fashion_classification_full.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from src.train import *\n",
    "import os\n",
    "\n",
    "DATA_PATH = '/home/ud/datasets/fashion-dataset/fashion-dataset/'\n",
    "CKPT_PATH_TOP20 = 'ckpts/best_val_top20.ckpt'\n",
    "CKPT_PATH_FT = 'ckpts/best_val_ft.ckpt'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train on top-20 classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Preprocessing data\n",
      "INFO:fashion:Cleaning csv\n",
      "INFO:fashion:Reading clean csv into df\n",
      "INFO:fashion:Splitting into test-train\n",
      "INFO:fashion:Sub-splitting based on top-20 classes\n",
      "INFO:fashion:Create maps for categoires\n",
      "INFO:fashion:Creating datasets\n",
      "INFO:fashion:Created train dataset with 14400 samples\n",
      "INFO:fashion:Created val dataset with 3600 samples\n",
      "INFO:fashion:Created test dataset with 15149 samples\n",
      "INFO:fashion:Creating dataloaders\n",
      "INFO:fashion:Creating model\n",
      "INFO:fashion:Creating trainer\n",
      "INFO:fashion:Loaded previous checkpoint trained on 13 epoch(s) with final loss=0.1611, acc=0.9403\n"
     ]
    }
   ],
   "source": [
    "processor, trainer_top20, dataloaders_top20 = setup_top20(ckpt_path=CKPT_PATH_TOP20, data_path=DATA_PATH, mt=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda\n",
      "True\n"
     ]
    }
   ],
   "source": [
    "print(trainer_top20.device)\n",
    "print(next(trainer_top20.model.parameters()).is_cuda)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 0/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1028d6dd01c34b5f8781198e674685bd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.8084 Acc: 0.7785\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0edbe29f0593425f9cc2211d62ab3a28",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.2698 Acc: 0.8986\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 1/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b407415e6ef2435782573259d9a1075b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.2389 Acc: 0.9102\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "91c1e5fce75449e4951604018a440310",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.2047 Acc: 0.9231\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 2/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "583f29d77eec4c6ebd36c8a1302510d8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.1750 Acc: 0.9328\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3520fce2fdf24375a44e18f795fa1988",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1911 Acc: 0.9217\n",
      "INFO:fashion:Running epoch 3/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1738f83fad71437caf181916c2ef3b1e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.1445 Acc: 0.9417\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4f7102b11d5340398445ffa6308f78b6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1756 Acc: 0.9339\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 4/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ba5c577505a34c18bc5b6c766db7a6b9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.1226 Acc: 0.9488\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e5d0587d3b62497381bd04c35f9529c0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1669 Acc: 0.9350\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 5/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c482f04ccd75488fb25b0f03142535c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.1125 Acc: 0.9536\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ebf26cfb0b4c4c76a2c774fa47f1223e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1660 Acc: 0.9342\n",
      "INFO:fashion:Running epoch 6/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c8d5b5e948744318ab88de0a00410945",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.1071 Acc: 0.9548\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3627c782bc8648af9cdaa123858edd14",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1668 Acc: 0.9339\n",
      "INFO:fashion:Running epoch 7/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "835b06835747463aa63fb9652b2fb5f1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.1017 Acc: 0.9584\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0db2b150e112463b99dac96266909637",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1660 Acc: 0.9319\n",
      "INFO:fashion:Running epoch 8/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0652117acd6141adb0ae7deefcb60bb1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0950 Acc: 0.9573\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3c04e7860f4d4737861aa2a4ae0cc5fe",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1620 Acc: 0.9383\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 9/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9dddc26f35d848c2996ffa25409880a1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0945 Acc: 0.9606\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "593e5fb4dd064e3a97c7a6ec4a454323",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1628 Acc: 0.9400\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 10/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0dbfa540203c47b38972c2b97959f130",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0937 Acc: 0.9597\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9f8c48b6dac94a9282fdd55d315de450",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1624 Acc: 0.9383\n",
      "INFO:fashion:Running epoch 11/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "46cb8d02848c48d29aac719e61184576",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0902 Acc: 0.9628\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ad5bbd9019d44da98e12ee35bb8d27b4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1642 Acc: 0.9378\n",
      "INFO:fashion:Running epoch 12/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9a8680a1ec0d4cdeaecc041de0b079ef",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0873 Acc: 0.9646\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d4927c2b32c14e44bc544282395096d1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1611 Acc: 0.9403\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 13/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "72e90b7a5b3f4518b6c5878a8df040a4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0916 Acc: 0.9605\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "689dde4386734ff79a36a8783c65a619",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1639 Acc: 0.9394\n",
      "INFO:fashion:Running epoch 14/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "341cd21188fd427ab6d74e4e72c42637",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0884 Acc: 0.9622\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "74820a7aac284909844a99f163decae2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1614 Acc: 0.9400\n",
      "INFO:fashion:Running epoch 15/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7775d64ead924887a08572fc7c86a434",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0895 Acc: 0.9624\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "95925252cd074fa78a1416774ce919c0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1629 Acc: 0.9403\n",
      "INFO:fashion:Running epoch 16/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f635c6ce6bf74c189bcc212ce7332c11",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0872 Acc: 0.9628\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c780c85e984e4c7d8e99e741f5c676b8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1635 Acc: 0.9386\n",
      "INFO:fashion:Running epoch 17/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f96260e299b74baca5872e31f3cd94f0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0869 Acc: 0.9635\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a765f5ac1a0f43e2bbea8287c18428ae",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1630 Acc: 0.9403\n",
      "INFO:fashion:Running epoch 18/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "559d25acfc3f4f64b5ae2013e624e0df",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0890 Acc: 0.9631\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7142c48a10d345939a93a18a6d6bf7e9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1657 Acc: 0.9397\n",
      "INFO:fashion:Running epoch 19/19\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0bd6edcf584540ec97a0e22341a21eef",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=225.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.0889 Acc: 0.9618\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6189e54d972d489da0cff4878291375d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=57.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 0.1624 Acc: 0.9397\n",
      "INFO:fashion:Training complete in 109m 58s\n",
      "INFO:fashion:Best val Acc: 0.940278\n",
      "INFO:fashion:Setting model to the best one found during training\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "trainer_top20.train(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeMAAAGDCAYAAAAYg3hiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdeXxcVf3/8ddnJpnJMkm6JKSU1raUtRtdQimLNAXEVgQEUfZFxX5RccOt8kV+iPoVERVRXEBFESTwxS9asVBRWxZZW5ZCWwqltFAK3dM2TbOf3x/3Jpmmk2TaTma5eT8fj3nkLufeOWcmyXvOnXvPNeccIiIikjmhTFdARESkv1MYi4iIZJjCWEREJMMUxiIiIhmmMBYREckwhbGIiEiGKYwlo8xspJk5M8tLouxlZvZEOurlP99ZZva2mdWZ2aR0Pa+kh5n93sy+m4bnuc7M7urr55HcpjCWpJnZajNrMrPyLstf9AN1ZGZq1mduAq50zsWccy9kujIiElwKY9lbbwLnt8+Y2XigMHPVSb24XvoIYOk+7iOcuhoFSzJHQUT6G4Wx7K0/ApfEzV8K3BlfwMzKzOxOM9toZmvM7BozC/nrwmZ2k5ltMrNVwGkJtv2tmb1rZu+Y2XeTCba4w92zzWydv/1X4taHzGyOmb1hZpvN7D4zG9Rl20+Z2VvA42ZWB4SBl8zsDb/ckWa20MxqzWypmZ0Rt//fm9kvzWyeme0EZvjLfmFmD/mHuv9jZkPM7GYz22pmr8Yf/o6r3w4zW2ZmZ8Wtu8zMnvBfu61m9qaZzYpbP8jM7vDbvtXM/hK37sP+0YtaM3vSzCb08DoeZ2bPmdk2/+dx/vLzzGxRl7JfNrO5/nTUr9tbZrbezH5lZoX+umozW2tm3zCz94A7unnuT5rZcr/+881sRNw6Z2ZfMLNV/u/OD+N+p0L+79gaM9vg/+6VxW17gt/uWvO+drgs7mkHmtnf/df8GTMb7W9jZvYTf3/bzGyJmY3r7nXbG2Z2hv/7U+v/Ph0Zt+4b/u/9DjNbYWYn+8unmtkiM9vuv74/TkVdJIs45/TQI6kHsBo4BVgBHIkXVm/j9SAdMNIvdyfwV6AEGAm8BnzKX3cF8CowHBgELPC3zfPX/wX4NVAMHAA8C/yXv+4y4Ilu6jbS3889/rbjgY3AKf76LwFPA8OAqP8c93TZ9k5/20J/uQMO8afzgZXA1UAEOAnYARzur/89sA04Hu9DboG/bBMwxZ//N96RhUv81+67wIK4NnwMGOpvfy6wEzgwru3NwKf9bT8DrAPMX/934F5goF/X6f7yycAG4Bh/u0v99zGa4DUcBGwFLgby8I6AbAUGA0V+ew+NK/8ccJ4/fTMw199HCfA34Pv+umqgBfiB/9oXJnjuj/iv75H+c18DPBm33uH9rgwC3of3O3W5v+6T/rYHAzHg/4A/+uve59f7fP91GQxMjHvPtgBT/ee8G6jx130QWAwMAMyv14H7+HdzHXCXP32Y/75+wK/P1/26R4DD8f6ehsb9Xo72p58CLvanY8C0TP8/0CO1j4xXQI/cedAZxtcA3wdmAo/4/8ic/88jDDQCY+K2+y9goT/9b+CKuHWn+tvmAZX+toVx68/HDyySC+Mj4pbdCPzWn14OnBy37kC8cMuL2/bgLvuMD+P3A+8Bobj19wDX+dO/B+7ssv3vgdvj5j8PLI+bHw/U9vB6vwicGdf2lXHrivz6DfHb0gYMTLCPXwLf6bJsBX5Yd1l+MfBsl2VPAZf503cB1/rTh+KFXBFeWO1sDw5//bHAm/50NdAEFPTQ1ofwP7D58yGgHhgR917MjFv/WeBf/vS/gM/GrTs87r39JvBAN8/5e+A3cfMfAl71p0/CC/xp8e/5Pv7dXEdnGH8LuK9LO9/xX6ND8D44nQLkd9nHY8C3gfK++NvWI/MPHaaWffFH4AK8gLizy7pyvE/5a+KWrQEO8qeH4n36j1/XbgReb+Fd/xBeLV4P9oC9qFvXfQ+N2/cDcftdDrTifQBItG1XQ4G3nXNtXfZ/UNx8ou3Xx03vSjAfa58xs0viDifXAuPwXs9277VPOOfq/ckY3lGGLc65rQmefwTwlfZ9+vsdTufr0rWNa7osi2/jn+g8X+AC4C9+PSrwQnlx3HM87C9vt9E515DgOePr+dO47bfghXx3r2/8e9u13mvo/HA3HHijh+d9L266Hv/9cM79G/g5cCuw3sxuM7PSrhub2fv9ryDqzCyZ8wt2q6v/+/Q2cJBzbiXeEZzrgA1mVmNm7W38FF6v+lX/64MPJ/FckkMUxrLXnHNr8A63fgjvkGC8TXi9khFxy96H9+kf4F28f5Dx69q9jdczLnfODfAfpc65sXtRva77Xhe371lx+x3gnCtwzr0TV76nW5itA4a3f0+ZoF29bd8j//vR24ErgcHOuQHAK3iB1Ju3gUFmNqCbdd/r0u4i59w9CcquY/f3DXZv4z+AcjObiBfKf/KXb8L7YDE27jnKnHOxuP309tq8jfd1RHw9C51zT8aV6e697Vrv9+EdFl/v73d0L8+dkHPuFufcFGAsXhB+LUGZx513tn0syd/T3epqZobXrnf8/f3JOXcCnV/9/MBf/rpz7ny8D6Y/AO43s+J9aZdkJ4Wx7KtPASc553bGL3TOtQL3Ad8zsxI/ZK7CO8SJv+4LZjbMzAYCc+K2fRfvH/6PzKzUPzFntJlN34t6fcvMisxsLPAJvO9RAX7l12kEgJlVmNmZe7HfZ/AOxX7dzPLNrBo4HajZi330pBjvn+9Gv36fwOsZ98p/3R4CfmFmA/36neivvh24wsyO8U9KKjaz08ysJMGu5gGHmdkFZpZnZucCY4AH/edpAe4Hfoj33e0j/vI2/3l+YmYH+PU/yMw+uBft/xXwTf99az+R72NdynzNb99w4It0vrf3AF82s1FmFgP+B7jXr+/dwClm9nG/TYP9DxM9MrOj/dcsH+99b8A7krK/7gNOM7OT/X1/Be8D6JNmdriZnWRmUf/5drU/p5ldZGYV/mtd6+8rFfWRLKEwln3inHvDObeom9Wfx/sHtgp4Aq8H9Tt/3e3AfOAl4Hn27FlfgneYexneyUP3430nmqxH8U6I+Rdwk3PuH/7yn+KdYPQPM9uBdzLXMcnu1DnXBJwBzMLrCf4CuMQ59+pe1K2n/S8DfoT3He16vO+T/7MXu7gY74jEq3jfO37J3+8ivJO+fo73eq7E+3ohUR02Ax/GC4jNeCcXfdg5tymu2J/wvtP8Xz/s2n3D3/fTZrYd+Cfed7dJcc49gNfjq/G3fwXvtY73V7yTql7EO2Htt/7y3+F9dfIY3hGbBrzfQZxzb+EdwfkK3qHvF4GjkqhSKd7v6la8w8qb8a473y/OuRXARcDP8H6PTgdO93+/osAN/vL38HrBV/ubzgSWmneW/0/xTpzr6bC/5Jj2MzFFcpp5A468iXfiS0vPpSXXmJnDO5N7ZabrItIX1DMWERHJMIWxiIhIhukwtYiISIapZywiIpJhCmMREZEMy9jdU8rLy93IkSNTtr+dO3dSXBy8a+CD2K4gtgmC2S61KXcEsV1Ba9PixYs3OecqEq3LWBiPHDmSRYu6u0x17y1cuJDq6uqU7S9bBLFdQWwTBLNdalPuCGK7gtYmM+s63GwHHaYWERHJMIWxiIhIhimMRUREMixj3xmLiEjympubWbt2LQ0NiYekLisrY/ny5WmuVd/K1TYVFBQwbNgw8vPzk95GYSwikgPWrl1LSUkJI0eOxLvz4u527NhBSUmim3Hlrlxsk3OOzZs3s3btWkaNGpX0djpMLSKSAxoaGhg8eHDCIJbsYWYMHjy42yMY3VEYi4jkCAVxbtiX90lhLCIivdq8eTMTJ05k4sSJDBkyhIMOOqhjvqmpKal9fOITn2DFihU9lrn11lu5++67U1FlTjjhBF588cWU7Kuv6TtjERHp1eDBgzuC7brrriMWi/HVr351tzLOOZxzhEKJ+3l33HFHr8/zuc99bv8rm4PUMxYRkX22cuVKxo0bxxVXXMHkyZN59913mT17NlVVVYwdO5brr7++o2x7T7WlpYUBAwYwZ84cjjrqKI499lg2bNgAwDXXXMPNN98MwKmnnsqcOXOYOnUqhx9+OE8++STgDZP50Y9+lKOOOorzzz+fqqqqXnvAd911F+PHj2fcuHFcffXVALS0tHDxxRd3LL/lllsA+MlPfsKYMWM46qijuOiii1L+miWinrGISI759t+Wsmzd9t2Wtba2Eg6H93mfY4aW8v9OH7tP2y5btow77riDX/3qVwDccMMNDBo0iJaWFmbMmME555zDmDFjdttm27ZtTJ8+nRtuuIGrrrqK3/3ud8yZM2ePfTvnePbZZ5k7dy7XX389Dz/8MD/72c8YMmQIf/7zn3nppZeYPHlyj/Vbu3Yt11xzDYsWLaKsrIxTTjmFBx98kIqKCjZt2sTLL78MQG1tLQA33ngja9asIRKJdCzra4HoGb+3rYEXNrTQ1NKW6aqIiPQ7o0eP5uijj+6Yv+eee5g8eTKTJ09m+fLlLFu2bI9tCgsLmTVrFgBTpkxh9erVCfd99tln71HmiSee4LzzzgPgqKOOYuzYnj9EPPPMM5x00kmUl5eTn5/PBRdcwGOPPcYhhxzCihUr+OIXv8j8+fMpKysDYOzYsVx00UXcfffde3Wt8P5IqmdsZjOBnwJh4DfOuRu6rH8f8AdggF9mjnNuXorr2q1HX9vAT59v5GMfaGDYwKJ0Pa2ISEYk6sFm8prc+Dsrvf766/z0pz/l2WefZcCAAVx00UUJL/OJRCId0+FwmJaWloT7jkaje5Rxzu1V/borP3jwYJYsWcJDDz3ELbfcwp///Gduu+025s+fz6OPPspf//pXvvvd7/LKK6/s11GHZPTaMzazMHArMAsYA5xvZmO6FLsGuM85Nwk4D/hFqivak/KY92ZtqkvujD4REekb27dvp6SkhNLSUt59913mz5+f8uc44YQTuO+++wB4+eWXE/a8402bNo0FCxawefNmWlpaqKmpYfr06WzcuBHnHB/72Mf49re/zfPPP09raytr167lpJNO4oc//CEbN26kvr4+5W3oKpme8VRgpXNuFYCZ1QBnAvGtd0CpP10GrEtlJXtTUeKH8Y7GdD6tiIh0MXnyZMaMGcO4ceM4+OCDOf7441P+HJ///Oe55JJLmDBhApMnT2bcuHEdh5gTGTZsGNdffz3V1dU45zj99NM57bTTeP755/nUpz6Fcw4z4wc/+AEtLS1ccMEF7Nixg7a2Nr7xjW+k54hD+6no3T2Ac/AOTbfPXwz8vEuZA4GXgbXAVmBKb/udMmWKS5V3tta7Ed940P3pmTUp22e2WLBgQaarkHJBbJNzwWyX2pQ9li1b1uP67du3p6km6dNdm5qbm92uXbucc8699tprbuTIka65uTmdVetVovcLWOS6ycRkesaJhhLpegD+fOD3zrkfmdmxwB/NbJxzbrczqsxsNjAboLKykoULFyb3iaEXzW1edZ57+VUOrF+Vkn1mi7q6upS9TtkiiG2CYLZLbcoeZWVl7Nixo9v1ra2tPa7PRd21qba2ljPOOIOWlhacc/zkJz9h165dGahh9xoaGvbq9yyZMF4LDI+bH8aeh6E/BcwEcM49ZWYFQDmwIb6Qc+424DaAqqoqV11dnXRFe1O84O+UlA+lunpcyvaZDRYuXEgqX6dsEMQ2QTDbpTZlj+XLl/d4uDQXb6rQm+7aVFJSwgsvvJCBGiWvoKCASZMmJV0+mUubngMONbNRZhbBO0FrbpcybwEnA5jZkUABsDHpWqRAacTYWKfvjEVEJPf0GsbOuRbgSmA+sBzvrOmlZna9mZ3hF/sK8Gkzewm4B7jMPz6eNqURY9MOnU0tIiK5J6nrjJ13zfC8LsuujZteBqT+lLm9UBZVz1hERHJTIEbgAi+MdWmTiIjkosCEcWnE2NHYQkNza6arIiISONXV1XsM4HHzzTfz2c9+tsftYrEYAOvWreOcc87pdt+LFi3qcT8333zzboNvfOhDH0rJuNHXXXcdN910037vZ38FJozLot4VWBvVOxYRSbnzzz+fmpqa3ZbV1NRw/vnnJ7X90KFDuf/++/f5+buG8bx58xgwYMA+7y/bBCaMSyNeGG/S98YiIil3zjnn8OCDD9LY6P2PXb16NevWreOEE06grq6Ok08+mcmTJzN+/Hj++te/7rH96tWrGTfOu/R0165dnHfeeUyYMIFzzz13t2uEP/OZz3TcfvF73/seALfccgvr1q1jxowZzJgxA4CRI0eyadMmAH784x8zbtw4xo0b13H7xdWrV3PkkUfy6U9/mrFjx3Lqqaf2ei3yiy++yLRp05gwYQJnnXUWW7du7Xj+MWPGMGHChI4bVDz66KNMnDiRiRMnMmnSpP2+xjswt1AcEG0PY51RLSIB99AceO/l3RYVtrZAeD/+pQ8ZD7Nu6Hb14MGDmTp1Kg8//DBnnnkmNTU1nHvuuZgZBQUFPPDAA5SWlrJp0yamTZvGGWecgVmiMaPgl7/8JUVFRSxZsoQlS5bsdgvE733vewwaNIjW1laqq6tZsmQJX/jCF/jxj3/MggULKC8v321fixcv5o477uCZZ57BOccxxxzD9OnTGThwIK+//jr33HMPt99+Ox//+Mf585//3OP9iS+55BJ+9rOfMX36dK699lq+/e1vc/PNN3PDDTfw5ptvEo1GOw6N33TTTdx6660cf/zx1NXVUVBQsDev9h6C0zPWYWoRkT4Vf6g6/hC1c46rr76aCRMmcMopp/DOO++wfv36bvfz2GOPdYTihAkTmDBhQse6++67j8mTJzNp0qRub78Y74knnuCss86iuLiYWCzG2WefzeOPPw7AqFGjmDhxItDzbRrBu79ybW0t06dPB+DSSy/lscce66jjhRdeyF133UVenveB5/jjj+eqq67illtuoba2tmP5vgpMz1iHqUWk30jQg92VhhG4PvKRj3DVVVfx/PPPs2vXro4e7d13383GjRtZvHgx+fn5jBw5MuFtE+Ml6jW/+eab3HTTTTz33HMMHDiQCy+8sNf99DSkRfvtF8G7BeO+Dpn597//nccee4y5c+fyne98h6VLlzJnzhxOO+005s2bx7Rp0/jnP//JEUccsU/7hwD1jPNCxoCifIWxiEgficViVFdX88lPfnK3E7e2bdvGAQccQH5+PgsWLGDNmjU97ufEE0/k7rvvBuCVV15hyZIlgHf7xeLiYsrKyli/fj2PPPJIxzYlJSUJv5c98cQT+ctf/kJ9fT07d+7kgQce4P3vf/9et62srIyBAwd29Kr/+Mc/Mn36dNra2nj77beZMWMGN954I7W1tdTV1fHGG28wfvx4vvGNb1BVVcWrr766188ZLzA9Y/Dua6zD1CIifef888/n7LPP3u3M6gsvvJDTTz+dqqoqJk6c2GsP8TOf+Qyf+MQnmDBhAhMnTmTq1KkAHHXUUUyaNImxY8dy8MEHM23atI5tZs+ezaxZszjwwANZsGBBx/LJkydz2WWXdezj8ssvZ9KkST0eku7OH/7wB6644grq6+s5+OCDueOOO2htbeWiiy5i27ZtOOf48pe/zIABA/jWt77FggULCIfDjBkzhlmzZu3188WzNI9a2aGqqsr1dl3Z3li4cCG/ei1Ka5vjf684LmX7zbRcHdS+J0FsEwSzXWpT9li+fDlHHnlkt+v7040ickGi98vMFjvnqhKVD8xhaoCKkgKdTS0iIjknUGFcHovoMLWIiOScgIVxlLrGFnY1aUhMERHJHYEK44oS7zR2nVEtIkGUqXN8ZO/sy/sUrDCOeWGsWymKSNAUFBSwefNmBXKWc86xefPmvR6RK1CXNnX0jPW9sYgEzLBhw1i7di0bN25MuL6hoWG/h2TMNrnapoKCAoYNG7ZX2wQqjMvVMxaRgMrPz2fUqFHdrl+4cCGTJk1KY436XhDb1J1AHaYeHIsAsGmHLm8SEZHcEagwzg+HGKghMUVEJMcEKoxBQ2KKiEjuCWQYq2csIiK5JHBhXFGiMBYRkdwSuDDWYWoREck1wQvjkgg7m1qpb2rJdFVERESSErgwbh+FS5c3iYhIrghcGJeXaOAPERHJLYEL446escJYRERyRPDCuL1nrJO4REQkRwQujAcV+0NiqmcsIiI5InBhnB8OMag4ojAWEZGcEbgwBiiPRXSYWkREckZAwzjKpjpd2iQiIrkhkGGsITFFRCSXJBXGZjbTzFaY2Uozm5Ng/U/M7EX/8ZqZ1aa+qsnTkJgiIpJL8norYGZh4FbgA8Ba4Dkzm+ucW9Zexjn35bjynwcm9UFdk1Yei1Lf1MrOxhaKo702UUREJKOS6RlPBVY651Y555qAGuDMHsqfD9yTisrtq/ZrjXWoWkREcoE553ouYHYOMNM5d7k/fzFwjHPuygRlRwBPA8Occ60J1s8GZgNUVlZOqamp2f8W+Orq6ojFYgAs2djCjxc38t/HFHDowHDKniMT4tsVFEFsEwSzXWpT7ghiu4LWphkzZix2zlUlWpfMMVxLsKy7BD8PuD9REAM4524DbgOoqqpy1dXVSTx9chYuXEj7/srf2caPFz/B8EPHUj1uSMqeIxPi2xUUQWwTBLNdalPuCGK7gtim7iRzmHotMDxufhiwrpuy55HhQ9QAB+hmESIikkOSCePngEPNbJSZRfACd27XQmZ2ODAQeCq1Vdx7g4ojmMEmnVEtIiI5oNcwds61AFcC84HlwH3OuaVmdr2ZnRFX9HygxvX2JXQa5IVDDCzSkJgiIpIbkrruxzk3D5jXZdm1XeavS1219l+FrjUWEZEcEcgRuADKS9QzFhGR3BDYMK7Q+NQiIpIjAhvGGhJTRERyRXDDuCTKrmZvSEwREZFsFtgwrohpSEwREckNgQ3j8vaBP3SoWkREslxwwzgWAdQzFhGR7BfYMK5Qz1hERHJEYMN4UJE3JOZGXd4kIiJZLrBhnBcOMUhDYoqISA4IbBiDd6hah6lFRCTbBTqMy2NR9YxFRCTrBTqMK0oUxiIikv0CHcblsQgbdzSSBXd1FBER6VbAwzhKQ3MbO5taM10VERGRbgU6jNuvNd6kk7hERCSLBTqMy/3xqTfqe2MREcli/SKM1TMWEZFsFugw7hgSUz1jERHJYoEO40HFEUKmnrGIiGS3QIdxOGQMKo5ofGoREclqgQ5j8L431pCYIiKSzQIfxhqFS0REsl3gw1jjU4uISLYLfBi337lJQ2KKiEi2CnwYl8ciNLa0UdfYkumqiIiIJBT4MO4YElNnVIuISJYKfBh3DImpM6pFRCRL9Zsw1klcIiKSrQIfxh1DYqpnLCIiWSrwYTywyB8SUz1jERHJUoEPY29ITF1rLCIi2SupMDazmWa2wsxWmtmcbsp83MyWmdlSM/tTaqu5f9qvNRYREclGeb0VMLMwcCvwAWAt8JyZzXXOLYsrcyjwTeB459xWMzugryq8L8pjulmEiIhkr2R6xlOBlc65Vc65JqAGOLNLmU8DtzrntgI45zaktpr7pyIW1W0URUQka1lvw0Sa2TnATOfc5f78xcAxzrkr48r8BXgNOB4IA9c55x5OsK/ZwGyAysrKKTU1NalqB3V1dcRisYTr7l3RxCNrmrn9A0WYWcqeMx16aleuCmKbIJjtUptyRxDbFbQ2zZgxY7FzrirRul4PUwOJ0qtrgucBhwLVwDDgcTMb55yr3W0j524DbgOoqqpy1dXVSTx9chYuXEh3+3s9tIqH3lzOlGNPoLQgP2XPmQ49tStXBbFNEMx2qU25I4jtCmKbupPMYeq1wPC4+WHAugRl/uqca3bOvQmswAvnrNAxJKYOVYuISBZKJoyfAw41s1FmFgHOA+Z2KfMXYAaAmZUDhwGrUlnR/aEhMUVEJJv1GsbOuRbgSmA+sBy4zzm31MyuN7Mz/GLzgc1mtgxYAHzNObe5ryq9t8pLIoBuFiEiItkpme+Mcc7NA+Z1WXZt3LQDrvIfWadC41OLiEgWC/wIXOANiRkOmQ5Ti4hIVuoXYRwKGYOKI+oZi4hIVuoXYQzeoWr1jEVEJBv1mzAuL9HNIkREJDv1nzCORXQ2tYiIZKV+E8btd27qbfhPERGRdOs/YRyL0tTaxvaGlkxXRUREZDf9JozLda2xiIhkqX4Txu3jU+uMahERyTb9JozVMxYRkWzVb8JYd24SEZFs1W/CeEBhvjckpnrGIiKSZfpNGIdCxuDiCJt26FpjERHJLv0mjMG/1lg9YxERyTL9KozLYxoSU0REsk//C2OdwCUiIlmmX4VxRUmUTXVNGhJTRESySr8K4/JYxBsSc5eGxBQRkezRr8K4YxQufW8sIiJZpH+FcUxDYoqISPbpV2FcXqIhMUVEJPv0rzDW+NQiIpKF+lUYDyjMJy9kOkwtIiJZpV+FcShkDI5F1DMWEZGs0q/CGPwhMdUzFhGRLNLvwtgbElM3ixARkezRT8NYPWMREcke/S6MvSExGzUkpoiIZI1+F8blsSjNrY5tu5ozXRURERGgX4ZxBNC1xiIikj36XRi3j0+9QWdUi4hIluh/YdwxCpfOqBYRkeyQVBib2UwzW2FmK81sToL1l5nZRjN70X9cnvqqpkbHkJjqGYuISJbI662AmYWBW4EPAGuB58xsrnNuWZei9zrnruyDOqZUWWE++WHTbRRFRCRrJNMzngqsdM6tcs41ATXAmX1brb4TChmDi6PqGYuISNaw3q63NbNzgJnOucv9+YuBY+J7wWZ2GfB9YCPwGvBl59zbCfY1G5gNUFlZOaWmpiZFzYC6ujpisVhSZa97chelUeOqKQUpe/6+sjftyhVBbBMEs11qU+4IYruC1qYZM2Ysds5VJVrX62FqwBIs65rgfwPucc41mtkVwB+Ak/bYyLnbgNsAqqqqXHV1dRJPn5yFCxeS7P5GvfksG+saqa5+f8qev6/sTbtyRRDbBMFsl9qUO4LYriC2qTvJHKZeCwyPmx8GrIsv4Jzb7JxrP+57OzAlNdXrG+WxKJt26GxqERHJDsmE8XPAoWY2yswiwHnA3PgCZkvnutkAACAASURBVHZg3OwZwPLUVTH12ofEbGvTkJgiIpJ5vR6mds61mNmVwHwgDPzOObfUzK4HFjnn5gJfMLMzgBZgC3BZH9Z5v5XHorS0eUNiDiyOZLo6IiLSzyXznTHOuXnAvC7Lro2b/ibwzdRWre+Ul7QP/NGoMBYRkYzrdyNwQecoXBt1eZOIiGSB/hnGJV5vWAN/iIhINuiXYVyu8alFRCSL9Msw7hgSU4epRUQkC/TLMDYz71pjHaYWEZEs0C/DGFAYi4hI1ui3YVxREtVhahERyQr9NozLYxH1jEVEJCv02zD2hsRs0pCYIiKScf02jMtjUVrbHLW7mjNdFRER6ef6dRgDOlQtIiIZ12/DuKJEQ2KKiEh26LdhrJ6xiIhki34bxrpZhIiIZIt+G8alhXlEwiHdLEJERDKu34axNyRmhE07dLMIERHJrH4bxgDlJRoSU0REMq9fh3FFTENiiohI5vXrMNbNIkREJBv06zCuKImyeaeGxBQRkczq12FcHovQ2ubYWq+TuEREJHP6dxiXtA/8oTAWEZHM6ddhrIE/REQkG/TrMO7sGSuMRUQkc/p3GGt8ahERyQL9OoxLC/KI5IV0mFpERDKqX4exmXkDf6hnLCIiGdSvwxi8y5t0NrWIiGRSvw/jihINiSkiIpnV78NYQ2KKiEimKYxjUbbsbKJVQ2KKiEiG9PswriiJakhMERHJqKTC2MxmmtkKM1tpZnN6KHeOmTkzq0pdFfuWrjUWEZFM6zWMzSwM3ArMAsYA55vZmATlSoAvAM+kupJ9qaJEQ2KKiEhmJdMzngqsdM6tcs41ATXAmQnKfQe4EWhIYf36XHksAqhnLCIimWPO9XzikpmdA8x0zl3uz18MHOOcuzKuzCTgGufcR81sIfBV59yiBPuaDcwGqKysnFJTU5OyhtTV1RGLxfZ6u/pmx2f/Vc+5h0eYNSo/ZfVJlX1tVzYLYpsgmO1Sm3JHENsVtDbNmDFjsXMu4de4eUlsbwmWdSS4mYWAnwCX9bYj59xtwG0AVVVVrrq6OomnT87ChQvZl/0554g++jBllcOorj4yZfVJlX1tVzYLYpsgmO1Sm3JHENsVxDZ1J5nD1GuB4XHzw4B1cfMlwDhgoZmtBqYBc3PlJC4z86411nfGIiKSIcmE8XPAoWY2yswiwHnA3PaVzrltzrly59xI59xI4GngjESHqbNVeYnGpxYRkczpNYydcy3AlcB8YDlwn3NuqZldb2Zn9HUF06EipiExRUQkc5L5zhjn3DxgXpdl13ZTtnr/q5VeFSURXny7NtPVEBGRfqrfj8AF7UNiNmpITBERyQiFMd7AH20OtuzUkJgiIpJ+CmM0JKaIiGSWwhgNiSkiIpmlMEY9YxERySyFMRqfWkREMkthDMSieRTkh3SYWkREMkJhTNyQmHU6m1pERNJPYezzwlg9YxERST+Fsa+iRENiiohIZiiMfeoZi4hIpiiMfRWxCFt2NmlITBERSTuFsa99SMzNO9U7FhGR9FIY+zoG/tihM6pFRCS9FMa+8vYhMfW9sYiIpJnC2FfR0TNWGIuISHopjH3tPWOdUS0iIummMPYVR8IU5od1rbGIiKSdwthnZpSXRNQzFhGRtFMYx9H41CIikgkK4zgVMQ2JKSIi6acwjlNeoiExRUQk/RTGccpjUbbUN9HS2pbpqoiISD+iMI5TURLFOdiyU98bi4hI+iiM41TEIoBG4RIRkfRSGMfpGJ9aZ1SLiEgaKYzjVLSPT60zqkVEJI0UxnE6e8YKYxERSR+FcZziaJ6GxBQRkbRTGHdRoWuNRUQkzRTGXZTHND61iIikl8K4i4oSDYkpIiLplVQYm9lMM1thZivNbE6C9VeY2ctm9qKZPWFmY1Jf1fTQzSJERCTdeg1jMwsDtwKzgDHA+QnC9k/OufHOuYnAjcCPU17TNCmPRdla30SzhsQUEZE0SaZnPBVY6Zxb5ZxrAmqAM+MLOOe2x80WAy51VUwvDYkpIiLpZs71nJtmdg4w0zl3uT9/MXCMc+7KLuU+B1wFRICTnHOvJ9jXbGA2QGVl5ZSampqUNAKgrq6OWCy23/tZvL6Fn73QyLePK2BEaTgFNds/qWpXNglimyCY7VKbckcQ2xW0Ns2YMWOxc64q4UrnXI8P4GPAb+LmLwZ+1kP5C4A/9LbfKVOmuFRasGBBSvazaPVmN+IbD7oFr65Pyf72V6ralU2C2CbngtkutSl3BLFdQWsTsMh1k4nJHKZeCwyPmx8GrOuhfA3wkST2m5UqYgWAhsQUEZH0SSaMnwMONbNRZhYBzgPmxhcws0PjZk8D9jhEnSvKS7w7N+mMahERSZe83go451rM7EpgPhAGfuecW2pm1+N1uecCV5rZKUAzsBW4tC8r3ZeKInkURTQkpoiIpE+vYQzgnJsHzOuy7Nq46S+muF4ZpSExRUQknTQCVwLewB8KYxERSQ+FcQIVMQ2JKSIi6aMwTqC8RDeLEBGR9FEYJ+ANidmsITFFRCQtFMYJVJREAdisy5tERCQNAhPG4ZadKdtXecwLYx2qFhGRdAhGGL90L8c8cwVsfiMlu2sP440KYxERSYNghPH7jsGcg3svgsa6/d7dAf5hap1RLSIi6RCMMB44kmVjvgobX4W/fg56uRNVb3SYWkRE0ikYYQxsHTQRTrkOlv0F/vPT/dpXYSRMsYbEFBGRNAlMGANw3Bdg7Fnwr2/Dyn/t1668ITF1NrWIiPS9YIWxGZx5K1QcCfd/Era8uc+7Ko9F2aSesYiIpEGwwhggUgzn3eVN33sRNO3bJU/lsajOphYRkbQIXhgDDDoYzvktrF8Kcz+/Tyd06c5NIiKSLsEMY4BDToGTr4VX/gxP/myvNy+PRamtb6apRUNiiohI3wpuGAOc8GUYcyb88//BGwv2atOOITF3qncsIiJ9K9hhbAZn/gLKD4f7PwFbVye9aXksAsCmHTqjWkRE+lawwxggGoPz7gbX5p/QVZ/UZuUlGvhDRETSI/hhDDB4NJz9G3jvFfjbF5M6oasipiExRUQkPfpHGAMcdiqc9N/w8n3w9C96LV5REiU/bDyyfD1uP4fXFBER6Un/CWOAE74CR3wY/vEtWPVoj0UL8sN8/YNH8Miy9dy6YGWaKigiIv1R/wrjUAjO+hUMPsQ7oav2rR6LX/7+UZw16SBu+sdrPLJsfZoqKSIi/U3/CmOAaAmc9ydobfZO6Gre1W1RM+P7Z49nwrAyvnzvi7y+fkcaKyoiIv1F/wtjgPJD4Ozb4d2X4G9f6vGEroL8ML++eAoF+WEuv3MRtfW61ElERFKrf4YxwOEzofpqWFIDz/y6x6IHlhXy64sn825tA5+/5wVaWjUql4iIpE7/DWOAE78Gh58G86+G1U/0WHTKiEF89yPjePz1Tfzg4VfTVEEREekP+ncYt5/QNehguO9S2La2x+IfP3o4lx03ktsff5M/L+65rIiISLL6dxgDFJR6J3S1NPZ6QhfAf592JMcePJhvPvAyL75dm6ZKiohIkCmMASoOg7N/DetegAev6vGErvxwiFsvnMwBJVH+64+L2LC9IY0VFRGRIFIYtzviNJj+DXjpT/Dcb3osOqg4wm8urWJHQwv/dddiGlta01RJEREJIoVxvOlz4LCZ8PAcWPNkj0WPGFLKjz52FC+8Vcs1D7yiITNFRGSfKYzjhUJw9m0wcCTcdwlse6fH4rPGH8gXTj6U/128lt8/uTotVRQRkeBJKozNbKaZrTCzlWY2J8H6q8xsmZktMbN/mdmI1Fc1TQrK4Ny7vRO5krjl4pdOPpRTx1Ty3b8v5z8rN6WpkiIiEiS9hrGZhYFbgVnAGOB8MxvTpdgLQJVzbgJwP3BjqiuaVgcc4V3ytO55uPFg+ONZ8MRP4J3noW3374dDIePH505kdEUxn/vT87y1Obn7JYuIiLRLpmc8FVjpnFvlnGsCaoAz4ws45xY459pT6GlgWGqrmQFHng6X/g2mXArb34V/Xge3z4AbR0HNhfDMbbBxBThHLJrH7ZdU4Rx8+s5F1DW2ZLr2IiKSQ/KSKHMQ8Hbc/FrgmB7Kfwp4aH8qlTVGneg9AHash9WPw6qF8Oaj8OqD3vLYEBh1IiMOns5vzhjPufet5Sv3vcgvL5xCKGQZq7qIiOQO6+0sYDP7GPBB59zl/vzFwFTn3OcTlL0IuBKY7pxrTLB+NjAboLKyckpNTc3+t8BXV1dHLBZL2f56U7BrPQO3vsSA2iUM3LqESPM2ADbnDWF+wxiaD5jAIUdMpjlStl/Pk+52pUMQ2wTBbJfalDuC2K6gtWnGjBmLnXNVidYlE8bHAtc55z7oz38TwDn3/S7lTgF+hhfEG3qrVFVVlVu0aFFyLUjCwoULqa6uTtn+9opzsGE5vPko7s1HaXj9MQrbdnrrKsf5PezpMOI4b8SvvZDRdvWRILYJgtkutSl3BLFdQWuTmXUbxskcpn4OONTMRgHvAOcBF3R5gknAr4GZyQRx4JhB5RioHINN+wzW2MjXfnkXQ7c8y+y8tRQv+h08/QuwMBw0GQ6e4V3PPHSSdzmViIj0a72GsXOuxcyuBOYDYeB3zrmlZnY9sMg5Nxf4IRAD/tfMAN5yzp3Rh/XOagXRKF/95IWc/rMR/N/WEHM/X8XALS/Cqke975sfvwkeuxGKD4DDToXDZsHoGRApznTVRUQkA5LpGeOcmwfM67Ls2rjpU1Jcr5xXWVrAry+ewrm/fprP3beMOz95AnmjTgS+BfVb4PVH4LWHYdnf4IW7IByFUe/3esyHfRAGvC/TTRARkTTRMdI+NOl9A/mfs8fz5Bub+d685Z0rigbBUefCx+6Ar7/hXUJ19OWwZRXM+yrcPB5+cRz863pKt63Y49pmEREJlqR6xrLvzpkyjOXvbue3T7zJkQeW8vGq4bsXCOd3XkI1839g0+uw4iF4bT48cTOTXSusuNHrLR/2QRh9EkRLMtMYERHpEwrjNPjmrCNY8d4OrnngFUZXxJgyYmD3hcsP9R7HfwHqt7Dsbz9nTHiNd13zi3dDKB9GngCHz/LCeeDItLVDRET6hg5Tp0FeOMTPL5jEkLICrrhrMe9tS/IeyEWD2FB5IpzzW/jaKrjs7zDtCti2Fh76Ovz0KLh1Gjzy/2D1f6Bhe982RERE+oR6xmkyoMi7B/JZt/6HC25/mu9+ZBzHHVKe/A7CeV6PeOQJcOp3YfMb3glgKx6Cp34O/7nZK1c6zBtbu8J/HHAklB+219c3i4hI+iiM0+iwyhJ+c+nRfP3PL3HBb55h1rghXP2hIxk+qGjvdzZ4NBz7Oe+xq9a7//LG5bDhVe/n6iegJa4HniikKw7X988iIllAYZxmx44ezCNfns7tj63i1oUr+ferG/hM9WiumD6agvzwvu20cAAc8SHv0a6tFbauho2veo9kQvqAI/2wVkiLiKSTwjgDCvLDfP7kQ/nolGH8z7zl3PzP1/nfRWu55rQjmTluCP7AKfsnFPZ6z4NHwxGndS6PD+kNy707T21cDm8+Dq1xw4mXDfdCuXCQd8Z3KM977DGd7z1Xx3Sed0g9lN/tdgO3vAyvt0Bb+6PZq1db3LLWBMvaWv2y8fNx6y0c9zzhzudO+Ajv2Z7dtvHbFS3xLkUrHOTd6zoV742IpEZLEzTVeY9wxPsbzSvIyb9ThXEGDR1QyM8vmMxF0zZz3dylfObu5zlu9GCuO2Msh1X2Uc806ZB+1QvqzSuhtaUzBHebbgZ6Hts8kaMAluxrAyxxgFoYXGvikHZt+/pkuwvleaFcNAiKBneGdNFgKBpM5Xsb4LVGb75woPcziAHuHLQ0QssuaG7wjrS0NEDzrl5+NnjbtLV65zBES73Xp6DMny7tnI7E0jNUbFsbNO+Exjpo3AFNO7yfjTu8ZThvZLxIsVenrtP5Rdnz/rq2zg+xrrXz78C17b6s42ei5W27zzvn/Y1ZyPsbC4XipuOWW8hf1z4d7jJtndsANO30X++6zte6Y7p9eR2HrXkNNv6+2/W7dSDatYdyr48BiefzC9L6trXr9UYRfSVQN4pIgZbWNv707Fv86B+vUdfYwsXTRvDlUw7jhWf/k93tav+Db23u7OF2THcNb2/6hecXM2nK1AS90956s3n79g+6ra1LDztBYCda39rs/eHv2gL1m+MeW/zH5s51bd3cwzqU1xnM7SEdLenl+dtfw651bd59vuvRAzP/n2P7P8X4+USPXtaHQtRtryUWDXcGaXv47sOHMKBz3929Xp0F/cAu2z2ku5uOlnr/mBvroHH7nv/oG7d3zO/cup7icFtnmX1tS3s9OwK6h9COn8+Leu9dS6NX5xb/0doUN93YZbqp97L71Y4sFInRSD7RknKIxrzXLlriPSIxf1n7fLH399GwzTuHpmFbN49a77XrSTjaGdCxSvjE31PWpP29UYSkQV44xCXHjuTDE4byo3+s4A9PrWbuS+s4cyS8v80RztZ7I4f8T7t50aQ32baqAYYf3YeV6iIUglAEiPTN/p2Dxu08vWAe08Yftntw7xbkW7yz4Jvqev7gEc6HSFEP67s57A5+b6fN+9ntw/Wyvq2j17SruYDYkOFebyGv0Huf8wu9Q4E9/ox65du3a/8Zzvc+BDQ3eP8cG7d7l+Q11MZNb0s8vW0tbNjmTTdu7/2Ih4U7/3m3/wMvGMDO4gKKh43q7IFHS7x/7HvMlwAGzfVeT66pzv/ZdTrBul1bvfrGr+s2BMx7vcJR/3WLer27juko5EW8+nRdllfQUfbNt9YyatTozh5qe++0/UPsbsvaf/ay3CyuB922e2864fL4nnVb4m2cf7QhPlSjJX6w+tP5xRAK8VRfdLLaf/e6hnSi8E7jUQ+FcZYZVBzhe2eN5/yp7+Pbf1vKHUu3sujW/3DdGWOYMmJQpqsniZhBQRkNhQfCsCmZrk1KLe2rI075Bd6jpHLftnfOC7mOwN7hhVR8+Hbz3eGyhQs5IBNHm1qavEPizQ1+gPphGspLyT/9NQsXMmp69f7XM+j293evjyiMs9S4g8q477+O5Qf3/Iu/rG7ko798irMnHcScWUdwQGlmvtMQyRpmnaFbdlCma5OcPD+ACzNdEclGGoEri5kZ04bm8a+vTOez1aN5cMm7zLhpIb969A0aW3TzCBGRoFAY54DiaB5fn3kE//jyiRw7ejA3PPQqM29+nAUrNmS6aiIikgIK4xwysryY31x6NHd84mgM+MQdz/Gp3z/H6k07M101ERHZDwrjHDTj8AN4+EsncvWHjuDpVZs59SePccNDr7KpLsE1dyIikvUUxjkqkhdi9omjWfDVak4/aii/evQNjvv+v/lSzQs8/9ZWMnX9uIiI7D2dTZ3jDigt4EcfP4rPzhjNH59aw/2L1/KXF9cx/qAyLjl2BKcfNXTfx7wWEZG0UM84IEZXxLjujLE8ffXJfOcj42hobuVr9y/h2O//ixseepW3t9RnuooiItIN9YwDJhbN4+JpI7jomPfx9Kot3PnUam5/fBW3PfYGJx1RyaXHjeCEQ8pTczMKERFJCYVxQJkZx44ezLGjB7Oudhd/euYt7nn2Lf65fD0HVxRzybQRfHTKMEoK8jNdVRGRfk+HqfuBoQMK+eoHD+fJb57EzedOpKwwn+v+toxp//MvvvWXV3h9/Y5MV1FEpF9Tz7gfieaF+cikg/jIpINYsraWO59aw72L3uaPT6/h2IMHc+lxIzjlyErywvqMJiKSTgrjfmrCsAHc9LEBXP2hI7n3ube56+k1XHHX8xxYVsBF00Zw7tHDKY8lfycmERHZdwrjfm5QcYTPVI9m9okH8+9XN3DnU6v54fwV/PSfr3PahAM5dUwl4w4qY9jAQp30JSLSRxTGAkA4ZHxgTCUfGFPJyg113PW0d83yAy+8A0BpQR7jDipj7NBS/2cZo8qLs/c+yyIiOURhLHs45ADvmuU5s47gtfU7eOWd7byybhtL39nGH55aQ1OLd1P3wvwwY4aWMm5oKWMPKmPc0DIOrYyRr++cRUT2isJYulWQH2bCsAFMGDagY1lzaxtvbKzzAvqdbSxbt537F6/lD0+tASASDnH4kBLGHVTKmKFljBtaypEHlmoUMBGRHiiMZa/kh0McMaSUI4aUcs6UYQC0tTlWb97JK+u2s3TdNpa+s52HXnmPe559G/AOgR9SEWOs34Nu2NTKsA07qCwt0HXOIiIojCUFQiHj4IoYB1fEOOOooQA453indhdL121n6TvbeGXddp5YuYn/87+D/uGixwAojoSpLC2gsrSAIWX+z9Kot6ysgCGlBVSURHXoW0QCTWEsfcLMGDawiGEDi/jg2CEdyzfsaOD/HvkPBx58BO9ta2D99kbWb2/gve0NPPvmFjbsaKC51XXZFwwujjKkLMoQP7i90O4M7CGlBZQW5umMbxHJSUmFsZnNBH4KhIHfOOdu6LL+ROBmYAJwnnPu/lRXVILhgJICjhgUpnriQQnXt7U5ttY38d72Bi+kt3lh3R7Ya7fuYvGarWytb95j24L8kBfUJV5IV5Z09rArS6IdPW99fy0i2abXMDazMHAr8AFgLfCcmc11zi2LK/YWcBnw1b6opPQfoZAxOBZlcCzK2KFl3ZZraG5l445G3tve4Pew2x/espfX1vLI9gYamtv22LasMJ/K9kPhpQVUlno97gPae9ulBZTHIhqJTETSJpme8VRgpXNuFYCZ1QBnAh1h7Jxb7a/b8z+fSB8oyA8zfFARwwcVdVvGOcf2hhY2+L3q9kPinT3tRlZu2MSGHY20tu1+aDxkUB6LMnRAofc8A9t/FjF8UCEHlhUSyVNYi0hqJBPGBwFvx82vBY7pm+qIpI6ZUVaYT1lhPodWlnRbrrXNsXlnIxu2N3q97B0NrN/mBfi62gaWrK3loZffpSUusEMGQ0oLGBYX0DvXN1P05haGDyqksqSAkAZEEZEkmXOu5wJmHwM+6Jy73J+/GJjqnPt8grK/Bx7s7jtjM5sNzAaorKycUlNTs3+1j1NXV0csFkvZ/rJFENuVi21qc46tDY6Nuxwb69vYtMub3rSrjY31jtpGR/xfUp7B4EKjojBEeZFRETc9MGqEzDDzbpsWsj0fBllxMlouvle9CWKbIJjtClqbZsyYsdg5V5VoXTI947XA8Lj5YcC6famIc+424DaAqqoqV11dvS+7SWjhwoWkcn/ZIojtCmKbGltaeWD+oxx46Hje3lLP21vrWbtlF29vreelzfUJTzjrjRmEzQiFjLAZ4ZB5Ye3P77Y8BHmhECHzfoZD1vHIi5vuOp8XChGKX2ZGONw5v/HdJsYfPpwBRfn+UYaI97MonwGF+RRFwlnxoWFvBPH3D4LZriC2qTvJhPFzwKFmNgp4BzgPuKBPayWSY6J5YYYUh5h+WEXC9XWNLV5Ib6lnY10jbW2ONucdIm9zjtY2R6tztLU5Wtu8nnhvyzvW+8u9ae/R0uZobWujxS/X0upoammj1S/X0uov7yjfRlsbtLS1de6j1bGzqYUHV73abbvzQtYRzu1fCQwo7JwuK4p0LosrU1KQR2F+ZoO8tc1RW9/E5p1NbK5rYvPORrbsbGJTXRNbdjayua6J1jZHqV/f0oJ8SgvzKS3Io6Qgn9JCf5k/HYvm6aQ/2We9hrFzrsXMrgTm413a9Dvn3FIzux5Y5Jyba2ZHAw8AA4HTzezbzrmxfVpzkRwSi+Zx5IHe0KC5ZMGCBRx93PuprW9i265m71Hf3DFdu6t5t+Wb65pYtXEntfVN7GhsoadvwULmvS4lBV7YxaJ5xPyg85bnJfjply3Io8TftiA/hJnR1ubYtquZzX6Qbt7ZHrReyLYH7lsb6ml8/BG21jfRlqB+ZjCgMJ/BsShhM3Y0NLOjoYUdjS29vl7FkbAf2F5AlxR44R0f6EXRPFpbvQ9Kza2O5tY2WlrbaGp1tPjLm/xlneu9n81tzl/eZV1bG80Nuzhg+ZMURcIUR/IoioQpirZP51EcDVMYt644mrfHfFEkTDQvlJEPSU0tbexqbqWhuZVdTa3UN7Wycmsr+Ss3efPNrTQ0tbKr2XvUN3WW3dXcSlv7m2ntP7yvguIWxc3Hreto6p7liyJh/vu0MX3X6DhJXWfsnJsHzOuy7Nq46efwDl+LSICYmReS0TyGDdy7bVvbHHUNLdTu6gzyWj/Idza2sKOhhbqOn17gbdnZxFub69nR2MKOhuaEl6Z1FQ4ZRZEw9U2te5wV366sMJ/BsQiDiyMcWBzi8JFDKC+OMKg44l1K5/8cVBxhYFF+wh5ue3u2NzR7j13+9C6v7vHLdvjT67c38PoGf/2u5oTh3y4/bOSHQ+SFvJ/54RB54fZp7yuF/LwQ+SEjL2wU5Id3W7dufQMF+SF2NHjPu7PRC6mdjS00tiR/oUvI8AI6GiaSF+oILgNC/kT7OQ3ez67hFrc8bp1BR/I1+mG6Ky5gW7p7cZ55ptu6FuSHKIrkUZAXIhy2jg9/8R8C28+LcnSuaz/Do6P8btt0rispyMuuMBYR2VvhkHmHpov2ffzx5ta2juBuD+/24O4M82Z2NrYSi+b54RphcHG0I3wHFkd2G07V+x5yfNrb45xjZ1Mr9Y0thEPmB6sXuHkh2+/eqNeuaQnXtbS2Ue/3Inc2tlDf/rO5lfrGVnY2tVC/x3wrTa1tOOc6wqrNn8YPNOc6w8372R5q8fOd27fnXWFplMJ8r6demJ9HYSREYX6YgvwwRZHO+deWL2Va1WS/bKhzfb7Xgw/SFQsKYxHJWvnhEAOKIgwoimS6Kvst/ihDuuWFQ5SGQ5Tm2I1ZCjatYOqoQZmuRlrobAMREZEMUxiLiIhkmMJYREQkwxTGIiIiGaYwFhERyTCFsYiISIYpjEVERDJMYSwiIpJhCmMREZEMUxiLiIhkmMJYREQkwxTGIiIiGaYwFhERvlm94QAAB0RJREFUyTBzPd39uy+f2GwjsCaFuywHNqVwf9kiiO0KYpsgmO1Sm3JHENsVtDaNcM5VJFqRsTBONTNb5JyrynQ9Ui2I7QpimyCY7VKbckcQ2xXENnVHh6lFREQyTGEsIiKSYUEK49syXYE+EsR2BbFNEMx2qU25I4jtCmKbEgrMd8YiIiK5Kkg9YxERkZyUc2FsZjPNbIWZrTSzOQnWR83sXn/9M2Y2Mv213DtmNtzMFpjZcjNbamZfTFCm2sy2mdmL/uPaTNR1b5jZajN72a/vogTrzcxu8d+rJWY2ORP1TJaZHR73+r9oZtvN7EtdyuTE+2RmvzOzDWb2StyyQWb2iJm97v8c2M22l/plXjezS9NX655106Yfmtmr/u/XA2Y2oJtte/xdzaRu2nWdmb0T93v2oW627fH/ZaZ006Z749qz2sxe7GbbrH2v9otzLmceQBh4AzgYiAAvAWO6lPks8Ct/+jzg3kzXO4l2HQhM9qdLgNcStKsaeDDTdd3Ldq0GyntY/yHgIcCAacAzma7zXrQtDLyHd91gzr1PwInAZOCVuGU3AnP86TnADxJsNwhY5f8c6E8PzHR7emjTqUCeP/2DRG3y1/X4u5qF7boO+Gov2/X6/zKb2tRl/Y+Aa3PtvdqfR671jKcCK51zq5xzTUANcGaXMmcCf/Cn7wdONjNLYx33mnPuXefc8/70DmA5cFBma5UWZwJ3Os/TwAAzOzDTlUrSycAbzrlUDlyTNs65x4AtXRbH/+38AfhIgk0/CDzinNvinNsKPALM7LOK7oVEbXLO/cM51+LPPg38//buLkSLKo7j+PePu9WWJaVkL1uZtVdCmSxS9nIRIRkh9EIqQqFCKEl1E154F910E2FKkCVBSET0thdWwgZBVBqJa0WREkGL26pBLlKIbf8uznliGGd2Z3fdZ+asvw8Mz8yZsw/n8J+ZM3Pm7HO6216wKSqJVRVVrpe1GKtO8Xr9GPB2WwtVs9Qa42uB3zLbg5zdaP2fJ56EJ4G5bSndORC71W8D9hXsvsPMBszsYzNb1NaCTY4De83sWzN7smB/lXg21WrKLxapxallvrsPQbhBBK4syJNyzNYTemKKjHesNtHm2P2+q+SVQqqxuhsYdvfDJftTjNW4UmuMi55w88PBq+RpJDObDbwHPOvuI7ndBwhdorcCrwAftrt8k3Cnuy8BVgBPmdk9uf1JxsrMLgBWAu8W7E4xThORasy2Av8Au0uyjHesNs2rwE3AYmCI0K2bl2SsgDWM/VScWqwqSa0xHgSuy2x3A0fL8phZBzCHyXXxtJWZdRIa4t3u/n5+v7uPuPupuL4H6DSzeW0u5oS4+9H4eQz4gNBtllUlnk20Ajjg7sP5HSnGKWO49Zogfh4ryJNczOIgsweBtR5fOuZVOFYbxd2H3X3U3f8FdlJc3hRj1QE8DLxTlie1WFWVWmP8DdBjZjfGp5PVQF8uTx/QGuH5KPBZ2QnYFPEdyRvAj+7+Ukmeq1rvvs1sKSF2f7SvlBNjZpeY2aWtdcJAmu9z2fqAx+Oo6tuBk61u0oYrvXNPLU452XPnCeCjgjyfAsvN7PLYNbo8pjWSmd0PbAFWuvtfJXmqHKuNkhtb8RDF5a1yvWya+4Cf3H2waGeKsaqs7hFkE10II3B/JowS3BrTniecbAAXEboPjwD7gYV1l7lCne4idB8dAg7G5QFgI7Ax5tkM/EAYEfk1sKzuco9Tp4WxrAOx3K1YZetkwI4Yy++A3rrLXaFeFxMa1zmZtOTiRLiZGALOEJ6gNhDGVvQDh+PnFTFvL/B65m/Xx/PrCLCu7rqMU6cjhPemrfOq9Z8W1wB7xjpWm7KU1OuteM4cIjSwV+frFbfPul42YSmqU0x/s3UuZfImE6upLPoFLhERkZql1k0tIiIy46gxFhERqZkaYxERkZqpMRYREamZGmMREZGaqTEWSZSZjeZmkTpns/KY2YLsjDoiMr066i6AiEza3+6+uO5CiMjU6clYZIaJ872+aGb743JzTL/BzPrj5AL9ZnZ9TJ8f5/odiMuy+FWzzGynhTm295pZV22VEpnh1BiLpKsr1029KrNvxN2XAtuBl2PadsKUlbcQJkzYFtO3AZ97mNxiCeGXjQB6gB3uvgj4E3hkmusjct7SL3CJJMrMTrn77IL0X4F73f2XOAHJ7+4+18xOEH428UxMH3L3eWZ2HOh299OZ71hAmLe4J25vATrd/YXpr5nI+UdPxiIzk5esl+UpcjqzPorGmIhMGzXGIjPTqsznV3H9S8LMPQBrgS/iej+wCcDMZpnZZe0qpIgEutMVSVeXmR3MbH/i7q1/b7rQzPYRbrjXxLSngV1m9hxwHFgX058BXjOzDYQn4E2EGXVEpE30zlhkhonvjHvd/UTdZRGRatRNLSIiUjM9GYuIiNRMT8YiIiI1U2MsIiJSMzXGIiIiNVNjLCIiUjM1xiIiIjVTYywiIlKz/wDbChhLYCXv3gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer_top20.plot_perf()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "588c2f42be434636a1fc25a32493ee9b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=237.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "acc_df = trainer_top20.get_test_accuracy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Top-1</th>\n",
       "      <th>Top-5</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Average (Micro)</th>\n",
       "      <td>88.05</td>\n",
       "      <td>95.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Backpacks</th>\n",
       "      <td>98.53</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Belts</th>\n",
       "      <td>99.07</td>\n",
       "      <td>99.07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Briefs</th>\n",
       "      <td>96.26</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Casual Shoes</th>\n",
       "      <td>81.62</td>\n",
       "      <td>99.77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Flip Flops</th>\n",
       "      <td>87.25</td>\n",
       "      <td>99.51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Formal Shoes</th>\n",
       "      <td>93.88</td>\n",
       "      <td>99.28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Handbags</th>\n",
       "      <td>96.15</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Heels</th>\n",
       "      <td>90.15</td>\n",
       "      <td>99.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Jeans</th>\n",
       "      <td>99.26</td>\n",
       "      <td>99.63</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Kurtas</th>\n",
       "      <td>94.46</td>\n",
       "      <td>99.86</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Perfume and Body Mist</th>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sandals</th>\n",
       "      <td>85.00</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Shirts</th>\n",
       "      <td>99.04</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Socks</th>\n",
       "      <td>96.69</td>\n",
       "      <td>98.90</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sports Shoes</th>\n",
       "      <td>86.84</td>\n",
       "      <td>99.48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sunglasses</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Tops</th>\n",
       "      <td>76.81</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Tshirts</th>\n",
       "      <td>92.25</td>\n",
       "      <td>99.91</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Wallets</th>\n",
       "      <td>97.02</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Watches</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        Top-1   Top-5\n",
       "Average (Micro)         88.05   95.78\n",
       "Backpacks               98.53  100.00\n",
       "Belts                   99.07   99.07\n",
       "Briefs                  96.26  100.00\n",
       "Casual Shoes            81.62   99.77\n",
       "Flip Flops              87.25   99.51\n",
       "Formal Shoes            93.88   99.28\n",
       "Handbags                96.15  100.00\n",
       "Heels                   90.15   99.82\n",
       "Jeans                   99.26   99.63\n",
       "Kurtas                  94.46   99.86\n",
       "Perfume and Body Mist    0.00    0.00\n",
       "Sandals                 85.00  100.00\n",
       "Shirts                  99.04  100.00\n",
       "Socks                   96.69   98.90\n",
       "Sports Shoes            86.84   99.48\n",
       "Sunglasses             100.00  100.00\n",
       "Tops                    76.81  100.00\n",
       "Tshirts                 92.25   99.91\n",
       "Wallets                 97.02  100.00\n",
       "Watches                100.00  100.00"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "acc_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|                       |   Top-1 |   Top-5 |\n",
      "|:----------------------|--------:|--------:|\n",
      "| Average (Micro)       |   88.05 |   95.78 |\n",
      "| Backpacks             |   98.53 |  100    |\n",
      "| Belts                 |   99.07 |   99.07 |\n",
      "| Briefs                |   96.26 |  100    |\n",
      "| Casual Shoes          |   81.62 |   99.77 |\n",
      "| Flip Flops            |   87.25 |   99.51 |\n",
      "| Formal Shoes          |   93.88 |   99.28 |\n",
      "| Handbags              |   96.15 |  100    |\n",
      "| Heels                 |   90.15 |   99.82 |\n",
      "| Jeans                 |   99.26 |   99.63 |\n",
      "| Kurtas                |   94.46 |   99.86 |\n",
      "| Perfume and Body Mist |    0    |    0    |\n",
      "| Sandals               |   85    |  100    |\n",
      "| Shirts                |   99.04 |  100    |\n",
      "| Socks                 |   96.69 |   98.9  |\n",
      "| Sports Shoes          |   86.84 |   99.48 |\n",
      "| Sunglasses            |  100    |  100    |\n",
      "| Tops                  |   76.81 |  100    |\n",
      "| Tshirts               |   92.25 |   99.91 |\n",
      "| Wallets               |   97.02 |  100    |\n",
      "| Watches               |  100    |  100    |\n"
     ]
    }
   ],
   "source": [
    "print(acc_df.to_markdown())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Progressive resizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "img_sizes = [480, 1080, 1440]\n",
    "img_epochs = [10, 10, 5]\n",
    "acc_dfs = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer_top20.data_loaders['train'].batch_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Progressively resize dataset images\n",
    "\n",
    "torch.cuda.empty_cache()\n",
    "trainer_top20 = setup_progressive(trainer_top20, img_sizes[0], batch_size=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 0/9\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6f6f85c3cb844173880fdcbb94a84fa1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=14400.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "ename": "RuntimeError",
     "evalue": "CUDA out of memory. Tried to allocate 16.00 MiB (GPU 0; 11.17 GiB total capacity; 10.79 GiB already allocated; 8.31 MiB free; 10.88 GiB reserved in total by PyTorch)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-21-8ae109aae74e>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtrainer_top20\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg_epochs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m~/notebooks/git/fashion-product-classification/src/train.py\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(self, num_epochs)\u001b[0m\n\u001b[1;32m    210\u001b[0m             \u001b[0mlogger\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minfo\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Running epoch {}/{}\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mepoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_epochs\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    211\u001b[0m             \u001b[0mlogger\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minfo\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'-'\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m10\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 212\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrun_epoch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    213\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    214\u001b[0m         \u001b[0mtime_since\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0msince\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/notebooks/git/fashion-product-classification/src/train.py\u001b[0m in \u001b[0;36mrun_epoch\u001b[0;34m(self, do_val)\u001b[0m\n\u001b[1;32m    153\u001b[0m                 \u001b[0;32mwith\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_grad_enabled\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mphase\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'train'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    154\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcriterion\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 155\u001b[0;31m                         \u001b[0moutputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mloss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    156\u001b[0m                     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    157\u001b[0m                         \u001b[0;31m# TODO: Add support for multiple criterions\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    530\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    531\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 532\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    533\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    534\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/notebooks/git/fashion-product-classification/src/models/fashion_classifier.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x, y)\u001b[0m\n\u001b[1;32m     39\u001b[0m         \u001b[0;32mand\u001b[0m \u001b[0mloss\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0;32mif\u001b[0m \u001b[0my\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mprovided\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     40\u001b[0m         \"\"\"\n\u001b[0;32m---> 41\u001b[0;31m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresnet\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     42\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     43\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    530\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    531\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 532\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    533\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    534\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torchvision/models/resnet.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m    214\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    215\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 216\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_impl\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    217\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    218\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torchvision/models/resnet.py\u001b[0m in \u001b[0;36m_forward_impl\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m    202\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmaxpool\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    203\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 204\u001b[0;31m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayer1\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    205\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayer2\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    206\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayer3\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    530\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    531\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 532\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    533\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    534\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/container.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m     98\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     99\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mmodule\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 100\u001b[0;31m             \u001b[0minput\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodule\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    101\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    102\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    530\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    531\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 532\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    533\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    534\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torchvision/models/resnet.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m    110\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    111\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownsample\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 112\u001b[0;31m             \u001b[0midentity\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownsample\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    113\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    114\u001b[0m         \u001b[0mout\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0midentity\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    530\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    531\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 532\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    533\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    534\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/container.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m     98\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     99\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mmodule\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 100\u001b[0;31m             \u001b[0minput\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodule\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    101\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    102\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    530\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    531\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 532\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    533\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    534\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/modules/batchnorm.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m    105\u001b[0m             \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrunning_mean\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrunning_var\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mweight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    106\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtraining\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrack_running_stats\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 107\u001b[0;31m             exponential_average_factor, self.eps)\n\u001b[0m\u001b[1;32m    108\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    109\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda/envs/py37_pytorch/lib/python3.7/site-packages/torch/nn/functional.py\u001b[0m in \u001b[0;36mbatch_norm\u001b[0;34m(input, running_mean, running_var, weight, bias, training, momentum, eps)\u001b[0m\n\u001b[1;32m   1668\u001b[0m     return torch.batch_norm(\n\u001b[1;32m   1669\u001b[0m         \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbias\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrunning_mean\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrunning_var\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1670\u001b[0;31m         \u001b[0mtraining\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmomentum\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0meps\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackends\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcudnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0menabled\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1671\u001b[0m     )\n\u001b[1;32m   1672\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mRuntimeError\u001b[0m: CUDA out of memory. Tried to allocate 16.00 MiB (GPU 0; 11.17 GiB total capacity; 10.79 GiB already allocated; 8.31 MiB free; 10.88 GiB reserved in total by PyTorch)"
     ]
    }
   ],
   "source": [
    "trainer_top20.train(img_epochs[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = trainer_top20.get_test_accuracy()\n",
    "acc_dfs.append(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(df.to_markdown())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Progressively resize dataset images\n",
    "\n",
    "torch.cuda.empty_cache()\n",
    "trainer_top20 = setup_progressive(trainer_top20, img_sizes[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer_top20.train(img_epochs[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = trainer_top20.get_test_accuracy()\n",
    "acc_dfs.append(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(df.to_markdown())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Progressively resize dataset images\n",
    "\n",
    "torch.cuda.empty_cache()\n",
    "trainer_top20 = setup_progressive(trainer_top20, img_sizes[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer_top20.train(img_epochs[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer_top20.plot_perf()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "acc_df = trainer_top20.get_test_accuracy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "acc_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(acc_df.to_markdown())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train on fine-tune data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Creating datasets\n",
      "INFO:fashion:Created train dataset with 4629 samples\n",
      "INFO:fashion:Created val dataset with 1158 samples\n",
      "INFO:fashion:Created test dataset with 5505 samples\n",
      "INFO:fashion:Creating dataloaders\n",
      "INFO:fashion:Creating model\n",
      "INFO:fashion:Creating trainer\n"
     ]
    }
   ],
   "source": [
    "processor, trainer_ft, dataloaders_ft = setup_ft(processor=processor, ckpt_path=CKPT_PATH_FT,\n",
    "                                                 model=trainer_top20.get_best_model(), mt=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 0/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "34cf6ce38de74e2bbb2afa6b9c3cb437",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 4.1167 Acc: 0.1951\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "54ab2d3e58d64a6fadd5ca24cac23fba",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 3.3175 Acc: 0.4983\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 1/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5562402013b04028bde71c73bb217288",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 2.4164 Acc: 0.5550\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "17834c5738a14cc2a9b20f9717c288e7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 2.3978 Acc: 0.6028\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 2/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "96050a70911e4847b7b7bd50fa462ea1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 1.6115 Acc: 0.6474\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fdc24f66d26749f48380ceeb69d4faa0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 2.1296 Acc: 0.7228\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 3/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ec718df5585e4df28468064c43fd8556",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 1.1739 Acc: 0.7202\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ce740de1aaf940dd9eb3fa6b9311452c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.9826 Acc: 0.7358\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 4/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2ae13eafabc1438fa0acc495d2814202",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.8823 Acc: 0.7531\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "091042c738d24852a941790f5a807bbf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.9004 Acc: 0.7642\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 5/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4f644ab583ee4ea18e19742a4c071343",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.7757 Acc: 0.7755\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c7906a22cc2642e8bb5b633fa6a8e76e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8658 Acc: 0.7789\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 6/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "374c5e278a1c4827962b8112a924a016",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.7363 Acc: 0.7959\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2e4f6288ddb2403b9a85a7227ad6b3f2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8651 Acc: 0.7832\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 7/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c372bfd74f804de682803d1c73134e97",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6942 Acc: 0.7905\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5bf69f7519f04094bd28186abe054232",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8376 Acc: 0.7867\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 8/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "eb10b1e2e6d7447ebcf1341a74643120",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6599 Acc: 0.7984\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "557f2e84342e4b5cb954c3e92281dab4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8325 Acc: 0.7902\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 9/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d649e328077c438d9980d2070050fd2a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6358 Acc: 0.8002\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "de46cc3f73614d81b0a1105a905548d0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8363 Acc: 0.7893\n",
      "INFO:fashion:Running epoch 10/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4a619819395f4873a572ade65c5d4f18",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6191 Acc: 0.8086\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "99714f35dde84076ab95f81f18adb891",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8276 Acc: 0.7971\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 11/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "408571ddb0db418eabe996240bc52543",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6223 Acc: 0.8077\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "54a6cfecad0142899b017e7b550bdf5e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8509 Acc: 0.7936\n",
      "INFO:fashion:Running epoch 12/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b32403db97e048b6a29490ffbfd2d329",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6303 Acc: 0.8067\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "03238a0a7a4e4955af27add0a8e3b768",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8469 Acc: 0.7902\n",
      "INFO:fashion:Running epoch 13/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1558f66a28b94cb5ba26ab4f6f867ec3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6058 Acc: 0.8097\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "09accdbfe7bd4d8cac02df5e14ea4e38",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8425 Acc: 0.7936\n",
      "INFO:fashion:Running epoch 14/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7178f1efbba14a7d950a35562c7160e1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6082 Acc: 0.8110\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "67c2ca198f8f45b88199a423a82927c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8467 Acc: 0.7997\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Running epoch 15/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5fb8c14a98244ec5828a71cb095a3028",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6080 Acc: 0.8062\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "98123def9b5e49b3a7fd82c153691360",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8223 Acc: 0.7953\n",
      "INFO:fashion:Running epoch 16/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4c331f3e9202427aa3a1dd7101e7029b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5948 Acc: 0.8127\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "68b0a8205c3d4c42a2a35926326d9eb2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8407 Acc: 0.7962\n",
      "INFO:fashion:Running epoch 17/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "71e897220dc94e6bab16b3f5094a271b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6134 Acc: 0.8062\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c0e76420936444ed907028dea108e6a8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8524 Acc: 0.7953\n",
      "INFO:fashion:Running epoch 18/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c90ee07e7d164477bfc47d8013e9154c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5891 Acc: 0.8149\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7cf7aaa5b6834de78c31ce338863224e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8388 Acc: 0.7945\n",
      "INFO:fashion:Running epoch 19/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3f7cbc4ab8248ab935259124fddf48e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6001 Acc: 0.8136\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "05d50adcd7bd4befb3d901c36fa066ca",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8256 Acc: 0.7979\n",
      "INFO:fashion:Running epoch 20/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fff0de6d924945d4b1084a6a777d02f3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6071 Acc: 0.8159\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9313d2eea64c46248cfe6437f7e7777e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8245 Acc: 0.7979\n",
      "INFO:fashion:Running epoch 21/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "02da2a78c6ee44a78ad7f436e932654a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5989 Acc: 0.8164\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4822425e1e2b44859486dbaf518bd352",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8300 Acc: 0.7962\n",
      "INFO:fashion:Running epoch 22/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dff8a9c5d48d435a85e34182178bfd20",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5913 Acc: 0.8090\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2328fb406b59471588f73650cc9a6382",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8271 Acc: 0.7927\n",
      "INFO:fashion:Running epoch 23/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ca2051e887524c4db29786daa4f96899",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5925 Acc: 0.8110\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f87def08bfba435ea1ab605394e9892d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8279 Acc: 0.7962\n",
      "INFO:fashion:Running epoch 24/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fa7c01f0fc5a4a97a392cf6469f588f6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5952 Acc: 0.8079\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1f440babbd444391a3e6f739cf9e70ac",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8319 Acc: 0.7953\n",
      "INFO:fashion:Running epoch 25/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f0c752ba7d814cc8b1f0317a360e9ba1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6040 Acc: 0.8168\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3c569c2e9c0e4b908fb48be5d1d1b74d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8392 Acc: 0.7979\n",
      "INFO:fashion:Running epoch 26/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a986935e26c444d89ef7aab7946e7cbd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.6233 Acc: 0.8101\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c40fcef0b7ba477c935f16e41809b73f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8398 Acc: 0.7979\n",
      "INFO:fashion:Running epoch 27/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8ddf282f9cae480ba33c00f66ee7ef6a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5990 Acc: 0.8123\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d4fed30099c9480ea8eb16422ef387d1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8493 Acc: 0.7945\n",
      "INFO:fashion:Running epoch 28/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "eabb5d7fe19a40fba97a770d8e1a216b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5900 Acc: 0.8144\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "13db0cc8112f4e768a25c6b39121806f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8344 Acc: 0.7953\n",
      "INFO:fashion:Running epoch 29/29\n",
      "INFO:fashion:----------\n",
      "INFO:fashion:Iterating through data for phase: train\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7c7211bdd39449a2adcf8cfecb44c802",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=73.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:train Loss: 0.5933 Acc: 0.8157\n",
      "INFO:fashion:Iterating through data for phase: val\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "05445fb69b6b45abbb9cf91b5d8459c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=19.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:val Loss: 1.8402 Acc: 0.8014\n",
      "INFO:fashion:--Saving checkpoint--\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:fashion:Training complete in 55m 38s\n",
      "INFO:fashion:Best val Acc: 0.801382\n",
      "INFO:fashion:Setting model to the best one found during training\n"
     ]
    }
   ],
   "source": [
    "trainer_ft.train(30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer_ft.plot_perf()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2514ed5f8d0d49d09c0ef1105e1cc4f5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=87.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "acc_df_ft = trainer_ft.get_test_accuracy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "acc_df_ft[acc_df_ft==-0] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|                           |   Top-1 |   Top-5 |\n",
      "|:--------------------------|--------:|--------:|\n",
      "| Average (Micro)           |   47.1  |   63.85 |\n",
      "| Accessory Gift Set        |  100    |  100    |\n",
      "| Baby Dolls                |    0    |    0    |\n",
      "| Bangle                    |   13.04 |   78.26 |\n",
      "| Basketballs               |   90.91 |  100    |\n",
      "| Bath Robe                 |    0    |    0    |\n",
      "| Beauty Accessory          |    0    |    0    |\n",
      "| Blazers                   |    0    |    0    |\n",
      "| Body Lotion               |    0    |    0    |\n",
      "| Body Wash and Scrub       |    0    |    0    |\n",
      "| Booties                   |    0    |    0    |\n",
      "| Boxers                    |    0    |    0    |\n",
      "| Bra                       |   96.96 |   99.7  |\n",
      "| Bracelet                  |   76.19 |  100    |\n",
      "| Camisoles                 |   41.18 |   82.35 |\n",
      "| Capris                    |   36.54 |   98.08 |\n",
      "| Caps                      |   90.08 |   96.95 |\n",
      "| Churidar                  |    0    |   75    |\n",
      "| Clothing Set              |    0    |    0    |\n",
      "| Clutches                  |   95.83 |  100    |\n",
      "| Compact                   |    0    |    0    |\n",
      "| Concealer                 |    0    |    0    |\n",
      "| Cufflinks                 |   96.67 |  100    |\n",
      "| Cushion Covers            |    0    |    0    |\n",
      "| Deodorant                 |    0    |    0    |\n",
      "| Dresses                   |   50.54 |   97.85 |\n",
      "| Duffel Bag                |   78.26 |   97.83 |\n",
      "| Dupatta                   |   12.5  |  100    |\n",
      "| Earrings                  |   90.12 |  100    |\n",
      "| Eye Cream                 |    0    |    0    |\n",
      "| Eyeshadow                 |    0    |    0    |\n",
      "| Face Moisturisers         |    0    |    0    |\n",
      "| Face Scrub and Exfoliator |    0    |    0    |\n",
      "| Face Serum and Gel        |    0    |    0    |\n",
      "| Face Wash and Cleanser    |    0    |    0    |\n",
      "| Flats                     |   92.41 |  100    |\n",
      "| Footballs                 |  100    |  100    |\n",
      "| Foundation and Primer     |    0    |    0    |\n",
      "| Fragrance Gift Set        |    0    |    0    |\n",
      "| Free Gifts                |    0    |   50    |\n",
      "| Gloves                    |  100    |  100    |\n",
      "| Hair Accessory            |    0    |    0    |\n",
      "| Hair Colour               |    0    |    0    |\n",
      "| Hat                       |    0    |    0    |\n",
      "| Headband                  |  100    |  100    |\n",
      "| Highlighter and Blush     |    0    |    0    |\n",
      "| Innerwear Vests           |  100    |  100    |\n",
      "| Ipad                      |    0    |    0    |\n",
      "| Jackets                   |   61.17 |   98.4  |\n",
      "| Jeggings                  |    0    |    0    |\n",
      "| Jewellery Set             |   38.46 |  100    |\n",
      "| Jumpsuit                  |   50    |  100    |\n",
      "| Kajal and Eyeliner        |    0    |    0    |\n",
      "| Key chain                 |    0    |    0    |\n",
      "| Kurta Sets                |   96.88 |  100    |\n",
      "| Kurtis                    |   78.41 |   98.86 |\n",
      "| Laptop Bag                |   65.71 |  100    |\n",
      "| Leggings                  |   37.5  |   81.25 |\n",
      "| Lehenga Choli             |    0    |    0    |\n",
      "| Lip Care                  |    0    |    0    |\n",
      "| Lip Gloss                 |    0    |    0    |\n",
      "| Lip Liner                 |    0    |    0    |\n",
      "| Lip Plumper               |    0    |    0    |\n",
      "| Lipstick                  |    0    |    0    |\n",
      "| Lounge Pants              |   36.84 |   78.95 |\n",
      "| Lounge Shorts             |   11.11 |   88.89 |\n",
      "| Lounge Tshirts            |    0    |    0    |\n",
      "| Makeup Remover            |  100    |  100    |\n",
      "| Mascara                   |    0    |    0    |\n",
      "| Mask and Peel             |    0    |    0    |\n",
      "| Mens Grooming Kit         |    0    |    0    |\n",
      "| Messenger Bag             |   66.67 |  100    |\n",
      "| Mobile Pouch              |   27.78 |  100    |\n",
      "| Mufflers                  |   27.59 |  100    |\n",
      "| Nail Essentials           |    0    |    0    |\n",
      "| Nail Polish               |    0    |    0    |\n",
      "| Necklace and Chains       |   97.5  |  100    |\n",
      "| Nehru Jackets             |    0    |    0    |\n",
      "| Night suits               |   38.57 |   75.71 |\n",
      "| Nightdress                |   46.25 |   84.38 |\n",
      "| Patiala                   |    0    |    0    |\n",
      "| Pendant                   |   54.32 |   98.77 |\n",
      "| Rain Jacket               |  100    |  100    |\n",
      "| Rain Trousers             |    0    |  100    |\n",
      "| Ring                      |   87.93 |  100    |\n",
      "| Robe                      |    0    |    0    |\n",
      "| Rompers                   |    0    |    0    |\n",
      "| Rucksacks                 |  100    |  100    |\n",
      "| Salwar                    |    0    |    0    |\n",
      "| Salwar and Dupatta        |    0    |    0    |\n",
      "| Sarees                    |    0    |    0    |\n",
      "| Scarves                   |   50    |   91.3  |\n",
      "| Shapewear                 |  100    |  100    |\n",
      "| Shoe Accessories          |    0    |   21.05 |\n",
      "| Shoe Laces                |    0    |    0    |\n",
      "| Shorts                    |   40.28 |   96.21 |\n",
      "| Shrug                     |    0    |  100    |\n",
      "| Skirts                    |   67.92 |   75.47 |\n",
      "| Sports Sandals            |   90    |  100    |\n",
      "| Stockings                 |   17.65 |  100    |\n",
      "| Stoles                    |   30    |   90    |\n",
      "| Sunscreen                 |    0    |    0    |\n",
      "| Suspenders                |    0    |    0    |\n",
      "| Sweaters                  |   54.51 |   97.25 |\n",
      "| Sweatshirts               |   80.93 |   98.44 |\n",
      "| Swimwear                  |   25    |   87.5  |\n",
      "| Tablet Sleeve             |    0    |    0    |\n",
      "| Ties                      |   98.8  |   98.8  |\n",
      "| Ties and Cufflinks        |    0    |    0    |\n",
      "| Tights                    |  100    |  100    |\n",
      "| Toner                     |    0    |    0    |\n",
      "| Track Pants               |   76.88 |   98.27 |\n",
      "| Tracksuits                |   95    |  100    |\n",
      "| Travel Accessory          |   30    |  100    |\n",
      "| Trolley Bag               |    0    |    0    |\n",
      "| Trousers                  |   93.1  |   98.47 |\n",
      "| Trunk                     |    0    |    0    |\n",
      "| Tunics                    |   42.22 |  100    |\n",
      "| Umbrellas                 |  100    |  100    |\n",
      "| Waist Pouch               |    0    |   84.62 |\n",
      "| Waistcoat                 |    0    |    0    |\n",
      "| Water Bottle              |  100    |  100    |\n",
      "| Wristbands                |  100    |  100    |\n"
     ]
    }
   ],
   "source": [
    "print(acc_df_ft.to_markdown())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Progressive resizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "img_sizes = [480, 1080, 1440]\n",
    "img_epochs = [10, 10, 5]\n",
    "acc_df_fts = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Progressively resize dataset images\n",
    "\n",
    "trainer_ft = setup_progressive(trainer_ft, img_sizes[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer_ft.train(img_epochs[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = trainer_top20.get_test_accuracy()\n",
    "df[df==-0] = 0\n",
    "acc_df_fts.append(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(df.to_markdown())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Progressively resize dataset images\n",
    "\n",
    "trainer_ft = setup_progressive(trainer_ft, img_sizes[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer_ft.train(img_epochs[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = trainer_top20.get_test_accuracy()\n",
    "df[df==-0] = 0\n",
    "acc_df_fts.append(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(df.to_markdown())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Progressively resize dataset images\n",
    "\n",
    "trainer_ft = setup_progressive(trainer_ft, img_sizes[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer_ft.train(img_epochs[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer_ft.plot_perf()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "acc_df_ft2 = trainer_ft.get_test_accuracy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "acc_df_ft2[acc_df_ft2==-0] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(acc_df_ft2.to_markdown())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py37_pytorch",
   "language": "python",
   "name": "conda-env-py37_pytorch-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}