baseline_net_train.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Almost Visual Inertial Odometry"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "from collections import OrderedDict\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torchvision.models import *\n",
    "from torchvision.models.resnet import BasicBlock\n",
    "from torch.utils.data import DataLoader\n",
    "from tqdm import tqdm\n",
    "from  matplotlib import pyplot as plt\n",
    "from matplotlib.image import imread\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Constants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 20197\n",
    "BATCH_SIZE = 32\n",
    "TEST_BATCH_SIZE = 32\n",
    "GAMMA = 0.1   # torch default\n",
    "LR=0.001 \n",
    "EPOCHS = 210\n",
    "EARLY_STOPPING = 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.manual_seed(SEED)\n",
    "np.random.seed(SEED)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Available cuda: True\n"
     ]
    }
   ],
   "source": [
    "available_cuda = torch.cuda.is_available()\n",
    "print(f\"Available cuda: {available_cuda}\")\n",
    "\n",
    "device = torch.device(\"cuda\" if available_cuda else \"cpu\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "class AdvioDataset(torch.utils.data.Dataset):\n",
    "      def __init__(self, frames, inertials, labels, inertials_buffer, choose_split=None, split={'train':0.6, 'val':0.2, 'test':0.2}, shuffle=True):\n",
    "            if shuffle:\n",
    "                  shuffled_index = torch.randperm(frames.shape[0])\n",
    "                  frames = frames[shuffled_index]\n",
    "                  inertials = inertials[shuffled_index]\n",
    "                  labels = labels[shuffled_index]\n",
    "                  inertials_buffer = inertials_buffer[shuffled_index]\n",
    "            \n",
    "            if choose_split is None:\n",
    "                  self.frames = frames\n",
    "                  self.inertials = inertials\n",
    "                  self.labels = labels\n",
    "                  self.inertials_buffer = inertials_buffer\n",
    "            else:\n",
    "                  assert split['train'] + split['val'] + split['test'] == 1\n",
    "                  length = frames.shape[0]\n",
    "                  if choose_split == \"train\":\n",
    "                        self.frames = frames[:round(split['train']*length)]\n",
    "                        self.inertials = inertials[:round(split['train']*length)]\n",
    "                        self.labels = labels[:round(split['train']*length)]\n",
    "                        self.inertials_buffer = inertials_buffer[:round(split['train']*length)]\n",
    "                  elif choose_split == \"val\":\n",
    "                        self.frames = frames[round(split['train']*length):-round(split['test']*length)]\n",
    "                        self.inertials = inertials[round(split['train']*length):-round(split['test']*length)]\n",
    "                        self.labels = labels[round(split['train']*length):-round(split['test']*length)]\n",
    "                        self.inertials_buffer = inertials_buffer[round(split['train']*length):-round(split['test']*length)]\n",
    "                  elif choose_split == \"test\":\n",
    "                        self.frames = frames[-round(split['test']*length):]\n",
    "                        self.inertials = inertials[-round(split['test']*length):]\n",
    "                        self.labels = labels[-round(split['test']*length):]\n",
    "                        self.inertials_buffer = inertials_buffer[-round(split['test']*length):]\n",
    "                  else:\n",
    "                        raise Exception(f\"The split name '{choose_split}' doesn't exists\")\n",
    "\n",
    "      def __len__(self):\n",
    "            return len(self.frames) - 1\n",
    "\n",
    "      def __getitem__(self, index):\n",
    "            # load sample of frames\n",
    "            frame_name = self.frames[index]\n",
    "            scene_name, frame_name = frame_name.split(\"_\")\n",
    "            frame_name, file_extension = frame_name.split(\".\")\n",
    "            sample_frame = torch.Tensor(np.load(f\"./data/{scene_name}/iphone/frames/{scene_name}_{frame_name}.npy\", allow_pickle=True))\n",
    "\n",
    "            # build buffer sample of inertials\n",
    "            sample_inertials_buffer = self.inertials_buffer[index]\n",
    "\n",
    "            # load labels\n",
    "            label_odometry = self.labels[index]\n",
    "\n",
    "            # load inertials\n",
    "            label_inertial = self.inertials[index]\n",
    "\n",
    "            return sample_frame, sample_inertials_buffer, label_odometry, label_inertial\n",
    "\n",
    "\n",
    "frames = np.load(\"dataset_frames.npy\", allow_pickle=True)\n",
    "inertials = torch.Tensor(np.load(\"dataset_inertials.npy\", allow_pickle=True))\n",
    "labels = torch.Tensor(np.load(\"dataset_labels.npy\", allow_pickle=True))\n",
    "buffer_inertials = torch.Tensor(np.load(\"dataset_buffer.npy\", allow_pickle=True))\n",
    "\n",
    "train_data = AdvioDataset(frames, inertials, labels, buffer_inertials, \"train\")\n",
    "val_data = AdvioDataset(frames, inertials, labels, buffer_inertials, \"val\")\n",
    "test_data = AdvioDataset(frames, inertials, labels, buffer_inertials, \"test\")\n",
    "\n",
    "del frames\n",
    "del inertials\n",
    "del labels\n",
    "del buffer_inertials\n",
    "\n",
    "params = {'batch_size': BATCH_SIZE,\n",
    "          'num_workers': 6,\n",
    "          'drop_last':True}\n",
    "\n",
    "train_loader = DataLoader(train_data, **params, shuffle=True)\n",
    "val_loader = DataLoader(val_data, **params, shuffle=True)\n",
    "test_loader = DataLoader(test_data, **params)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "def replace_head(model):\n",
    "    n_inputs = model.classifier[0].in_features\n",
    "    classifier = nn.Linear(n_inputs, 100)\n",
    "    model.classifier[0] = classifier\n",
    "\n",
    "    n_inputs = model.classifier[3].in_features\n",
    "    classifier = nn.Linear(100, 3)\n",
    "    model.classifier[3] = classifier\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BaselineNet2(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(BaselineNet2, self).__init__()\n",
    "        \n",
    "        self.conv2d_odometry = mobilenet_v3_small(weights=MobileNet_V3_Small_Weights.IMAGENET1K_V1)\n",
    "        self.conv2d_odometry = replace_head(self.conv2d_odometry)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.conv2d_odometry(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BaselineNet(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(BaselineNet, self).__init__()\n",
    "        \n",
    "        self.conv2d_odometry = MobileNetV2(num_classes=3)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.conv2d_odometry(x)\n",
    "        return x"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "def normalize(data):\n",
    "  data = data - data.min()\n",
    "  data = data / data.max()\n",
    "  return data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(model, device, train_loader, optimizer, epoch):\n",
    "    model.train()\n",
    "    history_loss = 0\n",
    "\n",
    "    for batch_idx, (data, _, labels_odometry, _) in enumerate(train_loader):\n",
    "        # data preparation\n",
    "        data = normalize(data)\n",
    "        labels_odometry = labels_odometry.mean(axis=1)\n",
    "        data, labels_odometry = data.to(device), labels_odometry.to(device)\n",
    "\n",
    "        #forward\n",
    "        optimizer.zero_grad()\n",
    "        output = model(data)\n",
    "\n",
    "        # loss\n",
    "        loss = torch.sqrt(F.mse_loss(output, labels_odometry))\n",
    "\n",
    "        # backward\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        #stats\n",
    "        history_loss += loss.item()\n",
    "\n",
    "        if batch_idx%(len(train_loader)//5)==0:\n",
    "            print(f\"\\t[# {batch_idx: 4}] train_loss: {loss.item():.6f}\")\n",
    "\n",
    "    history_loss /= len(train_loader) # average epoch loss\n",
    "    print(f\"\\ttrain_loss: {history_loss:.6f}\")\n",
    "\n",
    "    return history_loss\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "def validation(model, device, val_loader, epoch):\n",
    "    model.train()\n",
    "    history_loss=0\n",
    "\n",
    "    for batch_idx, (data, _, labels_odometry, _) in enumerate(val_loader):\n",
    "        # data preparation\n",
    "        data = normalize(data)\n",
    "        labels_odometry = labels_odometry.mean(axis=1)\n",
    "        data, labels_odometry = data.to(device), labels_odometry.to(device)\n",
    "\n",
    "        # forward\n",
    "        output = model(data)\n",
    "\n",
    "        # loss\n",
    "        loss = torch.sqrt(F.mse_loss(output, labels_odometry))\n",
    "\n",
    "        #stats\n",
    "        history_loss += loss.item()\n",
    "\n",
    "    history_loss /= len(val_loader) # average epoch loss\n",
    "    print(f\"\\tval_loss: {history_loss:.6f}\")\n",
    "\n",
    "    return history_loss\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/210\n",
      "\t[#    0] train_loss: 17.918837\n",
      "\t[#  180] train_loss: 11.736006\n",
      "\t[#  360] train_loss: 10.976809\n",
      "\t[#  540] train_loss: 6.604259\n",
      "\t[#  720] train_loss: 9.857423\n",
      "\t[#  900] train_loss: 8.324232\n",
      "\ttrain_loss: 11.029631\n",
      "\tval_loss: 7.096916\n",
      "\t*** Personal Best ***\n",
      "Epoch 2/210\n",
      "\t[#    0] train_loss: 8.886558\n",
      "\t[#  180] train_loss: 7.993495\n",
      "\t[#  360] train_loss: 5.226495\n",
      "\t[#  540] train_loss: 13.042361\n",
      "\t[#  720] train_loss: 6.399005\n",
      "\t[#  900] train_loss: 4.881711\n",
      "\ttrain_loss: 6.211379\n",
      "\tval_loss: 4.818977\n",
      "\t*** Personal Best ***\n",
      "Epoch 3/210\n",
      "\t[#    0] train_loss: 4.510308\n",
      "\t[#  180] train_loss: 4.349883\n",
      "\t[#  360] train_loss: 3.946164\n",
      "\t[#  540] train_loss: 3.972699\n",
      "\t[#  720] train_loss: 4.854510\n",
      "\t[#  900] train_loss: 4.204525\n",
      "\ttrain_loss: 4.392783\n",
      "\tval_loss: 4.337262\n",
      "\t*** Personal Best ***\n",
      "Epoch 4/210\n",
      "\t[#    0] train_loss: 3.894087\n",
      "\t[#  180] train_loss: 5.166529\n",
      "\t[#  360] train_loss: 4.684035\n",
      "\t[#  540] train_loss: 5.451829\n",
      "\t[#  720] train_loss: 3.817747\n",
      "\t[#  900] train_loss: 3.438318\n",
      "\ttrain_loss: 4.117295\n",
      "\tval_loss: 4.580567\n",
      "Epoch 5/210\n",
      "\t[#    0] train_loss: 4.969490\n",
      "\t[#  180] train_loss: 3.495045\n",
      "\t[#  360] train_loss: 2.917242\n",
      "\t[#  540] train_loss: 7.938910\n",
      "\t[#  720] train_loss: 3.398004\n",
      "\t[#  900] train_loss: 10.271050\n",
      "\ttrain_loss: 3.295819\n",
      "\tval_loss: 4.294551\n",
      "\t*** Personal Best ***\n",
      "Epoch 6/210\n",
      "\t[#    0] train_loss: 2.676186\n",
      "\t[#  180] train_loss: 3.417066\n",
      "\t[#  360] train_loss: 2.455113\n",
      "\t[#  540] train_loss: 4.130322\n",
      "\t[#  720] train_loss: 4.905964\n",
      "\t[#  900] train_loss: 2.600872\n",
      "\ttrain_loss: 3.644827\n",
      "\tval_loss: 3.083912\n",
      "\t*** Personal Best ***\n",
      "Epoch 7/210\n",
      "\t[#    0] train_loss: 3.143243\n",
      "\t[#  180] train_loss: 2.758914\n",
      "\t[#  360] train_loss: 2.973103\n",
      "\t[#  540] train_loss: 2.739216\n",
      "\t[#  720] train_loss: 2.686304\n",
      "\t[#  900] train_loss: 2.815245\n",
      "\ttrain_loss: 3.016169\n",
      "\tval_loss: 3.491713\n",
      "Epoch 8/210\n",
      "\t[#    0] train_loss: 2.980322\n",
      "\t[#  180] train_loss: 2.017887\n",
      "\t[#  360] train_loss: 2.424605\n",
      "\t[#  540] train_loss: 3.090542\n",
      "\t[#  720] train_loss: 2.755564\n",
      "\t[#  900] train_loss: 2.936854\n",
      "\ttrain_loss: 3.024763\n",
      "\tval_loss: 3.872621\n",
      "Epoch 9/210\n",
      "\t[#    0] train_loss: 2.490080\n",
      "\t[#  180] train_loss: 3.086742\n",
      "\t[#  360] train_loss: 3.910458\n",
      "\t[#  540] train_loss: 2.712559\n",
      "\t[#  720] train_loss: 4.102284\n",
      "\t[#  900] train_loss: 2.936087\n",
      "\ttrain_loss: 3.089929\n",
      "\tval_loss: 2.953919\n",
      "\t*** Personal Best ***\n",
      "Epoch 10/210\n",
      "\t[#    0] train_loss: 4.356492\n",
      "\t[#  180] train_loss: 2.923419\n",
      "\t[#  360] train_loss: 3.147014\n",
      "\t[#  540] train_loss: 3.516625\n",
      "\t[#  720] train_loss: 2.529983\n",
      "\t[#  900] train_loss: 2.257757\n",
      "\ttrain_loss: 2.961945\n",
      "\tval_loss: 2.636838\n",
      "\t*** Personal Best ***\n",
      "Epoch 11/210\n",
      "\t[#    0] train_loss: 2.529938\n",
      "\t[#  180] train_loss: 2.727136\n",
      "\t[#  360] train_loss: 1.949618\n",
      "\t[#  540] train_loss: 2.428995\n",
      "\t[#  720] train_loss: 2.549992\n",
      "\t[#  900] train_loss: 2.351092\n",
      "\ttrain_loss: 2.550133\n",
      "\tval_loss: 2.461853\n",
      "\t*** Personal Best ***\n",
      "Epoch 12/210\n",
      "\t[#    0] train_loss: 2.575003\n",
      "\t[#  180] train_loss: 2.926389\n",
      "\t[#  360] train_loss: 2.826078\n",
      "\t[#  540] train_loss: 2.784590\n",
      "\t[#  720] train_loss: 3.068324\n",
      "\t[#  900] train_loss: 2.520106\n",
      "\ttrain_loss: 2.523700\n",
      "\tval_loss: 3.160752\n",
      "Epoch 13/210\n",
      "\t[#    0] train_loss: 2.461168\n",
      "\t[#  180] train_loss: 3.281272\n",
      "\t[#  360] train_loss: 2.528756\n",
      "\t[#  540] train_loss: 4.793160\n",
      "\t[#  720] train_loss: 2.481047\n",
      "\t[#  900] train_loss: 2.697181\n",
      "\ttrain_loss: 3.129452\n",
      "\tval_loss: 2.767442\n",
      "Epoch 14/210\n",
      "\t[#    0] train_loss: 2.449944\n",
      "\t[#  180] train_loss: 2.291345\n",
      "\t[#  360] train_loss: 1.662775\n",
      "\t[#  540] train_loss: 1.635100\n",
      "\t[#  720] train_loss: 2.434587\n",
      "\t[#  900] train_loss: 2.270706\n",
      "\ttrain_loss: 2.589254\n",
      "\tval_loss: 2.514390\n",
      "Epoch 15/210\n",
      "\t[#    0] train_loss: 3.225894\n",
      "\t[#  180] train_loss: 2.340923\n",
      "\t[#  360] train_loss: 2.104408\n",
      "\t[#  540] train_loss: 2.353979\n",
      "\t[#  720] train_loss: 2.597520\n",
      "\t[#  900] train_loss: 1.859202\n",
      "\ttrain_loss: 2.481501\n",
      "\tval_loss: 2.394564\n",
      "\t*** Personal Best ***\n",
      "Epoch 16/210\n",
      "\t[#    0] train_loss: 2.176139\n",
      "\t[#  180] train_loss: 2.133820\n",
      "\t[#  360] train_loss: 2.030669\n",
      "\t[#  540] train_loss: 2.021345\n",
      "\t[#  720] train_loss: 2.465870\n",
      "\t[#  900] train_loss: 1.262237\n",
      "\ttrain_loss: 2.294004\n",
      "\tval_loss: 2.382744\n",
      "\t*** Personal Best ***\n",
      "Epoch 17/210\n",
      "\t[#    0] train_loss: 2.427579\n",
      "\t[#  180] train_loss: 2.515792\n",
      "\t[#  360] train_loss: 1.754889\n",
      "\t[#  540] train_loss: 2.379669\n",
      "\t[#  720] train_loss: 3.407969\n",
      "\t[#  900] train_loss: 9.733036\n",
      "\ttrain_loss: 2.601041\n",
      "\tval_loss: 3.718605\n",
      "Epoch 18/210\n",
      "\t[#    0] train_loss: 3.688837\n",
      "\t[#  180] train_loss: 2.361396\n",
      "\t[#  360] train_loss: 2.369853\n",
      "\t[#  540] train_loss: 2.981173\n",
      "\t[#  720] train_loss: 2.796386\n",
      "\t[#  900] train_loss: 2.561032\n",
      "\ttrain_loss: 2.821295\n",
      "\tval_loss: 2.561028\n",
      "Epoch 19/210\n",
      "\t[#    0] train_loss: 1.825998\n",
      "\t[#  180] train_loss: 2.977762\n",
      "\t[#  360] train_loss: 1.553585\n",
      "\t[#  540] train_loss: 2.022710\n",
      "\t[#  720] train_loss: 2.412845\n",
      "\t[#  900] train_loss: 2.171328\n",
      "\ttrain_loss: 2.331069\n",
      "\tval_loss: 2.286309\n",
      "\t*** Personal Best ***\n",
      "Epoch 20/210\n",
      "\t[#    0] train_loss: 2.336013\n",
      "\t[#  180] train_loss: 2.259724\n",
      "\t[#  360] train_loss: 2.154289\n",
      "\t[#  540] train_loss: 1.920299\n",
      "\t[#  720] train_loss: 2.506419\n",
      "\t[#  900] train_loss: 2.153826\n",
      "\ttrain_loss: 2.198567\n",
      "\tval_loss: 2.203015\n",
      "\t*** Personal Best ***\n",
      "Epoch 21/210\n",
      "\t[#    0] train_loss: 3.055532\n",
      "\t[#  180] train_loss: 1.594379\n",
      "\t[#  360] train_loss: 2.173181\n",
      "\t[#  540] train_loss: 2.499850\n",
      "\t[#  720] train_loss: 2.505737\n",
      "\t[#  900] train_loss: 1.612265\n",
      "\ttrain_loss: 2.630623\n",
      "\tval_loss: 2.345092\n",
      "Epoch 22/210\n",
      "\t[#    0] train_loss: 2.045248\n",
      "\t[#  180] train_loss: 2.296262\n",
      "\t[#  360] train_loss: 2.001731\n",
      "\t[#  540] train_loss: 6.030599\n",
      "\t[#  720] train_loss: 1.436195\n",
      "\t[#  900] train_loss: 2.642736\n",
      "\ttrain_loss: 2.422430\n",
      "\tval_loss: 2.329035\n",
      "Epoch 23/210\n",
      "\t[#    0] train_loss: 2.243934\n",
      "\t[#  180] train_loss: 2.664860\n",
      "\t[#  360] train_loss: 2.536849\n",
      "\t[#  540] train_loss: 2.267160\n",
      "\t[#  720] train_loss: 2.311478\n",
      "\t[#  900] train_loss: 2.055614\n",
      "\ttrain_loss: 2.316194\n",
      "\tval_loss: 2.294174\n",
      "Epoch 24/210\n",
      "\t[#    0] train_loss: 1.556529\n",
      "\t[#  180] train_loss: 2.274250\n",
      "\t[#  360] train_loss: 2.721782\n",
      "\t[#  540] train_loss: 1.816142\n",
      "\t[#  720] train_loss: 2.283037\n",
      "\t[#  900] train_loss: 1.807244\n",
      "\ttrain_loss: 2.152014\n",
      "\tval_loss: 2.149221\n",
      "\t*** Personal Best ***\n",
      "Epoch 25/210\n",
      "\t[#    0] train_loss: 2.542471\n",
      "\t[#  180] train_loss: 2.870723\n",
      "\t[#  360] train_loss: 1.544933\n",
      "\t[#  540] train_loss: 2.030470\n",
      "\t[#  720] train_loss: 2.451718\n",
      "\t[#  900] train_loss: 7.597232\n",
      "\ttrain_loss: 2.282184\n",
      "\tval_loss: 3.369845\n",
      "Epoch 26/210\n",
      "\t[#    0] train_loss: 2.118983\n",
      "\t[#  180] train_loss: 2.473483\n",
      "\t[#  360] train_loss: 2.110152\n",
      "\t[#  540] train_loss: 2.010916\n",
      "\t[#  720] train_loss: 2.641855\n",
      "\t[#  900] train_loss: 1.965598\n",
      "\ttrain_loss: 2.734427\n",
      "\tval_loss: 2.212069\n",
      "Epoch 27/210\n",
      "\t[#    0] train_loss: 2.092973\n",
      "\t[#  180] train_loss: 2.308636\n",
      "\t[#  360] train_loss: 2.656647\n",
      "\t[#  540] train_loss: 2.492698\n",
      "\t[#  720] train_loss: 2.764632\n",
      "\t[#  900] train_loss: 1.655694\n",
      "\ttrain_loss: 2.214156\n",
      "\tval_loss: 2.204552\n",
      "Epoch 28/210\n",
      "\t[#    0] train_loss: 2.063239\n",
      "\t[#  180] train_loss: 2.337773\n",
      "\t[#  360] train_loss: 2.033004\n",
      "\t[#  540] train_loss: 2.240756\n",
      "\t[#  720] train_loss: 2.061701\n",
      "\t[#  900] train_loss: 1.793094\n",
      "\ttrain_loss: 2.128143\n",
      "\tval_loss: 2.103290\n",
      "\t*** Personal Best ***\n",
      "Epoch 29/210\n",
      "\t[#    0] train_loss: 2.103392\n",
      "\t[#  180] train_loss: 2.075055\n",
      "\t[#  360] train_loss: 2.017971\n",
      "\t[#  540] train_loss: 1.690510\n",
      "\t[#  720] train_loss: 1.891149\n",
      "\t[#  900] train_loss: 2.597524\n",
      "\ttrain_loss: 2.107622\n",
      "\tval_loss: 2.082229\n",
      "\t*** Personal Best ***\n",
      "Epoch 30/210\n",
      "\t[#    0] train_loss: 2.421346\n",
      "\t[#  180] train_loss: 2.042867\n",
      "\t[#  360] train_loss: 1.807122\n",
      "\t[#  540] train_loss: 2.110179\n",
      "\t[#  720] train_loss: 4.976038\n",
      "\t[#  900] train_loss: 1.993923\n",
      "\ttrain_loss: 2.678159\n",
      "\tval_loss: 2.549631\n",
      "Epoch 31/210\n",
      "\t[#    0] train_loss: 2.451835\n",
      "\t[#  180] train_loss: 2.150352\n",
      "\t[#  360] train_loss: 2.276881\n",
      "\t[#  540] train_loss: 2.889389\n",
      "\t[#  720] train_loss: 2.251512\n",
      "\t[#  900] train_loss: 1.826003\n",
      "\ttrain_loss: 2.331616\n",
      "\tval_loss: 2.231311\n",
      "Epoch 32/210\n",
      "\t[#    0] train_loss: 2.125463\n",
      "\t[#  180] train_loss: 2.382970\n",
      "\t[#  360] train_loss: 1.982728\n",
      "\t[#  540] train_loss: 3.814385\n",
      "\t[#  720] train_loss: 2.410689\n",
      "\t[#  900] train_loss: 2.451536\n",
      "\ttrain_loss: 2.152839\n",
      "\tval_loss: 2.094632\n",
      "Epoch 33/210\n",
      "\t[#    0] train_loss: 1.743998\n",
      "\t[#  180] train_loss: 2.178468\n",
      "\t[#  360] train_loss: 2.181806\n",
      "\t[#  540] train_loss: 2.293178\n",
      "\t[#  720] train_loss: 1.844859\n",
      "\t[#  900] train_loss: 1.990313\n",
      "\ttrain_loss: 2.147749\n",
      "\tval_loss: 2.143296\n",
      "Epoch 34/210\n",
      "\t[#    0] train_loss: 2.404779\n",
      "\t[#  180] train_loss: 2.444100\n",
      "\t[#  360] train_loss: 1.970105\n",
      "\t[#  540] train_loss: 1.782790\n",
      "\t[#  720] train_loss: 1.837436\n",
      "\t[#  900] train_loss: 2.899019\n",
      "\ttrain_loss: 2.332060\n",
      "\tval_loss: 2.485352\n",
      "Epoch 35/210\n",
      "\t[#    0] train_loss: 2.838941\n",
      "\t[#  180] train_loss: 3.215563\n",
      "\t[#  360] train_loss: 2.605794\n",
      "\t[#  540] train_loss: 1.821440\n",
      "\t[#  720] train_loss: 4.631138\n",
      "\t[#  900] train_loss: 2.825739\n",
      "\ttrain_loss: 2.498343\n",
      "\tval_loss: 2.169215\n",
      "Epoch 36/210\n",
      "\t[#    0] train_loss: 1.687096\n",
      "\t[#  180] train_loss: 2.102468\n",
      "\t[#  360] train_loss: 2.131423\n",
      "\t[#  540] train_loss: 2.231681\n",
      "\t[#  720] train_loss: 2.132399\n",
      "\t[#  900] train_loss: 2.167326\n",
      "\ttrain_loss: 2.160887\n",
      "\tval_loss: 2.102989\n",
      "Epoch 37/210\n",
      "\t[#    0] train_loss: 2.525672\n",
      "\t[#  180] train_loss: 1.741071\n",
      "\t[#  360] train_loss: 1.951021\n",
      "\t[#  540] train_loss: 1.999202\n",
      "\t[#  720] train_loss: 1.586510\n",
      "\t[#  900] train_loss: 1.736132\n",
      "\ttrain_loss: 2.091332\n",
      "\tval_loss: 2.059792\n",
      "\t*** Personal Best ***\n",
      "Epoch 38/210\n",
      "\t[#    0] train_loss: 2.471259\n",
      "\t[#  180] train_loss: 2.524779\n",
      "\t[#  360] train_loss: 1.885141\n",
      "\t[#  540] train_loss: 2.011787\n",
      "\t[#  720] train_loss: 1.620318\n",
      "\t[#  900] train_loss: 2.265742\n",
      "\ttrain_loss: 2.089967\n",
      "\tval_loss: 2.356191\n",
      "Epoch 39/210\n",
      "\t[#    0] train_loss: 2.190802\n",
      "\t[#  180] train_loss: 1.935324\n",
      "\t[#  360] train_loss: 2.328206\n",
      "\t[#  540] train_loss: 1.905713\n",
      "\t[#  720] train_loss: 2.370659\n",
      "\t[#  900] train_loss: 1.959296\n",
      "\ttrain_loss: 2.068504\n",
      "\tval_loss: 2.139162\n",
      "Epoch 40/210\n",
      "\t[#    0] train_loss: 2.262669\n",
      "\t[#  180] train_loss: 1.755826\n",
      "\t[#  360] train_loss: 2.372665\n",
      "\t[#  540] train_loss: 2.245456\n",
      "\t[#  720] train_loss: 2.173336\n",
      "\t[#  900] train_loss: 1.613097\n",
      "\ttrain_loss: 2.421456\n",
      "\tval_loss: 2.419464\n",
      "Epoch 41/210\n",
      "\t[#    0] train_loss: 1.933471\n",
      "\t[#  180] train_loss: 2.244340\n",
      "\t[#  360] train_loss: 1.785707\n",
      "\t[#  540] train_loss: 4.585553\n",
      "\t[#  720] train_loss: 2.129589\n",
      "\t[#  900] train_loss: 2.124688\n",
      "\ttrain_loss: 2.173617\n",
      "\tval_loss: 2.280289\n",
      "Epoch 42/210\n",
      "\t[#    0] train_loss: 1.648138\n",
      "\t[#  180] train_loss: 1.957626\n",
      "\t[#  360] train_loss: 2.216222\n",
      "\t[#  540] train_loss: 2.991876\n",
      "\t[#  720] train_loss: 2.679080\n",
      "\t[#  900] train_loss: 2.163630\n",
      "\ttrain_loss: 2.315373\n",
      "\tval_loss: 2.337419\n",
      "Epoch 43/210\n",
      "\t[#    0] train_loss: 2.154421\n",
      "\t[#  180] train_loss: 2.304875\n",
      "\t[#  360] train_loss: 1.557097\n",
      "\t[#  540] train_loss: 1.693080\n",
      "\t[#  720] train_loss: 1.746929\n",
      "\t[#  900] train_loss: 2.291149\n",
      "\ttrain_loss: 2.174738\n",
      "\tval_loss: 2.140188\n",
      "Epoch 44/210\n",
      "\t[#    0] train_loss: 1.499319\n",
      "\t[#  180] train_loss: 1.958010\n",
      "\t[#  360] train_loss: 2.150815\n",
      "\t[#  540] train_loss: 1.646989\n",
      "\t[#  720] train_loss: 1.809168\n",
      "\t[#  900] train_loss: 2.376009\n",
      "\ttrain_loss: 2.039839\n",
      "\tval_loss: 2.119483\n",
      "Epoch 45/210\n",
      "\t[#    0] train_loss: 2.131037\n",
      "\t[#  180] train_loss: 1.976674\n",
      "\t[#  360] train_loss: 2.283673\n",
      "\t[#  540] train_loss: 1.773574\n",
      "\t[#  720] train_loss: 1.934892\n",
      "\t[#  900] train_loss: 2.225216\n",
      "\ttrain_loss: 2.068934\n",
      "\tval_loss: 2.100146\n",
      "Epoch 46/210\n",
      "\t[#    0] train_loss: 1.806616\n",
      "\t[#  180] train_loss: 1.769738\n",
      "\t[#  360] train_loss: 2.128777\n",
      "\t[#  540] train_loss: 2.130730\n",
      "\t[#  720] train_loss: 1.836895\n",
      "\t[#  900] train_loss: 2.628126\n",
      "\ttrain_loss: 2.009981\n",
      "\tval_loss: 2.090191\n",
      "Epoch 47/210\n",
      "\t[#    0] train_loss: 3.442436\n",
      "\t[#  180] train_loss: 5.733561\n",
      "\t[#  360] train_loss: 1.824950\n",
      "\t[#  540] train_loss: 1.575890\n",
      "\t[#  720] train_loss: 2.101818\n",
      "\t[#  900] train_loss: 2.241020\n",
      "\ttrain_loss: 2.379174\n",
      "\tval_loss: 2.188839\n",
      "Epoch 48/210\n",
      "\t[#    0] train_loss: 2.569854\n",
      "\t[#  180] train_loss: 2.317102\n",
      "\t[#  360] train_loss: 2.556738\n",
      "\t[#  540] train_loss: 2.063351\n",
      "\t[#  720] train_loss: 1.899104\n",
      "\t[#  900] train_loss: 1.846393\n",
      "\ttrain_loss: 2.079111\n",
      "\tval_loss: 2.119834\n",
      "Epoch 49/210\n",
      "\t[#    0] train_loss: 1.884488\n",
      "\t[#  180] train_loss: 2.022130\n",
      "\t[#  360] train_loss: 1.695119\n",
      "\t[#  540] train_loss: 2.080637\n",
      "\t[#  720] train_loss: 2.061921\n",
      "\t[#  900] train_loss: 2.001646\n",
      "\ttrain_loss: 2.227508\n",
      "\tval_loss: 2.423755\n",
      "Epoch 50/210\n",
      "\t[#    0] train_loss: 2.548824\n",
      "\t[#  180] train_loss: 1.457286\n",
      "\t[#  360] train_loss: 2.189777\n",
      "\t[#  540] train_loss: 2.017983\n",
      "\t[#  720] train_loss: 2.189452\n",
      "\t[#  900] train_loss: 1.871090\n",
      "\ttrain_loss: 2.363272\n",
      "\tval_loss: 2.126405\n",
      "Epoch 51/210\n",
      "\t[#    0] train_loss: 2.151089\n",
      "\t[#  180] train_loss: 2.362591\n",
      "\t[#  360] train_loss: 1.928079\n",
      "\t[#  540] train_loss: 1.613854\n",
      "\t[#  720] train_loss: 1.772964\n",
      "\t[#  900] train_loss: 2.173301\n",
      "\ttrain_loss: 2.040011\n",
      "\tval_loss: 2.233067\n",
      "Epoch 52/210\n",
      "\t[#    0] train_loss: 2.120137\n",
      "\t[#  180] train_loss: 2.098592\n",
      "\t[#  360] train_loss: 2.160226\n",
      "\t[#  540] train_loss: 2.057021\n",
      "\t[#  720] train_loss: 1.750157\n",
      "\t[#  900] train_loss: 1.315724\n",
      "\ttrain_loss: 2.043398\n",
      "\tval_loss: 1.982147\n",
      "\t*** Personal Best ***\n",
      "Epoch 53/210\n",
      "\t[#    0] train_loss: 1.558796\n",
      "\t[#  180] train_loss: 1.930827\n",
      "\t[#  360] train_loss: 1.873010\n",
      "\t[#  540] train_loss: 1.571473\n",
      "\t[#  720] train_loss: 2.354633\n",
      "\t[#  900] train_loss: 3.102155\n",
      "\ttrain_loss: 1.971864\n",
      "\tval_loss: 1.981169\n",
      "\t*** Personal Best ***\n",
      "Epoch 54/210\n",
      "\t[#    0] train_loss: 1.995285\n",
      "\t[#  180] train_loss: 1.414310\n",
      "\t[#  360] train_loss: 2.424404\n",
      "\t[#  540] train_loss: 1.829855\n",
      "\t[#  720] train_loss: 1.785821\n",
      "\t[#  900] train_loss: 1.553147\n",
      "\ttrain_loss: 1.993915\n",
      "\tval_loss: 1.954887\n",
      "\t*** Personal Best ***\n",
      "Epoch 55/210\n",
      "\t[#    0] train_loss: 2.045369\n",
      "\t[#  180] train_loss: 1.798120\n",
      "\t[#  360] train_loss: 1.595382\n",
      "\t[#  540] train_loss: 2.197838\n",
      "\t[#  720] train_loss: 2.283382\n",
      "\t[#  900] train_loss: 2.143681\n",
      "\ttrain_loss: 2.154055\n",
      "\tval_loss: 2.396756\n",
      "Epoch 56/210\n",
      "\t[#    0] train_loss: 3.666867\n",
      "\t[#  180] train_loss: 2.233668\n",
      "\t[#  360] train_loss: 2.205066\n",
      "\t[#  540] train_loss: 2.049256\n",
      "\t[#  720] train_loss: 1.921646\n",
      "\t[#  900] train_loss: 1.959374\n",
      "\ttrain_loss: 2.130450\n",
      "\tval_loss: 2.000030\n",
      "Epoch 57/210\n",
      "\t[#    0] train_loss: 2.058342\n",
      "\t[#  180] train_loss: 2.223937\n",
      "\t[#  360] train_loss: 1.661892\n",
      "\t[#  540] train_loss: 1.631283\n",
      "\t[#  720] train_loss: 1.882743\n",
      "\t[#  900] train_loss: 3.189343\n",
      "\ttrain_loss: 2.106782\n",
      "\tval_loss: 2.651317\n",
      "Epoch 58/210\n",
      "\t[#    0] train_loss: 2.073609\n",
      "\t[#  180] train_loss: 1.884960\n",
      "\t[#  360] train_loss: 1.473382\n",
      "\t[#  540] train_loss: 1.781156\n",
      "\t[#  720] train_loss: 2.270530\n",
      "\t[#  900] train_loss: 1.880214\n",
      "\ttrain_loss: 2.224893\n",
      "\tval_loss: 2.029234\n",
      "Epoch 59/210\n",
      "\t[#    0] train_loss: 1.586473\n",
      "\t[#  180] train_loss: 2.229960\n",
      "\t[#  360] train_loss: 2.055395\n",
      "\t[#  540] train_loss: 2.728280\n",
      "\t[#  720] train_loss: 1.462845\n",
      "\t[#  900] train_loss: 2.008673\n",
      "\ttrain_loss: 1.998388\n",
      "\tval_loss: 2.009392\n",
      "Epoch 60/210\n",
      "\t[#    0] train_loss: 2.179205\n",
      "\t[#  180] train_loss: 1.541502\n",
      "\t[#  360] train_loss: 1.371758\n",
      "\t[#  540] train_loss: 1.798224\n",
      "\t[#  720] train_loss: 2.203162\n",
      "\t[#  900] train_loss: 1.926520\n",
      "\ttrain_loss: 1.952390\n",
      "\tval_loss: 2.072877\n",
      "Epoch 61/210\n",
      "\t[#    0] train_loss: 1.445181\n",
      "\t[#  180] train_loss: 1.446972\n",
      "\t[#  360] train_loss: 2.198344\n",
      "\t[#  540] train_loss: 1.519892\n",
      "\t[#  720] train_loss: 2.191228\n",
      "\t[#  900] train_loss: 1.653159\n",
      "\ttrain_loss: 1.993955\n",
      "\tval_loss: 1.953397\n",
      "\t*** Personal Best ***\n",
      "Epoch 62/210\n",
      "\t[#    0] train_loss: 2.252438\n",
      "\t[#  180] train_loss: 1.642023\n",
      "\t[#  360] train_loss: 1.602114\n",
      "\t[#  540] train_loss: 2.106682\n",
      "\t[#  720] train_loss: 1.982329\n",
      "\t[#  900] train_loss: 1.715940\n",
      "\ttrain_loss: 1.968391\n",
      "\tval_loss: 2.019230\n",
      "Epoch 63/210\n",
      "\t[#    0] train_loss: 1.394713\n",
      "\t[#  180] train_loss: 2.246909\n",
      "\t[#  360] train_loss: 2.500904\n",
      "\t[#  540] train_loss: 1.979884\n",
      "\t[#  720] train_loss: 2.665135\n",
      "\t[#  900] train_loss: 1.945609\n",
      "\ttrain_loss: 2.184918\n",
      "\tval_loss: 2.100093\n",
      "Epoch 64/210\n",
      "\t[#    0] train_loss: 1.901423\n",
      "\t[#  180] train_loss: 2.265712\n",
      "\t[#  360] train_loss: 1.803062\n",
      "\t[#  540] train_loss: 1.869115\n",
      "\t[#  720] train_loss: 1.532642\n",
      "\t[#  900] train_loss: 1.389326\n",
      "\ttrain_loss: 2.088291\n",
      "\tval_loss: 2.141468\n",
      "Epoch 65/210\n",
      "\t[#    0] train_loss: 2.050993\n",
      "\t[#  180] train_loss: 1.625885\n",
      "\t[#  360] train_loss: 1.988644\n",
      "\t[#  540] train_loss: 2.161130\n",
      "\t[#  720] train_loss: 2.297791\n",
      "\t[#  900] train_loss: 1.809398\n",
      "\ttrain_loss: 2.080720\n",
      "\tval_loss: 2.017809\n",
      "Epoch 66/210\n",
      "\t[#    0] train_loss: 1.785858\n",
      "\t[#  180] train_loss: 2.034762\n",
      "\t[#  360] train_loss: 1.902623\n",
      "\t[#  540] train_loss: 2.284966\n",
      "\t[#  720] train_loss: 2.282911\n",
      "\t[#  900] train_loss: 1.532663\n",
      "\ttrain_loss: 1.984009\n",
      "\tval_loss: 2.238214\n",
      "Epoch 67/210\n",
      "\t[#    0] train_loss: 1.419207\n",
      "\t[#  180] train_loss: 1.208671\n",
      "\t[#  360] train_loss: 2.235378\n",
      "\t[#  540] train_loss: 1.417736\n",
      "\t[#  720] train_loss: 2.162975\n",
      "\t[#  900] train_loss: 2.196823\n",
      "\ttrain_loss: 2.015158\n",
      "\tval_loss: 1.977852\n",
      "Epoch 68/210\n",
      "\t[#    0] train_loss: 2.114936\n",
      "\t[#  180] train_loss: 2.210777\n",
      "\t[#  360] train_loss: 1.844673\n",
      "\t[#  540] train_loss: 1.703266\n",
      "\t[#  720] train_loss: 1.926909\n",
      "\t[#  900] train_loss: 2.043437\n",
      "\ttrain_loss: 1.971301\n",
      "\tval_loss: 1.925539\n",
      "\t*** Personal Best ***\n",
      "Epoch 69/210\n",
      "\t[#    0] train_loss: 1.623782\n",
      "\t[#  180] train_loss: 2.392005\n",
      "\t[#  360] train_loss: 2.545849\n",
      "\t[#  540] train_loss: 1.282565\n",
      "\t[#  720] train_loss: 1.650074\n",
      "\t[#  900] train_loss: 2.718168\n",
      "\ttrain_loss: 1.912035\n",
      "\tval_loss: 1.887552\n",
      "\t*** Personal Best ***\n",
      "Epoch 70/210\n",
      "\t[#    0] train_loss: 1.476197\n",
      "\t[#  180] train_loss: 1.746141\n",
      "\t[#  360] train_loss: 2.197520\n",
      "\t[#  540] train_loss: 2.295728\n",
      "\t[#  720] train_loss: 2.075559\n",
      "\t[#  900] train_loss: 2.179098\n",
      "\ttrain_loss: 1.915554\n",
      "\tval_loss: 1.945881\n",
      "Epoch 71/210\n",
      "\t[#    0] train_loss: 1.715232\n",
      "\t[#  180] train_loss: 2.090471\n",
      "\t[#  360] train_loss: 2.111333\n",
      "\t[#  540] train_loss: 2.573423\n",
      "\t[#  720] train_loss: 1.807173\n",
      "\t[#  900] train_loss: 1.601769\n",
      "\ttrain_loss: 2.140368\n",
      "\tval_loss: 3.165082\n",
      "Epoch 72/210\n",
      "\t[#    0] train_loss: 2.210079\n",
      "\t[#  180] train_loss: 1.810126\n",
      "\t[#  360] train_loss: 2.278148\n",
      "\t[#  540] train_loss: 2.096358\n",
      "\t[#  720] train_loss: 1.965205\n",
      "\t[#  900] train_loss: 2.054993\n",
      "\ttrain_loss: 2.038447\n",
      "\tval_loss: 1.918796\n",
      "Epoch 73/210\n",
      "\t[#    0] train_loss: 1.200438\n",
      "\t[#  180] train_loss: 2.065877\n",
      "\t[#  360] train_loss: 1.767961\n",
      "\t[#  540] train_loss: 1.649073\n",
      "\t[#  720] train_loss: 1.534938\n",
      "\t[#  900] train_loss: 1.910040\n",
      "\ttrain_loss: 1.926591\n",
      "\tval_loss: 1.987085\n",
      "Epoch 74/210\n",
      "\t[#    0] train_loss: 1.459054\n",
      "\t[#  180] train_loss: 2.396528\n",
      "\t[#  360] train_loss: 1.851185\n",
      "\t[#  540] train_loss: 2.012296\n",
      "\t[#  720] train_loss: 1.604017\n",
      "\t[#  900] train_loss: 1.493366\n",
      "\ttrain_loss: 1.923669\n",
      "\tval_loss: 1.961461\n",
      "Epoch 75/210\n",
      "\t[#    0] train_loss: 1.809596\n",
      "\t[#  180] train_loss: 2.006111\n",
      "\t[#  360] train_loss: 1.822201\n",
      "\t[#  540] train_loss: 2.398105\n",
      "\t[#  720] train_loss: 2.354964\n",
      "\t[#  900] train_loss: 2.204684\n",
      "\ttrain_loss: 1.903435\n",
      "\tval_loss: 1.915253\n",
      "Epoch 76/210\n",
      "\t[#    0] train_loss: 1.339764\n",
      "\t[#  180] train_loss: 2.142786\n",
      "\t[#  360] train_loss: 2.396152\n",
      "\t[#  540] train_loss: 1.743188\n",
      "\t[#  720] train_loss: 2.250613\n",
      "\t[#  900] train_loss: 2.736884\n",
      "\ttrain_loss: 1.913684\n",
      "\tval_loss: 1.904318\n",
      "Epoch 77/210\n",
      "\t[#    0] train_loss: 1.769378\n",
      "\t[#  180] train_loss: 2.123633\n",
      "\t[#  360] train_loss: 1.932895\n",
      "\t[#  540] train_loss: 1.530841\n",
      "\t[#  720] train_loss: 2.220631\n",
      "\t[#  900] train_loss: 2.192969\n",
      "\ttrain_loss: 2.118841\n",
      "\tval_loss: 2.532670\n",
      "Epoch 78/210\n",
      "\t[#    0] train_loss: 5.133906\n",
      "\t[#  180] train_loss: 1.860675\n",
      "\t[#  360] train_loss: 2.556842\n",
      "\t[#  540] train_loss: 2.356463\n",
      "\t[#  720] train_loss: 2.222289\n",
      "\t[#  900] train_loss: 1.741600\n",
      "\ttrain_loss: 2.027025\n",
      "\tval_loss: 2.012136\n",
      "Epoch 79/210\n",
      "\t[#    0] train_loss: 2.054100\n",
      "\t[#  180] train_loss: 1.670523\n",
      "\t[#  360] train_loss: 2.354132\n",
      "\t[#  540] train_loss: 1.829400\n",
      "\t[#  720] train_loss: 1.943270\n",
      "\t[#  900] train_loss: 2.042387\n",
      "\ttrain_loss: 1.908099\n",
      "\tval_loss: 1.999268\n",
      "Epoch 80/210\n",
      "\t[#    0] train_loss: 2.007375\n",
      "\t[#  180] train_loss: 1.678682\n",
      "\t[#  360] train_loss: 2.554313\n",
      "\t[#  540] train_loss: 2.090772\n",
      "\t[#  720] train_loss: 1.413099\n",
      "\t[#  900] train_loss: 1.597542\n",
      "\ttrain_loss: 1.927579\n",
      "\tval_loss: 1.907128\n",
      "Epoch 81/210\n",
      "\t[#    0] train_loss: 1.894513\n",
      "\t[#  180] train_loss: 1.955353\n",
      "\t[#  360] train_loss: 1.624412\n",
      "\t[#  540] train_loss: 2.525830\n",
      "\t[#  720] train_loss: 1.537734\n",
      "\t[#  900] train_loss: 1.335600\n",
      "\ttrain_loss: 1.895579\n",
      "\tval_loss: 1.900241\n",
      "Epoch 82/210\n",
      "\t[#    0] train_loss: 1.694063\n",
      "\t[#  180] train_loss: 2.017023\n",
      "\t[#  360] train_loss: 2.133806\n",
      "\t[#  540] train_loss: 1.565567\n",
      "\t[#  720] train_loss: 2.220481\n",
      "\t[#  900] train_loss: 2.185615\n",
      "\ttrain_loss: 1.892186\n",
      "\tval_loss: 2.081199\n",
      "Epoch 83/210\n",
      "\t[#    0] train_loss: 1.636241\n",
      "\t[#  180] train_loss: 2.024943\n",
      "\t[#  360] train_loss: 1.801418\n",
      "\t[#  540] train_loss: 1.889307\n",
      "\t[#  720] train_loss: 1.945568\n",
      "\t[#  900] train_loss: 2.286880\n",
      "\ttrain_loss: 2.101522\n",
      "\tval_loss: 2.406653\n",
      "Epoch 84/210\n",
      "\t[#    0] train_loss: 1.789518\n",
      "\t[#  180] train_loss: 2.212446\n",
      "\t[#  360] train_loss: 2.001074\n",
      "\t[#  540] train_loss: 1.675632\n",
      "\t[#  720] train_loss: 1.515437\n",
      "\t[#  900] train_loss: 2.096612\n",
      "\ttrain_loss: 2.079488\n",
      "\tval_loss: 1.949298\n",
      "Epoch 85/210\n",
      "\t[#    0] train_loss: 1.646951\n",
      "\t[#  180] train_loss: 2.371792\n",
      "\t[#  360] train_loss: 1.477550\n",
      "\t[#  540] train_loss: 1.616659\n",
      "\t[#  720] train_loss: 1.996232\n",
      "\t[#  900] train_loss: 1.931924\n",
      "\ttrain_loss: 1.913618\n",
      "\tval_loss: 1.896652\n",
      "Epoch 86/210\n",
      "\t[#    0] train_loss: 1.677185\n",
      "\t[#  180] train_loss: 1.874815\n",
      "\t[#  360] train_loss: 1.237877\n",
      "\t[#  540] train_loss: 2.580454\n",
      "\t[#  720] train_loss: 1.249330\n",
      "\t[#  900] train_loss: 2.588144\n",
      "\ttrain_loss: 2.075839\n",
      "\tval_loss: 1.959734\n",
      "Epoch 87/210\n",
      "\t[#    0] train_loss: 1.746819\n",
      "\t[#  180] train_loss: 1.945325\n",
      "\t[#  360] train_loss: 1.635092\n",
      "\t[#  540] train_loss: 1.951488\n",
      "\t[#  720] train_loss: 1.396071\n",
      "\t[#  900] train_loss: 1.980364\n",
      "\ttrain_loss: 1.888974\n",
      "\tval_loss: 2.039324\n",
      "Epoch 88/210\n",
      "\t[#    0] train_loss: 1.803839\n",
      "\t[#  180] train_loss: 2.138114\n",
      "\t[#  360] train_loss: 2.181559\n",
      "\t[#  540] train_loss: 1.181834\n",
      "\t[#  720] train_loss: 2.176669\n",
      "\t[#  900] train_loss: 1.953878\n",
      "\ttrain_loss: 1.865509\n",
      "\tval_loss: 1.934449\n",
      "Epoch 89/210\n",
      "\t[#    0] train_loss: 2.149844\n",
      "\t[#  180] train_loss: 1.529798\n",
      "\t[#  360] train_loss: 2.197546\n",
      "\t[#  540] train_loss: 1.764873\n",
      "\t[#  720] train_loss: 1.738083\n",
      "\t[#  900] train_loss: 1.589435\n",
      "\ttrain_loss: 1.861657\n",
      "\tval_loss: 1.843409\n",
      "\t*** Personal Best ***\n",
      "Epoch 90/210\n",
      "\t[#    0] train_loss: 2.523638\n",
      "\t[#  180] train_loss: 1.769453\n",
      "\t[#  360] train_loss: 6.425014\n",
      "\t[#  540] train_loss: 1.924152\n",
      "\t[#  720] train_loss: 1.662419\n",
      "\t[#  900] train_loss: 1.654886\n",
      "\ttrain_loss: 2.035256\n",
      "\tval_loss: 2.126137\n",
      "Epoch 91/210\n",
      "\t[#    0] train_loss: 1.768907\n",
      "\t[#  180] train_loss: 1.820773\n",
      "\t[#  360] train_loss: 1.526112\n",
      "\t[#  540] train_loss: 2.075573\n",
      "\t[#  720] train_loss: 2.045382\n",
      "\t[#  900] train_loss: 1.223635\n",
      "\ttrain_loss: 1.948984\n",
      "\tval_loss: 1.912244\n",
      "Epoch 92/210\n",
      "\t[#    0] train_loss: 1.715267\n",
      "\t[#  180] train_loss: 2.112574\n",
      "\t[#  360] train_loss: 1.860768\n",
      "\t[#  540] train_loss: 2.169668\n",
      "\t[#  720] train_loss: 1.884446\n",
      "\t[#  900] train_loss: 1.340164\n",
      "\ttrain_loss: 1.866358\n",
      "\tval_loss: 1.894214\n",
      "Epoch 93/210\n",
      "\t[#    0] train_loss: 2.222604\n",
      "\t[#  180] train_loss: 2.396227\n",
      "\t[#  360] train_loss: 2.074708\n",
      "\t[#  540] train_loss: 2.135385\n",
      "\t[#  720] train_loss: 1.874804\n",
      "\t[#  900] train_loss: 1.253303\n",
      "\ttrain_loss: 1.897378\n",
      "\tval_loss: 1.861347\n",
      "Epoch 94/210\n",
      "\t[#    0] train_loss: 2.105624\n",
      "\t[#  180] train_loss: 2.184878\n",
      "\t[#  360] train_loss: 1.489688\n",
      "\t[#  540] train_loss: 2.086138\n",
      "\t[#  720] train_loss: 1.427392\n",
      "\t[#  900] train_loss: 1.976637\n",
      "\ttrain_loss: 1.857120\n",
      "\tval_loss: 1.856115\n",
      "Epoch 95/210\n",
      "\t[#    0] train_loss: 2.133673\n",
      "\t[#  180] train_loss: 1.897512\n",
      "\t[#  360] train_loss: 1.348847\n",
      "\t[#  540] train_loss: 1.822204\n",
      "\t[#  720] train_loss: 1.692186\n",
      "\t[#  900] train_loss: 1.595519\n",
      "\ttrain_loss: 1.891371\n",
      "\tval_loss: 1.955227\n",
      "Epoch 96/210\n",
      "\t[#    0] train_loss: 2.354494\n",
      "\t[#  180] train_loss: 2.307179\n",
      "\t[#  360] train_loss: 1.649713\n",
      "\t[#  540] train_loss: 1.677511\n",
      "\t[#  720] train_loss: 1.804597\n",
      "\t[#  900] train_loss: 2.456541\n",
      "\ttrain_loss: 1.958428\n",
      "\tval_loss: 1.947505\n",
      "Epoch 97/210\n",
      "\t[#    0] train_loss: 2.369156\n",
      "\t[#  180] train_loss: 2.011541\n",
      "\t[#  360] train_loss: 1.716101\n",
      "\t[#  540] train_loss: 1.826337\n",
      "\t[#  720] train_loss: 2.101276\n",
      "\t[#  900] train_loss: 1.789410\n",
      "\ttrain_loss: 2.034820\n",
      "\tval_loss: 1.934387\n",
      "Epoch 98/210\n",
      "\t[#    0] train_loss: 2.126780\n",
      "\t[#  180] train_loss: 2.271571\n",
      "\t[#  360] train_loss: 1.303178\n",
      "\t[#  540] train_loss: 1.890170\n",
      "\t[#  720] train_loss: 1.434379\n",
      "\t[#  900] train_loss: 2.729702\n",
      "\ttrain_loss: 1.993391\n",
      "\tval_loss: 2.035148\n",
      "Epoch 99/210\n",
      "\t[#    0] train_loss: 1.767572\n",
      "\t[#  180] train_loss: 2.689816\n",
      "\t[#  360] train_loss: 2.119209\n",
      "\t[#  540] train_loss: 1.913913\n",
      "\t[#  720] train_loss: 1.447751\n",
      "\t[#  900] train_loss: 2.045656\n",
      "\ttrain_loss: 1.965738\n",
      "\tval_loss: 1.899836\n",
      "Epoch 100/210\n",
      "\t[#    0] train_loss: 1.531963\n",
      "\t[#  180] train_loss: 1.364941\n",
      "\t[#  360] train_loss: 1.430456\n",
      "\t[#  540] train_loss: 1.621835\n",
      "\t[#  720] train_loss: 1.612803\n",
      "\t[#  900] train_loss: 1.973206\n",
      "\ttrain_loss: 1.848486\n",
      "\tval_loss: 2.085849\n",
      "Epoch 101/210\n",
      "\t[#    0] train_loss: 2.447283\n",
      "\t[#  180] train_loss: 2.472352\n",
      "\t[#  360] train_loss: 2.021073\n",
      "\t[#  540] train_loss: 1.613837\n",
      "\t[#  720] train_loss: 1.702065\n",
      "\t[#  900] train_loss: 1.555207\n",
      "\ttrain_loss: 1.926140\n",
      "\tval_loss: 1.988658\n",
      "Epoch 102/210\n",
      "\t[#    0] train_loss: 1.225450\n",
      "\t[#  180] train_loss: 1.757270\n",
      "\t[#  360] train_loss: 2.254261\n",
      "\t[#  540] train_loss: 2.125194\n",
      "\t[#  720] train_loss: 1.801984\n",
      "\t[#  900] train_loss: 2.568453\n",
      "\ttrain_loss: 1.848332\n",
      "\tval_loss: 1.850439\n",
      "Epoch 103/210\n",
      "\t[#    0] train_loss: 1.453738\n",
      "\t[#  180] train_loss: 1.325713\n",
      "\t[#  360] train_loss: 1.934018\n",
      "\t[#  540] train_loss: 1.496286\n",
      "\t[#  720] train_loss: 1.745400\n",
      "\t[#  900] train_loss: 2.258219\n",
      "\ttrain_loss: 1.831529\n",
      "\tval_loss: 1.882627\n",
      "Epoch 104/210\n",
      "\t[#    0] train_loss: 1.358731\n",
      "\t[#  180] train_loss: 1.608015\n",
      "\t[#  360] train_loss: 1.814938\n",
      "\t[#  540] train_loss: 1.884860\n",
      "\t[#  720] train_loss: 2.050855\n",
      "\t[#  900] train_loss: 1.534241\n",
      "\ttrain_loss: 1.922867\n",
      "\tval_loss: 1.829738\n",
      "\t*** Personal Best ***\n",
      "Epoch 105/210\n",
      "\t[#    0] train_loss: 1.710204\n",
      "\t[#  180] train_loss: 1.256264\n",
      "\t[#  360] train_loss: 1.440403\n",
      "\t[#  540] train_loss: 1.657081\n",
      "\t[#  720] train_loss: 1.977456\n",
      "\t[#  900] train_loss: 1.908253\n",
      "\ttrain_loss: 1.848283\n",
      "\tval_loss: 1.830562\n",
      "Epoch 106/210\n",
      "\t[#    0] train_loss: 1.967799\n",
      "\t[#  180] train_loss: 1.800936\n",
      "\t[#  360] train_loss: 2.471648\n",
      "\t[#  540] train_loss: 2.052201\n",
      "\t[#  720] train_loss: 1.387058\n",
      "\t[#  900] train_loss: 1.948449\n",
      "\ttrain_loss: 2.095962\n",
      "\tval_loss: 1.919628\n",
      "Epoch 107/210\n",
      "\t[#    0] train_loss: 1.643210\n",
      "\t[#  180] train_loss: 1.996082\n",
      "\t[#  360] train_loss: 1.575629\n",
      "\t[#  540] train_loss: 1.428396\n",
      "\t[#  720] train_loss: 2.064957\n",
      "\t[#  900] train_loss: 2.145683\n",
      "\ttrain_loss: 1.890963\n",
      "\tval_loss: 1.838021\n",
      "Epoch 108/210\n",
      "\t[#    0] train_loss: 1.686891\n",
      "\t[#  180] train_loss: 2.010590\n",
      "\t[#  360] train_loss: 2.307358\n",
      "\t[#  540] train_loss: 2.365184\n",
      "\t[#  720] train_loss: 1.850703\n",
      "\t[#  900] train_loss: 2.375057\n",
      "\ttrain_loss: 1.841931\n",
      "\tval_loss: 1.895098\n",
      "Epoch 109/210\n",
      "\t[#    0] train_loss: 1.885424\n",
      "\t[#  180] train_loss: 2.204580\n",
      "\t[#  360] train_loss: 2.260753\n",
      "\t[#  540] train_loss: 1.848054\n",
      "\t[#  720] train_loss: 2.477968\n",
      "\t[#  900] train_loss: 2.727432\n",
      "\ttrain_loss: 1.823071\n",
      "\tval_loss: 1.856108\n",
      "Epoch 110/210\n",
      "\t[#    0] train_loss: 1.709519\n",
      "\t[#  180] train_loss: 1.376926\n",
      "\t[#  360] train_loss: 1.151952\n",
      "\t[#  540] train_loss: 2.460225\n",
      "\t[#  720] train_loss: 1.574048\n",
      "\t[#  900] train_loss: 1.762976\n",
      "\ttrain_loss: 1.943307\n",
      "\tval_loss: 1.936662\n",
      "Epoch 111/210\n",
      "\t[#    0] train_loss: 1.762924\n",
      "\t[#  180] train_loss: 2.854257\n",
      "\t[#  360] train_loss: 1.409088\n",
      "\t[#  540] train_loss: 1.766225\n",
      "\t[#  720] train_loss: 1.709859\n",
      "\t[#  900] train_loss: 1.390255\n",
      "\ttrain_loss: 1.845959\n",
      "\tval_loss: 1.831437\n",
      "Epoch 112/210\n",
      "\t[#    0] train_loss: 1.571600\n",
      "\t[#  180] train_loss: 1.818380\n",
      "\t[#  360] train_loss: 1.714535\n",
      "\t[#  540] train_loss: 1.605827\n",
      "\t[#  720] train_loss: 1.597710\n",
      "\t[#  900] train_loss: 1.068859\n",
      "\ttrain_loss: 1.834631\n",
      "\tval_loss: 1.804929\n",
      "\t*** Personal Best ***\n",
      "Epoch 113/210\n",
      "\t[#    0] train_loss: 1.548569\n",
      "\t[#  180] train_loss: 1.385803\n",
      "\t[#  360] train_loss: 1.943356\n",
      "\t[#  540] train_loss: 2.054380\n",
      "\t[#  720] train_loss: 1.688011\n",
      "\t[#  900] train_loss: 1.826649\n",
      "\ttrain_loss: 1.809127\n",
      "\tval_loss: 1.848318\n",
      "Epoch 114/210\n",
      "\t[#    0] train_loss: 1.410179\n",
      "\t[#  180] train_loss: 2.080628\n",
      "\t[#  360] train_loss: 1.421888\n",
      "\t[#  540] train_loss: 1.617034\n",
      "\t[#  720] train_loss: 2.473896\n",
      "\t[#  900] train_loss: 1.359871\n",
      "\ttrain_loss: 1.810216\n",
      "\tval_loss: 1.873950\n",
      "Epoch 115/210\n",
      "\t[#    0] train_loss: 1.824878\n",
      "\t[#  180] train_loss: 1.979656\n",
      "\t[#  360] train_loss: 2.498466\n",
      "\t[#  540] train_loss: 1.882145\n",
      "\t[#  720] train_loss: 1.559817\n",
      "\t[#  900] train_loss: 2.362199\n",
      "\ttrain_loss: 1.833642\n",
      "\tval_loss: 1.834499\n",
      "Epoch 116/210\n",
      "\t[#    0] train_loss: 1.931027\n",
      "\t[#  180] train_loss: 1.857996\n",
      "\t[#  360] train_loss: 1.417314\n",
      "\t[#  540] train_loss: 1.959825\n",
      "\t[#  720] train_loss: 1.439093\n",
      "\t[#  900] train_loss: 2.044488\n",
      "\ttrain_loss: 1.809786\n",
      "\tval_loss: 1.914208\n",
      "Epoch 117/210\n",
      "\t[#    0] train_loss: 2.096795\n",
      "\t[#  180] train_loss: 1.930873\n",
      "\t[#  360] train_loss: 1.789035\n",
      "\t[#  540] train_loss: 1.632483\n",
      "\t[#  720] train_loss: 1.432577\n",
      "\t[#  900] train_loss: 1.440924\n",
      "\ttrain_loss: 1.793305\n",
      "\tval_loss: 1.819356\n",
      "Epoch 118/210\n",
      "\t[#    0] train_loss: 2.188461\n",
      "\t[#  180] train_loss: 1.778816\n",
      "\t[#  360] train_loss: 1.857108\n",
      "\t[#  540] train_loss: 1.567063\n",
      "\t[#  720] train_loss: 2.083512\n",
      "\t[#  900] train_loss: 2.641369\n",
      "\ttrain_loss: 1.826579\n",
      "\tval_loss: 1.855928\n",
      "Epoch 119/210\n",
      "\t[#    0] train_loss: 2.244403\n",
      "\t[#  180] train_loss: 1.752421\n",
      "\t[#  360] train_loss: 1.455769\n",
      "\t[#  540] train_loss: 1.781990\n",
      "\t[#  720] train_loss: 1.967102\n",
      "\t[#  900] train_loss: 2.171026\n",
      "\ttrain_loss: 1.861147\n",
      "\tval_loss: 1.825448\n",
      "Epoch 120/210\n",
      "\t[#    0] train_loss: 1.873037\n",
      "\t[#  180] train_loss: 0.971103\n",
      "\t[#  360] train_loss: 2.002463\n",
      "\t[#  540] train_loss: 1.809020\n",
      "\t[#  720] train_loss: 1.870530\n",
      "\t[#  900] train_loss: 1.480245\n",
      "\ttrain_loss: 1.883360\n",
      "\tval_loss: 2.051936\n",
      "Epoch 121/210\n",
      "\t[#    0] train_loss: 1.963827\n",
      "\t[#  180] train_loss: 1.693394\n",
      "\t[#  360] train_loss: 2.089282\n",
      "\t[#  540] train_loss: 1.243766\n",
      "\t[#  720] train_loss: 1.835060\n",
      "\t[#  900] train_loss: 1.986403\n",
      "\ttrain_loss: 1.855976\n",
      "\tval_loss: 1.848004\n",
      "Epoch 122/210\n",
      "\t[#    0] train_loss: 2.131273\n",
      "\t[#  180] train_loss: 1.794785\n",
      "\t[#  360] train_loss: 1.422438\n",
      "\t[#  540] train_loss: 1.322024\n",
      "\t[#  720] train_loss: 1.863729\n",
      "\t[#  900] train_loss: 2.105660\n",
      "\ttrain_loss: 1.787963\n",
      "\tval_loss: 1.784906\n",
      "\t*** Personal Best ***\n",
      "Epoch 123/210\n",
      "\t[#    0] train_loss: 1.482960\n",
      "\t[#  180] train_loss: 1.633207\n",
      "\t[#  360] train_loss: 1.630449\n",
      "\t[#  540] train_loss: 2.072739\n",
      "\t[#  720] train_loss: 1.136884\n",
      "\t[#  900] train_loss: 1.659383\n",
      "\ttrain_loss: 1.778126\n",
      "\tval_loss: 1.831685\n",
      "Epoch 124/210\n",
      "\t[#    0] train_loss: 1.605117\n",
      "\t[#  180] train_loss: 1.654821\n",
      "\t[#  360] train_loss: 1.699279\n",
      "\t[#  540] train_loss: 1.155093\n",
      "\t[#  720] train_loss: 1.273040\n",
      "\t[#  900] train_loss: 1.895728\n",
      "\ttrain_loss: 1.806340\n",
      "\tval_loss: 1.799740\n",
      "Epoch 125/210\n",
      "\t[#    0] train_loss: 1.755196\n",
      "\t[#  180] train_loss: 1.441568\n",
      "\t[#  360] train_loss: 1.524629\n",
      "\t[#  540] train_loss: 1.291027\n",
      "\t[#  720] train_loss: 1.859562\n",
      "\t[#  900] train_loss: 1.749234\n",
      "\ttrain_loss: 1.771503\n",
      "\tval_loss: 1.810351\n",
      "Epoch 126/210\n",
      "\t[#    0] train_loss: 1.677076\n",
      "\t[#  180] train_loss: 2.166680\n",
      "\t[#  360] train_loss: 1.997837\n",
      "\t[#  540] train_loss: 1.527133\n",
      "\t[#  720] train_loss: 2.651521\n",
      "\t[#  900] train_loss: 1.836041\n",
      "\ttrain_loss: 2.034318\n",
      "\tval_loss: 1.927156\n",
      "Epoch 127/210\n",
      "\t[#    0] train_loss: 1.370822\n",
      "\t[#  180] train_loss: 2.425485\n",
      "\t[#  360] train_loss: 1.371384\n",
      "\t[#  540] train_loss: 1.979864\n",
      "\t[#  720] train_loss: 1.923063\n",
      "\t[#  900] train_loss: 1.471486\n",
      "\ttrain_loss: 1.795353\n",
      "\tval_loss: 1.793112\n",
      "Epoch 128/210\n",
      "\t[#    0] train_loss: 1.574884\n",
      "\t[#  180] train_loss: 1.195014\n",
      "\t[#  360] train_loss: 1.967336\n",
      "\t[#  540] train_loss: 1.852223\n",
      "\t[#  720] train_loss: 2.612158\n",
      "\t[#  900] train_loss: 1.767511\n",
      "\ttrain_loss: 1.775090\n",
      "\tval_loss: 1.794799\n",
      "Epoch 129/210\n",
      "\t[#    0] train_loss: 1.697084\n",
      "\t[#  180] train_loss: 1.699339\n",
      "\t[#  360] train_loss: 1.724746\n",
      "\t[#  540] train_loss: 2.069022\n",
      "\t[#  720] train_loss: 1.592550\n",
      "\t[#  900] train_loss: 1.530159\n",
      "\ttrain_loss: 1.773895\n",
      "\tval_loss: 1.719171\n",
      "\t*** Personal Best ***\n",
      "Epoch 130/210\n",
      "\t[#    0] train_loss: 2.084824\n",
      "\t[#  180] train_loss: 2.574271\n",
      "\t[#  360] train_loss: 1.626873\n",
      "\t[#  540] train_loss: 1.964145\n",
      "\t[#  720] train_loss: 1.485071\n",
      "\t[#  900] train_loss: 1.841521\n",
      "\ttrain_loss: 1.737871\n",
      "\tval_loss: 1.776863\n",
      "Epoch 131/210\n",
      "\t[#    0] train_loss: 1.552110\n",
      "\t[#  180] train_loss: 1.840124\n",
      "\t[#  360] train_loss: 1.995526\n",
      "\t[#  540] train_loss: 2.071971\n",
      "\t[#  720] train_loss: 1.686668\n",
      "\t[#  900] train_loss: 1.163264\n",
      "\ttrain_loss: 1.822596\n",
      "\tval_loss: 1.954054\n",
      "Epoch 132/210\n",
      "\t[#    0] train_loss: 1.683365\n",
      "\t[#  180] train_loss: 2.115183\n",
      "\t[#  360] train_loss: 1.885093\n",
      "\t[#  540] train_loss: 1.943812\n",
      "\t[#  720] train_loss: 2.032706\n",
      "\t[#  900] train_loss: 2.598458\n",
      "\ttrain_loss: 1.829997\n",
      "\tval_loss: 1.789296\n",
      "Epoch 133/210\n",
      "\t[#    0] train_loss: 2.608340\n",
      "\t[#  180] train_loss: 2.743245\n",
      "\t[#  360] train_loss: 2.245283\n",
      "\t[#  540] train_loss: 1.665536\n",
      "\t[#  720] train_loss: 1.317582\n",
      "\t[#  900] train_loss: 1.130663\n",
      "\ttrain_loss: 1.761384\n",
      "\tval_loss: 1.757884\n",
      "Epoch 134/210\n",
      "\t[#    0] train_loss: 1.875626\n",
      "\t[#  180] train_loss: 1.558617\n",
      "\t[#  360] train_loss: 1.951339\n",
      "\t[#  540] train_loss: 1.550204\n",
      "\t[#  720] train_loss: 1.513943\n",
      "\t[#  900] train_loss: 1.731839\n",
      "\ttrain_loss: 1.738129\n",
      "\tval_loss: 1.776950\n",
      "Epoch 135/210\n",
      "\t[#    0] train_loss: 1.277365\n",
      "\t[#  180] train_loss: 0.977698\n",
      "\t[#  360] train_loss: 1.427948\n",
      "\t[#  540] train_loss: 1.719268\n",
      "\t[#  720] train_loss: 1.930600\n",
      "\t[#  900] train_loss: 1.714342\n",
      "\ttrain_loss: 1.780602\n",
      "\tval_loss: 1.843772\n",
      "Epoch 136/210\n",
      "\t[#    0] train_loss: 1.878332\n",
      "\t[#  180] train_loss: 1.365379\n",
      "\t[#  360] train_loss: 2.070599\n",
      "\t[#  540] train_loss: 1.628741\n",
      "\t[#  720] train_loss: 1.234194\n",
      "\t[#  900] train_loss: 2.484812\n",
      "\ttrain_loss: 2.023332\n",
      "\tval_loss: 1.792258\n",
      "Epoch 137/210\n",
      "\t[#    0] train_loss: 1.768461\n",
      "\t[#  180] train_loss: 2.455092\n",
      "\t[#  360] train_loss: 1.856309\n",
      "\t[#  540] train_loss: 1.240579\n",
      "\t[#  720] train_loss: 1.660702\n",
      "\t[#  900] train_loss: 1.612703\n",
      "\ttrain_loss: 1.773087\n",
      "\tval_loss: 1.773644\n",
      "Epoch 138/210\n",
      "\t[#    0] train_loss: 1.614478\n",
      "\t[#  180] train_loss: 1.492382\n",
      "\t[#  360] train_loss: 1.877979\n",
      "\t[#  540] train_loss: 1.621570\n",
      "\t[#  720] train_loss: 1.689151\n",
      "\t[#  900] train_loss: 1.252065\n",
      "\ttrain_loss: 1.740807\n",
      "\tval_loss: 1.841509\n",
      "Epoch 139/210\n",
      "\t[#    0] train_loss: 1.918398\n",
      "\t[#  180] train_loss: 1.466106\n",
      "\t[#  360] train_loss: 1.644937\n",
      "\t[#  540] train_loss: 2.119476\n",
      "\t[#  720] train_loss: 1.910351\n",
      "\t[#  900] train_loss: 1.758243\n",
      "\ttrain_loss: 1.761668\n",
      "\tval_loss: 1.736555\n",
      "Epoch 140/210\n",
      "\t[#    0] train_loss: 1.402175\n",
      "\t[#  180] train_loss: 1.870861\n",
      "\t[#  360] train_loss: 1.595167\n",
      "\t[#  540] train_loss: 1.813200\n",
      "\t[#  720] train_loss: 1.545974\n",
      "\t[#  900] train_loss: 1.579711\n",
      "\ttrain_loss: 1.856312\n",
      "\tval_loss: 1.800857\n",
      "Epoch 141/210\n",
      "\t[#    0] train_loss: 1.574257\n",
      "\t[#  180] train_loss: 2.231651\n",
      "\t[#  360] train_loss: 1.921935\n",
      "\t[#  540] train_loss: 1.223820\n",
      "\t[#  720] train_loss: 1.908574\n",
      "\t[#  900] train_loss: 1.547090\n",
      "\ttrain_loss: 1.752279\n",
      "\tval_loss: 1.784885\n",
      "Epoch 142/210\n",
      "\t[#    0] train_loss: 2.274664\n",
      "\t[#  180] train_loss: 1.873684\n",
      "\t[#  360] train_loss: 2.332128\n",
      "\t[#  540] train_loss: 2.068326\n",
      "\t[#  720] train_loss: 1.372162\n",
      "\t[#  900] train_loss: 1.390235\n",
      "\ttrain_loss: 1.751675\n",
      "\tval_loss: 1.697835\n",
      "\t*** Personal Best ***\n",
      "Epoch 143/210\n",
      "\t[#    0] train_loss: 1.516534\n",
      "\t[#  180] train_loss: 1.823704\n",
      "\t[#  360] train_loss: 1.592739\n",
      "\t[#  540] train_loss: 1.646781\n",
      "\t[#  720] train_loss: 1.783669\n",
      "\t[#  900] train_loss: 2.339450\n",
      "\ttrain_loss: 1.757660\n",
      "\tval_loss: 1.752843\n",
      "Epoch 144/210\n",
      "\t[#    0] train_loss: 1.956182\n",
      "\t[#  180] train_loss: 2.067989\n",
      "\t[#  360] train_loss: 1.960169\n",
      "\t[#  540] train_loss: 2.222237\n",
      "\t[#  720] train_loss: 1.073587\n",
      "\t[#  900] train_loss: 1.574614\n",
      "\ttrain_loss: 1.707928\n",
      "\tval_loss: 1.748631\n",
      "Epoch 145/210\n",
      "\t[#    0] train_loss: 1.635641\n",
      "\t[#  180] train_loss: 1.441710\n",
      "\t[#  360] train_loss: 2.057420\n",
      "\t[#  540] train_loss: 1.468580\n",
      "\t[#  720] train_loss: 1.759737\n",
      "\t[#  900] train_loss: 6.205117\n",
      "\ttrain_loss: 1.808358\n",
      "\tval_loss: 1.886670\n",
      "Epoch 146/210\n",
      "\t[#    0] train_loss: 2.620660\n",
      "\t[#  180] train_loss: 2.015886\n",
      "\t[#  360] train_loss: 3.168919\n",
      "\t[#  540] train_loss: 1.680852\n",
      "\t[#  720] train_loss: 1.851460\n",
      "\t[#  900] train_loss: 1.465674\n",
      "\ttrain_loss: 1.910297\n",
      "\tval_loss: 1.787735\n",
      "Epoch 147/210\n",
      "\t[#    0] train_loss: 2.005604\n",
      "\t[#  180] train_loss: 1.953333\n",
      "\t[#  360] train_loss: 1.657248\n",
      "\t[#  540] train_loss: 2.315160\n",
      "\t[#  720] train_loss: 1.735093\n",
      "\t[#  900] train_loss: 1.647776\n",
      "\ttrain_loss: 1.734444\n",
      "\tval_loss: 1.738572\n",
      "Epoch 148/210\n",
      "\t[#    0] train_loss: 1.638661\n",
      "\t[#  180] train_loss: 1.742321\n",
      "\t[#  360] train_loss: 2.272345\n",
      "\t[#  540] train_loss: 1.963639\n",
      "\t[#  720] train_loss: 1.411264\n",
      "\t[#  900] train_loss: 1.583300\n",
      "\ttrain_loss: 1.696268\n",
      "\tval_loss: 1.717577\n",
      "Epoch 149/210\n",
      "\t[#    0] train_loss: 1.347474\n",
      "\t[#  180] train_loss: 2.061151\n",
      "\t[#  360] train_loss: 1.725963\n",
      "\t[#  540] train_loss: 1.985033\n",
      "\t[#  720] train_loss: 2.195984\n",
      "\t[#  900] train_loss: 2.049944\n",
      "\ttrain_loss: 1.719945\n",
      "\tval_loss: 1.732162\n",
      "Epoch 150/210\n",
      "\t[#    0] train_loss: 1.718284\n",
      "\t[#  180] train_loss: 1.342875\n",
      "\t[#  360] train_loss: 1.791695\n",
      "\t[#  540] train_loss: 1.908570\n",
      "\t[#  720] train_loss: 2.157586\n",
      "\t[#  900] train_loss: 1.518134\n",
      "\ttrain_loss: 1.724252\n",
      "\tval_loss: 1.758450\n",
      "Epoch 151/210\n",
      "\t[#    0] train_loss: 1.539328\n",
      "\t[#  180] train_loss: 1.954632\n",
      "\t[#  360] train_loss: 1.778599\n",
      "\t[#  540] train_loss: 1.371811\n",
      "\t[#  720] train_loss: 2.203748\n",
      "\t[#  900] train_loss: 1.564594\n",
      "\ttrain_loss: 1.743305\n",
      "\tval_loss: 2.640325\n",
      "Epoch 152/210\n",
      "\t[#    0] train_loss: 3.191070\n",
      "\t[#  180] train_loss: 1.642043\n",
      "\t[#  360] train_loss: 2.156427\n",
      "\t[#  540] train_loss: 1.636846\n",
      "\t[#  720] train_loss: 1.651385\n",
      "\t[#  900] train_loss: 1.818028\n",
      "\ttrain_loss: 1.769691\n",
      "\tval_loss: 1.713664\n",
      "Epoch 153/210\n",
      "\t[#    0] train_loss: 1.605585\n",
      "\t[#  180] train_loss: 1.386392\n",
      "\t[#  360] train_loss: 1.891562\n",
      "\t[#  540] train_loss: 1.326025\n",
      "\t[#  720] train_loss: 1.375234\n",
      "\t[#  900] train_loss: 1.629991\n",
      "\ttrain_loss: 1.698910\n",
      "\tval_loss: 1.710404\n",
      "Epoch 154/210\n",
      "\t[#    0] train_loss: 1.409439\n",
      "\t[#  180] train_loss: 1.336074\n",
      "\t[#  360] train_loss: 1.638585\n",
      "\t[#  540] train_loss: 1.925884\n",
      "\t[#  720] train_loss: 2.305129\n",
      "\t[#  900] train_loss: 1.430291\n",
      "\ttrain_loss: 1.701210\n",
      "\tval_loss: 1.723551\n",
      "Epoch 155/210\n",
      "\t[#    0] train_loss: 1.390908\n",
      "\t[#  180] train_loss: 1.508498\n",
      "\t[#  360] train_loss: 1.344174\n",
      "\t[#  540] train_loss: 1.709140\n",
      "\t[#  720] train_loss: 2.390825\n",
      "\t[#  900] train_loss: 1.340798\n",
      "\ttrain_loss: 1.693350\n",
      "\tval_loss: 1.722375\n",
      "Epoch 156/210\n",
      "\t[#    0] train_loss: 2.319691\n",
      "\t[#  180] train_loss: 1.619413\n",
      "\t[#  360] train_loss: 1.701549\n",
      "\t[#  540] train_loss: 2.031563\n",
      "\t[#  720] train_loss: 1.430470\n",
      "\t[#  900] train_loss: 1.571193\n",
      "\ttrain_loss: 1.701787\n",
      "\tval_loss: 1.699261\n",
      "Epoch 157/210\n",
      "\t[#    0] train_loss: 1.647362\n",
      "\t[#  180] train_loss: 1.310446\n",
      "\t[#  360] train_loss: 1.451550\n",
      "\t[#  540] train_loss: 1.808758\n",
      "\t[#  720] train_loss: 1.773042\n",
      "\t[#  900] train_loss: 2.036036\n",
      "\ttrain_loss: 1.815701\n",
      "\tval_loss: 1.780538\n",
      "Epoch 158/210\n",
      "\t[#    0] train_loss: 1.840647\n",
      "\t[#  180] train_loss: 1.763877\n",
      "\t[#  360] train_loss: 1.581706\n",
      "\t[#  540] train_loss: 1.983229\n",
      "\t[#  720] train_loss: 1.723473\n",
      "\t[#  900] train_loss: 1.900279\n",
      "\ttrain_loss: 1.707923\n",
      "\tval_loss: 1.710580\n",
      "Epoch 159/210\n",
      "\t[#    0] train_loss: 1.424091\n",
      "\t[#  180] train_loss: 1.308509\n",
      "\t[#  360] train_loss: 1.328943\n",
      "\t[#  540] train_loss: 2.054148\n",
      "\t[#  720] train_loss: 1.643373\n",
      "\t[#  900] train_loss: 2.048238\n",
      "\ttrain_loss: 1.688511\n",
      "\tval_loss: 1.679549\n",
      "\t*** Personal Best ***\n",
      "Epoch 160/210\n",
      "\t[#    0] train_loss: 1.610023\n",
      "\t[#  180] train_loss: 1.441442\n",
      "\t[#  360] train_loss: 1.429095\n",
      "\t[#  540] train_loss: 1.348130\n",
      "\t[#  720] train_loss: 1.721621\n",
      "\t[#  900] train_loss: 2.429826\n",
      "\ttrain_loss: 1.769559\n",
      "\tval_loss: 1.919983\n",
      "Epoch 161/210\n",
      "\t[#    0] train_loss: 1.395934\n",
      "\t[#  180] train_loss: 1.443449\n",
      "\t[#  360] train_loss: 1.179320\n",
      "\t[#  540] train_loss: 1.454084\n",
      "\t[#  720] train_loss: 1.742405\n",
      "\t[#  900] train_loss: 1.108070\n",
      "\ttrain_loss: 1.820008\n",
      "\tval_loss: 1.787318\n",
      "Epoch 162/210\n",
      "\t[#    0] train_loss: 1.694214\n",
      "\t[#  180] train_loss: 1.462974\n",
      "\t[#  360] train_loss: 2.027061\n",
      "\t[#  540] train_loss: 1.541722\n",
      "\t[#  720] train_loss: 2.262094\n",
      "\t[#  900] train_loss: 1.245681\n",
      "\ttrain_loss: 1.717837\n",
      "\tval_loss: 1.765163\n",
      "Epoch 163/210\n",
      "\t[#    0] train_loss: 1.480782\n",
      "\t[#  180] train_loss: 1.897097\n",
      "\t[#  360] train_loss: 2.030947\n",
      "\t[#  540] train_loss: 2.148205\n",
      "\t[#  720] train_loss: 1.919818\n",
      "\t[#  900] train_loss: 1.897989\n",
      "\ttrain_loss: 1.802051\n",
      "\tval_loss: 1.690843\n",
      "Epoch 164/210\n",
      "\t[#    0] train_loss: 1.604087\n",
      "\t[#  180] train_loss: 1.809341\n",
      "\t[#  360] train_loss: 1.331864\n",
      "\t[#  540] train_loss: 1.371410\n",
      "\t[#  720] train_loss: 1.759638\n",
      "\t[#  900] train_loss: 1.205891\n",
      "\ttrain_loss: 1.697612\n",
      "\tval_loss: 1.746990\n",
      "Epoch 165/210\n",
      "\t[#    0] train_loss: 1.350132\n",
      "\t[#  180] train_loss: 1.902920\n",
      "\t[#  360] train_loss: 1.254002\n",
      "\t[#  540] train_loss: 1.568721\n",
      "\t[#  720] train_loss: 1.383520\n",
      "\t[#  900] train_loss: 1.868232\n",
      "\ttrain_loss: 1.683568\n",
      "\tval_loss: 1.639281\n",
      "\t*** Personal Best ***\n",
      "Epoch 166/210\n",
      "\t[#    0] train_loss: 2.624826\n",
      "\t[#  180] train_loss: 1.684937\n",
      "\t[#  360] train_loss: 2.032820\n",
      "\t[#  540] train_loss: 1.929346\n",
      "\t[#  720] train_loss: 1.378105\n",
      "\t[#  900] train_loss: 1.388191\n",
      "\ttrain_loss: 1.689111\n",
      "\tval_loss: 1.738256\n",
      "Epoch 167/210\n",
      "\t[#    0] train_loss: 1.760567\n",
      "\t[#  180] train_loss: 1.654960\n",
      "\t[#  360] train_loss: 1.728762\n",
      "\t[#  540] train_loss: 2.068981\n",
      "\t[#  720] train_loss: 1.933904\n",
      "\t[#  900] train_loss: 1.155385\n",
      "\ttrain_loss: 1.666683\n",
      "\tval_loss: 1.682994\n",
      "Epoch 168/210\n",
      "\t[#    0] train_loss: 2.032700\n",
      "\t[#  180] train_loss: 1.148204\n",
      "\t[#  360] train_loss: 1.521716\n",
      "\t[#  540] train_loss: 1.669043\n",
      "\t[#  720] train_loss: 1.754251\n",
      "\t[#  900] train_loss: 1.672968\n",
      "\ttrain_loss: 1.729212\n",
      "\tval_loss: 1.710642\n",
      "Epoch 169/210\n",
      "\t[#    0] train_loss: 1.361986\n",
      "\t[#  180] train_loss: 1.148057\n",
      "\t[#  360] train_loss: 1.492878\n",
      "\t[#  540] train_loss: 1.649183\n",
      "\t[#  720] train_loss: 1.171570\n",
      "\t[#  900] train_loss: 1.708525\n",
      "\ttrain_loss: 1.723659\n",
      "\tval_loss: 1.693794\n",
      "Epoch 170/210\n",
      "\t[#    0] train_loss: 1.236550\n",
      "\t[#  180] train_loss: 1.807645\n",
      "\t[#  360] train_loss: 1.905952\n",
      "\t[#  540] train_loss: 2.078484\n",
      "\t[#  720] train_loss: 1.596524\n",
      "\t[#  900] train_loss: 1.717371\n",
      "\ttrain_loss: 1.669902\n",
      "\tval_loss: 1.635080\n",
      "\t*** Personal Best ***\n",
      "Epoch 171/210\n",
      "\t[#    0] train_loss: 2.008991\n",
      "\t[#  180] train_loss: 1.511087\n",
      "\t[#  360] train_loss: 1.746263\n",
      "\t[#  540] train_loss: 1.576649\n",
      "\t[#  720] train_loss: 1.742110\n",
      "\t[#  900] train_loss: 1.548063\n",
      "\ttrain_loss: 1.701388\n",
      "\tval_loss: 1.855164\n",
      "Epoch 172/210\n",
      "\t[#    0] train_loss: 1.773973\n",
      "\t[#  180] train_loss: 1.554936\n",
      "\t[#  360] train_loss: 1.844723\n",
      "\t[#  540] train_loss: 2.500636\n",
      "\t[#  720] train_loss: 1.705116\n",
      "\t[#  900] train_loss: 1.421953\n",
      "\ttrain_loss: 1.832369\n",
      "\tval_loss: 1.709619\n",
      "Epoch 173/210\n",
      "\t[#    0] train_loss: 1.685722\n",
      "\t[#  180] train_loss: 1.639339\n",
      "\t[#  360] train_loss: 1.700091\n",
      "\t[#  540] train_loss: 2.566360\n",
      "\t[#  720] train_loss: 1.627551\n",
      "\t[#  900] train_loss: 1.574278\n",
      "\ttrain_loss: 1.729908\n",
      "\tval_loss: 1.710974\n",
      "Epoch 174/210\n",
      "\t[#    0] train_loss: 1.438068\n",
      "\t[#  180] train_loss: 1.398618\n",
      "\t[#  360] train_loss: 1.274975\n",
      "\t[#  540] train_loss: 2.080630\n",
      "\t[#  720] train_loss: 1.673995\n",
      "\t[#  900] train_loss: 1.575443\n",
      "\ttrain_loss: 1.712825\n",
      "\tval_loss: 1.690183\n",
      "Epoch 175/210\n",
      "\t[#    0] train_loss: 1.707283\n",
      "\t[#  180] train_loss: 1.418838\n",
      "\t[#  360] train_loss: 1.571367\n",
      "\t[#  540] train_loss: 1.808573\n",
      "\t[#  720] train_loss: 1.260885\n",
      "\t[#  900] train_loss: 1.390883\n",
      "\ttrain_loss: 1.721139\n",
      "\tval_loss: 1.688240\n",
      "Epoch 176/210\n",
      "\t[#    0] train_loss: 1.769029\n",
      "\t[#  180] train_loss: 1.677682\n",
      "\t[#  360] train_loss: 1.434007\n",
      "\t[#  540] train_loss: 1.212576\n",
      "\t[#  720] train_loss: 1.559921\n",
      "\t[#  900] train_loss: 1.661796\n",
      "\ttrain_loss: 1.690674\n",
      "\tval_loss: 1.666591\n",
      "Epoch 177/210\n",
      "\t[#    0] train_loss: 1.771867\n",
      "\t[#  180] train_loss: 1.623122\n",
      "\t[#  360] train_loss: 1.424817\n",
      "\t[#  540] train_loss: 1.619604\n",
      "\t[#  720] train_loss: 1.818149\n",
      "\t[#  900] train_loss: 1.685414\n",
      "\ttrain_loss: 1.674667\n",
      "\tval_loss: 1.661907\n",
      "Epoch 178/210\n",
      "\t[#    0] train_loss: 1.660484\n",
      "\t[#  180] train_loss: 1.622112\n",
      "\t[#  360] train_loss: 1.234794\n",
      "\t[#  540] train_loss: 1.203323\n",
      "\t[#  720] train_loss: 1.835702\n",
      "\t[#  900] train_loss: 1.987475\n",
      "\ttrain_loss: 1.715047\n",
      "\tval_loss: 1.733809\n",
      "Epoch 179/210\n",
      "\t[#    0] train_loss: 1.607317\n",
      "\t[#  180] train_loss: 1.618250\n",
      "\t[#  360] train_loss: 1.602780\n",
      "\t[#  540] train_loss: 1.431923\n",
      "\t[#  720] train_loss: 1.261016\n",
      "\t[#  900] train_loss: 1.783595\n",
      "\ttrain_loss: 1.674626\n",
      "\tval_loss: 1.643705\n",
      "Epoch 180/210\n",
      "\t[#    0] train_loss: 1.322641\n",
      "\t[#  180] train_loss: 1.442841\n",
      "\t[#  360] train_loss: 1.710490\n",
      "\t[#  540] train_loss: 1.541270\n",
      "\t[#  720] train_loss: 1.265229\n",
      "\t[#  900] train_loss: 1.412056\n",
      "\ttrain_loss: 1.635148\n",
      "\tval_loss: 1.769316\n",
      "Epoch 181/210\n",
      "\t[#    0] train_loss: 2.031448\n",
      "\t[#  180] train_loss: 1.947421\n",
      "\t[#  360] train_loss: 2.211653\n",
      "\t[#  540] train_loss: 1.094131\n",
      "\t[#  720] train_loss: 1.529414\n",
      "\t[#  900] train_loss: 1.919815\n",
      "\ttrain_loss: 1.641059\n",
      "\tval_loss: 1.651750\n",
      "Epoch 182/210\n",
      "\t[#    0] train_loss: 1.711138\n",
      "\t[#  180] train_loss: 1.325413\n",
      "\t[#  360] train_loss: 1.608505\n",
      "\t[#  540] train_loss: 1.733324\n",
      "\t[#  720] train_loss: 1.512620\n",
      "\t[#  900] train_loss: 1.944757\n",
      "\ttrain_loss: 1.647743\n",
      "\tval_loss: 1.647711\n",
      "Epoch 183/210\n",
      "\t[#    0] train_loss: 1.726792\n",
      "\t[#  180] train_loss: 1.152284\n",
      "\t[#  360] train_loss: 2.268671\n",
      "\t[#  540] train_loss: 1.728000\n",
      "\t[#  720] train_loss: 1.417468\n",
      "\t[#  900] train_loss: 1.712129\n",
      "\ttrain_loss: 1.643127\n",
      "\tval_loss: 1.674252\n",
      "Epoch 184/210\n",
      "\t[#    0] train_loss: 2.169625\n",
      "\t[#  180] train_loss: 1.290866\n",
      "\t[#  360] train_loss: 1.593925\n",
      "\t[#  540] train_loss: 1.534376\n",
      "\t[#  720] train_loss: 1.678301\n",
      "\t[#  900] train_loss: 1.392216\n",
      "\ttrain_loss: 1.666254\n",
      "\tval_loss: 1.666630\n",
      "Epoch 185/210\n",
      "\t[#    0] train_loss: 1.648447\n",
      "\t[#  180] train_loss: 1.371264\n",
      "\t[#  360] train_loss: 2.069493\n",
      "\t[#  540] train_loss: 1.646047\n",
      "\t[#  720] train_loss: 1.534483\n",
      "\t[#  900] train_loss: 1.550661\n",
      "\ttrain_loss: 1.795916\n",
      "\tval_loss: 1.707776\n",
      "Epoch 186/210\n",
      "\t[#    0] train_loss: 1.692923\n",
      "\t[#  180] train_loss: 1.930266\n",
      "\t[#  360] train_loss: 1.930701\n",
      "\t[#  540] train_loss: 1.381406\n",
      "\t[#  720] train_loss: 1.956287\n",
      "\t[#  900] train_loss: 1.778501\n",
      "\ttrain_loss: 1.681573\n",
      "\tval_loss: 1.762861\n",
      "Epoch 187/210\n",
      "\t[#    0] train_loss: 1.584782\n",
      "\t[#  180] train_loss: 1.639585\n",
      "\t[#  360] train_loss: 1.284394\n",
      "\t[#  540] train_loss: 1.666881\n",
      "\t[#  720] train_loss: 1.847724\n",
      "\t[#  900] train_loss: 1.481168\n",
      "\ttrain_loss: 1.640108\n",
      "\tval_loss: 1.729630\n",
      "Epoch 188/210\n",
      "\t[#    0] train_loss: 1.526096\n",
      "\t[#  180] train_loss: 1.782249\n",
      "\t[#  360] train_loss: 1.808009\n",
      "\t[#  540] train_loss: 1.117897\n",
      "\t[#  720] train_loss: 1.483097\n",
      "\t[#  900] train_loss: 1.196391\n",
      "\ttrain_loss: 1.662278\n",
      "\tval_loss: 1.691561\n",
      "Epoch 189/210\n",
      "\t[#    0] train_loss: 1.952570\n",
      "\t[#  180] train_loss: 2.116829\n",
      "\t[#  360] train_loss: 1.925084\n",
      "\t[#  540] train_loss: 1.405768\n",
      "\t[#  720] train_loss: 1.774545\n",
      "\t[#  900] train_loss: 2.191550\n",
      "\ttrain_loss: 1.690315\n",
      "\tval_loss: 1.750122\n",
      "Epoch 190/210\n",
      "\t[#    0] train_loss: 1.659610\n",
      "\t[#  180] train_loss: 1.395861\n",
      "\t[#  360] train_loss: 2.042323\n",
      "\t[#  540] train_loss: 1.239140\n",
      "\t[#  720] train_loss: 1.366287\n",
      "\t[#  900] train_loss: 1.976684\n",
      "\ttrain_loss: 1.721817\n",
      "\tval_loss: 1.679597\n",
      "Epoch 191/210\n",
      "\t[#    0] train_loss: 1.783694\n",
      "\t[#  180] train_loss: 1.889015\n",
      "\t[#  360] train_loss: 1.477045\n",
      "\t[#  540] train_loss: 1.199385\n",
      "\t[#  720] train_loss: 2.021802\n",
      "\t[#  900] train_loss: 1.309970\n",
      "\ttrain_loss: 1.635872\n",
      "\tval_loss: 1.681618\n",
      "\n",
      "\tStopped by  early stopping!\n"
     ]
    }
   ],
   "source": [
    "model = BaselineNet2().to(device)\n",
    "optimizer = optim.Adam(model.parameters(), lr=LR)\n",
    "\n",
    "h_loss_train = []\n",
    "h_loss_val = []\n",
    "\n",
    "early_stopping = {\"epoch\":0, \"best\":10**3, \"epoch_threshold\":EARLY_STOPPING}\n",
    "\n",
    "for epoch in range(1, EPOCHS + 1):\n",
    "    print(f\"Epoch {epoch}/{EPOCHS}\")\n",
    "\n",
    "    loss_train = train(model, device, train_loader, optimizer, epoch)\n",
    "    loss_val = validation(model, device, val_loader, epoch)\n",
    "\n",
    "    #populating the history of the loss\n",
    "    h_loss_train.append(loss_train)\n",
    "    h_loss_val.append(loss_val)\n",
    "\n",
    "    # early stopping\n",
    "    if loss_val < early_stopping[\"best\"]:\n",
    "        early_stopping[\"best\"] = loss_val\n",
    "        early_stopping[\"epoch\"] = epoch\n",
    "        print(\"\\t*** Personal Best ***\")\n",
    "\n",
    "    if epoch - early_stopping[\"epoch\"] > early_stopping[\"epoch_threshold\"]:\n",
    "        print(\"\\n\\tStopped by  early stopping!\")\n",
    "        break\n",
    "\n",
    "np.savez(f\"train_loss_{EPOCHS}\", h_loss_train)\n",
    "np.savez(f\"val_loss_{EPOCHS}\", h_loss_val)\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plotting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(h_loss_train, label=\"train\")\n",
    "plt.plot(h_loss_val, label=\"validation\")\n",
    "plt.xlabel(\"Epochs\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.legend(loc=\"upper right\")\n",
    "plt.show()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Zoomed plot (last 30 epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.array(h_loss_train)[-50:], label=\"train\")\n",
    "plt.plot(np.array(h_loss_val)[-50:], label=\"validation\")\n",
    "plt.xlabel(\"Epochs\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.legend(loc=\"upper right\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "pytorch_gpu_env",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.9"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "2a60a9521ac24e310983a613c26989eaf1fc2b3e1d7e934fda12fd60cebe70c9"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}