVy+Q#f7L;*kgAbm}aU
zQLjtX$hZgOR-k_HUoS3iia#)AdY1tgf3kQQ(v{cI3i(cEV<ExTc|TM~Y}bE4wkg3D
z5T>a4&C~vRw*9@E<v?hJoY)wkNj~NFD^t1>#wEp+Ss(Q;Y_k5mO~P0q4(QMc{+Bxi
zN7XSW{Je2JgY)f?^Ih-wz_weZq$^+_KTV_|sNs|X(a7e!;yk5=)_7nne7MrWM=t_u
z^MAcBvZJ#=@5WHH>G`fg`Qi_Qp*YzqF*HCmGTSjZ)x)BEFSYUFlemej`@9Za_KGjE
zz{8Fn9mo!Vs2I1K%(?+h$C?!LV+@8ujZ!4IRI~f|Sq-Y%(Otx}^ajX8Piz0&(!2h2
zn2do}5QqGLVT(()>gA5n{^p+8NBi8MU#Q>f#z02>CAf$H@d?QWD{6}G<Ql!-sj`e?
zKd4U78}M`GAvP~iJoCP#7O>21kiUA6_cZ_Q;bccwNVXYLr>NWv*w8T6;_?X?v~bm$
zo><EG0K{|d&zzkZX(9`c4#oZ9O6fj72+jy2Bcr3Av&HuJRcQZ8`~w)6vVB8yT*4~B
zgWBH0H`s!LrobLs88?Dcdta``HJURVXaHk(Bi{Vv=B<5UEcqqbX8dD1LW=3;Y{ns3
zB}9s=Y=AaSxki7w?2fqM!ejUGZ>!f1=cV&8)t|y3POivM8&@PXzFO}Rwn1wOO2nLL
z?twWX{(iOv*o;z7^)kpZoATFBP5A-{)+2k1l5uZ?8|-i+8sXV;&%4Zx>|I5@J<C47
zgDxQ90|Tb9$Ef~KV`HOmD{R0J>Q0#!OY@E!qyPlvb}ZxfqP)xOv|3a%vLWJU7xbR{
z(UtL?G!3&wSkP6OdWGn0bjC<JCrUbS#0UsIwO94PT)9tx;eY{MKUD^&F$P=ZZ-NG{
zasXCtK9=Tfd3i-O4dUK!+Ymt)K_QJ^cMS6e=t6Y35BQqc!@zGmOA_LN0SbDulqn7b
znr~Q_BCWW)ny3*C9o$OlAu&7#bF~$1-wkaS4|e=c5&Q|aS#IzhXQ(oy#<@S1hEUPL
z^RY}!XZL^olz;f}X9^BF&?VeNhUVi<OiT=9qYs{-CE2vXMmSJ672M2`7KmOfa3K?D
zF(LGM3^v{ijOo#gE?J^2rh={x9Q<24N(wE2?(-=Vb(O8ItI^*eNF&oyC`r?U`lYsm
z2Z)U~!hyNY0<q*vlQ@XiSBREPbIQdkOy^OWXYB_wq$=P9ez<`F@NX*F3UNuM=#I>I
zT`br$X58TBI~l}409)AgkB0*r=0kOy5`6)9Q_CQFCkr_7{g5jw=C_!bEaZGwr3MFO
zn?lR)VB5!~L*3zN{O5}a5x*@1letDT%qSC^+fNuIy_gQ}h+$%4B7}d0jg5_9Dpsg8
zggS$7?K?L403+h+68w)cZZA|h$wL+_rM9M6t#0Og$j*zQb5kz}0u@jOpc$^>cRY`e
zib_j~{P5^~nv_50Ul=oaN&0@FHyKI^AFf+$Uq9S5YgH9X<z@8L(V;!}^cwUT{#`k}
zJAXv&wD|gR3stS5GzF6ydOmN<K>MWk9@3hvbJ@(q<d%`K@t|=(UYenQz^bZW5ogir
z7hk%~O<`$_Tll>m7$tIuzL<3CKtg)XWUGXAbQxQ`*3QAT?eAB2XSeW>cA-TpW7!<F
z6E2dq6Y4NJrgP=Sf46B%4XJ(Eyui3Ro5r*rrND|0>F$uvN2a}XiSHC<JJ@^N24vvu
zgGZ4xSw2PO$<J-?kMSWmZCvU7f;!)4b_P0Fkkx!oeD^MI>Vm2N0VP}tdCBUFa8OlM
zB8KmS6U#jM1Z_YMkY`vU`;V*W=@7E%^@I_{U%%*i@t78~h!9-Tava{DH{a(u7c1mi
zmF;dT2^7sd%l#S@)YIG8c3kkkgs+fqaCCKs?E6n=OxqZjNPN~=Zh24VAM~`G$J>Pd
zRb_5)z~_x#<7WA=tLJ3S^va0JLoo&iwJdR%NkQhsXp|mwoyz{s)QQud^dD~t%fXMY
z$3A~v)itky-RYda;bbxLn%}x31)5bxgm&w8XJ==NuRqq{-ff`xj~({10sWi0x;hbM
zWT->R<m7}kDb^A$Y=kduCGI>~92M#@SFYi6Z@Z2@?dZ<anT#8KS~1>V7vxetuQO8F
z+X~}O-|@O(bFQ%DV&`HpdGaIVOOyUD|0?II_SixC`BDR-4P<vn!Tk4~RZ=?x<-~8t
z!(DW_%i8BnMA^;~xNn@$A7Z|2Kcm@fiw)x1)>PTlP}yN?*fW&cjZWp>InWXsu8?$h
zt|JZ4cZbi<y*S_bDR{i_-KjcAWQBW`FU+Hc&re)|S;n*nsOMdu`GS27d$D_DZ*0gP
z(x8EusBsjQLSo}%ztj;&uvrp8si>Rvplv39!BN~Wp)Gk1R{ONOkgOBCKL)jOWgAga
z*$fVSLiOdRV(**!>4|hn(XszMiO0l3u8Rp!haa_66D)tS;fsQs|Gxw%iFo`X8Ra`3
zeDyEnkD%Q!k8^ToVA!{bdTi2zJ=IKR+omR>KY2gRv8fci%zQ?7$i44gyGZpS<RV-N
zT~6Jw+jP9A_G;e%K>+)N0KO1BAk%z&gjXZY7=x+cuYZF&jRFTPwnYzu8F!2|bbBgf
zur;L>&sbwdB}I2?Qf05Qu2-?pY-YcJekb3z&i~mzvVyF=T6gDy1CqDoIFec`zdbBz
zHPKy&<9q$5QF96R-oF84AV2=fZYp=E@c37FlZ?p%;CTuB9Z|$rX_U1#@V`daQ`sv%
z*Arrn$u(cFEm)pX_rUDP+{aZ>uvQEc`J^OI1gX$K#@xY1HB&Iqb{l|+=gI$tAunq)
z_J2dj`?M>zx?`@f6yIMA9J3JZ`X{PJnK~|pOM8Dcvc17%p5f4w7i-O^Yt-cxD_0vX
z?EjFoRlcS+WumzBAF>SmFG_US=I&g2DR;@Z1Qgn*AIhd_r2ZPQ2@u8qk3@oc5svau
zqE%<wuy-8OudYJUSHz7lluX|7=r96`VIY=fS@|Fty6T%45>iK}{$>=tF&|3^pQIy4
z14)O^P0JmC>n&v-%XdDF7bpJEw%%p*CDg}y?0MZ!FEg)!>ZxD$c_HpDn5=QE1>^e^
z*K9l<!D)8ZmW|!^KD(wrdTNtOmDIkKUzPDVQpwykS(qW|x5@QWO~Rih*ALGOWH5L@
zH#97$r!L*_hlshCoJ)O`dFt+G9th*9Kwm^6ah=j)xH0cuGtWDjuMdcz>uddtGa+C7
z`1<-fIx8>ky1BYG_Y8C8$RV$bL+n-QHh3W1DgS>E`3a1lDw(anhPgfEsaM9M+>eW@
z`nSw1G>?wse?(@HM~_n7!r%Ts$A%dW=Xt(RToOPnN0VGqOpuX~b&H8b^fnVEF>x9v
zQ-VhkM1WW&qMQZ!S@;BgYyYWY0^g0T1FKMYeEnrIA92N%i%h-~1^WJq-!pa^$1gcx
z+(%n)oqu+>n4o4^pBpvrD}3qq8?x`0l0M|Rp0vYwKD{1HPWI@qF^bqfb2TadFJteg
z_=j{lBm-$Uo)ve#r_2j|e@Mv?CnX{ESNzz<n6>4)NF05{w=VFV2P-Ia6GPcXM}^#W
zvz*blKQ3SZWiU7hfNs-<O)7ig6%-#T{~yCZQ^Oip@@VabZKb&`;mn>BV=2{Rmjnwh
z+@7l4Cot(JI{+GV@O%iEh0*0iuZ>djDcuzX`x!;M$x<8YC9@rEL%bT1z_Qk!jSjN`
zRZ9cfK4=CMK`@~vF@Li9Gj{#lu5EF7(MK)`7ktP5o$__7u1T1!+J1BdMql{AgZqjD
zPmgM7s>D~tw|=W7jViM@T@PY8xkuheJ484&mTtA^E<Ol2`=SHHWYj)nVL`e%yI*<U
z)z~XCKMw8%2ZLGjub(*;jGY37RaM~gnt_@d9ZjKhi=>&mxvQW<Ng_PRl-{_b<mtA%
z11)(`N*Ive;Cl*cTGZ>-2C3QPZ*$(<d}H&P&{&}@Jv-FC^2pJAIo>g_)kKWgbDHMZ
zxgBbKa_RK4NKd7w?Khg=L%b$U`|LyRH*(_WkS1K&FGWvI<`WdLXk#9zd9Hh4<T_%=
zmulK?P6&f~Ft1$(j52P-UyBWfWxF6S#?K<?(7B1ZNf#Yi*NyS%dP&poSf)S76y4t6
zuGsc4rG6f<7J2wNu50*C5<iM3K*3B;@HNO*;8>)Z<Uc_EG9TKxn}bH2R#Rtz+0`Tf
zyqK%7i*3;K&(c7ssl_@E+_gwU6$CHR8vT&BnOL*^`j3}Aq0bj1W_n$?t)_^u$LYf}
zb0Q7T0)khlugK`^MhX0Lira(5Selty|Bm*bCbvam;E1Zv)wMb_%tefG5q^I1h+`Ct
zgjylSM#FiD?L6-UyN|(ov<;wxtT;8uA@8#R`m+4Px-qZO=^pxSvZ5kd14%K&78L@W
zf(u<>EySDjxzXJGyOHf*j;=F#I3w4-BZhSP?L9{ot((js_Xs>}R6i9wm`xCMmN2O`
zU#^>K`$WQ=Z0yTQU*FFSR%1UnL}sbH6q)-ZGXv4oZBme31-E5Je~_WNXdqEYFPQ`3
z4wLU|fqbTeE;MIz(ox^Vb`gTT_0pfo_AMugHu*w_hN3gE8|00z<)P~<XWro|)X9<g
zD<@#F$(mo+^ilIv%@gh;dCa6|jRB*4r<vLDTFaVn;Ix1|;qw?~i}(iQ%sGNH2iypH
z24Wgg{{riAF&bXl=IN%8F%ASh9Z=gw6`Nrr+yID9p5^71QR3{^G8bOio>9E{uVw!T
zFfbVA`<u;K-)z+U8dZ?)6~<J;CxR%r!RL;({a7YmmH;SoThSA+<Yexnc2#n8(7*`;
z<Jk5u<~<OUJ_m~2$L#>M!dxr|e$<xKLi0c8WxHfH<3dRn{=^xBPcpb>R$~+YHZYT|
zbx8?7KG-X*KY{OCIG6LuB!Y;_f6tfe1pUi$@aC;vC^C~}qWTZZPHQF38``g4zcvit
zJqF#yQtiBHb79Y8`@FcHk00ARIvUUKi|p^s6JxS60duO!_yI1!u|J`{R1&_46agoG
zjCMy-!LJ>mT06;z6lJ;3O{SaJ1y7&lbG&hZpZ6Xlb}oiL85A-9@WVSPF2Qfj=bG8S
zA``l~p;j$Qp5-I-(jEd>X1LXKf1sOslw{AJJSnxer2n*EcLHUR0if^ZSYU@j(8jp=
zhY2y4TZ_5_t~c2`s3ad>%;Yeh0X;f4xW20bXX-#O%<1VzwZ|GnQ8i!`FNNG2tU=ok
zhPf5}XnkJ-i=nrR#G<RFAetB=jd4kCQi2<B5y}}?2pENbDJRr+F0jl`eJIm^AR?*B
z$#80rq3yIefEylO#{WiuPBf?I>mo~ShI^-Qk~mGGvT*1zx)*uc-x3ebF`h<%B%3-Y
zss8)d``+c;N9A<pox*5AxC>c2pRRu!_xC=M-t_EH{H7|#q5dTzS+uWFSM-2W{+ojA
zmu2OdU*FDRPoE@Q+rgm2W1r+S1`guY6BN+=cf0A1MZeSPGkgLsdaOFHuo_a2cx@v9
z0hE&{VA<YkB$_WI;0<cIaWHjE<W2oTlzl9MJhWeke$?Yb+6E)(&3NTTyMoU}B!$I%
zY|$56Paoo6Y$@Q|>5A!+gZo;(`kev5*}ku=TwOD!m+!z*QQzwm(cFx9#c!Q^Nqf)x
z?e4MTK;9Fq-;MdW;neJE9a1#b)WA_KEigK}u4F-qrBIBF3b?uI;K%EhwqF@lTN|53
zPPhWH(2D+Sf->0?7)!NWP07}IPcu9mpC4QESWj}G-@SZN{^GEGxYw|wsbYh9rEXoW
zemLE=xcrZ7u>S%1i<5>tV$OI=@U@{;<MZ4FF8QV#(FO+@ag^?V!mO4r*h=@J8?b)b
zdSLT$nvV-+b|z|8y4=zxn>V%p)+FH&12<rlSd*kPbN3GZ=Jl+Fqlco|RK`{qd$`g@
zWjOO#?<UjH<%_-jo`bS;m#n#V4#Fp(>YxE9=?V94f?`HSCguWpXssX8s~CxdjO##a
z0<l*1&~=Ih5hRO8Y9E9?j}HDQihnC=Ik?Z2ChCf-L@$vwNT%nKY5b7hrd57ICI6YS
z*po*mNj0>-l93kDkX@7eOW$DF%IY6&iDrIve{OpR8Oat1*otiJOIVAW=2?^)|MI3X
z(V$Nwa)WCMo_4#0d8@T5bNRD~wGvbc$sl^okp3#*mQss+mJy`^^|}a?5cALI=ysD!
zuVe0|p5#!VwN5t-x3r`>yF44IbjoiP*^;^cfO<_M3#f3`HHf#jw>#c(AoP_~0<acR
z_@lZ$ps`PnVMJ=|Nfq8rUy-mRZ=`<t>{7h1|Nk`}gE#jsDGpjA{f=<=Vd^WM^qVJG
zP+Y44!&ibuk#OhdjH;k-i_x$)kRv2$K>i$m>(wf!Tf-BD;XGX-i(X$-+L(7(u!wK5
z0hgb8attO$h)DW#NxZes%H7ROz;tVS=RoQzMSOb>?_!W=8+;@Jv`@rv(MU3MFcF=d
zvwkMaodXVXn{FfxJ~|~OWlYTUB<0~EM2)B#7AY}^L$HUp_CIhchQ30g_k<}C7F?gW
zx3e7n1J~tg`z-2wSyeHC8hX`x#C_?;m`YSD`;X+w{F(HK(O{uZrhG>42mG2ULp#<%
z>lad|goAg@=L~1>$U=AGPLt8)F1Ay_c{mOrnL*?CUVIEtpWO6wAElyt?PxI^ET_uJ
zaOgL&h>xhVtzyW<Z&YLUK4q(K@ZvVRM0;6(W~)|UUf0ANQ%OM2bIl90@fC_Hf#;QC
z3`g8&XTdMg1dU!JV@Gvj(4qidFJW;shXCvm45-O-PV99d0EIdHT07gY-JgS3;@qw;
zMCAefHrjbsCU66X#k=(*PHXR$zd*;wev%#~t&kJD=@{i|J2>5=KC4wR&%-VzMGx=m
zjZ>cJ@e4^osP54vHDk&Z+P9vL?RGO_)v)>Iys&S-voyYLV|?eSbFt#@bB8WHvlML0
z+&YhVwu&)U<Pj8Pi0ms}i;V!6kmm(F?|9MSyt~-$2b;;HXbXoO#*ulEED^MA34(HY
zP8mM;?m@g<k#yJ{op{$US4@f^%ea*f(Dv}qD6tcgMbVxewa8s*zdDj$V^~B`E&+07
z@sW9Co4{7H<CdrY4XP^Rm88@X?{QVnfqVh9a5|7QT>pMP@4;2uTMXrl{>@R!OHcYg
z2*2LwOV*BGTlWNwzGme-x_jOSfR=cbgJ?#P##dL~G9z23cOt~G&yFzHB}t4$y(OqB
zHTw|-!6Skm*Y)04(7b-O#YTMK_h~0u<_mXo?}$?;At=Dx<&Aj-ZAQ|j4&n^~aAqOf
zqWrlw!K&k^C=&Of-T9i+MW6fE%8wDT<<1RgCi2M0sz`9Sf+nr5f{FqtL2>;dowIY{
z+IQ+^@}x3(Y}<6Yv}#Cw+Slbq-Uufge+HwPPXwzc@9lT>S(Mf64&bMa0vLF2Z?`15
z;q6Sm!ALm1eCYbgi6Eg^8i}ziSRbi0so&<Y716%$^H(rv$gXge3)P26ABwGpi5tnX
z#eoIkCaf<KdFJOIy~h8YG4jp#mW)UMRQ7Apli`_!u5(H=#ib)hv;4^Zj;oc6qXBo@
z8V~&Q3ZEu*Ms`{tnqmW*PY!I8=OSHXaA~z$kltn7C{y%7Vc5Dm#>%vR&`Mi1d}Zg|
zu}6Dk0-v2k<F}k#!~S{};w-N?Lv(RpI|p@DR>gN(s}I9T<Cp?)z@0ugD5gk7Az1tc
z@$4G8!BC6;Sl>Z{sF$!at(dRohH$_-Z?-YG9D*{l5uca2+$M)po4|Z3<2867z5bwK
z4dN(nggSA_7TL_!vJa>jhG?3-g{44$Jfq(rfxFqkfbL$Mc%<O88y0<7YFN$p^}2^n
zf5XWMTF0~9s8b?Cf<<fmisv0ll8|m!A6zAQN{JGIY&g9H6SBhRdp`ICxl!vh+ow0Y
zs2Jde2z7OGBk-ZYO`}?e0J^S0*b)921yzZ=#j?ooLH?BKql-)|gNw9@1s5AYz!`x!
z84#hGTw7T-%5>{i?sBF@3^VX7`u-Jo+212qs9m8VDnv;pE<U?tNyEZHke|rph|{s_
zchCLWZS7NkZ7$B!9*~F;6cs=)m@3`Tv#abK@#e!7El|2h{h$^Y0E%NXHg?NMU{{Eh
z3*8-!Pl@&gW0)5TZ1>zTUh(7m<~w*#h)>KeEqxf$t(Zk~?|}qBOn;Udx7)958|t;1
z|KohF_@9jX*2iwHSG|(MCv;}Es1Q`e`^;1W4M(2?wZn7m_i1%+&>55h31Zer4;j1x
zZJ-9KljMj7sR%@>Qlg{m&laeN<PSH&?eqaA#d!&Bcq0GfHJR7Zr06;|Z-Au5z3rPW
z&sh!Hg3?0_GaDQ!9O$Q;Xs4b~krVIx0}lX{J7^^UHCkznaYGv_(#j+xa;^3tdS!Ep
zabv6TnWzFFvCQkzF=0ndK~Wr22@(}1ND*h{EU6`MZe*`G9N`+kQ7wO_U_%HAm8zPW
zT6h0D8*TnpK<()>n|bJZ5|_SCf)ISv6v}mfiv->Zo9j+~vfSDCv`9zG<`~RT_RxjC
z6GIC%kn2D!npH%BT3>HaxO03qA)ygN8$&Ac60bEb?2R#;o`coA01TO{S9I9&!j4N}
zl@y>~s=|Ul;DG^5GAwm^zo9tU;Tp^`Cb+z}4K9yD2TqIh?`A$s93K=$55j#&0FBKc
zBvJ{%mo5d%dU?x^=2}`0#ms^UP<9Eo)m?#4qNa`l>ivwGLbLPPFRjGz+`sg&l#a-S
zA4<daGY+wP)ro^dJ=8rufJeT4Gna_O9v~tQ8(_QGu!S?QN=f1XI7!BYy=uxiQ5~j4
zD;)NeL&Nf!%y2v@wod5}il~HyAoT2&eLX0zGO{r?jT&2?2%3$3NC?cW8D&WDHNI2$
z`TWPH_dojYPg@CO&@f7)4*-l#n=J;E-i{s&ewO*qH}Jq8daTvTah-t|ovQrKcP!cU
z(aeo+&rsPf@5A>A;k6_{XUm2n@JM~!eebveJ}p)bu9AI`y%U?1mpQ}(hFU&aDD*PE
zVOL`2z#=Twd*SA7{X;GryAu{@`?jI9<J$+Zam1){sELU03#&!VGHy_Ca9@OcF~m-7
z_oK-g&Wz41n3*{HA|o7sFGnp@ct@wW1!b{bsm?fB8@|ued|fc)_b{(0yZ=@qmKg3i
z%=vO_hri|Sg&)e*5)~zTyl~+%z;rGAx_hjXa0A+Hasgt%#`vmL7Yu7?@EQ<C%c{0?
z#8{C=FB;}RHejL#g8bkWWjOSdt1=Znf1(1@rye~VLkZqG3AJWtUQ6_(FW#w|^N*BJ
zjlXX_k%gmOVOwbHBeL=E!0np5yaD2?azkP<7Mg;v&t*O6DQ#~ft0Jq$wl$=YI{8>7
zrKowKDJo=){mA4S0tYhYyy}PAxT5VMh9Y#<iZ9M8s8xy|lZu^pWR9ti^2~>F5udJP
zx*E<PY$7x1lqLNoVEU?{N9ZSx9uf(rchSTMx^ln)3cv<4zzjdo`Hf14g4cuEN|L?c
z&Gx7fU?&=C*<iav>|JuYErp^e-kEIg1(SZdGjkEyrpuRZU=eeZXu2mkT*B+j!ee)6
zZ`7>fePJMSpE+$V9TPg1<qK{XC?SZtbEUgt0EVlF$2gEH%|aEx`ULZY>^?h5MPvgy
z>ifG5a5pjmaA<Q5Da&kAzPjQ1*_JVoToOv7nZ)^`bYewxaB*xC{Kl6}TvpE3(?y52
zxvJSAl`M9=)(CGx-Q&$k?eLKIU{2+x27_jU-HBL*39&&2rlKDm(d8j&0;i+fET{$D
z^`NGM?TtnFL1@<cy7j?k;`N>ovOmdtO1ci;)&Wm%V6RWDCdlr;Y}wB&Ry6kaUnS<}
zyDM(ilFLwE`m9fef_-<eZv9>WNeAV5lx(;HKYFhZ;V%Wc{*v@lN9+@=-BV!`LddUE
zE@{f=UGEF;p6>B&?6D>AxMgU^8^U-~5ik0DB0y|S+VgPg)yB1p+p-2%qaTQfgxvRY
zov)jZePYvzOP&7l<599Gf|+Q5<yx#gg8Mpa>faNc*{>R($0sD@UJ|OQdT%-Q{4|pN
zLnD&?;R6E`6Gl~)ZFpK598mc({ut`v!xV^VJbtT4Jcasu+JC%)ng(AwPMA}|zhkZ5
zBGznLq_A1jE?ZB#os%m{O`0<pl5*)t$%Nl4d5UY9``yvR9HB%R^)6WQXhT|670^E5
zaLAzlM-GCo`G&|OEQ;)o%*zHU%vj~R)>aab(PRT=*LtDkcT{qmZEW(b`y_`)mG}e$
zpiKYLNBmlDwKhalbtey`;{!AC&Nt9a@j9&>$nG$E4{EAi*UJ_RS|NLcI48BBZ9%v2
zO>x-7<iyP8W>E^3Q#GQf1}*m-G(fl`=tKiB57zdSa^|WXI02O8u{bL>wDauUpVF=C
zKAW`NZC}%E67O2t>yZCSU9--42N(p`zWk*f+ut*+<VGxs9D~|OrO@3e%+)xyHp#%-
zgbeylOht!F=s0ITwH0XW_9O3siL}J<0^g$$MM~G1;-;vmgj~9#qa%%O=|>fENU$i;
zpe`=sWqzGUQCqFtX`Z4V_-z4RyUU95hi&11q*|XYi=gx&nypCSHClR^b1ThZvGXq;
zU~NHC=*8JlruvWfxrI8K2~7B`_+H-7okKbL<+|&JkGY>v!kxtNqlgi0pP5-$7#SGu
zc}3(t%i*l|svm?P+%Z=gz>L3H2xTe$OF9E+QT43gXKBLNWGzOo@mt^4kvQs(Il&4B
z`j0}NkC6`ylHk6#-cIEO|9+}j*_~DxDyI#oH`e!`qI-IJ4h(0bLoF%cq?qi>h;DyR
zaTY>;OhB>zD_sR_VC-B9z*;GYp}djN*Om1wZ;BPWy1UHqdGjCfD_vhafJi6+dQw8q
z0DnkANYA5MNI{Vq2#N#w*q(BZUGq{nBj4V8J!n~2c4{#_C~pR*P?G0-Y;J6Bpp^f&
zKL%K3IQ>4qgqSA+{!nXOcyi-czx67HXz}j+42~HHUJr@6bOc|aiUvt<3-V`<a{m2}
z5GX;Mj~s;=g{#`WijOl6%mf2U#SW)Lr-c1IegzOs;wUPgQGx^7whO3x4<0iUcy6_p
zex*ZC@CAG9-J(O>UNl=RC2nSB*-0)F^%YztGL_F-Y`nLcwGz3xH6oD=oujD4mJn2$
zdWM6{ThepyO=|({gsTe%wb`gQ4B3PvC2LgnoDzB)JIp4ONX30NDdhgED4Sm-Gq+n|
z4hNV9TPSM{`x**s5=Tk`g6&3{FIRh!sY90q5xO}(N=>1ud#t|5AyGsszY4k<4)CiB
zY1pd+i6Q|!+}q-?d<S*7EdKEXaCUawOt}cwQem_%d3yFV3aTsiz3CNCFRx-kl3W8C
z=ClClZ-s0wEh=tFt?dEv_Px~Q-nAMDe6*0DshQzl<Et{&<e?(`@Vj@_MJ*FKnn&*k
zppO9c+6mhJ6cir=wb>`3HhZ;zZ^^&@j!TTHL0?(QGeqv@_E6dX>SqGo{82)7lV8df
z=lR+9Fvi88&c8(+kKIruLT*^UKT3U@j|n}APJzwVIa@q?$+&CzCEpQ)^*%FUSa8J9
z;XPgQez0lUoaY0XA{|s4{8{j}=zc4AGH+p<yL~Ujz=X~%1e0Pf)piTs2c2VON>(Oe
zVJaM;v*v>oE9kB2KtTxr^!q^e1j-Y)-xLk}_mVD~1Q<}Go~@>I_i*Yl@=0s}3}65q
zBq00A)<8JlotvBcWCoQ0{E^*VztH?p8zv|fsxJ+&Td41w67|M;W&M!<4XsEP01$Rp
z4(^y)AF@G~wnU?b22lyeHQ4p*n7Pxy^99l$P(`*{K!*(S8rpSrgTas5;kXa|AQkb=
zaU7_E=n##LjxKjySIg#nEe7kOA#!pX;$}jGQe3e|{$49LTQLuZEnkz;(iRM^xbN2;
z3P<vafznL&IWtP5yGaTffnb6F2a0}$HGPk^r})3m>H0rj9Wbue<=uPDVu);XN5+nf
zXl?|7;u9t5)j&~cL^KfH3TO(g`Zw5B5RgBs2@#e>qg;8+yB{gk6uxzLtH-y#nbRs|
z03k%df2It0d~o|eUYBHL1GG&n1la_Cf<y?m6y=)n+s3^Y{}R>+z6(>b(0ogA+9a1O
z;-vB<v7n^>^7d|NJg}57J#q8O&dD)WmU(qb#SDzxI6jg-)gIOPlBtSxtoI2pV^ib8
zz2-&#$UM;L9Ol-lzFvd3Hhe+!?5^#X)5&(EyUTpE99=78k61w6+EexAVpg0#4z)<~
zJKAYV4~q2OctKf|b){4;5U$scTeYK;U8!06^BHdCW>gj*$Bvz_W&%NGT=-F>-=Bw#
z2PSd!?2V@qu@CcQ?%%%&kRUhWl}rPvu{-ADI98j&@rl5=&0^cb!=p9NuMi8-y$T)v
zMSxI?E%?@6a=GWbGIa3dmBy-{M&l(t;g>J7+ijziV;Guza{^ssF=>qv1RLk-p8Y%O
zFI^pOeYm3;e5dZ{CNI1CNpjq^*BxF>#_hG)$EzC>)ki~uB@&zu&+hS+F<i`4t+<R+
z7ATj11SJMhXmlcl;arCVu)C2GSPgvHI6C^An0lvqEQm5|k1>H&8CXyKT$hLc)L)*S
z<UY+Ox5UEAcS&C)EM$#z8GylHB>g2fJut5`o(OT|mKOO;e9LSMOT#CUf2iD^>T{Xr
zZ5s@#5;cN?k6??*^^C|9M~h931a#o+{mpe$H=c}dn)datGwu0{>{JlMoT?)ZK{2$i
z`|E|ZmJGJ`eKgu>Vs?oiN}ev~;K>~{u<pORw>W`$(RhYy=ah#*VWaVay<+ciQ~{$V
zZ<^@Tw7uY;@fCB5Be7E_CGeButL?#+;uq&d-YZw~aufaB#B}oQC#~b9*1L7bo*Vc1
zy$@kwG#A&kF>0;!HR?zB0K4!IT*(-oL~|qD(K|BHM;)N~SWsZz2xa(%11Gpj3Wkkx
zAQgfH+$7)DOH%)=8fYsyNnId8&-y2YfZ6KoM<+(33&Gb|Cp?AQ*sO`Daonx2aeJll
zyD*&YyT)TH6}g9-XkC;$x5i$Xw?6pqc}e5e;><xszk|2l1$b>i8sri$9Q>Z1U(Bw~
zh2VCBIwk-!_aP2@q|Ddu_s9lEiI#Rlp4_8O`TE|@tubNfZMl-?=ZfoM(OK3>EuV8`
z#=RN$YU#bEZ-lT&*QLE6v)I&-1zlv<-N7Diuf@C8=ficb1zKdJU~|Y6S~c?&enFrp
zAm2hAxxx2*-g9fX@9b^;)kQLUm2A&VBE)Ht)}dE0PkF21;c-|;%1)6j;H8IahQ3Y>
zm$@=aZ4HH;&YhT@Id-WKF2Wd@@iVt5gDZBHl;jG3@dRlK(99p~nEpe^g|yk{9!+qi
zWuWqt#x|fp2h2iFAkF~x`yasmAo&p-0qt345DzSsoySvVM1{?-da6FpokR3iC#mE1
z+-siZhD04Y**GneS7$~y7229?R>5EPdL#%fW4+plZ0k$N94T{<U~U|wC#uXNqNj+t
z($nrSxLByjJy#L634Czt(pMm~%IKhEdAntVOei*K_#pV7?1e=q{KtgSuc1Q=u;ZiL
zoL;9c@@i))-(m``Kbm?jKe@BKMAQ#`u6wm#yy_5}zO9Jxdo#Jmz5iKClO#DpF;pvt
z1n*5`!56H_v-!kKiIrZL+e<}P1T;gBWVWJQ&-Z8H`<wQXm`anca`!pLh7)wuZ(Z(<
zUbl{{3md@y$zrLd^ne%~*W`F*6^rPmesa191Y}}za&~>LAOHd``VxvrDF~UU)$O}J
z;&>jh53dhn(aW4^+m89_Bz>HpB8oToR)xo7H~DhkPW^!>Vk!2|38g`O?}v|-J+x(!
zTkRhdvyFs9lFL~PK6md`!fMoV*FzD7AOS8YOkwzLP=}BqvR_k+#|Sc|J$*g($Bj_4
z@qzkU^zi?tzR8j@e$<}5!WcACkFOuT>Mxh#ibXT`!?WiZv#z%9<DT2E#%cDrjZKta
z+G^1c`dg;imJ_<37Yb`u91HL}Nxa^OIZZn&p(nOD-MsajHc(zRD*)P=xsM+u3<Eu(
zsBH8DGW3BIXg>$y9%PZyAHr@s#5Z2LE{BujX>XPD7Mp#BX?fu(mdn%;L_`3O|HZKD
zwV_vq2|UH@0Y?_TbNVW!)_F4&uVjTChuzc5SbS)LaKD4rAUgv^c&liBPv!~7#f}O9
ze@~~7>a0nIk+h1=_piP<HrLAf-2Rk_ac<AnToa5ar~7Vf-5JB05dKriqxKi7|A$$|
zO&7H-F?9N-L9*_Swyz~GT#xE5xa|V0`(Ff^2)P$SK%)e#Z?R4XzuIOb1Weq))qVbp
z@2P{|Iu+t8$Dh9T7o~ss+H17tOC8%Hw|&Z4v}2n?u9-f1?M`I8Uf|LzzHhOAu|Qfs
za(C+{-`_H_hOJ9+9$(^<U~2wT%N82B_?47EbUIR9D9|P*-XJi&9p5cu6fd;YYis(L
z!(FmvDZD)@XG=B)=~;W@z2Dz60<P-Vlp4+Z@+Ikx-h_va?0usx2<5SOGEy*4{EgTF
z4N{lE$(DOvOBIl}m<`Hz_VxlDF~+xn1`&B`PJz}jOiC7sC$ztos@*`cI@q=L0W(@0
z?@dcntK!-dSBc_a<{!=9ux_Ro8?_|;e{_9yTvc7y?Ky|;?rsHXkcLA@NF#!DH_|0}
zq>%=Zkdlxt=}zem>29RExtr&Czk9!1|B64@d+oL6nsdxC$Jq5KGthzG6mXY9%({Gv
zvaCMzhf(S|$eXPzCQG|iq94!4M4kzCmJ73y<&Wbk`lIQKP5y7LYt;k!obO>F_V3}}
zpEV2WcAO9YH><Pj9R8cyjsRsVn+J>1-?e|EI=w5AlHAE@6O<7a>!+D8)V&a1{Pslv
zKhIzR=qQ}&C-G_vfGjXWlw}<I-)HX=E&l(r_kX^g&FYGK?o-~M5AjD(UZMtL?nL{J
zT(*z8|6E_5kEWY7)k@>F!rJmPYiH@4>x{NnycrOT0u(=Mr{KJd2UpyL4W!)Au@V2{
z^v1EQMa5gnr`wy$ythb_sh(ACH(|ih?k}7(z#rBuM%7lcx;UCUVp`1Z03o5`X{xT*
zQ0{0Kkx9@@sZaU@#Mp3vQvvY#;lIWK=mKF{ZFDry8b(JeXMau%5mh<mL)*vTOAZgw
z>_@_ppVshR$+uvzdvjnI3?@el$P=#r<rAc_*&NuppCNuVi)WGk-o7;Tcgs^m=Rn3i
z=1~;8(S9O#!bl#%wcE+7h?Xl<jn-2TrH?BLW{SE0jw!iz1G0seScsdiK)}{6>3=C5
zKtBJM&_M_;Ub5y57MvPw1<U@|8dZKRqecggphZzdx@}_?6pPnN9ld3Kzs)Y7`v0Bg
zW%KjR3k6Zk*oaQwP#6#<RHzx}{urrWdw69V|NgS4_moNvJwnMOx3C8>qkIeBcF-=8
zuiZ*=J*!~cFpKc06MzVx%U7cw=lc>NdOtD**A;i&MYsAVd*PxlbdD)?Fc<SLP-ZAl
z1P&_}0p;6{I!e19;`)Sg1D)&d7w29N28>DPn%{08{)=$LUn`aV|DjKLw1hgnyfT74
z)7&PtU0Q~0BoA+h2C>?QY-S1*v$4Hmi~7C?=dl+SoB`^o8wyEgVALo{dO8PW09+m>
zh)IhKM?sM-2WDc230C&eH8nM>0fmEPb|@AizcB2rKQAkZKzJ6zLn!-e-gbGcsbOb3
zvb_G1vF+-^<0{+E^Unq=MA>D@hm@gPaEz-1oL8Bi=_VbEs?9(Te7S3uJpQp?t<`|M
z$D~t`CdOXTjSHLLk`obYCv23kdNgN@!o%}g`?{G6gS+QK=5FO}Q#wt*WIYz#E20rv
zwIZS=xm1C~A5(qD|8G8;p7`FTRPSkZVZ$TGfH(JZWY0o`a7jTsK)F)$SE$D);sBFh
z0D$@M5a)pU3K^P9bB3DY-rNk!MMztnK7U;mXT3nHpy%5=Q`CIz+Gev(DcS75b?aCZ
zU4!+Mp~iWe)B0=j=wA-h?4k&fJA6n>fYzBWbmq~LT*L=`o=-t)`nv4cW5AAB*hfSu
z$($HvERtn==({dPjRauJHn63~Gwl6;^4CU9;|us*4O>tBR{gCHzGwY>*#QE+%DPGb
z=s=YbvkB10ZTJEkUKX21g`h+76)dBKzQ2EY%=Yt{akIr#O;3J3jY-byL)L@W9!Vzt
zFxkHlNL8D`xpIs$uh#k>H-`bgH}t<8Pb5?-OLzE;7#T&=!!R;LLy*!_V5}8QD6*zy
z_aWd!HfxNU?~W*riQ`p*v1x-QGnX02I*avmw<2>A^^+Wez+;)HO2Sw)V?I_;3ki_{
zEa<VQ8BYX{_or%1nChBX4sOBWgvc92h`m>I%$RJvf9DU+(JaKSKk4NaZ0|$(9#9WQ
zkN6G$Y}z~Xce1}APMEs=v_D5-)r68YmFvrwwQCcd`RJ-4MJx0lv%F4{^8GsqJ4R|`
zDE%ci^>}j5I=8CPhnR8`h8Ce$yA?v#t^DxHyd|bWrl!CrN6frL9F+jl+GkST8T?Ec
zH83zjz>BQng`loNnX3udnE}PUde;7S)d_}lwZz5XT1?NvuSb5Payh??TwkOg%9Y>K
z?t6>>W1BZ)PEfg@7BS|0>0b*8vjr;RO=CsWo`s6DslL?IBTupk=_I3l;T{^o+KmlU
zauGLlnJ9AAfmJ|!d9^cSKv95#0D+IM*2&R*K@?jh5l@}c1Rx<T5Qr{mmjNJp4j)YI
z92-i-2t9w=Uqmv*pK3rl_LF6RSCf@M4qU3w^oOYByr}VLCd9@D<NWR?5XYB#gFx`o
z{|oB4R0S=qD0V%JS`k?hB@imXPJ<GIjVo)m>}_GuVJAyhWxDm`b+nN7%N4!x%*T1y
zz4T><%A}2DYiZ<0XDFYSYMZg#n?ZK_A6}J+JExQCwhB#w{v#*_{1#bJ!ZkS;*K6NF
zO`}CGuhH}%Z0(jag;6bWpC8^48JqNf@=5EROG2l1GL`SBA;IG`QAX7L-WqE4?l!wV
zdbVfSnS8C<VQY{C@Oa;<Y4*A*+<<CXg<w%ho12>_qkLt+<iG{zot>k`XoJ}{b)Ejg
z6N|u+dm;3)#zS0&zM{RU9LCrm5YXx!tp6%o1Gd)WHVVy0Xq!Oti~7i<ygs^i#bMoU
zhcEQT_-9rt*VGXccg(3Fk-DOM+sV2kLe;vg0vF<N`WvT?6STXvc%yC-qK`mFd66<~
z25<MxWazP!$_44|FGg1iTDP<@&CueK2*39YZ4>iIhqrUNMIbSIET7~TaE=3_9Rusr
zpg80tk*ZMI0LFI5(CKH=sN<EM`|6U{(R)^1o2#IZ#N<DvA9Wt$zbZ-FgYtmhm=v{;
z4g)tjpmt!52-x26y=Q5&{QZ)$cLSs+B5FBVN>`Y_DKg_%6}r|<vhC3Q0&n!xoCoE#
z7bzd8I)H`HX*?x>n+Y%G=Ql--bG~it*F!pAZbn?iLkRh@d_I>jP?Fz#f6QFwxIO$v
zbh*fhr^@)()>vLU1O=<A6R?W^+-ScShC-%<AB*jPA3ekGqwho;&a!ThP@4Zi-~-yO
zd_U2T1Xd2h)Gsth1wz}#U2VIv7*C+H7BcflDNqkhTwlHvjHg)x>Ma2oPHW^4oCM}e
zsYngWaHt1xH;O#L=$E?<j-Rg--Lu{4ZWaZib~@#P5MS2>2860rD1fvtY@-{@QM@T`
z{zy%1m0*in@(25gA{{LdX~-JQze?rzl+7@%KxlCPXq%qZYRAgSuZE@{O)&M^WKHDK
z@dV1Ju;6*#(bD_F(x}i|6*G?S;fvP1^4}sqqd!e0i7Gr?*hZg)w`p+A9=AJrso^g!
zHmkCkR9*5icv6K6{XSV5AU-C{yi9S5A-%xR_jQ5bHY#?c%K3~gv~D~o_TvHpj5`J%
z=J<V{1!^Op?VKtO1)`US=LRD<g0T}0J|66%_vs$3dC5aUh+D2LBb;wd)=hjD=0g6V
z6_;x{@~`5^+UDKgB<DekdcnH0U9wmbCnCzb5e#G}mW^c#jaM2ERdPBGl7>e_qm4mj
zK0?V~(R3F@$UPCgvOKF{62m6r4lg_tTZ&KWF(^{b>;q-L-YjGBPTMiuIcYeNekslS
z9#%Q`fS=)+uNBk@Upy4NU$;qob*`QxCQ88F?IvR~F^xQ$>`lIQxMMk(!U-3E)C{X#
z_I}5!m3UyPq!b2AC8yt*V987k*XJ2E`Tz<?RNGxaeNq}R?x-wr^gV%Z65_)xAo6LQ
zi>W@+j3BFp7ZArX+PL<uE6CFFt6WgMD9v;5n}lz=l6Mj9kx2A`FIw`$z^h4<gkIkz
z@$O*`hnje9s+@EWdt`oLFRaDv>FoTuR{5FlM^noc?Y43G<JAWr1`w;kVbtpcg=7i3
z22-;q1Ali#o+E5q10~t+<g70NTd5smFT7-Er=wU$M|lX$cgE#t4H3F8{ismjej4|$
zMCGR(C4}I270<!o%!^t~)*M66eDA@ldR0&9KYk?L_u)r)Zc|K|>eUSRGLuCVJ$@3n
zMC1T$1tsCRxVYH%08Qgf1YpcIPa9n5KphWO*RE3+zr)Ee*(wj7xt)vmFdC<|6>~t{
ztvQob77XouQfxoNYKdSS3RyctSRc&`v!~J9R%|OlJfsvs)w@)N9(~3;8JMHsk4gGs
zd-I?aDLELN!Fe;=b9{b|@8bKBZOiYJIOPMw4|_+p$6BM&9yk@=W5R{y61qKt^W{Z1
zQdMWM-&G<`^`k_Q88`6p8J68kF;We`s8DRgQdt|8P%j9uorzK23^ByJp*k{|d8R<C
z4m{~3JRU8yY9<e^Pp6X0F;b#X9#T5<GFyZ8iC_7;;|(X3)TPVy(DY9|;RHN|E1M5?
zi(D-g*;mKA&a&NY`l`PsK>^9shXIT#=)%c*eKUl92kRJ1#TBBC3k^n+9RbO~@AQGP
z)H+K@=URE-1|CyG?}T3)Z`p{9wrImGJao-pdj<#{rv8Eqn70O&x-zE-;bc0W6ox6W
z`$U!LTLtZ>p&id2_c+LWKu%e@c-1{JABL(Yvzn*ax_U+=NH;j}mdJyi#%X*>p52a$
z_UQ_Z(EhdbkbV1F#y2{rC(gS^yf`Q)86UhWaYS^+N#}Tb^YQszWrgV?CV8k(m^z8E
ziM-7CqvD6_c4}eKJwDU-hcFo?-w>^`Mg6uQDq0(^_S&CFh3*Gol<%uT>CIRkXN$1F
zy@?pfNN%*W?{&`W$f8}Ci-|Fklj$KXl&3UNS5(nBs~2cW@RqhA**Ih6m=YcpKPKa8
zKu662xE}@GT3+B7#vJ@IIP9T;GgB510~rCsJpF@R02OBPHO>bC?DX{XdR|CBi6Rv<
zRQ<NDv(^{M#mk6C^wy2!HC6cULWD|=!y${&5=ozHZMLMb=$;9C!-YYE@;wJ(x3W6~
z0|jq5Wsz}`A&Dkd57EQ`yn$t@o{)^+WM$FqOk%chb}eXnhbYt9c!u-Q*QSGiJ=SMK
z#!jZp_<xA?>k8vc{vuksF=AUnFAJ(l3+Fsr)I3qUa<lw!Kq&nB6s>rmj)Yy|iD2N5
zcZj=Z^{b{!_*jHw+MvC)WkjDxFCpCu0<>>3>mA5rO={&PH){I{ClX~Bk`@!nza?+_
zMFW?YpCBjAOTUkYr_eGq#e_&2HA=cFq^GJ|Ntq8?o-TD<=Npv`Ty?Nbp0+cTRBP2e
z3EZZZWFf?W+@G^kOhQ^@c25+IP9woM9w0q$0CaMsV5Fqfa^P$W*yDY-3iuM2I_p<E
z1`U>XVI1v*#c&4^Le(zXACs#=B%Ca<{;vznKRnNdcU~XAk4We0d`ymhV|-FO!D%<?
zH^$e*YkjAc&A~I4%O&T>Y(II#3(4+~tneF`^1e~#Vry~x8ZZfH03*fM&{E%HMC=Zn
zNPMFY3SRO82Uv`BRxPoPy{JOnBlu&@)Ac~dw;ntve);2ty7jYTfx$@9&u@Px3$pFd
zjQ0(A{L($i6GWs(*}_0{B-GWw9Vikm8G>2h1CXWirJw>{Ty2gXFhj^Q@u9e1IH0JB
zG<!NHnRtGkhw+LcctrEC-O5BeeV@Lo_?}$y);DmHlFL;WYx0WgsV{?7gyh6?>Z<NR
zHS~iZ*}HA2U9Uu=Zl<$xHfyo9^SdNxg=9D#t{P>s;JiG$uH306G>HRjpMF&N7vI;h
z-KC`W<YDeXsJCj5<X1wXQT>QLD~p`Z51FG06GuhVz$`8{$Byv?*Gt0FgEJ2-aQ}J8
zLE%17WVrG>7qTZjjUiuTg4ar&D=j|9lQ7?%d3WTWWsW^Ql^TYR*D|lTU3G_tlaY_+
zh{__!!&FTe4ZEYG*8AqN_fE``xE*<<t?q_Ieku^P;=WsYSjWW!X93w+rJ7jWF9A$W
z<4!~$6hJI~BJp&)Zu!Jf{dQ9kK6N@pf$l{ZAYx(KtDY2`nVGpD1)Deq9efjcHvH_}
z)2)gx2=|&yggL&K?zVc1U}3eiIbgDQdto_UQ$kv4<qD{v?RoInO(MJg{WDdoO%xr%
zO^IdEeUh)=ENE@n?)Ayu-lJ~5=Uz@FQFB^}qQWn_nL2aOMFpP(k=B~eGSIB(QGGx&
z;6#9)qr{*r{gzxpRNgluyNXLxW+;VMm6b>RV<Tn8?-M3@YxoVzR)IKlAT0%2JSr&Q
z)ptK2*3+f&LJ5YS4z6VF7B6xDN+3)Q+&{9^>{Y=cn6k~$5Q&2$tDqoNX+ES<DUIP%
z<a*-4zkcGNLD(V|vFz&GqxHSO6X@W$l7zwJ-;ei<U6`R`sEEK-&Brq-Z!tooR8%;0
z`|I<y!O~@*o&tCZJdoxKCZ*^8<4+C}sFYbcyno`l6P2>2viRo)+giEks#Nm#{Ksvd
z;chAOS;&_NnhxF}^<#rILpnas`Ch^U?FL$7E`X1>B4wP_qRy3g=h09TK?xqGr=jMS
z=pE&e-0V+q_JZ#%2fgP}QAW%Uge`HmRRB#(94(A^fQP1TEr7UJ$Ol_ePfOXiH8nL~
zOX@_2sC{pr=sc4cgPMfGco6>?fv*Qj06;ZhgqUsRD|&BVdwT%V`x7WlyPx4E)b+yP
z+te}0Bou`A54BC{-~Zc(e19enFPjF-5q^gPxb18ZqJ7ob#fZ^i$%L80P8NXImeS}n
z1>q!oS|VWHk!NtOkIpNivj?$%ZUmI^UfP;zw*#kf*K7YB%I(cfK>_@t$ijgLTtp!3
z59(inwm_iysx`Ij_jJIs5LF!12QZ<|IyySebK~k4dM2jwJmGf$<}Gu(iA?kwpvrNp
zPtCz0djX2+_#7n(mm|>s9l!Ux1F(X>`~hvQEI<<FS<v;*$N6%=Q*D2-QIlbzPw@Fo
zn9Xvy9{;fFoL6_qG5A=SGsuCmfe&0fr~p&|K<1yX^26V!1~lzdj{v7f9*o4X15k<@
zpq?YV)|1G!?z{fob0zc$Zf|5vepC5zyfmay0Jg}>^G~N%K7Eu5L|Yqu@wahj>u<v)
z0Z)Zkzvb}KG(qD(5hNc@=j~@o9-w3&hxVqDuGX=Lg?A8K?0hO8Uz|BA0K-P00z7Ia
zraLTh2WC>@oGVlZcXv%iRHU~o#Tw6j=~+LSzkf=K;vm4*4+yywvZhC%rK5{YPcL1L
z9}ESSiz|rn$-0nzHek0SRaRj#F_lVlZhsjs@@k1ZOkDX=#v?JRzQk|jkeKtj^#hL|
zrMaKj$k$GCda&U_q5o&kGT4Uj8@PY~|MzJit3o9hIu3A{;*du#xq+=LrOtDMQKFTN
zDCm9|LCCk569)klK|4+fSd#ELI2qwWd-3NM<}?f-<KyG!x%WVecSjz2w#Bz2Z#LkA
z<=)dD`F(7)X8xRY%XAV&rB>jw03Jh)XH=|hS)T<O2bwEE$Kv>jV@eA|y`Ax(;NUGr
z1b=^jr+5e%9c|3luQ|)%gRtjUM5q^B`e|ZO-Q;dx+`!rk6pecths))vbk(|~&6mTC
zN}a^?VCs3_`_^TSoA{SM*}e0Fk%@tQa|!DVN&A9`s9BJO@OW7OxA`SF7MsHyiz65X
zpd^mv#^*d6)r|t>U;;i66S^8M{wCvd;5@$>PjHj4!Xle)IdqY*YsP+|waRm?kBM_H
z%^ln8l|HW=sJDiYRcG&tgU`egHi;susQCLc9MZW6G-1W?E4#sxXhP;BYK$oiX!hcP
z6(GQ$(<1@=^O#?bp47}wszB&oQgdTFXm|XIe11a=l(^0Qn_w}`!Yr?w*yoJKA|bP(
zigOQ4td0c=F`q<MVP%qGfWJJL&s!z~m<QuZ3J~BYJ;0UCn0bY0DDk|8{c*d<xxFhC
zwacY(5G#w3EhYI6<2g`oZ(;M1^hmxX56Nmk)q|F8cq!n_>EV1r`TAnoKwkX_>Pf5g
z{WV6BMu~*E`OlW>bL|F~0?$^b{nK;{g6=ns&L>BD$UQ9lYJEB>?l-5*HuQS(^x3Aa
z<7v`SAqvYh?}bXDdOAsfK)>#xUZzjM+tH!2G_ONYiGvGID9wB<zuNr2vRx)f4sixO
zWCQiC+Rz(wS6{PLU*tj{(EFReXviQjbr2+%4{BxP9R&zuWocXH&uOHN#8KVMI-*aW
zJ)#RR!f{nmt<cFt@iV|#|H((m_#z7A^noEEGKa4qarnT3^R%0S^r9>emW}q8n)Bb~
zMWr;1?_FB;NGRPG1qV{A+=*5S0{4+H<uEQDu7fU`(|na($<J0^7L;Gnlp=*KEC79S
z(%w<zfLvpk!1P*QHD`Jm5LS6_v;t_FtQBAKRXMCms28fElr>+HKFgplSDM~l{6zw0
zg_{8XLSPJUqLwfkehN$E9Q&~GNEDq?xNh<Zv*~uP=tHGBhUncsH;?_2@Re@<3k5`7
zN%YIqfA1f2^5{5vxIp;Hzkf=9U-~_2XvW6JDgC&?;LFmxyR7@>3W*qLH-)y7`vsfM
zUW>%)gY(|09gm1T<;SDbv@&o7tLiJqfzv4|9VYO=zlJI-Kx3OT#5g}1I&g6kzUsb7
zWZ5xiI?2Gl9{P?SW2o<WMnF$T2N!}#p(cU}$gWNuJv}`vKdm-jEp;BRcAgyhJ?St!
z$QLSSh%%ELKi(ZBz92bngTu<dqV5BIN*~`((9~}S)T~Yb6R`DQGCM=AOdjL%?TG){
z2VZ$eRJ4gR0{US0n-3^)%W<}V))nuK$ZU&GV~UbW@B84}i@nZ8Nq;|g56@49^sp2)
z6&v72WrXa!<?uwDEP02tv`^B;()#K6pX=E?Asrl*lL(zG)vNDBV7A;wR(!f(hbKC{
zqTU9Llf!A&gQ(lxYwZ?~uh*lCoM9JpH$93#$ZAQ<CHgEZZ7vj(t`ilVu!?$jb)>nC
zUAyFIV;2qUhy%wpUH>g>E%&;zt6s_!hr5^ouK7aKHVXHJ&wu9pWJrdo%cgSV@D{w!
zhL~4rax4Oy#6D$wx*FKmI9-U7jhUV;Vv<jJ{ju<nE)0_-iux*6P}Yg9(qe>?_tlSB
zLabgN(?)SmxqchDRMwnr3E(Q9;)(~GfdWLHx1fh`DoIc0ULPnsq;pj=WKwi@Ri()7
zV^!15Hygo{%oBl94KAo%ZXdq?sSvGVFj6m8o;XknPK#$-8h&DDpJIXP&nwz=V^IM(
z2t)7k65GKt)risl>pc;MVyE277yteVt=&7xI1n&HY<(+EBkoRdApFMjB%$kcbKqoL
zo_#gVeMwdCJoj(oId05$^V8$q`{TRyXt6@5PB3xrr?4arBgo_RI+iO{{n4KioeJaM
zs01p0$L&ZrjBxbW8GQDXz`uq7zRiWs=r_WE|IoXuR^T8^-2}rg#(8`wN{f~|X4gBN
zud>P>j@;li^9Wc_zLbNq9kss8Q?k}u^0Kupe*3ETl?osuSnFwT-+b}S6LbBBp5)}X
z6DzyMZGS`b`|FQtk_@KTf#-n{qmPk34;Rz<KOXMlD2c?L?kgLP0KJ~cvwr8V4W=-4
zFSk_0{=FPD^ynT0D=Q0UZk-$ijm?2G4@(H#0m(4_!SY)`iASPH;|(yV@hTbJ!|${w
zbglxAz3JN6ZcCH68q0L9T~j*3p`X-%Rg;^<kOvrf_*MBUrtqP6Ao|DpF|3g0{rfUd
zU+O?~sf);0dYGzkOefsVTVt*^lK=TtwN~%w>~W#P&ZLuNep(rt>GqTPs?7Q+!k>OM
z@YmSRauMhbOnJ^SeJ-a*M+C}ePLu9@sGdJ?nEdn+x<tG#18d!Ue>tv83!il1?G-+T
zQ28d=a~Ynj{Vwz{D${SCbN~s|^84IY3FgImB960Uwp=D}oW!Be)O;ihMkmSAm(({H
zR?U%tm6esD&M@pGULBYDJ;>`7Bsw#FI&#Vg!(%F>1aoo6oqNZyJjK0`m6KDwiY<EN
zHXl@xL#c0V${@3E!m)3CsQH4G?sKU{upxg;g^QcD6fPX?E&WN%zU9Vz_kaZIb{f>S
z?kL`gh*9Ac-7JrJ;(;#(QAUB5B%!}39{mxEU3av4=;60;U6hVpm(mBidEmf>B+p^l
zXqYp)9WI2?AM~hfU}BBx_W-$SJ9OQ36G@pA57^9{BrCRb0U}KQ?%?E`JIP_Bob9Cq
z1xRm#Ydwg=ZT|y7pnQziQS(*dvL77@hwTi<`OcW69&$`(Bj69X_qGAP@7+%Y)n*WV
z(zg}8VI+FS1Zo4epS@>WiRdxg30(p3<4(U9V<*vL^bt=*H>z`so`4p>lvli=PEFrk
zJhlEFW$$R<&DnNQ3KCbB5%9|kul$>W={YudArOfAG5ZM3lYY)jMw4=V^m9zW-9fu4
zK3!@77^71xE3HotWvKT$)z-QCn)=F~$_u`C2gIDlfu7B3-_c{(5&A|yq}_vw#l<r_
z--{Lr`YrwRUw%u#QlO8g!DPTCOi2Xf@he?nc;6Hdd!vW$KZzM1$Naec7T)J>TiH-O
z`Rkl|=TR1`+Phl!z!Y|iG3Gl%PePqx(0xy0Jyo1*juk^CHbn+)OaQVuY>qgP0^(>w
zw<(Mp(d#+p05=Zz`c+6O(tcPN;2--tK^B5skIJTZk;??AJf0se>P*3Lpk9*n!2mVj
z!WDz=S$Pl}*d5SyRF8R84bJ%nqesXYX!?;4@gH|Jh(juCeQaoGIC{|v@sER#y`o(_
z+vGf5*3#G6Bk#S7J&R|M|CWD?s0Oc!O5gvG8&Y&ain}sULm}k0GH>7dew*zVn;Zz`
zFgGSEW}Q0YtnsL&^Xc(QjMvn#4F*=qP{w5%_ku<xDh%|dfOAtFxrR7b-8eZbn6FC+
zwORSl%*{@*QhL_?(Rw;P`(SIHcSC$97ZFz0pG-`@PN?{^99YxCMPyKskWzCa^_HaK
z_hRL63PXLH7_!+$*xOrU(PLJ@zu3gGq9S+R8gaGFN+WtH{b!LTu6m~4LXb(UslgTS
zlejx#7nARp9lkBQOB(foQ2cXsT675#Vcx^30?aCWy$L=7d6ay1x-EU6U48ZpnP4*A
zw_I(=hnp9^fbrmzym$)!V>~6uX_)p%?ho2qMA^@#DCZ28>@P^DMW}tRSCQWHIP8r6
zC`^Q*>p2LEo_bg@uu^|$@&f3o1rr+viWdyj#_Q5P1?TbfuvlajU=m9Ts^nI4yUL9V
z#y!_8P(#?%?cZdCQ>TU6Y*x84f`Nj)6jKf&8z;jCFG~ycH<z=kvYXob!oZ&+@dSLd
zLdXJ<fj7>@-jp+{OSt4P7X5WX0ft%rXWd95wSDE`mm3nU;`GVXdiQ+xOT~vw+l)?}
zZ-ZI!LsH9ngNR@9zo?X!jn<G;$G=Qrpm_td+9C$OQ9tQza1YeSG^Kkl4b<0+0JWhi
z$>zGV5x-nlFvhLgEVY;}Gfw8>+cGW^+H(6R-vmxR*=SG#$e|VsC5B4$2K*`y;Ype?
zH4Gp?XR$B=QPk7ZrRkuCB?X>Cc7=87JU65ba6y_%xMXj=SF3&#PAHIuhrcx}bfg{x
zpKIP$Wi>Imp4;Ebsb?U5SiepAeqXtCbYOksf<b^r67!4XmxsRkP~GOl>96__z<jWw
zctu!Dtx3h2uD2pZ(6x}+(~_L-#@<DsZ2R`xpp64U?{^NRLwYh=MWrRv{k#;yeP*Is
zX0#0~y(k$K!^}b|L)X*jH(SSFIZuJHNtXChDd6zx@>3GLpJuxejfJY<*ZzHHWq99M
zrt$;AZ1Z#&8BvRB&3@+G)!k*-SB5tAtM=0UOr_S&`o@|=G~~WE=7}*YCxhYYs$-Q;
zQc09Pz&>Wy(%a|32@kAB2mOhoUWv3*W<d`*{YybTG9m@m!@T+awTMkPq#{0ErCyt{
z+4cU^C9j(H-z2wx?JA^c`ye(pbg5bAq4q0%8!O*(G47e#%O~rYW#{@Vw)K_VyVm;}
zGR5JbP3yYJu{gU+Cs*A=mezw){#_0BcV~S*^m2{^thM<C;LT5_ejXlmq>b$0Mq$y`
z9GI|sJfouem5+dZ%ozqrKdUP|@bQ-`h08c~RhJUaV`3a{)f~zJ^|d_>N~Syd3WxcR
zwFR_(xNQT@{7cIom;#A$R}7uj8}}-~h*g$j5+dFFwQ*PU8Cr*Qev*-7d>WF}5!*40
z6PW#8@gs~qMW4>L3gt>#*3#iB%-eYBpL#bu%FSUEHi-NtLm&%{5|VQB3+tw{EwcQq
zhqm5V;HV(uQ!aCSjxNgZ>BLgub!XkM6mX#o?D0E|B>b>-D-l;l=~&5do<CNhW*H{X
zg1C@m7TZ51J%!u7rh8FB4aCh&W8gQF2{108{EK9n?cS}=RG3&t-SITrG~Hjt8-B1p
zkGXM70yg?f+A!FIS60R9NP@NW=QZnzipm!C2L7$@X0w*eG%@i3#rw8WM{^0^kz@92
zOxGeQfm&Nd4)n2c5``lUjD8Lb4ZKzIyze7EFPX0?kb=>J59=P>5=M<^+x~?y-q7P1
z0;1yAdaU*>n}k3p;Pii5pno1o{(|^+H9cp;^dViP*|^^;G30xZs4&#V+7fmS@9@&M
z9m<FXgacH@<~jxxY*NyGdRP!AP~)qAX*xb-`RA5CNOK=6a<})fc_a@<TynOrYf--?
z1~)mrS>6HWgt0Km`7@W%3e&$#4#wQcVS%#cqCx8w=%>Ilm!{5DTXUiq8Ud&XP6BPM
z>wchiwat1MQ^u>Rofy;e)^tuqR(Z$|qikvGdLAx-SsMZ7B8PtC2DD&VnBTS{O@_&C
z_J7u1DPw{x=(3+9J6JvgOD<-|gI_}P@vX*w<&siz*mR@&4E(0Y;aE#Bvexp%5q{Wr
z27Vp4H3Fd7@+?>*EnVmdme_;Ox32R{17%xg780yjE2HvXq8t9hkk1<Ed5%Ue>6#&3
zCe5COvk0bN-q6T_wEXtM9-qOHrHuQE6vxaeB^~<q)|m8X8E#KRQ!{f~M#gTdLqh<n
z5rLy$84(h0dvk=*wvquX=uZDLc9)MI7uZIYy(BT1E|PRQCX-L3jc~Q>%=h@Ag}nLh
z{C9Wn3m3l}nF3{a{xQgH7ao(t{k5YHoJzfunO7Q!$cq;0iL${6Zwd@cf&r)iM~~p2
zl~hwhg2q9zmp&m?^sH6bPGwl%qyv3yaTlvq{%^~~nAGY&pv@Y9L4mB2Qg^9d<D{tS
z4+#XoG_I&^d9hP8GLw`tlL?+Y1QD6>c|IewlpKIILU+EznZitHp+B(|fZ;~~dn;M)
z)MuhD_Dn*(&eebw7wLn?$)NTM&tsN2dLnFdayp_u0_<smyAvr9XbFxb<4psJJpfiR
z1^_45cXxlUs@RgH<)dWQr6YkjrUyW(zq`7OuhJLu3AGg?NopF@SwF0u)V1+b0;=SN
zB$|;Fa9wMK?i5eWa#Q!$*B;k*HwAuIo6XHrZdlFy`dUWbepAI-@9xMj-s?S1VIv$c
zNufTHJn5ZYqGE54PgH!If1M`?cKtgUr1^>66@B9bc-%Pf;hdjIwqYhuyA((d3&;R}
zm+tKRb;gWO3ryk-uP_`_yT54_E5D;(nJz|aa5?lJc3%n_<_qfhOZrp^6w05wx}<oC
zJ^=&R=OxoI2d~y@9>Sg;>;*JFMu+Hp(lkwF35nUN)k`#ST8x%?*}1;D-{9+qAAiy~
z%s(^-Fvmj`*fs7{Tl5cr>bZAtu<M2;-ko&{z;U}v1_`M%^nwxj86l=yAUROox5{Z`
zl}<>G06jPY9>7~oSp{SlYvI>DFFLzEF6~?oh9tE<yfBVF{CyOE_d{cl2fd*TkCO5d
zqlVFZCz?}CXHV&#r<klE;se6fu;lR#pEIY=7P<Gul&SDE8wJ|!a<p%@-x~$%s7GYy
z%3P|>KS2{5=9W}}d_}>YMjyv~+EgNLjA>pU%KQ7I6Nbma&-K?6Tad?(TG!o_ih7vs
z+rFChMBm+dLSbaaxkRrEp5bmI@d6_32*`pcN+!R7lvQ(b)4JhBxaMSlzrP9q;GCVK
zkLHhLJp=;8kU+p*1c++fNL}5uY-~y&@zm(3%bhX-M~BAmdk3opyb0(pG?hLkZgZt@
z`2I{H&a~N`GnlOeY)uimAFnh6n?+HZ?eo$3e4^Xkw%8S}-LVwr<H6HgoySw|>_4iX
zs&6Rk?KFKWjgKeJd^o&KgcNko*&gw>a*sG>Vyf*aAMaKMX2v}szSmVk){O=aM_>Lr
z3p>R$Y->h%Sux`_oXd=qmW*=DppoQVt{xe*ws}#p|9&|sb!TM4c{{6weAeJLeP24i
z9P&2v0V|{VR^W2+mjCvt5za232%bG>UFux>o`2QVyKV!YR8$Z(+u$IWP(1y1y_D?|
zJ@c2p$Ui5HGZ;}8l#!W6EXWAiFqOmw$H6<gRswWob9%_6x<95Ih*rLW!T8{-75~^>
zt~VaIBiC5|-22c@_Bx~T!6YJy@U!|ITfAU+Qhhe3(-;bmPm9=L^9vkm-4*?9QFM|8
zlsuxyZXvO5YWZaowz%`8Xd{uWw)M^AB(5qZLEf@}LYg1>Q#m4(#G({vt)Flr^{|W&
zu0=gOMz2CSA2uE2GE2HPG!rxX|5kFE^-5amiS>%=cKt^C9C_4M8PBx3i?}=UaDaWj
zcZTwsQ7iijS=I6RwvbAE1M#zpLyj^2JkGKL+=76U_Mao841vtI4g^3HvjH+Y_H9B8
z7g>0IM))LhYI)g5vR0o;(r`qPV5X(?XuTV^Pgl_D?{;tSs#8BpzVmB0V8pbul_NX8
z=wXQoaN8NH7TpKrHP<6O(+(xj=naPg?JI^OdgxgY&z4e9MpWZDnw&4Z<KCp_&+Q5{
z2{obV#zwvims!6kqm3;+;$>hQ4P_(yrz>LYwAd$$2&BEiUjY0hDh$NV&d%!$lNazR
zXP-a@!0*GuOW-xm1;Nq-;V59P#(kgIYr1zjy<xv^t*8tyZ}BcEK7B3P38D5ONmdON
zu>Iq}PR8_yy}Rk!Hq*(O2bHDCt5*I4UDY=d&2^&H#EX-?#y()9RbA?MdzKQVW0fV?
zj~3Nk|IjLX^k|dOP%;|uny5HW61bClp^%BA_xhh$97iGtGSMQyoJ_$-T>!ALIkp@m
z*kpUmEeOa`+*o)arhh>1`BiYi+L`wFnDoC2ogQf{V}$OPkqDa6&~IL?(LX+#L^coj
z6$v7<?m5&UWV;nXKdy3od#r%#yDVXJgnTC|kq$VdP%Q(D|JcB+4$6GZZ11;lz3W}5
z_5*4MLr{3K1}sjw*~xxIvvwENqW_MQLbCvy38t6Gb6D)Foco<~3iD=C-HOg-5NM#8
zN=gr6dF+;6l{3y3i~T_8<&WykHCWGfNBpsTz&iS4s`cc2wDx#xVY%X9)QTffNy1#3
zBN#!p9h%HZEh0j}q?{%J;4i;^?Q3Sy;{Y8vMPPcobE5Md0qJ?17H|W?mdIczAlgQ!
z>Z8v2cQR|M$1<Zpk$<fI9pn>TnHoqpQ^MZ4EzMK#Uf+N-xr*5KgSU!7J&#s6A2suk
z6cZl_Qu)DE+6V4B7Qw4zm{|~FWlO%h$vdAL^sJh$wVW37*UyVDdX^p}2WoO;)UUIk
z-l9^JF^}S<(^u+w2bsOA*oe29@m@n8HrL)f+7ceWWNp9B)9>IzV*c^|%82-B(f-=)
zkj$fe_~onb=gg0HhU`uEzXiz$%0KuAo&wFn(MOqTz|rg3_Z|a}BQQ9)kOn*_0w%gM
z(XCn$t2NC0gGx#WH5FiM$)|cW@aA~HYdY~D3#{9`I--RO0lMA!+HR7DSHpEbvILYj
zZBdYL5{wNlZM5N>hFTpSk!`FqcHv)!;JEio;q5a}NZQ7|N^OrNSm!)HYt5ue@vfuG
zv7@`*wGq&{T&<H+K1qVoHF$5`B2?tApO1;mG@QzKzb?jld{~ivrChaPTjZw2nIvWk
z3TC&7sNi1e&MqmMWhqAGGnm{vMi<pPMulv(!j>erO#Hs_;xP^nua1^wtY*8K4`pw&
z-;L5nd;)q*!Vfgv!Vgpnb9IRR0g~8%G6@cSeTB_^gh8%)(Z*+0;>6dt68j@7%U=iz
z;M+cZ-@^bDMk1bPHh{rY=2{upb8ndWR$aXrZt@&xN`aEq#$J$Ho}+WbL8R0Ji+!%o
z5|QtmRFoq;ezM*x-YPR~Pdaw(3?v?cqUcR`1{hZkLYNCNp(r`#9aVYm|D3M)dmGu`
zf3O;}^s7^VdC*P|lS8-6DG>Unrr4I{FZ>kBa;8)|LCG0UVENpbUspzBfk!JR<MSRZ
zdQ;vk_(-*0<Yl7M8h0JN?Ym-(Mhi`r2ipn7T{|caOKz0g&cj8NATHq*UC)L+9H5_%
zuJ#J>dBQr>USU^^1CQq@!RV1@=wSKu<BlFD=-v&7_}J&zOZSJ03mHfDjetz->{ruX
zhvPW8&OUs+NVGB+7DOeI;LB<_)rZT~YY!D2^>KH-S0fcg1R(jI7Q(lgI0=DrEtE}T
z8)5e4Zu55g{G?&OxqDXXW$Yxgp;$hx7Y>5yxaAa`dj)=Nc>_bYh%BT2ccB@aOb=jT
zUztdM7JQpLib$%RFrj9%8y+`><nZ?=8p;$}A+g9D-xz3AhI3Q*2+9G}I#sTjLp7cr
zVoECR>UR)0fPB5_$AFUgQDuIH43=MK+p8oJSR&REzKzHlI3k(NX)y+WZnakQsFxg1
zIGQdRR0&Wrg(9yfw$IJDk0oZ(*w+VKg(~*`CXN<Z8*#Iw{%%i;y{&MTOmK@-9s1P=
zVqq8!ecMfMfVH`@NfZ&o>`T<pu0xu)Uhgu?Uw<p~%5IcM#4awth#;@r=SF(+@v0>M
zz4{_WtN9MDerHQ+HSU=P1=q(Q<A#+iDd_O04b$zF{rXDl+qNbV$UWTs>h5O27sT7^
z>wFz6=_3$t0yQw5NkAisX1w&98_tuqNz0YT+5wG{qzV3UM7cXJ`z~AW0|FisWapRZ
z;qR1;+POqMp_P-n@8X*$tiQhiu`;g0sfPw{-9<R@)h9Z0KdMlsMqdA1r|#qsoO&m$
z*UeLS(df72Hc<qBN`BE=^M1@{&!aKF^Ue2^Hr1A<g=!Cfx*LTR@4o^@fY(;q17uxy
za8ydPd4Z17%UVL%LO<ZPDXM}C0{Z#sAO{5q2MzQqm2zNEbk}6^vWeq{Y8T~Pc=bTf
z5BW6<kIYY+!e@P4`xdJP4YHVVe^}cz$GBof@OaZbF&9ZS%a6E3fcx;Re;<f@4S3)k
z`qw`7@V+CDNRiF2E|qBoD3R3V)_BMb#0ZYOs}YDg8k-hybc-mN_iVQ5uxLJSXZ<vz
z2Ln4n4xac890LsiPSK=_c>RF0k0j;AUZDbEDL1QRCMGr{cmgP6OcyK`K8A5bUMjI|
z2#H3qVdVr%^miQ+SH=B9kAr(HRylLjl+F%hJ{ft?!;8?g#%Ef`pmT%f;1!Q(`t!mX
za<v8WQ;g*DM^BcVarC%RHSqB<u2saGCSa?NKL%JO|6#7Oq>-j=`|#Qv1N-I6mr4^z
z8W;s%`V&|>hZYqXz&Qmp;M-y_KnQ^cg&<<<f$gdtjxs;j{U+$0eZ)!kbt0`v=QM$>
zLzL|^B=#r>av}FktQvg%-vU2!Ga7fNqu0Im!<qPbF5{fBAa`Bzd@~vYfZ+2(F~~k3
zE&C+|_$n_z41D5q)osI;Ow!wjrXFy{FF!R+JelJQdyRcc6>`4qh`^hASm~@Ac8LIp
zxj=WACt0-I8)%zy8X}{hhy$`Y&T!rr{2yDj+39yIBw+bDKu*^RSXSFApKaTIhiT^7
z2PD7UoUP9?-ZW&*VSi~5s8RFw$|Lr3of*~Y=7|l!101B(mWmZet5ASM=9?y=b5H;`
zm$y~ujhQGe)qe1IpOO@jD+$XU3r?$5=D|BtYP%A(nZ8Z|@+At4JA)7wh<!mcwr@a^
zzX=4If{wxA3AbTtvF@Rp%1pQ;8;qcSUXWjUFRu&&9zFzsO3NtPD8W*iN%9<|HHFz_
zriM?GK{H9lEl#DSV4D0g_B;xaFtpsaXjy5k<RT^PNFA+estECewEqWRxmyZ`S40o&
z$KbrvpPjoNUwVQgM9?@d+9-%UbO;C&UPcweg?{+e*|I-}x_D$8aeB$ijQFyKsL{PF
z<%_a5082Bsu_=C>F4Ze6F!S;8GR^TULnL{k!Y2+y6(s;jlIEB(^su+^mVc$&p>!zJ
zr%SkFzis~)lYDS*JuGwYc94zsnMqDS!JNp;*3LU7m7yTUB7pJcN^7u7>%jMmvdCny
zagY&2KsNFgkzGs>9Ja`4G@K)9d1!FjPVmz#7M@g94q*EH`Pxh+Z%)~fro8Z_KSCLx
zXXo*;w%<38Z+F4{Xr`3$fcvYc;UZv)+A=dUvrq#Vz$TE_EwM^3DW|*A8PNnml+y%r
z{UBag5Tb*p93t4keNa05+|x{AF5l-)4oG!$=oi!fn<lQ?!P9yY8aaW6kR0~kNxH_h
zsqgqn%`&k1s!h;%iY+hS>5~0g|7Nl<@h<l9T78gso+6S}gAVqb?Exr=)~Epat?O&r
zxvP$YANy}LUlsVR&tv(iyhHhB!-;H<VQV%-ik2weFflOc_2~O}xmCtz(o8(G<7Qy#
zTtW2LS(~qBbCyYSFVgG{)*eIV)#ar|GYH*j%HZ;SvAXIIDK!eXfF?$VLj_Qj1Dr?T
zU^{9X)So8GlB)HN*3!?2%!lSZQ6}yYX<wDEShJgblP9R!@no5-RT(s*Pi3JMiF@YN
z=GIwJ&$%_@(ENNPjgx9ajW~D`#YI&3{k5G!b&9nWE7d>gJ(&abqIZ}U5avqJ8R2@J
zenppb>_@Vm42tj?CEx@lGLghCEB(H@<5n#l12u{LL?YE((d@~MYtu#;q&%#DcUa#z
zn;1S1VL{qjhEOC1q+TId{q27JP~!wOgNKLDXOu$}6BE1EIDNxlU~qkX-KN~=`U?NW
zk&eG_y?ySN1Qh^q@z6Y$zB$!Dp1dsBg`<c=mCkCnM7+eO?vDnz<kd%rK8mHr8fAna
zcAo=dD}MZDd4oUK3&EH}LrLiu(2G)$vFtCG@@c;ap7y6gKpO+WgK$DPG|uau3z0H{
zLCK!i8LX~CE#9t-wgEp)BwDu+h|Y$h+SlCqRwEi`qY&{dr8H4Tb`(P<PVoEP(2Md)
z#UH?WssYI`<6dY-1~-SfHavA03@bx=ULqX*o05%2nKZz=yQuyv39)|J!1TFJ;dh?@
zDfHtFuJ-5Ct{Ek9chQqt)bZ-4cUKd`IGdME2TctL$#4KsJ7PG2J`f*QUA+0e)dY5g
zy+cHBzFiv`DkY%aRKZP|zZSk%(s^=RUT|%@qKGhUDnJx)(b}`e0=B$ya<3{d9i9*%
zdV0NcwmiT3rQ}1oa6?R%{mYpO@0XYTCmtnKf=J)5ko0lE23R-PJ$KPPp~tsD!+Rz|
z<dLDluGq#6V{p%z-Ae_;B+R_^mmgj$l|EvIlbyGBVifais>r3u%pGQ&JQgj}mJHY=
z-Og5;ap9D2)Gm*<TDb2ouj^twX~$>IO-~c~KKd5kAYB^}Gph`KGw`QwB_!TRwM5Rh
zJwh;iHQ?OWaE^F<cJXckuA#nxhLu$j;Sq6`y%{V~6r2Sh*@b=1*BXU6qTz}9{G-Iy
zhV<~{cyNe$^bi~Y$CFl!08vU<;pf_wM?0Y8%B)zfnFvj}d-?MqHZYByRO}3eQEaR@
zhC09aO!`5ej)Pek|BgwqEAfebJAQdkd{1n7H9$@W2m2AR`<(aWr1yWP`5rAMrA6o!
zX{>$7-a{VcOGf(V+dnZRVFT;YwmWof{#lIDFO%lO%gh(su{9Q=(yhk=N{e@djLlON
zkv;=TL2J+Kh9yw+#ojYJY`Rw8U{AY=5?!9iyI!rd^qYai+0_|K)219{{!jZv_(1sM
zapoHhf8{q+ymT;^Co#a|xi631rTnYi(|CaRVV*Ud$Gi<3@denlH((MmMyH1c><<=G
zm8+x4?=<wjkD&U7c^H(poDs?J*KGy5AeUo4(E8*;f3O@FM1|h5p6x5Cv5D#7vgV$2
zF?WCvLBNMo*1f}P#n^qxAKB*<vU--_<)9rE)~h5j*gwLVb0i~sKr?mjmD!Rg7cFy%
zQ|x(!fSxXbHQ`w+l|9Owulb>94Erj2A%e^rY4$H$68iJ{AjX3kyLLTJopHaL`-F45
z*GeW-GR4His2B<*3stWP<fOXHBqpgvm{?)Xo)n$T78%fF-ysn{%{}BBNL#}<J#2m7
zqtAD(WK6oevA7-yz*=hwY5eO|#?Hs5NuVxoBuYi%T(+n&s(mAG>2#~}!C8TvxDSyi
zgym8VI?DMWF&*S(d`Z1I7$osl4GZcCf=56QpTZlBM^=-Ero2V=@az~`6ai|W6%J6=
zz-CzHx8~zvEaww*57wLm-H7C)mM@77wUV9IwJQ~|28MdwLmo!AWglv`gqkLb82ks*
zkw@#VdEV5Xh#5`oho~5>^PnHjBPnC-gtP!|^0%b!3w$;RThV@9O;hI|`({Kx5~rs(
z;1J!~5y)U$hwLx~{_KP&9ZOirN^8@L<>s#+XCTNoYuc@M-c~<jGobwYr%a}?=HpaC
zfej0Rj+i%gw8Bg&t9xfx3Dv3?&Ggk5cR(OakP>iplvr$Vy(rQAe8547*+e8s!~|mP
zG4AVl%s2S(D)9_U)W(lpR5%G^`ncf3#qWT8PmC5O6Jz5FuMs<Sw36VM7;K-v@8c>d
zxDWRO1b&yXIdisJoXkF@i&qx8l+v7y8dR>UhVEbUUc2;~c0E|2FL~rRf+dxP4WA6W
zR{mrmbO@IXQw;>I<9vCxR5YBCM1uyIBzVaHV{F}8tZg6jAIP|6{a5~yOiViMS9rL3
zv64I;=JpyidtRvMqy4N7)%78yi(xA*E`H_Hmwdv&_Qc@SI(kuddFE8ad*k~}3Wg3B
z<95EI%Ilg*S73&%#sFVxcWE5bed0EqxVhaiHKcff8nb!7)M{&S1mc2AestD!Id>Hd
zdFk_7T8f&J69=GhN)C}?PE=`4@`_OQ5E&YVTO5b9XnRi|-d{C5`k3ae$W9X4&R~D`
ztLa&}9(z%|<bil0SS>ygJak<4T_;<zkud+j{lJs|ThI~K`hDQC;~$OJY*%Bst1iMs
z&L5+gH17j=b-q~3AA2&bqn+qCwV3`5@?j~}r_G#{4ew}jZw)#KL-D9=lY%i?h-rAA
zW2wn&xrw}bHTV)oWllD;_>s52<PTfI`XD4q5#nmq{i_jvf3Ac_OLOY68l;~>Z|%N6
zv-%Y|<^h<hd)Ew3#bw2vv{CfGTc9>*<P8=eu8m4UrlxJ^kf7`$>REd)aW_7fH~k<n
z!@H`1GWFeEy7qIK@%dMhq!vRFH>&WiHTyi1h5V;1dyl_1Dah2!@<1lbpwuJD5L1ID
z)+~a7BC^zlRy!SB+ae%Vlg)YEXoms+T|~eMA;I}&-p^%j|I;|0Y#rYvMzF^cqa=jh
zK@$ZTWx?6W0dT;xU2dVU7}m|FUwhXs!=SnWxW>cr1arxbl%8tG8-cleGU-X_t*90C
zCp^SXwV%69>X#fOhH9x%A#8G&s`s3mHuRz#<Ri63+8#k`srX%w)s!a}603RkS;h^6
zZ36af83x>?GZyA%G$JB;i|L|1q))cXQ2=v?AA=VekZk`D$&rAq5(YwfydZ%@`~Z&+
zv~l7XHAVH#v$V+yKetv?cY91>R*r&{%s=<}i=RyCmmVECTPR%k1Z$hX6sR^Of-7N;
zPnnL9{XdIUysLXbV1gI&ZN>I@(jYw*7EC&RqCL*JhCPx@0YKmA@o}4qo=Iqg@EKBr
z*T7W&yQzC1hR|&i1pgSP9-|`@bu@6|8f237JYo>~mT!U{&UUviRE#oYT|0c9DkwC!
zlG9b`Yl}OHxi=Tq->geIrPh%Xyn-0j)1TcyuFk;3+9#?5y{Fl~J##Ofu~XMrjv@%x
zG@1Ufe)*6$&Q|foUMt35&w05$w0_<}e_2iVb3V>~Hv2nAOQT;RD&JsQ%Zw4AhRknN
zbky#)(nf7SC)D-gNnH)M#%|tC)gDBGNsXXj_6OvS1rP!~3ISx644R_-C&h#vc25ed
z7EZ#)pK^#B`>A2ar9LP9y|Ki+<`DDyI*hs-HkW!PU7G{di)*gIWPZ}@st0@FI}X8?
z26U$Lo|NQG;Br*MR4OG2IaF@`a<a~7bXgV)J&UTVen(Wo3v#>vGE3zHJE23&ylX6S
zRP+04MPJ;xul&crm=-!I{ej?;t{z{zmrb>)MT1Lv<*4^V;!^^(2$K-m-Q__iz<W!z
zhLg32Y4h6ru$Fya_s#izGq|c;{}=ux4G4GR?^VKf+(lF8^YBDyudsq}wc)Pw&`-iw
zbfvqK{$^0CWbN{*7k!idH!oS=p}oz0yydFSycD3X`sVK)l_A;kB{Ea!$VuEZgC|=W
zs;irtGW_BQKTwUV?HO*eyWB<JtdHD<Gn;*Z`2qpcOe=k&E`!4^tT^x;n&5nI0Ycip
z?8)u|go0wjAwZh!?~7K$WzK6i-Pe#OTN2Dng!|;=WM&j<Cc|+&YHgtt2L*UpBPgID
z7lur5+}PMYG4PHz`$HzD4vZdgLo^QFA@wEfnVCUL>-^mK+!5?&pLvZ7?_|Rx2O*x{
zKy49b<7LHort%eH#Kaj-Hk4zb<ZD+K?y0I4sO2trg5p3NKheIL(}N~mKF^)WLH^l~
z#h)#FrKEBLO;+p|`V)Lfd>KSQInc~k@~kkxpU=267=!=QczxU(LqqDlsglHbvbN5N
zbqOqx9I7lk&^=cmc>$p4dr&KBW!OC~{_|-zWqa<R>RI4kqtPL|qj&286&BA&tvFDl
z+d)T4wT&YX$?qs!029b(1;s*m&A#Riza_V&N8Ro-xnJUTXP+q6MV?P-EbsGFhYKqS
zxL(_Z5|Mf+iI$_54JwTor<5iEuu(reO&jog{tw>XGOViSZ5yVME~PuAr4*DDq(cxy
z8YHDta??t;v{KRy64D?ojg)k!boVnG{KdcSINtks*EbQjv)5WPbIp03@hk4i<jIV5
zeJ-1@iAi?#8N2Gq@CUD?q@A4SBKK@ri!5T7Q!Th$5P6RfutNi&c5T$0LeK_4A&wLr
z5b!3!V2i^EBVM0bdxH2!kYz|o#ew5-y^lrx(BfMOKcGil7%m5_>kX@D!NfBPzegZ!
zh0UlU;ekD<hOu<f6aGxp38=HscP!tp6Fg?KZb{U|w>`0*nwDiTQJ+G5eG$M-(FJd;
z={ki(B*L9M^33-L9sMv9r)U9)5hP<~6Y2MLx2i=O*(XRTM4@+Z1S;zoP@#sc)aQB+
z*HMaaS=9>5#WidWGf?B0mkOh(sL1Pjlz+wY*7i{6Jv_J-8mwK>khOJo0odB_u3D=&
zvl#@aLzaN#z<KUD!0agi<s<n^-rUZ2ix?L`IWtJRqv`+y3=l^K%RH|;B51}+ow**G
zPx?a@T$iEVaPTT{DK6#f*{@1oTgrjnspS6AQN<`?e?kIV|BoMnFFfBi)S|$&(|RtH
z?j}Jk7dviv64h|Qp5rBapy4_d{_%q^#P#^~<=NrbJ+hDwT*e0iZ*6&mXlW5JNFIt=
zTi<iY&~X+2`XOo#fxz2kfhX8FrXV-E<|J>*3DSP7XDy5V*K42*hpS95L>dJnE~B8J
zs(SUva?<~kq9Q>LDrNe)$E(A2ljjYv0CyZxfjA2elQd2Vih2^`&l)C^2_GVbTMQn6
zWzFD~C;tdwRO}&L?A3<GVFUFFbtyjdT|k{9T$Je|SzIxZGlC1Jf@ymI^QH~k(+AuO
z-Aj<#2P6v0$2PQH5TG&OHtB}}+7#iX#>UB?FtmKYVLA&H5h}4Ehc~_NNq$1|12H!N
zW?T^E>Y3gTq-AhQ1J}*V!5IP$0k0kH?JYHcI_1QrjT^%L;LW@kl$bKHC=Hp`ohgwj
zwwioQ2U9kyV;O?rIk0$myg3slm-t8@@O%9SwMB7ZoTX;T@7_&Eg_FjmEK}!E!4lTm
zRSke~6@XET74x=XE{6$rfmx!Rjb*+b-{-p(Ac?ikrF~k}=JzZ^7(yv^w`EanOqWv3
zbJ!UDtTj=wb9_9y{YH!PSsZCtTL^Kg<y7_A%Z<Q*fYvR6nqkc~PQmeY5I5}1PYUQ&
zso82@=)Z&3QS+46CRd^7%3%%u1(wS?3Z%d!On?vj^R&AJ;zeliVQuBg(o3^aHdaPH
zqAZ2vm)nORE`{LUwH@uYTH!(LajDMX@y<cNV3?l^x$2|Q$-L9t#D)r%#g`VN6znr4
zoLXQNva9``{D`P=5m4#_LPN!Gg$0;Q%GoJa<!m{2)B;JlrVR>SA|hbC^Sjn<3v(<S
zRIu#}3j>)BX26JfT|<U|hE4CIkp-0UFAG}ffGZpY+bbQ3eqVCBsLFd}_}a9q)RH9+
z!Bc#?4~R6vV`4I6G6;kfhqyh|pIFa&uCMFR<RdYijQOB0m^Ora>J%QC;Q-@TKws?l
zai?mkbL%7aY8XHd?Gm#;1qJ8g#1}vR?605}exJZ&x@wkz76nTi2b%eD=+;moC_Vhg
z&Rzd$vXJLMHL3bzUkXX{?$}19w8(kcW)%vI+S{?142WN+nM{9jL{g~oW(BWTL-^J;
z?m^eENBmx2&lE;hl!E?#OwYO<;OY|y-h(L5S_WUnjz;&<gxxFcTP2!3;WMm3SFo1F
z#2y&dKR^fTEN1ObE>BhW{n?v)6f6Lrov+I|OAKXBgS!Dn7L$mcO&nNEte0zApO}RH
z+^hI%c<BUj`M@EQmBh^UofhFnXJ@Br1Hu`uZkHL6{uTQ>oD0|3;|kS-<2`$K(eVjc
zzE5e9uquGwT~~dm+y@hl9!?BHPHw6O=l1!$A%Dffc?nE~!u&dQYnm4xd!%i>tBi^X
zMfTb<w@l+`P+HT&OKJ`SDtSkYWw`+Z;x{LLHc6Q$c~|ls@`Y}INPfV33<BJ98<T0p
z(~Uh5rU9$PG#T|hxQCMA$t3s9&ihtH>UpySW`1_@*DDRod`)Ao1bg+O@4{&I?it(C
zuO^7JH0ndEX>_8k{$f2ES5D!HB2+lyP7DE}@bg!G6BkKOifVNh5U2VgcH`zJV><$&
zPffQEDzO2yK3SyaYBq8%Na3F@^cQx&-o(jD^Zo>^-l!t_H<9RIaev?4jV`t|mbzV6
z4IF%J^G;-9At>Vl0>2i{$(E1?$L089ZNX#y$rcu6$oSyvn;-N%vMGvyo^|5+>v$*&
z`9q%xf0)h2*MfA?hVgw@HvQ|Rddvtlz{(D8N`j67+^sl3cbHG0g<q5rk@qZMqofoj
z!-`j8DYi8P44_X!Rxg^M0GV<TN?sJqhxTUq$0u1?S$TLtAa^lEz1-9ka=gVH#S)>7
zWLrz88@qdn+bv`0filg$zP>Gt*t>qcZ@(~lrIRERv&3hcSmFMd4w2-MrLq)8ym+y<
z8-eLWhNIVAjBZG%+ugEH{)K|zjpi7$19K0mK>;2DFN*;~uq6Su_Z6V8m?3Eb-3pNs
z4Pq@+?6TX@*(>69VFU5FqJ>dys_Qj+QiM2_bbm!%UO+N3GB;8<iQz#mnq>HP$DM=W
zx0}}?dDl#<v4$v%hYmS1Da)6!knsZb(U>TO<hJ)1Btym~W5ua;#-2x$(B=bNQ9oX^
zJan4>DTtv5ulm>Lxe5N+A0ma-s(+2<tI3oTj|YeY?yX9cHhf#Ded{sSj;fr?ymBFF
zwu_pRvrA69%;4bLX~By|4lJg=q6GXg8PCGy%h7}b%J9OA!SoeRzpRlmN^KqsK!(fy
z=B55(V#pfqykM8tag%S+nN}o4yZOp{-C;?9ut*NNquaAWIUJb`O<zs7jxxPnSxp>|
z`z`6RM>pUlqhNF_Gg3aQI&Y42x#uOgQ#c4o!I|{wal3jqf=!ISp*BgH#YD~<0w&ZW
zeOiPe>L{3-&xAnt^Na+9Zrta2l-g+18GsEz=Hv8P3Okvk!_}-y{e|l4Qi%(0qo@b@
zd8yrtV+T_L7v$_DI2SHlbn!;;wRZB<VO`igC#PW$5iV6tKHx*E{Q){4-1V2{LCK!d
zoKj9nBH1yMYnxtq$a?Z9Ap-rmTtkumEW1cuNq$29MpOOxN-+!T41y=OA7DnqJeR||
z+ON2Pz82tGHNE+y0_J-xPY!9rHHy%tE;7s-$|ei1)L(SDa0-u*MZqyq&93JryucMU
z3vCjDhd}EyrXC<1O;hT;lTFw0Fz}Q}bvGns+9b5e-M!~)X_5C?tydlQC2E6Bp7x8F
zeUu$EQ|RpwGeiO86XkW2x$=4g1ysVOD%8b8-RTk{tuvJ<hdQEj72WNzvK2e@6GJ|$
zE|Om)H`{UtI#i?j!%h}%QFXB>8*I4Hf%|uZ#vL_NbZwVw9*kxnvJt2XnYr+O{`{%r
ztz*G1zoY%}ZNUu`>Bk)}Dr|&1YYunwNa}7L$smAmhBkqIAtEa3vn83&$_3o{xA8-I
zQNq$CI}<rVl$$CvZNm#%FK&3<7WLJk(2-P&5B0;GKiztK^CS)%p@kriQg5sFitsr_
zxi;~8_S=%EYOJt7gOR6D)l6wtrtM7nZ-CL<_IA<CkLt+C$e~@6$!wf;pklQuFN}+O
z^=tYI$?(%WDq32x8B?vCoE&~~OwzFSnd=J-1%<xJHTaPA>%Bw6A%y$C@?@`iwAhXN
zl3Ia?c*u2UupY0ju1;~f3$2iIDfNoaKl|WkjUzjd<;Hs35ds3s<@<>1Q}|ErBz~^m
znunXCBW?b7XzA(G)V>&Ht9^c7e@I;V>Dgq3&~C#uYkThIt~K*^V6fd1_Ta@s`}%!0
z`+CN^+M^E|Tz*sh@(<iR?sogj5N4^7^gRveI9NnJb>RWsQ04ijI`q^`kqC_>#`Eu7
zkzQX7*c&XrvNMTbbzpEO9>LHO&-Zp+p*p!ptcnw~!>m6u<xDb*4Mn`%mS`Agpr#+z
zP8mKZjl%w7%hram>9b=r{G7dv*s?wkOPlBkiA)8#X!VjriWHm$hjyLe{TN#HMJw-Y
z-<ub*ue}}F($X^5(K8mbUK6(F=2Bxah$i+QItO#Bs*W;GFA5NLwV8p#SY0%`4mmjC
zK;ei_;|{<~KXuFa03kdb#t#L1;F<XvFT@_JqVNJI8Dqo8Lx>l*3-8O}C;Sihd`?A?
zbWcj#2%jui@Q+uIzaxjStj`{vBXX;ry@$~40bA$S-G0TwCGqWlLtHE&gp#vv=0_$W
zpg(r*r03`7!zJ1XNzrSry&TA*c4HcgWd>DM%<w~qR)V9m6<7tEntzC5SXtXSypyT2
z&Bejtllix-j*Bx4yXvc6YNIWaU`=~Cx#aCI*XG)T&Cq{Kn0si@{xAHuBtl@frn~TD
z=hvc?Q?<zBw;oRr>dxG}L!|Jln0i6eFJL$LgTJ)EJvY^Sf9YtZrH5+A$NLgM(T8Sk
zOYw%uq~TkEcNqSTSwW-2HtNwC%u-ig5}M_fC6HAgpz9XFP*=x2O5oqc7_poPA(Nmf
zaatziKkl$Fxx1RWT}iGnh~##;i@KOX=3-xf<p?=%tyL`gp=IKQkRQLpIug%ur{WL3
zAl*TID(B=uPpBqSg;?Lp)_dh>(BS3}mkkomg~Up6c!}Frb;qL5^5zDO4TkA$e0xKv
zPa0KRu_~y0zs;=#o8*|Tsk-WiouvN-{C;NbJF310Pd!ncmY_bYwm8NaeZxhYS7kn*
zqJEt$IX${yy|CFkOM0b(Qs@nKZpw6NNz16ilvzpyMEuD`L1=%Vdapo<jt5;hX6YrX
zOxjZoRVjbzdvh-aB<MJmpEfp&9-`|G2vdL8;d44gG@G$<A1%YyUW*RF+aS=_u6-M$
zRK-X9&Ut-ee+75sP64wS+HTNJ*O}8fccD?_#UuA~BOZJV7j%VM0g9Tvg0NXH(??T*
z^rn39ewkO^4YE^DZhD%AX<SO7pR42E${w*aX8WBl1ijKJ+dh71#4_Q#B-5Vicqw*q
zHX?jr=|Pb6@c<b!D|wVcALmU?$?B)9{)Nj!YpY}Q>1%t`Hie5j6d<VVQr%FuGrXO{
zX%GluzKq#)I$)w`G_-Dn)p^f^8B=aNPL|1rnE<csrQ7iE@%L!>*_D-8l3;?N^v8wV
zG_M-e<+Pa{e^OCViXwh@op43U)`N4g0REv+#{)$@`T1?3k+GVK%3PoR3$}MpY|+Fw
zUwVyy(_^{RM~>XokfU+cp}p!D!nb{0;y#WXS9j$Lksx1UI-`=*5jVQ~8Z&Ey*{B8J
z_Dh_dKkv6IYIVQQhwO9Yx05>6^5eaDe<YT*?C_k2_HKDXSv!1lmj2u$_Oy}>k$kcl
zWGt*Ra*mg44<qU*f~Lv6u||fPcTk#l8s(@u>ZWJaar6q#=IY}6nzk!PIaW8H@fySz
z;t@#(O!##;G`#M8;<Iz|f<kvHXK;60%%Q4i_S3|sHV<G{Lf#-+Xo$#d(a52mX|2)2
zy1P&8Sw4z{Svr985hsIhN(FlpqGV9h`&Q(`Xql*TC-}6%ZU^}ny+Q~Ww#Ly1BG0zd
zN@>Kzo^eTVdM;$M#oY68w(9X+$(ACVu5uA<Z^2V}BH-uf=Qt25^-!LbJY=*X&^La+
z{q0IIDB^;3B7Dy-UFMGZFKWoDhgSo=6s7~YnCbQRiq2&PKTu;t{BTJV?Rt5qVOvV}
z#F@bQ1v~mLZPgi%%c>7{B76j{lv?swnIY6}4-*R--)dV8KD$=u#$w%x&5S9h4}fue
z^eAf%HWciX)Zdd?=RjHTtIJ)^Ot|orpqD~VPINQqhLES(Fu7zp<5!s#+R{81mNMhI
zJ(u$k_Ke(^4@II_L+5>8h>22OZT(X*dk#BVc-~N^FjCrrOZ?=djdv@}W=n~rikcVw
z`KliLvoy?E#Y=T-uUxu6@;KSiS+y#B8JXv|Y`Ux*W;+>6;WwQc$GQF{Wu9=v^^h*H
zib2YmdzA-q-$Tb|N#m;VV?a^bOv6M<p~o7LT~_sIjaJ-<0_v`mCo8O@`wLX!o>eN!
zum*fN#A_D?7-tVa+I^(IE<z8200Lv|)_GqaEXg>+NZ`soq#bG{IBdQS!pijUenf8N
zgI#&yF3M@mYs(dyYxoG^?RGE;ldU8%&b{utGuWgO(W5Bl+GWHT)sA8U$KT*FV?FV|
zW+l2K^4g))YgQ|uedu9AW~gLowqhu2oG#*6f62^0=8*G};?e0FG=`XkW7dG9kW6LF
zq-Z!CQsuR8)&WXwcalQ1VS!iMy2o32c`sH8D{*Z1f<*m(D;AOsfZ2!d^Cm{3^OZBl
zbAOcO@X%=wGT3ZRZV(%b^>KOK!euenmrP2{<jWWd3q~i=25Wu}VHiQ*3haf@>#C(n
zVd)Ijl;qUAAt`FSwo%gIN`B$5PN28F=b<mB+<BSE=xIjz&GKLrf4<md%JO4{Rfdun
zoMqLS_VX^14?2C~Vq&ch<i-lg-_#(RzzpGH&F-vJC4FIMp6JAWiji`as`hj$@vBg`
z%#4KC2cers{#|X5qlLcn8xg4*2f3)wHQHHm%`IiK1JOF-iy=j5V!6b{&&#pWZ`00O
zN`^&afda;?yzGXIdI-msE(C7+UxgJ#j>=J6LWmJV_cZ#7K#dmt5#wi$U-4nP<L+e2
zyB%i}?q!w)PeD1iCT?z=d#$AK0lcB#<0~n0vme9Ig(n2CM71$g;Q8TvRt?D?OCdIR
zJpb*Kx=h3$!f|+iK8lsAKsYBV`jf@>S#osr#O3wH>HaCwXy~STC;M9XjEg&{Syu;5
zR?32A{tFHrKqIqX$r59%G)pgJGff40y?CDr*%Vf8fba^E58+22Xc2UAu@$|9=}w^G
z^LUJNeh6MVgm#4(w*zV|^5H2Rw2vW{*R7jZAjZT%FzF5&ML-T$L;HNkXMce*0<+jK
z(b$&09pU{#r{yTav1BU_3u<a(d?}<?7NpSx-?rR{bP6W|RCF4A<yFv_(H!xS9mUrd
z_9mZaXz43+E$OsExmv_@$;?z3QpU0`<Ksa1hO(_!RwdTtVY!Sdtow9^vG~Wl-v7j8
zTVWMb=8a+ilQ>s0-nC?TrQpbL6!?~38SPXQ#RtnBdfs=onB$Txdu{j{=orm<@7-fY
zHlu<_fg5exNugy_HT<p42WcK(mD2cJ!s)E@&6vx30R;bSU3yt0Yf_cnCCWEbaBlZl
z<MkDWoYwj5&CHu96rjfv6;1f3&vkU<pSa)}A1rxi%q=SFPbUkXsFu!mE3sJ@&EQ};
zVk(>QP0-H{uTAKdJKyuqK9P-smIL#pp8^_#D~;!TD}K3bn8op@ldXKF1H;?srTa>s
zu!<N~8u{b7z2C9y$KkE4lOfR_CVXQ0*$z5F42|IdRnB6M^{3{td__fgIDA|lJuso4
z!D@&yVI<oUEI?#!aYdaew-&Z6_)M-AnmuEW;WR3?kJ)sz<ZJMektwN%@7-s`PSXJ>
zL5W+9$3Vk{`gNrQ^^w?P)MVy%Jdt6u)vkmg2mdh@`vz3%Qu=)ccyv;UCCYzs_B5J>
z>)X5cd{5qII$ODZF(Vjy%EJYgfkB0pPqi}r`GS4zp2-p<-gxt&&fa?S$qxM(h$+Lw
zBRBB_0%Lf`nISy9hac}#Q_bEdFR@`f<HU*J+JG*~(y=~6pX@kZ&)TQI_waO{!T2If
zE&RbpL74LUhFNSO?o+sKHT4Max*Y8OChw4h|08!NBA4|ycPCp=`L`wJq01e<8&AyQ
zW3T(6!oG-St6gGZy(}Nl`(L?}s$eR8u5aoJlNWwnDHX?+gOzZuv101TH$yoz65H3x
zAks^O>WJUDEB<hbqF(SN3s`x?>>*!IDn*j7r!iH{TfLVC-f1>UHH9w8ln?;Ajk=IM
z=EEP`<u_)QnQ8}Oho-RioxFc>lLk&_u$#GAiux_rpP_>9`~v!&b9v1>LN!8d6cnDF
zExh2+H+ST&cohN@+xTE;G-}~zRv-RnLPh*u)8iMlW~nZe;P?QPVNa&BXUq`9-}$}1
zP7gTQOI~#A-M0-t>vKHO9ej!4+u+3kNxX-9Kla(bv!A>-`D0fTedL-vuLvP!scUK=
zGqSw!5=U`2`qT<J^fMEt6heOE*wr?TUZb)s(Ik$_)ZT`0>@DG|?nI(kcDL|-TLIHV
ztinfe^@vvgNE?2j9#Slnvm4@3<ABZ+t^aX<b9tDsj=ra5j692@YWsm(y)Bl!@x}_{
zPbq)3Q@-~Hmv?dm(%bc+mT)s=lK)~DY0uIAmtkb{Z^Ot38n(Xvh3TB+Q(!LJj44$v
zbg3GDduZ~Q{=f_~+1s7ri+=IRlpODZWj%#b4JCI?LB?g+l`yIN-RJ|}L|<Q8M$Wac
zm`a{ua)8{}AO7#!p0aXS=5N^q<pwM+6khS1Qz?(}V)&dNHNAj8OSPJzimPuUJMK~U
zI$faPunCj+cYcnW%FDD$bR4h$v!7!t#l9|F$m-tLG48WHLlAGsZ(A1g=~IZ^N^j1x
zRz3enflk<4i-~W*S+HRx+Mb+|St1p(&91xMy%{jhejTjiH>oLnob$t<@SKwT(kTlv
zY)oln?oRTlQ1Lh?J)PEppZJlfZ?(h5x<(P36oQ+80X62K3qkV}Rra!kdW-8vs{#X9
z6mlpH^j=o7vMg`6e}<T*4r2q03ah&Bk)wKUVIn7oEf`L=TXzr4G(~1?O7gY7?4IaH
z*>qfr)YhF~*yp%8C~$&)Leb^!{pZBgSs0CWIvrn|U3bqSIleNVj3*iJk%s{pYf~nq
zK{@+5?2@L|TqX|zfnD+}_P}8|;B)>846P4bd0^P}ca))7uOs7`R^>v(CvO@>jbc4f
zgsY$AJ_{0`i(M9ohBk(f<VYSiH&|MRU~a0XRLqo<jXTwaN*u)mnCzcyn+!7LcaURO
zycl>z_5j_Z*9}|soF$LN&4bVerC|XWA0I=P@u!9q-?q0syiQ)z!iE1y2pf;!;Ap^K
zJrZ->WH&~3*C7kChwQiZQo0;UlX|mX9X`b>tN9)P!!#Msbu>=CLg$!M&itX77R(fl
ze{gby|3Q%7zNH`W>SEJAnB-9B0K*E>g9kerrO}l-Mp}ig)AR8id!WhsD=i-}7)3Ru
zWg0y&Uxoi0HrC1kPr+XHWg}(n5FMd}k$h{I;^}ZOZ&+D2;)(e(=E+-Ulp=d81ZLTK
zNhpRTHlpX<HBTzxfb*xlqHnMhdp+e4W==YLVw^FAe`mzc>9C$F?sWc1+!sp<rC|8z
zsNl<u<=IUF<Jn&jE4@LDYY__tG<%6YNs-r2;c(N*#(f%8FFlJq#y@Xs`fM9rQ#{<{
zGnirDR34|h0vU$jQhcM-W`?w=DBZ3|RH)YDxq$YdxRvWfREM?RGVbdk%>~2acz2(x
zmMIc1pPz3>!_G=_-tK52p7o*aGoxxW@AY!AyeHLZ`d`A$#%X3<M$kI~J_dSVX}>#r
z^pbUtYa&t~XYQDES*{kj>8v8JY1c^^Gao$}80rAOtYI|kVf89odh&TuvC%&y?YCWt
z5+j)Cd&|;DA8bcId+NS<dl?0rA+yFr>2wK2bVu<(_k{2lA8;{*!rz@ROw3LXZxCNl
zI&nbz;5U~VI#d^TbKygKyEj)vD-<lex&A-=s`B!f`9sJ}|5d&8%a@1`AMDjwm%52A
ze|CI_0*cU!iKf@Y>jKDz+4G)>i?8>Izib18h(-y8z-nr09)L20n%ay}rye@-?Vly^
z6R3@<NFnWr0c@Y!JrDXFV70Zivo$KW?!DR5G|*DzCu+odOI<yv>>VK`4*l0@%x$f2
z-!a*I67nGD8!Y`|!zMrmz{7XIv&>Xh8nD=YgOcg5{7A9ZB5C4z^MY=>i;Ape#800-
zJt!E7PDuE^FxR;2gn)#Y_VgWUE^}gH9<yfo%R*8KKAeGJJTkJ7u{u}9N_jFTvl*e6
zl?qSr-;7sjscarIKDaLl5Y5{sMn_LO>*U5iUS=86*eDEPR$OZ~0PTp|&Kqc>bl*OD
z3MnKfSE$$CXv^&)vTg=MmW6-dmEy{04~VQWplz2<7|{3O?E>a8Nw9mL=0W%~5S4=F
zakV-SQi14bMKG9l=7!()E<mfx#9U~?DV)&l+G!ws^;g%Aj~t6fwsB$t!~}ukY&Fu6
zX;V+(s!3AwYH$*6?ckgC+XhO55NKGKc6x&KJhe797S(+zC{?L7@JygUF4WIW>Sd9(
z*34@y2KkT5O-Tl_%*e_kmFgtGa%qpuc6Em;8vPC`s<$Je0MYDsJvgv|?HWw~WsbCa
z?(;n`FA`lc*1&V5mR~?+N9HddA?P3Y$xovUgsr1>hq;6^8*Cixe9WqB!FxQQ(i-J#
z{HJAF+cp;XlCL<0lRzBijpi;Id#6AdMIwxRcv}Q;)WC$8U0x0aK2vLEcd@Y>7b38L
zO`9@cs2`&49PH?V4nrYIQumX;`ZVqyptbfa3U-jIQ4-0($cS3BMou6qcOALI#VBT^
zY~mYlJsD3WFHXF;`*l58WmU`<ELiJf>rprl1=oQ_)FX0OU^#qb;8~%A3kx0aaLd0A
z*a<upgbOt_HT9k33|!c?mBC?`mv3W;#h>4AcWl;#i8x)n1R1DiT}zr}lr!a|lbTZi
zx((f%UQ!8z!NS6FGTE?H2V4^9#8t&pSpM`^+w113r2#uJsabXbgwpa%7X%Ylh6zF6
zlnyZ0;uDToQ(G^`ulWwpttaV4>NPJN9iVh2T3|+ar{f_{l&3qz5h-<AUI4@X-xib^
z3gwrY0|uTC-lHx`vcqmZltH^1t{w?af>&Lo3&=)~^ti_DSa~GdIA%$UKzs5C7q(Q+
z2=~F3K7SJEzRY&k{$)F~aEu&%56oQ3%21Jp1I+MrPnIqL16MuL8BB$vfNqJakd@0y
zi;Egnk@`tfuZDHl6vXScV$Vs7@d6tiFtB%jAJ{Kb4-dljbLm;fkL6msV0>`yKj}AX
zOB8h2C8q`WhZ#zd7G~<D5eP-<5w1$jCrtr6G0hB^jQ;fsy)ADLN-JLx{1lm16djDd
zAulh#?D(-&D;j*-2-^5ocFLp19}sFY>kG_qkK@p$`gPWD<4-p|rN6m-fu?a7s9Ewg
z8j@yg@Eu#f+pGwPJd&SXlGav~tA<?}-K$|IAxS({*Sp?|zjL!3R__(APHKw7r4+{;
zizSDIgs@t5ZP$q2Fx`Mv<fwhx7Be@sm%r@Odb~hlX@f}yD-ie!*0{);%uX2XpET4G
zr&_yxRss#svj9j%ca-~cb+^CS75>wjZs?7%rl>Q&n<%cUrW9J=Lu|g*UhFW(9ZWQi
zJlB((s{KmCp5nn>{rXq_A?K0d1v4PJJunOkeSwT+N@gUXuuuNwyNtYvdeLesKv!=J
zD=sieq0@4f25G3|(Emx#qVpKDZi8Wdv=S-J@w{edF<r=-_p<hiFzO3vzKqbtalO5N
zCUi!aMBI-Bq1X&lZR>4|ENBn6JVkLd?+~mYESl!LD}!4Q%}rCSe7k!^yQ~y%x9<|x
z@r&qG!jY67^|Tk0+s*Fo$;uLq<9$jP6&VhHK4dybj)k5a`fjGyzpW6t<gFV8KmV$0
zOB_FLMaEW(LUVO$m%oZN2;xfi;35Eq)=jE9a&2~-gC!}~m#dG?R9n*3x$ne?b+*Mj
z%_Vuod2o}(^L-4A%YN{C5T*|eD6YsT!p<?&&hTYA0-I_pjlS6p(V}l_l3AWDqwy=2
z|FCm%bRzglO_GHl*0!PZj9P2P?D6>SKM8D`y5io9_7Idm=J!z`sP*<dWluQ{KOIXp
zj1V4+3H>2)th@1@V)tjU;K{KMO$mPlyo7(CPex(;HyD_wK~iEO%C}Eru=_1S7d?we
ztLwx%Nj@dAlvzV-){4vA>5lp-Eu1}@yU7TM=GuvfelfN;ZO<ZL)b^oMt)-hRBqB&Y
zWo`e`#wsqhekuZOa%0ku%IC@4LE3=RN;_c`Lzq&@VRK=&Y5~3a^xlK%xEeAO9vD$_
ztMqM#w6xoLLD|%h#^xtWx7itbk_lP9)faEPv04)zV*71>m#wy??TadevPOZa^>J0s
z*x9a(bKQz{*ot(Fle2|;9U@N-R@EuC&W?hdKWk^z^uwXN;+oUQ=W<Auvf_50C&{0x
zAbAoh{(85Y!)YsGqW^xs=!}BcGd-|egJbuHQo4e+5AjWelUYRHHzru97GNqa>+#Hx
z9G|TS;uZh0*86k~QQU6JJli7@vj}LFg7lD<+PuDe-5u12#Pmqw^8*El>y_p=;#3Di
z3o3UUcQp#EeN6oO$m|WezRJ5b4>Y}RyHsETWkj&zY-C+{d#bvmsW}Hddm`12ok)vj
zY;Dl$gnoI!o}Nv1Y9k>zCBE}x)VAVRv&{=FY@vXL>sR{Qdo#J1Pti!ZQtkc;+Yj((
z`*u@(Ddt4^aN+38z0i1cU@?c)rFC*}BzjK;IR!z9OD}ia`3=Wd@kvU(?VD13D`90~
z=5;(pu#ZDm4<<u%;|kpoV?HPM=TA;FO3>^xG&rokKPW3A^IxyTYH#-$Pm8SIc)eot
zkd7$+jMx-0B`u)W+5$+)GaQ3<{*sato9f8&1%G{!pRXWx3T=}^Q8202!tLdGr|tIs
zkr|a4BOdKLHT#BcRhA4dA(yz1RanhGUZ{~y66IVLS>MWd#f-o5Wrfm+>(tvLThmgB
zg4U=lw6=r#mg$MQ3}MA#lY~-fi55KKPod>Je%;fjvDsBfwsUgXGi#y4<;7n|&Zb5z
zpLzGCTZDf;sC1x9%^)>cYCOuYu8D?lM7j~~ivHHO+8$2)P2c)%pV#&5!}LZftvt=u
z0UspZzXNbTpi5uIH@;uzyD#$VATZyWYkwtXKd<1ziQ7G5`Pd($Wd*^A@_SX*28W}F
za{E5pPyGysoH)v=O##i{E0SkD<PyXhxD`Ed{zR^vv*Q=fWD^G|dLa4g`d$G1yz#;i
zv#<LZs{7OTp2@|0gJS!q*ovT-f6-`yhCxuLC+9pM44qsa&y)b`$8yb=2FAt<z<=#q
zB%O@)*})qKGq_h;;UO)CUmg^#2_L*s8tgnQp#3NPq|v;XCmMepWVWzKs~UA@|ENDU
z!2cr$(Q|QH28M-1tNJGJk%47p+>niOr~zp(AqN~_yAD>~4Woqz1zQM6WhDac>}Mn2
z$oOm#^xJ})!1;P;0hI0C-7#h4I&OHL1TH*?%!qgCkaHAgW|t%wC`%@2yW@wDdoE?j
znPT1lo=MZA{5-<Hkd_|)yFaLKgNb=i0crDHLqkLT;dlgX<ME6~SWAl-L}nKp4es^p
zNI!3|nh}iV=e1wO?oAddvH}G^CzS^EhzBP;Bi37y;W7kLA#zpWC!sC0pN(hLXc<Tw
zirEE;)*eD)bKb1nwdvy6TFa62aAPtG-VANS&&HW0#`ZotvmGCZ;{D7{z=FXLMqI}M
zKc^FL@4o+i4_TxSIJnAL>FL=ZRGdW=@uB&cXLj~|^Qr24*T){$eu!xJ*$y;2f_BTn
z)l3bL^S8i0>h(a{vvt9}uLlQccD>&hD$Rd-cgvo$RPgcfftvr(LyK`=D88`-jLC9W
zG`<*x5a$9RQOdN0Hg0WuOUq(?48NoJd)Z?<uiD-Ui;0~s_J-{93t;M5RfmL0&hLM?
zF}eVp!q*RhlVYw5YRA6xw*U3zw#Pch;3CA(&@iVRya&-ze164mrj@!q3eg^yeZ|>C
zGiTkM?U*@0zI;F6SXo=V9UAZdeZVn3h8FInuKD$G!~<QXO!O`PT+7qpmy#mgxfr~<
zX}|nstmU|K40L*><$d<mRNOgjUe+Yfzrf<jvj$FOOJU%pe2NQex7@8LQjD+%3i-yY
zWOpzq7M*%hP(Z8jAc}{Q1^|!7Y)|b%>R<{X?J48wTIY4?8chd4?>^rBp&*natkba%
zybWWknz3kK>C`%L7<I>afpR;i#RphC^U;~a3Ttge>2T_TF8Rso=P+eT_9FJqJMF%n
z2#14RU@3#!-Q8t6xn9pJ+JRcmj)nv8Hx*oBC4FIS8Ns63%6Di4)F5!2j@sZ(6Y%=X
zN=k&%R+zp_s=u~2;D7x)h{)^Ar&@1*<TJt1Gt6gbQ92jp8J6v_MZLZj{r0k}qojg}
z!>_u3OD5w3AGrJtmd^cQ(91{~BK<O`kW3>*?^)6oSY=N(pNmUke@aNmg|kzOp>O(v
zu%uP<w$&3BerM9IFK1Q_&bg`xVH@|k3~q`{EZe#uuwe#Ay*E)LuVjSTkE7}P7!48;
zmi8`6H`_?8!~p9W>dQ6hI0JE7jF5oJ{Xa0TT&&QbphzK{oSx3*WF>tA`2f4g*B26X
z63*=3T3ZoQQ&XLQ$?ooAl*0T^<K$M>%1Lji8EK0!V8ai)MyF!iPJA&|7%I>p0LS8W
z$nnOgg1|9zQEi$FrR28vhN7q!Sj`JDpxD`>DCxEkInB2Xf{%Cl$1o*m()7rD<K7&2
z{+8WJ1l~aA2QHFN-hGW4l;Os<sG0W1mX{Y1Lj=ao!Da(#h(P&>=;-JOY;560L_~7W
zYi)*gCLKn#PDUH95ET>@?kPL24@qh5&eS+s)vF%wN5V|sxmbFcVBWQ{-EjR97P8X&
z2?rM!4uljQPScLXj^5k*3LJ>@*scdfg>ebEjC+;CqK;Y4LzYj}_fBi;>TG5kVR_Q$
z3M!|SM5)vL5Q5Ngad6_5kEkxLt_b|M582*37KT23B5Yu`g8!E!(&vwW8HHphD9`$f
z+W#N&Wl-Vm06z;GtXjM%J9Mk~;<GGMt7J93s=&9LoSdL7NGgR*bt7X`*d`LTl}-Qy
zcs55fsmv}LU01~%(}>j8KBCgPYPdYo;6GdaZ2H!_+-&sQ=%}gyo0SEu#K6D+A-~fD
za@SuRvWXAnETYoA-H<+y!h%^HFZtg9WJPI-{y`oj2dCSc<16EVh+Us;<{}leZ{@Mn
zAfRc(ueXVXTv7}XkcWL~dEcV%g`^@P#p4J^!h;biHJwFMt6ndi(#|@P+RmPw4bsd3
z(_mpBZ&S8rM}bAzSt@x)N1xu-wA_=Zf;+bqtMl>@Zha`HSqyB{@gG0~kzw--V#}uJ
z;<bCze9sZM{rD_bZ}_N>DZXqAMWFsTjQB#Qe_p2N<gzZVInEA&st1RUWLrrNT|*kP
zcb@-B=El*vipsxyw;278;Vu)wFH~?9Q{&K!VHoe<U$}mUu%xu8&~^<-kS`FoVPRoS
z8-_@*2epLQ*m@cJW6oumS#*J4a>P0!)Zcs&^F_Ei!eGHnT%E8BOGiDhjMc(W4fplE
zOK0!K>Wo>nzI~#uWd8iUF#kcn=(+<iJ>LBY3V`LP|5aWO_q^)s?=>EpK=i{cylG00
z;Z^r9K{0FQmzs=EP&k-{+QaK9sLI=+dn5c|UAK^MV@y)86C|+bKSsrH+KAf|L2m(3
zBKy5_I+O>FTbe3s*U((Lcfjo0x}D51o9#?V`FZu`_0fct>AFyRDw5KJ=cc&%uS(!N
zE)Qdg_mdxS^yjKDs~l{P>Ui`(*s9XNll_eB9TtW~3=o5Lw0!k)=^kIoAIOi9j+&a9
zvbAa|5?((~7(xu6_u3<0Iy7_<jfb5B4s~ZVU~%YfgNFMAPHnh19mIdxc?N57=6yIO
zurK1f8patCj=cNshdFAiMe@85)$%6s_WX~#Q*GD2Dqkp`z6<&B;HwqIjW5&@nfIF6
zq|HqJy*QyO9PT5I3E_Z?rGM6Gt;TVih2^%#>h-n^5N2#XvX~EE<@{;_tJM_ik~0*}
ze9Y<S0jx1pcG$GGJ+4lTl?&8Wv<zb{k8q!R98Wo}e|LqtUJu*ntj++JcHp3|IgAp1
zE;%5AQke0&0bva{Uq%<$3+Ak2Nru3#iBP^fC3l*I<$2=C|1uK6mLdnmdgYU2?xLq6
zS|jBTl&hKR%j9)K%5wvHI4M%e*K<j(TZ(;j0Z3A^l2$NMn97BkYWD;_%)v#bgr9%^
z;7jt*{P}S#f)6cC{q>b=G-0?4!k#m2DrJDr)>KX2!c)+r$&ieg(i#|mrr?~_l|fTx
zOyl&RAu}(|FDrj14t)Q1?03+uRp~C!u)q=u-GQ&Tq{3O6Y7{>zz}cgGH3R`^5QD_<
z1Q=Xh8P3@e4J(y}{fy_yC4EkqKKzvk*L$SE&AD(<=AMT>a7dNYeo;a44vmaxyJ_`^
z{TQ!w%8BhKLmv$a^1^<{LuVavTaoFYNr{fP&aBSb`pYoRY|g>_+D9bYpnJJ04eVC_
zg5*F!Yl!HF@&QM@-wa`B(10FRf9h+Pd6bbkUfWxuAa+(OYT=Ik4qnJlu)U3bCoury
zY&dw@!Hz?@2dW<Ngax}!{UttwhT6<uu@3FPQJdT_nR}spk&iT4l|1VB3s}3sr8od2
zo*axP49&2_56!c^jX|Xky_yU>13&;sZqB)ZY+Y~<$X<b7xIpP2XI~V{Asl%}SJmbA
zhn)Wxcs2BZASU*IiGl|Q$$e(W179(M{_$ht64&E2%TJGY3ZIJrMEJZ+KjQ5XOD>3?
z_%9%0v$e<%AAJ9&g{h1jzWuHr+W#Fk{IVMSEtDY)4om#E7_bQ95&CTO4F9cr^$iAm
zdtm&oUxYsi+EvQ|rT<><Q0z&0^MG*vZ~6iM_xJa|>Rl+Cg0q|XZ>H?>>O_8fj61pp
zfqZjZDg=8vpLqRU`LknKQI6%HcVl1KQGT9Xf@AK<@1t~xxiL=V-+kcZX7uO)t@R7c
zXY#Eu_ScYs`k9Hi5_7y5b~3-|L%iJ$4fuhYOD!kcK?>fF?%E!~ot-nS{{^bmDij6x
zGo1X8j=bRwLJXbSqj+&>{Q=fre`z$0$MO_YJ_XDkEym4fvolzsEqh{l+Q$`CW;1T+
z8yv{w?0qdCQUwNboXM0iP@WagZ|*kkti3wv2{QTRf^F7*+|=w-FA>~HqxU~F1!aQQ
zq5B(#*TjONsC^4>X*uSMo$09#<-NpXwQ?r3>Yl<ZBle^Udo*~g5su%{<7!6FFCq60
z#LdlRTGcoHLC5oAF10b;HL1N)p&dM|%lu3h_K)OY;#$oE$iHI`=OH)z1?n-R)!sgj
z>aG85ayGn3`DU~7<Uo_1pF*2CS<JJy6|2YPupM5o)D~4^Ix#e(y>y7pj_)5#GcooD
zE1%eUxqdo-U|`MgQJb51`+^Xji#%Yygz9`}`^0I|3SSUO?Aa~)v0Bai2M%OiJ~}9`
zyPq~}(qBijUPa(8tO=(+n`G&*$s8}Uvor{Kpm7NE->`Po96r`=-ieWW<sEojG5iQy
z(f#&YrVY~lAD~tn82BCH^QCHWqX$I%1CL+Lh>M-xf@J>EvFZB$a*^-2cuQaNzjwbP
zP#Oxo_y=de95(;tmKl%y5=TcX735UJ(D?qUe=?7Z(UVaRVdV1R_^QD5VYCwG5BMo<
zuyq1UMVZVKSF*=9$8;s5RkwNG=j?k%r+O`{;v4*R5?z^=GsXtW+2|ybmq`m2{(-lM
z7fv)=v}J1}V{?D~Tgu!SK~f4V(;(@y!k|x3Lp+0Zv+R4Gf%K1?%jrnPK1{QID|&;`
z!E^oXlqUa8HiTp8!Mwbhq3L(hj{zsB$>Y_jR%jj%?g=Y%-%h6N54->DH@C3ZH$U4e
zx#{taB)7~g*c5zp51#`m$Jt)S6HgixqO;PfpS|Wy`F*xK7b$OdjOjhIU;TbSrObBh
zBNw&9z>EquY-R?1`iKXw<HyEaWX7C<v|cMg5B0s8`oy4xItoz@4^;>W&OK;&8Mv4}
z0}D_~*!sYjyqToz-66Zxk+0s6SVrLa_ao~-8;j0{csHwWt;@*pZ^#%Qzvn9M?~Tnc
za#I$;s8#@tq0I2sA?+8nC7&9{$?0dvNnc9nop}$}G?_ncs6Yj+c_If6*O5VE`Dswx
z!wjK>^Wvbd<sX~tDTG?bCy*)mySwTG`~#smQ_$h*b7EKiqN(L~2NG}3pQ1tcIVG3%
zMUUoYI1C*~YA<vxzOHEm{INQwasCfLD7iK$44p5$KyV%Mc#D)mSQ%{G(0;8VQH0M>
zw$;wwXT0z+3*2Sdb)V%cI;?0JAywsk()@P^bW1_~-;J2EO@6<|_LOJby91RVnG32d
z2iJF)e>JgsPSf~jkg$!o4%;hYdGcG=vB=ZM{S%QR3qe+JJ5P%;bu(bnB5$X)zck;!
z;S&)j{Wg%wEQ*Cdt-#VFA$-6Jj5FZ%++T1>oOzrlK!Jtgy*?n<q?oG1g;Inur;tg!
zj^tom_!Fk0Kf|M9f@Y$DD_i-?KjRM=IDLQM4{hPDf8h^&Ryr)t{{euA`n0t|`YqF0
zhU2=5ZMA=1yo|lLyv2vc;|Ik7iTKRa-naTEM$}9aGtUfe?AD<mD_wwRk)NCY!ezU@
znA4><9xU<1oQW8Bz7IB62)T8RH@=qDyE91+zX<C>f0^+&S=f%88X9-RCJG->x0sh7
z{#Rg%ypV9q4U~%$t@P7x=looUvVOD%O39x4(Qe3~KbjlOJCgkv?m%vN)@xnMHBI8)
z=8r7%Rw36zDwxw!s{KZJv2@cQa>s=TdA$c3Y@p+MV?qw@(m|)7x5{>%XVaGzJ@xGv
zle_AOhp1*e#Pe4Xi8i8}l;<k;cs9FMK7!YETIZS?CA!Ysy=xQfG2Ucdv9H<T`uED+
zS4q1;B&+g+-`gpyh70HXA)K+z`Dg4B`-hpm0fWbE&A|%GdRjy7Ekw`y<<p-&_DPIn
zC7Q^>X#Vly(fr{1x5%k){k1Lnlxr0mM(+fpVzpjL;q5L7b$)d!7RFS%KE<Sv9S`)}
zdGsLDitFV#L)z0@@K+i2e@Du?@uHBwkd}2m4wI2Z@JGtl>Pe?6R@Fa9;KtK;{n)lx
z-@J4BU4ZvfK^6j38MJ4IzIKkI<SGmDumzsabeX~<K;1XrD12!B^N{{D@Xtd!mj`tS
zAVC5;7ALfldvmlk6?A(ZVZFG#JS-R~=b(lQ2naA4$tMEIX{kVrwc)}KEq*o?F19_o
zN=#k(C%@NFFKF`la+~@b&0V*BVtr~;un&{5uuVhK?%|9>)=trC5dw#rAFcc3UdS+P
z^Uex13f8J-5FBhIl$3A*>8o#>H|TcG-3K~afKJc;js$}$AZH)kXwJYg6!d$a0)qY5
zw<|9AVJYG5a|Ba}<^An)_5bUymZn@HV*-<6EghP?KgQSb^m{COSV{^d>65qUU8?zw
z&{CX^Zq2om>5p-jVgqMskgI~aIbNY61@(~!jbLp$MBFmNQlvk=5laVP!qFi<o_<RL
zz59`H;2FKDvdviM5njISm4KteJC*!bP@ACUKO_8JZ&X-RNqPPHWqz4tK+pCtq+S+U
zTPLipnzVoi{McbV1&~yL?0h_n3=IYIzdK_5q<)5YlX8a)QtqIn0vCskG^3sbYdgCi
z+s_QiZd)hL9beVf*kr_%m25M#3QJ=R2We)3RVe`-A~f%A9aEFdS)M2c8xQaEh}mdi
zh=yeq;`H>i>7^1aEv@eUa*xS$Egy)}D70ct+F$9TVPcXNXb16w{+^yN5QVi5A3p{U
z=c#_yYkr>_efRhO2z=w$M{?k;{(gSp67BP>1=IwuJ#fe#_AsLg2n9iVvAm4o@$oru
z>dx|OUJmmuDA9yYtEqm_Y&y-oeap$OI3Syf{yipH;FImz7c=Vod)k@upK_D*Tl{gL
zSY^gG<EO<q48jL63WBG!-~*8f2nZm38MhDoei#sje_?*Ct}-42r}5JG;KPIF=w4>8
z+gQ6V9v+eyUbRxNRk$8G$vg}U$Tl`MAb(Jqi#R<!T^Be$kChstWy;2<;bO$T27Xvh
zOVO%3<7P!v{QR0=C_pF<MZt^RFI{Rck0!p(y(c*k$cQ?4y*ivn03ZlHbW+~pmQEOa
zY(UEcSK4b742)+0IJL{7<9AwpzE$Q6?cdY|F0ZfWM&8*Yed>bTu>+r@O4*F;OM53h
zD~2}l?dEvKd%TasQP^+ZykWWzlKUD;KzHNt>wNeMH+fbh@%TRP_nH|e0Uz|laLC}%
zub&HLLV|+Y0!P3Xp{lRMZbkm%6G!<5Wb3^}sl(d#FbdVmHV8;Kf^XkAh)fx)w4%5=
zU7<KUUA*G_I45ljk@<b@6_+>3=rRo{egNVxwZW-ZNm5qcDNi{EPDn@R&y-!G=xwZ1
z^WpMYEpKO(7CK!ax045vkx`1ht<QR%TQkyzUxCa-<%n;tTU8Rn7|>+C2W}_U&}5Is
zQ;@=V=SOIs{YLEh)y00_IuG5e7)mW%B&}+D5g#Os17in#5|SWbp70u&RE#xvG)QxI
zY=AHf{U4cfxyznz_-j)odd)~IuZm$nJ3<sZ9+f%-XDk5bl7MDSruIH)S*d!Qm(Dad
z!|S-8(oH)pVB8z)f-Iea_{;+BW~Wf_(`Sc*yta$}s7G{fD8r(XaDjgNMKjGJ9|4=D
zKQ!hFp#!Ah4Tqyj#dNqvHxxk;`!iHljS7RloAL{U%2pUS-K36w1^K0HQeGWoPll|$
z(>hf4h`5YqGh(&BAyc@exR4!ePCTmgOx9G4HHbp!nW}LFO&I~eZ}l#!Ka)nB2cvHW
zb8`3Bm%dyW^+*K7h+`henSC^O_^8uvD*>=*u>uep&v9Tatu1X8)xz$Q^HTW@c$NJ?
zGVSS+lvz5&D2Io#k1@SCnU>afmsrU|*c$?(%tC2VG07gMZ_y-6Y}suTROUA7L}J?}
z151S*17Q$GJM8W(WSdVLK`B5`c=wOxEnAFsRR=6*xlxp=^>mEZuCZ)P#q5FT7$!e5
zF|PKLud_AQV2jOvmIlFTGcw)xa?Fs94o^r31nEsVHkW&Gz{59h8V-cRwzo@ow6ubJ
zY|YJBU2Sq^6y*3i;e=W%#k`lS_MpK~=eU;I5AqmEd95jrFC~}Ut6a8MR<Z<-LH@5Y
zJFK$;(oe@xaFnIOwxw&Zn5Zxv^Bj^0&%1s)Z47#$L+3H)GjAs>)_G`msX9F=+)qR-
zV*406V~BZ7XEPHml>~s9BVKIKADa);L!PDCR8c0BVimOaTwh9UqhJQfrh9ti8jR=C
zWok9~Vqr9mu)iKpyQ&;}5HW1?3TngYFYqm))@U}x3Xk8J&q%CAmkEE2a2i3WSy(N@
zDdKIvc9u-xEHpSZLs`W5<$gy$?3x3m)E!27r}NFqyuzzJ?Zqm(u~>cT#a*Sl1FO~S
z8+&T$qn;q8Y&3S1bBm^k1_hC3jWPg@==%eM4(x+%kod^Ij}A>}jBdQYRl<t-=QEMS
z-I0D?_Y_TbiOF3m3g$Fo_En*Kmx>)@Hv{4Usy@QZ)56TI%b*v!+e+=DT^k+yGNV_}
zx9c8h?RNe71EX$2{_Bf<(=*8hLBt4%ICToJ)Ljy^)sdoyreQiA4y$sQ4&9fGZvZPP
z0{9n&(%zcDo%5D%s^|r5azMzBW)7_<P1QU*7fQ#gOYfu?)!j0?Po=~ORnF+bV_thA
z(qS)3*Ye93)_18mwu?Qr=~QuLJR1d_Eq;)-W_rhqzi2@gRy9!e2jjL1E-j~~$a{9l
zs52ySD}QFFa{cD2?idESfN|%b|E(U<Nx-S?4%RF1<dr+w@2^Hv!Gm^UTVIjd7jcJ~
z*dzpK`!_3ae@Vgx9&}&}U;v7K@8X^Eg-(Mzqy)}ot%ULr#h|;MmLfPF5!I!qiwXtV
z*`1O4zUbY>1|OC#V;1>szxm$>6n}r)B{?`J6EEkWM9b<l0oZicy1VO9Ticv$?qI{o
zm7Fyzs>@_LLnB&XpP+@E9jm+UWJqq&4-<G#;c=m;l>x>22l4=^-E>%5fL=(+Um=sx
z4fOpuK)~T)0_ZlYlp`&o*2?~5Xhk5)^7(LlUkV|f_<3Qs5D4%!HG(H1r+jg@!ejuB
z?6LjxBF1fn%+U|z@S%Nk8yk73Sj5R9xTvLUMP_dFjOq7(P)W<d{G{(Y2U%$#MSKYD
zf%Y-zPj(kPqU@vUbKui7e}`d73GQeqB_B!*-Ynqn=i*7`O@I{H%%#Y}JRY;5{yziV
z4M8MhB#hEska~kYsMX+&k08^p_Awq-`e|WDS3gWuH9P1l599KaAYK_X++B$hR)@2t
zal*HK1}3%YcRJpfS@r6@E(n>=ZtWt6tq_XOk`1n>3_HS4`eu;mBaS1oz(d;+Yje?d
z9-*&)7xjb19_d~J)WB!zoUTNd-%r;0ek**)G2J;>*sMU>rC?NTA%BsZyA=AJH5|%`
zV0{_^Jn+Hhi-$`yAKi@J(6!z&Qpo@A%kic3<dVwESXMSTfl63&lNKf#la&6J4s9S3
zi4gwO4U~Mq#7_#Y!`id=(-STnT_~*+T43X-`J*H2;(Gp71rmJ$F@{9#y|kVVw?_oP
zz7~;7<krJ&{PxFh{hJy$ZIBk)Qd)UE*)Mp<>LteE)4aLmnzPyHk}o9&8yx8xHcBQF
zYAqI8-hBFgY1f~_=5k;!`;P~%Mg-{rS2*Bi5~L|nA?KXO=xP?Z$9?S~Axl2R-mj#1
z_G_T-1Y#>R6d3<r<k;a`b8~&uNDKNU$Fi}^R6z232hmuZyxD2zJA&6nY?*x@eTwQ6
zf>*{f^hK`E2doZ?{H1yZDD+ewrb|Eq+O3Ga45daM<5wK$k5yX}w-KV0{ez^ON~3`S
zhwl#<4VvdeN8eD5prV}|!YE4qbmL59$9CXtl@nF*E}7HYJzD-LO0t>;GH-2&V3RmA
zg4@5wP=p0o?>{TR?3?|8SVXCNg|;eSAb4b~_BF`!ZN5AC6CAUJ7uFA(ncBfucg5Eo
zYQEfkmd`$Z>`KBFWASY-W1+vso!T8@M(I-7fTkdS2JiQyZzGI~8veHC98F}!G0T!|
z4wz<7J;sOzpyt~}r4l+9LB~g$<x3woDfC;)?xc@9#t;1uICBPTuK7!(N0+?hpJN$T
zH(u(Bj?|wLj_70ZYbOOTin^w~*ngZ&j2oyfH@@IC_oHMdkzio#ka8q++s6+X5=I!x
zKxKOi=?omk{IlqNb8qXPMDLSN8e_^}bv$a}_)PrBwKkfyX52ZfhBoX8)G1-j=-q-G
z@HeQ7T%t*D=!!k+0pa<C-G2eN`%)>+X{K2#J6k(!*3bS>Q;SvC$NJyCsMTAg$Is@&
zWG$!|@Hes`o<BDf#vB(F3)3jP4-5^&sg{e?L=(+FzG@)cpa0_Zy=UNl_lub%q1a+^
z6D>6je;r?)k#@7Wl6PZJeiU7{la&U<=iWb{w=+rHfk6}rhwEX;ZU#$tD{3E+$%LtW
z*}n4!L|WbZ{nRhT>)=||?fF1W<4zm(t!Tc;9x9qo!}Qa5U|NYfgpaAeM)4%ve8v>^
zr=;dXwEYA@9|m>gjfwnOgoo_vAxKkC(voTv#i%`QvTrJ@kD|NjPHfi%i(OzMexG+n
z)vCsi=QeX_cCAlnR>5ZOlq=_1->xSA4d>TL?3^7<!VptYQJkzg`a>4M>>|yC{8kd|
zXrqvNz-Szf<K!arIp_9~y2jm$wSk#%p9$dcsY$a-MtSwp1dHe&1;#X`k8gM9!bb!;
zLs-}Bf^k<U(@wP*Pbc?k#^b%30rH{DT0>RrYCQRT;A56;g|CmlbP4-8PXTPzvJS6@
zFAqjJdnT#)BqN$Ez46~r*7}{ih0G{>^b)C3Yb1%6!R`Hh*WOnvkywttLly`^5LV+Z
zLDEWfHIy{CTbxlD|DTl7R)d5O{xMd?DEjaG#Gni~7%L%VMF$vZN*=tINnVj1IJYw?
zL;Eo%dy<#J=qEUD@^6VVbphx5pG^DEdl$^5In$F~cc#x71Xh`~P5gzmX3E{UmpfQM
z$ZJSAK8h)JIN`UjAY!2?HRDji@1al*r#pzrBNiKceHV(SNwboazQEGk2-}{o5<#2W
z?>FgI<A#MKoFtr>7UTcF>aH`asbyUUP>M8>j)H)K2qJPIv>>3;tDyAWC85_)LQz3M
z4I*8TAZ!8Yz4xkY0g)a$h(PF_&~jIReYT#n&pr44xn=#y^W<5{%zQJm*33Ka`@Plg
zesvvD{*%5aU55Fs1<tY98M`)9`A%<ZOJ(k1N&XObCmt0(T=`6?jRwnu<z-~g;jZH!
z>e0N+V%59Gl0wPX8WXe?Qi;nZA{taCa0ga?5|z^EVC{U9GUspj!yF>+<`7H8f%`|S
zbP7LAva#M-h@lc9wz@jz>Lr+L4BlrD56n%4=C4M?No@GuKuU%U-t#T+#EQ8q9Z*mU
zuY}o3>4zMC%nr0s`U%^$;7(@x13zZD@F#xENOOuhWmaI>e^X+VJZea1+L7Uoslzb=
zK2`)eE%7yqY~qk-2x<SYk}_(Ua=QgNBmKXkJZci0%yC}w5$NRD9U_+(e6NcRuSgz#
z7q-7`*~&(Eb9Hk&sH)a;p0~bh_@1c<`NcwN9P)bB@Wc!WtVwNLIXIuoeE4N(T){P>
z+z!*`;_HXqBW;;rhUe?2?T()y+}UGYbGF^j-v2{~T&v~%GkZ|2pMK<L*tojbxR|1u
zf1gw!+>}U>(`@@IXBfo=eL&fS$2+WpfAyYlH8B;X9tb?&I}qE0OloFERG8U`d2bfi
zULwUQ+f;fc{TnJ}E1+CE4MSz;pB54M^TX@2fJX#)(^?H9A-l*c@$e?h=rJrMRLaFq
z$CV{47x_B02z^M0^2yL|k!(6+ue7k%B1!;%ulpchOGeZ6HOm&IiN8DHg(@`wwL^Rz
zsryc({`|{v;s`)x{pmjwAQy)IESJvQxc4h8oH&KPP;eB<tmvIjo|(oqvAb4Ih<3Tk
zJ!JbW)_W_G7HK-jh<dp>g}W1N4y3o?Z!1qv=R92d7$;Z?VJVUa;rTcCwOys7>+9-y
z-7n}|FPAPXdC?Dt;M>{!ct!Ht{DtxpLfAtdMmHw1*|6s}J?b3GR;8uqpO+dt$v)Q)
z>H9BW)9sr=n#!DJXIr-eE|U?ZS+q|eGcYmZm+zdj&~d0IfLQ@-_7hCG+hdwg*NggD
zJ2ol9gg^Q9^2E>f2#76up+6+n2@G`(e~l@lHNj+MF-(`3!v`(pN;=};;WhbJzf#hc
zvDM@Yxbhgp`Rx%HVm$q@7}Y^ywYI1J9g`2fUY7BsV5qAiv(<YVp?HG)%vD~07AET1
zN~G<lS>qwb#wJ3KqO2yH*{Wgu_cjLlv|9O=SgpnA={3Ca`bV>`4&Vg)FEDu|8@twN
zUm%SRoSKruNs^%x{N;6q3cY-f5{s6EQ-1h|+4=sKnDmEZz?WvZnSaEmvAtTK{=qO{
zV=eiG)u}erWW@qLi{TNbjQYDm6tE1S(f{L}7k%-u1Vw9G*RJ{egb94_a7eGvk%X%G
zoh-DJ;RyXLqBww-!T%O%c=#Trel%pFJ9ON<A6CYA>n2*Z`k#nVcH-QeKk=fTJw&bq
z?U^hdIWP|?<YQ_WZ^5@ERToAtyt<o}Y+{}xe{V$ERmd@e)hbW!j_m32b3hU)bclTA
z;1;7($iQ86XG<J*ISc;L7b2YPNzQf=d$!+t<RxPoF3cKM+XSgG;QX{03VQcK<AM7{
z9IZ}=KPkP4decNLVzoEWe{R@|G%e4i4N;_HcC@Q-Lq~ipWU)7Ldx&sqn`1qo&TP#l
zb@BTz{I=^kk9bS1>`@{V8XBX|RkB_&Jw<cuB}U7hVN8$*812B-F;?SSz$d{1;Jn8?
z=f28cGP?uDq5U3*;R+&1+&ZTOQaWsy<t&}vt!__VmMo%>$*y>6PT6r!u+he}iyr}T
zDV#W9F018Q3%zt>z};L&<Iuv@LqIr*XMk)#I;T!XNc)o$-QQ2rPBGQvqp|-}$oXGv
zYw!X4w<v#WOIkNTVh6GU|Dmf)S7!yp*8hX7nT6gUQD5TKb|<z|sjV#_wLKY<GX=Wj
zjzDs`4Ox8v2<GX~mB`~db=7X%Ap!EK>dCyuuTTDeb_H*Ap|gYyrT^P?ViR~iKTC+t
zohHxGwe`V@T{j%_;HIyB5dR@l>CMCt!F&7c)%xIGa@dPyq7aKOjN}C_rWXpg_JJkN
z+UdHaeNb!1V(6Jh3M_VqM)SQ;67C>i(EA2kH(%OhaA!TRLvQZoS7J~k{m$;uMflRE
zT)Ou8mfXI-lVxk6nqIB@C7%I}<~%1#*_Cat$<Imw6V7|yhk`ozL-JVoK^l!-OjN3q
zP{m@TSiUIDyxh^fd=3bf-2#c~83;X2DGek8f<C&VGJ@LO9WsG?B1KK~_E@Fn#?QXG
z=QokI<gs>6Ca1qf&}NGDIebRN9%B2$J&;m=|A(bsxcDmEn>&qSDO9~<AsMR30|ncK
zlZ0rrH9DEdS`IwXgVa*9v2EVA2t%Ih;QNU@V;(W>YyloyaK3_;7N(1|6{ydAHB93s
zM%5Os0(lv?lu)3)FS<DdLpZm+60p?wi#d|PbJ~tqniooY75(;qWbImMQ&PO$VgnAo
zgJ0BkEnlL#;eVNEuR0W56=Ux4Ftc{_Y4L>Tn4!?5IP?O*c+P#p(fKcQX8eEBY0BQ%
zUv)}+as3LaA(Gm&<#rB0LTC8q*m~RRS*j_nVRv<bZbZ=c0`b^=r2q;VvXecDG7%tn
zQ41hhxhWNb?xJp?<v-ZCN+>MuyE+`FQQ(hbc9>yLp}IC)mY@!3ar|FsbZs?mOqQ%G
zotgXRv{g{u0EbUf2M{T?or~~Mh=Azs=pG^W&3DE!KpsW(>FaOVUU3|J`LbU*v3Rc6
z^!ymV=30tsV4Be_Oj(w-si#ATA<LP;Shd+dcXGb|BvXXPX{8K}_mxjvOmAr4uE#GW
zPla$dT4!%=`W~K@q#I+V<Wng(rRg(#HvJxlY`I3>Z`MMcH!ZEkD?}p`1)NXf=Wed3
z;Uyh^DSAU!(2no&ohl659#YaMa|&5j+fruQ+3@89DmftX+#XfrXR$zsx|Qa)4aVIS
zU1x4UjPzy$e%;>*)!ARYnGo!)hH9YAd9vleFzrp$NKzZ*(sP=*h~b4cxsCbA>QQ;R
z2kUMneUd6?;onZ{Av7FOvbz@BgnlF7xHx!;Be+@MrWCjv(^=JBp!C{1&y)LCv(MKl
zp5NP8IXo^lXqf&!R7l;%ucbO3Q_S+fhS<^>Wy6yPC`Ng2p8WYyT@8c)Sy(+ScGUDV
zH3F1CNXFhE8~l`sF5k7ql}DL8Sj&Z%u$|*8=_h(#1w$zF<6w5e_sNs1AXYhc4+?Ef
zsvyz`85Nc{ERKX^Wcfka-)*X@ydS*a4l-M6tR3(AVso^z(b;K*`kdg68n0c!d1P_E
z+m*ZnF(x0>!O%_NMokE8rh62Ra-~3`^*Y%XgoJy_W_blgO#JH*T9eXK!%iTu0uu-X
zn$Vez-foe8Sy}SfpjQY=3;QZ(X_0=NGoxS`4zG++oy*MJuQ%<UQSs&#DRoar`Q1LE
zC{x8MIYHwQuak1KyRU+bNC2TE2;{{p@gd4Vf5rnL4c>G~Od8to@Y^@&K&@4CBiTN7
zYf$z-|0pp3s)Hm-`%%a?e5-f8o2RB0qCGtRHk+*j!!jLRkxl#YIufNmw!XrU$-w3x
z0_)P~$J9mHjXZTM1WX3Y?>r>{3R>Po?N75}M6`%d<jMw@a(3*Oz^>bT2EcyX#VN06
zJ!6@7hz`ihZZo&^y@(e)X5rJKQ0z8+w2ofjpj^3p6SdAyZT$cZfkof8%BVD`ekoW~
zmC~4Pg;+&js<~yJ*c#6Aeo}+7XIEck>nJz1Fn+R>W`g)UF;H~5=1}mZcfwFxT%ge9
zy#61)oZt_paV!H_V%k+|TW>wOGh$=s#8=ezngrVH^b57KUD~ZiYm}VQ7{xJ_=Pn&4
z5(oRt(J~k<@=9*zT2SruMU8b-+^(NjE~RozO#L*6fUbVkt0}jNx6sY$URcnX3s%UT
zYjQ#K;;>~NpKRXbzLB9ksYpakeVUtm&H<c%j3uQ&nmT(^iJ-&}0v&8_n|V<BKm}!-
z&8p)j;Jk=cGQdtj2M)YF;y>!9!FR1UxR}x9yBWl)bGh7C31+eb`a(7x!%T$bZ{JPG
zG-;{stu*HMJ(Ax@xjZmK*(2<?Ue(TXRf_27qB#(N`i_*{utuNT_d8P0)>x4_<WY65
zFB<S0pwcc3l#>nRNaVY|*YVfBn&(No1zjX9$y)gw+rc<D=lgr{!lWi<_zidDO^*Z~
z`(8r+__h4E-nyP&d$o(kG%)k$1&kHrQpaoJQ|?+)gYY9NFhMbn&3=CxDuB!FD~?7R
zBx=`cjgRIurB#b1E9x6E@If2YKQjXfJAGbpKPi)_Qr1BHHrNH>gPPs{p!+KB4cA@T
zS&IAjm!QZvi3ZHbiiv{DozLAad%G{~riCUTq(nd=BcH0|oUx*Z>Lv{O7*orSvokpo
z^tR4+`%%!!7PWZ$k1(*MYN9;?m%X*3y0OY$-;nf?Ul4WOeM*4NXta?rdR6mxk(`-I
zo8RDBQ(o5JU_`f0Iw(@(3U)vgBhNo$no70S79BrrpJdBmUC3&Xo{?@51$h)PrYEer
zbfe@IA}RA>g`o}%F~~#-GhBQ2cwv@HMMec%iE%?0iQe<hWdY$s9|-Al%B>oX63~wF
zQnpnaqsxbSM%y4Vv6i|S&b#Mw{N~AmecZzXT|HqbCWf==YfqYF(*sxXzOS@Jb#M@u
zpH!>Ip@V*`$iv%MCuOSetx@xc!5#iD`LgcB1gQBBG#gPu&|r&e9(gI|BV5+@<Ehh0
z-@s!@S<N5#JnYN_!21#kLqq!7)rHZp8vE+`!gbfex*;zSjidqD@wAt_(wNZIehk<r
zc>6~T%k%(8TN>dqEy(C3(3!P&D?MWCvW}G7)J;>zy)}c@`Os(F{bU>xeEcEeRZGbx
z0p`(*v{Q5&s*t`sOeyvo1=p8DEo_pjDPl+oQE_DH0MCtO!#LM*X8L*b*6Sy{5_RZL
zlwz$L+WExGx?b$mm=hY7P;5lfY1;ew>7$OZO#$N_h3X!S+U!6OBh^_Th-H~vsBMJs
z)#jly0*3SH*?l5nMYD-t->Y=CYfwuuzps)!u<v9(Qw^T6whq@zW=5=hlpuA^M=9%S
zC77Lsw~rEyv4XG9WK*Fk`e7)BqwIpUv^xgnxJ}W0t+dQ13bS(l=b#wFj+8W2?5x{n
zN!E5uK-v8gv$JQ)oq2%r%MIW8f_HbW^y!JCBl30QKWMiQ1WP29tL?K!>+UPcZBcJ8
zMpk@_+<P<zlYqc;({`#;_Y^SEKdw9Mi@YPNXey9=ssI~QHeX_UI8N3BMqn@QJ*bQ6
zlSCHD&Z?u_58N@y*L<F}w`mixQvl^TLj?4FWO=y@BXh4dKUzl94b?`H3~IY>5_O<(
zqS0V%v<*L8ehG2AJjlc@tyCd=;5YEX7UnhboZWI>RhIABQcaxxzPsuD$n~~Q4Q`_h
z*$QoAc~gOH!tV75sFod$^xkK4hX9<4!embCy;hjwvyho-&~b-5o+~+^7qqO<QG=x(
z3ZeRT<t}3;*2sB_O*P+dJ=Ky5?JEz`RCW~A{#~?VAitC{3oEk8VB6Kxpqp;D7*;b7
zAW*JCd-kmGEhsz}4c#7mv?WcKI=8#2{8;FT(EYcf4+KfheIM{j(&_S-Rg>IKCV(Pu
z0WF@enH;@K6R8XR9T|+#wawG%I4I#XDAqwm2w-w<mux*+v_%)%6#9_t`#7GTaQ95G
zn03rCKR;oWlFm2S87s}tzK{V_PPef(EPeJG-{hf2wCkcO_Q*}7FL~f7rwMW&@#Bk5
zo~0nm-)!Dbc^y7zb9;yC6BpHQA1u<$lcnk#ELXvZyXA=7V9hM$hSPD~r!85elBWc!
z_UwuRl7}C#ud4KnZm|%Q?|?aJy>WGOqi18w)d>n}&G4=WkW!9jmDU_(OzhXVcaTXD
zvdH*tt72SuNe_IVpmmVCp=Luf<pr~EvGsx{7Z(@be561W{h-_^#H&|BLM~%N12R7b
zLg8R=Prd_OzCjN9UEC*}SmYJ_SlDG~49ub-HhP6EXW4K8H-biZ%hiDIfh&>&e*^G%
zw>s`R5a<^3@jD#^QVqce6}bKvZ;3bXBuNEczky8=R}`XRl5$s9ox!VHVEGO{Fj(80
zQ2>)W0nD&7OsPLw{P4{)u(3%~Pn7)VXk689QNWangXm;wSJBeG%2}Cxw#VuMJDwZ6
zF=0U49DAutzIJwX4ZJTFwp`e^yU4<8N~qDpc^t8r#>Tt!baYIHJiNSGnl1&7J8Ka@
z8gN0V$(OdixbSvy#vy=;)ym+!@%I2eVkgCiEAaF2^A-d$``huN1q6ztz(2)-3rGon
h{Q2Lyh4)(>f$N23hHnCt=kdQ*l2ez>y=(UPKLF-!Cv*S+

literal 0
HcmV?d00001

diff --git a/figures/dlscratch_poolflatten.png b/figures/dlscratch_poolflatten.png
new file mode 100644
index 0000000000000000000000000000000000000000..d0a94d9c9c35441568c991c0b50a7a33177c5934
GIT binary patch
literal 72379
zcmc$G1yEhfnlA1EHtqy>4{iZAF2UUi5Zn_aNN{%z8k`W^-6aqp5ZqmZyT1iF_s+R9
zx86+EtEtyjsjwGo_gdZkclZBE6QZaf`2vXm2?7G*g|rk{83F=E7Xku`76A^pqDt^c
z3H%4?q%0{4Q94Yt1AKrn7m*i%fT)N<zB7UaJ|o&oX*oeaU^qYhgIqKtb%lU<%##L-
zsJiO^UVzudk|cyXf(MDfh>OF(gu{vp%XaJMOzVV8$Y+OCSTR-*dPE4;W`!8bhcjJt
zmsWg$Z7R%?04YI2?rhW_nrV&GzTBvv>)Lv}-)x#`fKD3PUmW(}IK(UBd=37qXvfKV
zQIh{vn2g#3g8vk$K|Fi1zb=ol!|DEYIk-9;68E3CoBaR2{fATu*;Ghyu=DC-i^ox`
zr_Q2stMNt8@?*EbB01FVWRagQEb@R3c-CdLTe=y=o6ZhDu+ns(bEcuXSsV<8f@TJT
zbDhYSF^<mY9yD5;=d-OJA8yer94csa3(J8|t(PB-EG<LnNxd2CXy>kxaTVvkBi7c|
zI)+uI3poWs;tr^jFUR)1it5mHx+A&m810Aw`X{@Nby*Y)=A(gP7U}3H)XE72s!C!p
zh#`4F_0s-)S1ma;HAlqn2B?LY#}%EKnOXT~5Bl%Ls-Qm=8e3Rcuuc(?l9Eo(%&-Af
zo#AZ$ua8D{niw%+TTa+$%Qc3s#&?@gtca~CwB<`)6Smwi!riHNVO|vZhR2Ny^;a*k
zLKj`g*<UtKqJwrYpE~jFGzK~P`UiO2i*FriizTDEdva;+7-6n8cg<tqo7MKT(PU52
z0D6Oap9VtIwWBWvY-cJoSC2x7UHtDmq@t=^GhVjom#u;}olppI!uz821xGdLJ=xqZ
z2!1zR5QR?nRO87(MA8!Cjn@PQ)0!JU1s;luFAa_wW1wY;he>C)umk59WT_fa!hM8|
zuW;m#%plo`Uc-ZS^}_A9uqs266&8^2@wd4&W~HBf-2*aK$4->+TGknT&+%U=2dzxr
zD7{;96YR7+>q<)~Ua62#QD^Qfm!Rh1`ikk!A`s~V6ChrT!UQ21PmHpoEe4MFdm0vS
zAl(Nwl$L`iS$Hm^nY)nfUkxC7zV)O`)ofP*;Z}lTxh5mUHRFD!x>mY66<HMGQRP!z
zo`aHG#NrXS?<3pVD0lDW#V(Z=?VU119?sJY%_xOU-=a2s)?=G!p((FNyJ<22*JJQm
z1>*To4B!MhmynSU%sfxbv4Cz#o(GSYLVLO|VAu0Mq-4EYP=9C2yxTh6vV)SbO_uz{
zs}hYP=;2d2b<2^>0=ZOVO|uzH75hGGmbb%v8aHuqbLaYdLH2qGA|)=iEUOnFC51j8
zWDqEZS)e6^gFA9<i>1-<$E<7-+De_XXhT$7qjbm?^U4LXUesuC_x?hvJ#8t3ZmLE=
zSJNnt7NlyS1}RRx&jjJf8Z5sRc)*Z#ihw*jE$Vr>AXbDG?jSB&4f}O4=+MrsZRxE+
zhE-;msj7C5+_qIA?V=ZQ+A}I|&WR1LN@tVaxt;h{qF-komek5pN$2sCIW175%%2+3
zAfDn4%x7qAKrkV{{sNN_YdBNcETFE3OutXsV&Z|@{-k!>b;D15vHEhfzZ=F4Om#ps
zyzr5(1{C9kKLEtfwpjb0hGN<J9BPbo`E+2i+6=M2V*Wf&;jRM%MM_0e&Is9;ZRV(l
zivfB#w3(0JRiHI9p`TDT-S<BU5bq(?7aHLdmIu3=qd&D04f@YJ#lQy)a(8Jg@PAR_
z-NVvAjq$(jJOyW|4ym;e3}_L`x-3Y6Ri}}`-*t9yQQM*sfMpe{1HF_J3?_wpUShE*
zNJbP6d-Um6l?ZGaJ)vmQy_EYX)~`7lGT34sIuMffz`4#ht-!SN#B;TvA}3c4P1d>c
z*bt#?$H;5N56P%0*`RS257-X=!E{VSN`RO@Psxt%-uC>3=eSY!PdZwxM%g1NRBX3F
z=)Eio=`3Oafx+?8&yyMHxgqWaqRCTmKdv_f7#DhtSK>!j(xOO5!MQfi_h6Kj0|Vfm
z5ZvE1$F>*rqFn=d`8dbuy-)b*CZvij2A$S$J@7z$41H5zX+EP}nE^2p-0@>H4foA=
zyfCu+Iud2ZF`dLnF~!kanahMU;6koUVyb5q94v&=Jca=R?MQc4#cA?ca-5*(1b<&B
z?xLPJ*Kwi3{eg_-+o0<G>E{tlQR3IaB!oov4eic~ELc1GU@RH8KO=D%0c=%gWWUMP
z!?oehQUi^cWGFJ{@3e`a9%g)?qnsNq-Y{$O#DnRINc^Ld4w38qSGD{Lr?WaG1Kd<Z
za>1cbZ3KGP4rpVpRm|e_%`g3RLOx>Xy;-simQ;$PGV_-|PE-5m8e@v<89p~ATUy2-
zc;Uy5CnHUq_SI9+_m&oJ8dCwm=1Y48tQ2QthEI!?RFbi0xzh#La;nkxruM4@)M$x$
zlle0yl({>;hr(eMq;vjyOZq5>(Jg#c=@L4{m5pD0uBAmO)z;wzwP`u8b7^IsM-jr+
z6d3lg1b>LMYg20Vw*G+V9Kv$P>q^xDWya5ZX#8Q*+0nzoMNEocy6WD2rom}#QdqNX
z*0%Ky<3mV7C6N@c%rX$>p2q*a56<<5b9h=&uEWO(9Mge!q&Zk<3d&kG9RH>yX8S$c
z9{~n-zN<of$>c}xoR0$q$|5VY-d8M1P}sMq=K7D<p=a-`tQZ8%Lbo9rqI#x<nJJWP
zWm4HfQE#AAJ>o}>90xVZ^aC0icnMw=vedm*eMNG|OJIP~5-7niX(VA)P}fSVLAJV!
zbY@uvp#(j(d`Y^SZHNQK43A3U9~Ooj%Fp|qX*SxYfIlc}(Wm6~%IMYU2|~T41^V!T
zpoe0e0x=e2ze@_OM;ea<Mq*-OO4H`{k#^jVzCQT(@88?MXJ)EH2gS5&a)Chjw8Fo4
z)o@Fi&R#EhEt$;6_we9pBiS<Pbd{v=+Loq&Iu^UhwkxyUR;}a2ag&+RJD1p~LPH)~
zfW$b-$xT<w=YOpghfBqGA6#?#b$PX2B)R66s}p1W(0p78fu{|QqNxN1j@hrTpHB;^
z_AN5V=d>wAg~pE3CP=^gd(|a6-f;$HJDAtU=qbt_xgwu7^YmXg2uarY2_R;^_;V2R
z_ovN|z~e#4|CAdV{RQW0Hm0FcHqK|4cTr}g7p%WT%f<OAbioh@vaaAHk&0%EeN`w?
zTlk!VkyN(ouL1LkY(y)fGfc7+x@SHVK)R;f`?SXG5J5A~$a+BHF7#*6iO)Rc7q%$%
z=%5XhH08fCKya?<)0IEP0{?jdc>b04a}CeeV-laYyXVXQqucR|IM?}ry+k3c%^vd8
zr%xw1p4%rnn|*^vNhgU$v^x2WY3Z_qe<fxam|!sP^Rk}nRA;x;`^$VpF&m87AU^Za
z)00ey*(P+lpJ`z8avII;WL+eM-E<wt)Z^+<OjApX%x<w6$f6b+Jc80?|H{SCkZ|dN
zY%R5ZlZTJzB<&@+JMvSba-A*@7Q__oSuH0E$$($u#Zcf4Jdp6XDk2#IKZFkFrKb}!
zGBN^{pPrq(Z4anS;Gtn;Vlo0^9JoCyD#{u46sYZn<$pgHBkxK?hsuo8sW6^>7fy04
zcyqb!uaJ4!y=^gAu{ZDOPgS^qOD`?Z`7zR{%l8;>teECTCXYp8MBmSIDuY7A3mc?#
zfzaa;)w*51iKFTW1cC+WQy;v|8ka)giR|ezH+*v>`Ht-)_2%#+{@|o_N9JBZPR@eq
zkL{l*`$46h*98?PBeXtzD0P!jqPN3)AoU#5&2PFdqSQWDI)VZ@**hwf(tmoaIKa@7
zDdt9s`fM3ox!k>Ww$caX5$|QGl!2}suGP~{5ZTJx(fO0X(ik8I8wQ+GfvA3-zz-q8
zm4C9v<m|F`lOKrjr2Nog!o_fsVe==LJ~3Ks<J*F$Yjc{*qZh~9T=vAaJ+wx-vj3xG
ziR6P8u>)Zf^;uWF82$KfL1L2Iven+2h~%>#N!PPMbmzz121h8d!yZ)XQ3jAk#ZE!Z
zk<ofgFzwuPgu_V%M8~Brd%eT&y}*k|$mnn_n6pzwdBMjO6q5gyuZe(g1dgU=FXraE
z&<~;;_|DH~opml@FL@_VK7)P?RCoD2^f)*zI)6Jsikov`TmfCYNeSn0F*EE*zOZLf
zLNJqPFjOz+q30s>&*R6*wg7pjeV&5%G$)qK8Z%-3ZHVz*)WLI&{`Xl}Y+Xn&E&rc)
zq9s7HbtNM~L};l+Je^Dvi@y6|)mzSJ?e|gnp5#K$t8j~sHKxAB0~56e+sEqktGz4t
zEIES@()n%|6mGfunp^WvH&zM~Nccp<7z^QWiw*y&wZK^Xj@-8%a-U}*sS*4x%T$)V
zeqT31vh*0j$I*u1?bIO84b5dGS$Q=8`{`e!<?~}%^9vu6^f?r*!FTD?M+>%qngA^%
z3K?hGVO&Y>^a)iW#k^`4f?H$%<{rJHihP`!b5hs9T~|$sRZk+_3oO0HDUlA<4`VRz
zZHa%5ijKEbxDBxhsK4zFAePJ4ATKpORabG|(FAQeVuC=XEKey>Kg5PaO#WKAO`m(c
z-E;`HAj8Wjp!vtH?uT5Wd5zFt`Edg(RHx8t9$%z2GsRwR0(swuzKG`U{oRy1havny
zPTU4EHjUXAZJ{|C@T?0Zus1A%aW~FngBXRn9CYh>eC}}EzF0g4LBw+8DR?_B=$i(0
z0aZo%Vk1CShV}dktoT~(6=s;#)|w=f4Jx~@{W~rPhH@<{-oMkSvRBq)dCn0j_aiVv
zXOd-VRJHiOG*>r09G$q8URx|E30`)+C|Zr^;=nO1_)tCmfJ$K%|7|r|Rb!#c&nG#z
z(YV}CIvyOrT2?Y`_CUuMFa7zgG#K32@w9wzNbTb4PuHDi<f|ju^nM5|ZV=*M2nj~=
z_!MpqLJTsH@>hUn?u}$FOcKzgjUu-GHbxQ=M+xU}Ec02V`zgYWaF1_M8JzdFeJ}mp
zLxTUH`s_4G;$RNhO%ok3Wc;lyb#EO*jOdAmUZ`aMskx(|3HiMrh<VDTqdy<9_fR{8
zZEdULzAWUO4a=mP%y(IVCQ{m|!7MNvM?)#ZQ^@){IP5#{T-Zj$677fs%k0x`qJ;WI
zLKMsjbzsn4%1gm&9fsZbc!SHO@Gh9#{m{G+oj9g}Dq~HIf{eXwJCZI`)i!Jqy6KB4
zRq8mj8q4AGj}C|;OWiV5`k8I++;f7b1w<90sT|gWnE#9XakC>Ku(M%2FK9>OBT>fs
zOZs@MZ?ULHlF_2h>sH3jkzX+wXpT_mIDU17A)1LS0cGo>8ho6>BBfi6^nHr8ZmF_`
z+U;WtX?^&R;4}XxkizaN!4<(eV8SV&9#V<e#B$L6Gd78Bp@d8O#YMYCn-s0%gs~X5
zCw40CzNkQVd<h-i{7;ECSz8(t_&6!*sI%P(A_W0_Aom1V0dn#IhwvZ2S7xu_f23$4
zKg`jr&T-{b9GAW#>RQIz8p@g4__p)~wWns{^mtDX8>uG4hqr5&!Ehk49hF)MBk~N$
z?2(>gk2a|Cb;C%jsBFcm!m_#?@daV2_?}f-82)eB#F#yN6Rj~E*p0@AB6LSDs@@Ty
zPf#W1M*E$mw02+o?Vr}&GW%RyP>k_29l@Iz<6@y&^p0*qEosVRu)(mX)T6;rKMVaw
z-VL&UAdOUws(1^yJf3a}BNCPg>%^!;w7cXSZ_W<;r#WYBgLg@^VHn)gP>he*JAOBM
ze#s{Y6Wo8)J>OA!Osq6iK?ck&zxHvs(slNr1o@NBb~K08d|}m-MB5GD05R$K*j*6>
zx9e6PS=tO~jx1}%m7Gqbt8#UUmG541hm9y9o*KqI^=9_3RQj1<o?ai;edTaphqX<N
zX}nCEH(?Azm;^aSpXA{91p##(w|?N>O~Y#p(1zwy%*A|CB3tja4A*kP_(E?-lu-e}
z-;%SjHD(;%>~=OqF`_n)Ew2YYqQU53F^MHm`QrY7_W|59jXk0JEtl(FiGpmI^v6LJ
zTz;H}Nrz8k`A|vLM&e8drTb7m)uy|C&-dzq0GBKE)^C~@dO;@N_aDzB#l@+@fx!9r
z{B$FZT5#*QgifJ;`=fecvl5PNLf8%qq)mo$;Zld1id%S?(8w5`MNa2nLDaFX3WvoE
z9;fv=zqJ^#BOI~^3%mq#XYPou;jzLuD%7aUZ{4NvV??wIKyU$g1}pSALRovuzbYcX
zIGxV2UbDq(wnh<RiV?8cZo@eASoztNyTjk>(v;m|1zA)|ic|P=Md#HJCi0SAfdjWt
zf}7&*OI<!!qT|9(Uib5c1$`ZqyTMkP+-~rU)Xcrf5~xM%l5U7gRU?MBJ=X$HiJsjb
zk|i?6Qw1e+CGc)|XtcpC1xrHjT_Lo?`!#A^+;ehbHa7`Lp49CR^x-j{$l)1t_yNpi
zeY1XEuH129NT8|rh8vr>w3qjIAB#sIl)|~WB6ZNtGJ_v#eyhBZAiPD_IKcWO=7oy=
z#5B79NeO%ym*-6hs1t+(NE_wb2L!iP)<}KvEPhHsk>8aXE_KG>)&7S2W{KA6%fxZ4
zhL5~MbI!mb<9q|M+EVN2`Q*oHb!p>I)lZ@KpEGmX!m=yHP7|V3ejH)=4bh{A1kGkc
zA)GwTP2-zj>1oAkyl?n+WXi46idSy>KIX1tX6t##NA6fig7OHI<O9e59wdDOcLQ{Z
z6OF>!`1LOyk0XQw7@GEau;443Y4jJVufQ#{kA5~d8sog)E52vR=lAkBQLLH2rcW=)
z(%SHS-W7sy(;>nzYJPqb^LXV{6QqUX%k%_)5)#xlFEn6u%urvqnQr}R$X2n%BCNxe
zgi{-K_DaK~QRo~VM#aIILC3-*{!y&M<`m6u`|cr2x76?WS9fIL`rCkbn(UU$f+U)O
zL^@>ituaUBnD-%0onvDdPwb`aWML_bSQFxwQfW1|a&vW|*zvNvyxeSNL$idCkkA<K
z4=3N??Y3veApMtqyEu0>sr`3-EyVi5(uS^*o~Wawx}^~U+qU09edduT?|V%p+o|~Z
zlTqpw-~W2wPSy03xWqrx#!_2&Pv0b1$%dcd>!czCmZF&mAeyt&KJfx;$Di`Q!75YK
z1nSZNP&wojm-!_ttKuTL&7fThDNC2ZfKEJd8d17pk+Qg3m{gNO8jpqjmi-L7>B?_i
z?L{CIKRwO1u?COJ12S43JWi|eXq|X@m|z&O8G})+s_JV0l9CdD(S1U<EiEk_Fs$zn
zXZ#vncC!pSe4jcRf0p<6+8Q)JKmXJ6xY$VtaLs?p{|l!2pF*$=qFt#gV!iN#p`_(T
z8C#WHSO06Gu0Yr|wp@)jX9+R};TbJ(x21xTzIXKG*^Ej7s~rIgnr-%cUTIvACV`jY
zZ9-i~Yn?7U%jf7U*q-T>zzQc}`JzDqz)+Zf$53xYL}7fP?i?Ox@rf252-Ocy1HT@<
zyK{4c(RCUu`<@MSyi}C-jDZWDtxa;ch_HBe+%5R5a1VxqVw5Dn2YIgI-58R972As{
zIBp+3+hcHV`X3};UQ?5x!!GSSDez|H$IhTDI!RdY&n*PR5%|87(!vL6=<!lrqU(r_
zDYK4=VhtlZyC|6i`T-`bmytPBM!)C1Ak10P+f#HB3&*~3(N7gd1lmR@4K}>2ys4dT
zF-QA{EWiYw&5Lm2S((wgt6hIN?jkH!eDmn%DRiAqdW)4*$K2ixTM%8=^D-R&g!k)o
zGPPimG%z>HtE&8}oGcmuP6y^0$9Kgqhlpc`ebL@qOfKJkqZ{GuV}&hAdy71WceVdH
z@F1soWm5?Ae(!q+PDek{d(I}~uO4p<9Z1hUv<2G>dQp+{Srz9q_6oBQjTRni$|avs
zn6VtSpHK;B)eO1RUtS}5Qrom|Y*n1R+0Nu!<$5o_9pn(CS~%muCvaiV6JbTjwi6b0
zw4Fq@{UGghrteAwiFj8U;#i-&rzyNgG=M_jx5z}_hFXh|Ut2h4iU7yW)fac~#Oi#<
zb=CcZjt5ZxY+nO<*kWb3Yh$0}HMzXWZ=7xYLs73cXcDCF-;$c4-a{M7yRXbOiXR<Z
zVN)b0SOh`dhplZk?`UpdQ`hY!lf9qBATfH3|IVXg4yPV&fi{H^U}zSPYz7;aV_YSo
z*=R!c8RxjK$j$TThe-#RN^faib@I;P!_L{bJyxr$sA*Rwu&uXW46$X6Ep5bw%-L*k
zdInu{AOJr8C{F4B!a0k5tqMTO5^cFDZ-@Q1=t{!kOoGSSJ*+JNh054L{>uyR*8p7r
zotOkG-3&=w{(B$9&gJ_(ta?%p)Lva$^U&U$Iqql(>`uGK0Mk@K5To#^HnS#HrG$Rb
z1A=6ku=>^ZYjX0g?msMX_x0*_e%0qt=&$j$3P#Sd>=f%v2fk69HH9UQ?HsC83ooI6
zf1_IL>o9#Yn&&}`C3oKbxqY|<nom2pk(;ZGH?S|fUNQ15oJZ0C?aK+K72Q(2F_POP
z_<)m?d<S`Q^6VeaJFOJ__%nCi{~;C)$8Cw&@0gkb6H^{nisCQ&dM9$lWpeyM!(sds
zBEPBT^<$p7awhXzEvN)5oY=2lbIk22f|qtw6yAn+W`0DAN;pWaeapG&sf9{?!co?<
zcHP#BkhEJ{s9TA^4+{p*_)Mt-&$dnqKJvC(%UAs%T5(x|VElJ}yF@l{N=xOx!7qGD
zk#T4jY6MwI`|PyI#vi$f9v7SARBLP^snzONaX?D%pZQPPIVy7c->}*1C1=P_xFr&r
z+1JQrGMrdX@eKh+ZcHC4m%2sNOBoEmde>pPY&6wv7owR{DHR9bx#c%XY%DZb03w8c
zT26y2^#nvSgyZ(`F?UEAqbQZo1Auz0KO5eaCg5>+k<34S4xYR?jQlQrf-!OcKXA@Y
ztMo&z6w^}*04hW5zoRO_MRvU_x0`fSpQ*{oZT@<_qQ7Gj+8_*mt9)AJ1}1m3-+_q<
z*2+tnNgwPt4SiHOOEusVtfz-^(SqpNVMa*xIzHv8u(^Eja!zZ&kqw%ZFhaEjI90Bv
zNd+49!1pp|B7u~wHL>z9OzOg2;_Q)ZM$Sg_;k$a37(o(P5tqMG&@)(W)Wjm=@-7I@
z6v+{w50$vqO5gzEEsws>c{01q_%vHDSn>&~A;EE1udvsjiLlb^;@)pDch6sEGj`c|
zXynxWa$$VPW@?OhgvQN$E6@aVS036kmxpAQ#*G%f(I~VYJp7ptYf62juk3>w?YKMC
zCQ5M>d3f#5pn-85>^aoVt3$nG=SKmQ$M<~i7o*}c0bSVUIW%eKQaW1S;sYW?B9+aT
zLRg&_rB(T&{nrV2!nZI;)foXb$W7VJ0YYM$V3}yNVJ2ZQu^5A!9(~5NO4vm<-I}mq
zsk!G|V&MJ%1#ny$+0dqQVw2mJ^i6ve`v?_R4BcD7=>*4OGk!ucJ(Zs5PMs^CV*r1}
z1jSrDBbxYq$Bxb~lrqxiFa4W88njbkjpVMEaupsa8JT)%GPG9ZOLWHFCoj6-W89W@
zhdK-`zw4bNxq;|j?E1v(`SWY<7GBi_hObIkz~(^M-K5+43-7eHQ!IM7u;AcozYiq9
zp3N8ie`Wc1<RZ7f6~&+*p_yH3LI3(*Tw(ZHZ+g>g4Ce-&lKl6)#E*wls7|$3p8ZxL
z_>a$M1cd|BggTa*{4Z7And}YS_~G)`B7t7<Omhnu8dcZpGqwyXOKK*<oWK(1DX0Pb
zbWWFke8%4`M0gDdaY9a?dieQqg8Dqvji;b37IBn@P1Dde^NZ#WibapdxAK~mkX#1l
zDokDj%x-h(lb+%aOYeMzBY)c6`C-apKH`n(cJRN50XrO_Mn1qpx;ek4HbdiJ;+n`&
z1gR#Ldi?<eri#cqn0aE%>mQ0jwW$M?Du=hO-cK67a>1b3TTHmxCQ!Jc3Gi}Kvx!Pf
zIFWS$QYVt<iR#?l-wG=+vL7xA;LUvz;pWuu;sL={{N;|P$nm#cD6K9`FYER~KYf@q
zZmw{cch7nRpOu+r%9M*!T#%K{(?JE@$z{p%Mk$TLZSyxWGs<vv7G%bYD(C0b!FD-q
zmF5<O1!x;hCv(wuTv32xCZ7|eKo&A65if3s-Svpv{hwhxZWkm_6pL|b>RTt--~!T$
z_xjknL_7#p@0TPreFetEi35C2tynAPWDJ+tUBkcm`MTpR7?-El`1kQ}ePHxL9SB&y
zXQ*)P_{b<z@uTMq6j%P`9h)Y{qA0JjQ2aJ4l34nPNpJMb^i>ot35-9P1(Y5XdFFg~
z%*HR)^C<pu=l#f#!M>HMSh~y|KxO0n_lZ;>>HG5=43be(MMgXAm=GBMc~psoOm35R
zYRC;tCc_5xF9pqPyqB>4TkJrikmjG4M`dkevym9?j+nVXqFKoTS;O~AtmGum&`tv)
zU)g*gb@*}r6D$c$%=XQTz{vdwGb*Ny%)zu`jk3<gMUn)1wRjzNsefXRlzMLN(1^J(
z3O{-Z4z{EPkBcnZbUs?m@~nB$Ts<w}pa5{UP#bjFDXzoHiMT7VaK!Bw`gO;`$eYiX
zb|p78N;#HV)45zWtk54nel#>Uhkr&RAP^!-eJ=TRIVCl<OqB7hXHaNR?AiDJ?S;jZ
zcOF%gb)cQp_fX7@Irl!Y7wpx;+Un5EU#$-8t&S<~Huf*80_Rw6SB_J1n$y{yLB}xf
zH~yz3H!mDANj11DHcMF~A=>Ba%9ZjR$qVs=`Gyyk-gLJqsi_o66>N_9_#K+3Qn_|0
z^>10k_P&|A`G-XU(en>r0#s3`N>BF_qkwL4aS;IpB|cNBm0Ua&Z~Q^|;x<<VIypI6
zs)Gk)ze0;U#to-ii3_MKVgI4604)=R(KR+UMlYeuHZjzzX9e<8D&b=5Y!k)anHfV{
zm>2uK6X=(9>LuF!R%U?$ZYNSAxNy2sOHWwo>80RwY5!m-4*P^`RmwMXp%L-}I|^_6
zyKgmNK$!RjfCvbbF+@0BWilvcawA(&0jDp~D=Jilu^6v8p1cV_$;AZTL<B~!yAXn0
zEJ>g-fCB+gWP#%B2bpyF(^+YQAZ8^NiV3CmkN&fR0?lk;1fSrD0OG+qG;7WI$KAzP
zd7_Mn#3Uq<(LWVoU|}5+vaP$-LMxM|mmD_+ItB)iHa9m}IXE}y#dAO4vgnCwzQeP&
zwtmuSF78s!H1p-~ca(NW0Ben#+-a8mmrg*lxn4Y-_&BWHm0xJoOj>Ht7OBl?)j2z>
zZA&)BTp<%fmfqzRb-%p5+^-h;grR(7>2rN#!7!qtqCh8^0i9H~?jtM0yKFq*gG}eM
zE%$8iiK*SF`!TYXIz+GjEqGNy;RFXjW2>B%)+v=w(a{nB46VI?*a7QuYzwd>Q8fZ8
zrgMbLq;4_FCqUsjo6ZvP1l|;tNPsrNviOg2$Jkp>R~Loa$LL)q^jlYOd(U4WI;=9p
zDi_d<o_1L~BUD^Zm7sPjBbrybj@Y3(Xwt+15F>3zdsm;fC!5J-_$NgQ^eK+kPD*$-
zFrj)_l1n1}mqvTG7e6AFGPqUH5fhvLLsl9i1_t#@>L+1(e~*%`aabC3P*PUmaQXi*
z`A6F8qyfL|KSX8zt>wqtqLU*%mBMpQiF1q;@oM{oCyb4WXzH)<AVjI(j^UffET&=V
zkk5-~m2&mJSD{{$OUKt_%NNv@#qv#2Vot_pj0_!uGunv^c&FV>TCjO8vZU1qD)cv$
z__zKWZp&@?K-aWCp+l>OXFrPTg1}sMgTMC}5l<LT{&yW>-f_?Bzvk$rkE6T|#m3t%
z-nx9{h=ZGNz?s7lilbz0cexn&+YUg|29#`PqjqtYZq;o0D+&V|loC^_Iy&&*WrnTH
z<o}d~!@Na_;&|3Gl_Gn6!bqf4LcF-e4(-0a%$(21vV1J&*>!{XwV<8#IL1-*#-xJr
z(*P&3wA#cdM?ENa`#XPV*P7h(yo#@4Th6u`Q_XIRDKnsn=T^~2QkifPo{y$q<Pk7|
zeD5wPyMM@9De;{iRoZ1jPyiuV`m0Lg5f7kh0A<4(==OLdvaKm2y;g;*#^zUVzz$D|
zFiTYPa~^zXhU=z*vKrrr5-ocC_b)~Bt-tbGP()wyQ~{YerjFswyz02pV_-A86R3aW
z1eNT|ubVGk{4?v7EAozE|Cb6v`OwW1NjfFP__M>I3NTvCLdg^6vM<|h-)ViP4w!Iy
z`!C#kuDL&4OYfRo{Jp5dLJkXGNkd*U)#`*2!qFXaTO?_)?GQtDwF$BFE+>>Q^HQja
z5Sdi8fdn{S5bL8Sz?g{uLyw3ps_+<TL)9u$5bL(VC598_LorvuDLOv55_b4`m7+DS
z5;BTIX_Jb4byl5G*Nr5`VF1aXkiynv%HIp`gH*uhmVz#mN45F-T2Nf;6+)C3zVAWi
zE6KxvK2UU)Kb`+E!YOay3*=rCsj`f1hJ;o;L5~*~dLE^B?a_P~UXr*u9;a0G!}?S)
zat%|G0O=ICz)c*Tk6}X>D3YEYGusXF!+56j$L0fZUpE})8`_Hz(z^YK>{xZ69%v3O
zo<94feMMUmmwtqBTOik)!pBi=Eay?5NDIS}qsz*mNravHqH~NGd9*rMSYc?-ueB<&
zpw%~pxv<OL>Yldxu$jb&^Tud+VB$zKkB$M?NRtASh;Tw$<kJ^!%6~UY<vxj1s_dQv
zf+V~#z@t&Up&}SVrIL4i4d|MoR>fOuQ%&?(mV%H@ffJG-3CQZGz7$&I{}2rm{yx3p
zJ5@y~lH;wTq*eIaVT_I*T#)9Y1cSU*Vc3r5IA1JCx^)eCQr}EDbr#Q*OKg@G&tR)8
zctC~SX0_dl`!kRw$NW2HLm}zq4O*T)X-KdmHJt7k86zXg@$vDAdT*M*P6Ul9eU@yf
zN(hld-t?AoiOqKD_^In#MMJ}v3YokVb{12+YC5Avy`Vc*8+zt%+kq!Iy(b)6IrsNN
z5rb~pG-`fZooa(cZz}h$s6oYACEZsuG>Td0jNkO8)vSP)ZefaTd@yQqBRq2)V);+=
z9lDUZ+-c7za!mC7;&k&S9?9N+@xalAmvZ=OrM$=c*~Vi#oEoX7Z{*lSvQZlPW$NBx
zr=6NHRT$*S$1N%jAJ76Hd?wTJ6f#7ikV)29L`2~>IuGE~+ROo_!2R_Q%w%2*fnNF4
zPZ=#9gC)5(pt?yD>o&F5nG@5S51f8XV0@4Rc52<{Y}vk0)}+*U(S2`hNkKHjMjhlc
z19$549nnN8#RfC7qOZ>aq_dy|1xDQmyUs85Q1&+Z*r-KtLI1*U@R$Zk&HR~HMZo?i
zA&bM(`vDrB%uc`SO}xj}h0WPjCFZ<FsZlu<VaS?5%N{%Bg<c5u+UJhZ@B&(;n~*Oj
z@J0_VIy|ZLcBmP<4WzbG*#FQ-@-9B$Y301d4$5uZB_V<7rkrr0`WHdBAtyY#t{z!v
zay+~C0eVDHVUbLk+0+0C0hDy@^b_tkzlTh7To{saYUXHYZ46#bkn6|Y&2rC)_vz5x
z-QBSkdJBJ>sfK0JsfpE*o{NpD+F75_WX)o>b~**zocF2-r?a6po0-{SDIXnEEAjXn
zv-C;aD4z99jwv=c2fpEpuNQ5f;_6H|Yd)6ur(OsVq$Iz-iCZ0BnIB40CIrz%oRmd@
zb3Mr+UWUr##mb0^`o>dJlwjcE_QlFb@3@AT(90MbuNIp;a8bxpz(W?iCK?Y8M)3C!
zd@>H(J30e?qp-Vg7s1P+NAibW$pSW?RY$%8sjgI%jrJg~T;)qXJ_1koi`Q+3a9Lh{
zaE^Q4OxmoZVnI>Sb?geMx3~G89+$@Uha}1im)|*dln5(Lyq5%qYdHI^7nl`@NDi-w
zGd9{pnlSPmb#8+1nKVj?khHZr=I4p>^YdBRjah@v_D(h!5YW(qs;a8yW5tcOmLGnZ
zRA(IV%E&A&3Uzz}=^d?Apg&1ZZMSNo>eO4;S65O|F9^nyNH}P4K53vCPZgnK6)!=_
zGtA`1Htiu^ys!-Dn1g<G9+`YtgST(Zkr=$&-r-k1mUj9~Xpb>~SfUgpT1rWRWk%YV
zFecP30;oC?3wcb?YR_p?i|ib`A3{f_fgH*)jEi!4XPB&z+mJ!g_0>bOfT-1VbGJ*)
z-(bMsaR+1n=3dUr(ip&m_1Vh(i77pkhta|iZ>J)ITwENuW*08_s#pOP?z9nmqh)td
z!_yg{Oyvt<A{iyaF-`6tK>t!J8`9&Jr(mAc^%;O>l-hNN2)*j4ufO0S@~BJX$CNSs
z=f;&{SuVb|L7c`*>&xWh(85>vHfQiu2&81h5`BCw+qr^^@#WfrTp35KK?oy;tA$`8
zHrl8D%JoY4SW@a;@R?~68CFFbT@6D;@EZLEw^E+F%luL|kZeA1K)sAJ@rgxq)uu`@
zW0=@tr)Lm;+-z$zx+mTKS%JAbQTCX5!9AEG85Q038nP%u>w6nRajf@9@zQ1Q;m2Fv
zool&)-H|q%wVU&+W|gg1e%isF!3-Zl+sTA6w=wRMkc@C^4a-bus_AU6t}Ao%a-yvJ
zCJ~faf>>IuLZh5-q=}=Vw^9f}csr!w_5k@V6aA|+3LFp0ZX=gy2u>mN&GE~5RXL3w
z8m-)f5yn&M`>JL;!prz1#~7OVd%Cvg4_?^^z2nuT@US3yr7E<Lkg^M^7OHDIsp97B
zqSkv|nKlkVDhA=qyQ*T1((ZiW$(8<<u4Fz7<*%mm+e);1%IbEvw`S50f@K4QF_j-O
ze#OUoNxI8Oc<KMx!lb%3U!;ya?X8QfO3prCV#oa@Dc2d8-D5c9iahEapx1raWHY+<
zE4C;KkBxFvqrz>JxjP49z>^!+k8n0OrkacY*#n>ycKjs<J;fkcip+5q*!0-J&JUQc
z?qjAkz4-Qgw{MF!beZjx&k%5r;XoZx$;o(N32`!Vc8nDeNSewGZ-me##tCSY%v~$i
ztayx9(pBB7sF=@^6=>wk6oDqDx!t`zd0NAn!Yq$@?F;;M4u=o-j{=!vhjMSE6&B;h
zoDLaQmjoTx5zOQ>eBeJEE=Q@6SR^{pM3O}nr7T`t4xMaPVD>qdrCEtz-5$1~6X0BA
zPUW8QkEx7RwS5<P4XTz!tw*qf(~C`-@v^_Hjeo`4z<6zOpkpfJHRlEErbiTAt;h(X
ziU}IHv_vgP`<bd3v^u+ceRW=n6rla4J<aG`$)5uvm{!Gc`l<DD&JW_klw?98wqJfz
zo?oxuq++GM(f|kN#!!ED6#d1P_gkq}PHVowLvm4BC?@ylEQRCAO{biTpsT01B%K|f
z#>(sYZRKLn%k^X8tG7%P$*+>)Y4@a^dK9m*HP7ntdp}6+%;`{356OoC`6o`cPC<a{
zq%qDU^Y73ToRp=tA2dS|K2yFI+jqY><pn9EojA`E2kkA@Jl<0G(8O(G+-pq6Zth;(
z(6U-E$5aM0x8luUU=v6Dj8>dKX{9>9eX`GK(53>i<JU&@gt~Znlxg()9ljbXSx+WR
zOiVn&w$`uoe9Pn#wP34(2B$5>3S;s0sgFl^P#|=4ZsWk0w6vP;e!-YO(KU5&xeC{A
zm4AKtcymTauU>exyFQp8TL=AgE`@5qv0eOIBike0to#jQf}Pz9y5;O=v%>WR<CS+S
zLR+{kl4R94J(@F+J=;M!K|x}*1n9ab9N>Dl+bXf_M~}~yfS;wb7E$W`{r%hL=2C%8
z{-o6>EZeERnRiPLMyER>f((l+?>$xG+tqar|5(PnHt`TDUMbrAmmzNP;9S^UbDoBE
zdsKY-vg>aim88F#n8l2j5;WOLwG#c9v`pY`V@=oHdo>fQVY|@4?uj{1lNVcg#P1%h
zMa8*2z-gGa09`t%M?fY3m6n2?@ss|Q$1To}Lq+8CVs(YQ>iE_c>`uW}k*@WN-a&gk
zDozF;0z}A1!}pn-tm(}J$Z+EUHnyHoLW8iUPyvR_69z9|-3exPD6xPTJz0zq`+RW*
z6i_73$okoy^InnBONneJ`g|M^J|CQRaCQIRO%0RuI~&IRcNOaw-ZweM>V3mjA$;p7
zI2Zb;;ky3PnGNl$>6)^@E~XWthvwYS`1(H^s8g!_YhKWQR80nSQ_n#%;DG~tJGH(R
zFlP$KmlC^JPWKr2c$`b$!DpZORtYBM7a_w!!r1a-#OP&fTm27b%vtNk_M3fld-O)f
zdsJ1zz(<40xxK1|8Nz|_?-!mN8tWuP0CJ-9Q)oMT@ef}kib=n}?S;k+ByJ?qOCYtU
z?j<DbD}ldx1Af%LD1y;>WlXv{kWOyjU$$13b{a2wk6QY9z(n`VEwNWV7>OlQsMkq;
zDA&|vT#BKA8K`|H-qvF*B)Gm78(AYm@@*PfF3jk{gX0+s;A1VN6zHvJFgbQoWQI93
z2<buIG*JpQWzLUE_J1KB)}g&6cK5H2>yPwx*8k$T=u;X5up?>yUyuS%h6|nepsW&_
zIn7A|493SML=qlBK0<jj8{X^;sUIc!Wl)Zm4x=M61^TDmj1~;BPDGyYP*1Gt;K#e#
zd!&;_o~-D*wZ^w^m!NXC-b;KL+^e|AB3-;cko+$UD`reK-T93a7>*HZio3K^p^er}
zXUDmz?b!j$>$`iyUTrZ2ozp8HaJ!jW{_QYu_QirC$D_QpmH;+p(B<ob{2QGBF&8U%
zY8L*?Pi~#+w0nvOH{Xp?{*57e<gE)9)Nolxzb(V3PlL;Bm8eMc`?oQGll*$2yC{x3
zwpNT~b$!MIA)M9UXc#<OaJEMQ!l!5BG{>z?trKCHx^qm8-SP5{7)R0*q7o7!Sy@>@
zKztEUSI50OIXeFnn~_#H6%mo_i1!vPlDGvTvmCGaBgkhtDUnvGfqIwADl~9HCd(%=
zacv$@dS|O`ne0r(0iUdb$;eWW{eu}qc3$m#xnpUBM2U8e1z<PZdMJpCiefpKZ}IzD
zs<we+6ViLSXYO!!g_w4Z>RXY57;cE_F=h5{As;mjIk~@V5RZdHB~8dlLbgvMAX=r(
zco^2I`kgXi!XQs})=Vl}KMYu)B}ga3-2P10>!Z-QQ~a=DFgqnB%m>ep%esxPPClnn
zM1>%RN4U&+VjT7X?Ae#B1neK2w}zIMNEDGI6oO69l+=g!?-O<6ha{ayke`#hgoD2)
zZ|r;fBV_-@BHp<H{$2JD<9y0WkX}oxzoauh)?n(S0@cg)_pG}Q&G=%)+Wj{86s5&l
ztt@0RT_Kn>Miv%`IyyRl-MX7|Y<O6~NuC98&c_IaFnh^aGiK*x>z@}40xlBdupgcW
zdGxF7>O(DlaBF*O5A~&Ev8Tr8RX|EQFor9Up`S7HQ+T33@E-7tM&JB$v7Ac0(WbUL
zzr-T64+#s4doQ#HiOY!rrC_-8t?kSETZ~NZYrIGtX(J?;aAD@s)JTyyw65$fydR&B
zURchCCg4{}FhJt|XZt|BvkF72R&zOV7<EZ|70xeKD|_f^Q!>SPV#p-_Wrpbrd+N!6
z2BU4Ci$a}J+So`-0sTecP|iUTi^X+8#k|+OOOvxN>i?0_7B|}<27Pu>%#U{U26sGm
zO3x!IK!iHC=SEZ=%U?rD4;Gugc5L4OU_yT!O_VAyU{rcAaimMs%z?9sgFFE9n@jt|
z*zUvXI*aiGMtk54WhEiN3C3eLA_9j?ZbCQekH5eCs78$fFX61@+D_$hap?(}l$QC5
z`vys%H|_Ox1qbU{m1UVPlkdN^K9fCQLAUAs`so?ZJwI=r?JdYVO^%n+RcfXo=W>Y1
zc;NSNfD0<kiq?8MHFI-+c_qI98)kkPFjs6CCh=P=?P?8;=*k?LE7Oynn_ilSBvaUj
z;`ZX*fYLl791<r9<Rw!Ia_w97*ROrl!|EaaFu@i<NJjcP6+?F_&a`oFY*pCr_w7*8
z>I=~w9Wm$8n7XD(aw^6+t`;^b!b87^^lhSJ>Z+P+n5=2Zv*?VBzqv5l{3(n+nTy+6
zqC*)F+4^6+(dUe-?30lMVI?A&#&?Cav@wTmb}SPMU00@4YW@^>QzUTaFQY`>vrD9)
zLjSxko-9=ERIW-VRL4&JBKBC6+;WYFT7n;tXj(~PU8Gx7iIwN#;*vm8tX2XyMK66`
z7k1TakoxB+tCUqI^XetB`u61L#(C17On%?}&<bSIJAwL@09VSFXS38lbevE{0EryG
ztFrt>?5gf%`Qb?ymb%H$JgND`7Q~=cr1ly&HJ;qrUo8LPslKI6vUjz+#@@5tl&J9{
zR$SO!5Yg(TvdZ=g%ga|LCbq;K8Dr(xXZs`;myyis%~_jv5pan4#A%{B(VR>6-3)6F
z?uk7Uif^y}!$g}O_)}x-@L}`FCk&s6K{Y99oYChbA??*;CXR}xPV$cRu2ksiM+fI$
zTJk-RYBD+n6)gX9R<6*iFRU;#z4LHzgkrRwk6_ACiZ{ism45sFR^&HtBT5T{1ESJ1
zg-YJ_MRYoVX?kxO(I#{&VJ=?o@iw&!nz=zKcZbJ%u7c!%@V&A42vR=<;Exc<SZs<=
zEIRPn9Tq?5I|29%Xd1gdOUrA+818+};BYy)*oTaIC!P<ddE@z{6<3$j20YkhMTy0s
zk29?v@dj*Mk`Es(9`&Bv3(jRVwj8$AIlHQUo2W53fe8EiDZJj{{fE(-P5}+k?=T@%
zdct`BXRX;vZlgXW`Q6f-)GQ$lR+ZZb3WpU<eA4@=KS%LOXhNbKEDqI*EcF{Gt(_j4
z0L=g>+tB21hIRmBE*!{4srSL_NYCUPndSGK`x_O80IoIut_j^t%uoyv_63yEiTX`I
z7Yn7RH)K*YeEk7&QPH{xC>2VXJllU+hv#OC{Rg2ilu15Q`|QxI!?8%AD8@$qYsR|$
z8nm{iH$B@zf}`Hu&fl(1X-kD8+_^{DpQv8YD7>Fa2>YNE6V8r$R((B>h5X*SOV{3C
z2a`uEL-^1O3cMrik^}+rQJ~yiR19lWwAu7BL-5_N#+v2?(CE>{6`V~$+W0uXaqE<E
zb(K5?Q4&DQsc9QYvl^aE@TH|Jq2LRO3JJB2RdyJ=r2SCVGYG6*@D{J%G$<2XLz826
z1GOD+N}K`|`r;x+Z-vNF%Ne&ozO{sF(<t6t4q4iwb_Z?Cj~{XlH$PJ3NS5lT0I*NW
z%C?UAs?iMr&GKWwux%-=jSZ4oH+0&q4Nf2nvSdM~H*h+(Xg$v@Dmof{{`~lpxKLQG
z%gkwVJ5Q1=O-YE$@4*CyAYXO$F`?wopF8`d%2Gmu?RR2nM7u<dRJtW>m1@mLGNq$T
z2jXK$5)=}$d}Z>c(J%RpO-zi8jiIWV%>ac+ug#0N#&XiXrl#geYP>uYW-WBm>`&!p
zAtDO>t^>|}lO0Byb8Q_F8R|Zf$qXDSdd44b6{mUYN4!Jd#|g=rR`+8P;f=xqw&72v
zOD<PU26>EmJLvaG-_G8+G_;YMkv(ry3cH6E2j<}s29LF9tjpOFqTR?aLTypV-i;=`
zH<}-B)>KjM$6;}_cm2$SmN6j9VE0F_3#>l7_d+qhYd|SbuL-gl<Jd_&_mL&<j2!7U
z@2Yz#0;Ag-A)3%fef8FmWBSp23KO(aL*#k$?X~L~-IeKPK?=ca-o4Kt`gp2kB<~#c
z7cuF<M`}6+)cz7y@u(LJtfgL~C^V{O?E$o=1#dEA04-FAl$}8&A&d7hB6^Mz6PH=A
z-zpIoIIGS&WP%>XNiIOHJFB6VZRE_lwzj6Uu@<HD4Ua!5Uhv7^sJLbSR+UmFiT=eO
z^>rr>&{F{f!|Md=+eY$fM46Z=JIg05@-hiBiMue9wC4@xWQSy9W|;3|z%A{(3ksa4
zK52ps?kC`USV~ss={@Gr^;OGfcM>cI%h1wOn$^=-y5@$7sB#Y}?jOAA!7@1-M#s3r
z_#~X!3qAnWESNNoljc7wOF0Quv}_X|tr|8=t3HS@K$8<!Ne$m;1NKY0rVocCzP^4+
zRT3|NT;y5wCU`oR_4!%Q=C~=!BK_mI6r@-CN(&sxx=I^KjHv^&i}43}zPjWC_$EN(
zqw@z^Hgaj?JjjN=w{4a@@!@L2&|CF~;U$eMBrfsc;Vl$1sVXWiMXCheMg;*$F?hy*
z0r!LPyYGX^2kL|+a30Im29&2RZ>5Q#!c@g0Fp9t!5usVuFU;ksP^Q!mdp8hTOw(iV
za|~Pbub2}I!wglwu@iV%BMRsQWj-_)T_gWdzJ$=ZnAoOH=@s(Tn7?(^sY|oqQb<Hn
z)jh@?8_FiqS$hj_=Qc=~UD)s6L2t@^gMHk`Jhn+4op$9G^_naP&IrWi8R=*lSsS11
zM!eOUz`rO$&FJ@o`hL3L7r8uLliO2|&7@(3RYvP5*S)$J=~Q3vYxLkt{lO@1`sBQ_
zZKctlC0-^5po5I@6l_OhMiOK{KSX<~yktaIanV!`;<9I;=41?5He{2OBt2Ql5VLD&
z<=o!^iynNH{uT1d%NqUDkK6;yuFT3_JYiCK*Ba5W#aCNx_!P~~EDa9``pB`5lV`}t
z9wGA$E-7LiDAy&YRBA0{E4GHB%7TF1>z|5F;|=$pK9SPF66_TJA&42W({1uT|1%x%
zCl_>O9arw$8uG0q%FD~+_wXsa6b!U@k+7osO!CAkxGSfGy<R4+4dWeqNlE=wfFK8m
zST8TH6Lt6b9_UBBhKwQL*wC*xRvRDx?4;lwiksn(kdWYr2r)Tl<wvSLgX)5gdJL~P
z8kSdX4F+}`Eb+9GZ|E@6cgossFyslneB)>o`b!Qm2B+!7N3+E@w8tI_Hiq}-9QWIV
z3#`<>wko`%jP1_vd8}yDg(i}TkF{!~kzn(jt%N0*_!h|IZJVf*K>kl;mHNEzNTrk-
z;D)tQUSD>20^)w-NC8j9m!s2Wh>@e*p?+E1<#SrB3h7ksq#S7Gx!SuBr@pAwj)l2o
zE{p8>uhu=nH&UgLN-;5x`}d1?ym}0KtK?HYxcJGD6ZIAx4f!6%SaGff2qbNH<bth9
z$B&AWbhmy^fZP6`%E+H2j;>G8zyLTmp+yhf=uJ@vl{AvsgKMMH;;ul~{ZRYgU_O9z
z800{4bZ*~}7vKNAKHIc7Jfr+2`ynxpQMaD8b!Xwt*QGXZ19v(3lhf7n)vdFh#a6tp
z<KLJjNCn+#1@k58%qxszyeOj(=#_e5<MQQ`cIlMiR~|O{lqc$yh*isn7a<+l*hZ(D
zC$RMR)cqUg0T>LFT*^F|B(D$WnOLF9!S@R-RxdV(41ycz5@4;$H_$cc2BN=Ce3d1(
z1Vtxve#n@=Db?Y&8>q=$4#>&LY0e|HnX8E-VM*hs5zOCgZ*3VfIoUfevZ^~6TQhz0
zW`~V)vCdXHiv0Dj{u1ES7<^1!!O&Lv(Nybw;ZoM8`5R`Jemz}*M^TkqyrnZ6m>_F~
z_|rbSwvM~KT6?Yu4GEGzlWGjE1y(l}b0uG0|GQ|Z?r6+|K~Jc?)fnZ(%TTGb#=EIp
zIx2D=j(7d&MWt^_)ryUO_K#}TIr26uBN_%@EX>tPm0^}m)qSq9{Bob9S7GwYJ?#yZ
zBIV{u$)jh9Dk+CZn@TfyF=K?^rMLHdFYoeoC<e*eRV>vsz7oVEAt$leF3$JK?r!6Q
zoB&fVXIhG~&_7uf7*4hf5;unT+EYBCP{v)gh+VgPWGGn>{*j2!pAM5|_3GFJaI<_o
zZ)$0=Gx|{&qnwd6Nz7-wVb_)eSl3e?E@NfYy@cGBZqD|iK2b=l1cADi-FC)`-z#<3
zzQY{}o1&^KIJ;fc(7*q-$zxti`IWOxOyvfRz~AoXqObqc{n3bU5a^ElYQkx2IK3nI
z5WZu!7mo087Fvg&zeq4|>9RF8rcJGxtLJq_AVUAl%v{c!5fWwPjrI00*N^Udtm-KH
zfnE$4M>E+>n*8;Aj8$y2<bYVa(2SVr@(xV}IEy$&M+PHAQ)k>$J&Gr|>Tz~BikL2R
z7&7`R(PXOO7D2YZRqcSCg~UjxuiU7^6=i1v(=%@jH^;oJqt$Apqh~pU)+1oGQPRL!
z+X-10J*J<vWu{NMB=})6buO?o`r_O({FS#{sJFah)pj`&U`z78y3O$Zb?cQWb9x8_
z=<;>d*3t2~NQrh$uAx+5^YQKBVoR8lxx(7jXimL7y+u#<YTjU$_#Ri`>TYP``L480
z8J=nDhZegfp=#?H<;8)ly2y~}L%-Y0wUstB!@TQTF)>layBv&nli9EX@pKezGKnFc
zIJfkw*Y%-ig)g}6=Gy~Kx9smSyN<lCURzqt?Z~P=T$zl2QzJApGgH<v14~G(mX0i3
zk}R!L#4u<u;}H^Cp+>~0`mguBB@Y@YCFZvWSGAE}QL9=^>6~V~^Y4w=%knL8%a^PT
zR!cp&xoRiRg#9rzWE9&6)7!9^j#uw^&Hl=A9{gD+c^wfoeA|~aDxkgckne7xOgTSE
z12`NFZgAXC{5c>q3$LQ`>q8`DOzA??2Q-n-PGx2h1dg(1g)!*%P_pLtFKXtJuA0C1
z*;n_cpUM3jh;Js8=JwDDyTfL6%b)`M6_&v7cxNh<d<}2FQH=_$m<73fNRYU-CZoC7
z08LF`_QDpTX+DKt<4qb5qO7d!^z7^a(=0WgRSw(GTUxGj+RTj&(lY%PBo7Y{)vvh$
zM3$>O?!>*r`|$P`V~A-DrXlo0mN||Qv+uSi@oLhL$#?W(-I^|ci!AWDuzMMZ#w*Nt
zpjsVNl$G>ewk#Z1YfK)n#a_QVaQXfF_ww(wkECK}D>4UbW*WkLDcftg5_fZ>Y;3Oh
z7Z@!1&FyVyj|sDSocH=-%NrXB2cGf=TOp4-e!*J*hq1Q|t7>1{w?!ocq&ua%q+2?r
zyO~ILhm>?9-5}lFAWXVLy1Toj;T^2C_p{g9``OR?KMue7K^V?C#&6vBb)HucS^dFe
zS=!_Z^VsF}@{M(jQ5%^+7>c;~CwF%qE7z}sqm&uDi>+xes{N7)^pN0QiD32j-?Dvu
z1&+!;{8lF&K!?E6hQb|3wLw-4Cg8ilwB&nKvO>nP$xQuB1Cg*hc$52V#W6|{p65jM
zR&g@2qIe}AtCIUwPpz?D@0|`Mg9zJ4nXeED0F8hbISRelmhbiS?d)nfy{JHRXlZ2(
zJO}RwfBVJ@zBH<E%rRhEeL|&81YoGWhLgXg#p5&Hgv!c2k<fgYMC8MK_PH;Z<3Le^
zzY-lDRA7_>X6RL#b?KeR36*wjdJj&_JcHL*{lJ9Y?K6&H`JPm$XH7vUpW}JKSBdlN
z2140Jnsoa0Rz6nHsi3qpe1~9X&QqU|t3IlwXMcJcy>4UbUF!hhN)~_2Af>9pGkwC7
z4}B6sLlAo)Oih<*h)eb2Pe-_C8eJVvmAR+KYa&5GA(k|YM{@Aza;;=`@wSP4D0_F2
zSQct(SOf$F;0+@9x-4xwpu4|c!tEk7MnQEM^W`EMJ#ye1P)PeF&80}m09iRs&FMGv
z)P8t)fm}<^pIRmH$Gmz-FN-v8N+ahsXG%nPB-e_sy!72j%x@dR#(=By_yrnjaxaqP
z{i5wn{fLe9nG$@r|6JOyEn{Z*yL&|VQjJemviv}>(3EU78VHW`gmqMkUNC0YPH1MU
zO$LF%kSq0s0Ycwy<&L=#OVWQKiF|aWdNB@Y{!Z)6Y-k0&?y=(Vtw^8-Vzf2V4yCyt
zLi3(!@5U%(NQX;)94clCLT-+v-7CKmodK2^$ns1~X#Hvv0_bU?rSQ4a-v$%PKk)Eq
z*^kYvRCQt_qoWu4XI?dta7hXlD&*#f<)Y$~9d!7wt?t604)+)2cn>x!21f(&$3LzE
z@~Orzf;6uta!YJywQxl#jqISBD-E9=_enEu)rsy`C8EdOjcdxxtZBmB5DDTgUoV9h
zs1oEomia7pHK^lG^e2P3Sn(aX7(_q`jvQ>|*szADv8+IV6S&4_Ftf}AOQSVyd4x-H
zjNKc#uEQg1Hn^wxc`69n)6c<$;T9jSw6WgM$1YXv;9+D>OzS^gWfg3);=<L~L`DH1
zNET_9IUS_t`g)G%C9pqQ-P|Rmo7IOAUi|?i{LbU6f`6t&J`l8bcZ=+=1o5Q<jBw#`
z)TNPpArdYwO+6Ae+f|CtE!wA&?JwrW8G;r5LCd4?5$HwP-<25tHe}KNv4{JA4p|fw
zG(tw^=`Gl;eRaxix7hr}e@6Nare{fAa{;S|x2Us?j-SNf<ga_SS$j{_I0@eg&oQi1
zvstr>Bt)P`-3h-X9-wAQ+a30-fvHndQ^OKQRxncqNg#m8x+qNUTn9eGrW&=046&km
zy+Srkz`Ct+zScDvFDYCU6vmFh6-kchtiwN8Yt^Svxh(PK`2+CSYgriI#X#OHIM!kr
zTv<2rpMQpuS;k=3{bNL9%O*-rF5xA+$eG%Ft6(<44e9)EFbPuPJa2;GJEgKDv^+(8
zUSC`RtqgxYG~n}i_F)KY1!t_-5|AEj$8$!;a6i^WGX5iIDlwLNn2DWuml)V9ZM~HY
zaV=ACe7)3k?M`D2hin)W1&L?QCNLPVrJn|RJ;`{Z)0m`0H;rIvjv-TYAVS>n+!CJC
zVk^j$a$S=61t3qxkMBAMcN(|zuRhPEvHmSlM?rV-{0BKtyhEt?Cpiz0CwYLUNX3Hr
z84Z76GiCu0CR7rZa<6mHI68!$E=$Omi8v|-+XdjKOlRL?W@LL#4HU-4D?DKkEpFhl
zNLyL|gr9LeBA3McV<tpxCle=_a+bDcRyO)cfH<o|=Y!<3B-rN0UwD^uHS1v{xiTAW
zpA6A<zyKPz`0~qsuz(iNiS6ziBqwC>Nrhm21byGb`i*d`{Gr%^D3Ewc=W%`Q<>kfm
z+bh-VoePdP@D^{@5`;(JeHhyR^`fli4LxiG8ZvfOOuFg>J?8Y9W^v9;h@Vq$^uqIA
zWXN4As=yb@$!{FhU-;@o(17uf5y-t6B}p>AnD-3E^+~$SSUSgxv16-A2kVpZ@V~SA
z*2Mf5bqOv`{BmUp0>L07yEMvt*q-j%-nH3gR!KW*K+Ph~Yw|nWQB+c{xVIA@7ZdK8
z;^}<4d6~Nd7@YPo0?%;dFz<WJO7jsI_yv&vG5E<pTwU!+^{Y})1k4M41L&n5bIaLd
zN0R>r@cWwoEkwA>4lfql@-KSdRz{yZE49#MENfgcusa0ZSBNQ4DKxD#t>!$XPHIii
zR90P<d$l9iAQ*5zyi6i~wze?p^-)TvGZ7LTSa;?gV4A43R-}v7VO6D0tA~oUMIFuJ
z5bbZBb~}GQLk2r#rAmj%ELesTWW;YeeiskSAz-;5Bb``@C3Ylzh@cm}-j+g3k0Hhx
zEhlU{vF`ziW3o?{k6}EDO1G~&Hor{vww;vpQ3O}*?Pye_al^+*TE(6T(thn-q+{_~
zayq3oM^?NgL$D4NOB`jgP-7$za=Q}NExhdBoindEnYU*Ivm>`eMtvp;J68jIDOnaB
z$D!OzHJgUg&c)#>zPvW`4AIPm%+&dg|3G_^W$~DXEnd)`v6GK$!gG=YY174=saTdo
z<GeCd#7S1*1&N?f-&gRpW~-s>-RebS%9w5UC1}x7s_Dlup|`_&4pD*Q;y(1yk_}cM
z;$ui=23Py@wFK63C7h^RYPY?J@Z0}q?aLi*>JtL6aVaJA|D85NQX<Mfq)z3W_;(xL
zE4sHKPKUGK(wd#tPL`T0zmBFW4X$}TNmTMOZ@q*P!08ZwzVFI%nhO1iQ5W@6K?(=-
z_6-q~rFpfb2G_2FJ&0Q^nQ?GWB!zI{BW#y>+|)+60s+J=v!Tw7&$FdiFmoD}(OAs`
zQkt1$fId|c82B}v%m@7A_%7w@3qcBc)roAO{t>bh8lj^Axs%bY%Y|0M3y)9#?A+Y0
zOz921hPDP(J8CF3hEVNv3zPm>Mgz{?s~PGDSP!8-ln2`{Pdr*hY*6;lb%G)kG>K*i
z3I9zKh|dzz`Z|({LeBFL9{#Ze2air?g$OqZT_}8cgRwOnlE9KM((293no1Z-me#-J
zD}B_@-&aI&9-29ID_y6zt@G`kcp|T<FK+!lTG*&|0_NMFacfMMxTeN)Y&Xyrm=ys;
zmE`&0#Cdc^RU_DBvZ2ss#T_;NYDum+6W44T984FQ4_ZGZ*qw|mMnlfcfE@7njjj<g
zW&XMzv|q7eL#LrT!}oW^+ESk>z*Y&4l>a4CiM~ab=#gs2C__ek=S}*rdCfoy=657_
zNKJD0&}9N=`Gzy+#9D`Uf4qiY<o_@}QaemL!=(@0IvfeN5^dQ0&-i5E!+DG(TI5cz
zszID7JG_||Z8u`^q6{6AgRs=F{aQo)F+~X<E-8ODVsc9V-*h`CqAKBkE|h>bsFAXs
zp6YgqBe$Hts|=~&kJ$bm<4ls`Fr}1?qo*qsj&(qm%%oNGr3-A3M3xuDQcn<#3yD|I
zc68;fOP!dhW%!Fm?s^gJ#dyI|GW=q^aJbZ`8<hWF2sBD#MgGoyvOvx_KpprJ>-5xs
zBZTasMM+QV8b1N|TJO`V@ia3m4xoFq%<KQ##%)`=DNW%EB?{r%&B?c$5_?v(?<vcM
zK3R^MnQ;!+hoMU;Bdfv7P~pNG?e7`7NFXw2*z}i^?OoU(4B!Vl*!n1@PV`r{0^LT!
zQGyqm+1KQ>AR5@pVFQ%{oiIQ7yItvjgHvlc<I$ni*F!+w(;dP;#92X}!o$r1LhLZo
zNAuqY@l@FQ2~Uy{uT!4faI}-)Fc8S(C5xbmHn)ILXr>GKuy>V!e<bX|aVM7;@a+?K
z`D>U38*kWUL5Nf92j<gqE93?DWVGmTo|eXXuL05cVCO&bDR?4MU)25;LPjV3kL6wl
zxTrbsY_O~tMvSqH7k+x8CCTb^goI4idVMH-<K{M$9w6})_9y&l)8vjfa=uHiH%c`;
z3hG*;GjpgEL<lOhyc$rolO&NNP3)NOkOud^-uOROV3a3sg8-2eYB7`1u!LXHPg|Fs
z*N2x8@flK~Oy}EUx@bhIwyG?S$n6>8Z^2;_X43xqw+4&I=mx4v_Lb4|4#WU<DXdfN
z#n?QGlPIGT4*6!kgRNeN|0>9#^N-w2oaj)V#>-gu+^`#XGscs$dJfDs35E|$l&twv
zoK!q+6yewTb+BB`nVG%l>cD$n<wQp<Oln#f8dK_XE*IA%eOaKEV^gQM6s%IaY_yfb
zKOsTI<|qcgb@aoSq9U)gYb{;V&-U5k6SlK;2E2>xhg~7*=pPg?9#sFfkGtzpxnF7P
zUs(Usd&&LjZ)ZIcVz=&Dlv@Sd#vg7F&zcgdd_QOxsWB&3Y8n2<9=fOeT~6YNMLIzA
zpX^~}JO|HmHA*u9BR(gf*82e@TK=5j#pct{1m4Y2J+waXpoAA^+tLNb=2Pqx`z(VA
zyVA1<e>E%23qisIzzh{*9Ygp&@m#PT$hYHm!oC_e(8Ytk(c%k4Z}R6~Rr;=K`8RYr
zSnlPE(nT2Nfpx5K6jN8k$ZOBbqT|aPnX|=fPlXjug07~{k84{OS{D<-m-@^B%KjQA
ziM>#UxXzuch6bmm!Lj8e!PhM|yZYmo_!~{yW6R1Jg!_Ltqtue!O$szPoN}zozj?fC
zY>#-u{81R+_Wc3*iFbEbH`WAY94_zAe`J*O838a<oCbhSd>Nmc=u$t&ImIA_dPP3^
zD}v54?Uc{>7n}njeu69_9&R1(IZGm^qOoiwB-WsAJg<1*v3MXRrs0J)viipo>P0@B
z%G=e&qS(-yk<Y}5J$As|*DX&>N@fn<jM%Ww?DQuE6;xpM-B@(K96(=aSH>#qkTvGT
z?<<X*T<i=n9WYSu9Y2g%E&3f~>Ew`*kM6=4qdqPtV$IdJ?KY;#@1DDr@a`rE($k?6
zmmReY3^T|b-!t8)ltU1|LE8uq-<W>$NRj15cRv~P)Wg>m$1Sq@ou2{`L=GgW#QtKV
z8W<_0{OfKN9mNu<xZRWg2Utf%3P2P@T;&?vncb&aP;ug_WgY|Q7BSN)Ugw&X?N7;#
zXk@ZmCCl_CkzG{MWxd2dQ8yOi#2>B4gI)Z1yvS#m$;bJz)&Tb;-xmy<<B#uS1~NJj
zm&o<D(l1WW5Cz6VuHOZnHv4$MiGgOTa4U*<Vc_b9x>OBmb228&e^QR^iIt?OLSh0B
z|B|J8<aiz~h5x`g@s^@J-$mM59IHLx#lp4UtAfsaZq;HzNuhfr<oZHmAi4csHu<{X
z_x>3>kS+u4ictylZ3ps{9F6oe>t~mn3Sq|+s2*QEf~V2JJ=T1tJA1?fES1r#e^|B7
z-MR_M-nn`oEJr0v5};L5eK;}@Jf*{Xw8!2KC;Y}VfuVP_{Rjw?S;7ma<{h6FIUSzY
z?I#OoJ~0%NF|9_Mp1C-=G5nV7WRcANTraztQ;=QwU%gr2Z>mIG#P(7=y@8_VdEP<M
zSm>V`BVtx7SEXVjeufRhwf64Bg=E%vFeyFSas7U)+PQmmU-|XX#a$!M+Y{9vEK$9(
z)jZM}A=D|qf6OJlKO<!yv;1bM`FCd+7&K;frF(M(%|1)T)L*Q?ZB9J<(f-vP9e3yt
zgW}4|{6@8Cv!uwCI$Z%BOlZzI9{G{{rdD{%`Kz`9TgA23TXNci=1eCc_Q@eHR80)o
z?yfc7Uxie_xIT5UqU@Y3Ko&Q|Xgkt|A`v>ON{W%bj2EGIxl~@k7MMH~q~EMZJHT%G
zwm9+$uPl8NIdIq4E~g0?O^lr5cM#$Yd}XnBPX<ND=_3|3$L%ceR#Oz=Fz~moQ+I-P
z)2?AHDU{Ynnr)0H9ap4x6#&~Wnq_Qhe*5lGk)J=r)#Iyv+jsfI5?6z-CE)i7Wx7Aq
zZ#&!w<^%zoZKIx%Ch~jk>t4UxgNmIde)kNX&a{bziZ4BS4i-wtbs<`?vu`0&akB<b
zvDJQ>0b;Py_WAb*EcZM}8NkXnspB|1c3E9<Y5rUP;E3_6Xg{nWKJ3=6GIG%J7p+Qu
zrxOhl#7x<L9!BK1v54wL^k?Ve1Po`^?Qo;&3Xz=j;eK;D38t(AY!sk<?+X0@m|Drt
zsh{Q(5VfH^LPCDW1M$neX-e&DP3hzbgiLc$P}}5^%jNNdSt-oBckNjj+mcrVE%)M)
z))U=k5<n1XK+45hBft?+iE@?g?Z*A2us-Ax9_}1V2PLBI8-@Y$!`22=At{|6hx&OF
z*4a1tkHne1m#rb2TYJG%G^&z@`#BQ3m#(s5>2-L`BTHVqi~Mk}-@G}yoKt-J_N~c4
zjFObN2~1WNajnKcV?*Cp0pVCq$&(UnRh+O5*>s8ay34qf)0-jrCsSZm;*Y8r?6i>8
zn@@)R0NI37ITmH-*(84RWQ*fEKOOkU&7BH|p{$W@zVrZdze6!Jw;dxW(_F$`VNFQD
z-vC`7Un*arC(w9B$E8LW43Pkj8XUr=H^Q)^zsH6Fm+y|SQy)N16&{tutHu57&53H>
z?sYHhzF@?{wjBOFgO#?&^}>Yp919<x4pJ?h_iKRD(L%QI$ua<U;5=#Z<KrLE8;AxP
zneaQHkMIxAd*AjIhy!aBaWbB86m0BZK!aO~ak5aq)$-wwrB5fZEZK|V4T@Qt2;rMZ
zQI=@1lIR8&y1*RHNU|hWS?O~4Ao6^g1{x}Ac<#lH_!S_X$?a7#y!hjPMVy%U#HzQ1
zknNQ&op*2K0Ak%wqi=Q@9{%wl!?L#0M6N@2t<dQi4$Qxt+7*IjjRqFgtAEvLbJti*
z!TQ62b*LBmO0di>2?Eb7X-`e=T^w*55bC3=Y>gz3G-~C_b$D3{bO|$9Oum~H>XhOD
zrqj%+KUvR!KBPQ=_XF0+)NOiytdmDx_xrw{wv&wbLBiog$5L2Erdm!~8cWK-x#wI2
z*g~<tpB}<synn(E*4{BVUJa;VXD#V8{-TYeOsyg5IJI>(Ye%nEbf80@h)Ma@zVPXF
z=GqCiFvb;1`}iwVDGrR9U-Ky{(j*$d-pdn<36DZH^}FVZ-k&Z_H8d?pCswxfshP7s
zE^ZcLgn%@eq!By~?pIPe70X~0k@NiO6^t-$PoiJjSP9VACxfg%5`~!d<Z0iR{`pwV
z_cNxbbeJ^S#I8e%X;uvkk6u1wa;`9U7gMsg2zrtl{lHlJ4Xbg8l|ta0Snb6_=T!%8
zSBU>_=U?-q&op_+?+}ldMI#xwC9a-71{mt&-e%b%-qZ6B#ozj0>b}Tn8Nd%dxl#%1
z>)oNVjrJ65mh*mq*?*uMNFbMdHK1A`$jHpxP#ssOd-qMX(tOYTq3KE1LhU26kl0h0
zhjVAN9hmIx#&o)}+F>OD_jI$t_^Nzj^Ak^cUi^VHesFI-J80~%`{ro$vl{n3beMVt
zI;KJ&S^n<@-~N`SOvR;ffUi=bHav*vApmpS-s0bsWvTho-Zsi4*6a}3k9(Z5oNQ#n
zG36vY@Hr&~888?i?cP#gFxd2VAh$4C?7e&Yu)~`il=w>aq1FM)Q|mnNQYW3~$@sA8
z?qTVk$C2ZSw@$|s@uaN`-)nV-bFjkT+AIva7T?qKCj_dX6=((4xkQ{VL3F4+;6hfR
zB|)i|rz7bql*xI`%;M1=eLDg#P{b>}cOwz4DRZBKqt*oNE)?7>rKn`y48>PDYHPUz
z3d80^huZ-XQ>`RS*O41H?8LRrkMMZV<A_(xsakfkKk`tvsO8?0<!D%KGLOcLpc9jv
zkt#KFie}vK$D4a6sp}j*9WdZM?o|AF0O9SR2Xz-Xegd0fO=s0CLU<|uL-l=G$(RRA
zM8Sgj$V!1+bl$Uu4OcO7uWYr;f<<CY@R-)v7}A_o7z*<wOssa%o`~@`eYdGle>B+^
z&I=7@N5`61QOBq_QdPw{3E(~zdhM?lfqhp*M8r!Lsm^LC1h7z%u&{Uty+2uO^p%Jq
zUz@Krmn)36`Y3FLcg9_3G3PU@Mo%jveCbBDfNY}8@ivk*VX!NoqS8tie*Y3T;-tw*
zG(o_C>C`KMv1a0}Me|q8-f03s;qY>%9Bb8@6D}B3=x4-)UWeWa$9mSK<$94e<B5-l
z?B*c?1_7%JBf44liKbWz)2J);*ujqVn;#;E`YKuT%I@&J2xcQBO1qETs2rYbSv0sq
z8h@?d#$)mQgb1T?>aD@oX>c=x%d4dy8Qnrm-}caI!R>kCKpUn_->?O}7G!u1em`oa
zulKwk#9xW%HJTXVn67_*V1R66_wLT?iaRSAq31?G*L-=<m&wqAmnSK2&!w~_AV}3(
zdibkLCoL^pDG4}mAzS;t{A_8rT%KqVPRuHhXlriUtc#Z~WaT;HQ!Rn-FN#~OA{$gi
z?uhogX8H4jWu%Sp;fsj2lBri=1k<Sa+}9nubC2@BI}mpnZx9#VxJ^J`d?UdG!TtO-
zA_6Cx$zOArZ#;fu@RZNvR(mFg>64HfjQH?TI=Z4HJE!^@Kv=Kbf7#^tzWnz8v`J2n
z4tx`xj<Bgiszq9K!HOMWxW4!y$ER4TGwN)i!7$^UE%qe97wW#dRQ~7LcgusgrSdAj
zrH|c#3u?Mn%8*j^?b8*!pQ|U+t?+0VT@*ttnVd4isuxY;gvN^w(v%g#GGLR+{|;Y^
zHLkZ|_kuP&M&3U`mT;pn&uF}g$!yk)L0oCN8}cjgJ-Q!Y#o~1;S-i`0;c0Qkl@J?(
zm`V%XKXLK+aBOC^JDScLYBZdj`8kEn#?DR<1R`EvU;m+!e-iWD;ScwME1g~LDef6V
zGI4E}gP)>(7~j2nSPlphx7Va)Fw}(IdNwYfYYjX)AjmY!$DxeozO7?t2vIHl7Wvw_
z{qa6yO|u2v3{8T7%KU9iWYLdSds2EU{y?vlJ3BdiJBWjA`rd%r=hgONokFU^WXDct
z!wCJJ0)87_e=R#v4D>O>ePMT<OC|&h9v^667B|Uodg5U`bmIuyG)bG+kUJrvyQ~vE
z2ZwvDnI$Aeu{olNVCjYBhwylxee~QF>&6N?w(6V|7(~9e^m5lBe=bHyj}kF-=PZCI
zlYV%&;2!;&cz>t2(&6Dv?EQ-7^|L4j<x<ug4Yi!#<Ub?9Y<vQ>u_I(t*oQnv*VB^#
zgg#=`LmatscAxpx<%;!-l^Dz30)FoA=b%dS@)jL4)o5IdX%T(ZCY}NAO~|Kkk-c)a
zAHfE#>`!@ByNvoDTV3kV9^DuMRb9@PoO&YO7;4dzkYJgHG&rM2W_=0yEH!lXGbHAO
zs(;>V?#l6=|4yCrRKkSGB4)<O*4iR{q&0?)I)OIY@>{S0@Da*JL*3aodO5a^qa66A
zRo6qTzzea@WW69O>jWEl<=dbPUBbW#EEROKOnm~}Sj)xtz6s&&`DD@70Z%k%t(XYB
z(r&oeGe;r6%jT*tmG9Jk2nJe@vB=W8JE!$5ZQ>Ty7y8=lt=e{FL)~53`)ZSys}>>)
zU`8>jhUK`Z{^hk=mtu<S2$e)J0-u{4Qj%V~$q%@{0Z%r}leZdOFz9ECz+dDL>)@iI
zl0X`k>9mAg?2Jks9A|LaLjlBR+vEK;KuXw}UWO%>k@_&LuY^R(-tkS(%wT@~eHi6A
z(;q{TJI3e(GZaMidEMFN>sJFHEirB`lfg$<Y9%5ZTkY3*9s^bI{dyQt<FhfFHD4kL
zhU~?R=tjuxmXo>_^X6cJm^}orS+7@341)2|mBS)9dVp4avx4;uT8EjD9T0T9MUuHC
z)wX_?y0D&S@9oQg{;}KIk5#?hl=lYb4H(R}8ltJ|uA*=&K?1~neDqbz$?>`tW(!^o
zIO2si#_)-8!U6AHk62;}wC!t{j%Thh3JMXeGc4B=&&3fMJ-rocKM-?*Eh?j66t&xz
zee0L8kMzGN!`7N(c6-SrsE%_bbl8sWynAX^DS$=(CX1BS6ztJl-V%lLfwKVHH)TP#
zlbt^9t09rKEm}Qm47(09wc49BH|3dBYcObex><r`?4uhVrI4@<ne*`?BO6-)mpt84
zZ{D?3XKLHu^Hj75`T_zH$`|tm%6#{Cxu#_bvBSk=??jW?KZTsU8goUKnho#^$dYUK
zX*k%(?nakIId#}Un#8R@r{ux$(jkCQr5mwobiak|=<HlOq(#snCwsdZcF!^N+fj5u
z(qBi!b0xP?oR2H8^9BPvY~nNI(4U<Z9shdd;Vj&H^zJOtp_~pQ%pAU1b?lNIFu=Oq
z&D=Q6q5lvUlV11;tvw*V(ehDR`Wi_xPI90l+I7t3bjQS<=@JL0R4ov@G;JBy%rxR?
z3VnCV2qFurto?+4Qlx$IStw`!!5^-IUJL7YcL-5raIgb}G~!jLFP7o~u0<%NKO4BD
zS2iO~oOd@KcI*suC&GSjywBV!J=059^F_M(H-0AhR{Sd)V05WsIogTjfBFPOLF9z&
z`IG?9XEFgA5|a0d=ly%&!<}H)baQ)4B%96ygrT~^aoJD1F?D~)XNxK<b#@BD_PIaZ
z?u-w}z0@*&5ZUxYCNa{evj}c-JjfFChgFaheOP{eJn<SoL2XY7L8r-bK3N_Q-0F)W
zu{~botTCTK0D`Ee*VhR;%zACFOwl5e=9<~rV*zz!>$q`PW>B~kPDE*!Jq6kNJP>n8
zX#XkJNz##P&*=43&vgk)R>nJ3q~iY~7zOo#R;%-%H6px%%BP?G_rf=oNxlU0nV;TL
z7}%8Glcy%n!93qG#JWI(Yt6&=oLf5vq!h<a`E_}^iBIR%wS-*V75cZsM@MU(TT~|n
zIQa9~o9KqaJ^5LMTx`(cpUb#t-*2D|Ri&_7OI)u9<j7?TCXbp@xJu1Luc(e;wLT-l
z7hE3mmfhVFz!#69HiyFZ25e-@cRrM|$2@qZKaPP?ZyH%cNN@BG+N>FkoRLyBJblX-
zwbn`+^KGa!U)Nz!Eg>7NX-T?79Q0a2#F=NW7t4RaF-<qXIr&OpCY+WLns<cxp&*zW
z;dvSx;Ttv&0%0F#{%4S6^Udp%x$2z;`)h<3eJubUT{VvjNfV)T!6d9@i%MM$odY2x
z<(Z;$mF?7-MeazvS%#xWdH#-Lq9w*12V!b?%?J_Q1fc;TC<TgRlI}CQGx>Lgkp|IX
z6QcuhlcSP+<Q?1KYsqgvVhc~*pWnLkd6wrL`^heObrx{>N2g2XZ$1_%Ham@b{+wEw
zc>Esrzwm2vzZku!W<O#WJf8~oH4fh_ZKyuPo1L}QJ+kYo%oEHH(nrhxFEg#EojCFj
z9Vf8A>F|Q}2d&zQet)2y{fMe-_;EH%+f_bmB2taJ9F`sd&riMCnuM0KFbRl2?ab(X
zsgG-GGd#sor%-H<msW&cABw)6#gKos)tVA~cJz_JA~WpvW+O89l|-xiZO)!BKDU!F
zkY-)kTMb$5V_$Li<@31H>-$bLexO>aLG~FJ<r|tjECwwlq{@7qB~6RY178%X=v$j6
z=M#EfK9|#1p%`@0cDI*@bEv>gUMz<EbA{Csmy4^b1c{#Ts!Eu5VQHgx$V?a(lb{Bl
z`$C<i^m>ZOdgtC=b^bULn;)?qOz?*Q7ZyKaQIl?pE)r_$fe#@!$6^ljig%i^YJQt|
z^07FIXoZtVR)()u6q`>|?pdVq_~F1Ma{&hIueWA$+WT-=p+0x~!eKuf?323DT^Jyg
zt`LDiKoFI)kqen2A;YlcRTcH7WH{gYych8~d>5MCM*c8s&R&bR0=3!c_{Mt95sh7F
z<2x3{J5!CwU}iqu3fSuA*l*9Rx{St3)|XbAh*txw0?&|N*J8{*&P(@M`i6YQX`9=H
zkU7x`%~k9*7`JeKsJjZPkxepwTP?^|F*d2RQ?{YgK)uoXjnda_x`+%?Q<wl9+5MfT
zu?xUU0U`3Gy#!$Fg4MZ@Fl<w?<*}$y7jG_+Z8a`-58SG`aMsrUhsl$2Jxrs8AgcG3
zzup<f(qO%S_*N2OBkx+?gQ?3PI*!A(Cy!KXY|*xe0YY%-G4$}0s``I1O`CD*j7xka
z(TeUS8QNj_{onz!V-58|gA_MZ##<y<!(=*>H@R`%*tE<^RIFjK(F3>Md88}ss`BQt
zw-D>D5KbX2@16i}Et-kA@Tfw+6ZiwgV)E+~tXa?}0DqI!3AcF%3me_4jgf}WG`zEu
z;T7-mVWHkSVL66IrNq0rncMSjkHT~^H0Wb!6cuPva;iv~$>~rHz=@WBVgsSjI3G&D
zFBcsV>|z~>%?!2Tah-;CiVXDHq^OY<3J2$;9w_EXM~OBLW$+UMY>50}r4m<ZnD8VK
z;jk+$i;0}Bn^MyoXv7Xl`w?^NFJ`|{?rB7`(VfR<>Nj4PM#qE6D4*s423)ZQ7ptzY
zOmn2GZsD-IJ5&zM;C{G0yfxy2jMZOV?TM8K7_{8aAycS0o{TGX8GmYrp_oG?wMe?f
z<z5(x&@wV`xp#2y%&O*2z1TI``qo%~!OYYfd1YgRl`;Ywnd_s9ZN95^N3B*@Tt4=)
z)y<U!+id|Bz~vu!b*$TdJN4O2MYMs(ccv)`>((BPiyq6r>?%$EQ<&Q5IZVRL^RZ)S
zYx0bAK=IC`Gy$|EgLZ^Xqvs<Il*6-VQD`BL0lD<hC+{`wd8{?!t}fvX#Nn1^tRHek
z7t#;EnwbXBr&ALEl!vqVJ$Ohhkb+fgwYTzUHI>2Mx}!$4fi{u-T6uku{bKRw!7wtt
z_dn1V_^l(QJ{U}~8qM52Icv2yk;=srlwgCwJzk);$i&ySF?}{U)rL}gOD)frZaxmq
zs1ALs!rn_d5sdJ)?7kCFn=Ep+yq7Y97wr{Aug0_JCm^R>T}09j7;OXXc~n4|vzJU9
zUVL%~Om6M*mnbWwpVL?*s8d+dJpNV7kVvA5CcNp4{4G`{jccu!(5q{sH!@eo5C}m9
zM6ZR+gtUeCJrYM>h9J{`SV^O*w_;>j(q0=In@UD~IKST<xVrm^j46-&9fL5Zu&X#Y
zh|k8XYf&RmAY-`87vX}@OAnQEHX7M(Ilia5(#?&A=N^c5&ff<OPAQOY`H0wJDj|7%
z7~XM~Hj_#M3p`%3Jn>u!Usz2q=EvKJbh2sxG9J=;@O<E@8@)lBtwu?qZ~3m9hK}4&
z>-M%ac{9?9W@o)w)G*qIP9u{OQ<o<JBZIL_89A5M-uQ;&`}yL<Rh%ki(F64?u@;{f
zhHj1J)jp>OiWF0*@;6WQW(4h!<+i)Aw9{at48A=cC~z9Pzy>yM#dh!SB!s%3vT;S_
zzM|`Jr<Y?-W)eOqUsr$%Nl|g>bhabqIS*TkfcT$bdsek$W);vf_omL4E(3T`NUC99
zOqZHi0}u%#faA6liT+2f0b$`Z5i2;R`JF)qaIEN5&_}aL%Aq7?Xe%qLlVOi`J-{$a
z4J3=|pAcP&in4@KBu|oH(fkZhch+YI*`aMim6e?I<DMyYSAJoWEx@-=(zn&wUQ}H0
z#l$OaZ`wJCTxj@v-y@!Ku+;8-Vsdd``$*7RoQVJq9Zq2&**%K9U-XwRH1~3Fe&?Ma
zD;J+n7=+%!_UMLI6sei59Oq-|{9A{hJ2qG>>Pccas=@h;Ha!zf!$V$El00>6<I{j*
zk;5Ju)T^JGu|GiFk|xuS&<_p!)_60A6KC2qXB&vTV|f9nLt8Unpeq?At1#Q!?T$RE
zj9=cTf)w<=7$MG#@av6qX`HU`A+OY7hle{?GqmR?_SBfPyR%r{y!*R~ufzIc`GXw=
zApS`_CW+n;86yX|jip@lbR&6<J&`kMh>&hbBMb!rA4ab4E@^6w;-m+(azDH;ahTBL
zEKrf<jYNq17?Ao55K~@ThodU<_m);KpNkGm0*P27Q!K8_f_Su&{v$t(?A}a+GVMfe
zb7*%RI_A%@5afox^oN{pGRzn}S2kj5K%<(IM8;?L?Q)wjVu_^3i(xHPbdj{Bzm!pG
z2Qi1;j3qV3XduW5=i2s5O+xSX=u(>r=&f^!?uZEZJlHov?ru#>bhrEkOQH2Y83g9&
z$&g39c}%z&Kk^`x_auFk8-E!?Bjpaf-r~E5r4w1<wxP*GxPL#BRqM5;)nb<mET{~N
zr#T!BIjpo6f+Td^{lviV0@t$1+1DK#Fst8XCFZIO*H%}(eW4L0rNl#dXo>aC$)kEj
zN@73c!D=_!i6q~V<Ut0%3Kfr~QDH<W#*1Un)xn{v{?d<fIA1HJaeVtLQ%FH3ac8u9
zRTHR86nG6H?k<Sp@DuB9VNF&H_uCZ@IyQ25K?AKkQA!4lX&G#=@Z<RJCbGBguvdvK
zG4T02E|TGa&mY29up__2<s;9o;!ZLnd^-Vu3yMVSA|Wx3twO8Mc^Bkj13wr5-&TH`
zdNzjZ`W$AlpJmRV`g4{UlEjNM6v>l*+=~<(nT?E{2H<i%6(#eCgKX~56B)grk#?^w
zNb<MCUwJoS+2|WXW$&iETI789327ZyLZPi&!I^?LTfQ7}S?sMvcSbw3_B+v0+H?*;
zg8eLiWL0t%_dfj;4->KV(<3sS9xbF&7V<O8E73T9pWKIN)i7UMrv;rwH<GAwu5G@j
zbZ$7YqR%~5Ng3cMb8^()_E)6PT*6Oi@;RG;PR$hK%wO9Ga8`i=Lc~FYnB}+XZz$Ji
zb0^cax>krIN~z`jg&niD@v|4rZD*{Oa(f5U)5i@$Ccib<3vBKI-A|A6*#XL;RrNWM
zXJ}CnA0fxvr+;P&N4yzRf9HAj8ln4RhPPOPa5{A52`D*;2#La7$)TnlZ)MZfCQH#A
zMz}+X$*ldOm!~Opf66V#axY~$&S->1ckruW=$(n)KIIGMhB)!(R~)uaLH79-+iGMK
za&8wZE9W~#UEMwUvq8uebuyA=@!aL;A@N1uFoyAMuTSVWoJ{yfx|IYcpa_X{$@y-C
z$Cm-A>WA{Aq#8s}O`rm5ONhOImi7KjIYSG$3)<MiqNFBT?rMK3*D$SnK!o@MG0I*@
zlUk2D=>i%*i3%;sP778kv1n&d#^Y{wWDXC*dhI$m@TM^*mnJk@qt<L3Oi_kSOc>_U
zs$OkGna7FQ%>?|A80Dy4=75}Zb$_4wkz0~CUeefPW)Cy@LJjEzSLj1DZ>Xtlz<DF_
zZ&j@~r7SzKL`XBr6k@%qGmxEVfOOdiv{{mSW3Z(DQwiL)#B98_x99&MaNcIhfh8aQ
z8F|oa!ABF+jdD`WvE_)I?~6ay$4U@L5z(`#JH0wu{7tUdYVFWeR*MTg1QcG}HbTly
zJ2)xUOrCsj#m`-NGGgy{v((ktBDQug3A<8L6dFRNyZ-A?WOmy!BCiwdeX&$~<GFf`
z#XbBc-$Mx5YYCA&qehVKD^rVNIOVABFY{A!zXAhf7bDy{yw;l@aKc4dCn6zVbY6?m
zjzZNuX;~C;zx7&2Q!Bgsy~hKV@il%Y*{H#+GZT9e3$^dVv+Gxr<&^H?Y?0=VK^!f}
z1Yp9RtAzwxw`;!C_f!EGynI5rKIa7`1<(y<=VYFD%MhyUWu3Y@&&0n@xMYsGylL{n
znLwu$6)~dTU)HpW%n)O=AXg8kao&eQuJ+XlHK5{!^|FuZJrnVB)IFon?W_1H&O!|7
z{(Q4c9?rD65UHOe^^`r?-q2+FM#hh=ljE8#h`$2cOScK}IR*N3Apn<h){P|((ZmNO
z-rAUSsLhrUiMS$j91R&BX-9Uv$WGrBBkBz!Ga0FHiB_T;a(Qq>Nc{0_s2XMzN+UV6
zPppn`22t*BRWc%~M6zSAn0uDHCQOX$C}Koof#RI8BL)7!T}OT?cM06Pk6M!H_uv8L
zoTF!yVzScCFx^p-gasJs$1*;P^Dht963d8{#AcF@y!fRGIhp}=cmBUssK(zPl`F)y
zjO15HMq#ZnI$cuoz`?ZTm`=!p1>W3FS)<Ze)V2kAfC|+AS?40-yuDweLWOew>msH6
z3;@vz7i)S+^mc3--rp!H(ge~UJa<isvSG_?ei1K53x;cK_#NbdhDmMfiXwn@MwxdT
z^eFSxBUlr>8s&mUSk=!@60{S;58pK!>~w~6*4(|3@{MAQT!$uTh78P?>!pcg*$*bU
zUC9pHpca`6Ek-jePpy?irb1Sw%5sqCr4}K4ZCg%$Ow|3TmAcJ$)D?_+&K#@VW$bj`
za-AZ8pe<=0m-WTEES|-3`Sq+yS&C&bS#F&qG7x2M13b-=RI9}2xqtzy?9x5>BmKBi
zdTa}}4U@<o)qO?&$#V_L3w?2Tcr;d}Ok>I|!Vn1p)iP8vy=JspkLUzD1{gpYkOiRG
zWFkWU;ZsgH?DmE;1pLY`7TKtv;}SHBu}j@-1CR~ibGFb9QD5j%7z(HZ=Ss!J-=Q~#
z*|(r>{))SNs}&u*RXv))>FdRC*xpYQ`6_U~p!^oEe5&BxVpU=rVBf&><dq`)nU##|
zr&Rctx+X&t?cX(cw>ykbK=TWo2tIz=KoQ&@p2qQ0tv@udu^McyFB{7&#29gflphKb
z6YW&@5+tot>e|^-&Ru2_pd>Tx@_14putS*h0xFVd0;nWn_4Y`N=1QE)3S1xo`Db>f
zvIX=b?b-z0)y1>y@r|aqLEA$Ug-3kRX-ajGbnAkqKxl{)qaPH1{6lhP!NvTa_PcW4
z*1kK_T2aL8LMu}uANZvS5I~60eRJpF5cIX_oZ#41N|F6fsY?y-W<*+msa*OP@`#4?
z5QPtKzc6U}IuuB&Ggt^0h5~*Yu?oX}G<ZS|m~2t_&Ryh<N|Q(18^7F|(_B8l@L~3@
zhj!M_V1|HrGYxn1#b@IS_jMXLQdyoV?>|;^z14x^&RmNwP6mtrD<)lfd-D7rHI=rG
zxOX40JFcR}KR+VflV6HtFtm2AFI0Y7B$s~JHorWoN70cdJZ7cmOn@`wM_$aj1mr1v
zYM?KJQJsOgRJ<=c<0>hN(Elw0RKs0vcL}&&bQ__7ZHbsVCT8Q6CBxaF=MOCgUzF}H
zv-iT6buT)Yk87HjzZ*h3!oi&-xl)YQPl7{f{E)ony`Yn@6&p_JckKFRT~MwD2eLxH
znnm6#F{nemUvL&$(cxv?lBy|F!y{~F`;=gdsduf68mq9GUjw_QbHM1>{!X}yjM79v
z9Prsh7@VDA&!RpcYrtm|0vCcgzI4uqe@0eJta&hTEE9VkQQQAQmUA<e4==#pBgcW6
zIwGFmO|LtGx#40H+dbpvrFG?0Z5g+0o6sqpkBmVvHAcKQDUeHhJdNf0B~Z;gpei%r
zdi6)cJ)h9R%1g<8d1D1GO64g1mex~D4ci?4a3#OTer&8FnQL;SZLpikd<78>n`w|+
z?K8Ay_0W~uxMXq)y*2r8Y5l8vc1Mw5_+!|Z?PwHJW>c#_`j*{!5Ik7$jyL#3g4wHw
zdhbBjsiuIB#Xu>!5}{i2i&fH>BeLPoZ4+ria@K7suJ@LW@&#u|(rL4ZbOUI8r$t}-
ztC&s8xE@}6KAoF4=py(|^|f%Y|12^puNI||u+L`LNO)>q<N3;xJfAcWZY{^(QRbNe
z>N<XfHIx1g|KMxpPxb8x#+YpdrwNsDEp07ANb+Ga50s_3#91|AF%bz5RcmYD#~&62
zd3E3F6zFBLpiUXphmj5s%bZ9hryP3^4r|t*g3q_MjN^`gF)xhEUF?0z=NC{6v;mTD
zkkbdW0uKcD5(m}!7&iSf>TnJ2E`x#-fkNgG2|<HuoO+C&Uw>n7TM-1zt3309m1Yt$
zEw6Qv%L+CB)NT$at(zY~X$55OtaTI^+##usrnPNBk5v@yLe)1p%4f?Pu#{@!K-w7}
z;SJY$hXPqp7sk7Cju1>DU}*B<&auJUcX)ntOYJE9j2-i<*lex7ETO+#y|^(izC-ok
zx#LoF`rID7^W~;j`c})q!2=CI={!JfBK2KQP>_%kGnB|$NK!)iqC5;Jher8$s5VSF
zhMC<}-2rt&RYXVSyOi$&YLqe?K459Jk=>zS)E;`1IS>B}G8=<7X!DO12e>U8mvZcH
zPg2Jnjt^!-2Xg025nefYonxvKYp1sD^)x%#&bdQmt<zS(?<d_M2iOE9pD5}2(gBsj
zS~@}Ep<$%iPLKY0>jy|kE#lSMAG}=84xi^g{?oK)dQkgz^tNEiiN0&zUVYB_#b$5F
zyPDN}kPS4i@5StngiHlzWo74nwSJN3i#0r`o_kJ~Y2)4sdojooP{G9E8TBvnydxga
z*_YGO<Rt_zD`(0byh~&*@yzp~6Xs0%x0EUO$ZgS&>l5h8o!u&0M{;rdpPjswch1b0
zPkFHC{gwmzP`$8IQ8huH5V7T(Ik(R>FDYW$3B|gg*oYaiyAQcM5#Hj)UEGYEzld8`
zG5i6KT{loB_#hdU;qe+$W3*%5Ut4}zkUz?Rcq^nE-22o)#Uxp)rTj3;CaYjC88#fn
zudC9l=e{~g)bMlPdh=k0YQj~vO4SO@lnY_OsU+||AWdzo`t`DhSl7m%oL1Y*HrL4Q
z?rl`n*_}9H`XO9x5PB=Z>NTeUQ%)%E{)|beCrO^(S<?~_A5^lRv7-E^gxm1beJV5X
zfXQL}59u(KgnO9O*K$eaivcqILD^~xQlGYi3VK}%cWs^a&a`maRkiAO8f>(P2Wa-I
zf2lE@U_AehY={Lobmx)xLf_bn8vK5TUgUoDX4{G*tTWKF`<HCm>hnD*{K+gN3Gc|d
ztIva5x1?}aZK+zrIS=s$8n_xtYSHzl>vcQObp}hSBAxt~(v`&isi+{kpMSOLAOW!a
z+m((n{sjHy8)PhIONeQ!<?=)844}2|bUtupDIesJk-p0h;!=5VMVrYGr83JEaqKgF
zXv+N1dY?xO@2c<iH<ORosZ_oKLOd#9S-gv1jU>6)SyQNGO60CUr{+>O9m&;Vq+9j2
z%ed?sQdX2J|D~v<!3Ue2kSbCU&DXGk@n@z#1JAAF{^qXjzOFvDd1ty*voOAy&pFp!
zah@LZ4WnO^^J9#?Q5Cyb4iaFXz?uXTN>h~g_&_-7r*tbz2HAa_Le<rZw!%AaU*;mr
zIyr93a%`<An?dh19_l?LY|2a*dR2G%h~gWM*Ph)cytfl8h;N6DEBfd?)e0$hUrfko
z+)YiJG>^XZn*E3BMEcF%t^wsbHzg4ArRDXD3~&<07Lg&7q&Rb?O^UQchLk&QPQqF#
z6V8j?LcxA51m;88)8$Dg)D2h8TVgD*twiHzu;Z@n`h`r#I_kf%+>}vqXYu_ZY#dOM
z^(fEoVUk(z@PHhPc6WDkP+~;Kco?Fi%$tLzY(nSEmnj<UJ1PCqQ-Ir#!Yf+{EN(Y6
z^3N#|f2troKmS7o3EHyELR2Rv)_F^$u`IZ${7}n4*aUxvhJHjBb8eTYk?pW#&15RW
z?N^2!458?}5d{(*>sj3$rO@ZnBkyr6dt+=K^>YfQkxnuzJ<n1<ABpya?UlU@-^7JD
z=D&I40VU6Ww#gdVb+Z$9>g@u)Nl0;bzYQ?-diBWExOkh@pLj)M!9D0YkJ%-wL7pB&
zJ@!AaW?d?Ud}+K2BnnM?$d5qEtS6`b1cr<nVMmn^wSILT@`7mkWRS?qvah{;45BC5
zqwi{(5|gPk2)HfqN+Hx;-dnxeXE*siP+!Kd20Sr_N^jC=qjtr6Ma2e4J`u0o_nK~+
zY7+iFNhN61lh0*snjVg<@bc_Pwd8}vVg=>w9W;QtstA8&=4^Rw-${mr{E&&%DTusK
zyLcSsDCrGe1DXZsx)}?qtGDaMEB-43$0$}OhhW?X{PzAXZm7mqc)g>#4(*Ax2`{Z+
zdwzC=6#idOMzuSRFkw@wQUAwA{(W7LGd3;`{bl=uUa910)?MU{4y%bd;!s&vCmKa8
zlGEGa51Xw=Gz>1J+P)pqH!=&+>giEf6^s3%l^euI`TTTWu0+7+9uyc!kgMV2<CDYs
zP%H0uzBM2bSvc3=$;)>#3CxposQ*&pbE#_!L#u$!Dw*@~b!P$0z_iV@QiqM@m6cY+
zuj^l~252QSa4S<ap6s*4%+xB_@A>QArk8fgdM!^r_`7}Qi8B%Bj%8u`x%T1z?&uZS
z>L(f?2ypTsRe(}HMO}=d{zvnYZaDT=DI&CTy$mJgM>z0~Bu+4W9tytS?TqD`MRIkZ
zW5}ma81Ep30vjT#4<PNXyEN}$WdRyAwDu(F_}40sIUurxhue$Jj*izr(@CEt$Y-&q
zr~JI`r$jGA!hh*kRQ3F#jsFgu9Xi%>i=w4!Hck}fH+I4Bz~%X&iM2dvJi_r;;S9_}
zgFB7PtsmBGJ?6Z2>*4v<V1wNib(uCe;=nXkDm|UUV>@hmfX>?56>+)NZA;SP!2mx3
zk1wn+hFrpM%;(lRt}&4hgu;eK2W8%)8ARHXj99mc$T(OuQXzY#LyQP(!skLJAZBx@
zF`m^H5?{<1qFOVi1hTd6<y@S&-3iZ>C38#oFBw|00d>2}t0-~kup_B#xAR{zsJLh&
z8R?Z#wn4ITl9amt!UXy$>bJBY&j-JK_ferF6>SQNq8K+xc#eO4dm+`1TpIV&Ro~1k
zgeTd&j1n}jVAn?G>QB5RL!|&XAMWH$R$O^I%q}}q;{cWX|8yk0Yr!NC^9C&!RQ<7i
zZTIIjXUumtXW8RI%}?(l*7>FzUikml6QR1zBop;Jfx&#Gm6s3L%v&zh2&7DxD$>y<
zv&#aoVlJRCKL<aR0SX|!?VK|gn`^e|Qmv<=ze}9u#Gll&5L1{+EDoa_;Pn3pr(Sf@
zVmUtuvZ!L_G7EW{-$M1@;~Ycnvng5iDF%=KBirlBGWE}FFPq<}P*zyq9`J3&r(Ars
zQl1_N;Y*=6pR8}9#ZoKV3G<duQYK#Wb%Y3dRlM%3$U|!<>=IX#cya9B#MQw}h3A^X
z1Bd179{7_`Pmh4LUjGA|wgVy^vpqO6rU7Bn^5KlsK9yz9=AaLN{752C%1%#Y(-K*v
zCRm=Xe<ElZpdhqn)?%{&3B+G2DLL1sK+ykL{IK>!@xLLH1KNBH|K4aXZXagfUZQ8K
zWWfd_<xDF)LHxwQT8?-(!WHo;cg0})bZgL$K-VeSd{5uM3c)`*dPzRwv};9oU_~e?
zruY-`aB28=+}ojx94_7+Ffowx%utj7Sk*_&nexZb3J2)cll5<2VOlc~SEzXDExmd`
zXG!G0w^+*zkd?)(yGiUBC(ejTGNhw6eTViRWbag6;?9)kJeaii^1tu65dNVwtlWY#
zKn0?7`xk!@sTd3bWmW1eYrTbp%93BHOk~^>8`4?*wk75H$vox2=HvqXPI}&ZP$MzG
zFm*u<)?IwUZPTxDye{<<d#P|!5>-D$Jo|zIY#r%@#ETf>uT_}}4X3#6CDPpRySjMn
zJD+>0%NZNBBkfeXe;qtEIOw6q4`&4NrQv2~`6^U%_mZr(q0oLj3S=rxRcN5ZYR~!d
zRXTYdf*i*8x?FM9^RUr}U&;RI+|cJF#u%HD{iXcsAtn~JM6DI?%3P`UlxSt+1pKr<
ztO$jz&fPI5I4Nkep)befyXGoKnPJMsO*o&Az!_p1zMth0+GN6aODP!Zg6!!;pgCRI
zo#RFba3f<%lB{8~376qIyRxOL(Th5^d_ON)ty#AboWf6n%p%^8dF>@lf|dFV5ol62
zPwolUyFPnuc}5AMcn#N{O<)X<J*Jdbj(5J<8BH^&xZ{Nf4z_y<?lLu|%MVkW$mi?+
z543wSJn3o6p)Cr|xh(Sz_itsIQn;gG(L3lZ+I4zmOcb@iiauVp&SDc7bievanGP>n
z#<*E)MdIj%7#k`%(+@LzGdJqnDW(|(Tly#b{MZlrml0}6C)yW?|6ZS=(+~I8dQkX{
z1{$HiwgW|$zdKyy`jLGJaTq>q9<f$Q9C;R4>5eU5g(?=Iw|`?>Zk3KPk};Z+f62%Y
zo=lgPQ$+odJT2JJQoivEviAEU_rU!p!$S{pVw86<Tz06D6U<I4H(rznIh`rw3%^gr
znRf3*U28CXVJTV`%{FxwWvsw_ugmt5w#yu;{ZiG!napK~mA1k=nKAV8o!z)Ld-_tu
z7`UaAY&HhaY3-k}-`rn5$yUtI%&2-owt&26#0DDjWgm3YP(SmVen>-2VY67}%%owe
zKfNCCO_p_#ioG);b(@aAAli!oUOX3oQ7v8J4Ek*sOBSF%FxZ}?8s|OY4m*@9b`&xt
zL~UUM<r|L+lFXC(5K^IAN*p-@tA)pzDGpt{B?lEn+fQ-hh2;Z3=6L^FbsD?EnVT^G
zOda7y(B@E&Vd_<F{8BzjZU`a&hb{uJ*%J1h8wyh)v&$aLl)3I~@7>j!pVREJ(kMCD
z3kiIt+pi<^QF#BMmxxw8UDr|vw|0PnZlwUR$KQfz`i;Qa-7Z%qtCj+;YY9Tf<6vZN
zk!?0tL7T-Kwt+>kyHVld@*mymcA#HXHpN<DimdOI>?-aCe`heMJtA~A^C?UR+pp)V
zGw0f8X>`MmbJ}iZc)7@Y2~9X%u7N)wD3EKjgz3314DOge7&MONi7QaG=h(Y#e#;at
z?((oo08mBwbJg|0mYB$#)FbWkr3scTXe=-3{VZ0YA`W6UGA3r=pRFO1>K-R|H$MQ|
z(RuY=zGnr2_C7+d$Yk5;V+!LB1)QYljN?zbGfy0^G2QQXWTZ<d_^3Y~0}5dnScH+u
zr#yNHS!V=+76?sNpg-bl$MDD_CSmIF^J!|NhUE_N9!3?PC&y0&kAeKlaFe$`F&@Az
z+}}d89y(t=y|VL7Mfu^9y}Lp(#&atE|1tK~QFXlAvS@$+!QCxLa1HJW?(XjH?gV#B
za7}P`cXxN!;O+~*o&3&uXYYOQd+)9>76blZtnOaj-#2Gf%_=%pW*WM(+EInB))*Fy
zbDEjfZrB8#&#W`8)2*M+ArQvUFk(<KkYe$^s@-**Giv64!}lEBzCL9>ZK)YKtE{eJ
zo@hTmKMgF`4gb*q{_OdC>-P|Uo}?^bm8Rf|YG-mOknU~&ZoMQil%S_Y|7hKf1FF1}
z;j44FmmGfH!wN>w_h$>xi{jU1x!&2y?Y14wI%=!a6ZqHX?!+=pKv0KZ&c!TGE1@(q
zZ`^GOzx`!fz25x^(JhINMweOIfd6}XI=rqfl|N-_q*y5~a-i)5%fO;E+7x|5w#(7+
z;^#)_c8>Tc4M8k}^(WSFr7tokqJfxnsgg;;-wA@0Uwnn<0C)6Brt_C?kFPS37SCWl
zGVX-DzK>vCV-9%XX&nUIuHEX_nG2`;Kc!|lO75vK&7yw_AtmMX9AQ1Bu~fAIeO#d&
z-@fKUx5CEqHy^Ma`?d!fvnLzd8;?3Jd<XZb>YT*wGAz-p$b9Rm8#pWVTVk&Wxe}2R
zRbhE2COH3T-jSz~JQ~$sE<hc@I#ccY26niZH#ndHwaO1`Ky<jfo3~V;ny^5*Q1hBu
zSp2YBS4W4#JZT;_vvd;)&m$)JV_ch*lvKVgb<eT$>(O$Bp#jv-KI~Oj6a!uSs3x=x
z{|fKRM@FIaP|Cf#@Lxg^Oy?q+?_I~sTu10KGE`v<Jo}#3=K3%aRZ<=CuHX2U6GoDz
zt}_~PT?$-V3^zmLWYM=&Ghl=VGB-$YR0<hypuOIFZUye&qp0`wNn%|-Ur$IfBQKe~
zmJ6R^eih&6@;KZ29`(C}elvZRKBoFDGHpD-(|tC+nQEgqtgxjtUgrxYpN|>2i8gX+
z>*khLtl)+dkJ7A?<iz}qu~6To885`t({4(csh&20E&OSe#u9gx(Sv+1Z<XZCiE6ZM
zZ@8eiodqv)50^5#O`Xqm@c~aIi^{|k!qlO=SGVz6XFDWZB~u^F8y*v6fMc1bNDT9}
zO@RK4Moz6Qx0AIh-9jbgktav>qbQZt*OUe({pA&5Sj|Bvui3UVCkI~hXSc(&SJPsZ
z#G+H8p^v?u0`Nm<=P>gzG~rgcIvOz({F;!h?9dVUnEX~#FhTI2ri&Hos1wT`aDJGM
zeeLjra7?*mM00&>jx;LLz6xPJ;<Xc(Y>nwo;*%C}@`8m(0{?vNhSQbS%}DKrDZmyZ
z2z}m~PcPY>QNpR+`pu$HNrhN32?jQ`J#rA;>=-9Bpg<P<={KK9eJ^u}_n0fA`TBs}
z8cO6Aa|6j|nan8)QcQWz9!4e*#m^Ux3@->}FO$lmB?MAUo$gw`ohq*h&Q&R~2`atJ
z1pk$N-=1w^UgTO1INNc}2xAD{Z#msp@0=gKpPhkekHH*=???o_Q<nE}HaE}fVC`$<
zhF)U<pwa=gKx+ii*)ccwmg-p47mE6^7j!|viSDMMccbdu1n0^#1MFG3LLF9WF>!)W
zlKOSn`43isv-$PDWP+&YFxD>|OgEuHg>dA@<LAlj7oJA0Sy3qvn%yfJVCQ(Z1o&zf
zX(Yj@yb^)**?Q$s(l%{H-!B)MA~>ZdT-=&6uRJ3#Lo(kaKBM^FN3YVS&!*Ws90|*s
zMtyeX&USqxG}9d*JvU-LBtblnj7!BF1D6{Rq&k-{meLh5l4!nv{=mFHHc;;URAtqN
z1h`;Sd5zTc;}U{>O>VFNYvy&uq!D%Linu-A?kmr!<XJ^!kk$;!Q{<*PFzUwD8H3jA
zQeE48Szt)!uO!t$W`0((i0~u!j9oT@pC`}_!O=eM&Wm@r3o$sb=*hPwseCI)Za+a!
zUv~L&d`EDReNE-Xf2#=p2$HTXwd_?7f6q)Y>*mRWIp&U~tSD<Rmj2bPL`GfdmOkR~
zT&8Qj5emD>fxEsGzW;bGebMm%XDZv-(u>O@k|lmShS~JerorZu4Ttc$l=Yifu-x&(
z{u^K?!{r5HzR9F)F6D;4xD92k#O1<5^Jst9moBo#?Wa3m(FZteOle)ej(eLkAW)!w
z$MXJ`xO~eHv>nM5UO<-Kb*@=qiMu=Hv3Xo*ty#?<cdOO2JC#aoYj_v-;MT*O)iEyG
zAC%)wiRiNcBaGAZSUP7%l`Pe3g&#cs;xk=q@pEN3i?Jv{G!Bu&S+nO;1ye{fp5-7e
zed(6MbSMOwV!;nwdZ^sLKNfK5J>*t#>4W=m>GQw7pQbP8`~jJqYOr9(#5ixqgpmK7
zIT*8U*+@n>&;Wt8qd!S@V|xUwMk`1tHJ$`APwhc2yKoz-9UxY%OaW(HeQW;CtM}Ns
z!IH1rc3x-i;Gc+bmHCo`mIBhyn>H(pX6^Jh74*mn8+kww*UPT`{*r8kmqR49I4h{e
z;UL$-na)p_lmaFj{dhYYUZr?iL$F$(r)Tnh=EUzkjE!Btk`Jwq%kZxhEFH$@cIJ`A
z_RHF;G3a`JFK|_FeVEr&=o+w?D^6&HAcLZR?CTWnkfO3<ayAIpWIC4c5n@PGVQ7nd
zygg3y`f5)HMIWEk-AZ9GTN+lZEcx0IrUQ{QD^{9zi8jr3E6o13e}ye>M<i-e{zYZ;
z3yUkgS4@>Dv7LNgKPmm0ZThn@jKI9<&#^(2gSB$%aK)w{!j9scvoIWOMr`eEx&D?R
zRUiYiYmyMU=DhU75qh1*Z|;sJhJ&48(_wac6<;FLtI$_uHiGCq1|tZ*)h;e}vSsBd
zQLS|#$E9vWNeK?fWB{w6lR0KK9k<I*SG}JsN`+)Tc)~(TN(N#=8-4+~#3<0w1Ow?0
zy1#agE0U5cg9<Ba&%-LPLSFWHFKGkmJQ7fXqL_W4_^`tX-Gk|Hk!&|tjoIB2_lj<J
zdXpx8k%k!DfQCR0K9nkD9g)--8($(U=-oRqgN#VqXXn!esczwn`(BmR7Bw(v+)3#$
zwtISzc?|i}xxAu{9b0np15kC;nc~gpd~Of@7pSpqR|_do>&a;0As(TAy_66zG8lo8
zOJzn1>t8e0CVaY!KJZ_6yeB3val;@kQ7Iz1Rk$e8(SF2augC+${)O3174%pN<pZ*~
zy&HtGxS^q$xn-K4z|G|wB;HSFLw~fCove}Y62As{3V4b4+y_P*^=)3T&s6<x=iI(8
z6<F+^<VmbxvYn>2qTPW}-J_F5=KcCKo*t1>(}x<q+d=>J{K}N^EhG8><ryM+gZlY~
zD+VnGT4<4E8}m>>zbhj1jn-;YyD4>*9)DmsQ%`hlTY-2N;0X{P5f5tK^M|xn#8wIm
z63--+=`^9v3_u%(?FAG!XJxgj^o%0vL^T^K3=VWY^Bv$h;17vp1qDcakINCg&h-mD
zpc+o3dy>_km}h9~{gRsLzI-1D`W}1^!;7LXIcOI7y7hiLro8JuqP_2mCrKFO{=~Vc
zbQ@gryc7OC?%P4AVQ8#rj`+(Cwv3kS7m;!0o|-$M(`X1}o<X99hKcS<qyoL2Ii-jG
z@8T1F5u7LKTKW`q=cL3BQ~}J>-;pU^j=8%B`uuh>!iA$sHCLGOJSnRodMe$dG(n4$
zzcG{tZ;TDF<34CYd2yivqYPKeqA6YZ%XkM14dCNrss86OWW&c{;W1!vC{r``y5cvw
zUa|*S)E0Pq&vvLwSGNT;QNudp;kX&UxX{8E)<GXg`=ta1QyuNB)^VjZ>P2#<_bXjq
z$s+OYP?~F*rfG#MLu{Yer=}HVTHk2)WV9j%Z@xc`)vrTfg*YU{F&7eJUW&`YX`$Lt
z2($3Wm1N^VB#|hR(wMND+^EAI!$>j~=qkEVn|a9RN$KA06e~{~zd$jFB{ew;l$nfe
zeKS<6=EA>U8u=R$QP^8uc*!&`oZjsR&fv*(jbZyxkkFOwqHDEOi(Zrai>Ha!G11_(
zyuHx6G=4A-)k1lKDkaVP%+!iA&T(B8^^kXlv0^iZ&Bwr_RH^q<-hSz5z|k3k7CHi)
zFUy6ZQ336bpV9o0O>J1<S>rt*_Lt2Iq4_Ih_WEG`8CFtf@@uo0l%}&%LiOmf&XB9D
zZ-x17sx3YZHZJhlE+gTLW0tu<2oACK-Z*?c#`8sMenPzsI|*teD#Sl=vQ*1un(R!4
zy1K3RE~bG*ywnwAK>(!QqB*2$L-M?B9Y0$``{GJcRqVf)@xmH~$3f^!k%grYl)J)i
zr#e$@^@fVO$L`ItE1uziOG9q;t*7w;d#+fE@K*l2-NPWkA}4=K#pP|sTcYNd2Uv{X
zDrcj#bi}8KgoJ%lj^fi?f7B2pB##Rfz+Ci?Em6CjWsK1$hy2z%b6Pz^b7G7(y0Fn%
zgY7FV9xsqw>Wapw@!8R~zofjp*=*%6FUze$Tk9_!-`W>%I{;UFL!#Nr(cN98jjnPO
z+sz-!#*CXuM2xxZKfDCD0F5DJmA$?2`ueKz)?I&>*}N^mCL1b$20o%M!7c{Z%K8n^
zL^Q#xdk<^|q{_IzO_>Pktv{id-5Y+iK>yV+IP}9;A}JHl$>r%B*tLTw$2Q7ya!7#&
zrfYZ_6%!M4@$R?+d+~|%@NT5xOf1QLc)C^RR;SOOLv%K?9)sUksjX8yFBw+4)zzy6
zn9#{s<Dak<)2z1<j>2W#+&+Ql9%EC})TD7)X7O+l64p-SGn5kU?@1^0%fDI=i;7CP
zV>NJjr}(jcVSFCb<&JYLBH>R<XeQUu?!ym#nBsD_>VBxU8sFI53{VeOSF9vGC<Qf#
zI5=*xiR2kAA|fK$9(6}{c6PDnzpPLwNTh5$+ge*UkG5eYX~vqO9y<2f4ED(7rPK-+
z3SrtG$(1blzd93+aJR)&re$-BAhb5Pp*%h$jW?Bj>$U$4RR1WGxnWMxhUS6SrT&a!
zo8Ua1by*_Y8N>xs7MHM6$8xV#nOM}h=^Bwb6=PpD-ts`NHTiZ_{Y<CKIE_YYo$H4X
zj7I_ptuvjxS%Gpvze}Dj;2!ml{ed*I3cASe$c?2h<_IE4IAm)&7oCuHuQ#gr1S%b!
z%rs1P#KRejF{|y4cy})O_Wl`nAEmi|)>Ku+El_G&SQN8JXBVNXqOZ^(#+R1X0+&hJ
z$~FKBVe?_t@d}q@ZMoy^8~TlgW|;oNa8hbwc(HPE_d-K)Yw(~Qb|bcmk6#3~2$@HH
z>trF*`sq5s`g+quc7pe*===7UCMB8P)>mxWr#%8C>fcSQ5J>|?X!bsH@8LhRhpWJu
z7);X4ZIQt?70)0$T49hw29vzDr6k{*)3{;Obe4=mB$cA~_G@*my9Y2@+(LF{V?dRs
z2HV_H|BMv8R7^6ktw6885tK7?0~=^h5LjaB*uDpIjzBj!;I%`UNy^X&IfM|fQ@Msq
zc|z(<+F2LLxEiM%))%$$8_B)gc`MpdNbc;rYQx~*X|eVzpf%kThzI;E#bspX@zbA%
zE#syCyZe=BFx;2nuw1`xe%b!&+k5x;@bF8(eFo&Bb)uRE^*&h&iR3dci5<!Seg1TW
z#0hC13^oMTXDBH8-~`=G5-DtMsEYPGE?ia{+6s3q{OAdSZ~FY{&|12E7eR)XCcE9f
zV5Zc55nJTopB)9p&T1<47Z{bAYQc8`Jcuh7L=TpDKg_V1Zq9v5t9q`P??cpdqQFh_
zU=<FrAtGipL++PKkI8^ho^1ApuC=iYaU97fcC-7r=X&5rFU_r|6L{I&Uovjn83-Oi
z=g(Z6<fuG5YwTZLz%UY*5^K@we*8~V0J_!9qCJO`9{=8GD%Gu96Xp-fEPi9++j-9C
zcDv8yFLo%=ue?k)TgfCrbosq@yjt*L{DR>(1GDuD9|gmm#FBWmCi&2zJB|bWU-x8>
zw@B)9Zbuk$y5JaoPv7#xQ8iR-diCm`aBAS%N_)U5M<B;lhrm&xcs3qj{qv4(uL^;a
z@2U!k(B~}+iI@j!S^MWjrqw2|_Y`*o-E;3TMn~Gd?WImk^pB<e)PFg;sHFFJL%Opg
z&MrXxH6eQL^7WJ&<MP$>Tmno=uOfzZq~$=a1w8ZK*3ZT;y6d<tNj~UY<?rJ88BDoY
zrLSeJS^olrOuOL?4oxDXzmd-<^>Sj^iYb?)rXNAIMTIX<4AfYM)|B&8-k!Pd_U^4U
zflMo{NC+)m-00Xk=Bq5{e?o6U(L?;;cz;;TKW42`|EW?S>610(AovKv7S@kQbRp6$
z8-(}Ue=r)3>6~bAfuEKzSrPv^4(bm~Jfg^gGE(|XbEkR!`UqH%dI(rA!)3uE^}4=&
zo$Ey64g7V%lGB8dmKXFY1svITwPm1(oFhmX>MWI>yyAODS>IO!0|YDul6ku+hJW6~
z-i8QR+5RJC!4VvLbjI?}OGXI9RGdS<>l;3@YTZa1VWeX&i(xk&sEU8kk`WU6X*r;p
zw{jo&21Z0gRFqM&f`@#tfz^=%yRO-Uqo~m|!p=hgft9_=1bo8Z;yvgl$`AOWb}<_(
z&B|ZVYt72rQzSOKiG^>ZfSB3rRq$yA#a)#%C0E<cY%?`bj$c+yZS>!ihEPg=c&&q!
z-+Oz(NFLDQrzNTt5=l*p#D{kXt8=xqD;<HsnfX(5F$XEx5}kF1nw1V`=u#Oxm^s0a
zB!v@lQW-7a+%nHmWC>pU!hptZR;M%-EwXER)W`s)cMwOi>hwn2Xz2dWrOj?N4>>mT
zyoex_b%)<(f`y{L89f5^<4x~Y<%s5uXcZY=4vc%JD&_=#fVDti;En2azu5FsQZ-pm
zU&3D6H@qR>gN*Q94q$GWd0h%ap88RJhTr9iwr_avO;ihuA62zeV9=i~tk$^63|N^3
z#PXg0ngA(txU&ns!a&MVY(Ed99Lzd@LdApHim3dgc9UTOlNR!;jJC>E2<s-Z1&K)=
z*6qGz1{tjiX9Mb9G{&V)evaauwg<w%EI>3dQtPC2$*J&UBbN4T50NrCfIi;56v8#P
z-ZF{E=!8m(R*=ZURM5ozN;di^Eu_S;{X-kCI&@Q5d>0M7aguqY!^hh^_#+cWP2u4c
z8HR$4?U%*Wr$Qz(biNzc?+eHCelJf<*Vhwy_Gw?UZhEjHI(q!Y*W`EBsc$T<nc7z!
z12$aHcFAAhReXHV=b$15&Q*`PS6<@?&E_);7pxhf+KoaqS2{XlH?vR@eg&ePItkgP
zMrM+dOmP`;J+jdt;Q<+3DFm&!&I7rzF@6u;lkC4azDG3<9mT^fvao$HbSauFyho>9
z_JK8(t>_22Qg1)92TWb}xeUVulUW$`o%B5G&s4IOljz{E5ipzj22JK!Ud{|txvd<N
z<vZI%$OyIzf~+AS?SwoGzWcGD$ipaZW;cZ(5+&Now*zS<p{;gb+w;AQ+eWtc%F?b_
zcbV%JRFPLswChjMge`?9z6Wji1eB$*??fy)<|2!bw9oO@!~OslM0s168hxt8XqCX=
zqoNa=T4nlCu(h+^w#ynLx6$*9>U^iCZysbeXAo)rsuh<XVlt(rkiCrsDCD-=58WIW
zJr*CGJE5}>dOV#88T5+VCNzwG@#r9_N27q1Us(s`lw>y@Y2r;X{ABR4Me*NAbPbX%
zx9xk6pEd=#)?VH9P1mP0_Av;_GeVn%{g|sh3fHf#%tT)!<O+1IC&Bu;qP*{?j9}Gq
zCa;B8^#{rRdZ8496BQ=~misPouNhZ3FVjdNJO)emUDDZwDyjo#_zhHu2=bXqePf2%
zXNps5d`v*zN8>WJdc&;7cf_tlxP~;AHFEl5hsMqKgskYg5zzEz`a$!8(5o!sl^i&^
z&*b)8BiT*Cr1c9$PUakzd=Pm8br0xc82k8-xa*|jJLRIfzS(BwZhW6Ij_ndjAQ|hH
zxTdEa!beqr5%6ZE<_p7t&9l8k%gt=(%Wg=kCuf@1W*2G4cc<Z&Nflkc6bx%mR{_kc
z?d=7sIW`lsQ3{uzkW@4#7?{Rx-Rqd0;@{f&=Et$Qj<<1^9LPo(R6*PW=;;bUPi;!+
z>WgqMn8FeZqq-?kw=+7yX-=oQ#mv7B7OyqQESIDGXf7n>2!4C3#j>xxr!NZCx?$py
z;R(YsvF<58Q@;8jKPSe`0STtvA>z`2vB+IQrv>+rNT;*;JuJDb>%bj^@yt=js^#Eq
zGv>{PD0u9{Z?3^QX}v3(JUg5t_N!9vxl=d@lwpWGc@4d|;Ev~A5G7jiXcJ={bH_Kr
zvZ4i2-zV(r47w?E^NY}jF!%F=!$E_<>Ooz>1ANg>bA$%<Z8GnF;s^#3fu&9ua*s8k
zn^KV|R34>5yb$*U8|n;#OC|H*jHr}#`kTL8dc0(Z&_NPs(o|FmGty{c(Pg>5KP&A_
z;?L+2bmv7`>RhRycX;&S=xz{Ek5627uktTxbB==m%W!kU@$})zHjfu<D?WR$iomg=
z0d2WU(^#+B2$5ko4hq~UpTM_8+3-SKeb9U_=~j0Z*^p=0BUY;3jKv<wReb<fp26!U
zHCANUeMrP)c_`mI-0(aC9&JNdnY650dHQ<;zg+y&EQD4EP-oqj30wt@eb7ep<9JB)
zP3h^HBj6PMxKOn}iZIp|779yFMJ28wL=h#LbaQj_MysNtqVPc>JrA%8c&imIG_W;A
za&>i8DprGeeSMuiHF*f+>+c6LgNH`f<bKtM<j48Z;6-o4#sr4~-i4m_$e2MvdRP9R
zl5cGg)PSR+`GX_V-r9ii$Tv9P<oFyE*=F|n!DXxX^Ws}wpO=~BbN;MEjS3O9ztQ?$
zJ1#fNE~_uwizsFO0{iIWjeUOwtAJ1Q%O)q?wpZB`amT8Z$yJW+qUaXfp}rr6^*BDM
zr?FM7XRWKh5ez?WJy{P}dC7%uY*2U|($mv(g>Q=IboiR%>9uif%5C~I06A@*utiI=
zM9lF)?DaFJVCty;s~STp57Z?#Bm@?TaLrfq`}A`|6nEUDv&$#7Y~Q7x79F@+3k3xF
zrai17(D7<tdcG-ltJXwv4}W#s^8#f0kL6wYdD5g**cb<fnAB-dVHyXDv+DIsa85H`
zRE+{DYKL@MKvW!3?yW4Nqv)>3uETOMO(W)y%o(<D1<Dx{c%Su3RN@jggUf+B2k?@&
z!QN9Etk{orBU_Lg`{69Dkzx;046)=2w94WshcsZOn2^~c)7M;wW90rw><#IRLU@sl
zElrY1#;YUGHKEgFHfO0T5mlk<w0}b7==*kB8k@x;GTI2NO3U<Z^obEHJ)F#bd>^K;
z+mJ9K%brwpa;o(z`$)gFENuuxFr*|00wzMo&eAzN8c1FQb=FZ2TWL|Xy-#94DloJ@
zXYVM&<2I^+K)+Jh$||q?Lma<ai019+=f`zOe|G#Po0BxvZY<rnuko7)CFyN6tX1v~
z!O7~zt&b@Cdv83rIf*efC$qOP>Fy2JC<2`5k>Hw_JB>#<2RCQW(2{tuDLLlUwa|N}
z<u5|Cc^NDXmh)JqC=<k(rtT;cbRG?%%kg=%bn?U90#mnCIE~0SU$wOGb!^{YHCQda
z)$y0bEuhAo^=^Tt<*Z~Kg%XhZ_2gOxOqZEzrFu8oNE_fNZcWmk>1y?p-CH9!rS-8f
zc;JZB27U$H|JaDy7X$c<``<3CO4OF54GH4jZ}02NTC&ZRyDYo3w`>bm1DHwL6qP<o
z=YBaft-QL<DP+?__vlc-TEC0I7gqDk)@kU%27M>6v3>&=^!F7}3U&yt<RRWg_jz`}
z$#z=A6P=Yu=?Y<q9PGo%%aD+ah{!J^Fd@Tax)FO%%B5{(ohSaK*NAKxM!eQ9!i5qg
z1RaHUCpsu(XE`sus1D2fI?Cxm3G+?x_R1Y=@e|3H1l*XGg}2`53JU${#A!Fr$Io88
zxiG7dKJ>r31LB36v^i2>iWpB^P3I{U8%RCw&;9P4d9DzPzteADjAIv>Ptm?}4rh!=
z?J7Lqy6|5DarVKm@})#?vQy)5ek1)%o(DNCB=IAv_d;6DMGumXhz+Aec3|X~NPjP=
zpFl(6JmUndq>70o>4{PRe7y7(5ibXBV}OmH1XNj7=NBbCZa|~U*B+YPlsztCk7->=
zLMZv<maQ7$f<2Yfrdex^NJZk&&wIb>IBlF~;e*ZZ9zHvsz=f$QyB?d0Sq3gQl7Rso
z>W~x79XJ^$$mH>C<i*L4UNWyMk}z)jQ<xPiQ|v&P+`vaS#+fQx8{*t+8VUo$Rp^(9
ziUhtvIpIJ)L$Kk2hUpEE#i9~p9rk&c6Or^Y_AV4~PKpX9kEnuSZ2E0$mIu38zK18>
z)z$TnvJ7Zc@Kai^6dzF*h*I<m8zb2C+?}*-^dqJj?TA17<FZ<A*EKZjUDuzQt!D?d
z-iaT9BsmCubZ0yWiQd->a1x@U><*8XQ9b^@!l=`y?A#Nr&;^wy>*PiPH`pnmessL(
ztUY0mXN04$CqX^|qI%u6OKP=O;Lpt+0rjkF*qu|%A$~0K`U$Zq?FuYebH}Aqr>D!6
zKU5FHQ$Vus@l$uFbUroF_1zv^{9Fx&J>2HJ2?M$w=Ch3s20CDX5=t{LmDNHlDW<xv
zu1{0D+2D3KLE&;$>pwWgw8X9WC?QDW)39X%=#Or({5G~(LHa~8UHv6RS5f(W-SZOa
zHHU?0b{<1kUY_ER3y6l7eUp3P$4mx$Q?R6%bY0o!y;r35_#2_b39<$f$K-C#n<V(*
zGC}^LD`<!45a!w4)G*B6VZVg9_3wQ|K^n)ef*N=ZW(9`x%8gen?z|!21+o6L102pr
z=rS36Q8Nadj^q=&!wEEv=oooFc2^&+If!tmY`c8JGU;^I*AK2QML8;c-#WS^RbXwh
z-Vh=}ZoW)z)3$Qj(j_8zfkh9m#UN9~^Lnjxge7iM9==v2bKBFA%tSe-T0@)<&;54U
z;!wE@bl5|ZC+6dU?_$2|DeBo*V{72H;G*;`%t?u2e{S&!X6R?_AG_VT*s>(8di>5p
z7=Ey%=npJoFHE~B?Mh-iRfzm6ZPDX%C%ewg6hZhR@aj@Qm*7~=)Y<o=q7yD~b|O>P
zNv0%B77Qw5F#Vf4M5J;W;7ktv3Rc|SSj{PFBXUmMZ_VR{{~tYvX@z}>O4($wv1BV~
zJHo9>ID#h$3IX@0d`~c!`m1V9{_{zb<=at7a>u5nkajZ}B!hBWmf)O(H4~!|l0sT!
zSM$y>V~(ZvSaha2oP1oPl0exJCp6yDzE6K5T0gagDIzz8YLn!-HyIhf@kC-`AV-lx
zIjNTxKKj*ZC@j^!s)5s_>zAxgAkN?3KR#%_?ta6yZ4bFK7(VtqW;8!=NcZXVp}#3r
zQ1~cq<R8y7qq=SH)zO8+yM7l?*DA44J-0M{LFioXMa6JK)<f81ocEFKktEt$JDkD8
zwR2>xU}v)o-iZ2kLnI^gnnv`e;zEL$6?Py)_wTcz|KLJK&;h*||64PCp(4HB%;Lw$
zr96S0b;U}vC#rPbHr$<!qR`_Fc##RLm6S<KEf!$RAJph-`&{`bs@3icos@%Qc-k<-
znENzRcxpbNFC^47)aXy3d(%~du+d#;?E8_vLObveWE{b4wME{|u+FBiX@v*??efAF
z10;Nh=6;gkXJQ15e}MKhaBwWvY8&f;__JSwi10nzOx9b5h8`J4W2+_EI@Y2^Wn4|D
zmN9(lP;mz-zmmVC4$e}#=QE7e)2%=gjWxL#^TK}`jymwm);<sq=F}Jt?`&N*p86?h
z8<}yBCX<&K;LA&7d|T}%z?ID|Ei!=;XGZXuz&kA=B<jv~@PA`aBqktk7~edqjI{^u
z!qDtE^l&j_j^4jXl&3B%;<M3X$hDll^J>$lUyg4I3#fVgcyaCBsOrFsMM52KMMkI{
zg~LxESbg*vF^IM|dc=KJ;1LOKQB4Dzu9xL6pugRc|1W^P7(oKv^Hk@Ax`rvwpRTY!
zl21~fS085Y#u8RG@3vX8BUHM}Hpfp^^5*5yZaoWY&&BR;(r4qI{VuDe=)w>sn}o3m
zxF9T)zWEFos4-=I6@1zk8bRp~KL$bYXpQm3F~p`e+c5<EBu;l6H;2oxbt8u^7S!5`
z8_U6mf<9liN7^p*%ljn8^b}7KZjLp<_D!U_Tq#g@ZzXcG9u^`f(3rR}^)j82McPs|
zsjZ+!hBFv_6in4fP~tXZga+4JXdT3aWUDJfGMovCOpD*IMoWCJQ2P2snqlV_^WXUa
zG(3#&HXEJV(j(O`l4b|ytLJ00mmy3I_X3+U?$~enMEIGYFJ7T|nmv_=sVq>#ACy8j
zI^O3b5`OWg{;#?Z5^mF~j6)PIfh8YD42>#fX!Pb&Fsk)suV{tp9=s8(fyav<0tp6I
z{d1@`QpV;+xeuS>CD$*NnmxNtCR%lZ=%;Wgh#Pj`+cnbd8=qV`m}h=R@7~^?cww=I
zW6K@tLN-sk7J(<F6)JW;rMz6hhpXBA%b`;fbaCxxrpR|MKfJUcyVE(AZr0Q4Ls45{
zIQsGv!ov9#*}C+zo%;;2*03c{85gGS$!drci;$e0r5nyAyx4z`AkcQ=jb6VIPtq|Q
z+)TGVJhcXM7oF|wJ-mOAaIP)aR8eRCf^wKz^nO|h)&!T?dq^<MqfXF0pDw)n#lnVA
zr$3V@#te5?uA<nJ&%EQ`x*f_&(tN^wT?WM4F~vo+DhNf5j_<zrwgVD{D}%}w#drAq
zOKL7IbkruwO5c-t7r^~U)4G2-(6qi9VEonz;73HkW8Z}zv-idj6m1KZ5aIj4bnOF4
zYU&bnJd~hT&&-vSy$OK#>)$F_@8u0ADZLKP$tu|W0g8mM2K9g&(HqZm|HWFpe*<)F
zfD=z+U3$<BK3<2Lla~Np+%XFQAt<_aE8+g$PE}TKVS}@xS$O8`z-V+KNsK{?HJ1D{
zad_HcgkV^Fq@N`HEc?WEC*&0CG9hC915rkrx`qJm%i{%QbmokY<WKa+4Bs{mCD$B#
z6$_rcr-l@6>IIsl85QlWy5#3$Lb>($TdXfiT6cxB?(c&^!G_D5IBIQAfhKRUq3K`g
z)WzDI5jz;tBgOf(0&{$2BQ-goY*pqrW<w!OBP6u*4VFqhIM#jN@ohMOg)L`q8uZs+
z<bi5Mu_yV_BlyNDpnqlpJN|k~&ZTlG=XQOF{%iVDa?<f<G{xYUf#Ut)?zK>D;nCI=
zy~Wyj%F`7@?G@z_<HaArkA%^suPCNMEu$;RjHlKh4TlM_PccP68I>>mmDw9dpE9l`
z4G;h1RA2w{JcK`DZ!g+!1nqzNqPwhZ4|hADEwahBN2AC-v+&WNf@gm7kXQfHLDb8n
zC1CNu&SLNqAYpLlB~+(`H3o5irM8#oxv$et=G;P1lH!gSix?oe_~gq{p-WWZ^BJpy
zVpez~7Wi)U7?)MXQ&L1wM$_eLo@lKWS>cuG%Q`_nbv~TjqIkF<b=QvJxH%7*(7u(j
zwni!^s|3=`JP}jgd@#7D?^@6u@Ac<NT2gC&m0f&H5R}msGPM6@+c1Q!d;O8`m<dx~
zvj@eQnUHg!U0p2f6>VkDC;Znh)2&sbcYCE7(f7sTJIY}a9~BXD!=xfnO&%3Cy8IwJ
z>?z~?ne5yJ*rrj>b1q0PzoQUjWv6%cy}Jg41}Ot0AJ)-><-2N-tKIRr+#=jK*u@(i
zZ!=fRgC&fuf3sl(Yrs=(uY#0FQhJtipi-Sq{JM4oE43UA>_2X!-<jVrDDppHS3u5w
z6V<teiyTIkX{VdAUSt`4vw^|%Zb1Kka0%~oU(-Hj@VN(sWMtfEf22((kz0rI{)!-j
zy+B;sn1O80^K6SF@zNCL54Lku5yk7q@^5QYi?|E#NL6hH5)-=I+EURxU*us-v;A1&
zP?M*nbnj(0lWr}US{x0>o?2`OB|W9q($JtjF+9Gbt2@?(?ZZ*AoaBb0mmL&(UgLIX
zu~V?^Cswz?OWFdSZ63?`q7{_3r~5BYT!ikfzLL$MJ=})5`NB2tW-s-eG?lh%{NvzI
zc%y_f{1#o05$UnqI75VO9|!rZt@bJ}OQRCCpA1e;Fcv-b@@u28kX|AoXP~xERqD1<
zRKIFxqYfxRn3EWOu1aa=SvJhy^QK#0ai<?p*%aJQZ{Y@%nmVVCYu@r{2|KsZo=X&k
z7a#M?6nRX^bBEVhql(Bdq=$t@&yTC<g=;N0Gh~;4W`n&NG`u|0c+_kDoawzEoXX*F
zyZCrV#PQAh>3+Wn|7`~;kcc~fMi_aAW()`n9oHBBn>!2HlhR(83}Tm`j?aBqAWH8}
zbx(_hp_m03kwDeoOln$q0cabQnfqItD^umYD2E#jL^Cbc^~IK;`OxU-b3w*Rwn9-Z
ziBhwc1?Zt`co-IQ&#UDxxxE|3>>X!w312Ei_4T@<<Uh-B%AAneKeWU26+rQeiHnOI
zC+Q^GPSHU_LvwI(bQjg4NpL47(i7Q6^v%y>0v04#Heo-KM9T*TC_YkA;o#sP>B|=@
z$!r?vcO7m0J(Y(|DcxL2crB;FZWTubng)alGK5bG|68Gau%9qD|1p`-{d2REH#AU{
zU(q%10!%Cb7&SJvbhSmc%qC(s(alHn_4Ps8*rd7;W2&+*qz^dH@Fbd`SgWe4Ha0eQ
zF7>IydX|=zz2T_*{cs_7Dqw9fCi}>>jZN}fAg9O|zbEV%29`45@?>*q0qtLeCw6fw
zP^t#Os70|3KA%LOy`gW^{j&tu7c>_Q#AL>06hK=irP?eDtPv5bm`Mgyrw1U}md=QX
z2&V^-Ar~j7`1Q`p@?z)kw8GILiAQ5)5*!NflJ}AF%x9QOk!!u{7sX~WK^2$>+>@{8
z&8DkaaTEAL>aHpDS1V&C^AJp_KEDU<dB%{8A2cVqcR&SgDqhBu_la00`~$((&bD4U
zR&DQSN4dy+*=h1LFZz<7z8X(HotMO*6)ND4Sy@vT)YdplTi>J=-Y4Qnbl4uGRvYus
zo9#|(iBdn<6DHK91rTZfkWw=lha>+j#xR-a8(mH0l&o^W`f~`G-n?D-h`ZD|;j_`g
zUehz+N$*$VOr|csl!e>$yMLAJeLnurLTy6fQvrntkb|jbZcas)-4Lc$d-_)D`P-*Y
zJ9mGUc3XivJ+w{pA9@Hr0NC9|X+SwqS%o4Q;f=XB=?U&~Yi)6gv_suQ2+zM^#u3Wi
z1E=|)>JIvg|5A6%;M-GzBN@^z#|8=h!)u1Zk4TjfH@pc-$)|n)?hgul-`L!^si1_8
z%W6frp`oST*k;d|daDWD7}YCF5z?a@BB)=k)#Sg}*l1v73p-BhMwF4C^<^$>rIPi+
z*B!k_nb>C^`~8g;R4Zoq29}GzD$2OIZ00lC(9qDu?QO!sf!`5J_p;BoT^LTkkw4OH
ztxa}1I_2h4;6f0|*%v9w81^Nf$1Pi9KeF77=Rir*-bHJNj&9{l_UGAY68fAZv+!2C
zE>$ZqprQ!<7M4Pw%MY5L+c5U>YQ0&%EYn~$9gMy0m>E9gdN<XcP!~f}&c8el<hx4P
zvGCx-q)uRQucU*eOC(XB-wS=UP~6lz|L7f^7G4W&K7BE8a$@BkRsAPC7~yCkq@bXn
zHCU`+JcHX-;avag2@#DFC4(;MXMFqgw9Gg^Xmj)FCQBnDvSQT|dTD%pSP~hb<kdSU
z7ITsWM&!1mYwiY;JG-HPe=qVzS$|Bow*2MS289IQL)OtNEMARD#`GRB(fGCROtMc@
z3X|^GKD$@J4=1p#vDe;B2)-wL{J=}-yE?EEvms+-<^KC67sB4&ULn?&8WjyKE;G|N
zMxk%(HU4IBcD<*{zKR?R2wcYj_iX%q2v+{Wf_e_m8Kn}Tss~UdF+pJeyM|)BYJqvB
z=hHFNoea@8c`S>~45+iW_o%W8#1BV&&;CA(%|9WA2QA?w%z3r2J&+GA_^mk%_(_t7
z@6I-S-vPe!<TDUEvqg#*cXuKgTy?vw!}Z3;Dl7;eVokzGr=b-+kwnd-om*R5Sq7?=
zcUgskdkE=+`v~cS3_*nS>Qtdod?5kQ<9k}N9MKS><(oRgis3W3>2%c2DdEO7M_X=h
zU?8s%NHUY5rFbp`rY*@!^BBgTzLc2W#heEUSDdw|`?^Wpi9R1-dHomLBrk50n5#Zc
z7_%i-yrI?m?CRlWJU9*{|J7E7N@dg!1m8s%p?H{MJu*N*(oZeUmQZmyJ$>-|8o+3>
z?R6qm=s$nT{nJ^KJ6)3vp(_9aWw8XhndS7GJIZt|f4}KoVCtHweowi6p@TVOKlJUi
z)y{uv5H1uBthgjHeYKr^AjkuQb2ep9Bae2qB35ZcVH%(Qt|Qsszof62L0~2_45q~e
zqaP*}IV-ad#*z?BJ5OnO@5GXj>~6)9ylnsTo@NJNI%*Dkt`Y(TCtb%A@2ze~M5<}h
zsy6B0#KN+`jRtU6f0iebY}9YMR=B}#2}Is`obi;lCtKO?fVb|+mOw7gKfnC_<FTKR
zUI>yx#+3Q<=iH^U-lpR}@wI1Xh?X>#4}`-m*YnapR@UE*nf}Zuql?(XeCOE++*80S
zZs;>Jdq|TB3fLF(foZqoOSwmZ6?9C+)pdnD3GoU}UIN`uBoN3%?@oYB8{n0zt6+;_
zX<4X5ookv)%1W0YNt^^cY;Uzhx%{y@k~EeyjZICzs6DGOU;*48VsLyM^Mzp7^6v(e
zOS5NUGChfjQoFdcA})B9o}TeXU&OA;P9LPzdz|lms->;Hb#ekP3N2bxS-}!1c0=iL
zGGfG38&Ri~*^3Lf0O8U<=EdXEXMb-1A?wKk^U`Mpb48)+c)|9}w2aRSV1;El`hCq6
zTi?Ta3k4S+N9Y5yBDx^VwW&Cs*uIvia{rVjKpLyd1~daNfRJFCJ)<o|8Kx1_9ytUl
zqqpuid@k9luvay&m&M~w5SHdo@ikb=vVa#d%f?uyuo?BsT_6*koG@J2k&x!7sjymU
zP{DZUKD(RoNmg8|bF~})56QDgVxf>YtrdPveYfq)1qB>5AOOa0M5c<7xu5y-bToyP
z6d)N+JnpJh)1gumw1w0;;WifjBNPIC#?Th%-^$6~)hBg<J7Vd@*Nk!B!P*(8`EmKs
ziFrY-T>BDKlu>#}XE&wzZUCU%q_NJoO1VWVy}D!y)7ryS)83HQ@2(J;cGzUf(*o=r
z5xZX#z_3|CbhFo#I#n`I^hQ$KPAas;Csu{k)v<XB;(~fYA966;MvEqE!MgZGiBqq!
z<Qs(3MZ;qA;`bD4``i&Gtglpmb_y#=`CNS5e_~hM+)*6QtzVd*Ph2T6HAw3t^6>GK
zbaG-bnRo`*^wEJ+tL^Soo|!DufsD7Oe8j=~qo+6|vl{}y7>|k_=<_`@xntkLjY1~s
z^%ue{sg0tVNhL{FKS=<Sf!ZnaF}-}0<^R~Io-)ubs6hHMeSfgxg1Zg<_Khf*CD0JX
z1t;I7I)4Tz{=x7eu%;UfIznb9^d1T~<lLS3S$ye}?h78tUA*py^UvC$9p5I9>o<8p
z8dxu&K!BA@V{<~0tZziuk5HTF%oEG}pv98g-J5!iZC#NlI$<uPiAQsxwci1Y-e^=(
zZGSl}{D8Szvb7&F2#59Ca(Li$j=yTG>DDfV#fJxId*UlzN}Qh%9V7<@o`X18UMvuT
zXI*0#FyQ~5aRG$^rj0Frq{+)X6`P&>-w7b^OV0@8=lWI}FNZY(vdzUv`h&X!1gPh$
z0F#a#vI+(cMQ-&O@lAuychP=uhnkn`JxgM%>;$%mnn;=*h@j0HeYcA79p94Wr_#9-
z&XJwl*j;RtP3jF5H+ZeV#oB_}?awhYn$+N@?jt!BMIW>X`yA7rF*xjisa>WbVupN{
z+`-x4FJY$`62MZVZ3t+<T1=Vk#D#xn$4uF8i?=db;;A@ST&bO7(y>k>%6$E@P!YFg
z4S6?i0k<G<#wK-YB>hJRb${Ram+d{@`oCnR(QiKh`1w?gFdp$m>)s)U<e#O7n5Q+8
zm!EKoW%!ql@;y&CX4k(-P?b1K5ca^zbvYtv3uC?XnFSVvKFYN3>h9UuIfs<Zwny`t
z+L)b53k~Z**mg7z9^&nP*uXWG*Ra^{gDt<Ce>nLOdZ2!7mKC}BXv%0t#Q{YRtU^#f
z+ubpt>5a@Z?(8bZI!c04^78i3#dfyj@p+TT%9!ypOHoqT;2PUFK16g2FuHf4j(Rje
z(e7%#CWN*4YA1R$yvr=qY=W1C4W(3d52AtBh7ErDqqAxo5xLUEvXik$5?w05NO1<5
zbA`9J|K==vf3T9wm?E8f&`lPLvo`)tJtyC(aCGn%`7_8u7&3q|)q$OVfEW`K3jc9I
zKwRhsen>=P+dFq?0hS(T>XHfyQR_W$WC><4rWa51$Em>VCn1Q-V&&?0c&Hc=u)5&9
zLoy7_bWREF@D#Pdk~tdVXrc)7zc7j#4D9=N;_hlvo)<{Z@&Im95<$HNo@~=}SR@71
zU6(B*@>~HWA9=(-mw-gsg6B-<4UM}!Gf-SpV(0X5<l;7#+-`>$7A0=|4|5Z^YN|^?
z@)?Imo%FNxQKR4o0#5Xt0onefYr!gu7h|QvXDAINM}0LQSJxC?D-`}qFI}Wo=pgzf
za7?u+O{~HjDUM8z^=blD&kLUpv|r)(I%ZIK{e3$xHG?YXfz}TaP(Nhi#KpyNRH)Ny
zHO8Jh+%B_f0al<CBY>%X?minkdkAL;28uiiAYAJbu)%|noe!13wv?Sc<9OE<@NDn-
zL<|392g{}-onT8wLZbcIYV-eBY4u6IfSBBXxgFswhVaxPr&#t~^9^ov5!}L-o)S;>
z?-nXaa@3qVgSm7)__JFk@VrmRq^7ie^)s#*&10Ycg`ODjMFMi0#$g1?b;ev#h{dh6
z(q$5Bp^-L~nptQ^Q>;;WSN^5nlAFc<0$fLGu#>HoSLs@G$M%gk3`dG`VI!`mZBzgP
z>>3BJ`M=X-sdDCl%~4@hnp+|FpD6s2Kc`ZweL)}3)QP%0SL(gBh058FIGyx7<Lfn|
zI)IS)4{{(;Z^||GlNDibeRc0TigtX%*q<G%J=9P$X**(_;KJel@K1l8!nH1xAn|F8
zis6=R$h4}mGWJY$Q}`)swVtuzTS6WNJ^}vas^dggn^cM~;g8Tv<-JNrhoFqv_L#um
z1wtF<;-NqI5dU;?KReF~|J*?C^a*k$&26)zeD}=t>7hx>K&8PQ%o!}nACQnmBwPoU
zGx?Ozd{OUfG}}-|apa5_>Bq3NHGjvzNCq+$GN#=8Ne4)XWYfR(fX!68DN%q_9fG1#
zIIXm(f4Z!IfjoM&YN;I2piOVTZtg$X$)XIT9EPs5;P8qHWjhZ#-B_#W)Qnw8o48z9
z%o6+%;Dg$`|GE*oBgS^eW-6?wb6RaS`HQ`O{0XK&VWg`g6X|AdYUpq4ZDZ%?Jxp?@
z7}_^ChGBDNW9RCaM6V50G22JrrdXyN92}j?%Nnp+Ldk+c6V#3z@g<Se3f%uOU62M2
zfrS7(bX1D{7$=}?BpquIrU=~=guOe#dCE$5ZA*gJ?jPP{euW7eWvXWQ-GJSa82m<{
zF|^9&wZ3B2qZ2)}k%^Ak>aRQ{FNUXH3#&-RyRVw$1}?OnJjWj+Jy;tK>Bscco6!1Y
zkKiTANUm1GJ}h}I90<$LHPq`r#uCMX92mS7Ya>l)h)ki<ENp-3ZoqP|Tf(+;%hys#
zIOi*Q!*e>K`JRAKe$@>8l3h?K)Gn{}$~g=BC<LptdaVoSKLClk)`f6;M@`KPf0B%h
zj9q0zNLds@DJ>lUg3R*s8iXD!8R>KpSwEMumi-Iu@n3Tcj*0U)@WAscmAu#ZbYL+G
zI>i32oMcaFMC_vF^Km1Q=nr%9+u;Pr(CgFx*17x7zA&b*_MaM@dW9#hyYwdg)qwhw
zl0_U3yfqJ~&QBAvg#XBW2!yy%EfdfB5OT=riyA9;MNd2q-oLrr%F|Dry8P?(MpOnW
zPOPFi1qVNF{O_tz9V!4=fYdp0R5vxXgacp%HTB&a8>SH)zzutXv4q-eeMxtw(QilI
z_39!0-!-1W>2RSGmO`JxgLqF$uQ^`jK5e=vefP?P1-I~^s$nNUiGyM8^qGTN_i<(|
zc<?9nvWypg^~NsUw6L|F4Pg5p0N=k|muTFxG1l<Yto3hd)_-(6D2CY~Nmcs;q^TmD
zvn5BF6UU}dE3iIpIx`ZQ4R|Op6XW-BE_W`Rk~UyNq!Ay)Y3b-9>f))DVG~n@wJbu=
zs$79>c^s^c%Quuw1*fs~5Se~gxln>2g5Tv#kQ|qUvJe1yn~pb!aO1+5M9n)M&8!ZC
zhrw*OdM2Q=rQN4$&cy$V2-*2^{g^S?9{uWpgNDLe0&fT8?KWUR>&vTsAZJkXS3jTs
zR#<hfo6`8SU`!@yEjwGL)n5p0UEDIoLBmRNxl+ryh@CRf;AwJJs8yjQqv;H}U~qL#
zQZ9RxW@^9bUs!2|1ikXyfxuc*8#oK|{Ab55;8ml45Q-(ilvn)JA&mDO-E}IM7GFY0
zNW)~kLhtuzxgBp4>WGWT{%6Z{SgHc{k5*&ev)C-KM34Jk+WH)SRvu@QiMlVFgZXxa
ztFRRvYQm02$!Q!yR;%&MPE)ta@C3KG)Hv151SnI?-Cvr5d0UU`zLb#9ZGJx<=fUyK
z_~G*K?XPU95lv6y5BBb&)(l|p>2Oyt@zEj8nJNXH|AX4klimmH_%<eqO0!QjhM~#3
zvhu%J#hQHdpnX4NO+{VzxZ(RYu(BK!dLxv2dV-Hg7`yH8af<^HzscW|N+RvU24+ID
zRA06FtzGNeJFoU#aMZe*{vv?lZb#3$`vCD6VlY+kfM7XC^k)O-@>DwF0T=Y@VY0q}
zFJ1QrUob$8V@5K=RIttHIftV#oJPXhI@mzl!Hn)oOK|ip0NlneDRb^^yK$b|Z=3Vw
z3Cp0{qnGRZPZ9L?E<gLT2k9$a&<0u&h9IuX-VF`K1#+vN51j&2$@3yMcp86@c~a?i
z6}zF(Xjyj?a4c_6mQwdt?N!E*PzSUq>9*8t!G;)QY?cp(<l_V5fz)Pe?GPFbbv;eS
zGsrv~$bM}eKP6`nld*`UpA5ZBWD1wOG1*RVgQ=5L^t|L4#@KuUGuHwbtsS9sw5xGO
z_$?7C6#AD<<G=l2{Q<KXdXvZErHdDwg0}vpueN5f)r}Th_e%8o@%HkF!tv5mK>G$k
zJ<J#mD#zMb^uZKJSzZ{Ob@w`b1wi85+}w|R#9&LsR=vHwoq!=myK>Th4vhjfI|Iq)
z=0%}V5rR3v2)nEv-)o_Fa9bnhQVrcAOdKm;2F?nV0zVuBx{AKlb!b3;2?ppdf0}nh
zEm7?4L_LSoh94#$w}nI6f7|t>v2%0LM9-IvTh#d_k&q4}<4gh9z>X+V<8%D3Ar5r-
z+YWTRi8u3)oHTeESWBt3HE;nCw_o}BB~w4_B~*`kr3Ekp;Narwo|#e9QAA%FZ1hB-
z=#JT09NOWiI!7)iFVxdHgJePa@MULDda}~T+3mOTP98E4iU0}NTms;ZCh%V*-#oD*
z2$ckvygC|({-=dWqTaw-1{@Zql&*0@)sI4NzOtX|_k@Lq`zIwO?P>i2hG9ZO-{;)^
znL^z;yXCLiITlZ@&0|^5Y=_rHqVicZ;7AF5@aCNt*G-=9<j$)7D!kK{>K3a|OO?rt
zrm$;E2Y{7?#)y_NxMtH;&`>c*T~})qc7M1SYuKLI3BG<yFjgnj`h#esb<i1deYr-s
z@CMzG=>wr9k~N*fA^=3>qMJ@+_5m}rami}6j$fzw1Bj@Kb`Xi3e#=D?^5f`^BlQ*E
zCdubr<C)2iL6WmRKXp-+rZm8$*a<|*k{jqrBFTh^{{<zK`~fVZRHqG>+vOxy#P47k
zaG*!V#SI5yBP1u3v0Gfv`)>|s?DxkB+{b3<bldSY>#gE4lWrz9{UTB{6>l;)v6xLG
zKB15JYWuJ6PUyXI6HJkUpz%ADK@wz81u~t7pK2~oKVfKbNA_U}rscrTVEz}f^NEHA
z4#>gj_Je?qO-Z4u01O+k7$p2;jEszaK>(srCo&2OEYSY;&Q6unp~A^pi|V8LH2=;$
zWc~kf^it&iKqROQHlz^3Y0@O0&Hppaw@Xh<47NX>(Uax-N=3(RyZ$W8VZYOlPy{uV
z!>cMpY5#C#oSUC7aS9}{0a9>J1?icY0$EZvHcWE>9`yz$%4m4|Q<zP22V$b56Txx6
z@=WOZz65murfvLRqYk1RZjY3yE*hPW6K|5m#l;(aUYeC=q<Cy~daup1jxPu1n!TUy
z`0r677FE7JX{>hm>Mlbu;fp`hXXlmxc|*FgJRP7E%Z2inYfrgLN~4Rqe}qN`pMO4e
z7@YOig-WSzqc;PPr*7EP*9|c5#LfNKdkob2#m*K7RzRB<oxlsH%$W2SU>7?S`pD$_
z>gDxx+Mtl}!+0cd>rY^XN3ofwE3Hl|c8*BY<3ud})RW;e=(5pn%OBnUWVJ~VmIer_
zBSYw$3Z^t1h)zg6?|9Skc>W}rxOqLn|F(bG7n`wEV=5W(!KT*d^}$pslSduaROsM;
z)%Mm=QLXPEr~wjELpL&XgM>5+GjxN3(qJJC(v1Q`gOqfal!&Bs3kcHP(vqU2aNhyX
z`FhUz#&6wqFMqfeu3^vK`>p4BKJnPBe5PeV0Ta7n=~F_MgC$J+rco2h>i%e@4c$G^
z6Y1wk#^qN|-5V1X3Kc;Nf`ZydRPYcIh(N%Ml4u>e7HIRyipmBsBlV75aF?<YbS@K!
z83|EQ%CWJ;-tx*Qjg6jjZD`!Ro5z#Aqqg%i67r8!>DGrG6j!MOzcz5QQx6b}3&=YI
zOTFbUm?C-ZG#DRM<md)~`u1+8{mZbR7e1%x7sx}P3<^E;HD2@cY0oovEV7#@V9rag
z_Ge|O!d?R9AKu)Hm8#vRRycjlFcfYmU*5e!a`Dpx6H!1bbei)e`CM!u0d^3XH6p0X
zD+FH4%#6VUuH+2BY<?|dm;F4!$*qr9dJpQ%M+lko2fv790fT&}^}>qG`*ommsXv(T
z@*C}dZ+4x8Gs-0&EH4?Cq8JVx`LLf(95HPn6bq3m;t`wQj@FA68oNQ;5w$#ZhMN6p
z*Z!}_a+7<@ma&o1<y*IIJ>Q%p1Jb=$9wcUFX6EDXS$enqF_q7j$H}zbE(D1?mCfmp
zmVL!-Ssb!yog~hhDjWDd4LfL^V4g%5a997ko;OBay^r`jL{F59H-V(=y>DQ+?3g(g
zjU*;b@Dfxak}%F|rqSl<x7zF6k3W}N=z#V`mx_vt)Mr0uD^cV*29gTXxj2Z1L`O#x
z_9MQbLnH-7L=@J$yKQ?=bt)`FvXv6j72|H^>Q=oY^g+x?LO-*}){^Mx=!{NGDBKB2
zN=n*0NdGBV^2r!I68#Aboqm()N~&}Sx!gq7;=;mJ2?>eH4<8in^n-@enfhGm7M7Nr
zO1HSE4Nig&YFo0hvKZXl-PP*coGU9UWm{e0Rup}aObl>g(@f*f->Mu0@WYDQf>T0i
z@7t64OA;HmFZyUom3xI``5Hg|sn$-@HaUIWTV>hkB*DYl$+($%zHOM>y!vqsM$D&k
zn98X%3(CoIc18Bh2ha|SkNtwKw6Ca|pOtx^zW-uBy%|7F>$CD{=cyO1%Ns<v)I@V;
zC9DDT_PAm2Vo;>Dtxej+O>gKnWSJ{nIe3ME&W8??h(5oMpB8|wh+=a**X-YyCWZF?
zX`dvyGQA%>U<1$B{u?<p{aU-*_aQBHs|0X{$-(X6y=0JbLi(um6S8u<7w4h1ll<K)
zmNE<<wcgxF&5nh-=;tEQr<d5)q_=1?9z=JUtgX@Oirb<&^hhU;j=#Z-cl<HJXF2OM
znB?YymKXl~_*x5R#|0rtJ<{gahN<upGOsY)E_DqekN7_CV8InVk7v}D(M#(`7hn$^
z>}y>x3kZj!z{1YNjRhkdDqo??5wmOG`2N^=r<Ej2m?xb$2tDb2y`v?}d#cQAL|)9o
zWq*@ns0u8%puuBb15J{YxFvY^_q&AA@$ff9DdAq`cgy#1nX1W&Vvc$Qf}Kr>Eo(pH
zf13Fky08DnCsCf`><_u=;J2nf<)+5O;W06Rqm_bu&qh`CHDId|+WfX;?~Xzn(fE^g
zH*F#z-p*SxsZ(LzxoR)f*wUXyaF%Pin_WFZ-D@p6Q)=X*I8OC@#HmwI_d<_ux`8Rb
z)3AX{!cf+o>Cn%IV1i9E?-egX&heeK>w+QQdIgR(ML>!&j(<JQZt6vx!e}2sp!dC{
zM=Dn2Hk7qS${&Z0mgh3@QZ~;ZmqK*bAr`k3i^2eUo`r>Z#18k>;d^I*Lj3Ma)I(F9
zyJMYXiitEfpLMnlm(|M6?>l2hSZ>UCHe^mW=f#_B0SWH29-=p$W{6+~tW^lVqOikP
z<{{C(AGTv<X0e(dzN{D5$M{_ZMIjus;ey6;i;2uM!KPp`-pu(*1GgPtO<Xt%VBb{8
z+`fHDI*ghFl^|Up*oo5SgZA>5C_82TN1cRuh)vMa5rJ&Hhz>fC`H>cCms}xYyBD=N
zQ19uvx(pXoN7_v}Mdor1<ZE%&9B+Q;56=R77#p2*aQIq&&9@$r{w=z-K`c`4+vEa!
z6IS(dX^Mu1se>ADQU4VSy1?1Mrn-3#F_*pY9`6=zHXcH`0+MAJs5!*f9{!V*=ZpN8
zOY`=S=GQG|Y3_QtPxqz+TE12hqf?F%;zaAreJ8QQ)$zK>OS)Un^O6S<&BRF`tG#P^
zWft<a?OH5wn@@yFdALR%F=k`Ygx&3p_WcmiOVM3p(AW{DH|Ju@5VH~77MV1vDJxWa
z*~n}&qb*hK;B)=9(RD}!r*_o4@HVL+yg@?ZlENk8rT*8Z8rrZO5yxAOrUm6J?8XU6
ziAl)^Vo}wyp3Mc<rClCA+cN517PS+TqrVx0NyIvv6|3I_Isww5;zRi$jqSi`bO)Sd
z3x~@ov3~OiVdnIaj>Z>gSw8R69Rgn4$P$(h;nVs$skF7W{^V_aHw+@@thZ8xqhhvS
z5PV(X=rEWV3D0|3(Yn`1_X!J$n(bh-rLOy%XN1+?fBcB?(0^{-VKy4(Bns(4%*dQj
z`!GLqoo@;7^dk+}hGALZ^Wp|r5c9GJ_*8}UG|-2d!UXBd<!Tsx&>y$_K%{}LN<uDs
zD+Cj=@SJ^WPN7$LWse}l=3SBtJr;eHkBOXIg&N6`<S@uwJ$q9k94&m~?6P0*W=MH7
z)nW=;yz#y@#HeczK$O^AhI7;E%pyx#C*rA`fwl)(`P7{F_RyZ<c(2QiT7AzN6|r2h
zgcD)j2KMB}p@mJE4XQ2mw+Oz8Tc8$0iBE?VaQ`a%{uE)(hNa7oJ+*fCHkto?7J&Q!
z)W@W9$`ZoFd9)m81c$@rVmTi0#d1blw)Z5dbEii35?#fcV5_^6uUc1ABRy7bJ{6dK
z`Evg<W`sC`no~V|Zk452O`s%<kjVmEO&<Dq6e1@2V{3*>k0i~H`PuPOf|Wen5^l{I
z3P(2BV==@tdU3#xe<ygw8~XTbp5BLY!l#JO(u!OhJ#;Mv@bs9#l{*@>F7~T6k}nD7
zOJG|N9!eQEaYPEiz#V^IUpem}r%9LZYjRSaEjMky$mWF?ql?4v-kY@Jtm?w|IP)HS
zuis+ep_fHrF%C4{aSb>gDmE~{l!M)dX4ezLty(pEXlZHt?SjIQ!=L@2TX}EXtXNJe
z3An(e`-dIqi65=dnI^!(D0hSd!evozr!(b@a_*2L+EdecRcE0O=@vHOiRcocCQM-q
zEp<|r@u@z}RV1Hv?LPNcIuq~}7*J6=4>_^CoWI8TW`OqT&rZ$BsA7pEkq4foR6)~w
z5yd2-%2!F`K&2PQXr1c==;?F~H?^%9?Y?#HAgP6jqoj_f)hHpI)n~m~e305sx2B^F
zsSXR&V|V^5i8`xbzIz9F5vWuD@&F>O46g&f_hYMRz)=Xtg_)qt5Z}i%-&fq<FJnS)
zLyn#Lh+$iVJgueZw6JhmmoS8pNO8?>;9v&a)sohdU~5VyVEU#pq*klNV!jUg8E+z;
z!-=iuL#qJd7_tBAhcgX>CBdDL^4O4-JfI3OulM)&-_xGAfj1i3uw0?vahrFrVR5NP
z90z@Kv!U78g8In@ER#q@wb&s*&)EVeu@<w`DQQ;nt3b$~Q3{-y(vJafx!2ze4@07t
zxK{yWI-72Qr7Jsw{5_%EuPeW}j`jN$h?o>ml)Mb^x(F^22zkPDs0A4gpwh3!BgWaq
zTz_=iAK*q11I*zf`?-QIg@769sEPPyVIp<XfBfz9V5H2#wh%|@Jeq)Iq8*}fex-&p
zpr6HGP+U4xcd>K@i9_TmPn8J7On{A`+Kp;<G2G}HX`Vq-M8IgU)r;c{nGfT+YXn}L
zwSFdRP4<GML}BcQmN3{++U!p{prA$Sx4){|se*DU9R*9yh>@kXpgIjJTb*i-ZbNp-
z9nw+-U+LN$!@eIHg03$7ZS@+Ek$SS=WUg|lt*T?Z$gG#{_x=MA5KV)Xrva&bA`W<N
zUc93D^siVP&tzem{jc-*zx?og`1mnYNQi6)1Eb^KBf2ALKDW-pjtbU<swy&L%bB*b
zXTwn=%o9J4_mwPN2m1R}i*$tmhb2kGlUhGH+1kOzJ1Ge%IU9Qs)X|}we4=p_ymQXO
z(e9Lc`JDo>xcq94wDL+l%8<9On}!<f&&?u;{Vy6@W`^RPH0h<GGSWd)W{0U#nGh9X
zqmQF8%o;DJhC-7>01O{2jZM*M7I+&#1o@1H4Y7GE_&0HA_&IPXN;uLTdrH2)zU9HQ
zS#y&t1=sv?v3l!7PT<D(Z{MVRL5mFj*6?Vb(box;E=Sxy3JgLSCZK7y7a_x$r@=OK
z;+;KP{56yY__r#_Ox)vr8ZY3LBFf*+C7iSHGvmgO2z`D1cz2h(&e#wL#^E(imb^S!
zSy?Y(9dO87c2qDMfX<(I7LpU|8by>9TD>g-6XRoz!1x3J@L(<lE7k)|K@U51P+vX{
z9gkdAY6AlSj}8!BZ?hfPJTt5OQccl6`DB6n{pNmT_T8&f_fB5|f*GdXi#G@V28J%b
z6SAi;lVLSsil8~U>OtC%7uq^nxb|NVZ43aA5oFXi!1s@3fEX5{wh89!JAk*x$M{7>
z5z;f#Ls^6{aH&OaySP-l;X)VjJlmi!NYGOqZS7Q$N4>2mjm6PU)JvQ*w890s4OmZ@
zso-+YqYb>ro+L{p5%6+9v8u7X68kks`?QVEum|e4Vn$co4gbSE;Sw>I|7^vg-ke5<
z-aVVEB;O0MS4cjS)K_$D9THfZOOrf5?!0!@<M6W7n#jlX>nb^7x~CDo6BK9}_v`He
zmi%Rf$-1xiRLyW!5*GDA^eRcTyf517s(J%dGIlUZ^Py;WxX){*r!?NnZy&OUonk*5
z2gc><E6^*%BCaAx*kicuT1-q#e^Y~RN97)IWoBZFBQIAv+}ci;YJf_Eh_Es@!?Q^y
z<l1~H9J@yJnQ->r^$z#S@FT8*^*-thhZ!{XX{nBYv8M*v);b0?4GcsPLE_Tw0Zcnh
z1s2*vwer+7ThB775`Dai4Cp>gV17*+A})eYmff_bJ>sfG;dB|7z(npUwy{rUUnM1l
z_~X*ZO9kEJiv=lmM`yQWOkiu7taD3B$zU`5TgroPDV2N`+_5gD2sjE?3}C+0>cj8=
z{s`_J<b6fE*KK~ZUUn*e0Cm9H%Fb<IU^5mVcTCJ1lTQ;8Kej+U8hjsJ^ivB{xqB6u
zohA>zr!XG<LDN?p>fWP-s#(ymILG4lO<tzEN|nf)Lcs80MvZIH2?6x}_m&E#oQx2K
zKK(~vh|jXXj_dj7j7cIMm$b8GRlW!By(9xDr$W(wK}(l5ci@zR0SpxrmDW5R=|U?+
zu{}M?XSzt%Wiz5>rZ_q)Yj2<3_a>GOLjDF3zDhxJuS0Q$U0IVafuHf*g8$=tyF_eU
zPJ$#~zDiKx^4k5>hgqFd?0vD6x{wrmtBknEGps-5al__Us_0FrAS@xf`(ZE`!`-?8
z$%LTKtxeDzf3Ky?q2b|s@r#Rna8_8SZ%_ouvF>odm8$uzdnZ2w>rHm{+5O1DMjPcs
zIb&rl9aizvkfT_goHr~MQp19k;q!#{;WxqO+vQPK&bs)Uy0uId`DyB7ZfmtGx&p7x
zs=e&gKZtg^h%OF2|DEL?k05ME`Eh<Tr`|!jXi`|sc+EWTzjIOR!Fn%myqeh1tt2(;
zJDqyw4+*G$^23srH2~Cd$HWizHoZEB<vl=)P{paxm8p8P@sS-TbNLk+dcBspBp&sF
z^!u0AoK~JvDtbSp<Jb}J|6+z38@zIWJjw!1SdV3(>91I3XyIX;^AGP*G+56Wvjzz~
z?BfLcdTzLr`7isqFZXa;2OmP0_P=Y#LdW^HdzdU#`%<aC{pL_98#4jRL}36vjj)F<
znD(TxGVXfdNL;|vTdZ)J&lq7_o-%2P5*762|0Y07pbTsYVOZv?5$#xaj?(`nK*ftT
zK*-X{kDo>$I#7sMt&MOmL!s(8Rrzkk4$KDlJ&gab0G!MTVT8K8!3Wlx(3@)4HgXj~
z6_tXaFuw+KXU+&cxr9cUO1%AD)s)&mS|vdleqU(-l2aPx1GX<nTr&R9U47rEZ)`lm
zD<Y}A_$QRRb#i_j#O3!4!DTKXA(u>2<a+CPpSv~N;4>pHtL6MpAy;?UV+tp&BF2jA
z?IrJ+biW8d$m9aXCGa6};>eKU;e*Yj2CnrHV~6For~365ba)D`DKY0hw5NG`XX?y7
zxtT+lsnYklExT7r35N+{43+D%x7i<kP3QOxVrI{a&UsJYA{T9o#F!giU_wpUc+Q8=
zyImUjHGt3lxP^tX!%C?93XSmc<P?uJ>W^n5Tbn`Sm}L9ue<$4jX{RRD^V8%X$~6sb
z*VKFPThy9(WI7mQ;9;T(MFqf+-sHCLG%a}j_A<`Gl8I)%t@Gf0*``H0N5f_FQJc$4
zL??Z1y$=TD;kxQBJ)NSto6`;OzPH;lS0xUy&JASf&!ZwN7BT$%_5zfnP^p==X=jqM
z?gI)>efAf&S=cAhrFV(=w!RjdPB!f1igdWX^v0B{u$nE(el&K<z_<6X5gpt{x(tXq
z!QDg^P>hP$@zn+L%S`^@?oQ#Q_-3p3*tm0hNBcYA1yj@n)&G^+stHud8KJE-BKx!i
z*bu@&a{V?=2`LbkWpX3$Ecf9@+1t-scwqF)x=I(ufL^q0X<y|3iyY)d{>g&h!=;~?
z(mh`>M<Uru%%AWj7ko*V>|Di0*RiAFzZTNuf2zFe0#)bB+N23U$TK77g(ft&=jD9a
zE%7D<PbA*EB}&SK3hBt0!oJ)+p@Jj{j0HdUmz);Cx32rIom^8y$L@WC=vhcqt4(AJ
z)@bhk2WnROC7BaY>t1rT=7=?Uou+B;nO@S38vek*hFa3MY`xtumbi}5ZZq3KG^yY$
z4>3%~mtQ@fJ)KByHCPl_^T5YC;@n-@00d`$%gC%99Hh@dB6##)Scthh+jod5S^1?W
zdXhd5&rk>mqwD-~*E-zYjZa6kW$SMl_q}+=Y?d%1{fPEdk}}?->>4<rJJvCubPX3i
zCC9r&LPGMZI)WtXZtv}ZXfa__#>Ls$^v5caj0>X&u?H0T!UV+X)z@{gq;3JPE%0CN
zfb0d+ylE|aUcyAuuR_n~1(+dFXg?3~HDCM)0~d;4(YS0oi1dGbeifK}oqrbmL&dw=
zz`((Q4O+9e7wLV-EYIuc>}-mllX+_~AwbW_*lBhVsv~kUxxps^<iMs?)B@__5%KY?
z#r%m1v7APGD}9LI;HVM8V02xOiiUwM6bkiHcP1-su4?2cx28$?XQm)K6e52XZ)%uz
zSVSTv9lG@%R9W)Tl2x{gnF><l?@bB(u*iHCyJ9ooLBzY`Ij-BYmzTQZUUbEBnKZq`
z$ki@McT<>#-KI-6Ye7JeMQFoEj~Heo!R8C1XM&2YNx#~pT97pNQ)k|B(~VC^Xxk@@
zi;Lst<>ftCHdX^vx#GgAIMbTC`)}Aa3sfOx&wi@tOJ(YSnDusBGnTf_{Vua84t@@z
zyZM^F#a0&)D<f4UKyhX#tL)x_rZ;8JhBB{~0~{se`D(wcic0&Zgj-qDH-_@HVl7F~
zk^z+-_@G3BX%mAZ4S*D@)n&k~-UjB`&B$WuR{|zw*;sImS54S}bAw6`aFu>x$Q~l~
zQC~WoRQNeFxqvyM+I}5bSngjCXB82l&??ex1TE^amUp+e8$zi>@TjPMs?0*0=ypCJ
zHXvv8pkke?k%ObFtD9*LY^mBRW<1z@ISL<?N!Gf2pWWEtu;?u@X+K?^(<ptcq*<g(
zh79HGp-_gS<E{06k6z7(j~~g${`ip%!3*dm0Ojk|Zw-XHq19MG8e^KE)cAd6eY<ci
zU#KpZv6!~M5PXGPdvZ3c#<MG$jm>}#3T@218Vjm9a?R$j4j!j%!<P1TIzd6wl9Cd2
zW?CN-NYP}is}`ofvyr00!%H6mjhg-BycHuEm1b90S-uNG#AK_VuR>AD=dVhMCZo$U
zswA`fQqPM+xAD?G9_%b+wd=lnh~aTChqkvibRC<V?{?|qPnmdMcC2nYZcT?^QwTJC
z`lN*E(Bcr7cE|WQTO}nFl&cm%4|746z1-)XJG%4L=r-V$GNQ@=Oum+IyJokO?dfz)
z)t+vz*nT1(yWC3-VCjem41A>@>GAWZab_kZ?K*@d20hJWVg9phX$mjuTzjZoZ1%fa
z5r@sFI&JRg4WgN)pk5hB^=lsl0VG$yP`LgRQ!Qi*Lq<ksRd9$9VWPT2z2xeHSgBX6
z_6tIXsjI7J`c$QJ5cj=+yP1sxe1QcWGAEU|l<KiQT-fj4T7jCbTDX$dUu|!;Dh@J6
znT58HELzEQlAw;%7WY;twFcfg3=c|eP(V;?1Y4o4+i=?`8~I#jrWM7Z^CRFb{3Qc`
z`C@V0j`vpSIIL=R{N9I*TXDj|NFjsI*9I|aUH99VQrvt<^qnqCoSl>oet_?NhD@;f
z(AXSqe%MNvk7iZ3qAZ!B@i<7^tE7_&jS}A$u$+7v6)*5>FNtaf<6MK7rZRpVx-y5X
zX;5>gKM6!vd=GFSCTwi{<_MT|-@8W&NF?xRxa>xf2TV2{TPxFMIeu{`4c3H&DSa71
zrDR4Ok<4s70ZUD%s-yQ%-g^T<eGb2@55aNYpK|Q)1;xG!vsjrFg%c0J(>VYIVX$v`
z2M_>nvcqrcxHQDW7kql?&18JCN(3Gpg~CT55l_ks_LGo9?csi3Qz;Y+QmYKxBGC(d
zhcEURhD!-p&^k4$=3(gaE#(--P6Kc5=4Z}H84qUww5;jlN4X;*wH!t+5>DN4%+)%u
zZnZ0|)7~qi+?sdfF0-ls^#&N^0<d&#>8>DTUJ!0@({rdD14u2sC=ea?9R5;beg~Gj
zyRw}rlVY$}06;SAH}?uzQJJC7A+&)jmm>`hN-d{^Pj=dA)K$N&=A_7fH-ve^{5XDH
zyX|{73TK#z>yHA>FgRXgI}#LfVJBkM48MZUqY$1u2VJX2H-4(3sMzh<gk|Gd@`WUD
zFjq}#?UU`=V4kNRUqCw%cmwEkh~!)I<+1uMa^&Omk~C`C`+dv!VYDvAqDQ)po0AcK
z3ibeK1I-N7dx<wp@GlO8h#k>V@1Wdh=;Kq!ouJpID`4F@0e~jU_&wCx`EW5dYg>}A
zpGeBh<L7rR)3yv5e}({PRl0QWYu}ZQ2s(A_nD%B8o`LY(nRm|LmOQBy_7ov$LFG&#
zucPmv@TO090rYtXSw<g_mLN|?(8@)#-faTlP!@d96oP8k8cw%!f`&sG{+I_}L+Jvm
zfkfEcS0s$Srxb;L3udFuLzn})bgL^7_OZcUlGN4JoehGocv-3P^XPlln5)O(B8h1X
zEhGjUIZ|t_&?aIvS*>nHdLKNsf!3)hstT)_n8)P!>#c*m^EP0?v0?8+Y7JYu??6xR
z!q~2ms=2}`GSSoD7FO2ZpjHqy_!Q5dWpA7So73MR1s2BYhIGjaShq0I(Zd7R1uVuj
zU8s5ZQ%9-$g5Z1I5Dm$o4TzX|%eR?odj(cvcF8zoy|g}sv<HM&uyhgwKto{88MBJN
zQes7{_{bH0*=a}U5$nY9{>(^LZ<UN$K1<xn&f4Y@&aGQo$bcYV|D|9kZ**RN>$cM$
z+er7|VTiw)V99EFdjPuN`?)}7H#Q#<8@&ROeaJeB%?+l<Y?xSBFtDv=y0AcR^i5~G
zV{z<^LO(;oXal#Oy&#07Kb@EtH-_}G44Vag7nlElJx(h*@8P3e?9OhJ;6%w&BX@j^
zus8njzSX|99C3_pT<gTSQ#no+&|(sh0+Had9I>Q>MD!zW-w@!$2_Y8%!HUj7iAjG}
zp7|CiGG%dLf@j;`lDNdh#T6-36)XtxUK@JTeD1d@TLBs@ppedAzpa5sM8r_*f0%p&
zdc+k;t)7WZ{65WByQSjt`1&~b{_**;2jkz<1p>9Hadl931k)`8EA=L|L`N3x_*I5g
z>bXU}))|G0p6bp&Y1v%?cSaOOYe~|}Vo5SmW)yUzTy2t(c*8EB-D%IJwO91?13pBI
zcMzybcxl8YI2y<pMsT(M3n^HzL;|v^1*%P$u-L5S#ynyKVsAMve-7$5ix_7@9`aW(
zLHKR+?}@aS8})h0RbEOaK&=#u8RDYSn7)x?{bTQ0uxAdm$di{;%)s3L+_78%qyF}P
zC^;&zlg?u>s)d?oChGfF8O_PMcR0l&WK)u>8d@hNbmAR=jTI;30APnBY$J%goy)q?
z@AyxXifl0fI1%m4MZ8lyuBSS5gMew)x`B_KdI;k*TGz<rF%Cs6X6}^3rMhWu5{u5s
zmd5+;cn62#le~|Y(v@2dqv<~9t;^F3U^S5#DZE_~`GhZkE2VSacniUaiRqW@$MhO@
zCM)DiQ?KjC%m|Wpm;G*3PGja!W_7kEVw?}O(3?hCYbG6W2?r+aVdzO>IPTTuk{h;J
zixkn!|MHr6K&^uFkpIP_MP*WnJYIH?@LDlJA8zT9GWE#H>}7vk5rX0uN@ZtQmF~Xn
z^LQ9P)|jv6sEBz;`g#V<T`H9B{%YKH74CqC7=oooBId!$k;Md%2yrnxWL_Jg_PhDR
z($5R?2L&o_x{}AM`jgy}uhZ(vum}84yjo&i=bE!0ZGTyc|A#SoGy*)S5KN_5N!x0l
z+~tr>D3KA$Ds&z@lX)F2v3v~}^=2tFxHD6XyDX;LQG&lF&>8pOZrL(-PaAo)jx;N$
z`61*3wQys-o_kamFQx78L{uW+BmV16^8aa51a95qBCWeLH1#wwnmppbVw}=E*zHE4
zr_b*m2z!a;X@8Mf=(<!qjcY9Tp-CQj-#@$|4fDCuzP^nTDa+{Mvhi~2m}<@}Y5zk!
z0`WZP{0&_`HzkX9Ywf&F-J^2P##vpMn%t?H;rpjjL2Fv6fp`8dEKVn$J~qWz)pBtn
zl<8%VTaS8yd?A0e@d0T!`b9$9(7o^o#R{`IclJK|m?#ca4acGmb?jg0gp$A2fV33~
zMXP@2OcoM~Hqsx-{Aq{NtTEjOc1@Fut2dE9T62JHor(b_stji-;m6;NxkNGh38b~&
zMZ`GK_;PV2AnQAeD4gJOuYgG6o%{Cc)aB^0fgnW3?D^oaNHhM`3Jd#4E-B#PgClS`
z%A5|CdN(P+nQVx^s7kvt`ru@30JZq;+vpK6HnitsqaOInld1`60fscC+GTl=kIO15
z%Q8;b3v9$d6(=|dIktFWwdKExKN%qR-wQl{_4xlVhGK5UA@^-2P=v-!0vK8(YB|9h
z(ed<Rq9y&Kj?*Yv7lMggg+im?nPgcmJ{SFW?;$K5cJo5Rn!+Uv#y*MjPx4%RGKBno
zMY)OmzUo6XKck)3<=Z7ZOERU{HEddu6TS=DV+5SG6SGb}a#2$5XlTSIZOQsD7-l0f
zd+cBy><AnSA#;yua%sZvUL7G!&I;Ch-%9DAT=X__d_N*Qk)doi>R+Bn!FH;R1d8dN
zzcnVQ@t6{g`SH5?mVUMCZaVH~^$(|6s)=fH`2h%7nTs3r7Cz-(`k{~RCXBl<v-hHz
z;C`3X<S4nH(g>E&ewYwt=Ur){JgT6B%p6osj7mBsqYdu|L{oo!m-GAfZiI$l_>8SN
zs`&GsD$vmg^XpL2%oIu4ePJN0$+`HF*TRc@_Y_Aj`P+Qsd2_c))UT0#ZZ|^Yt(|~*
z1@<@0)29nh3Qdu`7_ldhsDHBocDwfEDVSLxef2d4sW<GQEZ}5_J!bgqNO>giqvykt
z?4Rx^v#^HW*eFzArY^&N_s@5+6~zTXS6{}(4apB<e@zl1H)xHgvveh0>WvF=uHxk7
zd34}Z;8Mgsi9LO_`7d<tM{gFthpDyChb0|{2V6J+o5ZFdWvl9>=T*0(cj|Z)?l@?;
ziJz*Httrt$bsKA=at3ek&_l`<xBOE-(;=olpcr@x5nSKZeQa`gBw@6!g;TT}M)4|c
zpN&+glucKqsO4Q!pVRJLP1?o1YcC6{!pL7p&q;YQ0l!X`2hvy4=|X9tLd(zAXxgpk
zU9Mo=X9)!2EC3E55Q>_70^gqn;!KH48h*YJ94AO4LhvRP$m*U>4dX1wC`oP4-bVSc
z_RHQ|b86_R=JG8=$)`@@QqQTEkwRRfM0~pP+`&hzF`wS}FZ@KC{l-BO{+eO`Ia15x
z#v;A}xwePfnyzwnHvP*q2N)BdHAQB9*9E#@QeF&Nv!2EhXEQ$2G4c@kl;-inNN&qf
znU#J&cxC|N`KLcCK7gO_C;PXP80glYn*tIp(kWd;WZlq}H-gk2UB`3GciWD~=@LJ+
zc!!TvR@>8wDuVGN*Jf!nqpnkMV~dNZAVYRm+6f@&zSZWJA$|ABm+F8u6kU-4^o0tN
zI%~_80CdgbdsKCGpq4Y?<y&&;5_yUo?2aa^nqivY&x((Zre@UMuVfK9DIyw?e$`W`
zoPKFx4e*p-P&1U3Ofd5gVD&%Na*%bQi=SUx==onDaoAs#pZ{1WdYnBRJ^}vp8pPu(
z>E_f2CIdP>whe7^xsP(Y{QdsB;)`lI$`*&@7i69<)6GNqcl|+$<d7VIddS_nV;1PT
zye1qsQ-Rocm;BKOmqpAL|3@Iwx4p+aIpg^g`&>_4{n0FHV>h}Bh->`I%E~~o1~~Eh
zTirF#EY?>-)?%R6JX!nJpqt%6)HfsWIC~fK+8UHW?wIRRBE1g)|Dlgg2APF3nf>`1
zC-O9MK()i-rkM0E&V;p%O~eS^YD`|Oe@1#b7%$*zi&)_)nEQYGj{XnDwDSST=3t{%
zLmM{Q`4T}Y<c4?TF^H-+eMCyf#MP3Wp%w8uD`qQ~K2A1xJm{{S4^QPtJ9<>iAKiQ4
zF>YK?=ug>(IJo!9yFF1&(8mDUFLu7R(~%po|6y~sk(n;M=BByu<DOH@G=-7*oM?^h
z@)0@3^-1C+KR5sF39{W@W-npLvR3HroFcY}QEqgDmq`CXK7~<HEQ?10wRF{~`;+bn
zwTG&iS+kM+fX)IHSeTeSUFj<lcyol3s!%-ly4sowN4k}_8!9QFu3X$a`PMp9{ELVW
zh(*t*sj1606LH9RSbRo~QAYyy?1HL%-DU1lmLR{{_8SNMph*BtF!fKZS+{EZE5ZNp
z+2%cD<P~w^UkjC^rMdd;IMu4dhMw)Fo!aKdc;Uq>0wLH<uk~2@*2GYQ%2S%g2J(4M
zfvd*pC&N9rk|d0m_hBwEwwMwgRak0Y$0cEzj^{HFB@fSYR_5hb%+{4KX(&4sM<r1*
zAsQZ3G0~_#ph<P#XGmt=c9<gKXjnjzX7?(-!KN>6LX5Y7*<mc1`I}mA4|lfN$pV&x
zO;~!3{+%e7cQN)<6HLktb6%#}qHhRJ*s7Y^y8n|`Tzj`~vfN+e((}N9VDHjMPx5WX
zcj4AYGTjec2upT3&fa-tB%poG`axtaa##`JADKDd6hl+8o<?voF(gyPG(a3j@QmLK
z2aU4%)zDmJajxO*JSMX^#8}CtZv`*7*M@cvDvBdx$k^E&n-1iXi~?XV4-p8h>y|iK
zKhTw?JB~9JRJdZ-o+xy!1AWBBcwxNrRw*QzWi3lmEmd~@boR{z+Yv26*CmSdb92||
zw*`n&2Ibma``#xy-G(}ixV>f%OxGRgpm$jG<Oto~#kdoRNLgj*^j@m`a@h2sDc}A{
z&Bl%K2<_76T$bE6X;0)jSqkny39+;q@k^=!-39i}L(aI=zuWg3GR*82D^AwgUdB-W
zK&ij&jzg?IA$Va7rdBfoP)n5`jZRT<S&w?Qz9F5#6kZLJCHbff|JC(aaBEI7G+MPn
zO<6H#+>Z2#$^90eWFqk7%A>V2kh)>%kmLYFGRyhA;ENM#$m#2PI`fWkQen~IN(7?I
zGr7w&#{DW4Mb+Z_5Gch*Prchc!%;H#an16gnf7;cSu8e{5)sw;ho&YPKR(8<S@QMB
zhsd4?Sl|dqi-!+OG<H)(uSa*8<`>OAP;^mF$c1Up?z9M>nC}Y<Yju%51bPI&^XN&Q
z{`*gY(ell}?7OJNsE)q%Njxun%rhCLW#-sY=;?&TTRXgic2$Pi-A+C$PPBeRM&^h6
zH##Fi{o^P#V7*=~3kM!AcwK5X{BvNne}u-lU({xr=TJ*`>;eg%x!t5{w9@R%3V%cU
z2zt6n1*r3Rw-&<LD?*x3l!v=T^`*D7xq~s2eZxO3d~Vd~R*FubpR96GvDA{vdM99;
zkeeCUyz}UIcHqNRRy$@8@qz=07PwVNh7?2zw&;q5Dj8t4e=h4;Iox<$Kn=R81$!hT
z{d3cmb`s;lG1U%R1r-+`e)>IDpPqZr52u_a1Ev!3sHvg;HYS45pX&cUvX0ED`+!z@
z#Wiha7V?_gLV&oi=2HFUR;q>4ExLVko<|HD4><^U>WkZ}3Hyq?9mP#w1d3v&zf#;&
zo|q2^(=YD+c>@v=v90<7gdfRqGHk)x?@@y42)e>XZQbbnG{pm03KM3Byo0Qng^3D=
z)nvi-Hhu|6@S3~N^&<<)1j?COwp;J@H&lTEC?ui}iMs&PtnN07Q$2}hk_GaGr6Zez
z?Xn1Mce%B3Lt7AK2XiOcV~E_>V4B@~HFogh6=o4t+k8;kCnE0dU|9pfD6sI(qjWWl
zKioin#Q)(2YFOaGDX?9nt7J(MYVt^>vcsa8*_6qQwZ-3`)~U@y(Efn3Yso=I39VU1
z8+NTZ4IIpY7t@KIcmGYU)`Z%#eC48+@eCXFiS6Gs0{7L~RFHyP)<t2^qO0*O%`n5-
zL#4lBz2eQptYd?Y)`pE%=TI-+`5M{FG-%15t>|mWVYu)VU#m2>M0<j}@vzO)<g6+o
zUc=gi+h3}xGZ$L+FFrj)l{)us-zV#f@A*eJu<f`0@|CVk+g(RxucxYb;zx=T+WV<B
z_S+O3RrL2RKX^7$PZ-GCHg*T#tlFZxH=t0049SyI2GX*<OZb|9rorcm(FbH)@QdDA
z&=a|T;;c_hlkP;lj^yWmXY61ios#u7`7Mgw5x?&f7GjIECtfnH-NAysR}VRmmS@ft
z^MF?6TDI=E54Ui*L<j9>IX_>=*(r@uw0cp8+$Tj-ZW!c+4%>H;r(G)S{w<&f^z!js
zB)xp|*@Ku314${Bt9~$sFAtTdJW1_bxlqKD{CBL58s`EE7HKWbv758a`YSn*xK}Nb
zeFI}``=@o{xRDf~fqzGF_?g7()}C(a_<Yy#)p*;S*MlSp7)3so?(}o?*qh%-(X##m
z;C_9r#=I6apv_}+m9p(qd+xo(-CU{2mt8WO4LxyiP7wLYdY&KDA9?=T7&Vp}i~OQ9
zY9eIgrj<$4`-(X;)5*Q2?qj0h#4(~GCN1N!o~NSsKf-;Z1}N~#m&Bi8k^AC?kmJRr
z{>2qiZ|~M0c1TMH7~moMY0h7y59REH#m%-6QADpxFnn~_8a5RyQ$1)(f1Rbz(|BT1
zg@<pCJuv~RA+c;=N1q>Ul_{j_|Ay?2e#n}AXKA?A-FLt`OldRG%<b7@-rZ#%7r@fo
zZZc%|Xs)NmH<_Od0}(3dhxn41|H~L{b{2FW`^k-j9>HzCsI*<gGzQ>Ku91|plG-D>
zpFe~sr##WonsMhkdS+~RIIj~&$EPWE#XLQ<Rp2w*m2e`qbBG7z_{^?(t70Tu@V}Bt
z&ByA`E>TiZ%>jg~xDR6U&fytVHZFSOeImn62|R;PR|@-15xV_5dZ>15gDg(2t{KWn
z<Gb;a_<!(_XW2EgR2{mD58Q80L5KFnelhG?Tj9tMx!MCaA7JPqC3CrILu6H(4+b$R
z9kL;9R$LD_vU78pot&K;&&#er*>^r3*5}~yzITvE-1x@_fI&92iuc~*5vyhANJ<Y?
zVf|Pr(n)aSI%o#8nlLS?A%+aRqkvT*$aJl-boaUjm4^eTnp+$Gm5gXN|4(4~|AS=2
zX7@`<rSp2WTS&*=V38gLQD<yyF=TZZ<X_=YiS!1%*#AsJV6(3h6AOd{_dzdZWN!6*
zEFEsQs=n&*brR#rlPBkziVSmO*PkaRii{1Te|@!3Yn$iKUn1Sbq{mvs`@6c3QB0r7
z|F5an-_r^;jGar4y(gh%*BOnq&n}blxG_uP;I`F7ALB*8W|@(p{h!R(Y9mP`#{nO>
z<N&@;O+lqZiNja_mD&~jN&O>~6wlgArAT|cF@hbNNURfK3n693H(y*SV*gj<P36)%
z<^P(!?1TU_GqgRRO7F96BMbAwT_R0{MCGN7r4vML&Y83u=%}X82Kr`k&5sUe9T-pc
zR-PPuf-yv~Y#>=6<mc0>AG9TE7<qBcsH&P`sTUo#PTt%2pbnc195%QThV7eowe7x?
z*YqRa6`35jU<gujV+YCD`*<>FjlPa+Zp*sztb)p`_*OSO;pQzH>w7dP9(A|?>#qXl
zGRCNkpT@$X{O_ywFusX^Pz2Z;;7D(2k<a^f4t;*7fs69vNV~Kq2&ar~*7=NhMPP}v
zX4;kPt`iEZ_fjO1zZ892bP^mz1JSnlUDCmA)uSU2(R4c2v9o;>CHvJ|Zl$Y)uOR*d
zp6z5+w?I+x4`YThQTdN%XOI4ArHqNmXPX;I_x9n*l|+pblKGF7kh-h&RugaT|ClwV
zoIUEdep!&@jr&a4RNo>UzbEsY69-zZ0H8tbBSQC><6a`0@3lKgJ3ghpHKzu|(G+Q=
zDH+*2SA=inc`<Dn-5%2_;oTFc{*L9Pko@HMZ>Goc^16dNf`?OD+J{<iY9vkBRSWGT
z_S}lEuHEfF`H(^QZ(4k%J@F1WAF6Z0&cj!OjyUtjzH=SJiR?9lmAaf_(tKan)Asj8
zO6*mbT}EtPX_5Q?`^va&bo?7Je;a{&EHffOX)uX-v|~=v{?YsTIean9`@8L=BCZ(2
z)YZXV6-YB{Hf}pNgPN&H66Gw?vd^WBgXeF8==@FGAv-YJdQ3?eX*QS}cINzGv7jf!
zmr8zO*`?Y@^)I#qhqoV`q3}9t7HM(j6ju9~@vI72&_(*;yIZ}OrWt%uTN8?i>e&13
z(qdz`w%dzyGPCujI=A_}Rd}z{Nbn6L?`u0ZHdozX5?Xb7o#({^?^jJXFr*no<+m7B
zcq!YbXpJO-tJ{zO?p-$1DR~bpP#AiTd0~$JQ*{K9BekmsW^2Lh94WsuPE6rg=)wp-
zp$2;Y|1OsYfzmJA+ornf)5R=Av9wSI-bN(LT@G2iKq!W;4L>@eO=1~so|vM><+Nz1
zA&Q-wE)elkN(doK`F^|9mPec7fb$yr0o8~`X;um^=kLC<RU?{se<`5{PfQt5d7{r0
z*)}rK4EmL1q+HR+p<QYbaHF}2L3ivHT-TgqxTE4~z2p!-|JWVc&p$AUTp^0*F=NX^
zx<wE6(>$+NE|>`wV;K`(BWe>9JWsN*%^Wo?j%eFFcdp^bcm#Sxme0DxJh$@g`kRR?
zk3`<kJ`l+3<qCeW(BzFsgRLDzYV*(YvR8e5$++e2$3^Vvj~ZE>(N=qRWoXI7t!nw2
z;=GD?uB`knk}bJ+PLCqUkZyaSd6+3g6zeI7k{|kM?k7wJ&rWb~-D^`xfzD<>_F#UQ
zc`@aw#NUXRMlCGCcM%eMVje|%m#a0)qeocugnzi%UmE_IZK61G>7aVzslEhnlx=SC
z<9K_!4=pA}VIBwb4OBm#r{TXr*#f(nBHsDzG{UthR&?hph?2Qsc<<T}#nChJ0MNGi
zx^gyi<$8_qozjrM=*EK~BGIx^0mCt^UPW~s`QyRwUN??U(KJg*NG*4Us$9DpU+?P`
z=8`7VRPy|@*1-m+3Rhkx+a0i{LkWJ(1K+Gg2^z!D@9uL+Pq>XxnH={MAAj8%brq(~
zDHKfSwjlL+^3`9%zqiEP(sA&4xbPW;zXNmM>@C%?em*~Z(^J;j{sIzl(=W+Sif@#@
zv+vt=mYMCrbUPdn`n1Zd_EBROxj?cna(L3~g04=@`p-nt)#Lh6Pq>4cTUVTzw?uS_
z)O(6sR#LaT#1*#fI*!}0brfxx)h@?<NRza{RieA9B|zCqq}(UTpe#25rcIo{`POgS
zoz$87YLFGxRU}{3mG4B0)si8TwZOsupAf~SOzA2%11lCqltZr&5&Vm(Zgsm6Tyr-n
z5v+-b)$<60vAuf=tqv!hD6G`8J>8Is2x+#W*AR&-rTPz5@Xg)fJgd11G94vz=7>~h
z4uaFBxB*XEp8V9D;8^R8$JLXJtU~f_U~c6;JdfA2+lyGq%ec7s@8+$d7PQU<R7ZC)
zFTGItd$iA+<X>6BkCjmLsAB|sR#|YkNZW7c6Kgtf9Cl{M=62I5c)tIV%fL6q!F^*)
z&?pm!ZtXzq_2e5C;o5$}4aTo`MC-G2p4`V6MXoNw$zpt9Bx5VRF<dy8ZknZskTy?d
z8pVwuO(4K&D8e8gdERMC5%(7?Lr9dzB%dB5Hf+*6$!@b^&G!itdo^h2@J*-aEby|W
zkfS5$s;c|BNP?I0^xEX0OkPoPOR*ztJkob`rgu*I4yKy+3GPr=>|1hv82-RC@;1&n
zs~sqD&1su#P9p|(KGsJlJu{!8MkJ)&aY@49rsT@GM1vUENg?-x@5N9D44K@!;@+NS
zWwHhA-$H`xca~dkGHX4!JW*gxT$GF#Inh)oR2{Z)slG=4ragO6at$p%#{<;V7!d;<
z)&NB(<aXM0zMuyl>QPx?ZG=1+6DfIksMaF3!%>78(0<FfHH}`dj2dB-^rrbiUF*~T
zng}#1&aQT|^1<3~wHXf0@xLBSvrTUG7}|-o7CouU{w^wxZtiQ#u6>6gn&q6FdgFX<
zrgs||e4|o~-{;YFL@_z&E<^-J2Bki3fj+Y$WxRHV@}9=OieSd4xHdv!7dCpcmc~ai
z?1sY@zT3}5MfMK6xRAy5vzLViLY18o%7@&=fjw6hOupC5;3XxuJ3&9K*IPASF;}!{
zI+IpxoXviueIrVqn*g~qqE#mWRUrbS)KBI%-csQMFU#Jf*3JmMZR4NSP9^Ki?6?&z
z3Iq3yRtJ>>1>;ncX1p-c=1>!PVAe?2a#eLS68j_f6Z<bWLg~Ngs)#?wlPR8(sqPCJ
z1PJ%zTfyee7z0Pa?Ym;j*1WW0PXZ?j2q(qQD2Q=ay;EE7JK0S?V5c~^LK#=C8t_E;
zKhb0JcKznz@Bm@7+d;&ivJ|o=O6-^tlal(*lc=4C@NVgb)T_Dc%ki$o;a`s{pQj5g
z!3Ybji$?W<)LoM~g|}3`FE4_KCP0khmdVK0hn?fi0ok{?J}$*&k^_$mxBC??=PsIR
zeNE;-c1j-bx>Q+-c$W#QuK3Lz1qALitBZ?SV>w#UBLFJ_)`ft=8wxU;RblJCTS(Ld
z)Jc$SJ|Su!8^m8;fFUs0|Ah$qf5=zjfb2^ERLJfmiXJk}6U@(P^vLLEnD^aE&+2mN
zmEh82)dQ3m(0IAwLi?34I*qy5)r|CY$uI|99;MHps<Oij@!ag}^y=Gp<vt1d3Fo<n
z`qy4}_k5s>M#!G;JlB}6#3odL3=N?O<a+%~kV<+6rjH|vQs18qfwo={F~8qHof5zQ
n3Gn<q%eVl9erEzFT{*?2dcPr{%1caq3H->xm0?AahCcrfcRb9{

literal 0
HcmV?d00001

diff --git a/figures/lena_gray.png b/figures/lena_gray.png
new file mode 100644
index 0000000000000000000000000000000000000000..fe2fd39899ca9088a44bbdd98625528722d870e9
GIT binary patch
literal 42588
zcmV({K+?a7P)<h;3K|Lk000e1NJLTq009610096900000c^UV*00009a7bBm000XU
z000XU0RWnu7ytkO2XskIMF-#o4F?o2U@960001BWNkl<Zc-rK<S<iLravd~ARXxv~
zbFKBhU%T7W`kWISJ5C(Oia|ia4N3x0eB3Bbg5ZMRzzw(j3+^KE3y?tK8j%2o#0g1o
zoD(;v+57D7>)vLqHRpV);bQYIQmWgMN~1J1s-#i;9{@^hEo=k91XVhk)Zl(UefIcO
z?6)MVN3<#p;VwG+d|k}7Z3pP478QM%7dux`W@%|=i8&-JG$iJ$4lRaQCac&Sl5}gH
zZqXuTEIv4`4e*p2I>d)Q$f@77KA~G}TVuc^npX&sOij{UG^_M`w-XxV`udOl#ZVp<
z(pOT%CA5mufCTqLB5AP=5qbV?eJA*LJpP>rKz}Hi;iw2JWU!Ss6&6urJU%{*T^(-I
zDy)^<3vnE|3}Xj&rH;ONX;DMmBAQWB(L7bKgOr3Mh-67H1?o()l0Zp?pisj?YZNI`
zo+?OCA|s-jg|^Wp)fFXWOeqjqO$7CICOl;kB`L(iB|9OR!X6UQsHJWwH*={#!HV;=
zefEF-WBuEOY^B9Hpw5=ry7VfE%x&yrkJK`AkyIGWkGVCBi84B;lq;*NW@=DGssbp0
zQ6?Fus${W{38BSsLsJYYQ6)n~%F1=`aEmrfvIInxG9W|ByvR<HP!y3uXOfmp9933Q
z2mzDZ(k+=v`YTsq^+2dc^-*E0EQ+h9u~g7R_n-c&PcHhys;VMuD2B?e6qO-5qI}(C
zTZ49p(oAmbYjG3PG=m`!)oLE8m6m2}v05}*FbUa1BPuH)l#0%cQe|VLCBs6!2M1Cl
zRfdoSR)LzJRQXCNqzyDSF*Q*qpqMCtpujR_!L{hnNF3CJ5W#l0RcUr7$ttiarJN9P
z(xTWXcc1?FF-(LsVI-GgRg$xyv{n~o+CFqIc~=w&=N2NV9)jAjQBc!fJC6>MxvNML
zfh&W8WVeihAmO4`gB6BQsB4LLHn0on0t=-Bs+ekQ7*KehEdvbQF6+Mg*y*O!)PiXP
zL1C5<;w;q`cy~V{P+XpU<Di+zwMz;trl_2&SfxWOHfFEbU;pId4E>=5eHbbnfT)>i
zB~-$5-$!g3rY3+!a|WjYp(sc#t+7<tUR?xXVD+d3K&vQ2)-;4&Q`cT@P^gH6lxwxB
zCJ~xFtH$DN8Ds{dX-}_!CIsnHYO<JUs!>FxCWvH7mC-3xC|3EDQ71D9ggU;oLbZ)T
zZ>WkXcWu=*bd;$Q?|$~9P0&;SPHH7JlUmF?mb55M0())D$FURHxlT=FMQf{uLZr~j
zc2f~eOR%Jp%!Eo~9-Y;S7C>>Z_6dMQlL{B1i%?ojJH@54DCveu+l*qYD52J7u5omM
z8#{#()dOjoQLq#fMipq@c8$tx*FLL4sUmMLWAUYuLb^9}N6UgxCh5}g>Zd=-Yw9lg
zLq%w@i6kI%HK~;3Fvr-pZM}#<+i1D>O(qZ5>eWCiD%x(TDHO>~lq{r_csfe0+&c=X
zJv0l@T&_hl-6TUm2^34xj2QrxM5Pp??V=DvAhTqJG#IApu%JaMp(wbD8^m18QnV1c
z9=lqxpyM;LGNh$womv)dM|N16rM2~|Pky}4Xk_YdW{D_}kv&NQC5)6oX9lJX*D|VU
zA1zX18oD<!=AieW#Merfm|UR33bBe7yp|w}t%wv!NmtK9v5G?h1#*diZV{*RJhW0|
zKnk>ig>J<cqiT2`rnAldbrA{%b<h=<)<z_<L@Si@)efqpYT5PGoJ(1Truy!EjqJs0
z6zSWWpZ#>bg&+2U{&uEd7YMa>p*6x`$Vv>^a=jnuV2y9SlbzG03CpZ|_t|xCI<!bl
zstn5vWc5U8X^AmHEBYp=l1g!41k-~8N<w3RFl8{h>xiLcU~xby61;F^9=1YErd=SR
zRpwPlh54qtz>y#lDAt+Vu~Se~+rL<)w6Wk~tY~AwtgKS{+dusDs>2c`P5+KWBLoal
zG!vF33ab#6V>|B~V9|8|EDGqV6^SX-xHmR)8^<e<#Mgo+18NmTfSrlZRIfBo$svF!
zBxorZjjqvjNI<ld43x2gU{-eqLpEreO;ypL%4#(<#S=oPct=DFKnvAs*o}z_v>HAK
zg;tYWrHD(m5yWES^&jb7nQi6Tsee0(uQ^4hm68Gvf|Rx5aea(SsZmN<>Zq0?T8*k@
z)sm&8mzH-7RaA*ra(7w|nNZW-hGtfZD;q>d6>8X3T;YO}kStV!SX^6>Wl|7iY4@R#
z1YwFM)JUdlB_*YZ7y_q+C`yRXy39wDQc>ddDnL#-g)qt%tW@;E7eD>vzFl_<(aEm*
zyX;k9u|#2JV8RPkl8CX75hA@TNtB2<LZwuJ&P@%ZZ*}m3q)N+uW>!OKmtwJ8eNoQ<
zuq%R6L#v`e4dQ7f5LSVs1@uA&3UFmGp^-vXNuLhgw3T84Qc6<yX`qTmiWFNGQv=N6
z$qXPB`=`p%5Z4N^+Et+1;){R#C!+-wO6})^{w7hPXem%_LR2E82$Nca+eOmY!+OlC
zBq*$4rAX4CvRgGt<seuo>d6E(y);9HP_{w&dcdQ(2nlJ<RJAk~bf7qd=#{Ou2|+X&
zrhSJ85CXVL!vtuhrDRxipdhR$i-7deouLKhwschpRP5dBBilgsJw=i^X@|f0(NAd9
z(S}X%rusujNFl_6lOd(dqpBpcZ1{cdX?=9m?2T<BhYD>@g$2S5)S;SGFB=xAr7C4w
zrDRIFHD|AAgGwi26INAa5k!VU0|jn^qy`d%xXQ=~6@Rou!49C52$({EM3vRKi%3z0
zsiY{BQGS`si<CCKyqpK5SJupE(FoDUpZ%<DH#<t(90;nIhNwX`%rFXB2qj5~@9T=O
z_O`cS@Tlu<jH%FYHBYEXS`t$Y>gaS1&f0t%Kslnkl}rha^Qh<?X-M4E*rha4l@xV_
z)|_>aR7n6Wfs#OJQ*VoOktSYD$SB0lB(hQh3z$_%RCywjwJW$@b(%y=tj)Xcmusd5
zBoDNfU;fK~eZI)~Aj=Y-V)L%Qm9U7$Mz#dSb(1|Q!i2j#IQKY~d9qScoTdVq_8G~D
zK2oZXRq9zOZQs0!jIFY<z?5BF_6E}+O-d1r7NH1LhL#kEc#&QRB1@?bunpzLDrN*5
zyMgG)P$EmHmF1F{0(3zN%PUO=<#L!z1+?sM?nSK9N;{eJ`sJVgS)$8Klff!!p87*+
z!;pd!Vai|u#Rj2n8z19*?iD3CcuInl9FnLO+Oi4_fVdEmsp_e%v>aX3ljg}5OoHWf
zawEmFGTKnd%mSMWLX4oKRH+8akYI6B%##l<aH?k1A`}cl9Nt_|5osk=D9ut^6Xn6%
zCIYIo{pX{x)Ya*8-(LUZr>&}3#R(#iOPT&%tDq6e7KvRUF+>D}SFYPb4xc%97D_-Q
zq=AcQ8WcTY1E$U*su1Q~YOrsWY@x8C4Xu)zF9P<h5=5vJhkniK8bk)b&>BrBWf@JQ
zvN*yw>riP{LPapMg{52Q&eTc@1(H2?GLtS4j^kpKqU3US$7w*y*o$xe<mcCEZ-lme
zugH{O>TeDy2b!@)=s-ly7{(C9n&Z0BEjk>HMcWqFOsOQ4gOHXG-6tDBSK1=mKp!S-
zSsGc2%7n}{WCBoC5=eTIqFP$nl6#|U+NUASoR!t3lEr9V+Bd*}*?UqT7$lPnA%+O5
zvZz@#WQgus10kpO|06KINJMeKrh4;}pA1Gzo27#ZEg*;f4$(-lh8S5<Sw*EoQwFcT
z#@bX(vyT>~=8!;3rD_?h?XA>Vfg%cmp{OEguENwxG`FIZGBTt0m;Pk!UZs_K%`A^>
z0zp{?Q+9F+jY@OKuw7(SN!7w0S^|(3u!^!#t&EB=cu8TPh7<|gjx(tc$?5K^GBr-H
zFMsmWQIq%9U<hT&2wLjzhGi->qP9XL)5OS7WER%TTG^e_b8=DFARz}%sL2W283>g`
zuu~ExilV+)3sPY|4V!&$qoAkB?Az^Ew_m{zp8kcyx39bwkc^&{H7H#i0GLCgmAKN=
zZS<iblQyLabqE1~B~__RNl0>+D2HVwYVVTnCsQFQ<fjSgGV#mL|7@^30Zv}2^Sw?%
z3;pdvP$rs>^ig!{EL3zNxz@U3m}P+!`q#Uzl8%ItAR%IqXoOOF1OU`yTD2XeZtm>j
z-@Ux=ynVjh__1<2jt|E#zjyoIeMMxIF<U_kMhzVX5w9sSjRRl-X@oUMp(Lr2Kmsh1
zkW5HLMlXtHmq?EJL?b9@{uPu<KK$&DWg}v;B`igK)SkMP>2FK0nwXDpk3^U$Tm>v_
zt6$FRES#i-g{smd8%3hG5~Rh!Q9h*zVF5HLI57Loig^$2YVV!eOP%+4wmf;by&mHB
zVaa{|^6>oGx1UOkF?82RZOLTY$z-@zNJNPygHBbmC8<uNc^0dn1ki@o5Zl6>J3^co
zMcCyuDM?l6{bf1Wmw)={wZ3NU(4xAml2xXqKdfv%N&1LA!Y-34r2q}x!uK`9rEN1w
z2D*oq1I4B&+bxwQ1IX?Mrn03BiiB;T6MR0MCe63BF~8VOR~@IVXTz6br*|SZAAS5m
ztV6w0Gd(3;RN5d4YiYI|KuDssne}92wdg*T1jUgp0*>IyDjg+^YN0~S<Uy69Dh;oy
z)<64fXCu_1%`JECj3y-;^tV8EN{n3dyro*FP{~5r8khaL_MCDeELRmrk7lqX)6mG!
zAqsJW9<~(ltixOxgcH>dH<!xeSG5`X!_&K*pw7c1aD14i$KLOk`5PbIoh2Art%_}6
zPy<7RDOZWI3~;824NJ`ynpV<ETr%MyRm6Z<6rxHJi1Pc>!BkR=^TYM><DXp?!Gg|%
z(6^qo&YHsN)ZdX*vx?#S)zfFQ)_`mzWc4~<x4pS?@7@YV5;9s%qm}NuFQe67sv5XX
zLluN%X9$}e;PcJC&4>F_n`0cdsGGN^pT8u}H{<gDBaiRe>1@l*hflxS*mgr2HbY2e
z8kkv<)i#YOXPpH?x=>IE2^IoK-7P8rX4udoQ&C`Ie>$50Vxm9(@xOJ(db)_As-*cC
zx0N}T71SS6p;lu=-0v>)B1%e@R_t=wA1^yprVLcIgp6$=L!*Ux144Oe{o0(=0O95v
zVt?<&GR3>Im5yb8m@jyGIUtX3?GW?hvT?oLd2{VgA9Rx3=i_%D9#^xJ(i~w`#>~hZ
zmFc+~x=$h9Dugu+^Ax%i5>^=&^T;xkqz=$6m{RL>OC?E>pZ+6e9AV=iKFK&rO2v}V
zE*+`AW!==bJui2cH5Y45TNn_U-09~vR8K?t6da|yvVVQUHyg|rIoh<EE2D=+RzQEf
z-%k7fCZfl1<6}Llh&Nx(PadCvTCbOBJ3h|GYt9JkTk7)B@iVup`bKD3W|oYAv_g~*
zt=4r;Qio(pmYTUHIN%g8TeV0Q_KDe2#fn&;tC6Nv|J%Q05u&42jnaK{=r`Xnr+Gh>
z{x*ylV;k4|JD->7pau*}g?L<df2=_nIznnwSK3y!m|LXJQ56kISQJXCY#7^XtdHw!
z-Mu(1ZOzAnY*DhvWj<^&E~mMsO>g(}?S8q%z0O&Ay!qDt*7leaVUA>#NM)|joh7}s
zxlBn&2?YU9Wr7tcOt3~~BLRVENtUY1v8zedXa8`Ab~l-<(1<;$)9VcPh;I6O){vvd
zcz6HcOP^ZTvMLnSvo4RZd8G7KvRWemwaOgm(bf{ylmbUfR3_a1%Joa0H{X*v)lE(N
zFrBwUG#Pg{OuRW>j?1}jmh5$%4`)6-hqkfu$p`)C84+M&*F-t16!4y9iit2o2p~}m
zNiwZ=?-kjzLd60ZiWDO!1milNTBE=FAAi5^;5-`Oar8`-p$%8j1-nor10LJBTqBGD
zu&7yBJ+kICyqk?ZF>y!DH4NLvTBT(1-2!P^dp7Nr_GJ5m%{3o2R93Ei^=aeQ?s<7D
zlds!4;|JGQFDbmlay;wd{AQj!+FY0T!GHcF!z0EpFG2#?DkU8wf~o>tv?SCL5!_j7
zle=OUSdC<gfsn`=37PP-5xL#}pa09j7mdL-`DrQclu7(9OwroqoiK;z=KH?h?I|pp
zH({AVGFBOoYzYO~MA<KF7iTxoa$M28Br~bzHBfc@)7Ijumh*(#VrN@N-(r%n3>y#2
z`V6n0AKo0UZ)dFI<-_`H-fG6q2G;H0{132gm-F!2va$%Z5-hDHGfah928|jv^6XIq
zA}nf1q?&6Xsv%lIXILLU+xENP|L@N^hxVM;tL^C*Nzrmkmp<dV>E9Fj@cr`m=~nZT
z*=SnK>|oYe_v<@uw4BpsOhX)~W-Yz98X^L*%4}#~cd~qXcM4S<K3^?tYO2@MDyK7|
z?VmmAa=e_lJ-ckrwl~jk=ECi$SJTOdHcEf-?C0BVQRR}+hC&>=4L0<N$|8%wy#{61
zf>s5TXeFp5DI$g{O+ZpLpT7HF|KVk!u%v~BvM%k=nvm7>*yF47qJNL+*w*W-hqT*i
zo(^5nk{F?T#rb-!=yq6JANyQ2&}xn15(Uv>gpJWMB)CbmmtVG8)X%=SVp*=IN7-(-
z8`Rx=z>6_I>byRAh(*TD)4O^ho1Gqb@}THd)goSe`13c3NE=gfPj~G*kkmydm<Y-M
zWO}&+)S@~{q=ID9f(WI;ENQe?|M-7eTi+lZf}BPln=YAtjhp?+3HOhDDO!;+&JPbR
z5X(Sj8<K;%C%rWTmDlFi)HT}FQFiDoHHw5|3Y{QCqNV)CXRNkAUeHpOhOb^;ul<nF
zSGZYTw>R_WPaoI*#r$wL_lL_(zvHpa2fIvG+rG#ck^b`cK7NApdalfD0*N(B6j9KI
z?9_x}#CpB%g=|U+XMAlyDRHq7Ff0<<?DGHo?>^T(hU56G>B@&Q=M9>N=B@3o-r>E!
zp?`PmG4^+_A2rR6leK0@Wo0tM$Ggj@mNM7u&?=y`v|UlIKy;Z<D63&bOi%x9YY%ke
z!X0CM{!pjw_;%T!qf1rhHy@{~$-@U6(_B8cv-TX?*VJ|`-QSih%gtZ@)nEMNb%Ugc
zV^p(5_NgtZqA{_Ad+yt)(6S<^fN4>Rq#`L&RZ^;HpZ&orgfGBsxmw9y)1jW6uR+L@
zdbQTmzxI9oVMLuD?!GFsw5QX&kk(3z64>*C^<f@Q2HKN>W?M1I7S@#=iOO9-K!lb0
zjX&OFwcT{)&YZcNZHl0fm*w-9$!VM(Znov=>*u%U-VUE1509*yJMD6oei`kfo72>A
zJiYktKlPy~>`tk@#_V;|L}IB*n!tU3bQx?#^CAP8P-a0&T|pBgrq}=Clg*}$LoI;F
z4W*siUI#z%+2u}-zx;z2`deJL%e!~0k$G4SlbY&4N|ucD?Q+R;kKXE$5>hUPDCkfP
z8wbv??3whEru({sW3E=gTBN*g-pyCZww^8@e%i*fM?Sf}d~@?+AGdZdH|u&dTo3Yy
zTKDJA7cTD1R5|J&4iRx^a2Ql{T4*v4Lsse<<Fc=tcfh7JuPP}9ReI=_l9CGP?|%GG
zlGx+`9ySJ9-Ws=^D4m|R^EUm`UwFFc?~FJ<zI~-Bs@jgtR9lt}Nw3|v^L40AO0D^3
z)lCFyhL{TwfUbcOb2X9ae74Oy7iHUSc$LCsaVeSos_U~i&-Zday!s{{56?fZ+kE|W
zejIZ@pYz&3crvG8?*YjwIeq$h1F|{25BgA<7J+hD7R%enug<HhbS-fi0%n>xizUsp
z6=qp~`hTpp6&p)R&2nosn{5Y)X@cQg-~aP38uV}FdO1I=tyRf<)Tv1;WeU@4+b{BV
z15M`Wk-|KfXVTdy%_%t*Rg&xu+aLKjnRMH#mbgsZTX8?`H}!<Ay?*h~wsv^+!8_~M
z@l<wuQFl-D`ex(v4{y_uA(_0EmErX5f0j|QHn&VkE|}0&jDcdti0eb#Z*sEglA5Yy
zngkU!SO|*6=l|w4&uKTiVG@r(OcYrYX?mK+kdOb|3lr#n7VDe0Z#o6p_v5m(iI#@S
zO8D3>UNa_aqzzEI145G&ltIfOkVGwRa=z|!lLJ>dB?uYIrSsO0m&3iZ?e_7s&n=#v
z=fnNs{m(e>=lSr{?eXU07ti;q0hrUJ?_u8jbo!$WJ*#5bz+&sB6Ii88aqPZ*b{)qP
ztyGW|G*4(1T57sT*T;YTFRszWhsf>Zjr-bFZQXA2>66m;w)Mp?zIPmv`nUYu{ljQf
z9eJb2rbf0Ql!{z4uGjr4*`bjlDmj{)P^9mqU|2&zDx>>(@0G3AwmtJTtzX-@_h@^a
zVmb@%KX{mR93P%t^~rWQOs@~8cbEIaM;};oj-3`MLjqF-eX956kN46vMyS(fn@xKv
zgDxQ>*M~>$dCH|GK}d!r16379=}q=O`KOVGod>;b>n>S)&AryhY<->kx_{#rUo0|m
z$mM#u+gnPNBGjQNHN?XeE@LIheo$kpSyRz7GpM<$0Q=m`w1^<ru_Z*H`SiM##v^c}
zb$h{-JU^HA@;u9Yd#{hye%_yL<2e24m+wCQ-c9!Xdah-ZBHGYgO$;vl>)+wN*1E}5
zU^vqey^3Y%VW=bSAI^7|hY_nu2m(lBNfQB3pZ($G>Gw}fw2xhTVYAi@hIt(J>gXSQ
zw9KO8Zj$}-_}FBoT}@pejAWup2Ni8vB@*kdQrN~=eS~S%pvKU$${j-2$?mUY&6a2w
zf%(D*8C{NVj?&up;)*&?=<&|380$CZSBJX$>AUa!()&C7JnUBX=@MCkSDo7|syf+k
zAN_iDTpqV_wW>~%rP^euLLzG0eY*}9^fsv&Sl$E!Mt%6>PnS2JJgN@*9*G@%W)GX|
zG+#t@nO;0yEF*3gS?}LTYS(O=_Y&VQLQ@-3gjv(j+n5d77Yl}sxz?tsxaA-l5Gl%O
z)7qz7*208WT&C-dbve9!c7E!5dNb?v(r-V(wBKA#H=pNxdHqHGmFH%5j*?}wa$K*T
z)5}G+IQxE8`Q8r~T->wZlF<p~NOj4O+-*a7UEhR^K$h*|RZI2UVT6pI{b@W(dea5@
z6zlb&di4^W_pa@E9zOc`cG~J%$4URvey=7jnW~(RT_<XaWT@xRM_(VM!*EqzO<FWC
z+m)(dx4nu52<;l^g_u^gm^VqtTpCBeJb!H4{`~%ku-7Mf_w@edzP{*Rym|PAKPOs|
zQir|r%6;);Z-MnX0v<UkUoNj=VC%MqPCCIFfPxH-+8!V7+kF3`<*`Jna+k!$4%K3p
zSO5HcEyYx=mVx1!OP9BCs4*?I$>|%nH)0WaJE{K4qZpJ0jeLTGbZJ5miS+fFmn}S+
zv8kIT9V$z-!)6KzVp8+s?fuB6U)uw>!@DOnpP$?x$FsW|ySs61uik%adUpdX<8<Gj
zub+ML-Vff7X<S=td2Z`P_k&D%9@_w$X``fY`oXX0A6;cNVehjv(-g8&hwj_i@7<rg
zn(P=&1Y$s_pv<N7@lXEXYRtu7?(p)eYBP7l<nt)D_VIhSV)X6FQl`In#la;=^{U<u
zdJHj3^prjGIxZ{40E=T5teV4-*%+dkOO9oe<?v->s(DhQdCj3R4ZV3>#$?s8Kl7>9
z<qhySJw87?p5Ok(U$9Z@Z2MGIshm-J#w50+SgA7<K%Kr%`B{@N)48`LDKen);(lM(
z^|kwMNlZXTF;ESg!c@o4{?VqO#<TnBG(y^D2aQ!_#%MMzFTS<Rqs#Sh6bt=Jb8vQ1
z$xuBlvo=$t0Pa~=4Bv*SX--!elz`}NIYH4$$z+N=d3{|j%{8fZZF~2u<^IJ9{e0uc
z!{yNKp5969%XK<E-oBDI<5xcd1?}8ot1Pm;YgpE$m87;JSvmyRrG5Y5Un5jyy6K|E
zEVP^+w_L9ekIo(58%{7SA{i=8X#Vi~mn#}i<_%Z&4chf)9mhb~5c%-CADCNRreg!3
zzhrZ<y6&xNYKKXt3S}ah7{0FcsM^ya87J}1kn9z#1f`4>@y#!<V@u-1tZGxQnszy4
zf3qCp=~x@;@b-Xc`^|>KSI2iBfB)ImRj-C!v61P1)3eQ23JW5>2{Jn(J^MF4`w`K2
z?4nbW#Xv=k%k_L8$joeZlAtA0s?0^w|KES(GNo$twkJJ@aNKPcJnhN;(f3|V6>-&5
z(?m+GYn6Jo5!zIp$(F8t03{anU{X=qkn^?r*yN%km8uz#P0N>G(+ALi^I%optrxyK
zdK>r8#y0Kcmy|f9)nj~W=P#50{&)6Q-RnA?J05X(ciD7iwfj7Hy6zEdT}02SeeMtQ
z?>^P=%7nV2*n~bZ=hj>1ZRhpz`pG5tVaiN|g7UL}RX~=PZR>$77wOENn)ku!@bQyb
z_T3LRT|+@R^e$3<9#o{3DG^FvwG6}6mJ!Mpt1MMDY^q7js$#OfU&Wo@{P|WpmYj;E
zv9L{X6E~kvQ{dRKy?e1y_IAh9`RTsJzxTZ-(Vwg}JWuy}*zWs$)C%IX+VV!Fr6h(A
z#CeBv!Z&{9L%l}sF32JR3K2NV*w%{e-Rpf_*L};Z1V!}qAKvf9dfqd7cbVOKlbUQ=
zx?T5|Kltc~ja!?#xF$rh5=frlTaJKIgeizZl}_eoAZquXgF*p7hO#j-w9CHbU#$Cv
zo3n01-TU>1(*fj%D_XPbEBpAz&vU<drM|UK-rzS*r@LoIp0|gE<M!>^PPYOb60GS`
zCyY!-?T9kd(6t%=%a?cJTQjeS&}yZO&_uSJlDgi1_4?g<t(&Jy6g#g!{Zyr4y5gpd
z2wc2(JkWK1+~2`&KRn66cDS8WW%o++(?icz&;)9a^npnnR_2Ip)9#mrNTSY*OWQOI
zGNNsB*_PS9Gk@kfgt)%a001BWNkl<ZT~uYwJOpHDjMIi29M13J<*VaIul4xV;h~Rr
z_=Rs<)N%CJeP1>`fBODmIxk!?jj6!Fo+fdp6Xh5T*1$T;2mj!oU$nx<Y%b0$9(F?0
z)Yr+|<K0%<$A7&)<M~JOjZZ%N@gM9y?}xG+mB4hpT_w{bK(KxLJGZ*;UW=edR*0E|
zCHK%Q4iig>Rt`WsVbgR`St5>(jA`7gCF_Kz*Pa*MnW<jC@i=izmtN7L%UQqtXvOX0
zlgBZ?mfP*&>E-Qv<@wq3?Q#0c%b}KS+i@&i<IT(WEUxoji;sC-(oObx4LOAn>*AZ3
zw-V%szx+>5alPqPfMjNc%yHF$m_p9Xt^b4X+;H8`-@n1L_CG($>GI(hi&gl{<(y(8
zP?UYt)AzrBGbbo>Gr?}vM@U`~(hzBA5m)-8#f<Wz*Ko-B8ZmtJN%G?7eb)2av)8Ok
zIj(&2^8BZ3y9QXxT#q;P@~)$9#!d674I6LmeT^TFr+3$9|M3UB+**4WYc2Y4{fK?f
z^hQVXe1KrD*Lvh(+N*3AtwPqNoPPrmmXVMN35B3?36_Do+~{}zn_ql7zW5ZkbCrDW
zu=#55&~?&&$%m`Dv8Mhc_f6(+esC&JKb)5CZp4tZWv><+8$)ETw$CmMX?r$Tuc(J?
z>@oS2U>K`%k_>gM%o_OR{S&(xdC0c9Wsk%9{Pj(F+nSa5g-^?`uTGaQjz9SD{Nm*!
z-{`!mzP&!{yPc1)&Gqv9k{x?H=X}#*dYq5fG8>jzA=sDv?#GX7jst@`ErqC9w<z~2
z*Zn{JFW##+pMUWFVNvRkAO0v=zUYOVo@^~r#Ld>Li1EV@4jAxdI*>$YR*77)vK9fm
zT&5`zD6|SO6eFa(@7pC(Q?m&S;!r7}7D`0^>fMuL?Hx!vryceg*PE>m(%<RppU#Ip
z9;O$M@#HtY({2ve^^3l*$N5L+7xZ-;E>hGwyp!;4d@((lWDnG}n)^VO?_=}Yaq~OT
zvK;6kQqCl$Y=FMw@L&DcAC5o!<8R+QnabU#(_h5|saTuX^*kO%@vJF0M!x*kx1XAF
z=xpweVwq)%y6kW@qaI|J($dn&3?<s2`;?WIFdx#`9p3j%CilI;Mm-IWhh01|*7*>R
z%bU~V;|cS49Mg~Un`(BR9`AVbYY|6#*!Iia>G_+>N6%j8^|)+@75k7_wtl_f14*WG
zpdJD);z+Hj&V%~vzx*(0Tcd!Gig0P?>vkRW-T&@y9^d^F{L0+M7;Ek;j~_an8JnE8
zL`}yv1~)v6!~Ct62f3~_b*&0P(swDrqDK)SyuzoQQK6m@B{f3bYB4=en`o1fLwcbx
zcFUI5ir<3XX8_3REp#mNtLM)t<8I2gFP}@keuyv5U%mgUw{6D5Mcds6um9w9yNvm(
z+pvC}A8s~?>!;HbarLWbCAIG&Bj(*~Drjvy{th2A6g4S3CyI@0Vo-nezy03y<^TP`
z#|I4Gl9$baUt@KWBIl`sm$#1xnR=g&`r_M9YbcLL5s3<wf*}p1KJB2YU(IPo?HJ7|
zs^X@3X@268#vqI#(VWDQd-ukV;`|tC^J?Q}<mVsU#fO>8y&f+1LEj$s_pj&U!>_)l
z{NV9%RXM%?$&>fayV9TCzj!#-10PQ3ytZ$ttPc2ooHL>GFlx-KJ(ab0{l;JWXPGUy
zs}%RD?TYh=)8F{}FD`%l<6nC^MLKHudV{d<GDh{X>mf2N_nT-su-c0ckGe%p6F9)q
zp)Og?RHS>8h`6;99X`bnmQ+G&esCIQP=aBFtW?#U27?xR;RlcBZ~r9Q<sjJf({E1Y
z+wsAtw|H8Ji@e+(WBc<@OY>d%et2>I^x2HQQ=V`7X8G!cFL#LbMWlr7qc8o@Btz<m
z9KKI5V<kt)fAo1qv}h>T;=aY*Cg1)0|Ka@bNAc^=YA?oKiX+p>0Cbx7Tz8pw92~O8
z@cx}|_eAQ<D$8OP8DGE6x*9Y~uE-isGA&!Nk_D6zbWS;zj%$m`DOO!P&80iD+wu6X
zsM0Qd#Q9s9`;+x<+4oaEyzlt<Zk!+NJE!9n>s99`_zK^;xTH@<?xv4V-aL6be{%YG
z<T{)7bIiOw-w1qbkM~cd8CV*u+S`lY`~AD{X(+^_JkFQpKlz*A-M;$l`HP!OW{fe#
zL?db+YQ$(`7hEl6-omcy2hYBF#FjNrrCtT3B^lyYJ+tipW9hwmc1^A`y_Kq-@P;pk
z9XfZjo2eNk4J8aE&5SvgY>X@eJ~1%oJ~#XebZmQI7%*VN25eyfLjxNSJV}B)k|>iR
zMNuT1O?KXSe>;5XjZdhmwU!UI{=_P*Rrht>1=!u)t9}P9WDlUssmaa^<fV-XgraL*
z^^7SJyA)2^(CTbClz{}j-fg!I>sXyvO&In7hD_tx;TM`E9Y0%5XgWSRXczk{Yxb<~
zX1!I{yVv3Q;eNuAJtI#`W^dQ?;KPp&Z-^Hx=!L*i2)6f~A3cf#BcZ`yV><lZcWH5E
zM|)9=N;V75LO`Hb6AfP6N~Ow)EL#0MQ~lZwCvavHH!=cO6IO5ug~d&>fe~4rM3*cM
z%A~v{Gz8`*hd>rGRK<$}767oiAm8ZgZK`d+(9Zxjkgrbj;y83ZUGMhY&UkV7lJUXy
zCo?ZM9{KCC@r$WDcbS&+E?0|fIJC<huf`0Fm}|~9FQgClP83nVfn7C=_QB-KU8*UA
z(k5m9)~_6|uAPo2Vrm(RyQo+|BfH}Xq~u+20G*hR*<7c$Z_FH&D<pUFdMj*{LU1BN
zH?x8aVjP5%0jeXRI*3ES?74sy!3>-*vGvN1fI61#=fahGO;~l4G+DVVj!lnktfTE-
zYP~t1y?PvbwX!&_evJF2a20Ya=YdjOwbVV)(~4ouh=r|F8<5HP;%6`49J7-fh`T$b
zqNxn>wxQcdPJa9MzcYPyL3X@Dnr+}>E?I)Og5xL9Rs_&P84b4NUi%PsPYyhzN=4?b
zG_adzG4)X_x<hsh24oH(*6kVq8wO=V#LB^0X$S!=jFG(JPAG@IC{J>i=c|VLx#F|x
zIvyd&>Y#)n^*du3W{0!By<Eq-e%_x5Ti23phQJrM%X0dBvVEDaXIjC!?57n^61o$O
zSIgTWMS<e>_g|xwGmlN#Wn+}A^*0W1eK}hk?=`A!g$fWtMM%g{Qu5R&m7H5s&l*e8
z`Q&pa4$xPetAs|yfsNJK3yb+c6aZbZ>>w-fi-RDMp_7!57=RpvvzSL885BinLvfVT
z;aR(BHaJb#H+^f%*;<c#gnm3+BOlyMeKRWP>AlTM4V>>)OiRhV-`q}jFIL-IN$KJ^
zj|dmBS?E04YSNuGw`HVFg$(QoJE#CaIaw9weL+pH-rQR)w7t4L^J2vf5fO}xu>=5G
z83zg=(yyA)YB0naZ{4n|(TFgbpgUnAEaIwIgsdqFl7&PIF#rW)C37Z3#FDDN&!+dJ
z!5mBNPH2eg*2sADN!=q@)cth>IP04x*Lf{dt+w+wmeKmmb^1teghH+Gmc(#gJ-dAw
zuIj_VJ+)U&SniZ9?TotcvuT*O*muF43JIsw<}C;B3yugcy*GiwTa7=V-Scpx)&c_w
zsWZ5<WfwuCU)C9w!ny(5z;wV|RdqZ$klvhI0u>Z>XLhh0(3z03)tms?$wDG$Ap#+)
zAiALo0)?4ViKqz-2oV@G2kQMsvFlqJVZaRz6Jy*6c458VX<J>)Pu4L)mxM0l)-?6h
zuI}ejDjs?AF+4chPlBN_&YPWWH60>@KR?}{)Wh?|PQ3>bps4B=LrF=paNf>tmY2Uc
zUEZ%#I&DHJP?*3d5*QC7qAL>IU!n6Pwakq~rvxd`?y-;*WHW%ARb9cXKm;HZb1_s6
z$s>^oc*z0Dkw(ZYx!TxJjbl_t2x!G91t%pO74w^H$M>6cBkMWpYU)yZnWVf|js0eJ
zdsAB~rH=P_K5h;Kw-x7MzvAKPTvn$<ysV=Y0bb3Q`%`O{ewds^+O2Y~)~$BY9M#ac
zWKi_s=;eB|2#0;!axy50l-R|wP|g&=*~$J^a|vBM7&^(*<Skb7_LUte1!G_YQ3F&~
z03gUnOo$_TYAB*ck5(X50)*)7=$>P?8oeUMV(dx<u;7?T4P0IxWEHWk;K2I6A@o$$
zX}%Abmh=+UIrZgOKW%PgsJ8Waw{_X+N;RW%njXm3>e5vT>b#!z)-Sn^p|ypkNhv<G
z1Nn8w02>~2O3!d(+QNFAh|G6z)S_UX2?LrDb{3O?nfIT9QH}Ze0z1yM<KviicUwae
zL~t;K@<I~CU<~9`v8O&JO#mfiMJOW?pcx>UR+)>rfU(t{M}i0%$vA+=g1fl9y@^9z
z&#JaIaGqVR!m`jR#g}0e?RoDa?VXOCk$yQ-uWO#Io@_SL!(pXl)Re1<IV-H<)lk=E
zYr3tnV(=V>veGIgxA_&Mameev8*+WO7nMm}Ogl3t1LcB%;c7i90H}}msZi_?h!jZM
zaofw`VQZc<1pvzk5?+w^&V+(a+w24Y=-@VB1^~)n>g48yi8**M4#p(_17r=!D(N|l
zW{~11Q>;ZPM2(}^6K^Ii+Ya9vG;GMH&!5sPMcIuPlk*vmZRz9U`K=kwFRv?vGKrfY
zjw)9kQmRPwU?fyRTPiiSZnm7$0MAl7eWiYOr#hbVWMZTBV~u2PATG{yebuF8Wi-Hl
z7o#R8ffHw(3<@eodo>kkVjw3%1O#(Xmkfv)39#{!U<@%UI!gf{cOo^g;EJFKk~{+g
zauy-a?nbB}JmKXnpN(zAc0CzL$Gu<%4_57K>xP!xpM3Uwf9^%RtP#Rm^3$++E_)U<
z)p1?9j<qb$wZf%fmgTi)@Mg(aJeOX_xWC$a;u;q^efMbdkZy!V0_Wm`lbexJAapGG
zy7TRbtI_)O{~9w_nE9+CNx-q*->I!Q$Dob~qv7Zb5X>PkTH$Z1L1ZH*mD;CLS0Ny<
zA#l>Mtd`~!DiZVnY`~zAR1HuE{dO0g%+Fta^3tX^4x8)!o_D$}O>dwUWIOx?PIrZX
zy^L{kuwEW@%Wn6C%eF9puc=LI1WqNkrMFdOBqbnq4ufHCFqq`N-gNZ2lhwj^qf`}?
zoN@sq0rP;$$Qqt*b(Gx?v}XCqcj17;NL3h<yJ3CUzI`L0nZ$s|g%|^9l|w-fpzcJO
zvS@Mh&ZHD~0XBu=3L^lMhl<%r6>v<&(U~ncu>x0t?dk5r>78~FtDP|PhYQ>3yzJi!
zISi@W|Ins;RSC4#Zhu}L8!x7tc{MT=8;IKJxC)psh;#*(dCF{xTbDME=)jlRhy0k&
z^Q|}SL0{D*JDf*LJ%ATdr56x2TEAS_{Pbs+Irsd(PDYrQ5OW_rLf{x*KV;3B8yHP3
z8Q1`lMFCkH2%R{`lBqCq%(=pdSb$ZHlu$jRbv}5ppsjFbD>?f}YN6ObwAg?8($$yJ
zWSl|odmG%j(&zi9xVBBB$3MYIOr`0{XfyTA;r^_D^m;$S^;jV(4out2qcB<}`Lf@a
z(QCy`*F%9#S72zbi*2fI_S)X!y7ns8VJtTE5Q?!NF;H+}_4>{zv%OC)rnNV#vn^j<
z>)c8Q4LOE(den%T0004L%|#snwK$k&Fn21InONlxX3T^c+*HUx5JU*0y-2A;?^tCc
zi$=s%=NXCTXSbgmeYD%n*7eBa&U{UmJGV_`+49~`0vznH^<lHuUoE?x(fd;Sfj#fL
zGLhtYK_Q;;WW0Q|(+C2?svdHax5aDE09)Cnvh&s(S07QVcPBM!84?FXWF!PIOz0rK
z*vjtY)64z+P@12+BU-r56x1Db>ifGhBli|f6oZ-tQUeio%VI`OQbY^60Y)lh%8nQv
z(VWaQx>Z&=C^EQ>6bpMD+&vcM{qgF?lUEkE*5TQ#o6njw#^oEY85E`E?vHNNr`W~Z
zb)m0IO_;CC(T&9l8>sL?ZDS0fovp^l@X}^NY#WM>w4~LlN3p)@;ra65Hhx$i*E>xZ
zhM^x72%Lp6I+6ej3B^xArjM@=Z+P0N{t%tr#&A561%<$;2d!E`QVgZB&WmRR08smT
z#01GLxEBhDTWJ){8DIcFG6QEtBh%pK=nN4N)pJr!2;Rl@@$&WV=EbS)=ht{rv#f5C
zYDV8{{_d!sHw;WhRkz#=Vc2NvtH!pr*RA+$<sk;N)kXhO-AykYt7<bW$R^L)tSc>f
z)T<kB-*|Gdi!<bu^X9rZ5;8dxD<cr2pxJ{g(@%$254KPJ;Nps9-eOgr%sH~kK1_E4
zDM-Tzkem$=3c4AnqB=%la<)2fP?E-+SWHPOBy<8|z^W~|xUv>bYN{+CjvU<qJ{MOn
zT_0it(pG)5?QnVXuzg;18|dLK9QX!jTdOwrVNbex*&fsdi(m)4b&Tohr;k6rc&w+P
z4Y-_G2Qdzl#dJJc)SSk=Syyk}8t>1!*=wq9*sM!XW-<tY5W(CKkht=@a&(uD^~Vo(
zZ^VB+Aa=smq92k>ZM$<&mw`|i(uhf!ozRKUkkJImAQN~74x_tD6RKj&W=@$=&5^0^
zHQ>MjDG{l^a5$>vgrhFf{%Wt3-EBL#?0_umUDShxM*83c_G(eRcDm51IMyFI=T>ms
z4&m;_YKd#yl_!4+)8bt}lA>(Y;*u|6T@A+1Sr6W-pKe2a!c^8~p^l+Ngg_3A4#)t&
z3gy$d`{``_{{OSay*>RCNyOe}lO(?FYB}7Cj=*Tl0)$yjjf*O}1B?tgqf{eg0yZ*p
zALecjrUBL1Ij6)Rse)2TF-HXjgib0Axd6<j&u5!u`{3l__M_?f<hngbnUJLVc*pA9
zu~eDm&Th*?Y40D_g%{<s^IaZScyw^QekIp$*u8_wu^ATI=?okxcl;lIXWKm9<NV6O
zdLz?TThAGtxrn;}kYXGwCpSVw_U&T+eDZ_mu|9a^R&{5Q$xBAqLbbNQ)zN7Ug`Axg
zy+|Su%m5Y)u>?>+P<}UPEaK#h<P}#T1av}UFc~~;I~%m7!tMx!5JpCFOa!jCZ#CN(
z?;l*YcVB;WyrP3FMa$HeW6K8}L5kz~P6@cTc{-a`P}V;gz7_Y%JoX>!<I~%dx?X?$
z*|o3d^~BTrduwPv8_xdyG0W=kmEHT;-JFG(3adD=1>~Rz1c2y90D{c+OWfZ5L8+Sk
zH{RZ-|Ne<UtTPm>6thcw`Lr^%&=4A_7!rG7AplYY1qX}3-!O2p62d?l!X$>kH3}gV
z(A<xGzm<T1K(#_3L=r<Z!M^CNcb9?JhZlPvfAO>1Pj%c*`)&f^f<tqVbOc<k+Zd5;
z2zxMuboby@#nC-Qd%~Zey1?ehX65nEm8ui}^yzn(Ki`|*{l!1(_ut%p%5*ZY!H9?+
zWQ+i03`kB){^D7u{?qI2cdw$v*^M_(<^4Z|8pw#WG=!L{o40o%Ye1?Pj0sg$z>9;M
z!wcz&nE-KyP+V%nSc`F~JdPwrn5quFfC*qT;=r|*+^VAmrwqiaJn6TGk52Bq@r%>*
zqs?<V-_vn+zL{1#y^&Dv_C2*_3wE%nn#JX7bxzM#RGw}ytjn|g@@#i?+%5WY=gH%l
zuVDP)^iNLK;g`R@b9ao#ZCxA(i;_SwXmF$=4u%XMhDdnt`GW`77=nbOFC60b4-~t3
zjzcl*8_#rfaKPCCW3*h4pvHiOkPH9;(aFg`g|P?#>kwIoPzMr#B5pt>^oS^40%d?W
zQb(bf0%C!ZtD-AOo85=E?w!6rzkfR1w>%+kpDk~Wrz4DYeOB#qt`nK<5O?{4()994
zirO8$_Dt%_!>6Ncf3_0ZeE#KsH@O-VJ6?w~{n9rk@A3W#ifc6jgOY+FYXBF@2tibw
zvE2RS(b6<x1b+R^X$=2&s<bIwpzMIt+A2GW*_<m;<b<YD;z*8((TNPeG%B+w02Jh^
zgNg`^wQ_rbfpNBIVyNh0ob;_{GbqGJ=8a;Zt1#}Kzj^=GFHWB|;|41o^WnuXFMC7G
zjdfNPEmU{ufJbQVl~?(EMK8<6?yYmYW6wUgefrbIt@E+@&8L4_(cv#;<MEZxzTN(A
z{{GIZbH-9^EaYY+BLTRhiBiHaqWJiukIxlqkICxE?YE_lAOBHsgQ~BzSyy4}WxhWH
zAmb2wpbS`<Bw}STa3x1VDnUU2nB!nQHN}NY4Hz9Lvx7m3Cbb0*;4(6Ju1CrYp&y*3
zTRz%d-+prQ(aBHrd}rFOohR$Q4F_0q5r4E(B24-P?(%@u13T$GuWyud+&#Oz*A8cF
z@9$pTJ*waM-t5gwYbRT>{-fD9zIbP~|LQDDvKNyW>XgJ;9Sa}^L|_>H&BF~5AdlS8
zt<&Ru)c@}9a)BxqHm%s=Xj0#pLAE-AVl0Gg2nYZP7#x5cbKqhkPSxOyUV04V!APAz
z!59lNCbz7G3`byBvI5l*H(&-zfSYFb`SH`8_ix<QcOD&Oh?kcKJjwGB(q=dybj}Wf
z7Q0oxfw;Lmt*>Vv|HKV@YHluHdi<sJ$>qb}{o{*%^6pP@?+M-dFTU{Jz0ZBIA7R5{
zSu9t&se)oAawI16%I)5tEG0-mf_a#q+?efw{@2?q0`)*eoUO{N^{ts=wy43lxC0sy
zfU*WNaD+mXnaHF<jiyRu6{7;HL1DL;NJc|3pqQFc8w3I%$I*wVT!?__FaE2KZ$7{E
zXy>!ZtLMj`ymGeercF(gRe(*I6-G74?IioKxwxs*YwWxGSDp`6+m>yAHeb-E**^Qu
zXY=NK#I0}VU;O>O?_PZQO$u7H0)tYOK7^qfD+Vk?g>y)YpI@wntyu7**-@ybQ{DN4
z%TTMfL){64<ZX|<b2`frL#Po9xCBKR&=3L`fuaczsZkL!%20w6Cg7MR=s;wU#SOs~
zG@0mLAFwdy$Y5y1i3L-0b+)<veDCqWXE!bvlZ%^I)#EVpz2$m3dY&UY!G@>P0s6dt
zh2=UuySP^-H&?4};C%AUpU#K(6lb6RZg}hcr?Zt6`r3aIezN@L5%!~Eqyprf3l=HS
zfSlZjWclZxezSS$64tysJJ{!mPlzV}?Rym&HWo%sj?T;K^wHj4<r$g^O%gh(LndT%
zG$@1!{$kd<IDuGX2W8?M4A8Oya6txRCK6*>fK^I1SxOZ*ZE6hNQRDT7_nz%P`tp1H
zeAlaTcQ}HBb@bsXPjHBC0^@wkjtvNH&aR&i$8TNU+dw}})i<8K^Wb~X%)jt&Pe1>c
z&*A1Tr<?!iKVJT#`PPlmlTrmq3>7#tV{vA%00!&t|MhFLVXhNC-rEaJfn3vc`Uf8~
zv9?%Z>_Q)7y>tE~-fm431OeB8E~b*vy&xBH<SdHH3WWSE8zBG`1+<De1geOX$OueR
zzaB^RU_peu87tBfO|u~)ntK0hX9v%2UCc6cFMTpQZ+Q=aG(F#|p{mwY^p>eX+ir*I
zqh}ZX)zcrIb=Y(?esLc^{8^ygfBLtveR5}7e)T;6v)_L9N&VGRg{%MuMFA;eFaZH$
zWaRO0{<l0Hur$-xP7V$zmKJ2Pv;QxC>?QKlt?m=~6lGYX@W#nLP>M4HWEWHgAVu`(
zXhg+~5DOZbAVe?h%0N`u5LttV0ssi0LJB1^MK*vD%!N{62nq&5bk7c-oh<f;^LIX!
z$@^iKNJW>I2T~ld&fWgy!||0p80^vW#~Hu!>;a6k`O^~M<nhJY<n%Z1j=N`Pvr^yw
ztKa|H{re|hZat5b1CoJS(N<G%E)Yz}fB9$c&2Q~3CRe*t(ig~T(@xv}<HsSfQ9<k`
zot5AY<H=^e2bts45fs4OK@tZq4&b7ML}rY_g(-U=a0kbv8lr<)CKP0LB~nHuCo&y0
zuof+59ta{45Yy&rxG}VP=VbBr{nI>^D4fgUNQ@;Vu-UAA_vQ1yd^WIuf-l|t=rUB*
zjq{%D8}Py9vpRnBhubgH1OBI1o1K6Dg%8edeoYz7R}zR5p&}#}&PfHa^#9}kP`(nT
zyK!Dn-H|wiNn6+d>W`Dx!cGY&Lm_qLX@9yS*$Rs|vKu;xMR_3u_KZ$$B!r$x`RiB^
z44fUSM4_fS5@!T=;xyv6+pu`;#N^4aiX#JAt|<Uvw6ot@S9m?*$+OG%abLPD>!(Kz
zwhQRv7<PlFh2J>)Y}p5Gos*4U+@7P%`lAnGgx~$!=9N!>G5OCvIRD&NPyenvd3PF<
zK@uvE(UO5Ww5h?ML7)E15A^oox_@I+1*}JiEL4Tr_&@)L!oU>KLqRH_r8Y14#!2mk
zS}}p(;E3o74wM}M2o%}e4ULT?P%Q|^CP|5m!9)pw5C){ePGiDk39|?0VSoq)L_LcI
zw+Cla$(8Z-{%czA!_;$b!Wg^SrcV1h(sJ|D?>+_U;t5@EFBY%w+{GxjF5k-=i=UKd
zmrwK84}aGFr|Tc<>tCtCQ9Vi^BUJJ*MJ$z}rsd-wq*vcL+kW2fzr30~%mLBB>x0Mt
z%@0!ocJfe_T?F9Jlp6S85{6;G7^$cOn1V7p0t*6$;*lN2NkN&v&Wu6^0|8PUBL_AR
z6K5}3Q+KV*?2(Nlx)5Mc0cU4X2O97G#%5<5b2EE*byEwMaob3%oa)RwzIb{S{pBi_
z3%Px<yxQ97+xj8U?8f7*&mEtyt2cgHe*0niDqSwV`Ff=at~OxefH0=2&J;j#C|Ec9
z&ldH|j~uQ~p)as+p|1DmfB8T4Qx?LAD0wnyKiQHn!ruI}O`dlq2%%6CcLH>Fuot}m
zh)(VV08Bu$zYI!LfpuL1XY5BU*}w?`Fn~c#07S*W+1)C2bZ73>Bp6|47O0;7+r!>=
zrq$w?uTM2+8vp<x07*naR8|W?DNUqA9yfVS7LIRvyO`v4`)b~v4}}OHN{IdQNduU+
zFjC?DIQ-t*-^;gOb5--9WK=E)y@#LxW@SCbhB3UnKYaL=T-To*@xG2UpPi&X{NHoh
zy4Q_VFltXG8_^;dWv`*U4h#+kiAP2!2X!!moE#VkU$75?V$KWzU>sG=s4!H92EwB=
zXd=!IQUr}g0ssxp1%=2%j@a=Jf7=37^*!HvoJ&!f6dC(=vy-c{%h&nx$&**k>ch&`
zLx=<Kq^E{o`pNU)hdZx~a_^<bw{Kn@UE%(hpyyobBnDdb)tE76Lv-rf5I~&5?tYp5
z__y!puAj@onkIh#{~WQjWei^GF*k*~1OibC)VN<|;<|#6h#W<Mj5CoTxHzC{m5oVJ
zL*`!rXH_8pb*Cw{l_Pm=h6KI(c9lkPXB1Q77h+xYn$ejH5|-S&QwC@s$CYXsmr@C@
z@shZe>B;lW``3pZ>~4?oNd``n*?B$AUc1Xaq_G>Ha{GApbobF~+piu<qX|6gFk-^Q
z?CR*Cv6blJ6$Qg(4?q9XgYxsw=f<k#&wudH#*mr-UO5T~Cq*Ilj_RXRE`@fc45m(u
z9*r_6At5**fD~XSLn_F>2ACiub)^EeOd}#Pstb71)a7+nlpIkklDn(DIIT*dfC?GQ
z^qpBRWjk!C*5zaZVH~^t1JB(jH|zCKd^wgOrZS9~_J;h@UVcAVL9ns!Hh^{IDQehG
zLeA03h?zp-%%Cv?%_bS0(Etdxo8iJg{YKdwM9Fcn=pSQvA+Hlahfp%BG+uCja%+Y|
zuIIHCHgPWq0POxk$s56o3l=m$1t#+%k^-Q(kFMD>vpXP|0gF2s##{v{f*z}i$rXdq
z7^>_+0>a4KU6H-f-8r^awio*+JEKPRXHSMr><gdeW=u!>dr`*CWL_1EA%Kud3@?4D
zmG;Kz;c@HM5Dd^E<wm&2n7Y}NfJm4`#UwO*q>I1WG~b7Dx2pYm3Ocff9ZFaeajC?0
zTf8ary_le$Kt?olL?=~K2?nNyh8EO}7z=u3BqExE8WgwaHRh-ifEkE`VTEWfm~4Rp
zu1X*!XK?W1L`hLpIW*~KPe;*VkMMfE8|qSr^<s1J;KRduJ6!AV*!J=(z7iR(ZXOa1
zhFXRVO!vQb$L6ojZr*4jHMKi;0W3($o3e$QZ6!pc;DlLhI+xVS^W*9m;&y~;Jr-nz
z4lHUy?`@|@Vbp$hPz5gDHlEO+SSBR`EiMX7q?s^Q2&TbtFj6bAA{7o0RWXxs{(FD7
z6PRfbYn+2H1Ox}vX=H)GQlr<57C<)D#np!LP?>E?wVTN<*KL3H$-PoP8A1_*`k8O6
zH7-}%H+C}tu=cE9_`><{#;=*}m*7IBYzOPJLh<W3?Cg*e6B8GdSZPF$tkC?8EkCHt
zrjHg)9Yrv1BlE1P+@4DjG-&3NHffwo52!*6L~d@ClLY}&U`*PBK?D$CDrRN0)>8&^
z%YxvD;A(0G4JlZTBPC1_aEOcmW`>w8XGIW<S20EVY#CCY1{i67y2^u>x`*LnHhrog
z3Fqg0G^AZ!?&Zau>g(TpH(cj6>DPZ_xbx0GBvWRT7|9GUMik3+U9kbC><C;3Cu~AJ
z6PDuFQ7*(==&8>ckd4Reo%z;mjEy=~^O+jVD=poaTmcvfO)@Eg3oBq@Bt&pCD@sbi
zxEn2mVjLNrSu1pQA<P8GIS4uzR<clp3=)Y!aN$}DRn6u)hSj~mL*Jn1X%pyTa}Bwh
z0@3ZqT}RbO_9@3D98Rh>+>Mv_u9}!gf8{qXhqwOGh1MhFB^4d|(O|6Fbh)mI5TgSF
z5rS8dq;B>qj+EPH%~mI~(piP<2^~Vk6pNTt5R+CrJ8dzjYet8>NySjrncP54)uS7k
zkrN_;1`|;bY&>KVC{!FAC>xJa8O7PHSp`PpRAW&>4?tjKUNvh(M%w0&eodD_gQ<_p
zTl0$&`5tf9fc1DnXdKdPwDI}0jEJ3iPVxMVOQP5R*`s@}d}S**Tb;)uT}NCGF5q`|
z+r}nVVV%*b5Qw8tX?WTUyp=1iAAT+nkO>{H0ZPF*U>3}gvmxwH!Z7SqRdzAUbw<;e
z5XhZE!301CkPy8Hdk$1oGNC$|5G7{<B^4AW4Uh_muor_=6{6U9<w4Ce%I|cBk#nx)
zgUIJnsEW(+cGXSrM;Ga8GW5G<d(VKw{!~V{?Ni;Z4-ZcY{ooCV3Uc&co<BT#w;PD6
zS<~%cH5n5$lSy~yfP=ad`QTuR#H^r()J`T<oRqwp{#e)Tsmz6zgM%>eL^&4$aYWgh
zC#yn0ATy+rZAe`bH9#1~ECgPH2^va4ltPiSY*oOiWHSVRu^Ez~y8!}P9Faqdq9jNb
zM#2L8>wgXtHESE5uRg>u)}TRex@FpV_vmDHzU@C+H_~f3z!h?JK#3h5&wuqsKC2;n
zKKke5#s05v+Xfn-*s79)kizOhxN>iVdtx+Va<&M<2+dARfre7XYZ+lKt5zWZ5Tcbn
zyK#)jw%4d?Jz-)I0^(p95XU@@mQ1)P8Cfh&YQfZ;&{zoysv>CRu7*??GH@)4&^ncQ
z1sR16b+BS+)>QD*1SudViq`!lzsyor<Bc$0H$%4=E}k#zL&NoD!R${bJ<&Ep?&{Mg
zKR$Y=zV#hVlizC}!`FeGTAtxjn`V-n#=KfJ4H6kTSLPT1kpLCUsN#q`QOfJ?u$Q<X
z8k(pe&J|GORaHtFXnr&ejsYVZIs$kh8;gxuyS}8H5fxk$fdZAJ00gW75}>kSm31UT
zWrx7#5)8%yEeMm9j82GwDhJ2{0B#_n80+PuYEBj#8QU9Uc(O^4OL^(S3S8a1{pLAL
z6u{x84U5BvcJa9D9u)q)x9-^6vL>DjuN)KkD)|PtI&B&vsKJQ@T@e|~fCOWRJef7M
zh%9fuM+QeFB^6c4iONt$L5}7JHzx&1>XJfiW6i)&ivwzA%V{iH)Ll6j@`4^6N!=Zo
z3MPlnkU^+oM?|-j3n=H(qasQ$RX{Apk|@oYQN-%nwQ2r>fRjun%!w7B{wOSd>uA6U
zEe<!Ig_os7?r@rjo?Nwy<;B1H0)6M5pT?K#O>ohQ$E#*u<#p=otH{m}K@<z*5datp
zB0&apjMYq<rXt7gVXTJ57>PZ4l3a2Q-nYIbswaRY)))l9$eF7OBO^m_&!EG0NTXwM
zR0U^6ATQJ#6&ESMQ391w4ZMH>OC<(bK_deeM+s~Qh!Xso05W;@y0&`!!OFw2Y!6CT
z9^8HM%C`L79d1Hg{qS=<nXHW*n#;XO^YI~B4^QaZzw%@oZl<oU(Zvt@p@FOQ)>pMw
z5hW8rQ43XpG8Qm15-=laVpHu1(q@%<oy%>Dy+G4?$mU#Cs>zREWz$ijW_2dT*t*m)
z1PYuAGh-e{Ta8L!jxUTA%1J6k#aI^_tr!Ji1Mmo_RIJ1pOI9-ritZQ;fB*+)=$Avk
zI=h-4eDG)W_32pZXW)-6pTO?Mcb&f0HNN`AEAKurqb(d?JZ&f}4qpe|{^eJnJwN#e
zS5CP%P~N}I^ki_vwjwn*K?}-SoYf(tmr_b`GMzV}2o?{EY6~|Hx4MXldk(=<QV#?s
z^O{|&U`PfjJGuu8Ld4_|-5i-P(b$j40;Lx<Z6+`<srRhTAS6^qLKHWk(E+#z&SWOe
z;|OJ7U~s_#)oj{U)6?VqKb`#Ao~4ph`Z%lLt1Hv$=-@qH-#FTP^QZk)pqy_ce+^sR
z|Lca^Uw`sp{VR_vQX8?(7;nz69r|kO94JZyAX-KPGk|1jfJ%mj5^EL#grfdH=RtXA
zw?*m1C`wQ*z<4ypag@NO01yM1V*rlKktGnL8YwVJ&Lw9CBqoJoL7hh&BD@F^gQ${u
zMJ`AJ?j<;o4Q>&Q9Fro8`=CIAIUWASYx<3Mw*?>@UVJ#wHR2Gv%bhZwynVOLl~W@E
zsp9jm|C_p%Z@vC(`Wwr7+B8I1n0?yg*`2)Y)o6;MftfMnT);tG&Ak`_r~}o$Qq+J2
zE8a|mF>F18jVTT?;oKZg(H$!RF@`WIP=P{V4C*lul@N^3-NA;w?|Vg);Ptk6fu;Z`
zW*Vk|Kx7W3gbrbJXHHNjp*j~5bz&>x0L<!8e`zuNaST6sX|Jc$?=C(HZ{GMp)lA!K
zgxS%>-Mz;^yl(YFt<i1wR(JO1w?2r!9X8B)TVk#S>VEx4;f>8~;5wF=S^!I2*cn($
zAu@zQ1-0}!JzY2>4sSk&YIj4l+%3*K7gAW!2KHG?o`nRH`-{C!oJR&UK)`4f6(-UE
zq>QAr6_+veVqjvxAkH>XOEn=P6duTf1j&IY2Mn1)Hd3%y3L$|J+ShOW%O{sb9(-f+
z*3UNE@hAB9Lj;<<b)KhNdJ?H;pHNcwYWj$Fe(`Xaf9LVDub)0@WR{rfz%luEKbbVk
z?X1weIXGyY-Pp6XY6S{IqDZO$>2i5?K6qC2P~B<x#`~~So^#8qlrbS@$7&vB(5X-X
zEF=oW#{g2$V=lopYN;|H^O$1-hYTYMI;v<jA}3VU>`-cRc9Sq*;ADdkU<HH;4KQRv
zC}b5Bi0+XeZSS>x`<B06@w4-Zwe`3EuSt1jtO@$PI^xuqVVmvnb?6?yb$w0ocMt#U
z)z3ewWG;hqWnR*s)2|Y(x^Dd_hMnV=PbUBZgO0TmpgJQO`q({xe0{xi0&>Ue&3ZO5
zTkrR=<P4gpLpu;oD#?IRf-A6QbjspTfXuac;6gwo>RCjix^pThbwYt`VkpIpP`MT=
z{GzOKq)OBy@F2u(hl%<K?#4kWRgSU#X!fNC=cZXd`kTEcP3?TOmbe@K>Yt#>jqUTd
z@4dE=k(X(H@W!9Dw307>`KPD<<VlE4yzH<I0k6I{n^!o#u^G?9DnGq<@ANANR0>7)
zF*=rjnv*Tg?w^%@(wxG*!Q7VxQry{OZb&_CS{zkHy*Z9!Ur{rNfuS1%sxh$=XaxqK
zF-HrAlE4K9DFjlCN|=pV6b;ZV=*S3&B~YR;Ktgi!C~PUJ5f76T$<(|?Je+p08qTNl
zM>vG-@|}AR+c(Ge(7$x|deWFB<&S2~!Kz@5_g}va7em_l_Qrp8fEcH`@!5WiAN=$X
z284%p5EftfSzRZ8GQRyXAu1OE1S(kc{Lu$p<NNT&Vz{bsWsa<nPkI_Ufe7`8v2m=+
z40DD$lsF(8AebYdnX>{?iW%KR2^~uSEfg%6NePPBF*=wPbu(ZD&O#u?$vto%i6=uO
z8wC`jd4>Rt$#1j}3jS)pe)a!5*6OYAg=aLce3y@w@9~ox=XrY3%TT}ax8l_}-g*2F
z|JvnuChvdk=KZ#+p)d7LRsG4modwtDlZTIgqr3l--L$o%tXBaWkBtL?gRS+G=W=8E
zxRJBPlKiBjBxyNwwzAYlwFDy@c$n-Q2_pb|V#_be%@slxP;xG@kbwdN5}9ET%ZlVu
z$l0i{n`TD@00UqJNNnU}Xf##@DN$qt1>}L^NLEkgUTX7NzI{dC{_q!f@6ZR$%ZG=5
z^d+AZ^@m^hE~C|5llEpM*)%2n-szp!-~Pe=J9h#%HKi!A{$CceHJo_`XSDO+;p4sP
zv(Kjy;d<7TO#K|iGf;UF_tQtiCD$GR>os-_6V%IDhFVazC4*7~D+g239A{jBp&+rL
zIXXF_8BxJ3PJjhKvwARR8I**{3=~ufi5drU%g7j#IDnfFwjDHXf+;A2ayLmC(pzz7
z0bU&B(}(ct`S)QfCvQY~<1coTS5K!ld*IsG;P%Ns{@UMU3|U7$oBZcz<+tv|IEg;U
ztZDzn*47(W99{H>86G`ZygA)K)3aqFB1DHKcUKQD&(0&})YV<n;&b*95TMSl-t$H&
z3=+49W~r`ZsDrtf8@nsl1E8xoJEjs12(kw@qmV*zHz06Ag0Ya<K&r4E%qWmsC=8~m
zq}Fs0)H0`7v;dc|9y>ia`PQq0U_ptx_|d<<C<EX82RnB3i`$c1<6e{0K6*P85Zs*o
z`@?_y|FiUFznW&(dEc7$eur;3W7Vm-s>kYq&2DycXbPsKNPtbrK@1rVoB%;C0^}w@
z{(uB<E{rG_xp0gKhyW)x0>eNeM}h<+iEL4JNQu<Iq1j}!=c%e|t}}k;eBbxJdsu7b
zqV!kng=hbs-*Y)1!L)LI{mSQGZ6HRY3e3&_^q@XktP`!#vdu0<FOPq6r)@xzm9<rZ
zKtalb@Bih;i%8@mFj)@=;9W2rr5T1xITsshE#4L&Yo`X-PY4~PvSV512nt?ntrU!@
z8i%T`IyMUa<VB%;rd~!XpfwlrO4SiDyE-|LlCKDXT?(z)tJP4nrujeE=zbVV@sRe{
z2Z#Q{q?^C`;C)$sv~G{uy+J3Zc=sk>;5l@wAGEK$BIAmN-EV$>@725F+_Tqa`rBtV
z=3-@q33%u4{N1l@Qa$^{VPh1FQdRWOG;rsie!i@wz51&!>!_z(Y9oakg|M-Np_{^3
z8H*TzoZCSVW>a-_%3fJy6akM2<ZQ&mmQkGyGFa7QqZOj)NTOczx_7OpY(zi=h)l*n
zqKFs(fE@#d<PLagoZmY&j-kAh#`MhRXY=kqy!GSo%%4BJf4SSaf6(V6Uw^QFoxSVu
z^n2gke|2IGzuX_c@c=vUM!ms*IrQ;3`D_=aB|W$K?AO2jUtEQ!>zK!c$OU*M+{ymG
z&Zo2r|MBq)U%CzhnL{yE@<GwJfD$q)#wO{2R~m3_1P~DckkLn{Tu9bM4GJfq0<LVO
zlv*>SQq_nU%>+T9(omrZ21q8LXw)bfl)`}<X^nXR7h({z_cQuMAg@Y${AqEy_-p_E
zr<<?-75CehmPcbnIxN8sZ<3@n*P$NX8rzpHJ^#ZmUs%p#-v)2x|2>VzOA6uoUjI}}
z7r!`$^EZFv^46<IW64Q}0yGU19{zX<dm)FP;>M-Tt+1IXcp=>Y9TDc<Rm;HBRJF$K
z&54P$O|8|jxEng7Ay6WMEP$k`CdJ}vOl|{GwWNf=h*kUv23`$47%NF&sKFdst_oP8
zWojaa8bI*YLHeCPdihIie764hJNe`F&TF6VcQ<}IZM%gIJ6^C#n^X^X2PTuMw$SyR
zFMkwX`+Rk08Tyj#hYv6AY@Ta$hS&SyxvzZCzd+O3&F3F&;yUq|5ZM?lf3pAd<nl9D
zUwOEtb(i#hIV)L!1<xR^VYapEa}Dc=F;HyKGFp|0PJ~>@UCqf6kwSq|sv|*x5xwTI
z!dN}bw<QkD9>Wr`dWl5^3q~{cC!-~xA%()NaU6+Pv5(=CY`1S;xp4f#wL5qJ@EBWd
zFF*T})24sGV=vq7&ZV_;^nUc=qXuI2P`VMCVf(4S_~zZ!)=m`F@Uyo!`)-<des;8%
zzWkGscmI<KcuspuzfgKZaCG4r+h-el8$WtJKCHi?ni}p0NU=i8csydea+MR>Sm10(
zh+FfJT*yeti;@yTsDQasFM?cvvc=ja<q-|gjHqk-R!WyhL#0ZjC{i6v9D;+PLI@hM
z(LRWkI7uaXhMzQx?zyv@n|-%i|IzM2f6DU0!N%@Ky0@5KOX2pT*@f<On!1(vqf&Pt
zP>#ply4m~ohv$Cj=jS7ln7{k;-2yR&=Bbk=gkL(`c=hQV+#XEd_-OB#0zlS?lgMY|
zs5k2U;oRTZ9lUs(h~w!T;4CgM&JX%-;}Dma$Dj%Om>s06ri9K0WZs}cXA~-;s$d|a
z(Hy0iR|tJOp*RsAVnsA1<8=gJ1$HF>HgvSoD2P#x961Ny{Qhy7PQE$a-8^^uwfaxK
zQ=a~8ob0^W#P7p$<MuP<^XcW;gO~K#lP|_eh5iI~b3?oPF}uG$(O)lfYTHoL+wbgA
z?_rYn;<5eW_3f)o*q?N(qcA*dOP@7@P9j&L3gPp7WiM>c)9Ib%0yakF4kKAW-RQ+0
z4w%s*S_)g+q!!y^Dhl9OBFfM$IWZa(qmWTu(5*x@uI*;Sy)Xx9YYjQ9!HY|&8Og~7
zKny(qhcogTC89V*)Gz<8@9e#P<t&`m`td(~FyxQE2IsckTFzGxFTdT6b6>ym)co+m
zr*Hgd#cq9C7O+P}9X7xGXHQ@IB(@T~{^;HFZRu1C!B!3D509kZnCr~L*2eV*$I#R&
z;>Fa0zk0LWV>p`}?;qa3pPw6xV!Cj|tMxE!ZGJZ8xDtS&bMDVcTmcY*7&v0($`Ch~
zrV&(tN(x|wG;2%*+Hc1g0->N(4+vJ3poFow2|5-rCZilGM0ZqWsJ(;H5c)T+T^W~0
z^c?+T-#(0sFZ1_S7vA`N40G#`F8=xT$^H#KOW*#{*kjF3k~h}s=JIQI_4PZ_k%0Z;
zul73X7#gJ7?&65M2Mw<7%;<88>vc(?8&MU?;25+HQ$0I6IjNuF#cggtE$_#z7_)WL
zR@#FC0j6H5+vy6SL{~IIK=SICAAnPapsr}@E?FUD#<@&7iGqk`0_0@wm{~?rqSAr0
zR7Z$v%FbNaRALp*aQKbgbHfLF^Do}|XuNQ1k>bIh|7YL+@f(j`$cKw%U-`y^XTI1w
z+1h#d;RX&99zv$V24nlibAP%QPBuNn`oVWLh=2l6B+lIM`M4>byGsR2Dw9+L5a(KJ
zs}TAXXx5!PA6B!o&B480A&vqAA;ihzWSpF7npG^bjW*YArdcYfD1sPM77%ktqK;0b
zBIIBIo6V#TAb{WsVpJ8@<Y;E&T+>_;AqzT>><DBflX-<$E2YQ3d~_b`=bs-;9z1t&
zCJ>kD|M&;rTJLWD81#$hj>fCwgOe@TJ>B0-l{ClFj)SYj@TJ@G^^-PePq%+>XV!Sb
z)CxNFCBs(PB51axjg)39YgkA_Eexf)Y@fxo)Ub|QF+xD*Ge$g|a&lGL-zYUg+munD
z1A+(^YKa7jF62r%j3P?a(3%RN)^_x;H*2K<!s5<^!Wk2)Tje^Au+dd5!I;5NMnVJk
zsc~fiCY!7ln{w|#d9;cz9p9^Y&-dEtk6TzhI9ls;P4E25SAKD6b#mqZzTP^kabd$o
zeNErG^n<<aqBIZ3zkRV~A!R{SnYd>4PCM(AgVq+;NKH*hh;^tQB4yW*PsR3@4TS-)
z1#fji)?%i}y5`gjIrA=}t2-75HH@)xF$NHeUJ%`l5-Sy4!shNq&qxkNhE6V4O*vzA
zsG;GSJ7#NKoejy2&AUZ|41#8{;N8vrvHA3DdVc->l)H<I<|@)#<D1J{vT*`e7vKHO
zZ=Za6?@zw5^)`|<SwJ@zNZst^x8dgTr0@FSx3;EJaVB*S*0N(y*$tx9ux_<IvDjLP
z#m%CjqognrhIS;<qG15%=&>H>Y_{RGSu+~e#FIuUwk;HtL{DV2Y%UeuDu{<#gHdq}
z{ob^10Nnw=5rC)wIU-rmYS@q(6)R}MhK9kAl+eghMlfu`$9xl>`%i!4(&z0u#=Ur{
z>lWcn_{0>Z&)vMyeD6<udHsbq|4rA14*Q~Z!WQ+-iC(+X_2~7txMy!DGgdD_r5Lrr
zkQg}=G|NJu^@%rmfyGHw+ExKWW1*Q5x20Ar(57r<V^Y#G<QN=i-ilj{p=TAW2Iyb_
zMa@}I^I$fb8d-l~K4~~0n4`KGq71=(Knt##Ae)+nNXnG~0E<~=R!~k|C5qemmHE#3
z#rpmG7UxxGIpUoQUGx4JDvTebYm@%q$A5b={;#jDPU}ZFaVjBOzWmD1?CD46>ZJYs
zhdbUuWggoa3@|s2m1n3G#~H#pd#Di1sy?yx2jRqCD>(&c<+eK`Zr05T3a+6mM>Jg-
zR<^wjKxi<NQIWQaT2KTmmAEjlqoX?H+3r>&p@IV#l9M|N`Q+HaVQ|rmeGsoTFgZ%a
z0v_VAHBbZV^r?+pl~<4dXxyD23s%`U4BZDio9WYg1hgi*vl_m9Q{VmB8vlcOyLOD5
ztEgP<l_70jSwho)@P`u-VWJv<s~|yih+}TKEmLLZ#%xku5@Sch01Awnc`~kHbmu}k
zg5lh`Y!22g&XsAG5HU+L@2cbm*dP-n9Fa0cwTe|qM!<ncLut=%L}GJwKr{qE0`XC)
zq^5FVAkQYw4AlrER;Ko>5PA+W-}Qsroz&s;yZNQlhmB44cXvMDgByF@|5ajG_iDK8
zw!YWB{DakOJ5RTM-mYcoL<x7F`)u>l!;93X-%{#Zn;3hcR@qCN8Y>itMill`jk}RS
zON^e?SdrZ-FtUoqVuH{v>@3gf7*iVNd|DUU4X!&=l@gs5ncT=?g`f#-vKE&S2^fmz
zurqH0_>;|UKoBPkG|;LRFfml+VxW#rB*^9kta_B}7<7H_!5M^Me|7HWGd8c`ba~<C
zhv9s<_LskZwX~%%BxvqFdg)+ke0FIcha0W~X2y~JCKUhHJZJfD>XuUzjY=U>z~U7F
zfdgbSEu#QD$!dBsXV%<A1~W@S0t}(4<2vc2$ZqvI!eQJBqwgv^Ok-mjTQDo`URB9z
zj?N>IXRc~k0KMC68^kB&VE{s45hN*O#MKpTBR7tPsu?-DD?q`vXl%rsM=y@6<!6r$
zw$AL_$?U>&kCy53>APoqu0P9Z!a|PtvyF9E%iwAEYDO#Oxz4YAGI?n+#o7ONo7ZzP
zE7-W@5v{0t^{T)vV=0z%B2*b6t4mX{Ruy$-DPn}x0L<B<QmaZMs_dptH-*Wkonafn
z$*pJ=(L~r~S0o?+aa2x&>`a8+9o!IEpy8I8$hoev!iYmX1#@O6QdIR?O`{u?JXmcq
zJNsh1^XmC;jQP3Eof}Vo_P%U?xI4V$Z(-uygpL^-Jv@lME%Cwm&6_D@Um`-I1^@sc
z07*naRIJ)J2P!v}y76B$^KO7ZT%bo}8m)L4Goh4<2m;QS>e|UMX-QI-YB)hrrYu;Y
zI1E-`?Epwcd_yl(wWTPvTVfPAARq@sGZxDZz);Ox9a2%o?O7l|2PZ;sZzX^b&^iMk
zfma~~&=jmxRMg4_EdZ(sw%wVksq4WnU;WAl_nu$eI{)IkkGJMu+?t1tC3zdAj5r#7
z@zV5e)8>(uKiK-^>n#y>^Q)iHl~vbF|IxyvHe)Ts6#%V>NGj&0qpDYMH$#KE&T0&V
zsx>i#R!b%h)Ib@=xRPW}s`-%Pc)fxFEAGq!q`EWBMoLb=YEIQrt7T&XK+e9s*|4Ca
zqbiU@VgUyuv>2HggCw$n(HaVfpq8bi5iR;a`RT@`rgrb!v+w>dC$H?>+I#tr9&^7F
zqn__~2I#nnO|qP1;eBS>kf&Go>fi4>X#JH~!dLqyzV{h5y$Gr1AO(;uum`IG>Y$)l
zpsHGi0#+Pyw3Lf_SIpfoBS26ahtNpVYBpG<&V5=Xlh)mi95+u-nOqVWRVS1R2I#%0
zJy{2vbdk-fDG(A9Mn+b5>S^sxnaTj`(gg72*{T9YYK*NCc0=Hlee+wdw(tJ)JMpP6
zt?{|v9ae8TjK8todhcbb+U4=AphN91UDU}__4#w#zjyoGH%@%%UwY8&O(Ko|WAI^?
zP8rAzkS$wPsMIdkiHbRcrGNm{+?;ZOv<eK!EwQs1CMYJU<^dS3L~7Tgn_dWwCXCZq
zP$0M%LIlJp;EIOol#HxWb?Ub!63EfTorpOAf?G9m3&a+v21iT@&4Hc}=3^ysT3OX$
zw%xq&%BO!hJOY3D+=EYE_-+%YHSF%?+wFs!JPX4B0K<BI{bC>1m-V@gD>s*)ethg}
ze`EVBUMPV7@T4O1gk6-Ks3?$H2qiUE*F`fFz*0@o9IFygZB*AoY;&<nN}>TviAfBb
zO|MkXyiseH2{lcpjbiYs8Py74E#mAUBD-p?3MHFtO^Dnb*eC?yC*U_CQDxVXw8RlZ
zQ4b)DRnRL0a58nNwEp0gr}7UEch7M1;@1}6z4-nMoqqM>m%o_Y_ipcOdv!`Gz;ex&
z@#1NJ_VUn$Exx_~7IX8`$9;1>(04u;6Oi1E2`r2ji8M9KjH2#Ey_j+^$c5Y$tb)4h
zFapa2O&wjl=#U;c6cIaWYB|f{lmx<Q<#SV%+6PmD5m=DXS;-7s*hx9mz_Ymss5?6$
ziJ-eXxiAwo6A;BvW6e@5S1=>hrbS?IfoQSf$vda}QyQCDzIS`<9`S?mnZs>w`C!&G
zf83qcljI$1qRD3`_S!eor~mx%*q!|#I$Yly^J}f`|FghcJ;u<u;Xqheabt9ZF<Bk8
zNX{u)M$c9~5;s`@Fah9BLBbA5P{*^(E>yGyb5XA0Tt_u;&1VHkp8N)LQgtatgydc|
z3esrUZg;|N?g&IA#0CnEghX7Ddn@FX%s?s_VSq*z*Z~PGK!|p<oG#97=F_-)>+Xf=
z&hgFq@akI`7vp+%^lIj&D9`4pTsuBFe|cCQo&L$cT21|ozw+WI8<VHXJO6zxOduVq
z4FKu@7KhV924-syP^~o83Ml1;vqaWL&D=9ktOhdZm{s_o0**@UYAZrINlm*O0czh-
zG*YeZrKzkH9UV;Fn1Z6@3V9Nn)`2}ZOVd1YYq|pxk*NljI(Sv6pseWNkb%I7-HcOm
zp554aOphL%em0v!zDyrJeSg0>e|UVJ%W$dTCNbRDI65A>lkpjQ_U8QBi*=pPj{U{2
z&K3vTcmJ$hn`kB_Zp0g?fVEocfG`n2tm-_Vi-GBCjSLz$TR=z#06-;~8bI{bItQP5
zA5uKbxS!hHHr7&_=|*72C+J8~5G+=aNPwy_*22h5Gh;$9F(xK6RdZ592N2=v?m!^L
zu~udPR}A3fM9x^zx=WYG2k4RWe0}cVD}3wP!SvH#|A}Xv-TO*OKf3(5z5H<4JXBBr
zmA&;l7cYFa?<3ESr~i39yKwYi>$RW1Kg50!jrSg7VGg-LvKm-v6w_(7%AB3EgRkeY
zL9^Zl7v(%!Ep4?71<k)$iY`Mz3f7KkO|rm?i!t?$I|K%0D1<OFn5a1_BUN(GZnKuj
z0MM0$5zqm#1_yK%GRjl{8N0!!Z8bmuHwz}YJ*zlfU)%OihRA*6i*ED_$E$PuN4&N7
z9*&j9!yC=ffAlYXzcv@ohSROz84~%6U-@@KY>vyD%gK4YF6~eMbRZw*t2ma`W<)18
zHmcnapfRPjKqUZIL?DTnq4sOCu^Dnbk%ysxw}Weaa2g1v3ES1S#_Ef)Jzi`aGSC(v
zPR-mMNGY0gCD-6&R7{yNZcaoBpt%HbKtMo4BS2&EkQ-&^3UyLQs0Tn3MQ5kAE7jTk
zFIo}ODi`iouF`a+9-RMspYZih4))HU?mqMJUYpB}gQ;?RwEYO4``UNDa14XWuS`%n
zTp#@5exfEWrN-twQVm%UN^B|;s5cra3pik)H8kzS!jErnU7Jlb9na=ramI$0)gj*=
ztVco)*+q;N>#6Q;W^Oi#D!M=g!qLT%0DX+=iIl<2Fob~4?kFOFjujDz(9s;I4#5*}
zL2NO!lNQJo71@C*N&s1m4rmy;e)joxX;#;uCgaK7FRuCmrttW>ok+^-8~cr)9o#v0
zZR2ukLX7I0WQ{bzfBPN=LGBHF>X95t(WPo74j!tNiU4W=IkwWA*?`gwcRwD+vK)1P
zVuF>4?C7kC3JJkF=DKC~Q(R1-T-cd0Sis^AP+d%&Dp-ID=7<2OaY7LR5d;ws(TRWw
zfRNCsaxFrNwWZk4BLxm(YzA6ISh*O43If52=6z%KXLsn)g}?WU<8vOk$N2t@c6@qS
zAAhj*?K#WYt?9KtJdV5V>Gb+9-8)Iy%R7rmGAB)esCq+FRYrusMmm=25F^$q0v-TE
z%-LS}mDirVaqfic!j4hJQ;@{NEuyKawcMvtY|2ZpZJf*=otpwhM>GO60Y+D^PwK&z
z2d4rWqo5fmQeXrocLD@5R|WB!+M0oQP0|UoxoOA&9K}*4U@(;&b$xCeFTN`%b^pG6
z^Wmpg>fNn8eFvu-S5BV#GwDzEzVhDw{jTmEd@x+gH`4~$qTIT{fAD?7t&Mu$d}Bc&
zq*;l;V69d(56%@v>&B=QM^@r#sAG&w0j3+fdy|!#RID|o^)YA-hT0|wl20sU@i9+^
ziyeBN`x9ZdQw0PJme9dbDpf8`Xo1{8n2;)fJz0x3LnNeXE!MU-^w723H-gSp)EKOS
zJ78hO2C67Fw--whAM$JMJ74U}>Lwpdw$sN?_0vslINRR`Pv;hYb!l_+Cf|7NTf@Pl
z_RZ;^GV=QLC=eV`SdNb3R)BH_S1p><2|SXO(qxh@bd%`>sN30i{*9OBmQ{-jVM*@-
z#8DfG!M7J}NRo%@c0%0@stJLz)WXC7wN_Q3V37@TRfizth{y!+*J&^*Kty<Q2&8qd
zm7%SMJ!(X(2tWl5TmnJsjzOiK#b?he=KG0XeB<4_*bU?Kh4CCee*0pz7y7l_zxVU4
zcJta>dsn`)C(piq^$(`9@gn}WUBBHm94Yi|4;Yb+16e6e^*{is)zFB7NC}8JHr!D^
zId^lH(SW17Yf07X7^PSo`qU3$qBCdRFCF;oQWOVd)xijj6A>elsVbY31QQAkAgDP#
z*~n4^H*&~;L?E$(R;~qdB?6$KcSHwdWD^Ij#nGL$nV#p`H|;E5e*G^FBd2M6{}MKG
z?)L15Z$5vznvJ*O(WT8_=>8i4F3#h-;nCHnZv5}1ooIs9q^@JZ(Sm3L9?iw5Edgo(
zBEjm08Qqh(HHCO~esbzgie{RfF{ZaFK@+WOmK4_3Hq!wgW&+)ww1DV}Z8f9{p#ZXv
zEb0hY$)M<j1RO}1@ktowXaMBo<WAW#<rOffSET?!kR4bZ4XS%W7QnF@UG@(Vc(r=^
z>Q6sec)^3_c)We@v%a}!oO|Bc9FN+?GEaZ)<>vaF^Y?o^e*OplW>L>h>WV0m+7h=n
zk<<vy2geZ@Lei2`0)vDOhGeGB)J($$1;9|zur>0trD)D$mCDuGM$5(b$4QC1jHn#J
zlUdcasG5~*YM_-XxCIOJM8Zh~Xifmm2!H?q0Fa0=Q~)!p92nirjG&@A6C($w(W-iT
zqkJ+S8$SK&?!W%orc8pYBUTHSKZal2Jn`+b-E;SDKRPVepS^S(!(`*VizoiV;(r?(
zss^CFQqW#%_t!UQ+a-vm1=&Fhm>Sp0;45ltY$w!85NTN1)Qfulc`h*oY+8y-7{bON
z1;>nOQYP|57R$&Wl}#H7Y@VI55CB%JEQm-*On~4dPf8I;mCcdLKpD!ade_rBR|X>m
zj0~<|1XVH!Ti$B#%<54eo(&I2SWP;~JdwwKj}FUac_2^Yiw_o$E}U<ko_)UF4AZms
zAH&73{SUcos3r2=MM3g$ou0Wl8rQ%Mu3QxiEV-*$)*_hmgp5LjE}xpIl|s2fe%QcD
z2{oLan|c^vuppD6J(9^33VO(15y{lJEdUsiu?QAQV__E-HWWlOM<j4Ca3lmJ2NJR5
z0nJBmxaQFj2&#MIjAVuk4iNU9`Ngn-iEb?XfwaTbw4-!)f&uTnbojl)OUH}9jh~$H
z!(Y8Q=DDPe`Y#x#zy9ByAY9XoJgsSWUb3~mrc}qi0%b=d<0r$twkoQ%BrU<A7#1v?
z`t7ELpjjd0pXEX4l5I2dYT=T#D-4T)V52~B?2NdBf`sS}hTv!_8CxYn6<{|K4DJ9x
zWT3=^28hl;I;=b=Y>0TJE)0%X3Edcg1QC>~#%F(SZp*Oq)78<zB=0@KX1zPxY!}Vy
z<hk)L$4hYS-yP@XZ|%JK^%vV`F311#V%~k>)<d+NEy3F4;xyw3t7_e-xQb}P-~=Oj
zg(q}l8FV1a3ZZS9;qJlo`s$M}GA(sfPRE#`qQ#Z>%#IV9N`|unO*fqefP!RM8C=vM
zVsS1424INfnFJgnu``e%7!o2PqdSAU)CwM|CxptvOu*KFk1e7B8bd}#e(vr98nfn$
z<+aIYR9Bp`oP6!m?S9n5SDt_Pz4Pbpz3}$(srCJ<uX~=<e>*Pu+&BJ9$N}b-y%N<`
zt-!0&cvWOA(KVES%r&TTskmkVEX-m{8d)PX{gAcig-%&NJHt-Y0@W6lGp@5mXx7&7
zQ0F7W0uIInR6x0sDiVl?f)yN{AgDMIA}Jao0vG|gIk_^JBz6S=*Nn4Ts{|$hi4KTb
z+yF_tz4m>ZPd4tHmj3o$b37sB2cd0oE_xc?|GGWi9q_Z?{OlX6<8i$=_{X1b?qB)V
zKOAlD$9Iy)T&A|#D#Vkc<$!J}HeO(CD*^^|XqSiUc?<*<t*Umq2A;!se1FzHvU4FU
zJ^`rh+Ptk>54#o357t}Lh3IJ&32hF;CZf89#Ncdd<ba@#M5bsaQJcU>28d)vPyE1u
z4vfZ<0SPD$75gAT(4RDf7|dAEE0|yT($5O@*EVxsUVQh?EQZh>YB$}wd6stL>FGPK
zY|h?N|M!2lJ2-dQw7K(pPoGV`y!c{H?Lk~`9CRzuw1r$~wdli9Y#WDQ1*-K36`Atz
zFx@S)X~Q1I5$d|@X1X384W=_bAM)qRCU@eo2xpre3wOt*?M6B+eLJc{2X(SW1OWh2
zmqI9NhDfET2{92NJt-gf6QisfM0Er-K{be2Ts%!l6pRSG2^B!x)rfrK_4kfUF3<0#
z#nT^+iFam?VLeZ4-?(o+**Hp{eSKdKZ=F9V4Yrrc^6kwB-Rn30>se_=@#Beb6qJgy
z(*jarE4gKLbj0A`0R@z=^Dvz0IE~cw#WtmK8^XNQsX_GmPN=Pzs;tkgz3m4#AGc8P
z*e1DcMydn}fB~Qe1p*;8^Tynq44pxknZO(!5s{xL<c$m*Kq`=UVaht0a6}+ti#CeI
zg4NN|&Nn{&#AbW#Cm7lxwCV?6Y*BOboni06{$5qzvGt$ZtM6aB@l-io{`}d+r|+IW
z`N|vrcx)QUWH8{!;**lQGf7}dDOh!OWUc72bYwgU!*sIcim|2ChuTx@+IaXNbwO$T
z!AK<WG$QuhLF|rppcmQtVmj0^JB;KiZsfHyAVy$U2S#>xZA+-8B1{O5sEF=>jxG)W
z3g7_3B$)^+x~D#=c_k^#C=e?X>&BG_Ki39#(tvS3oa?6#`Wge3^?c)&fB4LXoh<eq
z;8S~7wlq?GxVrZjTmH;TAACL?deq`c`ZI1$dmXE_(1g*3J_8tVV_K++qd^M=$Dmw_
zPvUHQzTO^ZAsuRf>dTWaTmz~no2(Kv(Q0@MeWDdnBOxe|DKS(712%-1EoKKq2#MCj
z<Uxc{(FuSY(4W9D22UCxRYvim!XuFe_a_cY8v9DcILOY!pVq*e+rQ9!7S7)M<Hzc7
z{kENKvQ}A~=AEl(?hmF{k{hNM(jRjB+V;cW##Nt1yre$G#0@lpJT7CvyF0lvK(|2|
zCJBe3C6fp#*RGRs;Ie-zr$w52%(3-E->)il?*I_W<+~btXWE3pq^XzzV;s2<R}Y9%
zfxMbYu56^04daM85;GecR{|$+AanpQ2S)=5rZv_E2gKY~fD#mqt#x7n+qTa?JZmIA
zy)c@**UmePP35?H@$FYA9e@3P?tZcNrT4yh&{p3*ym+VmrHAY5uU`GhKP$pjNK_ha
zv)8>-2vM+|pGC`<B{#}I5i3>i3|&bZ9E%6>GGdCTW?|I?Na@yDbb}1R`gJAjwXElP
znkr#ml=^^Vq>cbaNQgBQK}HJ|gg6@(auEhHb}|xw0!OHUxv@G>UVG36Gmrq(q6*}J
z2*`=c>fWQ{aoWJkTenJ|eR*xzUtrmj)%rqx>0YI0d1rp-XD`ZgPhGgv+{lR^zWvlQ
zzW8aG)dpH?CT)|?RKk8|qymGjPb4i<O6#Hpuwp>2Lr%yZJ%q+bg;gGN&cpJMTFrTN
zZ*&V$;~Zkp7I`$}DQM%-hl-VrxdW+C9b5>5BV=%_;B|1QSRKR>5S&m536b1QwN^JZ
zbDD)*Kq_0cc}iBy0T98I&6u-sn^*n$pB4&tjCWx<OT*>k=KXfOx?cbMjcfDsTSvF%
z&7FRHcr=FDyuESq^1pf~*bI7~d*UDgyBRSJ8l<N6R*YpVqFJl1G}l5E1xzJ_aingY
zmZv8N4<A0(^E1;j{K+z^t#n!QbO@>7WH}!z7h8ptOKwMVH>bpepaxJ7$fKd51Ik$3
zD!4EqAQ%DyA-if(a|9#_ObZQ^$8u)Up|H8+ifT@XV1PhuMf!HY_V6LZ{HQYw;SQvO
z*|vnc($)!XKaRZdF+6;<?|Z!1`KABoN2|IlWwd2Ur@7LWuDACQ%{VEGF0n4xfR>%q
zj0!bH4XcxwA<sIgjIqCZk!jR=%afr7<tDd8wqc2!w+C!3%v!gukH*y@C^I6vMTlfT
zz|KS#m<o(YOpK27*Fq9AHzyVjL`YHyKn$q#S6A4JFDH(qVnFU@M4l0<6*xJ#@VIaP
zGU(ZKIZpWgD|B}4@V#eY|J<hh@M-98y)wBSR(_*>?{}FgAT=)3fd@SDZrzjA*}4pA
z-3)a?1*4G$Vz59(3L={~j6gHnps0YMHVbn9@%9jkus1{mxrjRi%SqAaw|RwQK(9>>
zfVHBa0$^3K3WgPc2r~qQ3V=us#A>8&js)ZYi0D9VP%5gR{kjwKV8xRHGS(^>kT8Jf
zgPD|#afkQ!xpL@>*K6wG!*3k!<<qeB6AHF@_+Xf9o||tpe=w}G6whl|nrc5}wq=I0
z(zUM7x~1@Pc{HT402EaxaW3F)eXtPQX+PcG?E9|cI0nI=`#C!W@n9V#6G!Qy*KrbB
zt`rBBwpt+~6Ci~GsOD5qFc_hWCBV^;01S{2)D6uHorM?)+`y=s)d)^~N13omVjuus
z9i<e53VrMO(KO%+_U5V8N|QV+y5r9t9fXgc|HYhl?oRQejqXe54_5a_yZldXQ{U1I
zL8vd)lsP(_xvR}g6q4JJR>xV3Rvke#k%1tE0!<&;sFT={C@Ev`pVc$y3EdESHI~sf
zK%1pg)2syg6eW1T2A+Uzs8kG*l^|nrOO-Ik8iWafOwHA@V<aX51#{O7R6S$~1ZY62
z7@$W-a7VL1Vn{WMou1{su(u-sc;~3Lk;Cbthn4O^dwsnQCG`3zzIW?)znBupdNm?9
z6%n;+ff`pXs0e2H*iHJQgT=5KlO|_O%7Bj8lk}55G=a6XPNIc+zhX=RMM9VAda{8A
zMVinex-9xyV^K>`2q~Kh06_rP%K1rcmWw!A+eC6lM{!3$Mkj#k?oi!BCbsOuT1;D)
zPzrlRCoryt&Y8>#7`es8(d;8^6HjeOh-ETg%+KEW$<{#=`~FHkn2a0E{u%ZklT<SC
zp1~s|sG*i1&PBt}vqb4T^M+6N55~I6CDmk_ks>j3>}CzCr%aL9bMUuAH!yZ0+L)xW
z7Hlz2EY~J2>HNqds@iBZjF3xm9jh4zLNx%Vqy}7N3MA%Q3m`B$gKGhGHFBy%?jfoV
z!hX6OLQq2s#;Am7AeBi1LtY;)!`XOY86xz6(7^H3R(`Vh=_1Zfy9>`8PxkoyGl#eT
z=X+mUG|-uQLzQ9!QM=L+F~yayLl|*@cC9l!J3KfU*9j9LvSxEEu|gxlDq?M~Ms{{e
z;Y?Z}gmL2xyi{<rC3+BPvUh0>L|hsoCy$jNwBklsfgRi-P;H!11fA8;{E1)E9Xz3F
zt+k{Tkr@lq*sY2NKn})k2t=glCRU1CwTpKSbrG8FK{qg+HhDOj*u`zwSS@ZXHy||A
zgB`!Wxq0<=_Y<l(HeAPQysTXxN+Y5*r@(2F07~SPB}+Ox%xO7bE#?Y2sOEIG&IFN>
zG^G_R4o0140%4Dpp<$QQ&Ah;nAPy~73JR69<Z4zkk*ljoHq+!#)s@P+dK)o-0~jMR
zBN(VDnOn{4lY>u%sn!D6ue1U0LJk7p)j*ISWQa&Ze|F&k545#<RHI<GIl9Gi?{&R$
zyA9p?!)MAabe{+M8^1T)dHxG)5x7d3a_CR#Y?3*giEF6FF>fx~#=V*6%_Pav1ze@j
zKyLG7DXam<){Ki6$chfeq{3`X#X=GWYB^S(F9SKZZQJIem6Vo&5FmIWQh+fiO6H8L
zPqME`L`47tL^nVK!%D7}&(h<^w?2~O3|+~pS_m~qc%oBRG6kcO6iR*O;(J#<O7OxR
zoR?N6$Dma1O<|)woqfLb41e<YjgO|QZF^3%zgI`;2MAtOP=#W!)L3*G+6t5i#Zgb%
z-qgq0u^fAy&Sa}iFwGGhJW8?^q2#P~R(Vj)h(3hVB2ug?N)A$b<x~NtRLO@V4FXkS
zQjV*r03gcloY~B9hY4IA$N|CK6%0$Q<H>`M-(P6qi)=J%ko8>Ww(dKTz(Nj2(NapQ
z#l!IO-&!qy^)Iw^c@h0|e!PsTpx=b8<FIwS-W%h+t=s3eb`H98&(Wn1)oOI6x(i{M
z;L5&Jj%EN1m89Y7l5*E^!=X{eelNtfh#0z=m;)N*%)?4cIas4JAcfojIgBy&Hk+Ob
zd*W`bP`|W0jKjFF#HC=x%$CU5NCArE>a|prXEy??8X_9Ff;p5joILpWFMjaBkr@b>
z6h&Rx@_dTC>?cTwsDO|wfaM{+_M7j1w|q5>>%9HA%nL&tK|KtmzV_!US=@X5XnF7X
zxBlia?f>!1&s_c+KRz9690A*vJJ4w`P>YMES#_w0#G^<xX!3k|!7<LXg(f-1Jk#Bw
ztpX8RLR+$&EbdXpN!CG%5Ev9|&QL65!jQnVZH&i7gmpmlapRQK7#KtKCkzNURGfA!
zZVF^*fKbHBFf8uA|M9pkf^Fc-|9?ww@~l~QUH9$ooOAE}n)khz`Ce9LR#sLOs?Zb)
zAOI31NQ#{r%Rx$1XNBz0z#e#D`;YO+13SV)S>e#c4#{>{6l79_2tptM5CjTUM0J@}
zmF?@l?%vbvy*)_&5Br>m6>F{Enjxd2g%Gjrmka4h5EvX#Y92TB>;L$N-@dxNda&3m
zW*`u+>%2L3{aa_JchA;o++A<w?c?*0-Z*FvKK<D*p1k?#8k<W<X<BPKx-3RuiVB2I
zR6;AwGN*ZWlV)j-#Ir^WjP8!u%A}>~6o!M<-7kFgr94eq`s~ps+AnO8=lzz{qAhcB
z{>DB%8__dO%06y!GPUZ~OkE0Mal`h~ojG}Sa0_Pcpk=z=efr*eYp}!8-xPjDaYf`{
zgR^PBVxfeH;E}B5k-qS|-}#pz)3Xn=$;r6Eqjh?DTTs$n)o|zH`fOT#DZcvd*}&=P
zo9+C`*H^DRs1k%}u?FexJeC+Vxw-@=HqEPCVqQdCb={0TlQ%N}3)qn&RMe&_BG`Le
z-20{bdfv>6dohQ&@8+xLQVT#BS<5eFx~LKq*=SO->{Btwrr@fvDr&{sw<DRl12-UO
z&`O@3fAYa~E`ZC!KJlBfEL91-#l8s-LXiYPp<*@HTVMO1|F7fn{GWe&ggje~QuUh4
zy=e@*f4P4!KY4O>aO39P{@<SL%Cz|5NiMU~qqlxgG~DnTXH%7atidXzpdEBxy3VI|
zFs4<4Va@@~h=LLbD7i~5W~$Dihb}NWbhp0tMY(n)PM$jL@8aSN!qi*LlHhM#wowM1
zX;dI{Fm0AcueFU0w1HxUmk$H0IsmyNwK}!&>Z5lbPhB_fmvhYgTU=79955_#ox3h|
zT}-A<NaT9&^?&uVA{W2-#-C=ibi+PDa1JkTL&}$@_W~04^#S809NGW?AOJ~3K~(#{
zxb^bq?jPJg`~%s)dAd8i^@Eew-uUd#zk2Jl9Gdgl4*KKM)>X=}YVcibqK<J&Erh=F
z&aFvvWhY|-kTP*YvV;SNh=}ZfzWVZ)<>|QNb=VU0`<EUe0JBb9;9J+alLV45ku(~i
zkyca4ShY-=aj$&)2D=&oA{#l>Y4`Zuciz{o@8gh!2malI7(+)2Y<nvn`<MibO+ZtB
zu>J1YB)5lO`?pWJV6Z!c#WQ7psX{$EJ-u_-eefe#(an#3obJ5;TH0cF^OY}uxc~qD
z;lrnH{c>%Q4<92Isl+%DItEfI%#|@LX0S_$2mr&21wm_~4nP>9Bqnh}1aw-x^_BE=
z(pH^Tdh|3|6R3N!;TN{H6;W7ZB6L7-%?|m>OU_xdO?%tDGEb;rj9>_$)Bf4_zkeCK
zB?~V>QTcDj12&9^3(pQ+8s<q73ZwB-o^5t{EwB8eKlsU{YLfBwdw;2cKW9^W<L7$)
zMn7v?|Md7@zm;s>Zm1^v==L9a{QlqhsgSJBbU!~nHM9T$8C)SYX$#3%d^sfZo*KF^
zfI0$G?bR`Qh|<MK;s7p=WaNi$zj1w;$@=anoRX1CLF#5-y7n>$lqQUfGhvG$1;wFE
zSZ1;0bGrFbhwjYeXkN<p>5smrc%$p6@5O<PW`Ry_-WnFO)CFu702zuE*T%amtKPyl
z9=-eyB#h#1cd<m=HCftXIbU4<L7xBez5GU){p9u_$A9*PG#tJ9@jsb9eEA3efSk5e
z0WX5|h>C<=FJf4h%p?-5H!rmU1`w`-;$)cJxdu++YG6=Hvu1`3jK1{W|Fb1xXtkg<
z*J(<1>~}s!N+=QmJwXcdVTdyVu7o9YG>~OyP%tEL@V38r^s~Q$*~xOry)grZ$bT1-
zB@{qL-qu-K%!bI!;DV-F_T}J9m-Z)_4V`X%(e0B4$8SGbzy9ON;p=DrX}36Dd?(+2
z_bXP?`s+`+@@f9sk00Ot+Pl{6nuQS8ge<0lvn@{19r<9dhBsPR1T}IGQ|+Avh}|nQ
zQnVC>Q52lP5rKdlhSy*DxKy~lxs{;(vT5d5Zj>pqM=+?#5Rr%>5CaTjUC9k@cp+c7
zmC!K|0ND8KgFpTLdbXS)ChCLr+A9Ce93(Qcb=B4y7Dq!!%%<$DHjU%GlOL~s_-O<#
ztxfmeO0V4f;x9h>8Jx{8?H6zUgt|Ab-a9*Ni_ebxaI&B8kN@bi`Ip`~t-t+)z=i8!
zP!w-vMvjhb1E=*M+hICp5f4tiH|r=636=_i#4!W_Bwzz_Cj&7+FqAu=|M0q&n^#cx
zZcVS|Z;IN?nVcI5yC5MWtE5l^)*%hkwXQyQO9%rXIKcG>-+O*Je`$!#8%LV$p#=U-
zPhk*2hBlQVheyjUp&$YQs!r{f@Bb^Gs%6Jsef;E++*v;O>|os-zwz21U(der-g_{P
z!;LM5d9bpy;U`c2>E_|hU-_|(BT$?91{{l9GJuAn%X)ZlwZ)@y7!VVZ+aiFonMV)+
z<U|bONdCgV2Q+pdCZMCYKDffe>(b@EU+cEL-Qm77F@u(nxe+NfW&#_Q(<}{~bG`rO
zVFE-%x0WA0c<21aOUnd^v`W#qQb#2ixb(}_(5oPLDFLG~uoD_sfBfL)lSd(Sf<+cL
zcOU=sXMgn3<paO{#;?Rbxtfk&d(iUzbWL}6kJfH@GJZW>y#LP+<lp>e9vxl66dhzn
zrLq=uPnR>M-f)~4jZ(bc_k;-98Z=i&Ns6fI)~sP|pblOgv6=zj`yXD3QyQ8?oCC?c
ziC{t_4-vqb8^ysfg#fLO3Do-ejk!BHkb521R}a=ak5sGTJPBIsdEmb_S`36(2CdV!
zpReXgBm%fdF}L|^|M9hN{=d^c%?1g5(AwYIjxmPQvyVO=osUm09;ZR*;?6X-<>6S!
z51&1M^-K2S|LTv(go2C!5nJklRN@dj?^c~`3Jo_=O!s=zkqD6ljUWmV2dC&7j0y-K
zIfxhuk~^<n{nL$H$exRIr|CC0R8j>*G*<&RVrWj_%q=1#aS5-V^oR(?O~>=6Kd;@1
zH0jZ261-qWqQ}>@>@vjp{9s9_!QD{|gS_bBUhf`!`wg$$O5L|r<<#E0&fPuXacqm^
zAN|KS=XrN+;e(e~n^8G$e|GTOcix3hx7YV8m+Us6)jV#qhQh5(DiC9c%?dHW*xvv^
zFk}jam`u>b4Jr`S3O2T)-b~4wI-%3&|GQlBmiw_ZnRU1qB4TuSK|w??AS&FP86cp^
z>M#lc5Chcx#f5bzv*cKP=x6BUJWurC)6Kqx6j!s<2SW!A=!mM3oZS6q!{&b?OgJjF
zaz4&>HETE26bqHl!}c_Py=@l3?fhofx(3VPV2tnn;J;|!``zu<3sEr`1xW+O+$Il+
zalQ!MW;x|Nbu-KfNt}G-+JYDK$*rk^#s)}01_EeK2nuNU&EGf4Z5sFMY^*E|+?<J_
z5St^jOC$y&iGm!vr2sm)sg{e2Cwh>mkU=-Y;>`>l>FoJ>(kjA@`UP530wPurQ$$_h
z{rTcC(WJnz>?-YNnP?R@2TQVYCwAM*i(el1#P$GE`gFhUhZ}Fae(}!H+n;@<a|<<#
z(BK|At=b0%-^QrcDcXM89nBe8QK7ktvs40SZ0x43s5`m^_hPkzx|yp{_{Vn%%~}n7
zSiChiLpQe=8iR<t8CF37WQtwCVi9+9FxTyqOOCT*b54DXva3i7Hd^m5ryP(%=psRN
zA@XX8k-#<X{b-%vXqFrhZa?A8%3|dK?V^ZnPd}%5ad!T->%*IKeH7X!yK>!R_5A$&
zfBcmanob3ZBx}XTeGk-n9@o=6#jYF2o$x_3$KD+ZLqK6QcL#7qa=;S3DP#l$Bn1UC
zK*jlggS8c_TRq5#%%Qp)!DOL07J$%Ur_l|pROG{m0Lbp3<JB1q<Rv<Cpn5S9L}eq(
zj~;G32nfc`wIUcYHg&Q<WcB~}YB)BF5lS@Aa&C<=q;afF^}X4>I)=ymg^P6K<tJB;
z@#bg!v&f6`{^b4QFbelmG*gOlbu#+ads_u483R-vS|Do*#Y6Dw039=O)@I~F4b>6Q
z4N2Vyz!91e;p@K|bX#Q)vt5PAB{V<<Xa!uIh$<;(^6YVWa`ynpf!&+y<+HJi{XRgP
z3y=Gq)yP7L+Sty=YL3t~_Pj$kb8u)41TpZVUufTr-82Te_31>Ujkraidizje|AX7#
z{3z~EX+ACQeD&-}8J7n~@#%MuX{q1Pm(I;FKnN{3c|40i(CRX88ulHg%E?;=C(B+U
z2gkAF07z`gUYO98!BOo61_oFOJzM%G?(WkR?okm?Ge;*0#DLlip|~}kpr`rr==Ndf
zie?0Dd%5Ah2Zr9cL(j|&2<u3xHx)7APG<W;YGNQpfsvG*&Hm3XU%3cVx(XaUsOzu-
zhuNs<C3~M<KX!j`c;-`Ej`bS8*d*Hf->8q?E@zbMd{gE{sAb9OIUWV!Ww)J>SFsU_
zYlRv+rm6*6$WSNohBaqs)f}i0m^VjfBM)xM2zP(Q&5FXxM9I(-5ko-_Y@vXg0%UhO
zzHwtQ6CxpGrD=0MEl3Pz%49{=W8Y!R9c5JA)P_LRreM7px<&*7YT($-&VCyQHtFI<
zp<)YCkB@Sj&gSO3slETRrwDwx+Kjo(O7ENS(a~3@%iIpud+azK@&QHfb&ONZdKu<J
z+|;4d%u}HdgLojWppGsLGl?M)IfE-X7hrX8MRagA1TaMThn5?HG?_=0IGX@X15;_z
zq$!#yA)nklT80n=4cx4Z&z=VK1~ZOKnD-SFfl5CSWgehaN6i&$nH)i|X+yyT1nsq3
ze@LvE(lvUWrhq=X*G9G7tQ*3T>HPWX@tx;Ko5S7l?wU(`cagipGduTpsx=vUx@hPr
zNuWUc#k4G^aAV;NtK^~7Mhxai1q)7!J%XB=nj=ZY38gw9x-~@yCvY?8*Y7)Jm!?{L
zQVY;PtvOM4H3L&+pWi-S#>9?-K;ZfOYBED4b|KLTLu9nVvpIoZpBK}q63f5>Fo6St
zcr%Y|Fl=FrNJm>YNW9qY$9QAf((Kuuh8ia5bX~@EOSV{ty)M>%(-;4Fkelm&_d8u^
zZH)V1Qz6Y$=mV0*gPXyrv@&;Mk%RzX)+ttFpPH+JGJvVq##SqkD&UJ?le)Qrb^m2R
z#~JNg;z$iWGN2h$_7~)wKJ)R>d`P6=4&b2st7ic0xib=I6Do2i25mo)k+>FsfFZUf
zqXGjs0}|_`?pKH3Hp{bVnpz&~WMsb@ZRT6w41EdfOi6j->Fnmg*(dV6>+iw$zw_E}
z{MK)u(m%VaQL|l3hzm#c8hg93JRGoZ2ys$!#Q=su!JCi~IFd7f8#tjub%0<F;D(^U
zq=X1c^yRuYz16VI-L-jrk?yvDWD1U>BClpiVh|(%0?oUtiAXKAAjaORD|saDS|dQS
z&FBoO&`AR#Fvh=zDm&wKmaeUaOH<IeG?cgR{in+hK_qi`?PZHN<T1VT^wH6*f92?u
z2Kq|3{Rw;c&F5zw>qQ&qz{z7P5)$=?U6NfG`BoRCU<D4%U&PjsUfjOBvw$HgB2^~?
zbX7xAM|XEnTzx4yFR}%!<TzqQCvx&ulM7CqRc1j5kplv{x9RL!$svbE7}^dA`vg_F
zrmdJdA|NwvxeE;om4zEMfhxd;Iz~IZC@4qg4!NXn-u^eb_eIe(&yf3oz<PCe-yK|h
z;mOHPy??y;o71oV#<PF%nDk4hUi+{uu=ReMO1smAIgqNm1+N4u1}-lU;uN(K(2J9;
zjH(1BqN5_cDEnZ;7k?kp@BKjJL|Ke7x1?ZWaH1M`0^kN9LqbDsDBz|#ZMI`n&NCf6
zXf1_8bavauQz0Z(Caq8zI<~6b8z8U(x>b}m&Ld6}!Zotfjf3g`etQ4uv4l<+)=jZh
zKZ5pBn||~`{DqT4P31>_oc(@#bc_9)pABnXZD^UWx+zkQbHfT9UYk_%2vDmKC2is`
zDNfC@Lec7MPL6D>#Sqo383H;H7=Y1RZY@}oO+bv!S_X1(K>(m4E-4bY7DFH)to!o`
zfP?hVdI#?#VD@Z9D->c#uwkHzz;S??w@hl&RJYS)*2-{G)@zKRX&+(!1^%mi7^jZA
zop?(QC3KG#qKEaf$A7V(t?GDwCp>&!|HJ+N`oI0Xm-wLCwx=7aVg$BuVqgV`Yk)z-
z#1IvNM`*fpL!lO35gScF&8&LM%79=<pv~D0kj!xLrcs!SMuADKp)(R1cnjKML8mSv
zsF?w{s$Q+HWA78>0L9EU6H2dXs(pnfYJKzbn*_iBYDJCA#mz-96Iq{juRc^%^c>Be
zp2cBm^TyNAy*I>J9>UeV(_87HJF7$EuJ+sgy?>40`NDsAY5Sk)Kdl#|ZM%2F8qLjP
zjEPH-*5+%Ri9uw{$lh8sZiYz63J^kt08QKxloD13ba!KOAv7dv@z)+OI;u6~3J5$B
zYN!DK>m=gVk-Iqn0I6-SwvK=T&;nK1I8*I;BCVwlgq%fyMv0Or^vs0jElNrO69X;Z
zgGq<7qCOeq(|#H3qFxm_oU{bucHLjCSF%6SC0>+2?>@f%!sVCi^ttCBr2qAQ`Qvwf
zeDGU`f_;CeQ<_7HkuX@$)={GYogqavm*OahZhIpEm;kK3;K=~GdNVawcSJBVFa!RU
z4g~005)nEmSrf;C&7kkts59WdZb#cRuE(NowA09-BYLLZV5+<)bOK8C`M!WDW0|#i
zqvjqeBRV>;BTUZ;W@{*B*x7O{ecr6yYU}%MyH^Y6Ar|=Lm-{2Xn(vRx=H*}SQEAGV
zwx4W%_`m(gf5&Mg+OGsVF3E(~(~Y6cAW{}FDj+esxg$kuP2IiKfB@u9lq))+Iyf*X
z7`dzaZ3&GV5Qze?z7RwR29#=Z!aAv%HU`kvc6*49jA3e8r;TYZ0cz?z00-)j62?-i
zRY=#-MG?^iUi59udV(&}#OA3;y-k&;Jh99#S_MXmVTE&k|63J~>*4xFeo{-5-+^l$
zK8?4G$M6Zi^JBJl4GSHH0DTKsucp{B0(dfaj7*p-lCz=%ya+o?0PYB_qN`VTXRFP?
zP!Zj%e`Cnhg^23_UVH0>&CLKg#|Xi$wq;a@7m288uc$d#=?u4Z=ZBq)xm7Ss6{%pI
zcxURaPRF#Lnt4Mra}Wewwnf5;Vk^Ktyy~Mvla~8km*TXZxpC%`N#)D$r<dQkcy^=1
z?>$?8>CMZ9WE|vjuP58-@wLDB>V^heu2u77wPP@5LNY*Ylvu1KP<AJXS%V?C0(vo4
zfMCvG=Eews2>AOe@Kzf_1(Y2*VQ>;+hzLSq^Yrp+I|70!+Q>SR=bcp2@(hI7n40Gr
z5<3kcvW^8w6Yvls1gGGJ4VAKoEf02oCYd7bU;X0lbgJVFDQw3XC1_u#IPCiVi9dU7
z|Gm3^{|n#vv-#}agX?#H<%axzPvd?$Y>*qd?>HSq4#_)JE4zapI(K3U0HURm0=6Vu
zXQiM3ov53eWd*URVK!^c(8&$l%^iMWIXGQyCp4neGnyC?S#uVoz<mDk)3Y^@IF@mz
zP4`rq&opBfW2EZ48X#uv3c0mB<${Hk`XSPbT^3NT)Igg6#h&)o{QeiyhflOP^jEX7
z^xbkLn~&3M+I;!Zx?OC)VgK?+@8py4^kF+YIbQwh7wddiH&Rav=e^E}DD>7b8u;0@
z#3hToU>I@|0(MlW<k_hyHOS!L1Z-9jMNtue6r3F0-2i8Y!Se26v!1qXL_yF*Bm$v>
zz#Te0fBIBe49xeXy6g3xpwYNR4hEi$Gf2ekB<6ac)81@kZ*dB(sUe!bSXZENJ*!#*
zVY<ASf@?npB#0&I4(8BS7yCO0Q)|8SuRi%<`}D2P-+6TX@cA!(3;&<khK(L(X;Qo?
z#n_exM2dEmU&)lHQD_(qJD)&YOq~VM0Gt~cIV%(-ji!jkiiiM=_F^}F_SV!cu7|5m
zhTOG4;tDMQrre`hROg9%W^kW!RkARau7_)~l+dbUnN`v<z+N2Vj?;lVXBAylE3TD5
z98sk}*bE4$V0Eo&H0by9_43-{Y8>NwwcWPRKlE5`9=3<y{PWetAJ<>~lkn2{cLRKT
z<~6eytnQ3N5toLb(fR61x<!wSrVb{!7a@Zt>ZXX8N{#BI&;km2bmxYw0*KxaiGa)r
zzwqZz$Frld?Kg22m<A<;BEhH)jom9F_2}qzzuuVcwWgpjN*88O4JWoia?0yVF=`a1
z^&!ooY-*JhfVvjk*x8j~>kAn!Ed~hE9M@^KP1j{60rqgwA!+{rR~I{zzx%(x_Fz2t
z*lvCje(l4*{EAYamwFkDi{We*Kq7XHmCEz&;9*hJET9XxFjk-n#MY>p_6^xk05o!A
zbs0N%Lux{TFIxZLe7t`=<qL~>Ub95raR^9`5<JLQOd;I}0j!j~gIc%QXR%#pm4z2$
z4T(g3ubUYkp*x48yN3uc?oUg97&|~VMhK0YPG|cvKh>_&Qh;f>xJbAa*_~_<+qf)X
z9ObRm>Aima@aWF@!P(Wkd(MXsPf7FCk<}LK;5V`F7c2&ByT8yo@vZ_OFd`BbZ6q3~
z6*gd24b7uNLnMO$<e`Zm8weu-fPpa*z4tysziu$`uGA6swT&`zmA-TEYo*zpgT$>}
zKDpTLwhd_o`wa}st{N8wh@lzpwn{ct>Pz8?1LCX!A}O;I0w}11%OVd|YVXVrDjOeK
z3mn&xA#^qjoBgkU{Nz@@d-v~s{@U*2?C$LEoPT!tAOxJcI#q#*le8FC6g^<&54VeW
z<ZRK@BSJHFVbxllQOH2BRG+L6qB*FuD^z!LH1ijN84Vm?gQP<m5cb<VZk}8{e{%ZC
zv!~lUH3qx+Qo<Anb$4+!UT<rzo#o>FQbGkslul~#%c4yJ>tu_06GP&zQ)+~YAgUdp
zp>K^22=nSRrxM_1T-PG8cIb`!iv~<g5B#Hh&*bIHzjgNy=3k5P&%blQ>n>O$K(YqQ
z{WvfTu{T%N)6XCs4Gf_|UwoK|sx|h|q!HE@ry#YTl5<Ib<n9Iyq#%+5C=w!=02;j}
zFyNtE6Wz1#5;=BdG3sFzz9GBWq9<u>zuul+k7GCZ*hZPr>}g;_Xc0C;30uxW&BbwK
zfRG2{saYjJ1VZ4-sDJg<?ezeQGzK+X!e*iyPoZ#016>WfZZm}XBUyasTl@Oy>i%!^
zkKyyL{B%qVv+I^{(&Vs>9WA|=iIH*q%iZcI3=yc22A|lPMmFBFlQfI~1f;#GlOxn#
zv9gIfyEz!LLlaa)^5twl)6QEz4k;^SZ_RMM*<a6ZzPYH?mJGF(-R5k)pY*Wc74X4|
z#~{sB5V5bi14ruTi{levS2PZ*h#?9hU{gaQ1|uKWrxYdw48xGi46m)ok~h0u=x3F8
zyod+2{#n01Ir`~a<t~2v@Xe?Md+VZ7XL}^YUXy4b#r*z0-91=RgNRIoU=0E`r~!!_
z2%OzoQxgnCjE&q26IO6&O|7Vc0|A)J*I8J?q67|10kq`WuAe;E)_ccY>iSN|ZNEN0
zA9sy2ZBdTqLJnCc@lGbJo4uAsZ8%;HD;p6x%*H-YBQv7_q|89u!?dtd)D5(7bN@W^
zXjdJ?Z4TnycB!>o%4(X;=#MWS4sU<_`n$t#FWpRFs?%s~Y{Am<W*6szZcBXh2yPrK
zgEMCYQ+pxV9T)-3O~DXM++DF%ETGzq0Fd0A{YBTUIRXH{FMx|cSq_3p7IWh0yWMv4
z=;b?IhoklamdA@Hmsipq2$meGw%u6k*f>NB#WU2zbocIJSRVKUh-`5|MH{)egM&3F
zXCV!e8mEK%x2|hxUC)AjKj~b-@Vw#*o|C*D!~A=X9zVI4@9Wva&q?t8Y!TTykOA9@
zma`7oX#4ylz~dVWA}}Q)3Jek5-5r3rp#c810B-J@Bawny(^9oG2k@p~Z02sx$Emd`
zh0w<^NC-oRwI6IR%P3)-Rspf@AN}Nc9xYO7+Qr2p?#x*@2kf=D41MY8r0-;&7T(;w
z#vU2XNZkop5Q1nuKfDlKrrlGWaoSkm3A=qqBCZsW4obQ6AbMGhe>Cjgyw@&o9S>aE
zG@(hUZK74MN(hR){{D{MIO>tm6hOFD4W?Nf9NZN-fS3};V&v$m22@#z8#WH^fJTmn
z6_mhcue2$GnsRhcfroSil{y2nnp5YNx1YZEV81_Du<pk+U&!vXOk*ukLNM+dutk?$
zpwwX((SW+t;jE&PFo2<Ru|3t+lG3il(2B9ocD3MhBg`D;*W8WAc$Pe3mv+m4KiB1(
zpM3U%oc7p)#WjYAF<CtElnu}RBzLbIpOAxz0aiwGgVq@VYI2BP(TJRs!M&lN=M<b(
zS=nkdbW$fY1a64*_B+dwJ4FUuRfynLJc*3^fXg|CsXh4V4@Sij`n~q47u)al6iOch
zdYyK(*t-uzL3<oJCpRD#E^LmbfPvg8?iURv=hnCg0UEandaT^`G!3@~_DruFU+b-L
z7hivN(#AiS-oEkN0aG%o(Ztg-<++N3Jo;`~zIy8b-Jv0JXcmASMH~I~69+_SU<3gb
z4O*)J3Pz?b)(W=ffS`of9FRUQT1tzq?xvJL!s0L@T^Yc#lkKw){_;I<B161tv6}~-
zkPN*?3aO1sR<bQwHn8Y$Kth1pKv=*@1(eYdpxV6Q^R~*v7%s!&nzbR0m=0|9q#r_S
zLhIw#-mkeFtUn*Gzkc)b{YSIaW)y2_G&?ZrW^^~?P&a>hsy9EkjM2M{<e(rx2Fhr}
zPEjGaVQN6YriLae=2TgwLEs%z$kCOW3jnaX;r#CNR1E>$Yv<U2jkqn?C9tm_uOIxp
zj5I@54g(Ki4-*oY9;TrcCCS<%ZD?O}?6s=3X0A25BRK)OIjH*41<Xfjx#v}wH_s45
zb&{8KNb|kF)^ghM#f|jvo9p?X-oeL*zZCxT(L6S@el)ht<UK^`_{n$w{p0?vFCPp8
zxOajIp1}<x8L&4sHX{J8WN7M!FI;gncTsRNFsz^kN~Q(P5Q5PgsLrFSNRKRmX?Za3
z(#%OTKmO#;-XHe?F-W6>IES1!yEO-1Xo5yy7S16`kg1WkT0%*Rt|(Z=Km&k-iw1cv
z8MDaRod~eylXQmele3M><<;>iFv6o?FRjYK5KgadeI{-G&`O`xA~WCWQ#^lq@F5-?
z-bo=PXV0Aj11q@oS(+vV1n*Q38)~NnDroMgY{-xa06-c#xT7nw0}!a8z4`BEaT<yl
z*VLkH9Z67YtG<2q^M@NZPs?F%axe^3weGb}%O2)5av|StVJIcbHZqE(R@#7&=4?F0
z0gV8Gv--@gwe@+S-s*BIvEtg;lmSZL^{4l*-QN4g$A0qtzxOYEfAH#)Z{+vu!?a7S
zcEu%z>r3C8Ud~_XVi%IkvSDXjQ9v+`1)C$eA%a;=82~Wk8pr@WWn?rDOy*vbL+-qC
z@X8A4=ni+5R;VH;R~oW~0lGbx?G^I+-485PUWrf8LJZx0Dr3Jh!&%23GwZl-Gp|5=
ziyBDau3sXwdO3QU%>o4v0EDRHyXj++48}HuZsYM_v+FeZFkR5A`0_9U<@;sz3~%b)
z)03BLIZlsv{kF3<1V!80Sm(2w_m7rwHJc4g7!d&jVeO~VkyFFw3XDzMz(aEn1w5dV
ziW@;7FhnC^4*=|lz`zIq1n}Xxyx@X4yCVkX(7HzBvyVQ~ZKKe0bsb_He9EUaH=iAM
zoRZd(ubi?%9w{~7wcYjDvNZs@Aq<GY#Er0@{@MTl1;0r|K~xY&Ta^92?p;Dt9CwF?
z+y{vZIoUPMpqFxe_Lb+q@$<F(;TsR{KJ4<xf6=|Io4I@1UAvZ9xOH;(&TU#QPD0dD
zt1B2lwx$z^)@no?lc6!9AQp&*y;DVYs>X<_hzelMfKZ&kP+g4>6v9{bxgmoAVPv7$
zBX+$#e)#bz;(QjDh%&j%f}dSppyr1CK;#rC8Bi0|hUDF(Tg@~A1OgENYo=y^&6M2`
zWiBb9n#DQx@$6n6bvN^U8qd3*z4Y=(?{?!?$=>|l?|gduv+4Fvuq=Xs*NJd`<JR4i
z+l!-cayT<I=7=oZuwf2~z^bB%S~bpE&9di+iedzAO3e`qRM$ogNx^b4XvN*Y+}xS$
zewT492pE!4Jcs}u1K4#@jEnh^0NKGT^fKixN(#7~IS<`_Yq?M>8kBp1zWE+J2MqJt
zB3VNw3X0kxl6{i)L%+{jVpP1O2)lmKba%BJ35u`ljidFZzq<eNr<dO<w?3_Jm0$nl
z^6;URzAlz)Kx3Ho;!ufMsE|^YSQ%$Z#A1q-1x=BuNhMF7)LU>!G%<sj5Q4Fz*O-ND
z*OPf81tJ3{S#_J%15%5b0s%zJSfEJ_)Ot{&B1E08!JbEE_YgW@FX4K#*{ip`<DT!T
zk7<jGo3{^>%=97;!z^?lL?BxK>icKjw9P{H-eU&sMml*C^J_PD74i6LygOW%S$|N^
zl`ekcFRr`k@aLzzU$nYtM(O5kKAS;|F34<VNFEHukc*(A2{>4ZMx+R8Wa5VGB<lDA
zdf{x$7Eqi8IglGi1Rw%1QTpuDS(6y>#bk`p6dIC)H(Z3^6PWb_7Txa}*I2vbMMu^1
zuC`(&bWJo+1GYkuXc>jW&6T+UHgs}qV8a^+?s>7@BP$lEXn20pYX9=~+)}XaY<55L
zPtWbat9BM2|5|<Y`0A^#O+#kWXz6g6_e24K!3_YJy*Z<I*r--e2K5H1SS^mIAtVQq
z;DV4u`Gw!@uHI15Dq&SZQ!p?!fJ*qKp+Mf9HlJz`PqjlKPQY>21^2)-4@gC)$c1+O
zA_SH&LfLG#JMtZ1I3&;sy<Wkb4Fu#Ra8>i>=3t7mdnb?!hb6A=a#4!D+Mk}J>GnBU
zJMJ!S{d{?1^6>J0zBl2|`jaVK)}vp~c>@LJ2V$|0UKJuKVL-GOgUX7H0lk5PBP&K`
zqavP5G_bvZxfy$?NW$KT83B~N0$8wG6N5W2Ip5ot7FVmV>5!GHu({<L(vV^xZ*4YX
z8`TeS+4-tFS|;M=yR6#+4TR;OBM@|XQC3hlicIW4V$GU?T;x>Q#@x#O6JiUukC(&a
zRh(HB=vz7OPB(A&E#3O1X|{TP`zNn#B_D0$@AW;oP#i)Xof1(awq^j08O4Yk#{q&=
zbqD}h9g+!42pdDgik!R`Gh!g>q=}o2R9ulT17cBgaB#DOJHhkWq*Brz8gfDu={o9r
z_7{fYcE2Wsd4Wz2`W|U2`?}wysj_ysygQI(gG%G2p}ASZT8K>*fq_Y{Ke^_^dM*;B
zJ9LnJSBq{Icl=4hfb(!ReYgISU{}LOSLnkB;ln4NjSm<9Xjm<djsjwz0+V~dkQ^LI
z8kjZ*=n6Do0jn)IGBh!t6e)TL0O)4WS4Sx7#0mmQp@3IIKvQsVw~E}oq8wuzC!V-^
zms;0Rl5Vw#(0ml!?@BX?=pA<<^&}p~{qB0)cfqh8E_vYN4o!C4-@Oe0tz0~vb+ZU8
zXow9jt}e%I!}<Jaw?0r4*tL)k?tJa17dM{tbmO!2;rQTN^7xI1pLQWb!qruPlOJ|J
z{~Mp)NHli_M%Uorq#|T&?7-0oz?_MyJ2()MIUs?%8!0#dl7WdBK$n_h(AZoY5ecG`
z0RR#LEADeN04L)%>m;JJAQ%&SvFn}XT=KMJ<_a=Pi-^$1-PM)Xv3OX_ZbXXzKV6m+
Uqlh+af&c&j07*qoM6N<$f=oBo-2eap

literal 0
HcmV?d00001