LR_adaptive_Hint.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2019-08-24T13:09:14.577Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/data/yaliu/jupyterbooks/multi-KD/models/teacher/resnet20.py:36: UserWarning: nn.init.kaiming_normal is now deprecated in favor of nn.init.kaiming_normal_.\n",
      "  init.kaiming_normal(m.weight)\n",
      "WARNING:root:Setting up a new session...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "StudentNet:\n",
      "\n",
      "ResNet(\n",
      "  (conv1): Conv2d(3, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "  (bn1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "    (2): BasicBlock(\n",
      "      (conv1): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(16, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): LambdaLayer()\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "    (2): BasicBlock(\n",
      "      (conv1): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(32, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): LambdaLayer()\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "    (2): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (shortcut): Sequential()\n",
      "    )\n",
      "  )\n",
      "  (linear): Linear(in_features=64, out_features=10, bias=True)\n",
      ")\n",
      "Training adapter:\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/ipykernel_launcher.py:248: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/ipykernel_launcher.py:183: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/ipykernel_launcher.py:256: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/torch/nn/_reduction.py:15: UserWarning: reduction='elementwise_mean' is deprecated, please use reduction='mean' instead.\n",
      "  warnings.warn(\"reduction='elementwise_mean' is deprecated, please use reduction='mean' instead.\")\n",
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/ipykernel_launcher.py:154: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument.\n",
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/torch/nn/functional.py:1992: UserWarning: reduction: 'mean' divides the total loss by both the batch size and the support size.'batchmean' divides only by the batch size, and aligns with the KL div math definition.'mean' will be changed to behave the same as 'batchmean' in the next major release.\n",
      "  warnings.warn(\"reduction: 'mean' divides the total loss by both the batch size and the support size.\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch[0/80]adapter Loss: 1.6207\n",
      "epoch[1/80]adapter Loss: 1.3406\n",
      "epoch[2/80]adapter Loss: 1.1132\n",
      "epoch[3/80]adapter Loss: 0.9417\n",
      "epoch[4/80]adapter Loss: 0.8863\n",
      "epoch[5/80]adapter Loss: 1.0321\n",
      "epoch[6/80]adapter Loss: 0.7275\n",
      "epoch[7/80]adapter Loss: 0.6101\n",
      "epoch[8/80]adapter Loss: 0.7095\n",
      "epoch[9/80]adapter Loss: 0.8646\n",
      "epoch[10/80]adapter Loss: 0.7112\n",
      "epoch[11/80]adapter Loss: 0.6941\n",
      "epoch[12/80]adapter Loss: 0.8521\n",
      "epoch[13/80]adapter Loss: 0.6496\n",
      "epoch[14/80]adapter Loss: 0.5462\n",
      "epoch[15/80]adapter Loss: 0.6668\n",
      "epoch[16/80]adapter Loss: 0.7586\n",
      "epoch[17/80]adapter Loss: 0.4472\n",
      "epoch[18/80]adapter Loss: 0.5592\n",
      "epoch[19/80]adapter Loss: 0.5962\n",
      "epoch[20/80]adapter Loss: 0.4775\n",
      "epoch[21/80]adapter Loss: 0.6044\n",
      "epoch[22/80]adapter Loss: 0.5636\n",
      "epoch[23/80]adapter Loss: 0.4450\n",
      "epoch[24/80]adapter Loss: 0.7001\n",
      "epoch[25/80]adapter Loss: 0.5509\n",
      "epoch[26/80]adapter Loss: 0.4664\n",
      "epoch[27/80]adapter Loss: 0.5320\n",
      "epoch[28/80]adapter Loss: 0.6596\n",
      "epoch[29/80]adapter Loss: 0.5473\n",
      "epoch[30/80]adapter Loss: 0.6754\n",
      "epoch[31/80]adapter Loss: 0.4818\n",
      "epoch[32/80]adapter Loss: 0.6309\n",
      "epoch[33/80]adapter Loss: 0.4247\n",
      "epoch[34/80]adapter Loss: 0.4216\n",
      "epoch[35/80]adapter Loss: 0.3893\n",
      "epoch[36/80]adapter Loss: 0.6945\n",
      "epoch[37/80]adapter Loss: 0.4900\n",
      "epoch[38/80]adapter Loss: 0.6010\n",
      "epoch[39/80]adapter Loss: 0.5811\n",
      "epoch[40/80]adapter Loss: 0.6155\n",
      "epoch[41/80]adapter Loss: 0.4617\n",
      "epoch[42/80]adapter Loss: 0.6094\n",
      "epoch[43/80]adapter Loss: 0.8036\n",
      "epoch[44/80]adapter Loss: 0.5804\n",
      "epoch[45/80]adapter Loss: 0.4308\n",
      "epoch[46/80]adapter Loss: 0.4140\n",
      "epoch[47/80]adapter Loss: 0.6365\n",
      "epoch[48/80]adapter Loss: 0.5714\n",
      "epoch[49/80]adapter Loss: 0.6430\n",
      "epoch[50/80]adapter Loss: 0.5169\n",
      "epoch[51/80]adapter Loss: 0.6014\n",
      "epoch[52/80]adapter Loss: 0.5421\n",
      "epoch[53/80]adapter Loss: 0.5178\n",
      "epoch[54/80]adapter Loss: 0.5058\n",
      "epoch[55/80]adapter Loss: 0.6119\n",
      "epoch[56/80]adapter Loss: 0.7975\n",
      "epoch[57/80]adapter Loss: 0.6275\n",
      "epoch[58/80]adapter Loss: 0.5234\n",
      "epoch[59/80]adapter Loss: 0.4424\n",
      "epoch[60/80]adapter Loss: 0.7151\n",
      "epoch[61/80]adapter Loss: 0.4268\n",
      "epoch[62/80]adapter Loss: 0.5795\n",
      "epoch[63/80]adapter Loss: 0.6317\n",
      "epoch[64/80]adapter Loss: 0.4817\n",
      "epoch[65/80]adapter Loss: 0.4904\n",
      "epoch[66/80]adapter Loss: 0.5337\n",
      "epoch[67/80]adapter Loss: 0.4633\n",
      "epoch[68/80]adapter Loss: 0.6332\n",
      "epoch[69/80]adapter Loss: 0.3780\n",
      "epoch[70/80]adapter Loss: 0.4549\n",
      "epoch[71/80]adapter Loss: 0.3549\n",
      "epoch[72/80]adapter Loss: 0.4550\n",
      "epoch[73/80]adapter Loss: 0.4081\n",
      "epoch[74/80]adapter Loss: 0.4513\n",
      "epoch[75/80]adapter Loss: 0.4656\n",
      "epoch[76/80]adapter Loss: 0.4968\n",
      "epoch[77/80]adapter Loss: 0.3992\n",
      "epoch[78/80]adapter Loss: 0.4354\n",
      "epoch[79/80]adapter Loss: 0.4593\n",
      "--- adapter training cost 68.822 mins ---\n",
      "\n",
      "===> epoch: 1/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.135 (0.135)\tData 0.038 (0.038)\tLoss 0.4320 (0.4320)\tPrec@1 91.406 (91.406)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/ipykernel_launcher.py:315: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
      "/home/yaliu/Dev/anaconda3/envs/py37/lib/python3.7/site-packages/ipykernel_launcher.py:323: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[10/391]\tTime 0.153 (0.155)\tData 0.034 (0.040)\tLoss 0.4179 (0.4835)\tPrec@1 89.844 (87.500)\n",
      "[20/391]\tTime 0.158 (0.154)\tData 0.037 (0.040)\tLoss 0.4267 (0.4809)\tPrec@1 88.281 (87.426)\n",
      "[30/391]\tTime 0.143 (0.145)\tData 0.033 (0.038)\tLoss 0.4615 (0.4712)\tPrec@1 88.281 (87.450)\n",
      "[40/391]\tTime 0.153 (0.145)\tData 0.033 (0.038)\tLoss 0.4179 (0.4580)\tPrec@1 87.500 (87.919)\n",
      "[50/391]\tTime 0.133 (0.149)\tData 0.028 (0.039)\tLoss 0.4177 (0.4562)\tPrec@1 89.062 (88.006)\n",
      "[60/391]\tTime 0.140 (0.147)\tData 0.034 (0.037)\tLoss 0.5587 (0.4601)\tPrec@1 85.938 (87.935)\n",
      "[70/391]\tTime 0.168 (0.148)\tData 0.032 (0.038)\tLoss 0.5440 (0.4642)\tPrec@1 85.938 (87.709)\n",
      "[80/391]\tTime 0.147 (0.150)\tData 0.032 (0.038)\tLoss 0.4663 (0.4649)\tPrec@1 86.719 (87.703)\n",
      "[90/391]\tTime 0.185 (0.150)\tData 0.039 (0.038)\tLoss 0.4062 (0.4625)\tPrec@1 87.500 (87.740)\n",
      "[100/391]\tTime 0.151 (0.151)\tData 0.035 (0.038)\tLoss 0.4191 (0.4678)\tPrec@1 89.062 (87.515)\n",
      "[110/391]\tTime 0.212 (0.152)\tData 0.052 (0.038)\tLoss 0.5544 (0.4721)\tPrec@1 82.812 (87.331)\n",
      "[120/391]\tTime 0.136 (0.152)\tData 0.035 (0.038)\tLoss 0.6135 (0.4793)\tPrec@1 82.031 (87.119)\n",
      "[130/391]\tTime 0.131 (0.152)\tData 0.038 (0.038)\tLoss 0.5091 (0.4847)\tPrec@1 84.375 (87.005)\n",
      "[140/391]\tTime 0.193 (0.151)\tData 0.071 (0.038)\tLoss 0.5260 (0.4851)\tPrec@1 85.938 (86.968)\n",
      "[150/391]\tTime 0.154 (0.149)\tData 0.051 (0.038)\tLoss 0.4758 (0.4883)\tPrec@1 89.844 (86.802)\n",
      "[160/391]\tTime 0.135 (0.148)\tData 0.035 (0.038)\tLoss 0.4494 (0.4903)\tPrec@1 89.062 (86.787)\n",
      "[170/391]\tTime 0.150 (0.149)\tData 0.030 (0.038)\tLoss 0.4744 (0.4910)\tPrec@1 86.719 (86.696)\n",
      "[180/391]\tTime 0.199 (0.149)\tData 0.080 (0.038)\tLoss 0.4450 (0.4907)\tPrec@1 89.844 (86.697)\n",
      "[190/391]\tTime 0.129 (0.148)\tData 0.039 (0.038)\tLoss 0.5911 (0.4906)\tPrec@1 82.812 (86.706)\n",
      "[200/391]\tTime 0.153 (0.148)\tData 0.036 (0.038)\tLoss 0.5057 (0.4896)\tPrec@1 87.500 (86.727)\n",
      "[210/391]\tTime 0.161 (0.148)\tData 0.036 (0.038)\tLoss 0.4421 (0.4896)\tPrec@1 88.281 (86.715)\n",
      "[220/391]\tTime 0.156 (0.148)\tData 0.035 (0.038)\tLoss 0.3043 (0.4893)\tPrec@1 94.531 (86.754)\n",
      "[230/391]\tTime 0.150 (0.148)\tData 0.036 (0.038)\tLoss 0.5836 (0.4902)\tPrec@1 80.469 (86.746)\n",
      "[240/391]\tTime 0.131 (0.148)\tData 0.033 (0.038)\tLoss 0.5791 (0.4912)\tPrec@1 83.594 (86.735)\n",
      "[250/391]\tTime 0.156 (0.149)\tData 0.038 (0.038)\tLoss 0.6232 (0.4919)\tPrec@1 82.031 (86.744)\n",
      "[260/391]\tTime 0.199 (0.149)\tData 0.028 (0.038)\tLoss 0.5227 (0.4929)\tPrec@1 86.719 (86.707)\n",
      "[270/391]\tTime 0.202 (0.149)\tData 0.075 (0.038)\tLoss 0.5890 (0.4947)\tPrec@1 83.594 (86.647)\n",
      "[280/391]\tTime 0.153 (0.149)\tData 0.045 (0.038)\tLoss 0.4669 (0.4963)\tPrec@1 86.719 (86.616)\n",
      "[290/391]\tTime 0.135 (0.150)\tData 0.029 (0.039)\tLoss 0.4965 (0.4971)\tPrec@1 88.281 (86.611)\n",
      "[300/391]\tTime 0.130 (0.150)\tData 0.039 (0.039)\tLoss 0.4048 (0.4977)\tPrec@1 90.625 (86.594)\n",
      "[310/391]\tTime 0.187 (0.150)\tData 0.080 (0.039)\tLoss 0.3779 (0.4974)\tPrec@1 89.062 (86.626)\n",
      "[320/391]\tTime 0.160 (0.151)\tData 0.035 (0.039)\tLoss 0.4948 (0.4982)\tPrec@1 89.844 (86.617)\n",
      "[330/391]\tTime 0.156 (0.151)\tData 0.043 (0.039)\tLoss 0.4421 (0.5002)\tPrec@1 88.281 (86.542)\n",
      "[340/391]\tTime 0.145 (0.150)\tData 0.036 (0.039)\tLoss 0.4673 (0.4988)\tPrec@1 85.156 (86.568)\n",
      "[350/391]\tTime 0.159 (0.150)\tData 0.061 (0.039)\tLoss 0.5489 (0.4987)\tPrec@1 85.938 (86.556)\n",
      "[360/391]\tTime 0.154 (0.150)\tData 0.033 (0.039)\tLoss 0.4239 (0.4979)\tPrec@1 87.500 (86.574)\n",
      "[370/391]\tTime 0.191 (0.151)\tData 0.068 (0.039)\tLoss 0.4425 (0.4981)\tPrec@1 90.625 (86.584)\n",
      "[380/391]\tTime 0.144 (0.151)\tData 0.020 (0.039)\tLoss 0.4386 (0.4975)\tPrec@1 89.844 (86.606)\n",
      "[390/391]\tTime 0.131 (0.151)\tData 0.024 (0.039)\tLoss 0.4956 (0.4978)\tPrec@1 87.500 (86.566)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.039 (0.039)\tLoss 1.0137 (1.0137)\tPrec@1 76.562 (76.562)\n",
      "Test: [10/79]\tTime 0.034 (0.038)\tLoss 1.0205 (1.2793)\tPrec@1 76.562 (74.219)\n",
      "Test: [20/79]\tTime 0.038 (0.038)\tLoss 1.2753 (1.3438)\tPrec@1 75.000 (73.661)\n",
      "Test: [30/79]\tTime 0.036 (0.039)\tLoss 1.2246 (1.3408)\tPrec@1 73.438 (73.614)\n",
      "Test: [40/79]\tTime 0.038 (0.039)\tLoss 1.4252 (1.3148)\tPrec@1 69.531 (74.219)\n",
      "Test: [50/79]\tTime 0.043 (0.039)\tLoss 1.4711 (1.3197)\tPrec@1 69.531 (73.851)\n",
      "Test: [60/79]\tTime 0.048 (0.038)\tLoss 1.3068 (1.3063)\tPrec@1 78.906 (74.065)\n",
      "Test: [70/79]\tTime 0.029 (0.037)\tLoss 1.3965 (1.3135)\tPrec@1 68.750 (73.999)\n",
      " * Prec@1 74.160\n",
      "\n",
      "===> epoch: 2/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.118 (0.118)\tData 0.022 (0.022)\tLoss 0.5177 (0.5177)\tPrec@1 88.281 (88.281)\n",
      "[10/391]\tTime 0.157 (0.142)\tData 0.037 (0.041)\tLoss 0.4578 (0.4767)\tPrec@1 90.625 (87.784)\n",
      "[20/391]\tTime 0.129 (0.145)\tData 0.029 (0.038)\tLoss 0.5267 (0.4663)\tPrec@1 86.719 (88.281)\n",
      "[30/391]\tTime 0.150 (0.150)\tData 0.027 (0.040)\tLoss 0.4617 (0.4698)\tPrec@1 87.500 (87.979)\n",
      "[40/391]\tTime 0.121 (0.149)\tData 0.035 (0.042)\tLoss 0.5000 (0.4657)\tPrec@1 84.375 (87.862)\n",
      "[50/391]\tTime 0.118 (0.145)\tData 0.033 (0.041)\tLoss 0.5353 (0.4731)\tPrec@1 82.812 (87.515)\n",
      "[60/391]\tTime 0.140 (0.143)\tData 0.030 (0.040)\tLoss 0.4120 (0.4712)\tPrec@1 89.844 (87.513)\n",
      "[70/391]\tTime 0.123 (0.142)\tData 0.032 (0.039)\tLoss 0.5444 (0.4741)\tPrec@1 85.938 (87.313)\n",
      "[80/391]\tTime 0.141 (0.143)\tData 0.034 (0.039)\tLoss 0.5153 (0.4814)\tPrec@1 85.938 (87.153)\n",
      "[90/391]\tTime 0.167 (0.144)\tData 0.048 (0.039)\tLoss 0.5359 (0.4850)\tPrec@1 85.938 (86.959)\n",
      "[100/391]\tTime 0.178 (0.144)\tData 0.051 (0.040)\tLoss 0.5496 (0.4879)\tPrec@1 88.281 (86.928)\n",
      "[110/391]\tTime 0.178 (0.146)\tData 0.039 (0.040)\tLoss 0.5102 (0.4885)\tPrec@1 89.844 (86.930)\n",
      "[120/391]\tTime 0.158 (0.147)\tData 0.031 (0.040)\tLoss 0.4287 (0.4892)\tPrec@1 85.156 (86.835)\n",
      "[130/391]\tTime 0.171 (0.148)\tData 0.037 (0.040)\tLoss 0.5059 (0.4912)\tPrec@1 84.375 (86.808)\n",
      "[140/391]\tTime 0.135 (0.149)\tData 0.034 (0.040)\tLoss 0.5182 (0.4907)\tPrec@1 85.156 (86.769)\n",
      "[150/391]\tTime 0.172 (0.149)\tData 0.038 (0.039)\tLoss 0.5347 (0.4925)\tPrec@1 85.156 (86.677)\n",
      "[160/391]\tTime 0.156 (0.149)\tData 0.035 (0.039)\tLoss 0.4121 (0.4924)\tPrec@1 86.719 (86.665)\n",
      "[170/391]\tTime 0.152 (0.150)\tData 0.037 (0.040)\tLoss 0.6249 (0.4953)\tPrec@1 80.469 (86.563)\n",
      "[180/391]\tTime 0.152 (0.150)\tData 0.038 (0.040)\tLoss 0.4049 (0.4962)\tPrec@1 88.281 (86.490)\n",
      "[190/391]\tTime 0.132 (0.152)\tData 0.031 (0.040)\tLoss 0.4024 (0.4949)\tPrec@1 92.188 (86.576)\n",
      "[200/391]\tTime 0.165 (0.152)\tData 0.033 (0.040)\tLoss 0.5384 (0.4931)\tPrec@1 85.938 (86.610)\n",
      "[210/391]\tTime 0.163 (0.152)\tData 0.043 (0.040)\tLoss 0.4021 (0.4924)\tPrec@1 89.062 (86.634)\n",
      "[220/391]\tTime 0.132 (0.152)\tData 0.035 (0.040)\tLoss 0.3825 (0.4914)\tPrec@1 91.406 (86.662)\n",
      "[230/391]\tTime 0.096 (0.151)\tData 0.030 (0.040)\tLoss 0.5588 (0.4904)\tPrec@1 86.719 (86.688)\n",
      "[240/391]\tTime 0.190 (0.151)\tData 0.057 (0.040)\tLoss 0.5008 (0.4900)\tPrec@1 84.375 (86.690)\n",
      "[250/391]\tTime 0.140 (0.151)\tData 0.038 (0.040)\tLoss 0.4976 (0.4906)\tPrec@1 83.594 (86.644)\n",
      "[260/391]\tTime 0.139 (0.151)\tData 0.028 (0.040)\tLoss 0.4695 (0.4911)\tPrec@1 87.500 (86.638)\n",
      "[270/391]\tTime 0.152 (0.152)\tData 0.041 (0.040)\tLoss 0.4539 (0.4916)\tPrec@1 87.500 (86.592)\n",
      "[280/391]\tTime 0.156 (0.152)\tData 0.034 (0.040)\tLoss 0.4329 (0.4910)\tPrec@1 88.281 (86.619)\n",
      "[290/391]\tTime 0.143 (0.152)\tData 0.031 (0.040)\tLoss 0.4276 (0.4929)\tPrec@1 89.844 (86.603)\n",
      "[300/391]\tTime 0.162 (0.151)\tData 0.041 (0.040)\tLoss 0.5329 (0.4944)\tPrec@1 88.281 (86.581)\n",
      "[310/391]\tTime 0.136 (0.152)\tData 0.031 (0.040)\tLoss 0.5033 (0.4943)\tPrec@1 85.156 (86.576)\n",
      "[320/391]\tTime 0.152 (0.152)\tData 0.034 (0.040)\tLoss 0.6396 (0.4948)\tPrec@1 85.938 (86.587)\n",
      "[330/391]\tTime 0.142 (0.152)\tData 0.032 (0.040)\tLoss 0.5163 (0.4949)\tPrec@1 85.156 (86.603)\n",
      "[340/391]\tTime 0.146 (0.152)\tData 0.032 (0.040)\tLoss 0.5854 (0.4943)\tPrec@1 82.812 (86.639)\n",
      "[350/391]\tTime 0.152 (0.152)\tData 0.045 (0.040)\tLoss 0.5521 (0.4956)\tPrec@1 82.031 (86.605)\n",
      "[360/391]\tTime 0.133 (0.151)\tData 0.036 (0.040)\tLoss 0.4704 (0.4950)\tPrec@1 85.938 (86.630)\n",
      "[370/391]\tTime 0.113 (0.151)\tData 0.052 (0.040)\tLoss 0.4964 (0.4965)\tPrec@1 89.062 (86.601)\n",
      "[380/391]\tTime 0.127 (0.151)\tData 0.033 (0.040)\tLoss 0.4594 (0.4964)\tPrec@1 84.375 (86.579)\n",
      "[390/391]\tTime 0.147 (0.151)\tData 0.042 (0.040)\tLoss 0.4881 (0.4962)\tPrec@1 86.250 (86.590)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.080 (0.080)\tLoss 0.5376 (0.5376)\tPrec@1 88.281 (88.281)\n",
      "Test: [10/79]\tTime 0.039 (0.047)\tLoss 0.7562 (0.7198)\tPrec@1 79.688 (82.457)\n",
      "Test: [20/79]\tTime 0.034 (0.045)\tLoss 0.9103 (0.7791)\tPrec@1 81.250 (81.473)\n",
      "Test: [30/79]\tTime 0.037 (0.043)\tLoss 0.8252 (0.7854)\tPrec@1 81.250 (81.527)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test: [40/79]\tTime 0.047 (0.043)\tLoss 0.7483 (0.7840)\tPrec@1 77.344 (81.612)\n",
      "Test: [50/79]\tTime 0.039 (0.041)\tLoss 0.6333 (0.7726)\tPrec@1 85.938 (81.847)\n",
      "Test: [60/79]\tTime 0.031 (0.040)\tLoss 0.7182 (0.7812)\tPrec@1 79.688 (81.532)\n",
      "Test: [70/79]\tTime 0.032 (0.039)\tLoss 0.5295 (0.7783)\tPrec@1 85.156 (81.382)\n",
      " * Prec@1 81.400\n",
      "\n",
      "===> epoch: 3/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.134 (0.134)\tData 0.029 (0.029)\tLoss 0.4404 (0.4404)\tPrec@1 87.500 (87.500)\n",
      "[10/391]\tTime 0.129 (0.152)\tData 0.032 (0.038)\tLoss 0.3935 (0.4561)\tPrec@1 88.281 (86.790)\n",
      "[20/391]\tTime 0.146 (0.153)\tData 0.030 (0.039)\tLoss 0.3633 (0.4431)\tPrec@1 89.062 (87.946)\n",
      "[30/391]\tTime 0.183 (0.151)\tData 0.062 (0.038)\tLoss 0.4267 (0.4484)\tPrec@1 89.844 (87.853)\n",
      "[40/391]\tTime 0.226 (0.154)\tData 0.043 (0.039)\tLoss 0.5196 (0.4541)\tPrec@1 85.156 (87.786)\n",
      "[50/391]\tTime 0.149 (0.156)\tData 0.032 (0.039)\tLoss 0.4562 (0.4578)\tPrec@1 85.938 (87.592)\n",
      "[60/391]\tTime 0.162 (0.153)\tData 0.040 (0.038)\tLoss 0.4844 (0.4612)\tPrec@1 88.281 (87.705)\n",
      "[70/391]\tTime 0.139 (0.152)\tData 0.034 (0.038)\tLoss 0.4860 (0.4588)\tPrec@1 86.719 (87.786)\n",
      "[80/391]\tTime 0.141 (0.151)\tData 0.029 (0.037)\tLoss 0.5347 (0.4610)\tPrec@1 86.719 (87.751)\n",
      "[90/391]\tTime 0.142 (0.150)\tData 0.037 (0.037)\tLoss 0.4937 (0.4623)\tPrec@1 86.719 (87.620)\n",
      "[100/391]\tTime 0.159 (0.152)\tData 0.035 (0.038)\tLoss 0.5411 (0.4666)\tPrec@1 85.938 (87.539)\n",
      "[110/391]\tTime 0.145 (0.151)\tData 0.032 (0.038)\tLoss 0.4290 (0.4695)\tPrec@1 88.281 (87.465)\n",
      "[120/391]\tTime 0.133 (0.150)\tData 0.035 (0.038)\tLoss 0.5937 (0.4709)\tPrec@1 81.250 (87.461)\n",
      "[130/391]\tTime 0.130 (0.151)\tData 0.040 (0.038)\tLoss 0.4197 (0.4708)\tPrec@1 91.406 (87.440)\n",
      "[140/391]\tTime 0.172 (0.151)\tData 0.032 (0.039)\tLoss 0.5306 (0.4738)\tPrec@1 86.719 (87.400)\n",
      "[150/391]\tTime 0.132 (0.152)\tData 0.036 (0.039)\tLoss 0.5484 (0.4775)\tPrec@1 84.375 (87.272)\n",
      "[160/391]\tTime 0.181 (0.152)\tData 0.048 (0.039)\tLoss 0.3890 (0.4778)\tPrec@1 91.406 (87.286)\n",
      "[170/391]\tTime 0.169 (0.153)\tData 0.051 (0.039)\tLoss 0.4495 (0.4791)\tPrec@1 85.156 (87.189)\n",
      "[180/391]\tTime 0.162 (0.153)\tData 0.036 (0.039)\tLoss 0.5104 (0.4790)\tPrec@1 83.594 (87.150)\n",
      "[190/391]\tTime 0.194 (0.153)\tData 0.046 (0.039)\tLoss 0.6286 (0.4800)\tPrec@1 81.250 (87.095)\n",
      "[200/391]\tTime 0.142 (0.153)\tData 0.029 (0.039)\tLoss 0.5232 (0.4814)\tPrec@1 84.375 (87.034)\n",
      "[210/391]\tTime 0.144 (0.153)\tData 0.034 (0.039)\tLoss 0.4377 (0.4821)\tPrec@1 89.844 (87.030)\n",
      "[220/391]\tTime 0.197 (0.154)\tData 0.075 (0.039)\tLoss 0.4617 (0.4829)\tPrec@1 89.844 (87.055)\n",
      "[230/391]\tTime 0.135 (0.153)\tData 0.028 (0.038)\tLoss 0.4490 (0.4844)\tPrec@1 88.281 (86.996)\n",
      "[240/391]\tTime 0.141 (0.153)\tData 0.039 (0.038)\tLoss 0.6069 (0.4848)\tPrec@1 83.594 (87.011)\n",
      "[250/391]\tTime 0.146 (0.153)\tData 0.057 (0.038)\tLoss 0.5599 (0.4859)\tPrec@1 87.500 (86.983)\n",
      "[260/391]\tTime 0.141 (0.151)\tData 0.045 (0.039)\tLoss 0.4302 (0.4854)\tPrec@1 90.625 (87.042)\n",
      "[270/391]\tTime 0.145 (0.151)\tData 0.034 (0.039)\tLoss 0.4893 (0.4851)\tPrec@1 83.594 (87.016)\n",
      "[280/391]\tTime 0.163 (0.151)\tData 0.044 (0.039)\tLoss 0.4143 (0.4854)\tPrec@1 91.406 (86.969)\n",
      "[290/391]\tTime 0.140 (0.151)\tData 0.030 (0.039)\tLoss 0.5486 (0.4851)\tPrec@1 86.719 (86.979)\n",
      "[300/391]\tTime 0.155 (0.151)\tData 0.036 (0.039)\tLoss 0.3520 (0.4868)\tPrec@1 93.750 (86.937)\n",
      "[310/391]\tTime 0.145 (0.151)\tData 0.034 (0.039)\tLoss 0.5993 (0.4880)\tPrec@1 84.375 (86.922)\n",
      "[320/391]\tTime 0.138 (0.151)\tData 0.032 (0.039)\tLoss 0.4486 (0.4894)\tPrec@1 86.719 (86.865)\n",
      "[330/391]\tTime 0.180 (0.151)\tData 0.052 (0.039)\tLoss 0.4765 (0.4903)\tPrec@1 85.156 (86.841)\n",
      "[340/391]\tTime 0.148 (0.151)\tData 0.050 (0.039)\tLoss 0.5230 (0.4906)\tPrec@1 83.594 (86.833)\n",
      "[350/391]\tTime 0.163 (0.150)\tData 0.036 (0.038)\tLoss 0.4637 (0.4907)\tPrec@1 85.156 (86.823)\n",
      "[360/391]\tTime 0.144 (0.151)\tData 0.036 (0.039)\tLoss 0.4815 (0.4899)\tPrec@1 88.281 (86.859)\n",
      "[370/391]\tTime 0.141 (0.151)\tData 0.035 (0.039)\tLoss 0.5257 (0.4908)\tPrec@1 83.594 (86.801)\n",
      "[380/391]\tTime 0.150 (0.151)\tData 0.037 (0.039)\tLoss 0.5093 (0.4915)\tPrec@1 85.156 (86.780)\n",
      "[390/391]\tTime 0.126 (0.151)\tData 0.024 (0.039)\tLoss 0.5507 (0.4927)\tPrec@1 78.750 (86.746)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.037 (0.037)\tLoss 1.1766 (1.1766)\tPrec@1 72.656 (72.656)\n",
      "Test: [10/79]\tTime 0.035 (0.036)\tLoss 0.7605 (1.0200)\tPrec@1 77.344 (75.852)\n",
      "Test: [20/79]\tTime 0.037 (0.037)\tLoss 1.2752 (1.0828)\tPrec@1 71.875 (75.074)\n",
      "Test: [30/79]\tTime 0.034 (0.037)\tLoss 1.1129 (1.1006)\tPrec@1 75.781 (75.050)\n",
      "Test: [40/79]\tTime 0.035 (0.036)\tLoss 1.3916 (1.1076)\tPrec@1 68.750 (75.438)\n",
      "Test: [50/79]\tTime 0.033 (0.036)\tLoss 1.3991 (1.1061)\tPrec@1 66.406 (75.153)\n",
      "Test: [60/79]\tTime 0.031 (0.035)\tLoss 1.1404 (1.1051)\tPrec@1 77.344 (75.307)\n",
      "Test: [70/79]\tTime 0.027 (0.034)\tLoss 1.4206 (1.1054)\tPrec@1 73.438 (75.429)\n",
      " * Prec@1 75.540\n",
      "\n",
      "===> epoch: 4/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.159 (0.159)\tData 0.040 (0.040)\tLoss 0.6251 (0.6251)\tPrec@1 82.812 (82.812)\n",
      "[10/391]\tTime 0.169 (0.144)\tData 0.050 (0.041)\tLoss 0.4974 (0.5168)\tPrec@1 85.938 (85.938)\n",
      "[20/391]\tTime 0.125 (0.142)\tData 0.040 (0.038)\tLoss 0.5270 (0.5154)\tPrec@1 85.156 (86.272)\n",
      "[30/391]\tTime 0.177 (0.144)\tData 0.040 (0.036)\tLoss 0.4774 (0.4968)\tPrec@1 85.938 (86.719)\n",
      "[40/391]\tTime 0.149 (0.146)\tData 0.031 (0.036)\tLoss 0.4375 (0.4892)\tPrec@1 90.625 (86.833)\n",
      "[50/391]\tTime 0.145 (0.149)\tData 0.050 (0.038)\tLoss 0.4457 (0.4902)\tPrec@1 85.938 (86.811)\n",
      "[60/391]\tTime 0.170 (0.150)\tData 0.043 (0.038)\tLoss 0.5269 (0.4902)\tPrec@1 86.719 (86.719)\n",
      "[70/391]\tTime 0.145 (0.151)\tData 0.041 (0.038)\tLoss 0.5557 (0.4962)\tPrec@1 86.719 (86.609)\n",
      "[80/391]\tTime 0.155 (0.152)\tData 0.046 (0.039)\tLoss 0.4691 (0.4974)\tPrec@1 85.156 (86.526)\n",
      "[90/391]\tTime 0.142 (0.151)\tData 0.037 (0.039)\tLoss 0.5085 (0.5015)\tPrec@1 89.062 (86.324)\n",
      "[100/391]\tTime 0.146 (0.150)\tData 0.036 (0.038)\tLoss 0.4315 (0.5024)\tPrec@1 88.281 (86.216)\n",
      "[110/391]\tTime 0.146 (0.150)\tData 0.031 (0.038)\tLoss 0.4979 (0.5019)\tPrec@1 84.375 (86.233)\n",
      "[120/391]\tTime 0.148 (0.151)\tData 0.051 (0.039)\tLoss 0.5632 (0.5043)\tPrec@1 82.031 (86.273)\n",
      "[130/391]\tTime 0.151 (0.151)\tData 0.035 (0.039)\tLoss 0.5827 (0.5023)\tPrec@1 84.375 (86.415)\n",
      "[140/391]\tTime 0.146 (0.151)\tData 0.041 (0.039)\tLoss 0.4807 (0.5009)\tPrec@1 86.719 (86.514)\n",
      "[150/391]\tTime 0.098 (0.148)\tData 0.033 (0.039)\tLoss 0.5087 (0.5028)\tPrec@1 86.719 (86.413)\n",
      "[160/391]\tTime 0.114 (0.148)\tData 0.036 (0.040)\tLoss 0.4916 (0.5029)\tPrec@1 85.938 (86.394)\n",
      "[170/391]\tTime 0.196 (0.147)\tData 0.034 (0.040)\tLoss 0.4827 (0.5034)\tPrec@1 87.500 (86.417)\n",
      "[180/391]\tTime 0.189 (0.147)\tData 0.060 (0.039)\tLoss 0.4763 (0.5007)\tPrec@1 90.625 (86.525)\n",
      "[190/391]\tTime 0.130 (0.147)\tData 0.046 (0.040)\tLoss 0.3941 (0.4975)\tPrec@1 90.625 (86.641)\n",
      "[200/391]\tTime 0.170 (0.148)\tData 0.039 (0.039)\tLoss 0.5440 (0.4970)\tPrec@1 85.156 (86.637)\n",
      "[210/391]\tTime 0.131 (0.148)\tData 0.033 (0.039)\tLoss 0.5038 (0.4955)\tPrec@1 84.375 (86.630)\n",
      "[220/391]\tTime 0.127 (0.148)\tData 0.036 (0.039)\tLoss 0.4755 (0.4963)\tPrec@1 85.156 (86.602)\n",
      "[230/391]\tTime 0.149 (0.148)\tData 0.041 (0.039)\tLoss 0.5746 (0.4951)\tPrec@1 85.156 (86.627)\n",
      "[240/391]\tTime 0.140 (0.148)\tData 0.034 (0.039)\tLoss 0.5311 (0.4956)\tPrec@1 85.938 (86.576)\n",
      "[250/391]\tTime 0.156 (0.148)\tData 0.034 (0.039)\tLoss 0.4852 (0.4937)\tPrec@1 82.031 (86.588)\n",
      "[260/391]\tTime 0.143 (0.148)\tData 0.032 (0.039)\tLoss 0.4781 (0.4941)\tPrec@1 85.938 (86.575)\n",
      "[270/391]\tTime 0.131 (0.148)\tData 0.036 (0.039)\tLoss 0.4676 (0.4943)\tPrec@1 86.719 (86.580)\n",
      "[280/391]\tTime 0.135 (0.148)\tData 0.033 (0.039)\tLoss 0.4978 (0.4954)\tPrec@1 85.938 (86.571)\n",
      "[290/391]\tTime 0.131 (0.148)\tData 0.029 (0.039)\tLoss 0.4321 (0.4949)\tPrec@1 89.062 (86.571)\n",
      "[300/391]\tTime 0.142 (0.147)\tData 0.028 (0.039)\tLoss 0.5480 (0.4960)\tPrec@1 86.719 (86.532)\n",
      "[310/391]\tTime 0.144 (0.147)\tData 0.041 (0.038)\tLoss 0.3886 (0.4962)\tPrec@1 92.188 (86.528)\n",
      "[320/391]\tTime 0.163 (0.147)\tData 0.055 (0.039)\tLoss 0.6553 (0.4965)\tPrec@1 81.250 (86.539)\n",
      "[330/391]\tTime 0.227 (0.147)\tData 0.044 (0.039)\tLoss 0.4077 (0.4959)\tPrec@1 84.375 (86.575)\n",
      "[340/391]\tTime 0.173 (0.147)\tData 0.033 (0.039)\tLoss 0.4407 (0.4954)\tPrec@1 89.844 (86.602)\n",
      "[350/391]\tTime 0.135 (0.147)\tData 0.032 (0.039)\tLoss 0.4771 (0.4953)\tPrec@1 87.500 (86.610)\n",
      "[360/391]\tTime 0.136 (0.147)\tData 0.026 (0.039)\tLoss 0.5179 (0.4949)\tPrec@1 84.375 (86.613)\n",
      "[370/391]\tTime 0.148 (0.147)\tData 0.034 (0.039)\tLoss 0.5868 (0.4950)\tPrec@1 86.719 (86.601)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[380/391]\tTime 0.132 (0.148)\tData 0.033 (0.039)\tLoss 0.5097 (0.4945)\tPrec@1 85.156 (86.614)\n",
      "[390/391]\tTime 0.121 (0.148)\tData 0.029 (0.039)\tLoss 0.4362 (0.4948)\tPrec@1 93.750 (86.612)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.047 (0.047)\tLoss 0.9255 (0.9255)\tPrec@1 84.375 (84.375)\n",
      "Test: [10/79]\tTime 0.039 (0.048)\tLoss 0.6879 (0.7546)\tPrec@1 84.375 (82.386)\n",
      "Test: [20/79]\tTime 0.034 (0.047)\tLoss 0.7703 (0.7800)\tPrec@1 82.031 (81.622)\n",
      "Test: [30/79]\tTime 0.045 (0.046)\tLoss 0.4677 (0.7836)\tPrec@1 89.062 (81.401)\n",
      "Test: [40/79]\tTime 0.054 (0.045)\tLoss 1.0797 (0.7756)\tPrec@1 77.344 (81.555)\n",
      "Test: [50/79]\tTime 0.034 (0.044)\tLoss 0.7944 (0.7575)\tPrec@1 82.812 (81.909)\n",
      "Test: [60/79]\tTime 0.032 (0.043)\tLoss 0.7346 (0.7510)\tPrec@1 85.156 (82.070)\n",
      "Test: [70/79]\tTime 0.035 (0.041)\tLoss 0.6467 (0.7463)\tPrec@1 84.375 (82.218)\n",
      " * Prec@1 82.300\n",
      "\n",
      "===> epoch: 5/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.149 (0.149)\tData 0.034 (0.034)\tLoss 0.6049 (0.6049)\tPrec@1 80.469 (80.469)\n",
      "[10/391]\tTime 0.146 (0.169)\tData 0.037 (0.043)\tLoss 0.6279 (0.5013)\tPrec@1 78.125 (84.517)\n",
      "[20/391]\tTime 0.167 (0.166)\tData 0.046 (0.041)\tLoss 0.4442 (0.5107)\tPrec@1 90.625 (84.970)\n",
      "[30/391]\tTime 0.137 (0.160)\tData 0.020 (0.039)\tLoss 0.5407 (0.4948)\tPrec@1 83.594 (85.761)\n",
      "[40/391]\tTime 0.148 (0.157)\tData 0.032 (0.038)\tLoss 0.3523 (0.4796)\tPrec@1 92.188 (86.643)\n",
      "[50/391]\tTime 0.131 (0.151)\tData 0.032 (0.038)\tLoss 0.6096 (0.4823)\tPrec@1 82.031 (86.749)\n",
      "[60/391]\tTime 0.134 (0.145)\tData 0.031 (0.037)\tLoss 0.5117 (0.4859)\tPrec@1 83.594 (86.578)\n",
      "[70/391]\tTime 0.146 (0.145)\tData 0.033 (0.038)\tLoss 0.4257 (0.4886)\tPrec@1 89.062 (86.576)\n",
      "[80/391]\tTime 0.138 (0.146)\tData 0.036 (0.039)\tLoss 0.4040 (0.4866)\tPrec@1 90.625 (86.709)\n",
      "[90/391]\tTime 0.136 (0.146)\tData 0.048 (0.038)\tLoss 0.4635 (0.4888)\tPrec@1 87.500 (86.641)\n",
      "[100/391]\tTime 0.163 (0.146)\tData 0.034 (0.038)\tLoss 0.5110 (0.4885)\tPrec@1 87.500 (86.734)\n",
      "[110/391]\tTime 0.136 (0.147)\tData 0.040 (0.039)\tLoss 0.3559 (0.4873)\tPrec@1 94.531 (86.824)\n",
      "[120/391]\tTime 0.185 (0.148)\tData 0.029 (0.038)\tLoss 0.5159 (0.4886)\tPrec@1 85.938 (86.770)\n",
      "[130/391]\tTime 0.151 (0.148)\tData 0.033 (0.038)\tLoss 0.4328 (0.4886)\tPrec@1 89.844 (86.701)\n",
      "[140/391]\tTime 0.136 (0.149)\tData 0.041 (0.039)\tLoss 0.5062 (0.4888)\tPrec@1 82.812 (86.630)\n",
      "[150/391]\tTime 0.182 (0.150)\tData 0.054 (0.039)\tLoss 0.4795 (0.4876)\tPrec@1 83.594 (86.662)\n",
      "[160/391]\tTime 0.171 (0.151)\tData 0.048 (0.039)\tLoss 0.4802 (0.4876)\tPrec@1 88.281 (86.641)\n",
      "[170/391]\tTime 0.161 (0.151)\tData 0.040 (0.039)\tLoss 0.6038 (0.4879)\tPrec@1 81.250 (86.609)\n",
      "[180/391]\tTime 0.189 (0.151)\tData 0.052 (0.039)\tLoss 0.4515 (0.4893)\tPrec@1 86.719 (86.598)\n",
      "[190/391]\tTime 0.176 (0.152)\tData 0.049 (0.039)\tLoss 0.5009 (0.4905)\tPrec@1 87.500 (86.563)\n",
      "[200/391]\tTime 0.149 (0.152)\tData 0.034 (0.039)\tLoss 0.5744 (0.4912)\tPrec@1 85.156 (86.590)\n",
      "[210/391]\tTime 0.190 (0.152)\tData 0.037 (0.040)\tLoss 0.4054 (0.4930)\tPrec@1 90.625 (86.548)\n",
      "[220/391]\tTime 0.194 (0.152)\tData 0.067 (0.040)\tLoss 0.4789 (0.4937)\tPrec@1 88.281 (86.521)\n",
      "[230/391]\tTime 0.160 (0.152)\tData 0.028 (0.040)\tLoss 0.4959 (0.4937)\tPrec@1 87.500 (86.523)\n",
      "[240/391]\tTime 0.142 (0.152)\tData 0.043 (0.040)\tLoss 0.4029 (0.4950)\tPrec@1 90.625 (86.518)\n",
      "[250/391]\tTime 0.188 (0.153)\tData 0.057 (0.040)\tLoss 0.4592 (0.4963)\tPrec@1 87.500 (86.504)\n",
      "[260/391]\tTime 0.149 (0.153)\tData 0.034 (0.040)\tLoss 0.6532 (0.4975)\tPrec@1 82.031 (86.464)\n",
      "[270/391]\tTime 0.156 (0.153)\tData 0.037 (0.040)\tLoss 0.6249 (0.4970)\tPrec@1 81.250 (86.485)\n",
      "[280/391]\tTime 0.151 (0.153)\tData 0.036 (0.040)\tLoss 0.5464 (0.4969)\tPrec@1 85.156 (86.494)\n",
      "[290/391]\tTime 0.137 (0.153)\tData 0.034 (0.040)\tLoss 0.4330 (0.4987)\tPrec@1 90.625 (86.456)\n",
      "[300/391]\tTime 0.158 (0.153)\tData 0.040 (0.040)\tLoss 0.4746 (0.4973)\tPrec@1 85.156 (86.521)\n",
      "[310/391]\tTime 0.156 (0.153)\tData 0.032 (0.040)\tLoss 0.5016 (0.4964)\tPrec@1 85.156 (86.555)\n",
      "[320/391]\tTime 0.146 (0.153)\tData 0.020 (0.040)\tLoss 0.4793 (0.4972)\tPrec@1 85.938 (86.563)\n",
      "[330/391]\tTime 0.127 (0.153)\tData 0.037 (0.039)\tLoss 0.5765 (0.4967)\tPrec@1 85.938 (86.579)\n",
      "[340/391]\tTime 0.146 (0.153)\tData 0.031 (0.039)\tLoss 0.5134 (0.4969)\tPrec@1 85.938 (86.572)\n",
      "[350/391]\tTime 0.128 (0.153)\tData 0.047 (0.039)\tLoss 0.6231 (0.4971)\tPrec@1 83.594 (86.556)\n",
      "[360/391]\tTime 0.105 (0.152)\tData 0.043 (0.040)\tLoss 0.3872 (0.4960)\tPrec@1 90.625 (86.580)\n",
      "[370/391]\tTime 0.127 (0.152)\tData 0.035 (0.040)\tLoss 0.4841 (0.4953)\tPrec@1 84.375 (86.586)\n",
      "[380/391]\tTime 0.150 (0.152)\tData 0.028 (0.040)\tLoss 0.7233 (0.4965)\tPrec@1 78.906 (86.555)\n",
      "[390/391]\tTime 0.107 (0.152)\tData 0.027 (0.039)\tLoss 0.7001 (0.4974)\tPrec@1 85.000 (86.554)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.040 (0.040)\tLoss 0.8180 (0.8180)\tPrec@1 77.344 (77.344)\n",
      "Test: [10/79]\tTime 0.038 (0.037)\tLoss 0.7801 (0.7241)\tPrec@1 82.812 (81.960)\n",
      "Test: [20/79]\tTime 0.040 (0.036)\tLoss 0.7228 (0.7601)\tPrec@1 82.812 (81.548)\n",
      "Test: [30/79]\tTime 0.061 (0.038)\tLoss 0.8416 (0.7552)\tPrec@1 81.250 (81.300)\n",
      "Test: [40/79]\tTime 0.040 (0.038)\tLoss 0.8322 (0.7499)\tPrec@1 78.906 (81.574)\n",
      "Test: [50/79]\tTime 0.032 (0.037)\tLoss 0.8401 (0.7379)\tPrec@1 78.906 (81.801)\n",
      "Test: [60/79]\tTime 0.026 (0.039)\tLoss 0.8702 (0.7369)\tPrec@1 83.594 (81.942)\n",
      "Test: [70/79]\tTime 0.026 (0.038)\tLoss 0.7472 (0.7341)\tPrec@1 82.031 (81.965)\n",
      " * Prec@1 81.940\n",
      "\n",
      "===> epoch: 6/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.137 (0.137)\tData 0.021 (0.021)\tLoss 0.4873 (0.4873)\tPrec@1 85.156 (85.156)\n",
      "[10/391]\tTime 0.129 (0.142)\tData 0.030 (0.037)\tLoss 0.4538 (0.4652)\tPrec@1 88.281 (88.068)\n",
      "[20/391]\tTime 0.177 (0.150)\tData 0.043 (0.038)\tLoss 0.5752 (0.4882)\tPrec@1 84.375 (86.979)\n",
      "[30/391]\tTime 0.164 (0.155)\tData 0.042 (0.039)\tLoss 0.5344 (0.4904)\tPrec@1 84.375 (86.719)\n",
      "[40/391]\tTime 0.161 (0.157)\tData 0.061 (0.041)\tLoss 0.4943 (0.4774)\tPrec@1 87.500 (87.157)\n",
      "[50/391]\tTime 0.171 (0.156)\tData 0.033 (0.040)\tLoss 0.4785 (0.4740)\tPrec@1 87.500 (87.485)\n",
      "[60/391]\tTime 0.134 (0.154)\tData 0.034 (0.039)\tLoss 0.4376 (0.4713)\tPrec@1 85.156 (87.359)\n",
      "[70/391]\tTime 0.173 (0.156)\tData 0.040 (0.040)\tLoss 0.5734 (0.4767)\tPrec@1 79.688 (86.950)\n",
      "[80/391]\tTime 2.118 (0.179)\tData 0.034 (0.039)\tLoss 0.5193 (0.4787)\tPrec@1 85.156 (87.037)\n",
      "[90/391]\tTime 0.120 (0.177)\tData 0.033 (0.043)\tLoss 0.5154 (0.4778)\tPrec@1 86.719 (87.054)\n",
      "[100/391]\tTime 0.193 (0.174)\tData 0.067 (0.042)\tLoss 0.4984 (0.4833)\tPrec@1 86.719 (86.858)\n",
      "[110/391]\tTime 0.173 (0.174)\tData 0.036 (0.042)\tLoss 0.5400 (0.4846)\tPrec@1 87.500 (86.902)\n",
      "[120/391]\tTime 0.139 (0.174)\tData 0.037 (0.043)\tLoss 0.5046 (0.4856)\tPrec@1 85.938 (86.874)\n",
      "[130/391]\tTime 0.113 (0.171)\tData 0.020 (0.041)\tLoss 0.5528 (0.4856)\tPrec@1 85.156 (86.904)\n",
      "[140/391]\tTime 0.137 (0.170)\tData 0.032 (0.041)\tLoss 0.6922 (0.4885)\tPrec@1 75.781 (86.746)\n",
      "[150/391]\tTime 0.147 (0.168)\tData 0.041 (0.041)\tLoss 0.4488 (0.4866)\tPrec@1 88.281 (86.833)\n",
      "[160/391]\tTime 0.153 (0.167)\tData 0.031 (0.040)\tLoss 0.4272 (0.4878)\tPrec@1 92.188 (86.782)\n",
      "[170/391]\tTime 0.129 (0.166)\tData 0.028 (0.040)\tLoss 0.4953 (0.4868)\tPrec@1 83.594 (86.801)\n",
      "[180/391]\tTime 0.173 (0.166)\tData 0.076 (0.040)\tLoss 0.5398 (0.4894)\tPrec@1 82.031 (86.667)\n",
      "[190/391]\tTime 0.157 (0.165)\tData 0.040 (0.040)\tLoss 0.5973 (0.4900)\tPrec@1 82.812 (86.653)\n",
      "[200/391]\tTime 0.140 (0.165)\tData 0.037 (0.040)\tLoss 0.5777 (0.4912)\tPrec@1 82.812 (86.614)\n",
      "[210/391]\tTime 0.169 (0.164)\tData 0.047 (0.040)\tLoss 0.5572 (0.4919)\tPrec@1 83.594 (86.556)\n",
      "[220/391]\tTime 0.146 (0.164)\tData 0.034 (0.040)\tLoss 0.4308 (0.4916)\tPrec@1 88.281 (86.581)\n",
      "[230/391]\tTime 0.166 (0.163)\tData 0.043 (0.040)\tLoss 0.5640 (0.4905)\tPrec@1 85.938 (86.665)\n",
      "[240/391]\tTime 0.188 (0.163)\tData 0.058 (0.040)\tLoss 0.3914 (0.4901)\tPrec@1 87.500 (86.686)\n",
      "[250/391]\tTime 0.180 (0.162)\tData 0.070 (0.040)\tLoss 0.4549 (0.4892)\tPrec@1 88.281 (86.716)\n",
      "[260/391]\tTime 0.173 (0.162)\tData 0.070 (0.041)\tLoss 0.4069 (0.4883)\tPrec@1 90.625 (86.731)\n",
      "[270/391]\tTime 0.138 (0.161)\tData 0.042 (0.041)\tLoss 0.5266 (0.4868)\tPrec@1 85.938 (86.802)\n",
      "[280/391]\tTime 0.157 (0.160)\tData 0.037 (0.041)\tLoss 0.5593 (0.4880)\tPrec@1 86.719 (86.785)\n",
      "[290/391]\tTime 0.147 (0.160)\tData 0.030 (0.040)\tLoss 0.6011 (0.4893)\tPrec@1 88.281 (86.799)\n",
      "[300/391]\tTime 0.154 (0.160)\tData 0.037 (0.040)\tLoss 0.5533 (0.4898)\tPrec@1 83.594 (86.789)\n",
      "[310/391]\tTime 0.162 (0.159)\tData 0.038 (0.040)\tLoss 0.4355 (0.4898)\tPrec@1 89.062 (86.782)\n",
      "[320/391]\tTime 0.123 (0.159)\tData 0.022 (0.040)\tLoss 0.4833 (0.4902)\tPrec@1 88.281 (86.804)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[330/391]\tTime 0.150 (0.159)\tData 0.050 (0.040)\tLoss 0.3138 (0.4902)\tPrec@1 92.188 (86.792)\n",
      "[340/391]\tTime 0.153 (0.160)\tData 0.038 (0.041)\tLoss 0.4818 (0.4905)\tPrec@1 88.281 (86.785)\n",
      "[350/391]\tTime 0.126 (0.160)\tData 0.035 (0.041)\tLoss 0.6273 (0.4915)\tPrec@1 82.031 (86.728)\n",
      "[360/391]\tTime 0.157 (0.160)\tData 0.019 (0.041)\tLoss 0.4574 (0.4915)\tPrec@1 88.281 (86.717)\n",
      "[370/391]\tTime 0.146 (0.160)\tData 0.036 (0.040)\tLoss 0.5925 (0.4917)\tPrec@1 81.250 (86.687)\n",
      "[380/391]\tTime 0.148 (0.160)\tData 0.033 (0.041)\tLoss 0.3699 (0.4920)\tPrec@1 87.500 (86.670)\n",
      "[390/391]\tTime 0.097 (0.159)\tData 0.029 (0.040)\tLoss 0.4741 (0.4922)\tPrec@1 88.750 (86.658)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.030 (0.030)\tLoss 0.5221 (0.5221)\tPrec@1 84.375 (84.375)\n",
      "Test: [10/79]\tTime 0.028 (0.034)\tLoss 0.6655 (0.6337)\tPrec@1 85.156 (84.020)\n",
      "Test: [20/79]\tTime 0.031 (0.033)\tLoss 0.7171 (0.6772)\tPrec@1 81.250 (82.999)\n",
      "Test: [30/79]\tTime 0.040 (0.035)\tLoss 0.7091 (0.6993)\tPrec@1 82.812 (82.560)\n",
      "Test: [40/79]\tTime 0.043 (0.036)\tLoss 0.8724 (0.6923)\tPrec@1 79.688 (82.736)\n",
      "Test: [50/79]\tTime 0.048 (0.036)\tLoss 0.7211 (0.6807)\tPrec@1 81.250 (82.889)\n",
      "Test: [60/79]\tTime 0.042 (0.035)\tLoss 0.8219 (0.6789)\tPrec@1 81.250 (82.877)\n",
      "Test: [70/79]\tTime 0.027 (0.035)\tLoss 0.5676 (0.6788)\tPrec@1 85.156 (82.812)\n",
      " * Prec@1 82.670\n",
      "\n",
      "===> epoch: 7/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.148 (0.148)\tData 0.055 (0.055)\tLoss 0.5208 (0.5208)\tPrec@1 86.719 (86.719)\n",
      "[10/391]\tTime 0.273 (0.172)\tData 0.072 (0.046)\tLoss 0.5195 (0.5185)\tPrec@1 85.938 (85.511)\n",
      "[20/391]\tTime 0.131 (0.162)\tData 0.038 (0.045)\tLoss 0.4983 (0.4999)\tPrec@1 88.281 (86.161)\n",
      "[30/391]\tTime 0.140 (0.160)\tData 0.020 (0.041)\tLoss 0.5606 (0.5009)\tPrec@1 82.031 (86.064)\n",
      "[40/391]\tTime 0.152 (0.155)\tData 0.035 (0.039)\tLoss 0.4808 (0.4917)\tPrec@1 91.406 (86.357)\n",
      "[50/391]\tTime 0.165 (0.153)\tData 0.036 (0.038)\tLoss 0.4536 (0.4922)\tPrec@1 89.062 (86.336)\n",
      "[60/391]\tTime 0.149 (0.153)\tData 0.027 (0.037)\tLoss 0.4863 (0.4885)\tPrec@1 82.812 (86.514)\n",
      "[70/391]\tTime 0.143 (0.152)\tData 0.042 (0.038)\tLoss 0.4580 (0.4895)\tPrec@1 89.062 (86.510)\n",
      "[80/391]\tTime 0.188 (0.153)\tData 0.070 (0.039)\tLoss 0.5557 (0.4828)\tPrec@1 85.938 (86.854)\n",
      "[90/391]\tTime 0.158 (0.154)\tData 0.035 (0.040)\tLoss 0.4200 (0.4835)\tPrec@1 88.281 (86.839)\n",
      "[100/391]\tTime 0.161 (0.156)\tData 0.039 (0.041)\tLoss 0.5719 (0.4865)\tPrec@1 86.719 (86.773)\n",
      "[110/391]\tTime 0.151 (0.155)\tData 0.049 (0.040)\tLoss 0.5288 (0.4893)\tPrec@1 85.938 (86.691)\n",
      "[120/391]\tTime 0.143 (0.156)\tData 0.038 (0.040)\tLoss 0.5288 (0.4949)\tPrec@1 79.688 (86.435)\n",
      "[130/391]\tTime 0.152 (0.155)\tData 0.038 (0.040)\tLoss 0.5501 (0.4966)\tPrec@1 85.938 (86.379)\n",
      "[140/391]\tTime 0.152 (0.154)\tData 0.033 (0.040)\tLoss 0.5055 (0.4966)\tPrec@1 82.812 (86.431)\n",
      "[150/391]\tTime 0.110 (0.152)\tData 0.035 (0.040)\tLoss 0.5070 (0.4946)\tPrec@1 88.281 (86.486)\n",
      "[160/391]\tTime 0.127 (0.150)\tData 0.047 (0.040)\tLoss 0.5318 (0.4971)\tPrec@1 85.156 (86.428)\n",
      "[170/391]\tTime 0.117 (0.149)\tData 0.034 (0.041)\tLoss 0.5082 (0.4998)\tPrec@1 83.594 (86.276)\n",
      "[180/391]\tTime 0.159 (0.149)\tData 0.038 (0.041)\tLoss 0.4832 (0.5006)\tPrec@1 86.719 (86.287)\n",
      "[190/391]\tTime 0.121 (0.149)\tData 0.024 (0.041)\tLoss 0.5451 (0.5018)\tPrec@1 86.719 (86.293)\n",
      "[200/391]\tTime 0.150 (0.149)\tData 0.039 (0.040)\tLoss 0.4572 (0.5014)\tPrec@1 87.500 (86.315)\n",
      "[210/391]\tTime 0.171 (0.149)\tData 0.050 (0.040)\tLoss 0.5515 (0.5015)\tPrec@1 82.812 (86.315)\n",
      "[220/391]\tTime 0.143 (0.149)\tData 0.038 (0.040)\tLoss 0.5725 (0.5040)\tPrec@1 86.719 (86.273)\n",
      "[230/391]\tTime 0.166 (0.149)\tData 0.038 (0.040)\tLoss 0.4494 (0.5034)\tPrec@1 82.812 (86.269)\n",
      "[240/391]\tTime 0.149 (0.149)\tData 0.039 (0.040)\tLoss 0.5715 (0.5041)\tPrec@1 83.594 (86.281)\n",
      "[250/391]\tTime 0.161 (0.150)\tData 0.038 (0.040)\tLoss 0.4375 (0.5040)\tPrec@1 88.281 (86.292)\n",
      "[260/391]\tTime 0.166 (0.151)\tData 0.032 (0.040)\tLoss 0.6096 (0.5046)\tPrec@1 83.594 (86.279)\n",
      "[270/391]\tTime 0.131 (0.150)\tData 0.033 (0.040)\tLoss 0.4374 (0.5056)\tPrec@1 87.500 (86.237)\n",
      "[280/391]\tTime 0.145 (0.151)\tData 0.054 (0.040)\tLoss 0.5715 (0.5053)\tPrec@1 82.812 (86.246)\n",
      "[290/391]\tTime 0.150 (0.151)\tData 0.034 (0.040)\tLoss 0.4973 (0.5057)\tPrec@1 86.719 (86.262)\n",
      "[300/391]\tTime 0.174 (0.151)\tData 0.029 (0.040)\tLoss 0.5132 (0.5061)\tPrec@1 88.281 (86.270)\n",
      "[310/391]\tTime 0.170 (0.151)\tData 0.060 (0.040)\tLoss 0.4887 (0.5066)\tPrec@1 86.719 (86.272)\n",
      "[320/391]\tTime 0.121 (0.151)\tData 0.037 (0.040)\tLoss 0.5535 (0.5063)\tPrec@1 85.938 (86.281)\n",
      "[330/391]\tTime 0.142 (0.152)\tData 0.037 (0.040)\tLoss 0.4279 (0.5061)\tPrec@1 88.281 (86.296)\n",
      "[340/391]\tTime 0.154 (0.151)\tData 0.028 (0.040)\tLoss 0.6371 (0.5060)\tPrec@1 83.594 (86.320)\n",
      "[350/391]\tTime 0.157 (0.151)\tData 0.034 (0.040)\tLoss 0.4921 (0.5065)\tPrec@1 87.500 (86.309)\n",
      "[360/391]\tTime 0.157 (0.151)\tData 0.042 (0.040)\tLoss 0.5687 (0.5072)\tPrec@1 83.594 (86.292)\n",
      "[370/391]\tTime 0.191 (0.151)\tData 0.028 (0.040)\tLoss 0.4624 (0.5058)\tPrec@1 89.062 (86.340)\n",
      "[380/391]\tTime 0.139 (0.151)\tData 0.026 (0.040)\tLoss 0.3871 (0.5055)\tPrec@1 89.844 (86.362)\n",
      "[390/391]\tTime 0.113 (0.151)\tData 0.022 (0.040)\tLoss 0.7816 (0.5057)\tPrec@1 81.250 (86.332)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.035 (0.035)\tLoss 0.8818 (0.8818)\tPrec@1 78.906 (78.906)\n",
      "Test: [10/79]\tTime 0.036 (0.042)\tLoss 1.0350 (0.9986)\tPrec@1 82.031 (79.048)\n",
      "Test: [20/79]\tTime 0.036 (0.044)\tLoss 1.1896 (1.0254)\tPrec@1 75.781 (78.199)\n",
      "Test: [30/79]\tTime 0.049 (0.045)\tLoss 1.1460 (1.0136)\tPrec@1 72.656 (78.377)\n",
      "Test: [40/79]\tTime 0.051 (0.045)\tLoss 1.2865 (1.0010)\tPrec@1 72.656 (78.697)\n",
      "Test: [50/79]\tTime 0.037 (0.047)\tLoss 0.7900 (0.9947)\tPrec@1 81.250 (78.523)\n",
      "Test: [60/79]\tTime 0.039 (0.047)\tLoss 0.6056 (0.9803)\tPrec@1 83.594 (78.727)\n",
      "Test: [70/79]\tTime 0.035 (0.045)\tLoss 1.0224 (0.9847)\tPrec@1 79.688 (78.609)\n",
      " * Prec@1 78.600\n",
      "\n",
      "===> epoch: 8/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.150 (0.150)\tData 0.022 (0.022)\tLoss 0.4849 (0.4849)\tPrec@1 86.719 (86.719)\n",
      "[10/391]\tTime 0.147 (0.145)\tData 0.041 (0.031)\tLoss 0.5273 (0.4901)\tPrec@1 87.500 (86.080)\n",
      "[20/391]\tTime 0.124 (0.151)\tData 0.031 (0.035)\tLoss 0.4420 (0.5118)\tPrec@1 89.062 (86.161)\n",
      "[30/391]\tTime 0.178 (0.152)\tData 0.051 (0.035)\tLoss 0.4873 (0.5013)\tPrec@1 85.938 (86.341)\n",
      "[40/391]\tTime 0.134 (0.153)\tData 0.023 (0.037)\tLoss 0.3904 (0.4896)\tPrec@1 90.625 (86.719)\n",
      "[50/391]\tTime 0.138 (0.153)\tData 0.057 (0.038)\tLoss 0.4255 (0.4831)\tPrec@1 89.062 (87.102)\n",
      "[60/391]\tTime 0.099 (0.148)\tData 0.033 (0.039)\tLoss 0.4549 (0.4787)\tPrec@1 87.500 (87.167)\n",
      "[70/391]\tTime 0.153 (0.149)\tData 0.054 (0.042)\tLoss 0.5632 (0.4770)\tPrec@1 82.031 (87.214)\n",
      "[80/391]\tTime 0.140 (0.150)\tData 0.020 (0.041)\tLoss 0.5350 (0.4770)\tPrec@1 82.031 (87.133)\n",
      "[90/391]\tTime 0.128 (0.150)\tData 0.036 (0.041)\tLoss 0.5390 (0.4765)\tPrec@1 86.719 (87.208)\n",
      "[100/391]\tTime 0.187 (0.150)\tData 0.035 (0.041)\tLoss 0.5869 (0.4834)\tPrec@1 83.594 (87.051)\n",
      "[110/391]\tTime 0.146 (0.151)\tData 0.045 (0.041)\tLoss 0.4351 (0.4821)\tPrec@1 89.844 (87.085)\n",
      "[120/391]\tTime 0.145 (0.151)\tData 0.043 (0.041)\tLoss 0.5687 (0.4846)\tPrec@1 84.375 (86.977)\n",
      "[130/391]\tTime 0.149 (0.151)\tData 0.037 (0.041)\tLoss 0.4505 (0.4847)\tPrec@1 89.844 (86.999)\n",
      "[140/391]\tTime 0.157 (0.153)\tData 0.039 (0.042)\tLoss 0.5917 (0.4864)\tPrec@1 85.938 (87.001)\n",
      "[150/391]\tTime 0.181 (0.155)\tData 0.053 (0.042)\tLoss 0.5546 (0.4885)\tPrec@1 87.500 (86.895)\n",
      "[160/391]\tTime 0.139 (0.155)\tData 0.036 (0.042)\tLoss 0.4402 (0.4895)\tPrec@1 89.844 (86.893)\n",
      "[170/391]\tTime 0.132 (0.156)\tData 0.033 (0.042)\tLoss 0.5204 (0.4884)\tPrec@1 84.375 (86.906)\n",
      "[180/391]\tTime 0.113 (0.155)\tData 0.027 (0.042)\tLoss 0.5140 (0.4894)\tPrec@1 85.938 (86.866)\n",
      "[190/391]\tTime 0.144 (0.154)\tData 0.031 (0.041)\tLoss 0.4709 (0.4891)\tPrec@1 89.062 (86.948)\n",
      "[200/391]\tTime 0.135 (0.154)\tData 0.047 (0.041)\tLoss 0.5670 (0.4904)\tPrec@1 83.594 (86.909)\n",
      "[210/391]\tTime 0.135 (0.153)\tData 0.037 (0.041)\tLoss 0.4714 (0.4914)\tPrec@1 86.719 (86.867)\n",
      "[220/391]\tTime 0.229 (0.154)\tData 0.101 (0.042)\tLoss 0.4646 (0.4925)\tPrec@1 85.156 (86.857)\n",
      "[230/391]\tTime 0.141 (0.153)\tData 0.019 (0.042)\tLoss 0.4559 (0.4942)\tPrec@1 88.281 (86.786)\n",
      "[240/391]\tTime 0.150 (0.153)\tData 0.031 (0.041)\tLoss 0.4646 (0.4930)\tPrec@1 87.500 (86.819)\n",
      "[250/391]\tTime 0.122 (0.153)\tData 0.029 (0.041)\tLoss 0.5298 (0.4936)\tPrec@1 86.719 (86.825)\n",
      "[260/391]\tTime 0.148 (0.152)\tData 0.028 (0.041)\tLoss 0.5697 (0.4930)\tPrec@1 82.812 (86.803)\n",
      "[270/391]\tTime 0.190 (0.153)\tData 0.065 (0.041)\tLoss 0.6484 (0.4933)\tPrec@1 78.906 (86.759)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[280/391]\tTime 0.193 (0.153)\tData 0.067 (0.041)\tLoss 0.4679 (0.4932)\tPrec@1 87.500 (86.758)\n",
      "[290/391]\tTime 0.123 (0.154)\tData 0.040 (0.041)\tLoss 0.5199 (0.4917)\tPrec@1 86.719 (86.818)\n",
      "[300/391]\tTime 0.163 (0.154)\tData 0.047 (0.042)\tLoss 0.4818 (0.4918)\tPrec@1 85.938 (86.817)\n",
      "[310/391]\tTime 0.166 (0.154)\tData 0.040 (0.041)\tLoss 0.6068 (0.4917)\tPrec@1 82.812 (86.827)\n",
      "[320/391]\tTime 0.128 (0.154)\tData 0.034 (0.041)\tLoss 0.5752 (0.4918)\tPrec@1 84.375 (86.811)\n",
      "[330/391]\tTime 0.132 (0.154)\tData 0.027 (0.041)\tLoss 0.6355 (0.4907)\tPrec@1 80.469 (86.865)\n",
      "[340/391]\tTime 0.129 (0.154)\tData 0.032 (0.041)\tLoss 0.5325 (0.4909)\tPrec@1 82.812 (86.865)\n",
      "[350/391]\tTime 0.150 (0.153)\tData 0.054 (0.041)\tLoss 0.4729 (0.4914)\tPrec@1 89.844 (86.839)\n",
      "[360/391]\tTime 0.168 (0.153)\tData 0.051 (0.041)\tLoss 0.4648 (0.4909)\tPrec@1 85.156 (86.851)\n",
      "[370/391]\tTime 0.123 (0.153)\tData 0.041 (0.041)\tLoss 0.4850 (0.4909)\tPrec@1 89.062 (86.843)\n",
      "[380/391]\tTime 0.146 (0.152)\tData 0.035 (0.041)\tLoss 0.4026 (0.4913)\tPrec@1 89.844 (86.811)\n",
      "[390/391]\tTime 0.144 (0.152)\tData 0.034 (0.041)\tLoss 0.5752 (0.4917)\tPrec@1 77.500 (86.792)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.043 (0.043)\tLoss 0.6794 (0.6794)\tPrec@1 82.031 (82.031)\n",
      "Test: [10/79]\tTime 0.042 (0.041)\tLoss 0.5455 (0.7973)\tPrec@1 85.938 (80.682)\n",
      "Test: [20/79]\tTime 0.039 (0.041)\tLoss 0.8074 (0.8623)\tPrec@1 80.469 (80.506)\n",
      "Test: [30/79]\tTime 0.032 (0.041)\tLoss 0.7136 (0.8284)\tPrec@1 81.250 (80.771)\n",
      "Test: [40/79]\tTime 0.049 (0.040)\tLoss 0.7052 (0.8276)\tPrec@1 84.375 (80.659)\n",
      "Test: [50/79]\tTime 0.037 (0.040)\tLoss 0.9499 (0.8201)\tPrec@1 79.688 (80.653)\n",
      "Test: [60/79]\tTime 0.038 (0.040)\tLoss 0.9710 (0.8269)\tPrec@1 81.250 (80.456)\n",
      "Test: [70/79]\tTime 0.032 (0.040)\tLoss 0.8494 (0.8303)\tPrec@1 80.469 (80.403)\n",
      " * Prec@1 80.310\n",
      "\n",
      "===> epoch: 9/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.131 (0.131)\tData 0.030 (0.030)\tLoss 0.4389 (0.4389)\tPrec@1 89.844 (89.844)\n",
      "[10/391]\tTime 0.151 (0.153)\tData 0.036 (0.036)\tLoss 0.3815 (0.4443)\tPrec@1 92.969 (88.991)\n",
      "[20/391]\tTime 0.171 (0.156)\tData 0.053 (0.037)\tLoss 0.4782 (0.4586)\tPrec@1 91.406 (88.244)\n",
      "[30/391]\tTime 0.155 (0.159)\tData 0.039 (0.041)\tLoss 0.6219 (0.4690)\tPrec@1 83.594 (87.828)\n",
      "[40/391]\tTime 0.145 (0.160)\tData 0.046 (0.041)\tLoss 0.3996 (0.4756)\tPrec@1 87.500 (87.290)\n",
      "[50/391]\tTime 0.129 (0.159)\tData 0.020 (0.041)\tLoss 0.4601 (0.4762)\tPrec@1 85.938 (87.224)\n",
      "[60/391]\tTime 0.168 (0.159)\tData 0.050 (0.041)\tLoss 0.4698 (0.4765)\tPrec@1 88.281 (87.180)\n",
      "[70/391]\tTime 0.135 (0.160)\tData 0.034 (0.042)\tLoss 0.5075 (0.4753)\tPrec@1 84.375 (87.203)\n",
      "[80/391]\tTime 0.131 (0.158)\tData 0.026 (0.040)\tLoss 0.5531 (0.4806)\tPrec@1 85.156 (87.027)\n",
      "[90/391]\tTime 0.112 (0.155)\tData 0.028 (0.039)\tLoss 0.4684 (0.4805)\tPrec@1 89.844 (87.011)\n",
      "[100/391]\tTime 0.137 (0.155)\tData 0.039 (0.040)\tLoss 0.5916 (0.4854)\tPrec@1 85.938 (86.966)\n",
      "[110/391]\tTime 0.171 (0.156)\tData 0.057 (0.041)\tLoss 0.4609 (0.4855)\tPrec@1 84.375 (86.874)\n",
      "[120/391]\tTime 0.132 (0.158)\tData 0.035 (0.042)\tLoss 0.4984 (0.4828)\tPrec@1 85.938 (86.938)\n",
      "[130/391]\tTime 0.144 (0.158)\tData 0.041 (0.042)\tLoss 0.3894 (0.4786)\tPrec@1 88.281 (87.148)\n",
      "[140/391]\tTime 0.122 (0.157)\tData 0.039 (0.041)\tLoss 0.4965 (0.4809)\tPrec@1 85.156 (87.051)\n",
      "[150/391]\tTime 0.171 (0.156)\tData 0.043 (0.041)\tLoss 0.4863 (0.4855)\tPrec@1 85.938 (86.967)\n",
      "[160/391]\tTime 0.132 (0.155)\tData 0.038 (0.040)\tLoss 0.4700 (0.4896)\tPrec@1 86.719 (86.796)\n",
      "[170/391]\tTime 0.140 (0.155)\tData 0.029 (0.040)\tLoss 0.4661 (0.4912)\tPrec@1 89.062 (86.769)\n",
      "[180/391]\tTime 0.131 (0.156)\tData 0.046 (0.041)\tLoss 0.4681 (0.4923)\tPrec@1 85.938 (86.792)\n",
      "[190/391]\tTime 0.139 (0.156)\tData 0.043 (0.041)\tLoss 0.4145 (0.4921)\tPrec@1 89.844 (86.825)\n",
      "[200/391]\tTime 0.146 (0.155)\tData 0.024 (0.041)\tLoss 0.5565 (0.4913)\tPrec@1 82.812 (86.804)\n",
      "[210/391]\tTime 0.159 (0.155)\tData 0.038 (0.040)\tLoss 0.4521 (0.4917)\tPrec@1 87.500 (86.767)\n",
      "[220/391]\tTime 0.150 (0.155)\tData 0.032 (0.041)\tLoss 0.5047 (0.4933)\tPrec@1 85.938 (86.715)\n",
      "[230/391]\tTime 0.165 (0.154)\tData 0.033 (0.040)\tLoss 0.4584 (0.4939)\tPrec@1 87.500 (86.719)\n",
      "[240/391]\tTime 0.151 (0.154)\tData 0.029 (0.040)\tLoss 0.5953 (0.4945)\tPrec@1 82.812 (86.673)\n",
      "[250/391]\tTime 0.121 (0.154)\tData 0.043 (0.040)\tLoss 0.5067 (0.4948)\tPrec@1 85.156 (86.657)\n",
      "[260/391]\tTime 0.122 (0.154)\tData 0.037 (0.041)\tLoss 0.5199 (0.4981)\tPrec@1 88.281 (86.545)\n",
      "[270/391]\tTime 0.118 (0.153)\tData 0.038 (0.041)\tLoss 0.6032 (0.4991)\tPrec@1 85.156 (86.531)\n",
      "[280/391]\tTime 0.128 (0.153)\tData 0.020 (0.041)\tLoss 0.5462 (0.5014)\tPrec@1 86.719 (86.488)\n",
      "[290/391]\tTime 0.164 (0.153)\tData 0.034 (0.041)\tLoss 0.6047 (0.5031)\tPrec@1 81.250 (86.445)\n",
      "[300/391]\tTime 0.133 (0.153)\tData 0.026 (0.041)\tLoss 0.5409 (0.5042)\tPrec@1 86.719 (86.392)\n",
      "[310/391]\tTime 0.161 (0.153)\tData 0.038 (0.041)\tLoss 0.5189 (0.5035)\tPrec@1 85.156 (86.405)\n",
      "[320/391]\tTime 0.150 (0.153)\tData 0.035 (0.041)\tLoss 0.4077 (0.5028)\tPrec@1 92.969 (86.436)\n",
      "[330/391]\tTime 0.200 (0.153)\tData 0.081 (0.041)\tLoss 0.5239 (0.5033)\tPrec@1 81.250 (86.414)\n",
      "[340/391]\tTime 0.207 (0.153)\tData 0.034 (0.041)\tLoss 0.5120 (0.5032)\tPrec@1 82.812 (86.423)\n",
      "[350/391]\tTime 0.145 (0.154)\tData 0.039 (0.041)\tLoss 0.4357 (0.5031)\tPrec@1 85.156 (86.394)\n",
      "[360/391]\tTime 0.133 (0.154)\tData 0.035 (0.041)\tLoss 0.5156 (0.5028)\tPrec@1 85.156 (86.405)\n",
      "[370/391]\tTime 0.143 (0.154)\tData 0.032 (0.041)\tLoss 0.6904 (0.5025)\tPrec@1 82.812 (86.413)\n",
      "[380/391]\tTime 0.165 (0.154)\tData 0.044 (0.040)\tLoss 0.5060 (0.5023)\tPrec@1 88.281 (86.456)\n",
      "[390/391]\tTime 0.168 (0.154)\tData 0.036 (0.040)\tLoss 0.4410 (0.5014)\tPrec@1 87.500 (86.482)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.052 (0.052)\tLoss 0.6959 (0.6959)\tPrec@1 82.812 (82.812)\n",
      "Test: [10/79]\tTime 0.032 (0.038)\tLoss 0.5194 (0.7217)\tPrec@1 83.594 (81.534)\n",
      "Test: [20/79]\tTime 0.030 (0.034)\tLoss 0.7525 (0.7200)\tPrec@1 79.688 (81.510)\n",
      "Test: [30/79]\tTime 0.037 (0.034)\tLoss 0.7085 (0.7247)\tPrec@1 85.938 (81.880)\n",
      "Test: [40/79]\tTime 0.040 (0.035)\tLoss 0.9196 (0.7099)\tPrec@1 77.344 (82.279)\n",
      "Test: [50/79]\tTime 0.033 (0.036)\tLoss 0.9708 (0.7007)\tPrec@1 78.906 (82.292)\n",
      "Test: [60/79]\tTime 0.033 (0.037)\tLoss 0.5887 (0.7000)\tPrec@1 85.938 (82.236)\n",
      "Test: [70/79]\tTime 0.034 (0.037)\tLoss 0.6001 (0.7016)\tPrec@1 81.250 (82.075)\n",
      " * Prec@1 82.010\n",
      "\n",
      "===> epoch: 10/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.132 (0.132)\tData 0.060 (0.060)\tLoss 0.4232 (0.4232)\tPrec@1 88.281 (88.281)\n",
      "[10/391]\tTime 0.145 (0.142)\tData 0.033 (0.039)\tLoss 0.5939 (0.4882)\tPrec@1 82.812 (86.861)\n",
      "[20/391]\tTime 0.159 (0.147)\tData 0.020 (0.039)\tLoss 0.4478 (0.5185)\tPrec@1 89.844 (85.640)\n",
      "[30/391]\tTime 0.161 (0.148)\tData 0.035 (0.037)\tLoss 0.4177 (0.5071)\tPrec@1 88.281 (86.013)\n",
      "[40/391]\tTime 0.168 (0.150)\tData 0.039 (0.038)\tLoss 0.3824 (0.5019)\tPrec@1 91.406 (86.242)\n",
      "[50/391]\tTime 0.160 (0.149)\tData 0.043 (0.038)\tLoss 0.5440 (0.4954)\tPrec@1 83.594 (86.535)\n",
      "[60/391]\tTime 0.135 (0.150)\tData 0.035 (0.037)\tLoss 0.5291 (0.4964)\tPrec@1 83.594 (86.488)\n",
      "[70/391]\tTime 0.156 (0.152)\tData 0.037 (0.038)\tLoss 0.4204 (0.4930)\tPrec@1 89.062 (86.532)\n",
      "[80/391]\tTime 0.174 (0.153)\tData 0.050 (0.039)\tLoss 0.4319 (0.4950)\tPrec@1 88.281 (86.400)\n",
      "[90/391]\tTime 0.125 (0.155)\tData 0.034 (0.040)\tLoss 0.4936 (0.4986)\tPrec@1 87.500 (86.298)\n",
      "[100/391]\tTime 0.138 (0.155)\tData 0.036 (0.040)\tLoss 0.4560 (0.5002)\tPrec@1 85.938 (86.231)\n",
      "[110/391]\tTime 0.184 (0.154)\tData 0.032 (0.039)\tLoss 0.4462 (0.4983)\tPrec@1 87.500 (86.325)\n",
      "[120/391]\tTime 0.147 (0.154)\tData 0.030 (0.039)\tLoss 0.4923 (0.4968)\tPrec@1 92.969 (86.480)\n",
      "[130/391]\tTime 0.155 (0.154)\tData 0.035 (0.039)\tLoss 0.4229 (0.4969)\tPrec@1 88.281 (86.432)\n",
      "[140/391]\tTime 0.204 (0.154)\tData 0.054 (0.039)\tLoss 0.4662 (0.4963)\tPrec@1 88.281 (86.436)\n",
      "[150/391]\tTime 0.203 (0.155)\tData 0.062 (0.040)\tLoss 0.5063 (0.4952)\tPrec@1 86.719 (86.522)\n",
      "[160/391]\tTime 0.121 (0.153)\tData 0.060 (0.040)\tLoss 0.5361 (0.4959)\tPrec@1 81.250 (86.500)\n",
      "[170/391]\tTime 0.125 (0.153)\tData 0.029 (0.041)\tLoss 0.4106 (0.4949)\tPrec@1 94.531 (86.568)\n",
      "[180/391]\tTime 0.125 (0.152)\tData 0.035 (0.041)\tLoss 0.5798 (0.4938)\tPrec@1 81.250 (86.650)\n",
      "[190/391]\tTime 0.169 (0.152)\tData 0.057 (0.041)\tLoss 0.3600 (0.4927)\tPrec@1 92.969 (86.719)\n",
      "[200/391]\tTime 0.131 (0.151)\tData 0.029 (0.040)\tLoss 0.5035 (0.4913)\tPrec@1 86.719 (86.742)\n",
      "[210/391]\tTime 0.181 (0.152)\tData 0.053 (0.040)\tLoss 0.4734 (0.4913)\tPrec@1 82.812 (86.726)\n",
      "[220/391]\tTime 0.148 (0.152)\tData 0.033 (0.040)\tLoss 0.5726 (0.4939)\tPrec@1 85.156 (86.602)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[230/391]\tTime 0.132 (0.152)\tData 0.030 (0.040)\tLoss 0.4741 (0.4925)\tPrec@1 88.281 (86.621)\n",
      "[240/391]\tTime 0.155 (0.152)\tData 0.035 (0.041)\tLoss 0.5153 (0.4937)\tPrec@1 82.031 (86.573)\n",
      "[250/391]\tTime 0.175 (0.152)\tData 0.037 (0.040)\tLoss 0.4595 (0.4937)\tPrec@1 88.281 (86.560)\n",
      "[260/391]\tTime 0.128 (0.152)\tData 0.034 (0.040)\tLoss 0.4391 (0.4919)\tPrec@1 89.062 (86.626)\n",
      "[270/391]\tTime 0.137 (0.152)\tData 0.036 (0.040)\tLoss 0.5281 (0.4918)\tPrec@1 84.375 (86.644)\n",
      "[280/391]\tTime 0.149 (0.152)\tData 0.027 (0.040)\tLoss 0.5347 (0.4928)\tPrec@1 86.719 (86.602)\n",
      "[290/391]\tTime 0.148 (0.151)\tData 0.035 (0.040)\tLoss 0.4944 (0.4931)\tPrec@1 86.719 (86.598)\n",
      "[300/391]\tTime 0.131 (0.152)\tData 0.039 (0.040)\tLoss 0.4158 (0.4927)\tPrec@1 90.625 (86.594)\n",
      "[310/391]\tTime 0.138 (0.152)\tData 0.034 (0.040)\tLoss 0.4127 (0.4916)\tPrec@1 85.156 (86.638)\n",
      "[320/391]\tTime 0.165 (0.151)\tData 0.033 (0.040)\tLoss 0.4793 (0.4910)\tPrec@1 87.500 (86.702)\n",
      "[330/391]\tTime 0.138 (0.151)\tData 0.033 (0.040)\tLoss 0.4640 (0.4923)\tPrec@1 87.500 (86.653)\n",
      "[340/391]\tTime 0.143 (0.151)\tData 0.038 (0.040)\tLoss 0.4379 (0.4918)\tPrec@1 88.281 (86.682)\n",
      "[350/391]\tTime 0.132 (0.151)\tData 0.040 (0.040)\tLoss 0.5204 (0.4908)\tPrec@1 85.938 (86.692)\n",
      "[360/391]\tTime 0.163 (0.151)\tData 0.047 (0.040)\tLoss 0.4760 (0.4908)\tPrec@1 89.844 (86.693)\n",
      "[370/391]\tTime 0.169 (0.152)\tData 0.029 (0.040)\tLoss 0.5166 (0.4905)\tPrec@1 87.500 (86.700)\n",
      "[380/391]\tTime 0.126 (0.152)\tData 0.032 (0.040)\tLoss 0.6557 (0.4908)\tPrec@1 84.375 (86.700)\n",
      "[390/391]\tTime 0.092 (0.152)\tData 0.023 (0.040)\tLoss 0.3766 (0.4910)\tPrec@1 88.750 (86.686)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.058 (0.058)\tLoss 0.5554 (0.5554)\tPrec@1 84.375 (84.375)\n",
      "Test: [10/79]\tTime 0.030 (0.035)\tLoss 0.6466 (0.7403)\tPrec@1 81.250 (81.605)\n",
      "Test: [20/79]\tTime 0.028 (0.036)\tLoss 0.5026 (0.7841)\tPrec@1 86.719 (80.766)\n",
      "Test: [30/79]\tTime 0.030 (0.035)\tLoss 0.7854 (0.7829)\tPrec@1 82.812 (81.048)\n",
      "Test: [40/79]\tTime 0.027 (0.037)\tLoss 0.8696 (0.7905)\tPrec@1 72.656 (80.793)\n",
      "Test: [50/79]\tTime 0.027 (0.037)\tLoss 0.8029 (0.7948)\tPrec@1 84.375 (80.836)\n",
      "Test: [60/79]\tTime 0.037 (0.038)\tLoss 0.7053 (0.7881)\tPrec@1 80.469 (80.943)\n",
      "Test: [70/79]\tTime 0.055 (0.039)\tLoss 0.6249 (0.7866)\tPrec@1 82.031 (80.920)\n",
      " * Prec@1 81.010\n",
      "\n",
      "===> epoch: 11/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.132 (0.132)\tData 0.036 (0.036)\tLoss 0.4590 (0.4590)\tPrec@1 88.281 (88.281)\n",
      "[10/391]\tTime 0.153 (0.143)\tData 0.037 (0.040)\tLoss 0.4341 (0.4678)\tPrec@1 87.500 (87.074)\n",
      "[20/391]\tTime 0.136 (0.150)\tData 0.033 (0.039)\tLoss 0.4728 (0.4654)\tPrec@1 89.844 (86.942)\n",
      "[30/391]\tTime 0.165 (0.154)\tData 0.040 (0.041)\tLoss 0.3743 (0.4548)\tPrec@1 90.625 (87.475)\n",
      "[40/391]\tTime 0.102 (0.152)\tData 0.027 (0.040)\tLoss 0.4009 (0.4513)\tPrec@1 89.844 (87.671)\n",
      "[50/391]\tTime 0.157 (0.149)\tData 0.069 (0.040)\tLoss 0.3872 (0.4504)\tPrec@1 91.406 (87.868)\n",
      "[60/391]\tTime 0.134 (0.149)\tData 0.038 (0.040)\tLoss 0.3887 (0.4544)\tPrec@1 85.938 (87.526)\n",
      "[70/391]\tTime 0.117 (0.145)\tData 0.033 (0.039)\tLoss 0.4523 (0.4571)\tPrec@1 85.938 (87.555)\n",
      "[80/391]\tTime 0.128 (0.143)\tData 0.051 (0.039)\tLoss 0.4255 (0.4624)\tPrec@1 90.625 (87.471)\n",
      "[90/391]\tTime 0.142 (0.143)\tData 0.033 (0.039)\tLoss 0.5945 (0.4693)\tPrec@1 85.938 (87.268)\n",
      "[100/391]\tTime 0.134 (0.144)\tData 0.042 (0.039)\tLoss 0.3916 (0.4681)\tPrec@1 89.062 (87.299)\n",
      "[110/391]\tTime 0.143 (0.145)\tData 0.028 (0.039)\tLoss 0.4149 (0.4705)\tPrec@1 89.844 (87.303)\n",
      "[120/391]\tTime 0.147 (0.145)\tData 0.038 (0.039)\tLoss 0.4192 (0.4739)\tPrec@1 88.281 (87.203)\n",
      "[130/391]\tTime 0.173 (0.146)\tData 0.036 (0.039)\tLoss 0.4782 (0.4771)\tPrec@1 86.719 (87.184)\n",
      "[140/391]\tTime 0.135 (0.146)\tData 0.034 (0.039)\tLoss 0.5696 (0.4780)\tPrec@1 84.375 (87.223)\n",
      "[150/391]\tTime 0.144 (0.146)\tData 0.034 (0.039)\tLoss 0.4955 (0.4780)\tPrec@1 89.844 (87.241)\n",
      "[160/391]\tTime 0.140 (0.146)\tData 0.036 (0.039)\tLoss 0.5291 (0.4775)\tPrec@1 84.375 (87.282)\n",
      "[170/391]\tTime 0.140 (0.147)\tData 0.038 (0.039)\tLoss 0.4626 (0.4781)\tPrec@1 88.281 (87.258)\n",
      "[180/391]\tTime 0.171 (0.148)\tData 0.044 (0.039)\tLoss 0.4576 (0.4774)\tPrec@1 89.062 (87.263)\n",
      "[190/391]\tTime 0.188 (0.149)\tData 0.063 (0.039)\tLoss 0.4677 (0.4775)\tPrec@1 90.625 (87.259)\n",
      "[200/391]\tTime 0.146 (0.148)\tData 0.032 (0.039)\tLoss 0.5607 (0.4807)\tPrec@1 85.156 (87.150)\n",
      "[210/391]\tTime 0.164 (0.148)\tData 0.060 (0.039)\tLoss 0.5465 (0.4818)\tPrec@1 84.375 (87.137)\n",
      "[220/391]\tTime 0.140 (0.148)\tData 0.041 (0.039)\tLoss 0.5208 (0.4834)\tPrec@1 87.500 (87.079)\n",
      "[230/391]\tTime 0.142 (0.148)\tData 0.030 (0.039)\tLoss 0.5925 (0.4829)\tPrec@1 84.375 (87.135)\n",
      "[240/391]\tTime 0.143 (0.148)\tData 0.033 (0.039)\tLoss 0.5359 (0.4846)\tPrec@1 85.156 (87.098)\n",
      "[250/391]\tTime 0.170 (0.149)\tData 0.038 (0.039)\tLoss 0.4913 (0.4856)\tPrec@1 89.062 (87.080)\n",
      "[260/391]\tTime 0.134 (0.149)\tData 0.040 (0.040)\tLoss 0.5408 (0.4873)\tPrec@1 85.938 (87.033)\n",
      "[270/391]\tTime 0.152 (0.149)\tData 0.035 (0.040)\tLoss 0.5512 (0.4861)\tPrec@1 84.375 (87.099)\n",
      "[280/391]\tTime 0.143 (0.149)\tData 0.036 (0.040)\tLoss 0.3588 (0.4855)\tPrec@1 89.844 (87.100)\n",
      "[290/391]\tTime 0.128 (0.149)\tData 0.032 (0.040)\tLoss 0.5011 (0.4860)\tPrec@1 85.938 (87.105)\n",
      "[300/391]\tTime 0.138 (0.150)\tData 0.032 (0.040)\tLoss 0.4340 (0.4867)\tPrec@1 88.281 (87.082)\n",
      "[310/391]\tTime 0.126 (0.150)\tData 0.036 (0.040)\tLoss 0.5376 (0.4867)\tPrec@1 86.719 (87.080)\n",
      "[320/391]\tTime 0.122 (0.150)\tData 0.034 (0.039)\tLoss 0.5538 (0.4869)\tPrec@1 84.375 (87.067)\n",
      "[330/391]\tTime 0.133 (0.149)\tData 0.057 (0.039)\tLoss 0.4017 (0.4869)\tPrec@1 89.062 (87.054)\n",
      "[340/391]\tTime 0.137 (0.150)\tData 0.030 (0.040)\tLoss 0.4882 (0.4881)\tPrec@1 90.625 (87.037)\n",
      "[350/391]\tTime 0.106 (0.149)\tData 0.032 (0.039)\tLoss 0.5034 (0.4895)\tPrec@1 88.281 (86.975)\n",
      "[360/391]\tTime 0.129 (0.150)\tData 0.034 (0.040)\tLoss 0.5122 (0.4888)\tPrec@1 87.500 (86.989)\n",
      "[370/391]\tTime 0.126 (0.149)\tData 0.035 (0.040)\tLoss 0.5454 (0.4897)\tPrec@1 86.719 (86.952)\n",
      "[380/391]\tTime 0.132 (0.149)\tData 0.036 (0.039)\tLoss 0.4293 (0.4896)\tPrec@1 87.500 (86.969)\n",
      "[390/391]\tTime 0.140 (0.149)\tData 0.041 (0.039)\tLoss 0.4956 (0.4899)\tPrec@1 87.500 (86.980)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.046 (0.046)\tLoss 0.7040 (0.7040)\tPrec@1 86.719 (86.719)\n",
      "Test: [10/79]\tTime 0.036 (0.038)\tLoss 0.7261 (0.7622)\tPrec@1 81.250 (82.102)\n",
      "Test: [20/79]\tTime 0.033 (0.037)\tLoss 0.6768 (0.8347)\tPrec@1 82.031 (80.469)\n",
      "Test: [30/79]\tTime 0.037 (0.036)\tLoss 0.6257 (0.8044)\tPrec@1 82.031 (80.696)\n",
      "Test: [40/79]\tTime 0.033 (0.036)\tLoss 0.8126 (0.7875)\tPrec@1 78.125 (80.888)\n",
      "Test: [50/79]\tTime 0.032 (0.036)\tLoss 0.8505 (0.7921)\tPrec@1 81.250 (80.744)\n",
      "Test: [60/79]\tTime 0.033 (0.035)\tLoss 0.8496 (0.7985)\tPrec@1 80.469 (80.494)\n",
      "Test: [70/79]\tTime 0.032 (0.034)\tLoss 0.6669 (0.7836)\tPrec@1 79.688 (80.634)\n",
      " * Prec@1 80.430\n",
      "\n",
      "===> epoch: 12/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.162 (0.162)\tData 0.043 (0.043)\tLoss 0.5027 (0.5027)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.135 (0.150)\tData 0.034 (0.038)\tLoss 0.5056 (0.5181)\tPrec@1 86.719 (86.435)\n",
      "[20/391]\tTime 0.165 (0.150)\tData 0.031 (0.038)\tLoss 0.4049 (0.4871)\tPrec@1 90.625 (87.165)\n",
      "[30/391]\tTime 0.145 (0.152)\tData 0.037 (0.039)\tLoss 0.5220 (0.4754)\tPrec@1 85.938 (87.399)\n",
      "[40/391]\tTime 0.154 (0.153)\tData 0.046 (0.038)\tLoss 0.5117 (0.4726)\tPrec@1 86.719 (87.309)\n",
      "[50/391]\tTime 0.124 (0.152)\tData 0.029 (0.039)\tLoss 0.5269 (0.4695)\tPrec@1 84.375 (87.347)\n",
      "[60/391]\tTime 0.126 (0.152)\tData 0.041 (0.039)\tLoss 0.5623 (0.4699)\tPrec@1 82.812 (87.295)\n",
      "[70/391]\tTime 0.198 (0.152)\tData 0.061 (0.039)\tLoss 0.4696 (0.4737)\tPrec@1 88.281 (87.269)\n",
      "[80/391]\tTime 0.132 (0.152)\tData 0.042 (0.039)\tLoss 0.3890 (0.4746)\tPrec@1 89.844 (87.355)\n",
      "[90/391]\tTime 0.113 (0.151)\tData 0.037 (0.040)\tLoss 0.4561 (0.4736)\tPrec@1 89.844 (87.406)\n",
      "[100/391]\tTime 0.141 (0.150)\tData 0.038 (0.039)\tLoss 0.3665 (0.4779)\tPrec@1 92.188 (87.345)\n",
      "[110/391]\tTime 0.162 (0.150)\tData 0.043 (0.040)\tLoss 0.5542 (0.4800)\tPrec@1 88.281 (87.303)\n",
      "[120/391]\tTime 0.127 (0.150)\tData 0.040 (0.040)\tLoss 0.4662 (0.4829)\tPrec@1 87.500 (87.145)\n",
      "[130/391]\tTime 0.143 (0.149)\tData 0.049 (0.040)\tLoss 0.4736 (0.4813)\tPrec@1 87.500 (87.178)\n",
      "[140/391]\tTime 0.128 (0.149)\tData 0.050 (0.040)\tLoss 0.4655 (0.4833)\tPrec@1 87.500 (87.162)\n",
      "[150/391]\tTime 0.130 (0.149)\tData 0.033 (0.039)\tLoss 0.4738 (0.4832)\tPrec@1 85.938 (87.210)\n",
      "[160/391]\tTime 0.174 (0.149)\tData 0.040 (0.039)\tLoss 0.6162 (0.4846)\tPrec@1 81.250 (87.160)\n",
      "[170/391]\tTime 0.105 (0.149)\tData 0.027 (0.039)\tLoss 0.5355 (0.4868)\tPrec@1 89.062 (87.116)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[180/391]\tTime 0.146 (0.150)\tData 0.037 (0.039)\tLoss 0.3988 (0.4876)\tPrec@1 90.625 (87.116)\n",
      "[190/391]\tTime 0.142 (0.150)\tData 0.035 (0.039)\tLoss 0.6400 (0.4890)\tPrec@1 84.375 (87.083)\n",
      "[200/391]\tTime 0.125 (0.150)\tData 0.029 (0.039)\tLoss 0.4797 (0.4892)\tPrec@1 85.156 (87.041)\n",
      "[210/391]\tTime 0.170 (0.150)\tData 0.051 (0.039)\tLoss 0.5224 (0.4901)\tPrec@1 83.594 (86.937)\n",
      "[220/391]\tTime 0.124 (0.150)\tData 0.034 (0.039)\tLoss 0.5396 (0.4909)\tPrec@1 83.594 (86.920)\n",
      "[230/391]\tTime 0.158 (0.150)\tData 0.027 (0.039)\tLoss 0.4610 (0.4931)\tPrec@1 89.844 (86.874)\n",
      "[240/391]\tTime 0.174 (0.150)\tData 0.055 (0.039)\tLoss 0.4001 (0.4935)\tPrec@1 91.406 (86.894)\n",
      "[250/391]\tTime 0.132 (0.151)\tData 0.041 (0.039)\tLoss 0.5223 (0.4945)\tPrec@1 84.375 (86.853)\n",
      "[260/391]\tTime 0.131 (0.150)\tData 0.041 (0.039)\tLoss 0.4829 (0.4942)\tPrec@1 88.281 (86.886)\n",
      "[270/391]\tTime 0.146 (0.149)\tData 0.037 (0.039)\tLoss 0.5497 (0.4934)\tPrec@1 84.375 (86.892)\n",
      "[280/391]\tTime 0.156 (0.150)\tData 0.048 (0.039)\tLoss 0.4680 (0.4926)\tPrec@1 85.156 (86.883)\n",
      "[290/391]\tTime 0.126 (0.149)\tData 0.030 (0.039)\tLoss 0.5559 (0.4912)\tPrec@1 84.375 (86.904)\n",
      "[300/391]\tTime 0.134 (0.148)\tData 0.032 (0.039)\tLoss 0.4558 (0.4910)\tPrec@1 89.844 (86.903)\n",
      "[310/391]\tTime 0.156 (0.148)\tData 0.035 (0.039)\tLoss 0.6167 (0.4929)\tPrec@1 82.812 (86.832)\n",
      "[320/391]\tTime 0.161 (0.148)\tData 0.036 (0.039)\tLoss 0.4178 (0.4919)\tPrec@1 90.625 (86.862)\n",
      "[330/391]\tTime 0.161 (0.148)\tData 0.036 (0.039)\tLoss 0.3997 (0.4917)\tPrec@1 89.844 (86.849)\n",
      "[340/391]\tTime 0.153 (0.148)\tData 0.037 (0.039)\tLoss 0.4309 (0.4916)\tPrec@1 85.938 (86.831)\n",
      "[350/391]\tTime 0.145 (0.149)\tData 0.029 (0.039)\tLoss 0.4737 (0.4924)\tPrec@1 86.719 (86.812)\n",
      "[360/391]\tTime 0.170 (0.149)\tData 0.047 (0.039)\tLoss 0.6263 (0.4926)\tPrec@1 82.812 (86.825)\n",
      "[370/391]\tTime 0.183 (0.149)\tData 0.068 (0.039)\tLoss 0.4568 (0.4929)\tPrec@1 85.156 (86.780)\n",
      "[380/391]\tTime 0.151 (0.149)\tData 0.028 (0.039)\tLoss 0.5242 (0.4936)\tPrec@1 85.938 (86.750)\n",
      "[390/391]\tTime 0.121 (0.149)\tData 0.020 (0.039)\tLoss 0.4793 (0.4946)\tPrec@1 86.250 (86.716)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.041 (0.041)\tLoss 0.5649 (0.5649)\tPrec@1 86.719 (86.719)\n",
      "Test: [10/79]\tTime 0.034 (0.044)\tLoss 0.5813 (0.6088)\tPrec@1 83.594 (84.730)\n",
      "Test: [20/79]\tTime 0.037 (0.046)\tLoss 0.7336 (0.6575)\tPrec@1 82.031 (83.222)\n",
      "Test: [30/79]\tTime 0.038 (0.043)\tLoss 0.6156 (0.6690)\tPrec@1 82.812 (82.989)\n",
      "Test: [40/79]\tTime 0.031 (0.041)\tLoss 0.9336 (0.6581)\tPrec@1 78.125 (83.155)\n",
      "Test: [50/79]\tTime 0.032 (0.039)\tLoss 0.6698 (0.6576)\tPrec@1 84.375 (83.395)\n",
      "Test: [60/79]\tTime 0.033 (0.039)\tLoss 0.6350 (0.6562)\tPrec@1 84.375 (83.594)\n",
      "Test: [70/79]\tTime 0.031 (0.039)\tLoss 0.6175 (0.6556)\tPrec@1 85.156 (83.561)\n",
      " * Prec@1 83.410\n",
      "\n",
      "===> epoch: 13/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.114 (0.114)\tData 0.030 (0.030)\tLoss 0.5136 (0.5136)\tPrec@1 89.062 (89.062)\n",
      "[10/391]\tTime 0.150 (0.136)\tData 0.035 (0.040)\tLoss 0.6138 (0.4936)\tPrec@1 84.375 (87.571)\n",
      "[20/391]\tTime 0.184 (0.145)\tData 0.037 (0.039)\tLoss 0.4412 (0.4911)\tPrec@1 88.281 (86.868)\n",
      "[30/391]\tTime 0.139 (0.149)\tData 0.034 (0.040)\tLoss 0.4901 (0.4926)\tPrec@1 85.938 (86.643)\n",
      "[40/391]\tTime 0.169 (0.150)\tData 0.035 (0.040)\tLoss 0.5492 (0.4982)\tPrec@1 83.594 (86.509)\n",
      "[50/391]\tTime 0.129 (0.150)\tData 0.027 (0.040)\tLoss 0.5237 (0.4940)\tPrec@1 83.594 (86.581)\n",
      "[60/391]\tTime 0.167 (0.148)\tData 0.041 (0.038)\tLoss 0.5136 (0.4975)\tPrec@1 88.281 (86.488)\n",
      "[70/391]\tTime 0.149 (0.147)\tData 0.037 (0.038)\tLoss 0.4955 (0.4969)\tPrec@1 88.281 (86.532)\n",
      "[80/391]\tTime 0.151 (0.149)\tData 0.037 (0.039)\tLoss 0.5474 (0.4964)\tPrec@1 86.719 (86.642)\n",
      "[90/391]\tTime 0.135 (0.148)\tData 0.035 (0.038)\tLoss 0.4972 (0.4969)\tPrec@1 86.719 (86.573)\n",
      "[100/391]\tTime 0.167 (0.149)\tData 0.032 (0.038)\tLoss 0.4650 (0.4999)\tPrec@1 84.375 (86.417)\n",
      "[110/391]\tTime 0.163 (0.149)\tData 0.042 (0.038)\tLoss 0.4805 (0.4975)\tPrec@1 89.062 (86.585)\n",
      "[120/391]\tTime 0.128 (0.148)\tData 0.036 (0.037)\tLoss 0.4466 (0.5002)\tPrec@1 88.281 (86.557)\n",
      "[130/391]\tTime 0.129 (0.147)\tData 0.033 (0.037)\tLoss 0.6029 (0.4981)\tPrec@1 82.031 (86.611)\n",
      "[140/391]\tTime 0.141 (0.148)\tData 0.033 (0.038)\tLoss 0.5383 (0.4955)\tPrec@1 82.031 (86.658)\n",
      "[150/391]\tTime 0.129 (0.148)\tData 0.033 (0.038)\tLoss 0.5495 (0.4924)\tPrec@1 87.500 (86.791)\n",
      "[160/391]\tTime 0.129 (0.147)\tData 0.037 (0.038)\tLoss 0.6421 (0.4952)\tPrec@1 83.594 (86.656)\n",
      "[170/391]\tTime 0.131 (0.147)\tData 0.029 (0.038)\tLoss 0.4613 (0.4962)\tPrec@1 87.500 (86.600)\n",
      "[180/391]\tTime 0.139 (0.148)\tData 0.038 (0.038)\tLoss 0.4285 (0.4959)\tPrec@1 90.625 (86.663)\n",
      "[190/391]\tTime 0.117 (0.147)\tData 0.033 (0.038)\tLoss 0.4687 (0.4961)\tPrec@1 85.156 (86.584)\n",
      "[200/391]\tTime 0.126 (0.146)\tData 0.039 (0.038)\tLoss 0.4122 (0.4972)\tPrec@1 90.625 (86.594)\n",
      "[210/391]\tTime 0.194 (0.146)\tData 0.069 (0.038)\tLoss 0.4537 (0.4951)\tPrec@1 89.844 (86.704)\n",
      "[220/391]\tTime 0.132 (0.146)\tData 0.037 (0.038)\tLoss 0.5056 (0.4947)\tPrec@1 87.500 (86.733)\n",
      "[230/391]\tTime 0.122 (0.147)\tData 0.035 (0.038)\tLoss 0.5558 (0.4932)\tPrec@1 85.156 (86.780)\n",
      "[240/391]\tTime 0.143 (0.147)\tData 0.034 (0.038)\tLoss 0.4412 (0.4922)\tPrec@1 87.500 (86.774)\n",
      "[250/391]\tTime 0.144 (0.147)\tData 0.039 (0.038)\tLoss 0.5006 (0.4913)\tPrec@1 91.406 (86.781)\n",
      "[260/391]\tTime 0.170 (0.147)\tData 0.029 (0.038)\tLoss 0.4646 (0.4921)\tPrec@1 86.719 (86.713)\n",
      "[270/391]\tTime 0.181 (0.147)\tData 0.066 (0.038)\tLoss 0.5011 (0.4925)\tPrec@1 85.938 (86.684)\n",
      "[280/391]\tTime 0.158 (0.148)\tData 0.034 (0.038)\tLoss 0.5925 (0.4947)\tPrec@1 83.594 (86.596)\n",
      "[290/391]\tTime 0.206 (0.148)\tData 0.052 (0.039)\tLoss 0.4205 (0.4948)\tPrec@1 88.281 (86.617)\n",
      "[300/391]\tTime 0.133 (0.148)\tData 0.033 (0.038)\tLoss 0.4718 (0.4951)\tPrec@1 88.281 (86.641)\n",
      "[310/391]\tTime 0.137 (0.148)\tData 0.042 (0.039)\tLoss 0.4473 (0.4938)\tPrec@1 89.844 (86.694)\n",
      "[320/391]\tTime 0.136 (0.148)\tData 0.031 (0.039)\tLoss 0.4231 (0.4943)\tPrec@1 88.281 (86.682)\n",
      "[330/391]\tTime 0.139 (0.148)\tData 0.036 (0.039)\tLoss 0.6170 (0.4951)\tPrec@1 85.156 (86.681)\n",
      "[340/391]\tTime 0.152 (0.148)\tData 0.034 (0.039)\tLoss 0.5431 (0.4962)\tPrec@1 88.281 (86.634)\n",
      "[350/391]\tTime 0.183 (0.148)\tData 0.056 (0.039)\tLoss 0.4868 (0.4958)\tPrec@1 84.375 (86.621)\n",
      "[360/391]\tTime 0.126 (0.148)\tData 0.033 (0.039)\tLoss 0.4513 (0.4956)\tPrec@1 90.625 (86.643)\n",
      "[370/391]\tTime 0.152 (0.148)\tData 0.033 (0.039)\tLoss 0.5076 (0.4960)\tPrec@1 85.938 (86.609)\n",
      "[380/391]\tTime 0.182 (0.148)\tData 0.032 (0.039)\tLoss 0.6599 (0.4962)\tPrec@1 82.031 (86.600)\n",
      "[390/391]\tTime 0.104 (0.148)\tData 0.026 (0.039)\tLoss 0.3976 (0.4960)\tPrec@1 91.250 (86.584)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.045 (0.045)\tLoss 1.0166 (1.0166)\tPrec@1 78.125 (78.125)\n",
      "Test: [10/79]\tTime 0.036 (0.043)\tLoss 0.9036 (0.9551)\tPrec@1 82.031 (78.267)\n",
      "Test: [20/79]\tTime 0.036 (0.042)\tLoss 0.7070 (0.9744)\tPrec@1 84.375 (78.199)\n",
      "Test: [30/79]\tTime 0.031 (0.040)\tLoss 1.0489 (0.9832)\tPrec@1 78.906 (77.974)\n",
      "Test: [40/79]\tTime 0.054 (0.040)\tLoss 1.1407 (0.9877)\tPrec@1 77.344 (77.858)\n",
      "Test: [50/79]\tTime 0.041 (0.040)\tLoss 1.1056 (0.9903)\tPrec@1 71.875 (77.451)\n",
      "Test: [60/79]\tTime 0.036 (0.039)\tLoss 1.0843 (0.9968)\tPrec@1 74.219 (77.126)\n",
      "Test: [70/79]\tTime 0.034 (0.039)\tLoss 0.9154 (0.9955)\tPrec@1 78.125 (77.168)\n",
      " * Prec@1 77.220\n",
      "\n",
      "===> epoch: 14/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.135 (0.135)\tData 0.029 (0.029)\tLoss 0.4572 (0.4572)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.120 (0.144)\tData 0.037 (0.036)\tLoss 0.4247 (0.4625)\tPrec@1 88.281 (88.139)\n",
      "[20/391]\tTime 0.143 (0.146)\tData 0.031 (0.038)\tLoss 0.4189 (0.4455)\tPrec@1 85.938 (88.430)\n",
      "[30/391]\tTime 0.145 (0.145)\tData 0.027 (0.036)\tLoss 0.4611 (0.4559)\tPrec@1 89.844 (88.029)\n",
      "[40/391]\tTime 0.131 (0.148)\tData 0.034 (0.038)\tLoss 0.5473 (0.4594)\tPrec@1 83.594 (87.671)\n",
      "[50/391]\tTime 0.153 (0.148)\tData 0.033 (0.037)\tLoss 0.5286 (0.4588)\tPrec@1 86.719 (87.623)\n",
      "[60/391]\tTime 0.124 (0.146)\tData 0.033 (0.037)\tLoss 0.3916 (0.4603)\tPrec@1 92.188 (87.590)\n",
      "[70/391]\tTime 0.208 (0.145)\tData 0.054 (0.038)\tLoss 0.4213 (0.4666)\tPrec@1 89.062 (87.456)\n",
      "[80/391]\tTime 0.167 (0.148)\tData 0.038 (0.038)\tLoss 0.5823 (0.4706)\tPrec@1 82.031 (87.230)\n",
      "[90/391]\tTime 0.125 (0.148)\tData 0.030 (0.039)\tLoss 0.4024 (0.4675)\tPrec@1 92.188 (87.431)\n",
      "[100/391]\tTime 0.151 (0.146)\tData 0.044 (0.039)\tLoss 0.3986 (0.4651)\tPrec@1 89.844 (87.430)\n",
      "[110/391]\tTime 0.160 (0.146)\tData 0.038 (0.039)\tLoss 0.3710 (0.4622)\tPrec@1 88.281 (87.444)\n",
      "[120/391]\tTime 0.158 (0.147)\tData 0.030 (0.039)\tLoss 0.6154 (0.4662)\tPrec@1 81.250 (87.261)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[130/391]\tTime 0.153 (0.148)\tData 0.033 (0.039)\tLoss 0.3817 (0.4684)\tPrec@1 89.844 (87.232)\n",
      "[140/391]\tTime 0.157 (0.148)\tData 0.039 (0.040)\tLoss 0.4643 (0.4711)\tPrec@1 86.719 (87.134)\n",
      "[150/391]\tTime 0.210 (0.149)\tData 0.062 (0.040)\tLoss 0.3832 (0.4736)\tPrec@1 87.500 (87.034)\n",
      "[160/391]\tTime 0.147 (0.149)\tData 0.050 (0.040)\tLoss 0.6264 (0.4774)\tPrec@1 82.812 (86.952)\n",
      "[170/391]\tTime 0.129 (0.149)\tData 0.028 (0.040)\tLoss 0.4879 (0.4822)\tPrec@1 85.156 (86.796)\n",
      "[180/391]\tTime 0.120 (0.150)\tData 0.035 (0.040)\tLoss 0.4971 (0.4845)\tPrec@1 85.156 (86.758)\n",
      "[190/391]\tTime 0.136 (0.149)\tData 0.036 (0.040)\tLoss 0.6012 (0.4863)\tPrec@1 79.688 (86.739)\n",
      "[200/391]\tTime 0.156 (0.149)\tData 0.030 (0.040)\tLoss 0.4625 (0.4874)\tPrec@1 84.375 (86.703)\n",
      "[210/391]\tTime 0.148 (0.149)\tData 0.034 (0.040)\tLoss 0.4735 (0.4882)\tPrec@1 89.062 (86.685)\n",
      "[220/391]\tTime 0.120 (0.148)\tData 0.033 (0.040)\tLoss 0.4236 (0.4880)\tPrec@1 88.281 (86.683)\n",
      "[230/391]\tTime 0.140 (0.148)\tData 0.057 (0.040)\tLoss 0.3713 (0.4885)\tPrec@1 92.188 (86.688)\n",
      "[240/391]\tTime 0.174 (0.148)\tData 0.060 (0.040)\tLoss 0.5571 (0.4894)\tPrec@1 85.938 (86.670)\n",
      "[250/391]\tTime 0.127 (0.148)\tData 0.034 (0.040)\tLoss 0.4453 (0.4900)\tPrec@1 90.625 (86.660)\n",
      "[260/391]\tTime 0.137 (0.148)\tData 0.032 (0.040)\tLoss 0.4980 (0.4900)\tPrec@1 88.281 (86.695)\n",
      "[270/391]\tTime 0.168 (0.148)\tData 0.028 (0.039)\tLoss 0.4624 (0.4896)\tPrec@1 86.719 (86.707)\n",
      "[280/391]\tTime 0.135 (0.148)\tData 0.032 (0.039)\tLoss 0.5921 (0.4900)\tPrec@1 85.156 (86.747)\n",
      "[290/391]\tTime 0.158 (0.147)\tData 0.039 (0.039)\tLoss 0.4648 (0.4906)\tPrec@1 88.281 (86.684)\n",
      "[300/391]\tTime 0.159 (0.147)\tData 0.048 (0.039)\tLoss 0.3787 (0.4897)\tPrec@1 88.281 (86.719)\n",
      "[310/391]\tTime 0.148 (0.148)\tData 0.032 (0.039)\tLoss 0.5519 (0.4915)\tPrec@1 84.375 (86.671)\n",
      "[320/391]\tTime 0.202 (0.148)\tData 0.042 (0.039)\tLoss 0.6519 (0.4931)\tPrec@1 82.031 (86.638)\n",
      "[330/391]\tTime 0.140 (0.148)\tData 0.033 (0.039)\tLoss 0.4943 (0.4930)\tPrec@1 86.719 (86.657)\n",
      "[340/391]\tTime 0.124 (0.148)\tData 0.036 (0.039)\tLoss 0.3747 (0.4935)\tPrec@1 87.500 (86.659)\n",
      "[350/391]\tTime 0.151 (0.148)\tData 0.040 (0.039)\tLoss 0.5451 (0.4939)\tPrec@1 86.719 (86.676)\n",
      "[360/391]\tTime 0.156 (0.148)\tData 0.041 (0.039)\tLoss 0.4289 (0.4940)\tPrec@1 89.844 (86.686)\n",
      "[370/391]\tTime 0.129 (0.148)\tData 0.042 (0.039)\tLoss 0.4931 (0.4950)\tPrec@1 85.938 (86.639)\n",
      "[380/391]\tTime 0.166 (0.148)\tData 0.037 (0.039)\tLoss 0.4020 (0.4943)\tPrec@1 89.062 (86.663)\n",
      "[390/391]\tTime 0.114 (0.148)\tData 0.025 (0.039)\tLoss 0.4317 (0.4933)\tPrec@1 90.000 (86.684)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.038 (0.038)\tLoss 0.4744 (0.4744)\tPrec@1 87.500 (87.500)\n",
      "Test: [10/79]\tTime 0.032 (0.037)\tLoss 0.5867 (0.6456)\tPrec@1 85.156 (84.233)\n",
      "Test: [20/79]\tTime 0.033 (0.037)\tLoss 0.6164 (0.6641)\tPrec@1 83.594 (82.999)\n",
      "Test: [30/79]\tTime 0.030 (0.037)\tLoss 0.7234 (0.6645)\tPrec@1 81.250 (83.417)\n",
      "Test: [40/79]\tTime 0.040 (0.037)\tLoss 0.7895 (0.6668)\tPrec@1 81.250 (83.556)\n",
      "Test: [50/79]\tTime 0.042 (0.038)\tLoss 0.6178 (0.6624)\tPrec@1 82.812 (83.655)\n",
      "Test: [60/79]\tTime 0.033 (0.039)\tLoss 0.6549 (0.6579)\tPrec@1 84.375 (83.619)\n",
      "Test: [70/79]\tTime 0.043 (0.039)\tLoss 0.5776 (0.6607)\tPrec@1 87.500 (83.484)\n",
      " * Prec@1 83.540\n",
      "\n",
      "===> epoch: 15/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.153 (0.153)\tData 0.049 (0.049)\tLoss 0.3937 (0.3937)\tPrec@1 87.500 (87.500)\n",
      "[10/391]\tTime 0.131 (0.135)\tData 0.036 (0.038)\tLoss 0.5496 (0.4922)\tPrec@1 85.156 (86.861)\n",
      "[20/391]\tTime 0.153 (0.140)\tData 0.033 (0.037)\tLoss 0.4185 (0.4953)\tPrec@1 89.062 (86.533)\n",
      "[30/391]\tTime 0.133 (0.142)\tData 0.029 (0.036)\tLoss 0.4013 (0.4967)\tPrec@1 90.625 (86.618)\n",
      "[40/391]\tTime 0.135 (0.145)\tData 0.039 (0.037)\tLoss 0.5210 (0.4941)\tPrec@1 85.938 (86.681)\n",
      "[50/391]\tTime 0.204 (0.150)\tData 0.064 (0.038)\tLoss 0.5593 (0.4867)\tPrec@1 78.906 (86.780)\n",
      "[60/391]\tTime 0.135 (0.147)\tData 0.030 (0.037)\tLoss 0.4293 (0.4845)\tPrec@1 88.281 (86.744)\n",
      "[70/391]\tTime 0.173 (0.149)\tData 0.052 (0.037)\tLoss 0.4830 (0.4825)\tPrec@1 83.594 (86.587)\n",
      "[80/391]\tTime 0.146 (0.150)\tData 0.033 (0.038)\tLoss 0.6053 (0.4869)\tPrec@1 80.469 (86.535)\n",
      "[90/391]\tTime 0.146 (0.151)\tData 0.034 (0.038)\tLoss 0.4164 (0.4893)\tPrec@1 88.281 (86.607)\n",
      "[100/391]\tTime 0.170 (0.152)\tData 0.034 (0.038)\tLoss 0.4987 (0.4869)\tPrec@1 87.500 (86.765)\n",
      "[110/391]\tTime 0.125 (0.150)\tData 0.041 (0.038)\tLoss 0.4878 (0.4880)\tPrec@1 88.281 (86.754)\n",
      "[120/391]\tTime 0.115 (0.149)\tData 0.032 (0.038)\tLoss 0.4789 (0.4903)\tPrec@1 86.719 (86.674)\n",
      "[130/391]\tTime 0.175 (0.150)\tData 0.039 (0.039)\tLoss 0.4318 (0.4880)\tPrec@1 87.500 (86.725)\n",
      "[140/391]\tTime 0.136 (0.152)\tData 0.034 (0.039)\tLoss 0.4684 (0.4891)\tPrec@1 86.719 (86.763)\n",
      "[150/391]\tTime 0.152 (0.151)\tData 0.027 (0.038)\tLoss 0.5490 (0.4900)\tPrec@1 86.719 (86.776)\n",
      "[160/391]\tTime 0.157 (0.151)\tData 0.039 (0.038)\tLoss 0.5597 (0.4921)\tPrec@1 81.250 (86.743)\n",
      "[170/391]\tTime 0.162 (0.151)\tData 0.038 (0.038)\tLoss 0.4444 (0.4941)\tPrec@1 89.844 (86.705)\n",
      "[180/391]\tTime 0.131 (0.150)\tData 0.029 (0.038)\tLoss 0.5112 (0.4931)\tPrec@1 85.938 (86.788)\n",
      "[190/391]\tTime 0.175 (0.150)\tData 0.055 (0.038)\tLoss 0.4794 (0.4943)\tPrec@1 88.281 (86.780)\n",
      "[200/391]\tTime 0.146 (0.151)\tData 0.034 (0.038)\tLoss 0.4663 (0.4930)\tPrec@1 85.938 (86.828)\n",
      "[210/391]\tTime 0.180 (0.151)\tData 0.043 (0.038)\tLoss 0.3922 (0.4917)\tPrec@1 91.406 (86.856)\n",
      "[220/391]\tTime 0.144 (0.150)\tData 0.033 (0.038)\tLoss 0.4103 (0.4895)\tPrec@1 89.844 (86.899)\n",
      "[230/391]\tTime 0.149 (0.150)\tData 0.031 (0.038)\tLoss 0.5083 (0.4876)\tPrec@1 87.500 (86.972)\n",
      "[240/391]\tTime 0.142 (0.150)\tData 0.027 (0.038)\tLoss 0.4593 (0.4880)\tPrec@1 87.500 (86.933)\n",
      "[250/391]\tTime 0.160 (0.150)\tData 0.035 (0.038)\tLoss 0.5557 (0.4886)\tPrec@1 85.938 (86.934)\n",
      "[260/391]\tTime 0.178 (0.151)\tData 0.048 (0.038)\tLoss 0.3597 (0.4910)\tPrec@1 91.406 (86.871)\n",
      "[270/391]\tTime 0.132 (0.150)\tData 0.036 (0.038)\tLoss 0.5447 (0.4917)\tPrec@1 85.938 (86.846)\n",
      "[280/391]\tTime 0.156 (0.150)\tData 0.033 (0.038)\tLoss 0.4881 (0.4906)\tPrec@1 82.031 (86.866)\n",
      "[290/391]\tTime 0.142 (0.149)\tData 0.030 (0.038)\tLoss 0.5620 (0.4903)\tPrec@1 82.812 (86.883)\n",
      "[300/391]\tTime 0.126 (0.150)\tData 0.036 (0.038)\tLoss 0.5169 (0.4893)\tPrec@1 82.031 (86.887)\n",
      "[310/391]\tTime 0.127 (0.149)\tData 0.035 (0.039)\tLoss 0.5371 (0.4904)\tPrec@1 83.594 (86.804)\n",
      "[320/391]\tTime 0.155 (0.149)\tData 0.060 (0.039)\tLoss 0.3700 (0.4906)\tPrec@1 92.188 (86.789)\n",
      "[330/391]\tTime 0.181 (0.149)\tData 0.064 (0.039)\tLoss 0.5709 (0.4906)\tPrec@1 85.156 (86.816)\n",
      "[340/391]\tTime 0.156 (0.149)\tData 0.033 (0.039)\tLoss 0.4057 (0.4918)\tPrec@1 88.281 (86.787)\n",
      "[350/391]\tTime 0.154 (0.149)\tData 0.050 (0.039)\tLoss 0.4211 (0.4920)\tPrec@1 87.500 (86.770)\n",
      "[360/391]\tTime 0.154 (0.149)\tData 0.038 (0.039)\tLoss 0.5143 (0.4925)\tPrec@1 85.156 (86.773)\n",
      "[370/391]\tTime 0.190 (0.149)\tData 0.061 (0.039)\tLoss 0.4517 (0.4937)\tPrec@1 89.844 (86.750)\n",
      "[380/391]\tTime 0.141 (0.149)\tData 0.037 (0.039)\tLoss 0.4871 (0.4936)\tPrec@1 87.500 (86.758)\n",
      "[390/391]\tTime 0.147 (0.149)\tData 0.050 (0.039)\tLoss 0.4068 (0.4930)\tPrec@1 87.500 (86.784)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.069 (0.069)\tLoss 1.3985 (1.3985)\tPrec@1 70.312 (70.312)\n",
      "Test: [10/79]\tTime 0.037 (0.043)\tLoss 1.4588 (1.5370)\tPrec@1 73.438 (69.602)\n",
      "Test: [20/79]\tTime 0.037 (0.042)\tLoss 1.7973 (1.6004)\tPrec@1 64.062 (69.234)\n",
      "Test: [30/79]\tTime 0.036 (0.042)\tLoss 1.2562 (1.5413)\tPrec@1 74.219 (69.758)\n",
      "Test: [40/79]\tTime 0.041 (0.042)\tLoss 1.8458 (1.5318)\tPrec@1 67.969 (69.741)\n",
      "Test: [50/79]\tTime 0.035 (0.041)\tLoss 1.9348 (1.5186)\tPrec@1 64.844 (69.868)\n",
      "Test: [60/79]\tTime 0.036 (0.041)\tLoss 1.5208 (1.4965)\tPrec@1 73.438 (70.146)\n",
      "Test: [70/79]\tTime 0.036 (0.041)\tLoss 1.1692 (1.5061)\tPrec@1 76.562 (69.960)\n",
      " * Prec@1 70.150\n",
      "\n",
      "===> epoch: 16/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.136 (0.136)\tData 0.028 (0.028)\tLoss 0.4188 (0.4188)\tPrec@1 89.062 (89.062)\n",
      "[10/391]\tTime 0.110 (0.137)\tData 0.050 (0.034)\tLoss 0.4053 (0.4381)\tPrec@1 90.625 (88.565)\n",
      "[20/391]\tTime 0.140 (0.150)\tData 0.035 (0.040)\tLoss 0.5132 (0.4620)\tPrec@1 85.938 (87.649)\n",
      "[30/391]\tTime 0.165 (0.154)\tData 0.042 (0.040)\tLoss 0.5497 (0.4796)\tPrec@1 85.938 (87.097)\n",
      "[40/391]\tTime 0.176 (0.155)\tData 0.055 (0.040)\tLoss 0.6459 (0.4845)\tPrec@1 81.250 (86.833)\n",
      "[50/391]\tTime 0.155 (0.154)\tData 0.039 (0.041)\tLoss 0.5636 (0.4877)\tPrec@1 84.375 (86.795)\n",
      "[60/391]\tTime 0.161 (0.155)\tData 0.038 (0.041)\tLoss 0.4234 (0.4891)\tPrec@1 86.719 (86.565)\n",
      "[70/391]\tTime 0.151 (0.155)\tData 0.035 (0.040)\tLoss 0.4791 (0.4949)\tPrec@1 86.719 (86.389)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[80/391]\tTime 0.186 (0.157)\tData 0.059 (0.040)\tLoss 0.5290 (0.4966)\tPrec@1 87.500 (86.449)\n",
      "[90/391]\tTime 0.161 (0.156)\tData 0.046 (0.039)\tLoss 0.4412 (0.4979)\tPrec@1 88.281 (86.444)\n",
      "[100/391]\tTime 0.157 (0.155)\tData 0.045 (0.040)\tLoss 0.4880 (0.4980)\tPrec@1 84.375 (86.417)\n",
      "[110/391]\tTime 0.133 (0.154)\tData 0.028 (0.040)\tLoss 0.4616 (0.4964)\tPrec@1 89.062 (86.508)\n",
      "[120/391]\tTime 0.190 (0.155)\tData 0.061 (0.040)\tLoss 0.4905 (0.4960)\tPrec@1 85.156 (86.493)\n",
      "[130/391]\tTime 0.142 (0.154)\tData 0.053 (0.040)\tLoss 0.3859 (0.4935)\tPrec@1 91.406 (86.594)\n",
      "[140/391]\tTime 0.155 (0.154)\tData 0.042 (0.040)\tLoss 0.4751 (0.4930)\tPrec@1 86.719 (86.553)\n",
      "[150/391]\tTime 0.156 (0.154)\tData 0.035 (0.040)\tLoss 0.5164 (0.4920)\tPrec@1 85.938 (86.657)\n",
      "[160/391]\tTime 0.157 (0.154)\tData 0.037 (0.040)\tLoss 0.5244 (0.4928)\tPrec@1 83.594 (86.607)\n",
      "[170/391]\tTime 0.196 (0.153)\tData 0.070 (0.040)\tLoss 0.4985 (0.4945)\tPrec@1 87.500 (86.518)\n",
      "[180/391]\tTime 0.187 (0.152)\tData 0.037 (0.040)\tLoss 0.5681 (0.4947)\tPrec@1 83.594 (86.572)\n",
      "[190/391]\tTime 0.182 (0.153)\tData 0.049 (0.040)\tLoss 0.3289 (0.4912)\tPrec@1 91.406 (86.666)\n",
      "[200/391]\tTime 0.127 (0.152)\tData 0.027 (0.040)\tLoss 0.4910 (0.4930)\tPrec@1 89.844 (86.594)\n",
      "[210/391]\tTime 0.118 (0.151)\tData 0.035 (0.040)\tLoss 0.4013 (0.4930)\tPrec@1 91.406 (86.589)\n",
      "[220/391]\tTime 0.114 (0.150)\tData 0.034 (0.040)\tLoss 0.4816 (0.4936)\tPrec@1 86.719 (86.584)\n",
      "[230/391]\tTime 0.118 (0.149)\tData 0.033 (0.040)\tLoss 0.5309 (0.4934)\tPrec@1 83.594 (86.594)\n",
      "[240/391]\tTime 0.121 (0.149)\tData 0.028 (0.040)\tLoss 0.4181 (0.4937)\tPrec@1 85.156 (86.566)\n",
      "[250/391]\tTime 0.158 (0.149)\tData 0.043 (0.040)\tLoss 0.4445 (0.4939)\tPrec@1 84.375 (86.541)\n",
      "[260/391]\tTime 0.149 (0.150)\tData 0.030 (0.040)\tLoss 0.4678 (0.4936)\tPrec@1 85.938 (86.551)\n",
      "[270/391]\tTime 0.153 (0.149)\tData 0.031 (0.039)\tLoss 0.4176 (0.4943)\tPrec@1 87.500 (86.528)\n",
      "[280/391]\tTime 0.132 (0.150)\tData 0.045 (0.040)\tLoss 0.4865 (0.4949)\tPrec@1 87.500 (86.521)\n",
      "[290/391]\tTime 0.158 (0.150)\tData 0.034 (0.039)\tLoss 0.5072 (0.4941)\tPrec@1 84.375 (86.550)\n",
      "[300/391]\tTime 0.145 (0.150)\tData 0.040 (0.039)\tLoss 0.5126 (0.4923)\tPrec@1 85.938 (86.631)\n",
      "[310/391]\tTime 0.122 (0.150)\tData 0.038 (0.039)\tLoss 0.5728 (0.4932)\tPrec@1 82.812 (86.571)\n",
      "[320/391]\tTime 0.110 (0.150)\tData 0.034 (0.039)\tLoss 0.5233 (0.4928)\tPrec@1 84.375 (86.587)\n",
      "[330/391]\tTime 0.113 (0.150)\tData 0.037 (0.039)\tLoss 0.4205 (0.4912)\tPrec@1 87.500 (86.631)\n",
      "[340/391]\tTime 0.187 (0.149)\tData 0.053 (0.039)\tLoss 0.5647 (0.4914)\tPrec@1 85.938 (86.606)\n",
      "[350/391]\tTime 0.133 (0.149)\tData 0.031 (0.039)\tLoss 0.4390 (0.4918)\tPrec@1 87.500 (86.592)\n",
      "[360/391]\tTime 0.160 (0.150)\tData 0.036 (0.039)\tLoss 0.4432 (0.4920)\tPrec@1 89.062 (86.606)\n",
      "[370/391]\tTime 0.141 (0.150)\tData 0.038 (0.039)\tLoss 0.4195 (0.4921)\tPrec@1 86.719 (86.578)\n",
      "[380/391]\tTime 0.202 (0.150)\tData 0.037 (0.039)\tLoss 0.5161 (0.4935)\tPrec@1 85.938 (86.536)\n",
      "[390/391]\tTime 0.104 (0.150)\tData 0.037 (0.039)\tLoss 0.4525 (0.4929)\tPrec@1 88.750 (86.556)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.038 (0.038)\tLoss 0.7209 (0.7209)\tPrec@1 82.812 (82.812)\n",
      "Test: [10/79]\tTime 0.051 (0.041)\tLoss 0.5966 (0.6347)\tPrec@1 82.812 (83.665)\n",
      "Test: [20/79]\tTime 0.040 (0.044)\tLoss 0.5430 (0.6520)\tPrec@1 86.719 (83.743)\n",
      "Test: [30/79]\tTime 0.038 (0.043)\tLoss 0.6781 (0.6528)\tPrec@1 83.594 (83.896)\n",
      "Test: [40/79]\tTime 0.049 (0.043)\tLoss 0.5780 (0.6678)\tPrec@1 80.469 (83.975)\n",
      "Test: [50/79]\tTime 0.026 (0.042)\tLoss 0.6966 (0.6659)\tPrec@1 84.375 (84.007)\n",
      "Test: [60/79]\tTime 0.038 (0.041)\tLoss 0.7827 (0.6742)\tPrec@1 82.031 (83.837)\n",
      "Test: [70/79]\tTime 0.068 (0.041)\tLoss 0.5588 (0.6788)\tPrec@1 86.719 (83.715)\n",
      " * Prec@1 83.740\n",
      "\n",
      "===> epoch: 17/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.178 (0.178)\tData 0.053 (0.053)\tLoss 0.4868 (0.4868)\tPrec@1 86.719 (86.719)\n",
      "[10/391]\tTime 0.142 (0.146)\tData 0.034 (0.036)\tLoss 0.5482 (0.4999)\tPrec@1 83.594 (85.653)\n",
      "[20/391]\tTime 0.131 (0.146)\tData 0.034 (0.036)\tLoss 0.4693 (0.4764)\tPrec@1 89.062 (86.942)\n",
      "[30/391]\tTime 0.126 (0.147)\tData 0.032 (0.036)\tLoss 0.4634 (0.4775)\tPrec@1 90.625 (87.223)\n",
      "[40/391]\tTime 0.161 (0.151)\tData 0.034 (0.038)\tLoss 0.5078 (0.4716)\tPrec@1 83.594 (87.367)\n",
      "[50/391]\tTime 0.155 (0.153)\tData 0.043 (0.039)\tLoss 0.3505 (0.4654)\tPrec@1 92.969 (87.638)\n",
      "[60/391]\tTime 0.096 (0.150)\tData 0.029 (0.038)\tLoss 0.4244 (0.4592)\tPrec@1 88.281 (87.999)\n",
      "[70/391]\tTime 0.128 (0.148)\tData 0.036 (0.039)\tLoss 0.6055 (0.4637)\tPrec@1 82.031 (87.720)\n",
      "[80/391]\tTime 0.115 (0.145)\tData 0.033 (0.039)\tLoss 0.6824 (0.4647)\tPrec@1 80.469 (87.654)\n",
      "[90/391]\tTime 0.162 (0.146)\tData 0.049 (0.039)\tLoss 0.5005 (0.4665)\tPrec@1 85.156 (87.569)\n",
      "[100/391]\tTime 0.149 (0.146)\tData 0.042 (0.039)\tLoss 0.3300 (0.4646)\tPrec@1 92.188 (87.631)\n",
      "[110/391]\tTime 0.126 (0.146)\tData 0.035 (0.039)\tLoss 0.4331 (0.4636)\tPrec@1 87.500 (87.683)\n",
      "[120/391]\tTime 0.130 (0.145)\tData 0.049 (0.039)\tLoss 0.5124 (0.4627)\tPrec@1 84.375 (87.694)\n",
      "[130/391]\tTime 0.159 (0.144)\tData 0.033 (0.039)\tLoss 0.5950 (0.4625)\tPrec@1 83.594 (87.697)\n",
      "[140/391]\tTime 0.152 (0.145)\tData 0.038 (0.039)\tLoss 0.4337 (0.4646)\tPrec@1 92.188 (87.683)\n",
      "[150/391]\tTime 0.151 (0.145)\tData 0.046 (0.039)\tLoss 0.5648 (0.4682)\tPrec@1 84.375 (87.567)\n",
      "[160/391]\tTime 0.154 (0.145)\tData 0.034 (0.039)\tLoss 0.4205 (0.4696)\tPrec@1 85.156 (87.432)\n",
      "[170/391]\tTime 0.135 (0.146)\tData 0.036 (0.039)\tLoss 0.5233 (0.4704)\tPrec@1 86.719 (87.431)\n",
      "[180/391]\tTime 0.162 (0.147)\tData 0.044 (0.039)\tLoss 0.6207 (0.4744)\tPrec@1 85.156 (87.336)\n",
      "[190/391]\tTime 0.139 (0.147)\tData 0.037 (0.039)\tLoss 0.4347 (0.4755)\tPrec@1 89.062 (87.300)\n",
      "[200/391]\tTime 0.176 (0.147)\tData 0.049 (0.039)\tLoss 0.6003 (0.4767)\tPrec@1 83.594 (87.263)\n",
      "[210/391]\tTime 0.112 (0.147)\tData 0.027 (0.039)\tLoss 0.4161 (0.4777)\tPrec@1 88.281 (87.245)\n",
      "[220/391]\tTime 0.118 (0.147)\tData 0.037 (0.039)\tLoss 0.4757 (0.4784)\tPrec@1 85.938 (87.150)\n",
      "[230/391]\tTime 0.144 (0.147)\tData 0.046 (0.039)\tLoss 0.5648 (0.4796)\tPrec@1 86.719 (87.114)\n",
      "[240/391]\tTime 0.126 (0.146)\tData 0.028 (0.039)\tLoss 0.5627 (0.4817)\tPrec@1 87.500 (87.075)\n",
      "[250/391]\tTime 0.223 (0.147)\tData 0.074 (0.039)\tLoss 0.5549 (0.4832)\tPrec@1 84.375 (87.046)\n",
      "[260/391]\tTime 0.146 (0.147)\tData 0.039 (0.039)\tLoss 0.5339 (0.4850)\tPrec@1 84.375 (86.970)\n",
      "[270/391]\tTime 0.107 (0.147)\tData 0.028 (0.039)\tLoss 0.5829 (0.4847)\tPrec@1 84.375 (86.987)\n",
      "[280/391]\tTime 0.125 (0.147)\tData 0.034 (0.039)\tLoss 0.4554 (0.4841)\tPrec@1 90.625 (87.022)\n",
      "[290/391]\tTime 0.163 (0.147)\tData 0.034 (0.039)\tLoss 0.5217 (0.4843)\tPrec@1 86.719 (87.014)\n",
      "[300/391]\tTime 0.116 (0.147)\tData 0.027 (0.039)\tLoss 0.4259 (0.4845)\tPrec@1 89.844 (86.994)\n",
      "[310/391]\tTime 0.132 (0.147)\tData 0.039 (0.039)\tLoss 0.5053 (0.4850)\tPrec@1 85.156 (86.993)\n",
      "[320/391]\tTime 0.202 (0.148)\tData 0.070 (0.039)\tLoss 0.4937 (0.4850)\tPrec@1 85.938 (86.969)\n",
      "[330/391]\tTime 0.177 (0.148)\tData 0.044 (0.039)\tLoss 0.4298 (0.4851)\tPrec@1 87.500 (86.974)\n",
      "[340/391]\tTime 0.132 (0.149)\tData 0.033 (0.039)\tLoss 0.5635 (0.4857)\tPrec@1 83.594 (86.959)\n",
      "[350/391]\tTime 0.149 (0.149)\tData 0.026 (0.039)\tLoss 0.4620 (0.4852)\tPrec@1 85.938 (86.968)\n",
      "[360/391]\tTime 0.153 (0.149)\tData 0.033 (0.039)\tLoss 0.4934 (0.4853)\tPrec@1 87.500 (86.952)\n",
      "[370/391]\tTime 0.166 (0.149)\tData 0.048 (0.039)\tLoss 0.4255 (0.4857)\tPrec@1 88.281 (86.944)\n",
      "[380/391]\tTime 0.105 (0.149)\tData 0.030 (0.039)\tLoss 0.4236 (0.4854)\tPrec@1 86.719 (86.948)\n",
      "[390/391]\tTime 0.118 (0.149)\tData 0.022 (0.039)\tLoss 0.4335 (0.4860)\tPrec@1 90.000 (86.926)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.042 (0.042)\tLoss 1.3322 (1.3322)\tPrec@1 75.781 (75.781)\n",
      "Test: [10/79]\tTime 0.032 (0.034)\tLoss 0.9855 (1.2461)\tPrec@1 79.688 (75.071)\n",
      "Test: [20/79]\tTime 0.034 (0.037)\tLoss 1.5551 (1.3053)\tPrec@1 75.781 (74.330)\n",
      "Test: [30/79]\tTime 0.043 (0.038)\tLoss 0.9851 (1.2957)\tPrec@1 76.562 (74.471)\n",
      "Test: [40/79]\tTime 0.031 (0.037)\tLoss 1.4918 (1.2809)\tPrec@1 67.188 (74.543)\n",
      "Test: [50/79]\tTime 0.027 (0.039)\tLoss 1.3200 (1.2728)\tPrec@1 71.094 (74.571)\n",
      "Test: [60/79]\tTime 0.032 (0.038)\tLoss 1.0123 (1.2614)\tPrec@1 75.781 (74.654)\n",
      "Test: [70/79]\tTime 0.029 (0.039)\tLoss 1.1218 (1.2614)\tPrec@1 72.656 (74.384)\n",
      " * Prec@1 74.420\n",
      "\n",
      "===> epoch: 18/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.137 (0.137)\tData 0.045 (0.045)\tLoss 0.6282 (0.6282)\tPrec@1 80.469 (80.469)\n",
      "[10/391]\tTime 0.111 (0.130)\tData 0.031 (0.038)\tLoss 0.5592 (0.5396)\tPrec@1 84.375 (86.080)\n",
      "[20/391]\tTime 0.166 (0.133)\tData 0.033 (0.037)\tLoss 0.4415 (0.5002)\tPrec@1 90.625 (87.016)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[30/391]\tTime 0.155 (0.137)\tData 0.030 (0.036)\tLoss 0.5720 (0.4981)\tPrec@1 82.812 (86.971)\n",
      "[40/391]\tTime 0.147 (0.141)\tData 0.033 (0.037)\tLoss 0.4680 (0.4912)\tPrec@1 86.719 (87.119)\n",
      "[50/391]\tTime 0.130 (0.142)\tData 0.039 (0.037)\tLoss 0.4059 (0.4894)\tPrec@1 92.188 (87.102)\n",
      "[60/391]\tTime 0.167 (0.144)\tData 0.030 (0.037)\tLoss 0.5474 (0.4941)\tPrec@1 86.719 (86.949)\n",
      "[70/391]\tTime 0.139 (0.145)\tData 0.037 (0.037)\tLoss 0.5426 (0.4991)\tPrec@1 83.594 (86.708)\n",
      "[80/391]\tTime 0.149 (0.146)\tData 0.034 (0.037)\tLoss 0.4590 (0.4954)\tPrec@1 89.062 (86.873)\n",
      "[90/391]\tTime 0.138 (0.146)\tData 0.027 (0.037)\tLoss 0.4809 (0.4939)\tPrec@1 85.156 (86.925)\n",
      "[100/391]\tTime 0.157 (0.148)\tData 0.034 (0.037)\tLoss 0.5258 (0.4889)\tPrec@1 85.938 (87.044)\n",
      "[110/391]\tTime 0.163 (0.148)\tData 0.033 (0.037)\tLoss 0.3962 (0.4889)\tPrec@1 89.844 (86.923)\n",
      "[120/391]\tTime 0.102 (0.148)\tData 0.029 (0.037)\tLoss 0.4216 (0.4859)\tPrec@1 89.062 (87.087)\n",
      "[130/391]\tTime 0.175 (0.148)\tData 0.039 (0.037)\tLoss 0.4581 (0.4834)\tPrec@1 87.500 (87.172)\n",
      "[140/391]\tTime 0.133 (0.147)\tData 0.039 (0.038)\tLoss 0.4898 (0.4857)\tPrec@1 86.719 (87.062)\n",
      "[150/391]\tTime 0.177 (0.149)\tData 0.027 (0.038)\tLoss 0.3998 (0.4847)\tPrec@1 92.188 (87.081)\n",
      "[160/391]\tTime 0.128 (0.149)\tData 0.034 (0.038)\tLoss 0.4958 (0.4844)\tPrec@1 86.719 (87.107)\n",
      "[170/391]\tTime 0.157 (0.150)\tData 0.033 (0.038)\tLoss 0.5957 (0.4853)\tPrec@1 79.688 (87.066)\n",
      "[180/391]\tTime 0.162 (0.150)\tData 0.036 (0.038)\tLoss 0.5608 (0.4851)\tPrec@1 83.594 (87.047)\n",
      "[190/391]\tTime 0.151 (0.150)\tData 0.032 (0.038)\tLoss 0.4667 (0.4876)\tPrec@1 89.844 (86.940)\n",
      "[200/391]\tTime 0.155 (0.150)\tData 0.025 (0.038)\tLoss 0.5817 (0.4892)\tPrec@1 82.812 (86.901)\n",
      "[210/391]\tTime 0.147 (0.149)\tData 0.032 (0.038)\tLoss 0.4441 (0.4881)\tPrec@1 88.281 (86.874)\n",
      "[220/391]\tTime 0.160 (0.150)\tData 0.051 (0.038)\tLoss 0.6093 (0.4867)\tPrec@1 85.156 (86.924)\n",
      "[230/391]\tTime 0.208 (0.150)\tData 0.063 (0.038)\tLoss 0.4446 (0.4862)\tPrec@1 86.719 (86.908)\n",
      "[240/391]\tTime 0.172 (0.150)\tData 0.038 (0.038)\tLoss 0.4836 (0.4874)\tPrec@1 87.500 (86.884)\n",
      "[250/391]\tTime 0.164 (0.150)\tData 0.040 (0.038)\tLoss 0.4833 (0.4891)\tPrec@1 87.500 (86.843)\n",
      "[260/391]\tTime 0.160 (0.151)\tData 0.044 (0.038)\tLoss 0.4002 (0.4909)\tPrec@1 89.844 (86.764)\n",
      "[270/391]\tTime 0.138 (0.151)\tData 0.039 (0.038)\tLoss 0.5107 (0.4933)\tPrec@1 89.062 (86.725)\n",
      "[280/391]\tTime 0.158 (0.151)\tData 0.033 (0.038)\tLoss 0.6187 (0.4945)\tPrec@1 81.250 (86.677)\n",
      "[290/391]\tTime 0.178 (0.151)\tData 0.036 (0.038)\tLoss 0.5092 (0.4954)\tPrec@1 89.062 (86.657)\n",
      "[300/391]\tTime 0.158 (0.151)\tData 0.053 (0.038)\tLoss 0.5617 (0.4956)\tPrec@1 84.375 (86.620)\n",
      "[310/391]\tTime 0.149 (0.151)\tData 0.034 (0.038)\tLoss 0.4512 (0.4971)\tPrec@1 89.844 (86.578)\n",
      "[320/391]\tTime 0.127 (0.151)\tData 0.042 (0.039)\tLoss 0.5637 (0.4967)\tPrec@1 86.719 (86.561)\n",
      "[330/391]\tTime 0.148 (0.150)\tData 0.036 (0.039)\tLoss 0.4698 (0.4957)\tPrec@1 85.938 (86.582)\n",
      "[340/391]\tTime 0.159 (0.150)\tData 0.031 (0.039)\tLoss 0.3968 (0.4954)\tPrec@1 88.281 (86.595)\n",
      "[350/391]\tTime 0.125 (0.150)\tData 0.027 (0.039)\tLoss 0.4448 (0.4952)\tPrec@1 90.625 (86.570)\n",
      "[360/391]\tTime 0.195 (0.150)\tData 0.054 (0.038)\tLoss 0.3725 (0.4950)\tPrec@1 91.406 (86.593)\n",
      "[370/391]\tTime 0.150 (0.150)\tData 0.036 (0.039)\tLoss 0.4210 (0.4950)\tPrec@1 88.281 (86.601)\n",
      "[380/391]\tTime 0.135 (0.150)\tData 0.027 (0.038)\tLoss 0.5057 (0.4957)\tPrec@1 88.281 (86.592)\n",
      "[390/391]\tTime 0.127 (0.151)\tData 0.026 (0.038)\tLoss 0.2950 (0.4952)\tPrec@1 93.750 (86.596)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.042 (0.042)\tLoss 1.3813 (1.3813)\tPrec@1 71.875 (71.875)\n",
      "Test: [10/79]\tTime 0.040 (0.044)\tLoss 1.4314 (1.4006)\tPrec@1 68.750 (69.815)\n",
      "Test: [20/79]\tTime 0.034 (0.042)\tLoss 1.6481 (1.4498)\tPrec@1 67.969 (70.015)\n",
      "Test: [30/79]\tTime 0.034 (0.040)\tLoss 1.4373 (1.4367)\tPrec@1 61.719 (69.632)\n",
      "Test: [40/79]\tTime 0.042 (0.039)\tLoss 1.6506 (1.4054)\tPrec@1 66.406 (70.179)\n",
      "Test: [50/79]\tTime 0.047 (0.039)\tLoss 1.5869 (1.4008)\tPrec@1 68.750 (70.374)\n",
      "Test: [60/79]\tTime 0.044 (0.038)\tLoss 1.5392 (1.3872)\tPrec@1 73.438 (70.799)\n",
      "Test: [70/79]\tTime 0.028 (0.037)\tLoss 1.5619 (1.3895)\tPrec@1 66.406 (70.643)\n",
      " * Prec@1 70.790\n",
      "\n",
      "===> epoch: 19/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.150 (0.150)\tData 0.035 (0.035)\tLoss 0.5696 (0.5696)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.133 (0.153)\tData 0.037 (0.044)\tLoss 0.4043 (0.4243)\tPrec@1 89.062 (88.636)\n",
      "[20/391]\tTime 0.165 (0.151)\tData 0.059 (0.043)\tLoss 0.4178 (0.4567)\tPrec@1 89.062 (87.723)\n",
      "[30/391]\tTime 0.158 (0.151)\tData 0.047 (0.044)\tLoss 0.3293 (0.4469)\tPrec@1 91.406 (87.853)\n",
      "[40/391]\tTime 0.157 (0.148)\tData 0.035 (0.042)\tLoss 0.4395 (0.4530)\tPrec@1 86.719 (87.500)\n",
      "[50/391]\tTime 0.169 (0.148)\tData 0.033 (0.040)\tLoss 0.5465 (0.4577)\tPrec@1 86.719 (87.653)\n",
      "[60/391]\tTime 0.148 (0.148)\tData 0.032 (0.039)\tLoss 0.5155 (0.4647)\tPrec@1 84.375 (87.359)\n",
      "[70/391]\tTime 0.156 (0.149)\tData 0.033 (0.039)\tLoss 0.5462 (0.4643)\tPrec@1 86.719 (87.456)\n",
      "[80/391]\tTime 0.160 (0.150)\tData 0.025 (0.039)\tLoss 0.4458 (0.4629)\tPrec@1 88.281 (87.490)\n",
      "[90/391]\tTime 0.134 (0.150)\tData 0.037 (0.039)\tLoss 0.4736 (0.4675)\tPrec@1 84.375 (87.294)\n",
      "[100/391]\tTime 0.164 (0.150)\tData 0.038 (0.039)\tLoss 0.5328 (0.4700)\tPrec@1 85.938 (87.338)\n",
      "[110/391]\tTime 0.173 (0.151)\tData 0.039 (0.039)\tLoss 0.4015 (0.4707)\tPrec@1 89.844 (87.261)\n",
      "[120/391]\tTime 0.150 (0.151)\tData 0.034 (0.039)\tLoss 0.4450 (0.4708)\tPrec@1 85.938 (87.345)\n",
      "[130/391]\tTime 0.191 (0.151)\tData 0.032 (0.039)\tLoss 0.4635 (0.4738)\tPrec@1 86.719 (87.202)\n",
      "[140/391]\tTime 0.137 (0.150)\tData 0.035 (0.038)\tLoss 0.7211 (0.4783)\tPrec@1 79.688 (87.084)\n",
      "[150/391]\tTime 0.130 (0.150)\tData 0.034 (0.038)\tLoss 0.6443 (0.4837)\tPrec@1 81.250 (86.931)\n",
      "[160/391]\tTime 0.163 (0.151)\tData 0.036 (0.038)\tLoss 0.4021 (0.4840)\tPrec@1 90.625 (86.937)\n",
      "[170/391]\tTime 0.140 (0.151)\tData 0.026 (0.038)\tLoss 0.6478 (0.4848)\tPrec@1 83.594 (86.915)\n",
      "[180/391]\tTime 0.150 (0.150)\tData 0.047 (0.038)\tLoss 0.4161 (0.4827)\tPrec@1 88.281 (86.965)\n",
      "[190/391]\tTime 0.116 (0.151)\tData 0.032 (0.039)\tLoss 0.4296 (0.4826)\tPrec@1 90.625 (86.993)\n",
      "[200/391]\tTime 0.149 (0.151)\tData 0.024 (0.039)\tLoss 0.4165 (0.4829)\tPrec@1 90.625 (87.002)\n",
      "[210/391]\tTime 0.135 (0.151)\tData 0.036 (0.039)\tLoss 0.5026 (0.4849)\tPrec@1 85.156 (86.937)\n",
      "[220/391]\tTime 0.124 (0.150)\tData 0.032 (0.039)\tLoss 0.6394 (0.4856)\tPrec@1 78.906 (86.927)\n",
      "[230/391]\tTime 0.164 (0.150)\tData 0.045 (0.039)\tLoss 0.4387 (0.4866)\tPrec@1 92.188 (86.888)\n",
      "[240/391]\tTime 0.155 (0.150)\tData 0.035 (0.039)\tLoss 0.4670 (0.4872)\tPrec@1 85.156 (86.858)\n",
      "[250/391]\tTime 0.122 (0.150)\tData 0.033 (0.039)\tLoss 0.5276 (0.4882)\tPrec@1 85.938 (86.818)\n",
      "[260/391]\tTime 0.140 (0.150)\tData 0.034 (0.039)\tLoss 0.4580 (0.4887)\tPrec@1 85.938 (86.785)\n",
      "[270/391]\tTime 0.123 (0.150)\tData 0.033 (0.039)\tLoss 0.4627 (0.4877)\tPrec@1 88.281 (86.840)\n",
      "[280/391]\tTime 0.135 (0.149)\tData 0.033 (0.039)\tLoss 0.4338 (0.4879)\tPrec@1 89.844 (86.838)\n",
      "[290/391]\tTime 0.134 (0.149)\tData 0.028 (0.039)\tLoss 0.4161 (0.4877)\tPrec@1 89.844 (86.826)\n",
      "[300/391]\tTime 0.136 (0.149)\tData 0.034 (0.038)\tLoss 0.4515 (0.4874)\tPrec@1 88.281 (86.807)\n",
      "[310/391]\tTime 0.188 (0.150)\tData 0.042 (0.039)\tLoss 0.4741 (0.4868)\tPrec@1 85.938 (86.812)\n",
      "[320/391]\tTime 0.108 (0.150)\tData 0.022 (0.038)\tLoss 0.5806 (0.4866)\tPrec@1 84.375 (86.819)\n",
      "[330/391]\tTime 0.166 (0.150)\tData 0.047 (0.038)\tLoss 0.3981 (0.4861)\tPrec@1 89.844 (86.834)\n",
      "[340/391]\tTime 0.127 (0.149)\tData 0.035 (0.038)\tLoss 0.5704 (0.4864)\tPrec@1 84.375 (86.815)\n",
      "[350/391]\tTime 0.115 (0.149)\tData 0.027 (0.038)\tLoss 0.6207 (0.4871)\tPrec@1 82.812 (86.814)\n",
      "[360/391]\tTime 0.160 (0.149)\tData 0.043 (0.038)\tLoss 0.5298 (0.4882)\tPrec@1 87.500 (86.786)\n",
      "[370/391]\tTime 0.116 (0.149)\tData 0.032 (0.038)\tLoss 0.5232 (0.4895)\tPrec@1 86.719 (86.763)\n",
      "[380/391]\tTime 0.171 (0.149)\tData 0.043 (0.038)\tLoss 0.5257 (0.4899)\tPrec@1 86.719 (86.752)\n",
      "[390/391]\tTime 0.113 (0.149)\tData 0.021 (0.038)\tLoss 0.4736 (0.4899)\tPrec@1 90.000 (86.778)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.037 (0.037)\tLoss 0.5658 (0.5658)\tPrec@1 85.156 (85.156)\n",
      "Test: [10/79]\tTime 0.041 (0.044)\tLoss 0.7544 (0.9239)\tPrec@1 85.156 (78.906)\n",
      "Test: [20/79]\tTime 0.036 (0.042)\tLoss 0.8620 (0.9278)\tPrec@1 82.812 (78.795)\n",
      "Test: [30/79]\tTime 0.036 (0.041)\tLoss 0.6578 (0.9152)\tPrec@1 84.375 (79.335)\n",
      "Test: [40/79]\tTime 0.030 (0.040)\tLoss 0.9310 (0.9247)\tPrec@1 76.562 (79.421)\n",
      "Test: [50/79]\tTime 0.066 (0.040)\tLoss 1.0374 (0.9309)\tPrec@1 79.688 (79.366)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test: [60/79]\tTime 0.029 (0.039)\tLoss 0.7579 (0.9374)\tPrec@1 82.031 (79.226)\n",
      "Test: [70/79]\tTime 0.036 (0.039)\tLoss 0.7759 (0.9391)\tPrec@1 85.938 (79.126)\n",
      " * Prec@1 79.060\n",
      "\n",
      "===> epoch: 20/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.166 (0.166)\tData 0.069 (0.069)\tLoss 0.5354 (0.5354)\tPrec@1 85.156 (85.156)\n",
      "[10/391]\tTime 0.145 (0.149)\tData 0.041 (0.040)\tLoss 0.5286 (0.4952)\tPrec@1 85.156 (87.500)\n",
      "[20/391]\tTime 0.142 (0.146)\tData 0.029 (0.037)\tLoss 0.6155 (0.4720)\tPrec@1 80.469 (87.835)\n",
      "[30/391]\tTime 0.141 (0.142)\tData 0.033 (0.035)\tLoss 0.4117 (0.4667)\tPrec@1 88.281 (88.029)\n",
      "[40/391]\tTime 0.123 (0.142)\tData 0.032 (0.035)\tLoss 0.3083 (0.4666)\tPrec@1 91.406 (88.167)\n",
      "[50/391]\tTime 0.182 (0.145)\tData 0.061 (0.037)\tLoss 0.4547 (0.4693)\tPrec@1 88.281 (88.021)\n",
      "[60/391]\tTime 0.157 (0.148)\tData 0.036 (0.037)\tLoss 0.4119 (0.4698)\tPrec@1 92.969 (87.743)\n",
      "[70/391]\tTime 0.136 (0.148)\tData 0.034 (0.038)\tLoss 0.5062 (0.4769)\tPrec@1 88.281 (87.456)\n",
      "[80/391]\tTime 0.130 (0.148)\tData 0.026 (0.038)\tLoss 0.5059 (0.4787)\tPrec@1 85.938 (87.355)\n",
      "[90/391]\tTime 0.178 (0.147)\tData 0.067 (0.038)\tLoss 0.4730 (0.4810)\tPrec@1 85.938 (87.303)\n",
      "[100/391]\tTime 0.147 (0.147)\tData 0.040 (0.039)\tLoss 0.6029 (0.4811)\tPrec@1 85.156 (87.198)\n",
      "[110/391]\tTime 0.125 (0.147)\tData 0.035 (0.039)\tLoss 0.5376 (0.4805)\tPrec@1 85.156 (87.233)\n",
      "[120/391]\tTime 0.120 (0.146)\tData 0.027 (0.039)\tLoss 0.4675 (0.4788)\tPrec@1 88.281 (87.177)\n",
      "[130/391]\tTime 0.148 (0.145)\tData 0.034 (0.039)\tLoss 0.3875 (0.4774)\tPrec@1 90.625 (87.100)\n",
      "[140/391]\tTime 0.116 (0.146)\tData 0.033 (0.040)\tLoss 0.5998 (0.4776)\tPrec@1 85.938 (87.118)\n",
      "[150/391]\tTime 0.190 (0.147)\tData 0.059 (0.040)\tLoss 0.4496 (0.4814)\tPrec@1 89.844 (87.065)\n",
      "[160/391]\tTime 0.154 (0.147)\tData 0.038 (0.040)\tLoss 0.4658 (0.4793)\tPrec@1 90.625 (87.155)\n",
      "[170/391]\tTime 0.149 (0.148)\tData 0.029 (0.040)\tLoss 0.4754 (0.4813)\tPrec@1 88.281 (87.130)\n",
      "[180/391]\tTime 0.138 (0.149)\tData 0.035 (0.040)\tLoss 0.4627 (0.4825)\tPrec@1 89.844 (87.090)\n",
      "[190/391]\tTime 0.204 (0.149)\tData 0.072 (0.040)\tLoss 0.4555 (0.4827)\tPrec@1 89.062 (87.099)\n",
      "[200/391]\tTime 0.147 (0.148)\tData 0.036 (0.040)\tLoss 0.6097 (0.4864)\tPrec@1 83.594 (86.987)\n",
      "[210/391]\tTime 0.121 (0.148)\tData 0.038 (0.040)\tLoss 0.4836 (0.4882)\tPrec@1 87.500 (86.948)\n",
      "[220/391]\tTime 0.164 (0.149)\tData 0.033 (0.040)\tLoss 0.5287 (0.4906)\tPrec@1 87.500 (86.896)\n",
      "[230/391]\tTime 0.119 (0.149)\tData 0.035 (0.040)\tLoss 0.6820 (0.4917)\tPrec@1 78.125 (86.847)\n",
      "[240/391]\tTime 0.114 (0.148)\tData 0.046 (0.040)\tLoss 0.3954 (0.4922)\tPrec@1 92.969 (86.829)\n",
      "[250/391]\tTime 0.147 (0.148)\tData 0.030 (0.040)\tLoss 0.4936 (0.4935)\tPrec@1 86.719 (86.778)\n",
      "[260/391]\tTime 0.131 (0.147)\tData 0.029 (0.040)\tLoss 0.3760 (0.4930)\tPrec@1 89.844 (86.797)\n",
      "[270/391]\tTime 0.157 (0.147)\tData 0.035 (0.040)\tLoss 0.6155 (0.4934)\tPrec@1 82.812 (86.736)\n",
      "[280/391]\tTime 0.179 (0.147)\tData 0.058 (0.040)\tLoss 0.5153 (0.4953)\tPrec@1 85.938 (86.683)\n",
      "[290/391]\tTime 0.141 (0.148)\tData 0.033 (0.040)\tLoss 0.5072 (0.4954)\tPrec@1 86.719 (86.676)\n",
      "[300/391]\tTime 0.133 (0.148)\tData 0.038 (0.040)\tLoss 0.5637 (0.4956)\tPrec@1 86.719 (86.680)\n",
      "[310/391]\tTime 0.145 (0.148)\tData 0.042 (0.040)\tLoss 0.4333 (0.4962)\tPrec@1 89.844 (86.656)\n",
      "[320/391]\tTime 0.158 (0.149)\tData 0.035 (0.040)\tLoss 0.3598 (0.4962)\tPrec@1 91.406 (86.702)\n",
      "[330/391]\tTime 0.207 (0.149)\tData 0.046 (0.040)\tLoss 0.4940 (0.4960)\tPrec@1 88.281 (86.728)\n",
      "[340/391]\tTime 0.143 (0.149)\tData 0.036 (0.040)\tLoss 0.4504 (0.4953)\tPrec@1 88.281 (86.746)\n",
      "[350/391]\tTime 0.133 (0.149)\tData 0.032 (0.040)\tLoss 0.4524 (0.4952)\tPrec@1 89.844 (86.748)\n",
      "[360/391]\tTime 0.150 (0.149)\tData 0.033 (0.040)\tLoss 0.4235 (0.4950)\tPrec@1 89.062 (86.756)\n",
      "[370/391]\tTime 0.130 (0.149)\tData 0.035 (0.040)\tLoss 0.4214 (0.4948)\tPrec@1 85.938 (86.717)\n",
      "[380/391]\tTime 0.153 (0.149)\tData 0.028 (0.040)\tLoss 0.4784 (0.4957)\tPrec@1 90.625 (86.696)\n",
      "[390/391]\tTime 0.114 (0.149)\tData 0.022 (0.040)\tLoss 0.4665 (0.4955)\tPrec@1 86.250 (86.724)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.050 (0.050)\tLoss 0.5450 (0.5450)\tPrec@1 86.719 (86.719)\n",
      "Test: [10/79]\tTime 0.040 (0.037)\tLoss 0.5883 (0.8062)\tPrec@1 86.719 (81.037)\n",
      "Test: [20/79]\tTime 0.036 (0.037)\tLoss 0.8114 (0.8508)\tPrec@1 79.688 (80.432)\n",
      "Test: [30/79]\tTime 0.032 (0.037)\tLoss 0.7620 (0.8598)\tPrec@1 78.906 (80.116)\n",
      "Test: [40/79]\tTime 0.036 (0.039)\tLoss 0.9657 (0.8445)\tPrec@1 81.250 (80.373)\n",
      "Test: [50/79]\tTime 0.041 (0.040)\tLoss 1.0496 (0.8408)\tPrec@1 78.125 (80.423)\n",
      "Test: [60/79]\tTime 0.059 (0.041)\tLoss 0.8963 (0.8359)\tPrec@1 80.469 (80.379)\n",
      "Test: [70/79]\tTime 0.032 (0.040)\tLoss 0.7496 (0.8329)\tPrec@1 81.250 (80.337)\n",
      " * Prec@1 80.430\n",
      "\n",
      "===> epoch: 21/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.113 (0.113)\tData 0.028 (0.028)\tLoss 0.5113 (0.5113)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.129 (0.149)\tData 0.035 (0.043)\tLoss 0.4783 (0.5254)\tPrec@1 83.594 (85.298)\n",
      "[20/391]\tTime 0.126 (0.140)\tData 0.037 (0.041)\tLoss 0.4596 (0.5142)\tPrec@1 86.719 (85.565)\n",
      "[30/391]\tTime 0.138 (0.139)\tData 0.033 (0.041)\tLoss 0.4917 (0.5068)\tPrec@1 86.719 (85.988)\n",
      "[40/391]\tTime 0.144 (0.143)\tData 0.037 (0.041)\tLoss 0.4731 (0.4979)\tPrec@1 84.375 (86.185)\n",
      "[50/391]\tTime 0.133 (0.145)\tData 0.034 (0.042)\tLoss 0.4938 (0.4943)\tPrec@1 87.500 (86.489)\n",
      "[60/391]\tTime 0.163 (0.145)\tData 0.031 (0.041)\tLoss 0.5918 (0.4924)\tPrec@1 84.375 (86.668)\n",
      "[70/391]\tTime 0.151 (0.145)\tData 0.036 (0.040)\tLoss 0.4642 (0.4900)\tPrec@1 89.062 (86.730)\n",
      "[80/391]\tTime 0.151 (0.145)\tData 0.032 (0.040)\tLoss 0.4819 (0.4899)\tPrec@1 84.375 (86.632)\n",
      "[90/391]\tTime 0.121 (0.146)\tData 0.034 (0.040)\tLoss 0.4040 (0.4859)\tPrec@1 89.844 (86.650)\n",
      "[100/391]\tTime 0.186 (0.148)\tData 0.069 (0.041)\tLoss 0.3687 (0.4804)\tPrec@1 93.750 (86.866)\n",
      "[110/391]\tTime 0.143 (0.150)\tData 0.034 (0.041)\tLoss 0.4406 (0.4798)\tPrec@1 86.719 (86.888)\n",
      "[120/391]\tTime 0.149 (0.151)\tData 0.027 (0.041)\tLoss 0.3582 (0.4776)\tPrec@1 92.188 (86.919)\n",
      "[130/391]\tTime 0.158 (0.150)\tData 0.036 (0.040)\tLoss 0.4940 (0.4776)\tPrec@1 84.375 (87.011)\n",
      "[140/391]\tTime 0.154 (0.150)\tData 0.047 (0.040)\tLoss 0.5201 (0.4796)\tPrec@1 84.375 (87.018)\n",
      "[150/391]\tTime 0.105 (0.150)\tData 0.028 (0.041)\tLoss 0.5298 (0.4814)\tPrec@1 85.938 (86.957)\n",
      "[160/391]\tTime 0.126 (0.148)\tData 0.033 (0.040)\tLoss 0.5655 (0.4835)\tPrec@1 85.156 (86.908)\n",
      "[170/391]\tTime 0.133 (0.148)\tData 0.036 (0.040)\tLoss 0.6518 (0.4867)\tPrec@1 82.031 (86.796)\n",
      "[180/391]\tTime 0.149 (0.148)\tData 0.029 (0.040)\tLoss 0.4841 (0.4891)\tPrec@1 85.938 (86.710)\n",
      "[190/391]\tTime 0.147 (0.148)\tData 0.034 (0.040)\tLoss 0.5691 (0.4906)\tPrec@1 82.031 (86.661)\n",
      "[200/391]\tTime 0.148 (0.148)\tData 0.038 (0.039)\tLoss 0.3531 (0.4881)\tPrec@1 92.969 (86.769)\n",
      "[210/391]\tTime 0.156 (0.148)\tData 0.045 (0.040)\tLoss 0.5196 (0.4877)\tPrec@1 84.375 (86.759)\n",
      "[220/391]\tTime 0.145 (0.148)\tData 0.034 (0.039)\tLoss 0.4950 (0.4881)\tPrec@1 89.062 (86.747)\n",
      "[230/391]\tTime 0.155 (0.149)\tData 0.039 (0.040)\tLoss 0.4375 (0.4882)\tPrec@1 86.719 (86.705)\n",
      "[240/391]\tTime 0.161 (0.149)\tData 0.030 (0.039)\tLoss 0.5381 (0.4884)\tPrec@1 85.156 (86.716)\n",
      "[250/391]\tTime 0.134 (0.149)\tData 0.032 (0.040)\tLoss 0.5756 (0.4874)\tPrec@1 82.031 (86.747)\n",
      "[260/391]\tTime 0.167 (0.149)\tData 0.049 (0.039)\tLoss 0.6322 (0.4872)\tPrec@1 82.031 (86.755)\n",
      "[270/391]\tTime 0.142 (0.149)\tData 0.031 (0.039)\tLoss 0.4808 (0.4886)\tPrec@1 86.719 (86.730)\n",
      "[280/391]\tTime 0.166 (0.149)\tData 0.050 (0.040)\tLoss 0.5192 (0.4896)\tPrec@1 83.594 (86.697)\n",
      "[290/391]\tTime 0.154 (0.150)\tData 0.033 (0.039)\tLoss 0.4469 (0.4891)\tPrec@1 89.844 (86.687)\n",
      "[300/391]\tTime 0.135 (0.149)\tData 0.058 (0.040)\tLoss 0.4290 (0.4896)\tPrec@1 87.500 (86.669)\n",
      "[310/391]\tTime 0.099 (0.149)\tData 0.034 (0.040)\tLoss 0.5588 (0.4901)\tPrec@1 82.031 (86.658)\n",
      "[320/391]\tTime 0.126 (0.149)\tData 0.030 (0.040)\tLoss 0.4949 (0.4894)\tPrec@1 88.281 (86.660)\n",
      "[330/391]\tTime 0.112 (0.149)\tData 0.027 (0.040)\tLoss 0.5392 (0.4897)\tPrec@1 85.156 (86.667)\n",
      "[340/391]\tTime 0.121 (0.149)\tData 0.036 (0.040)\tLoss 0.4670 (0.4908)\tPrec@1 89.844 (86.629)\n",
      "[350/391]\tTime 0.152 (0.148)\tData 0.049 (0.040)\tLoss 0.3991 (0.4898)\tPrec@1 89.844 (86.670)\n",
      "[360/391]\tTime 0.136 (0.148)\tData 0.020 (0.040)\tLoss 0.5404 (0.4903)\tPrec@1 85.156 (86.649)\n",
      "[370/391]\tTime 0.183 (0.148)\tData 0.061 (0.040)\tLoss 0.5541 (0.4901)\tPrec@1 83.594 (86.647)\n",
      "[380/391]\tTime 0.161 (0.149)\tData 0.039 (0.040)\tLoss 0.4971 (0.4902)\tPrec@1 87.500 (86.629)\n",
      "[390/391]\tTime 0.073 (0.148)\tData 0.018 (0.040)\tLoss 0.6229 (0.4913)\tPrec@1 83.750 (86.590)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.032 (0.032)\tLoss 0.6800 (0.6800)\tPrec@1 81.250 (81.250)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test: [10/79]\tTime 0.050 (0.038)\tLoss 0.6261 (0.6875)\tPrec@1 84.375 (82.670)\n",
      "Test: [20/79]\tTime 0.040 (0.038)\tLoss 0.6574 (0.7218)\tPrec@1 82.031 (81.734)\n",
      "Test: [30/79]\tTime 0.032 (0.039)\tLoss 0.6226 (0.7083)\tPrec@1 84.375 (82.056)\n",
      "Test: [40/79]\tTime 0.033 (0.040)\tLoss 0.5517 (0.6983)\tPrec@1 83.594 (82.679)\n",
      "Test: [50/79]\tTime 0.034 (0.040)\tLoss 0.8032 (0.6988)\tPrec@1 82.812 (82.767)\n",
      "Test: [60/79]\tTime 0.034 (0.040)\tLoss 0.8316 (0.7032)\tPrec@1 78.125 (82.518)\n",
      "Test: [70/79]\tTime 0.048 (0.041)\tLoss 0.4893 (0.7073)\tPrec@1 82.812 (82.361)\n",
      " * Prec@1 82.360\n",
      "\n",
      "===> epoch: 22/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.169 (0.169)\tData 0.033 (0.033)\tLoss 0.3273 (0.3273)\tPrec@1 92.969 (92.969)\n",
      "[10/391]\tTime 0.159 (0.171)\tData 0.046 (0.039)\tLoss 0.5136 (0.4975)\tPrec@1 84.375 (87.571)\n",
      "[20/391]\tTime 0.170 (0.163)\tData 0.035 (0.040)\tLoss 0.4187 (0.4929)\tPrec@1 87.500 (87.054)\n",
      "[30/391]\tTime 0.149 (0.163)\tData 0.034 (0.041)\tLoss 0.4522 (0.4866)\tPrec@1 88.281 (86.946)\n",
      "[40/391]\tTime 0.192 (0.159)\tData 0.060 (0.041)\tLoss 0.4138 (0.4796)\tPrec@1 92.188 (87.481)\n",
      "[50/391]\tTime 0.113 (0.155)\tData 0.034 (0.040)\tLoss 0.4233 (0.4731)\tPrec@1 90.625 (87.714)\n",
      "[60/391]\tTime 0.174 (0.156)\tData 0.035 (0.042)\tLoss 0.5013 (0.4796)\tPrec@1 85.156 (87.526)\n",
      "[70/391]\tTime 0.138 (0.157)\tData 0.040 (0.042)\tLoss 0.4590 (0.4797)\tPrec@1 86.719 (87.522)\n",
      "[80/391]\tTime 0.160 (0.157)\tData 0.035 (0.042)\tLoss 0.4422 (0.4798)\tPrec@1 88.281 (87.481)\n",
      "[90/391]\tTime 0.164 (0.157)\tData 0.029 (0.041)\tLoss 0.5864 (0.4815)\tPrec@1 82.812 (87.260)\n",
      "[100/391]\tTime 0.141 (0.158)\tData 0.034 (0.041)\tLoss 0.5572 (0.4796)\tPrec@1 85.156 (87.268)\n",
      "[110/391]\tTime 0.135 (0.157)\tData 0.032 (0.041)\tLoss 0.4664 (0.4785)\tPrec@1 86.719 (87.233)\n",
      "[120/391]\tTime 0.160 (0.156)\tData 0.042 (0.040)\tLoss 0.5088 (0.4780)\tPrec@1 85.156 (87.300)\n",
      "[130/391]\tTime 0.148 (0.156)\tData 0.035 (0.040)\tLoss 0.3932 (0.4779)\tPrec@1 86.719 (87.226)\n",
      "[140/391]\tTime 0.149 (0.156)\tData 0.027 (0.040)\tLoss 0.5729 (0.4804)\tPrec@1 85.156 (87.145)\n",
      "[150/391]\tTime 0.119 (0.155)\tData 0.033 (0.040)\tLoss 0.5374 (0.4853)\tPrec@1 86.719 (86.977)\n",
      "[160/391]\tTime 0.127 (0.155)\tData 0.033 (0.040)\tLoss 0.4227 (0.4875)\tPrec@1 91.406 (86.903)\n",
      "[170/391]\tTime 0.170 (0.156)\tData 0.033 (0.040)\tLoss 0.5393 (0.4884)\tPrec@1 86.719 (86.906)\n",
      "[180/391]\tTime 0.148 (0.155)\tData 0.033 (0.040)\tLoss 0.5183 (0.4881)\tPrec@1 85.938 (86.874)\n",
      "[190/391]\tTime 0.118 (0.155)\tData 0.030 (0.040)\tLoss 0.6581 (0.4895)\tPrec@1 78.125 (86.837)\n",
      "[200/391]\tTime 0.141 (0.153)\tData 0.028 (0.039)\tLoss 0.4759 (0.4883)\tPrec@1 85.938 (86.866)\n",
      "[210/391]\tTime 0.144 (0.152)\tData 0.027 (0.039)\tLoss 0.5895 (0.4886)\tPrec@1 82.031 (86.826)\n",
      "[220/391]\tTime 0.172 (0.153)\tData 0.051 (0.039)\tLoss 0.5320 (0.4878)\tPrec@1 85.156 (86.857)\n",
      "[230/391]\tTime 0.145 (0.152)\tData 0.035 (0.039)\tLoss 0.4856 (0.4870)\tPrec@1 87.500 (86.881)\n",
      "[240/391]\tTime 0.163 (0.152)\tData 0.063 (0.039)\tLoss 0.4519 (0.4880)\tPrec@1 85.938 (86.816)\n",
      "[250/391]\tTime 0.167 (0.151)\tData 0.048 (0.039)\tLoss 0.3893 (0.4885)\tPrec@1 90.625 (86.831)\n",
      "[260/391]\tTime 0.127 (0.152)\tData 0.028 (0.039)\tLoss 0.4007 (0.4874)\tPrec@1 89.062 (86.856)\n",
      "[270/391]\tTime 0.147 (0.152)\tData 0.039 (0.039)\tLoss 0.4222 (0.4855)\tPrec@1 89.062 (86.981)\n",
      "[280/391]\tTime 0.156 (0.152)\tData 0.033 (0.039)\tLoss 0.5862 (0.4863)\tPrec@1 82.812 (86.961)\n",
      "[290/391]\tTime 0.152 (0.152)\tData 0.037 (0.039)\tLoss 0.4568 (0.4859)\tPrec@1 85.938 (86.958)\n",
      "[300/391]\tTime 0.154 (0.152)\tData 0.034 (0.039)\tLoss 0.5053 (0.4864)\tPrec@1 84.375 (86.924)\n",
      "[310/391]\tTime 0.167 (0.152)\tData 0.038 (0.039)\tLoss 0.4745 (0.4867)\tPrec@1 86.719 (86.915)\n",
      "[320/391]\tTime 0.146 (0.152)\tData 0.032 (0.039)\tLoss 0.4597 (0.4865)\tPrec@1 87.500 (86.930)\n",
      "[330/391]\tTime 0.157 (0.152)\tData 0.038 (0.039)\tLoss 0.6042 (0.4865)\tPrec@1 82.812 (86.924)\n",
      "[340/391]\tTime 0.144 (0.151)\tData 0.032 (0.039)\tLoss 0.4415 (0.4874)\tPrec@1 86.719 (86.909)\n",
      "[350/391]\tTime 0.109 (0.151)\tData 0.033 (0.039)\tLoss 0.5404 (0.4872)\tPrec@1 83.594 (86.910)\n",
      "[360/391]\tTime 0.135 (0.151)\tData 0.036 (0.039)\tLoss 0.4231 (0.4882)\tPrec@1 87.500 (86.859)\n",
      "[370/391]\tTime 0.154 (0.151)\tData 0.031 (0.039)\tLoss 0.5131 (0.4885)\tPrec@1 85.938 (86.858)\n",
      "[380/391]\tTime 0.176 (0.151)\tData 0.033 (0.039)\tLoss 0.5558 (0.4893)\tPrec@1 85.938 (86.844)\n",
      "[390/391]\tTime 0.106 (0.151)\tData 0.021 (0.039)\tLoss 0.4947 (0.4901)\tPrec@1 82.500 (86.810)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.039 (0.039)\tLoss 0.9145 (0.9145)\tPrec@1 77.344 (77.344)\n",
      "Test: [10/79]\tTime 0.035 (0.036)\tLoss 0.8468 (0.9827)\tPrec@1 83.594 (78.267)\n",
      "Test: [20/79]\tTime 0.033 (0.038)\tLoss 1.1142 (0.9885)\tPrec@1 76.562 (77.790)\n",
      "Test: [30/79]\tTime 0.042 (0.039)\tLoss 0.8485 (0.9653)\tPrec@1 83.594 (77.999)\n",
      "Test: [40/79]\tTime 0.041 (0.039)\tLoss 1.1120 (0.9617)\tPrec@1 75.781 (78.201)\n",
      "Test: [50/79]\tTime 0.033 (0.040)\tLoss 1.0701 (0.9689)\tPrec@1 71.875 (77.972)\n",
      "Test: [60/79]\tTime 0.034 (0.039)\tLoss 0.9979 (0.9851)\tPrec@1 75.000 (77.882)\n",
      "Test: [70/79]\tTime 0.029 (0.038)\tLoss 0.8425 (0.9825)\tPrec@1 81.250 (77.894)\n",
      " * Prec@1 77.880\n",
      "\n",
      "===> epoch: 23/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.137 (0.137)\tData 0.029 (0.029)\tLoss 0.5682 (0.5682)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.215 (0.167)\tData 0.078 (0.047)\tLoss 0.3934 (0.5042)\tPrec@1 90.625 (86.080)\n",
      "[20/391]\tTime 0.124 (0.156)\tData 0.033 (0.042)\tLoss 0.4764 (0.4991)\tPrec@1 89.844 (86.384)\n",
      "[30/391]\tTime 0.149 (0.155)\tData 0.033 (0.043)\tLoss 0.4399 (0.4966)\tPrec@1 89.062 (86.467)\n",
      "[40/391]\tTime 0.149 (0.158)\tData 0.034 (0.042)\tLoss 0.4727 (0.4989)\tPrec@1 86.719 (86.376)\n",
      "[50/391]\tTime 0.135 (0.157)\tData 0.029 (0.042)\tLoss 0.5270 (0.4948)\tPrec@1 85.938 (86.657)\n",
      "[60/391]\tTime 0.147 (0.158)\tData 0.030 (0.041)\tLoss 0.4147 (0.4895)\tPrec@1 86.719 (86.770)\n",
      "[70/391]\tTime 0.175 (0.158)\tData 0.035 (0.042)\tLoss 0.4938 (0.4864)\tPrec@1 83.594 (86.785)\n",
      "[80/391]\tTime 0.162 (0.158)\tData 0.030 (0.042)\tLoss 0.4239 (0.4877)\tPrec@1 88.281 (86.777)\n",
      "[90/391]\tTime 0.123 (0.155)\tData 0.032 (0.041)\tLoss 0.5126 (0.4920)\tPrec@1 83.594 (86.745)\n",
      "[100/391]\tTime 0.139 (0.154)\tData 0.036 (0.041)\tLoss 0.5218 (0.4939)\tPrec@1 85.156 (86.672)\n",
      "[110/391]\tTime 0.137 (0.154)\tData 0.028 (0.040)\tLoss 0.5611 (0.4927)\tPrec@1 83.594 (86.712)\n",
      "[120/391]\tTime 0.129 (0.153)\tData 0.035 (0.040)\tLoss 0.4812 (0.4925)\tPrec@1 85.938 (86.757)\n",
      "[130/391]\tTime 0.134 (0.152)\tData 0.039 (0.040)\tLoss 0.3790 (0.4923)\tPrec@1 92.969 (86.689)\n",
      "[140/391]\tTime 0.123 (0.150)\tData 0.037 (0.040)\tLoss 0.5250 (0.4949)\tPrec@1 85.156 (86.602)\n",
      "[150/391]\tTime 0.158 (0.149)\tData 0.036 (0.040)\tLoss 0.3217 (0.4923)\tPrec@1 90.625 (86.657)\n",
      "[160/391]\tTime 0.140 (0.148)\tData 0.039 (0.040)\tLoss 0.2905 (0.4901)\tPrec@1 92.969 (86.753)\n",
      "[170/391]\tTime 0.158 (0.148)\tData 0.039 (0.040)\tLoss 0.4695 (0.4888)\tPrec@1 84.375 (86.751)\n",
      "[180/391]\tTime 0.153 (0.148)\tData 0.032 (0.040)\tLoss 0.3560 (0.4871)\tPrec@1 89.844 (86.822)\n",
      "[190/391]\tTime 0.186 (0.149)\tData 0.053 (0.040)\tLoss 0.4460 (0.4849)\tPrec@1 90.625 (86.968)\n",
      "[200/391]\tTime 0.154 (0.149)\tData 0.032 (0.039)\tLoss 0.5267 (0.4842)\tPrec@1 85.156 (86.952)\n",
      "[210/391]\tTime 0.106 (0.148)\tData 0.031 (0.039)\tLoss 0.6006 (0.4849)\tPrec@1 85.156 (86.945)\n",
      "[220/391]\tTime 0.137 (0.148)\tData 0.034 (0.039)\tLoss 0.5045 (0.4857)\tPrec@1 85.938 (86.934)\n",
      "[230/391]\tTime 0.175 (0.149)\tData 0.036 (0.039)\tLoss 0.4869 (0.4864)\tPrec@1 85.938 (86.888)\n",
      "[240/391]\tTime 0.128 (0.150)\tData 0.028 (0.039)\tLoss 0.5217 (0.4859)\tPrec@1 88.281 (86.939)\n",
      "[250/391]\tTime 0.126 (0.149)\tData 0.034 (0.039)\tLoss 0.4748 (0.4865)\tPrec@1 88.281 (86.937)\n",
      "[260/391]\tTime 0.157 (0.149)\tData 0.056 (0.039)\tLoss 0.5822 (0.4870)\tPrec@1 82.031 (86.883)\n",
      "[270/391]\tTime 0.138 (0.149)\tData 0.027 (0.039)\tLoss 0.5522 (0.4879)\tPrec@1 85.938 (86.857)\n",
      "[280/391]\tTime 0.154 (0.150)\tData 0.035 (0.039)\tLoss 0.4314 (0.4885)\tPrec@1 88.281 (86.838)\n",
      "[290/391]\tTime 0.123 (0.150)\tData 0.030 (0.039)\tLoss 0.4241 (0.4880)\tPrec@1 90.625 (86.861)\n",
      "[300/391]\tTime 0.165 (0.150)\tData 0.037 (0.039)\tLoss 0.5342 (0.4883)\tPrec@1 87.500 (86.846)\n",
      "[310/391]\tTime 0.154 (0.150)\tData 0.035 (0.039)\tLoss 0.5199 (0.4884)\tPrec@1 87.500 (86.837)\n",
      "[320/391]\tTime 0.168 (0.150)\tData 0.043 (0.039)\tLoss 0.5763 (0.4884)\tPrec@1 82.812 (86.850)\n",
      "[330/391]\tTime 0.168 (0.150)\tData 0.050 (0.039)\tLoss 0.4284 (0.4870)\tPrec@1 89.844 (86.912)\n",
      "[340/391]\tTime 0.123 (0.150)\tData 0.034 (0.039)\tLoss 0.4915 (0.4867)\tPrec@1 84.375 (86.916)\n",
      "[350/391]\tTime 0.184 (0.151)\tData 0.047 (0.039)\tLoss 0.6308 (0.4867)\tPrec@1 83.594 (86.890)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[360/391]\tTime 0.146 (0.151)\tData 0.031 (0.039)\tLoss 0.4589 (0.4856)\tPrec@1 86.719 (86.909)\n",
      "[370/391]\tTime 0.126 (0.151)\tData 0.033 (0.039)\tLoss 0.5162 (0.4845)\tPrec@1 87.500 (86.946)\n",
      "[380/391]\tTime 0.153 (0.151)\tData 0.036 (0.039)\tLoss 0.4228 (0.4847)\tPrec@1 89.844 (86.953)\n",
      "[390/391]\tTime 0.110 (0.151)\tData 0.040 (0.039)\tLoss 0.5174 (0.4863)\tPrec@1 85.000 (86.900)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.037 (0.037)\tLoss 0.7478 (0.7478)\tPrec@1 82.031 (82.031)\n",
      "Test: [10/79]\tTime 0.034 (0.040)\tLoss 0.7745 (0.9157)\tPrec@1 82.812 (79.688)\n",
      "Test: [20/79]\tTime 0.034 (0.038)\tLoss 0.8160 (0.9539)\tPrec@1 81.250 (78.757)\n",
      "Test: [30/79]\tTime 0.037 (0.041)\tLoss 0.9557 (0.9438)\tPrec@1 82.031 (78.931)\n",
      "Test: [40/79]\tTime 0.028 (0.039)\tLoss 0.9310 (0.9296)\tPrec@1 73.438 (79.287)\n",
      "Test: [50/79]\tTime 0.035 (0.040)\tLoss 0.7606 (0.9187)\tPrec@1 83.594 (79.366)\n",
      "Test: [60/79]\tTime 0.083 (0.041)\tLoss 0.8018 (0.9337)\tPrec@1 79.688 (79.188)\n",
      "Test: [70/79]\tTime 0.038 (0.040)\tLoss 0.7722 (0.9331)\tPrec@1 83.594 (79.159)\n",
      " * Prec@1 79.100\n",
      "\n",
      "===> epoch: 24/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.139 (0.139)\tData 0.027 (0.027)\tLoss 0.5951 (0.5951)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.128 (0.142)\tData 0.035 (0.034)\tLoss 0.3906 (0.4910)\tPrec@1 91.406 (86.577)\n",
      "[20/391]\tTime 0.169 (0.147)\tData 0.049 (0.038)\tLoss 0.3853 (0.4743)\tPrec@1 91.406 (86.868)\n",
      "[30/391]\tTime 0.135 (0.144)\tData 0.020 (0.037)\tLoss 0.5178 (0.4864)\tPrec@1 87.500 (86.517)\n",
      "[40/391]\tTime 0.163 (0.143)\tData 0.050 (0.039)\tLoss 0.5376 (0.4905)\tPrec@1 83.594 (86.280)\n",
      "[50/391]\tTime 0.158 (0.141)\tData 0.041 (0.039)\tLoss 0.5046 (0.4942)\tPrec@1 86.719 (86.290)\n",
      "[60/391]\tTime 0.145 (0.140)\tData 0.032 (0.038)\tLoss 0.2825 (0.4902)\tPrec@1 93.750 (86.591)\n",
      "[70/391]\tTime 0.168 (0.142)\tData 0.031 (0.038)\tLoss 0.4434 (0.4872)\tPrec@1 87.500 (86.697)\n",
      "[80/391]\tTime 0.149 (0.145)\tData 0.034 (0.038)\tLoss 0.4876 (0.4917)\tPrec@1 87.500 (86.564)\n",
      "[90/391]\tTime 0.131 (0.146)\tData 0.035 (0.038)\tLoss 0.4642 (0.4912)\tPrec@1 89.844 (86.736)\n",
      "[100/391]\tTime 0.150 (0.148)\tData 0.032 (0.039)\tLoss 0.4474 (0.4905)\tPrec@1 88.281 (86.719)\n",
      "[110/391]\tTime 0.156 (0.149)\tData 0.051 (0.039)\tLoss 0.6569 (0.4926)\tPrec@1 78.906 (86.515)\n",
      "[120/391]\tTime 0.145 (0.149)\tData 0.036 (0.039)\tLoss 0.4982 (0.4984)\tPrec@1 88.281 (86.389)\n",
      "[130/391]\tTime 0.123 (0.148)\tData 0.033 (0.039)\tLoss 0.5668 (0.5022)\tPrec@1 82.812 (86.277)\n",
      "[140/391]\tTime 0.119 (0.147)\tData 0.044 (0.039)\tLoss 0.3893 (0.5010)\tPrec@1 89.844 (86.342)\n",
      "[150/391]\tTime 0.114 (0.146)\tData 0.026 (0.038)\tLoss 0.3987 (0.4991)\tPrec@1 91.406 (86.445)\n",
      "[160/391]\tTime 0.166 (0.147)\tData 0.034 (0.038)\tLoss 0.5586 (0.5012)\tPrec@1 86.719 (86.350)\n",
      "[170/391]\tTime 0.164 (0.148)\tData 0.033 (0.039)\tLoss 0.4569 (0.4995)\tPrec@1 85.156 (86.385)\n",
      "[180/391]\tTime 0.137 (0.148)\tData 0.035 (0.038)\tLoss 0.5351 (0.5002)\tPrec@1 82.812 (86.369)\n",
      "[190/391]\tTime 0.155 (0.148)\tData 0.033 (0.038)\tLoss 0.3829 (0.4991)\tPrec@1 89.844 (86.416)\n",
      "[200/391]\tTime 0.113 (0.148)\tData 0.031 (0.038)\tLoss 0.6117 (0.4991)\tPrec@1 81.250 (86.458)\n",
      "[210/391]\tTime 0.128 (0.148)\tData 0.036 (0.038)\tLoss 0.4631 (0.4996)\tPrec@1 88.281 (86.456)\n",
      "[220/391]\tTime 0.130 (0.148)\tData 0.034 (0.038)\tLoss 0.4314 (0.4972)\tPrec@1 90.625 (86.546)\n",
      "[230/391]\tTime 0.215 (0.149)\tData 0.046 (0.038)\tLoss 0.4146 (0.4964)\tPrec@1 88.281 (86.631)\n",
      "[240/391]\tTime 0.152 (0.149)\tData 0.031 (0.038)\tLoss 0.5389 (0.4970)\tPrec@1 86.719 (86.589)\n",
      "[250/391]\tTime 0.161 (0.149)\tData 0.033 (0.038)\tLoss 0.4050 (0.4962)\tPrec@1 88.281 (86.625)\n",
      "[260/391]\tTime 0.120 (0.149)\tData 0.027 (0.038)\tLoss 0.5151 (0.4969)\tPrec@1 85.156 (86.584)\n",
      "[270/391]\tTime 0.157 (0.149)\tData 0.034 (0.038)\tLoss 0.5910 (0.4959)\tPrec@1 82.031 (86.624)\n",
      "[280/391]\tTime 0.138 (0.149)\tData 0.032 (0.038)\tLoss 0.5307 (0.4966)\tPrec@1 82.031 (86.580)\n",
      "[290/391]\tTime 0.156 (0.148)\tData 0.047 (0.038)\tLoss 0.5920 (0.4964)\tPrec@1 82.812 (86.563)\n",
      "[300/391]\tTime 0.156 (0.148)\tData 0.034 (0.038)\tLoss 0.3376 (0.4968)\tPrec@1 92.188 (86.542)\n",
      "[310/391]\tTime 0.116 (0.148)\tData 0.033 (0.038)\tLoss 0.4926 (0.4974)\tPrec@1 85.938 (86.540)\n",
      "[320/391]\tTime 0.116 (0.148)\tData 0.030 (0.038)\tLoss 0.4999 (0.4969)\tPrec@1 82.812 (86.517)\n",
      "[330/391]\tTime 0.138 (0.148)\tData 0.033 (0.038)\tLoss 0.4804 (0.4973)\tPrec@1 89.062 (86.532)\n",
      "[340/391]\tTime 0.164 (0.148)\tData 0.035 (0.038)\tLoss 0.4573 (0.4979)\tPrec@1 88.281 (86.538)\n",
      "[350/391]\tTime 0.129 (0.148)\tData 0.037 (0.038)\tLoss 0.4510 (0.4982)\tPrec@1 85.938 (86.530)\n",
      "[360/391]\tTime 0.133 (0.147)\tData 0.033 (0.038)\tLoss 0.5858 (0.4976)\tPrec@1 85.156 (86.561)\n",
      "[370/391]\tTime 0.136 (0.147)\tData 0.033 (0.038)\tLoss 0.6291 (0.4983)\tPrec@1 81.250 (86.538)\n",
      "[380/391]\tTime 0.139 (0.147)\tData 0.031 (0.038)\tLoss 0.5696 (0.4983)\tPrec@1 84.375 (86.542)\n",
      "[390/391]\tTime 0.103 (0.147)\tData 0.024 (0.038)\tLoss 0.6231 (0.4985)\tPrec@1 82.500 (86.532)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.053 (0.053)\tLoss 0.7971 (0.7971)\tPrec@1 80.469 (80.469)\n",
      "Test: [10/79]\tTime 0.037 (0.043)\tLoss 0.4816 (0.8562)\tPrec@1 82.812 (80.043)\n",
      "Test: [20/79]\tTime 0.034 (0.043)\tLoss 0.9746 (0.9404)\tPrec@1 79.688 (79.204)\n",
      "Test: [30/79]\tTime 0.039 (0.043)\tLoss 0.7973 (0.9215)\tPrec@1 75.000 (78.982)\n",
      "Test: [40/79]\tTime 0.033 (0.041)\tLoss 1.0805 (0.9199)\tPrec@1 78.125 (79.249)\n",
      "Test: [50/79]\tTime 0.040 (0.040)\tLoss 1.0699 (0.9306)\tPrec@1 76.562 (79.121)\n",
      "Test: [60/79]\tTime 0.036 (0.040)\tLoss 0.9719 (0.9289)\tPrec@1 82.031 (79.124)\n",
      "Test: [70/79]\tTime 0.035 (0.040)\tLoss 0.7552 (0.9388)\tPrec@1 81.250 (78.862)\n",
      " * Prec@1 78.770\n",
      "\n",
      "===> epoch: 25/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.180 (0.180)\tData 0.044 (0.044)\tLoss 0.5434 (0.5434)\tPrec@1 85.156 (85.156)\n",
      "[10/391]\tTime 0.147 (0.160)\tData 0.037 (0.040)\tLoss 0.5659 (0.4884)\tPrec@1 84.375 (87.145)\n",
      "[20/391]\tTime 0.166 (0.156)\tData 0.035 (0.038)\tLoss 0.4973 (0.4743)\tPrec@1 88.281 (87.723)\n",
      "[30/391]\tTime 0.152 (0.154)\tData 0.027 (0.039)\tLoss 0.4304 (0.4726)\tPrec@1 85.938 (87.500)\n",
      "[40/391]\tTime 0.160 (0.153)\tData 0.031 (0.039)\tLoss 0.5598 (0.4888)\tPrec@1 82.812 (86.909)\n",
      "[50/391]\tTime 0.120 (0.150)\tData 0.035 (0.039)\tLoss 0.4174 (0.4848)\tPrec@1 87.500 (87.148)\n",
      "[60/391]\tTime 0.181 (0.150)\tData 0.046 (0.038)\tLoss 0.5817 (0.4855)\tPrec@1 83.594 (87.308)\n",
      "[70/391]\tTime 0.146 (0.149)\tData 0.031 (0.037)\tLoss 0.5073 (0.4930)\tPrec@1 87.500 (86.807)\n",
      "[80/391]\tTime 0.148 (0.150)\tData 0.028 (0.037)\tLoss 0.3087 (0.4920)\tPrec@1 90.625 (86.844)\n",
      "[90/391]\tTime 0.150 (0.150)\tData 0.044 (0.037)\tLoss 0.4774 (0.4909)\tPrec@1 87.500 (86.830)\n",
      "[100/391]\tTime 0.203 (0.152)\tData 0.069 (0.038)\tLoss 0.3952 (0.4926)\tPrec@1 88.281 (86.757)\n",
      "[110/391]\tTime 0.150 (0.151)\tData 0.034 (0.038)\tLoss 0.3791 (0.4880)\tPrec@1 89.844 (86.845)\n",
      "[120/391]\tTime 0.151 (0.152)\tData 0.037 (0.038)\tLoss 0.4446 (0.4885)\tPrec@1 89.062 (86.854)\n",
      "[130/391]\tTime 0.168 (0.152)\tData 0.034 (0.038)\tLoss 0.4631 (0.4906)\tPrec@1 84.375 (86.766)\n",
      "[140/391]\tTime 0.150 (0.151)\tData 0.025 (0.038)\tLoss 0.5360 (0.4906)\tPrec@1 82.812 (86.691)\n",
      "[150/391]\tTime 0.138 (0.151)\tData 0.033 (0.037)\tLoss 0.5006 (0.4924)\tPrec@1 85.938 (86.667)\n",
      "[160/391]\tTime 0.177 (0.151)\tData 0.067 (0.037)\tLoss 0.4289 (0.4912)\tPrec@1 87.500 (86.719)\n",
      "[170/391]\tTime 0.137 (0.151)\tData 0.029 (0.037)\tLoss 0.4468 (0.4923)\tPrec@1 88.281 (86.646)\n",
      "[180/391]\tTime 0.130 (0.151)\tData 0.034 (0.037)\tLoss 0.5417 (0.4936)\tPrec@1 87.500 (86.637)\n",
      "[190/391]\tTime 0.169 (0.151)\tData 0.046 (0.038)\tLoss 0.4572 (0.4914)\tPrec@1 88.281 (86.731)\n",
      "[200/391]\tTime 0.161 (0.151)\tData 0.029 (0.038)\tLoss 0.4744 (0.4927)\tPrec@1 85.938 (86.695)\n",
      "[210/391]\tTime 0.139 (0.150)\tData 0.039 (0.038)\tLoss 0.5602 (0.4917)\tPrec@1 87.500 (86.734)\n",
      "[220/391]\tTime 0.166 (0.149)\tData 0.034 (0.037)\tLoss 0.4850 (0.4926)\tPrec@1 88.281 (86.687)\n",
      "[230/391]\tTime 0.162 (0.150)\tData 0.038 (0.038)\tLoss 0.4774 (0.4934)\tPrec@1 85.156 (86.604)\n",
      "[240/391]\tTime 0.137 (0.150)\tData 0.035 (0.038)\tLoss 0.4387 (0.4917)\tPrec@1 86.719 (86.657)\n",
      "[250/391]\tTime 0.123 (0.150)\tData 0.034 (0.038)\tLoss 0.5581 (0.4893)\tPrec@1 84.375 (86.688)\n",
      "[260/391]\tTime 0.121 (0.149)\tData 0.035 (0.038)\tLoss 0.4586 (0.4903)\tPrec@1 87.500 (86.671)\n",
      "[270/391]\tTime 0.135 (0.148)\tData 0.038 (0.038)\tLoss 0.4563 (0.4897)\tPrec@1 89.062 (86.667)\n",
      "[280/391]\tTime 0.137 (0.148)\tData 0.034 (0.038)\tLoss 0.5194 (0.4897)\tPrec@1 85.938 (86.677)\n",
      "[290/391]\tTime 0.139 (0.148)\tData 0.031 (0.038)\tLoss 0.4365 (0.4893)\tPrec@1 89.844 (86.729)\n",
      "[300/391]\tTime 0.187 (0.149)\tData 0.071 (0.038)\tLoss 0.3772 (0.4890)\tPrec@1 89.844 (86.732)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[310/391]\tTime 0.146 (0.149)\tData 0.037 (0.038)\tLoss 0.5303 (0.4887)\tPrec@1 83.594 (86.761)\n",
      "[320/391]\tTime 0.111 (0.149)\tData 0.025 (0.038)\tLoss 0.4572 (0.4892)\tPrec@1 85.156 (86.736)\n",
      "[330/391]\tTime 0.146 (0.149)\tData 0.049 (0.038)\tLoss 0.4672 (0.4896)\tPrec@1 89.844 (86.740)\n",
      "[340/391]\tTime 0.161 (0.149)\tData 0.033 (0.038)\tLoss 0.6009 (0.4914)\tPrec@1 82.812 (86.675)\n",
      "[350/391]\tTime 0.149 (0.149)\tData 0.028 (0.038)\tLoss 0.5416 (0.4924)\tPrec@1 87.500 (86.641)\n",
      "[360/391]\tTime 0.160 (0.149)\tData 0.065 (0.038)\tLoss 0.5137 (0.4931)\tPrec@1 84.375 (86.604)\n",
      "[370/391]\tTime 0.141 (0.149)\tData 0.034 (0.038)\tLoss 0.4526 (0.4930)\tPrec@1 87.500 (86.605)\n",
      "[380/391]\tTime 0.154 (0.149)\tData 0.032 (0.038)\tLoss 0.4787 (0.4933)\tPrec@1 86.719 (86.579)\n",
      "[390/391]\tTime 0.110 (0.150)\tData 0.025 (0.038)\tLoss 0.5694 (0.4936)\tPrec@1 78.750 (86.568)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.059 (0.059)\tLoss 0.5303 (0.5303)\tPrec@1 83.594 (83.594)\n",
      "Test: [10/79]\tTime 0.036 (0.037)\tLoss 0.5705 (0.6285)\tPrec@1 86.719 (83.594)\n",
      "Test: [20/79]\tTime 0.037 (0.037)\tLoss 0.6979 (0.6473)\tPrec@1 82.031 (83.482)\n",
      "Test: [30/79]\tTime 0.038 (0.039)\tLoss 0.5976 (0.6399)\tPrec@1 82.812 (83.342)\n",
      "Test: [40/79]\tTime 0.043 (0.040)\tLoss 0.7378 (0.6411)\tPrec@1 81.250 (83.651)\n",
      "Test: [50/79]\tTime 0.036 (0.040)\tLoss 0.7664 (0.6549)\tPrec@1 81.250 (83.609)\n",
      "Test: [60/79]\tTime 0.026 (0.039)\tLoss 0.6599 (0.6625)\tPrec@1 79.688 (83.453)\n",
      "Test: [70/79]\tTime 0.028 (0.038)\tLoss 0.6272 (0.6667)\tPrec@1 84.375 (83.231)\n",
      " * Prec@1 83.250\n",
      "\n",
      "===> epoch: 26/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.133 (0.133)\tData 0.030 (0.030)\tLoss 0.4233 (0.4233)\tPrec@1 85.156 (85.156)\n",
      "[10/391]\tTime 0.213 (0.159)\tData 0.094 (0.045)\tLoss 0.4336 (0.4641)\tPrec@1 89.062 (86.151)\n",
      "[20/391]\tTime 0.136 (0.159)\tData 0.039 (0.044)\tLoss 0.4619 (0.4612)\tPrec@1 86.719 (86.942)\n",
      "[30/391]\tTime 0.163 (0.151)\tData 0.027 (0.039)\tLoss 0.5550 (0.4829)\tPrec@1 85.938 (86.593)\n",
      "[40/391]\tTime 0.143 (0.150)\tData 0.042 (0.039)\tLoss 0.4476 (0.4874)\tPrec@1 89.844 (86.585)\n",
      "[50/391]\tTime 0.155 (0.153)\tData 0.038 (0.040)\tLoss 0.5067 (0.4865)\tPrec@1 85.156 (86.627)\n",
      "[60/391]\tTime 0.143 (0.152)\tData 0.028 (0.039)\tLoss 0.4232 (0.4842)\tPrec@1 87.500 (86.680)\n",
      "[70/391]\tTime 0.186 (0.153)\tData 0.070 (0.039)\tLoss 0.4665 (0.4824)\tPrec@1 87.500 (86.774)\n",
      "[80/391]\tTime 0.127 (0.152)\tData 0.029 (0.038)\tLoss 0.4349 (0.4818)\tPrec@1 86.719 (86.651)\n",
      "[90/391]\tTime 0.140 (0.150)\tData 0.031 (0.038)\tLoss 0.4155 (0.4817)\tPrec@1 89.062 (86.573)\n",
      "[100/391]\tTime 0.165 (0.149)\tData 0.038 (0.038)\tLoss 0.4279 (0.4845)\tPrec@1 86.719 (86.518)\n",
      "[110/391]\tTime 0.153 (0.150)\tData 0.030 (0.038)\tLoss 0.4463 (0.4851)\tPrec@1 85.938 (86.522)\n",
      "[120/391]\tTime 0.156 (0.150)\tData 0.039 (0.037)\tLoss 0.6048 (0.4877)\tPrec@1 82.812 (86.428)\n",
      "[130/391]\tTime 0.138 (0.150)\tData 0.034 (0.038)\tLoss 0.4005 (0.4849)\tPrec@1 89.844 (86.546)\n",
      "[140/391]\tTime 0.169 (0.149)\tData 0.054 (0.038)\tLoss 0.4571 (0.4835)\tPrec@1 88.281 (86.625)\n",
      "[150/391]\tTime 0.165 (0.149)\tData 0.051 (0.038)\tLoss 0.6504 (0.4865)\tPrec@1 78.906 (86.584)\n",
      "[160/391]\tTime 0.145 (0.148)\tData 0.039 (0.038)\tLoss 0.5537 (0.4887)\tPrec@1 84.375 (86.568)\n",
      "[170/391]\tTime 0.177 (0.148)\tData 0.062 (0.038)\tLoss 0.4588 (0.4893)\tPrec@1 87.500 (86.563)\n",
      "[180/391]\tTime 0.138 (0.148)\tData 0.037 (0.038)\tLoss 0.5898 (0.4907)\tPrec@1 84.375 (86.507)\n",
      "[190/391]\tTime 0.136 (0.149)\tData 0.035 (0.039)\tLoss 0.6097 (0.4921)\tPrec@1 82.812 (86.510)\n",
      "[200/391]\tTime 0.162 (0.150)\tData 0.034 (0.039)\tLoss 0.4028 (0.4932)\tPrec@1 88.281 (86.454)\n",
      "[210/391]\tTime 0.126 (0.150)\tData 0.039 (0.039)\tLoss 0.4439 (0.4934)\tPrec@1 89.062 (86.430)\n",
      "[220/391]\tTime 0.163 (0.150)\tData 0.051 (0.039)\tLoss 0.3844 (0.4909)\tPrec@1 89.062 (86.524)\n",
      "[230/391]\tTime 0.145 (0.150)\tData 0.029 (0.039)\tLoss 0.5684 (0.4908)\tPrec@1 82.812 (86.533)\n",
      "[240/391]\tTime 0.147 (0.150)\tData 0.035 (0.039)\tLoss 0.5849 (0.4916)\tPrec@1 83.594 (86.544)\n",
      "[250/391]\tTime 0.160 (0.150)\tData 0.039 (0.039)\tLoss 0.5019 (0.4918)\tPrec@1 82.031 (86.526)\n",
      "[260/391]\tTime 0.128 (0.150)\tData 0.028 (0.039)\tLoss 0.4879 (0.4932)\tPrec@1 85.156 (86.443)\n",
      "[270/391]\tTime 0.158 (0.150)\tData 0.039 (0.039)\tLoss 0.4270 (0.4933)\tPrec@1 87.500 (86.456)\n",
      "[280/391]\tTime 0.118 (0.149)\tData 0.027 (0.038)\tLoss 0.4685 (0.4928)\tPrec@1 86.719 (86.471)\n",
      "[290/391]\tTime 0.146 (0.150)\tData 0.048 (0.039)\tLoss 0.5613 (0.4927)\tPrec@1 85.156 (86.458)\n",
      "[300/391]\tTime 0.163 (0.150)\tData 0.036 (0.039)\tLoss 0.6637 (0.4919)\tPrec@1 84.375 (86.493)\n",
      "[310/391]\tTime 0.192 (0.150)\tData 0.037 (0.039)\tLoss 0.4917 (0.4916)\tPrec@1 86.719 (86.528)\n",
      "[320/391]\tTime 0.131 (0.150)\tData 0.038 (0.039)\tLoss 0.4829 (0.4910)\tPrec@1 87.500 (86.556)\n",
      "[330/391]\tTime 0.114 (0.150)\tData 0.031 (0.039)\tLoss 0.4361 (0.4914)\tPrec@1 88.281 (86.568)\n",
      "[340/391]\tTime 0.135 (0.150)\tData 0.036 (0.039)\tLoss 0.5115 (0.4918)\tPrec@1 87.500 (86.563)\n",
      "[350/391]\tTime 0.166 (0.150)\tData 0.038 (0.039)\tLoss 0.3789 (0.4916)\tPrec@1 90.625 (86.599)\n",
      "[360/391]\tTime 0.162 (0.150)\tData 0.034 (0.039)\tLoss 0.5177 (0.4912)\tPrec@1 87.500 (86.652)\n",
      "[370/391]\tTime 0.130 (0.150)\tData 0.033 (0.038)\tLoss 0.4422 (0.4913)\tPrec@1 85.938 (86.664)\n",
      "[380/391]\tTime 0.157 (0.150)\tData 0.031 (0.038)\tLoss 0.4973 (0.4926)\tPrec@1 86.719 (86.626)\n",
      "[390/391]\tTime 0.157 (0.150)\tData 0.034 (0.038)\tLoss 0.3344 (0.4929)\tPrec@1 93.750 (86.628)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.033 (0.033)\tLoss 0.8718 (0.8718)\tPrec@1 77.344 (77.344)\n",
      "Test: [10/79]\tTime 0.030 (0.031)\tLoss 1.0923 (1.0931)\tPrec@1 79.688 (75.213)\n",
      "Test: [20/79]\tTime 0.026 (0.029)\tLoss 0.7877 (1.1417)\tPrec@1 79.688 (75.372)\n",
      "Test: [30/79]\tTime 0.041 (0.032)\tLoss 1.1078 (1.1481)\tPrec@1 73.438 (75.176)\n",
      "Test: [40/79]\tTime 0.050 (0.035)\tLoss 1.4698 (1.1559)\tPrec@1 71.094 (75.457)\n",
      "Test: [50/79]\tTime 0.035 (0.037)\tLoss 1.3570 (1.1349)\tPrec@1 71.094 (75.797)\n",
      "Test: [60/79]\tTime 0.033 (0.037)\tLoss 1.2082 (1.1328)\tPrec@1 79.688 (75.768)\n",
      "Test: [70/79]\tTime 0.045 (0.038)\tLoss 0.9966 (1.1426)\tPrec@1 74.219 (75.605)\n",
      " * Prec@1 75.720\n",
      "\n",
      "===> epoch: 27/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.150 (0.150)\tData 0.036 (0.036)\tLoss 0.4777 (0.4777)\tPrec@1 86.719 (86.719)\n",
      "[10/391]\tTime 0.182 (0.145)\tData 0.037 (0.038)\tLoss 0.4714 (0.5240)\tPrec@1 86.719 (85.795)\n",
      "[20/391]\tTime 0.128 (0.144)\tData 0.032 (0.035)\tLoss 0.4980 (0.4953)\tPrec@1 85.156 (86.682)\n",
      "[30/391]\tTime 0.157 (0.146)\tData 0.036 (0.037)\tLoss 0.5268 (0.4760)\tPrec@1 83.594 (87.298)\n",
      "[40/391]\tTime 0.126 (0.144)\tData 0.024 (0.037)\tLoss 0.4162 (0.4715)\tPrec@1 90.625 (87.443)\n",
      "[50/391]\tTime 0.143 (0.148)\tData 0.032 (0.038)\tLoss 0.4153 (0.4717)\tPrec@1 90.625 (87.393)\n",
      "[60/391]\tTime 0.131 (0.144)\tData 0.036 (0.038)\tLoss 0.4695 (0.4703)\tPrec@1 85.938 (87.500)\n",
      "[70/391]\tTime 0.143 (0.142)\tData 0.047 (0.038)\tLoss 0.4823 (0.4646)\tPrec@1 85.156 (87.555)\n",
      "[80/391]\tTime 0.231 (0.144)\tData 0.059 (0.039)\tLoss 0.4827 (0.4617)\tPrec@1 87.500 (87.751)\n",
      "[90/391]\tTime 0.132 (0.145)\tData 0.032 (0.039)\tLoss 0.5199 (0.4602)\tPrec@1 87.500 (87.749)\n",
      "[100/391]\tTime 0.137 (0.145)\tData 0.027 (0.038)\tLoss 0.7348 (0.4665)\tPrec@1 77.344 (87.485)\n",
      "[110/391]\tTime 0.127 (0.144)\tData 0.033 (0.038)\tLoss 0.5738 (0.4731)\tPrec@1 85.938 (87.401)\n",
      "[120/391]\tTime 0.187 (0.146)\tData 0.060 (0.038)\tLoss 0.5022 (0.4773)\tPrec@1 89.844 (87.306)\n",
      "[130/391]\tTime 0.174 (0.147)\tData 0.044 (0.039)\tLoss 0.4448 (0.4770)\tPrec@1 87.500 (87.267)\n",
      "[140/391]\tTime 0.140 (0.146)\tData 0.039 (0.038)\tLoss 0.4480 (0.4795)\tPrec@1 89.844 (87.112)\n",
      "[150/391]\tTime 0.130 (0.147)\tData 0.038 (0.039)\tLoss 0.5141 (0.4793)\tPrec@1 84.375 (87.071)\n",
      "[160/391]\tTime 0.151 (0.146)\tData 0.036 (0.038)\tLoss 0.4238 (0.4771)\tPrec@1 87.500 (87.088)\n",
      "[170/391]\tTime 0.118 (0.146)\tData 0.036 (0.038)\tLoss 0.4450 (0.4766)\tPrec@1 87.500 (87.052)\n",
      "[180/391]\tTime 0.133 (0.146)\tData 0.041 (0.038)\tLoss 0.5130 (0.4787)\tPrec@1 85.156 (86.960)\n",
      "[190/391]\tTime 0.135 (0.147)\tData 0.035 (0.039)\tLoss 0.5742 (0.4801)\tPrec@1 82.812 (86.911)\n",
      "[200/391]\tTime 0.142 (0.147)\tData 0.038 (0.038)\tLoss 0.4175 (0.4803)\tPrec@1 91.406 (86.905)\n",
      "[210/391]\tTime 0.138 (0.148)\tData 0.039 (0.038)\tLoss 0.5158 (0.4808)\tPrec@1 86.719 (86.896)\n",
      "[220/391]\tTime 0.128 (0.147)\tData 0.030 (0.038)\tLoss 0.4444 (0.4801)\tPrec@1 86.719 (86.899)\n",
      "[230/391]\tTime 0.162 (0.148)\tData 0.041 (0.038)\tLoss 0.3988 (0.4793)\tPrec@1 89.844 (86.932)\n",
      "[240/391]\tTime 0.141 (0.148)\tData 0.033 (0.038)\tLoss 0.5188 (0.4800)\tPrec@1 86.719 (86.907)\n",
      "[250/391]\tTime 0.146 (0.148)\tData 0.050 (0.039)\tLoss 0.3924 (0.4789)\tPrec@1 88.281 (86.971)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[260/391]\tTime 0.146 (0.148)\tData 0.042 (0.039)\tLoss 0.4098 (0.4796)\tPrec@1 90.625 (86.934)\n",
      "[270/391]\tTime 0.149 (0.149)\tData 0.031 (0.039)\tLoss 0.4976 (0.4800)\tPrec@1 86.719 (86.921)\n",
      "[280/391]\tTime 0.144 (0.149)\tData 0.030 (0.039)\tLoss 0.3984 (0.4799)\tPrec@1 92.188 (86.950)\n",
      "[290/391]\tTime 0.165 (0.149)\tData 0.044 (0.039)\tLoss 0.4600 (0.4817)\tPrec@1 86.719 (86.907)\n",
      "[300/391]\tTime 0.148 (0.148)\tData 0.049 (0.039)\tLoss 0.4584 (0.4808)\tPrec@1 90.625 (86.950)\n",
      "[310/391]\tTime 0.131 (0.148)\tData 0.030 (0.039)\tLoss 0.4855 (0.4810)\tPrec@1 86.719 (86.952)\n",
      "[320/391]\tTime 0.186 (0.148)\tData 0.046 (0.039)\tLoss 0.4341 (0.4809)\tPrec@1 84.375 (86.938)\n",
      "[330/391]\tTime 0.138 (0.148)\tData 0.036 (0.039)\tLoss 0.3860 (0.4798)\tPrec@1 89.844 (86.981)\n",
      "[340/391]\tTime 0.128 (0.148)\tData 0.027 (0.039)\tLoss 0.5789 (0.4801)\tPrec@1 88.281 (87.007)\n",
      "[350/391]\tTime 0.156 (0.148)\tData 0.040 (0.039)\tLoss 0.3782 (0.4798)\tPrec@1 89.844 (87.019)\n",
      "[360/391]\tTime 0.128 (0.148)\tData 0.032 (0.039)\tLoss 0.6022 (0.4808)\tPrec@1 83.594 (86.978)\n",
      "[370/391]\tTime 0.181 (0.148)\tData 0.044 (0.039)\tLoss 0.6158 (0.4810)\tPrec@1 78.906 (86.965)\n",
      "[380/391]\tTime 0.115 (0.148)\tData 0.035 (0.039)\tLoss 0.5584 (0.4817)\tPrec@1 83.594 (86.948)\n",
      "[390/391]\tTime 0.101 (0.147)\tData 0.026 (0.039)\tLoss 0.3801 (0.4830)\tPrec@1 90.000 (86.906)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.046 (0.046)\tLoss 0.9002 (0.9002)\tPrec@1 78.906 (78.906)\n",
      "Test: [10/79]\tTime 0.046 (0.037)\tLoss 1.1345 (0.9503)\tPrec@1 76.562 (79.688)\n",
      "Test: [20/79]\tTime 0.043 (0.036)\tLoss 0.7077 (0.9803)\tPrec@1 79.688 (78.869)\n",
      "Test: [30/79]\tTime 0.029 (0.037)\tLoss 0.8075 (0.9874)\tPrec@1 79.688 (78.553)\n",
      "Test: [40/79]\tTime 0.030 (0.035)\tLoss 1.2205 (0.9734)\tPrec@1 70.312 (78.487)\n",
      "Test: [50/79]\tTime 0.031 (0.034)\tLoss 1.0975 (0.9682)\tPrec@1 79.688 (78.585)\n",
      "Test: [60/79]\tTime 0.032 (0.035)\tLoss 1.1138 (0.9750)\tPrec@1 75.781 (78.445)\n",
      "Test: [70/79]\tTime 0.036 (0.035)\tLoss 0.8378 (0.9676)\tPrec@1 78.906 (78.510)\n",
      " * Prec@1 78.410\n",
      "\n",
      "===> epoch: 28/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.151 (0.151)\tData 0.036 (0.036)\tLoss 0.5481 (0.5481)\tPrec@1 84.375 (84.375)\n",
      "[10/391]\tTime 0.167 (0.143)\tData 0.046 (0.036)\tLoss 0.4062 (0.4552)\tPrec@1 89.062 (87.997)\n",
      "[20/391]\tTime 0.143 (0.145)\tData 0.029 (0.035)\tLoss 0.5475 (0.4641)\tPrec@1 85.156 (87.872)\n",
      "[30/391]\tTime 0.168 (0.144)\tData 0.035 (0.035)\tLoss 0.4865 (0.4636)\tPrec@1 88.281 (87.777)\n",
      "[40/391]\tTime 0.144 (0.146)\tData 0.033 (0.038)\tLoss 0.4121 (0.4621)\tPrec@1 90.625 (87.786)\n",
      "[50/391]\tTime 0.167 (0.143)\tData 0.035 (0.037)\tLoss 0.4755 (0.4688)\tPrec@1 85.938 (87.377)\n",
      "[60/391]\tTime 0.143 (0.143)\tData 0.033 (0.036)\tLoss 0.4155 (0.4721)\tPrec@1 87.500 (87.423)\n",
      "[70/391]\tTime 0.139 (0.144)\tData 0.033 (0.036)\tLoss 0.4497 (0.4720)\tPrec@1 87.500 (87.313)\n",
      "[80/391]\tTime 0.146 (0.144)\tData 0.031 (0.036)\tLoss 0.4208 (0.4731)\tPrec@1 88.281 (87.346)\n",
      "[90/391]\tTime 0.158 (0.147)\tData 0.036 (0.037)\tLoss 0.4887 (0.4770)\tPrec@1 84.375 (87.191)\n",
      "[100/391]\tTime 0.158 (0.146)\tData 0.028 (0.037)\tLoss 0.6176 (0.4788)\tPrec@1 85.938 (87.167)\n",
      "[110/391]\tTime 0.143 (0.146)\tData 0.030 (0.037)\tLoss 0.5370 (0.4817)\tPrec@1 85.156 (87.071)\n",
      "[120/391]\tTime 0.215 (0.148)\tData 0.063 (0.038)\tLoss 0.3297 (0.4825)\tPrec@1 92.969 (87.113)\n",
      "[130/391]\tTime 0.164 (0.148)\tData 0.037 (0.038)\tLoss 0.5073 (0.4809)\tPrec@1 85.156 (87.166)\n",
      "[140/391]\tTime 0.142 (0.149)\tData 0.032 (0.038)\tLoss 0.6777 (0.4819)\tPrec@1 78.906 (87.101)\n",
      "[150/391]\tTime 0.167 (0.149)\tData 0.051 (0.038)\tLoss 0.4485 (0.4818)\tPrec@1 89.062 (87.159)\n",
      "[160/391]\tTime 0.191 (0.150)\tData 0.047 (0.038)\tLoss 0.6488 (0.4838)\tPrec@1 83.594 (87.039)\n",
      "[170/391]\tTime 0.156 (0.149)\tData 0.039 (0.038)\tLoss 0.5499 (0.4832)\tPrec@1 85.938 (87.075)\n",
      "[180/391]\tTime 0.157 (0.150)\tData 0.036 (0.039)\tLoss 0.5969 (0.4841)\tPrec@1 85.156 (87.038)\n",
      "[190/391]\tTime 0.101 (0.150)\tData 0.031 (0.039)\tLoss 0.4897 (0.4857)\tPrec@1 87.500 (86.960)\n",
      "[200/391]\tTime 0.127 (0.149)\tData 0.034 (0.039)\tLoss 0.4520 (0.4881)\tPrec@1 89.062 (86.901)\n",
      "[210/391]\tTime 0.150 (0.149)\tData 0.033 (0.039)\tLoss 0.4736 (0.4896)\tPrec@1 89.844 (86.908)\n",
      "[220/391]\tTime 0.135 (0.149)\tData 0.033 (0.039)\tLoss 0.4250 (0.4895)\tPrec@1 89.844 (86.888)\n",
      "[230/391]\tTime 0.153 (0.149)\tData 0.036 (0.039)\tLoss 0.3544 (0.4883)\tPrec@1 90.625 (86.952)\n",
      "[240/391]\tTime 0.193 (0.150)\tData 0.045 (0.039)\tLoss 0.4077 (0.4885)\tPrec@1 89.062 (86.939)\n",
      "[250/391]\tTime 0.169 (0.150)\tData 0.055 (0.039)\tLoss 0.5975 (0.4889)\tPrec@1 83.594 (86.930)\n",
      "[260/391]\tTime 0.162 (0.150)\tData 0.032 (0.039)\tLoss 0.3836 (0.4907)\tPrec@1 91.406 (86.886)\n",
      "[270/391]\tTime 0.117 (0.150)\tData 0.033 (0.039)\tLoss 0.5231 (0.4905)\tPrec@1 85.938 (86.892)\n",
      "[280/391]\tTime 0.122 (0.149)\tData 0.029 (0.039)\tLoss 0.4228 (0.4924)\tPrec@1 91.406 (86.877)\n",
      "[290/391]\tTime 0.133 (0.149)\tData 0.040 (0.039)\tLoss 0.4571 (0.4939)\tPrec@1 86.719 (86.818)\n",
      "[300/391]\tTime 0.142 (0.148)\tData 0.031 (0.039)\tLoss 0.4557 (0.4943)\tPrec@1 86.719 (86.815)\n",
      "[310/391]\tTime 0.144 (0.149)\tData 0.032 (0.039)\tLoss 0.4433 (0.4936)\tPrec@1 89.062 (86.842)\n",
      "[320/391]\tTime 0.167 (0.149)\tData 0.065 (0.039)\tLoss 0.7299 (0.4946)\tPrec@1 81.250 (86.819)\n",
      "[330/391]\tTime 0.178 (0.149)\tData 0.055 (0.039)\tLoss 0.4282 (0.4940)\tPrec@1 88.281 (86.820)\n",
      "[340/391]\tTime 0.160 (0.149)\tData 0.043 (0.039)\tLoss 0.5351 (0.4932)\tPrec@1 82.031 (86.875)\n",
      "[350/391]\tTime 0.132 (0.148)\tData 0.044 (0.039)\tLoss 0.4266 (0.4934)\tPrec@1 89.062 (86.830)\n",
      "[360/391]\tTime 0.112 (0.148)\tData 0.034 (0.039)\tLoss 0.6461 (0.4934)\tPrec@1 78.906 (86.810)\n",
      "[370/391]\tTime 0.156 (0.148)\tData 0.039 (0.039)\tLoss 0.5294 (0.4933)\tPrec@1 86.719 (86.803)\n",
      "[380/391]\tTime 0.180 (0.148)\tData 0.056 (0.039)\tLoss 0.6052 (0.4933)\tPrec@1 81.250 (86.799)\n",
      "[390/391]\tTime 0.133 (0.148)\tData 0.026 (0.039)\tLoss 0.4871 (0.4944)\tPrec@1 87.500 (86.780)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.033 (0.033)\tLoss 0.7624 (0.7624)\tPrec@1 78.125 (78.125)\n",
      "Test: [10/79]\tTime 0.034 (0.037)\tLoss 1.0783 (1.0705)\tPrec@1 74.219 (74.574)\n",
      "Test: [20/79]\tTime 0.059 (0.040)\tLoss 1.2644 (1.1786)\tPrec@1 71.875 (72.991)\n",
      "Test: [30/79]\tTime 0.027 (0.038)\tLoss 1.0439 (1.1835)\tPrec@1 76.562 (72.984)\n",
      "Test: [40/79]\tTime 0.035 (0.036)\tLoss 1.3449 (1.1957)\tPrec@1 71.875 (73.056)\n",
      "Test: [50/79]\tTime 0.063 (0.037)\tLoss 1.2072 (1.1980)\tPrec@1 68.750 (73.131)\n",
      "Test: [60/79]\tTime 0.040 (0.036)\tLoss 1.3008 (1.1918)\tPrec@1 72.656 (73.220)\n",
      "Test: [70/79]\tTime 0.035 (0.036)\tLoss 1.3353 (1.1867)\tPrec@1 72.656 (73.261)\n",
      " * Prec@1 73.150\n",
      "\n",
      "===> epoch: 29/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.144 (0.144)\tData 0.039 (0.039)\tLoss 0.4159 (0.4159)\tPrec@1 91.406 (91.406)\n",
      "[10/391]\tTime 0.150 (0.157)\tData 0.054 (0.042)\tLoss 0.4962 (0.4931)\tPrec@1 85.938 (86.932)\n",
      "[20/391]\tTime 0.150 (0.151)\tData 0.036 (0.039)\tLoss 0.4611 (0.4735)\tPrec@1 83.594 (87.463)\n",
      "[30/391]\tTime 0.130 (0.150)\tData 0.034 (0.037)\tLoss 0.5486 (0.4788)\tPrec@1 87.500 (87.349)\n",
      "[40/391]\tTime 0.177 (0.153)\tData 0.045 (0.039)\tLoss 0.3868 (0.4761)\tPrec@1 92.188 (87.519)\n",
      "[50/391]\tTime 0.190 (0.155)\tData 0.057 (0.040)\tLoss 0.5384 (0.4772)\tPrec@1 86.719 (87.331)\n",
      "[60/391]\tTime 0.139 (0.156)\tData 0.036 (0.041)\tLoss 0.4673 (0.4774)\tPrec@1 86.719 (87.359)\n",
      "[70/391]\tTime 0.130 (0.154)\tData 0.027 (0.040)\tLoss 0.5856 (0.4817)\tPrec@1 82.812 (87.181)\n",
      "[80/391]\tTime 0.170 (0.154)\tData 0.034 (0.039)\tLoss 0.6504 (0.4835)\tPrec@1 84.375 (87.162)\n",
      "[90/391]\tTime 0.136 (0.152)\tData 0.038 (0.039)\tLoss 0.5064 (0.4816)\tPrec@1 85.938 (87.165)\n",
      "[100/391]\tTime 0.101 (0.150)\tData 0.033 (0.039)\tLoss 0.5750 (0.4815)\tPrec@1 82.812 (87.106)\n",
      "[110/391]\tTime 0.138 (0.148)\tData 0.035 (0.039)\tLoss 0.6221 (0.4872)\tPrec@1 80.469 (86.930)\n",
      "[120/391]\tTime 0.170 (0.149)\tData 0.049 (0.039)\tLoss 0.5500 (0.4917)\tPrec@1 85.938 (86.816)\n",
      "[130/391]\tTime 0.149 (0.150)\tData 0.035 (0.039)\tLoss 0.4882 (0.4957)\tPrec@1 90.625 (86.760)\n",
      "[140/391]\tTime 0.151 (0.150)\tData 0.036 (0.039)\tLoss 0.4092 (0.4961)\tPrec@1 90.625 (86.724)\n",
      "[150/391]\tTime 0.183 (0.151)\tData 0.070 (0.039)\tLoss 0.4704 (0.4957)\tPrec@1 88.281 (86.698)\n",
      "[160/391]\tTime 0.168 (0.151)\tData 0.031 (0.039)\tLoss 0.5641 (0.4949)\tPrec@1 86.719 (86.728)\n",
      "[170/391]\tTime 0.177 (0.151)\tData 0.051 (0.039)\tLoss 0.4632 (0.4931)\tPrec@1 85.156 (86.760)\n",
      "[180/391]\tTime 0.111 (0.151)\tData 0.032 (0.039)\tLoss 0.5303 (0.4921)\tPrec@1 84.375 (86.775)\n",
      "[190/391]\tTime 0.135 (0.150)\tData 0.043 (0.039)\tLoss 0.6477 (0.4942)\tPrec@1 78.906 (86.674)\n",
      "[200/391]\tTime 0.207 (0.150)\tData 0.061 (0.039)\tLoss 0.5426 (0.4942)\tPrec@1 86.719 (86.657)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[210/391]\tTime 0.151 (0.151)\tData 0.032 (0.040)\tLoss 0.4953 (0.4933)\tPrec@1 88.281 (86.678)\n",
      "[220/391]\tTime 0.156 (0.152)\tData 0.035 (0.040)\tLoss 0.4811 (0.4941)\tPrec@1 84.375 (86.645)\n",
      "[230/391]\tTime 0.146 (0.152)\tData 0.043 (0.040)\tLoss 0.4653 (0.4937)\tPrec@1 85.156 (86.624)\n",
      "[240/391]\tTime 0.143 (0.151)\tData 0.035 (0.040)\tLoss 0.4727 (0.4935)\tPrec@1 86.719 (86.625)\n",
      "[250/391]\tTime 0.188 (0.151)\tData 0.025 (0.040)\tLoss 0.3442 (0.4926)\tPrec@1 90.625 (86.669)\n",
      "[260/391]\tTime 0.144 (0.152)\tData 0.034 (0.040)\tLoss 0.5672 (0.4907)\tPrec@1 86.719 (86.755)\n",
      "[270/391]\tTime 0.133 (0.153)\tData 0.032 (0.040)\tLoss 0.5315 (0.4908)\tPrec@1 88.281 (86.750)\n",
      "[280/391]\tTime 0.167 (0.153)\tData 0.039 (0.040)\tLoss 0.5936 (0.4923)\tPrec@1 82.812 (86.652)\n",
      "[290/391]\tTime 0.161 (0.153)\tData 0.036 (0.040)\tLoss 0.4875 (0.4940)\tPrec@1 85.938 (86.598)\n",
      "[300/391]\tTime 0.117 (0.153)\tData 0.028 (0.040)\tLoss 0.4848 (0.4936)\tPrec@1 87.500 (86.625)\n",
      "[310/391]\tTime 0.128 (0.152)\tData 0.037 (0.040)\tLoss 0.5441 (0.4932)\tPrec@1 85.938 (86.653)\n",
      "[320/391]\tTime 0.156 (0.152)\tData 0.033 (0.040)\tLoss 0.5878 (0.4939)\tPrec@1 82.812 (86.638)\n",
      "[330/391]\tTime 0.189 (0.152)\tData 0.052 (0.040)\tLoss 0.4305 (0.4936)\tPrec@1 89.062 (86.648)\n",
      "[340/391]\tTime 0.150 (0.153)\tData 0.039 (0.040)\tLoss 0.5140 (0.4933)\tPrec@1 85.156 (86.675)\n",
      "[350/391]\tTime 0.141 (0.153)\tData 0.033 (0.040)\tLoss 0.4290 (0.4941)\tPrec@1 89.062 (86.663)\n",
      "[360/391]\tTime 0.153 (0.153)\tData 0.033 (0.040)\tLoss 0.5170 (0.4943)\tPrec@1 87.500 (86.671)\n",
      "[370/391]\tTime 0.168 (0.153)\tData 0.038 (0.040)\tLoss 0.3900 (0.4926)\tPrec@1 86.719 (86.712)\n",
      "[380/391]\tTime 0.139 (0.153)\tData 0.035 (0.040)\tLoss 0.5600 (0.4922)\tPrec@1 82.812 (86.704)\n",
      "[390/391]\tTime 0.101 (0.152)\tData 0.021 (0.040)\tLoss 0.5827 (0.4938)\tPrec@1 81.250 (86.644)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.041 (0.041)\tLoss 0.4435 (0.4435)\tPrec@1 89.062 (89.062)\n",
      "Test: [10/79]\tTime 0.097 (0.045)\tLoss 0.5696 (0.6518)\tPrec@1 84.375 (83.381)\n",
      "Test: [20/79]\tTime 0.037 (0.045)\tLoss 0.7886 (0.6810)\tPrec@1 82.812 (83.110)\n",
      "Test: [30/79]\tTime 0.051 (0.044)\tLoss 0.5428 (0.6697)\tPrec@1 86.719 (83.543)\n",
      "Test: [40/79]\tTime 0.037 (0.043)\tLoss 0.6567 (0.6680)\tPrec@1 84.375 (83.765)\n",
      "Test: [50/79]\tTime 0.037 (0.042)\tLoss 0.5167 (0.6536)\tPrec@1 82.031 (83.900)\n",
      "Test: [60/79]\tTime 0.038 (0.041)\tLoss 0.5582 (0.6501)\tPrec@1 85.938 (83.888)\n",
      "Test: [70/79]\tTime 0.042 (0.042)\tLoss 0.5958 (0.6544)\tPrec@1 82.812 (83.682)\n",
      " * Prec@1 83.770\n",
      "\n",
      "===> epoch: 30/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.134 (0.134)\tData 0.039 (0.039)\tLoss 0.4391 (0.4391)\tPrec@1 89.062 (89.062)\n",
      "[10/391]\tTime 0.184 (0.160)\tData 0.077 (0.042)\tLoss 0.4224 (0.4949)\tPrec@1 89.844 (87.216)\n",
      "[20/391]\tTime 0.134 (0.164)\tData 0.039 (0.045)\tLoss 0.5419 (0.4850)\tPrec@1 85.938 (87.612)\n",
      "[30/391]\tTime 0.120 (0.158)\tData 0.027 (0.042)\tLoss 0.4146 (0.4716)\tPrec@1 90.625 (88.054)\n",
      "[40/391]\tTime 0.157 (0.153)\tData 0.048 (0.041)\tLoss 0.3022 (0.4663)\tPrec@1 94.531 (88.110)\n",
      "[50/391]\tTime 0.125 (0.152)\tData 0.037 (0.040)\tLoss 0.4448 (0.4633)\tPrec@1 86.719 (88.036)\n",
      "[60/391]\tTime 0.135 (0.151)\tData 0.029 (0.039)\tLoss 0.4927 (0.4653)\tPrec@1 85.938 (87.846)\n",
      "[70/391]\tTime 0.140 (0.150)\tData 0.049 (0.039)\tLoss 0.4874 (0.4670)\tPrec@1 87.500 (87.775)\n",
      "[80/391]\tTime 0.111 (0.146)\tData 0.034 (0.039)\tLoss 0.4205 (0.4652)\tPrec@1 86.719 (87.712)\n",
      "[90/391]\tTime 0.123 (0.145)\tData 0.033 (0.039)\tLoss 0.4707 (0.4685)\tPrec@1 84.375 (87.534)\n",
      "[100/391]\tTime 0.146 (0.144)\tData 0.036 (0.038)\tLoss 0.5481 (0.4752)\tPrec@1 84.375 (87.399)\n",
      "[110/391]\tTime 0.143 (0.144)\tData 0.034 (0.038)\tLoss 0.4781 (0.4750)\tPrec@1 87.500 (87.416)\n",
      "[120/391]\tTime 0.142 (0.144)\tData 0.033 (0.038)\tLoss 0.5353 (0.4753)\tPrec@1 86.719 (87.397)\n",
      "[130/391]\tTime 0.169 (0.145)\tData 0.029 (0.038)\tLoss 0.4891 (0.4731)\tPrec@1 86.719 (87.434)\n",
      "[140/391]\tTime 0.179 (0.145)\tData 0.047 (0.038)\tLoss 0.4744 (0.4744)\tPrec@1 86.719 (87.356)\n",
      "[150/391]\tTime 0.150 (0.147)\tData 0.035 (0.038)\tLoss 0.5800 (0.4743)\tPrec@1 84.375 (87.324)\n",
      "[160/391]\tTime 0.151 (0.147)\tData 0.030 (0.039)\tLoss 0.3969 (0.4735)\tPrec@1 91.406 (87.384)\n",
      "[170/391]\tTime 0.141 (0.148)\tData 0.033 (0.039)\tLoss 0.4136 (0.4752)\tPrec@1 90.625 (87.345)\n",
      "[180/391]\tTime 0.137 (0.148)\tData 0.032 (0.039)\tLoss 0.4201 (0.4747)\tPrec@1 87.500 (87.345)\n",
      "[190/391]\tTime 0.143 (0.148)\tData 0.034 (0.039)\tLoss 0.4429 (0.4752)\tPrec@1 89.062 (87.328)\n",
      "[200/391]\tTime 0.150 (0.149)\tData 0.037 (0.039)\tLoss 0.4765 (0.4764)\tPrec@1 85.156 (87.240)\n",
      "[210/391]\tTime 0.164 (0.149)\tData 0.038 (0.039)\tLoss 0.4785 (0.4789)\tPrec@1 89.062 (87.148)\n",
      "[220/391]\tTime 0.126 (0.149)\tData 0.036 (0.039)\tLoss 0.5388 (0.4798)\tPrec@1 85.938 (87.146)\n",
      "[230/391]\tTime 0.158 (0.150)\tData 0.034 (0.040)\tLoss 0.4395 (0.4804)\tPrec@1 84.375 (87.070)\n",
      "[240/391]\tTime 0.134 (0.150)\tData 0.025 (0.039)\tLoss 0.5257 (0.4784)\tPrec@1 83.594 (87.105)\n",
      "[250/391]\tTime 0.121 (0.149)\tData 0.035 (0.039)\tLoss 0.3482 (0.4793)\tPrec@1 92.188 (87.102)\n",
      "[260/391]\tTime 0.128 (0.149)\tData 0.038 (0.039)\tLoss 0.5982 (0.4806)\tPrec@1 82.031 (87.075)\n",
      "[270/391]\tTime 0.159 (0.149)\tData 0.045 (0.039)\tLoss 0.3986 (0.4810)\tPrec@1 89.062 (87.047)\n",
      "[280/391]\tTime 0.132 (0.148)\tData 0.036 (0.039)\tLoss 0.4992 (0.4816)\tPrec@1 85.156 (87.011)\n",
      "[290/391]\tTime 0.110 (0.149)\tData 0.039 (0.039)\tLoss 0.6154 (0.4814)\tPrec@1 84.375 (87.027)\n",
      "[300/391]\tTime 0.171 (0.148)\tData 0.033 (0.039)\tLoss 0.5111 (0.4817)\tPrec@1 85.938 (87.012)\n",
      "[310/391]\tTime 0.130 (0.148)\tData 0.033 (0.039)\tLoss 0.5354 (0.4817)\tPrec@1 83.594 (87.053)\n",
      "[320/391]\tTime 0.150 (0.148)\tData 0.038 (0.039)\tLoss 0.5442 (0.4819)\tPrec@1 86.719 (87.096)\n",
      "[330/391]\tTime 0.161 (0.149)\tData 0.041 (0.039)\tLoss 0.3908 (0.4815)\tPrec@1 89.844 (87.106)\n",
      "[340/391]\tTime 0.151 (0.149)\tData 0.035 (0.039)\tLoss 0.4476 (0.4822)\tPrec@1 90.625 (87.111)\n",
      "[350/391]\tTime 0.160 (0.149)\tData 0.036 (0.039)\tLoss 0.4855 (0.4823)\tPrec@1 85.938 (87.108)\n",
      "[360/391]\tTime 0.207 (0.149)\tData 0.064 (0.039)\tLoss 0.5041 (0.4829)\tPrec@1 86.719 (87.056)\n",
      "[370/391]\tTime 0.130 (0.149)\tData 0.037 (0.039)\tLoss 0.6251 (0.4847)\tPrec@1 81.250 (87.014)\n",
      "[380/391]\tTime 0.182 (0.149)\tData 0.064 (0.039)\tLoss 0.4982 (0.4844)\tPrec@1 88.281 (87.030)\n",
      "[390/391]\tTime 0.101 (0.149)\tData 0.033 (0.039)\tLoss 0.4618 (0.4843)\tPrec@1 86.250 (87.032)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.045 (0.045)\tLoss 0.8579 (0.8579)\tPrec@1 79.688 (79.688)\n",
      "Test: [10/79]\tTime 0.028 (0.029)\tLoss 0.6637 (0.8487)\tPrec@1 83.594 (79.119)\n",
      "Test: [20/79]\tTime 0.046 (0.031)\tLoss 1.0658 (0.8942)\tPrec@1 77.344 (78.943)\n",
      "Test: [30/79]\tTime 0.039 (0.035)\tLoss 0.8120 (0.8931)\tPrec@1 81.250 (79.007)\n",
      "Test: [40/79]\tTime 0.037 (0.038)\tLoss 1.0504 (0.9010)\tPrec@1 78.125 (79.078)\n",
      "Test: [50/79]\tTime 0.037 (0.038)\tLoss 0.9799 (0.8805)\tPrec@1 78.906 (79.167)\n",
      "Test: [60/79]\tTime 0.065 (0.038)\tLoss 0.9607 (0.8810)\tPrec@1 80.469 (79.098)\n",
      "Test: [70/79]\tTime 0.037 (0.038)\tLoss 0.7793 (0.8779)\tPrec@1 81.250 (79.159)\n",
      " * Prec@1 79.320\n",
      "\n",
      "===> epoch: 31/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.167 (0.167)\tData 0.038 (0.038)\tLoss 0.5364 (0.5364)\tPrec@1 82.812 (82.812)\n",
      "[10/391]\tTime 0.115 (0.142)\tData 0.026 (0.034)\tLoss 0.5212 (0.4474)\tPrec@1 82.031 (87.997)\n",
      "[20/391]\tTime 0.127 (0.136)\tData 0.037 (0.034)\tLoss 0.3700 (0.4637)\tPrec@1 94.531 (87.537)\n",
      "[30/391]\tTime 0.106 (0.135)\tData 0.034 (0.035)\tLoss 0.4135 (0.4791)\tPrec@1 89.844 (87.198)\n",
      "[40/391]\tTime 0.170 (0.140)\tData 0.029 (0.036)\tLoss 0.4134 (0.4861)\tPrec@1 89.062 (86.738)\n",
      "[50/391]\tTime 0.152 (0.143)\tData 0.033 (0.037)\tLoss 0.3575 (0.4799)\tPrec@1 89.844 (87.132)\n",
      "[60/391]\tTime 0.153 (0.145)\tData 0.036 (0.037)\tLoss 0.4855 (0.4805)\tPrec@1 87.500 (87.269)\n",
      "[70/391]\tTime 0.148 (0.147)\tData 0.031 (0.038)\tLoss 0.4664 (0.4864)\tPrec@1 85.156 (86.961)\n",
      "[80/391]\tTime 0.194 (0.148)\tData 0.057 (0.038)\tLoss 0.3901 (0.4819)\tPrec@1 88.281 (87.047)\n",
      "[90/391]\tTime 0.119 (0.149)\tData 0.032 (0.038)\tLoss 0.4013 (0.4808)\tPrec@1 89.062 (87.096)\n",
      "[100/391]\tTime 0.143 (0.149)\tData 0.026 (0.038)\tLoss 0.4302 (0.4842)\tPrec@1 88.281 (87.005)\n",
      "[110/391]\tTime 0.186 (0.150)\tData 0.050 (0.038)\tLoss 0.5566 (0.4830)\tPrec@1 78.906 (87.021)\n",
      "[120/391]\tTime 0.132 (0.150)\tData 0.026 (0.038)\tLoss 0.4301 (0.4821)\tPrec@1 89.062 (87.061)\n",
      "[130/391]\tTime 0.179 (0.149)\tData 0.048 (0.038)\tLoss 0.4182 (0.4807)\tPrec@1 89.062 (87.130)\n",
      "[140/391]\tTime 0.134 (0.150)\tData 0.034 (0.038)\tLoss 0.5216 (0.4819)\tPrec@1 86.719 (87.029)\n",
      "[150/391]\tTime 0.126 (0.150)\tData 0.032 (0.038)\tLoss 0.4836 (0.4837)\tPrec@1 91.406 (86.977)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[160/391]\tTime 0.139 (0.150)\tData 0.035 (0.038)\tLoss 0.4604 (0.4829)\tPrec@1 89.844 (86.981)\n",
      "[170/391]\tTime 0.131 (0.150)\tData 0.037 (0.038)\tLoss 0.4087 (0.4843)\tPrec@1 89.062 (86.915)\n",
      "[180/391]\tTime 0.203 (0.150)\tData 0.065 (0.039)\tLoss 0.5858 (0.4842)\tPrec@1 84.375 (86.870)\n",
      "[190/391]\tTime 0.136 (0.149)\tData 0.034 (0.038)\tLoss 0.4213 (0.4838)\tPrec@1 89.062 (86.882)\n",
      "[200/391]\tTime 0.145 (0.148)\tData 0.033 (0.038)\tLoss 0.5131 (0.4871)\tPrec@1 87.500 (86.715)\n",
      "[210/391]\tTime 0.181 (0.150)\tData 0.048 (0.039)\tLoss 0.5074 (0.4898)\tPrec@1 86.719 (86.619)\n",
      "[220/391]\tTime 0.167 (0.150)\tData 0.037 (0.039)\tLoss 0.5367 (0.4922)\tPrec@1 85.156 (86.538)\n",
      "[230/391]\tTime 0.183 (0.150)\tData 0.039 (0.039)\tLoss 0.6738 (0.4952)\tPrec@1 86.719 (86.499)\n",
      "[240/391]\tTime 0.138 (0.150)\tData 0.028 (0.039)\tLoss 0.5282 (0.4948)\tPrec@1 82.031 (86.524)\n",
      "[250/391]\tTime 0.166 (0.151)\tData 0.064 (0.039)\tLoss 0.4885 (0.4944)\tPrec@1 86.719 (86.579)\n",
      "[260/391]\tTime 0.183 (0.151)\tData 0.065 (0.039)\tLoss 0.4236 (0.4952)\tPrec@1 89.062 (86.587)\n",
      "[270/391]\tTime 0.162 (0.152)\tData 0.043 (0.039)\tLoss 0.4383 (0.4951)\tPrec@1 88.281 (86.603)\n",
      "[280/391]\tTime 0.163 (0.152)\tData 0.038 (0.039)\tLoss 0.3671 (0.4943)\tPrec@1 90.625 (86.616)\n",
      "[290/391]\tTime 0.133 (0.152)\tData 0.031 (0.039)\tLoss 0.4817 (0.4952)\tPrec@1 89.844 (86.593)\n",
      "[300/391]\tTime 0.147 (0.151)\tData 0.059 (0.039)\tLoss 0.4653 (0.4944)\tPrec@1 88.281 (86.607)\n",
      "[310/391]\tTime 0.134 (0.150)\tData 0.034 (0.039)\tLoss 0.4447 (0.4933)\tPrec@1 86.719 (86.618)\n",
      "[320/391]\tTime 0.126 (0.150)\tData 0.037 (0.039)\tLoss 0.4596 (0.4929)\tPrec@1 85.938 (86.631)\n",
      "[330/391]\tTime 0.180 (0.150)\tData 0.042 (0.039)\tLoss 0.5988 (0.4929)\tPrec@1 79.688 (86.591)\n",
      "[340/391]\tTime 0.123 (0.149)\tData 0.039 (0.039)\tLoss 0.4804 (0.4934)\tPrec@1 85.938 (86.568)\n",
      "[350/391]\tTime 0.191 (0.149)\tData 0.072 (0.039)\tLoss 0.4922 (0.4947)\tPrec@1 86.719 (86.525)\n",
      "[360/391]\tTime 0.163 (0.149)\tData 0.035 (0.039)\tLoss 0.5673 (0.4956)\tPrec@1 82.031 (86.472)\n",
      "[370/391]\tTime 0.155 (0.149)\tData 0.037 (0.039)\tLoss 0.6268 (0.4961)\tPrec@1 82.031 (86.451)\n",
      "[380/391]\tTime 0.164 (0.150)\tData 0.037 (0.039)\tLoss 0.4067 (0.4956)\tPrec@1 89.062 (86.493)\n",
      "[390/391]\tTime 0.148 (0.150)\tData 0.034 (0.039)\tLoss 0.4203 (0.4952)\tPrec@1 90.000 (86.498)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.052 (0.052)\tLoss 1.5797 (1.5797)\tPrec@1 71.094 (71.094)\n",
      "Test: [10/79]\tTime 0.051 (0.039)\tLoss 1.4200 (1.5845)\tPrec@1 75.781 (70.170)\n",
      "Test: [20/79]\tTime 0.033 (0.041)\tLoss 1.5896 (1.6163)\tPrec@1 71.094 (70.126)\n",
      "Test: [30/79]\tTime 0.038 (0.040)\tLoss 1.7490 (1.6235)\tPrec@1 68.750 (69.909)\n",
      "Test: [40/79]\tTime 0.039 (0.040)\tLoss 1.7055 (1.5764)\tPrec@1 64.062 (70.293)\n",
      "Test: [50/79]\tTime 0.043 (0.039)\tLoss 1.7647 (1.5702)\tPrec@1 67.188 (70.450)\n",
      "Test: [60/79]\tTime 0.033 (0.041)\tLoss 1.4520 (1.5719)\tPrec@1 71.094 (70.479)\n",
      "Test: [70/79]\tTime 0.039 (0.041)\tLoss 1.5836 (1.5879)\tPrec@1 67.969 (70.158)\n",
      " * Prec@1 70.280\n",
      "\n",
      "===> epoch: 32/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.142 (0.142)\tData 0.037 (0.037)\tLoss 0.5218 (0.5218)\tPrec@1 85.156 (85.156)\n",
      "[10/391]\tTime 0.194 (0.143)\tData 0.073 (0.038)\tLoss 0.5552 (0.4880)\tPrec@1 81.250 (86.435)\n",
      "[20/391]\tTime 0.140 (0.153)\tData 0.035 (0.042)\tLoss 0.6056 (0.4916)\tPrec@1 82.031 (86.310)\n",
      "[30/391]\tTime 0.175 (0.153)\tData 0.057 (0.041)\tLoss 0.3316 (0.4822)\tPrec@1 91.406 (86.593)\n",
      "[40/391]\tTime 0.150 (0.157)\tData 0.038 (0.041)\tLoss 0.5302 (0.4825)\tPrec@1 86.719 (86.623)\n",
      "[50/391]\tTime 0.130 (0.157)\tData 0.032 (0.042)\tLoss 0.4590 (0.4830)\tPrec@1 87.500 (86.673)\n",
      "[60/391]\tTime 0.131 (0.153)\tData 0.029 (0.041)\tLoss 0.4052 (0.4777)\tPrec@1 88.281 (86.860)\n",
      "[70/391]\tTime 0.125 (0.151)\tData 0.037 (0.040)\tLoss 0.4516 (0.4757)\tPrec@1 87.500 (87.093)\n",
      "[80/391]\tTime 0.157 (0.152)\tData 0.035 (0.040)\tLoss 0.4879 (0.4696)\tPrec@1 86.719 (87.317)\n",
      "[90/391]\tTime 0.153 (0.152)\tData 0.026 (0.040)\tLoss 0.4993 (0.4729)\tPrec@1 85.938 (87.303)\n",
      "[100/391]\tTime 0.153 (0.152)\tData 0.035 (0.041)\tLoss 0.4760 (0.4737)\tPrec@1 89.844 (87.291)\n",
      "[110/391]\tTime 0.152 (0.152)\tData 0.033 (0.040)\tLoss 0.4748 (0.4728)\tPrec@1 85.938 (87.380)\n",
      "[120/391]\tTime 0.173 (0.153)\tData 0.028 (0.040)\tLoss 0.4478 (0.4721)\tPrec@1 85.938 (87.351)\n",
      "[130/391]\tTime 0.123 (0.153)\tData 0.032 (0.040)\tLoss 0.4684 (0.4724)\tPrec@1 88.281 (87.375)\n",
      "[140/391]\tTime 0.173 (0.153)\tData 0.035 (0.040)\tLoss 0.4871 (0.4726)\tPrec@1 87.500 (87.367)\n",
      "[150/391]\tTime 0.152 (0.153)\tData 0.060 (0.040)\tLoss 0.4587 (0.4711)\tPrec@1 89.844 (87.422)\n",
      "[160/391]\tTime 0.136 (0.154)\tData 0.033 (0.040)\tLoss 0.5403 (0.4716)\tPrec@1 87.500 (87.451)\n",
      "[170/391]\tTime 0.166 (0.154)\tData 0.029 (0.039)\tLoss 0.4885 (0.4699)\tPrec@1 83.594 (87.445)\n",
      "[180/391]\tTime 0.156 (0.154)\tData 0.036 (0.039)\tLoss 0.5106 (0.4719)\tPrec@1 84.375 (87.349)\n",
      "[190/391]\tTime 0.144 (0.154)\tData 0.044 (0.040)\tLoss 0.4578 (0.4720)\tPrec@1 88.281 (87.361)\n",
      "[200/391]\tTime 0.113 (0.152)\tData 0.024 (0.039)\tLoss 0.4461 (0.4726)\tPrec@1 88.281 (87.348)\n",
      "[210/391]\tTime 0.163 (0.150)\tData 0.047 (0.039)\tLoss 0.4321 (0.4722)\tPrec@1 87.500 (87.315)\n",
      "[220/391]\tTime 0.153 (0.151)\tData 0.039 (0.039)\tLoss 0.5492 (0.4732)\tPrec@1 85.156 (87.263)\n",
      "[230/391]\tTime 0.169 (0.151)\tData 0.057 (0.040)\tLoss 0.4519 (0.4720)\tPrec@1 85.938 (87.328)\n",
      "[240/391]\tTime 0.138 (0.152)\tData 0.033 (0.040)\tLoss 0.4969 (0.4727)\tPrec@1 86.719 (87.341)\n",
      "[250/391]\tTime 0.146 (0.152)\tData 0.037 (0.040)\tLoss 0.5183 (0.4729)\tPrec@1 86.719 (87.329)\n",
      "[260/391]\tTime 0.160 (0.152)\tData 0.032 (0.040)\tLoss 0.4375 (0.4743)\tPrec@1 87.500 (87.281)\n",
      "[270/391]\tTime 0.176 (0.152)\tData 0.054 (0.040)\tLoss 0.4779 (0.4749)\tPrec@1 87.500 (87.272)\n",
      "[280/391]\tTime 0.136 (0.152)\tData 0.031 (0.040)\tLoss 0.5249 (0.4761)\tPrec@1 83.594 (87.253)\n",
      "[290/391]\tTime 0.168 (0.152)\tData 0.064 (0.040)\tLoss 0.4536 (0.4777)\tPrec@1 86.719 (87.189)\n",
      "[300/391]\tTime 0.144 (0.152)\tData 0.038 (0.040)\tLoss 0.5101 (0.4795)\tPrec@1 85.938 (87.121)\n",
      "[310/391]\tTime 0.144 (0.152)\tData 0.035 (0.040)\tLoss 0.4591 (0.4806)\tPrec@1 85.938 (87.086)\n",
      "[320/391]\tTime 0.148 (0.152)\tData 0.022 (0.040)\tLoss 0.5320 (0.4823)\tPrec@1 86.719 (87.025)\n",
      "[330/391]\tTime 0.140 (0.152)\tData 0.039 (0.040)\tLoss 0.4576 (0.4841)\tPrec@1 90.625 (86.974)\n",
      "[340/391]\tTime 0.128 (0.152)\tData 0.035 (0.040)\tLoss 0.5386 (0.4836)\tPrec@1 87.500 (87.007)\n",
      "[350/391]\tTime 0.164 (0.153)\tData 0.056 (0.040)\tLoss 0.5077 (0.4838)\tPrec@1 86.719 (86.988)\n",
      "[360/391]\tTime 0.176 (0.152)\tData 0.063 (0.040)\tLoss 0.5724 (0.4861)\tPrec@1 86.719 (86.914)\n",
      "[370/391]\tTime 0.103 (0.152)\tData 0.029 (0.040)\tLoss 0.5826 (0.4865)\tPrec@1 85.938 (86.925)\n",
      "[380/391]\tTime 0.152 (0.152)\tData 0.042 (0.040)\tLoss 0.5052 (0.4858)\tPrec@1 88.281 (86.940)\n",
      "[390/391]\tTime 0.094 (0.152)\tData 0.024 (0.040)\tLoss 0.4764 (0.4861)\tPrec@1 82.500 (86.930)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.032 (0.032)\tLoss 1.3589 (1.3589)\tPrec@1 67.969 (67.969)\n",
      "Test: [10/79]\tTime 0.034 (0.043)\tLoss 1.4247 (1.5416)\tPrec@1 67.969 (69.176)\n",
      "Test: [20/79]\tTime 0.033 (0.040)\tLoss 1.7815 (1.6068)\tPrec@1 66.406 (68.527)\n",
      "Test: [30/79]\tTime 0.028 (0.038)\tLoss 1.2555 (1.5641)\tPrec@1 71.094 (69.103)\n",
      "Test: [40/79]\tTime 0.032 (0.038)\tLoss 1.3547 (1.5550)\tPrec@1 71.094 (69.284)\n",
      "Test: [50/79]\tTime 0.033 (0.037)\tLoss 1.5904 (1.5290)\tPrec@1 71.094 (69.669)\n",
      "Test: [60/79]\tTime 0.044 (0.037)\tLoss 1.4976 (1.5180)\tPrec@1 71.875 (69.839)\n",
      "Test: [70/79]\tTime 0.032 (0.037)\tLoss 1.4754 (1.5116)\tPrec@1 71.875 (69.828)\n",
      " * Prec@1 69.920\n",
      "\n",
      "===> epoch: 33/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.161 (0.161)\tData 0.041 (0.041)\tLoss 0.3909 (0.3909)\tPrec@1 86.719 (86.719)\n",
      "[10/391]\tTime 0.191 (0.148)\tData 0.051 (0.037)\tLoss 0.4165 (0.4786)\tPrec@1 88.281 (87.358)\n",
      "[20/391]\tTime 0.143 (0.157)\tData 0.030 (0.042)\tLoss 0.4493 (0.4741)\tPrec@1 89.062 (87.314)\n",
      "[30/391]\tTime 0.155 (0.157)\tData 0.042 (0.041)\tLoss 0.3887 (0.4586)\tPrec@1 89.844 (87.349)\n",
      "[40/391]\tTime 0.150 (0.157)\tData 0.034 (0.041)\tLoss 0.4085 (0.4561)\tPrec@1 89.844 (87.367)\n",
      "[50/391]\tTime 0.133 (0.155)\tData 0.026 (0.040)\tLoss 0.5995 (0.4558)\tPrec@1 83.594 (87.454)\n",
      "[60/391]\tTime 0.167 (0.154)\tData 0.060 (0.039)\tLoss 0.4618 (0.4591)\tPrec@1 87.500 (87.538)\n",
      "[70/391]\tTime 0.195 (0.153)\tData 0.055 (0.039)\tLoss 0.4858 (0.4588)\tPrec@1 90.625 (87.720)\n",
      "[80/391]\tTime 0.153 (0.154)\tData 0.023 (0.039)\tLoss 0.5137 (0.4598)\tPrec@1 87.500 (87.828)\n",
      "[90/391]\tTime 0.121 (0.150)\tData 0.033 (0.039)\tLoss 0.4665 (0.4584)\tPrec@1 86.719 (87.929)\n",
      "[100/391]\tTime 0.146 (0.148)\tData 0.057 (0.039)\tLoss 0.3862 (0.4584)\tPrec@1 92.188 (87.871)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[110/391]\tTime 0.165 (0.147)\tData 0.064 (0.040)\tLoss 0.5497 (0.4611)\tPrec@1 90.625 (87.810)\n",
      "[120/391]\tTime 0.121 (0.149)\tData 0.039 (0.040)\tLoss 0.4177 (0.4607)\tPrec@1 89.062 (87.797)\n",
      "[130/391]\tTime 0.136 (0.149)\tData 0.041 (0.040)\tLoss 0.4864 (0.4618)\tPrec@1 86.719 (87.750)\n",
      "[140/391]\tTime 0.155 (0.150)\tData 0.027 (0.040)\tLoss 0.4967 (0.4654)\tPrec@1 84.375 (87.616)\n",
      "[150/391]\tTime 0.224 (0.150)\tData 0.068 (0.040)\tLoss 0.4239 (0.4693)\tPrec@1 89.062 (87.448)\n",
      "[160/391]\tTime 0.155 (0.151)\tData 0.031 (0.040)\tLoss 0.3575 (0.4693)\tPrec@1 90.625 (87.398)\n",
      "[170/391]\tTime 0.158 (0.151)\tData 0.044 (0.039)\tLoss 0.7208 (0.4713)\tPrec@1 80.469 (87.340)\n",
      "[180/391]\tTime 0.135 (0.152)\tData 0.039 (0.039)\tLoss 0.3764 (0.4740)\tPrec@1 93.750 (87.310)\n",
      "[190/391]\tTime 0.169 (0.152)\tData 0.035 (0.040)\tLoss 0.4300 (0.4756)\tPrec@1 88.281 (87.193)\n",
      "[200/391]\tTime 0.160 (0.152)\tData 0.036 (0.039)\tLoss 0.5936 (0.4764)\tPrec@1 84.375 (87.150)\n",
      "[210/391]\tTime 0.173 (0.153)\tData 0.057 (0.039)\tLoss 0.4504 (0.4775)\tPrec@1 85.156 (87.085)\n",
      "[220/391]\tTime 0.143 (0.152)\tData 0.027 (0.039)\tLoss 0.4465 (0.4765)\tPrec@1 85.156 (87.076)\n",
      "[230/391]\tTime 0.157 (0.152)\tData 0.032 (0.039)\tLoss 0.5319 (0.4763)\tPrec@1 87.500 (87.098)\n",
      "[240/391]\tTime 0.157 (0.152)\tData 0.039 (0.039)\tLoss 0.6190 (0.4771)\tPrec@1 81.250 (87.053)\n",
      "[250/391]\tTime 0.125 (0.152)\tData 0.043 (0.039)\tLoss 0.5228 (0.4767)\tPrec@1 85.938 (87.046)\n",
      "[260/391]\tTime 0.112 (0.151)\tData 0.035 (0.039)\tLoss 0.5106 (0.4776)\tPrec@1 85.156 (87.003)\n",
      "[270/391]\tTime 0.162 (0.151)\tData 0.035 (0.039)\tLoss 0.4840 (0.4775)\tPrec@1 86.719 (86.993)\n",
      "[280/391]\tTime 0.139 (0.151)\tData 0.035 (0.039)\tLoss 0.5802 (0.4773)\tPrec@1 82.812 (87.005)\n",
      "[290/391]\tTime 0.160 (0.151)\tData 0.044 (0.039)\tLoss 0.4391 (0.4769)\tPrec@1 89.844 (87.006)\n",
      "[300/391]\tTime 0.158 (0.151)\tData 0.034 (0.039)\tLoss 0.4399 (0.4760)\tPrec@1 91.406 (87.033)\n",
      "[310/391]\tTime 0.110 (0.151)\tData 0.030 (0.039)\tLoss 0.3504 (0.4762)\tPrec@1 90.625 (87.060)\n",
      "[320/391]\tTime 0.172 (0.151)\tData 0.047 (0.039)\tLoss 0.4849 (0.4765)\tPrec@1 90.625 (87.059)\n",
      "[330/391]\tTime 0.144 (0.151)\tData 0.032 (0.039)\tLoss 0.4190 (0.4755)\tPrec@1 88.281 (87.101)\n",
      "[340/391]\tTime 0.172 (0.151)\tData 0.037 (0.039)\tLoss 0.4400 (0.4758)\tPrec@1 88.281 (87.101)\n",
      "[350/391]\tTime 0.147 (0.151)\tData 0.034 (0.039)\tLoss 0.3837 (0.4763)\tPrec@1 90.625 (87.097)\n",
      "[360/391]\tTime 0.130 (0.152)\tData 0.034 (0.039)\tLoss 0.4222 (0.4767)\tPrec@1 90.625 (87.100)\n",
      "[370/391]\tTime 0.154 (0.152)\tData 0.027 (0.039)\tLoss 0.4791 (0.4775)\tPrec@1 88.281 (87.073)\n",
      "[380/391]\tTime 0.146 (0.152)\tData 0.036 (0.039)\tLoss 0.4261 (0.4798)\tPrec@1 89.844 (87.016)\n",
      "[390/391]\tTime 0.104 (0.152)\tData 0.020 (0.039)\tLoss 0.3807 (0.4799)\tPrec@1 91.250 (87.014)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.072 (0.072)\tLoss 0.5944 (0.5944)\tPrec@1 82.812 (82.812)\n",
      "Test: [10/79]\tTime 0.034 (0.046)\tLoss 0.5939 (0.8209)\tPrec@1 84.375 (80.043)\n",
      "Test: [20/79]\tTime 0.028 (0.042)\tLoss 0.8132 (0.8217)\tPrec@1 78.906 (80.246)\n",
      "Test: [30/79]\tTime 0.022 (0.037)\tLoss 0.6651 (0.8272)\tPrec@1 82.812 (80.343)\n",
      "Test: [40/79]\tTime 0.029 (0.036)\tLoss 0.8973 (0.8258)\tPrec@1 76.562 (80.297)\n",
      "Test: [50/79]\tTime 0.036 (0.039)\tLoss 0.8988 (0.8161)\tPrec@1 76.562 (80.714)\n",
      "Test: [60/79]\tTime 0.039 (0.039)\tLoss 0.8881 (0.8171)\tPrec@1 78.125 (80.584)\n",
      "Test: [70/79]\tTime 0.040 (0.039)\tLoss 0.6092 (0.8152)\tPrec@1 84.375 (80.491)\n",
      " * Prec@1 80.440\n",
      "\n",
      "===> epoch: 34/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.200 (0.200)\tData 0.069 (0.069)\tLoss 0.5660 (0.5660)\tPrec@1 85.938 (85.938)\n",
      "[10/391]\tTime 0.176 (0.151)\tData 0.059 (0.041)\tLoss 0.4224 (0.4994)\tPrec@1 89.844 (87.713)\n",
      "[20/391]\tTime 0.176 (0.152)\tData 0.054 (0.040)\tLoss 0.5596 (0.4780)\tPrec@1 83.594 (87.500)\n",
      "[30/391]\tTime 0.149 (0.156)\tData 0.050 (0.042)\tLoss 0.3328 (0.4694)\tPrec@1 92.188 (87.525)\n",
      "[40/391]\tTime 0.133 (0.153)\tData 0.032 (0.041)\tLoss 0.5122 (0.4654)\tPrec@1 83.594 (87.462)\n",
      "[50/391]\tTime 0.195 (0.155)\tData 0.068 (0.040)\tLoss 0.5140 (0.4692)\tPrec@1 82.031 (87.347)\n",
      "[60/391]\tTime 0.169 (0.154)\tData 0.033 (0.040)\tLoss 0.5719 (0.4729)\tPrec@1 82.031 (87.167)\n",
      "[70/391]\tTime 0.186 (0.156)\tData 0.044 (0.041)\tLoss 0.5152 (0.4754)\tPrec@1 85.938 (87.214)\n",
      "[80/391]\tTime 0.168 (0.156)\tData 0.036 (0.040)\tLoss 0.4730 (0.4770)\tPrec@1 84.375 (87.153)\n",
      "[90/391]\tTime 0.122 (0.155)\tData 0.034 (0.040)\tLoss 0.3720 (0.4763)\tPrec@1 92.969 (87.148)\n",
      "[100/391]\tTime 0.158 (0.154)\tData 0.035 (0.039)\tLoss 0.3277 (0.4761)\tPrec@1 92.969 (87.252)\n",
      "[110/391]\tTime 0.182 (0.154)\tData 0.054 (0.039)\tLoss 0.4656 (0.4788)\tPrec@1 86.719 (87.148)\n",
      "[120/391]\tTime 0.169 (0.153)\tData 0.053 (0.039)\tLoss 0.5265 (0.4770)\tPrec@1 85.156 (87.177)\n",
      "[130/391]\tTime 0.166 (0.153)\tData 0.044 (0.039)\tLoss 0.4543 (0.4789)\tPrec@1 85.938 (86.987)\n",
      "[140/391]\tTime 0.105 (0.152)\tData 0.031 (0.039)\tLoss 0.5139 (0.4779)\tPrec@1 84.375 (87.079)\n",
      "[150/391]\tTime 0.145 (0.153)\tData 0.042 (0.040)\tLoss 0.3681 (0.4770)\tPrec@1 90.625 (87.117)\n",
      "[160/391]\tTime 0.107 (0.152)\tData 0.023 (0.040)\tLoss 0.4920 (0.4758)\tPrec@1 85.938 (87.165)\n",
      "[170/391]\tTime 0.149 (0.152)\tData 0.035 (0.040)\tLoss 0.5977 (0.4753)\tPrec@1 86.719 (87.226)\n",
      "[180/391]\tTime 0.156 (0.152)\tData 0.039 (0.040)\tLoss 0.3857 (0.4760)\tPrec@1 89.062 (87.202)\n",
      "[190/391]\tTime 0.188 (0.152)\tData 0.050 (0.040)\tLoss 0.4771 (0.4783)\tPrec@1 89.844 (87.169)\n",
      "[200/391]\tTime 0.136 (0.152)\tData 0.033 (0.040)\tLoss 0.6348 (0.4801)\tPrec@1 82.812 (87.084)\n",
      "[210/391]\tTime 0.178 (0.152)\tData 0.046 (0.040)\tLoss 0.5782 (0.4806)\tPrec@1 80.469 (87.071)\n",
      "[220/391]\tTime 0.154 (0.151)\tData 0.052 (0.040)\tLoss 0.4152 (0.4815)\tPrec@1 88.281 (87.040)\n",
      "[230/391]\tTime 0.146 (0.152)\tData 0.039 (0.040)\tLoss 0.4648 (0.4820)\tPrec@1 89.062 (87.027)\n",
      "[240/391]\tTime 0.131 (0.151)\tData 0.034 (0.040)\tLoss 0.5175 (0.4812)\tPrec@1 84.375 (87.027)\n",
      "[250/391]\tTime 0.190 (0.151)\tData 0.061 (0.039)\tLoss 0.5289 (0.4811)\tPrec@1 85.156 (87.027)\n",
      "[260/391]\tTime 0.167 (0.151)\tData 0.036 (0.039)\tLoss 0.5393 (0.4811)\tPrec@1 82.812 (87.024)\n",
      "[270/391]\tTime 0.126 (0.151)\tData 0.033 (0.040)\tLoss 0.4372 (0.4810)\tPrec@1 87.500 (87.021)\n",
      "[280/391]\tTime 0.157 (0.151)\tData 0.038 (0.039)\tLoss 0.5681 (0.4810)\tPrec@1 82.812 (87.011)\n",
      "[290/391]\tTime 0.128 (0.151)\tData 0.039 (0.039)\tLoss 0.5236 (0.4827)\tPrec@1 85.938 (86.968)\n",
      "[300/391]\tTime 0.107 (0.151)\tData 0.031 (0.039)\tLoss 0.6150 (0.4850)\tPrec@1 86.719 (86.913)\n",
      "[310/391]\tTime 0.082 (0.149)\tData 0.025 (0.039)\tLoss 0.3844 (0.4856)\tPrec@1 89.844 (86.877)\n",
      "[320/391]\tTime 0.165 (0.149)\tData 0.033 (0.039)\tLoss 0.6768 (0.4859)\tPrec@1 81.250 (86.877)\n",
      "[330/391]\tTime 0.224 (0.150)\tData 0.080 (0.039)\tLoss 0.3925 (0.4855)\tPrec@1 87.500 (86.893)\n",
      "[340/391]\tTime 0.124 (0.150)\tData 0.035 (0.039)\tLoss 0.6398 (0.4859)\tPrec@1 86.719 (86.911)\n",
      "[350/391]\tTime 0.156 (0.150)\tData 0.034 (0.039)\tLoss 0.5002 (0.4854)\tPrec@1 83.594 (86.906)\n",
      "[360/391]\tTime 0.141 (0.150)\tData 0.040 (0.039)\tLoss 0.4266 (0.4859)\tPrec@1 85.156 (86.851)\n",
      "[370/391]\tTime 0.120 (0.150)\tData 0.031 (0.039)\tLoss 0.4180 (0.4871)\tPrec@1 90.625 (86.814)\n",
      "[380/391]\tTime 0.184 (0.150)\tData 0.049 (0.039)\tLoss 0.5303 (0.4870)\tPrec@1 83.594 (86.815)\n",
      "[390/391]\tTime 0.096 (0.150)\tData 0.019 (0.039)\tLoss 0.5402 (0.4864)\tPrec@1 81.250 (86.828)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.032 (0.032)\tLoss 0.8757 (0.8757)\tPrec@1 82.031 (82.031)\n",
      "Test: [10/79]\tTime 0.032 (0.032)\tLoss 0.7726 (0.8952)\tPrec@1 78.906 (80.469)\n",
      "Test: [20/79]\tTime 0.029 (0.033)\tLoss 0.9601 (0.9534)\tPrec@1 79.688 (78.832)\n",
      "Test: [30/79]\tTime 0.036 (0.033)\tLoss 0.4388 (0.9331)\tPrec@1 87.500 (79.083)\n",
      "Test: [40/79]\tTime 0.028 (0.032)\tLoss 1.2937 (0.9073)\tPrec@1 74.219 (79.383)\n",
      "Test: [50/79]\tTime 0.040 (0.033)\tLoss 0.9351 (0.8873)\tPrec@1 77.344 (79.795)\n",
      "Test: [60/79]\tTime 0.037 (0.035)\tLoss 0.6489 (0.8903)\tPrec@1 82.812 (79.713)\n",
      "Test: [70/79]\tTime 0.036 (0.035)\tLoss 0.8640 (0.8964)\tPrec@1 80.469 (79.467)\n",
      " * Prec@1 79.510\n",
      "\n",
      "===> epoch: 35/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.160 (0.160)\tData 0.065 (0.065)\tLoss 0.4500 (0.4500)\tPrec@1 85.938 (85.938)\n",
      "[10/391]\tTime 0.138 (0.166)\tData 0.026 (0.044)\tLoss 0.5200 (0.4717)\tPrec@1 87.500 (87.074)\n",
      "[20/391]\tTime 0.167 (0.160)\tData 0.034 (0.039)\tLoss 0.4202 (0.4668)\tPrec@1 89.062 (87.314)\n",
      "[30/391]\tTime 0.141 (0.158)\tData 0.034 (0.039)\tLoss 0.3380 (0.4723)\tPrec@1 92.969 (86.920)\n",
      "[40/391]\tTime 0.161 (0.152)\tData 0.037 (0.039)\tLoss 0.4151 (0.4836)\tPrec@1 88.281 (86.528)\n",
      "[50/391]\tTime 0.154 (0.149)\tData 0.044 (0.038)\tLoss 0.4609 (0.4849)\tPrec@1 85.156 (86.550)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[60/391]\tTime 0.183 (0.152)\tData 0.039 (0.038)\tLoss 0.5627 (0.4843)\tPrec@1 85.156 (86.629)\n",
      "[70/391]\tTime 0.186 (0.152)\tData 0.063 (0.039)\tLoss 0.4377 (0.4853)\tPrec@1 92.188 (86.565)\n",
      "[80/391]\tTime 0.180 (0.152)\tData 0.055 (0.039)\tLoss 0.4618 (0.4861)\tPrec@1 89.844 (86.526)\n",
      "[90/391]\tTime 0.164 (0.154)\tData 0.039 (0.039)\tLoss 0.4893 (0.4845)\tPrec@1 88.281 (86.684)\n",
      "[100/391]\tTime 0.148 (0.153)\tData 0.029 (0.039)\tLoss 0.4757 (0.4807)\tPrec@1 92.188 (86.920)\n",
      "[110/391]\tTime 0.165 (0.153)\tData 0.036 (0.038)\tLoss 0.3960 (0.4807)\tPrec@1 92.188 (86.909)\n",
      "[120/391]\tTime 0.171 (0.153)\tData 0.049 (0.039)\tLoss 0.7037 (0.4835)\tPrec@1 84.375 (86.900)\n",
      "[130/391]\tTime 0.174 (0.153)\tData 0.044 (0.039)\tLoss 0.5689 (0.4846)\tPrec@1 85.156 (86.838)\n",
      "[140/391]\tTime 0.136 (0.153)\tData 0.034 (0.038)\tLoss 0.4238 (0.4850)\tPrec@1 88.281 (86.708)\n",
      "[150/391]\tTime 0.148 (0.153)\tData 0.039 (0.038)\tLoss 0.5453 (0.4857)\tPrec@1 85.938 (86.698)\n",
      "[160/391]\tTime 0.136 (0.153)\tData 0.032 (0.038)\tLoss 0.4113 (0.4842)\tPrec@1 89.062 (86.787)\n",
      "[170/391]\tTime 0.126 (0.152)\tData 0.036 (0.038)\tLoss 0.4383 (0.4842)\tPrec@1 88.281 (86.783)\n",
      "[180/391]\tTime 0.196 (0.153)\tData 0.065 (0.038)\tLoss 0.5644 (0.4859)\tPrec@1 83.594 (86.740)\n",
      "[190/391]\tTime 0.155 (0.153)\tData 0.030 (0.038)\tLoss 0.5766 (0.4851)\tPrec@1 83.594 (86.821)\n",
      "[200/391]\tTime 0.118 (0.152)\tData 0.032 (0.038)\tLoss 0.5342 (0.4856)\tPrec@1 87.500 (86.789)\n",
      "[210/391]\tTime 0.109 (0.150)\tData 0.034 (0.038)\tLoss 0.5074 (0.4862)\tPrec@1 85.938 (86.782)\n",
      "[220/391]\tTime 0.162 (0.149)\tData 0.073 (0.038)\tLoss 0.6248 (0.4862)\tPrec@1 81.250 (86.786)\n",
      "[230/391]\tTime 0.144 (0.149)\tData 0.041 (0.038)\tLoss 0.5680 (0.4861)\tPrec@1 83.594 (86.769)\n",
      "[240/391]\tTime 0.159 (0.149)\tData 0.042 (0.038)\tLoss 0.4606 (0.4886)\tPrec@1 86.719 (86.683)\n",
      "[250/391]\tTime 0.140 (0.149)\tData 0.038 (0.038)\tLoss 0.6128 (0.4894)\tPrec@1 83.594 (86.691)\n",
      "[260/391]\tTime 0.150 (0.150)\tData 0.035 (0.038)\tLoss 0.4767 (0.4911)\tPrec@1 87.500 (86.656)\n",
      "[270/391]\tTime 0.179 (0.149)\tData 0.036 (0.038)\tLoss 0.4840 (0.4909)\tPrec@1 85.156 (86.650)\n",
      "[280/391]\tTime 0.152 (0.149)\tData 0.029 (0.038)\tLoss 0.4241 (0.4891)\tPrec@1 91.406 (86.719)\n",
      "[290/391]\tTime 0.161 (0.149)\tData 0.034 (0.038)\tLoss 0.5187 (0.4885)\tPrec@1 87.500 (86.732)\n",
      "[300/391]\tTime 0.153 (0.150)\tData 0.032 (0.038)\tLoss 0.4184 (0.4891)\tPrec@1 87.500 (86.682)\n",
      "[310/391]\tTime 0.179 (0.150)\tData 0.068 (0.038)\tLoss 0.5114 (0.4904)\tPrec@1 85.156 (86.679)\n",
      "[320/391]\tTime 0.129 (0.150)\tData 0.040 (0.039)\tLoss 0.3753 (0.4906)\tPrec@1 92.969 (86.694)\n",
      "[330/391]\tTime 0.144 (0.150)\tData 0.029 (0.039)\tLoss 0.4808 (0.4908)\tPrec@1 85.938 (86.669)\n",
      "[340/391]\tTime 0.177 (0.150)\tData 0.047 (0.039)\tLoss 0.5387 (0.4913)\tPrec@1 83.594 (86.639)\n",
      "[350/391]\tTime 0.139 (0.150)\tData 0.035 (0.039)\tLoss 0.4145 (0.4917)\tPrec@1 92.188 (86.630)\n",
      "[360/391]\tTime 0.131 (0.150)\tData 0.036 (0.039)\tLoss 0.5824 (0.4934)\tPrec@1 85.938 (86.580)\n",
      "[370/391]\tTime 0.159 (0.149)\tData 0.032 (0.039)\tLoss 0.4883 (0.4940)\tPrec@1 85.938 (86.565)\n",
      "[380/391]\tTime 0.133 (0.150)\tData 0.035 (0.039)\tLoss 0.4141 (0.4936)\tPrec@1 89.062 (86.581)\n",
      "[390/391]\tTime 0.103 (0.150)\tData 0.024 (0.039)\tLoss 0.6208 (0.4934)\tPrec@1 85.000 (86.598)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.057 (0.057)\tLoss 0.7647 (0.7647)\tPrec@1 81.250 (81.250)\n",
      "Test: [10/79]\tTime 0.028 (0.043)\tLoss 0.9641 (0.8761)\tPrec@1 75.781 (79.759)\n",
      "Test: [20/79]\tTime 0.040 (0.038)\tLoss 0.7061 (0.9150)\tPrec@1 82.031 (79.092)\n",
      "Test: [30/79]\tTime 0.036 (0.037)\tLoss 0.8009 (0.8965)\tPrec@1 82.031 (79.461)\n",
      "Test: [40/79]\tTime 0.039 (0.037)\tLoss 1.1468 (0.8856)\tPrec@1 75.000 (79.287)\n",
      "Test: [50/79]\tTime 0.036 (0.037)\tLoss 1.0041 (0.8869)\tPrec@1 74.219 (79.350)\n",
      "Test: [60/79]\tTime 0.044 (0.037)\tLoss 0.7458 (0.8783)\tPrec@1 81.250 (79.636)\n",
      "Test: [70/79]\tTime 0.042 (0.038)\tLoss 0.6642 (0.8792)\tPrec@1 85.156 (79.621)\n",
      " * Prec@1 79.670\n",
      "\n",
      "===> epoch: 36/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.163 (0.163)\tData 0.040 (0.040)\tLoss 0.5564 (0.5564)\tPrec@1 82.031 (82.031)\n",
      "[10/391]\tTime 0.158 (0.151)\tData 0.033 (0.039)\tLoss 0.5434 (0.4771)\tPrec@1 85.938 (87.784)\n",
      "[20/391]\tTime 0.161 (0.151)\tData 0.029 (0.035)\tLoss 0.6338 (0.4985)\tPrec@1 84.375 (87.054)\n",
      "[30/391]\tTime 0.173 (0.151)\tData 0.051 (0.036)\tLoss 0.3730 (0.4801)\tPrec@1 91.406 (87.702)\n",
      "[40/391]\tTime 0.185 (0.154)\tData 0.031 (0.036)\tLoss 0.4522 (0.4758)\tPrec@1 90.625 (87.748)\n",
      "[50/391]\tTime 0.146 (0.154)\tData 0.035 (0.036)\tLoss 0.5758 (0.4802)\tPrec@1 85.938 (87.531)\n",
      "[60/391]\tTime 0.136 (0.155)\tData 0.044 (0.037)\tLoss 0.5134 (0.4795)\tPrec@1 85.938 (87.474)\n",
      "[70/391]\tTime 0.145 (0.154)\tData 0.035 (0.038)\tLoss 0.5554 (0.4840)\tPrec@1 85.156 (87.357)\n",
      "[80/391]\tTime 0.123 (0.154)\tData 0.034 (0.037)\tLoss 0.4870 (0.4876)\tPrec@1 85.938 (87.172)\n",
      "[90/391]\tTime 0.149 (0.152)\tData 0.037 (0.038)\tLoss 0.5033 (0.4889)\tPrec@1 85.156 (87.019)\n",
      "[100/391]\tTime 0.170 (0.151)\tData 0.046 (0.038)\tLoss 0.4553 (0.4918)\tPrec@1 87.500 (86.889)\n",
      "[110/391]\tTime 0.155 (0.150)\tData 0.039 (0.038)\tLoss 0.5408 (0.4936)\tPrec@1 89.844 (86.838)\n",
      "[120/391]\tTime 0.148 (0.150)\tData 0.065 (0.040)\tLoss 0.4359 (0.4893)\tPrec@1 85.938 (86.912)\n",
      "[130/391]\tTime 0.175 (0.151)\tData 0.057 (0.040)\tLoss 0.4268 (0.4906)\tPrec@1 89.062 (86.802)\n",
      "[140/391]\tTime 0.158 (0.151)\tData 0.034 (0.040)\tLoss 0.5121 (0.4922)\tPrec@1 85.938 (86.686)\n",
      "[150/391]\tTime 0.169 (0.151)\tData 0.037 (0.040)\tLoss 0.4179 (0.4922)\tPrec@1 92.969 (86.683)\n",
      "[160/391]\tTime 0.118 (0.151)\tData 0.031 (0.040)\tLoss 0.4080 (0.4914)\tPrec@1 89.844 (86.675)\n",
      "[170/391]\tTime 0.142 (0.151)\tData 0.037 (0.039)\tLoss 0.5560 (0.4914)\tPrec@1 85.156 (86.668)\n",
      "[180/391]\tTime 0.138 (0.152)\tData 0.047 (0.040)\tLoss 0.5275 (0.4913)\tPrec@1 86.719 (86.654)\n",
      "[190/391]\tTime 0.169 (0.151)\tData 0.031 (0.039)\tLoss 0.4790 (0.4894)\tPrec@1 86.719 (86.719)\n",
      "[200/391]\tTime 0.151 (0.151)\tData 0.032 (0.039)\tLoss 0.4381 (0.4884)\tPrec@1 86.719 (86.808)\n",
      "[210/391]\tTime 0.174 (0.151)\tData 0.052 (0.039)\tLoss 0.5389 (0.4887)\tPrec@1 85.156 (86.797)\n",
      "[220/391]\tTime 0.154 (0.151)\tData 0.035 (0.039)\tLoss 0.4348 (0.4885)\tPrec@1 89.062 (86.860)\n",
      "[230/391]\tTime 0.167 (0.151)\tData 0.035 (0.039)\tLoss 0.5461 (0.4883)\tPrec@1 86.719 (86.861)\n",
      "[240/391]\tTime 0.127 (0.151)\tData 0.034 (0.039)\tLoss 0.4685 (0.4870)\tPrec@1 89.062 (86.907)\n",
      "[250/391]\tTime 0.135 (0.151)\tData 0.042 (0.039)\tLoss 0.5721 (0.4863)\tPrec@1 84.375 (86.940)\n",
      "[260/391]\tTime 0.147 (0.150)\tData 0.039 (0.039)\tLoss 0.4324 (0.4861)\tPrec@1 88.281 (86.943)\n",
      "[270/391]\tTime 0.179 (0.150)\tData 0.058 (0.039)\tLoss 0.4894 (0.4869)\tPrec@1 88.281 (86.903)\n",
      "[280/391]\tTime 0.151 (0.151)\tData 0.033 (0.039)\tLoss 0.4302 (0.4872)\tPrec@1 89.062 (86.872)\n",
      "[290/391]\tTime 0.141 (0.151)\tData 0.042 (0.039)\tLoss 0.5722 (0.4895)\tPrec@1 78.906 (86.778)\n",
      "[300/391]\tTime 0.184 (0.151)\tData 0.066 (0.039)\tLoss 0.6179 (0.4909)\tPrec@1 82.812 (86.714)\n",
      "[310/391]\tTime 0.151 (0.151)\tData 0.037 (0.039)\tLoss 0.4762 (0.4894)\tPrec@1 87.500 (86.794)\n",
      "[320/391]\tTime 0.161 (0.152)\tData 0.039 (0.039)\tLoss 0.5464 (0.4891)\tPrec@1 82.812 (86.794)\n",
      "[330/391]\tTime 0.144 (0.152)\tData 0.028 (0.039)\tLoss 0.5166 (0.4907)\tPrec@1 85.156 (86.723)\n",
      "[340/391]\tTime 0.165 (0.152)\tData 0.041 (0.039)\tLoss 0.4707 (0.4901)\tPrec@1 86.719 (86.735)\n",
      "[350/391]\tTime 0.144 (0.152)\tData 0.038 (0.039)\tLoss 0.5397 (0.4905)\tPrec@1 85.156 (86.714)\n",
      "[360/391]\tTime 0.148 (0.152)\tData 0.055 (0.039)\tLoss 0.4346 (0.4909)\tPrec@1 89.062 (86.704)\n",
      "[370/391]\tTime 0.142 (0.152)\tData 0.034 (0.039)\tLoss 0.4211 (0.4908)\tPrec@1 89.062 (86.691)\n",
      "[380/391]\tTime 0.202 (0.152)\tData 0.078 (0.039)\tLoss 0.4491 (0.4907)\tPrec@1 85.938 (86.708)\n",
      "[390/391]\tTime 0.090 (0.153)\tData 0.016 (0.039)\tLoss 0.4419 (0.4897)\tPrec@1 90.000 (86.762)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.031 (0.031)\tLoss 0.9581 (0.9581)\tPrec@1 79.688 (79.688)\n",
      "Test: [10/79]\tTime 0.039 (0.034)\tLoss 0.8116 (0.9071)\tPrec@1 82.031 (78.835)\n",
      "Test: [20/79]\tTime 0.052 (0.041)\tLoss 0.6899 (0.9313)\tPrec@1 82.031 (77.939)\n",
      "Test: [30/79]\tTime 0.031 (0.041)\tLoss 0.9915 (0.9150)\tPrec@1 75.781 (78.125)\n",
      "Test: [40/79]\tTime 0.036 (0.040)\tLoss 1.1996 (0.9227)\tPrec@1 75.781 (78.316)\n",
      "Test: [50/79]\tTime 0.029 (0.040)\tLoss 0.9077 (0.9193)\tPrec@1 78.125 (78.248)\n",
      "Test: [60/79]\tTime 0.037 (0.039)\tLoss 1.1651 (0.9273)\tPrec@1 73.438 (78.163)\n",
      "Test: [70/79]\tTime 0.034 (0.039)\tLoss 0.7433 (0.9202)\tPrec@1 80.469 (78.400)\n",
      " * Prec@1 78.540\n",
      "\n",
      "===> epoch: 37/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.129 (0.129)\tData 0.040 (0.040)\tLoss 0.4949 (0.4949)\tPrec@1 86.719 (86.719)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[10/391]\tTime 0.133 (0.124)\tData 0.046 (0.037)\tLoss 0.5561 (0.4444)\tPrec@1 85.938 (88.565)\n",
      "[20/391]\tTime 0.178 (0.132)\tData 0.050 (0.035)\tLoss 0.4427 (0.4656)\tPrec@1 88.281 (87.574)\n",
      "[30/391]\tTime 0.158 (0.140)\tData 0.036 (0.037)\tLoss 0.3749 (0.4658)\tPrec@1 89.844 (87.450)\n",
      "[40/391]\tTime 0.136 (0.142)\tData 0.023 (0.036)\tLoss 0.5203 (0.4679)\tPrec@1 88.281 (87.595)\n",
      "[50/391]\tTime 0.169 (0.146)\tData 0.037 (0.036)\tLoss 0.6479 (0.4623)\tPrec@1 79.688 (87.883)\n",
      "[60/391]\tTime 0.158 (0.147)\tData 0.048 (0.037)\tLoss 0.5078 (0.4663)\tPrec@1 85.938 (87.666)\n",
      "[70/391]\tTime 0.139 (0.146)\tData 0.027 (0.036)\tLoss 0.4675 (0.4704)\tPrec@1 86.719 (87.335)\n",
      "[80/391]\tTime 0.158 (0.147)\tData 0.039 (0.036)\tLoss 0.3824 (0.4692)\tPrec@1 91.406 (87.365)\n",
      "[90/391]\tTime 0.132 (0.146)\tData 0.033 (0.036)\tLoss 0.5312 (0.4660)\tPrec@1 81.250 (87.509)\n",
      "[100/391]\tTime 0.142 (0.146)\tData 0.036 (0.037)\tLoss 0.4674 (0.4669)\tPrec@1 88.281 (87.430)\n",
      "[110/391]\tTime 0.138 (0.146)\tData 0.034 (0.037)\tLoss 0.4443 (0.4712)\tPrec@1 85.938 (87.261)\n",
      "[120/391]\tTime 0.144 (0.147)\tData 0.056 (0.038)\tLoss 0.5009 (0.4690)\tPrec@1 92.188 (87.390)\n",
      "[130/391]\tTime 0.117 (0.148)\tData 0.030 (0.038)\tLoss 0.4426 (0.4672)\tPrec@1 90.625 (87.446)\n",
      "[140/391]\tTime 0.163 (0.148)\tData 0.039 (0.038)\tLoss 0.5189 (0.4691)\tPrec@1 86.719 (87.361)\n",
      "[150/391]\tTime 0.140 (0.147)\tData 0.045 (0.039)\tLoss 0.4929 (0.4722)\tPrec@1 85.156 (87.288)\n",
      "[160/391]\tTime 0.142 (0.148)\tData 0.033 (0.040)\tLoss 0.4690 (0.4714)\tPrec@1 87.500 (87.291)\n",
      "[170/391]\tTime 0.192 (0.149)\tData 0.047 (0.040)\tLoss 0.4828 (0.4735)\tPrec@1 84.375 (87.230)\n",
      "[180/391]\tTime 0.163 (0.150)\tData 0.031 (0.040)\tLoss 0.4601 (0.4748)\tPrec@1 92.188 (87.254)\n",
      "[190/391]\tTime 0.151 (0.150)\tData 0.033 (0.040)\tLoss 0.4574 (0.4767)\tPrec@1 87.500 (87.205)\n",
      "[200/391]\tTime 0.143 (0.150)\tData 0.034 (0.040)\tLoss 0.6137 (0.4774)\tPrec@1 79.688 (87.181)\n",
      "[210/391]\tTime 0.145 (0.150)\tData 0.036 (0.040)\tLoss 0.4564 (0.4781)\tPrec@1 87.500 (87.167)\n",
      "[220/391]\tTime 0.138 (0.150)\tData 0.035 (0.040)\tLoss 0.5665 (0.4785)\tPrec@1 87.500 (87.146)\n",
      "[230/391]\tTime 0.155 (0.150)\tData 0.033 (0.040)\tLoss 0.4946 (0.4788)\tPrec@1 84.375 (87.169)\n",
      "[240/391]\tTime 0.144 (0.149)\tData 0.035 (0.039)\tLoss 0.4572 (0.4779)\tPrec@1 89.062 (87.182)\n",
      "[250/391]\tTime 0.152 (0.150)\tData 0.040 (0.039)\tLoss 0.3877 (0.4775)\tPrec@1 89.844 (87.198)\n",
      "[260/391]\tTime 0.138 (0.150)\tData 0.033 (0.040)\tLoss 0.4608 (0.4776)\tPrec@1 89.062 (87.219)\n",
      "[270/391]\tTime 0.133 (0.150)\tData 0.030 (0.039)\tLoss 0.5591 (0.4779)\tPrec@1 78.906 (87.163)\n",
      "[280/391]\tTime 0.137 (0.149)\tData 0.032 (0.039)\tLoss 0.4811 (0.4788)\tPrec@1 84.375 (87.125)\n",
      "[290/391]\tTime 0.110 (0.149)\tData 0.037 (0.039)\tLoss 0.5839 (0.4790)\tPrec@1 82.812 (87.130)\n",
      "[300/391]\tTime 0.163 (0.149)\tData 0.062 (0.039)\tLoss 0.3797 (0.4789)\tPrec@1 90.625 (87.129)\n",
      "[310/391]\tTime 0.150 (0.149)\tData 0.033 (0.039)\tLoss 0.6027 (0.4815)\tPrec@1 85.156 (87.058)\n",
      "[320/391]\tTime 0.121 (0.149)\tData 0.034 (0.039)\tLoss 0.4374 (0.4814)\tPrec@1 86.719 (87.052)\n",
      "[330/391]\tTime 0.116 (0.148)\tData 0.026 (0.039)\tLoss 0.4517 (0.4815)\tPrec@1 89.062 (87.059)\n",
      "[340/391]\tTime 0.140 (0.148)\tData 0.040 (0.039)\tLoss 0.4899 (0.4823)\tPrec@1 85.938 (87.003)\n",
      "[350/391]\tTime 0.145 (0.149)\tData 0.045 (0.039)\tLoss 0.5241 (0.4820)\tPrec@1 85.156 (87.015)\n",
      "[360/391]\tTime 0.112 (0.148)\tData 0.029 (0.039)\tLoss 0.6113 (0.4829)\tPrec@1 85.938 (86.981)\n",
      "[370/391]\tTime 0.162 (0.149)\tData 0.045 (0.039)\tLoss 0.5242 (0.4832)\tPrec@1 84.375 (86.976)\n",
      "[380/391]\tTime 0.124 (0.149)\tData 0.032 (0.039)\tLoss 0.4075 (0.4846)\tPrec@1 87.500 (86.940)\n",
      "[390/391]\tTime 0.109 (0.149)\tData 0.025 (0.039)\tLoss 0.5211 (0.4858)\tPrec@1 90.000 (86.914)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.034 (0.034)\tLoss 0.6229 (0.6229)\tPrec@1 79.688 (79.688)\n",
      "Test: [10/79]\tTime 0.029 (0.032)\tLoss 1.2789 (1.0247)\tPrec@1 74.219 (76.420)\n",
      "Test: [20/79]\tTime 0.035 (0.036)\tLoss 1.1382 (1.1257)\tPrec@1 73.438 (74.814)\n",
      "Test: [30/79]\tTime 0.033 (0.037)\tLoss 1.1090 (1.1440)\tPrec@1 67.188 (74.269)\n",
      "Test: [40/79]\tTime 0.041 (0.038)\tLoss 1.1123 (1.1453)\tPrec@1 74.219 (74.238)\n",
      "Test: [50/79]\tTime 0.038 (0.038)\tLoss 1.1318 (1.1499)\tPrec@1 71.875 (73.928)\n",
      "Test: [60/79]\tTime 0.050 (0.038)\tLoss 1.1041 (1.1270)\tPrec@1 77.344 (74.372)\n",
      "Test: [70/79]\tTime 0.035 (0.038)\tLoss 1.2006 (1.1363)\tPrec@1 75.000 (74.098)\n",
      " * Prec@1 74.020\n",
      "\n",
      "===> epoch: 38/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.179 (0.179)\tData 0.055 (0.055)\tLoss 0.4208 (0.4208)\tPrec@1 89.844 (89.844)\n",
      "[10/391]\tTime 0.140 (0.159)\tData 0.021 (0.038)\tLoss 0.4212 (0.4623)\tPrec@1 90.625 (88.210)\n",
      "[20/391]\tTime 0.193 (0.156)\tData 0.063 (0.039)\tLoss 0.4976 (0.4866)\tPrec@1 84.375 (87.277)\n",
      "[30/391]\tTime 0.145 (0.154)\tData 0.029 (0.040)\tLoss 0.5403 (0.4928)\tPrec@1 83.594 (87.198)\n",
      "[40/391]\tTime 0.138 (0.150)\tData 0.032 (0.039)\tLoss 0.4888 (0.5055)\tPrec@1 84.375 (86.643)\n",
      "[50/391]\tTime 0.138 (0.151)\tData 0.045 (0.040)\tLoss 0.4096 (0.4933)\tPrec@1 88.281 (86.857)\n",
      "[60/391]\tTime 0.132 (0.152)\tData 0.034 (0.041)\tLoss 0.4959 (0.4870)\tPrec@1 88.281 (86.988)\n",
      "[70/391]\tTime 0.131 (0.150)\tData 0.027 (0.039)\tLoss 0.4545 (0.4812)\tPrec@1 86.719 (87.115)\n",
      "[80/391]\tTime 0.140 (0.150)\tData 0.039 (0.039)\tLoss 0.4409 (0.4733)\tPrec@1 91.406 (87.355)\n",
      "[90/391]\tTime 0.173 (0.152)\tData 0.027 (0.039)\tLoss 0.5123 (0.4729)\tPrec@1 85.938 (87.388)\n",
      "[100/391]\tTime 0.174 (0.151)\tData 0.035 (0.039)\tLoss 0.4581 (0.4721)\tPrec@1 89.062 (87.314)\n",
      "[110/391]\tTime 0.166 (0.152)\tData 0.034 (0.039)\tLoss 0.3807 (0.4677)\tPrec@1 92.188 (87.444)\n",
      "[120/391]\tTime 0.160 (0.152)\tData 0.033 (0.039)\tLoss 0.4664 (0.4672)\tPrec@1 83.594 (87.410)\n",
      "[130/391]\tTime 0.146 (0.153)\tData 0.037 (0.039)\tLoss 0.4343 (0.4699)\tPrec@1 89.844 (87.309)\n",
      "[140/391]\tTime 0.156 (0.152)\tData 0.035 (0.039)\tLoss 0.4217 (0.4696)\tPrec@1 88.281 (87.306)\n",
      "[150/391]\tTime 0.132 (0.153)\tData 0.027 (0.039)\tLoss 0.5522 (0.4709)\tPrec@1 85.156 (87.241)\n",
      "[160/391]\tTime 0.166 (0.153)\tData 0.034 (0.039)\tLoss 0.4836 (0.4723)\tPrec@1 85.938 (87.199)\n",
      "[170/391]\tTime 0.138 (0.154)\tData 0.036 (0.039)\tLoss 0.6116 (0.4743)\tPrec@1 85.156 (87.162)\n",
      "[180/391]\tTime 0.132 (0.152)\tData 0.029 (0.039)\tLoss 0.4149 (0.4736)\tPrec@1 87.500 (87.194)\n",
      "[190/391]\tTime 0.160 (0.151)\tData 0.038 (0.039)\tLoss 0.5164 (0.4729)\tPrec@1 89.062 (87.238)\n",
      "[200/391]\tTime 0.132 (0.151)\tData 0.039 (0.038)\tLoss 0.4054 (0.4736)\tPrec@1 89.844 (87.189)\n",
      "[210/391]\tTime 0.170 (0.151)\tData 0.048 (0.038)\tLoss 0.5934 (0.4751)\tPrec@1 82.812 (87.108)\n",
      "[220/391]\tTime 0.116 (0.150)\tData 0.035 (0.038)\tLoss 0.4263 (0.4761)\tPrec@1 87.500 (87.033)\n",
      "[230/391]\tTime 0.124 (0.149)\tData 0.035 (0.038)\tLoss 0.5428 (0.4789)\tPrec@1 81.250 (86.905)\n",
      "[240/391]\tTime 0.142 (0.149)\tData 0.032 (0.038)\tLoss 0.4703 (0.4786)\tPrec@1 80.469 (86.871)\n",
      "[250/391]\tTime 0.134 (0.148)\tData 0.030 (0.038)\tLoss 0.4284 (0.4796)\tPrec@1 89.844 (86.846)\n",
      "[260/391]\tTime 0.214 (0.149)\tData 0.072 (0.038)\tLoss 0.4297 (0.4808)\tPrec@1 90.625 (86.832)\n",
      "[270/391]\tTime 0.155 (0.148)\tData 0.035 (0.038)\tLoss 0.4974 (0.4810)\tPrec@1 85.938 (86.843)\n",
      "[280/391]\tTime 0.139 (0.149)\tData 0.036 (0.038)\tLoss 0.5384 (0.4824)\tPrec@1 86.719 (86.772)\n",
      "[290/391]\tTime 0.128 (0.149)\tData 0.035 (0.038)\tLoss 0.5790 (0.4813)\tPrec@1 87.500 (86.781)\n",
      "[300/391]\tTime 0.182 (0.149)\tData 0.045 (0.038)\tLoss 0.4493 (0.4809)\tPrec@1 88.281 (86.802)\n",
      "[310/391]\tTime 0.192 (0.149)\tData 0.035 (0.038)\tLoss 0.3868 (0.4826)\tPrec@1 92.188 (86.771)\n",
      "[320/391]\tTime 0.140 (0.150)\tData 0.034 (0.038)\tLoss 0.6243 (0.4836)\tPrec@1 81.250 (86.736)\n",
      "[330/391]\tTime 0.138 (0.150)\tData 0.031 (0.038)\tLoss 0.4330 (0.4831)\tPrec@1 87.500 (86.745)\n",
      "[340/391]\tTime 0.136 (0.149)\tData 0.038 (0.038)\tLoss 0.5564 (0.4835)\tPrec@1 83.594 (86.749)\n",
      "[350/391]\tTime 0.162 (0.149)\tData 0.034 (0.038)\tLoss 0.4941 (0.4837)\tPrec@1 83.594 (86.748)\n",
      "[360/391]\tTime 0.116 (0.149)\tData 0.030 (0.038)\tLoss 0.4666 (0.4838)\tPrec@1 87.500 (86.758)\n",
      "[370/391]\tTime 0.145 (0.149)\tData 0.035 (0.038)\tLoss 0.5833 (0.4859)\tPrec@1 84.375 (86.708)\n",
      "[380/391]\tTime 0.122 (0.149)\tData 0.031 (0.038)\tLoss 0.4829 (0.4864)\tPrec@1 84.375 (86.692)\n",
      "[390/391]\tTime 0.107 (0.149)\tData 0.025 (0.038)\tLoss 0.5808 (0.4870)\tPrec@1 81.250 (86.684)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.031 (0.031)\tLoss 1.4434 (1.4434)\tPrec@1 71.875 (71.875)\n",
      "Test: [10/79]\tTime 0.031 (0.032)\tLoss 1.0316 (1.3249)\tPrec@1 77.344 (73.580)\n",
      "Test: [20/79]\tTime 0.043 (0.036)\tLoss 1.4542 (1.3261)\tPrec@1 68.750 (72.024)\n",
      "Test: [30/79]\tTime 0.038 (0.038)\tLoss 1.2147 (1.3259)\tPrec@1 75.781 (72.127)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test: [40/79]\tTime 0.033 (0.039)\tLoss 1.5837 (1.3121)\tPrec@1 66.406 (72.199)\n",
      "Test: [50/79]\tTime 0.041 (0.038)\tLoss 1.3296 (1.2980)\tPrec@1 71.875 (72.457)\n",
      "Test: [60/79]\tTime 0.035 (0.039)\tLoss 0.9588 (1.3066)\tPrec@1 79.688 (72.515)\n",
      "Test: [70/79]\tTime 0.044 (0.038)\tLoss 1.2082 (1.2991)\tPrec@1 76.562 (72.766)\n",
      " * Prec@1 72.840\n",
      "\n",
      "===> epoch: 39/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.177 (0.177)\tData 0.040 (0.040)\tLoss 0.4595 (0.4595)\tPrec@1 86.719 (86.719)\n",
      "[10/391]\tTime 0.124 (0.146)\tData 0.031 (0.040)\tLoss 0.5559 (0.4819)\tPrec@1 82.031 (87.500)\n",
      "[20/391]\tTime 0.112 (0.148)\tData 0.037 (0.042)\tLoss 0.4056 (0.4674)\tPrec@1 89.844 (87.723)\n",
      "[30/391]\tTime 0.158 (0.153)\tData 0.041 (0.042)\tLoss 0.4373 (0.4728)\tPrec@1 88.281 (87.550)\n",
      "[40/391]\tTime 0.183 (0.150)\tData 0.044 (0.039)\tLoss 0.3546 (0.4636)\tPrec@1 91.406 (87.900)\n",
      "[50/391]\tTime 0.152 (0.152)\tData 0.038 (0.040)\tLoss 0.4209 (0.4631)\tPrec@1 87.500 (88.006)\n",
      "[60/391]\tTime 0.137 (0.153)\tData 0.028 (0.040)\tLoss 0.5139 (0.4635)\tPrec@1 88.281 (88.025)\n",
      "[70/391]\tTime 0.146 (0.152)\tData 0.056 (0.040)\tLoss 0.3854 (0.4629)\tPrec@1 90.625 (88.006)\n",
      "[80/391]\tTime 0.145 (0.151)\tData 0.033 (0.040)\tLoss 0.4864 (0.4674)\tPrec@1 87.500 (87.770)\n",
      "[90/391]\tTime 0.134 (0.151)\tData 0.038 (0.041)\tLoss 0.4884 (0.4689)\tPrec@1 84.375 (87.646)\n",
      "[100/391]\tTime 0.141 (0.151)\tData 0.042 (0.040)\tLoss 0.5583 (0.4754)\tPrec@1 82.031 (87.353)\n",
      "[110/391]\tTime 0.118 (0.150)\tData 0.032 (0.040)\tLoss 0.4848 (0.4783)\tPrec@1 89.844 (87.218)\n",
      "[120/391]\tTime 0.188 (0.150)\tData 0.069 (0.040)\tLoss 0.5715 (0.4804)\tPrec@1 82.812 (87.171)\n",
      "[130/391]\tTime 0.130 (0.149)\tData 0.031 (0.040)\tLoss 0.3889 (0.4813)\tPrec@1 90.625 (87.166)\n",
      "[140/391]\tTime 0.125 (0.147)\tData 0.036 (0.040)\tLoss 0.5322 (0.4807)\tPrec@1 87.500 (87.245)\n",
      "[150/391]\tTime 0.173 (0.148)\tData 0.046 (0.040)\tLoss 0.4044 (0.4798)\tPrec@1 89.062 (87.262)\n",
      "[160/391]\tTime 0.123 (0.149)\tData 0.026 (0.039)\tLoss 0.4193 (0.4805)\tPrec@1 89.062 (87.248)\n",
      "[170/391]\tTime 0.136 (0.149)\tData 0.034 (0.039)\tLoss 0.5709 (0.4827)\tPrec@1 80.469 (87.112)\n",
      "[180/391]\tTime 0.160 (0.149)\tData 0.030 (0.039)\tLoss 0.3299 (0.4809)\tPrec@1 93.750 (87.155)\n",
      "[190/391]\tTime 0.187 (0.149)\tData 0.048 (0.039)\tLoss 0.4623 (0.4830)\tPrec@1 86.719 (87.038)\n",
      "[200/391]\tTime 0.161 (0.150)\tData 0.040 (0.039)\tLoss 0.5023 (0.4835)\tPrec@1 89.844 (87.022)\n",
      "[210/391]\tTime 0.158 (0.149)\tData 0.032 (0.039)\tLoss 0.3675 (0.4820)\tPrec@1 91.406 (87.100)\n",
      "[220/391]\tTime 0.165 (0.150)\tData 0.034 (0.039)\tLoss 0.4375 (0.4815)\tPrec@1 90.625 (87.129)\n",
      "[230/391]\tTime 0.142 (0.149)\tData 0.035 (0.039)\tLoss 0.4331 (0.4815)\tPrec@1 89.062 (87.108)\n",
      "[240/391]\tTime 0.121 (0.148)\tData 0.034 (0.039)\tLoss 0.5337 (0.4837)\tPrec@1 85.938 (87.066)\n",
      "[250/391]\tTime 0.167 (0.148)\tData 0.045 (0.039)\tLoss 0.5241 (0.4842)\tPrec@1 89.844 (87.049)\n",
      "[260/391]\tTime 0.124 (0.149)\tData 0.034 (0.039)\tLoss 0.6187 (0.4853)\tPrec@1 79.688 (86.979)\n",
      "[270/391]\tTime 0.137 (0.149)\tData 0.029 (0.039)\tLoss 0.3730 (0.4860)\tPrec@1 89.062 (86.944)\n",
      "[280/391]\tTime 0.166 (0.149)\tData 0.040 (0.039)\tLoss 0.5119 (0.4867)\tPrec@1 86.719 (86.930)\n",
      "[290/391]\tTime 0.175 (0.149)\tData 0.033 (0.039)\tLoss 0.6154 (0.4872)\tPrec@1 78.906 (86.904)\n",
      "[300/391]\tTime 0.142 (0.149)\tData 0.029 (0.039)\tLoss 0.5044 (0.4859)\tPrec@1 87.500 (86.929)\n",
      "[310/391]\tTime 0.151 (0.150)\tData 0.038 (0.039)\tLoss 0.4386 (0.4853)\tPrec@1 88.281 (86.960)\n",
      "[320/391]\tTime 0.125 (0.150)\tData 0.037 (0.039)\tLoss 0.4929 (0.4854)\tPrec@1 83.594 (86.943)\n",
      "[330/391]\tTime 0.167 (0.149)\tData 0.047 (0.039)\tLoss 0.4667 (0.4844)\tPrec@1 88.281 (86.962)\n",
      "[340/391]\tTime 0.151 (0.150)\tData 0.045 (0.039)\tLoss 0.5320 (0.4846)\tPrec@1 87.500 (86.966)\n",
      "[350/391]\tTime 0.176 (0.150)\tData 0.047 (0.040)\tLoss 0.4327 (0.4840)\tPrec@1 88.281 (86.977)\n",
      "[360/391]\tTime 0.167 (0.150)\tData 0.028 (0.040)\tLoss 0.4911 (0.4842)\tPrec@1 85.156 (86.961)\n",
      "[370/391]\tTime 0.153 (0.150)\tData 0.037 (0.040)\tLoss 0.5268 (0.4838)\tPrec@1 85.938 (86.969)\n",
      "[380/391]\tTime 0.127 (0.150)\tData 0.035 (0.040)\tLoss 0.4762 (0.4842)\tPrec@1 90.625 (86.957)\n",
      "[390/391]\tTime 0.106 (0.150)\tData 0.036 (0.039)\tLoss 0.5775 (0.4856)\tPrec@1 86.250 (86.892)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.033 (0.033)\tLoss 0.8654 (0.8654)\tPrec@1 79.688 (79.688)\n",
      "Test: [10/79]\tTime 0.023 (0.029)\tLoss 0.9336 (0.8423)\tPrec@1 79.688 (79.688)\n",
      "Test: [20/79]\tTime 0.038 (0.030)\tLoss 0.7352 (0.8726)\tPrec@1 81.250 (79.204)\n",
      "Test: [30/79]\tTime 0.039 (0.033)\tLoss 0.6435 (0.8443)\tPrec@1 82.031 (79.612)\n",
      "Test: [40/79]\tTime 0.035 (0.034)\tLoss 0.7155 (0.8383)\tPrec@1 78.906 (79.745)\n",
      "Test: [50/79]\tTime 0.035 (0.035)\tLoss 0.7560 (0.8217)\tPrec@1 79.688 (80.009)\n",
      "Test: [60/79]\tTime 0.036 (0.036)\tLoss 1.0040 (0.8201)\tPrec@1 75.781 (80.046)\n",
      "Test: [70/79]\tTime 0.042 (0.037)\tLoss 0.9204 (0.8210)\tPrec@1 80.469 (79.908)\n",
      " * Prec@1 79.850\n",
      "\n",
      "===> epoch: 40/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.170 (0.170)\tData 0.036 (0.036)\tLoss 0.4083 (0.4083)\tPrec@1 91.406 (91.406)\n",
      "[10/391]\tTime 0.131 (0.151)\tData 0.027 (0.032)\tLoss 0.5957 (0.4892)\tPrec@1 82.031 (86.293)\n",
      "[20/391]\tTime 0.105 (0.153)\tData 0.031 (0.036)\tLoss 0.5028 (0.4781)\tPrec@1 88.281 (86.570)\n",
      "[30/391]\tTime 0.117 (0.147)\tData 0.033 (0.038)\tLoss 0.5701 (0.4797)\tPrec@1 85.938 (86.694)\n",
      "[40/391]\tTime 0.123 (0.144)\tData 0.024 (0.039)\tLoss 0.3518 (0.4713)\tPrec@1 91.406 (86.928)\n",
      "[50/391]\tTime 0.156 (0.146)\tData 0.033 (0.038)\tLoss 0.5529 (0.4703)\tPrec@1 82.031 (86.903)\n",
      "[60/391]\tTime 0.127 (0.146)\tData 0.033 (0.038)\tLoss 0.5328 (0.4742)\tPrec@1 87.500 (86.975)\n",
      "[70/391]\tTime 0.122 (0.145)\tData 0.027 (0.037)\tLoss 0.6235 (0.4744)\tPrec@1 85.938 (87.137)\n",
      "[80/391]\tTime 0.138 (0.146)\tData 0.040 (0.037)\tLoss 0.4543 (0.4719)\tPrec@1 88.281 (87.288)\n",
      "[90/391]\tTime 0.154 (0.148)\tData 0.034 (0.037)\tLoss 0.4017 (0.4710)\tPrec@1 86.719 (87.251)\n",
      "[100/391]\tTime 0.158 (0.148)\tData 0.040 (0.037)\tLoss 0.6284 (0.4724)\tPrec@1 82.812 (87.229)\n",
      "[110/391]\tTime 0.169 (0.148)\tData 0.036 (0.037)\tLoss 0.4559 (0.4737)\tPrec@1 85.938 (87.211)\n",
      "[120/391]\tTime 0.137 (0.149)\tData 0.031 (0.037)\tLoss 0.4565 (0.4761)\tPrec@1 86.719 (87.106)\n",
      "[130/391]\tTime 0.148 (0.148)\tData 0.036 (0.037)\tLoss 0.4959 (0.4783)\tPrec@1 85.938 (87.011)\n",
      "[140/391]\tTime 0.179 (0.150)\tData 0.033 (0.038)\tLoss 0.5018 (0.4775)\tPrec@1 86.719 (86.996)\n",
      "[150/391]\tTime 0.181 (0.150)\tData 0.060 (0.038)\tLoss 0.3799 (0.4772)\tPrec@1 86.719 (86.983)\n",
      "[160/391]\tTime 0.143 (0.150)\tData 0.035 (0.038)\tLoss 0.3948 (0.4784)\tPrec@1 91.406 (87.005)\n",
      "[170/391]\tTime 0.130 (0.150)\tData 0.032 (0.038)\tLoss 0.3921 (0.4785)\tPrec@1 90.625 (87.043)\n",
      "[180/391]\tTime 0.157 (0.150)\tData 0.034 (0.038)\tLoss 0.4734 (0.4780)\tPrec@1 86.719 (87.068)\n",
      "[190/391]\tTime 0.138 (0.150)\tData 0.037 (0.038)\tLoss 0.4270 (0.4769)\tPrec@1 89.062 (87.144)\n",
      "[200/391]\tTime 0.146 (0.152)\tData 0.053 (0.039)\tLoss 0.4871 (0.4751)\tPrec@1 86.719 (87.236)\n",
      "[210/391]\tTime 0.165 (0.151)\tData 0.036 (0.039)\tLoss 0.3892 (0.4734)\tPrec@1 91.406 (87.278)\n",
      "[220/391]\tTime 0.143 (0.150)\tData 0.033 (0.039)\tLoss 0.4314 (0.4738)\tPrec@1 88.281 (87.231)\n",
      "[230/391]\tTime 0.168 (0.150)\tData 0.057 (0.039)\tLoss 0.5784 (0.4751)\tPrec@1 83.594 (87.240)\n",
      "[240/391]\tTime 0.146 (0.151)\tData 0.028 (0.039)\tLoss 0.5548 (0.4757)\tPrec@1 85.156 (87.192)\n",
      "[250/391]\tTime 0.127 (0.150)\tData 0.034 (0.038)\tLoss 0.4729 (0.4764)\tPrec@1 85.938 (87.183)\n",
      "[260/391]\tTime 0.136 (0.151)\tData 0.036 (0.039)\tLoss 0.6468 (0.4774)\tPrec@1 78.125 (87.138)\n",
      "[270/391]\tTime 0.154 (0.151)\tData 0.035 (0.039)\tLoss 0.4915 (0.4797)\tPrec@1 85.938 (87.068)\n",
      "[280/391]\tTime 0.147 (0.151)\tData 0.033 (0.039)\tLoss 0.4914 (0.4812)\tPrec@1 85.938 (87.013)\n",
      "[290/391]\tTime 0.196 (0.151)\tData 0.045 (0.039)\tLoss 0.4365 (0.4830)\tPrec@1 87.500 (86.995)\n",
      "[300/391]\tTime 0.148 (0.151)\tData 0.029 (0.039)\tLoss 0.5281 (0.4823)\tPrec@1 89.062 (87.033)\n",
      "[310/391]\tTime 0.145 (0.151)\tData 0.034 (0.039)\tLoss 0.5477 (0.4829)\tPrec@1 83.594 (87.005)\n",
      "[320/391]\tTime 0.144 (0.151)\tData 0.031 (0.039)\tLoss 0.4408 (0.4835)\tPrec@1 87.500 (87.008)\n",
      "[330/391]\tTime 0.142 (0.151)\tData 0.031 (0.039)\tLoss 0.4215 (0.4836)\tPrec@1 88.281 (87.011)\n",
      "[340/391]\tTime 0.120 (0.150)\tData 0.034 (0.039)\tLoss 0.5199 (0.4843)\tPrec@1 84.375 (86.996)\n",
      "[350/391]\tTime 0.136 (0.150)\tData 0.038 (0.039)\tLoss 0.4280 (0.4837)\tPrec@1 87.500 (87.039)\n",
      "[360/391]\tTime 0.124 (0.149)\tData 0.034 (0.039)\tLoss 0.4685 (0.4836)\tPrec@1 89.062 (87.046)\n",
      "[370/391]\tTime 0.137 (0.149)\tData 0.035 (0.039)\tLoss 0.5201 (0.4837)\tPrec@1 87.500 (87.037)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[380/391]\tTime 0.142 (0.150)\tData 0.051 (0.039)\tLoss 0.4654 (0.4828)\tPrec@1 86.719 (87.061)\n",
      "[390/391]\tTime 0.151 (0.150)\tData 0.052 (0.039)\tLoss 0.4035 (0.4832)\tPrec@1 86.250 (87.042)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.036 (0.036)\tLoss 0.8753 (0.8753)\tPrec@1 82.031 (82.031)\n",
      "Test: [10/79]\tTime 0.036 (0.035)\tLoss 1.5283 (1.0207)\tPrec@1 71.875 (76.918)\n",
      "Test: [20/79]\tTime 0.039 (0.036)\tLoss 1.0634 (1.1188)\tPrec@1 80.469 (76.339)\n",
      "Test: [30/79]\tTime 0.041 (0.037)\tLoss 0.7629 (1.0949)\tPrec@1 78.125 (76.462)\n",
      "Test: [40/79]\tTime 0.040 (0.039)\tLoss 1.3979 (1.0664)\tPrec@1 72.656 (77.153)\n",
      "Test: [50/79]\tTime 0.034 (0.039)\tLoss 0.8051 (1.0538)\tPrec@1 79.688 (77.191)\n",
      "Test: [60/79]\tTime 0.034 (0.040)\tLoss 1.4290 (1.0476)\tPrec@1 75.000 (77.241)\n",
      "Test: [70/79]\tTime 0.032 (0.039)\tLoss 0.9584 (1.0440)\tPrec@1 77.344 (77.212)\n",
      " * Prec@1 77.290\n",
      "\n",
      "===> epoch: 41/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.140 (0.140)\tData 0.037 (0.037)\tLoss 0.4678 (0.4678)\tPrec@1 90.625 (90.625)\n",
      "[10/391]\tTime 0.139 (0.145)\tData 0.029 (0.038)\tLoss 0.5328 (0.4833)\tPrec@1 84.375 (87.145)\n",
      "[20/391]\tTime 0.139 (0.143)\tData 0.050 (0.040)\tLoss 0.4070 (0.4909)\tPrec@1 89.062 (86.942)\n",
      "[30/391]\tTime 0.176 (0.146)\tData 0.034 (0.041)\tLoss 0.4083 (0.4819)\tPrec@1 89.844 (87.072)\n",
      "[40/391]\tTime 0.143 (0.147)\tData 0.026 (0.040)\tLoss 0.4574 (0.4844)\tPrec@1 86.719 (86.966)\n",
      "[50/391]\tTime 0.164 (0.146)\tData 0.034 (0.039)\tLoss 0.4514 (0.4817)\tPrec@1 88.281 (86.994)\n",
      "[60/391]\tTime 0.135 (0.149)\tData 0.036 (0.041)\tLoss 0.4197 (0.4822)\tPrec@1 87.500 (87.077)\n",
      "[70/391]\tTime 0.141 (0.148)\tData 0.037 (0.040)\tLoss 0.4336 (0.4819)\tPrec@1 89.062 (86.983)\n",
      "[80/391]\tTime 0.216 (0.148)\tData 0.080 (0.040)\tLoss 0.5111 (0.4810)\tPrec@1 85.156 (87.018)\n",
      "[90/391]\tTime 0.171 (0.149)\tData 0.055 (0.040)\tLoss 0.3892 (0.4787)\tPrec@1 86.719 (87.079)\n",
      "[100/391]\tTime 0.130 (0.149)\tData 0.035 (0.039)\tLoss 0.4681 (0.4804)\tPrec@1 85.938 (87.144)\n",
      "[110/391]\tTime 0.151 (0.149)\tData 0.050 (0.039)\tLoss 0.4746 (0.4794)\tPrec@1 89.062 (87.155)\n",
      "[120/391]\tTime 0.117 (0.149)\tData 0.033 (0.040)\tLoss 0.4757 (0.4768)\tPrec@1 87.500 (87.268)\n",
      "[130/391]\tTime 0.117 (0.149)\tData 0.027 (0.040)\tLoss 0.6510 (0.4792)\tPrec@1 81.250 (87.184)\n",
      "[140/391]\tTime 0.160 (0.150)\tData 0.035 (0.040)\tLoss 0.3783 (0.4788)\tPrec@1 89.844 (87.212)\n",
      "[150/391]\tTime 0.170 (0.150)\tData 0.045 (0.040)\tLoss 0.5712 (0.4810)\tPrec@1 85.156 (87.133)\n",
      "[160/391]\tTime 0.133 (0.150)\tData 0.028 (0.040)\tLoss 0.4665 (0.4807)\tPrec@1 91.406 (87.160)\n",
      "[170/391]\tTime 0.117 (0.150)\tData 0.037 (0.040)\tLoss 0.5935 (0.4818)\tPrec@1 82.031 (87.112)\n",
      "[180/391]\tTime 0.112 (0.150)\tData 0.035 (0.040)\tLoss 0.3719 (0.4815)\tPrec@1 92.969 (87.107)\n",
      "[190/391]\tTime 0.114 (0.149)\tData 0.035 (0.040)\tLoss 0.5779 (0.4807)\tPrec@1 82.031 (87.148)\n",
      "[200/391]\tTime 0.128 (0.149)\tData 0.035 (0.039)\tLoss 0.5517 (0.4803)\tPrec@1 83.594 (87.170)\n",
      "[210/391]\tTime 0.143 (0.149)\tData 0.029 (0.039)\tLoss 0.4888 (0.4800)\tPrec@1 85.938 (87.156)\n",
      "[220/391]\tTime 0.185 (0.149)\tData 0.085 (0.039)\tLoss 0.5297 (0.4785)\tPrec@1 83.594 (87.210)\n",
      "[230/391]\tTime 0.174 (0.149)\tData 0.051 (0.039)\tLoss 0.5184 (0.4775)\tPrec@1 82.812 (87.223)\n",
      "[240/391]\tTime 0.108 (0.149)\tData 0.028 (0.039)\tLoss 0.4579 (0.4780)\tPrec@1 86.719 (87.186)\n",
      "[250/391]\tTime 0.131 (0.148)\tData 0.035 (0.039)\tLoss 0.4203 (0.4793)\tPrec@1 91.406 (87.167)\n",
      "[260/391]\tTime 0.143 (0.147)\tData 0.034 (0.039)\tLoss 0.4431 (0.4812)\tPrec@1 89.844 (87.120)\n",
      "[270/391]\tTime 0.140 (0.147)\tData 0.033 (0.039)\tLoss 0.4122 (0.4835)\tPrec@1 88.281 (87.010)\n",
      "[280/391]\tTime 0.150 (0.147)\tData 0.030 (0.039)\tLoss 0.5097 (0.4856)\tPrec@1 87.500 (86.961)\n",
      "[290/391]\tTime 0.152 (0.148)\tData 0.035 (0.039)\tLoss 0.4751 (0.4847)\tPrec@1 87.500 (86.971)\n",
      "[300/391]\tTime 0.147 (0.148)\tData 0.029 (0.039)\tLoss 0.4746 (0.4858)\tPrec@1 85.938 (86.950)\n",
      "[310/391]\tTime 0.208 (0.149)\tData 0.053 (0.039)\tLoss 0.4893 (0.4869)\tPrec@1 86.719 (86.935)\n",
      "[320/391]\tTime 0.155 (0.149)\tData 0.037 (0.039)\tLoss 0.4558 (0.4861)\tPrec@1 89.844 (86.974)\n",
      "[330/391]\tTime 0.171 (0.148)\tData 0.056 (0.039)\tLoss 0.6337 (0.4862)\tPrec@1 82.812 (86.967)\n",
      "[340/391]\tTime 0.136 (0.148)\tData 0.027 (0.039)\tLoss 0.7343 (0.4868)\tPrec@1 82.812 (86.943)\n",
      "[350/391]\tTime 0.215 (0.148)\tData 0.071 (0.039)\tLoss 0.5093 (0.4862)\tPrec@1 82.812 (86.939)\n",
      "[360/391]\tTime 0.155 (0.148)\tData 0.034 (0.039)\tLoss 0.4587 (0.4858)\tPrec@1 87.500 (86.935)\n",
      "[370/391]\tTime 0.127 (0.148)\tData 0.030 (0.039)\tLoss 0.5773 (0.4868)\tPrec@1 81.250 (86.902)\n",
      "[380/391]\tTime 0.159 (0.148)\tData 0.037 (0.039)\tLoss 0.6227 (0.4870)\tPrec@1 84.375 (86.903)\n",
      "[390/391]\tTime 0.121 (0.148)\tData 0.022 (0.039)\tLoss 0.4078 (0.4867)\tPrec@1 92.500 (86.918)\n",
      "Testing:\n",
      "Test: [0/79]\tTime 0.031 (0.031)\tLoss 1.0942 (1.0942)\tPrec@1 71.875 (71.875)\n",
      "Test: [10/79]\tTime 0.034 (0.035)\tLoss 1.1158 (1.0430)\tPrec@1 73.438 (76.349)\n",
      "Test: [20/79]\tTime 0.030 (0.039)\tLoss 0.9795 (1.0642)\tPrec@1 72.656 (75.558)\n",
      "Test: [30/79]\tTime 0.032 (0.038)\tLoss 0.9321 (1.0712)\tPrec@1 78.906 (75.832)\n",
      "Test: [40/79]\tTime 0.032 (0.039)\tLoss 1.4034 (1.0833)\tPrec@1 71.094 (75.686)\n",
      "Test: [50/79]\tTime 0.035 (0.040)\tLoss 1.2785 (1.0976)\tPrec@1 75.781 (75.827)\n",
      "Test: [60/79]\tTime 0.035 (0.040)\tLoss 1.0911 (1.0962)\tPrec@1 75.000 (75.781)\n",
      "Test: [70/79]\tTime 0.039 (0.039)\tLoss 0.8475 (1.0926)\tPrec@1 80.469 (75.792)\n",
      " * Prec@1 75.940\n",
      "\n",
      "===> epoch: 42/200\n",
      "current lr 1.00000e-01\n",
      "Training:\n",
      "[0/391]\tTime 0.168 (0.168)\tData 0.058 (0.058)\tLoss 0.5761 (0.5761)\tPrec@1 85.156 (85.156)\n",
      "[10/391]\tTime 0.160 (0.157)\tData 0.031 (0.039)\tLoss 0.5016 (0.4501)\tPrec@1 86.719 (88.494)\n",
      "[20/391]\tTime 0.180 (0.156)\tData 0.036 (0.038)\tLoss 0.4606 (0.4628)\tPrec@1 87.500 (87.946)\n",
      "[30/391]\tTime 0.165 (0.161)\tData 0.033 (0.039)\tLoss 0.3927 (0.4469)\tPrec@1 86.719 (88.231)\n",
      "[40/391]\tTime 0.146 (0.160)\tData 0.027 (0.040)\tLoss 0.4232 (0.4484)\tPrec@1 87.500 (88.014)\n",
      "[50/391]\tTime 0.133 (0.159)\tData 0.037 (0.039)\tLoss 0.5576 (0.4564)\tPrec@1 85.938 (87.730)\n",
      "[60/391]\tTime 0.142 (0.158)\tData 0.040 (0.039)\tLoss 0.3930 (0.4611)\tPrec@1 92.188 (87.641)\n",
      "[70/391]\tTime 0.109 (0.156)\tData 0.020 (0.039)\tLoss 0.4074 (0.4600)\tPrec@1 85.938 (87.566)\n",
      "[80/391]\tTime 0.176 (0.154)\tData 0.066 (0.039)\tLoss 0.5876 (0.4663)\tPrec@1 82.812 (87.355)\n",
      "[90/391]\tTime 0.127 (0.153)\tData 0.031 (0.039)\tLoss 0.5576 (0.4734)\tPrec@1 84.375 (87.182)\n",
      "[100/391]\tTime 0.119 (0.153)\tData 0.020 (0.039)\tLoss 0.5588 (0.4767)\tPrec@1 83.594 (87.013)\n",
      "[110/391]\tTime 0.159 (0.153)\tData 0.032 (0.038)\tLoss 0.6157 (0.4803)\tPrec@1 83.594 (86.881)\n",
      "[120/391]\tTime 0.152 (0.153)\tData 0.033 (0.039)\tLoss 0.4879 (0.4800)\tPrec@1 88.281 (86.861)\n",
      "[130/391]\tTime 0.172 (0.152)\tData 0.043 (0.038)\tLoss 0.5101 (0.4807)\tPrec@1 88.281 (86.826)\n",
      "[140/391]\tTime 0.152 (0.151)\tData 0.035 (0.038)\tLoss 0.5774 (0.4789)\tPrec@1 85.156 (86.913)\n",
      "[150/391]\tTime 0.139 (0.149)\tData 0.040 (0.038)\tLoss 0.4731 (0.4779)\tPrec@1 89.844 (86.941)\n",
      "[160/391]\tTime 0.147 (0.147)\tData 0.028 (0.038)\tLoss 0.4297 (0.4799)\tPrec@1 91.406 (86.874)\n",
      "[170/391]\tTime 0.132 (0.147)\tData 0.035 (0.037)\tLoss 0.5513 (0.4779)\tPrec@1 85.156 (86.933)\n",
      "[180/391]\tTime 0.162 (0.147)\tData 0.033 (0.038)\tLoss 0.5872 (0.4803)\tPrec@1 85.938 (86.878)\n",
      "[190/391]\tTime 0.113 (0.148)\tData 0.030 (0.038)\tLoss 0.5740 (0.4813)\tPrec@1 83.594 (86.870)\n",
      "[200/391]\tTime 0.157 (0.148)\tData 0.037 (0.038)\tLoss 0.5022 (0.4825)\tPrec@1 88.281 (86.870)\n",
      "[210/391]\tTime 0.175 (0.148)\tData 0.039 (0.038)\tLoss 0.3989 (0.4818)\tPrec@1 92.969 (86.941)\n",
      "[220/391]\tTime 0.133 (0.148)\tData 0.025 (0.038)\tLoss 0.4003 (0.4816)\tPrec@1 92.188 (86.949)\n",
      "[230/391]\tTime 0.147 (0.148)\tData 0.034 (0.038)\tLoss 0.5663 (0.4831)\tPrec@1 85.156 (86.945)\n",
      "[240/391]\tTime 0.124 (0.147)\tData 0.037 (0.037)\tLoss 0.5669 (0.4835)\tPrec@1 85.156 (86.916)\n",
      "[250/391]\tTime 0.135 (0.148)\tData 0.028 (0.038)\tLoss 0.5218 (0.4839)\tPrec@1 87.500 (86.940)\n",
      "[260/391]\tTime 0.158 (0.148)\tData 0.053 (0.038)\tLoss 0.3733 (0.4842)\tPrec@1 92.188 (86.937)\n",
      "[270/391]\tTime 0.157 (0.148)\tData 0.036 (0.038)\tLoss 0.4556 (0.4851)\tPrec@1 88.281 (86.886)\n",
      "[280/391]\tTime 0.166 (0.148)\tData 0.032 (0.038)\tLoss 0.5215 (0.4842)\tPrec@1 85.938 (86.938)\n",
      "[290/391]\tTime 0.142 (0.148)\tData 0.032 (0.038)\tLoss 0.5983 (0.4856)\tPrec@1 82.031 (86.904)\n",
      "[300/391]\tTime 0.162 (0.148)\tData 0.034 (0.038)\tLoss 0.5759 (0.4864)\tPrec@1 85.156 (86.869)\n",
      "[310/391]\tTime 0.150 (0.148)\tData 0.022 (0.038)\tLoss 0.6096 (0.4891)\tPrec@1 86.719 (86.792)\n",
      "[320/391]\tTime 0.123 (0.148)\tData 0.036 (0.038)\tLoss 0.4937 (0.4895)\tPrec@1 88.281 (86.780)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[330/391]\tTime 0.143 (0.149)\tData 0.030 (0.038)\tLoss 0.4928 (0.4903)\tPrec@1 88.281 (86.728)\n"
     ]
    }
   ],
   "source": [
    "from __future__ import print_function\n",
    "import os\n",
    "import time\n",
    "import logging\n",
    "import argparse\n",
    "import numpy as np\n",
    "from visdom import Visdom\n",
    "from PIL import Image\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torch.autograd import Variable\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision import datasets, transforms\n",
    "from utils import *\n",
    "from metric.loss import FitNet, AttentionTransfer, RKdAngle, RkdDistance\n",
    "\n",
    "# Teacher models:\n",
    "# VGG11/VGG13/VGG16/VGG19, GoogLeNet, AlxNet, ResNet18, ResNet34, \n",
    "# ResNet50, ResNet101, ResNet152, ResNeXt29_2x64d, ResNeXt29_4x64d, \n",
    "# ResNeXt29_8x64d, ResNeXt29_32x64d, PreActResNet18, PreActResNet34, \n",
    "# PreActhttps://www.bing.com/?mkt=zh-CNResNet50, PreActResNet101, PreActResNet152, \n",
    "# DenseNet121, DenseNet161, DenseNet169, DenseNet201, \n",
    "import models\n",
    "\n",
    "# Student models:\n",
    "# myNet, LeNet, FitNet\n",
    "\n",
    "start_time = time.time()\n",
    "# os.makedirs('./checkpoint', exist_ok=True)\n",
    "\n",
    "# Training settings\n",
    "parser = argparse.ArgumentParser(description='PyTorch ada. FitNet')\n",
    "\n",
    "parser.add_argument('--dataset',\n",
    "                    choices=['CIFAR10',\n",
    "                             'CIFAR100'\n",
    "                            ],\n",
    "                    default='CIFAR10')\n",
    "parser.add_argument('--teachers',\n",
    "                    choices=['ResNet32',\n",
    "                             'ResNet50',\n",
    "                             'ResNet56',\n",
    "                             'ResNet110',\n",
    "                             'DenseNet121'\n",
    "                            ],\n",
    "                    default=['ResNet32', 'ResNet56', 'ResNet110'],\n",
    "                    nargs='+')\n",
    "parser.add_argument('--student',\n",
    "                    choices=['ResNet8',\n",
    "                             'ResNet15',\n",
    "                             'ResNet16',\n",
    "                             'ResNet20',\n",
    "                             'myNet'\n",
    "                            ],\n",
    "                    default='ResNet20')\n",
    "\n",
    "parser.add_argument('--kd_ratio', default=0.7, type=float)\n",
    "parser.add_argument('--n_class', type=int, default=10, metavar='N', help='num of classes')\n",
    "parser.add_argument('--T', type=float, default=20.0, metavar='Temputure', help='Temputure for distillation')\n",
    "parser.add_argument('--batch_size', type=int, default=128, metavar='N', help='input batch size for training')\n",
    "parser.add_argument('--test_batch_size', type=int, default=128, metavar='N', help='input test batch size for training')\n",
    "parser.add_argument('--epochs', type=int, default=20, metavar='N', help='number of epochs to train (default: 20)')\n",
    "parser.add_argument('--lr', type=float, default=0.1, metavar='LR', help='learning rate (default: 0.01)')\n",
    "parser.add_argument('--momentum', type=float, default=0.9, metavar='M', help='SGD momentum (default: 0.5)')\n",
    "parser.add_argument('--device', default='cuda:0', type=str, help='device: cuda or cpu')\n",
    "parser.add_argument('--print_freq', type=int, default=10, metavar='N', help='how many batches to wait before logging training status')\n",
    "\n",
    "config = ['--epochs', '200', '--T', '5.0', '--device', 'cuda:0']\n",
    "args = parser.parse_args(config)\n",
    "\n",
    "device = args.device if torch.cuda.is_available() else 'cpu'\n",
    "load_dir = './checkpoint/' + args.dataset + '/'\n",
    "\n",
    "# teachers model\n",
    "teacher_models = []\n",
    "for te in args.teachers:\n",
    "    te_model = getattr(models, te)(num_classes=args.n_class)\n",
    "#     print(te_model)\n",
    "    te_model.load_state_dict(torch.load(load_dir + te_model.model_name + '.pth'))\n",
    "    te_model.to(device)\n",
    "    teacher_models.append(te_model)\n",
    "\n",
    "st_model = getattr(models, args.student)(num_classes=args.n_class)  # args.student()\n",
    "st_model.to(device)\n",
    "\n",
    "# logging\n",
    "logfile = load_dir + 'ada_fitnet_' + st_model.model_name + '.log'\n",
    "if os.path.exists(logfile):\n",
    "    os.remove(logfile)\n",
    "def log_out(info):\n",
    "    f = open(logfile, mode='a')\n",
    "    f.write(info)\n",
    "    f.write('\\n')\n",
    "    f.close()\n",
    "    print(info)\n",
    "    \n",
    "# visualizer\n",
    "vis = Visdom(env='distill')\n",
    "loss_win = vis.line(\n",
    "    X=np.array([0]),\n",
    "    Y=np.array([0]),\n",
    "    opts=dict(\n",
    "        title='FitNet ada. loss',\n",
    "        xtickmin=0,\n",
    "#         xtickmax=1,\n",
    "#         xtickstep=5,\n",
    "        ytickmin=0,\n",
    "#         ytickmax=1,\n",
    "        ytickstep=0.5,\n",
    "#         markers=True,\n",
    "#         markersymbol='dot',\n",
    "#         markersize=5,\n",
    "    ),\n",
    "    name=\"loss\"\n",
    ")\n",
    "\n",
    "acc_win = vis.line(\n",
    "    X=np.column_stack((0, 0)),\n",
    "    Y=np.column_stack((0, 0)),\n",
    "    opts=dict(\n",
    "        title='FitNet ada. ACC',\n",
    "        xtickmin=0,\n",
    "#         xtickstep=5,\n",
    "        ytickmin=0,\n",
    "        ytickmax=100,\n",
    "#         markers=True,\n",
    "#         markersymbol='dot',\n",
    "#         markersize=5,\n",
    "        legend=['train_acc', 'test_acc']\n",
    "    ),\n",
    "    name=\"acc\"\n",
    ")\n",
    "\n",
    "\n",
    "# adapter model\n",
    "class Adapter():\n",
    "    def __init__(self, in_models, pool_size):\n",
    "        # representations of teachers\n",
    "        pool_ch = pool_size[1]  # 64\n",
    "        pool_w = pool_size[2]   # 8\n",
    "        LR_list = []\n",
    "        torch.manual_seed(1)\n",
    "        self.theta = torch.randn(len(in_models), pool_ch).to(device)  # [3, 64]\n",
    "        self.theta.requires_grad_(True)\n",
    "   \n",
    "        self.max_feat = nn.MaxPool2d(kernel_size=(pool_w, pool_w), stride=pool_w).to(device)\n",
    "        self.W = torch.randn(pool_ch, 1).to(device)\n",
    "        self.W.requires_grad_(True)\n",
    "        self.val = False\n",
    "\n",
    "    def loss(self, y, labels, weighted_logits, T=10.0, alpha=0.7):\n",
    "        ls = nn.KLDivLoss()(F.log_softmax(y/T), weighted_logits) * (T*T * 2.0 * alpha) + F.cross_entropy(y, labels) * (1. - alpha)\n",
    "        if not self.val:\n",
    "            ls += 0.1 * (torch.sum(self.W * self.W) + torch.sum(torch.sum(self.theta * self.theta, dim=1), dim=0))\n",
    "        return ls\n",
    "        \n",
    "    def gradient(self, lr=0.01):\n",
    "        self.W.data = self.W.data - lr * self.W.grad.data\n",
    "        # Manually zero the gradients after updating weights\n",
    "        self.W.grad.data.zero_()\n",
    "        \n",
    "    def eval(self):\n",
    "        self.val = True\n",
    "        self.theta.detach()\n",
    "        self.W.detach()\n",
    "    \n",
    "    # input size: [64, 8, 8], [128, 3, 10]\n",
    "    def forward(self, conv_map, te_logits_list):\n",
    "        beta = self.max_feat(conv_map)\n",
    "        beta = torch.squeeze(beta)  # [128, 64]\n",
    "        \n",
    "        latent_factor = []\n",
    "        for t in self.theta:\n",
    "            latent_factor.append(beta * t)\n",
    "#         latent_factor = torch.stack(latent_factor, dim=0)  # [3, 128, 64]\n",
    "        alpha = []\n",
    "        for lf in latent_factor:  # lf.size:[128, 64]\n",
    "            alpha.append(lf.mm(self.W))\n",
    "        alpha = torch.stack(alpha, dim=0)  # [3, 128, 1]\n",
    "        alpha = torch.squeeze(alpha).transpose(0, 1) # [128, 3]\n",
    "        weight = F.softmax(alpha)  # [128, 3]\n",
    "\n",
    "        return weight\n",
    "\n",
    "# adapter instance\n",
    "_,_,_,pool_m,_ = st_model(torch.randn(1,3, 128, 128).to(device))  # get pool_size of student\n",
    "# reate adapter instance\n",
    "adapter = Adapter(teacher_models, pool_m.size())\n",
    "\n",
    "\n",
    "# data\n",
    "normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n",
    "train_transform = transforms.Compose([\n",
    "    transforms.RandomHorizontalFlip(),\n",
    "    transforms.RandomCrop(32, 4),\n",
    "    transforms.ToTensor(),\n",
    "    normalize,\n",
    "])\n",
    "test_transform = transforms.Compose([transforms.ToTensor(), normalize])\n",
    "train_set = getattr(datasets, args.dataset)(root='../data', train=True, download=True, transform=train_transform)\n",
    "test_set = getattr(datasets, args.dataset)(root='../data', train=False, download=False, transform=test_transform)\n",
    "train_loader = DataLoader(train_set, batch_size=args.batch_size, shuffle=True)\n",
    "test_loader = DataLoader(test_set, batch_size=args.test_batch_size, shuffle=False)\n",
    "# optim\n",
    "optimizer_W = optim.SGD([adapter.W], lr=args.lr, momentum=0.9)\n",
    "optimizer_theta = optim.SGD([adapter.theta], lr=args.lr, momentum=0.9)\n",
    "optimizer_sgd = optim.SGD(st_model.parameters(), lr=args.lr, momentum=0.9, weight_decay=5e-4)\n",
    "lr_scheduler = optim.lr_scheduler.MultiStepLR(optimizer_sgd, gamma=0.1, milestones=[100, 150])\n",
    "lr_scheduler2 = optim.lr_scheduler.MultiStepLR(optimizer_W, milestones=[40, 50])\n",
    "lr_scheduler3 = optim.lr_scheduler.MultiStepLR(optimizer_theta, milestones=[40, 50])\n",
    "\n",
    "# loss\n",
    "dist_criterion = RkdDistance().to(device)\n",
    "angle_criterion = RKdAngle().to(device)\n",
    "fitnet_criterion = [FitNet(32, 64), FitNet(64, 64),FitNet(64, 64)]\n",
    "[f.to(device) for f in fitnet_criterion]\n",
    "\n",
    "\n",
    "def train_adapter(n_epochs=70, model=st_model):\n",
    "    print('Training adapter:')\n",
    "    start_time = time.time()\n",
    "    model.train()\n",
    "    adapter.eval()\n",
    "    for ep in range(n_epochs):\n",
    "        lr_scheduler2.step()\n",
    "        lr_scheduler3.step()\n",
    "        for i, (input, target) in enumerate(train_loader):\n",
    "\n",
    "            input, target = input.to(device), target.to(device)\n",
    "            # compute outputs\n",
    "            b1, b2, b3, pool, output = model(input) # out_feat: 16, 32, 64, 64, - \n",
    "            st_maps = [b1, b2, b3, pool]\n",
    "#             print('b1:{}, b2:{}, b3{}, pool:{}'.format(b1.size(), b2.size(), b3.size(), pool.size()))\n",
    "# b1:torch.Size([128, 16, 32, 32]), b2:torch.Size([128, 32, 16, 16]), b3torch.Size([128, 64, 8, 8]), pool:torch.Size([128, 64, 1, 1])\n",
    "\n",
    "            te_scores_list = []\n",
    "            hint_maps = []\n",
    "            fit_loss = 0\n",
    "            for j,te in enumerate(teacher_models):\n",
    "                te.eval()\n",
    "                with torch.no_grad():\n",
    "                    t_b1, t_b2, t_b3, t_pool, t_output = te(input)\n",
    "#                 print('t_b1:{}, t_b2:{}, t_b3:{}, t_pool:{}'.format(t_b1.size(), t_b2.size(), t_b3.size(), t_pool.size()))\n",
    "# t_b1:torch.Size([128, 16, 32, 32]), t_b2:torch.Size([128, 32, 16, 16]), t_b3:torch.Size([128, 64, 8, 8]), t_pool:torch.Size([128, 64, 1, 1])\n",
    "                hint_maps.append(t_pool)\n",
    "                t_output = F.softmax(t_output/args.T)\n",
    "                te_scores_list.append(t_output)\n",
    "            te_scores_Tensor = torch.stack(te_scores_list, dim=1)  # size: [128, 3, 10]\n",
    "            \n",
    "            weight = adapter.forward(pool, te_scores_Tensor)\n",
    "            weight_t = torch.unsqueeze(weight, dim=2)\n",
    "            weighted_logits = weight_t * te_scores_Tensor  # [128, 3, 10]\n",
    "            weighted_logits = torch.sum(weighted_logits, dim=1)\n",
    "            weight_f = F.softmax(torch.mean(weight, dim=0))\n",
    "            \n",
    "            optimizer_sgd.zero_grad()\n",
    "            optimizer_W.zero_grad()\n",
    "            optimizer_theta.zero_grad()\n",
    "            \n",
    "            angle_loss = angle_criterion(output, weighted_logits)\n",
    "            dist_loss = dist_criterion(output, weighted_logits)\n",
    "            # compute gradient and do SGD step\n",
    "            ada_loss = adapter.loss(output, target, weighted_logits, T=args.T, alpha=args.kd_ratio)\n",
    "            \n",
    "            for j in range(len(teacher_models)-1):\n",
    "                fit_loss += fitnet_criterion[j](st_maps[j+1], hint_maps[j]) #weight_f[j] * \n",
    "#             fit_loss = fitnet_criterion[0](b2, hint_maps[0][3]) + fitnet_criterion[1](b3, hint_maps[1][3]) + fitnet_criterion(pool, hint_maps[2][3])\n",
    "            loss = ada_loss + fit_loss #+ dist_loss + angle_loss\n",
    "            \n",
    "            loss.backward(retain_graph=True)\n",
    "            optimizer_sgd.step()\n",
    "            optimizer_W.step()\n",
    "            optimizer_theta.step()\n",
    "            \n",
    "#          vis.line(np.array([loss.item()]), np.array([ep]), loss_win, update=\"append\")\n",
    "        log_out('epoch[{}/{}]adapter Loss: {:.4f}'.format(ep, n_epochs, loss.item()))\n",
    "    end_time = time.time()\n",
    "    log_out(\"--- adapter training cost {:.3f} mins ---\".format((end_time - start_time)/60))\n",
    "\n",
    "\n",
    "# train with multi-teacher\n",
    "def train(epoch, model):\n",
    "    print('Training:')\n",
    "    # switch to train mode\n",
    "    model.train()\n",
    "    adapter.eval()\n",
    "    batch_time = AverageMeter()\n",
    "    data_time = AverageMeter()\n",
    "    losses = AverageMeter()\n",
    "    top1 = AverageMeter()\n",
    "    \n",
    "    end = time.time()\n",
    "    for i, (input, target) in enumerate(train_loader):\n",
    "\n",
    "        # measure data loading time\n",
    "        data_time.update(time.time() - end)\n",
    "\n",
    "        input, target = input.to(device), target.to(device)\n",
    "        \n",
    "        # compute outputs\n",
    "        b1, b2, b3, pool, output = model(input)\n",
    "        st_maps = [b1, b2, b3, pool]\n",
    "        \n",
    "        te_scores_list = []\n",
    "        hint_maps = []\n",
    "        fit_loss = 0\n",
    "        for j,te in enumerate(teacher_models):\n",
    "            te.eval()\n",
    "            with torch.no_grad():\n",
    "                t_b1, t_b2, t_b3, t_pool, t_output = te(input)\n",
    "        \n",
    "            hint_maps.append(t_pool)\n",
    "            t_output = F.softmax(t_output/args.T)\n",
    "            te_scores_list.append(t_output)\n",
    "        te_scores_Tensor = torch.stack(te_scores_list, dim=1)  # size: [128, 3, 10]\n",
    "        \n",
    "        weight = adapter.forward(pool, te_scores_Tensor)\n",
    "        weight_t = torch.unsqueeze(weight, dim=2)\n",
    "        weighted_logits = weight_t * te_scores_Tensor  # [128, 3, 10]\n",
    "        weighted_logits = torch.sum(weighted_logits, dim=1)\n",
    "        weight_f = F.softmax(torch.mean(weight, dim=0))\n",
    "        \n",
    "        optimizer_sgd.zero_grad()\n",
    "        \n",
    "        angle_loss = angle_criterion(output, weighted_logits)\n",
    "        dist_loss = dist_criterion(output, weighted_logits)\n",
    "        \n",
    "        # compute gradient and do SGD step\n",
    "        ada_loss = adapter.loss(output, target, weighted_logits, T=args.T, alpha=args.kd_ratio)\n",
    "        for j in range(len(teacher_models)-1):\n",
    "            fit_loss += fitnet_criterion[j](st_maps[j+1], hint_maps[j])\n",
    "        \n",
    "        loss = ada_loss + fit_loss #+ dist_loss + angle_loss\n",
    "\n",
    "        loss.backward(retain_graph=True)\n",
    "        optimizer_sgd.step()\n",
    "\n",
    "        output = output.float()\n",
    "        loss = loss.float()\n",
    "        # measure accuracy and record loss\n",
    "        train_acc = accuracy(output.data, target.data)[0]\n",
    "        losses.update(loss.item(), input.size(0))\n",
    "        top1.update(train_acc, input.size(0))\n",
    "\n",
    "        # measure elapsed time\n",
    "        batch_time.update(time.time() - end)\n",
    "        end = time.time()\n",
    "\n",
    "        if i % args.print_freq == 0:\n",
    "            log_out('[{0}/{1}]\\t'\n",
    "                  'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\\t'\n",
    "                  'Data {data_time.val:.3f} ({data_time.avg:.3f})\\t'\n",
    "                  'Loss {loss.val:.4f} ({loss.avg:.4f})\\t'\n",
    "                  'Prec@1 {top1.val:.3f} ({top1.avg:.3f})'.format(\n",
    "                      i, len(train_loader), batch_time=batch_time,\n",
    "                      data_time=data_time, loss=losses, top1=top1))\n",
    "    return losses.avg, train_acc.cpu().numpy()\n",
    "\n",
    "\n",
    "def test(model):\n",
    "    print('Testing:')\n",
    "    # switch to evaluate mode\n",
    "    model.eval()\n",
    "    batch_time = AverageMeter()\n",
    "    losses = AverageMeter()\n",
    "    top1 = AverageMeter()\n",
    "\n",
    "    end = time.time()\n",
    "    with torch.no_grad():\n",
    "        for i, (input, target) in enumerate(test_loader):\n",
    "            input, target = input.to(device), target.to(device)\n",
    "\n",
    "            # compute output\n",
    "            _,_,_,_,output = model(input)\n",
    "            loss = F.cross_entropy(output, target)\n",
    "\n",
    "            output = output.float()\n",
    "            loss = loss.float()\n",
    "\n",
    "            # measure accuracy and record loss\n",
    "            test_acc = accuracy(output.data, target.data)[0]\n",
    "            losses.update(loss.item(), input.size(0))\n",
    "            top1.update(test_acc, input.size(0))\n",
    "\n",
    "            # measure elapsed time\n",
    "            batch_time.update(time.time() - end)\n",
    "            end = time.time()\n",
    "\n",
    "            if i % args.print_freq == 0:\n",
    "                log_out('Test: [{0}/{1}]\\t'\n",
    "                      'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\\t'\n",
    "                      'Loss {loss.val:.4f} ({loss.avg:.4f})\\t'\n",
    "                      'Prec@1 {top1.val:.3f} ({top1.avg:.3f})'.format(\n",
    "                          i, len(test_loader), batch_time=batch_time, loss=losses,\n",
    "                          top1=top1))\n",
    "\n",
    "    log_out(' * Prec@1 {top1.avg:.3f}'.format(top1=top1))\n",
    "\n",
    "    return losses.avg, test_acc.cpu().numpy(), top1.avg.cpu().numpy()\n",
    "\n",
    "# \"\"\"\n",
    "print('StudentNet:\\n')\n",
    "print(st_model)\n",
    "st_model.apply(weights_init_normal)\n",
    "train_adapter(n_epochs=80)\n",
    "# st_model.apply(weights_init_normal)\n",
    "best_acc = 0\n",
    "for epoch in range(1, args.epochs + 1):\n",
    "    log_out(\"\\n===> epoch: {}/{}\".format(epoch, args.epochs))\n",
    "    log_out('current lr {:.5e}'.format(optimizer_sgd.param_groups[0]['lr']))\n",
    "    lr_scheduler.step(epoch)\n",
    "    train_loss, train_acc = train(epoch, st_model)\n",
    "    # visaulize loss\n",
    "    vis.line(np.array([train_loss]), np.array([epoch]), loss_win, update=\"append\")\n",
    "    _, test_acc, top1 = test(st_model)\n",
    "    vis.line(np.column_stack((train_acc, top1)), np.column_stack((epoch, epoch)), acc_win, update=\"append\")\n",
    "    if top1 > best_acc:\n",
    "        best_acc = top1\n",
    "            \n",
    "# release GPU memory\n",
    "torch.cuda.empty_cache()\n",
    "log_out(\"BEST ACC: {:.3f}\".format(best_acc))\n",
    "log_out(\"--- {:.3f} mins ---\".format((time.time() - start_time)/60))\n",
    "# \"\"\""
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}