MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "# basic libraries\n",
    "import torch\n",
    "import torchvision\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "# data loading and transforming\n",
    "from torchvision.datasets import MNIST\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision import transforms"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Download dataset from torch database\n",
    "\n",
    "The data can also be augmented here using the transforms variable. Augmentations can include random or uniform crop, brightness, hue, saturation, etc.\n",
    "\n",
    "Both the training data and testing data are loaded and split here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train data, number of images:  60000\n",
      "Test data, number of images:  10000\n"
     ]
    }
   ],
   "source": [
    "transforms = transforms.ToTensor()\n",
    "\n",
    "# Get training and test datasets\n",
    "train_data = MNIST(\"../data\",\n",
    "                   train=True,\n",
    "                   download=True,\n",
    "                   transform=transforms)\n",
    "test_data = MNIST(\"../data\",\n",
    "                  train=False, \n",
    "                  transform=transforms)\n",
    "\n",
    "# Print out some stats about the training and test data\n",
    "print('Train data, number of images: ', len(train_data))\n",
    "print('Test data, number of images: ', len(test_data))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prepare the data\n",
    "\n",
    "This creates batches of data into a `DataLoader` object. The `batch_size` is a hyperparameter than can be tuned, but generally it is a balance of model performance vs hardware performance. Generally it is increased by a factor of two (keeping note of accuracy) until performance or training time suffers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# prepare data loaders, set the batch_size\n",
    "batch_size = 64\n",
    "\n",
    "train_loader = DataLoader(train_data, batch_size=batch_size, shuffle=True)\n",
    "test_loader = DataLoader(test_data, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "# specify the image classes\n",
    "classes = ['zero', 'one', 'two', 'three', 'four', \n",
    "           'five', 'six', 'seven', 'eight', 'nine']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize some training data\n",
    "\n",
    "This cell iterates over the training dataset, loading a random batch of image/label data, using `dataiter.next()`. It then plots the batch of images and labels in a `batch_size/8 x 8` grid."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x1536 with 64 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# obtain one batch of training images\n",
    "dataiter = iter(train_loader)\n",
    "images, labels = dataiter.next()\n",
    "images = images.numpy()\n",
    "\n",
    "# plot the images in the batch, along with the corresponding labels\n",
    "fig = plt.figure(figsize=(25, batch_size/3))\n",
    "plt.rcParams.update({'font.size': 22})\n",
    "for idx in np.arange(batch_size):\n",
    "    ax = fig.add_subplot(batch_size/8, 8, idx+1, xticks=[], yticks=[])\n",
    "    ax.imshow(np.squeeze(images[idx]), cmap='gray')\n",
    "    ax.set_title(classes[labels[idx]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import model\n",
    "\n",
    "There are a few models in `models.py` generally ranging from least to most complex. The model is loaded using `.cuda()`, which assumes hardware acceleration is supported. This is a flag for the model to expect cuda tensors to be found on the GPU instead of CPU memory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Net2(\n",
      "  (conv1): Conv2d(1, 32, kernel_size=(4, 4), stride=(1, 1))\n",
      "  (bn1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (pool1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (dropout1): Dropout(p=0.1)\n",
      "  (conv2): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1))\n",
      "  (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (pool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (dropout2): Dropout(p=0.2)\n",
      "  (conv3): Conv2d(64, 128, kernel_size=(2, 2), stride=(1, 1))\n",
      "  (bn3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (pool3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (dropout3): Dropout(p=0.3)\n",
      "  (fc1): Linear(in_features=512, out_features=1000, bias=True)\n",
      "  (dropout4): Dropout(p=0.5)\n",
      "  (fc2): Linear(in_features=1000, out_features=1000, bias=True)\n",
      "  (dropout5): Dropout(p=0.6)\n",
      "  (fc3): Linear(in_features=1000, out_features=136, bias=True)\n",
      "  (fc4): Linear(in_features=136, out_features=10, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "from models import Net2 as net\n",
    "\n",
    "# Use CUDA version\n",
    "net = net().cuda()\n",
    "print(net)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Set the optimization parameters\n",
    "\n",
    "These are the functions that will be used when training the model. `criterion` is the function that is used to calculate loss, and `optimizer` is the function that will determine how the weights are updated according to the loss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "import torch.nn as nn\n",
    "\n",
    "# cross entropy loss combines softmax and nn.NLLLoss() in one single class.\n",
    "criterion = nn.NLLLoss()\n",
    "\n",
    "# stochastic gradient descent with a small learning rate\n",
    "optimizer = optim.SGD(net.parameters(), lr=0.001)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy before training\n",
    "\n",
    "This cell is mainly to test our network and make sure it is syntactically correct. The images and labels are obtained from each batch in the test_loader, then they are transferred to GPU memory. The predicted labels are found by passing them through the network, and this value is compared to the actual label.\n",
    "\n",
    "It is expected that the performance will be bad here since the model is initialized with random weights. If some sort of transfer learning was necessary, it would be beneficial to ensure the model changes do not negatively affect the initialized results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy before training:  1.0\n"
     ]
    }
   ],
   "source": [
    "# Calculate accuracy before training\n",
    "correct = 0\n",
    "total = 0\n",
    "\n",
    "# Iterate through test dataset\n",
    "for images, labels in test_loader:\n",
    "    images, labels = images.cuda(), labels.cuda()\n",
    "\n",
    "    # forward pass to get outputs\n",
    "    # the outputs are a series of class scores\n",
    "    outputs = net(images)\n",
    "\n",
    "    # get the predicted class from the maximum value in the output-list of class scores\n",
    "    _, predicted = torch.max(outputs.data, 1)\n",
    "\n",
    "    # count up total number of correct labels\n",
    "    # for which the predicted and true labels are equal\n",
    "    total += labels.size(0)\n",
    "    correct += (predicted == labels).sum()\n",
    "\n",
    "# calculate the accuracy\n",
    "# to convert `correct` from a Tensor into a scalar, use .item()\n",
    "accuracy = 100.0 * correct.item() / total\n",
    "\n",
    "# print it out!\n",
    "print('Accuracy before training: ', accuracy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train function\n",
    "\n",
    "This function actually controls the adjusting of weights for the training portion. It first sets the network to train to turn on dropout, batch_norm and other parameters that use randomization to avoid overfitting. It keeps track of the running loss for each batch and the loss over time for visualization purposes (can help avoid overfitting). \n",
    "\n",
    "For each epoch, every batch is passed through the network. For each batch, the gradient is calculated by performing a forward pass, calculating the loss, performing a backwards pass, and updating the weights accordingly. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(n_epochs):\n",
    "    net.train()\n",
    "    loss_over_time = [] # to track the loss as the network trains\n",
    "    \n",
    "    for epoch in range(n_epochs):  # loop over the dataset multiple times\n",
    "        \n",
    "        running_loss = 0.0\n",
    "        \n",
    "        for batch_i, data in enumerate(train_loader):\n",
    "            # get the input images and their corresponding labels\n",
    "            inputs, labels = data\n",
    "            inputs, labels = inputs.cuda(), labels.cuda()\n",
    "\n",
    "            # zero the parameter (weight) gradients\n",
    "            optimizer.zero_grad()\n",
    "\n",
    "            # forward pass to get outputs\n",
    "            outputs = net(inputs)\n",
    "\n",
    "            # calculate the loss\n",
    "            loss = criterion(outputs, labels)\n",
    "\n",
    "            # backward pass to calculate the parameter gradients\n",
    "            loss.backward()\n",
    "\n",
    "            # update the parameters\n",
    "            optimizer.step()\n",
    "\n",
    "            # print loss statistics\n",
    "            # to convert loss into a scalar and add it to running_loss, we use .item()\n",
    "            running_loss += loss.item()\n",
    "            \n",
    "            if batch_i % 100 == 99:    # print every 100 batches\n",
    "                avg_loss = running_loss/100\n",
    "                # record and print the avg loss over the 100 batches\n",
    "                loss_over_time.append(avg_loss)\n",
    "                print('Epoch: {}, Batch: {}, Avg. Loss: {}'.format(epoch + 1, batch_i+1, avg_loss))\n",
    "                running_loss = 0.0\n",
    "\n",
    "    print('Finished Training')\n",
    "    return loss_over_time"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Start training\n",
    "\n",
    "This simply calls the above function with the given number of epochs. A state_dict file can be called to load weights from a previous training section by uncommenting the `net.load_state_dict(torch.load(filename))` line. Note, this needs to be the exact same model architecture.\n",
    "\n",
    "To start, `n_epochs` should be small (<10) to make sure the training function works and the loss is decreasing. If the loss is increasing or negative, there is likely an error in the training function of the model itself. Through each epoch, the loss should decrease, so increasing the numebr of epochs will yield better results and take longer to train. If the number of epochs is too high, the model may be overfitting to the training data. There are techniques to stop the training when the loss appears constant, but that is not implemented here. \n",
    "\n",
    "One technique is to use verification data to compare each epoch results to the training data (saving test data for the end) and stopping training when the training data seems to be fitting more closely, and the verification data is getting worse. In reality, every model trains different, and it is often a function of the complexity of the model vs the complexity of the features. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1, Batch: 100, Avg. Loss: 4.452904884815216\n",
      "Epoch: 1, Batch: 200, Avg. Loss: 3.3944355702400206\n",
      "Epoch: 1, Batch: 300, Avg. Loss: 2.6595975804328917\n",
      "Epoch: 1, Batch: 400, Avg. Loss: 2.40438138961792\n",
      "Epoch: 1, Batch: 500, Avg. Loss: 2.285378303527832\n",
      "Epoch: 1, Batch: 600, Avg. Loss: 2.182758469581604\n",
      "Epoch: 1, Batch: 700, Avg. Loss: 2.0946541106700898\n",
      "Epoch: 1, Batch: 800, Avg. Loss: 2.0295510149002074\n",
      "Epoch: 1, Batch: 900, Avg. Loss: 1.929506126642227\n",
      "Epoch: 2, Batch: 100, Avg. Loss: 1.8017039287090302\n",
      "Epoch: 2, Batch: 200, Avg. Loss: 1.7208531284332276\n",
      "Epoch: 2, Batch: 300, Avg. Loss: 1.6325329458713531\n",
      "Epoch: 2, Batch: 400, Avg. Loss: 1.539190912246704\n",
      "Epoch: 2, Batch: 500, Avg. Loss: 1.4421944475173951\n",
      "Epoch: 2, Batch: 600, Avg. Loss: 1.3691272616386414\n",
      "Epoch: 2, Batch: 700, Avg. Loss: 1.278710606098175\n",
      "Epoch: 2, Batch: 800, Avg. Loss: 1.2198312675952911\n",
      "Epoch: 2, Batch: 900, Avg. Loss: 1.1386039900779723\n",
      "Epoch: 3, Batch: 100, Avg. Loss: 1.0742890560626983\n",
      "Epoch: 3, Batch: 200, Avg. Loss: 0.9991979217529297\n",
      "Epoch: 3, Batch: 300, Avg. Loss: 0.9388084888458252\n",
      "Epoch: 3, Batch: 400, Avg. Loss: 0.9052234816551209\n",
      "Epoch: 3, Batch: 500, Avg. Loss: 0.8504151850938797\n",
      "Epoch: 3, Batch: 600, Avg. Loss: 0.8178828018903732\n",
      "Epoch: 3, Batch: 700, Avg. Loss: 0.7643716758489609\n",
      "Epoch: 3, Batch: 800, Avg. Loss: 0.7538913130760193\n",
      "Epoch: 3, Batch: 900, Avg. Loss: 0.7089526295661926\n",
      "Epoch: 4, Batch: 100, Avg. Loss: 0.6788873100280761\n",
      "Epoch: 4, Batch: 200, Avg. Loss: 0.6521614748239517\n",
      "Epoch: 4, Batch: 300, Avg. Loss: 0.6211563909053802\n",
      "Epoch: 4, Batch: 400, Avg. Loss: 0.5985895574092865\n",
      "Epoch: 4, Batch: 500, Avg. Loss: 0.5862664344906807\n",
      "Epoch: 4, Batch: 600, Avg. Loss: 0.5514826411008835\n",
      "Epoch: 4, Batch: 700, Avg. Loss: 0.5528908106684685\n",
      "Epoch: 4, Batch: 800, Avg. Loss: 0.5390749329328537\n",
      "Epoch: 4, Batch: 900, Avg. Loss: 0.5184370592236519\n",
      "Epoch: 5, Batch: 100, Avg. Loss: 0.4990367770195007\n",
      "Epoch: 5, Batch: 200, Avg. Loss: 0.48198725819587707\n",
      "Epoch: 5, Batch: 300, Avg. Loss: 0.46682755708694457\n",
      "Epoch: 5, Batch: 400, Avg. Loss: 0.4549971607327461\n",
      "Epoch: 5, Batch: 500, Avg. Loss: 0.454181766808033\n",
      "Epoch: 5, Batch: 600, Avg. Loss: 0.4587581020593643\n",
      "Epoch: 5, Batch: 700, Avg. Loss: 0.44925232604146004\n",
      "Epoch: 5, Batch: 800, Avg. Loss: 0.4351923832297325\n",
      "Epoch: 5, Batch: 900, Avg. Loss: 0.4183114337921143\n",
      "Epoch: 6, Batch: 100, Avg. Loss: 0.40680658906698225\n",
      "Epoch: 6, Batch: 200, Avg. Loss: 0.4027291417121887\n",
      "Epoch: 6, Batch: 300, Avg. Loss: 0.4035393799841404\n",
      "Epoch: 6, Batch: 400, Avg. Loss: 0.37699789077043533\n",
      "Epoch: 6, Batch: 500, Avg. Loss: 0.3869954039156437\n",
      "Epoch: 6, Batch: 600, Avg. Loss: 0.38966534614562987\n",
      "Epoch: 6, Batch: 700, Avg. Loss: 0.36866431057453153\n",
      "Epoch: 6, Batch: 800, Avg. Loss: 0.3828067433834076\n",
      "Epoch: 6, Batch: 900, Avg. Loss: 0.369012154340744\n",
      "Epoch: 7, Batch: 100, Avg. Loss: 0.3380245593190193\n",
      "Epoch: 7, Batch: 200, Avg. Loss: 0.35733420073986055\n",
      "Epoch: 7, Batch: 300, Avg. Loss: 0.3540596139431\n",
      "Epoch: 7, Batch: 400, Avg. Loss: 0.33989472962915895\n",
      "Epoch: 7, Batch: 500, Avg. Loss: 0.3317305186390877\n",
      "Epoch: 7, Batch: 600, Avg. Loss: 0.33328957214951516\n",
      "Epoch: 7, Batch: 700, Avg. Loss: 0.3362000568211079\n",
      "Epoch: 7, Batch: 800, Avg. Loss: 0.3255192992091179\n",
      "Epoch: 7, Batch: 900, Avg. Loss: 0.30759395599365236\n",
      "Epoch: 8, Batch: 100, Avg. Loss: 0.31477876864373683\n",
      "Epoch: 8, Batch: 200, Avg. Loss: 0.316655638217926\n",
      "Epoch: 8, Batch: 300, Avg. Loss: 0.3100934347510338\n",
      "Epoch: 8, Batch: 400, Avg. Loss: 0.3133776764571667\n",
      "Epoch: 8, Batch: 500, Avg. Loss: 0.2949063365161419\n",
      "Epoch: 8, Batch: 600, Avg. Loss: 0.29550796836614607\n",
      "Epoch: 8, Batch: 700, Avg. Loss: 0.30229863315820693\n",
      "Epoch: 8, Batch: 800, Avg. Loss: 0.2940409205853939\n",
      "Epoch: 8, Batch: 900, Avg. Loss: 0.3103906278312206\n",
      "Epoch: 9, Batch: 100, Avg. Loss: 0.28861013263463975\n",
      "Epoch: 9, Batch: 200, Avg. Loss: 0.28970234528183936\n",
      "Epoch: 9, Batch: 300, Avg. Loss: 0.2796951450407505\n",
      "Epoch: 9, Batch: 400, Avg. Loss: 0.2834067068994045\n",
      "Epoch: 9, Batch: 500, Avg. Loss: 0.27230313904583453\n",
      "Epoch: 9, Batch: 600, Avg. Loss: 0.2726021483540535\n",
      "Epoch: 9, Batch: 700, Avg. Loss: 0.2736139383912086\n",
      "Epoch: 9, Batch: 800, Avg. Loss: 0.26832761310040953\n",
      "Epoch: 9, Batch: 900, Avg. Loss: 0.2764669622480869\n",
      "Epoch: 10, Batch: 100, Avg. Loss: 0.27974444031715395\n",
      "Epoch: 10, Batch: 200, Avg. Loss: 0.2782619944959879\n",
      "Epoch: 10, Batch: 300, Avg. Loss: 0.2590805070847273\n",
      "Epoch: 10, Batch: 400, Avg. Loss: 0.2523219507187605\n",
      "Epoch: 10, Batch: 500, Avg. Loss: 0.24666364066302776\n",
      "Epoch: 10, Batch: 600, Avg. Loss: 0.24182543516159058\n",
      "Epoch: 10, Batch: 700, Avg. Loss: 0.2555550033599138\n",
      "Epoch: 10, Batch: 800, Avg. Loss: 0.2540617118030786\n",
      "Epoch: 10, Batch: 900, Avg. Loss: 0.25729816876351835\n",
      "Epoch: 11, Batch: 100, Avg. Loss: 0.2529861114919186\n",
      "Epoch: 11, Batch: 200, Avg. Loss: 0.2392948667705059\n",
      "Epoch: 11, Batch: 300, Avg. Loss: 0.24095241621136665\n",
      "Epoch: 11, Batch: 400, Avg. Loss: 0.242050771266222\n",
      "Epoch: 11, Batch: 500, Avg. Loss: 0.2521316257864237\n",
      "Epoch: 11, Batch: 600, Avg. Loss: 0.22758775994181632\n",
      "Epoch: 11, Batch: 700, Avg. Loss: 0.24226700082421304\n",
      "Epoch: 11, Batch: 800, Avg. Loss: 0.24072364218533038\n",
      "Epoch: 11, Batch: 900, Avg. Loss: 0.23857666328549385\n",
      "Epoch: 12, Batch: 100, Avg. Loss: 0.24289526551961899\n",
      "Epoch: 12, Batch: 200, Avg. Loss: 0.2426660969853401\n",
      "Epoch: 12, Batch: 300, Avg. Loss: 0.22953109681606293\n",
      "Epoch: 12, Batch: 400, Avg. Loss: 0.22819897644221782\n",
      "Epoch: 12, Batch: 500, Avg. Loss: 0.2175408274680376\n",
      "Epoch: 12, Batch: 600, Avg. Loss: 0.22997009061276913\n",
      "Epoch: 12, Batch: 700, Avg. Loss: 0.22951605185866356\n",
      "Epoch: 12, Batch: 800, Avg. Loss: 0.24367329448461533\n",
      "Epoch: 12, Batch: 900, Avg. Loss: 0.20336619764566422\n",
      "Epoch: 13, Batch: 100, Avg. Loss: 0.2260273377597332\n",
      "Epoch: 13, Batch: 200, Avg. Loss: 0.22099106453359127\n",
      "Epoch: 13, Batch: 300, Avg. Loss: 0.2153104943037033\n",
      "Epoch: 13, Batch: 400, Avg. Loss: 0.21764347851276397\n",
      "Epoch: 13, Batch: 500, Avg. Loss: 0.2180968426167965\n",
      "Epoch: 13, Batch: 600, Avg. Loss: 0.2163806324452162\n",
      "Epoch: 13, Batch: 700, Avg. Loss: 0.22777654975652695\n",
      "Epoch: 13, Batch: 800, Avg. Loss: 0.21871960379183292\n",
      "Epoch: 13, Batch: 900, Avg. Loss: 0.20940325126051904\n",
      "Epoch: 14, Batch: 100, Avg. Loss: 0.21453174874186515\n",
      "Epoch: 14, Batch: 200, Avg. Loss: 0.20652234233915806\n",
      "Epoch: 14, Batch: 300, Avg. Loss: 0.22011771194636823\n",
      "Epoch: 14, Batch: 400, Avg. Loss: 0.2254501624405384\n",
      "Epoch: 14, Batch: 500, Avg. Loss: 0.2030575006455183\n",
      "Epoch: 14, Batch: 600, Avg. Loss: 0.19624248504638672\n",
      "Epoch: 14, Batch: 700, Avg. Loss: 0.2130947780609131\n",
      "Epoch: 14, Batch: 800, Avg. Loss: 0.20936587005853652\n",
      "Epoch: 14, Batch: 900, Avg. Loss: 0.21401029519736767\n",
      "Epoch: 15, Batch: 100, Avg. Loss: 0.20410148963332175\n",
      "Epoch: 15, Batch: 200, Avg. Loss: 0.20620932929217817\n",
      "Epoch: 15, Batch: 300, Avg. Loss: 0.2052878015488386\n",
      "Epoch: 15, Batch: 400, Avg. Loss: 0.19996532700955869\n",
      "Epoch: 15, Batch: 500, Avg. Loss: 0.20589647941291334\n",
      "Epoch: 15, Batch: 600, Avg. Loss: 0.20019807338714599\n",
      "Epoch: 15, Batch: 700, Avg. Loss: 0.1957202360779047\n",
      "Epoch: 15, Batch: 800, Avg. Loss: 0.1877109459042549\n",
      "Epoch: 15, Batch: 900, Avg. Loss: 0.19291630811989308\n",
      "Epoch: 16, Batch: 100, Avg. Loss: 0.19868743434548378\n",
      "Epoch: 16, Batch: 200, Avg. Loss: 0.19360115751624107\n",
      "Epoch: 16, Batch: 300, Avg. Loss: 0.1923595830798149\n",
      "Epoch: 16, Batch: 400, Avg. Loss: 0.18169549837708474\n",
      "Epoch: 16, Batch: 500, Avg. Loss: 0.2056161390990019\n",
      "Epoch: 16, Batch: 600, Avg. Loss: 0.19033964447677135\n",
      "Epoch: 16, Batch: 700, Avg. Loss: 0.1940733204036951\n",
      "Epoch: 16, Batch: 800, Avg. Loss: 0.1977187830209732\n",
      "Epoch: 16, Batch: 900, Avg. Loss: 0.19437365286052227\n",
      "Epoch: 17, Batch: 100, Avg. Loss: 0.1892612384259701\n",
      "Epoch: 17, Batch: 200, Avg. Loss: 0.1935679444670677\n",
      "Epoch: 17, Batch: 300, Avg. Loss: 0.2025440240651369\n",
      "Epoch: 17, Batch: 400, Avg. Loss: 0.18561015188694\n",
      "Epoch: 17, Batch: 500, Avg. Loss: 0.1811131574958563\n",
      "Epoch: 17, Batch: 600, Avg. Loss: 0.20191813096404077\n",
      "Epoch: 17, Batch: 700, Avg. Loss: 0.1899220993369818\n",
      "Epoch: 17, Batch: 800, Avg. Loss: 0.17432502694427968\n",
      "Epoch: 17, Batch: 900, Avg. Loss: 0.18778503194451332\n",
      "Epoch: 18, Batch: 100, Avg. Loss: 0.19076591782271862\n",
      "Epoch: 18, Batch: 200, Avg. Loss: 0.17402383878827096\n",
      "Epoch: 18, Batch: 300, Avg. Loss: 0.18372008696198464\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 18, Batch: 400, Avg. Loss: 0.1864746291190386\n",
      "Epoch: 18, Batch: 500, Avg. Loss: 0.17618222303688527\n",
      "Epoch: 18, Batch: 600, Avg. Loss: 0.18813206970691682\n",
      "Epoch: 18, Batch: 700, Avg. Loss: 0.16496420614421367\n",
      "Epoch: 18, Batch: 800, Avg. Loss: 0.18336844056844712\n",
      "Epoch: 18, Batch: 900, Avg. Loss: 0.17272150136530398\n",
      "Epoch: 19, Batch: 100, Avg. Loss: 0.17566022127866746\n",
      "Epoch: 19, Batch: 200, Avg. Loss: 0.1893272802978754\n",
      "Epoch: 19, Batch: 300, Avg. Loss: 0.17954964727163314\n",
      "Epoch: 19, Batch: 400, Avg. Loss: 0.1721625320613384\n",
      "Epoch: 19, Batch: 500, Avg. Loss: 0.1761695421487093\n",
      "Epoch: 19, Batch: 600, Avg. Loss: 0.17001292370259763\n",
      "Epoch: 19, Batch: 700, Avg. Loss: 0.1858087360858917\n",
      "Epoch: 19, Batch: 800, Avg. Loss: 0.1732622457295656\n",
      "Epoch: 19, Batch: 900, Avg. Loss: 0.16919793367385863\n",
      "Epoch: 20, Batch: 100, Avg. Loss: 0.16712391324341297\n",
      "Epoch: 20, Batch: 200, Avg. Loss: 0.1614630115032196\n",
      "Epoch: 20, Batch: 300, Avg. Loss: 0.1876898258179426\n",
      "Epoch: 20, Batch: 400, Avg. Loss: 0.15716021560132504\n",
      "Epoch: 20, Batch: 500, Avg. Loss: 0.18381944082677365\n",
      "Epoch: 20, Batch: 600, Avg. Loss: 0.1614443685859442\n",
      "Epoch: 20, Batch: 700, Avg. Loss: 0.18437214158475398\n",
      "Epoch: 20, Batch: 800, Avg. Loss: 0.17112751580774785\n",
      "Epoch: 20, Batch: 900, Avg. Loss: 0.16168083801865576\n",
      "Epoch: 21, Batch: 100, Avg. Loss: 0.16831272326409816\n",
      "Epoch: 21, Batch: 200, Avg. Loss: 0.16806177884340287\n",
      "Epoch: 21, Batch: 300, Avg. Loss: 0.16588120579719542\n",
      "Epoch: 21, Batch: 400, Avg. Loss: 0.1609013494849205\n",
      "Epoch: 21, Batch: 500, Avg. Loss: 0.17262725345790386\n",
      "Epoch: 21, Batch: 600, Avg. Loss: 0.16798861645162105\n",
      "Epoch: 21, Batch: 700, Avg. Loss: 0.1601853221654892\n",
      "Epoch: 21, Batch: 800, Avg. Loss: 0.17750794917345047\n",
      "Epoch: 21, Batch: 900, Avg. Loss: 0.16879391632974147\n",
      "Epoch: 22, Batch: 100, Avg. Loss: 0.16533922456204891\n",
      "Epoch: 22, Batch: 200, Avg. Loss: 0.15851567052304744\n",
      "Epoch: 22, Batch: 300, Avg. Loss: 0.16292994938790797\n",
      "Epoch: 22, Batch: 400, Avg. Loss: 0.1698658771812916\n",
      "Epoch: 22, Batch: 500, Avg. Loss: 0.16446180403232574\n",
      "Epoch: 22, Batch: 600, Avg. Loss: 0.15663994885981083\n",
      "Epoch: 22, Batch: 700, Avg. Loss: 0.1617712099850178\n",
      "Epoch: 22, Batch: 800, Avg. Loss: 0.16517185285687447\n",
      "Epoch: 22, Batch: 900, Avg. Loss: 0.1764623372256756\n",
      "Epoch: 23, Batch: 100, Avg. Loss: 0.1707948675751686\n",
      "Epoch: 23, Batch: 200, Avg. Loss: 0.17154281571507454\n",
      "Epoch: 23, Batch: 300, Avg. Loss: 0.15515071511268616\n",
      "Epoch: 23, Batch: 400, Avg. Loss: 0.1581327050179243\n",
      "Epoch: 23, Batch: 500, Avg. Loss: 0.15648425728082657\n",
      "Epoch: 23, Batch: 600, Avg. Loss: 0.15857239499688147\n",
      "Epoch: 23, Batch: 700, Avg. Loss: 0.149945802167058\n",
      "Epoch: 23, Batch: 800, Avg. Loss: 0.16912967592477798\n",
      "Epoch: 23, Batch: 900, Avg. Loss: 0.16099749863147736\n",
      "Epoch: 24, Batch: 100, Avg. Loss: 0.16464153371751308\n",
      "Epoch: 24, Batch: 200, Avg. Loss: 0.1621289436519146\n",
      "Epoch: 24, Batch: 300, Avg. Loss: 0.16139940224587918\n",
      "Epoch: 24, Batch: 400, Avg. Loss: 0.1585185608267784\n",
      "Epoch: 24, Batch: 500, Avg. Loss: 0.15187773905694485\n",
      "Epoch: 24, Batch: 600, Avg. Loss: 0.15333766497671605\n",
      "Epoch: 24, Batch: 700, Avg. Loss: 0.154221992790699\n",
      "Epoch: 24, Batch: 800, Avg. Loss: 0.15167732797563077\n",
      "Epoch: 24, Batch: 900, Avg. Loss: 0.163991072550416\n",
      "Epoch: 25, Batch: 100, Avg. Loss: 0.1618982558697462\n",
      "Epoch: 25, Batch: 200, Avg. Loss: 0.1498812051862478\n",
      "Epoch: 25, Batch: 300, Avg. Loss: 0.16220719307661058\n",
      "Epoch: 25, Batch: 400, Avg. Loss: 0.16233924880623818\n",
      "Epoch: 25, Batch: 500, Avg. Loss: 0.16246338836848737\n",
      "Epoch: 25, Batch: 600, Avg. Loss: 0.15608014591038227\n",
      "Epoch: 25, Batch: 700, Avg. Loss: 0.14377227306365967\n",
      "Epoch: 25, Batch: 800, Avg. Loss: 0.15703944310545923\n",
      "Epoch: 25, Batch: 900, Avg. Loss: 0.14161856204271317\n",
      "Epoch: 26, Batch: 100, Avg. Loss: 0.15923247516155242\n",
      "Epoch: 26, Batch: 200, Avg. Loss: 0.1492968063056469\n",
      "Epoch: 26, Batch: 300, Avg. Loss: 0.14244347654283046\n",
      "Epoch: 26, Batch: 400, Avg. Loss: 0.15216007009148597\n",
      "Epoch: 26, Batch: 500, Avg. Loss: 0.15078130371868612\n",
      "Epoch: 26, Batch: 600, Avg. Loss: 0.15600503854453562\n",
      "Epoch: 26, Batch: 700, Avg. Loss: 0.14196600325405598\n",
      "Epoch: 26, Batch: 800, Avg. Loss: 0.14397292889654636\n",
      "Epoch: 26, Batch: 900, Avg. Loss: 0.15067409425973893\n",
      "Epoch: 27, Batch: 100, Avg. Loss: 0.17097903579473495\n",
      "Epoch: 27, Batch: 200, Avg. Loss: 0.15548818089067937\n",
      "Epoch: 27, Batch: 300, Avg. Loss: 0.12852534785866737\n",
      "Epoch: 27, Batch: 400, Avg. Loss: 0.1414043530076742\n",
      "Epoch: 27, Batch: 500, Avg. Loss: 0.15033156417310237\n",
      "Epoch: 27, Batch: 600, Avg. Loss: 0.14744801111519337\n",
      "Epoch: 27, Batch: 700, Avg. Loss: 0.13959409572184087\n",
      "Epoch: 27, Batch: 800, Avg. Loss: 0.1601352258771658\n",
      "Epoch: 27, Batch: 900, Avg. Loss: 0.1321632868796587\n",
      "Epoch: 28, Batch: 100, Avg. Loss: 0.1366673044115305\n",
      "Epoch: 28, Batch: 200, Avg. Loss: 0.15824978917837143\n",
      "Epoch: 28, Batch: 300, Avg. Loss: 0.14265555389225482\n",
      "Epoch: 28, Batch: 400, Avg. Loss: 0.14770953327417374\n",
      "Epoch: 28, Batch: 500, Avg. Loss: 0.14579518295824528\n",
      "Epoch: 28, Batch: 600, Avg. Loss: 0.1507534620165825\n",
      "Epoch: 28, Batch: 700, Avg. Loss: 0.14006050705909728\n",
      "Epoch: 28, Batch: 800, Avg. Loss: 0.15021577320992946\n",
      "Epoch: 28, Batch: 900, Avg. Loss: 0.13525341354310513\n",
      "Epoch: 29, Batch: 100, Avg. Loss: 0.14816692002117635\n",
      "Epoch: 29, Batch: 200, Avg. Loss: 0.1438108717650175\n",
      "Epoch: 29, Batch: 300, Avg. Loss: 0.14900452971458436\n",
      "Epoch: 29, Batch: 400, Avg. Loss: 0.14772954158484936\n",
      "Epoch: 29, Batch: 500, Avg. Loss: 0.13695890679955483\n",
      "Epoch: 29, Batch: 600, Avg. Loss: 0.1474641691148281\n",
      "Epoch: 29, Batch: 700, Avg. Loss: 0.13873767480254173\n",
      "Epoch: 29, Batch: 800, Avg. Loss: 0.14908418156206607\n",
      "Epoch: 29, Batch: 900, Avg. Loss: 0.13620849885046482\n",
      "Epoch: 30, Batch: 100, Avg. Loss: 0.1477499832957983\n",
      "Epoch: 30, Batch: 200, Avg. Loss: 0.1357337299734354\n",
      "Epoch: 30, Batch: 300, Avg. Loss: 0.14688115656375886\n",
      "Epoch: 30, Batch: 400, Avg. Loss: 0.13071004286408425\n",
      "Epoch: 30, Batch: 500, Avg. Loss: 0.13655042625963687\n",
      "Epoch: 30, Batch: 600, Avg. Loss: 0.13592492938041686\n",
      "Epoch: 30, Batch: 700, Avg. Loss: 0.13885549508035183\n",
      "Epoch: 30, Batch: 800, Avg. Loss: 0.1417798902094364\n",
      "Epoch: 30, Batch: 900, Avg. Loss: 0.13373843535780908\n",
      "Epoch: 31, Batch: 100, Avg. Loss: 0.1259208633005619\n",
      "Epoch: 31, Batch: 200, Avg. Loss: 0.13814419880509377\n",
      "Epoch: 31, Batch: 300, Avg. Loss: 0.13522887475788592\n",
      "Epoch: 31, Batch: 400, Avg. Loss: 0.1532077670097351\n",
      "Epoch: 31, Batch: 500, Avg. Loss: 0.1523856393992901\n",
      "Epoch: 31, Batch: 600, Avg. Loss: 0.13379419803619386\n",
      "Epoch: 31, Batch: 700, Avg. Loss: 0.14250969216227533\n",
      "Epoch: 31, Batch: 800, Avg. Loss: 0.1435604226589203\n",
      "Epoch: 31, Batch: 900, Avg. Loss: 0.13427016891539098\n",
      "Epoch: 32, Batch: 100, Avg. Loss: 0.14798625789582728\n",
      "Epoch: 32, Batch: 200, Avg. Loss: 0.1332807829231024\n",
      "Epoch: 32, Batch: 300, Avg. Loss: 0.136566424369812\n",
      "Epoch: 32, Batch: 400, Avg. Loss: 0.12889219969511032\n",
      "Epoch: 32, Batch: 500, Avg. Loss: 0.12867543801665307\n",
      "Epoch: 32, Batch: 600, Avg. Loss: 0.13162265181541444\n",
      "Epoch: 32, Batch: 700, Avg. Loss: 0.14118993379175662\n",
      "Epoch: 32, Batch: 800, Avg. Loss: 0.1258566030114889\n",
      "Epoch: 32, Batch: 900, Avg. Loss: 0.14743091113865375\n",
      "Epoch: 33, Batch: 100, Avg. Loss: 0.12471259720623493\n",
      "Epoch: 33, Batch: 200, Avg. Loss: 0.1463499641418457\n",
      "Epoch: 33, Batch: 300, Avg. Loss: 0.13412920333445072\n",
      "Epoch: 33, Batch: 400, Avg. Loss: 0.12410645119845867\n",
      "Epoch: 33, Batch: 500, Avg. Loss: 0.13247244931757451\n",
      "Epoch: 33, Batch: 600, Avg. Loss: 0.13929821155965327\n",
      "Epoch: 33, Batch: 700, Avg. Loss: 0.13672243773937226\n",
      "Epoch: 33, Batch: 800, Avg. Loss: 0.1330166294425726\n",
      "Epoch: 33, Batch: 900, Avg. Loss: 0.12973139591515065\n",
      "Epoch: 34, Batch: 100, Avg. Loss: 0.13074936158955097\n",
      "Epoch: 34, Batch: 200, Avg. Loss: 0.13513037897646427\n",
      "Epoch: 34, Batch: 300, Avg. Loss: 0.1328313621878624\n",
      "Epoch: 34, Batch: 400, Avg. Loss: 0.129883688762784\n",
      "Epoch: 34, Batch: 500, Avg. Loss: 0.13890107572078705\n",
      "Epoch: 34, Batch: 600, Avg. Loss: 0.137378611266613\n",
      "Epoch: 34, Batch: 700, Avg. Loss: 0.1349439709633589\n",
      "Epoch: 34, Batch: 800, Avg. Loss: 0.1299125587195158\n",
      "Epoch: 34, Batch: 900, Avg. Loss: 0.13882759176194667\n",
      "Epoch: 35, Batch: 100, Avg. Loss: 0.14046737149357796\n",
      "Epoch: 35, Batch: 200, Avg. Loss: 0.13902981013059615\n",
      "Epoch: 35, Batch: 300, Avg. Loss: 0.13034107856452465\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 35, Batch: 400, Avg. Loss: 0.13103775069117546\n",
      "Epoch: 35, Batch: 500, Avg. Loss: 0.13081665761768818\n",
      "Epoch: 35, Batch: 600, Avg. Loss: 0.12897710256278516\n",
      "Epoch: 35, Batch: 700, Avg. Loss: 0.13226525858044624\n",
      "Epoch: 35, Batch: 800, Avg. Loss: 0.13663984715938568\n",
      "Epoch: 35, Batch: 900, Avg. Loss: 0.13361125573515892\n",
      "Epoch: 36, Batch: 100, Avg. Loss: 0.12189697697758675\n",
      "Epoch: 36, Batch: 200, Avg. Loss: 0.13141385197639466\n",
      "Epoch: 36, Batch: 300, Avg. Loss: 0.12843492768704892\n",
      "Epoch: 36, Batch: 400, Avg. Loss: 0.11375123031437397\n",
      "Epoch: 36, Batch: 500, Avg. Loss: 0.13705809988081455\n",
      "Epoch: 36, Batch: 600, Avg. Loss: 0.13111661188304424\n",
      "Epoch: 36, Batch: 700, Avg. Loss: 0.1486022900789976\n",
      "Epoch: 36, Batch: 800, Avg. Loss: 0.12838820666074752\n",
      "Epoch: 36, Batch: 900, Avg. Loss: 0.12321610890328884\n",
      "Epoch: 37, Batch: 100, Avg. Loss: 0.1361247768253088\n",
      "Epoch: 37, Batch: 200, Avg. Loss: 0.12874410748481752\n",
      "Epoch: 37, Batch: 300, Avg. Loss: 0.13993053577840328\n",
      "Epoch: 37, Batch: 400, Avg. Loss: 0.11543634787201881\n",
      "Epoch: 37, Batch: 500, Avg. Loss: 0.1314834164083004\n",
      "Epoch: 37, Batch: 600, Avg. Loss: 0.12060063607990741\n",
      "Epoch: 37, Batch: 700, Avg. Loss: 0.12840520471334457\n",
      "Epoch: 37, Batch: 800, Avg. Loss: 0.12313172303140163\n",
      "Epoch: 37, Batch: 900, Avg. Loss: 0.12397181116044521\n",
      "Epoch: 38, Batch: 100, Avg. Loss: 0.11437939792871475\n",
      "Epoch: 38, Batch: 200, Avg. Loss: 0.12593537665903567\n",
      "Epoch: 38, Batch: 300, Avg. Loss: 0.1301279465854168\n",
      "Epoch: 38, Batch: 400, Avg. Loss: 0.11725646026432514\n",
      "Epoch: 38, Batch: 500, Avg. Loss: 0.12658131502568723\n",
      "Epoch: 38, Batch: 600, Avg. Loss: 0.14241009682416916\n",
      "Epoch: 38, Batch: 700, Avg. Loss: 0.13128772825002671\n",
      "Epoch: 38, Batch: 800, Avg. Loss: 0.11158737063407898\n",
      "Epoch: 38, Batch: 900, Avg. Loss: 0.12625795632600784\n",
      "Epoch: 39, Batch: 100, Avg. Loss: 0.1167235454916954\n",
      "Epoch: 39, Batch: 200, Avg. Loss: 0.12432456195354462\n",
      "Epoch: 39, Batch: 300, Avg. Loss: 0.1287697445601225\n",
      "Epoch: 39, Batch: 400, Avg. Loss: 0.1296608105301857\n",
      "Epoch: 39, Batch: 500, Avg. Loss: 0.1214782327413559\n",
      "Epoch: 39, Batch: 600, Avg. Loss: 0.13256143063306808\n",
      "Epoch: 39, Batch: 700, Avg. Loss: 0.11645789124071598\n",
      "Epoch: 39, Batch: 800, Avg. Loss: 0.13325499027967452\n",
      "Epoch: 39, Batch: 900, Avg. Loss: 0.12356514997780323\n",
      "Epoch: 40, Batch: 100, Avg. Loss: 0.11823084190487862\n",
      "Epoch: 40, Batch: 200, Avg. Loss: 0.12409479342401028\n",
      "Epoch: 40, Batch: 300, Avg. Loss: 0.11599623121321201\n",
      "Epoch: 40, Batch: 400, Avg. Loss: 0.12673843197524548\n",
      "Epoch: 40, Batch: 500, Avg. Loss: 0.12248835548758506\n",
      "Epoch: 40, Batch: 600, Avg. Loss: 0.12810735933482648\n",
      "Epoch: 40, Batch: 700, Avg. Loss: 0.11544319912791252\n",
      "Epoch: 40, Batch: 800, Avg. Loss: 0.11254662223160267\n",
      "Epoch: 40, Batch: 900, Avg. Loss: 0.12263120196759701\n",
      "Epoch: 41, Batch: 100, Avg. Loss: 0.11445234797894954\n",
      "Epoch: 41, Batch: 200, Avg. Loss: 0.1271531792730093\n",
      "Epoch: 41, Batch: 300, Avg. Loss: 0.11446333207190036\n",
      "Epoch: 41, Batch: 400, Avg. Loss: 0.12595909424126148\n",
      "Epoch: 41, Batch: 500, Avg. Loss: 0.1323857568949461\n",
      "Epoch: 41, Batch: 600, Avg. Loss: 0.10588117726147175\n",
      "Epoch: 41, Batch: 700, Avg. Loss: 0.12439990922808647\n",
      "Epoch: 41, Batch: 800, Avg. Loss: 0.11449366964399815\n",
      "Epoch: 41, Batch: 900, Avg. Loss: 0.11858232110738755\n",
      "Epoch: 42, Batch: 100, Avg. Loss: 0.12074431903660297\n",
      "Epoch: 42, Batch: 200, Avg. Loss: 0.12196481108665466\n",
      "Epoch: 42, Batch: 300, Avg. Loss: 0.11783020786941051\n",
      "Epoch: 42, Batch: 400, Avg. Loss: 0.1149186272919178\n",
      "Epoch: 42, Batch: 500, Avg. Loss: 0.12671395242214203\n",
      "Epoch: 42, Batch: 600, Avg. Loss: 0.12241858288645745\n",
      "Epoch: 42, Batch: 700, Avg. Loss: 0.12191286958754062\n",
      "Epoch: 42, Batch: 800, Avg. Loss: 0.11775115646421909\n",
      "Epoch: 42, Batch: 900, Avg. Loss: 0.11464918330311775\n",
      "Epoch: 43, Batch: 100, Avg. Loss: 0.12904856815934182\n",
      "Epoch: 43, Batch: 200, Avg. Loss: 0.12854594871401787\n",
      "Epoch: 43, Batch: 300, Avg. Loss: 0.11999309502542019\n",
      "Epoch: 43, Batch: 400, Avg. Loss: 0.12042133443057537\n",
      "Epoch: 43, Batch: 500, Avg. Loss: 0.11243280477821826\n",
      "Epoch: 43, Batch: 600, Avg. Loss: 0.10523272030055524\n",
      "Epoch: 43, Batch: 700, Avg. Loss: 0.11700047589838505\n",
      "Epoch: 43, Batch: 800, Avg. Loss: 0.11671417623758316\n",
      "Epoch: 43, Batch: 900, Avg. Loss: 0.11445893727242946\n",
      "Epoch: 44, Batch: 100, Avg. Loss: 0.11195388004183769\n",
      "Epoch: 44, Batch: 200, Avg. Loss: 0.1043837796151638\n",
      "Epoch: 44, Batch: 300, Avg. Loss: 0.12047468312084675\n",
      "Epoch: 44, Batch: 400, Avg. Loss: 0.11795284919440746\n",
      "Epoch: 44, Batch: 500, Avg. Loss: 0.1211869066953659\n",
      "Epoch: 44, Batch: 600, Avg. Loss: 0.12474909380078315\n",
      "Epoch: 44, Batch: 700, Avg. Loss: 0.11989805035293102\n",
      "Epoch: 44, Batch: 800, Avg. Loss: 0.12002933204174042\n",
      "Epoch: 44, Batch: 900, Avg. Loss: 0.10613567098975181\n",
      "Epoch: 45, Batch: 100, Avg. Loss: 0.11188558258116245\n",
      "Epoch: 45, Batch: 200, Avg. Loss: 0.11409630745649338\n",
      "Epoch: 45, Batch: 300, Avg. Loss: 0.10728642024099827\n",
      "Epoch: 45, Batch: 400, Avg. Loss: 0.1165685173124075\n",
      "Epoch: 45, Batch: 500, Avg. Loss: 0.11428305312991142\n",
      "Epoch: 45, Batch: 600, Avg. Loss: 0.11458631068468093\n",
      "Epoch: 45, Batch: 700, Avg. Loss: 0.11813215896487236\n",
      "Epoch: 45, Batch: 800, Avg. Loss: 0.10502526670694351\n",
      "Epoch: 45, Batch: 900, Avg. Loss: 0.11955584734678268\n",
      "Epoch: 46, Batch: 100, Avg. Loss: 0.10285226069390774\n",
      "Epoch: 46, Batch: 200, Avg. Loss: 0.12352134816348553\n",
      "Epoch: 46, Batch: 300, Avg. Loss: 0.11420834317803383\n",
      "Epoch: 46, Batch: 400, Avg. Loss: 0.10767446078360081\n",
      "Epoch: 46, Batch: 500, Avg. Loss: 0.11461737811565399\n",
      "Epoch: 46, Batch: 600, Avg. Loss: 0.11833911545574666\n",
      "Epoch: 46, Batch: 700, Avg. Loss: 0.1261049474775791\n",
      "Epoch: 46, Batch: 800, Avg. Loss: 0.10840509705245495\n",
      "Epoch: 46, Batch: 900, Avg. Loss: 0.11118643999099731\n",
      "Epoch: 47, Batch: 100, Avg. Loss: 0.11718692556023598\n",
      "Epoch: 47, Batch: 200, Avg. Loss: 0.11197028286755085\n",
      "Epoch: 47, Batch: 300, Avg. Loss: 0.11293893165886403\n",
      "Epoch: 47, Batch: 400, Avg. Loss: 0.10638359189033508\n",
      "Epoch: 47, Batch: 500, Avg. Loss: 0.11416540496051311\n",
      "Epoch: 47, Batch: 600, Avg. Loss: 0.11902564860880376\n",
      "Epoch: 47, Batch: 700, Avg. Loss: 0.11596152625977993\n",
      "Epoch: 47, Batch: 800, Avg. Loss: 0.11625082485377788\n",
      "Epoch: 47, Batch: 900, Avg. Loss: 0.1001155460625887\n",
      "Epoch: 48, Batch: 100, Avg. Loss: 0.11578108161687851\n",
      "Epoch: 48, Batch: 200, Avg. Loss: 0.10090942651033402\n",
      "Epoch: 48, Batch: 300, Avg. Loss: 0.09811174497008324\n",
      "Epoch: 48, Batch: 400, Avg. Loss: 0.10636496610939503\n",
      "Epoch: 48, Batch: 500, Avg. Loss: 0.12197919882833957\n",
      "Epoch: 48, Batch: 600, Avg. Loss: 0.12025030173361301\n",
      "Epoch: 48, Batch: 700, Avg. Loss: 0.12287114910781384\n",
      "Epoch: 48, Batch: 800, Avg. Loss: 0.11880241207778454\n",
      "Epoch: 48, Batch: 900, Avg. Loss: 0.10425219409167767\n",
      "Epoch: 49, Batch: 100, Avg. Loss: 0.10109337016940118\n",
      "Epoch: 49, Batch: 200, Avg. Loss: 0.10089979268610477\n",
      "Epoch: 49, Batch: 300, Avg. Loss: 0.10261613272130489\n",
      "Epoch: 49, Batch: 400, Avg. Loss: 0.1131567656993866\n",
      "Epoch: 49, Batch: 500, Avg. Loss: 0.11347507759928703\n",
      "Epoch: 49, Batch: 600, Avg. Loss: 0.11531580321490764\n",
      "Epoch: 49, Batch: 700, Avg. Loss: 0.11212289705872536\n",
      "Epoch: 49, Batch: 800, Avg. Loss: 0.1174753212928772\n",
      "Epoch: 49, Batch: 900, Avg. Loss: 0.09906448021531106\n",
      "Epoch: 50, Batch: 100, Avg. Loss: 0.10396762147545814\n",
      "Epoch: 50, Batch: 200, Avg. Loss: 0.10459547467529774\n",
      "Epoch: 50, Batch: 300, Avg. Loss: 0.09927496328949928\n",
      "Epoch: 50, Batch: 400, Avg. Loss: 0.11100475117564201\n",
      "Epoch: 50, Batch: 500, Avg. Loss: 0.1186522275954485\n",
      "Epoch: 50, Batch: 600, Avg. Loss: 0.11237650334835053\n",
      "Epoch: 50, Batch: 700, Avg. Loss: 0.10119534388184548\n",
      "Epoch: 50, Batch: 800, Avg. Loss: 0.10981936283409595\n",
      "Epoch: 50, Batch: 900, Avg. Loss: 0.10995014697313309\n",
      "Epoch: 51, Batch: 100, Avg. Loss: 0.11109675385057927\n",
      "Epoch: 51, Batch: 200, Avg. Loss: 0.11570743270218373\n",
      "Epoch: 51, Batch: 300, Avg. Loss: 0.10987597398459911\n",
      "Epoch: 51, Batch: 400, Avg. Loss: 0.10939781241118908\n",
      "Epoch: 51, Batch: 500, Avg. Loss: 0.10479812979698182\n",
      "Epoch: 51, Batch: 600, Avg. Loss: 0.10832008242607116\n",
      "Epoch: 51, Batch: 700, Avg. Loss: 0.1084766735136509\n",
      "Epoch: 51, Batch: 800, Avg. Loss: 0.09662243135273457\n",
      "Epoch: 51, Batch: 900, Avg. Loss: 0.09631622403860092\n",
      "Epoch: 52, Batch: 100, Avg. Loss: 0.11024138309061528\n",
      "Epoch: 52, Batch: 200, Avg. Loss: 0.10415311112999916\n",
      "Epoch: 52, Batch: 300, Avg. Loss: 0.11644300736486912\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 52, Batch: 400, Avg. Loss: 0.11599929720163345\n",
      "Epoch: 52, Batch: 500, Avg. Loss: 0.09752122640609741\n",
      "Epoch: 52, Batch: 600, Avg. Loss: 0.09038237519562245\n",
      "Epoch: 52, Batch: 700, Avg. Loss: 0.1054922579973936\n",
      "Epoch: 52, Batch: 800, Avg. Loss: 0.10840094909071922\n",
      "Epoch: 52, Batch: 900, Avg. Loss: 0.11567365877330303\n",
      "Epoch: 53, Batch: 100, Avg. Loss: 0.11148622982203961\n",
      "Epoch: 53, Batch: 200, Avg. Loss: 0.10556715726852417\n",
      "Epoch: 53, Batch: 300, Avg. Loss: 0.09587929971516132\n",
      "Epoch: 53, Batch: 400, Avg. Loss: 0.12363133490085602\n",
      "Epoch: 53, Batch: 500, Avg. Loss: 0.10403505489230155\n",
      "Epoch: 53, Batch: 600, Avg. Loss: 0.10433159790933132\n",
      "Epoch: 53, Batch: 700, Avg. Loss: 0.10960706524550914\n",
      "Epoch: 53, Batch: 800, Avg. Loss: 0.10538345634937286\n",
      "Epoch: 53, Batch: 900, Avg. Loss: 0.10653051927685737\n",
      "Epoch: 54, Batch: 100, Avg. Loss: 0.10644967287778855\n",
      "Epoch: 54, Batch: 200, Avg. Loss: 0.10561265476047993\n",
      "Epoch: 54, Batch: 300, Avg. Loss: 0.10764876939356327\n",
      "Epoch: 54, Batch: 400, Avg. Loss: 0.10929196961224079\n",
      "Epoch: 54, Batch: 500, Avg. Loss: 0.11529449604451657\n",
      "Epoch: 54, Batch: 600, Avg. Loss: 0.09751341477036476\n",
      "Epoch: 54, Batch: 700, Avg. Loss: 0.10317671336233616\n",
      "Epoch: 54, Batch: 800, Avg. Loss: 0.10640360541641712\n",
      "Epoch: 54, Batch: 900, Avg. Loss: 0.10987936608493327\n",
      "Epoch: 55, Batch: 100, Avg. Loss: 0.10311668433248997\n",
      "Epoch: 55, Batch: 200, Avg. Loss: 0.1127403063327074\n",
      "Epoch: 55, Batch: 300, Avg. Loss: 0.10649671614170074\n",
      "Epoch: 55, Batch: 400, Avg. Loss: 0.11129917860031129\n",
      "Epoch: 55, Batch: 500, Avg. Loss: 0.10134692624211311\n",
      "Epoch: 55, Batch: 600, Avg. Loss: 0.0992819156497717\n",
      "Epoch: 55, Batch: 700, Avg. Loss: 0.11144880406558513\n",
      "Epoch: 55, Batch: 800, Avg. Loss: 0.10511614076793194\n",
      "Epoch: 55, Batch: 900, Avg. Loss: 0.11285345852375031\n",
      "Epoch: 56, Batch: 100, Avg. Loss: 0.10207914270460605\n",
      "Epoch: 56, Batch: 200, Avg. Loss: 0.1071371316164732\n",
      "Epoch: 56, Batch: 300, Avg. Loss: 0.10410498268902302\n",
      "Epoch: 56, Batch: 400, Avg. Loss: 0.09255743727087974\n",
      "Epoch: 56, Batch: 500, Avg. Loss: 0.10241529755294324\n",
      "Epoch: 56, Batch: 600, Avg. Loss: 0.12011821009218693\n",
      "Epoch: 56, Batch: 700, Avg. Loss: 0.10220868423581124\n",
      "Epoch: 56, Batch: 800, Avg. Loss: 0.09993568629026413\n",
      "Epoch: 56, Batch: 900, Avg. Loss: 0.10785711407661439\n",
      "Epoch: 57, Batch: 100, Avg. Loss: 0.10355366595089435\n",
      "Epoch: 57, Batch: 200, Avg. Loss: 0.09672941528260708\n",
      "Epoch: 57, Batch: 300, Avg. Loss: 0.10636041946709156\n",
      "Epoch: 57, Batch: 400, Avg. Loss: 0.10139992281794548\n",
      "Epoch: 57, Batch: 500, Avg. Loss: 0.09441510066390038\n",
      "Epoch: 57, Batch: 600, Avg. Loss: 0.09628898836672306\n",
      "Epoch: 57, Batch: 700, Avg. Loss: 0.1145245074480772\n",
      "Epoch: 57, Batch: 800, Avg. Loss: 0.10405027449131012\n",
      "Epoch: 57, Batch: 900, Avg. Loss: 0.09800728216767311\n",
      "Epoch: 58, Batch: 100, Avg. Loss: 0.11354926489293575\n",
      "Epoch: 58, Batch: 200, Avg. Loss: 0.10104992195963859\n",
      "Epoch: 58, Batch: 300, Avg. Loss: 0.10138593755662441\n",
      "Epoch: 58, Batch: 400, Avg. Loss: 0.10239555068314075\n",
      "Epoch: 58, Batch: 500, Avg. Loss: 0.10713963404297829\n",
      "Epoch: 58, Batch: 600, Avg. Loss: 0.10410033747553825\n",
      "Epoch: 58, Batch: 700, Avg. Loss: 0.10163531444966793\n",
      "Epoch: 58, Batch: 800, Avg. Loss: 0.11296196475625038\n",
      "Epoch: 58, Batch: 900, Avg. Loss: 0.11292346864938736\n",
      "Epoch: 59, Batch: 100, Avg. Loss: 0.10354667037725448\n",
      "Epoch: 59, Batch: 200, Avg. Loss: 0.10243370227515697\n",
      "Epoch: 59, Batch: 300, Avg. Loss: 0.09843822456896305\n",
      "Epoch: 59, Batch: 400, Avg. Loss: 0.10115301474928856\n",
      "Epoch: 59, Batch: 500, Avg. Loss: 0.10306475915014744\n",
      "Epoch: 59, Batch: 600, Avg. Loss: 0.10340026184916497\n",
      "Epoch: 59, Batch: 700, Avg. Loss: 0.10031690098345279\n",
      "Epoch: 59, Batch: 800, Avg. Loss: 0.10506871692836285\n",
      "Epoch: 59, Batch: 900, Avg. Loss: 0.09660112284123898\n",
      "Epoch: 60, Batch: 100, Avg. Loss: 0.09595418229699135\n",
      "Epoch: 60, Batch: 200, Avg. Loss: 0.10369134709239006\n",
      "Epoch: 60, Batch: 300, Avg. Loss: 0.07941712886095047\n",
      "Epoch: 60, Batch: 400, Avg. Loss: 0.10391822919249534\n",
      "Epoch: 60, Batch: 500, Avg. Loss: 0.10083338871598244\n",
      "Epoch: 60, Batch: 600, Avg. Loss: 0.09495173938572407\n",
      "Epoch: 60, Batch: 700, Avg. Loss: 0.09600453078746796\n",
      "Epoch: 60, Batch: 800, Avg. Loss: 0.09380949713289738\n",
      "Epoch: 60, Batch: 900, Avg. Loss: 0.10530453257262706\n",
      "Epoch: 61, Batch: 100, Avg. Loss: 0.10429764948785306\n",
      "Epoch: 61, Batch: 200, Avg. Loss: 0.10887350514531136\n",
      "Epoch: 61, Batch: 300, Avg. Loss: 0.09843886695802212\n",
      "Epoch: 61, Batch: 400, Avg. Loss: 0.10223176650702953\n",
      "Epoch: 61, Batch: 500, Avg. Loss: 0.09928927175700665\n",
      "Epoch: 61, Batch: 600, Avg. Loss: 0.10042372860014438\n",
      "Epoch: 61, Batch: 700, Avg. Loss: 0.08872990764677524\n",
      "Epoch: 61, Batch: 800, Avg. Loss: 0.11086638934910298\n",
      "Epoch: 61, Batch: 900, Avg. Loss: 0.09101366482675076\n",
      "Epoch: 62, Batch: 100, Avg. Loss: 0.09371142454445362\n",
      "Epoch: 62, Batch: 200, Avg. Loss: 0.09399753890931606\n",
      "Epoch: 62, Batch: 300, Avg. Loss: 0.10303262643516063\n",
      "Epoch: 62, Batch: 400, Avg. Loss: 0.10879376567900181\n",
      "Epoch: 62, Batch: 500, Avg. Loss: 0.09832620568573475\n",
      "Epoch: 62, Batch: 600, Avg. Loss: 0.10770081229507923\n",
      "Epoch: 62, Batch: 700, Avg. Loss: 0.1050569698959589\n",
      "Epoch: 62, Batch: 800, Avg. Loss: 0.09950632140040398\n",
      "Epoch: 62, Batch: 900, Avg. Loss: 0.09494788989424706\n",
      "Epoch: 63, Batch: 100, Avg. Loss: 0.09354018375277519\n",
      "Epoch: 63, Batch: 200, Avg. Loss: 0.09162298910319805\n",
      "Epoch: 63, Batch: 300, Avg. Loss: 0.0989491180330515\n",
      "Epoch: 63, Batch: 400, Avg. Loss: 0.09894723728299142\n",
      "Epoch: 63, Batch: 500, Avg. Loss: 0.09215710818767547\n",
      "Epoch: 63, Batch: 600, Avg. Loss: 0.09857776686549187\n",
      "Epoch: 63, Batch: 700, Avg. Loss: 0.1013101476430893\n",
      "Epoch: 63, Batch: 800, Avg. Loss: 0.10125440023839474\n",
      "Epoch: 63, Batch: 900, Avg. Loss: 0.10360147140920162\n",
      "Epoch: 64, Batch: 100, Avg. Loss: 0.0985853674262762\n",
      "Epoch: 64, Batch: 200, Avg. Loss: 0.09756648235023022\n",
      "Epoch: 64, Batch: 300, Avg. Loss: 0.09556674413383007\n",
      "Epoch: 64, Batch: 400, Avg. Loss: 0.09969094917178153\n",
      "Epoch: 64, Batch: 500, Avg. Loss: 0.10048834554851055\n",
      "Epoch: 64, Batch: 600, Avg. Loss: 0.10897926144301891\n",
      "Epoch: 64, Batch: 700, Avg. Loss: 0.09402478694915771\n",
      "Epoch: 64, Batch: 800, Avg. Loss: 0.09706608146429062\n",
      "Epoch: 64, Batch: 900, Avg. Loss: 0.09080192394554615\n",
      "Epoch: 65, Batch: 100, Avg. Loss: 0.10448486939072608\n",
      "Epoch: 65, Batch: 200, Avg. Loss: 0.09338868938386441\n",
      "Epoch: 65, Batch: 300, Avg. Loss: 0.09109797589480877\n",
      "Epoch: 65, Batch: 400, Avg. Loss: 0.10344871908426284\n",
      "Epoch: 65, Batch: 500, Avg. Loss: 0.08704232551157474\n",
      "Epoch: 65, Batch: 600, Avg. Loss: 0.10054645799100399\n",
      "Epoch: 65, Batch: 700, Avg. Loss: 0.09400277353823185\n",
      "Epoch: 65, Batch: 800, Avg. Loss: 0.09640114955604076\n",
      "Epoch: 65, Batch: 900, Avg. Loss: 0.09385459341108798\n",
      "Epoch: 66, Batch: 100, Avg. Loss: 0.0986055114120245\n",
      "Epoch: 66, Batch: 200, Avg. Loss: 0.09813191033899785\n",
      "Epoch: 66, Batch: 300, Avg. Loss: 0.09880328319966793\n",
      "Epoch: 66, Batch: 400, Avg. Loss: 0.09633123755455017\n",
      "Epoch: 66, Batch: 500, Avg. Loss: 0.09365133136510849\n",
      "Epoch: 66, Batch: 600, Avg. Loss: 0.08590740233659744\n",
      "Epoch: 66, Batch: 700, Avg. Loss: 0.09888152927160263\n",
      "Epoch: 66, Batch: 800, Avg. Loss: 0.10117038011550904\n",
      "Epoch: 66, Batch: 900, Avg. Loss: 0.09022499680519104\n",
      "Epoch: 67, Batch: 100, Avg. Loss: 0.08273420758545398\n",
      "Epoch: 67, Batch: 200, Avg. Loss: 0.09632266119122505\n",
      "Epoch: 67, Batch: 300, Avg. Loss: 0.09118950046598912\n",
      "Epoch: 67, Batch: 400, Avg. Loss: 0.09623851709067821\n",
      "Epoch: 67, Batch: 500, Avg. Loss: 0.0954651603102684\n",
      "Epoch: 67, Batch: 600, Avg. Loss: 0.09814319968223571\n",
      "Epoch: 67, Batch: 700, Avg. Loss: 0.09654993854463101\n",
      "Epoch: 67, Batch: 800, Avg. Loss: 0.09492956638336182\n",
      "Epoch: 67, Batch: 900, Avg. Loss: 0.09176367171108722\n",
      "Epoch: 68, Batch: 100, Avg. Loss: 0.08865092538297176\n",
      "Epoch: 68, Batch: 200, Avg. Loss: 0.08815545596182346\n",
      "Epoch: 68, Batch: 300, Avg. Loss: 0.0966100500524044\n",
      "Epoch: 68, Batch: 400, Avg. Loss: 0.09581943601369858\n",
      "Epoch: 68, Batch: 500, Avg. Loss: 0.09019666828215123\n",
      "Epoch: 68, Batch: 600, Avg. Loss: 0.08597878437489272\n",
      "Epoch: 68, Batch: 700, Avg. Loss: 0.0900061471760273\n",
      "Epoch: 68, Batch: 800, Avg. Loss: 0.09686620458960533\n",
      "Epoch: 68, Batch: 900, Avg. Loss: 0.0988680898398161\n",
      "Epoch: 69, Batch: 100, Avg. Loss: 0.09311451055109501\n",
      "Epoch: 69, Batch: 200, Avg. Loss: 0.10892253115773201\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 69, Batch: 300, Avg. Loss: 0.09430132851004601\n",
      "Epoch: 69, Batch: 400, Avg. Loss: 0.0965896037966013\n",
      "Epoch: 69, Batch: 500, Avg. Loss: 0.09290353938937188\n",
      "Epoch: 69, Batch: 600, Avg. Loss: 0.11124850541353226\n",
      "Epoch: 69, Batch: 700, Avg. Loss: 0.09433628745377064\n",
      "Epoch: 69, Batch: 800, Avg. Loss: 0.08261031292378902\n",
      "Epoch: 69, Batch: 900, Avg. Loss: 0.09100077465176583\n",
      "Epoch: 70, Batch: 100, Avg. Loss: 0.0912116875499487\n",
      "Epoch: 70, Batch: 200, Avg. Loss: 0.09183600723743439\n",
      "Epoch: 70, Batch: 300, Avg. Loss: 0.08217784464359283\n",
      "Epoch: 70, Batch: 400, Avg. Loss: 0.10397103480994702\n",
      "Epoch: 70, Batch: 500, Avg. Loss: 0.08067858465015888\n",
      "Epoch: 70, Batch: 600, Avg. Loss: 0.10307347171008586\n",
      "Epoch: 70, Batch: 700, Avg. Loss: 0.08933373242616653\n",
      "Epoch: 70, Batch: 800, Avg. Loss: 0.10130013704299927\n",
      "Epoch: 70, Batch: 900, Avg. Loss: 0.09008154839277267\n",
      "Epoch: 71, Batch: 100, Avg. Loss: 0.08379532724618911\n",
      "Epoch: 71, Batch: 200, Avg. Loss: 0.09198853775858878\n",
      "Epoch: 71, Batch: 300, Avg. Loss: 0.08757059417665004\n",
      "Epoch: 71, Batch: 400, Avg. Loss: 0.09892251558601856\n",
      "Epoch: 71, Batch: 500, Avg. Loss: 0.0919396309554577\n",
      "Epoch: 71, Batch: 600, Avg. Loss: 0.09334353081882\n",
      "Epoch: 71, Batch: 700, Avg. Loss: 0.08769536308944226\n",
      "Epoch: 71, Batch: 800, Avg. Loss: 0.0879382561892271\n",
      "Epoch: 71, Batch: 900, Avg. Loss: 0.1009044023603201\n",
      "Epoch: 72, Batch: 100, Avg. Loss: 0.09684274449944497\n",
      "Epoch: 72, Batch: 200, Avg. Loss: 0.0994208986312151\n",
      "Epoch: 72, Batch: 300, Avg. Loss: 0.08476799815893173\n",
      "Epoch: 72, Batch: 400, Avg. Loss: 0.09239469766616822\n",
      "Epoch: 72, Batch: 500, Avg. Loss: 0.08312648192048072\n",
      "Epoch: 72, Batch: 600, Avg. Loss: 0.0965862326323986\n",
      "Epoch: 72, Batch: 700, Avg. Loss: 0.09416570760309696\n",
      "Epoch: 72, Batch: 800, Avg. Loss: 0.09556364767253399\n",
      "Epoch: 72, Batch: 900, Avg. Loss: 0.09821764633059502\n",
      "Epoch: 73, Batch: 100, Avg. Loss: 0.09297935597598553\n",
      "Epoch: 73, Batch: 200, Avg. Loss: 0.0759263638406992\n",
      "Epoch: 73, Batch: 300, Avg. Loss: 0.0900664246082306\n",
      "Epoch: 73, Batch: 400, Avg. Loss: 0.09697813041508198\n",
      "Epoch: 73, Batch: 500, Avg. Loss: 0.0940529652684927\n",
      "Epoch: 73, Batch: 600, Avg. Loss: 0.09535381555557251\n",
      "Epoch: 73, Batch: 700, Avg. Loss: 0.09312535390257835\n",
      "Epoch: 73, Batch: 800, Avg. Loss: 0.08632180474698543\n",
      "Epoch: 73, Batch: 900, Avg. Loss: 0.08888283096253873\n",
      "Epoch: 74, Batch: 100, Avg. Loss: 0.08178275555372239\n",
      "Epoch: 74, Batch: 200, Avg. Loss: 0.08566500678658485\n",
      "Epoch: 74, Batch: 300, Avg. Loss: 0.08248842351138591\n",
      "Epoch: 74, Batch: 400, Avg. Loss: 0.08661402545869351\n",
      "Epoch: 74, Batch: 500, Avg. Loss: 0.09332641422748565\n",
      "Epoch: 74, Batch: 600, Avg. Loss: 0.08757479563355446\n",
      "Epoch: 74, Batch: 700, Avg. Loss: 0.09546999625861645\n",
      "Epoch: 74, Batch: 800, Avg. Loss: 0.097351683229208\n",
      "Epoch: 74, Batch: 900, Avg. Loss: 0.09568680260330438\n",
      "Epoch: 75, Batch: 100, Avg. Loss: 0.0913899403065443\n",
      "Epoch: 75, Batch: 200, Avg. Loss: 0.09644075073301792\n",
      "Epoch: 75, Batch: 300, Avg. Loss: 0.09651996947824955\n",
      "Epoch: 75, Batch: 400, Avg. Loss: 0.08482513777911663\n",
      "Epoch: 75, Batch: 500, Avg. Loss: 0.08421830780804157\n",
      "Epoch: 75, Batch: 600, Avg. Loss: 0.09184074886143208\n",
      "Epoch: 75, Batch: 700, Avg. Loss: 0.09610087603330612\n",
      "Epoch: 75, Batch: 800, Avg. Loss: 0.09273304134607314\n",
      "Epoch: 75, Batch: 900, Avg. Loss: 0.09051196984946727\n",
      "Epoch: 76, Batch: 100, Avg. Loss: 0.08548838451504708\n",
      "Epoch: 76, Batch: 200, Avg. Loss: 0.09028274029493331\n",
      "Epoch: 76, Batch: 300, Avg. Loss: 0.09211287856101989\n",
      "Epoch: 76, Batch: 400, Avg. Loss: 0.09644161388278008\n",
      "Epoch: 76, Batch: 500, Avg. Loss: 0.09077881596982479\n",
      "Epoch: 76, Batch: 600, Avg. Loss: 0.08125195860862731\n",
      "Epoch: 76, Batch: 700, Avg. Loss: 0.08971736684441567\n",
      "Epoch: 76, Batch: 800, Avg. Loss: 0.08538171067833901\n",
      "Epoch: 76, Batch: 900, Avg. Loss: 0.07736496105790139\n",
      "Epoch: 77, Batch: 100, Avg. Loss: 0.09364143408834934\n",
      "Epoch: 77, Batch: 200, Avg. Loss: 0.08044580318033695\n",
      "Epoch: 77, Batch: 300, Avg. Loss: 0.0841747472435236\n",
      "Epoch: 77, Batch: 400, Avg. Loss: 0.094134701192379\n",
      "Epoch: 77, Batch: 500, Avg. Loss: 0.08650468870997428\n",
      "Epoch: 77, Batch: 600, Avg. Loss: 0.08376674264669419\n",
      "Epoch: 77, Batch: 700, Avg. Loss: 0.09135563395917416\n",
      "Epoch: 77, Batch: 800, Avg. Loss: 0.08648567996919156\n",
      "Epoch: 77, Batch: 900, Avg. Loss: 0.0796619314700365\n",
      "Epoch: 78, Batch: 100, Avg. Loss: 0.08361300624907017\n",
      "Epoch: 78, Batch: 200, Avg. Loss: 0.08056649029254913\n",
      "Epoch: 78, Batch: 300, Avg. Loss: 0.08552569635212422\n",
      "Epoch: 78, Batch: 400, Avg. Loss: 0.08738149896264076\n",
      "Epoch: 78, Batch: 500, Avg. Loss: 0.09373987957835198\n",
      "Epoch: 78, Batch: 600, Avg. Loss: 0.07674202810972929\n",
      "Epoch: 78, Batch: 700, Avg. Loss: 0.09773576192557812\n",
      "Epoch: 78, Batch: 800, Avg. Loss: 0.07917075790464878\n",
      "Epoch: 78, Batch: 900, Avg. Loss: 0.08261831730604172\n",
      "Epoch: 79, Batch: 100, Avg. Loss: 0.06958654351532459\n",
      "Epoch: 79, Batch: 200, Avg. Loss: 0.08176872134208679\n",
      "Epoch: 79, Batch: 300, Avg. Loss: 0.07567189015448093\n",
      "Epoch: 79, Batch: 400, Avg. Loss: 0.09537867441773415\n",
      "Epoch: 79, Batch: 500, Avg. Loss: 0.09371434211730957\n",
      "Epoch: 79, Batch: 600, Avg. Loss: 0.08321986965835095\n",
      "Epoch: 79, Batch: 700, Avg. Loss: 0.08556830443441868\n",
      "Epoch: 79, Batch: 800, Avg. Loss: 0.09399012483656406\n",
      "Epoch: 79, Batch: 900, Avg. Loss: 0.09367853082716465\n",
      "Epoch: 80, Batch: 100, Avg. Loss: 0.08039464257657528\n",
      "Epoch: 80, Batch: 200, Avg. Loss: 0.07591810129582882\n",
      "Epoch: 80, Batch: 300, Avg. Loss: 0.09638601660728455\n",
      "Epoch: 80, Batch: 400, Avg. Loss: 0.09298188090324402\n",
      "Epoch: 80, Batch: 500, Avg. Loss: 0.0866419917345047\n",
      "Epoch: 80, Batch: 600, Avg. Loss: 0.08000588186085224\n",
      "Epoch: 80, Batch: 700, Avg. Loss: 0.09447656087577343\n",
      "Epoch: 80, Batch: 800, Avg. Loss: 0.09541752748191357\n",
      "Epoch: 80, Batch: 900, Avg. Loss: 0.07888172261416912\n",
      "Epoch: 81, Batch: 100, Avg. Loss: 0.08895292334258556\n",
      "Epoch: 81, Batch: 200, Avg. Loss: 0.09302670046687127\n",
      "Epoch: 81, Batch: 300, Avg. Loss: 0.08512339971959591\n",
      "Epoch: 81, Batch: 400, Avg. Loss: 0.07870825462043285\n",
      "Epoch: 81, Batch: 500, Avg. Loss: 0.09633647255599499\n",
      "Epoch: 81, Batch: 600, Avg. Loss: 0.08686820767819882\n",
      "Epoch: 81, Batch: 700, Avg. Loss: 0.09218739286065102\n",
      "Epoch: 81, Batch: 800, Avg. Loss: 0.08838045932352542\n",
      "Epoch: 81, Batch: 900, Avg. Loss: 0.08742604188621045\n",
      "Epoch: 82, Batch: 100, Avg. Loss: 0.07621222518384457\n",
      "Epoch: 82, Batch: 200, Avg. Loss: 0.08484068438410759\n",
      "Epoch: 82, Batch: 300, Avg. Loss: 0.0765729808062315\n",
      "Epoch: 82, Batch: 400, Avg. Loss: 0.08890795476734638\n",
      "Epoch: 82, Batch: 500, Avg. Loss: 0.08617064602673054\n",
      "Epoch: 82, Batch: 600, Avg. Loss: 0.09531883127987385\n",
      "Epoch: 82, Batch: 700, Avg. Loss: 0.08621696598827838\n",
      "Epoch: 82, Batch: 800, Avg. Loss: 0.08961372457444668\n",
      "Epoch: 82, Batch: 900, Avg. Loss: 0.08164013564586639\n",
      "Epoch: 83, Batch: 100, Avg. Loss: 0.0879923764616251\n",
      "Epoch: 83, Batch: 200, Avg. Loss: 0.0983397625386715\n",
      "Epoch: 83, Batch: 300, Avg. Loss: 0.09011871546506882\n",
      "Epoch: 83, Batch: 400, Avg. Loss: 0.07068143099546433\n",
      "Epoch: 83, Batch: 500, Avg. Loss: 0.07691395230591297\n",
      "Epoch: 83, Batch: 600, Avg. Loss: 0.0850284269452095\n",
      "Epoch: 83, Batch: 700, Avg. Loss: 0.08974927999079227\n",
      "Epoch: 83, Batch: 800, Avg. Loss: 0.09172146528959274\n",
      "Epoch: 83, Batch: 900, Avg. Loss: 0.09225967817008496\n",
      "Epoch: 84, Batch: 100, Avg. Loss: 0.07962762631475925\n",
      "Epoch: 84, Batch: 200, Avg. Loss: 0.08483371511101723\n",
      "Epoch: 84, Batch: 300, Avg. Loss: 0.08444297783076764\n",
      "Epoch: 84, Batch: 400, Avg. Loss: 0.09092368200421333\n",
      "Epoch: 84, Batch: 500, Avg. Loss: 0.08262479417026043\n",
      "Epoch: 84, Batch: 600, Avg. Loss: 0.0878732766956091\n",
      "Epoch: 84, Batch: 700, Avg. Loss: 0.0847813805192709\n",
      "Epoch: 84, Batch: 800, Avg. Loss: 0.07597923971712589\n",
      "Epoch: 84, Batch: 900, Avg. Loss: 0.07766496129333973\n",
      "Epoch: 85, Batch: 100, Avg. Loss: 0.08836082018911838\n",
      "Epoch: 85, Batch: 200, Avg. Loss: 0.09056740425527096\n",
      "Epoch: 85, Batch: 300, Avg. Loss: 0.07930388025939465\n",
      "Epoch: 85, Batch: 400, Avg. Loss: 0.07926387414336204\n",
      "Epoch: 85, Batch: 500, Avg. Loss: 0.08093431949615479\n",
      "Epoch: 85, Batch: 600, Avg. Loss: 0.09154650993645191\n",
      "Epoch: 85, Batch: 700, Avg. Loss: 0.08553424842655659\n",
      "Epoch: 85, Batch: 800, Avg. Loss: 0.08026093021035194\n",
      "Epoch: 85, Batch: 900, Avg. Loss: 0.08153683014214039\n",
      "Epoch: 86, Batch: 100, Avg. Loss: 0.0858393743634224\n",
      "Epoch: 86, Batch: 200, Avg. Loss: 0.07801412917673588\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 86, Batch: 300, Avg. Loss: 0.08686496429145336\n",
      "Epoch: 86, Batch: 400, Avg. Loss: 0.07939108066260815\n",
      "Epoch: 86, Batch: 500, Avg. Loss: 0.08184866778552533\n",
      "Epoch: 86, Batch: 600, Avg. Loss: 0.09023882023990154\n",
      "Epoch: 86, Batch: 700, Avg. Loss: 0.08042468510568142\n",
      "Epoch: 86, Batch: 800, Avg. Loss: 0.0901332774013281\n",
      "Epoch: 86, Batch: 900, Avg. Loss: 0.08729609198868275\n",
      "Epoch: 87, Batch: 100, Avg. Loss: 0.090861132517457\n",
      "Epoch: 87, Batch: 200, Avg. Loss: 0.08258832819759845\n",
      "Epoch: 87, Batch: 300, Avg. Loss: 0.07971448831260204\n",
      "Epoch: 87, Batch: 400, Avg. Loss: 0.08434415318071842\n",
      "Epoch: 87, Batch: 500, Avg. Loss: 0.08906706631183624\n",
      "Epoch: 87, Batch: 600, Avg. Loss: 0.07619240760803223\n",
      "Epoch: 87, Batch: 700, Avg. Loss: 0.08897438123822213\n",
      "Epoch: 87, Batch: 800, Avg. Loss: 0.08686970263719558\n",
      "Epoch: 87, Batch: 900, Avg. Loss: 0.08268445990979671\n",
      "Epoch: 88, Batch: 100, Avg. Loss: 0.07957580752670765\n",
      "Epoch: 88, Batch: 200, Avg. Loss: 0.08696162894368172\n",
      "Epoch: 88, Batch: 300, Avg. Loss: 0.09069046013057232\n",
      "Epoch: 88, Batch: 400, Avg. Loss: 0.0765650200843811\n",
      "Epoch: 88, Batch: 500, Avg. Loss: 0.07990098111331463\n",
      "Epoch: 88, Batch: 600, Avg. Loss: 0.08049642138183116\n",
      "Epoch: 88, Batch: 700, Avg. Loss: 0.07847566045820713\n",
      "Epoch: 88, Batch: 800, Avg. Loss: 0.08249866545200348\n",
      "Epoch: 88, Batch: 900, Avg. Loss: 0.08412843890488148\n",
      "Epoch: 89, Batch: 100, Avg. Loss: 0.08968796364963055\n",
      "Epoch: 89, Batch: 200, Avg. Loss: 0.0864513248205185\n",
      "Epoch: 89, Batch: 300, Avg. Loss: 0.08883466571569443\n",
      "Epoch: 89, Batch: 400, Avg. Loss: 0.08333682864904404\n",
      "Epoch: 89, Batch: 500, Avg. Loss: 0.08520014889538288\n",
      "Epoch: 89, Batch: 600, Avg. Loss: 0.07006197974085808\n",
      "Epoch: 89, Batch: 700, Avg. Loss: 0.08090024404227733\n",
      "Epoch: 89, Batch: 800, Avg. Loss: 0.08662227042019367\n",
      "Epoch: 89, Batch: 900, Avg. Loss: 0.07905520915985108\n",
      "Epoch: 90, Batch: 100, Avg. Loss: 0.07789214737713337\n",
      "Epoch: 90, Batch: 200, Avg. Loss: 0.0826767048984766\n",
      "Epoch: 90, Batch: 300, Avg. Loss: 0.08207570552825928\n",
      "Epoch: 90, Batch: 400, Avg. Loss: 0.07766814202070237\n",
      "Epoch: 90, Batch: 500, Avg. Loss: 0.07913859754800796\n",
      "Epoch: 90, Batch: 600, Avg. Loss: 0.07568119294941425\n",
      "Epoch: 90, Batch: 700, Avg. Loss: 0.0800759468972683\n",
      "Epoch: 90, Batch: 800, Avg. Loss: 0.08175529558211565\n",
      "Epoch: 90, Batch: 900, Avg. Loss: 0.08060787543654442\n",
      "Epoch: 91, Batch: 100, Avg. Loss: 0.0761522263288498\n",
      "Epoch: 91, Batch: 200, Avg. Loss: 0.07652141191065312\n",
      "Epoch: 91, Batch: 300, Avg. Loss: 0.08224038437008857\n",
      "Epoch: 91, Batch: 400, Avg. Loss: 0.08540616571903228\n",
      "Epoch: 91, Batch: 500, Avg. Loss: 0.0979804202914238\n",
      "Epoch: 91, Batch: 600, Avg. Loss: 0.07628080211579799\n",
      "Epoch: 91, Batch: 700, Avg. Loss: 0.07641186714172363\n",
      "Epoch: 91, Batch: 800, Avg. Loss: 0.084579571262002\n",
      "Epoch: 91, Batch: 900, Avg. Loss: 0.09359907791018486\n",
      "Epoch: 92, Batch: 100, Avg. Loss: 0.07491824433207511\n",
      "Epoch: 92, Batch: 200, Avg. Loss: 0.08148106276988983\n",
      "Epoch: 92, Batch: 300, Avg. Loss: 0.07906774744391441\n",
      "Epoch: 92, Batch: 400, Avg. Loss: 0.07353968173265457\n",
      "Epoch: 92, Batch: 500, Avg. Loss: 0.0864991220086813\n",
      "Epoch: 92, Batch: 600, Avg. Loss: 0.08185872048139572\n",
      "Epoch: 92, Batch: 700, Avg. Loss: 0.08422725722193718\n",
      "Epoch: 92, Batch: 800, Avg. Loss: 0.07915057353675366\n",
      "Epoch: 92, Batch: 900, Avg. Loss: 0.09089731968939305\n",
      "Epoch: 93, Batch: 100, Avg. Loss: 0.07649995386600494\n",
      "Epoch: 93, Batch: 200, Avg. Loss: 0.0761596718430519\n",
      "Epoch: 93, Batch: 300, Avg. Loss: 0.07228268541395665\n",
      "Epoch: 93, Batch: 400, Avg. Loss: 0.08719336405396462\n",
      "Epoch: 93, Batch: 500, Avg. Loss: 0.08599462114274502\n",
      "Epoch: 93, Batch: 600, Avg. Loss: 0.07429470524191856\n",
      "Epoch: 93, Batch: 700, Avg. Loss: 0.08507794491946698\n",
      "Epoch: 93, Batch: 800, Avg. Loss: 0.08253847606480122\n",
      "Epoch: 93, Batch: 900, Avg. Loss: 0.08001551821827889\n",
      "Epoch: 94, Batch: 100, Avg. Loss: 0.08840486563742161\n",
      "Epoch: 94, Batch: 200, Avg. Loss: 0.07327625438570977\n",
      "Epoch: 94, Batch: 300, Avg. Loss: 0.0843181262910366\n",
      "Epoch: 94, Batch: 400, Avg. Loss: 0.08206186555325985\n",
      "Epoch: 94, Batch: 500, Avg. Loss: 0.07598735280334949\n",
      "Epoch: 94, Batch: 600, Avg. Loss: 0.0715178219974041\n",
      "Epoch: 94, Batch: 700, Avg. Loss: 0.07970416210591794\n",
      "Epoch: 94, Batch: 800, Avg. Loss: 0.08087074220180511\n",
      "Epoch: 94, Batch: 900, Avg. Loss: 0.08088410183787346\n",
      "Epoch: 95, Batch: 100, Avg. Loss: 0.08514286123216153\n",
      "Epoch: 95, Batch: 200, Avg. Loss: 0.0743800663203001\n",
      "Epoch: 95, Batch: 300, Avg. Loss: 0.07363688245415688\n",
      "Epoch: 95, Batch: 400, Avg. Loss: 0.07503628201782703\n",
      "Epoch: 95, Batch: 500, Avg. Loss: 0.0903602722287178\n",
      "Epoch: 95, Batch: 600, Avg. Loss: 0.08175933092832566\n",
      "Epoch: 95, Batch: 700, Avg. Loss: 0.07574377618730069\n",
      "Epoch: 95, Batch: 800, Avg. Loss: 0.0787751903384924\n",
      "Epoch: 95, Batch: 900, Avg. Loss: 0.07895507104694843\n",
      "Epoch: 96, Batch: 100, Avg. Loss: 0.08348740711808204\n",
      "Epoch: 96, Batch: 200, Avg. Loss: 0.08175951525568963\n",
      "Epoch: 96, Batch: 300, Avg. Loss: 0.07658334352076053\n",
      "Epoch: 96, Batch: 400, Avg. Loss: 0.08675104185938835\n",
      "Epoch: 96, Batch: 500, Avg. Loss: 0.08216813303530217\n",
      "Epoch: 96, Batch: 600, Avg. Loss: 0.07741275086998939\n",
      "Epoch: 96, Batch: 700, Avg. Loss: 0.07785656735301018\n",
      "Epoch: 96, Batch: 800, Avg. Loss: 0.07314306162297726\n",
      "Epoch: 96, Batch: 900, Avg. Loss: 0.07539861135184765\n",
      "Epoch: 97, Batch: 100, Avg. Loss: 0.07821284577250481\n",
      "Epoch: 97, Batch: 200, Avg. Loss: 0.08335631094872951\n",
      "Epoch: 97, Batch: 300, Avg. Loss: 0.07666182965040207\n",
      "Epoch: 97, Batch: 400, Avg. Loss: 0.07557642944157124\n",
      "Epoch: 97, Batch: 500, Avg. Loss: 0.073813197016716\n",
      "Epoch: 97, Batch: 600, Avg. Loss: 0.0797989708185196\n",
      "Epoch: 97, Batch: 700, Avg. Loss: 0.07765541322529317\n",
      "Epoch: 97, Batch: 800, Avg. Loss: 0.08014209248125553\n",
      "Epoch: 97, Batch: 900, Avg. Loss: 0.08192169994115829\n",
      "Epoch: 98, Batch: 100, Avg. Loss: 0.07770783640444279\n",
      "Epoch: 98, Batch: 200, Avg. Loss: 0.07866012543439865\n",
      "Epoch: 98, Batch: 300, Avg. Loss: 0.09110039830207825\n",
      "Epoch: 98, Batch: 400, Avg. Loss: 0.06770319551229477\n",
      "Epoch: 98, Batch: 500, Avg. Loss: 0.07642909914255142\n",
      "Epoch: 98, Batch: 600, Avg. Loss: 0.09176859930157662\n",
      "Epoch: 98, Batch: 700, Avg. Loss: 0.07256216742098331\n",
      "Epoch: 98, Batch: 800, Avg. Loss: 0.08437281742691993\n",
      "Epoch: 98, Batch: 900, Avg. Loss: 0.0746835207939148\n",
      "Epoch: 99, Batch: 100, Avg. Loss: 0.08002020679414272\n",
      "Epoch: 99, Batch: 200, Avg. Loss: 0.07538539797067642\n",
      "Epoch: 99, Batch: 300, Avg. Loss: 0.07207652613520622\n",
      "Epoch: 99, Batch: 400, Avg. Loss: 0.07326294802129268\n",
      "Epoch: 99, Batch: 500, Avg. Loss: 0.08389605708420277\n",
      "Epoch: 99, Batch: 600, Avg. Loss: 0.0829875761270523\n",
      "Epoch: 99, Batch: 700, Avg. Loss: 0.07719128280878067\n",
      "Epoch: 99, Batch: 800, Avg. Loss: 0.08248918123543263\n",
      "Epoch: 99, Batch: 900, Avg. Loss: 0.07423068821430207\n",
      "Epoch: 100, Batch: 100, Avg. Loss: 0.07856469690799713\n",
      "Epoch: 100, Batch: 200, Avg. Loss: 0.07524305671453475\n",
      "Epoch: 100, Batch: 300, Avg. Loss: 0.07725321866571903\n",
      "Epoch: 100, Batch: 400, Avg. Loss: 0.07192257910966873\n",
      "Epoch: 100, Batch: 500, Avg. Loss: 0.07802159063518048\n",
      "Epoch: 100, Batch: 600, Avg. Loss: 0.08322900965809822\n",
      "Epoch: 100, Batch: 700, Avg. Loss: 0.08363508187234402\n",
      "Epoch: 100, Batch: 800, Avg. Loss: 0.0796539507061243\n",
      "Epoch: 100, Batch: 900, Avg. Loss: 0.07137255057692528\n",
      "Epoch: 101, Batch: 100, Avg. Loss: 0.08167535580694675\n",
      "Epoch: 101, Batch: 200, Avg. Loss: 0.08520837433636189\n",
      "Epoch: 101, Batch: 300, Avg. Loss: 0.0838951537758112\n",
      "Epoch: 101, Batch: 400, Avg. Loss: 0.08148821376264095\n",
      "Epoch: 101, Batch: 500, Avg. Loss: 0.08145734384655952\n",
      "Epoch: 101, Batch: 600, Avg. Loss: 0.07926038287580013\n",
      "Epoch: 101, Batch: 700, Avg. Loss: 0.09211097471415997\n",
      "Epoch: 101, Batch: 800, Avg. Loss: 0.0746624556183815\n",
      "Epoch: 101, Batch: 900, Avg. Loss: 0.07075779661536216\n",
      "Epoch: 102, Batch: 100, Avg. Loss: 0.07659875564277172\n",
      "Epoch: 102, Batch: 200, Avg. Loss: 0.08183759413659572\n",
      "Epoch: 102, Batch: 300, Avg. Loss: 0.06199640922248364\n",
      "Epoch: 102, Batch: 400, Avg. Loss: 0.07923632986843586\n",
      "Epoch: 102, Batch: 500, Avg. Loss: 0.08348719768226147\n",
      "Epoch: 102, Batch: 600, Avg. Loss: 0.07365227654576302\n",
      "Epoch: 102, Batch: 700, Avg. Loss: 0.07845898874104024\n",
      "Epoch: 102, Batch: 800, Avg. Loss: 0.07298297710716724\n",
      "Epoch: 102, Batch: 900, Avg. Loss: 0.08140171840786933\n",
      "Epoch: 103, Batch: 100, Avg. Loss: 0.07587585307657718\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 103, Batch: 200, Avg. Loss: 0.07682436674833298\n",
      "Epoch: 103, Batch: 300, Avg. Loss: 0.08629803583025933\n",
      "Epoch: 103, Batch: 400, Avg. Loss: 0.07072281911969185\n",
      "Epoch: 103, Batch: 500, Avg. Loss: 0.07915310770273208\n",
      "Epoch: 103, Batch: 600, Avg. Loss: 0.08290607042610645\n",
      "Epoch: 103, Batch: 700, Avg. Loss: 0.07813271686434746\n",
      "Epoch: 103, Batch: 800, Avg. Loss: 0.07658857315778732\n",
      "Epoch: 103, Batch: 900, Avg. Loss: 0.07540241628885269\n",
      "Epoch: 104, Batch: 100, Avg. Loss: 0.08395576551556587\n",
      "Epoch: 104, Batch: 200, Avg. Loss: 0.0740804585069418\n",
      "Epoch: 104, Batch: 300, Avg. Loss: 0.07644367791712284\n",
      "Epoch: 104, Batch: 400, Avg. Loss: 0.07574278943240642\n",
      "Epoch: 104, Batch: 500, Avg. Loss: 0.07588396742939948\n",
      "Epoch: 104, Batch: 600, Avg. Loss: 0.07471557475626468\n",
      "Epoch: 104, Batch: 700, Avg. Loss: 0.07189608909189701\n",
      "Epoch: 104, Batch: 800, Avg. Loss: 0.08148979678750039\n",
      "Epoch: 104, Batch: 900, Avg. Loss: 0.07536472007632256\n",
      "Epoch: 105, Batch: 100, Avg. Loss: 0.07365230955183506\n",
      "Epoch: 105, Batch: 200, Avg. Loss: 0.08040524698793888\n",
      "Epoch: 105, Batch: 300, Avg. Loss: 0.0776745406538248\n",
      "Epoch: 105, Batch: 400, Avg. Loss: 0.07814049325883389\n",
      "Epoch: 105, Batch: 500, Avg. Loss: 0.08553849264979363\n",
      "Epoch: 105, Batch: 600, Avg. Loss: 0.07553422592580318\n",
      "Epoch: 105, Batch: 700, Avg. Loss: 0.07355174586176873\n",
      "Epoch: 105, Batch: 800, Avg. Loss: 0.08134490326046943\n",
      "Epoch: 105, Batch: 900, Avg. Loss: 0.0730147398263216\n",
      "Epoch: 106, Batch: 100, Avg. Loss: 0.07907881893217564\n",
      "Epoch: 106, Batch: 200, Avg. Loss: 0.08175262928009033\n",
      "Epoch: 106, Batch: 300, Avg. Loss: 0.07224666297435761\n",
      "Epoch: 106, Batch: 400, Avg. Loss: 0.07707610785961151\n",
      "Epoch: 106, Batch: 500, Avg. Loss: 0.07751775480806827\n",
      "Epoch: 106, Batch: 600, Avg. Loss: 0.07895012184977532\n",
      "Epoch: 106, Batch: 700, Avg. Loss: 0.07852794326841832\n",
      "Epoch: 106, Batch: 800, Avg. Loss: 0.06858939476311207\n",
      "Epoch: 106, Batch: 900, Avg. Loss: 0.06671018324792385\n",
      "Epoch: 107, Batch: 100, Avg. Loss: 0.06619036488234997\n",
      "Epoch: 107, Batch: 200, Avg. Loss: 0.08776650883257389\n",
      "Epoch: 107, Batch: 300, Avg. Loss: 0.0733894769847393\n",
      "Epoch: 107, Batch: 400, Avg. Loss: 0.07193864040076732\n",
      "Epoch: 107, Batch: 500, Avg. Loss: 0.09413019157946109\n",
      "Epoch: 107, Batch: 600, Avg. Loss: 0.07889170967042446\n",
      "Epoch: 107, Batch: 700, Avg. Loss: 0.06808089390397072\n",
      "Epoch: 107, Batch: 800, Avg. Loss: 0.08129457958042621\n",
      "Epoch: 107, Batch: 900, Avg. Loss: 0.07404189184308052\n",
      "Epoch: 108, Batch: 100, Avg. Loss: 0.07354947865009308\n",
      "Epoch: 108, Batch: 200, Avg. Loss: 0.07522918235510588\n",
      "Epoch: 108, Batch: 300, Avg. Loss: 0.07028101280331611\n",
      "Epoch: 108, Batch: 400, Avg. Loss: 0.06985181704163551\n",
      "Epoch: 108, Batch: 500, Avg. Loss: 0.07374994538724422\n",
      "Epoch: 108, Batch: 600, Avg. Loss: 0.0664146326482296\n",
      "Epoch: 108, Batch: 700, Avg. Loss: 0.07176158860325814\n",
      "Epoch: 108, Batch: 800, Avg. Loss: 0.07966437071561813\n",
      "Epoch: 108, Batch: 900, Avg. Loss: 0.08327088475227357\n",
      "Epoch: 109, Batch: 100, Avg. Loss: 0.06725056633353234\n",
      "Epoch: 109, Batch: 200, Avg. Loss: 0.07902947060763836\n",
      "Epoch: 109, Batch: 300, Avg. Loss: 0.07496672689914703\n",
      "Epoch: 109, Batch: 400, Avg. Loss: 0.07760019600391388\n",
      "Epoch: 109, Batch: 500, Avg. Loss: 0.07539880208671093\n",
      "Epoch: 109, Batch: 600, Avg. Loss: 0.08130309402942658\n",
      "Epoch: 109, Batch: 700, Avg. Loss: 0.07500743575394153\n",
      "Epoch: 109, Batch: 800, Avg. Loss: 0.07503978446125985\n",
      "Epoch: 109, Batch: 900, Avg. Loss: 0.07238471779972315\n",
      "Epoch: 110, Batch: 100, Avg. Loss: 0.08136938720941543\n",
      "Epoch: 110, Batch: 200, Avg. Loss: 0.07544424436986447\n",
      "Epoch: 110, Batch: 300, Avg. Loss: 0.07820133425295353\n",
      "Epoch: 110, Batch: 400, Avg. Loss: 0.07044435866177082\n",
      "Epoch: 110, Batch: 500, Avg. Loss: 0.0819370073080063\n",
      "Epoch: 110, Batch: 600, Avg. Loss: 0.06995787151157856\n",
      "Epoch: 110, Batch: 700, Avg. Loss: 0.06770204566419125\n",
      "Epoch: 110, Batch: 800, Avg. Loss: 0.07783538028597832\n",
      "Epoch: 110, Batch: 900, Avg. Loss: 0.07169955745339393\n",
      "Epoch: 111, Batch: 100, Avg. Loss: 0.0816642926633358\n",
      "Epoch: 111, Batch: 200, Avg. Loss: 0.07867605715990067\n",
      "Epoch: 111, Batch: 300, Avg. Loss: 0.06990408912301063\n",
      "Epoch: 111, Batch: 400, Avg. Loss: 0.07911447376012802\n",
      "Epoch: 111, Batch: 500, Avg. Loss: 0.07321421802043915\n",
      "Epoch: 111, Batch: 600, Avg. Loss: 0.07069057539105415\n",
      "Epoch: 111, Batch: 700, Avg. Loss: 0.07721589639782905\n",
      "Epoch: 111, Batch: 800, Avg. Loss: 0.07099058248102665\n",
      "Epoch: 111, Batch: 900, Avg. Loss: 0.07902992814779282\n",
      "Epoch: 112, Batch: 100, Avg. Loss: 0.08022459685802459\n",
      "Epoch: 112, Batch: 200, Avg. Loss: 0.07058383427560329\n",
      "Epoch: 112, Batch: 300, Avg. Loss: 0.0749061656743288\n",
      "Epoch: 112, Batch: 400, Avg. Loss: 0.06563160389661789\n",
      "Epoch: 112, Batch: 500, Avg. Loss: 0.07379179447889328\n",
      "Epoch: 112, Batch: 600, Avg. Loss: 0.07400281935930252\n",
      "Epoch: 112, Batch: 700, Avg. Loss: 0.06949330888688564\n",
      "Epoch: 112, Batch: 800, Avg. Loss: 0.08026807323098183\n",
      "Epoch: 112, Batch: 900, Avg. Loss: 0.06804024182260036\n",
      "Epoch: 113, Batch: 100, Avg. Loss: 0.08425365760922432\n",
      "Epoch: 113, Batch: 200, Avg. Loss: 0.07176376797258854\n",
      "Epoch: 113, Batch: 300, Avg. Loss: 0.08636749424040317\n",
      "Epoch: 113, Batch: 400, Avg. Loss: 0.06337762907147408\n",
      "Epoch: 113, Batch: 500, Avg. Loss: 0.07517019376158714\n",
      "Epoch: 113, Batch: 600, Avg. Loss: 0.07061209566891194\n",
      "Epoch: 113, Batch: 700, Avg. Loss: 0.07135394543409347\n",
      "Epoch: 113, Batch: 800, Avg. Loss: 0.06295356601476669\n",
      "Epoch: 113, Batch: 900, Avg. Loss: 0.07430858373641967\n",
      "Epoch: 114, Batch: 100, Avg. Loss: 0.06764345586299897\n",
      "Epoch: 114, Batch: 200, Avg. Loss: 0.07786424547433853\n",
      "Epoch: 114, Batch: 300, Avg. Loss: 0.08177361495792866\n",
      "Epoch: 114, Batch: 400, Avg. Loss: 0.08171088084578514\n",
      "Epoch: 114, Batch: 500, Avg. Loss: 0.06580545626580715\n",
      "Epoch: 114, Batch: 600, Avg. Loss: 0.06633518695831299\n",
      "Epoch: 114, Batch: 700, Avg. Loss: 0.07658221460878849\n",
      "Epoch: 114, Batch: 800, Avg. Loss: 0.06331944420933723\n",
      "Epoch: 114, Batch: 900, Avg. Loss: 0.07060964077711106\n",
      "Epoch: 115, Batch: 100, Avg. Loss: 0.0739161230623722\n",
      "Epoch: 115, Batch: 200, Avg. Loss: 0.07574106365442276\n",
      "Epoch: 115, Batch: 300, Avg. Loss: 0.07081130050122737\n",
      "Epoch: 115, Batch: 400, Avg. Loss: 0.06348075322806836\n",
      "Epoch: 115, Batch: 500, Avg. Loss: 0.07017737627029419\n",
      "Epoch: 115, Batch: 600, Avg. Loss: 0.06891433343291282\n",
      "Epoch: 115, Batch: 700, Avg. Loss: 0.07238820374011994\n",
      "Epoch: 115, Batch: 800, Avg. Loss: 0.05789982572197914\n",
      "Epoch: 115, Batch: 900, Avg. Loss: 0.06913986824452877\n",
      "Epoch: 116, Batch: 100, Avg. Loss: 0.0604795228689909\n",
      "Epoch: 116, Batch: 200, Avg. Loss: 0.07082220271229744\n",
      "Epoch: 116, Batch: 300, Avg. Loss: 0.06509189583361148\n",
      "Epoch: 116, Batch: 400, Avg. Loss: 0.06984334036707879\n",
      "Epoch: 116, Batch: 500, Avg. Loss: 0.07616009809076786\n",
      "Epoch: 116, Batch: 600, Avg. Loss: 0.06707836240530014\n",
      "Epoch: 116, Batch: 700, Avg. Loss: 0.074463145211339\n",
      "Epoch: 116, Batch: 800, Avg. Loss: 0.06987421177327632\n",
      "Epoch: 116, Batch: 900, Avg. Loss: 0.0679595746845007\n",
      "Epoch: 117, Batch: 100, Avg. Loss: 0.06443539403378963\n",
      "Epoch: 117, Batch: 200, Avg. Loss: 0.06478295169770718\n",
      "Epoch: 117, Batch: 300, Avg. Loss: 0.06936023220419883\n",
      "Epoch: 117, Batch: 400, Avg. Loss: 0.07466841392219066\n",
      "Epoch: 117, Batch: 500, Avg. Loss: 0.0723423657566309\n",
      "Epoch: 117, Batch: 600, Avg. Loss: 0.07517700977623462\n",
      "Epoch: 117, Batch: 700, Avg. Loss: 0.08345717757940292\n",
      "Epoch: 117, Batch: 800, Avg. Loss: 0.0743456769734621\n",
      "Epoch: 117, Batch: 900, Avg. Loss: 0.07913797676563263\n",
      "Epoch: 118, Batch: 100, Avg. Loss: 0.06553767308592796\n",
      "Epoch: 118, Batch: 200, Avg. Loss: 0.06755980893969536\n",
      "Epoch: 118, Batch: 300, Avg. Loss: 0.07195237815380097\n",
      "Epoch: 118, Batch: 400, Avg. Loss: 0.06956743411719798\n",
      "Epoch: 118, Batch: 500, Avg. Loss: 0.06519652090966702\n",
      "Epoch: 118, Batch: 600, Avg. Loss: 0.06620361328125\n",
      "Epoch: 118, Batch: 700, Avg. Loss: 0.08366641268134117\n",
      "Epoch: 118, Batch: 800, Avg. Loss: 0.07659854672849178\n",
      "Epoch: 118, Batch: 900, Avg. Loss: 0.07344917394220829\n",
      "Epoch: 119, Batch: 100, Avg. Loss: 0.06534670285880566\n",
      "Epoch: 119, Batch: 200, Avg. Loss: 0.07172932282090187\n",
      "Epoch: 119, Batch: 300, Avg. Loss: 0.07394148372113704\n",
      "Epoch: 119, Batch: 400, Avg. Loss: 0.06643546976149083\n",
      "Epoch: 119, Batch: 500, Avg. Loss: 0.0763529348373413\n",
      "Epoch: 119, Batch: 600, Avg. Loss: 0.0780513446033001\n",
      "Epoch: 119, Batch: 700, Avg. Loss: 0.07507905796170235\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 119, Batch: 800, Avg. Loss: 0.06725186362862587\n",
      "Epoch: 119, Batch: 900, Avg. Loss: 0.07036679781973362\n",
      "Epoch: 120, Batch: 100, Avg. Loss: 0.06347520589828491\n",
      "Epoch: 120, Batch: 200, Avg. Loss: 0.07426285333931446\n",
      "Epoch: 120, Batch: 300, Avg. Loss: 0.06632909417152405\n",
      "Epoch: 120, Batch: 400, Avg. Loss: 0.0783884820342064\n",
      "Epoch: 120, Batch: 500, Avg. Loss: 0.07442217491567135\n",
      "Epoch: 120, Batch: 600, Avg. Loss: 0.07467479825019836\n",
      "Epoch: 120, Batch: 700, Avg. Loss: 0.07491760410368442\n",
      "Epoch: 120, Batch: 800, Avg. Loss: 0.07252309702336789\n",
      "Epoch: 120, Batch: 900, Avg. Loss: 0.06722900390625\n",
      "Epoch: 121, Batch: 100, Avg. Loss: 0.0784424439817667\n",
      "Epoch: 121, Batch: 200, Avg. Loss: 0.06825517579913139\n",
      "Epoch: 121, Batch: 300, Avg. Loss: 0.06598586022853851\n",
      "Epoch: 121, Batch: 400, Avg. Loss: 0.07673633560538291\n",
      "Epoch: 121, Batch: 500, Avg. Loss: 0.06168938964605331\n",
      "Epoch: 121, Batch: 600, Avg. Loss: 0.06685384973883629\n",
      "Epoch: 121, Batch: 700, Avg. Loss: 0.07229442790150642\n",
      "Epoch: 121, Batch: 800, Avg. Loss: 0.07124564856290817\n",
      "Epoch: 121, Batch: 900, Avg. Loss: 0.0631640162318945\n",
      "Epoch: 122, Batch: 100, Avg. Loss: 0.06795256420969963\n",
      "Epoch: 122, Batch: 200, Avg. Loss: 0.0630437808483839\n",
      "Epoch: 122, Batch: 300, Avg. Loss: 0.06922665014863014\n",
      "Epoch: 122, Batch: 400, Avg. Loss: 0.07390657268464565\n",
      "Epoch: 122, Batch: 500, Avg. Loss: 0.07004376277327537\n",
      "Epoch: 122, Batch: 600, Avg. Loss: 0.0774279810488224\n",
      "Epoch: 122, Batch: 700, Avg. Loss: 0.07208406321704387\n",
      "Epoch: 122, Batch: 800, Avg. Loss: 0.06928742587566376\n",
      "Epoch: 122, Batch: 900, Avg. Loss: 0.07981852024793624\n",
      "Epoch: 123, Batch: 100, Avg. Loss: 0.07226617321372032\n",
      "Epoch: 123, Batch: 200, Avg. Loss: 0.0730302556604147\n",
      "Epoch: 123, Batch: 300, Avg. Loss: 0.0731358328461647\n",
      "Epoch: 123, Batch: 400, Avg. Loss: 0.07044607311487198\n",
      "Epoch: 123, Batch: 500, Avg. Loss: 0.0766354263573885\n",
      "Epoch: 123, Batch: 600, Avg. Loss: 0.07079072870314121\n",
      "Epoch: 123, Batch: 700, Avg. Loss: 0.07238785825669765\n",
      "Epoch: 123, Batch: 800, Avg. Loss: 0.07520100831985474\n",
      "Epoch: 123, Batch: 900, Avg. Loss: 0.0632480301707983\n",
      "Epoch: 124, Batch: 100, Avg. Loss: 0.06711502090096473\n",
      "Epoch: 124, Batch: 200, Avg. Loss: 0.06816273093223572\n",
      "Epoch: 124, Batch: 300, Avg. Loss: 0.06256666243076324\n",
      "Epoch: 124, Batch: 400, Avg. Loss: 0.0668516544252634\n",
      "Epoch: 124, Batch: 500, Avg. Loss: 0.06918772265315055\n",
      "Epoch: 124, Batch: 600, Avg. Loss: 0.07640997409820556\n",
      "Epoch: 124, Batch: 700, Avg. Loss: 0.07163873180747032\n",
      "Epoch: 124, Batch: 800, Avg. Loss: 0.06671496972441673\n",
      "Epoch: 124, Batch: 900, Avg. Loss: 0.0709060452133417\n",
      "Epoch: 125, Batch: 100, Avg. Loss: 0.06293701998889446\n",
      "Epoch: 125, Batch: 200, Avg. Loss: 0.07184813819825649\n",
      "Epoch: 125, Batch: 300, Avg. Loss: 0.07899818040430545\n",
      "Epoch: 125, Batch: 400, Avg. Loss: 0.07034320779144763\n",
      "Epoch: 125, Batch: 500, Avg. Loss: 0.07926050856709481\n",
      "Epoch: 125, Batch: 600, Avg. Loss: 0.06331148229539395\n",
      "Epoch: 125, Batch: 700, Avg. Loss: 0.07120996452867985\n",
      "Epoch: 125, Batch: 800, Avg. Loss: 0.05889168471097946\n",
      "Epoch: 125, Batch: 900, Avg. Loss: 0.06722719378769398\n",
      "Epoch: 126, Batch: 100, Avg. Loss: 0.05806387938559055\n",
      "Epoch: 126, Batch: 200, Avg. Loss: 0.07456321619451046\n",
      "Epoch: 126, Batch: 300, Avg. Loss: 0.0682410428673029\n",
      "Epoch: 126, Batch: 400, Avg. Loss: 0.07342550113797187\n",
      "Epoch: 126, Batch: 500, Avg. Loss: 0.06615974679589272\n",
      "Epoch: 126, Batch: 600, Avg. Loss: 0.07880991168320178\n",
      "Epoch: 126, Batch: 700, Avg. Loss: 0.06900617450475693\n",
      "Epoch: 126, Batch: 800, Avg. Loss: 0.07023652948439121\n",
      "Epoch: 126, Batch: 900, Avg. Loss: 0.06481550171971322\n",
      "Epoch: 127, Batch: 100, Avg. Loss: 0.06418684795498848\n",
      "Epoch: 127, Batch: 200, Avg. Loss: 0.0619710098952055\n",
      "Epoch: 127, Batch: 300, Avg. Loss: 0.07838575322180986\n",
      "Epoch: 127, Batch: 400, Avg. Loss: 0.08361786775290966\n",
      "Epoch: 127, Batch: 500, Avg. Loss: 0.06953663140535354\n",
      "Epoch: 127, Batch: 600, Avg. Loss: 0.068621876090765\n",
      "Epoch: 127, Batch: 700, Avg. Loss: 0.07048601850867271\n",
      "Epoch: 127, Batch: 800, Avg. Loss: 0.06141094289720059\n",
      "Epoch: 127, Batch: 900, Avg. Loss: 0.0684944960474968\n",
      "Epoch: 128, Batch: 100, Avg. Loss: 0.06657374799251556\n",
      "Epoch: 128, Batch: 200, Avg. Loss: 0.07215653903782368\n",
      "Epoch: 128, Batch: 300, Avg. Loss: 0.06300534464418889\n",
      "Epoch: 128, Batch: 400, Avg. Loss: 0.06761721104383468\n",
      "Epoch: 128, Batch: 500, Avg. Loss: 0.06243236809968948\n",
      "Epoch: 128, Batch: 600, Avg. Loss: 0.056114470735192296\n",
      "Epoch: 128, Batch: 700, Avg. Loss: 0.07255187422037125\n",
      "Epoch: 128, Batch: 800, Avg. Loss: 0.07440149277448654\n",
      "Epoch: 128, Batch: 900, Avg. Loss: 0.06702285699546337\n",
      "Epoch: 129, Batch: 100, Avg. Loss: 0.08067350372672082\n",
      "Epoch: 129, Batch: 200, Avg. Loss: 0.06774172462522983\n",
      "Epoch: 129, Batch: 300, Avg. Loss: 0.0701944500952959\n",
      "Epoch: 129, Batch: 400, Avg. Loss: 0.06389572247862815\n",
      "Epoch: 129, Batch: 500, Avg. Loss: 0.06436630919575691\n",
      "Epoch: 129, Batch: 600, Avg. Loss: 0.063001374527812\n",
      "Epoch: 129, Batch: 700, Avg. Loss: 0.07488754250109196\n",
      "Epoch: 129, Batch: 800, Avg. Loss: 0.07058239169418812\n",
      "Epoch: 129, Batch: 900, Avg. Loss: 0.06298529617488384\n",
      "Epoch: 130, Batch: 100, Avg. Loss: 0.05399598754942417\n",
      "Epoch: 130, Batch: 200, Avg. Loss: 0.06500491417944432\n",
      "Epoch: 130, Batch: 300, Avg. Loss: 0.06588956013321877\n",
      "Epoch: 130, Batch: 400, Avg. Loss: 0.07317130111157893\n",
      "Epoch: 130, Batch: 500, Avg. Loss: 0.06458618491888046\n",
      "Epoch: 130, Batch: 600, Avg. Loss: 0.07506037540733815\n",
      "Epoch: 130, Batch: 700, Avg. Loss: 0.06539667740464211\n",
      "Epoch: 130, Batch: 800, Avg. Loss: 0.07341618932783604\n",
      "Epoch: 130, Batch: 900, Avg. Loss: 0.06114599071443081\n",
      "Epoch: 131, Batch: 100, Avg. Loss: 0.07068077817559243\n",
      "Epoch: 131, Batch: 200, Avg. Loss: 0.07039971448481083\n",
      "Epoch: 131, Batch: 300, Avg. Loss: 0.07100127220153808\n",
      "Epoch: 131, Batch: 400, Avg. Loss: 0.061926794201135636\n",
      "Epoch: 131, Batch: 500, Avg. Loss: 0.06651947438716889\n",
      "Epoch: 131, Batch: 600, Avg. Loss: 0.06244409792125225\n",
      "Epoch: 131, Batch: 700, Avg. Loss: 0.06143596112728119\n",
      "Epoch: 131, Batch: 800, Avg. Loss: 0.0726386358588934\n",
      "Epoch: 131, Batch: 900, Avg. Loss: 0.06775645852088928\n",
      "Epoch: 132, Batch: 100, Avg. Loss: 0.07287359692156314\n",
      "Epoch: 132, Batch: 200, Avg. Loss: 0.07675149716436863\n",
      "Epoch: 132, Batch: 300, Avg. Loss: 0.0745134861022234\n",
      "Epoch: 132, Batch: 400, Avg. Loss: 0.058036531433463094\n",
      "Epoch: 132, Batch: 500, Avg. Loss: 0.06569284111261368\n",
      "Epoch: 132, Batch: 600, Avg. Loss: 0.07212257646024227\n",
      "Epoch: 132, Batch: 700, Avg. Loss: 0.06370239414274692\n",
      "Epoch: 132, Batch: 800, Avg. Loss: 0.07162324190139771\n",
      "Epoch: 132, Batch: 900, Avg. Loss: 0.06823568589985371\n",
      "Epoch: 133, Batch: 100, Avg. Loss: 0.07680459879338741\n",
      "Epoch: 133, Batch: 200, Avg. Loss: 0.05971161186695099\n",
      "Epoch: 133, Batch: 300, Avg. Loss: 0.06464518889784814\n",
      "Epoch: 133, Batch: 400, Avg. Loss: 0.0716592675447464\n",
      "Epoch: 133, Batch: 500, Avg. Loss: 0.06781615920364857\n",
      "Epoch: 133, Batch: 600, Avg. Loss: 0.06528641119599342\n",
      "Epoch: 133, Batch: 700, Avg. Loss: 0.07004026480019093\n",
      "Epoch: 133, Batch: 800, Avg. Loss: 0.07162967994809151\n",
      "Epoch: 133, Batch: 900, Avg. Loss: 0.06810532577335834\n",
      "Epoch: 134, Batch: 100, Avg. Loss: 0.05714050143957138\n",
      "Epoch: 134, Batch: 200, Avg. Loss: 0.07502423085272313\n",
      "Epoch: 134, Batch: 300, Avg. Loss: 0.07258702620863915\n",
      "Epoch: 134, Batch: 400, Avg. Loss: 0.06292165540158749\n",
      "Epoch: 134, Batch: 500, Avg. Loss: 0.06583533719182015\n",
      "Epoch: 134, Batch: 600, Avg. Loss: 0.07007198311388492\n",
      "Epoch: 134, Batch: 700, Avg. Loss: 0.055191625878214835\n",
      "Epoch: 134, Batch: 800, Avg. Loss: 0.05827026836574078\n",
      "Epoch: 134, Batch: 900, Avg. Loss: 0.07446887657046318\n",
      "Epoch: 135, Batch: 100, Avg. Loss: 0.06613794699311257\n",
      "Epoch: 135, Batch: 200, Avg. Loss: 0.06259878143668175\n",
      "Epoch: 135, Batch: 300, Avg. Loss: 0.05788875378668308\n",
      "Epoch: 135, Batch: 400, Avg. Loss: 0.06785036444664001\n",
      "Epoch: 135, Batch: 500, Avg. Loss: 0.06546083405613899\n",
      "Epoch: 135, Batch: 600, Avg. Loss: 0.07324361443519592\n",
      "Epoch: 135, Batch: 700, Avg. Loss: 0.06023036979138851\n",
      "Epoch: 135, Batch: 800, Avg. Loss: 0.06445863530039787\n",
      "Epoch: 135, Batch: 900, Avg. Loss: 0.06495969772338867\n",
      "Epoch: 136, Batch: 100, Avg. Loss: 0.0652741389721632\n",
      "Epoch: 136, Batch: 200, Avg. Loss: 0.05671932369470596\n",
      "Epoch: 136, Batch: 300, Avg. Loss: 0.06746159754693508\n",
      "Epoch: 136, Batch: 400, Avg. Loss: 0.06540403217077255\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 136, Batch: 500, Avg. Loss: 0.0752901667356491\n",
      "Epoch: 136, Batch: 600, Avg. Loss: 0.06310301728546619\n",
      "Epoch: 136, Batch: 700, Avg. Loss: 0.06653887890279293\n",
      "Epoch: 136, Batch: 800, Avg. Loss: 0.07363317415118217\n",
      "Epoch: 136, Batch: 900, Avg. Loss: 0.06184304941445589\n",
      "Epoch: 137, Batch: 100, Avg. Loss: 0.06889468163251877\n",
      "Epoch: 137, Batch: 200, Avg. Loss: 0.07244569204747676\n",
      "Epoch: 137, Batch: 300, Avg. Loss: 0.0655221989005804\n",
      "Epoch: 137, Batch: 400, Avg. Loss: 0.06582921624183655\n",
      "Epoch: 137, Batch: 500, Avg. Loss: 0.06395795904099941\n",
      "Epoch: 137, Batch: 600, Avg. Loss: 0.06892592951655388\n",
      "Epoch: 137, Batch: 700, Avg. Loss: 0.08145619891583919\n",
      "Epoch: 137, Batch: 800, Avg. Loss: 0.0630673298984766\n",
      "Epoch: 137, Batch: 900, Avg. Loss: 0.07130118250846863\n",
      "Epoch: 138, Batch: 100, Avg. Loss: 0.06437689766287803\n",
      "Epoch: 138, Batch: 200, Avg. Loss: 0.0714372131973505\n",
      "Epoch: 138, Batch: 300, Avg. Loss: 0.06927613005042076\n",
      "Epoch: 138, Batch: 400, Avg. Loss: 0.07560664817690849\n",
      "Epoch: 138, Batch: 500, Avg. Loss: 0.06581292353570461\n",
      "Epoch: 138, Batch: 600, Avg. Loss: 0.06312725096940994\n",
      "Epoch: 138, Batch: 700, Avg. Loss: 0.0658270214498043\n",
      "Epoch: 138, Batch: 800, Avg. Loss: 0.06810354992747307\n",
      "Epoch: 138, Batch: 900, Avg. Loss: 0.062065240517258646\n",
      "Epoch: 139, Batch: 100, Avg. Loss: 0.06864201076328755\n",
      "Epoch: 139, Batch: 200, Avg. Loss: 0.06202627956867218\n",
      "Epoch: 139, Batch: 300, Avg. Loss: 0.06693285048007964\n",
      "Epoch: 139, Batch: 400, Avg. Loss: 0.06460695467889309\n",
      "Epoch: 139, Batch: 500, Avg. Loss: 0.0761235798895359\n",
      "Epoch: 139, Batch: 600, Avg. Loss: 0.06477062173187732\n",
      "Epoch: 139, Batch: 700, Avg. Loss: 0.06961966320872306\n",
      "Epoch: 139, Batch: 800, Avg. Loss: 0.06421896338462829\n",
      "Epoch: 139, Batch: 900, Avg. Loss: 0.062503364905715\n",
      "Epoch: 140, Batch: 100, Avg. Loss: 0.06834047257900239\n",
      "Epoch: 140, Batch: 200, Avg. Loss: 0.06582529373466968\n",
      "Epoch: 140, Batch: 300, Avg. Loss: 0.0659245815128088\n",
      "Epoch: 140, Batch: 400, Avg. Loss: 0.062197150737047194\n",
      "Epoch: 140, Batch: 500, Avg. Loss: 0.055197421461343765\n",
      "Epoch: 140, Batch: 600, Avg. Loss: 0.0668510165065527\n",
      "Epoch: 140, Batch: 700, Avg. Loss: 0.07483389690518379\n",
      "Epoch: 140, Batch: 800, Avg. Loss: 0.06915616020560264\n",
      "Epoch: 140, Batch: 900, Avg. Loss: 0.0639802097529173\n",
      "Epoch: 141, Batch: 100, Avg. Loss: 0.06970328219234943\n",
      "Epoch: 141, Batch: 200, Avg. Loss: 0.05952899225056171\n",
      "Epoch: 141, Batch: 300, Avg. Loss: 0.06518504031002521\n",
      "Epoch: 141, Batch: 400, Avg. Loss: 0.06565343052148818\n",
      "Epoch: 141, Batch: 500, Avg. Loss: 0.06355020739138126\n",
      "Epoch: 141, Batch: 600, Avg. Loss: 0.05930693440139294\n",
      "Epoch: 141, Batch: 700, Avg. Loss: 0.07303442664444447\n",
      "Epoch: 141, Batch: 800, Avg. Loss: 0.07734613515436649\n",
      "Epoch: 141, Batch: 900, Avg. Loss: 0.0671638411283493\n",
      "Epoch: 142, Batch: 100, Avg. Loss: 0.06996617130935193\n",
      "Epoch: 142, Batch: 200, Avg. Loss: 0.06978702500462532\n",
      "Epoch: 142, Batch: 300, Avg. Loss: 0.05865725792944431\n",
      "Epoch: 142, Batch: 400, Avg. Loss: 0.06361532330513\n",
      "Epoch: 142, Batch: 500, Avg. Loss: 0.05897309176623821\n",
      "Epoch: 142, Batch: 600, Avg. Loss: 0.06605988264083862\n",
      "Epoch: 142, Batch: 700, Avg. Loss: 0.059456108808517455\n",
      "Epoch: 142, Batch: 800, Avg. Loss: 0.07183109186589717\n",
      "Epoch: 142, Batch: 900, Avg. Loss: 0.0693423543870449\n",
      "Epoch: 143, Batch: 100, Avg. Loss: 0.06875168360769748\n",
      "Epoch: 143, Batch: 200, Avg. Loss: 0.07420275181531906\n",
      "Epoch: 143, Batch: 300, Avg. Loss: 0.06134214513003826\n",
      "Epoch: 143, Batch: 400, Avg. Loss: 0.05822446398437023\n",
      "Epoch: 143, Batch: 500, Avg. Loss: 0.058848361745476724\n",
      "Epoch: 143, Batch: 600, Avg. Loss: 0.07248529322445392\n",
      "Epoch: 143, Batch: 700, Avg. Loss: 0.06342444140464068\n",
      "Epoch: 143, Batch: 800, Avg. Loss: 0.05761261522769928\n",
      "Epoch: 143, Batch: 900, Avg. Loss: 0.07741094440221787\n",
      "Epoch: 144, Batch: 100, Avg. Loss: 0.06566946811974049\n",
      "Epoch: 144, Batch: 200, Avg. Loss: 0.06896743796765804\n",
      "Epoch: 144, Batch: 300, Avg. Loss: 0.07135217420756818\n",
      "Epoch: 144, Batch: 400, Avg. Loss: 0.0688166557252407\n",
      "Epoch: 144, Batch: 500, Avg. Loss: 0.08116576455533504\n",
      "Epoch: 144, Batch: 600, Avg. Loss: 0.06337637722492218\n",
      "Epoch: 144, Batch: 700, Avg. Loss: 0.06589015111327172\n",
      "Epoch: 144, Batch: 800, Avg. Loss: 0.05889700204133987\n",
      "Epoch: 144, Batch: 900, Avg. Loss: 0.06942634604871273\n",
      "Epoch: 145, Batch: 100, Avg. Loss: 0.07312155611813069\n",
      "Epoch: 145, Batch: 200, Avg. Loss: 0.06287626713514329\n",
      "Epoch: 145, Batch: 300, Avg. Loss: 0.06960189551115036\n",
      "Epoch: 145, Batch: 400, Avg. Loss: 0.06450586818158627\n",
      "Epoch: 145, Batch: 500, Avg. Loss: 0.06013934887945652\n",
      "Epoch: 145, Batch: 600, Avg. Loss: 0.061400501206517216\n",
      "Epoch: 145, Batch: 700, Avg. Loss: 0.05929086677730083\n",
      "Epoch: 145, Batch: 800, Avg. Loss: 0.061161757484078406\n",
      "Epoch: 145, Batch: 900, Avg. Loss: 0.05899295561015606\n",
      "Epoch: 146, Batch: 100, Avg. Loss: 0.06556549154222012\n",
      "Epoch: 146, Batch: 200, Avg. Loss: 0.05777898259460926\n",
      "Epoch: 146, Batch: 300, Avg. Loss: 0.060944818928837775\n",
      "Epoch: 146, Batch: 400, Avg. Loss: 0.06143047578632831\n",
      "Epoch: 146, Batch: 500, Avg. Loss: 0.07265424683690071\n",
      "Epoch: 146, Batch: 600, Avg. Loss: 0.061995808966457844\n",
      "Epoch: 146, Batch: 700, Avg. Loss: 0.06881885327398778\n",
      "Epoch: 146, Batch: 800, Avg. Loss: 0.055632017627358435\n",
      "Epoch: 146, Batch: 900, Avg. Loss: 0.06764163181185723\n",
      "Epoch: 147, Batch: 100, Avg. Loss: 0.06068831361830235\n",
      "Epoch: 147, Batch: 200, Avg. Loss: 0.06623504862189293\n",
      "Epoch: 147, Batch: 300, Avg. Loss: 0.06263650842010975\n",
      "Epoch: 147, Batch: 400, Avg. Loss: 0.06907007105648517\n",
      "Epoch: 147, Batch: 500, Avg. Loss: 0.06905520021915436\n",
      "Epoch: 147, Batch: 600, Avg. Loss: 0.05136862702667713\n",
      "Epoch: 147, Batch: 700, Avg. Loss: 0.05531946554780007\n",
      "Epoch: 147, Batch: 800, Avg. Loss: 0.06185692258179188\n",
      "Epoch: 147, Batch: 900, Avg. Loss: 0.06864888451993466\n",
      "Epoch: 148, Batch: 100, Avg. Loss: 0.06686909355223179\n",
      "Epoch: 148, Batch: 200, Avg. Loss: 0.06435250490903854\n",
      "Epoch: 148, Batch: 300, Avg. Loss: 0.05325794011354446\n",
      "Epoch: 148, Batch: 400, Avg. Loss: 0.06466174140572548\n",
      "Epoch: 148, Batch: 500, Avg. Loss: 0.056829545348882675\n",
      "Epoch: 148, Batch: 600, Avg. Loss: 0.061281043142080306\n",
      "Epoch: 148, Batch: 700, Avg. Loss: 0.06744832672178745\n",
      "Epoch: 148, Batch: 800, Avg. Loss: 0.0634994514286518\n",
      "Epoch: 148, Batch: 900, Avg. Loss: 0.06800701834261418\n",
      "Epoch: 149, Batch: 100, Avg. Loss: 0.05893488138914108\n",
      "Epoch: 149, Batch: 200, Avg. Loss: 0.05960126794874668\n",
      "Epoch: 149, Batch: 300, Avg. Loss: 0.06145140148699284\n",
      "Epoch: 149, Batch: 400, Avg. Loss: 0.06152530118823051\n",
      "Epoch: 149, Batch: 500, Avg. Loss: 0.06718853630125522\n",
      "Epoch: 149, Batch: 600, Avg. Loss: 0.06406247697770595\n",
      "Epoch: 149, Batch: 700, Avg. Loss: 0.06819423303008079\n",
      "Epoch: 149, Batch: 800, Avg. Loss: 0.06397769503295421\n",
      "Epoch: 149, Batch: 900, Avg. Loss: 0.06003906607627869\n",
      "Epoch: 150, Batch: 100, Avg. Loss: 0.061683632507920264\n",
      "Epoch: 150, Batch: 200, Avg. Loss: 0.05600216880440712\n",
      "Epoch: 150, Batch: 300, Avg. Loss: 0.05795300588011742\n",
      "Epoch: 150, Batch: 400, Avg. Loss: 0.06468823380768299\n",
      "Epoch: 150, Batch: 500, Avg. Loss: 0.06721796326339245\n",
      "Epoch: 150, Batch: 600, Avg. Loss: 0.06461235709488392\n",
      "Epoch: 150, Batch: 700, Avg. Loss: 0.0599170146137476\n",
      "Epoch: 150, Batch: 800, Avg. Loss: 0.05424511574208737\n",
      "Epoch: 150, Batch: 900, Avg. Loss: 0.06784129314124585\n",
      "Epoch: 151, Batch: 100, Avg. Loss: 0.06638340901583434\n",
      "Epoch: 151, Batch: 200, Avg. Loss: 0.06295843146741391\n",
      "Epoch: 151, Batch: 300, Avg. Loss: 0.0599877567589283\n",
      "Epoch: 151, Batch: 400, Avg. Loss: 0.07116353511810303\n",
      "Epoch: 151, Batch: 500, Avg. Loss: 0.06558686830103397\n",
      "Epoch: 151, Batch: 600, Avg. Loss: 0.06430424496531487\n",
      "Epoch: 151, Batch: 700, Avg. Loss: 0.06718838617205619\n",
      "Epoch: 151, Batch: 800, Avg. Loss: 0.0695201188325882\n",
      "Epoch: 151, Batch: 900, Avg. Loss: 0.060060298889875414\n",
      "Epoch: 152, Batch: 100, Avg. Loss: 0.05853890113532543\n",
      "Epoch: 152, Batch: 200, Avg. Loss: 0.0578738072514534\n",
      "Epoch: 152, Batch: 300, Avg. Loss: 0.05843774080276489\n",
      "Epoch: 152, Batch: 400, Avg. Loss: 0.0631519179791212\n",
      "Epoch: 152, Batch: 500, Avg. Loss: 0.061049806848168374\n",
      "Epoch: 152, Batch: 600, Avg. Loss: 0.07190700829029083\n",
      "Epoch: 152, Batch: 700, Avg. Loss: 0.05970068745315075\n",
      "Epoch: 152, Batch: 800, Avg. Loss: 0.06463659189641476\n",
      "Epoch: 152, Batch: 900, Avg. Loss: 0.06279176786541939\n",
      "Epoch: 153, Batch: 100, Avg. Loss: 0.06355479069054126\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 153, Batch: 200, Avg. Loss: 0.06295179642736912\n",
      "Epoch: 153, Batch: 300, Avg. Loss: 0.059618789702653885\n",
      "Epoch: 153, Batch: 400, Avg. Loss: 0.07032025150954724\n",
      "Epoch: 153, Batch: 500, Avg. Loss: 0.06578468464314938\n",
      "Epoch: 153, Batch: 600, Avg. Loss: 0.05223383605480194\n",
      "Epoch: 153, Batch: 700, Avg. Loss: 0.06117271050810814\n",
      "Epoch: 153, Batch: 800, Avg. Loss: 0.05948520727455616\n",
      "Epoch: 153, Batch: 900, Avg. Loss: 0.06330914929509163\n",
      "Epoch: 154, Batch: 100, Avg. Loss: 0.05700251489877701\n",
      "Epoch: 154, Batch: 200, Avg. Loss: 0.061202216073870656\n",
      "Epoch: 154, Batch: 300, Avg. Loss: 0.061496431529521944\n",
      "Epoch: 154, Batch: 400, Avg. Loss: 0.061718795970082285\n",
      "Epoch: 154, Batch: 500, Avg. Loss: 0.06637732535600663\n",
      "Epoch: 154, Batch: 600, Avg. Loss: 0.06101634901016951\n",
      "Epoch: 154, Batch: 700, Avg. Loss: 0.05829626373946667\n",
      "Epoch: 154, Batch: 800, Avg. Loss: 0.07796517752110958\n",
      "Epoch: 154, Batch: 900, Avg. Loss: 0.06030705764889717\n",
      "Epoch: 155, Batch: 100, Avg. Loss: 0.06792815178632736\n",
      "Epoch: 155, Batch: 200, Avg. Loss: 0.0661602695286274\n",
      "Epoch: 155, Batch: 300, Avg. Loss: 0.06221213258802891\n",
      "Epoch: 155, Batch: 400, Avg. Loss: 0.05742317646741867\n",
      "Epoch: 155, Batch: 500, Avg. Loss: 0.05668342463672161\n",
      "Epoch: 155, Batch: 600, Avg. Loss: 0.06293027341365814\n",
      "Epoch: 155, Batch: 700, Avg. Loss: 0.057535371594130996\n",
      "Epoch: 155, Batch: 800, Avg. Loss: 0.06526266105473041\n",
      "Epoch: 155, Batch: 900, Avg. Loss: 0.058229640647768977\n",
      "Epoch: 156, Batch: 100, Avg. Loss: 0.0561261223256588\n",
      "Epoch: 156, Batch: 200, Avg. Loss: 0.06923918299376965\n",
      "Epoch: 156, Batch: 300, Avg. Loss: 0.05328394398093224\n",
      "Epoch: 156, Batch: 400, Avg. Loss: 0.0657001481950283\n",
      "Epoch: 156, Batch: 500, Avg. Loss: 0.06282964624464511\n",
      "Epoch: 156, Batch: 600, Avg. Loss: 0.068438760638237\n",
      "Epoch: 156, Batch: 700, Avg. Loss: 0.0622829744964838\n",
      "Epoch: 156, Batch: 800, Avg. Loss: 0.06389536872506142\n",
      "Epoch: 156, Batch: 900, Avg. Loss: 0.06212347872555256\n",
      "Epoch: 157, Batch: 100, Avg. Loss: 0.05860265776515007\n",
      "Epoch: 157, Batch: 200, Avg. Loss: 0.05906131718307733\n",
      "Epoch: 157, Batch: 300, Avg. Loss: 0.059307461753487584\n",
      "Epoch: 157, Batch: 400, Avg. Loss: 0.05677076660096645\n",
      "Epoch: 157, Batch: 500, Avg. Loss: 0.057708462700247765\n",
      "Epoch: 157, Batch: 600, Avg. Loss: 0.0644379536062479\n",
      "Epoch: 157, Batch: 700, Avg. Loss: 0.05899364531040192\n",
      "Epoch: 157, Batch: 800, Avg. Loss: 0.06827102988958358\n",
      "Epoch: 157, Batch: 900, Avg. Loss: 0.07327713251113892\n",
      "Epoch: 158, Batch: 100, Avg. Loss: 0.06669031620025635\n",
      "Epoch: 158, Batch: 200, Avg. Loss: 0.06568980731070041\n",
      "Epoch: 158, Batch: 300, Avg. Loss: 0.06253703117370606\n",
      "Epoch: 158, Batch: 400, Avg. Loss: 0.06606008715927601\n",
      "Epoch: 158, Batch: 500, Avg. Loss: 0.060775330662727355\n",
      "Epoch: 158, Batch: 600, Avg. Loss: 0.05089426189661026\n",
      "Epoch: 158, Batch: 700, Avg. Loss: 0.055132598429918286\n",
      "Epoch: 158, Batch: 800, Avg. Loss: 0.06561734609305858\n",
      "Epoch: 158, Batch: 900, Avg. Loss: 0.07399036101996899\n",
      "Epoch: 159, Batch: 100, Avg. Loss: 0.06618556626141071\n",
      "Epoch: 159, Batch: 200, Avg. Loss: 0.06530977062880992\n",
      "Epoch: 159, Batch: 300, Avg. Loss: 0.057178245410323146\n",
      "Epoch: 159, Batch: 400, Avg. Loss: 0.049314408674836156\n",
      "Epoch: 159, Batch: 500, Avg. Loss: 0.0581886900216341\n",
      "Epoch: 159, Batch: 600, Avg. Loss: 0.0652896012365818\n",
      "Epoch: 159, Batch: 700, Avg. Loss: 0.06027035877108574\n",
      "Epoch: 159, Batch: 800, Avg. Loss: 0.06089627705514431\n",
      "Epoch: 159, Batch: 900, Avg. Loss: 0.06458696112036705\n",
      "Epoch: 160, Batch: 100, Avg. Loss: 0.062061293944716454\n",
      "Epoch: 160, Batch: 200, Avg. Loss: 0.06536636494100094\n",
      "Epoch: 160, Batch: 300, Avg. Loss: 0.060092444643378255\n",
      "Epoch: 160, Batch: 400, Avg. Loss: 0.054411829933524135\n",
      "Epoch: 160, Batch: 500, Avg. Loss: 0.05839580096304417\n",
      "Epoch: 160, Batch: 600, Avg. Loss: 0.05797695182263851\n",
      "Epoch: 160, Batch: 700, Avg. Loss: 0.07274129912257195\n",
      "Epoch: 160, Batch: 800, Avg. Loss: 0.05290874592959881\n",
      "Epoch: 160, Batch: 900, Avg. Loss: 0.061014150828123094\n",
      "Epoch: 161, Batch: 100, Avg. Loss: 0.06296625673770904\n",
      "Epoch: 161, Batch: 200, Avg. Loss: 0.06241356611251831\n",
      "Epoch: 161, Batch: 300, Avg. Loss: 0.057752591893076896\n",
      "Epoch: 161, Batch: 400, Avg. Loss: 0.06170785762369633\n",
      "Epoch: 161, Batch: 500, Avg. Loss: 0.0574410405755043\n",
      "Epoch: 161, Batch: 600, Avg. Loss: 0.05647663578391075\n",
      "Epoch: 161, Batch: 700, Avg. Loss: 0.057090566977858546\n",
      "Epoch: 161, Batch: 800, Avg. Loss: 0.06593452408909797\n",
      "Epoch: 161, Batch: 900, Avg. Loss: 0.05226677216589451\n",
      "Epoch: 162, Batch: 100, Avg. Loss: 0.05246004417538643\n",
      "Epoch: 162, Batch: 200, Avg. Loss: 0.06856278367340565\n",
      "Epoch: 162, Batch: 300, Avg. Loss: 0.05980122275650501\n",
      "Epoch: 162, Batch: 400, Avg. Loss: 0.0620052395761013\n",
      "Epoch: 162, Batch: 500, Avg. Loss: 0.05500999815762043\n",
      "Epoch: 162, Batch: 600, Avg. Loss: 0.06478793792426586\n",
      "Epoch: 162, Batch: 700, Avg. Loss: 0.06546111941337586\n",
      "Epoch: 162, Batch: 800, Avg. Loss: 0.06250864677131177\n",
      "Epoch: 162, Batch: 900, Avg. Loss: 0.0556025954335928\n",
      "Epoch: 163, Batch: 100, Avg. Loss: 0.06036895863711834\n",
      "Epoch: 163, Batch: 200, Avg. Loss: 0.06632501974701882\n",
      "Epoch: 163, Batch: 300, Avg. Loss: 0.05646895311772823\n",
      "Epoch: 163, Batch: 400, Avg. Loss: 0.06012155249714851\n",
      "Epoch: 163, Batch: 500, Avg. Loss: 0.06565108634531498\n",
      "Epoch: 163, Batch: 600, Avg. Loss: 0.0509444859623909\n",
      "Epoch: 163, Batch: 700, Avg. Loss: 0.05383781209588051\n",
      "Epoch: 163, Batch: 800, Avg. Loss: 0.07293275065720081\n",
      "Epoch: 163, Batch: 900, Avg. Loss: 0.05871815953403711\n",
      "Epoch: 164, Batch: 100, Avg. Loss: 0.06036493308842182\n",
      "Epoch: 164, Batch: 200, Avg. Loss: 0.060780057311058046\n",
      "Epoch: 164, Batch: 300, Avg. Loss: 0.05816265232861042\n",
      "Epoch: 164, Batch: 400, Avg. Loss: 0.062398421838879586\n",
      "Epoch: 164, Batch: 500, Avg. Loss: 0.045082395747303966\n",
      "Epoch: 164, Batch: 600, Avg. Loss: 0.059985561147332195\n",
      "Epoch: 164, Batch: 700, Avg. Loss: 0.07792673975229264\n",
      "Epoch: 164, Batch: 800, Avg. Loss: 0.0734520860761404\n",
      "Epoch: 164, Batch: 900, Avg. Loss: 0.06637150004506111\n",
      "Epoch: 165, Batch: 100, Avg. Loss: 0.06313456781208515\n",
      "Epoch: 165, Batch: 200, Avg. Loss: 0.06690101757645607\n",
      "Epoch: 165, Batch: 300, Avg. Loss: 0.05386080250144005\n",
      "Epoch: 165, Batch: 400, Avg. Loss: 0.06255491212010383\n",
      "Epoch: 165, Batch: 500, Avg. Loss: 0.05957041673362255\n",
      "Epoch: 165, Batch: 600, Avg. Loss: 0.05587472412735224\n",
      "Epoch: 165, Batch: 700, Avg. Loss: 0.06413019709289074\n",
      "Epoch: 165, Batch: 800, Avg. Loss: 0.05987948626279831\n",
      "Epoch: 165, Batch: 900, Avg. Loss: 0.054341351985931395\n",
      "Epoch: 166, Batch: 100, Avg. Loss: 0.0611830298602581\n",
      "Epoch: 166, Batch: 200, Avg. Loss: 0.05696335576474667\n",
      "Epoch: 166, Batch: 300, Avg. Loss: 0.05632258161902428\n",
      "Epoch: 166, Batch: 400, Avg. Loss: 0.05417931325733662\n",
      "Epoch: 166, Batch: 500, Avg. Loss: 0.058907154798507694\n",
      "Epoch: 166, Batch: 600, Avg. Loss: 0.06252852477133274\n",
      "Epoch: 166, Batch: 700, Avg. Loss: 0.06001369722187519\n",
      "Epoch: 166, Batch: 800, Avg. Loss: 0.059770165607333184\n",
      "Epoch: 166, Batch: 900, Avg. Loss: 0.061153248623013495\n",
      "Epoch: 167, Batch: 100, Avg. Loss: 0.051566163450479506\n",
      "Epoch: 167, Batch: 200, Avg. Loss: 0.06850608997046947\n",
      "Epoch: 167, Batch: 300, Avg. Loss: 0.05491748668253422\n",
      "Epoch: 167, Batch: 400, Avg. Loss: 0.05094578541815281\n",
      "Epoch: 167, Batch: 500, Avg. Loss: 0.05268615961074829\n",
      "Epoch: 167, Batch: 600, Avg. Loss: 0.0568871733546257\n",
      "Epoch: 167, Batch: 700, Avg. Loss: 0.06379789307713508\n",
      "Epoch: 167, Batch: 800, Avg. Loss: 0.061011391654610635\n",
      "Epoch: 167, Batch: 900, Avg. Loss: 0.06475005440413951\n",
      "Epoch: 168, Batch: 100, Avg. Loss: 0.05282448902726174\n",
      "Epoch: 168, Batch: 200, Avg. Loss: 0.07020750567317009\n",
      "Epoch: 168, Batch: 300, Avg. Loss: 0.06389647290110588\n",
      "Epoch: 168, Batch: 400, Avg. Loss: 0.06576761335134507\n",
      "Epoch: 168, Batch: 500, Avg. Loss: 0.06063389874994755\n",
      "Epoch: 168, Batch: 600, Avg. Loss: 0.05995866626501083\n",
      "Epoch: 168, Batch: 700, Avg. Loss: 0.05834092326462269\n",
      "Epoch: 168, Batch: 800, Avg. Loss: 0.06028256542980671\n",
      "Epoch: 168, Batch: 900, Avg. Loss: 0.0635804782062769\n",
      "Epoch: 169, Batch: 100, Avg. Loss: 0.05160005643963814\n",
      "Epoch: 169, Batch: 200, Avg. Loss: 0.056403139904141426\n",
      "Epoch: 169, Batch: 300, Avg. Loss: 0.05843253508210182\n",
      "Epoch: 169, Batch: 400, Avg. Loss: 0.06585593961179256\n",
      "Epoch: 169, Batch: 500, Avg. Loss: 0.0588939843326807\n",
      "Epoch: 169, Batch: 600, Avg. Loss: 0.062152086198329924\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 169, Batch: 700, Avg. Loss: 0.0654049164801836\n",
      "Epoch: 169, Batch: 800, Avg. Loss: 0.05798391595482826\n",
      "Epoch: 169, Batch: 900, Avg. Loss: 0.057504934072494504\n",
      "Epoch: 170, Batch: 100, Avg. Loss: 0.053991583585739134\n",
      "Epoch: 170, Batch: 200, Avg. Loss: 0.06679007850587368\n",
      "Epoch: 170, Batch: 300, Avg. Loss: 0.06099972106516361\n",
      "Epoch: 170, Batch: 400, Avg. Loss: 0.06510869607329368\n",
      "Epoch: 170, Batch: 500, Avg. Loss: 0.0616281258314848\n",
      "Epoch: 170, Batch: 600, Avg. Loss: 0.061297760829329494\n",
      "Epoch: 170, Batch: 700, Avg. Loss: 0.06000059276819229\n",
      "Epoch: 170, Batch: 800, Avg. Loss: 0.05709680952131748\n",
      "Epoch: 170, Batch: 900, Avg. Loss: 0.05003064446151256\n",
      "Epoch: 171, Batch: 100, Avg. Loss: 0.05969335541129112\n",
      "Epoch: 171, Batch: 200, Avg. Loss: 0.06649347268044949\n",
      "Epoch: 171, Batch: 300, Avg. Loss: 0.054024682566523555\n",
      "Epoch: 171, Batch: 400, Avg. Loss: 0.05314726382493973\n",
      "Epoch: 171, Batch: 500, Avg. Loss: 0.05897950179874897\n",
      "Epoch: 171, Batch: 600, Avg. Loss: 0.06563225977122783\n",
      "Epoch: 171, Batch: 700, Avg. Loss: 0.06160526379942894\n",
      "Epoch: 171, Batch: 800, Avg. Loss: 0.0513689124584198\n",
      "Epoch: 171, Batch: 900, Avg. Loss: 0.06284961186349391\n",
      "Epoch: 172, Batch: 100, Avg. Loss: 0.05118809577077627\n",
      "Epoch: 172, Batch: 200, Avg. Loss: 0.0692526113241911\n",
      "Epoch: 172, Batch: 300, Avg. Loss: 0.060545573756098744\n",
      "Epoch: 172, Batch: 400, Avg. Loss: 0.06343320779502391\n",
      "Epoch: 172, Batch: 500, Avg. Loss: 0.0567205673456192\n",
      "Epoch: 172, Batch: 600, Avg. Loss: 0.05599205486476421\n",
      "Epoch: 172, Batch: 700, Avg. Loss: 0.05786860175430775\n",
      "Epoch: 172, Batch: 800, Avg. Loss: 0.05680822864174843\n",
      "Epoch: 172, Batch: 900, Avg. Loss: 0.061457356289029125\n",
      "Epoch: 173, Batch: 100, Avg. Loss: 0.05458930864930153\n",
      "Epoch: 173, Batch: 200, Avg. Loss: 0.060206276029348374\n",
      "Epoch: 173, Batch: 300, Avg. Loss: 0.056303013414144516\n",
      "Epoch: 173, Batch: 400, Avg. Loss: 0.05358950279653072\n",
      "Epoch: 173, Batch: 500, Avg. Loss: 0.059072755202651026\n",
      "Epoch: 173, Batch: 600, Avg. Loss: 0.05801167078316212\n",
      "Epoch: 173, Batch: 700, Avg. Loss: 0.0672726758569479\n",
      "Epoch: 173, Batch: 800, Avg. Loss: 0.060749659687280654\n",
      "Epoch: 173, Batch: 900, Avg. Loss: 0.06311099119484424\n",
      "Epoch: 174, Batch: 100, Avg. Loss: 0.058544959872961044\n",
      "Epoch: 174, Batch: 200, Avg. Loss: 0.054078133627772335\n",
      "Epoch: 174, Batch: 300, Avg. Loss: 0.061433408856391904\n",
      "Epoch: 174, Batch: 400, Avg. Loss: 0.06211258478462696\n",
      "Epoch: 174, Batch: 500, Avg. Loss: 0.05517732009291649\n",
      "Epoch: 174, Batch: 600, Avg. Loss: 0.05963301114737987\n",
      "Epoch: 174, Batch: 700, Avg. Loss: 0.06416617788374424\n",
      "Epoch: 174, Batch: 800, Avg. Loss: 0.06210880905389786\n",
      "Epoch: 174, Batch: 900, Avg. Loss: 0.057097019851207735\n",
      "Epoch: 175, Batch: 100, Avg. Loss: 0.06967797093093395\n",
      "Epoch: 175, Batch: 200, Avg. Loss: 0.06245253302156925\n",
      "Epoch: 175, Batch: 300, Avg. Loss: 0.06872936680912972\n",
      "Epoch: 175, Batch: 400, Avg. Loss: 0.053488798812031745\n",
      "Epoch: 175, Batch: 500, Avg. Loss: 0.056877851113677026\n",
      "Epoch: 175, Batch: 600, Avg. Loss: 0.06193395122885704\n",
      "Epoch: 175, Batch: 700, Avg. Loss: 0.06660307817161083\n",
      "Epoch: 175, Batch: 800, Avg. Loss: 0.059159092232584955\n",
      "Epoch: 175, Batch: 900, Avg. Loss: 0.05912631779909134\n",
      "Epoch: 176, Batch: 100, Avg. Loss: 0.06045635424554348\n",
      "Epoch: 176, Batch: 200, Avg. Loss: 0.060215872824192045\n",
      "Epoch: 176, Batch: 300, Avg. Loss: 0.059868770837783816\n",
      "Epoch: 176, Batch: 400, Avg. Loss: 0.056859551072120665\n",
      "Epoch: 176, Batch: 500, Avg. Loss: 0.06538819395005703\n",
      "Epoch: 176, Batch: 600, Avg. Loss: 0.05907546766102314\n",
      "Epoch: 176, Batch: 700, Avg. Loss: 0.06137863665819168\n",
      "Epoch: 176, Batch: 800, Avg. Loss: 0.06629897907376289\n",
      "Epoch: 176, Batch: 900, Avg. Loss: 0.05552573852241039\n",
      "Epoch: 177, Batch: 100, Avg. Loss: 0.07001748241484165\n",
      "Epoch: 177, Batch: 200, Avg. Loss: 0.052183981165289876\n",
      "Epoch: 177, Batch: 300, Avg. Loss: 0.062007455229759215\n",
      "Epoch: 177, Batch: 400, Avg. Loss: 0.04951945461332798\n",
      "Epoch: 177, Batch: 500, Avg. Loss: 0.05657733432948589\n",
      "Epoch: 177, Batch: 600, Avg. Loss: 0.05873151555657387\n",
      "Epoch: 177, Batch: 700, Avg. Loss: 0.058767445981502534\n",
      "Epoch: 177, Batch: 800, Avg. Loss: 0.053543604388833044\n",
      "Epoch: 177, Batch: 900, Avg. Loss: 0.06408048614859581\n",
      "Epoch: 178, Batch: 100, Avg. Loss: 0.06176084712147713\n",
      "Epoch: 178, Batch: 200, Avg. Loss: 0.06656313002109528\n",
      "Epoch: 178, Batch: 300, Avg. Loss: 0.05844153366982937\n",
      "Epoch: 178, Batch: 400, Avg. Loss: 0.05483278878033161\n",
      "Epoch: 178, Batch: 500, Avg. Loss: 0.05407602198421955\n",
      "Epoch: 178, Batch: 600, Avg. Loss: 0.06596428461372852\n",
      "Epoch: 178, Batch: 700, Avg. Loss: 0.05651401199400425\n",
      "Epoch: 178, Batch: 800, Avg. Loss: 0.06288903735578061\n",
      "Epoch: 178, Batch: 900, Avg. Loss: 0.06575783275067806\n",
      "Epoch: 179, Batch: 100, Avg. Loss: 0.05386230029165745\n",
      "Epoch: 179, Batch: 200, Avg. Loss: 0.05511788196861744\n",
      "Epoch: 179, Batch: 300, Avg. Loss: 0.05681047983467579\n",
      "Epoch: 179, Batch: 400, Avg. Loss: 0.055040905848145485\n",
      "Epoch: 179, Batch: 500, Avg. Loss: 0.06467813171446324\n",
      "Epoch: 179, Batch: 600, Avg. Loss: 0.05045266062021256\n",
      "Epoch: 179, Batch: 700, Avg. Loss: 0.058709882497787476\n",
      "Epoch: 179, Batch: 800, Avg. Loss: 0.06066141165792942\n",
      "Epoch: 179, Batch: 900, Avg. Loss: 0.05971714988350868\n",
      "Epoch: 180, Batch: 100, Avg. Loss: 0.05683459766209126\n",
      "Epoch: 180, Batch: 200, Avg. Loss: 0.06191774696111679\n",
      "Epoch: 180, Batch: 300, Avg. Loss: 0.053275193944573404\n",
      "Epoch: 180, Batch: 400, Avg. Loss: 0.05886628814041615\n",
      "Epoch: 180, Batch: 500, Avg. Loss: 0.057199164479970935\n",
      "Epoch: 180, Batch: 600, Avg. Loss: 0.0604141590744257\n",
      "Epoch: 180, Batch: 700, Avg. Loss: 0.059602914080023765\n",
      "Epoch: 180, Batch: 800, Avg. Loss: 0.05874586507678032\n",
      "Epoch: 180, Batch: 900, Avg. Loss: 0.063059092015028\n",
      "Epoch: 181, Batch: 100, Avg. Loss: 0.06224163740873337\n",
      "Epoch: 181, Batch: 200, Avg. Loss: 0.0556486739218235\n",
      "Epoch: 181, Batch: 300, Avg. Loss: 0.05469832867383957\n",
      "Epoch: 181, Batch: 400, Avg. Loss: 0.05720728740096092\n",
      "Epoch: 181, Batch: 500, Avg. Loss: 0.05538108162581921\n",
      "Epoch: 181, Batch: 600, Avg. Loss: 0.05693063713610172\n",
      "Epoch: 181, Batch: 700, Avg. Loss: 0.06852111957967282\n",
      "Epoch: 181, Batch: 800, Avg. Loss: 0.05306626692414284\n",
      "Epoch: 181, Batch: 900, Avg. Loss: 0.06411444202065468\n",
      "Epoch: 182, Batch: 100, Avg. Loss: 0.05902511656284332\n",
      "Epoch: 182, Batch: 200, Avg. Loss: 0.05732918992638588\n",
      "Epoch: 182, Batch: 300, Avg. Loss: 0.04965013846755028\n",
      "Epoch: 182, Batch: 400, Avg. Loss: 0.057781901210546494\n",
      "Epoch: 182, Batch: 500, Avg. Loss: 0.05240130014717579\n",
      "Epoch: 182, Batch: 600, Avg. Loss: 0.05680088721215725\n",
      "Epoch: 182, Batch: 700, Avg. Loss: 0.06557595305144787\n",
      "Epoch: 182, Batch: 800, Avg. Loss: 0.057916870713233946\n",
      "Epoch: 182, Batch: 900, Avg. Loss: 0.05745098635554314\n",
      "Epoch: 183, Batch: 100, Avg. Loss: 0.05243149757385254\n",
      "Epoch: 183, Batch: 200, Avg. Loss: 0.0565536081045866\n",
      "Epoch: 183, Batch: 300, Avg. Loss: 0.05296230830252171\n",
      "Epoch: 183, Batch: 400, Avg. Loss: 0.06445021100342274\n",
      "Epoch: 183, Batch: 500, Avg. Loss: 0.0669716488569975\n",
      "Epoch: 183, Batch: 600, Avg. Loss: 0.05475886650383473\n",
      "Epoch: 183, Batch: 700, Avg. Loss: 0.058805918619036675\n",
      "Epoch: 183, Batch: 800, Avg. Loss: 0.051977620124816895\n",
      "Epoch: 183, Batch: 900, Avg. Loss: 0.05950354732573032\n",
      "Epoch: 184, Batch: 100, Avg. Loss: 0.054427553117275235\n",
      "Epoch: 184, Batch: 200, Avg. Loss: 0.0594830010086298\n",
      "Epoch: 184, Batch: 300, Avg. Loss: 0.05490344390273094\n",
      "Epoch: 184, Batch: 400, Avg. Loss: 0.05693390399217606\n",
      "Epoch: 184, Batch: 500, Avg. Loss: 0.05712200455367565\n",
      "Epoch: 184, Batch: 600, Avg. Loss: 0.04814353831112385\n",
      "Epoch: 184, Batch: 700, Avg. Loss: 0.05902147971093655\n",
      "Epoch: 184, Batch: 800, Avg. Loss: 0.05567671962082386\n",
      "Epoch: 184, Batch: 900, Avg. Loss: 0.05469281107187271\n",
      "Epoch: 185, Batch: 100, Avg. Loss: 0.04885767735540867\n",
      "Epoch: 185, Batch: 200, Avg. Loss: 0.06243185184895992\n",
      "Epoch: 185, Batch: 300, Avg. Loss: 0.05467801958322525\n",
      "Epoch: 185, Batch: 400, Avg. Loss: 0.06229550667107105\n",
      "Epoch: 185, Batch: 500, Avg. Loss: 0.05175211362540722\n",
      "Epoch: 185, Batch: 600, Avg. Loss: 0.051709686815738676\n",
      "Epoch: 185, Batch: 700, Avg. Loss: 0.05219517394900322\n",
      "Epoch: 185, Batch: 800, Avg. Loss: 0.06372504822909832\n",
      "Epoch: 185, Batch: 900, Avg. Loss: 0.055638949871063235\n",
      "Epoch: 186, Batch: 100, Avg. Loss: 0.057668366879224775\n",
      "Epoch: 186, Batch: 200, Avg. Loss: 0.05901206396520138\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 186, Batch: 300, Avg. Loss: 0.06739446587860584\n",
      "Epoch: 186, Batch: 400, Avg. Loss: 0.048790399804711344\n",
      "Epoch: 186, Batch: 500, Avg. Loss: 0.05661242678761482\n",
      "Epoch: 186, Batch: 600, Avg. Loss: 0.06060613542795181\n",
      "Epoch: 186, Batch: 700, Avg. Loss: 0.05188808932900429\n",
      "Epoch: 186, Batch: 800, Avg. Loss: 0.0567309457808733\n",
      "Epoch: 186, Batch: 900, Avg. Loss: 0.048833827152848244\n",
      "Epoch: 187, Batch: 100, Avg. Loss: 0.051054080948233604\n",
      "Epoch: 187, Batch: 200, Avg. Loss: 0.06356025904417038\n",
      "Epoch: 187, Batch: 300, Avg. Loss: 0.05402003102004528\n",
      "Epoch: 187, Batch: 400, Avg. Loss: 0.057300465553998946\n",
      "Epoch: 187, Batch: 500, Avg. Loss: 0.05276441387832165\n",
      "Epoch: 187, Batch: 600, Avg. Loss: 0.05683871269226074\n",
      "Epoch: 187, Batch: 700, Avg. Loss: 0.05237342596054077\n",
      "Epoch: 187, Batch: 800, Avg. Loss: 0.05790216416120529\n",
      "Epoch: 187, Batch: 900, Avg. Loss: 0.06104411393404007\n",
      "Epoch: 188, Batch: 100, Avg. Loss: 0.05509056515991688\n",
      "Epoch: 188, Batch: 200, Avg. Loss: 0.05235688574612141\n",
      "Epoch: 188, Batch: 300, Avg. Loss: 0.05886736154556274\n",
      "Epoch: 188, Batch: 400, Avg. Loss: 0.05436340115964413\n",
      "Epoch: 188, Batch: 500, Avg. Loss: 0.05994970858097076\n",
      "Epoch: 188, Batch: 600, Avg. Loss: 0.056922266408801075\n",
      "Epoch: 188, Batch: 700, Avg. Loss: 0.056701976731419565\n",
      "Epoch: 188, Batch: 800, Avg. Loss: 0.05573224514722824\n",
      "Epoch: 188, Batch: 900, Avg. Loss: 0.05287982575595379\n",
      "Epoch: 189, Batch: 100, Avg. Loss: 0.05254409149289131\n",
      "Epoch: 189, Batch: 200, Avg. Loss: 0.04979601413011551\n",
      "Epoch: 189, Batch: 300, Avg. Loss: 0.05697724349796772\n",
      "Epoch: 189, Batch: 400, Avg. Loss: 0.05620078429579735\n",
      "Epoch: 189, Batch: 500, Avg. Loss: 0.05157770231366157\n",
      "Epoch: 189, Batch: 600, Avg. Loss: 0.05438764549791813\n",
      "Epoch: 189, Batch: 700, Avg. Loss: 0.058473284617066384\n",
      "Epoch: 189, Batch: 800, Avg. Loss: 0.06502529099583626\n",
      "Epoch: 189, Batch: 900, Avg. Loss: 0.05631408296525478\n",
      "Epoch: 190, Batch: 100, Avg. Loss: 0.0527132361382246\n",
      "Epoch: 190, Batch: 200, Avg. Loss: 0.0602541846036911\n",
      "Epoch: 190, Batch: 300, Avg. Loss: 0.052087765708565714\n",
      "Epoch: 190, Batch: 400, Avg. Loss: 0.05711142435669899\n",
      "Epoch: 190, Batch: 500, Avg. Loss: 0.05452348832041025\n",
      "Epoch: 190, Batch: 600, Avg. Loss: 0.05593467354774475\n",
      "Epoch: 190, Batch: 700, Avg. Loss: 0.05383265629410744\n",
      "Epoch: 190, Batch: 800, Avg. Loss: 0.056061483770608905\n",
      "Epoch: 190, Batch: 900, Avg. Loss: 0.0601990782469511\n",
      "Epoch: 191, Batch: 100, Avg. Loss: 0.05856095425784588\n",
      "Epoch: 191, Batch: 200, Avg. Loss: 0.05361665792763233\n",
      "Epoch: 191, Batch: 300, Avg. Loss: 0.058075961619615556\n",
      "Epoch: 191, Batch: 400, Avg. Loss: 0.05065147466957569\n",
      "Epoch: 191, Batch: 500, Avg. Loss: 0.05686917468905449\n",
      "Epoch: 191, Batch: 600, Avg. Loss: 0.05759013347327709\n",
      "Epoch: 191, Batch: 700, Avg. Loss: 0.060425479412078854\n",
      "Epoch: 191, Batch: 800, Avg. Loss: 0.05887604132294655\n",
      "Epoch: 191, Batch: 900, Avg. Loss: 0.05856767639517784\n",
      "Epoch: 192, Batch: 100, Avg. Loss: 0.058191855400800706\n",
      "Epoch: 192, Batch: 200, Avg. Loss: 0.06256723076105118\n",
      "Epoch: 192, Batch: 300, Avg. Loss: 0.057016384825110435\n",
      "Epoch: 192, Batch: 400, Avg. Loss: 0.056896453723311424\n",
      "Epoch: 192, Batch: 500, Avg. Loss: 0.05675101958215237\n",
      "Epoch: 192, Batch: 600, Avg. Loss: 0.053307707533240316\n",
      "Epoch: 192, Batch: 700, Avg. Loss: 0.05591925822198391\n",
      "Epoch: 192, Batch: 800, Avg. Loss: 0.05165972523391247\n",
      "Epoch: 192, Batch: 900, Avg. Loss: 0.04640487439930439\n",
      "Epoch: 193, Batch: 100, Avg. Loss: 0.0556135243922472\n",
      "Epoch: 193, Batch: 200, Avg. Loss: 0.059312563091516494\n",
      "Epoch: 193, Batch: 300, Avg. Loss: 0.052712939754128454\n",
      "Epoch: 193, Batch: 400, Avg. Loss: 0.04684925958514213\n",
      "Epoch: 193, Batch: 500, Avg. Loss: 0.0681251646578312\n",
      "Epoch: 193, Batch: 600, Avg. Loss: 0.05717313349246979\n",
      "Epoch: 193, Batch: 700, Avg. Loss: 0.05403338328003884\n",
      "Epoch: 193, Batch: 800, Avg. Loss: 0.0542211326956749\n",
      "Epoch: 193, Batch: 900, Avg. Loss: 0.053999201506376264\n",
      "Epoch: 194, Batch: 100, Avg. Loss: 0.06006038472056389\n",
      "Epoch: 194, Batch: 200, Avg. Loss: 0.05287828117609024\n",
      "Epoch: 194, Batch: 300, Avg. Loss: 0.05522327803075314\n",
      "Epoch: 194, Batch: 400, Avg. Loss: 0.04727191805839539\n",
      "Epoch: 194, Batch: 500, Avg. Loss: 0.05207956805825233\n",
      "Epoch: 194, Batch: 600, Avg. Loss: 0.05348314933478832\n",
      "Epoch: 194, Batch: 700, Avg. Loss: 0.0569811137765646\n",
      "Epoch: 194, Batch: 800, Avg. Loss: 0.060072841867804525\n",
      "Epoch: 194, Batch: 900, Avg. Loss: 0.059241990447044375\n",
      "Epoch: 195, Batch: 100, Avg. Loss: 0.04815074935555458\n",
      "Epoch: 195, Batch: 200, Avg. Loss: 0.049074341729283336\n",
      "Epoch: 195, Batch: 300, Avg. Loss: 0.05951243065297604\n",
      "Epoch: 195, Batch: 400, Avg. Loss: 0.047224951758980754\n",
      "Epoch: 195, Batch: 500, Avg. Loss: 0.05839487470686436\n",
      "Epoch: 195, Batch: 600, Avg. Loss: 0.05552735820412636\n",
      "Epoch: 195, Batch: 700, Avg. Loss: 0.06777351684868335\n",
      "Epoch: 195, Batch: 800, Avg. Loss: 0.062363949567079545\n",
      "Epoch: 195, Batch: 900, Avg. Loss: 0.047980440482497215\n",
      "Epoch: 196, Batch: 100, Avg. Loss: 0.054105701223015785\n",
      "Epoch: 196, Batch: 200, Avg. Loss: 0.06114606231451034\n",
      "Epoch: 196, Batch: 300, Avg. Loss: 0.05666742250323296\n",
      "Epoch: 196, Batch: 400, Avg. Loss: 0.05108551748096943\n",
      "Epoch: 196, Batch: 500, Avg. Loss: 0.06127856690436602\n",
      "Epoch: 196, Batch: 600, Avg. Loss: 0.05691051594913006\n",
      "Epoch: 196, Batch: 700, Avg. Loss: 0.051667410656809805\n",
      "Epoch: 196, Batch: 800, Avg. Loss: 0.06287336111068725\n",
      "Epoch: 196, Batch: 900, Avg. Loss: 0.050120131522417066\n",
      "Epoch: 197, Batch: 100, Avg. Loss: 0.0644952617585659\n",
      "Epoch: 197, Batch: 200, Avg. Loss: 0.05394494295120239\n",
      "Epoch: 197, Batch: 300, Avg. Loss: 0.055619241446256636\n",
      "Epoch: 197, Batch: 400, Avg. Loss: 0.05380984850227833\n",
      "Epoch: 197, Batch: 500, Avg. Loss: 0.05943986900150776\n",
      "Epoch: 197, Batch: 600, Avg. Loss: 0.055504227504134175\n",
      "Epoch: 197, Batch: 700, Avg. Loss: 0.05549032099545002\n",
      "Epoch: 197, Batch: 800, Avg. Loss: 0.04975523099303245\n",
      "Epoch: 197, Batch: 900, Avg. Loss: 0.04698704808950424\n",
      "Epoch: 198, Batch: 100, Avg. Loss: 0.05585992209613323\n",
      "Epoch: 198, Batch: 200, Avg. Loss: 0.05691495209932327\n",
      "Epoch: 198, Batch: 300, Avg. Loss: 0.05050647459924221\n",
      "Epoch: 198, Batch: 400, Avg. Loss: 0.05254918642342091\n",
      "Epoch: 198, Batch: 500, Avg. Loss: 0.049957565665245056\n",
      "Epoch: 198, Batch: 600, Avg. Loss: 0.057667276263237\n",
      "Epoch: 198, Batch: 700, Avg. Loss: 0.05866521954536438\n",
      "Epoch: 198, Batch: 800, Avg. Loss: 0.04609151884913445\n",
      "Epoch: 198, Batch: 900, Avg. Loss: 0.05604123316705227\n",
      "Epoch: 199, Batch: 100, Avg. Loss: 0.05145429566502571\n",
      "Epoch: 199, Batch: 200, Avg. Loss: 0.0494625137001276\n",
      "Epoch: 199, Batch: 300, Avg. Loss: 0.05568924754858017\n",
      "Epoch: 199, Batch: 400, Avg. Loss: 0.04864248678088188\n",
      "Epoch: 199, Batch: 500, Avg. Loss: 0.05679601594805717\n",
      "Epoch: 199, Batch: 600, Avg. Loss: 0.05385196596384048\n",
      "Epoch: 199, Batch: 700, Avg. Loss: 0.058756529539823535\n",
      "Epoch: 199, Batch: 800, Avg. Loss: 0.05970119148492813\n",
      "Epoch: 199, Batch: 900, Avg. Loss: 0.05802773274481297\n",
      "Epoch: 200, Batch: 100, Avg. Loss: 0.06267897859215736\n",
      "Epoch: 200, Batch: 200, Avg. Loss: 0.05143707506358623\n",
      "Epoch: 200, Batch: 300, Avg. Loss: 0.06588361710309983\n",
      "Epoch: 200, Batch: 400, Avg. Loss: 0.05866435751318932\n",
      "Epoch: 200, Batch: 500, Avg. Loss: 0.06239733636379242\n",
      "Epoch: 200, Batch: 600, Avg. Loss: 0.05165240727365017\n",
      "Epoch: 200, Batch: 700, Avg. Loss: 0.05651536568999291\n",
      "Epoch: 200, Batch: 800, Avg. Loss: 0.04454174019396305\n",
      "Epoch: 200, Batch: 900, Avg. Loss: 0.053425138741731645\n",
      "Epoch: 201, Batch: 100, Avg. Loss: 0.05909088529646397\n",
      "Epoch: 201, Batch: 200, Avg. Loss: 0.05109731391072273\n",
      "Epoch: 201, Batch: 300, Avg. Loss: 0.05473703570663929\n",
      "Epoch: 201, Batch: 400, Avg. Loss: 0.05671507053077221\n",
      "Epoch: 201, Batch: 500, Avg. Loss: 0.051807061433792115\n",
      "Epoch: 201, Batch: 600, Avg. Loss: 0.057829762771725655\n",
      "Epoch: 201, Batch: 700, Avg. Loss: 0.04781572796404362\n",
      "Epoch: 201, Batch: 800, Avg. Loss: 0.05306077226996422\n",
      "Epoch: 201, Batch: 900, Avg. Loss: 0.06181289613246918\n",
      "Epoch: 202, Batch: 100, Avg. Loss: 0.06416965246200562\n",
      "Epoch: 202, Batch: 200, Avg. Loss: 0.05174818679690361\n",
      "Epoch: 202, Batch: 300, Avg. Loss: 0.05304269067943096\n",
      "Epoch: 202, Batch: 400, Avg. Loss: 0.059134476631879804\n",
      "Epoch: 202, Batch: 500, Avg. Loss: 0.058099354729056356\n",
      "Epoch: 202, Batch: 600, Avg. Loss: 0.057437163591384885\n",
      "Epoch: 202, Batch: 700, Avg. Loss: 0.05672607786953449\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 202, Batch: 800, Avg. Loss: 0.05031961746513844\n",
      "Epoch: 202, Batch: 900, Avg. Loss: 0.05521466240286827\n",
      "Epoch: 203, Batch: 100, Avg. Loss: 0.05734890386462212\n",
      "Epoch: 203, Batch: 200, Avg. Loss: 0.058144305497407914\n",
      "Epoch: 203, Batch: 300, Avg. Loss: 0.05620974116027355\n",
      "Epoch: 203, Batch: 400, Avg. Loss: 0.05666744202375412\n",
      "Epoch: 203, Batch: 500, Avg. Loss: 0.0601208983361721\n",
      "Epoch: 203, Batch: 600, Avg. Loss: 0.05110423766076565\n",
      "Epoch: 203, Batch: 700, Avg. Loss: 0.0551623997092247\n",
      "Epoch: 203, Batch: 800, Avg. Loss: 0.04826949156820774\n",
      "Epoch: 203, Batch: 900, Avg. Loss: 0.05272821366786957\n",
      "Epoch: 204, Batch: 100, Avg. Loss: 0.04707358233630657\n",
      "Epoch: 204, Batch: 200, Avg. Loss: 0.05616152510046959\n",
      "Epoch: 204, Batch: 300, Avg. Loss: 0.0512540727108717\n",
      "Epoch: 204, Batch: 400, Avg. Loss: 0.04976569473743439\n",
      "Epoch: 204, Batch: 500, Avg. Loss: 0.055158839672803876\n",
      "Epoch: 204, Batch: 600, Avg. Loss: 0.06651044577360153\n",
      "Epoch: 204, Batch: 700, Avg. Loss: 0.04208490520715714\n",
      "Epoch: 204, Batch: 800, Avg. Loss: 0.06748191513121128\n",
      "Epoch: 204, Batch: 900, Avg. Loss: 0.04656487300992012\n",
      "Epoch: 205, Batch: 100, Avg. Loss: 0.0514456956833601\n",
      "Epoch: 205, Batch: 200, Avg. Loss: 0.0422115533053875\n",
      "Epoch: 205, Batch: 300, Avg. Loss: 0.054196634143590924\n",
      "Epoch: 205, Batch: 400, Avg. Loss: 0.057407229468226434\n",
      "Epoch: 205, Batch: 500, Avg. Loss: 0.05344809629023075\n",
      "Epoch: 205, Batch: 600, Avg. Loss: 0.06315478183329105\n",
      "Epoch: 205, Batch: 700, Avg. Loss: 0.05088915541768074\n",
      "Epoch: 205, Batch: 800, Avg. Loss: 0.06093312755227089\n",
      "Epoch: 205, Batch: 900, Avg. Loss: 0.05980674110352993\n",
      "Epoch: 206, Batch: 100, Avg. Loss: 0.05057616025209427\n",
      "Epoch: 206, Batch: 200, Avg. Loss: 0.05108581647276878\n",
      "Epoch: 206, Batch: 300, Avg. Loss: 0.05422957956790924\n",
      "Epoch: 206, Batch: 400, Avg. Loss: 0.05419299304485321\n",
      "Epoch: 206, Batch: 500, Avg. Loss: 0.05046916380524635\n",
      "Epoch: 206, Batch: 600, Avg. Loss: 0.05729660019278526\n",
      "Epoch: 206, Batch: 700, Avg. Loss: 0.05406962588429451\n",
      "Epoch: 206, Batch: 800, Avg. Loss: 0.05204700388014317\n",
      "Epoch: 206, Batch: 900, Avg. Loss: 0.06068303160369396\n",
      "Epoch: 207, Batch: 100, Avg. Loss: 0.051107062324881554\n",
      "Epoch: 207, Batch: 200, Avg. Loss: 0.04635474607348442\n",
      "Epoch: 207, Batch: 300, Avg. Loss: 0.050899375975131986\n",
      "Epoch: 207, Batch: 400, Avg. Loss: 0.05687058873474598\n",
      "Epoch: 207, Batch: 500, Avg. Loss: 0.055616169571876525\n",
      "Epoch: 207, Batch: 600, Avg. Loss: 0.04874149575829506\n",
      "Epoch: 207, Batch: 700, Avg. Loss: 0.05893580920994282\n",
      "Epoch: 207, Batch: 800, Avg. Loss: 0.055236876010894775\n",
      "Epoch: 207, Batch: 900, Avg. Loss: 0.05523519165813923\n",
      "Epoch: 208, Batch: 100, Avg. Loss: 0.051539641618728635\n",
      "Epoch: 208, Batch: 200, Avg. Loss: 0.05037639528512955\n",
      "Epoch: 208, Batch: 300, Avg. Loss: 0.04906199023127556\n",
      "Epoch: 208, Batch: 400, Avg. Loss: 0.052809467911720274\n",
      "Epoch: 208, Batch: 500, Avg. Loss: 0.04921321287751198\n",
      "Epoch: 208, Batch: 600, Avg. Loss: 0.05328560508787632\n",
      "Epoch: 208, Batch: 700, Avg. Loss: 0.05223205417394638\n",
      "Epoch: 208, Batch: 800, Avg. Loss: 0.058375987708568576\n",
      "Epoch: 208, Batch: 900, Avg. Loss: 0.045285243541002274\n",
      "Epoch: 209, Batch: 100, Avg. Loss: 0.05390632301568985\n",
      "Epoch: 209, Batch: 200, Avg. Loss: 0.05051750957965851\n",
      "Epoch: 209, Batch: 300, Avg. Loss: 0.05142470598220825\n",
      "Epoch: 209, Batch: 400, Avg. Loss: 0.05164141125977039\n",
      "Epoch: 209, Batch: 500, Avg. Loss: 0.055918980911374094\n",
      "Epoch: 209, Batch: 600, Avg. Loss: 0.05061978884041309\n",
      "Epoch: 209, Batch: 700, Avg. Loss: 0.06292913414537907\n",
      "Epoch: 209, Batch: 800, Avg. Loss: 0.05189493753015995\n",
      "Epoch: 209, Batch: 900, Avg. Loss: 0.05690416149795055\n",
      "Epoch: 210, Batch: 100, Avg. Loss: 0.051195703968405726\n",
      "Epoch: 210, Batch: 200, Avg. Loss: 0.05116522409021854\n",
      "Epoch: 210, Batch: 300, Avg. Loss: 0.05684408865869045\n",
      "Epoch: 210, Batch: 400, Avg. Loss: 0.053011020123958585\n",
      "Epoch: 210, Batch: 500, Avg. Loss: 0.05322557881474495\n",
      "Epoch: 210, Batch: 600, Avg. Loss: 0.05645736299455166\n",
      "Epoch: 210, Batch: 700, Avg. Loss: 0.056267640590667724\n",
      "Epoch: 210, Batch: 800, Avg. Loss: 0.04991621010005474\n",
      "Epoch: 210, Batch: 900, Avg. Loss: 0.05584401622414589\n",
      "Epoch: 211, Batch: 100, Avg. Loss: 0.05345474988222122\n",
      "Epoch: 211, Batch: 200, Avg. Loss: 0.04828434675931931\n",
      "Epoch: 211, Batch: 300, Avg. Loss: 0.057565038278698924\n",
      "Epoch: 211, Batch: 400, Avg. Loss: 0.053626832664012906\n",
      "Epoch: 211, Batch: 500, Avg. Loss: 0.053451567366719244\n",
      "Epoch: 211, Batch: 600, Avg. Loss: 0.051431934610009195\n",
      "Epoch: 211, Batch: 700, Avg. Loss: 0.05230860762298107\n",
      "Epoch: 211, Batch: 800, Avg. Loss: 0.04911488056182861\n",
      "Epoch: 211, Batch: 900, Avg. Loss: 0.06019549265503883\n",
      "Epoch: 212, Batch: 100, Avg. Loss: 0.05276944473385811\n",
      "Epoch: 212, Batch: 200, Avg. Loss: 0.048028124794363976\n",
      "Epoch: 212, Batch: 300, Avg. Loss: 0.053138716965913775\n",
      "Epoch: 212, Batch: 400, Avg. Loss: 0.0476164448261261\n",
      "Epoch: 212, Batch: 500, Avg. Loss: 0.055974246561527254\n",
      "Epoch: 212, Batch: 600, Avg. Loss: 0.05729047052562237\n",
      "Epoch: 212, Batch: 700, Avg. Loss: 0.052944545298814774\n",
      "Epoch: 212, Batch: 800, Avg. Loss: 0.05404424868524074\n",
      "Epoch: 212, Batch: 900, Avg. Loss: 0.049933238998055456\n",
      "Epoch: 213, Batch: 100, Avg. Loss: 0.05057854227721691\n",
      "Epoch: 213, Batch: 200, Avg. Loss: 0.04866200916469097\n",
      "Epoch: 213, Batch: 300, Avg. Loss: 0.053185680881142616\n",
      "Epoch: 213, Batch: 400, Avg. Loss: 0.06260527648031712\n",
      "Epoch: 213, Batch: 500, Avg. Loss: 0.056552522704005244\n",
      "Epoch: 213, Batch: 600, Avg. Loss: 0.05168802909553051\n",
      "Epoch: 213, Batch: 700, Avg. Loss: 0.05577154368162155\n",
      "Epoch: 213, Batch: 800, Avg. Loss: 0.04780167236924172\n",
      "Epoch: 213, Batch: 900, Avg. Loss: 0.0544606077671051\n",
      "Epoch: 214, Batch: 100, Avg. Loss: 0.05590085074305534\n",
      "Epoch: 214, Batch: 200, Avg. Loss: 0.056218010857701305\n",
      "Epoch: 214, Batch: 300, Avg. Loss: 0.04905995987355709\n",
      "Epoch: 214, Batch: 400, Avg. Loss: 0.04530158266425133\n",
      "Epoch: 214, Batch: 500, Avg. Loss: 0.051188117116689684\n",
      "Epoch: 214, Batch: 600, Avg. Loss: 0.056436029225587846\n",
      "Epoch: 214, Batch: 700, Avg. Loss: 0.053649205043911935\n",
      "Epoch: 214, Batch: 800, Avg. Loss: 0.05105438001453876\n",
      "Epoch: 214, Batch: 900, Avg. Loss: 0.052353294119238857\n",
      "Epoch: 215, Batch: 100, Avg. Loss: 0.04968454271554947\n",
      "Epoch: 215, Batch: 200, Avg. Loss: 0.05469689324498177\n",
      "Epoch: 215, Batch: 300, Avg. Loss: 0.052299459278583524\n",
      "Epoch: 215, Batch: 400, Avg. Loss: 0.05509813874959946\n",
      "Epoch: 215, Batch: 500, Avg. Loss: 0.04782262340188027\n",
      "Epoch: 215, Batch: 600, Avg. Loss: 0.04800046421587467\n",
      "Epoch: 215, Batch: 700, Avg. Loss: 0.051563336327672\n",
      "Epoch: 215, Batch: 800, Avg. Loss: 0.058338313698768615\n",
      "Epoch: 215, Batch: 900, Avg. Loss: 0.051168968975543974\n",
      "Epoch: 216, Batch: 100, Avg. Loss: 0.05079751491546631\n",
      "Epoch: 216, Batch: 200, Avg. Loss: 0.04588949307799339\n",
      "Epoch: 216, Batch: 300, Avg. Loss: 0.047898639664053916\n",
      "Epoch: 216, Batch: 400, Avg. Loss: 0.05458367139101029\n",
      "Epoch: 216, Batch: 500, Avg. Loss: 0.053279553800821305\n",
      "Epoch: 216, Batch: 600, Avg. Loss: 0.05541250601410866\n",
      "Epoch: 216, Batch: 700, Avg. Loss: 0.053333111256361004\n",
      "Epoch: 216, Batch: 800, Avg. Loss: 0.04741669848561287\n",
      "Epoch: 216, Batch: 900, Avg. Loss: 0.05008215770125389\n",
      "Epoch: 217, Batch: 100, Avg. Loss: 0.058515749648213385\n",
      "Epoch: 217, Batch: 200, Avg. Loss: 0.05537046931684017\n",
      "Epoch: 217, Batch: 300, Avg. Loss: 0.05513736605644226\n",
      "Epoch: 217, Batch: 400, Avg. Loss: 0.04326724141836166\n",
      "Epoch: 217, Batch: 500, Avg. Loss: 0.04621673472225666\n",
      "Epoch: 217, Batch: 600, Avg. Loss: 0.05489349536597729\n",
      "Epoch: 217, Batch: 700, Avg. Loss: 0.05295475572347641\n",
      "Epoch: 217, Batch: 800, Avg. Loss: 0.0648676810413599\n",
      "Epoch: 217, Batch: 900, Avg. Loss: 0.05425698906183243\n",
      "Epoch: 218, Batch: 100, Avg. Loss: 0.050804036036133766\n",
      "Epoch: 218, Batch: 200, Avg. Loss: 0.05267488084733486\n",
      "Epoch: 218, Batch: 300, Avg. Loss: 0.05295935764908791\n",
      "Epoch: 218, Batch: 400, Avg. Loss: 0.04944665495306253\n",
      "Epoch: 218, Batch: 500, Avg. Loss: 0.05025999642908573\n",
      "Epoch: 218, Batch: 600, Avg. Loss: 0.048945776745676996\n",
      "Epoch: 218, Batch: 700, Avg. Loss: 0.052354052215814594\n",
      "Epoch: 218, Batch: 800, Avg. Loss: 0.043373335152864456\n",
      "Epoch: 218, Batch: 900, Avg. Loss: 0.04879963122308254\n",
      "Epoch: 219, Batch: 100, Avg. Loss: 0.051285747438669205\n",
      "Epoch: 219, Batch: 200, Avg. Loss: 0.054093682765960695\n",
      "Epoch: 219, Batch: 300, Avg. Loss: 0.052308687493205074\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 219, Batch: 400, Avg. Loss: 0.054972888976335524\n",
      "Epoch: 219, Batch: 500, Avg. Loss: 0.05258723683655262\n",
      "Epoch: 219, Batch: 600, Avg. Loss: 0.05168424472212792\n",
      "Epoch: 219, Batch: 700, Avg. Loss: 0.051123807430267336\n",
      "Epoch: 219, Batch: 800, Avg. Loss: 0.05110387034714222\n",
      "Epoch: 219, Batch: 900, Avg. Loss: 0.051077077239751814\n",
      "Epoch: 220, Batch: 100, Avg. Loss: 0.05060171782970428\n",
      "Epoch: 220, Batch: 200, Avg. Loss: 0.04945165164768696\n",
      "Epoch: 220, Batch: 300, Avg. Loss: 0.05449155293405056\n",
      "Epoch: 220, Batch: 400, Avg. Loss: 0.04953581668436527\n",
      "Epoch: 220, Batch: 500, Avg. Loss: 0.05838124305009842\n",
      "Epoch: 220, Batch: 600, Avg. Loss: 0.05673025622963905\n",
      "Epoch: 220, Batch: 700, Avg. Loss: 0.051813599318265915\n",
      "Epoch: 220, Batch: 800, Avg. Loss: 0.05501490764319897\n",
      "Epoch: 220, Batch: 900, Avg. Loss: 0.043794428706169126\n",
      "Epoch: 221, Batch: 100, Avg. Loss: 0.05245931766927242\n",
      "Epoch: 221, Batch: 200, Avg. Loss: 0.054466294050216676\n",
      "Epoch: 221, Batch: 300, Avg. Loss: 0.04906670212745667\n",
      "Epoch: 221, Batch: 400, Avg. Loss: 0.056340844109654425\n",
      "Epoch: 221, Batch: 500, Avg. Loss: 0.0524162295460701\n",
      "Epoch: 221, Batch: 600, Avg. Loss: 0.05028139151632786\n",
      "Epoch: 221, Batch: 700, Avg. Loss: 0.047247878313064574\n",
      "Epoch: 221, Batch: 800, Avg. Loss: 0.0501462446898222\n",
      "Epoch: 221, Batch: 900, Avg. Loss: 0.05477357588708401\n",
      "Epoch: 222, Batch: 100, Avg. Loss: 0.056534376367926596\n",
      "Epoch: 222, Batch: 200, Avg. Loss: 0.05218781337141991\n",
      "Epoch: 222, Batch: 300, Avg. Loss: 0.06970811940729618\n",
      "Epoch: 222, Batch: 400, Avg. Loss: 0.04027258701622486\n",
      "Epoch: 222, Batch: 500, Avg. Loss: 0.0494480923563242\n",
      "Epoch: 222, Batch: 600, Avg. Loss: 0.04344948515295982\n",
      "Epoch: 222, Batch: 700, Avg. Loss: 0.054495868682861326\n",
      "Epoch: 222, Batch: 800, Avg. Loss: 0.05029702618718147\n",
      "Epoch: 222, Batch: 900, Avg. Loss: 0.058039479181170466\n",
      "Epoch: 223, Batch: 100, Avg. Loss: 0.050809982717037204\n",
      "Epoch: 223, Batch: 200, Avg. Loss: 0.04927099846303463\n",
      "Epoch: 223, Batch: 300, Avg. Loss: 0.0470235139131546\n",
      "Epoch: 223, Batch: 400, Avg. Loss: 0.04810985743999481\n",
      "Epoch: 223, Batch: 500, Avg. Loss: 0.049367035031318666\n",
      "Epoch: 223, Batch: 600, Avg. Loss: 0.04927688777446747\n",
      "Epoch: 223, Batch: 700, Avg. Loss: 0.05414985992014408\n",
      "Epoch: 223, Batch: 800, Avg. Loss: 0.05371998094022274\n",
      "Epoch: 223, Batch: 900, Avg. Loss: 0.060651163086295125\n",
      "Epoch: 224, Batch: 100, Avg. Loss: 0.044608349800109866\n",
      "Epoch: 224, Batch: 200, Avg. Loss: 0.04544955462217331\n",
      "Epoch: 224, Batch: 300, Avg. Loss: 0.05402463130652904\n",
      "Epoch: 224, Batch: 400, Avg. Loss: 0.051126285418868066\n",
      "Epoch: 224, Batch: 500, Avg. Loss: 0.047544770687818524\n",
      "Epoch: 224, Batch: 600, Avg. Loss: 0.051619653701782224\n",
      "Epoch: 224, Batch: 700, Avg. Loss: 0.06091789282858372\n",
      "Epoch: 224, Batch: 800, Avg. Loss: 0.06557517230510712\n",
      "Epoch: 224, Batch: 900, Avg. Loss: 0.05196041814982891\n",
      "Epoch: 225, Batch: 100, Avg. Loss: 0.05572853319346905\n",
      "Epoch: 225, Batch: 200, Avg. Loss: 0.048230762034654616\n",
      "Epoch: 225, Batch: 300, Avg. Loss: 0.0463445296138525\n",
      "Epoch: 225, Batch: 400, Avg. Loss: 0.05476322114467621\n",
      "Epoch: 225, Batch: 500, Avg. Loss: 0.04953358270227909\n",
      "Epoch: 225, Batch: 600, Avg. Loss: 0.056115598157048224\n",
      "Epoch: 225, Batch: 700, Avg. Loss: 0.06076075695455074\n",
      "Epoch: 225, Batch: 800, Avg. Loss: 0.05083938382565975\n",
      "Epoch: 225, Batch: 900, Avg. Loss: 0.04601959832012653\n",
      "Epoch: 226, Batch: 100, Avg. Loss: 0.048903341740369796\n",
      "Epoch: 226, Batch: 200, Avg. Loss: 0.048721941113471987\n",
      "Epoch: 226, Batch: 300, Avg. Loss: 0.057871493697166446\n",
      "Epoch: 226, Batch: 400, Avg. Loss: 0.05441945403814316\n",
      "Epoch: 226, Batch: 500, Avg. Loss: 0.04570303954184055\n",
      "Epoch: 226, Batch: 600, Avg. Loss: 0.04790837123990059\n",
      "Epoch: 226, Batch: 700, Avg. Loss: 0.052548966333270074\n",
      "Epoch: 226, Batch: 800, Avg. Loss: 0.048762233778834346\n",
      "Epoch: 226, Batch: 900, Avg. Loss: 0.061032163798809054\n",
      "Epoch: 227, Batch: 100, Avg. Loss: 0.054284576922655106\n",
      "Epoch: 227, Batch: 200, Avg. Loss: 0.048115913048386576\n",
      "Epoch: 227, Batch: 300, Avg. Loss: 0.06465912640094756\n",
      "Epoch: 227, Batch: 400, Avg. Loss: 0.04783674344420433\n",
      "Epoch: 227, Batch: 500, Avg. Loss: 0.05073860205709934\n",
      "Epoch: 227, Batch: 600, Avg. Loss: 0.04655774503946304\n",
      "Epoch: 227, Batch: 700, Avg. Loss: 0.05171797282993793\n",
      "Epoch: 227, Batch: 800, Avg. Loss: 0.05913052655756473\n",
      "Epoch: 227, Batch: 900, Avg. Loss: 0.04526607610285282\n",
      "Epoch: 228, Batch: 100, Avg. Loss: 0.05374700583517551\n",
      "Epoch: 228, Batch: 200, Avg. Loss: 0.05071936421096325\n",
      "Epoch: 228, Batch: 300, Avg. Loss: 0.049427941143512726\n",
      "Epoch: 228, Batch: 400, Avg. Loss: 0.06335881814360618\n",
      "Epoch: 228, Batch: 500, Avg. Loss: 0.04781078197062016\n",
      "Epoch: 228, Batch: 600, Avg. Loss: 0.0491649054735899\n",
      "Epoch: 228, Batch: 700, Avg. Loss: 0.050578170716762544\n",
      "Epoch: 228, Batch: 800, Avg. Loss: 0.05101240150630474\n",
      "Epoch: 228, Batch: 900, Avg. Loss: 0.04630708996206522\n",
      "Epoch: 229, Batch: 100, Avg. Loss: 0.041239162236452104\n",
      "Epoch: 229, Batch: 200, Avg. Loss: 0.05713268265128136\n",
      "Epoch: 229, Batch: 300, Avg. Loss: 0.052735960260033604\n",
      "Epoch: 229, Batch: 400, Avg. Loss: 0.04544205479323864\n",
      "Epoch: 229, Batch: 500, Avg. Loss: 0.05536122530698776\n",
      "Epoch: 229, Batch: 600, Avg. Loss: 0.04918322201818228\n",
      "Epoch: 229, Batch: 700, Avg. Loss: 0.05719940729439259\n",
      "Epoch: 229, Batch: 800, Avg. Loss: 0.049560737162828446\n",
      "Epoch: 229, Batch: 900, Avg. Loss: 0.05175289824604988\n",
      "Epoch: 230, Batch: 100, Avg. Loss: 0.03802556492388248\n",
      "Epoch: 230, Batch: 200, Avg. Loss: 0.04744250275194645\n",
      "Epoch: 230, Batch: 300, Avg. Loss: 0.057078655138611795\n",
      "Epoch: 230, Batch: 400, Avg. Loss: 0.05927410915493965\n",
      "Epoch: 230, Batch: 500, Avg. Loss: 0.04702320367097854\n",
      "Epoch: 230, Batch: 600, Avg. Loss: 0.050489641651511194\n",
      "Epoch: 230, Batch: 700, Avg. Loss: 0.050548529401421544\n",
      "Epoch: 230, Batch: 800, Avg. Loss: 0.04668942950665951\n",
      "Epoch: 230, Batch: 900, Avg. Loss: 0.054086062982678415\n",
      "Epoch: 231, Batch: 100, Avg. Loss: 0.047442482709884645\n",
      "Epoch: 231, Batch: 200, Avg. Loss: 0.0427079451084137\n",
      "Epoch: 231, Batch: 300, Avg. Loss: 0.05039488080888987\n",
      "Epoch: 231, Batch: 400, Avg. Loss: 0.045201587602496145\n",
      "Epoch: 231, Batch: 500, Avg. Loss: 0.05421608962118626\n",
      "Epoch: 231, Batch: 600, Avg. Loss: 0.054013496339321135\n",
      "Epoch: 231, Batch: 700, Avg. Loss: 0.053016590476036074\n",
      "Epoch: 231, Batch: 800, Avg. Loss: 0.0443938035517931\n",
      "Epoch: 231, Batch: 900, Avg. Loss: 0.05266621127724647\n",
      "Epoch: 232, Batch: 100, Avg. Loss: 0.04579488508403301\n",
      "Epoch: 232, Batch: 200, Avg. Loss: 0.04481708720326424\n",
      "Epoch: 232, Batch: 300, Avg. Loss: 0.04593021847307682\n",
      "Epoch: 232, Batch: 400, Avg. Loss: 0.05255649961531162\n",
      "Epoch: 232, Batch: 500, Avg. Loss: 0.06586623042821885\n",
      "Epoch: 232, Batch: 600, Avg. Loss: 0.05651526071131229\n",
      "Epoch: 232, Batch: 700, Avg. Loss: 0.05305895484983921\n",
      "Epoch: 232, Batch: 800, Avg. Loss: 0.050389984622597694\n",
      "Epoch: 232, Batch: 900, Avg. Loss: 0.0504464266449213\n",
      "Epoch: 233, Batch: 100, Avg. Loss: 0.05426154270768166\n",
      "Epoch: 233, Batch: 200, Avg. Loss: 0.045625541210174564\n",
      "Epoch: 233, Batch: 300, Avg. Loss: 0.05142054475843907\n",
      "Epoch: 233, Batch: 400, Avg. Loss: 0.05028976500034332\n",
      "Epoch: 233, Batch: 500, Avg. Loss: 0.0511640477925539\n",
      "Epoch: 233, Batch: 600, Avg. Loss: 0.05083303771913052\n",
      "Epoch: 233, Batch: 700, Avg. Loss: 0.04627449974417686\n",
      "Epoch: 233, Batch: 800, Avg. Loss: 0.05415091924369335\n",
      "Epoch: 233, Batch: 900, Avg. Loss: 0.042027156054973605\n",
      "Epoch: 234, Batch: 100, Avg. Loss: 0.050345659628510474\n",
      "Epoch: 234, Batch: 200, Avg. Loss: 0.04504109390079975\n",
      "Epoch: 234, Batch: 300, Avg. Loss: 0.048561601862311364\n",
      "Epoch: 234, Batch: 400, Avg. Loss: 0.051903503760695455\n",
      "Epoch: 234, Batch: 500, Avg. Loss: 0.05722019508481026\n",
      "Epoch: 234, Batch: 600, Avg. Loss: 0.04894548527896404\n",
      "Epoch: 234, Batch: 700, Avg. Loss: 0.05113194637000561\n",
      "Epoch: 234, Batch: 800, Avg. Loss: 0.04531568750739098\n",
      "Epoch: 234, Batch: 900, Avg. Loss: 0.04930929429829121\n",
      "Epoch: 235, Batch: 100, Avg. Loss: 0.04709586106240749\n",
      "Epoch: 235, Batch: 200, Avg. Loss: 0.045772060081362725\n",
      "Epoch: 235, Batch: 300, Avg. Loss: 0.04707988873124123\n",
      "Epoch: 235, Batch: 400, Avg. Loss: 0.05470398485660553\n",
      "Epoch: 235, Batch: 500, Avg. Loss: 0.04653886742889881\n",
      "Epoch: 235, Batch: 600, Avg. Loss: 0.04782206609845161\n",
      "Epoch: 235, Batch: 700, Avg. Loss: 0.04461221106350422\n",
      "Epoch: 235, Batch: 800, Avg. Loss: 0.04608320824801922\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 235, Batch: 900, Avg. Loss: 0.05115117453038692\n",
      "Epoch: 236, Batch: 100, Avg. Loss: 0.04421246960759163\n",
      "Epoch: 236, Batch: 200, Avg. Loss: 0.04902679681777954\n",
      "Epoch: 236, Batch: 300, Avg. Loss: 0.04780309818685055\n",
      "Epoch: 236, Batch: 400, Avg. Loss: 0.051520792469382286\n",
      "Epoch: 236, Batch: 500, Avg. Loss: 0.04614505372941494\n",
      "Epoch: 236, Batch: 600, Avg. Loss: 0.0577887286990881\n",
      "Epoch: 236, Batch: 700, Avg. Loss: 0.05465751729905605\n",
      "Epoch: 236, Batch: 800, Avg. Loss: 0.04639388412237167\n",
      "Epoch: 236, Batch: 900, Avg. Loss: 0.051935472786426545\n",
      "Epoch: 237, Batch: 100, Avg. Loss: 0.05244341388344765\n",
      "Epoch: 237, Batch: 200, Avg. Loss: 0.05443190678954125\n",
      "Epoch: 237, Batch: 300, Avg. Loss: 0.05785876542329788\n",
      "Epoch: 237, Batch: 400, Avg. Loss: 0.0514554363489151\n",
      "Epoch: 237, Batch: 500, Avg. Loss: 0.046065316796302797\n",
      "Epoch: 237, Batch: 600, Avg. Loss: 0.051220756992697715\n",
      "Epoch: 237, Batch: 700, Avg. Loss: 0.04941694878041744\n",
      "Epoch: 237, Batch: 800, Avg. Loss: 0.04686622507870197\n",
      "Epoch: 237, Batch: 900, Avg. Loss: 0.04533179618418217\n",
      "Epoch: 238, Batch: 100, Avg. Loss: 0.0463244204223156\n",
      "Epoch: 238, Batch: 200, Avg. Loss: 0.04261682346463203\n",
      "Epoch: 238, Batch: 300, Avg. Loss: 0.04408312574028969\n",
      "Epoch: 238, Batch: 400, Avg. Loss: 0.05239842422306538\n",
      "Epoch: 238, Batch: 500, Avg. Loss: 0.055199777111411094\n",
      "Epoch: 238, Batch: 600, Avg. Loss: 0.05840069241821766\n",
      "Epoch: 238, Batch: 700, Avg. Loss: 0.05259533546864986\n",
      "Epoch: 238, Batch: 800, Avg. Loss: 0.049626110047101976\n",
      "Epoch: 238, Batch: 900, Avg. Loss: 0.05365302100777626\n",
      "Epoch: 239, Batch: 100, Avg. Loss: 0.04648628726601601\n",
      "Epoch: 239, Batch: 200, Avg. Loss: 0.043179882243275645\n",
      "Epoch: 239, Batch: 300, Avg. Loss: 0.0535713654011488\n",
      "Epoch: 239, Batch: 400, Avg. Loss: 0.05044218309223652\n",
      "Epoch: 239, Batch: 500, Avg. Loss: 0.04568504735827446\n",
      "Epoch: 239, Batch: 600, Avg. Loss: 0.0567783622443676\n",
      "Epoch: 239, Batch: 700, Avg. Loss: 0.05251612551510334\n",
      "Epoch: 239, Batch: 800, Avg. Loss: 0.0503297733515501\n",
      "Epoch: 239, Batch: 900, Avg. Loss: 0.05201228611171246\n",
      "Epoch: 240, Batch: 100, Avg. Loss: 0.05819086566567421\n",
      "Epoch: 240, Batch: 200, Avg. Loss: 0.046825592666864396\n",
      "Epoch: 240, Batch: 300, Avg. Loss: 0.045654291138052944\n",
      "Epoch: 240, Batch: 400, Avg. Loss: 0.04835158243775368\n",
      "Epoch: 240, Batch: 500, Avg. Loss: 0.04665295727550983\n",
      "Epoch: 240, Batch: 600, Avg. Loss: 0.05057941049337387\n",
      "Epoch: 240, Batch: 700, Avg. Loss: 0.047036968916654584\n",
      "Epoch: 240, Batch: 800, Avg. Loss: 0.055906731113791464\n",
      "Epoch: 240, Batch: 900, Avg. Loss: 0.04993442192673683\n",
      "Epoch: 241, Batch: 100, Avg. Loss: 0.04097192734479904\n",
      "Epoch: 241, Batch: 200, Avg. Loss: 0.05218494474887848\n",
      "Epoch: 241, Batch: 300, Avg. Loss: 0.04834993571043014\n",
      "Epoch: 241, Batch: 400, Avg. Loss: 0.052239067107439044\n",
      "Epoch: 241, Batch: 500, Avg. Loss: 0.047588525414466856\n",
      "Epoch: 241, Batch: 600, Avg. Loss: 0.05455503404140472\n",
      "Epoch: 241, Batch: 700, Avg. Loss: 0.03993617668747902\n",
      "Epoch: 241, Batch: 800, Avg. Loss: 0.052131162360310554\n",
      "Epoch: 241, Batch: 900, Avg. Loss: 0.06013454668223858\n",
      "Epoch: 242, Batch: 100, Avg. Loss: 0.050952497869729996\n",
      "Epoch: 242, Batch: 200, Avg. Loss: 0.05318909890949726\n",
      "Epoch: 242, Batch: 300, Avg. Loss: 0.04844103269279003\n",
      "Epoch: 242, Batch: 400, Avg. Loss: 0.041408345326781275\n",
      "Epoch: 242, Batch: 500, Avg. Loss: 0.05458058558404446\n",
      "Epoch: 242, Batch: 600, Avg. Loss: 0.04115030869841576\n",
      "Epoch: 242, Batch: 700, Avg. Loss: 0.048148105964064596\n",
      "Epoch: 242, Batch: 800, Avg. Loss: 0.04732693932950496\n",
      "Epoch: 242, Batch: 900, Avg. Loss: 0.05635447330772877\n",
      "Epoch: 243, Batch: 100, Avg. Loss: 0.054950230047106745\n",
      "Epoch: 243, Batch: 200, Avg. Loss: 0.05082550697028637\n",
      "Epoch: 243, Batch: 300, Avg. Loss: 0.048722112849354744\n",
      "Epoch: 243, Batch: 400, Avg. Loss: 0.053607138022780415\n",
      "Epoch: 243, Batch: 500, Avg. Loss: 0.051298524886369705\n",
      "Epoch: 243, Batch: 600, Avg. Loss: 0.045635016933083535\n",
      "Epoch: 243, Batch: 700, Avg. Loss: 0.039510421901941296\n",
      "Epoch: 243, Batch: 800, Avg. Loss: 0.042342609763145446\n",
      "Epoch: 243, Batch: 900, Avg. Loss: 0.054250500425696374\n",
      "Epoch: 244, Batch: 100, Avg. Loss: 0.04762507535517216\n",
      "Epoch: 244, Batch: 200, Avg. Loss: 0.04745911970734596\n",
      "Epoch: 244, Batch: 300, Avg. Loss: 0.041724153906106946\n",
      "Epoch: 244, Batch: 400, Avg. Loss: 0.0481563251465559\n",
      "Epoch: 244, Batch: 500, Avg. Loss: 0.05487887494266033\n",
      "Epoch: 244, Batch: 600, Avg. Loss: 0.05192769676446915\n",
      "Epoch: 244, Batch: 700, Avg. Loss: 0.04653795354068279\n",
      "Epoch: 244, Batch: 800, Avg. Loss: 0.05189185410737991\n",
      "Epoch: 244, Batch: 900, Avg. Loss: 0.04807799130678177\n",
      "Epoch: 245, Batch: 100, Avg. Loss: 0.04460938856005669\n",
      "Epoch: 245, Batch: 200, Avg. Loss: 0.04709860973060131\n",
      "Epoch: 245, Batch: 300, Avg. Loss: 0.04755992144346237\n",
      "Epoch: 245, Batch: 400, Avg. Loss: 0.0496480667963624\n",
      "Epoch: 245, Batch: 500, Avg. Loss: 0.045043719783425334\n",
      "Epoch: 245, Batch: 600, Avg. Loss: 0.05044492222368717\n",
      "Epoch: 245, Batch: 700, Avg. Loss: 0.05298146314918995\n",
      "Epoch: 245, Batch: 800, Avg. Loss: 0.04694543279707432\n",
      "Epoch: 245, Batch: 900, Avg. Loss: 0.05397259987890721\n",
      "Epoch: 246, Batch: 100, Avg. Loss: 0.04942922793328762\n",
      "Epoch: 246, Batch: 200, Avg. Loss: 0.039722595438361166\n",
      "Epoch: 246, Batch: 300, Avg. Loss: 0.05253880769014359\n",
      "Epoch: 246, Batch: 400, Avg. Loss: 0.0501689650863409\n",
      "Epoch: 246, Batch: 500, Avg. Loss: 0.0502240826934576\n",
      "Epoch: 246, Batch: 600, Avg. Loss: 0.04389041960239411\n",
      "Epoch: 246, Batch: 700, Avg. Loss: 0.047849244475364684\n",
      "Epoch: 246, Batch: 800, Avg. Loss: 0.05270266801118851\n",
      "Epoch: 246, Batch: 900, Avg. Loss: 0.05314689286053181\n",
      "Epoch: 247, Batch: 100, Avg. Loss: 0.04782201647758484\n",
      "Epoch: 247, Batch: 200, Avg. Loss: 0.04763925939798355\n",
      "Epoch: 247, Batch: 300, Avg. Loss: 0.053140803799033166\n",
      "Epoch: 247, Batch: 400, Avg. Loss: 0.0477761360257864\n",
      "Epoch: 247, Batch: 500, Avg. Loss: 0.049832829684019087\n",
      "Epoch: 247, Batch: 600, Avg. Loss: 0.046019702404737475\n",
      "Epoch: 247, Batch: 700, Avg. Loss: 0.04654379062354565\n",
      "Epoch: 247, Batch: 800, Avg. Loss: 0.04979193091392517\n",
      "Epoch: 247, Batch: 900, Avg. Loss: 0.04504373595118523\n",
      "Epoch: 248, Batch: 100, Avg. Loss: 0.05239190578460693\n",
      "Epoch: 248, Batch: 200, Avg. Loss: 0.05690284185111523\n",
      "Epoch: 248, Batch: 300, Avg. Loss: 0.05134371004998684\n",
      "Epoch: 248, Batch: 400, Avg. Loss: 0.0469762771576643\n",
      "Epoch: 248, Batch: 500, Avg. Loss: 0.046497987061738966\n",
      "Epoch: 248, Batch: 600, Avg. Loss: 0.044089944809675215\n",
      "Epoch: 248, Batch: 700, Avg. Loss: 0.049141141772270205\n",
      "Epoch: 248, Batch: 800, Avg. Loss: 0.05197597295045853\n",
      "Epoch: 248, Batch: 900, Avg. Loss: 0.04219215415418148\n",
      "Epoch: 249, Batch: 100, Avg. Loss: 0.051326942071318625\n",
      "Epoch: 249, Batch: 200, Avg. Loss: 0.049995971471071245\n",
      "Epoch: 249, Batch: 300, Avg. Loss: 0.05570248708128929\n",
      "Epoch: 249, Batch: 400, Avg. Loss: 0.047055703774094584\n",
      "Epoch: 249, Batch: 500, Avg. Loss: 0.04613965891301632\n",
      "Epoch: 249, Batch: 600, Avg. Loss: 0.04901173375546932\n",
      "Epoch: 249, Batch: 700, Avg. Loss: 0.04739262118935585\n",
      "Epoch: 249, Batch: 800, Avg. Loss: 0.045143096819519994\n",
      "Epoch: 249, Batch: 900, Avg. Loss: 0.04644420735538006\n",
      "Epoch: 250, Batch: 100, Avg. Loss: 0.04591670274734497\n",
      "Epoch: 250, Batch: 200, Avg. Loss: 0.04544826626777649\n",
      "Epoch: 250, Batch: 300, Avg. Loss: 0.05257702797651291\n",
      "Epoch: 250, Batch: 400, Avg. Loss: 0.04682226553559303\n",
      "Epoch: 250, Batch: 500, Avg. Loss: 0.04780074708163738\n",
      "Epoch: 250, Batch: 600, Avg. Loss: 0.04832334481179714\n",
      "Epoch: 250, Batch: 700, Avg. Loss: 0.05310218840837479\n",
      "Epoch: 250, Batch: 800, Avg. Loss: 0.04612400360405445\n",
      "Epoch: 250, Batch: 900, Avg. Loss: 0.045434322729706766\n",
      "Epoch: 251, Batch: 100, Avg. Loss: 0.04482062041759491\n",
      "Epoch: 251, Batch: 200, Avg. Loss: 0.04777005344629288\n",
      "Epoch: 251, Batch: 300, Avg. Loss: 0.045277881398797036\n",
      "Epoch: 251, Batch: 400, Avg. Loss: 0.043993990942835806\n",
      "Epoch: 251, Batch: 500, Avg. Loss: 0.05059345528483391\n",
      "Epoch: 251, Batch: 600, Avg. Loss: 0.04980190061032772\n",
      "Epoch: 251, Batch: 700, Avg. Loss: 0.04592783913016319\n",
      "Epoch: 251, Batch: 800, Avg. Loss: 0.05663726381957531\n",
      "Epoch: 251, Batch: 900, Avg. Loss: 0.03990691915154457\n",
      "Epoch: 252, Batch: 100, Avg. Loss: 0.047002692967653276\n",
      "Epoch: 252, Batch: 200, Avg. Loss: 0.04604366131126881\n",
      "Epoch: 252, Batch: 300, Avg. Loss: 0.04939925119280815\n",
      "Epoch: 252, Batch: 400, Avg. Loss: 0.04346365489065647\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 252, Batch: 500, Avg. Loss: 0.04852987743914127\n",
      "Epoch: 252, Batch: 600, Avg. Loss: 0.05790821634232998\n",
      "Epoch: 252, Batch: 700, Avg. Loss: 0.04894481845200062\n",
      "Epoch: 252, Batch: 800, Avg. Loss: 0.04925081692636013\n",
      "Epoch: 252, Batch: 900, Avg. Loss: 0.04745954267680645\n",
      "Epoch: 253, Batch: 100, Avg. Loss: 0.04764762908220291\n",
      "Epoch: 253, Batch: 200, Avg. Loss: 0.05341980680823326\n",
      "Epoch: 253, Batch: 300, Avg. Loss: 0.05056893415749073\n",
      "Epoch: 253, Batch: 400, Avg. Loss: 0.04886864267289639\n",
      "Epoch: 253, Batch: 500, Avg. Loss: 0.05392579190433025\n",
      "Epoch: 253, Batch: 600, Avg. Loss: 0.039083441495895384\n",
      "Epoch: 253, Batch: 700, Avg. Loss: 0.046789835169911385\n",
      "Epoch: 253, Batch: 800, Avg. Loss: 0.04334992378950119\n",
      "Epoch: 253, Batch: 900, Avg. Loss: 0.043959902822971346\n",
      "Epoch: 254, Batch: 100, Avg. Loss: 0.050108747482299806\n",
      "Epoch: 254, Batch: 200, Avg. Loss: 0.04844709411263466\n",
      "Epoch: 254, Batch: 300, Avg. Loss: 0.04739964351058006\n",
      "Epoch: 254, Batch: 400, Avg. Loss: 0.054097937792539595\n",
      "Epoch: 254, Batch: 500, Avg. Loss: 0.044662837386131284\n",
      "Epoch: 254, Batch: 600, Avg. Loss: 0.0495687984675169\n",
      "Epoch: 254, Batch: 700, Avg. Loss: 0.05065563194453716\n",
      "Epoch: 254, Batch: 800, Avg. Loss: 0.047161817476153375\n",
      "Epoch: 254, Batch: 900, Avg. Loss: 0.04787991151213646\n",
      "Epoch: 255, Batch: 100, Avg. Loss: 0.043168969228863714\n",
      "Epoch: 255, Batch: 200, Avg. Loss: 0.057635296806693076\n",
      "Epoch: 255, Batch: 300, Avg. Loss: 0.04579835526645184\n",
      "Epoch: 255, Batch: 400, Avg. Loss: 0.04446308508515358\n",
      "Epoch: 255, Batch: 500, Avg. Loss: 0.04232879914343357\n",
      "Epoch: 255, Batch: 600, Avg. Loss: 0.04581661708652973\n",
      "Epoch: 255, Batch: 700, Avg. Loss: 0.05535494536161423\n",
      "Epoch: 255, Batch: 800, Avg. Loss: 0.04623794272542\n",
      "Epoch: 255, Batch: 900, Avg. Loss: 0.04371262438595295\n",
      "Epoch: 256, Batch: 100, Avg. Loss: 0.04320333644747734\n",
      "Epoch: 256, Batch: 200, Avg. Loss: 0.05371433779597282\n",
      "Epoch: 256, Batch: 300, Avg. Loss: 0.04632247649133205\n",
      "Epoch: 256, Batch: 400, Avg. Loss: 0.05425165198743343\n",
      "Epoch: 256, Batch: 500, Avg. Loss: 0.05462874539196491\n",
      "Epoch: 256, Batch: 600, Avg. Loss: 0.049371080249547956\n",
      "Epoch: 256, Batch: 700, Avg. Loss: 0.04090628154575825\n",
      "Epoch: 256, Batch: 800, Avg. Loss: 0.037912814915180205\n",
      "Epoch: 256, Batch: 900, Avg. Loss: 0.0476208571344614\n",
      "Epoch: 257, Batch: 100, Avg. Loss: 0.052041737586259844\n",
      "Epoch: 257, Batch: 200, Avg. Loss: 0.05774164661765099\n",
      "Epoch: 257, Batch: 300, Avg. Loss: 0.04335217751562595\n",
      "Epoch: 257, Batch: 400, Avg. Loss: 0.04404546946287155\n",
      "Epoch: 257, Batch: 500, Avg. Loss: 0.04348199643194675\n",
      "Epoch: 257, Batch: 600, Avg. Loss: 0.04976449064910412\n",
      "Epoch: 257, Batch: 700, Avg. Loss: 0.048784384354949\n",
      "Epoch: 257, Batch: 800, Avg. Loss: 0.04860784001648426\n",
      "Epoch: 257, Batch: 900, Avg. Loss: 0.045895499512553214\n",
      "Epoch: 258, Batch: 100, Avg. Loss: 0.04442562147974968\n",
      "Epoch: 258, Batch: 200, Avg. Loss: 0.042715899422764776\n",
      "Epoch: 258, Batch: 300, Avg. Loss: 0.047993482872843746\n",
      "Epoch: 258, Batch: 400, Avg. Loss: 0.05003799498081207\n",
      "Epoch: 258, Batch: 500, Avg. Loss: 0.047785751819610596\n",
      "Epoch: 258, Batch: 600, Avg. Loss: 0.03878006562590599\n",
      "Epoch: 258, Batch: 700, Avg. Loss: 0.04486025951802731\n",
      "Epoch: 258, Batch: 800, Avg. Loss: 0.05085277117788792\n",
      "Epoch: 258, Batch: 900, Avg. Loss: 0.048393372371792795\n",
      "Epoch: 259, Batch: 100, Avg. Loss: 0.04526707746088505\n",
      "Epoch: 259, Batch: 200, Avg. Loss: 0.043804613128304484\n",
      "Epoch: 259, Batch: 300, Avg. Loss: 0.044914938360452655\n",
      "Epoch: 259, Batch: 400, Avg. Loss: 0.05368333756923675\n",
      "Epoch: 259, Batch: 500, Avg. Loss: 0.05134222209453583\n",
      "Epoch: 259, Batch: 600, Avg. Loss: 0.047536955773830415\n",
      "Epoch: 259, Batch: 700, Avg. Loss: 0.04826714165508747\n",
      "Epoch: 259, Batch: 800, Avg. Loss: 0.04203288927674294\n",
      "Epoch: 259, Batch: 900, Avg. Loss: 0.04302659846842289\n",
      "Epoch: 260, Batch: 100, Avg. Loss: 0.04220355957746506\n",
      "Epoch: 260, Batch: 200, Avg. Loss: 0.04303230956196785\n",
      "Epoch: 260, Batch: 300, Avg. Loss: 0.03993506647646427\n",
      "Epoch: 260, Batch: 400, Avg. Loss: 0.04199549525976181\n",
      "Epoch: 260, Batch: 500, Avg. Loss: 0.046814990267157554\n",
      "Epoch: 260, Batch: 600, Avg. Loss: 0.04282768614590168\n",
      "Epoch: 260, Batch: 700, Avg. Loss: 0.04413449674844742\n",
      "Epoch: 260, Batch: 800, Avg. Loss: 0.05060347393155098\n",
      "Epoch: 260, Batch: 900, Avg. Loss: 0.057740568444132806\n",
      "Epoch: 261, Batch: 100, Avg. Loss: 0.051036784425377846\n",
      "Epoch: 261, Batch: 200, Avg. Loss: 0.04625266633927822\n",
      "Epoch: 261, Batch: 300, Avg. Loss: 0.04607775881886482\n",
      "Epoch: 261, Batch: 400, Avg. Loss: 0.045650970861315726\n",
      "Epoch: 261, Batch: 500, Avg. Loss: 0.04573902159929275\n",
      "Epoch: 261, Batch: 600, Avg. Loss: 0.05412843756377697\n",
      "Epoch: 261, Batch: 700, Avg. Loss: 0.048150670379400254\n",
      "Epoch: 261, Batch: 800, Avg. Loss: 0.04907560184597969\n",
      "Epoch: 261, Batch: 900, Avg. Loss: 0.0554189620167017\n",
      "Epoch: 262, Batch: 100, Avg. Loss: 0.045495222136378285\n",
      "Epoch: 262, Batch: 200, Avg. Loss: 0.053247243911027906\n",
      "Epoch: 262, Batch: 300, Avg. Loss: 0.04050493434071541\n",
      "Epoch: 262, Batch: 400, Avg. Loss: 0.04158739753067493\n",
      "Epoch: 262, Batch: 500, Avg. Loss: 0.0526065631210804\n",
      "Epoch: 262, Batch: 600, Avg. Loss: 0.050211247354745865\n",
      "Epoch: 262, Batch: 700, Avg. Loss: 0.04554531000554562\n",
      "Epoch: 262, Batch: 800, Avg. Loss: 0.04624024540185928\n",
      "Epoch: 262, Batch: 900, Avg. Loss: 0.05449050232768059\n",
      "Epoch: 263, Batch: 100, Avg. Loss: 0.04265257380902767\n",
      "Epoch: 263, Batch: 200, Avg. Loss: 0.04625471755862236\n",
      "Epoch: 263, Batch: 300, Avg. Loss: 0.042429342344403265\n",
      "Epoch: 263, Batch: 400, Avg. Loss: 0.04621566370129585\n",
      "Epoch: 263, Batch: 500, Avg. Loss: 0.04204360596835613\n",
      "Epoch: 263, Batch: 600, Avg. Loss: 0.04259191624820233\n",
      "Epoch: 263, Batch: 700, Avg. Loss: 0.052516688406467435\n",
      "Epoch: 263, Batch: 800, Avg. Loss: 0.043886187672615054\n",
      "Epoch: 263, Batch: 900, Avg. Loss: 0.05023494370281696\n",
      "Epoch: 264, Batch: 100, Avg. Loss: 0.04843604505062103\n",
      "Epoch: 264, Batch: 200, Avg. Loss: 0.051800438389182094\n",
      "Epoch: 264, Batch: 300, Avg. Loss: 0.04475025333464146\n",
      "Epoch: 264, Batch: 400, Avg. Loss: 0.04411455675959587\n",
      "Epoch: 264, Batch: 500, Avg. Loss: 0.04887237176299095\n",
      "Epoch: 264, Batch: 600, Avg. Loss: 0.04977364405989647\n",
      "Epoch: 264, Batch: 700, Avg. Loss: 0.04584877330809831\n",
      "Epoch: 264, Batch: 800, Avg. Loss: 0.0484994412958622\n",
      "Epoch: 264, Batch: 900, Avg. Loss: 0.04190119236707687\n",
      "Epoch: 265, Batch: 100, Avg. Loss: 0.047969036623835566\n",
      "Epoch: 265, Batch: 200, Avg. Loss: 0.04959087781608105\n",
      "Epoch: 265, Batch: 300, Avg. Loss: 0.04101281054317951\n",
      "Epoch: 265, Batch: 400, Avg. Loss: 0.044069283977150916\n",
      "Epoch: 265, Batch: 500, Avg. Loss: 0.049082129076123235\n",
      "Epoch: 265, Batch: 600, Avg. Loss: 0.05188099287450314\n",
      "Epoch: 265, Batch: 700, Avg. Loss: 0.04780925378203392\n",
      "Epoch: 265, Batch: 800, Avg. Loss: 0.04526256218552589\n",
      "Epoch: 265, Batch: 900, Avg. Loss: 0.04436219200491905\n",
      "Epoch: 266, Batch: 100, Avg. Loss: 0.0490559159964323\n",
      "Epoch: 266, Batch: 200, Avg. Loss: 0.047320780232548716\n",
      "Epoch: 266, Batch: 300, Avg. Loss: 0.044531077817082404\n",
      "Epoch: 266, Batch: 400, Avg. Loss: 0.0454937668889761\n",
      "Epoch: 266, Batch: 500, Avg. Loss: 0.04789859876036644\n",
      "Epoch: 266, Batch: 600, Avg. Loss: 0.04184414453804493\n",
      "Epoch: 266, Batch: 700, Avg. Loss: 0.03945218771696091\n",
      "Epoch: 266, Batch: 800, Avg. Loss: 0.04474570438265801\n",
      "Epoch: 266, Batch: 900, Avg. Loss: 0.05738242633640766\n",
      "Epoch: 267, Batch: 100, Avg. Loss: 0.04788981191813946\n",
      "Epoch: 267, Batch: 200, Avg. Loss: 0.05766625702381134\n",
      "Epoch: 267, Batch: 300, Avg. Loss: 0.05145384632050991\n",
      "Epoch: 267, Batch: 400, Avg. Loss: 0.04368338689208031\n",
      "Epoch: 267, Batch: 500, Avg. Loss: 0.046226813644170764\n",
      "Epoch: 267, Batch: 600, Avg. Loss: 0.04596978209912777\n",
      "Epoch: 267, Batch: 700, Avg. Loss: 0.044388677552342416\n",
      "Epoch: 267, Batch: 800, Avg. Loss: 0.04791665211319923\n",
      "Epoch: 267, Batch: 900, Avg. Loss: 0.04543134704232216\n",
      "Epoch: 268, Batch: 100, Avg. Loss: 0.04204820603132248\n",
      "Epoch: 268, Batch: 200, Avg. Loss: 0.045795729756355284\n",
      "Epoch: 268, Batch: 300, Avg. Loss: 0.053535428568720816\n",
      "Epoch: 268, Batch: 400, Avg. Loss: 0.04517521910369396\n",
      "Epoch: 268, Batch: 500, Avg. Loss: 0.04529741078615188\n",
      "Epoch: 268, Batch: 600, Avg. Loss: 0.04901506863534451\n",
      "Epoch: 268, Batch: 700, Avg. Loss: 0.03995371110737324\n",
      "Epoch: 268, Batch: 800, Avg. Loss: 0.04608338594436646\n",
      "Epoch: 268, Batch: 900, Avg. Loss: 0.04843086309731007\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 269, Batch: 100, Avg. Loss: 0.037358610332012175\n",
      "Epoch: 269, Batch: 200, Avg. Loss: 0.047261426225304604\n",
      "Epoch: 269, Batch: 300, Avg. Loss: 0.05243849165737629\n",
      "Epoch: 269, Batch: 400, Avg. Loss: 0.04432303793728352\n",
      "Epoch: 269, Batch: 500, Avg. Loss: 0.048581401258707045\n",
      "Epoch: 269, Batch: 600, Avg. Loss: 0.045930656865239144\n",
      "Epoch: 269, Batch: 700, Avg. Loss: 0.04493627406656742\n",
      "Epoch: 269, Batch: 800, Avg. Loss: 0.04049670472741127\n",
      "Epoch: 269, Batch: 900, Avg. Loss: 0.05361320286989212\n",
      "Epoch: 270, Batch: 100, Avg. Loss: 0.052213387414813045\n",
      "Epoch: 270, Batch: 200, Avg. Loss: 0.04239418245851993\n",
      "Epoch: 270, Batch: 300, Avg. Loss: 0.04405348278582096\n",
      "Epoch: 270, Batch: 400, Avg. Loss: 0.04168601796030998\n",
      "Epoch: 270, Batch: 500, Avg. Loss: 0.049927261620759965\n",
      "Epoch: 270, Batch: 600, Avg. Loss: 0.04059226714074612\n",
      "Epoch: 270, Batch: 700, Avg. Loss: 0.04619424737989902\n",
      "Epoch: 270, Batch: 800, Avg. Loss: 0.04818159855902195\n",
      "Epoch: 270, Batch: 900, Avg. Loss: 0.04583793357014656\n",
      "Epoch: 271, Batch: 100, Avg. Loss: 0.04269906602799892\n",
      "Epoch: 271, Batch: 200, Avg. Loss: 0.042622246891260145\n",
      "Epoch: 271, Batch: 300, Avg. Loss: 0.03948898211121559\n",
      "Epoch: 271, Batch: 400, Avg. Loss: 0.04947626825422049\n",
      "Epoch: 271, Batch: 500, Avg. Loss: 0.04531824253499508\n",
      "Epoch: 271, Batch: 600, Avg. Loss: 0.05341378092765808\n",
      "Epoch: 271, Batch: 700, Avg. Loss: 0.0460182011500001\n",
      "Epoch: 271, Batch: 800, Avg. Loss: 0.05128596596419811\n",
      "Epoch: 271, Batch: 900, Avg. Loss: 0.056499500423669816\n",
      "Epoch: 272, Batch: 100, Avg. Loss: 0.04703043766319752\n",
      "Epoch: 272, Batch: 200, Avg. Loss: 0.049555510804057123\n",
      "Epoch: 272, Batch: 300, Avg. Loss: 0.04849544756114483\n",
      "Epoch: 272, Batch: 400, Avg. Loss: 0.05169179819524288\n",
      "Epoch: 272, Batch: 500, Avg. Loss: 0.04146069273352623\n",
      "Epoch: 272, Batch: 600, Avg. Loss: 0.04580728657543659\n",
      "Epoch: 272, Batch: 700, Avg. Loss: 0.04390042014420033\n",
      "Epoch: 272, Batch: 800, Avg. Loss: 0.050771505534648896\n",
      "Epoch: 272, Batch: 900, Avg. Loss: 0.04650507591664791\n",
      "Epoch: 273, Batch: 100, Avg. Loss: 0.05066727831959725\n",
      "Epoch: 273, Batch: 200, Avg. Loss: 0.04443423129618168\n",
      "Epoch: 273, Batch: 300, Avg. Loss: 0.04807992190122604\n",
      "Epoch: 273, Batch: 400, Avg. Loss: 0.05256272904574871\n",
      "Epoch: 273, Batch: 500, Avg. Loss: 0.0442861545085907\n",
      "Epoch: 273, Batch: 600, Avg. Loss: 0.05103537529706955\n",
      "Epoch: 273, Batch: 700, Avg. Loss: 0.04691799283027649\n",
      "Epoch: 273, Batch: 800, Avg. Loss: 0.037455366626381874\n",
      "Epoch: 273, Batch: 900, Avg. Loss: 0.04145022362470627\n",
      "Epoch: 274, Batch: 100, Avg. Loss: 0.03924906514585018\n",
      "Epoch: 274, Batch: 200, Avg. Loss: 0.04797408476471901\n",
      "Epoch: 274, Batch: 300, Avg. Loss: 0.05313446529209614\n",
      "Epoch: 274, Batch: 400, Avg. Loss: 0.04631584212183952\n",
      "Epoch: 274, Batch: 500, Avg. Loss: 0.043685311004519466\n",
      "Epoch: 274, Batch: 600, Avg. Loss: 0.03928630836308002\n",
      "Epoch: 274, Batch: 700, Avg. Loss: 0.05288198493421078\n",
      "Epoch: 274, Batch: 800, Avg. Loss: 0.05315133526921272\n",
      "Epoch: 274, Batch: 900, Avg. Loss: 0.04907258100807667\n",
      "Epoch: 275, Batch: 100, Avg. Loss: 0.04726190812885761\n",
      "Epoch: 275, Batch: 200, Avg. Loss: 0.04642600148916245\n",
      "Epoch: 275, Batch: 300, Avg. Loss: 0.053423399701714516\n",
      "Epoch: 275, Batch: 400, Avg. Loss: 0.040661673098802566\n",
      "Epoch: 275, Batch: 500, Avg. Loss: 0.045822038054466246\n",
      "Epoch: 275, Batch: 600, Avg. Loss: 0.04883997201919556\n",
      "Epoch: 275, Batch: 700, Avg. Loss: 0.0509365589171648\n",
      "Epoch: 275, Batch: 800, Avg. Loss: 0.04026577107608318\n",
      "Epoch: 275, Batch: 900, Avg. Loss: 0.04137235812842846\n",
      "Epoch: 276, Batch: 100, Avg. Loss: 0.039742064401507377\n",
      "Epoch: 276, Batch: 200, Avg. Loss: 0.056305795535445215\n",
      "Epoch: 276, Batch: 300, Avg. Loss: 0.052170741111040114\n",
      "Epoch: 276, Batch: 400, Avg. Loss: 0.04509027607738972\n",
      "Epoch: 276, Batch: 500, Avg. Loss: 0.041538304314017296\n",
      "Epoch: 276, Batch: 600, Avg. Loss: 0.050035739168524745\n",
      "Epoch: 276, Batch: 700, Avg. Loss: 0.04170819878578186\n",
      "Epoch: 276, Batch: 800, Avg. Loss: 0.03937943495810032\n",
      "Epoch: 276, Batch: 900, Avg. Loss: 0.044090159609913825\n",
      "Epoch: 277, Batch: 100, Avg. Loss: 0.03942046739161015\n",
      "Epoch: 277, Batch: 200, Avg. Loss: 0.047240354791283606\n",
      "Epoch: 277, Batch: 300, Avg. Loss: 0.04194063723087311\n",
      "Epoch: 277, Batch: 400, Avg. Loss: 0.049502047225832936\n",
      "Epoch: 277, Batch: 500, Avg. Loss: 0.04823306836187839\n",
      "Epoch: 277, Batch: 600, Avg. Loss: 0.04574290938675404\n",
      "Epoch: 277, Batch: 700, Avg. Loss: 0.046323953196406364\n",
      "Epoch: 277, Batch: 800, Avg. Loss: 0.045991814360022545\n",
      "Epoch: 277, Batch: 900, Avg. Loss: 0.04538259617984295\n",
      "Epoch: 278, Batch: 100, Avg. Loss: 0.03851311929523945\n",
      "Epoch: 278, Batch: 200, Avg. Loss: 0.04917678013443947\n",
      "Epoch: 278, Batch: 300, Avg. Loss: 0.04500759094953537\n",
      "Epoch: 278, Batch: 400, Avg. Loss: 0.04603458233177662\n",
      "Epoch: 278, Batch: 500, Avg. Loss: 0.051081448942422863\n",
      "Epoch: 278, Batch: 600, Avg. Loss: 0.0471443922072649\n",
      "Epoch: 278, Batch: 700, Avg. Loss: 0.0488958965241909\n",
      "Epoch: 278, Batch: 800, Avg. Loss: 0.046452569141983986\n",
      "Epoch: 278, Batch: 900, Avg. Loss: 0.04617133393883705\n",
      "Epoch: 279, Batch: 100, Avg. Loss: 0.0479220425337553\n",
      "Epoch: 279, Batch: 200, Avg. Loss: 0.045272549390792845\n",
      "Epoch: 279, Batch: 300, Avg. Loss: 0.04666603945195675\n",
      "Epoch: 279, Batch: 400, Avg. Loss: 0.04387513771653175\n",
      "Epoch: 279, Batch: 500, Avg. Loss: 0.046577897891402244\n",
      "Epoch: 279, Batch: 600, Avg. Loss: 0.04127171456813812\n",
      "Epoch: 279, Batch: 700, Avg. Loss: 0.043581625819206236\n",
      "Epoch: 279, Batch: 800, Avg. Loss: 0.04843387693166733\n",
      "Epoch: 279, Batch: 900, Avg. Loss: 0.04153146155178547\n",
      "Epoch: 280, Batch: 100, Avg. Loss: 0.04241593927145004\n",
      "Epoch: 280, Batch: 200, Avg. Loss: 0.050691324546933175\n",
      "Epoch: 280, Batch: 300, Avg. Loss: 0.04371030531823635\n",
      "Epoch: 280, Batch: 400, Avg. Loss: 0.043015396744012835\n",
      "Epoch: 280, Batch: 500, Avg. Loss: 0.04893794514238834\n",
      "Epoch: 280, Batch: 600, Avg. Loss: 0.05054116941988468\n",
      "Epoch: 280, Batch: 700, Avg. Loss: 0.044042327105998993\n",
      "Epoch: 280, Batch: 800, Avg. Loss: 0.04164531081914902\n",
      "Epoch: 280, Batch: 900, Avg. Loss: 0.043470828980207446\n",
      "Epoch: 281, Batch: 100, Avg. Loss: 0.04782615691423416\n",
      "Epoch: 281, Batch: 200, Avg. Loss: 0.044432863891124726\n",
      "Epoch: 281, Batch: 300, Avg. Loss: 0.04433267004787922\n",
      "Epoch: 281, Batch: 400, Avg. Loss: 0.04579751074314117\n",
      "Epoch: 281, Batch: 500, Avg. Loss: 0.045202293023467066\n",
      "Epoch: 281, Batch: 600, Avg. Loss: 0.050281426161527636\n",
      "Epoch: 281, Batch: 700, Avg. Loss: 0.03962090261280537\n",
      "Epoch: 281, Batch: 800, Avg. Loss: 0.05378570944070816\n",
      "Epoch: 281, Batch: 900, Avg. Loss: 0.05124463863670826\n",
      "Epoch: 282, Batch: 100, Avg. Loss: 0.039115429297089575\n",
      "Epoch: 282, Batch: 200, Avg. Loss: 0.044832984283566475\n",
      "Epoch: 282, Batch: 300, Avg. Loss: 0.04292883366346359\n",
      "Epoch: 282, Batch: 400, Avg. Loss: 0.04316143274307251\n",
      "Epoch: 282, Batch: 500, Avg. Loss: 0.05411990717053414\n",
      "Epoch: 282, Batch: 600, Avg. Loss: 0.05280338741838932\n",
      "Epoch: 282, Batch: 700, Avg. Loss: 0.04188644878566265\n",
      "Epoch: 282, Batch: 800, Avg. Loss: 0.038202520459890366\n",
      "Epoch: 282, Batch: 900, Avg. Loss: 0.04123977690935135\n",
      "Epoch: 283, Batch: 100, Avg. Loss: 0.03631398685276508\n",
      "Epoch: 283, Batch: 200, Avg. Loss: 0.043521726205945015\n",
      "Epoch: 283, Batch: 300, Avg. Loss: 0.040628773272037504\n",
      "Epoch: 283, Batch: 400, Avg. Loss: 0.039534739404916766\n",
      "Epoch: 283, Batch: 500, Avg. Loss: 0.048426556885242465\n",
      "Epoch: 283, Batch: 600, Avg. Loss: 0.042198825255036355\n",
      "Epoch: 283, Batch: 700, Avg. Loss: 0.052304830104112625\n",
      "Epoch: 283, Batch: 800, Avg. Loss: 0.05003111980855465\n",
      "Epoch: 283, Batch: 900, Avg. Loss: 0.04119757853448391\n",
      "Epoch: 284, Batch: 100, Avg. Loss: 0.04438628625124693\n",
      "Epoch: 284, Batch: 200, Avg. Loss: 0.03774375483393669\n",
      "Epoch: 284, Batch: 300, Avg. Loss: 0.041601664200425145\n",
      "Epoch: 284, Batch: 400, Avg. Loss: 0.04442177802324295\n",
      "Epoch: 284, Batch: 500, Avg. Loss: 0.04351200170814991\n",
      "Epoch: 284, Batch: 600, Avg. Loss: 0.04708165116608143\n",
      "Epoch: 284, Batch: 700, Avg. Loss: 0.04078251875936985\n",
      "Epoch: 284, Batch: 800, Avg. Loss: 0.04701774977147579\n",
      "Epoch: 284, Batch: 900, Avg. Loss: 0.03836577154695988\n",
      "Epoch: 285, Batch: 100, Avg. Loss: 0.040348990485072136\n",
      "Epoch: 285, Batch: 200, Avg. Loss: 0.056086807996034625\n",
      "Epoch: 285, Batch: 300, Avg. Loss: 0.04121176265180111\n",
      "Epoch: 285, Batch: 400, Avg. Loss: 0.0425835321098566\n",
      "Epoch: 285, Batch: 500, Avg. Loss: 0.04245794713497162\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 285, Batch: 600, Avg. Loss: 0.05218890890479088\n",
      "Epoch: 285, Batch: 700, Avg. Loss: 0.04753051146864891\n",
      "Epoch: 285, Batch: 800, Avg. Loss: 0.043908670470118524\n",
      "Epoch: 285, Batch: 900, Avg. Loss: 0.03937255188822746\n",
      "Epoch: 286, Batch: 100, Avg. Loss: 0.04146633841097355\n",
      "Epoch: 286, Batch: 200, Avg. Loss: 0.03974572122097016\n",
      "Epoch: 286, Batch: 300, Avg. Loss: 0.04745774246752262\n",
      "Epoch: 286, Batch: 400, Avg. Loss: 0.04611464150249958\n",
      "Epoch: 286, Batch: 500, Avg. Loss: 0.03707671046257019\n",
      "Epoch: 286, Batch: 600, Avg. Loss: 0.041953701227903366\n",
      "Epoch: 286, Batch: 700, Avg. Loss: 0.04313722476363182\n",
      "Epoch: 286, Batch: 800, Avg. Loss: 0.0495529343932867\n",
      "Epoch: 286, Batch: 900, Avg. Loss: 0.04432898983359337\n",
      "Epoch: 287, Batch: 100, Avg. Loss: 0.04039498552680015\n",
      "Epoch: 287, Batch: 200, Avg. Loss: 0.04388798169791699\n",
      "Epoch: 287, Batch: 300, Avg. Loss: 0.04554081909358501\n",
      "Epoch: 287, Batch: 400, Avg. Loss: 0.04944391183555126\n",
      "Epoch: 287, Batch: 500, Avg. Loss: 0.04064493209123612\n",
      "Epoch: 287, Batch: 600, Avg. Loss: 0.04014606706798077\n",
      "Epoch: 287, Batch: 700, Avg. Loss: 0.050106611996889115\n",
      "Epoch: 287, Batch: 800, Avg. Loss: 0.04172508433461189\n",
      "Epoch: 287, Batch: 900, Avg. Loss: 0.050959686413407326\n",
      "Epoch: 288, Batch: 100, Avg. Loss: 0.04622654773294926\n",
      "Epoch: 288, Batch: 200, Avg. Loss: 0.043471596091985705\n",
      "Epoch: 288, Batch: 300, Avg. Loss: 0.04301909871399403\n",
      "Epoch: 288, Batch: 400, Avg. Loss: 0.04244098074734211\n",
      "Epoch: 288, Batch: 500, Avg. Loss: 0.042723610624670984\n",
      "Epoch: 288, Batch: 600, Avg. Loss: 0.04058642752468586\n",
      "Epoch: 288, Batch: 700, Avg. Loss: 0.047333640307188035\n",
      "Epoch: 288, Batch: 800, Avg. Loss: 0.044119228944182395\n",
      "Epoch: 288, Batch: 900, Avg. Loss: 0.0493339604139328\n",
      "Epoch: 289, Batch: 100, Avg. Loss: 0.048692505657672885\n",
      "Epoch: 289, Batch: 200, Avg. Loss: 0.045305010378360745\n",
      "Epoch: 289, Batch: 300, Avg. Loss: 0.04085144549608231\n",
      "Epoch: 289, Batch: 400, Avg. Loss: 0.04441323779523373\n",
      "Epoch: 289, Batch: 500, Avg. Loss: 0.04350270077586174\n",
      "Epoch: 289, Batch: 600, Avg. Loss: 0.047385052368044855\n",
      "Epoch: 289, Batch: 700, Avg. Loss: 0.04172483742237091\n",
      "Epoch: 289, Batch: 800, Avg. Loss: 0.04119131401181221\n",
      "Epoch: 289, Batch: 900, Avg. Loss: 0.03971910156309605\n",
      "Epoch: 290, Batch: 100, Avg. Loss: 0.05024605497717857\n",
      "Epoch: 290, Batch: 200, Avg. Loss: 0.04822545528411865\n",
      "Epoch: 290, Batch: 300, Avg. Loss: 0.04712700836360455\n",
      "Epoch: 290, Batch: 400, Avg. Loss: 0.04402224414050579\n",
      "Epoch: 290, Batch: 500, Avg. Loss: 0.04600126013159752\n",
      "Epoch: 290, Batch: 600, Avg. Loss: 0.03436729907989502\n",
      "Epoch: 290, Batch: 700, Avg. Loss: 0.04629983216524124\n",
      "Epoch: 290, Batch: 800, Avg. Loss: 0.042671195417642596\n",
      "Epoch: 290, Batch: 900, Avg. Loss: 0.03764058507978916\n",
      "Epoch: 291, Batch: 100, Avg. Loss: 0.04851364240050316\n",
      "Epoch: 291, Batch: 200, Avg. Loss: 0.04596170082688331\n",
      "Epoch: 291, Batch: 300, Avg. Loss: 0.04820225059986114\n",
      "Epoch: 291, Batch: 400, Avg. Loss: 0.03897265292704105\n",
      "Epoch: 291, Batch: 500, Avg. Loss: 0.04474650226533413\n",
      "Epoch: 291, Batch: 600, Avg. Loss: 0.04417037956416607\n",
      "Epoch: 291, Batch: 700, Avg. Loss: 0.04534843571484089\n",
      "Epoch: 291, Batch: 800, Avg. Loss: 0.05354103699326515\n",
      "Epoch: 291, Batch: 900, Avg. Loss: 0.03803111605346203\n",
      "Epoch: 292, Batch: 100, Avg. Loss: 0.051986095309257505\n",
      "Epoch: 292, Batch: 200, Avg. Loss: 0.044795247912406924\n",
      "Epoch: 292, Batch: 300, Avg. Loss: 0.04569808661937714\n",
      "Epoch: 292, Batch: 400, Avg. Loss: 0.04564321793615818\n",
      "Epoch: 292, Batch: 500, Avg. Loss: 0.03974165763705969\n",
      "Epoch: 292, Batch: 600, Avg. Loss: 0.046178158670663834\n",
      "Epoch: 292, Batch: 700, Avg. Loss: 0.04219448864459992\n",
      "Epoch: 292, Batch: 800, Avg. Loss: 0.04063454769551754\n",
      "Epoch: 292, Batch: 900, Avg. Loss: 0.04060457177460194\n",
      "Epoch: 293, Batch: 100, Avg. Loss: 0.04644950337707996\n",
      "Epoch: 293, Batch: 200, Avg. Loss: 0.05016253150999546\n",
      "Epoch: 293, Batch: 300, Avg. Loss: 0.05111175626516342\n",
      "Epoch: 293, Batch: 400, Avg. Loss: 0.04372223876416683\n",
      "Epoch: 293, Batch: 500, Avg. Loss: 0.04305032156407833\n",
      "Epoch: 293, Batch: 600, Avg. Loss: 0.04223044201731682\n",
      "Epoch: 293, Batch: 700, Avg. Loss: 0.04680626355111599\n",
      "Epoch: 293, Batch: 800, Avg. Loss: 0.047705128341913226\n",
      "Epoch: 293, Batch: 900, Avg. Loss: 0.03974893972277641\n",
      "Epoch: 294, Batch: 100, Avg. Loss: 0.045919172167778015\n",
      "Epoch: 294, Batch: 200, Avg. Loss: 0.04219045199453831\n",
      "Epoch: 294, Batch: 300, Avg. Loss: 0.03281198106706142\n",
      "Epoch: 294, Batch: 400, Avg. Loss: 0.042715275436639784\n",
      "Epoch: 294, Batch: 500, Avg. Loss: 0.052043353468179704\n",
      "Epoch: 294, Batch: 600, Avg. Loss: 0.0451193405687809\n",
      "Epoch: 294, Batch: 700, Avg. Loss: 0.0437622656673193\n",
      "Epoch: 294, Batch: 800, Avg. Loss: 0.04911187425255775\n",
      "Epoch: 294, Batch: 900, Avg. Loss: 0.04896755367517471\n",
      "Epoch: 295, Batch: 100, Avg. Loss: 0.04418873131275177\n",
      "Epoch: 295, Batch: 200, Avg. Loss: 0.04607625804841518\n",
      "Epoch: 295, Batch: 300, Avg. Loss: 0.04568653255701065\n",
      "Epoch: 295, Batch: 400, Avg. Loss: 0.03815294332802296\n",
      "Epoch: 295, Batch: 500, Avg. Loss: 0.04006529949605465\n",
      "Epoch: 295, Batch: 600, Avg. Loss: 0.04982909880578518\n",
      "Epoch: 295, Batch: 700, Avg. Loss: 0.045606679469347\n",
      "Epoch: 295, Batch: 800, Avg. Loss: 0.04797672510147095\n",
      "Epoch: 295, Batch: 900, Avg. Loss: 0.05087060034275055\n",
      "Epoch: 296, Batch: 100, Avg. Loss: 0.04222698912024498\n",
      "Epoch: 296, Batch: 200, Avg. Loss: 0.05227268286049366\n",
      "Epoch: 296, Batch: 300, Avg. Loss: 0.0379877707362175\n",
      "Epoch: 296, Batch: 400, Avg. Loss: 0.047534164860844615\n",
      "Epoch: 296, Batch: 500, Avg. Loss: 0.03935276053845883\n",
      "Epoch: 296, Batch: 600, Avg. Loss: 0.04222245186567306\n",
      "Epoch: 296, Batch: 700, Avg. Loss: 0.04617525152862072\n",
      "Epoch: 296, Batch: 800, Avg. Loss: 0.039524144753813745\n",
      "Epoch: 296, Batch: 900, Avg. Loss: 0.038654592484235764\n",
      "Epoch: 297, Batch: 100, Avg. Loss: 0.03968862429261208\n",
      "Epoch: 297, Batch: 200, Avg. Loss: 0.047656843662261965\n",
      "Epoch: 297, Batch: 300, Avg. Loss: 0.03474342621862889\n",
      "Epoch: 297, Batch: 400, Avg. Loss: 0.05159516021609306\n",
      "Epoch: 297, Batch: 500, Avg. Loss: 0.04563855156302452\n",
      "Epoch: 297, Batch: 600, Avg. Loss: 0.04292251653969288\n",
      "Epoch: 297, Batch: 700, Avg. Loss: 0.04318172611296177\n",
      "Epoch: 297, Batch: 800, Avg. Loss: 0.04583840221166611\n",
      "Epoch: 297, Batch: 900, Avg. Loss: 0.04090155377984047\n",
      "Epoch: 298, Batch: 100, Avg. Loss: 0.04593988515436649\n",
      "Epoch: 298, Batch: 200, Avg. Loss: 0.04139391884207726\n",
      "Epoch: 298, Batch: 300, Avg. Loss: 0.0511650837212801\n",
      "Epoch: 298, Batch: 400, Avg. Loss: 0.042445258498191835\n",
      "Epoch: 298, Batch: 500, Avg. Loss: 0.04732897385954857\n",
      "Epoch: 298, Batch: 600, Avg. Loss: 0.04485895320773125\n",
      "Epoch: 298, Batch: 700, Avg. Loss: 0.038066639378666875\n",
      "Epoch: 298, Batch: 800, Avg. Loss: 0.03945623137056828\n",
      "Epoch: 298, Batch: 900, Avg. Loss: 0.047677216082811356\n",
      "Epoch: 299, Batch: 100, Avg. Loss: 0.04359876736998558\n",
      "Epoch: 299, Batch: 200, Avg. Loss: 0.04183120705187321\n",
      "Epoch: 299, Batch: 300, Avg. Loss: 0.05303568705916405\n",
      "Epoch: 299, Batch: 400, Avg. Loss: 0.04734786920249462\n",
      "Epoch: 299, Batch: 500, Avg. Loss: 0.04933814845979214\n",
      "Epoch: 299, Batch: 600, Avg. Loss: 0.04773022569715977\n",
      "Epoch: 299, Batch: 700, Avg. Loss: 0.04417364098131657\n",
      "Epoch: 299, Batch: 800, Avg. Loss: 0.040801502764225006\n",
      "Epoch: 299, Batch: 900, Avg. Loss: 0.043425913378596304\n",
      "Epoch: 300, Batch: 100, Avg. Loss: 0.047017658427357674\n",
      "Epoch: 300, Batch: 200, Avg. Loss: 0.04595360212028027\n",
      "Epoch: 300, Batch: 300, Avg. Loss: 0.0503045129776001\n",
      "Epoch: 300, Batch: 400, Avg. Loss: 0.04173225127160549\n",
      "Epoch: 300, Batch: 500, Avg. Loss: 0.044311155751347545\n",
      "Epoch: 300, Batch: 600, Avg. Loss: 0.03866192311048508\n",
      "Epoch: 300, Batch: 700, Avg. Loss: 0.04627736009657383\n",
      "Epoch: 300, Batch: 800, Avg. Loss: 0.041640434488654135\n",
      "Epoch: 300, Batch: 900, Avg. Loss: 0.03987531676888466\n",
      "Epoch: 301, Batch: 100, Avg. Loss: 0.043908878564834594\n",
      "Epoch: 301, Batch: 200, Avg. Loss: 0.03991628579795361\n",
      "Epoch: 301, Batch: 300, Avg. Loss: 0.044589057192206384\n",
      "Epoch: 301, Batch: 400, Avg. Loss: 0.05052224263548851\n",
      "Epoch: 301, Batch: 500, Avg. Loss: 0.041599359288811685\n",
      "Epoch: 301, Batch: 600, Avg. Loss: 0.04220127180218697\n",
      "Epoch: 301, Batch: 700, Avg. Loss: 0.04529561497271061\n",
      "Epoch: 301, Batch: 800, Avg. Loss: 0.045119146555662154\n",
      "Epoch: 301, Batch: 900, Avg. Loss: 0.041515839248895646\n",
      "Epoch: 302, Batch: 100, Avg. Loss: 0.036470084711909295\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 302, Batch: 200, Avg. Loss: 0.046755027435719965\n",
      "Epoch: 302, Batch: 300, Avg. Loss: 0.04069695897400379\n",
      "Epoch: 302, Batch: 400, Avg. Loss: 0.045254823490977286\n",
      "Epoch: 302, Batch: 500, Avg. Loss: 0.04401885975152254\n",
      "Epoch: 302, Batch: 600, Avg. Loss: 0.04188823886215687\n",
      "Epoch: 302, Batch: 700, Avg. Loss: 0.051166751310229304\n",
      "Epoch: 302, Batch: 800, Avg. Loss: 0.04153670474886894\n",
      "Epoch: 302, Batch: 900, Avg. Loss: 0.037947098240256306\n",
      "Epoch: 303, Batch: 100, Avg. Loss: 0.043889447897672657\n",
      "Epoch: 303, Batch: 200, Avg. Loss: 0.03907962761819363\n",
      "Epoch: 303, Batch: 300, Avg. Loss: 0.04654176130890846\n",
      "Epoch: 303, Batch: 400, Avg. Loss: 0.04114008769392967\n",
      "Epoch: 303, Batch: 500, Avg. Loss: 0.04899621345102787\n",
      "Epoch: 303, Batch: 600, Avg. Loss: 0.04467306807637215\n",
      "Epoch: 303, Batch: 700, Avg. Loss: 0.040064990743994715\n",
      "Epoch: 303, Batch: 800, Avg. Loss: 0.04173124074935913\n",
      "Epoch: 303, Batch: 900, Avg. Loss: 0.044960406795144084\n",
      "Epoch: 304, Batch: 100, Avg. Loss: 0.05167577899992466\n",
      "Epoch: 304, Batch: 200, Avg. Loss: 0.0378619135171175\n",
      "Epoch: 304, Batch: 300, Avg. Loss: 0.04673697203397751\n",
      "Epoch: 304, Batch: 400, Avg. Loss: 0.04621073037385941\n",
      "Epoch: 304, Batch: 500, Avg. Loss: 0.03965398073196411\n",
      "Epoch: 304, Batch: 600, Avg. Loss: 0.04626933075487614\n",
      "Epoch: 304, Batch: 700, Avg. Loss: 0.031173483356833457\n",
      "Epoch: 304, Batch: 800, Avg. Loss: 0.04224152397364378\n",
      "Epoch: 304, Batch: 900, Avg. Loss: 0.04424926705658436\n",
      "Epoch: 305, Batch: 100, Avg. Loss: 0.044420011043548584\n",
      "Epoch: 305, Batch: 200, Avg. Loss: 0.04066330261528492\n",
      "Epoch: 305, Batch: 300, Avg. Loss: 0.04174036018550396\n",
      "Epoch: 305, Batch: 400, Avg. Loss: 0.04117296017706394\n",
      "Epoch: 305, Batch: 500, Avg. Loss: 0.04071610361337662\n",
      "Epoch: 305, Batch: 600, Avg. Loss: 0.04471313066780567\n",
      "Epoch: 305, Batch: 700, Avg. Loss: 0.04502858355641365\n",
      "Epoch: 305, Batch: 800, Avg. Loss: 0.03716241367161274\n",
      "Epoch: 305, Batch: 900, Avg. Loss: 0.04535394638776779\n",
      "Epoch: 306, Batch: 100, Avg. Loss: 0.03844154067337513\n",
      "Epoch: 306, Batch: 200, Avg. Loss: 0.04345601923763752\n",
      "Epoch: 306, Batch: 300, Avg. Loss: 0.04645195856690407\n",
      "Epoch: 306, Batch: 400, Avg. Loss: 0.04196935676038265\n",
      "Epoch: 306, Batch: 500, Avg. Loss: 0.04103594504296779\n",
      "Epoch: 306, Batch: 600, Avg. Loss: 0.04344627551734447\n",
      "Epoch: 306, Batch: 700, Avg. Loss: 0.03996942214667797\n",
      "Epoch: 306, Batch: 800, Avg. Loss: 0.04232294663786888\n",
      "Epoch: 306, Batch: 900, Avg. Loss: 0.04389035247266292\n",
      "Epoch: 307, Batch: 100, Avg. Loss: 0.04633911550045013\n",
      "Epoch: 307, Batch: 200, Avg. Loss: 0.039217890612781046\n",
      "Epoch: 307, Batch: 300, Avg. Loss: 0.04171835690736771\n",
      "Epoch: 307, Batch: 400, Avg. Loss: 0.041802466437220576\n",
      "Epoch: 307, Batch: 500, Avg. Loss: 0.04873715326189995\n",
      "Epoch: 307, Batch: 600, Avg. Loss: 0.046696639806032184\n",
      "Epoch: 307, Batch: 700, Avg. Loss: 0.040987751260399816\n",
      "Epoch: 307, Batch: 800, Avg. Loss: 0.040920359268784526\n",
      "Epoch: 307, Batch: 900, Avg. Loss: 0.04892998993396759\n",
      "Epoch: 308, Batch: 100, Avg. Loss: 0.04427163973450661\n",
      "Epoch: 308, Batch: 200, Avg. Loss: 0.04717811737209558\n",
      "Epoch: 308, Batch: 300, Avg. Loss: 0.04680940128862858\n",
      "Epoch: 308, Batch: 400, Avg. Loss: 0.04443633906543255\n",
      "Epoch: 308, Batch: 500, Avg. Loss: 0.04481495138257742\n",
      "Epoch: 308, Batch: 600, Avg. Loss: 0.04225809559226036\n",
      "Epoch: 308, Batch: 700, Avg. Loss: 0.04954419732093811\n",
      "Epoch: 308, Batch: 800, Avg. Loss: 0.036979874074459074\n",
      "Epoch: 308, Batch: 900, Avg. Loss: 0.04637498803436756\n",
      "Epoch: 309, Batch: 100, Avg. Loss: 0.043546183556318285\n",
      "Epoch: 309, Batch: 200, Avg. Loss: 0.04420742303133011\n",
      "Epoch: 309, Batch: 300, Avg. Loss: 0.04166862174868584\n",
      "Epoch: 309, Batch: 400, Avg. Loss: 0.052616001218557355\n",
      "Epoch: 309, Batch: 500, Avg. Loss: 0.0426475091278553\n",
      "Epoch: 309, Batch: 600, Avg. Loss: 0.048561443462967875\n",
      "Epoch: 309, Batch: 700, Avg. Loss: 0.03670994870364666\n",
      "Epoch: 309, Batch: 800, Avg. Loss: 0.042797214314341546\n",
      "Epoch: 309, Batch: 900, Avg. Loss: 0.04279866583645344\n",
      "Epoch: 310, Batch: 100, Avg. Loss: 0.03887444540858269\n",
      "Epoch: 310, Batch: 200, Avg. Loss: 0.044212789386510847\n",
      "Epoch: 310, Batch: 300, Avg. Loss: 0.043535165414214136\n",
      "Epoch: 310, Batch: 400, Avg. Loss: 0.05076739452779293\n",
      "Epoch: 310, Batch: 500, Avg. Loss: 0.04886900708079338\n",
      "Epoch: 310, Batch: 600, Avg. Loss: 0.041010655984282496\n",
      "Epoch: 310, Batch: 700, Avg. Loss: 0.04697898864746094\n",
      "Epoch: 310, Batch: 800, Avg. Loss: 0.04050230044871569\n",
      "Epoch: 310, Batch: 900, Avg. Loss: 0.03924605801701546\n",
      "Epoch: 311, Batch: 100, Avg. Loss: 0.040814659893512725\n",
      "Epoch: 311, Batch: 200, Avg. Loss: 0.04147160492837429\n",
      "Epoch: 311, Batch: 300, Avg. Loss: 0.04585380047559738\n",
      "Epoch: 311, Batch: 400, Avg. Loss: 0.04519957341253757\n",
      "Epoch: 311, Batch: 500, Avg. Loss: 0.04428167834877968\n",
      "Epoch: 311, Batch: 600, Avg. Loss: 0.048244227692484855\n",
      "Epoch: 311, Batch: 700, Avg. Loss: 0.047227618023753165\n",
      "Epoch: 311, Batch: 800, Avg. Loss: 0.03720271289348602\n",
      "Epoch: 311, Batch: 900, Avg. Loss: 0.0495960957556963\n",
      "Epoch: 312, Batch: 100, Avg. Loss: 0.037085995525121686\n",
      "Epoch: 312, Batch: 200, Avg. Loss: 0.04147406451404095\n",
      "Epoch: 312, Batch: 300, Avg. Loss: 0.035247145295143126\n",
      "Epoch: 312, Batch: 400, Avg. Loss: 0.04573805928230286\n",
      "Epoch: 312, Batch: 500, Avg. Loss: 0.05251038148999214\n",
      "Epoch: 312, Batch: 600, Avg. Loss: 0.043894794434309006\n",
      "Epoch: 312, Batch: 700, Avg. Loss: 0.05082765758037567\n",
      "Epoch: 312, Batch: 800, Avg. Loss: 0.039427962079644205\n",
      "Epoch: 312, Batch: 900, Avg. Loss: 0.03627323433756828\n",
      "Epoch: 313, Batch: 100, Avg. Loss: 0.0338718443363905\n",
      "Epoch: 313, Batch: 200, Avg. Loss: 0.04864291772246361\n",
      "Epoch: 313, Batch: 300, Avg. Loss: 0.042830773890018464\n",
      "Epoch: 313, Batch: 400, Avg. Loss: 0.03990346789360046\n",
      "Epoch: 313, Batch: 500, Avg. Loss: 0.038516354709863664\n",
      "Epoch: 313, Batch: 600, Avg. Loss: 0.04438328631222248\n",
      "Epoch: 313, Batch: 700, Avg. Loss: 0.04089944049715996\n",
      "Epoch: 313, Batch: 800, Avg. Loss: 0.04057491257786751\n",
      "Epoch: 313, Batch: 900, Avg. Loss: 0.04861576564610005\n",
      "Epoch: 314, Batch: 100, Avg. Loss: 0.040072150006890295\n",
      "Epoch: 314, Batch: 200, Avg. Loss: 0.03975346952676773\n",
      "Epoch: 314, Batch: 300, Avg. Loss: 0.04438886255025864\n",
      "Epoch: 314, Batch: 400, Avg. Loss: 0.03991815224289894\n",
      "Epoch: 314, Batch: 500, Avg. Loss: 0.04564599841833115\n",
      "Epoch: 314, Batch: 600, Avg. Loss: 0.0432888738065958\n",
      "Epoch: 314, Batch: 700, Avg. Loss: 0.04106827758252621\n",
      "Epoch: 314, Batch: 800, Avg. Loss: 0.04380167223513126\n",
      "Epoch: 314, Batch: 900, Avg. Loss: 0.04527540758252144\n",
      "Epoch: 315, Batch: 100, Avg. Loss: 0.04121966123580933\n",
      "Epoch: 315, Batch: 200, Avg. Loss: 0.04482856817543507\n",
      "Epoch: 315, Batch: 300, Avg. Loss: 0.046762275695800784\n",
      "Epoch: 315, Batch: 400, Avg. Loss: 0.0442507054656744\n",
      "Epoch: 315, Batch: 500, Avg. Loss: 0.04090587832033634\n",
      "Epoch: 315, Batch: 600, Avg. Loss: 0.04787896394729614\n",
      "Epoch: 315, Batch: 700, Avg. Loss: 0.047114502638578415\n",
      "Epoch: 315, Batch: 800, Avg. Loss: 0.034309440329670905\n",
      "Epoch: 315, Batch: 900, Avg. Loss: 0.03936900086700916\n",
      "Epoch: 316, Batch: 100, Avg. Loss: 0.036998574510216715\n",
      "Epoch: 316, Batch: 200, Avg. Loss: 0.0443167918920517\n",
      "Epoch: 316, Batch: 300, Avg. Loss: 0.03961716383695602\n",
      "Epoch: 316, Batch: 400, Avg. Loss: 0.034892956763505935\n",
      "Epoch: 316, Batch: 500, Avg. Loss: 0.04914393439888954\n",
      "Epoch: 316, Batch: 600, Avg. Loss: 0.04830370739102364\n",
      "Epoch: 316, Batch: 700, Avg. Loss: 0.045908106416463854\n",
      "Epoch: 316, Batch: 800, Avg. Loss: 0.04773863829672337\n",
      "Epoch: 316, Batch: 900, Avg. Loss: 0.0460944864153862\n",
      "Epoch: 317, Batch: 100, Avg. Loss: 0.04483523108065128\n",
      "Epoch: 317, Batch: 200, Avg. Loss: 0.043459095656871796\n",
      "Epoch: 317, Batch: 300, Avg. Loss: 0.03921502575278282\n",
      "Epoch: 317, Batch: 400, Avg. Loss: 0.04845026507973671\n",
      "Epoch: 317, Batch: 500, Avg. Loss: 0.047185311168432234\n",
      "Epoch: 317, Batch: 600, Avg. Loss: 0.039871352687478066\n",
      "Epoch: 317, Batch: 700, Avg. Loss: 0.04464488409459591\n",
      "Epoch: 317, Batch: 800, Avg. Loss: 0.04154242105782032\n",
      "Epoch: 317, Batch: 900, Avg. Loss: 0.048165454044938084\n",
      "Epoch: 318, Batch: 100, Avg. Loss: 0.03865685127675533\n",
      "Epoch: 318, Batch: 200, Avg. Loss: 0.046897921487689016\n",
      "Epoch: 318, Batch: 300, Avg. Loss: 0.037193619683384894\n",
      "Epoch: 318, Batch: 400, Avg. Loss: 0.03855087406933308\n",
      "Epoch: 318, Batch: 500, Avg. Loss: 0.04857370980083942\n",
      "Epoch: 318, Batch: 600, Avg. Loss: 0.04593007348477841\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 318, Batch: 700, Avg. Loss: 0.04445295788347721\n",
      "Epoch: 318, Batch: 800, Avg. Loss: 0.04415410503745079\n",
      "Epoch: 318, Batch: 900, Avg. Loss: 0.03905699424445629\n",
      "Epoch: 319, Batch: 100, Avg. Loss: 0.04475659117102623\n",
      "Epoch: 319, Batch: 200, Avg. Loss: 0.05248472437262535\n",
      "Epoch: 319, Batch: 300, Avg. Loss: 0.038828912973403934\n",
      "Epoch: 319, Batch: 400, Avg. Loss: 0.0437610699981451\n",
      "Epoch: 319, Batch: 500, Avg. Loss: 0.03881831116974354\n",
      "Epoch: 319, Batch: 600, Avg. Loss: 0.04134479723870754\n",
      "Epoch: 319, Batch: 700, Avg. Loss: 0.05221743673086166\n",
      "Epoch: 319, Batch: 800, Avg. Loss: 0.047194728925824166\n",
      "Epoch: 319, Batch: 900, Avg. Loss: 0.03295864876359701\n",
      "Epoch: 320, Batch: 100, Avg. Loss: 0.031871629506349565\n",
      "Epoch: 320, Batch: 200, Avg. Loss: 0.0427094566076994\n",
      "Epoch: 320, Batch: 300, Avg. Loss: 0.045774006620049476\n",
      "Epoch: 320, Batch: 400, Avg. Loss: 0.0412100650370121\n",
      "Epoch: 320, Batch: 500, Avg. Loss: 0.040416683480143545\n",
      "Epoch: 320, Batch: 600, Avg. Loss: 0.04742576390504837\n",
      "Epoch: 320, Batch: 700, Avg. Loss: 0.04180264338850975\n",
      "Epoch: 320, Batch: 800, Avg. Loss: 0.04853084519505501\n",
      "Epoch: 320, Batch: 900, Avg. Loss: 0.0463240896910429\n",
      "Epoch: 321, Batch: 100, Avg. Loss: 0.04431398153305054\n",
      "Epoch: 321, Batch: 200, Avg. Loss: 0.05093966037034989\n",
      "Epoch: 321, Batch: 300, Avg. Loss: 0.047338964939117434\n",
      "Epoch: 321, Batch: 400, Avg. Loss: 0.040050524175167086\n",
      "Epoch: 321, Batch: 500, Avg. Loss: 0.04019328534603119\n",
      "Epoch: 321, Batch: 600, Avg. Loss: 0.03601042613387108\n",
      "Epoch: 321, Batch: 700, Avg. Loss: 0.05383138947188854\n",
      "Epoch: 321, Batch: 800, Avg. Loss: 0.04093440152704716\n",
      "Epoch: 321, Batch: 900, Avg. Loss: 0.042823300659656525\n",
      "Epoch: 322, Batch: 100, Avg. Loss: 0.04319977544248104\n",
      "Epoch: 322, Batch: 200, Avg. Loss: 0.03480757087469101\n",
      "Epoch: 322, Batch: 300, Avg. Loss: 0.048599830046296116\n",
      "Epoch: 322, Batch: 400, Avg. Loss: 0.04246000126004219\n",
      "Epoch: 322, Batch: 500, Avg. Loss: 0.037701235711574556\n",
      "Epoch: 322, Batch: 600, Avg. Loss: 0.048995330706238745\n",
      "Epoch: 322, Batch: 700, Avg. Loss: 0.04179702416062355\n",
      "Epoch: 322, Batch: 800, Avg. Loss: 0.04381014950573445\n",
      "Epoch: 322, Batch: 900, Avg. Loss: 0.04122799888253212\n",
      "Epoch: 323, Batch: 100, Avg. Loss: 0.04401812605559826\n",
      "Epoch: 323, Batch: 200, Avg. Loss: 0.03791617371141911\n",
      "Epoch: 323, Batch: 300, Avg. Loss: 0.0496781413257122\n",
      "Epoch: 323, Batch: 400, Avg. Loss: 0.04235952563583851\n",
      "Epoch: 323, Batch: 500, Avg. Loss: 0.047090356573462486\n",
      "Epoch: 323, Batch: 600, Avg. Loss: 0.04156680516898632\n",
      "Epoch: 323, Batch: 700, Avg. Loss: 0.03607428036630154\n",
      "Epoch: 323, Batch: 800, Avg. Loss: 0.03716670446097851\n",
      "Epoch: 323, Batch: 900, Avg. Loss: 0.04253686837852001\n",
      "Epoch: 324, Batch: 100, Avg. Loss: 0.0390466932207346\n",
      "Epoch: 324, Batch: 200, Avg. Loss: 0.04196951635181904\n",
      "Epoch: 324, Batch: 300, Avg. Loss: 0.04786436647176742\n",
      "Epoch: 324, Batch: 400, Avg. Loss: 0.043448940441012385\n",
      "Epoch: 324, Batch: 500, Avg. Loss: 0.045857389271259305\n",
      "Epoch: 324, Batch: 600, Avg. Loss: 0.04630033738911152\n",
      "Epoch: 324, Batch: 700, Avg. Loss: 0.047727664113044736\n",
      "Epoch: 324, Batch: 800, Avg. Loss: 0.037867190539836885\n",
      "Epoch: 324, Batch: 900, Avg. Loss: 0.035977212488651274\n",
      "Epoch: 325, Batch: 100, Avg. Loss: 0.04242600068449974\n",
      "Epoch: 325, Batch: 200, Avg. Loss: 0.04270567126572132\n",
      "Epoch: 325, Batch: 300, Avg. Loss: 0.04124722950160503\n",
      "Epoch: 325, Batch: 400, Avg. Loss: 0.0358369380235672\n",
      "Epoch: 325, Batch: 500, Avg. Loss: 0.03931257419288158\n",
      "Epoch: 325, Batch: 600, Avg. Loss: 0.04764977194368839\n",
      "Epoch: 325, Batch: 700, Avg. Loss: 0.03971898719668388\n",
      "Epoch: 325, Batch: 800, Avg. Loss: 0.04729472272098065\n",
      "Epoch: 325, Batch: 900, Avg. Loss: 0.0401225620508194\n",
      "Epoch: 326, Batch: 100, Avg. Loss: 0.04905542261898518\n",
      "Epoch: 326, Batch: 200, Avg. Loss: 0.04443663090467453\n",
      "Epoch: 326, Batch: 300, Avg. Loss: 0.0399274168908596\n",
      "Epoch: 326, Batch: 400, Avg. Loss: 0.03902982108294964\n",
      "Epoch: 326, Batch: 500, Avg. Loss: 0.04259187512099743\n",
      "Epoch: 326, Batch: 600, Avg. Loss: 0.043709037229418754\n",
      "Epoch: 326, Batch: 700, Avg. Loss: 0.04125649631023407\n",
      "Epoch: 326, Batch: 800, Avg. Loss: 0.037577083334326744\n",
      "Epoch: 326, Batch: 900, Avg. Loss: 0.04024365521967411\n",
      "Epoch: 327, Batch: 100, Avg. Loss: 0.046913413405418394\n",
      "Epoch: 327, Batch: 200, Avg. Loss: 0.03798082165420055\n",
      "Epoch: 327, Batch: 300, Avg. Loss: 0.03746112242341042\n",
      "Epoch: 327, Batch: 400, Avg. Loss: 0.045177394598722456\n",
      "Epoch: 327, Batch: 500, Avg. Loss: 0.041828756257891654\n",
      "Epoch: 327, Batch: 600, Avg. Loss: 0.04534790448844433\n",
      "Epoch: 327, Batch: 700, Avg. Loss: 0.042605443224310875\n",
      "Epoch: 327, Batch: 800, Avg. Loss: 0.045656029134988785\n",
      "Epoch: 327, Batch: 900, Avg. Loss: 0.04020532459020615\n",
      "Epoch: 328, Batch: 100, Avg. Loss: 0.048881477415561675\n",
      "Epoch: 328, Batch: 200, Avg. Loss: 0.047576273158192635\n",
      "Epoch: 328, Batch: 300, Avg. Loss: 0.03745927631855011\n",
      "Epoch: 328, Batch: 400, Avg. Loss: 0.043680848628282545\n",
      "Epoch: 328, Batch: 500, Avg. Loss: 0.04139168240129948\n",
      "Epoch: 328, Batch: 600, Avg. Loss: 0.04662911012768745\n",
      "Epoch: 328, Batch: 700, Avg. Loss: 0.04266234077513218\n",
      "Epoch: 328, Batch: 800, Avg. Loss: 0.04138631969690323\n",
      "Epoch: 328, Batch: 900, Avg. Loss: 0.037046967819333074\n",
      "Epoch: 329, Batch: 100, Avg. Loss: 0.04133524462580681\n",
      "Epoch: 329, Batch: 200, Avg. Loss: 0.034989988207817076\n",
      "Epoch: 329, Batch: 300, Avg. Loss: 0.038100906237959864\n",
      "Epoch: 329, Batch: 400, Avg. Loss: 0.03918580941855907\n",
      "Epoch: 329, Batch: 500, Avg. Loss: 0.041878807544708255\n",
      "Epoch: 329, Batch: 600, Avg. Loss: 0.03836781106889248\n",
      "Epoch: 329, Batch: 700, Avg. Loss: 0.04172270596027374\n",
      "Epoch: 329, Batch: 800, Avg. Loss: 0.0378009931743145\n",
      "Epoch: 329, Batch: 900, Avg. Loss: 0.048292830660939215\n",
      "Epoch: 330, Batch: 100, Avg. Loss: 0.04392765253782272\n",
      "Epoch: 330, Batch: 200, Avg. Loss: 0.035681065544486044\n",
      "Epoch: 330, Batch: 300, Avg. Loss: 0.043788951709866526\n",
      "Epoch: 330, Batch: 400, Avg. Loss: 0.04893409512937069\n",
      "Epoch: 330, Batch: 500, Avg. Loss: 0.04633359044790268\n",
      "Epoch: 330, Batch: 600, Avg. Loss: 0.041024784967303274\n",
      "Epoch: 330, Batch: 700, Avg. Loss: 0.039839595779776575\n",
      "Epoch: 330, Batch: 800, Avg. Loss: 0.04590515844523907\n",
      "Epoch: 330, Batch: 900, Avg. Loss: 0.03817488260567188\n",
      "Epoch: 331, Batch: 100, Avg. Loss: 0.04567682333290577\n",
      "Epoch: 331, Batch: 200, Avg. Loss: 0.04584679000079632\n",
      "Epoch: 331, Batch: 300, Avg. Loss: 0.037968640252947806\n",
      "Epoch: 331, Batch: 400, Avg. Loss: 0.04296206995844841\n",
      "Epoch: 331, Batch: 500, Avg. Loss: 0.040623654276132584\n",
      "Epoch: 331, Batch: 600, Avg. Loss: 0.041248901039361956\n",
      "Epoch: 331, Batch: 700, Avg. Loss: 0.03677428297698498\n",
      "Epoch: 331, Batch: 800, Avg. Loss: 0.039072677418589594\n",
      "Epoch: 331, Batch: 900, Avg. Loss: 0.04223172977566719\n",
      "Epoch: 332, Batch: 100, Avg. Loss: 0.03675145708024502\n",
      "Epoch: 332, Batch: 200, Avg. Loss: 0.041292827650904654\n",
      "Epoch: 332, Batch: 300, Avg. Loss: 0.040478358045220375\n",
      "Epoch: 332, Batch: 400, Avg. Loss: 0.04445358455181122\n",
      "Epoch: 332, Batch: 500, Avg. Loss: 0.04273927815258503\n",
      "Epoch: 332, Batch: 600, Avg. Loss: 0.03852988213300705\n",
      "Epoch: 332, Batch: 700, Avg. Loss: 0.039673157036304474\n",
      "Epoch: 332, Batch: 800, Avg. Loss: 0.040690333880484104\n",
      "Epoch: 332, Batch: 900, Avg. Loss: 0.03982702113687992\n",
      "Epoch: 333, Batch: 100, Avg. Loss: 0.0348694359511137\n",
      "Epoch: 333, Batch: 200, Avg. Loss: 0.051021657139062884\n",
      "Epoch: 333, Batch: 300, Avg. Loss: 0.0358096794039011\n",
      "Epoch: 333, Batch: 400, Avg. Loss: 0.035069880709052086\n",
      "Epoch: 333, Batch: 500, Avg. Loss: 0.04561255909502506\n",
      "Epoch: 333, Batch: 600, Avg. Loss: 0.03395263843238354\n",
      "Epoch: 333, Batch: 700, Avg. Loss: 0.04580233126878738\n",
      "Epoch: 333, Batch: 800, Avg. Loss: 0.036159728467464444\n",
      "Epoch: 333, Batch: 900, Avg. Loss: 0.05124085195362568\n",
      "Epoch: 334, Batch: 100, Avg. Loss: 0.044943169951438905\n",
      "Epoch: 334, Batch: 200, Avg. Loss: 0.03670910008251667\n",
      "Epoch: 334, Batch: 300, Avg. Loss: 0.035513573586940766\n",
      "Epoch: 334, Batch: 400, Avg. Loss: 0.042528987526893616\n",
      "Epoch: 334, Batch: 500, Avg. Loss: 0.04200397878885269\n",
      "Epoch: 334, Batch: 600, Avg. Loss: 0.04208903588354587\n",
      "Epoch: 334, Batch: 700, Avg. Loss: 0.037613193392753604\n",
      "Epoch: 334, Batch: 800, Avg. Loss: 0.04095231793820858\n",
      "Epoch: 334, Batch: 900, Avg. Loss: 0.04770793333649635\n",
      "Epoch: 335, Batch: 100, Avg. Loss: 0.04322224013507366\n",
      "Epoch: 335, Batch: 200, Avg. Loss: 0.04333972409367561\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 335, Batch: 300, Avg. Loss: 0.04349137857556343\n",
      "Epoch: 335, Batch: 400, Avg. Loss: 0.040310622602701185\n",
      "Epoch: 335, Batch: 500, Avg. Loss: 0.035898295044898984\n",
      "Epoch: 335, Batch: 600, Avg. Loss: 0.04577283024787903\n",
      "Epoch: 335, Batch: 700, Avg. Loss: 0.04323216147720814\n",
      "Epoch: 335, Batch: 800, Avg. Loss: 0.039215007424354555\n",
      "Epoch: 335, Batch: 900, Avg. Loss: 0.03466269940137863\n",
      "Epoch: 336, Batch: 100, Avg. Loss: 0.04399633273482323\n",
      "Epoch: 336, Batch: 200, Avg. Loss: 0.041803672537207605\n",
      "Epoch: 336, Batch: 300, Avg. Loss: 0.04347770273685456\n",
      "Epoch: 336, Batch: 400, Avg. Loss: 0.04076117753982544\n",
      "Epoch: 336, Batch: 500, Avg. Loss: 0.03879204638302326\n",
      "Epoch: 336, Batch: 600, Avg. Loss: 0.0399371500313282\n",
      "Epoch: 336, Batch: 700, Avg. Loss: 0.04038645826280117\n",
      "Epoch: 336, Batch: 800, Avg. Loss: 0.030769608840346338\n",
      "Epoch: 336, Batch: 900, Avg. Loss: 0.04325932785868645\n",
      "Epoch: 337, Batch: 100, Avg. Loss: 0.03954856127500534\n",
      "Epoch: 337, Batch: 200, Avg. Loss: 0.042448391020298\n",
      "Epoch: 337, Batch: 300, Avg. Loss: 0.04243464663624764\n",
      "Epoch: 337, Batch: 400, Avg. Loss: 0.03712335243821144\n",
      "Epoch: 337, Batch: 500, Avg. Loss: 0.03367517992854118\n",
      "Epoch: 337, Batch: 600, Avg. Loss: 0.04753383435308933\n",
      "Epoch: 337, Batch: 700, Avg. Loss: 0.039847925081849096\n",
      "Epoch: 337, Batch: 800, Avg. Loss: 0.04406239964067936\n",
      "Epoch: 337, Batch: 900, Avg. Loss: 0.04405924566090107\n",
      "Epoch: 338, Batch: 100, Avg. Loss: 0.04121459886431694\n",
      "Epoch: 338, Batch: 200, Avg. Loss: 0.03847367323935032\n",
      "Epoch: 338, Batch: 300, Avg. Loss: 0.033378486931324\n",
      "Epoch: 338, Batch: 400, Avg. Loss: 0.04840061046183109\n",
      "Epoch: 338, Batch: 500, Avg. Loss: 0.03776983007788658\n",
      "Epoch: 338, Batch: 600, Avg. Loss: 0.04426343314349651\n",
      "Epoch: 338, Batch: 700, Avg. Loss: 0.044025643691420556\n",
      "Epoch: 338, Batch: 800, Avg. Loss: 0.043747009560465816\n",
      "Epoch: 338, Batch: 900, Avg. Loss: 0.039172222167253495\n",
      "Epoch: 339, Batch: 100, Avg. Loss: 0.03984496206045151\n",
      "Epoch: 339, Batch: 200, Avg. Loss: 0.04160278104245663\n",
      "Epoch: 339, Batch: 300, Avg. Loss: 0.041497974023222924\n",
      "Epoch: 339, Batch: 400, Avg. Loss: 0.04118319846689701\n",
      "Epoch: 339, Batch: 500, Avg. Loss: 0.04844652980566025\n",
      "Epoch: 339, Batch: 600, Avg. Loss: 0.03290520876646042\n",
      "Epoch: 339, Batch: 700, Avg. Loss: 0.04696913532912731\n",
      "Epoch: 339, Batch: 800, Avg. Loss: 0.04351797394454479\n",
      "Epoch: 339, Batch: 900, Avg. Loss: 0.04392771206796169\n",
      "Epoch: 340, Batch: 100, Avg. Loss: 0.03736781798303127\n",
      "Epoch: 340, Batch: 200, Avg. Loss: 0.03646623633801937\n",
      "Epoch: 340, Batch: 300, Avg. Loss: 0.042529742121696475\n",
      "Epoch: 340, Batch: 400, Avg. Loss: 0.038756216689944264\n",
      "Epoch: 340, Batch: 500, Avg. Loss: 0.04813874565064907\n",
      "Epoch: 340, Batch: 600, Avg. Loss: 0.039211339578032495\n",
      "Epoch: 340, Batch: 700, Avg. Loss: 0.040673643499612805\n",
      "Epoch: 340, Batch: 800, Avg. Loss: 0.03787903226912022\n",
      "Epoch: 340, Batch: 900, Avg. Loss: 0.03747023053467274\n",
      "Epoch: 341, Batch: 100, Avg. Loss: 0.04442570634186268\n",
      "Epoch: 341, Batch: 200, Avg. Loss: 0.039613322988152504\n",
      "Epoch: 341, Batch: 300, Avg. Loss: 0.04194084204733372\n",
      "Epoch: 341, Batch: 400, Avg. Loss: 0.039366764053702354\n",
      "Epoch: 341, Batch: 500, Avg. Loss: 0.0388948829472065\n",
      "Epoch: 341, Batch: 600, Avg. Loss: 0.043198593854904176\n",
      "Epoch: 341, Batch: 700, Avg. Loss: 0.04319772101938724\n",
      "Epoch: 341, Batch: 800, Avg. Loss: 0.04062383405864239\n",
      "Epoch: 341, Batch: 900, Avg. Loss: 0.039770178496837616\n",
      "Epoch: 342, Batch: 100, Avg. Loss: 0.03927283823490143\n",
      "Epoch: 342, Batch: 200, Avg. Loss: 0.03565532192587852\n",
      "Epoch: 342, Batch: 300, Avg. Loss: 0.04392517827451229\n",
      "Epoch: 342, Batch: 400, Avg. Loss: 0.033915297985076906\n",
      "Epoch: 342, Batch: 500, Avg. Loss: 0.039014587700366976\n",
      "Epoch: 342, Batch: 600, Avg. Loss: 0.04519175320863724\n",
      "Epoch: 342, Batch: 700, Avg. Loss: 0.040801638215780256\n",
      "Epoch: 342, Batch: 800, Avg. Loss: 0.03945981547236443\n",
      "Epoch: 342, Batch: 900, Avg. Loss: 0.042175886780023576\n",
      "Epoch: 343, Batch: 100, Avg. Loss: 0.03898937627673149\n",
      "Epoch: 343, Batch: 200, Avg. Loss: 0.03900418497622013\n",
      "Epoch: 343, Batch: 300, Avg. Loss: 0.03249141685664654\n",
      "Epoch: 343, Batch: 400, Avg. Loss: 0.04390730105340481\n",
      "Epoch: 343, Batch: 500, Avg. Loss: 0.05064930260181427\n",
      "Epoch: 343, Batch: 600, Avg. Loss: 0.0386873435229063\n",
      "Epoch: 343, Batch: 700, Avg. Loss: 0.03759474642574787\n",
      "Epoch: 343, Batch: 800, Avg. Loss: 0.042117486745119094\n",
      "Epoch: 343, Batch: 900, Avg. Loss: 0.03938252188265324\n",
      "Epoch: 344, Batch: 100, Avg. Loss: 0.04228358283638954\n",
      "Epoch: 344, Batch: 200, Avg. Loss: 0.0402056884020567\n",
      "Epoch: 344, Batch: 300, Avg. Loss: 0.040525200888514516\n",
      "Epoch: 344, Batch: 400, Avg. Loss: 0.042644179686903955\n",
      "Epoch: 344, Batch: 500, Avg. Loss: 0.043256331011652945\n",
      "Epoch: 344, Batch: 600, Avg. Loss: 0.04497185312211514\n",
      "Epoch: 344, Batch: 700, Avg. Loss: 0.03467533968389034\n",
      "Epoch: 344, Batch: 800, Avg. Loss: 0.03874983489513397\n",
      "Epoch: 344, Batch: 900, Avg. Loss: 0.03415574222803116\n",
      "Epoch: 345, Batch: 100, Avg. Loss: 0.04543597236275673\n",
      "Epoch: 345, Batch: 200, Avg. Loss: 0.039027360305190086\n",
      "Epoch: 345, Batch: 300, Avg. Loss: 0.03866441488265991\n",
      "Epoch: 345, Batch: 400, Avg. Loss: 0.04036780275404453\n",
      "Epoch: 345, Batch: 500, Avg. Loss: 0.04319106616079807\n",
      "Epoch: 345, Batch: 600, Avg. Loss: 0.03722648985683918\n",
      "Epoch: 345, Batch: 700, Avg. Loss: 0.041419135332107546\n",
      "Epoch: 345, Batch: 800, Avg. Loss: 0.04478031888604164\n",
      "Epoch: 345, Batch: 900, Avg. Loss: 0.04911377929151058\n",
      "Epoch: 346, Batch: 100, Avg. Loss: 0.03458570636808872\n",
      "Epoch: 346, Batch: 200, Avg. Loss: 0.04183040700852871\n",
      "Epoch: 346, Batch: 300, Avg. Loss: 0.03736718490719795\n",
      "Epoch: 346, Batch: 400, Avg. Loss: 0.04101645119488239\n",
      "Epoch: 346, Batch: 500, Avg. Loss: 0.041432282626628875\n",
      "Epoch: 346, Batch: 600, Avg. Loss: 0.03129181906580925\n",
      "Epoch: 346, Batch: 700, Avg. Loss: 0.04742877334356308\n",
      "Epoch: 346, Batch: 800, Avg. Loss: 0.041167106032371524\n",
      "Epoch: 346, Batch: 900, Avg. Loss: 0.04202892042696476\n",
      "Epoch: 347, Batch: 100, Avg. Loss: 0.04896751068532467\n",
      "Epoch: 347, Batch: 200, Avg. Loss: 0.04268329069018364\n",
      "Epoch: 347, Batch: 300, Avg. Loss: 0.04081255003809929\n",
      "Epoch: 347, Batch: 400, Avg. Loss: 0.03531943380832672\n",
      "Epoch: 347, Batch: 500, Avg. Loss: 0.040271276012063026\n",
      "Epoch: 347, Batch: 600, Avg. Loss: 0.03924697168171406\n",
      "Epoch: 347, Batch: 700, Avg. Loss: 0.04311989173293114\n",
      "Epoch: 347, Batch: 800, Avg. Loss: 0.03320994794368744\n",
      "Epoch: 347, Batch: 900, Avg. Loss: 0.03155722610652447\n",
      "Epoch: 348, Batch: 100, Avg. Loss: 0.038226066380739214\n",
      "Epoch: 348, Batch: 200, Avg. Loss: 0.03905292347073555\n",
      "Epoch: 348, Batch: 300, Avg. Loss: 0.0431581774353981\n",
      "Epoch: 348, Batch: 400, Avg. Loss: 0.041044958680868146\n",
      "Epoch: 348, Batch: 500, Avg. Loss: 0.039267907217144966\n",
      "Epoch: 348, Batch: 600, Avg. Loss: 0.04086823500692845\n",
      "Epoch: 348, Batch: 700, Avg. Loss: 0.044452432990074155\n",
      "Epoch: 348, Batch: 800, Avg. Loss: 0.04052448101341724\n",
      "Epoch: 348, Batch: 900, Avg. Loss: 0.036431469693779944\n",
      "Epoch: 349, Batch: 100, Avg. Loss: 0.03827652670443058\n",
      "Epoch: 349, Batch: 200, Avg. Loss: 0.03854322358965874\n",
      "Epoch: 349, Batch: 300, Avg. Loss: 0.03662439726293087\n",
      "Epoch: 349, Batch: 400, Avg. Loss: 0.04143141977488995\n",
      "Epoch: 349, Batch: 500, Avg. Loss: 0.0476269119232893\n",
      "Epoch: 349, Batch: 600, Avg. Loss: 0.04372313156723976\n",
      "Epoch: 349, Batch: 700, Avg. Loss: 0.04719438172876835\n",
      "Epoch: 349, Batch: 800, Avg. Loss: 0.038442915380001066\n",
      "Epoch: 349, Batch: 900, Avg. Loss: 0.038719580620527265\n",
      "Epoch: 350, Batch: 100, Avg. Loss: 0.03962551444768905\n",
      "Epoch: 350, Batch: 200, Avg. Loss: 0.03577846385538578\n",
      "Epoch: 350, Batch: 300, Avg. Loss: 0.044250286146998406\n",
      "Epoch: 350, Batch: 400, Avg. Loss: 0.03679041191935539\n",
      "Epoch: 350, Batch: 500, Avg. Loss: 0.0353466984629631\n",
      "Epoch: 350, Batch: 600, Avg. Loss: 0.04251723825931549\n",
      "Epoch: 350, Batch: 700, Avg. Loss: 0.03908075526356697\n",
      "Epoch: 350, Batch: 800, Avg. Loss: 0.032491137236356736\n",
      "Epoch: 350, Batch: 900, Avg. Loss: 0.04517829820513725\n",
      "Epoch: 351, Batch: 100, Avg. Loss: 0.042666920945048335\n",
      "Epoch: 351, Batch: 200, Avg. Loss: 0.044409413337707516\n",
      "Epoch: 351, Batch: 300, Avg. Loss: 0.03787675604224205\n",
      "Epoch: 351, Batch: 400, Avg. Loss: 0.036712630093097685\n",
      "Epoch: 351, Batch: 500, Avg. Loss: 0.038082359954714776\n",
      "Epoch: 351, Batch: 600, Avg. Loss: 0.03484307184815407\n",
      "Epoch: 351, Batch: 700, Avg. Loss: 0.04393167838454246\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 351, Batch: 800, Avg. Loss: 0.038528555035591126\n",
      "Epoch: 351, Batch: 900, Avg. Loss: 0.039132950231432914\n",
      "Epoch: 352, Batch: 100, Avg. Loss: 0.03406048662960529\n",
      "Epoch: 352, Batch: 200, Avg. Loss: 0.04037475526332855\n",
      "Epoch: 352, Batch: 300, Avg. Loss: 0.044294521734118464\n",
      "Epoch: 352, Batch: 400, Avg. Loss: 0.045253471210598945\n",
      "Epoch: 352, Batch: 500, Avg. Loss: 0.04035529427230358\n",
      "Epoch: 352, Batch: 600, Avg. Loss: 0.038835993334650994\n",
      "Epoch: 352, Batch: 700, Avg. Loss: 0.043260916098952294\n",
      "Epoch: 352, Batch: 800, Avg. Loss: 0.04546688042581081\n",
      "Epoch: 352, Batch: 900, Avg. Loss: 0.03589436337351799\n",
      "Epoch: 353, Batch: 100, Avg. Loss: 0.04029320031404495\n",
      "Epoch: 353, Batch: 200, Avg. Loss: 0.043048424273729326\n",
      "Epoch: 353, Batch: 300, Avg. Loss: 0.03785359174013138\n",
      "Epoch: 353, Batch: 400, Avg. Loss: 0.04523733295500278\n",
      "Epoch: 353, Batch: 500, Avg. Loss: 0.037900039330124854\n",
      "Epoch: 353, Batch: 600, Avg. Loss: 0.03916176587343216\n",
      "Epoch: 353, Batch: 700, Avg. Loss: 0.03660316966474056\n",
      "Epoch: 353, Batch: 800, Avg. Loss: 0.04463356703519821\n",
      "Epoch: 353, Batch: 900, Avg. Loss: 0.039108037501573566\n",
      "Epoch: 354, Batch: 100, Avg. Loss: 0.03298729449510574\n",
      "Epoch: 354, Batch: 200, Avg. Loss: 0.03820489674806595\n",
      "Epoch: 354, Batch: 300, Avg. Loss: 0.047136293873190876\n",
      "Epoch: 354, Batch: 400, Avg. Loss: 0.042738639190793036\n",
      "Epoch: 354, Batch: 500, Avg. Loss: 0.03730434983968735\n",
      "Epoch: 354, Batch: 600, Avg. Loss: 0.04164064064621925\n",
      "Epoch: 354, Batch: 700, Avg. Loss: 0.04411849409341812\n",
      "Epoch: 354, Batch: 800, Avg. Loss: 0.034339490309357644\n",
      "Epoch: 354, Batch: 900, Avg. Loss: 0.04007585644721985\n",
      "Epoch: 355, Batch: 100, Avg. Loss: 0.03955605886876583\n",
      "Epoch: 355, Batch: 200, Avg. Loss: 0.0347496035695076\n",
      "Epoch: 355, Batch: 300, Avg. Loss: 0.033842018097639086\n",
      "Epoch: 355, Batch: 400, Avg. Loss: 0.03707991950213909\n",
      "Epoch: 355, Batch: 500, Avg. Loss: 0.040311682224273684\n",
      "Epoch: 355, Batch: 600, Avg. Loss: 0.04599924929440022\n",
      "Epoch: 355, Batch: 700, Avg. Loss: 0.038038723170757294\n",
      "Epoch: 355, Batch: 800, Avg. Loss: 0.04147487722337246\n",
      "Epoch: 355, Batch: 900, Avg. Loss: 0.040325030982494355\n",
      "Epoch: 356, Batch: 100, Avg. Loss: 0.04287684381008148\n",
      "Epoch: 356, Batch: 200, Avg. Loss: 0.028816455602645875\n",
      "Epoch: 356, Batch: 300, Avg. Loss: 0.04278508551418781\n",
      "Epoch: 356, Batch: 400, Avg. Loss: 0.04873830169439316\n",
      "Epoch: 356, Batch: 500, Avg. Loss: 0.0336024534702301\n",
      "Epoch: 356, Batch: 600, Avg. Loss: 0.03159067191183567\n",
      "Epoch: 356, Batch: 700, Avg. Loss: 0.04161105379462242\n",
      "Epoch: 356, Batch: 800, Avg. Loss: 0.04027957409620285\n",
      "Epoch: 356, Batch: 900, Avg. Loss: 0.04270150616765022\n",
      "Epoch: 357, Batch: 100, Avg. Loss: 0.03596697948873043\n",
      "Epoch: 357, Batch: 200, Avg. Loss: 0.03884724035859108\n",
      "Epoch: 357, Batch: 300, Avg. Loss: 0.049339624792337416\n",
      "Epoch: 357, Batch: 400, Avg. Loss: 0.03886727809906006\n",
      "Epoch: 357, Batch: 500, Avg. Loss: 0.04126287765800953\n",
      "Epoch: 357, Batch: 600, Avg. Loss: 0.04428431235253811\n",
      "Epoch: 357, Batch: 700, Avg. Loss: 0.03417512707412243\n",
      "Epoch: 357, Batch: 800, Avg. Loss: 0.04033321723341942\n",
      "Epoch: 357, Batch: 900, Avg. Loss: 0.04266416542232036\n",
      "Epoch: 358, Batch: 100, Avg. Loss: 0.048756546303629876\n",
      "Epoch: 358, Batch: 200, Avg. Loss: 0.03735511131584644\n",
      "Epoch: 358, Batch: 300, Avg. Loss: 0.03869879186153412\n",
      "Epoch: 358, Batch: 400, Avg. Loss: 0.0318257337808609\n",
      "Epoch: 358, Batch: 500, Avg. Loss: 0.04023030497133732\n",
      "Epoch: 358, Batch: 600, Avg. Loss: 0.04057147666811943\n",
      "Epoch: 358, Batch: 700, Avg. Loss: 0.044103020578622816\n",
      "Epoch: 358, Batch: 800, Avg. Loss: 0.04421406537294388\n",
      "Epoch: 358, Batch: 900, Avg. Loss: 0.036883839666843415\n",
      "Epoch: 359, Batch: 100, Avg. Loss: 0.03731438063085079\n",
      "Epoch: 359, Batch: 200, Avg. Loss: 0.04036013074219227\n",
      "Epoch: 359, Batch: 300, Avg. Loss: 0.04211065970361233\n",
      "Epoch: 359, Batch: 400, Avg. Loss: 0.04598727360367775\n",
      "Epoch: 359, Batch: 500, Avg. Loss: 0.03574633710086346\n",
      "Epoch: 359, Batch: 600, Avg. Loss: 0.04231271810829639\n",
      "Epoch: 359, Batch: 700, Avg. Loss: 0.0482520580291748\n",
      "Epoch: 359, Batch: 800, Avg. Loss: 0.042705494314432144\n",
      "Epoch: 359, Batch: 900, Avg. Loss: 0.04327727593481541\n",
      "Epoch: 360, Batch: 100, Avg. Loss: 0.0366048513352871\n",
      "Epoch: 360, Batch: 200, Avg. Loss: 0.03669483445584774\n",
      "Epoch: 360, Batch: 300, Avg. Loss: 0.04278451211750507\n",
      "Epoch: 360, Batch: 400, Avg. Loss: 0.036654363945126535\n",
      "Epoch: 360, Batch: 500, Avg. Loss: 0.03444221578538418\n",
      "Epoch: 360, Batch: 600, Avg. Loss: 0.03810002461075783\n",
      "Epoch: 360, Batch: 700, Avg. Loss: 0.04124591335654259\n",
      "Epoch: 360, Batch: 800, Avg. Loss: 0.03329758636653423\n",
      "Epoch: 360, Batch: 900, Avg. Loss: 0.04266252294182778\n",
      "Epoch: 361, Batch: 100, Avg. Loss: 0.045385049432516096\n",
      "Epoch: 361, Batch: 200, Avg. Loss: 0.04213628113269806\n",
      "Epoch: 361, Batch: 300, Avg. Loss: 0.04003825232386589\n",
      "Epoch: 361, Batch: 400, Avg. Loss: 0.04191157273948193\n",
      "Epoch: 361, Batch: 500, Avg. Loss: 0.04228651024401188\n",
      "Epoch: 361, Batch: 600, Avg. Loss: 0.045004210323095324\n",
      "Epoch: 361, Batch: 700, Avg. Loss: 0.0348189078271389\n",
      "Epoch: 361, Batch: 800, Avg. Loss: 0.0404610875248909\n",
      "Epoch: 361, Batch: 900, Avg. Loss: 0.03228314571082592\n",
      "Epoch: 362, Batch: 100, Avg. Loss: 0.039200570210814474\n",
      "Epoch: 362, Batch: 200, Avg. Loss: 0.041810483261942866\n",
      "Epoch: 362, Batch: 300, Avg. Loss: 0.04179918825626373\n",
      "Epoch: 362, Batch: 400, Avg. Loss: 0.040850506350398066\n",
      "Epoch: 362, Batch: 500, Avg. Loss: 0.03995596334338188\n",
      "Epoch: 362, Batch: 600, Avg. Loss: 0.04060984805226326\n",
      "Epoch: 362, Batch: 700, Avg. Loss: 0.04092919707298279\n",
      "Epoch: 362, Batch: 800, Avg. Loss: 0.032650922387838364\n",
      "Epoch: 362, Batch: 900, Avg. Loss: 0.04256011553108692\n",
      "Epoch: 363, Batch: 100, Avg. Loss: 0.04112912543118\n",
      "Epoch: 363, Batch: 200, Avg. Loss: 0.0423769698292017\n",
      "Epoch: 363, Batch: 300, Avg. Loss: 0.034504179954528806\n",
      "Epoch: 363, Batch: 400, Avg. Loss: 0.03906561657786369\n",
      "Epoch: 363, Batch: 500, Avg. Loss: 0.041965817362070085\n",
      "Epoch: 363, Batch: 600, Avg. Loss: 0.04951960444450378\n",
      "Epoch: 363, Batch: 700, Avg. Loss: 0.04255395174026489\n",
      "Epoch: 363, Batch: 800, Avg. Loss: 0.04065071783959866\n",
      "Epoch: 363, Batch: 900, Avg. Loss: 0.03385865449905395\n",
      "Epoch: 364, Batch: 100, Avg. Loss: 0.038709283247590064\n",
      "Epoch: 364, Batch: 200, Avg. Loss: 0.036869593039155005\n",
      "Epoch: 364, Batch: 300, Avg. Loss: 0.04011942632496357\n",
      "Epoch: 364, Batch: 400, Avg. Loss: 0.04662340261042118\n",
      "Epoch: 364, Batch: 500, Avg. Loss: 0.039269381761550905\n",
      "Epoch: 364, Batch: 600, Avg. Loss: 0.04338187992572785\n",
      "Epoch: 364, Batch: 700, Avg. Loss: 0.0351702269166708\n",
      "Epoch: 364, Batch: 800, Avg. Loss: 0.03878991343080997\n",
      "Epoch: 364, Batch: 900, Avg. Loss: 0.03946870982646942\n",
      "Epoch: 365, Batch: 100, Avg. Loss: 0.03427265889942646\n",
      "Epoch: 365, Batch: 200, Avg. Loss: 0.03789249002933502\n",
      "Epoch: 365, Batch: 300, Avg. Loss: 0.04562345705926418\n",
      "Epoch: 365, Batch: 400, Avg. Loss: 0.043230246156454086\n",
      "Epoch: 365, Batch: 500, Avg. Loss: 0.034774234667420384\n",
      "Epoch: 365, Batch: 600, Avg. Loss: 0.042639864087104795\n",
      "Epoch: 365, Batch: 700, Avg. Loss: 0.03202496387064457\n",
      "Epoch: 365, Batch: 800, Avg. Loss: 0.04023600287735462\n",
      "Epoch: 365, Batch: 900, Avg. Loss: 0.034630781933665276\n",
      "Epoch: 366, Batch: 100, Avg. Loss: 0.037429555207490924\n",
      "Epoch: 366, Batch: 200, Avg. Loss: 0.035063055753707884\n",
      "Epoch: 366, Batch: 300, Avg. Loss: 0.035755762606859205\n",
      "Epoch: 366, Batch: 400, Avg. Loss: 0.0368471172451973\n",
      "Epoch: 366, Batch: 500, Avg. Loss: 0.039984553009271624\n",
      "Epoch: 366, Batch: 600, Avg. Loss: 0.04748626559972763\n",
      "Epoch: 366, Batch: 700, Avg. Loss: 0.03963626146316528\n",
      "Epoch: 366, Batch: 800, Avg. Loss: 0.034800552353262904\n",
      "Epoch: 366, Batch: 900, Avg. Loss: 0.03848809838294983\n",
      "Epoch: 367, Batch: 100, Avg. Loss: 0.033390622958540915\n",
      "Epoch: 367, Batch: 200, Avg. Loss: 0.04522612564265728\n",
      "Epoch: 367, Batch: 300, Avg. Loss: 0.03779989711940288\n",
      "Epoch: 367, Batch: 400, Avg. Loss: 0.032915515378117564\n",
      "Epoch: 367, Batch: 500, Avg. Loss: 0.041427124366164206\n",
      "Epoch: 367, Batch: 600, Avg. Loss: 0.039394247755408285\n",
      "Epoch: 367, Batch: 700, Avg. Loss: 0.03666877254843712\n",
      "Epoch: 367, Batch: 800, Avg. Loss: 0.03720013923943043\n",
      "Epoch: 367, Batch: 900, Avg. Loss: 0.03923178769648075\n",
      "Epoch: 368, Batch: 100, Avg. Loss: 0.03647778145968914\n",
      "Epoch: 368, Batch: 200, Avg. Loss: 0.04087799280881882\n",
      "Epoch: 368, Batch: 300, Avg. Loss: 0.04244629874825478\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 368, Batch: 400, Avg. Loss: 0.04318962164223194\n",
      "Epoch: 368, Batch: 500, Avg. Loss: 0.038367625474929806\n",
      "Epoch: 368, Batch: 600, Avg. Loss: 0.03839585952460766\n",
      "Epoch: 368, Batch: 700, Avg. Loss: 0.0381106524169445\n",
      "Epoch: 368, Batch: 800, Avg. Loss: 0.03982769705355167\n",
      "Epoch: 368, Batch: 900, Avg. Loss: 0.043124937638640407\n",
      "Epoch: 369, Batch: 100, Avg. Loss: 0.04276150800287724\n",
      "Epoch: 369, Batch: 200, Avg. Loss: 0.03610868223011494\n",
      "Epoch: 369, Batch: 300, Avg. Loss: 0.04000809364020824\n",
      "Epoch: 369, Batch: 400, Avg. Loss: 0.04042885899543762\n",
      "Epoch: 369, Batch: 500, Avg. Loss: 0.03717653468251228\n",
      "Epoch: 369, Batch: 600, Avg. Loss: 0.04344816274940968\n",
      "Epoch: 369, Batch: 700, Avg. Loss: 0.039560638070106506\n",
      "Epoch: 369, Batch: 800, Avg. Loss: 0.03537798218429088\n",
      "Epoch: 369, Batch: 900, Avg. Loss: 0.03597965747117996\n",
      "Epoch: 370, Batch: 100, Avg. Loss: 0.03805237665772438\n",
      "Epoch: 370, Batch: 200, Avg. Loss: 0.03762213990092277\n",
      "Epoch: 370, Batch: 300, Avg. Loss: 0.036833720728754994\n",
      "Epoch: 370, Batch: 400, Avg. Loss: 0.04334892965853214\n",
      "Epoch: 370, Batch: 500, Avg. Loss: 0.03847784675657749\n",
      "Epoch: 370, Batch: 600, Avg. Loss: 0.04424450308084488\n",
      "Epoch: 370, Batch: 700, Avg. Loss: 0.02973508432507515\n",
      "Epoch: 370, Batch: 800, Avg. Loss: 0.043286645263433454\n",
      "Epoch: 370, Batch: 900, Avg. Loss: 0.041261442154645916\n",
      "Epoch: 371, Batch: 100, Avg. Loss: 0.035414887592196465\n",
      "Epoch: 371, Batch: 200, Avg. Loss: 0.03675430074334145\n",
      "Epoch: 371, Batch: 300, Avg. Loss: 0.0447279492020607\n",
      "Epoch: 371, Batch: 400, Avg. Loss: 0.04346748456358909\n",
      "Epoch: 371, Batch: 500, Avg. Loss: 0.05101544015109539\n",
      "Epoch: 371, Batch: 600, Avg. Loss: 0.034142036586999894\n",
      "Epoch: 371, Batch: 700, Avg. Loss: 0.03941069215536117\n",
      "Epoch: 371, Batch: 800, Avg. Loss: 0.034007236659526825\n",
      "Epoch: 371, Batch: 900, Avg. Loss: 0.042725045531988144\n",
      "Epoch: 372, Batch: 100, Avg. Loss: 0.03550216011703014\n",
      "Epoch: 372, Batch: 200, Avg. Loss: 0.036061801761388776\n",
      "Epoch: 372, Batch: 300, Avg. Loss: 0.039713474810123446\n",
      "Epoch: 372, Batch: 400, Avg. Loss: 0.041824864447116854\n",
      "Epoch: 372, Batch: 500, Avg. Loss: 0.03822333510965109\n",
      "Epoch: 372, Batch: 600, Avg. Loss: 0.041975072026252745\n",
      "Epoch: 372, Batch: 700, Avg. Loss: 0.03864925317466259\n",
      "Epoch: 372, Batch: 800, Avg. Loss: 0.02948641449213028\n",
      "Epoch: 372, Batch: 900, Avg. Loss: 0.03875394694507122\n",
      "Epoch: 373, Batch: 100, Avg. Loss: 0.03619807995855808\n",
      "Epoch: 373, Batch: 200, Avg. Loss: 0.035132653266191485\n",
      "Epoch: 373, Batch: 300, Avg. Loss: 0.042803619801998136\n",
      "Epoch: 373, Batch: 400, Avg. Loss: 0.0381501067429781\n",
      "Epoch: 373, Batch: 500, Avg. Loss: 0.042252754941582676\n",
      "Epoch: 373, Batch: 600, Avg. Loss: 0.035817225500941276\n",
      "Epoch: 373, Batch: 700, Avg. Loss: 0.04038658514618874\n",
      "Epoch: 373, Batch: 800, Avg. Loss: 0.03478336296975613\n",
      "Epoch: 373, Batch: 900, Avg. Loss: 0.04082994744181633\n",
      "Epoch: 374, Batch: 100, Avg. Loss: 0.0393155287206173\n",
      "Epoch: 374, Batch: 200, Avg. Loss: 0.03978299088776112\n",
      "Epoch: 374, Batch: 300, Avg. Loss: 0.03535218887031078\n",
      "Epoch: 374, Batch: 400, Avg. Loss: 0.037716721370816234\n",
      "Epoch: 374, Batch: 500, Avg. Loss: 0.03474252715706825\n",
      "Epoch: 374, Batch: 600, Avg. Loss: 0.0332952792942524\n",
      "Epoch: 374, Batch: 700, Avg. Loss: 0.037370288968086245\n",
      "Epoch: 374, Batch: 800, Avg. Loss: 0.032501919865608214\n",
      "Epoch: 374, Batch: 900, Avg. Loss: 0.040932269543409346\n",
      "Epoch: 375, Batch: 100, Avg. Loss: 0.03482278622686863\n",
      "Epoch: 375, Batch: 200, Avg. Loss: 0.03819600164890289\n",
      "Epoch: 375, Batch: 300, Avg. Loss: 0.0337881837785244\n",
      "Epoch: 375, Batch: 400, Avg. Loss: 0.03509562484920025\n",
      "Epoch: 375, Batch: 500, Avg. Loss: 0.038533420413732526\n",
      "Epoch: 375, Batch: 600, Avg. Loss: 0.03577725917100907\n",
      "Epoch: 375, Batch: 700, Avg. Loss: 0.0383589705824852\n",
      "Epoch: 375, Batch: 800, Avg. Loss: 0.039049281775951385\n",
      "Epoch: 375, Batch: 900, Avg. Loss: 0.03325958758592606\n",
      "Epoch: 376, Batch: 100, Avg. Loss: 0.03292783834040165\n",
      "Epoch: 376, Batch: 200, Avg. Loss: 0.0363394608348608\n",
      "Epoch: 376, Batch: 300, Avg. Loss: 0.03260379374027252\n",
      "Epoch: 376, Batch: 400, Avg. Loss: 0.03878753311932087\n",
      "Epoch: 376, Batch: 500, Avg. Loss: 0.03854971304535866\n",
      "Epoch: 376, Batch: 600, Avg. Loss: 0.03359453782439232\n",
      "Epoch: 376, Batch: 700, Avg. Loss: 0.03026449203491211\n",
      "Epoch: 376, Batch: 800, Avg. Loss: 0.0437136959284544\n",
      "Epoch: 376, Batch: 900, Avg. Loss: 0.040395238995552064\n",
      "Epoch: 377, Batch: 100, Avg. Loss: 0.03492626242339611\n",
      "Epoch: 377, Batch: 200, Avg. Loss: 0.04189479306340218\n",
      "Epoch: 377, Batch: 300, Avg. Loss: 0.03918461203575134\n",
      "Epoch: 377, Batch: 400, Avg. Loss: 0.04281670086085796\n",
      "Epoch: 377, Batch: 500, Avg. Loss: 0.03409649476408958\n",
      "Epoch: 377, Batch: 600, Avg. Loss: 0.033689054399728774\n",
      "Epoch: 377, Batch: 700, Avg. Loss: 0.037339220866560933\n",
      "Epoch: 377, Batch: 800, Avg. Loss: 0.03864401027560234\n",
      "Epoch: 377, Batch: 900, Avg. Loss: 0.04146033741533756\n",
      "Epoch: 378, Batch: 100, Avg. Loss: 0.041746788248419764\n",
      "Epoch: 378, Batch: 200, Avg. Loss: 0.038806116208434105\n",
      "Epoch: 378, Batch: 300, Avg. Loss: 0.037933920621871946\n",
      "Epoch: 378, Batch: 400, Avg. Loss: 0.03806771293282509\n",
      "Epoch: 378, Batch: 500, Avg. Loss: 0.03977463066577911\n",
      "Epoch: 378, Batch: 600, Avg. Loss: 0.037630863562226294\n",
      "Epoch: 378, Batch: 700, Avg. Loss: 0.04002385765314102\n",
      "Epoch: 378, Batch: 800, Avg. Loss: 0.045798756405711176\n",
      "Epoch: 378, Batch: 900, Avg. Loss: 0.031021662726998328\n",
      "Epoch: 379, Batch: 100, Avg. Loss: 0.0367007315903902\n",
      "Epoch: 379, Batch: 200, Avg. Loss: 0.03940950244665146\n",
      "Epoch: 379, Batch: 300, Avg. Loss: 0.03631818525493145\n",
      "Epoch: 379, Batch: 400, Avg. Loss: 0.03754621423780918\n",
      "Epoch: 379, Batch: 500, Avg. Loss: 0.04124679535627365\n",
      "Epoch: 379, Batch: 600, Avg. Loss: 0.042299738228321074\n",
      "Epoch: 379, Batch: 700, Avg. Loss: 0.033252817541360856\n",
      "Epoch: 379, Batch: 800, Avg. Loss: 0.043057601675391195\n",
      "Epoch: 379, Batch: 900, Avg. Loss: 0.0307452604919672\n",
      "Epoch: 380, Batch: 100, Avg. Loss: 0.036314487382769586\n",
      "Epoch: 380, Batch: 200, Avg. Loss: 0.03316852577030659\n",
      "Epoch: 380, Batch: 300, Avg. Loss: 0.04289198577404022\n",
      "Epoch: 380, Batch: 400, Avg. Loss: 0.041713253483176234\n",
      "Epoch: 380, Batch: 500, Avg. Loss: 0.039044173806905745\n",
      "Epoch: 380, Batch: 600, Avg. Loss: 0.03171712592244148\n",
      "Epoch: 380, Batch: 700, Avg. Loss: 0.038282273486256596\n",
      "Epoch: 380, Batch: 800, Avg. Loss: 0.041957078874111174\n",
      "Epoch: 380, Batch: 900, Avg. Loss: 0.03511750034987927\n",
      "Epoch: 381, Batch: 100, Avg. Loss: 0.04441148795187473\n",
      "Epoch: 381, Batch: 200, Avg. Loss: 0.04001244947314262\n",
      "Epoch: 381, Batch: 300, Avg. Loss: 0.037289857044816015\n",
      "Epoch: 381, Batch: 400, Avg. Loss: 0.041705539748072624\n",
      "Epoch: 381, Batch: 500, Avg. Loss: 0.03686525605618954\n",
      "Epoch: 381, Batch: 600, Avg. Loss: 0.03634575814008713\n",
      "Epoch: 381, Batch: 700, Avg. Loss: 0.040592023059725764\n",
      "Epoch: 381, Batch: 800, Avg. Loss: 0.03692869186401367\n",
      "Epoch: 381, Batch: 900, Avg. Loss: 0.038530408442020415\n",
      "Epoch: 382, Batch: 100, Avg. Loss: 0.033336040079593655\n",
      "Epoch: 382, Batch: 200, Avg. Loss: 0.04113490797579288\n",
      "Epoch: 382, Batch: 300, Avg. Loss: 0.03523678213357925\n",
      "Epoch: 382, Batch: 400, Avg. Loss: 0.04087468728423119\n",
      "Epoch: 382, Batch: 500, Avg. Loss: 0.038569934368133545\n",
      "Epoch: 382, Batch: 600, Avg. Loss: 0.0455987137556076\n",
      "Epoch: 382, Batch: 700, Avg. Loss: 0.0321075676381588\n",
      "Epoch: 382, Batch: 800, Avg. Loss: 0.04275983579456806\n",
      "Epoch: 382, Batch: 900, Avg. Loss: 0.038207502514123914\n",
      "Epoch: 383, Batch: 100, Avg. Loss: 0.0392062383890152\n",
      "Epoch: 383, Batch: 200, Avg. Loss: 0.043594360053539276\n",
      "Epoch: 383, Batch: 300, Avg. Loss: 0.033956949785351756\n",
      "Epoch: 383, Batch: 400, Avg. Loss: 0.0375525451824069\n",
      "Epoch: 383, Batch: 500, Avg. Loss: 0.04074308447539807\n",
      "Epoch: 383, Batch: 600, Avg. Loss: 0.038558706864714624\n",
      "Epoch: 383, Batch: 700, Avg. Loss: 0.037488210275769235\n",
      "Epoch: 383, Batch: 800, Avg. Loss: 0.041316309124231336\n",
      "Epoch: 383, Batch: 900, Avg. Loss: 0.03731026843190193\n",
      "Epoch: 384, Batch: 100, Avg. Loss: 0.034073445722460745\n",
      "Epoch: 384, Batch: 200, Avg. Loss: 0.038168174028396604\n",
      "Epoch: 384, Batch: 300, Avg. Loss: 0.029534218534827232\n",
      "Epoch: 384, Batch: 400, Avg. Loss: 0.0432586582750082\n",
      "Epoch: 384, Batch: 500, Avg. Loss: 0.03874064572155476\n",
      "Epoch: 384, Batch: 600, Avg. Loss: 0.041642519012093544\n",
      "Epoch: 384, Batch: 700, Avg. Loss: 0.030123710706830023\n",
      "Epoch: 384, Batch: 800, Avg. Loss: 0.03199950903654099\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 384, Batch: 900, Avg. Loss: 0.03790405377745629\n",
      "Epoch: 385, Batch: 100, Avg. Loss: 0.03765949569642544\n",
      "Epoch: 385, Batch: 200, Avg. Loss: 0.03340772591531277\n",
      "Epoch: 385, Batch: 300, Avg. Loss: 0.04318219944834709\n",
      "Epoch: 385, Batch: 400, Avg. Loss: 0.03872257456183434\n",
      "Epoch: 385, Batch: 500, Avg. Loss: 0.031448947191238405\n",
      "Epoch: 385, Batch: 600, Avg. Loss: 0.03544173322618008\n",
      "Epoch: 385, Batch: 700, Avg. Loss: 0.040776981115341185\n",
      "Epoch: 385, Batch: 800, Avg. Loss: 0.0339517530053854\n",
      "Epoch: 385, Batch: 900, Avg. Loss: 0.04171112567186355\n",
      "Epoch: 386, Batch: 100, Avg. Loss: 0.03565416932106018\n",
      "Epoch: 386, Batch: 200, Avg. Loss: 0.04460712969303131\n",
      "Epoch: 386, Batch: 300, Avg. Loss: 0.03704617388546467\n",
      "Epoch: 386, Batch: 400, Avg. Loss: 0.03605372428894043\n",
      "Epoch: 386, Batch: 500, Avg. Loss: 0.03874512143433094\n",
      "Epoch: 386, Batch: 600, Avg. Loss: 0.040200257822871205\n",
      "Epoch: 386, Batch: 700, Avg. Loss: 0.04039537861943245\n",
      "Epoch: 386, Batch: 800, Avg. Loss: 0.03662567757070065\n",
      "Epoch: 386, Batch: 900, Avg. Loss: 0.0352174036949873\n",
      "Epoch: 387, Batch: 100, Avg. Loss: 0.04129038669168949\n",
      "Epoch: 387, Batch: 200, Avg. Loss: 0.03372144192457199\n",
      "Epoch: 387, Batch: 300, Avg. Loss: 0.03626753687858582\n",
      "Epoch: 387, Batch: 400, Avg. Loss: 0.03652584441006183\n",
      "Epoch: 387, Batch: 500, Avg. Loss: 0.03917364507913589\n",
      "Epoch: 387, Batch: 600, Avg. Loss: 0.03911464586853981\n",
      "Epoch: 387, Batch: 700, Avg. Loss: 0.03924553528428078\n",
      "Epoch: 387, Batch: 800, Avg. Loss: 0.03769698478281498\n",
      "Epoch: 387, Batch: 900, Avg. Loss: 0.041849232465028766\n",
      "Epoch: 388, Batch: 100, Avg. Loss: 0.028957418352365493\n",
      "Epoch: 388, Batch: 200, Avg. Loss: 0.04035658322274685\n",
      "Epoch: 388, Batch: 300, Avg. Loss: 0.03958594910800457\n",
      "Epoch: 388, Batch: 400, Avg. Loss: 0.03270571686327457\n",
      "Epoch: 388, Batch: 500, Avg. Loss: 0.040809810981154444\n",
      "Epoch: 388, Batch: 600, Avg. Loss: 0.04029334343969822\n",
      "Epoch: 388, Batch: 700, Avg. Loss: 0.036939287781715395\n",
      "Epoch: 388, Batch: 800, Avg. Loss: 0.04240737564861775\n",
      "Epoch: 388, Batch: 900, Avg. Loss: 0.03101633585989475\n",
      "Epoch: 389, Batch: 100, Avg. Loss: 0.03389816515147686\n",
      "Epoch: 389, Batch: 200, Avg. Loss: 0.030927981659770013\n",
      "Epoch: 389, Batch: 300, Avg. Loss: 0.03495614714920521\n",
      "Epoch: 389, Batch: 400, Avg. Loss: 0.036565170213580134\n",
      "Epoch: 389, Batch: 500, Avg. Loss: 0.038151938542723654\n",
      "Epoch: 389, Batch: 600, Avg. Loss: 0.03829359270632267\n",
      "Epoch: 389, Batch: 700, Avg. Loss: 0.03456459164619446\n",
      "Epoch: 389, Batch: 800, Avg. Loss: 0.04196251407265663\n",
      "Epoch: 389, Batch: 900, Avg. Loss: 0.0377563488483429\n",
      "Epoch: 390, Batch: 100, Avg. Loss: 0.029865406453609467\n",
      "Epoch: 390, Batch: 200, Avg. Loss: 0.04232201658189297\n",
      "Epoch: 390, Batch: 300, Avg. Loss: 0.03431248851120472\n",
      "Epoch: 390, Batch: 400, Avg. Loss: 0.03323179565370083\n",
      "Epoch: 390, Batch: 500, Avg. Loss: 0.04485047399997711\n",
      "Epoch: 390, Batch: 600, Avg. Loss: 0.035303368866443634\n",
      "Epoch: 390, Batch: 700, Avg. Loss: 0.030759008675813673\n",
      "Epoch: 390, Batch: 800, Avg. Loss: 0.03517572857439518\n",
      "Epoch: 390, Batch: 900, Avg. Loss: 0.0417081368714571\n",
      "Epoch: 391, Batch: 100, Avg. Loss: 0.04119819365441799\n",
      "Epoch: 391, Batch: 200, Avg. Loss: 0.038255390599370005\n",
      "Epoch: 391, Batch: 300, Avg. Loss: 0.03739616207778454\n",
      "Epoch: 391, Batch: 400, Avg. Loss: 0.03449326753616333\n",
      "Epoch: 391, Batch: 500, Avg. Loss: 0.03779693864285946\n",
      "Epoch: 391, Batch: 600, Avg. Loss: 0.03637958861887455\n",
      "Epoch: 391, Batch: 700, Avg. Loss: 0.039101465120911595\n",
      "Epoch: 391, Batch: 800, Avg. Loss: 0.038596135303378104\n",
      "Epoch: 391, Batch: 900, Avg. Loss: 0.033694780915975574\n",
      "Epoch: 392, Batch: 100, Avg. Loss: 0.04449209831655025\n",
      "Epoch: 392, Batch: 200, Avg. Loss: 0.035188935920596125\n",
      "Epoch: 392, Batch: 300, Avg. Loss: 0.03909955583512783\n",
      "Epoch: 392, Batch: 400, Avg. Loss: 0.03892589829862118\n",
      "Epoch: 392, Batch: 500, Avg. Loss: 0.04127003706991673\n",
      "Epoch: 392, Batch: 600, Avg. Loss: 0.04125965148210525\n",
      "Epoch: 392, Batch: 700, Avg. Loss: 0.03858248151838779\n",
      "Epoch: 392, Batch: 800, Avg. Loss: 0.02895036831498146\n",
      "Epoch: 392, Batch: 900, Avg. Loss: 0.03402555070817471\n",
      "Epoch: 393, Batch: 100, Avg. Loss: 0.033790353238582614\n",
      "Epoch: 393, Batch: 200, Avg. Loss: 0.0374064426869154\n",
      "Epoch: 393, Batch: 300, Avg. Loss: 0.043770841211080554\n",
      "Epoch: 393, Batch: 400, Avg. Loss: 0.03921015731990338\n",
      "Epoch: 393, Batch: 500, Avg. Loss: 0.03810826122760773\n",
      "Epoch: 393, Batch: 600, Avg. Loss: 0.038822382241487506\n",
      "Epoch: 393, Batch: 700, Avg. Loss: 0.03431291826069355\n",
      "Epoch: 393, Batch: 800, Avg. Loss: 0.03607465140521526\n",
      "Epoch: 393, Batch: 900, Avg. Loss: 0.03716706275939941\n",
      "Epoch: 394, Batch: 100, Avg. Loss: 0.041003364324569705\n",
      "Epoch: 394, Batch: 200, Avg. Loss: 0.03391398385167122\n",
      "Epoch: 394, Batch: 300, Avg. Loss: 0.03526203349232673\n",
      "Epoch: 394, Batch: 400, Avg. Loss: 0.03228594295680523\n",
      "Epoch: 394, Batch: 500, Avg. Loss: 0.03643142394721508\n",
      "Epoch: 394, Batch: 600, Avg. Loss: 0.0396563757956028\n",
      "Epoch: 394, Batch: 700, Avg. Loss: 0.03777992159128189\n",
      "Epoch: 394, Batch: 800, Avg. Loss: 0.0444006709754467\n",
      "Epoch: 394, Batch: 900, Avg. Loss: 0.0351150069385767\n",
      "Epoch: 395, Batch: 100, Avg. Loss: 0.04214277751743793\n",
      "Epoch: 395, Batch: 200, Avg. Loss: 0.032501120790839194\n",
      "Epoch: 395, Batch: 300, Avg. Loss: 0.03710404485464096\n",
      "Epoch: 395, Batch: 400, Avg. Loss: 0.036140208840370176\n",
      "Epoch: 395, Batch: 500, Avg. Loss: 0.037768732272088525\n",
      "Epoch: 395, Batch: 600, Avg. Loss: 0.03498449422419071\n",
      "Epoch: 395, Batch: 700, Avg. Loss: 0.0389368187636137\n",
      "Epoch: 395, Batch: 800, Avg. Loss: 0.040128056555986405\n",
      "Epoch: 395, Batch: 900, Avg. Loss: 0.03390981666743755\n",
      "Epoch: 396, Batch: 100, Avg. Loss: 0.03386860780417919\n",
      "Epoch: 396, Batch: 200, Avg. Loss: 0.03309770710766315\n",
      "Epoch: 396, Batch: 300, Avg. Loss: 0.032344722524285316\n",
      "Epoch: 396, Batch: 400, Avg. Loss: 0.03859934762120247\n",
      "Epoch: 396, Batch: 500, Avg. Loss: 0.039673033505678176\n",
      "Epoch: 396, Batch: 600, Avg. Loss: 0.034634905606508254\n",
      "Epoch: 396, Batch: 700, Avg. Loss: 0.030598800629377365\n",
      "Epoch: 396, Batch: 800, Avg. Loss: 0.037808562144637105\n",
      "Epoch: 396, Batch: 900, Avg. Loss: 0.04690829306840896\n",
      "Epoch: 397, Batch: 100, Avg. Loss: 0.03963819995522499\n",
      "Epoch: 397, Batch: 200, Avg. Loss: 0.03689816005527973\n",
      "Epoch: 397, Batch: 300, Avg. Loss: 0.03519094094634056\n",
      "Epoch: 397, Batch: 400, Avg. Loss: 0.035298590734601024\n",
      "Epoch: 397, Batch: 500, Avg. Loss: 0.03933296643197536\n",
      "Epoch: 397, Batch: 600, Avg. Loss: 0.03382826365530491\n",
      "Epoch: 397, Batch: 700, Avg. Loss: 0.03894121825695038\n",
      "Epoch: 397, Batch: 800, Avg. Loss: 0.036048412695527075\n",
      "Epoch: 397, Batch: 900, Avg. Loss: 0.03789950609207153\n",
      "Epoch: 398, Batch: 100, Avg. Loss: 0.03500084988772869\n",
      "Epoch: 398, Batch: 200, Avg. Loss: 0.0446200542896986\n",
      "Epoch: 398, Batch: 300, Avg. Loss: 0.03180843584239483\n",
      "Epoch: 398, Batch: 400, Avg. Loss: 0.04001449666917324\n",
      "Epoch: 398, Batch: 500, Avg. Loss: 0.03488645069301129\n",
      "Epoch: 398, Batch: 600, Avg. Loss: 0.03882843859493732\n",
      "Epoch: 398, Batch: 700, Avg. Loss: 0.03944096513092518\n",
      "Epoch: 398, Batch: 800, Avg. Loss: 0.040689761340618136\n",
      "Epoch: 398, Batch: 900, Avg. Loss: 0.029988399147987364\n",
      "Epoch: 399, Batch: 100, Avg. Loss: 0.035613501742482186\n",
      "Epoch: 399, Batch: 200, Avg. Loss: 0.032684495747089384\n",
      "Epoch: 399, Batch: 300, Avg. Loss: 0.035667847692966465\n",
      "Epoch: 399, Batch: 400, Avg. Loss: 0.03878212094306946\n",
      "Epoch: 399, Batch: 500, Avg. Loss: 0.03738234378397465\n",
      "Epoch: 399, Batch: 600, Avg. Loss: 0.03801949895918369\n",
      "Epoch: 399, Batch: 700, Avg. Loss: 0.03423926144838333\n",
      "Epoch: 399, Batch: 800, Avg. Loss: 0.03925556801259518\n",
      "Epoch: 399, Batch: 900, Avg. Loss: 0.03783909358084202\n",
      "Epoch: 400, Batch: 100, Avg. Loss: 0.03638181872665882\n",
      "Epoch: 400, Batch: 200, Avg. Loss: 0.0363678614795208\n",
      "Epoch: 400, Batch: 300, Avg. Loss: 0.036448387801647185\n",
      "Epoch: 400, Batch: 400, Avg. Loss: 0.029322003349661827\n",
      "Epoch: 400, Batch: 500, Avg. Loss: 0.045850752666592595\n",
      "Epoch: 400, Batch: 600, Avg. Loss: 0.03472543962299824\n",
      "Epoch: 400, Batch: 700, Avg. Loss: 0.036197915375232696\n",
      "Epoch: 400, Batch: 800, Avg. Loss: 0.038159783631563186\n",
      "Epoch: 400, Batch: 900, Avg. Loss: 0.04069291412830353\n",
      "Epoch: 401, Batch: 100, Avg. Loss: 0.036315012574195865\n",
      "Epoch: 401, Batch: 200, Avg. Loss: 0.038162129893898965\n",
      "Epoch: 401, Batch: 300, Avg. Loss: 0.03864987313747406\n",
      "Epoch: 401, Batch: 400, Avg. Loss: 0.03614454165101051\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 401, Batch: 500, Avg. Loss: 0.029860832318663596\n",
      "Epoch: 401, Batch: 600, Avg. Loss: 0.03752687580883503\n",
      "Epoch: 401, Batch: 700, Avg. Loss: 0.0401410236209631\n",
      "Epoch: 401, Batch: 800, Avg. Loss: 0.038925648406147956\n",
      "Epoch: 401, Batch: 900, Avg. Loss: 0.03560574807226658\n",
      "Epoch: 402, Batch: 100, Avg. Loss: 0.03671613171696663\n",
      "Epoch: 402, Batch: 200, Avg. Loss: 0.03674784742295742\n",
      "Epoch: 402, Batch: 300, Avg. Loss: 0.041908386573195455\n",
      "Epoch: 402, Batch: 400, Avg. Loss: 0.02898252584040165\n",
      "Epoch: 402, Batch: 500, Avg. Loss: 0.03761499390006065\n",
      "Epoch: 402, Batch: 600, Avg. Loss: 0.03617672421038151\n",
      "Epoch: 402, Batch: 700, Avg. Loss: 0.04226261481642723\n",
      "Epoch: 402, Batch: 800, Avg. Loss: 0.03441673181951046\n",
      "Epoch: 402, Batch: 900, Avg. Loss: 0.03586927093565464\n",
      "Epoch: 403, Batch: 100, Avg. Loss: 0.031418367587029934\n",
      "Epoch: 403, Batch: 200, Avg. Loss: 0.03768875390291214\n",
      "Epoch: 403, Batch: 300, Avg. Loss: 0.0385917205363512\n",
      "Epoch: 403, Batch: 400, Avg. Loss: 0.03860759720206261\n",
      "Epoch: 403, Batch: 500, Avg. Loss: 0.040063132643699643\n",
      "Epoch: 403, Batch: 600, Avg. Loss: 0.03831642080098391\n",
      "Epoch: 403, Batch: 700, Avg. Loss: 0.03398155927658081\n",
      "Epoch: 403, Batch: 800, Avg. Loss: 0.03464954368770123\n",
      "Epoch: 403, Batch: 900, Avg. Loss: 0.041110627725720404\n",
      "Epoch: 404, Batch: 100, Avg. Loss: 0.03880363576114178\n",
      "Epoch: 404, Batch: 200, Avg. Loss: 0.035042174234986304\n",
      "Epoch: 404, Batch: 300, Avg. Loss: 0.03625259093940258\n",
      "Epoch: 404, Batch: 400, Avg. Loss: 0.03854533098638058\n",
      "Epoch: 404, Batch: 500, Avg. Loss: 0.04117443658411503\n",
      "Epoch: 404, Batch: 600, Avg. Loss: 0.04100006483495235\n",
      "Epoch: 404, Batch: 700, Avg. Loss: 0.037160256505012514\n",
      "Epoch: 404, Batch: 800, Avg. Loss: 0.03716355971992016\n",
      "Epoch: 404, Batch: 900, Avg. Loss: 0.04028072401881218\n",
      "Epoch: 405, Batch: 100, Avg. Loss: 0.03214643284678459\n",
      "Epoch: 405, Batch: 200, Avg. Loss: 0.040027300715446475\n",
      "Epoch: 405, Batch: 300, Avg. Loss: 0.0372014582157135\n",
      "Epoch: 405, Batch: 400, Avg. Loss: 0.03610138017684221\n",
      "Epoch: 405, Batch: 500, Avg. Loss: 0.034076028540730476\n",
      "Epoch: 405, Batch: 600, Avg. Loss: 0.03377759337425232\n",
      "Epoch: 405, Batch: 700, Avg. Loss: 0.031431640610098836\n",
      "Epoch: 405, Batch: 800, Avg. Loss: 0.034347953870892525\n",
      "Epoch: 405, Batch: 900, Avg. Loss: 0.03865235112607479\n",
      "Epoch: 406, Batch: 100, Avg. Loss: 0.032385732382535934\n",
      "Epoch: 406, Batch: 200, Avg. Loss: 0.027176208272576333\n",
      "Epoch: 406, Batch: 300, Avg. Loss: 0.04107852779328823\n",
      "Epoch: 406, Batch: 400, Avg. Loss: 0.03956442266702652\n",
      "Epoch: 406, Batch: 500, Avg. Loss: 0.042882211431860925\n",
      "Epoch: 406, Batch: 600, Avg. Loss: 0.03012528456747532\n",
      "Epoch: 406, Batch: 700, Avg. Loss: 0.04235177889466286\n",
      "Epoch: 406, Batch: 800, Avg. Loss: 0.03267258234322071\n",
      "Epoch: 406, Batch: 900, Avg. Loss: 0.0382379774004221\n",
      "Epoch: 407, Batch: 100, Avg. Loss: 0.04222877934575081\n",
      "Epoch: 407, Batch: 200, Avg. Loss: 0.03448665291070938\n",
      "Epoch: 407, Batch: 300, Avg. Loss: 0.03649056471884251\n",
      "Epoch: 407, Batch: 400, Avg. Loss: 0.029899378344416618\n",
      "Epoch: 407, Batch: 500, Avg. Loss: 0.03393295831978321\n",
      "Epoch: 407, Batch: 600, Avg. Loss: 0.04007771484553814\n",
      "Epoch: 407, Batch: 700, Avg. Loss: 0.03938382238149643\n",
      "Epoch: 407, Batch: 800, Avg. Loss: 0.03626340851187706\n",
      "Epoch: 407, Batch: 900, Avg. Loss: 0.03392492219805718\n",
      "Epoch: 408, Batch: 100, Avg. Loss: 0.0336253321915865\n",
      "Epoch: 408, Batch: 200, Avg. Loss: 0.03464173465967178\n",
      "Epoch: 408, Batch: 300, Avg. Loss: 0.041208972185850144\n",
      "Epoch: 408, Batch: 400, Avg. Loss: 0.037825321033596995\n",
      "Epoch: 408, Batch: 500, Avg. Loss: 0.0421513482183218\n",
      "Epoch: 408, Batch: 600, Avg. Loss: 0.04002854719758034\n",
      "Epoch: 408, Batch: 700, Avg. Loss: 0.033861550092697146\n",
      "Epoch: 408, Batch: 800, Avg. Loss: 0.03495748430490494\n",
      "Epoch: 408, Batch: 900, Avg. Loss: 0.039397963434457776\n",
      "Epoch: 409, Batch: 100, Avg. Loss: 0.042643748596310614\n",
      "Epoch: 409, Batch: 200, Avg. Loss: 0.03378302603960037\n",
      "Epoch: 409, Batch: 300, Avg. Loss: 0.0362299582362175\n",
      "Epoch: 409, Batch: 400, Avg. Loss: 0.037925858721137046\n",
      "Epoch: 409, Batch: 500, Avg. Loss: 0.036916166618466376\n",
      "Epoch: 409, Batch: 600, Avg. Loss: 0.038983896225690845\n",
      "Epoch: 409, Batch: 700, Avg. Loss: 0.03578735426068306\n",
      "Epoch: 409, Batch: 800, Avg. Loss: 0.04638852164149285\n",
      "Epoch: 409, Batch: 900, Avg. Loss: 0.04046373799443245\n",
      "Epoch: 410, Batch: 100, Avg. Loss: 0.0349588467925787\n",
      "Epoch: 410, Batch: 200, Avg. Loss: 0.033869551420211794\n",
      "Epoch: 410, Batch: 300, Avg. Loss: 0.035796456709504126\n",
      "Epoch: 410, Batch: 400, Avg. Loss: 0.040985728725790975\n",
      "Epoch: 410, Batch: 500, Avg. Loss: 0.03829517774283886\n",
      "Epoch: 410, Batch: 600, Avg. Loss: 0.034168132990598676\n",
      "Epoch: 410, Batch: 700, Avg. Loss: 0.04553468763828278\n",
      "Epoch: 410, Batch: 800, Avg. Loss: 0.03416327685117722\n",
      "Epoch: 410, Batch: 900, Avg. Loss: 0.040622880905866625\n",
      "Epoch: 411, Batch: 100, Avg. Loss: 0.03658040784299374\n",
      "Epoch: 411, Batch: 200, Avg. Loss: 0.04008066147565842\n",
      "Epoch: 411, Batch: 300, Avg. Loss: 0.03439994648098946\n",
      "Epoch: 411, Batch: 400, Avg. Loss: 0.03703079618513584\n",
      "Epoch: 411, Batch: 500, Avg. Loss: 0.029335691034793852\n",
      "Epoch: 411, Batch: 600, Avg. Loss: 0.03722277991473675\n",
      "Epoch: 411, Batch: 700, Avg. Loss: 0.035609485656023024\n",
      "Epoch: 411, Batch: 800, Avg. Loss: 0.036471476033329966\n",
      "Epoch: 411, Batch: 900, Avg. Loss: 0.03489815793931484\n",
      "Epoch: 412, Batch: 100, Avg. Loss: 0.036018038317561146\n",
      "Epoch: 412, Batch: 200, Avg. Loss: 0.03781579341739416\n",
      "Epoch: 412, Batch: 300, Avg. Loss: 0.034184606298804286\n",
      "Epoch: 412, Batch: 400, Avg. Loss: 0.04063644394278526\n",
      "Epoch: 412, Batch: 500, Avg. Loss: 0.03244818277657032\n",
      "Epoch: 412, Batch: 600, Avg. Loss: 0.03793876104056835\n",
      "Epoch: 412, Batch: 700, Avg. Loss: 0.03802990011870861\n",
      "Epoch: 412, Batch: 800, Avg. Loss: 0.03425084628164768\n",
      "Epoch: 412, Batch: 900, Avg. Loss: 0.04013752669095993\n",
      "Epoch: 413, Batch: 100, Avg. Loss: 0.030090838819742203\n",
      "Epoch: 413, Batch: 200, Avg. Loss: 0.03831105552613735\n",
      "Epoch: 413, Batch: 300, Avg. Loss: 0.03311421476304531\n",
      "Epoch: 413, Batch: 400, Avg. Loss: 0.03369685970246792\n",
      "Epoch: 413, Batch: 500, Avg. Loss: 0.03367943972349167\n",
      "Epoch: 413, Batch: 600, Avg. Loss: 0.03680505521595478\n",
      "Epoch: 413, Batch: 700, Avg. Loss: 0.03073643200099468\n",
      "Epoch: 413, Batch: 800, Avg. Loss: 0.0453703010827303\n",
      "Epoch: 413, Batch: 900, Avg. Loss: 0.03801698513329029\n",
      "Epoch: 414, Batch: 100, Avg. Loss: 0.0355890129506588\n",
      "Epoch: 414, Batch: 200, Avg. Loss: 0.03699237927794456\n",
      "Epoch: 414, Batch: 300, Avg. Loss: 0.03310266435146332\n",
      "Epoch: 414, Batch: 400, Avg. Loss: 0.0347214363515377\n",
      "Epoch: 414, Batch: 500, Avg. Loss: 0.03819369986653328\n",
      "Epoch: 414, Batch: 600, Avg. Loss: 0.04076365858316421\n",
      "Epoch: 414, Batch: 700, Avg. Loss: 0.038155666962265966\n",
      "Epoch: 414, Batch: 800, Avg. Loss: 0.03491817906498909\n",
      "Epoch: 414, Batch: 900, Avg. Loss: 0.031063096970319747\n",
      "Epoch: 415, Batch: 100, Avg. Loss: 0.030497573912143708\n",
      "Epoch: 415, Batch: 200, Avg. Loss: 0.033562389686703685\n",
      "Epoch: 415, Batch: 300, Avg. Loss: 0.03303619384765625\n",
      "Epoch: 415, Batch: 400, Avg. Loss: 0.032383580878376964\n",
      "Epoch: 415, Batch: 500, Avg. Loss: 0.04081362567842007\n",
      "Epoch: 415, Batch: 600, Avg. Loss: 0.03495933085680008\n",
      "Epoch: 415, Batch: 700, Avg. Loss: 0.03256533652544022\n",
      "Epoch: 415, Batch: 800, Avg. Loss: 0.03302528992295265\n",
      "Epoch: 415, Batch: 900, Avg. Loss: 0.03747787579894066\n",
      "Epoch: 416, Batch: 100, Avg. Loss: 0.03177838727831841\n",
      "Epoch: 416, Batch: 200, Avg. Loss: 0.03126181125640869\n",
      "Epoch: 416, Batch: 300, Avg. Loss: 0.040140861123800275\n",
      "Epoch: 416, Batch: 400, Avg. Loss: 0.03611132606863975\n",
      "Epoch: 416, Batch: 500, Avg. Loss: 0.040395763143897055\n",
      "Epoch: 416, Batch: 600, Avg. Loss: 0.03738401211798191\n",
      "Epoch: 416, Batch: 700, Avg. Loss: 0.03468883320689201\n",
      "Epoch: 416, Batch: 800, Avg. Loss: 0.03480598457157612\n",
      "Epoch: 416, Batch: 900, Avg. Loss: 0.03986699379980564\n",
      "Epoch: 417, Batch: 100, Avg. Loss: 0.0327926056087017\n",
      "Epoch: 417, Batch: 200, Avg. Loss: 0.03517795853316784\n",
      "Epoch: 417, Batch: 300, Avg. Loss: 0.042121202424168586\n",
      "Epoch: 417, Batch: 400, Avg. Loss: 0.03630759224295616\n",
      "Epoch: 417, Batch: 500, Avg. Loss: 0.03401221632957459\n",
      "Epoch: 417, Batch: 600, Avg. Loss: 0.03823840864002705\n",
      "Epoch: 417, Batch: 700, Avg. Loss: 0.03491659797728062\n",
      "Epoch: 417, Batch: 800, Avg. Loss: 0.03803555004298687\n",
      "Epoch: 417, Batch: 900, Avg. Loss: 0.03790898978710175\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 418, Batch: 100, Avg. Loss: 0.037467639967799185\n",
      "Epoch: 418, Batch: 200, Avg. Loss: 0.02954632230103016\n",
      "Epoch: 418, Batch: 300, Avg. Loss: 0.031470222547650334\n",
      "Epoch: 418, Batch: 400, Avg. Loss: 0.03485345080494881\n",
      "Epoch: 418, Batch: 500, Avg. Loss: 0.033726009279489516\n",
      "Epoch: 418, Batch: 600, Avg. Loss: 0.03460591867566109\n",
      "Epoch: 418, Batch: 700, Avg. Loss: 0.03088112309575081\n",
      "Epoch: 418, Batch: 800, Avg. Loss: 0.03531106859445572\n",
      "Epoch: 418, Batch: 900, Avg. Loss: 0.03950517326593399\n",
      "Epoch: 419, Batch: 100, Avg. Loss: 0.031816696599125865\n",
      "Epoch: 419, Batch: 200, Avg. Loss: 0.03952875182032585\n",
      "Epoch: 419, Batch: 300, Avg. Loss: 0.04339223563671112\n",
      "Epoch: 419, Batch: 400, Avg. Loss: 0.03959009848535061\n",
      "Epoch: 419, Batch: 500, Avg. Loss: 0.036813657507300375\n",
      "Epoch: 419, Batch: 600, Avg. Loss: 0.03462826535105705\n",
      "Epoch: 419, Batch: 700, Avg. Loss: 0.034365902543067935\n",
      "Epoch: 419, Batch: 800, Avg. Loss: 0.035249690525233746\n",
      "Epoch: 419, Batch: 900, Avg. Loss: 0.035518428161740306\n",
      "Epoch: 420, Batch: 100, Avg. Loss: 0.035611243546009065\n",
      "Epoch: 420, Batch: 200, Avg. Loss: 0.03431585870683193\n",
      "Epoch: 420, Batch: 300, Avg. Loss: 0.03518654510378837\n",
      "Epoch: 420, Batch: 400, Avg. Loss: 0.04089685447514057\n",
      "Epoch: 420, Batch: 500, Avg. Loss: 0.03459845893085003\n",
      "Epoch: 420, Batch: 600, Avg. Loss: 0.03718790046870708\n",
      "Epoch: 420, Batch: 700, Avg. Loss: 0.03635511219501495\n",
      "Epoch: 420, Batch: 800, Avg. Loss: 0.03497964948415756\n",
      "Epoch: 420, Batch: 900, Avg. Loss: 0.03543400138616562\n",
      "Epoch: 421, Batch: 100, Avg. Loss: 0.04087147556245327\n",
      "Epoch: 421, Batch: 200, Avg. Loss: 0.04174263063818216\n",
      "Epoch: 421, Batch: 300, Avg. Loss: 0.03241146884858608\n",
      "Epoch: 421, Batch: 400, Avg. Loss: 0.04269080527126789\n",
      "Epoch: 421, Batch: 500, Avg. Loss: 0.035804793685674664\n",
      "Epoch: 421, Batch: 600, Avg. Loss: 0.036993855983018874\n",
      "Epoch: 421, Batch: 700, Avg. Loss: 0.03857445754110813\n",
      "Epoch: 421, Batch: 800, Avg. Loss: 0.03380774311721325\n",
      "Epoch: 421, Batch: 900, Avg. Loss: 0.03011675417423248\n",
      "Epoch: 422, Batch: 100, Avg. Loss: 0.0354854029417038\n",
      "Epoch: 422, Batch: 200, Avg. Loss: 0.03972873628139496\n",
      "Epoch: 422, Batch: 300, Avg. Loss: 0.03847994402050972\n",
      "Epoch: 422, Batch: 400, Avg. Loss: 0.034783797711133956\n",
      "Epoch: 422, Batch: 500, Avg. Loss: 0.03853753216564655\n",
      "Epoch: 422, Batch: 600, Avg. Loss: 0.03618317499756813\n",
      "Epoch: 422, Batch: 700, Avg. Loss: 0.03590447209775448\n",
      "Epoch: 422, Batch: 800, Avg. Loss: 0.03033001109957695\n",
      "Epoch: 422, Batch: 900, Avg. Loss: 0.03339540496468544\n",
      "Epoch: 423, Batch: 100, Avg. Loss: 0.04381612539291382\n",
      "Epoch: 423, Batch: 200, Avg. Loss: 0.036746777966618537\n",
      "Epoch: 423, Batch: 300, Avg. Loss: 0.032887758836150166\n",
      "Epoch: 423, Batch: 400, Avg. Loss: 0.033483456149697305\n",
      "Epoch: 423, Batch: 500, Avg. Loss: 0.027611174881458284\n",
      "Epoch: 423, Batch: 600, Avg. Loss: 0.02720691740512848\n",
      "Epoch: 423, Batch: 700, Avg. Loss: 0.028117457777261733\n",
      "Epoch: 423, Batch: 800, Avg. Loss: 0.04631597883999348\n",
      "Epoch: 423, Batch: 900, Avg. Loss: 0.040948222428560256\n",
      "Epoch: 424, Batch: 100, Avg. Loss: 0.036382525861263275\n",
      "Epoch: 424, Batch: 200, Avg. Loss: 0.03447078667581081\n",
      "Epoch: 424, Batch: 300, Avg. Loss: 0.04149578370153904\n",
      "Epoch: 424, Batch: 400, Avg. Loss: 0.03619362771511078\n",
      "Epoch: 424, Batch: 500, Avg. Loss: 0.03309537164866924\n",
      "Epoch: 424, Batch: 600, Avg. Loss: 0.03902823857963085\n",
      "Epoch: 424, Batch: 700, Avg. Loss: 0.031595044806599616\n",
      "Epoch: 424, Batch: 800, Avg. Loss: 0.03939770702272653\n",
      "Epoch: 424, Batch: 900, Avg. Loss: 0.03394028030335903\n",
      "Epoch: 425, Batch: 100, Avg. Loss: 0.03439323358237743\n",
      "Epoch: 425, Batch: 200, Avg. Loss: 0.03161658424884081\n",
      "Epoch: 425, Batch: 300, Avg. Loss: 0.03085258662700653\n",
      "Epoch: 425, Batch: 400, Avg. Loss: 0.03474080860614777\n",
      "Epoch: 425, Batch: 500, Avg. Loss: 0.025423001050949096\n",
      "Epoch: 425, Batch: 600, Avg. Loss: 0.04104584217071533\n",
      "Epoch: 425, Batch: 700, Avg. Loss: 0.03405910275876522\n",
      "Epoch: 425, Batch: 800, Avg. Loss: 0.04088196128606796\n",
      "Epoch: 425, Batch: 900, Avg. Loss: 0.03961680263280869\n",
      "Epoch: 426, Batch: 100, Avg. Loss: 0.036997769549489024\n",
      "Epoch: 426, Batch: 200, Avg. Loss: 0.03545321062207222\n",
      "Epoch: 426, Batch: 300, Avg. Loss: 0.035364235192537306\n",
      "Epoch: 426, Batch: 400, Avg. Loss: 0.03609861813485622\n",
      "Epoch: 426, Batch: 500, Avg. Loss: 0.033603511452674865\n",
      "Epoch: 426, Batch: 600, Avg. Loss: 0.03663422353565693\n",
      "Epoch: 426, Batch: 700, Avg. Loss: 0.03319962166249752\n",
      "Epoch: 426, Batch: 800, Avg. Loss: 0.03092267632484436\n",
      "Epoch: 426, Batch: 900, Avg. Loss: 0.03235480219125748\n",
      "Epoch: 427, Batch: 100, Avg. Loss: 0.03976783402264118\n",
      "Epoch: 427, Batch: 200, Avg. Loss: 0.033947537019848824\n",
      "Epoch: 427, Batch: 300, Avg. Loss: 0.03675519958138466\n",
      "Epoch: 427, Batch: 400, Avg. Loss: 0.032561976686120035\n",
      "Epoch: 427, Batch: 500, Avg. Loss: 0.03667911544442177\n",
      "Epoch: 427, Batch: 600, Avg. Loss: 0.031017146930098535\n",
      "Epoch: 427, Batch: 700, Avg. Loss: 0.03265928663313389\n",
      "Epoch: 427, Batch: 800, Avg. Loss: 0.03715209119021892\n",
      "Epoch: 427, Batch: 900, Avg. Loss: 0.042919012159109114\n",
      "Epoch: 428, Batch: 100, Avg. Loss: 0.03711902156472206\n",
      "Epoch: 428, Batch: 200, Avg. Loss: 0.03401505544781685\n",
      "Epoch: 428, Batch: 300, Avg. Loss: 0.03091945931315422\n",
      "Epoch: 428, Batch: 400, Avg. Loss: 0.03634143382310867\n",
      "Epoch: 428, Batch: 500, Avg. Loss: 0.03638935551047325\n",
      "Epoch: 428, Batch: 600, Avg. Loss: 0.04424768976867199\n",
      "Epoch: 428, Batch: 700, Avg. Loss: 0.037239867821335794\n",
      "Epoch: 428, Batch: 800, Avg. Loss: 0.03130663216114044\n",
      "Epoch: 428, Batch: 900, Avg. Loss: 0.03660014308989048\n",
      "Epoch: 429, Batch: 100, Avg. Loss: 0.03343455672264099\n",
      "Epoch: 429, Batch: 200, Avg. Loss: 0.037360545843839646\n",
      "Epoch: 429, Batch: 300, Avg. Loss: 0.032613110467791556\n",
      "Epoch: 429, Batch: 400, Avg. Loss: 0.03341271951794624\n",
      "Epoch: 429, Batch: 500, Avg. Loss: 0.03800674371421337\n",
      "Epoch: 429, Batch: 600, Avg. Loss: 0.03971296258270741\n",
      "Epoch: 429, Batch: 700, Avg. Loss: 0.036854704990983006\n",
      "Epoch: 429, Batch: 800, Avg. Loss: 0.03678780794143677\n",
      "Epoch: 429, Batch: 900, Avg. Loss: 0.03396169424057007\n",
      "Epoch: 430, Batch: 100, Avg. Loss: 0.03647419884800911\n",
      "Epoch: 430, Batch: 200, Avg. Loss: 0.031788361370563505\n",
      "Epoch: 430, Batch: 300, Avg. Loss: 0.04038802847266197\n",
      "Epoch: 430, Batch: 400, Avg. Loss: 0.034888435006141666\n",
      "Epoch: 430, Batch: 500, Avg. Loss: 0.03212329372763634\n",
      "Epoch: 430, Batch: 600, Avg. Loss: 0.0332674191147089\n",
      "Epoch: 430, Batch: 700, Avg. Loss: 0.028067354559898377\n",
      "Epoch: 430, Batch: 800, Avg. Loss: 0.03536317355930805\n",
      "Epoch: 430, Batch: 900, Avg. Loss: 0.03631654739379883\n",
      "Epoch: 431, Batch: 100, Avg. Loss: 0.03163199283182621\n",
      "Epoch: 431, Batch: 200, Avg. Loss: 0.037446058094501494\n",
      "Epoch: 431, Batch: 300, Avg. Loss: 0.04304661080241203\n",
      "Epoch: 431, Batch: 400, Avg. Loss: 0.033977663666009905\n",
      "Epoch: 431, Batch: 500, Avg. Loss: 0.03698016852140427\n",
      "Epoch: 431, Batch: 600, Avg. Loss: 0.03161028504371643\n",
      "Epoch: 431, Batch: 700, Avg. Loss: 0.03473641261458397\n",
      "Epoch: 431, Batch: 800, Avg. Loss: 0.036855792179703716\n",
      "Epoch: 431, Batch: 900, Avg. Loss: 0.038442583084106446\n",
      "Epoch: 432, Batch: 100, Avg. Loss: 0.04000423099845648\n",
      "Epoch: 432, Batch: 200, Avg. Loss: 0.03512497201561928\n",
      "Epoch: 432, Batch: 300, Avg. Loss: 0.03790481232106686\n",
      "Epoch: 432, Batch: 400, Avg. Loss: 0.030967810153961182\n",
      "Epoch: 432, Batch: 500, Avg. Loss: 0.03532694876194\n",
      "Epoch: 432, Batch: 600, Avg. Loss: 0.030608040019869803\n",
      "Epoch: 432, Batch: 700, Avg. Loss: 0.040024928525090216\n",
      "Epoch: 432, Batch: 800, Avg. Loss: 0.028392557948827744\n",
      "Epoch: 432, Batch: 900, Avg. Loss: 0.036617989800870415\n",
      "Epoch: 433, Batch: 100, Avg. Loss: 0.03454511307179928\n",
      "Epoch: 433, Batch: 200, Avg. Loss: 0.03450229577720165\n",
      "Epoch: 433, Batch: 300, Avg. Loss: 0.030243069976568223\n",
      "Epoch: 433, Batch: 400, Avg. Loss: 0.03251567646861076\n",
      "Epoch: 433, Batch: 500, Avg. Loss: 0.031001949533820152\n",
      "Epoch: 433, Batch: 600, Avg. Loss: 0.037974324971437455\n",
      "Epoch: 433, Batch: 700, Avg. Loss: 0.044003769159317016\n",
      "Epoch: 433, Batch: 800, Avg. Loss: 0.0336008333414793\n",
      "Epoch: 433, Batch: 900, Avg. Loss: 0.029607177376747132\n",
      "Epoch: 434, Batch: 100, Avg. Loss: 0.03221734024584293\n",
      "Epoch: 434, Batch: 200, Avg. Loss: 0.038264558836817744\n",
      "Epoch: 434, Batch: 300, Avg. Loss: 0.03237274698913097\n",
      "Epoch: 434, Batch: 400, Avg. Loss: 0.02984572686254978\n",
      "Epoch: 434, Batch: 500, Avg. Loss: 0.028961685821413993\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 434, Batch: 600, Avg. Loss: 0.03759843215346336\n",
      "Epoch: 434, Batch: 700, Avg. Loss: 0.03162607721984387\n",
      "Epoch: 434, Batch: 800, Avg. Loss: 0.03532833851873875\n",
      "Epoch: 434, Batch: 900, Avg. Loss: 0.03521813556551933\n",
      "Epoch: 435, Batch: 100, Avg. Loss: 0.03588084004819393\n",
      "Epoch: 435, Batch: 200, Avg. Loss: 0.03407292231917381\n",
      "Epoch: 435, Batch: 300, Avg. Loss: 0.034345275685191155\n",
      "Epoch: 435, Batch: 400, Avg. Loss: 0.030963514074683188\n",
      "Epoch: 435, Batch: 500, Avg. Loss: 0.033335729092359545\n",
      "Epoch: 435, Batch: 600, Avg. Loss: 0.0398233388364315\n",
      "Epoch: 435, Batch: 700, Avg. Loss: 0.03292271040380001\n",
      "Epoch: 435, Batch: 800, Avg. Loss: 0.03896590769290924\n",
      "Epoch: 435, Batch: 900, Avg. Loss: 0.0391028168797493\n",
      "Epoch: 436, Batch: 100, Avg. Loss: 0.034499925300478934\n",
      "Epoch: 436, Batch: 200, Avg. Loss: 0.03554232977330685\n",
      "Epoch: 436, Batch: 300, Avg. Loss: 0.033940893113613126\n",
      "Epoch: 436, Batch: 400, Avg. Loss: 0.03481871049851179\n",
      "Epoch: 436, Batch: 500, Avg. Loss: 0.03745764158666134\n",
      "Epoch: 436, Batch: 600, Avg. Loss: 0.03510938107967377\n",
      "Epoch: 436, Batch: 700, Avg. Loss: 0.034984545409679414\n",
      "Epoch: 436, Batch: 800, Avg. Loss: 0.03626905471086502\n",
      "Epoch: 436, Batch: 900, Avg. Loss: 0.04016291275620461\n",
      "Epoch: 437, Batch: 100, Avg. Loss: 0.042304133623838426\n",
      "Epoch: 437, Batch: 200, Avg. Loss: 0.04195093080401421\n",
      "Epoch: 437, Batch: 300, Avg. Loss: 0.03849020607769489\n",
      "Epoch: 437, Batch: 400, Avg. Loss: 0.033070068955421444\n",
      "Epoch: 437, Batch: 500, Avg. Loss: 0.033622026816010474\n",
      "Epoch: 437, Batch: 600, Avg. Loss: 0.030906269177794456\n",
      "Epoch: 437, Batch: 700, Avg. Loss: 0.04081474594771862\n",
      "Epoch: 437, Batch: 800, Avg. Loss: 0.029419299736618994\n",
      "Epoch: 437, Batch: 900, Avg. Loss: 0.033926557004451754\n",
      "Epoch: 438, Batch: 100, Avg. Loss: 0.03627219922840595\n",
      "Epoch: 438, Batch: 200, Avg. Loss: 0.03691521920263767\n",
      "Epoch: 438, Batch: 300, Avg. Loss: 0.03184874646365642\n",
      "Epoch: 438, Batch: 400, Avg. Loss: 0.030296945571899415\n",
      "Epoch: 438, Batch: 500, Avg. Loss: 0.03342381484806538\n",
      "Epoch: 438, Batch: 600, Avg. Loss: 0.032519905492663384\n",
      "Epoch: 438, Batch: 700, Avg. Loss: 0.038040110096335414\n",
      "Epoch: 438, Batch: 800, Avg. Loss: 0.03772075362503528\n",
      "Epoch: 438, Batch: 900, Avg. Loss: 0.0379530294239521\n",
      "Epoch: 439, Batch: 100, Avg. Loss: 0.03743632152676582\n",
      "Epoch: 439, Batch: 200, Avg. Loss: 0.03946262337267399\n",
      "Epoch: 439, Batch: 300, Avg. Loss: 0.03825601398944855\n",
      "Epoch: 439, Batch: 400, Avg. Loss: 0.0414983281493187\n",
      "Epoch: 439, Batch: 500, Avg. Loss: 0.037003278359770776\n",
      "Epoch: 439, Batch: 600, Avg. Loss: 0.02506262168288231\n",
      "Epoch: 439, Batch: 700, Avg. Loss: 0.03827895700931549\n",
      "Epoch: 439, Batch: 800, Avg. Loss: 0.0335152143239975\n",
      "Epoch: 439, Batch: 900, Avg. Loss: 0.030479188710451126\n",
      "Epoch: 440, Batch: 100, Avg. Loss: 0.03709655061364174\n",
      "Epoch: 440, Batch: 200, Avg. Loss: 0.03137689858675003\n",
      "Epoch: 440, Batch: 300, Avg. Loss: 0.031792838349938395\n",
      "Epoch: 440, Batch: 400, Avg. Loss: 0.03329678170382976\n",
      "Epoch: 440, Batch: 500, Avg. Loss: 0.031685622856020924\n",
      "Epoch: 440, Batch: 600, Avg. Loss: 0.03168890364468098\n",
      "Epoch: 440, Batch: 700, Avg. Loss: 0.03931256748735905\n",
      "Epoch: 440, Batch: 800, Avg. Loss: 0.03595177762210369\n",
      "Epoch: 440, Batch: 900, Avg. Loss: 0.030545969903469087\n",
      "Epoch: 441, Batch: 100, Avg. Loss: 0.03315810807049274\n",
      "Epoch: 441, Batch: 200, Avg. Loss: 0.03553422324359417\n",
      "Epoch: 441, Batch: 300, Avg. Loss: 0.02979052931070328\n",
      "Epoch: 441, Batch: 400, Avg. Loss: 0.026692568361759185\n",
      "Epoch: 441, Batch: 500, Avg. Loss: 0.0379635863751173\n",
      "Epoch: 441, Batch: 600, Avg. Loss: 0.03781484983861447\n",
      "Epoch: 441, Batch: 700, Avg. Loss: 0.034616659730672836\n",
      "Epoch: 441, Batch: 800, Avg. Loss: 0.042367406487464905\n",
      "Epoch: 441, Batch: 900, Avg. Loss: 0.04067408911883831\n",
      "Epoch: 442, Batch: 100, Avg. Loss: 0.030816374942660332\n",
      "Epoch: 442, Batch: 200, Avg. Loss: 0.037372661530971525\n",
      "Epoch: 442, Batch: 300, Avg. Loss: 0.028465479239821435\n",
      "Epoch: 442, Batch: 400, Avg. Loss: 0.03311680614948273\n",
      "Epoch: 442, Batch: 500, Avg. Loss: 0.040023030191659925\n",
      "Epoch: 442, Batch: 600, Avg. Loss: 0.04139978140592575\n",
      "Epoch: 442, Batch: 700, Avg. Loss: 0.028858117163181304\n",
      "Epoch: 442, Batch: 800, Avg. Loss: 0.03064442180097103\n",
      "Epoch: 442, Batch: 900, Avg. Loss: 0.030339093506336214\n",
      "Epoch: 443, Batch: 100, Avg. Loss: 0.036072218269109724\n",
      "Epoch: 443, Batch: 200, Avg. Loss: 0.036601160243153574\n",
      "Epoch: 443, Batch: 300, Avg. Loss: 0.03588514044880867\n",
      "Epoch: 443, Batch: 400, Avg. Loss: 0.0336296358704567\n",
      "Epoch: 443, Batch: 500, Avg. Loss: 0.03583375483751297\n",
      "Epoch: 443, Batch: 600, Avg. Loss: 0.04037817068397999\n",
      "Epoch: 443, Batch: 700, Avg. Loss: 0.03308611594140529\n",
      "Epoch: 443, Batch: 800, Avg. Loss: 0.02936107136309147\n",
      "Epoch: 443, Batch: 900, Avg. Loss: 0.031847446858882904\n",
      "Epoch: 444, Batch: 100, Avg. Loss: 0.03137485936284065\n",
      "Epoch: 444, Batch: 200, Avg. Loss: 0.03287506818771362\n",
      "Epoch: 444, Batch: 300, Avg. Loss: 0.031015440598130228\n",
      "Epoch: 444, Batch: 400, Avg. Loss: 0.036211127638816835\n",
      "Epoch: 444, Batch: 500, Avg. Loss: 0.03908550217747688\n",
      "Epoch: 444, Batch: 600, Avg. Loss: 0.029277003183960915\n",
      "Epoch: 444, Batch: 700, Avg. Loss: 0.0360244307667017\n",
      "Epoch: 444, Batch: 800, Avg. Loss: 0.03351702652871609\n",
      "Epoch: 444, Batch: 900, Avg. Loss: 0.03471410065889358\n",
      "Epoch: 445, Batch: 100, Avg. Loss: 0.03615262165665627\n",
      "Epoch: 445, Batch: 200, Avg. Loss: 0.03836341843008995\n",
      "Epoch: 445, Batch: 300, Avg. Loss: 0.036540194004774096\n",
      "Epoch: 445, Batch: 400, Avg. Loss: 0.03591185353696346\n",
      "Epoch: 445, Batch: 500, Avg. Loss: 0.03465680696070194\n",
      "Epoch: 445, Batch: 600, Avg. Loss: 0.033218804448843\n",
      "Epoch: 445, Batch: 700, Avg. Loss: 0.02982311651110649\n",
      "Epoch: 445, Batch: 800, Avg. Loss: 0.03305126458406448\n",
      "Epoch: 445, Batch: 900, Avg. Loss: 0.03819029092788696\n",
      "Epoch: 446, Batch: 100, Avg. Loss: 0.0338224521279335\n",
      "Epoch: 446, Batch: 200, Avg. Loss: 0.030635650604963302\n",
      "Epoch: 446, Batch: 300, Avg. Loss: 0.0354617927968502\n",
      "Epoch: 446, Batch: 400, Avg. Loss: 0.03404110305011272\n",
      "Epoch: 446, Batch: 500, Avg. Loss: 0.03316533721983433\n",
      "Epoch: 446, Batch: 600, Avg. Loss: 0.03711581692099571\n",
      "Epoch: 446, Batch: 700, Avg. Loss: 0.03443887315690517\n",
      "Epoch: 446, Batch: 800, Avg. Loss: 0.039774895459413526\n",
      "Epoch: 446, Batch: 900, Avg. Loss: 0.03432388529181481\n",
      "Epoch: 447, Batch: 100, Avg. Loss: 0.03311900027096271\n",
      "Epoch: 447, Batch: 200, Avg. Loss: 0.03783395417034626\n",
      "Epoch: 447, Batch: 300, Avg. Loss: 0.03669456779956817\n",
      "Epoch: 447, Batch: 400, Avg. Loss: 0.03358533784747124\n",
      "Epoch: 447, Batch: 500, Avg. Loss: 0.03722304686903954\n",
      "Epoch: 447, Batch: 600, Avg. Loss: 0.02941657990217209\n",
      "Epoch: 447, Batch: 700, Avg. Loss: 0.03591091550886631\n",
      "Epoch: 447, Batch: 800, Avg. Loss: 0.034659990444779396\n",
      "Epoch: 447, Batch: 900, Avg. Loss: 0.03375882267951966\n",
      "Epoch: 448, Batch: 100, Avg. Loss: 0.037894132733345035\n",
      "Epoch: 448, Batch: 200, Avg. Loss: 0.036884899884462356\n",
      "Epoch: 448, Batch: 300, Avg. Loss: 0.03036181777715683\n",
      "Epoch: 448, Batch: 400, Avg. Loss: 0.029825686737895013\n",
      "Epoch: 448, Batch: 500, Avg. Loss: 0.03683479629456997\n",
      "Epoch: 448, Batch: 600, Avg. Loss: 0.03037092387676239\n",
      "Epoch: 448, Batch: 700, Avg. Loss: 0.029329621121287345\n",
      "Epoch: 448, Batch: 800, Avg. Loss: 0.033428808972239495\n",
      "Epoch: 448, Batch: 900, Avg. Loss: 0.030999903678894044\n",
      "Epoch: 449, Batch: 100, Avg. Loss: 0.03176100052893162\n",
      "Epoch: 449, Batch: 200, Avg. Loss: 0.03251493372023106\n",
      "Epoch: 449, Batch: 300, Avg. Loss: 0.03514077097177506\n",
      "Epoch: 449, Batch: 400, Avg. Loss: 0.03552250497043133\n",
      "Epoch: 449, Batch: 500, Avg. Loss: 0.030499203056097032\n",
      "Epoch: 449, Batch: 600, Avg. Loss: 0.03461513824760914\n",
      "Epoch: 449, Batch: 700, Avg. Loss: 0.03117863304913044\n",
      "Epoch: 449, Batch: 800, Avg. Loss: 0.03664221808314323\n",
      "Epoch: 449, Batch: 900, Avg. Loss: 0.04017404153943062\n",
      "Epoch: 450, Batch: 100, Avg. Loss: 0.03409838616847992\n",
      "Epoch: 450, Batch: 200, Avg. Loss: 0.033632595613598826\n",
      "Epoch: 450, Batch: 300, Avg. Loss: 0.03500872381031513\n",
      "Epoch: 450, Batch: 400, Avg. Loss: 0.03594200648367405\n",
      "Epoch: 450, Batch: 500, Avg. Loss: 0.031000891700387\n",
      "Epoch: 450, Batch: 600, Avg. Loss: 0.03472816027700901\n",
      "Epoch: 450, Batch: 700, Avg. Loss: 0.031252841353416445\n",
      "Epoch: 450, Batch: 800, Avg. Loss: 0.031268469393253326\n",
      "Epoch: 450, Batch: 900, Avg. Loss: 0.031115912944078446\n",
      "Epoch: 451, Batch: 100, Avg. Loss: 0.03967563517391682\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 451, Batch: 200, Avg. Loss: 0.030187827050685883\n",
      "Epoch: 451, Batch: 300, Avg. Loss: 0.030296858102083206\n",
      "Epoch: 451, Batch: 400, Avg. Loss: 0.033452105671167374\n",
      "Epoch: 451, Batch: 500, Avg. Loss: 0.029868753999471663\n",
      "Epoch: 451, Batch: 600, Avg. Loss: 0.030297781452536584\n",
      "Epoch: 451, Batch: 700, Avg. Loss: 0.035859079509973524\n",
      "Epoch: 451, Batch: 800, Avg. Loss: 0.03670395269989967\n",
      "Epoch: 451, Batch: 900, Avg. Loss: 0.043332158401608464\n",
      "Epoch: 452, Batch: 100, Avg. Loss: 0.03374995902180672\n",
      "Epoch: 452, Batch: 200, Avg. Loss: 0.03728009760379791\n",
      "Epoch: 452, Batch: 300, Avg. Loss: 0.029784632250666618\n",
      "Epoch: 452, Batch: 400, Avg. Loss: 0.037464526370167735\n",
      "Epoch: 452, Batch: 500, Avg. Loss: 0.031911644265055654\n",
      "Epoch: 452, Batch: 600, Avg. Loss: 0.028487772718071937\n",
      "Epoch: 452, Batch: 700, Avg. Loss: 0.03786466859281063\n",
      "Epoch: 452, Batch: 800, Avg. Loss: 0.03432586677372455\n",
      "Epoch: 452, Batch: 900, Avg. Loss: 0.03236649356782437\n",
      "Epoch: 453, Batch: 100, Avg. Loss: 0.03496130093932152\n",
      "Epoch: 453, Batch: 200, Avg. Loss: 0.04015329994261265\n",
      "Epoch: 453, Batch: 300, Avg. Loss: 0.031835912093520165\n",
      "Epoch: 453, Batch: 400, Avg. Loss: 0.03288859628140926\n",
      "Epoch: 453, Batch: 500, Avg. Loss: 0.031481636092066764\n",
      "Epoch: 453, Batch: 600, Avg. Loss: 0.036034936383366585\n",
      "Epoch: 453, Batch: 700, Avg. Loss: 0.04215722642838955\n",
      "Epoch: 453, Batch: 800, Avg. Loss: 0.03614026255905628\n",
      "Epoch: 453, Batch: 900, Avg. Loss: 0.02958928607404232\n",
      "Epoch: 454, Batch: 100, Avg. Loss: 0.037067656740546226\n",
      "Epoch: 454, Batch: 200, Avg. Loss: 0.030022511184215547\n",
      "Epoch: 454, Batch: 300, Avg. Loss: 0.034813925176858905\n",
      "Epoch: 454, Batch: 400, Avg. Loss: 0.03205900520086288\n",
      "Epoch: 454, Batch: 500, Avg. Loss: 0.03354030571877956\n",
      "Epoch: 454, Batch: 600, Avg. Loss: 0.0316000346839428\n",
      "Epoch: 454, Batch: 700, Avg. Loss: 0.035889556258916856\n",
      "Epoch: 454, Batch: 800, Avg. Loss: 0.035318219810724256\n",
      "Epoch: 454, Batch: 900, Avg. Loss: 0.032995970621705055\n",
      "Epoch: 455, Batch: 100, Avg. Loss: 0.031594738736748694\n",
      "Epoch: 455, Batch: 200, Avg. Loss: 0.02592642366886139\n",
      "Epoch: 455, Batch: 300, Avg. Loss: 0.03390548326075077\n",
      "Epoch: 455, Batch: 400, Avg. Loss: 0.03732816070318222\n",
      "Epoch: 455, Batch: 500, Avg. Loss: 0.03007726989686489\n",
      "Epoch: 455, Batch: 600, Avg. Loss: 0.03321678064763546\n",
      "Epoch: 455, Batch: 700, Avg. Loss: 0.03683647610247135\n",
      "Epoch: 455, Batch: 800, Avg. Loss: 0.039180717840790746\n",
      "Epoch: 455, Batch: 900, Avg. Loss: 0.0324434507638216\n",
      "Epoch: 456, Batch: 100, Avg. Loss: 0.037149217054247854\n",
      "Epoch: 456, Batch: 200, Avg. Loss: 0.02722857542335987\n",
      "Epoch: 456, Batch: 300, Avg. Loss: 0.034338629692792895\n",
      "Epoch: 456, Batch: 400, Avg. Loss: 0.028720259815454483\n",
      "Epoch: 456, Batch: 500, Avg. Loss: 0.03886097528040409\n",
      "Epoch: 456, Batch: 600, Avg. Loss: 0.0399422363191843\n",
      "Epoch: 456, Batch: 700, Avg. Loss: 0.029920910522341727\n",
      "Epoch: 456, Batch: 800, Avg. Loss: 0.03849089987576008\n",
      "Epoch: 456, Batch: 900, Avg. Loss: 0.0312495743483305\n",
      "Epoch: 457, Batch: 100, Avg. Loss: 0.032173752561211584\n",
      "Epoch: 457, Batch: 200, Avg. Loss: 0.036900857537984846\n",
      "Epoch: 457, Batch: 300, Avg. Loss: 0.036298836097121236\n",
      "Epoch: 457, Batch: 400, Avg. Loss: 0.038466624766588214\n",
      "Epoch: 457, Batch: 500, Avg. Loss: 0.03308902092278004\n",
      "Epoch: 457, Batch: 600, Avg. Loss: 0.04414551742374897\n",
      "Epoch: 457, Batch: 700, Avg. Loss: 0.03177410028874874\n",
      "Epoch: 457, Batch: 800, Avg. Loss: 0.029205556437373163\n",
      "Epoch: 457, Batch: 900, Avg. Loss: 0.029568818435072898\n",
      "Epoch: 458, Batch: 100, Avg. Loss: 0.035283460766077045\n",
      "Epoch: 458, Batch: 200, Avg. Loss: 0.03000144995748997\n",
      "Epoch: 458, Batch: 300, Avg. Loss: 0.030681999400258064\n",
      "Epoch: 458, Batch: 400, Avg. Loss: 0.030080616399645807\n",
      "Epoch: 458, Batch: 500, Avg. Loss: 0.029518861770629883\n",
      "Epoch: 458, Batch: 600, Avg. Loss: 0.03373728200793266\n",
      "Epoch: 458, Batch: 700, Avg. Loss: 0.03305752210319042\n",
      "Epoch: 458, Batch: 800, Avg. Loss: 0.03883510552346706\n",
      "Epoch: 458, Batch: 900, Avg. Loss: 0.03564500324428081\n",
      "Epoch: 459, Batch: 100, Avg. Loss: 0.0426420347392559\n",
      "Epoch: 459, Batch: 200, Avg. Loss: 0.02982386074960232\n",
      "Epoch: 459, Batch: 300, Avg. Loss: 0.029948551282286642\n",
      "Epoch: 459, Batch: 400, Avg. Loss: 0.03668322332203388\n",
      "Epoch: 459, Batch: 500, Avg. Loss: 0.03321328483521938\n",
      "Epoch: 459, Batch: 600, Avg. Loss: 0.030365545600652695\n",
      "Epoch: 459, Batch: 700, Avg. Loss: 0.03668258644640446\n",
      "Epoch: 459, Batch: 800, Avg. Loss: 0.03171138018369675\n",
      "Epoch: 459, Batch: 900, Avg. Loss: 0.029979907497763632\n",
      "Epoch: 460, Batch: 100, Avg. Loss: 0.03506784476339817\n",
      "Epoch: 460, Batch: 200, Avg. Loss: 0.0364983044564724\n",
      "Epoch: 460, Batch: 300, Avg. Loss: 0.03709889993071556\n",
      "Epoch: 460, Batch: 400, Avg. Loss: 0.039519877806305885\n",
      "Epoch: 460, Batch: 500, Avg. Loss: 0.0310627181828022\n",
      "Epoch: 460, Batch: 600, Avg. Loss: 0.03479216657578945\n",
      "Epoch: 460, Batch: 700, Avg. Loss: 0.036820183023810384\n",
      "Epoch: 460, Batch: 800, Avg. Loss: 0.031649457961320876\n",
      "Epoch: 460, Batch: 900, Avg. Loss: 0.032573940828442574\n",
      "Epoch: 461, Batch: 100, Avg. Loss: 0.03424338474869728\n",
      "Epoch: 461, Batch: 200, Avg. Loss: 0.028837836608290673\n",
      "Epoch: 461, Batch: 300, Avg. Loss: 0.030263228863477706\n",
      "Epoch: 461, Batch: 400, Avg. Loss: 0.036476344913244245\n",
      "Epoch: 461, Batch: 500, Avg. Loss: 0.03542318612337112\n",
      "Epoch: 461, Batch: 600, Avg. Loss: 0.0300674469769001\n",
      "Epoch: 461, Batch: 700, Avg. Loss: 0.03034315213561058\n",
      "Epoch: 461, Batch: 800, Avg. Loss: 0.04188450783491135\n",
      "Epoch: 461, Batch: 900, Avg. Loss: 0.0339741849899292\n",
      "Epoch: 462, Batch: 100, Avg. Loss: 0.03329972162842751\n",
      "Epoch: 462, Batch: 200, Avg. Loss: 0.0334481056779623\n",
      "Epoch: 462, Batch: 300, Avg. Loss: 0.02970760904252529\n",
      "Epoch: 462, Batch: 400, Avg. Loss: 0.02949443891644478\n",
      "Epoch: 462, Batch: 500, Avg. Loss: 0.03582882076501846\n",
      "Epoch: 462, Batch: 600, Avg. Loss: 0.03656612820923329\n",
      "Epoch: 462, Batch: 700, Avg. Loss: 0.031030440926551817\n",
      "Epoch: 462, Batch: 800, Avg. Loss: 0.04019300565123558\n",
      "Epoch: 462, Batch: 900, Avg. Loss: 0.028831783831119537\n",
      "Epoch: 463, Batch: 100, Avg. Loss: 0.03317043147981167\n",
      "Epoch: 463, Batch: 200, Avg. Loss: 0.03704508602619171\n",
      "Epoch: 463, Batch: 300, Avg. Loss: 0.032170208320021626\n",
      "Epoch: 463, Batch: 400, Avg. Loss: 0.03461186312139034\n",
      "Epoch: 463, Batch: 500, Avg. Loss: 0.028371051102876663\n",
      "Epoch: 463, Batch: 600, Avg. Loss: 0.03721496358513832\n",
      "Epoch: 463, Batch: 700, Avg. Loss: 0.0344545129686594\n",
      "Epoch: 463, Batch: 800, Avg. Loss: 0.033331936374306675\n",
      "Epoch: 463, Batch: 900, Avg. Loss: 0.03291933611035347\n",
      "Epoch: 464, Batch: 100, Avg. Loss: 0.03711646497249603\n",
      "Epoch: 464, Batch: 200, Avg. Loss: 0.03314696110785007\n",
      "Epoch: 464, Batch: 300, Avg. Loss: 0.0299737960845232\n",
      "Epoch: 464, Batch: 400, Avg. Loss: 0.03002537101507187\n",
      "Epoch: 464, Batch: 500, Avg. Loss: 0.03027550332248211\n",
      "Epoch: 464, Batch: 600, Avg. Loss: 0.035495412424206736\n",
      "Epoch: 464, Batch: 700, Avg. Loss: 0.02996915824711323\n",
      "Epoch: 464, Batch: 800, Avg. Loss: 0.036175726652145385\n",
      "Epoch: 464, Batch: 900, Avg. Loss: 0.04101828292012215\n",
      "Epoch: 465, Batch: 100, Avg. Loss: 0.035614788606762884\n",
      "Epoch: 465, Batch: 200, Avg. Loss: 0.036720897182822225\n",
      "Epoch: 465, Batch: 300, Avg. Loss: 0.04179443962872029\n",
      "Epoch: 465, Batch: 400, Avg. Loss: 0.031520538479089734\n",
      "Epoch: 465, Batch: 500, Avg. Loss: 0.037790853157639505\n",
      "Epoch: 465, Batch: 600, Avg. Loss: 0.03069341242313385\n",
      "Epoch: 465, Batch: 700, Avg. Loss: 0.03527180179953575\n",
      "Epoch: 465, Batch: 800, Avg. Loss: 0.03517067052423954\n",
      "Epoch: 465, Batch: 900, Avg. Loss: 0.032106444016098974\n",
      "Epoch: 466, Batch: 100, Avg. Loss: 0.03281439498066902\n",
      "Epoch: 466, Batch: 200, Avg. Loss: 0.034835171103477475\n",
      "Epoch: 466, Batch: 300, Avg. Loss: 0.03743403315544128\n",
      "Epoch: 466, Batch: 400, Avg. Loss: 0.03659186124801636\n",
      "Epoch: 466, Batch: 500, Avg. Loss: 0.03677967607975006\n",
      "Epoch: 466, Batch: 600, Avg. Loss: 0.031176807060837745\n",
      "Epoch: 466, Batch: 700, Avg. Loss: 0.03436242453753948\n",
      "Epoch: 466, Batch: 800, Avg. Loss: 0.025705256164073945\n",
      "Epoch: 466, Batch: 900, Avg. Loss: 0.0326853597164154\n",
      "Epoch: 467, Batch: 100, Avg. Loss: 0.03426372185349465\n",
      "Epoch: 467, Batch: 200, Avg. Loss: 0.03700468719005585\n",
      "Epoch: 467, Batch: 300, Avg. Loss: 0.02851747438311577\n",
      "Epoch: 467, Batch: 400, Avg. Loss: 0.030595663264393808\n",
      "Epoch: 467, Batch: 500, Avg. Loss: 0.02711513489484787\n",
      "Epoch: 467, Batch: 600, Avg. Loss: 0.03658491484820843\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 467, Batch: 700, Avg. Loss: 0.03370600521564484\n",
      "Epoch: 467, Batch: 800, Avg. Loss: 0.03228825248777866\n",
      "Epoch: 467, Batch: 900, Avg. Loss: 0.029989357888698578\n",
      "Epoch: 468, Batch: 100, Avg. Loss: 0.02666848063468933\n",
      "Epoch: 468, Batch: 200, Avg. Loss: 0.03559501357376575\n",
      "Epoch: 468, Batch: 300, Avg. Loss: 0.03535238526761532\n",
      "Epoch: 468, Batch: 400, Avg. Loss: 0.044463379755616185\n",
      "Epoch: 468, Batch: 500, Avg. Loss: 0.03779508464038372\n",
      "Epoch: 468, Batch: 600, Avg. Loss: 0.02711050823330879\n",
      "Epoch: 468, Batch: 700, Avg. Loss: 0.0357121492177248\n",
      "Epoch: 468, Batch: 800, Avg. Loss: 0.034538553208112714\n",
      "Epoch: 468, Batch: 900, Avg. Loss: 0.03070373460650444\n",
      "Epoch: 469, Batch: 100, Avg. Loss: 0.030614581778645514\n",
      "Epoch: 469, Batch: 200, Avg. Loss: 0.0334002885594964\n",
      "Epoch: 469, Batch: 300, Avg. Loss: 0.036535964384675025\n",
      "Epoch: 469, Batch: 400, Avg. Loss: 0.033521135300397874\n",
      "Epoch: 469, Batch: 500, Avg. Loss: 0.0327301599085331\n",
      "Epoch: 469, Batch: 600, Avg. Loss: 0.03990406945347786\n",
      "Epoch: 469, Batch: 700, Avg. Loss: 0.03513751849532128\n",
      "Epoch: 469, Batch: 800, Avg. Loss: 0.025622823908925056\n",
      "Epoch: 469, Batch: 900, Avg. Loss: 0.03975366670638323\n",
      "Epoch: 470, Batch: 100, Avg. Loss: 0.0339975032582879\n",
      "Epoch: 470, Batch: 200, Avg. Loss: 0.03332554221153259\n",
      "Epoch: 470, Batch: 300, Avg. Loss: 0.03230839446187019\n",
      "Epoch: 470, Batch: 400, Avg. Loss: 0.03158503994345665\n",
      "Epoch: 470, Batch: 500, Avg. Loss: 0.03019206166267395\n",
      "Epoch: 470, Batch: 600, Avg. Loss: 0.03630982927978039\n",
      "Epoch: 470, Batch: 700, Avg. Loss: 0.04153025791049004\n",
      "Epoch: 470, Batch: 800, Avg. Loss: 0.027001625597476958\n",
      "Epoch: 470, Batch: 900, Avg. Loss: 0.0323780795186758\n",
      "Epoch: 471, Batch: 100, Avg. Loss: 0.036967485323548314\n",
      "Epoch: 471, Batch: 200, Avg. Loss: 0.03945488221943379\n",
      "Epoch: 471, Batch: 300, Avg. Loss: 0.02906462624669075\n",
      "Epoch: 471, Batch: 400, Avg. Loss: 0.03741492591798305\n",
      "Epoch: 471, Batch: 500, Avg. Loss: 0.03060591571033001\n",
      "Epoch: 471, Batch: 600, Avg. Loss: 0.03073519840836525\n",
      "Epoch: 471, Batch: 700, Avg. Loss: 0.03536595702171326\n",
      "Epoch: 471, Batch: 800, Avg. Loss: 0.0371679025888443\n",
      "Epoch: 471, Batch: 900, Avg. Loss: 0.031567615568637845\n",
      "Epoch: 472, Batch: 100, Avg. Loss: 0.03331861466169357\n",
      "Epoch: 472, Batch: 200, Avg. Loss: 0.02847864530980587\n",
      "Epoch: 472, Batch: 300, Avg. Loss: 0.03874398276209831\n",
      "Epoch: 472, Batch: 400, Avg. Loss: 0.03089477613568306\n",
      "Epoch: 472, Batch: 500, Avg. Loss: 0.028778409585356712\n",
      "Epoch: 472, Batch: 600, Avg. Loss: 0.028866681382060052\n",
      "Epoch: 472, Batch: 700, Avg. Loss: 0.03653801493346691\n",
      "Epoch: 472, Batch: 800, Avg. Loss: 0.03172256596386433\n",
      "Epoch: 472, Batch: 900, Avg. Loss: 0.033377030193805696\n",
      "Epoch: 473, Batch: 100, Avg. Loss: 0.03381426692008972\n",
      "Epoch: 473, Batch: 200, Avg. Loss: 0.032092871740460396\n",
      "Epoch: 473, Batch: 300, Avg. Loss: 0.033133328929543494\n",
      "Epoch: 473, Batch: 400, Avg. Loss: 0.034891752153635026\n",
      "Epoch: 473, Batch: 500, Avg. Loss: 0.03032576188445091\n",
      "Epoch: 473, Batch: 600, Avg. Loss: 0.03418102227151394\n",
      "Epoch: 473, Batch: 700, Avg. Loss: 0.03587566278874874\n",
      "Epoch: 473, Batch: 800, Avg. Loss: 0.03530541159212589\n",
      "Epoch: 473, Batch: 900, Avg. Loss: 0.02710720643401146\n",
      "Epoch: 474, Batch: 100, Avg. Loss: 0.03921303994953632\n",
      "Epoch: 474, Batch: 200, Avg. Loss: 0.03279894553124905\n",
      "Epoch: 474, Batch: 300, Avg. Loss: 0.03850741796195507\n",
      "Epoch: 474, Batch: 400, Avg. Loss: 0.03207393243908882\n",
      "Epoch: 474, Batch: 500, Avg. Loss: 0.03361601784825325\n",
      "Epoch: 474, Batch: 600, Avg. Loss: 0.034927355945110324\n",
      "Epoch: 474, Batch: 700, Avg. Loss: 0.04011491261422634\n",
      "Epoch: 474, Batch: 800, Avg. Loss: 0.03194964587688446\n",
      "Epoch: 474, Batch: 900, Avg. Loss: 0.032743251621723174\n",
      "Epoch: 475, Batch: 100, Avg. Loss: 0.031109063550829888\n",
      "Epoch: 475, Batch: 200, Avg. Loss: 0.03733755543828011\n",
      "Epoch: 475, Batch: 300, Avg. Loss: 0.03407733343541622\n",
      "Epoch: 475, Batch: 400, Avg. Loss: 0.02943958207964897\n",
      "Epoch: 475, Batch: 500, Avg. Loss: 0.03045119598507881\n",
      "Epoch: 475, Batch: 600, Avg. Loss: 0.030759120881557463\n",
      "Epoch: 475, Batch: 700, Avg. Loss: 0.04003354728221893\n",
      "Epoch: 475, Batch: 800, Avg. Loss: 0.03653278589248657\n",
      "Epoch: 475, Batch: 900, Avg. Loss: 0.03119903400540352\n",
      "Epoch: 476, Batch: 100, Avg. Loss: 0.02967467524111271\n",
      "Epoch: 476, Batch: 200, Avg. Loss: 0.033614401891827586\n",
      "Epoch: 476, Batch: 300, Avg. Loss: 0.030743564516305923\n",
      "Epoch: 476, Batch: 400, Avg. Loss: 0.03407256320118904\n",
      "Epoch: 476, Batch: 500, Avg. Loss: 0.029755381941795347\n",
      "Epoch: 476, Batch: 600, Avg. Loss: 0.032202005088329315\n",
      "Epoch: 476, Batch: 700, Avg. Loss: 0.0372726059705019\n",
      "Epoch: 476, Batch: 800, Avg. Loss: 0.03373475596308708\n",
      "Epoch: 476, Batch: 900, Avg. Loss: 0.0333703825622797\n",
      "Epoch: 477, Batch: 100, Avg. Loss: 0.03758566305041313\n",
      "Epoch: 477, Batch: 200, Avg. Loss: 0.03752689026296139\n",
      "Epoch: 477, Batch: 300, Avg. Loss: 0.030270877331495284\n",
      "Epoch: 477, Batch: 400, Avg. Loss: 0.03675972156226635\n",
      "Epoch: 477, Batch: 500, Avg. Loss: 0.0338512160629034\n",
      "Epoch: 477, Batch: 600, Avg. Loss: 0.03863092556595802\n",
      "Epoch: 477, Batch: 700, Avg. Loss: 0.03117676466703415\n",
      "Epoch: 477, Batch: 800, Avg. Loss: 0.03233154997229576\n",
      "Epoch: 477, Batch: 900, Avg. Loss: 0.0341607990115881\n",
      "Epoch: 478, Batch: 100, Avg. Loss: 0.03684794373810291\n",
      "Epoch: 478, Batch: 200, Avg. Loss: 0.03192196719348431\n",
      "Epoch: 478, Batch: 300, Avg. Loss: 0.030605427026748656\n",
      "Epoch: 478, Batch: 400, Avg. Loss: 0.03127440012991428\n",
      "Epoch: 478, Batch: 500, Avg. Loss: 0.033339824005961416\n",
      "Epoch: 478, Batch: 600, Avg. Loss: 0.03189224451780319\n",
      "Epoch: 478, Batch: 700, Avg. Loss: 0.042119638323783876\n",
      "Epoch: 478, Batch: 800, Avg. Loss: 0.03710698306560516\n",
      "Epoch: 478, Batch: 900, Avg. Loss: 0.03418744772672653\n",
      "Epoch: 479, Batch: 100, Avg. Loss: 0.03211729489266872\n",
      "Epoch: 479, Batch: 200, Avg. Loss: 0.03184891931712627\n",
      "Epoch: 479, Batch: 300, Avg. Loss: 0.032562905177474025\n",
      "Epoch: 479, Batch: 400, Avg. Loss: 0.034227628633379935\n",
      "Epoch: 479, Batch: 500, Avg. Loss: 0.02929749436676502\n",
      "Epoch: 479, Batch: 600, Avg. Loss: 0.03918987825512886\n",
      "Epoch: 479, Batch: 700, Avg. Loss: 0.03309629015624523\n",
      "Epoch: 479, Batch: 800, Avg. Loss: 0.0330019860714674\n",
      "Epoch: 479, Batch: 900, Avg. Loss: 0.03517461553215981\n",
      "Epoch: 480, Batch: 100, Avg. Loss: 0.03293618366122246\n",
      "Epoch: 480, Batch: 200, Avg. Loss: 0.02883305720984936\n",
      "Epoch: 480, Batch: 300, Avg. Loss: 0.036083810925483704\n",
      "Epoch: 480, Batch: 400, Avg. Loss: 0.030831022635102272\n",
      "Epoch: 480, Batch: 500, Avg. Loss: 0.035298355668783185\n",
      "Epoch: 480, Batch: 600, Avg. Loss: 0.03253962017595768\n",
      "Epoch: 480, Batch: 700, Avg. Loss: 0.027956619709730148\n",
      "Epoch: 480, Batch: 800, Avg. Loss: 0.03636612117290497\n",
      "Epoch: 480, Batch: 900, Avg. Loss: 0.032945941984653476\n",
      "Epoch: 481, Batch: 100, Avg. Loss: 0.031685177609324454\n",
      "Epoch: 481, Batch: 200, Avg. Loss: 0.03564813144505024\n",
      "Epoch: 481, Batch: 300, Avg. Loss: 0.029269775673747063\n",
      "Epoch: 481, Batch: 400, Avg. Loss: 0.034723425135016445\n",
      "Epoch: 481, Batch: 500, Avg. Loss: 0.03304923415184021\n",
      "Epoch: 481, Batch: 600, Avg. Loss: 0.03727734826505184\n",
      "Epoch: 481, Batch: 700, Avg. Loss: 0.03207273200154304\n",
      "Epoch: 481, Batch: 800, Avg. Loss: 0.03027120515704155\n",
      "Epoch: 481, Batch: 900, Avg. Loss: 0.030455530881881715\n",
      "Epoch: 482, Batch: 100, Avg. Loss: 0.03231918565928936\n",
      "Epoch: 482, Batch: 200, Avg. Loss: 0.033823674246668815\n",
      "Epoch: 482, Batch: 300, Avg. Loss: 0.030493400916457178\n",
      "Epoch: 482, Batch: 400, Avg. Loss: 0.030020137280225755\n",
      "Epoch: 482, Batch: 500, Avg. Loss: 0.03643619857728481\n",
      "Epoch: 482, Batch: 600, Avg. Loss: 0.033967056721448896\n",
      "Epoch: 482, Batch: 700, Avg. Loss: 0.03429098054766655\n",
      "Epoch: 482, Batch: 800, Avg. Loss: 0.03491751059889794\n",
      "Epoch: 482, Batch: 900, Avg. Loss: 0.03244578763842583\n",
      "Epoch: 483, Batch: 100, Avg. Loss: 0.0324641464650631\n",
      "Epoch: 483, Batch: 200, Avg. Loss: 0.03850511878728866\n",
      "Epoch: 483, Batch: 300, Avg. Loss: 0.027471697106957436\n",
      "Epoch: 483, Batch: 400, Avg. Loss: 0.0360607423633337\n",
      "Epoch: 483, Batch: 500, Avg. Loss: 0.036987104639410975\n",
      "Epoch: 483, Batch: 600, Avg. Loss: 0.033912440687417986\n",
      "Epoch: 483, Batch: 700, Avg. Loss: 0.03460496678948402\n",
      "Epoch: 483, Batch: 800, Avg. Loss: 0.025383949279785156\n",
      "Epoch: 483, Batch: 900, Avg. Loss: 0.041764979064464566\n",
      "Epoch: 484, Batch: 100, Avg. Loss: 0.03288247235119343\n",
      "Epoch: 484, Batch: 200, Avg. Loss: 0.027540258467197418\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 484, Batch: 300, Avg. Loss: 0.03148471549153328\n",
      "Epoch: 484, Batch: 400, Avg. Loss: 0.03306518159806728\n",
      "Epoch: 484, Batch: 500, Avg. Loss: 0.03293107524514198\n",
      "Epoch: 484, Batch: 600, Avg. Loss: 0.026367773562669755\n",
      "Epoch: 484, Batch: 700, Avg. Loss: 0.030453127697110175\n",
      "Epoch: 484, Batch: 800, Avg. Loss: 0.03399967290461063\n",
      "Epoch: 484, Batch: 900, Avg. Loss: 0.032560868859291076\n",
      "Epoch: 485, Batch: 100, Avg. Loss: 0.03216876789927483\n",
      "Epoch: 485, Batch: 200, Avg. Loss: 0.041308405846357345\n",
      "Epoch: 485, Batch: 300, Avg. Loss: 0.029454485923051835\n",
      "Epoch: 485, Batch: 400, Avg. Loss: 0.02693646676838398\n",
      "Epoch: 485, Batch: 500, Avg. Loss: 0.03675421811640263\n",
      "Epoch: 485, Batch: 600, Avg. Loss: 0.03728627420961857\n",
      "Epoch: 485, Batch: 700, Avg. Loss: 0.02771132580935955\n",
      "Epoch: 485, Batch: 800, Avg. Loss: 0.03138982713222504\n",
      "Epoch: 485, Batch: 900, Avg. Loss: 0.029576789289712906\n",
      "Epoch: 486, Batch: 100, Avg. Loss: 0.03706654779613018\n",
      "Epoch: 486, Batch: 200, Avg. Loss: 0.02750933587551117\n",
      "Epoch: 486, Batch: 300, Avg. Loss: 0.027521500512957573\n",
      "Epoch: 486, Batch: 400, Avg. Loss: 0.02863748922944069\n",
      "Epoch: 486, Batch: 500, Avg. Loss: 0.029599116519093513\n",
      "Epoch: 486, Batch: 600, Avg. Loss: 0.03791645109653473\n",
      "Epoch: 486, Batch: 700, Avg. Loss: 0.03874465107917786\n",
      "Epoch: 486, Batch: 800, Avg. Loss: 0.03546362049877644\n",
      "Epoch: 486, Batch: 900, Avg. Loss: 0.026757908836007118\n",
      "Epoch: 487, Batch: 100, Avg. Loss: 0.02720307804644108\n",
      "Epoch: 487, Batch: 200, Avg. Loss: 0.039721493273973466\n",
      "Epoch: 487, Batch: 300, Avg. Loss: 0.02718920722603798\n",
      "Epoch: 487, Batch: 400, Avg. Loss: 0.028859967067837715\n",
      "Epoch: 487, Batch: 500, Avg. Loss: 0.022585323750972747\n",
      "Epoch: 487, Batch: 600, Avg. Loss: 0.038721716925501824\n",
      "Epoch: 487, Batch: 700, Avg. Loss: 0.03163380555808544\n",
      "Epoch: 487, Batch: 800, Avg. Loss: 0.03598747707903385\n",
      "Epoch: 487, Batch: 900, Avg. Loss: 0.03380918130278587\n",
      "Epoch: 488, Batch: 100, Avg. Loss: 0.03744226112961769\n",
      "Epoch: 488, Batch: 200, Avg. Loss: 0.03332863189280033\n",
      "Epoch: 488, Batch: 300, Avg. Loss: 0.032297375574707986\n",
      "Epoch: 488, Batch: 400, Avg. Loss: 0.03705339163541794\n",
      "Epoch: 488, Batch: 500, Avg. Loss: 0.03158799275755882\n",
      "Epoch: 488, Batch: 600, Avg. Loss: 0.031168902963399886\n",
      "Epoch: 488, Batch: 700, Avg. Loss: 0.03446333020925522\n",
      "Epoch: 488, Batch: 800, Avg. Loss: 0.03159078486263752\n",
      "Epoch: 488, Batch: 900, Avg. Loss: 0.033221326395869256\n",
      "Epoch: 489, Batch: 100, Avg. Loss: 0.033477318435907365\n",
      "Epoch: 489, Batch: 200, Avg. Loss: 0.035103209018707275\n",
      "Epoch: 489, Batch: 300, Avg. Loss: 0.03093423254787922\n",
      "Epoch: 489, Batch: 400, Avg. Loss: 0.024786091521382333\n",
      "Epoch: 489, Batch: 500, Avg. Loss: 0.03671266421675682\n",
      "Epoch: 489, Batch: 600, Avg. Loss: 0.03561167612671852\n",
      "Epoch: 489, Batch: 700, Avg. Loss: 0.031188918948173522\n",
      "Epoch: 489, Batch: 800, Avg. Loss: 0.034417523331940175\n",
      "Epoch: 489, Batch: 900, Avg. Loss: 0.028691871203482152\n",
      "Epoch: 490, Batch: 100, Avg. Loss: 0.038509889915585516\n",
      "Epoch: 490, Batch: 200, Avg. Loss: 0.0361959183961153\n",
      "Epoch: 490, Batch: 300, Avg. Loss: 0.02857340782880783\n",
      "Epoch: 490, Batch: 400, Avg. Loss: 0.03206390000879764\n",
      "Epoch: 490, Batch: 500, Avg. Loss: 0.03269071415066719\n",
      "Epoch: 490, Batch: 600, Avg. Loss: 0.029229820296168326\n",
      "Epoch: 490, Batch: 700, Avg. Loss: 0.031598695740103724\n",
      "Epoch: 490, Batch: 800, Avg. Loss: 0.02873162128031254\n",
      "Epoch: 490, Batch: 900, Avg. Loss: 0.0440357743948698\n",
      "Epoch: 491, Batch: 100, Avg. Loss: 0.03204293265938759\n",
      "Epoch: 491, Batch: 200, Avg. Loss: 0.03319120198488235\n",
      "Epoch: 491, Batch: 300, Avg. Loss: 0.040452817752957344\n",
      "Epoch: 491, Batch: 400, Avg. Loss: 0.04012786515057087\n",
      "Epoch: 491, Batch: 500, Avg. Loss: 0.024849628433585166\n",
      "Epoch: 491, Batch: 600, Avg. Loss: 0.03029474064707756\n",
      "Epoch: 491, Batch: 700, Avg. Loss: 0.028349572345614433\n",
      "Epoch: 491, Batch: 800, Avg. Loss: 0.03908315062522888\n",
      "Epoch: 491, Batch: 900, Avg. Loss: 0.033144617825746535\n",
      "Epoch: 492, Batch: 100, Avg. Loss: 0.03340080119669438\n",
      "Epoch: 492, Batch: 200, Avg. Loss: 0.03361541122198105\n",
      "Epoch: 492, Batch: 300, Avg. Loss: 0.03756373092532158\n",
      "Epoch: 492, Batch: 400, Avg. Loss: 0.030636507757008075\n",
      "Epoch: 492, Batch: 500, Avg. Loss: 0.03150131911039353\n",
      "Epoch: 492, Batch: 600, Avg. Loss: 0.03401425361633301\n",
      "Epoch: 492, Batch: 700, Avg. Loss: 0.03733962081372738\n",
      "Epoch: 492, Batch: 800, Avg. Loss: 0.028424646705389023\n",
      "Epoch: 492, Batch: 900, Avg. Loss: 0.03374760746955872\n",
      "Epoch: 493, Batch: 100, Avg. Loss: 0.03842291072010994\n",
      "Epoch: 493, Batch: 200, Avg. Loss: 0.030729502737522126\n",
      "Epoch: 493, Batch: 300, Avg. Loss: 0.03269294947385788\n",
      "Epoch: 493, Batch: 400, Avg. Loss: 0.03389975033700466\n",
      "Epoch: 493, Batch: 500, Avg. Loss: 0.032609320729970935\n",
      "Epoch: 493, Batch: 600, Avg. Loss: 0.028437616154551506\n",
      "Epoch: 493, Batch: 700, Avg. Loss: 0.029569712430238725\n",
      "Epoch: 493, Batch: 800, Avg. Loss: 0.03134087897837162\n",
      "Epoch: 493, Batch: 900, Avg. Loss: 0.03267661854624748\n",
      "Epoch: 494, Batch: 100, Avg. Loss: 0.03556757874786854\n",
      "Epoch: 494, Batch: 200, Avg. Loss: 0.03205146111547947\n",
      "Epoch: 494, Batch: 300, Avg. Loss: 0.029657029658555985\n",
      "Epoch: 494, Batch: 400, Avg. Loss: 0.029534935504198074\n",
      "Epoch: 494, Batch: 500, Avg. Loss: 0.03307362951338291\n",
      "Epoch: 494, Batch: 600, Avg. Loss: 0.04016672931611538\n",
      "Epoch: 494, Batch: 700, Avg. Loss: 0.03429444409906864\n",
      "Epoch: 494, Batch: 800, Avg. Loss: 0.0345526198297739\n",
      "Epoch: 494, Batch: 900, Avg. Loss: 0.032965769916772844\n",
      "Epoch: 495, Batch: 100, Avg. Loss: 0.03015605576336384\n",
      "Epoch: 495, Batch: 200, Avg. Loss: 0.035449755191802976\n",
      "Epoch: 495, Batch: 300, Avg. Loss: 0.03299947530031204\n",
      "Epoch: 495, Batch: 400, Avg. Loss: 0.028830692321062088\n",
      "Epoch: 495, Batch: 500, Avg. Loss: 0.025635980144143103\n",
      "Epoch: 495, Batch: 600, Avg. Loss: 0.028803392499685287\n",
      "Epoch: 495, Batch: 700, Avg. Loss: 0.028957537189126015\n",
      "Epoch: 495, Batch: 800, Avg. Loss: 0.025656939521431922\n",
      "Epoch: 495, Batch: 900, Avg. Loss: 0.0353762386739254\n",
      "Epoch: 496, Batch: 100, Avg. Loss: 0.023845976442098616\n",
      "Epoch: 496, Batch: 200, Avg. Loss: 0.03621980391442776\n",
      "Epoch: 496, Batch: 300, Avg. Loss: 0.031033799052238464\n",
      "Epoch: 496, Batch: 400, Avg. Loss: 0.03838593564927578\n",
      "Epoch: 496, Batch: 500, Avg. Loss: 0.0395061694830656\n",
      "Epoch: 496, Batch: 600, Avg. Loss: 0.02933982275426388\n",
      "Epoch: 496, Batch: 700, Avg. Loss: 0.03082088403403759\n",
      "Epoch: 496, Batch: 800, Avg. Loss: 0.04151125095784664\n",
      "Epoch: 496, Batch: 900, Avg. Loss: 0.029339242205023767\n",
      "Epoch: 497, Batch: 100, Avg. Loss: 0.03583661966025829\n",
      "Epoch: 497, Batch: 200, Avg. Loss: 0.029337932467460633\n",
      "Epoch: 497, Batch: 300, Avg. Loss: 0.03142209328711033\n",
      "Epoch: 497, Batch: 400, Avg. Loss: 0.03151416100561619\n",
      "Epoch: 497, Batch: 500, Avg. Loss: 0.03340281054377556\n",
      "Epoch: 497, Batch: 600, Avg. Loss: 0.034600134044885635\n",
      "Epoch: 497, Batch: 700, Avg. Loss: 0.030865618363022804\n",
      "Epoch: 497, Batch: 800, Avg. Loss: 0.031534749418497085\n",
      "Epoch: 497, Batch: 900, Avg. Loss: 0.03640516869723797\n",
      "Epoch: 498, Batch: 100, Avg. Loss: 0.03175779156386852\n",
      "Epoch: 498, Batch: 200, Avg. Loss: 0.03248060598969459\n",
      "Epoch: 498, Batch: 300, Avg. Loss: 0.029971272349357606\n",
      "Epoch: 498, Batch: 400, Avg. Loss: 0.0245314472168684\n",
      "Epoch: 498, Batch: 500, Avg. Loss: 0.037687788382172585\n",
      "Epoch: 498, Batch: 600, Avg. Loss: 0.02795522704720497\n",
      "Epoch: 498, Batch: 700, Avg. Loss: 0.03342022627592087\n",
      "Epoch: 498, Batch: 800, Avg. Loss: 0.03579775325953961\n",
      "Epoch: 498, Batch: 900, Avg. Loss: 0.02922780692577362\n",
      "Epoch: 499, Batch: 100, Avg. Loss: 0.035183440670371054\n",
      "Epoch: 499, Batch: 200, Avg. Loss: 0.03295309133827686\n",
      "Epoch: 499, Batch: 300, Avg. Loss: 0.032217738553881645\n",
      "Epoch: 499, Batch: 400, Avg. Loss: 0.026080975383520125\n",
      "Epoch: 499, Batch: 500, Avg. Loss: 0.027374173700809478\n",
      "Epoch: 499, Batch: 600, Avg. Loss: 0.03075080841779709\n",
      "Epoch: 499, Batch: 700, Avg. Loss: 0.03548664458096027\n",
      "Epoch: 499, Batch: 800, Avg. Loss: 0.033555611073970794\n",
      "Epoch: 499, Batch: 900, Avg. Loss: 0.03386339738965034\n",
      "Epoch: 500, Batch: 100, Avg. Loss: 0.03604136221110821\n",
      "Epoch: 500, Batch: 200, Avg. Loss: 0.025564977899193764\n",
      "Epoch: 500, Batch: 300, Avg. Loss: 0.026829675287008286\n",
      "Epoch: 500, Batch: 400, Avg. Loss: 0.030814188197255135\n",
      "Epoch: 500, Batch: 500, Avg. Loss: 0.03817312851548195\n",
      "Epoch: 500, Batch: 600, Avg. Loss: 0.03667913176119328\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 500, Batch: 700, Avg. Loss: 0.03421690262854099\n",
      "Epoch: 500, Batch: 800, Avg. Loss: 0.029340810626745224\n",
      "Epoch: 500, Batch: 900, Avg. Loss: 0.030830513536930084\n",
      "Epoch: 501, Batch: 100, Avg. Loss: 0.031415420174598696\n",
      "Epoch: 501, Batch: 200, Avg. Loss: 0.03542245887219906\n",
      "Epoch: 501, Batch: 300, Avg. Loss: 0.0341936132311821\n",
      "Epoch: 501, Batch: 400, Avg. Loss: 0.03653231106698513\n",
      "Epoch: 501, Batch: 500, Avg. Loss: 0.030905277654528617\n",
      "Epoch: 501, Batch: 600, Avg. Loss: 0.03205330811440945\n",
      "Epoch: 501, Batch: 700, Avg. Loss: 0.03135377675294876\n",
      "Epoch: 501, Batch: 800, Avg. Loss: 0.03169741004705429\n",
      "Epoch: 501, Batch: 900, Avg. Loss: 0.030675241202116014\n",
      "Epoch: 502, Batch: 100, Avg. Loss: 0.03247989647090435\n",
      "Epoch: 502, Batch: 200, Avg. Loss: 0.031167054772377013\n",
      "Epoch: 502, Batch: 300, Avg. Loss: 0.030530308708548545\n",
      "Epoch: 502, Batch: 400, Avg. Loss: 0.03447220340371132\n",
      "Epoch: 502, Batch: 500, Avg. Loss: 0.029895118102431298\n",
      "Epoch: 502, Batch: 600, Avg. Loss: 0.030458860993385316\n",
      "Epoch: 502, Batch: 700, Avg. Loss: 0.03407738879323006\n",
      "Epoch: 502, Batch: 800, Avg. Loss: 0.03523179374635219\n",
      "Epoch: 502, Batch: 900, Avg. Loss: 0.03182783730328083\n",
      "Epoch: 503, Batch: 100, Avg. Loss: 0.03359477050602436\n",
      "Epoch: 503, Batch: 200, Avg. Loss: 0.02584815178066492\n",
      "Epoch: 503, Batch: 300, Avg. Loss: 0.03167452529072762\n",
      "Epoch: 503, Batch: 400, Avg. Loss: 0.039323066398501395\n",
      "Epoch: 503, Batch: 500, Avg. Loss: 0.029818305224180223\n",
      "Epoch: 503, Batch: 600, Avg. Loss: 0.03095658600330353\n",
      "Epoch: 503, Batch: 700, Avg. Loss: 0.03391565769910812\n",
      "Epoch: 503, Batch: 800, Avg. Loss: 0.026554996594786644\n",
      "Epoch: 503, Batch: 900, Avg. Loss: 0.029448852017521857\n",
      "Epoch: 504, Batch: 100, Avg. Loss: 0.03268239378929138\n",
      "Epoch: 504, Batch: 200, Avg. Loss: 0.0326072196662426\n",
      "Epoch: 504, Batch: 300, Avg. Loss: 0.03172744005918503\n",
      "Epoch: 504, Batch: 400, Avg. Loss: 0.03374527171254158\n",
      "Epoch: 504, Batch: 500, Avg. Loss: 0.030990070030093193\n",
      "Epoch: 504, Batch: 600, Avg. Loss: 0.02938609056174755\n",
      "Epoch: 504, Batch: 700, Avg. Loss: 0.026238459125161172\n",
      "Epoch: 504, Batch: 800, Avg. Loss: 0.033856900930404665\n",
      "Epoch: 504, Batch: 900, Avg. Loss: 0.034416005089879034\n",
      "Epoch: 505, Batch: 100, Avg. Loss: 0.029959855154156684\n",
      "Epoch: 505, Batch: 200, Avg. Loss: 0.02886534173041582\n",
      "Epoch: 505, Batch: 300, Avg. Loss: 0.03467484824359417\n",
      "Epoch: 505, Batch: 400, Avg. Loss: 0.03203708089888096\n",
      "Epoch: 505, Batch: 500, Avg. Loss: 0.03888765409588814\n",
      "Epoch: 505, Batch: 600, Avg. Loss: 0.03691954500973225\n",
      "Epoch: 505, Batch: 700, Avg. Loss: 0.032151293829083445\n",
      "Epoch: 505, Batch: 800, Avg. Loss: 0.028327571302652357\n",
      "Epoch: 505, Batch: 900, Avg. Loss: 0.024120906740427016\n",
      "Epoch: 506, Batch: 100, Avg. Loss: 0.03178690552711487\n",
      "Epoch: 506, Batch: 200, Avg. Loss: 0.03390415176749229\n",
      "Epoch: 506, Batch: 300, Avg. Loss: 0.031589591354131696\n",
      "Epoch: 506, Batch: 400, Avg. Loss: 0.029774255678057672\n",
      "Epoch: 506, Batch: 500, Avg. Loss: 0.029852930307388306\n",
      "Epoch: 506, Batch: 600, Avg. Loss: 0.03141006827354431\n",
      "Epoch: 506, Batch: 700, Avg. Loss: 0.039532957077026365\n",
      "Epoch: 506, Batch: 800, Avg. Loss: 0.033668863251805306\n",
      "Epoch: 506, Batch: 900, Avg. Loss: 0.024148355573415756\n",
      "Epoch: 507, Batch: 100, Avg. Loss: 0.03427059203386307\n",
      "Epoch: 507, Batch: 200, Avg. Loss: 0.029835967123508455\n",
      "Epoch: 507, Batch: 300, Avg. Loss: 0.03389627382159233\n",
      "Epoch: 507, Batch: 400, Avg. Loss: 0.029987107887864115\n",
      "Epoch: 507, Batch: 500, Avg. Loss: 0.03430241122841835\n",
      "Epoch: 507, Batch: 600, Avg. Loss: 0.035687245205044744\n",
      "Epoch: 507, Batch: 700, Avg. Loss: 0.031538436263799666\n",
      "Epoch: 507, Batch: 800, Avg. Loss: 0.02767796866595745\n",
      "Epoch: 507, Batch: 900, Avg. Loss: 0.032017665728926656\n",
      "Epoch: 508, Batch: 100, Avg. Loss: 0.03277701213955879\n",
      "Epoch: 508, Batch: 200, Avg. Loss: 0.03191202685236931\n",
      "Epoch: 508, Batch: 300, Avg. Loss: 0.032878224849700925\n",
      "Epoch: 508, Batch: 400, Avg. Loss: 0.03162354178726673\n",
      "Epoch: 508, Batch: 500, Avg. Loss: 0.02503418378531933\n",
      "Epoch: 508, Batch: 600, Avg. Loss: 0.03614947959780693\n",
      "Epoch: 508, Batch: 700, Avg. Loss: 0.03019811250269413\n",
      "Epoch: 508, Batch: 800, Avg. Loss: 0.029428879097104073\n",
      "Epoch: 508, Batch: 900, Avg. Loss: 0.027906184569001197\n",
      "Epoch: 509, Batch: 100, Avg. Loss: 0.03115756928920746\n",
      "Epoch: 509, Batch: 200, Avg. Loss: 0.034676666185259816\n",
      "Epoch: 509, Batch: 300, Avg. Loss: 0.03128070957958698\n",
      "Epoch: 509, Batch: 400, Avg. Loss: 0.02933338053524494\n",
      "Epoch: 509, Batch: 500, Avg. Loss: 0.03008800610899925\n",
      "Epoch: 509, Batch: 600, Avg. Loss: 0.03270683407783508\n",
      "Epoch: 509, Batch: 700, Avg. Loss: 0.0345850083231926\n",
      "Epoch: 509, Batch: 800, Avg. Loss: 0.028533881306648256\n",
      "Epoch: 509, Batch: 900, Avg. Loss: 0.030471930652856825\n",
      "Epoch: 510, Batch: 100, Avg. Loss: 0.029693027287721635\n",
      "Epoch: 510, Batch: 200, Avg. Loss: 0.03861448742449283\n",
      "Epoch: 510, Batch: 300, Avg. Loss: 0.03002347342669964\n",
      "Epoch: 510, Batch: 400, Avg. Loss: 0.02962996244430542\n",
      "Epoch: 510, Batch: 500, Avg. Loss: 0.029861811995506287\n",
      "Epoch: 510, Batch: 600, Avg. Loss: 0.03546864539384842\n",
      "Epoch: 510, Batch: 700, Avg. Loss: 0.03444181263446808\n",
      "Epoch: 510, Batch: 800, Avg. Loss: 0.033626554533839226\n",
      "Epoch: 510, Batch: 900, Avg. Loss: 0.03660862132906914\n",
      "Epoch: 511, Batch: 100, Avg. Loss: 0.030438352450728415\n",
      "Epoch: 511, Batch: 200, Avg. Loss: 0.034407355189323426\n",
      "Epoch: 511, Batch: 300, Avg. Loss: 0.03220135100185871\n",
      "Epoch: 511, Batch: 400, Avg. Loss: 0.03788114182651043\n",
      "Epoch: 511, Batch: 500, Avg. Loss: 0.03105174109339714\n",
      "Epoch: 511, Batch: 600, Avg. Loss: 0.027260914370417597\n",
      "Epoch: 511, Batch: 700, Avg. Loss: 0.033347439840435984\n",
      "Epoch: 511, Batch: 800, Avg. Loss: 0.02585712753236294\n",
      "Epoch: 511, Batch: 900, Avg. Loss: 0.04053698979318142\n",
      "Epoch: 512, Batch: 100, Avg. Loss: 0.03432097710669041\n",
      "Epoch: 512, Batch: 200, Avg. Loss: 0.030024009123444557\n",
      "Epoch: 512, Batch: 300, Avg. Loss: 0.03299189835786819\n",
      "Epoch: 512, Batch: 400, Avg. Loss: 0.03315897300839424\n",
      "Epoch: 512, Batch: 500, Avg. Loss: 0.02900585487484932\n",
      "Epoch: 512, Batch: 600, Avg. Loss: 0.027269857823848723\n",
      "Epoch: 512, Batch: 700, Avg. Loss: 0.036581624895334244\n",
      "Epoch: 512, Batch: 800, Avg. Loss: 0.03126765996217728\n",
      "Epoch: 512, Batch: 900, Avg. Loss: 0.03816830843687057\n",
      "Epoch: 513, Batch: 100, Avg. Loss: 0.031571401953697203\n",
      "Epoch: 513, Batch: 200, Avg. Loss: 0.03385669998824596\n",
      "Epoch: 513, Batch: 300, Avg. Loss: 0.02917962945997715\n",
      "Epoch: 513, Batch: 400, Avg. Loss: 0.039952940121293065\n",
      "Epoch: 513, Batch: 500, Avg. Loss: 0.031933419704437256\n",
      "Epoch: 513, Batch: 600, Avg. Loss: 0.032925241217017176\n",
      "Epoch: 513, Batch: 700, Avg. Loss: 0.03279974021017552\n",
      "Epoch: 513, Batch: 800, Avg. Loss: 0.032956866323947905\n",
      "Epoch: 513, Batch: 900, Avg. Loss: 0.026325609534978867\n",
      "Epoch: 514, Batch: 100, Avg. Loss: 0.029838905856013297\n",
      "Epoch: 514, Batch: 200, Avg. Loss: 0.03128032322973013\n",
      "Epoch: 514, Batch: 300, Avg. Loss: 0.029467097595334054\n",
      "Epoch: 514, Batch: 400, Avg. Loss: 0.02916110187768936\n",
      "Epoch: 514, Batch: 500, Avg. Loss: 0.03246814265847206\n",
      "Epoch: 514, Batch: 600, Avg. Loss: 0.03444448471069336\n",
      "Epoch: 514, Batch: 700, Avg. Loss: 0.03369708865880966\n",
      "Epoch: 514, Batch: 800, Avg. Loss: 0.03529382519423962\n",
      "Epoch: 514, Batch: 900, Avg. Loss: 0.030765225142240525\n",
      "Epoch: 515, Batch: 100, Avg. Loss: 0.023752076849341393\n",
      "Epoch: 515, Batch: 200, Avg. Loss: 0.030885673761367798\n",
      "Epoch: 515, Batch: 300, Avg. Loss: 0.03182343430817127\n",
      "Epoch: 515, Batch: 400, Avg. Loss: 0.03426854223012924\n",
      "Epoch: 515, Batch: 500, Avg. Loss: 0.030589792132377624\n",
      "Epoch: 515, Batch: 600, Avg. Loss: 0.030552637353539466\n",
      "Epoch: 515, Batch: 700, Avg. Loss: 0.02748728893697262\n",
      "Epoch: 515, Batch: 800, Avg. Loss: 0.03299414053559303\n",
      "Epoch: 515, Batch: 900, Avg. Loss: 0.046108670830726624\n",
      "Epoch: 516, Batch: 100, Avg. Loss: 0.027257478535175322\n",
      "Epoch: 516, Batch: 200, Avg. Loss: 0.03189149141311645\n",
      "Epoch: 516, Batch: 300, Avg. Loss: 0.02842060096561909\n",
      "Epoch: 516, Batch: 400, Avg. Loss: 0.031595930233597753\n",
      "Epoch: 516, Batch: 500, Avg. Loss: 0.029141825214028357\n",
      "Epoch: 516, Batch: 600, Avg. Loss: 0.04102120451629162\n",
      "Epoch: 516, Batch: 700, Avg. Loss: 0.029183312207460402\n",
      "Epoch: 516, Batch: 800, Avg. Loss: 0.03252272464334965\n",
      "Epoch: 516, Batch: 900, Avg. Loss: 0.025937761887907983\n",
      "Epoch: 517, Batch: 100, Avg. Loss: 0.02995181255042553\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 517, Batch: 200, Avg. Loss: 0.03120243363082409\n",
      "Epoch: 517, Batch: 300, Avg. Loss: 0.026513572856783865\n",
      "Epoch: 517, Batch: 400, Avg. Loss: 0.03460786148905754\n",
      "Epoch: 517, Batch: 500, Avg. Loss: 0.029211668893694878\n",
      "Epoch: 517, Batch: 600, Avg. Loss: 0.026225324943661688\n",
      "Epoch: 517, Batch: 700, Avg. Loss: 0.03509119227528572\n",
      "Epoch: 517, Batch: 800, Avg. Loss: 0.034938308969140054\n",
      "Epoch: 517, Batch: 900, Avg. Loss: 0.03264895215630531\n",
      "Epoch: 518, Batch: 100, Avg. Loss: 0.036228020191192624\n",
      "Epoch: 518, Batch: 200, Avg. Loss: 0.032956842184066776\n",
      "Epoch: 518, Batch: 300, Avg. Loss: 0.032882844135165214\n",
      "Epoch: 518, Batch: 400, Avg. Loss: 0.03410136371850967\n",
      "Epoch: 518, Batch: 500, Avg. Loss: 0.032674304693937305\n",
      "Epoch: 518, Batch: 600, Avg. Loss: 0.027959391847252844\n",
      "Epoch: 518, Batch: 700, Avg. Loss: 0.03447682783007622\n",
      "Epoch: 518, Batch: 800, Avg. Loss: 0.028694893419742584\n",
      "Epoch: 518, Batch: 900, Avg. Loss: 0.026921391263604166\n",
      "Epoch: 519, Batch: 100, Avg. Loss: 0.025220606103539468\n",
      "Epoch: 519, Batch: 200, Avg. Loss: 0.034727085232734684\n",
      "Epoch: 519, Batch: 300, Avg. Loss: 0.032490192428231236\n",
      "Epoch: 519, Batch: 400, Avg. Loss: 0.021278325766324997\n",
      "Epoch: 519, Batch: 500, Avg. Loss: 0.036687122583389284\n",
      "Epoch: 519, Batch: 600, Avg. Loss: 0.03833575651049614\n",
      "Epoch: 519, Batch: 700, Avg. Loss: 0.027322113662958145\n",
      "Epoch: 519, Batch: 800, Avg. Loss: 0.02793960824608803\n",
      "Epoch: 519, Batch: 900, Avg. Loss: 0.034429770782589915\n",
      "Epoch: 520, Batch: 100, Avg. Loss: 0.03547953180968762\n",
      "Epoch: 520, Batch: 200, Avg. Loss: 0.02632114052772522\n",
      "Epoch: 520, Batch: 300, Avg. Loss: 0.028503937795758248\n",
      "Epoch: 520, Batch: 400, Avg. Loss: 0.02961058184504509\n",
      "Epoch: 520, Batch: 500, Avg. Loss: 0.0337092836946249\n",
      "Epoch: 520, Batch: 600, Avg. Loss: 0.032283920273184774\n",
      "Epoch: 520, Batch: 700, Avg. Loss: 0.03865226849913597\n",
      "Epoch: 520, Batch: 800, Avg. Loss: 0.028483802899718285\n",
      "Epoch: 520, Batch: 900, Avg. Loss: 0.025813065841794013\n",
      "Epoch: 521, Batch: 100, Avg. Loss: 0.03014986529946327\n",
      "Epoch: 521, Batch: 200, Avg. Loss: 0.03628901019692421\n",
      "Epoch: 521, Batch: 300, Avg. Loss: 0.028721673861145974\n",
      "Epoch: 521, Batch: 400, Avg. Loss: 0.0343190485984087\n",
      "Epoch: 521, Batch: 500, Avg. Loss: 0.03193818360567093\n",
      "Epoch: 521, Batch: 600, Avg. Loss: 0.02705093167722225\n",
      "Epoch: 521, Batch: 700, Avg. Loss: 0.029217202588915824\n",
      "Epoch: 521, Batch: 800, Avg. Loss: 0.034542222768068315\n",
      "Epoch: 521, Batch: 900, Avg. Loss: 0.0298970540612936\n",
      "Epoch: 522, Batch: 100, Avg. Loss: 0.028656964525580406\n",
      "Epoch: 522, Batch: 200, Avg. Loss: 0.03147667959332466\n",
      "Epoch: 522, Batch: 300, Avg. Loss: 0.03425741322338581\n",
      "Epoch: 522, Batch: 400, Avg. Loss: 0.038245751410722735\n",
      "Epoch: 522, Batch: 500, Avg. Loss: 0.03134864017367363\n",
      "Epoch: 522, Batch: 600, Avg. Loss: 0.030579453855752944\n",
      "Epoch: 522, Batch: 700, Avg. Loss: 0.024192879125475884\n",
      "Epoch: 522, Batch: 800, Avg. Loss: 0.034704566076397894\n",
      "Epoch: 522, Batch: 900, Avg. Loss: 0.03553728401660919\n",
      "Epoch: 523, Batch: 100, Avg. Loss: 0.033329300954937933\n",
      "Epoch: 523, Batch: 200, Avg. Loss: 0.03736768245697022\n",
      "Epoch: 523, Batch: 300, Avg. Loss: 0.029571148604154586\n",
      "Epoch: 523, Batch: 400, Avg. Loss: 0.027500999197363853\n",
      "Epoch: 523, Batch: 500, Avg. Loss: 0.029150997027754783\n",
      "Epoch: 523, Batch: 600, Avg. Loss: 0.028057311102747916\n",
      "Epoch: 523, Batch: 700, Avg. Loss: 0.030469431802630426\n",
      "Epoch: 523, Batch: 800, Avg. Loss: 0.028560263365507127\n",
      "Epoch: 523, Batch: 900, Avg. Loss: 0.030245893746614457\n",
      "Epoch: 524, Batch: 100, Avg. Loss: 0.03394655115902424\n",
      "Epoch: 524, Batch: 200, Avg. Loss: 0.03135291948914528\n",
      "Epoch: 524, Batch: 300, Avg. Loss: 0.026308239698410035\n",
      "Epoch: 524, Batch: 400, Avg. Loss: 0.034542817249894144\n",
      "Epoch: 524, Batch: 500, Avg. Loss: 0.03105523020029068\n",
      "Epoch: 524, Batch: 600, Avg. Loss: 0.03389489844441414\n",
      "Epoch: 524, Batch: 700, Avg. Loss: 0.030351965054869652\n",
      "Epoch: 524, Batch: 800, Avg. Loss: 0.0314720506221056\n",
      "Epoch: 524, Batch: 900, Avg. Loss: 0.03093200594186783\n",
      "Epoch: 525, Batch: 100, Avg. Loss: 0.02550327330827713\n",
      "Epoch: 525, Batch: 200, Avg. Loss: 0.032695158198475835\n",
      "Epoch: 525, Batch: 300, Avg. Loss: 0.027008465826511383\n",
      "Epoch: 525, Batch: 400, Avg. Loss: 0.033089949563145635\n",
      "Epoch: 525, Batch: 500, Avg. Loss: 0.030804543793201446\n",
      "Epoch: 525, Batch: 600, Avg. Loss: 0.03233057744801045\n",
      "Epoch: 525, Batch: 700, Avg. Loss: 0.03400098279118538\n",
      "Epoch: 525, Batch: 800, Avg. Loss: 0.025380750745534898\n",
      "Epoch: 525, Batch: 900, Avg. Loss: 0.027142804563045502\n",
      "Epoch: 526, Batch: 100, Avg. Loss: 0.03554653815925121\n",
      "Epoch: 526, Batch: 200, Avg. Loss: 0.027862608954310416\n",
      "Epoch: 526, Batch: 300, Avg. Loss: 0.028087793961167336\n",
      "Epoch: 526, Batch: 400, Avg. Loss: 0.028364722356200217\n",
      "Epoch: 526, Batch: 500, Avg. Loss: 0.030941902250051497\n",
      "Epoch: 526, Batch: 600, Avg. Loss: 0.030089390203356742\n",
      "Epoch: 526, Batch: 700, Avg. Loss: 0.029670579954981804\n",
      "Epoch: 526, Batch: 800, Avg. Loss: 0.031785495802760126\n",
      "Epoch: 526, Batch: 900, Avg. Loss: 0.02829745091497898\n",
      "Epoch: 527, Batch: 100, Avg. Loss: 0.027521011531352998\n",
      "Epoch: 527, Batch: 200, Avg. Loss: 0.03045420557260513\n",
      "Epoch: 527, Batch: 300, Avg. Loss: 0.03072810620069504\n",
      "Epoch: 527, Batch: 400, Avg. Loss: 0.03127283588051796\n",
      "Epoch: 527, Batch: 500, Avg. Loss: 0.031042595878243446\n",
      "Epoch: 527, Batch: 600, Avg. Loss: 0.03029761016368866\n",
      "Epoch: 527, Batch: 700, Avg. Loss: 0.02919540949165821\n",
      "Epoch: 527, Batch: 800, Avg. Loss: 0.03154283247888088\n",
      "Epoch: 527, Batch: 900, Avg. Loss: 0.03355092890560627\n",
      "Epoch: 528, Batch: 100, Avg. Loss: 0.028287529796361923\n",
      "Epoch: 528, Batch: 200, Avg. Loss: 0.02533429853618145\n",
      "Epoch: 528, Batch: 300, Avg. Loss: 0.032284599617123606\n",
      "Epoch: 528, Batch: 400, Avg. Loss: 0.03303081348538399\n",
      "Epoch: 528, Batch: 500, Avg. Loss: 0.03622037447988987\n",
      "Epoch: 528, Batch: 600, Avg. Loss: 0.031893967390060424\n",
      "Epoch: 528, Batch: 700, Avg. Loss: 0.027498157024383546\n",
      "Epoch: 528, Batch: 800, Avg. Loss: 0.027514557987451553\n",
      "Epoch: 528, Batch: 900, Avg. Loss: 0.02952722690999508\n",
      "Epoch: 529, Batch: 100, Avg. Loss: 0.02845270775258541\n",
      "Epoch: 529, Batch: 200, Avg. Loss: 0.024984274357557297\n",
      "Epoch: 529, Batch: 300, Avg. Loss: 0.028769171610474587\n",
      "Epoch: 529, Batch: 400, Avg. Loss: 0.03813952334225178\n",
      "Epoch: 529, Batch: 500, Avg. Loss: 0.03292346782982349\n",
      "Epoch: 529, Batch: 600, Avg. Loss: 0.031925424337387084\n",
      "Epoch: 529, Batch: 700, Avg. Loss: 0.031920433044433594\n",
      "Epoch: 529, Batch: 800, Avg. Loss: 0.035704002603888514\n",
      "Epoch: 529, Batch: 900, Avg. Loss: 0.03813126921653748\n",
      "Epoch: 530, Batch: 100, Avg. Loss: 0.027615570574998857\n",
      "Epoch: 530, Batch: 200, Avg. Loss: 0.03266042299568653\n",
      "Epoch: 530, Batch: 300, Avg. Loss: 0.03653095632791519\n",
      "Epoch: 530, Batch: 400, Avg. Loss: 0.025703770741820335\n",
      "Epoch: 530, Batch: 500, Avg. Loss: 0.025820124670863153\n",
      "Epoch: 530, Batch: 600, Avg. Loss: 0.032735316902399066\n",
      "Epoch: 530, Batch: 700, Avg. Loss: 0.030881326571106912\n",
      "Epoch: 530, Batch: 800, Avg. Loss: 0.03214301906526089\n",
      "Epoch: 530, Batch: 900, Avg. Loss: 0.032897689417004584\n",
      "Epoch: 531, Batch: 100, Avg. Loss: 0.02573905348777771\n",
      "Epoch: 531, Batch: 200, Avg. Loss: 0.030492555871605872\n",
      "Epoch: 531, Batch: 300, Avg. Loss: 0.03530367188155651\n",
      "Epoch: 531, Batch: 400, Avg. Loss: 0.02856226071715355\n",
      "Epoch: 531, Batch: 500, Avg. Loss: 0.03010111942887306\n",
      "Epoch: 531, Batch: 600, Avg. Loss: 0.03256560511887074\n",
      "Epoch: 531, Batch: 700, Avg. Loss: 0.02751978762447834\n",
      "Epoch: 531, Batch: 800, Avg. Loss: 0.03118618845939636\n",
      "Epoch: 531, Batch: 900, Avg. Loss: 0.03336898900568485\n",
      "Epoch: 532, Batch: 100, Avg. Loss: 0.03307171795517206\n",
      "Epoch: 532, Batch: 200, Avg. Loss: 0.037328004017472265\n",
      "Epoch: 532, Batch: 300, Avg. Loss: 0.03125871054828167\n",
      "Epoch: 532, Batch: 400, Avg. Loss: 0.031520597189664844\n",
      "Epoch: 532, Batch: 500, Avg. Loss: 0.028803297579288484\n",
      "Epoch: 532, Batch: 600, Avg. Loss: 0.030517448261380194\n",
      "Epoch: 532, Batch: 700, Avg. Loss: 0.029383955448865892\n",
      "Epoch: 532, Batch: 800, Avg. Loss: 0.031186433658003807\n",
      "Epoch: 532, Batch: 900, Avg. Loss: 0.024852185770869256\n",
      "Epoch: 533, Batch: 100, Avg. Loss: 0.029583549350500105\n",
      "Epoch: 533, Batch: 200, Avg. Loss: 0.025289478451013564\n",
      "Epoch: 533, Batch: 300, Avg. Loss: 0.03241742674261332\n",
      "Epoch: 533, Batch: 400, Avg. Loss: 0.030542831942439078\n",
      "Epoch: 533, Batch: 500, Avg. Loss: 0.0342148444801569\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 533, Batch: 600, Avg. Loss: 0.03176404617726803\n",
      "Epoch: 533, Batch: 700, Avg. Loss: 0.03667725577950478\n",
      "Epoch: 533, Batch: 800, Avg. Loss: 0.0272427224367857\n",
      "Epoch: 533, Batch: 900, Avg. Loss: 0.027252156808972357\n",
      "Epoch: 534, Batch: 100, Avg. Loss: 0.026310502514243127\n",
      "Epoch: 534, Batch: 200, Avg. Loss: 0.031968814954161644\n",
      "Epoch: 534, Batch: 300, Avg. Loss: 0.022636979520320892\n",
      "Epoch: 534, Batch: 400, Avg. Loss: 0.029887729957699775\n",
      "Epoch: 534, Batch: 500, Avg. Loss: 0.03388749994337559\n",
      "Epoch: 534, Batch: 600, Avg. Loss: 0.029931445494294166\n",
      "Epoch: 534, Batch: 700, Avg. Loss: 0.03485952436923981\n",
      "Epoch: 534, Batch: 800, Avg. Loss: 0.028297079876065254\n",
      "Epoch: 534, Batch: 900, Avg. Loss: 0.0332968345284462\n",
      "Epoch: 535, Batch: 100, Avg. Loss: 0.026871842220425605\n",
      "Epoch: 535, Batch: 200, Avg. Loss: 0.03321067117154598\n",
      "Epoch: 535, Batch: 300, Avg. Loss: 0.033512062281370165\n",
      "Epoch: 535, Batch: 400, Avg. Loss: 0.032718420475721356\n",
      "Epoch: 535, Batch: 500, Avg. Loss: 0.03005244702100754\n",
      "Epoch: 535, Batch: 600, Avg. Loss: 0.02543818987905979\n",
      "Epoch: 535, Batch: 700, Avg. Loss: 0.031534691601991655\n",
      "Epoch: 535, Batch: 800, Avg. Loss: 0.03505239017307758\n",
      "Epoch: 535, Batch: 900, Avg. Loss: 0.03148506388068199\n",
      "Epoch: 536, Batch: 100, Avg. Loss: 0.03005436673760414\n",
      "Epoch: 536, Batch: 200, Avg. Loss: 0.040682716369628905\n",
      "Epoch: 536, Batch: 300, Avg. Loss: 0.03161817371845246\n",
      "Epoch: 536, Batch: 400, Avg. Loss: 0.027700992822647093\n",
      "Epoch: 536, Batch: 500, Avg. Loss: 0.03363135993480682\n",
      "Epoch: 536, Batch: 600, Avg. Loss: 0.03256693482398987\n",
      "Epoch: 536, Batch: 700, Avg. Loss: 0.02842950403690338\n",
      "Epoch: 536, Batch: 800, Avg. Loss: 0.027364706620573997\n",
      "Epoch: 536, Batch: 900, Avg. Loss: 0.0291746760904789\n",
      "Epoch: 537, Batch: 100, Avg. Loss: 0.038483186066150664\n",
      "Epoch: 537, Batch: 200, Avg. Loss: 0.028815347775816916\n",
      "Epoch: 537, Batch: 300, Avg. Loss: 0.030772485435009003\n",
      "Epoch: 537, Batch: 400, Avg. Loss: 0.03179392546415329\n",
      "Epoch: 537, Batch: 500, Avg. Loss: 0.028685981035232545\n",
      "Epoch: 537, Batch: 600, Avg. Loss: 0.03133645668625831\n",
      "Epoch: 537, Batch: 700, Avg. Loss: 0.03096174955368042\n",
      "Epoch: 537, Batch: 800, Avg. Loss: 0.030187832787632943\n",
      "Epoch: 537, Batch: 900, Avg. Loss: 0.029276012927293777\n",
      "Epoch: 538, Batch: 100, Avg. Loss: 0.029389326423406602\n",
      "Epoch: 538, Batch: 200, Avg. Loss: 0.02964799292385578\n",
      "Epoch: 538, Batch: 300, Avg. Loss: 0.02430313266813755\n",
      "Epoch: 538, Batch: 400, Avg. Loss: 0.026850203722715377\n",
      "Epoch: 538, Batch: 500, Avg. Loss: 0.027456297278404235\n",
      "Epoch: 538, Batch: 600, Avg. Loss: 0.03509085841476917\n",
      "Epoch: 538, Batch: 700, Avg. Loss: 0.026247500851750375\n",
      "Epoch: 538, Batch: 800, Avg. Loss: 0.03603975392878055\n",
      "Epoch: 538, Batch: 900, Avg. Loss: 0.028807078674435615\n",
      "Epoch: 539, Batch: 100, Avg. Loss: 0.03190166771411896\n",
      "Epoch: 539, Batch: 200, Avg. Loss: 0.03186205714941025\n",
      "Epoch: 539, Batch: 300, Avg. Loss: 0.02869003176689148\n",
      "Epoch: 539, Batch: 400, Avg. Loss: 0.026429781094193458\n",
      "Epoch: 539, Batch: 500, Avg. Loss: 0.029963350147008894\n",
      "Epoch: 539, Batch: 600, Avg. Loss: 0.030690564662218093\n",
      "Epoch: 539, Batch: 700, Avg. Loss: 0.02877072982490063\n",
      "Epoch: 539, Batch: 800, Avg. Loss: 0.03442604213953018\n",
      "Epoch: 539, Batch: 900, Avg. Loss: 0.028214432671666144\n",
      "Epoch: 540, Batch: 100, Avg. Loss: 0.03276558630168438\n",
      "Epoch: 540, Batch: 200, Avg. Loss: 0.030787161886692047\n",
      "Epoch: 540, Batch: 300, Avg. Loss: 0.025905811339616776\n",
      "Epoch: 540, Batch: 400, Avg. Loss: 0.0323052167147398\n",
      "Epoch: 540, Batch: 500, Avg. Loss: 0.027526729851961137\n",
      "Epoch: 540, Batch: 600, Avg. Loss: 0.03282554104924202\n",
      "Epoch: 540, Batch: 700, Avg. Loss: 0.02664202854037285\n",
      "Epoch: 540, Batch: 800, Avg. Loss: 0.02373651970177889\n",
      "Epoch: 540, Batch: 900, Avg. Loss: 0.032621942833065985\n",
      "Epoch: 541, Batch: 100, Avg. Loss: 0.031507602855563165\n",
      "Epoch: 541, Batch: 200, Avg. Loss: 0.03015588901937008\n",
      "Epoch: 541, Batch: 300, Avg. Loss: 0.027513462677598\n",
      "Epoch: 541, Batch: 400, Avg. Loss: 0.030759770274162292\n",
      "Epoch: 541, Batch: 500, Avg. Loss: 0.02950003296136856\n",
      "Epoch: 541, Batch: 600, Avg. Loss: 0.029729222282767295\n",
      "Epoch: 541, Batch: 700, Avg. Loss: 0.02910269409418106\n",
      "Epoch: 541, Batch: 800, Avg. Loss: 0.02617343470454216\n",
      "Epoch: 541, Batch: 900, Avg. Loss: 0.02635458990931511\n",
      "Epoch: 542, Batch: 100, Avg. Loss: 0.029228182435035707\n",
      "Epoch: 542, Batch: 200, Avg. Loss: 0.025548107773065567\n",
      "Epoch: 542, Batch: 300, Avg. Loss: 0.030796034932136535\n",
      "Epoch: 542, Batch: 400, Avg. Loss: 0.03267910592257976\n",
      "Epoch: 542, Batch: 500, Avg. Loss: 0.030144811198115348\n",
      "Epoch: 542, Batch: 600, Avg. Loss: 0.031195410415530204\n",
      "Epoch: 542, Batch: 700, Avg. Loss: 0.03029587335884571\n",
      "Epoch: 542, Batch: 800, Avg. Loss: 0.025300205424427985\n",
      "Epoch: 542, Batch: 900, Avg. Loss: 0.032991948127746584\n",
      "Epoch: 543, Batch: 100, Avg. Loss: 0.029665017500519753\n",
      "Epoch: 543, Batch: 200, Avg. Loss: 0.03246897362172604\n",
      "Epoch: 543, Batch: 300, Avg. Loss: 0.030005667731165885\n",
      "Epoch: 543, Batch: 400, Avg. Loss: 0.023479501381516457\n",
      "Epoch: 543, Batch: 500, Avg. Loss: 0.027531741335988045\n",
      "Epoch: 543, Batch: 600, Avg. Loss: 0.03273402132093906\n",
      "Epoch: 543, Batch: 700, Avg. Loss: 0.03385364852845669\n",
      "Epoch: 543, Batch: 800, Avg. Loss: 0.03596285350620747\n",
      "Epoch: 543, Batch: 900, Avg. Loss: 0.03126692935824394\n",
      "Epoch: 544, Batch: 100, Avg. Loss: 0.03664912797510624\n",
      "Epoch: 544, Batch: 200, Avg. Loss: 0.025639708042144774\n",
      "Epoch: 544, Batch: 300, Avg. Loss: 0.03677725866436958\n",
      "Epoch: 544, Batch: 400, Avg. Loss: 0.031370746120810505\n",
      "Epoch: 544, Batch: 500, Avg. Loss: 0.024085975214838982\n",
      "Epoch: 544, Batch: 600, Avg. Loss: 0.02664103463292122\n",
      "Epoch: 544, Batch: 700, Avg. Loss: 0.02907472774386406\n",
      "Epoch: 544, Batch: 800, Avg. Loss: 0.036167584359645844\n",
      "Epoch: 544, Batch: 900, Avg. Loss: 0.02325392931699753\n",
      "Epoch: 545, Batch: 100, Avg. Loss: 0.03393354773521423\n",
      "Epoch: 545, Batch: 200, Avg. Loss: 0.03586975000798702\n",
      "Epoch: 545, Batch: 300, Avg. Loss: 0.03087210305035114\n",
      "Epoch: 545, Batch: 400, Avg. Loss: 0.027963280454277994\n",
      "Epoch: 545, Batch: 500, Avg. Loss: 0.028234258219599725\n",
      "Epoch: 545, Batch: 600, Avg. Loss: 0.0317587323486805\n",
      "Epoch: 545, Batch: 700, Avg. Loss: 0.02598075456917286\n",
      "Epoch: 545, Batch: 800, Avg. Loss: 0.04111418753862381\n",
      "Epoch: 545, Batch: 900, Avg. Loss: 0.03759301111102104\n",
      "Epoch: 546, Batch: 100, Avg. Loss: 0.026117712929844855\n",
      "Epoch: 546, Batch: 200, Avg. Loss: 0.030945042297244072\n",
      "Epoch: 546, Batch: 300, Avg. Loss: 0.02892496354877949\n",
      "Epoch: 546, Batch: 400, Avg. Loss: 0.03371851444244385\n",
      "Epoch: 546, Batch: 500, Avg. Loss: 0.028405931815505027\n",
      "Epoch: 546, Batch: 600, Avg. Loss: 0.034584219455719\n",
      "Epoch: 546, Batch: 700, Avg. Loss: 0.03061911679804325\n",
      "Epoch: 546, Batch: 800, Avg. Loss: 0.026275454759597777\n",
      "Epoch: 546, Batch: 900, Avg. Loss: 0.028280830979347228\n",
      "Epoch: 547, Batch: 100, Avg. Loss: 0.031782299280166626\n",
      "Epoch: 547, Batch: 200, Avg. Loss: 0.03358738079667092\n",
      "Epoch: 547, Batch: 300, Avg. Loss: 0.035233036503195765\n",
      "Epoch: 547, Batch: 400, Avg. Loss: 0.029410209879279135\n",
      "Epoch: 547, Batch: 500, Avg. Loss: 0.029186575412750243\n",
      "Epoch: 547, Batch: 600, Avg. Loss: 0.03125989712774754\n",
      "Epoch: 547, Batch: 700, Avg. Loss: 0.027765373513102532\n",
      "Epoch: 547, Batch: 800, Avg. Loss: 0.027137105613946916\n",
      "Epoch: 547, Batch: 900, Avg. Loss: 0.025275075137615205\n",
      "Epoch: 548, Batch: 100, Avg. Loss: 0.02977283425629139\n",
      "Epoch: 548, Batch: 200, Avg. Loss: 0.029185889586806298\n",
      "Epoch: 548, Batch: 300, Avg. Loss: 0.028841842859983445\n",
      "Epoch: 548, Batch: 400, Avg. Loss: 0.027322111651301384\n",
      "Epoch: 548, Batch: 500, Avg. Loss: 0.025465745627880097\n",
      "Epoch: 548, Batch: 600, Avg. Loss: 0.02957301966845989\n",
      "Epoch: 548, Batch: 700, Avg. Loss: 0.029654699489474298\n",
      "Epoch: 548, Batch: 800, Avg. Loss: 0.03229606255888939\n",
      "Epoch: 548, Batch: 900, Avg. Loss: 0.030991250276565553\n",
      "Epoch: 549, Batch: 100, Avg. Loss: 0.027492010891437532\n",
      "Epoch: 549, Batch: 200, Avg. Loss: 0.02466999091207981\n",
      "Epoch: 549, Batch: 300, Avg. Loss: 0.031443719640374186\n",
      "Epoch: 549, Batch: 400, Avg. Loss: 0.032565603405237196\n",
      "Epoch: 549, Batch: 500, Avg. Loss: 0.028109831362962724\n",
      "Epoch: 549, Batch: 600, Avg. Loss: 0.032327702715992924\n",
      "Epoch: 549, Batch: 700, Avg. Loss: 0.024977393224835397\n",
      "Epoch: 549, Batch: 800, Avg. Loss: 0.02696732699871063\n",
      "Epoch: 549, Batch: 900, Avg. Loss: 0.028261693492531777\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 550, Batch: 100, Avg. Loss: 0.029335937201976775\n",
      "Epoch: 550, Batch: 200, Avg. Loss: 0.03164926409721375\n",
      "Epoch: 550, Batch: 300, Avg. Loss: 0.0330347429215908\n",
      "Epoch: 550, Batch: 400, Avg. Loss: 0.03256269566714764\n",
      "Epoch: 550, Batch: 500, Avg. Loss: 0.03346891552209854\n",
      "Epoch: 550, Batch: 600, Avg. Loss: 0.022904559522867202\n",
      "Epoch: 550, Batch: 700, Avg. Loss: 0.03193883776664734\n",
      "Epoch: 550, Batch: 800, Avg. Loss: 0.026512805968523026\n",
      "Epoch: 550, Batch: 900, Avg. Loss: 0.03117857463657856\n",
      "Epoch: 551, Batch: 100, Avg. Loss: 0.02853252589702606\n",
      "Epoch: 551, Batch: 200, Avg. Loss: 0.032882905527949334\n",
      "Epoch: 551, Batch: 300, Avg. Loss: 0.02471749857068062\n",
      "Epoch: 551, Batch: 400, Avg. Loss: 0.025028052181005477\n",
      "Epoch: 551, Batch: 500, Avg. Loss: 0.026834820806980134\n",
      "Epoch: 551, Batch: 600, Avg. Loss: 0.031851385906338694\n",
      "Epoch: 551, Batch: 700, Avg. Loss: 0.0332801441848278\n",
      "Epoch: 551, Batch: 800, Avg. Loss: 0.03084165759384632\n",
      "Epoch: 551, Batch: 900, Avg. Loss: 0.03014270171523094\n",
      "Epoch: 552, Batch: 100, Avg. Loss: 0.03885935164988041\n",
      "Epoch: 552, Batch: 200, Avg. Loss: 0.033271922618150714\n",
      "Epoch: 552, Batch: 300, Avg. Loss: 0.026532692164182664\n",
      "Epoch: 552, Batch: 400, Avg. Loss: 0.030468227565288542\n",
      "Epoch: 552, Batch: 500, Avg. Loss: 0.0258614481985569\n",
      "Epoch: 552, Batch: 600, Avg. Loss: 0.0313443673402071\n",
      "Epoch: 552, Batch: 700, Avg. Loss: 0.02604430191218853\n",
      "Epoch: 552, Batch: 800, Avg. Loss: 0.032019102871418\n",
      "Epoch: 552, Batch: 900, Avg. Loss: 0.02983724057674408\n",
      "Epoch: 553, Batch: 100, Avg. Loss: 0.03189145632088184\n",
      "Epoch: 553, Batch: 200, Avg. Loss: 0.027537939324975013\n",
      "Epoch: 553, Batch: 300, Avg. Loss: 0.025973876789212227\n",
      "Epoch: 553, Batch: 400, Avg. Loss: 0.02564884603023529\n",
      "Epoch: 553, Batch: 500, Avg. Loss: 0.03211712554097176\n",
      "Epoch: 553, Batch: 600, Avg. Loss: 0.02934788927435875\n",
      "Epoch: 553, Batch: 700, Avg. Loss: 0.02888328894972801\n",
      "Epoch: 553, Batch: 800, Avg. Loss: 0.028047094792127608\n",
      "Epoch: 553, Batch: 900, Avg. Loss: 0.03018188238143921\n",
      "Epoch: 554, Batch: 100, Avg. Loss: 0.02888105794787407\n",
      "Epoch: 554, Batch: 200, Avg. Loss: 0.032105588018894196\n",
      "Epoch: 554, Batch: 300, Avg. Loss: 0.028515181243419646\n",
      "Epoch: 554, Batch: 400, Avg. Loss: 0.029079894199967385\n",
      "Epoch: 554, Batch: 500, Avg. Loss: 0.03179316848516464\n",
      "Epoch: 554, Batch: 600, Avg. Loss: 0.028650744780898095\n",
      "Epoch: 554, Batch: 700, Avg. Loss: 0.033980870768427845\n",
      "Epoch: 554, Batch: 800, Avg. Loss: 0.030715564712882042\n",
      "Epoch: 554, Batch: 900, Avg. Loss: 0.028951272293925287\n",
      "Epoch: 555, Batch: 100, Avg. Loss: 0.02741163909435272\n",
      "Epoch: 555, Batch: 200, Avg. Loss: 0.030649380162358283\n",
      "Epoch: 555, Batch: 300, Avg. Loss: 0.027911588177084924\n",
      "Epoch: 555, Batch: 400, Avg. Loss: 0.03228165298700333\n",
      "Epoch: 555, Batch: 500, Avg. Loss: 0.030839942693710327\n",
      "Epoch: 555, Batch: 600, Avg. Loss: 0.029343501925468446\n",
      "Epoch: 555, Batch: 700, Avg. Loss: 0.021925494968891145\n",
      "Epoch: 555, Batch: 800, Avg. Loss: 0.029828443974256515\n",
      "Epoch: 555, Batch: 900, Avg. Loss: 0.028197279945015908\n",
      "Epoch: 556, Batch: 100, Avg. Loss: 0.028694818168878554\n",
      "Epoch: 556, Batch: 200, Avg. Loss: 0.035521590560674665\n",
      "Epoch: 556, Batch: 300, Avg. Loss: 0.03329927444458008\n",
      "Epoch: 556, Batch: 400, Avg. Loss: 0.03341587610542774\n",
      "Epoch: 556, Batch: 500, Avg. Loss: 0.029900918528437615\n",
      "Epoch: 556, Batch: 600, Avg. Loss: 0.02967190757393837\n",
      "Epoch: 556, Batch: 700, Avg. Loss: 0.02788291022181511\n",
      "Epoch: 556, Batch: 800, Avg. Loss: 0.02650505371391773\n",
      "Epoch: 556, Batch: 900, Avg. Loss: 0.03505663946270943\n",
      "Epoch: 557, Batch: 100, Avg. Loss: 0.025265349671244622\n",
      "Epoch: 557, Batch: 200, Avg. Loss: 0.03290823884308338\n",
      "Epoch: 557, Batch: 300, Avg. Loss: 0.02536766283214092\n",
      "Epoch: 557, Batch: 400, Avg. Loss: 0.02876827761530876\n",
      "Epoch: 557, Batch: 500, Avg. Loss: 0.02812453478574753\n",
      "Epoch: 557, Batch: 600, Avg. Loss: 0.03640211343765259\n",
      "Epoch: 557, Batch: 700, Avg. Loss: 0.030886817649006845\n",
      "Epoch: 557, Batch: 800, Avg. Loss: 0.037296461313962935\n",
      "Epoch: 557, Batch: 900, Avg. Loss: 0.029582280069589614\n",
      "Epoch: 558, Batch: 100, Avg. Loss: 0.03241157628595829\n",
      "Epoch: 558, Batch: 200, Avg. Loss: 0.030181590169668197\n",
      "Epoch: 558, Batch: 300, Avg. Loss: 0.03615428104996681\n",
      "Epoch: 558, Batch: 400, Avg. Loss: 0.03634930968284607\n",
      "Epoch: 558, Batch: 500, Avg. Loss: 0.03062301464378834\n",
      "Epoch: 558, Batch: 600, Avg. Loss: 0.02981001190841198\n",
      "Epoch: 558, Batch: 700, Avg. Loss: 0.02498447723686695\n",
      "Epoch: 558, Batch: 800, Avg. Loss: 0.02659558556973934\n",
      "Epoch: 558, Batch: 900, Avg. Loss: 0.027617647796869278\n",
      "Epoch: 559, Batch: 100, Avg. Loss: 0.030585458055138588\n",
      "Epoch: 559, Batch: 200, Avg. Loss: 0.026006137803196907\n",
      "Epoch: 559, Batch: 300, Avg. Loss: 0.03908635444939137\n",
      "Epoch: 559, Batch: 400, Avg. Loss: 0.029715247750282288\n",
      "Epoch: 559, Batch: 500, Avg. Loss: 0.031950693279504776\n",
      "Epoch: 559, Batch: 600, Avg. Loss: 0.02871662974357605\n",
      "Epoch: 559, Batch: 700, Avg. Loss: 0.027288301810622216\n",
      "Epoch: 559, Batch: 800, Avg. Loss: 0.02556499816477299\n",
      "Epoch: 559, Batch: 900, Avg. Loss: 0.029840357080101968\n",
      "Epoch: 560, Batch: 100, Avg. Loss: 0.025365046709775924\n",
      "Epoch: 560, Batch: 200, Avg. Loss: 0.035858515948057175\n",
      "Epoch: 560, Batch: 300, Avg. Loss: 0.029666745886206627\n",
      "Epoch: 560, Batch: 400, Avg. Loss: 0.023077895417809486\n",
      "Epoch: 560, Batch: 500, Avg. Loss: 0.03096359834074974\n",
      "Epoch: 560, Batch: 600, Avg. Loss: 0.030014379620552062\n",
      "Epoch: 560, Batch: 700, Avg. Loss: 0.03142125345766544\n",
      "Epoch: 560, Batch: 800, Avg. Loss: 0.02282028578221798\n",
      "Epoch: 560, Batch: 900, Avg. Loss: 0.03401271060109139\n",
      "Epoch: 561, Batch: 100, Avg. Loss: 0.028266742900013922\n",
      "Epoch: 561, Batch: 200, Avg. Loss: 0.03259532660245895\n",
      "Epoch: 561, Batch: 300, Avg. Loss: 0.03217829555273056\n",
      "Epoch: 561, Batch: 400, Avg. Loss: 0.03183566771447659\n",
      "Epoch: 561, Batch: 500, Avg. Loss: 0.03113421604037285\n",
      "Epoch: 561, Batch: 600, Avg. Loss: 0.028842074126005174\n",
      "Epoch: 561, Batch: 700, Avg. Loss: 0.027822018191218375\n",
      "Epoch: 561, Batch: 800, Avg. Loss: 0.030260315462946892\n",
      "Epoch: 561, Batch: 900, Avg. Loss: 0.027453977391123773\n",
      "Epoch: 562, Batch: 100, Avg. Loss: 0.033148570209741594\n",
      "Epoch: 562, Batch: 200, Avg. Loss: 0.02865039899945259\n",
      "Epoch: 562, Batch: 300, Avg. Loss: 0.037238968163728715\n",
      "Epoch: 562, Batch: 400, Avg. Loss: 0.02935393266379833\n",
      "Epoch: 562, Batch: 500, Avg. Loss: 0.027313295379281045\n",
      "Epoch: 562, Batch: 600, Avg. Loss: 0.028427734971046448\n",
      "Epoch: 562, Batch: 700, Avg. Loss: 0.03502737320959568\n",
      "Epoch: 562, Batch: 800, Avg. Loss: 0.023592704236507417\n",
      "Epoch: 562, Batch: 900, Avg. Loss: 0.022224638387560846\n",
      "Epoch: 563, Batch: 100, Avg. Loss: 0.032026381865143776\n",
      "Epoch: 563, Batch: 200, Avg. Loss: 0.0244964037835598\n",
      "Epoch: 563, Batch: 300, Avg. Loss: 0.029798843264579773\n",
      "Epoch: 563, Batch: 400, Avg. Loss: 0.029536199122667313\n",
      "Epoch: 563, Batch: 500, Avg. Loss: 0.028968588188290595\n",
      "Epoch: 563, Batch: 600, Avg. Loss: 0.02666901558637619\n",
      "Epoch: 563, Batch: 700, Avg. Loss: 0.02889547899365425\n",
      "Epoch: 563, Batch: 800, Avg. Loss: 0.027667498961091042\n",
      "Epoch: 563, Batch: 900, Avg. Loss: 0.03334428735077381\n",
      "Epoch: 564, Batch: 100, Avg. Loss: 0.03212920002639294\n",
      "Epoch: 564, Batch: 200, Avg. Loss: 0.027246159687638282\n",
      "Epoch: 564, Batch: 300, Avg. Loss: 0.03246216550469398\n",
      "Epoch: 564, Batch: 400, Avg. Loss: 0.02936315380036831\n",
      "Epoch: 564, Batch: 500, Avg. Loss: 0.030652713477611542\n",
      "Epoch: 564, Batch: 600, Avg. Loss: 0.029884208887815476\n",
      "Epoch: 564, Batch: 700, Avg. Loss: 0.03538143254816532\n",
      "Epoch: 564, Batch: 800, Avg. Loss: 0.02482820048928261\n",
      "Epoch: 564, Batch: 900, Avg. Loss: 0.030646784603595732\n",
      "Epoch: 565, Batch: 100, Avg. Loss: 0.02080389842391014\n",
      "Epoch: 565, Batch: 200, Avg. Loss: 0.028154057189822196\n",
      "Epoch: 565, Batch: 300, Avg. Loss: 0.0332976895570755\n",
      "Epoch: 565, Batch: 400, Avg. Loss: 0.03329698748886585\n",
      "Epoch: 565, Batch: 500, Avg. Loss: 0.03356701508164406\n",
      "Epoch: 565, Batch: 600, Avg. Loss: 0.020828296318650245\n",
      "Epoch: 565, Batch: 700, Avg. Loss: 0.03656754150986671\n",
      "Epoch: 565, Batch: 800, Avg. Loss: 0.030117028802633287\n",
      "Epoch: 565, Batch: 900, Avg. Loss: 0.02606277659535408\n",
      "Epoch: 566, Batch: 100, Avg. Loss: 0.02910837635397911\n",
      "Epoch: 566, Batch: 200, Avg. Loss: 0.027438910007476808\n",
      "Epoch: 566, Batch: 300, Avg. Loss: 0.02955807887017727\n",
      "Epoch: 566, Batch: 400, Avg. Loss: 0.03348858550190926\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 566, Batch: 500, Avg. Loss: 0.03465044319629669\n",
      "Epoch: 566, Batch: 600, Avg. Loss: 0.02747876591980457\n",
      "Epoch: 566, Batch: 700, Avg. Loss: 0.02475099839270115\n",
      "Epoch: 566, Batch: 800, Avg. Loss: 0.030818182453513147\n",
      "Epoch: 566, Batch: 900, Avg. Loss: 0.026373773738741876\n",
      "Epoch: 567, Batch: 100, Avg. Loss: 0.027551349848508835\n",
      "Epoch: 567, Batch: 200, Avg. Loss: 0.026294900476932524\n",
      "Epoch: 567, Batch: 300, Avg. Loss: 0.03191612005233765\n",
      "Epoch: 567, Batch: 400, Avg. Loss: 0.02715898647904396\n",
      "Epoch: 567, Batch: 500, Avg. Loss: 0.0386285200715065\n",
      "Epoch: 567, Batch: 600, Avg. Loss: 0.03191085807979107\n",
      "Epoch: 567, Batch: 700, Avg. Loss: 0.03170505180954933\n",
      "Epoch: 567, Batch: 800, Avg. Loss: 0.02861050009727478\n",
      "Epoch: 567, Batch: 900, Avg. Loss: 0.03223761029541492\n",
      "Epoch: 568, Batch: 100, Avg. Loss: 0.02917781539261341\n",
      "Epoch: 568, Batch: 200, Avg. Loss: 0.027633612602949144\n",
      "Epoch: 568, Batch: 300, Avg. Loss: 0.024804603457450867\n",
      "Epoch: 568, Batch: 400, Avg. Loss: 0.03140037208795547\n",
      "Epoch: 568, Batch: 500, Avg. Loss: 0.029413594156503676\n",
      "Epoch: 568, Batch: 600, Avg. Loss: 0.028906433656811715\n",
      "Epoch: 568, Batch: 700, Avg. Loss: 0.027484124824404717\n",
      "Epoch: 568, Batch: 800, Avg. Loss: 0.033755169957876204\n",
      "Epoch: 568, Batch: 900, Avg. Loss: 0.026910713836550713\n",
      "Epoch: 569, Batch: 100, Avg. Loss: 0.026674629002809525\n",
      "Epoch: 569, Batch: 200, Avg. Loss: 0.02743470698595047\n",
      "Epoch: 569, Batch: 300, Avg. Loss: 0.03395943101495504\n",
      "Epoch: 569, Batch: 400, Avg. Loss: 0.02836468316614628\n",
      "Epoch: 569, Batch: 500, Avg. Loss: 0.03299292683601379\n",
      "Epoch: 569, Batch: 600, Avg. Loss: 0.0306993817538023\n",
      "Epoch: 569, Batch: 700, Avg. Loss: 0.026230295449495317\n",
      "Epoch: 569, Batch: 800, Avg. Loss: 0.029933903217315674\n",
      "Epoch: 569, Batch: 900, Avg. Loss: 0.029351166188716887\n",
      "Epoch: 570, Batch: 100, Avg. Loss: 0.031751553118228915\n",
      "Epoch: 570, Batch: 200, Avg. Loss: 0.026831653267145157\n",
      "Epoch: 570, Batch: 300, Avg. Loss: 0.029796998277306558\n",
      "Epoch: 570, Batch: 400, Avg. Loss: 0.024388333857059477\n",
      "Epoch: 570, Batch: 500, Avg. Loss: 0.024837056696414946\n",
      "Epoch: 570, Batch: 600, Avg. Loss: 0.03062451347708702\n",
      "Epoch: 570, Batch: 700, Avg. Loss: 0.02879285253584385\n",
      "Epoch: 570, Batch: 800, Avg. Loss: 0.029008321464061737\n",
      "Epoch: 570, Batch: 900, Avg. Loss: 0.030603734776377678\n",
      "Epoch: 571, Batch: 100, Avg. Loss: 0.03652619570493698\n",
      "Epoch: 571, Batch: 200, Avg. Loss: 0.03051460601389408\n",
      "Epoch: 571, Batch: 300, Avg. Loss: 0.028221134692430497\n",
      "Epoch: 571, Batch: 400, Avg. Loss: 0.03130045726895332\n",
      "Epoch: 571, Batch: 500, Avg. Loss: 0.0265571591258049\n",
      "Epoch: 571, Batch: 600, Avg. Loss: 0.03249685309827328\n",
      "Epoch: 571, Batch: 700, Avg. Loss: 0.025927717238664626\n",
      "Epoch: 571, Batch: 800, Avg. Loss: 0.02754311092197895\n",
      "Epoch: 571, Batch: 900, Avg. Loss: 0.030577929392457008\n",
      "Epoch: 572, Batch: 100, Avg. Loss: 0.027818656340241433\n",
      "Epoch: 572, Batch: 200, Avg. Loss: 0.03050278779119253\n",
      "Epoch: 572, Batch: 300, Avg. Loss: 0.029880706071853638\n",
      "Epoch: 572, Batch: 400, Avg. Loss: 0.029907417297363282\n",
      "Epoch: 572, Batch: 500, Avg. Loss: 0.029468260630965235\n",
      "Epoch: 572, Batch: 600, Avg. Loss: 0.03504732958972454\n",
      "Epoch: 572, Batch: 700, Avg. Loss: 0.025905863642692567\n",
      "Epoch: 572, Batch: 800, Avg. Loss: 0.023634827360510827\n",
      "Epoch: 572, Batch: 900, Avg. Loss: 0.027728310003876686\n",
      "Epoch: 573, Batch: 100, Avg. Loss: 0.0308633354306221\n",
      "Epoch: 573, Batch: 200, Avg. Loss: 0.026684456020593644\n",
      "Epoch: 573, Batch: 300, Avg. Loss: 0.02958673447370529\n",
      "Epoch: 573, Batch: 400, Avg. Loss: 0.02872786484658718\n",
      "Epoch: 573, Batch: 500, Avg. Loss: 0.038638664931058886\n",
      "Epoch: 573, Batch: 600, Avg. Loss: 0.03389625370502472\n",
      "Epoch: 573, Batch: 700, Avg. Loss: 0.03055143654346466\n",
      "Epoch: 573, Batch: 800, Avg. Loss: 0.03294795647263527\n",
      "Epoch: 573, Batch: 900, Avg. Loss: 0.03230826243758202\n",
      "Epoch: 574, Batch: 100, Avg. Loss: 0.02965564578771591\n",
      "Epoch: 574, Batch: 200, Avg. Loss: 0.03286027818918228\n",
      "Epoch: 574, Batch: 300, Avg. Loss: 0.02742648556828499\n",
      "Epoch: 574, Batch: 400, Avg. Loss: 0.03221328504383564\n",
      "Epoch: 574, Batch: 500, Avg. Loss: 0.027918122559785843\n",
      "Epoch: 574, Batch: 600, Avg. Loss: 0.025177228078246117\n",
      "Epoch: 574, Batch: 700, Avg. Loss: 0.028716384172439574\n",
      "Epoch: 574, Batch: 800, Avg. Loss: 0.026086729913949967\n",
      "Epoch: 574, Batch: 900, Avg. Loss: 0.028566240295767785\n",
      "Epoch: 575, Batch: 100, Avg. Loss: 0.03336874037981033\n",
      "Epoch: 575, Batch: 200, Avg. Loss: 0.031694984287023546\n",
      "Epoch: 575, Batch: 300, Avg. Loss: 0.027244163602590563\n",
      "Epoch: 575, Batch: 400, Avg. Loss: 0.0327343551069498\n",
      "Epoch: 575, Batch: 500, Avg. Loss: 0.027534684538841246\n",
      "Epoch: 575, Batch: 600, Avg. Loss: 0.03085384152829647\n",
      "Epoch: 575, Batch: 700, Avg. Loss: 0.038209078758955003\n",
      "Epoch: 575, Batch: 800, Avg. Loss: 0.03269179709255696\n",
      "Epoch: 575, Batch: 900, Avg. Loss: 0.028826382756233216\n",
      "Epoch: 576, Batch: 100, Avg. Loss: 0.030488471761345864\n",
      "Epoch: 576, Batch: 200, Avg. Loss: 0.029715794548392297\n",
      "Epoch: 576, Batch: 300, Avg. Loss: 0.025959062725305557\n",
      "Epoch: 576, Batch: 400, Avg. Loss: 0.02388665549457073\n",
      "Epoch: 576, Batch: 500, Avg. Loss: 0.026526402793824674\n",
      "Epoch: 576, Batch: 600, Avg. Loss: 0.030856757909059524\n",
      "Epoch: 576, Batch: 700, Avg. Loss: 0.026634981483221055\n",
      "Epoch: 576, Batch: 800, Avg. Loss: 0.026461407318711282\n",
      "Epoch: 576, Batch: 900, Avg. Loss: 0.026625250577926637\n",
      "Epoch: 577, Batch: 100, Avg. Loss: 0.032326141148805616\n",
      "Epoch: 577, Batch: 200, Avg. Loss: 0.024854895621538163\n",
      "Epoch: 577, Batch: 300, Avg. Loss: 0.026662451550364494\n",
      "Epoch: 577, Batch: 400, Avg. Loss: 0.02830447070300579\n",
      "Epoch: 577, Batch: 500, Avg. Loss: 0.03288996078073978\n",
      "Epoch: 577, Batch: 600, Avg. Loss: 0.0321504083275795\n",
      "Epoch: 577, Batch: 700, Avg. Loss: 0.029090497717261313\n",
      "Epoch: 577, Batch: 800, Avg. Loss: 0.027280540615320207\n",
      "Epoch: 577, Batch: 900, Avg. Loss: 0.028881620541214945\n",
      "Epoch: 578, Batch: 100, Avg. Loss: 0.031176506876945495\n",
      "Epoch: 578, Batch: 200, Avg. Loss: 0.026200105622410774\n",
      "Epoch: 578, Batch: 300, Avg. Loss: 0.025664560794830323\n",
      "Epoch: 578, Batch: 400, Avg. Loss: 0.02562697932124138\n",
      "Epoch: 578, Batch: 500, Avg. Loss: 0.031152992844581603\n",
      "Epoch: 578, Batch: 600, Avg. Loss: 0.03537826806306839\n",
      "Epoch: 578, Batch: 700, Avg. Loss: 0.028576319217681886\n",
      "Epoch: 578, Batch: 800, Avg. Loss: 0.033252677395939824\n",
      "Epoch: 578, Batch: 900, Avg. Loss: 0.03685295537114144\n",
      "Epoch: 579, Batch: 100, Avg. Loss: 0.030008778125047684\n",
      "Epoch: 579, Batch: 200, Avg. Loss: 0.02671822525560856\n",
      "Epoch: 579, Batch: 300, Avg. Loss: 0.02454486519098282\n",
      "Epoch: 579, Batch: 400, Avg. Loss: 0.03355206362903118\n",
      "Epoch: 579, Batch: 500, Avg. Loss: 0.03185217574238777\n",
      "Epoch: 579, Batch: 600, Avg. Loss: 0.03156043618917465\n",
      "Epoch: 579, Batch: 700, Avg. Loss: 0.024289350733160974\n",
      "Epoch: 579, Batch: 800, Avg. Loss: 0.024155318960547446\n",
      "Epoch: 579, Batch: 900, Avg. Loss: 0.025283229649066926\n",
      "Epoch: 580, Batch: 100, Avg. Loss: 0.02329915463924408\n",
      "Epoch: 580, Batch: 200, Avg. Loss: 0.025268932804465294\n",
      "Epoch: 580, Batch: 300, Avg. Loss: 0.025515724048018454\n",
      "Epoch: 580, Batch: 400, Avg. Loss: 0.026792286410927773\n",
      "Epoch: 580, Batch: 500, Avg. Loss: 0.031394280418753626\n",
      "Epoch: 580, Batch: 600, Avg. Loss: 0.03655832938849926\n",
      "Epoch: 580, Batch: 700, Avg. Loss: 0.028401289135217667\n",
      "Epoch: 580, Batch: 800, Avg. Loss: 0.030059413313865663\n",
      "Epoch: 580, Batch: 900, Avg. Loss: 0.02588490389287472\n",
      "Epoch: 581, Batch: 100, Avg. Loss: 0.03592214114964008\n",
      "Epoch: 581, Batch: 200, Avg. Loss: 0.026768610179424286\n",
      "Epoch: 581, Batch: 300, Avg. Loss: 0.028402853533625604\n",
      "Epoch: 581, Batch: 400, Avg. Loss: 0.03022258497774601\n",
      "Epoch: 581, Batch: 500, Avg. Loss: 0.027421915382146837\n",
      "Epoch: 581, Batch: 600, Avg. Loss: 0.026335143446922303\n",
      "Epoch: 581, Batch: 700, Avg. Loss: 0.03204719826579094\n",
      "Epoch: 581, Batch: 800, Avg. Loss: 0.028057956174015998\n",
      "Epoch: 581, Batch: 900, Avg. Loss: 0.026389576718211172\n",
      "Epoch: 582, Batch: 100, Avg. Loss: 0.029260151162743567\n",
      "Epoch: 582, Batch: 200, Avg. Loss: 0.0280348814278841\n",
      "Epoch: 582, Batch: 300, Avg. Loss: 0.02692060723900795\n",
      "Epoch: 582, Batch: 400, Avg. Loss: 0.03250852137804031\n",
      "Epoch: 582, Batch: 500, Avg. Loss: 0.028197400644421576\n",
      "Epoch: 582, Batch: 600, Avg. Loss: 0.03202307291328907\n",
      "Epoch: 582, Batch: 700, Avg. Loss: 0.027904314324259757\n",
      "Epoch: 582, Batch: 800, Avg. Loss: 0.027195859849452973\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 582, Batch: 900, Avg. Loss: 0.028809460550546645\n",
      "Epoch: 583, Batch: 100, Avg. Loss: 0.02404893547296524\n",
      "Epoch: 583, Batch: 200, Avg. Loss: 0.026478735134005546\n",
      "Epoch: 583, Batch: 300, Avg. Loss: 0.026757359206676483\n",
      "Epoch: 583, Batch: 400, Avg. Loss: 0.025364454835653305\n",
      "Epoch: 583, Batch: 500, Avg. Loss: 0.030176864489912986\n",
      "Epoch: 583, Batch: 600, Avg. Loss: 0.030141921490430833\n",
      "Epoch: 583, Batch: 700, Avg. Loss: 0.029410625472664834\n",
      "Epoch: 583, Batch: 800, Avg. Loss: 0.025595121011137964\n",
      "Epoch: 583, Batch: 900, Avg. Loss: 0.0291243676841259\n",
      "Epoch: 584, Batch: 100, Avg. Loss: 0.02556942284107208\n",
      "Epoch: 584, Batch: 200, Avg. Loss: 0.026174988225102426\n",
      "Epoch: 584, Batch: 300, Avg. Loss: 0.02758873350918293\n",
      "Epoch: 584, Batch: 400, Avg. Loss: 0.03346522852778435\n",
      "Epoch: 584, Batch: 500, Avg. Loss: 0.028782136142253875\n",
      "Epoch: 584, Batch: 600, Avg. Loss: 0.026431237012147904\n",
      "Epoch: 584, Batch: 700, Avg. Loss: 0.027411796301603317\n",
      "Epoch: 584, Batch: 800, Avg. Loss: 0.026759704500436785\n",
      "Epoch: 584, Batch: 900, Avg. Loss: 0.029875781685113907\n",
      "Epoch: 585, Batch: 100, Avg. Loss: 0.02763185903429985\n",
      "Epoch: 585, Batch: 200, Avg. Loss: 0.036290732845664024\n",
      "Epoch: 585, Batch: 300, Avg. Loss: 0.02599496766924858\n",
      "Epoch: 585, Batch: 400, Avg. Loss: 0.02737527348101139\n",
      "Epoch: 585, Batch: 500, Avg. Loss: 0.032879739478230474\n",
      "Epoch: 585, Batch: 600, Avg. Loss: 0.026212189346551895\n",
      "Epoch: 585, Batch: 700, Avg. Loss: 0.026099321842193605\n",
      "Epoch: 585, Batch: 800, Avg. Loss: 0.027366947457194327\n",
      "Epoch: 585, Batch: 900, Avg. Loss: 0.027402841225266455\n",
      "Epoch: 586, Batch: 100, Avg. Loss: 0.02984660230576992\n",
      "Epoch: 586, Batch: 200, Avg. Loss: 0.023768257349729538\n",
      "Epoch: 586, Batch: 300, Avg. Loss: 0.027260458916425704\n",
      "Epoch: 586, Batch: 400, Avg. Loss: 0.029798583462834358\n",
      "Epoch: 586, Batch: 500, Avg. Loss: 0.0336026494204998\n",
      "Epoch: 586, Batch: 600, Avg. Loss: 0.03110940471291542\n",
      "Epoch: 586, Batch: 700, Avg. Loss: 0.03132559210062027\n",
      "Epoch: 586, Batch: 800, Avg. Loss: 0.026806788966059685\n",
      "Epoch: 586, Batch: 900, Avg. Loss: 0.023351467549800872\n",
      "Epoch: 587, Batch: 100, Avg. Loss: 0.02761852040886879\n",
      "Epoch: 587, Batch: 200, Avg. Loss: 0.023008744567632675\n",
      "Epoch: 587, Batch: 300, Avg. Loss: 0.028210841715335847\n",
      "Epoch: 587, Batch: 400, Avg. Loss: 0.02940609447658062\n",
      "Epoch: 587, Batch: 500, Avg. Loss: 0.02556110352277756\n",
      "Epoch: 587, Batch: 600, Avg. Loss: 0.029317516423761843\n",
      "Epoch: 587, Batch: 700, Avg. Loss: 0.0341846489906311\n",
      "Epoch: 587, Batch: 800, Avg. Loss: 0.028025303110480307\n",
      "Epoch: 587, Batch: 900, Avg. Loss: 0.02562890902161598\n",
      "Epoch: 588, Batch: 100, Avg. Loss: 0.029782889857888222\n",
      "Epoch: 588, Batch: 200, Avg. Loss: 0.029188553988933563\n",
      "Epoch: 588, Batch: 300, Avg. Loss: 0.02605686441063881\n",
      "Epoch: 588, Batch: 400, Avg. Loss: 0.02812986433506012\n",
      "Epoch: 588, Batch: 500, Avg. Loss: 0.03154904969036579\n",
      "Epoch: 588, Batch: 600, Avg. Loss: 0.031712405383586884\n",
      "Epoch: 588, Batch: 700, Avg. Loss: 0.030841682106256485\n",
      "Epoch: 588, Batch: 800, Avg. Loss: 0.027462608590722085\n",
      "Epoch: 588, Batch: 900, Avg. Loss: 0.030248231031000616\n",
      "Epoch: 589, Batch: 100, Avg. Loss: 0.027904933020472526\n",
      "Epoch: 589, Batch: 200, Avg. Loss: 0.03022538773715496\n",
      "Epoch: 589, Batch: 300, Avg. Loss: 0.03267898455262184\n",
      "Epoch: 589, Batch: 400, Avg. Loss: 0.031168132647871973\n",
      "Epoch: 589, Batch: 500, Avg. Loss: 0.02629764921963215\n",
      "Epoch: 589, Batch: 600, Avg. Loss: 0.02755321465432644\n",
      "Epoch: 589, Batch: 700, Avg. Loss: 0.02964790016412735\n",
      "Epoch: 589, Batch: 800, Avg. Loss: 0.03082751452922821\n",
      "Epoch: 589, Batch: 900, Avg. Loss: 0.02382124714553356\n",
      "Epoch: 590, Batch: 100, Avg. Loss: 0.027187394425272942\n",
      "Epoch: 590, Batch: 200, Avg. Loss: 0.025294920802116393\n",
      "Epoch: 590, Batch: 300, Avg. Loss: 0.02582628883421421\n",
      "Epoch: 590, Batch: 400, Avg. Loss: 0.02718554899096489\n",
      "Epoch: 590, Batch: 500, Avg. Loss: 0.04155047565698624\n",
      "Epoch: 590, Batch: 600, Avg. Loss: 0.02893886521458626\n",
      "Epoch: 590, Batch: 700, Avg. Loss: 0.019055411368608475\n",
      "Epoch: 590, Batch: 800, Avg. Loss: 0.030475942194461824\n",
      "Epoch: 590, Batch: 900, Avg. Loss: 0.029859929233789443\n",
      "Epoch: 591, Batch: 100, Avg. Loss: 0.0311643385887146\n",
      "Epoch: 591, Batch: 200, Avg. Loss: 0.032012683227658274\n",
      "Epoch: 591, Batch: 300, Avg. Loss: 0.030805419981479645\n",
      "Epoch: 591, Batch: 400, Avg. Loss: 0.023402392715215683\n",
      "Epoch: 591, Batch: 500, Avg. Loss: 0.025853070020675658\n",
      "Epoch: 591, Batch: 600, Avg. Loss: 0.02781925082206726\n",
      "Epoch: 591, Batch: 700, Avg. Loss: 0.029793713167309763\n",
      "Epoch: 591, Batch: 800, Avg. Loss: 0.0313231211155653\n",
      "Epoch: 591, Batch: 900, Avg. Loss: 0.02687821850180626\n",
      "Epoch: 592, Batch: 100, Avg. Loss: 0.029593399837613107\n",
      "Epoch: 592, Batch: 200, Avg. Loss: 0.033937121108174324\n",
      "Epoch: 592, Batch: 300, Avg. Loss: 0.0259975253790617\n",
      "Epoch: 592, Batch: 400, Avg. Loss: 0.026266385465860367\n",
      "Epoch: 592, Batch: 500, Avg. Loss: 0.02888812206685543\n",
      "Epoch: 592, Batch: 600, Avg. Loss: 0.025726403146982192\n",
      "Epoch: 592, Batch: 700, Avg. Loss: 0.026022309735417366\n",
      "Epoch: 592, Batch: 800, Avg. Loss: 0.026407680287957192\n",
      "Epoch: 592, Batch: 900, Avg. Loss: 0.030314598232507706\n",
      "Epoch: 593, Batch: 100, Avg. Loss: 0.029286495223641396\n",
      "Epoch: 593, Batch: 200, Avg. Loss: 0.025679290145635605\n",
      "Epoch: 593, Batch: 300, Avg. Loss: 0.030644669756293298\n",
      "Epoch: 593, Batch: 400, Avg. Loss: 0.037517487071454525\n",
      "Epoch: 593, Batch: 500, Avg. Loss: 0.02921848714351654\n",
      "Epoch: 593, Batch: 600, Avg. Loss: 0.02829404592514038\n",
      "Epoch: 593, Batch: 700, Avg. Loss: 0.028202342092990874\n",
      "Epoch: 593, Batch: 800, Avg. Loss: 0.025501757860183716\n",
      "Epoch: 593, Batch: 900, Avg. Loss: 0.022801578491926194\n",
      "Epoch: 594, Batch: 100, Avg. Loss: 0.03444950141012668\n",
      "Epoch: 594, Batch: 200, Avg. Loss: 0.030020234808325767\n",
      "Epoch: 594, Batch: 300, Avg. Loss: 0.02705315612256527\n",
      "Epoch: 594, Batch: 400, Avg. Loss: 0.025634992346167566\n",
      "Epoch: 594, Batch: 500, Avg. Loss: 0.028716631010174753\n",
      "Epoch: 594, Batch: 600, Avg. Loss: 0.033309276252985\n",
      "Epoch: 594, Batch: 700, Avg. Loss: 0.02930991046130657\n",
      "Epoch: 594, Batch: 800, Avg. Loss: 0.026693111285567284\n",
      "Epoch: 594, Batch: 900, Avg. Loss: 0.02500762403011322\n",
      "Epoch: 595, Batch: 100, Avg. Loss: 0.026980390176177024\n",
      "Epoch: 595, Batch: 200, Avg. Loss: 0.030198637172579767\n",
      "Epoch: 595, Batch: 300, Avg. Loss: 0.025703954473137856\n",
      "Epoch: 595, Batch: 400, Avg. Loss: 0.029170435816049577\n",
      "Epoch: 595, Batch: 500, Avg. Loss: 0.02983985424041748\n",
      "Epoch: 595, Batch: 600, Avg. Loss: 0.03008120507001877\n",
      "Epoch: 595, Batch: 700, Avg. Loss: 0.02583041101694107\n",
      "Epoch: 595, Batch: 800, Avg. Loss: 0.0323271519690752\n",
      "Epoch: 595, Batch: 900, Avg. Loss: 0.029936517179012297\n",
      "Epoch: 596, Batch: 100, Avg. Loss: 0.02833461031317711\n",
      "Epoch: 596, Batch: 200, Avg. Loss: 0.03252200998365879\n",
      "Epoch: 596, Batch: 300, Avg. Loss: 0.025954022705554962\n",
      "Epoch: 596, Batch: 400, Avg. Loss: 0.029873741790652275\n",
      "Epoch: 596, Batch: 500, Avg. Loss: 0.02658434845507145\n",
      "Epoch: 596, Batch: 600, Avg. Loss: 0.026452381685376166\n",
      "Epoch: 596, Batch: 700, Avg. Loss: 0.033402131721377376\n",
      "Epoch: 596, Batch: 800, Avg. Loss: 0.029557169452309608\n",
      "Epoch: 596, Batch: 900, Avg. Loss: 0.024343664050102232\n",
      "Epoch: 597, Batch: 100, Avg. Loss: 0.03467860884964466\n",
      "Epoch: 597, Batch: 200, Avg. Loss: 0.024295664802193643\n",
      "Epoch: 597, Batch: 300, Avg. Loss: 0.023965008556842804\n",
      "Epoch: 597, Batch: 400, Avg. Loss: 0.029255149587988853\n",
      "Epoch: 597, Batch: 500, Avg. Loss: 0.028684385642409323\n",
      "Epoch: 597, Batch: 600, Avg. Loss: 0.03242331512272358\n",
      "Epoch: 597, Batch: 700, Avg. Loss: 0.03010255306959152\n",
      "Epoch: 597, Batch: 800, Avg. Loss: 0.025706878155469893\n",
      "Epoch: 597, Batch: 900, Avg. Loss: 0.02670911930501461\n",
      "Epoch: 598, Batch: 100, Avg. Loss: 0.03364550456404686\n",
      "Epoch: 598, Batch: 200, Avg. Loss: 0.028064796328544618\n",
      "Epoch: 598, Batch: 300, Avg. Loss: 0.024307246953248977\n",
      "Epoch: 598, Batch: 400, Avg. Loss: 0.036029505878686904\n",
      "Epoch: 598, Batch: 500, Avg. Loss: 0.02818297579884529\n",
      "Epoch: 598, Batch: 600, Avg. Loss: 0.03153220325708389\n",
      "Epoch: 598, Batch: 700, Avg. Loss: 0.02756702110171318\n",
      "Epoch: 598, Batch: 800, Avg. Loss: 0.028373524397611618\n",
      "Epoch: 598, Batch: 900, Avg. Loss: 0.0246504557877779\n",
      "Epoch: 599, Batch: 100, Avg. Loss: 0.02958871714770794\n",
      "Epoch: 599, Batch: 200, Avg. Loss: 0.022549715042114258\n",
      "Epoch: 599, Batch: 300, Avg. Loss: 0.03318224735558033\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 599, Batch: 400, Avg. Loss: 0.02919818624854088\n",
      "Epoch: 599, Batch: 500, Avg. Loss: 0.029692224264144897\n",
      "Epoch: 599, Batch: 600, Avg. Loss: 0.026598791405558587\n",
      "Epoch: 599, Batch: 700, Avg. Loss: 0.03306779384613037\n",
      "Epoch: 599, Batch: 800, Avg. Loss: 0.0257914125174284\n",
      "Epoch: 599, Batch: 900, Avg. Loss: 0.030098736435174942\n",
      "Epoch: 600, Batch: 100, Avg. Loss: 0.029181786328554154\n",
      "Epoch: 600, Batch: 200, Avg. Loss: 0.02616166666150093\n",
      "Epoch: 600, Batch: 300, Avg. Loss: 0.027542053312063216\n",
      "Epoch: 600, Batch: 400, Avg. Loss: 0.026425950825214387\n",
      "Epoch: 600, Batch: 500, Avg. Loss: 0.02378357619047165\n",
      "Epoch: 600, Batch: 600, Avg. Loss: 0.02717715822160244\n",
      "Epoch: 600, Batch: 700, Avg. Loss: 0.029935334846377372\n",
      "Epoch: 600, Batch: 800, Avg. Loss: 0.029331948906183243\n",
      "Epoch: 600, Batch: 900, Avg. Loss: 0.027979906275868417\n",
      "Epoch: 601, Batch: 100, Avg. Loss: 0.027221840173006058\n",
      "Epoch: 601, Batch: 200, Avg. Loss: 0.029695684164762495\n",
      "Epoch: 601, Batch: 300, Avg. Loss: 0.022711679935455323\n",
      "Epoch: 601, Batch: 400, Avg. Loss: 0.04345458082854748\n",
      "Epoch: 601, Batch: 500, Avg. Loss: 0.024266652390360834\n",
      "Epoch: 601, Batch: 600, Avg. Loss: 0.03178558081388474\n",
      "Epoch: 601, Batch: 700, Avg. Loss: 0.03059012696146965\n",
      "Epoch: 601, Batch: 800, Avg. Loss: 0.03193653978407383\n",
      "Epoch: 601, Batch: 900, Avg. Loss: 0.02077056050300598\n",
      "Epoch: 602, Batch: 100, Avg. Loss: 0.02653341755270958\n",
      "Epoch: 602, Batch: 200, Avg. Loss: 0.03473432831466198\n",
      "Epoch: 602, Batch: 300, Avg. Loss: 0.030664201602339746\n",
      "Epoch: 602, Batch: 400, Avg. Loss: 0.023839902207255362\n",
      "Epoch: 602, Batch: 500, Avg. Loss: 0.02890393413603306\n",
      "Epoch: 602, Batch: 600, Avg. Loss: 0.026753531694412233\n",
      "Epoch: 602, Batch: 700, Avg. Loss: 0.02756042830646038\n",
      "Epoch: 602, Batch: 800, Avg. Loss: 0.033400252163410186\n",
      "Epoch: 602, Batch: 900, Avg. Loss: 0.024776904881000518\n",
      "Epoch: 603, Batch: 100, Avg. Loss: 0.030131739526987077\n",
      "Epoch: 603, Batch: 200, Avg. Loss: 0.027190464064478875\n",
      "Epoch: 603, Batch: 300, Avg. Loss: 0.022048639953136443\n",
      "Epoch: 603, Batch: 400, Avg. Loss: 0.028363067358732223\n",
      "Epoch: 603, Batch: 500, Avg. Loss: 0.02535191610455513\n",
      "Epoch: 603, Batch: 600, Avg. Loss: 0.028681235909461974\n",
      "Epoch: 603, Batch: 700, Avg. Loss: 0.023985706567764283\n",
      "Epoch: 603, Batch: 800, Avg. Loss: 0.026360965445637704\n",
      "Epoch: 603, Batch: 900, Avg. Loss: 0.02720102496445179\n",
      "Epoch: 604, Batch: 100, Avg. Loss: 0.026786082312464713\n",
      "Epoch: 604, Batch: 200, Avg. Loss: 0.036468412652611734\n",
      "Epoch: 604, Batch: 300, Avg. Loss: 0.030450672209262848\n",
      "Epoch: 604, Batch: 400, Avg. Loss: 0.028740351423621177\n",
      "Epoch: 604, Batch: 500, Avg. Loss: 0.02492935411632061\n",
      "Epoch: 604, Batch: 600, Avg. Loss: 0.02329135127365589\n",
      "Epoch: 604, Batch: 700, Avg. Loss: 0.030602308735251427\n",
      "Epoch: 604, Batch: 800, Avg. Loss: 0.022265117987990378\n",
      "Epoch: 604, Batch: 900, Avg. Loss: 0.02900520145893097\n",
      "Epoch: 605, Batch: 100, Avg. Loss: 0.03166098229587078\n",
      "Epoch: 605, Batch: 200, Avg. Loss: 0.02277066610753536\n",
      "Epoch: 605, Batch: 300, Avg. Loss: 0.021495476961135865\n",
      "Epoch: 605, Batch: 400, Avg. Loss: 0.02906173437833786\n",
      "Epoch: 605, Batch: 500, Avg. Loss: 0.029493687078356742\n",
      "Epoch: 605, Batch: 600, Avg. Loss: 0.02940483368933201\n",
      "Epoch: 605, Batch: 700, Avg. Loss: 0.023931362926959993\n",
      "Epoch: 605, Batch: 800, Avg. Loss: 0.023690852969884872\n",
      "Epoch: 605, Batch: 900, Avg. Loss: 0.034658987000584604\n",
      "Epoch: 606, Batch: 100, Avg. Loss: 0.032684201449155806\n",
      "Epoch: 606, Batch: 200, Avg. Loss: 0.021846213787794114\n",
      "Epoch: 606, Batch: 300, Avg. Loss: 0.028492747098207472\n",
      "Epoch: 606, Batch: 400, Avg. Loss: 0.0241874560713768\n",
      "Epoch: 606, Batch: 500, Avg. Loss: 0.025976172164082528\n",
      "Epoch: 606, Batch: 600, Avg. Loss: 0.030876916199922562\n",
      "Epoch: 606, Batch: 700, Avg. Loss: 0.03116398960351944\n",
      "Epoch: 606, Batch: 800, Avg. Loss: 0.03473105490207672\n",
      "Epoch: 606, Batch: 900, Avg. Loss: 0.027308787256479262\n",
      "Epoch: 607, Batch: 100, Avg. Loss: 0.021169984489679338\n",
      "Epoch: 607, Batch: 200, Avg. Loss: 0.02518964812159538\n",
      "Epoch: 607, Batch: 300, Avg. Loss: 0.027757123932242393\n",
      "Epoch: 607, Batch: 400, Avg. Loss: 0.027677243426442146\n",
      "Epoch: 607, Batch: 500, Avg. Loss: 0.035318576991558076\n",
      "Epoch: 607, Batch: 600, Avg. Loss: 0.02534311555325985\n",
      "Epoch: 607, Batch: 700, Avg. Loss: 0.03265863493084908\n",
      "Epoch: 607, Batch: 800, Avg. Loss: 0.02897649735212326\n",
      "Epoch: 607, Batch: 900, Avg. Loss: 0.02990175113081932\n",
      "Epoch: 608, Batch: 100, Avg. Loss: 0.03197760261595249\n",
      "Epoch: 608, Batch: 200, Avg. Loss: 0.029466554820537567\n",
      "Epoch: 608, Batch: 300, Avg. Loss: 0.027351601347327233\n",
      "Epoch: 608, Batch: 400, Avg. Loss: 0.026038087159395217\n",
      "Epoch: 608, Batch: 500, Avg. Loss: 0.02653360053896904\n",
      "Epoch: 608, Batch: 600, Avg. Loss: 0.03296704903244972\n",
      "Epoch: 608, Batch: 700, Avg. Loss: 0.03334968410432339\n",
      "Epoch: 608, Batch: 800, Avg. Loss: 0.03091928854584694\n",
      "Epoch: 608, Batch: 900, Avg. Loss: 0.024989083260297775\n",
      "Epoch: 609, Batch: 100, Avg. Loss: 0.028176588118076325\n",
      "Epoch: 609, Batch: 200, Avg. Loss: 0.02777836449444294\n",
      "Epoch: 609, Batch: 300, Avg. Loss: 0.02418693706393242\n",
      "Epoch: 609, Batch: 400, Avg. Loss: 0.036434075683355334\n",
      "Epoch: 609, Batch: 500, Avg. Loss: 0.030812538266181945\n",
      "Epoch: 609, Batch: 600, Avg. Loss: 0.027768467217683793\n",
      "Epoch: 609, Batch: 700, Avg. Loss: 0.02686888463795185\n",
      "Epoch: 609, Batch: 800, Avg. Loss: 0.02482365682721138\n",
      "Epoch: 609, Batch: 900, Avg. Loss: 0.026491600200533868\n",
      "Epoch: 610, Batch: 100, Avg. Loss: 0.02867830127477646\n",
      "Epoch: 610, Batch: 200, Avg. Loss: 0.03268372938036919\n",
      "Epoch: 610, Batch: 300, Avg. Loss: 0.026218131929636002\n",
      "Epoch: 610, Batch: 400, Avg. Loss: 0.029507273361086845\n",
      "Epoch: 610, Batch: 500, Avg. Loss: 0.024803600758314132\n",
      "Epoch: 610, Batch: 600, Avg. Loss: 0.029947863891720772\n",
      "Epoch: 610, Batch: 700, Avg. Loss: 0.030039396584033966\n",
      "Epoch: 610, Batch: 800, Avg. Loss: 0.031006036847829817\n",
      "Epoch: 610, Batch: 900, Avg. Loss: 0.025986803993582726\n",
      "Epoch: 611, Batch: 100, Avg. Loss: 0.03097903564572334\n",
      "Epoch: 611, Batch: 200, Avg. Loss: 0.025617806166410448\n",
      "Epoch: 611, Batch: 300, Avg. Loss: 0.03319859191775322\n",
      "Epoch: 611, Batch: 400, Avg. Loss: 0.03591946430504322\n",
      "Epoch: 611, Batch: 500, Avg. Loss: 0.028051459640264512\n",
      "Epoch: 611, Batch: 600, Avg. Loss: 0.025604656487703322\n",
      "Epoch: 611, Batch: 700, Avg. Loss: 0.025497366338968278\n",
      "Epoch: 611, Batch: 800, Avg. Loss: 0.027549133971333503\n",
      "Epoch: 611, Batch: 900, Avg. Loss: 0.027805093973875046\n",
      "Epoch: 612, Batch: 100, Avg. Loss: 0.0269683787971735\n",
      "Epoch: 612, Batch: 200, Avg. Loss: 0.03318377673625946\n",
      "Epoch: 612, Batch: 300, Avg. Loss: 0.02727012075483799\n",
      "Epoch: 612, Batch: 400, Avg. Loss: 0.027354345172643662\n",
      "Epoch: 612, Batch: 500, Avg. Loss: 0.02560209684073925\n",
      "Epoch: 612, Batch: 600, Avg. Loss: 0.028671954944729804\n",
      "Epoch: 612, Batch: 700, Avg. Loss: 0.02780682422220707\n",
      "Epoch: 612, Batch: 800, Avg. Loss: 0.024174473360180856\n",
      "Epoch: 612, Batch: 900, Avg. Loss: 0.0295416534692049\n",
      "Epoch: 613, Batch: 100, Avg. Loss: 0.0293668445199728\n",
      "Epoch: 613, Batch: 200, Avg. Loss: 0.02891420565545559\n",
      "Epoch: 613, Batch: 300, Avg. Loss: 0.024068819358944893\n",
      "Epoch: 613, Batch: 400, Avg. Loss: 0.023364178016781808\n",
      "Epoch: 613, Batch: 500, Avg. Loss: 0.035902363136410716\n",
      "Epoch: 613, Batch: 600, Avg. Loss: 0.027059821635484694\n",
      "Epoch: 613, Batch: 700, Avg. Loss: 0.027521951496601103\n",
      "Epoch: 613, Batch: 800, Avg. Loss: 0.024445947110652923\n",
      "Epoch: 613, Batch: 900, Avg. Loss: 0.032626328356564045\n",
      "Epoch: 614, Batch: 100, Avg. Loss: 0.022055512517690657\n",
      "Epoch: 614, Batch: 200, Avg. Loss: 0.025330120101571083\n",
      "Epoch: 614, Batch: 300, Avg. Loss: 0.030428261756896974\n",
      "Epoch: 614, Batch: 400, Avg. Loss: 0.030778873637318613\n",
      "Epoch: 614, Batch: 500, Avg. Loss: 0.028312646001577377\n",
      "Epoch: 614, Batch: 600, Avg. Loss: 0.030163127332925796\n",
      "Epoch: 614, Batch: 700, Avg. Loss: 0.021941536813974382\n",
      "Epoch: 614, Batch: 800, Avg. Loss: 0.023621329218149186\n",
      "Epoch: 614, Batch: 900, Avg. Loss: 0.028438435047864915\n",
      "Epoch: 615, Batch: 100, Avg. Loss: 0.027354443818330763\n",
      "Epoch: 615, Batch: 200, Avg. Loss: 0.027289632111787795\n",
      "Epoch: 615, Batch: 300, Avg. Loss: 0.02955433540046215\n",
      "Epoch: 615, Batch: 400, Avg. Loss: 0.02476230274885893\n",
      "Epoch: 615, Batch: 500, Avg. Loss: 0.03261779703199864\n",
      "Epoch: 615, Batch: 600, Avg. Loss: 0.027987273558974266\n",
      "Epoch: 615, Batch: 700, Avg. Loss: 0.028253613486886025\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 615, Batch: 800, Avg. Loss: 0.026765456572175026\n",
      "Epoch: 615, Batch: 900, Avg. Loss: 0.03642466850578785\n",
      "Epoch: 616, Batch: 100, Avg. Loss: 0.025642205253243445\n",
      "Epoch: 616, Batch: 200, Avg. Loss: 0.027302045077085495\n",
      "Epoch: 616, Batch: 300, Avg. Loss: 0.030058636143803597\n",
      "Epoch: 616, Batch: 400, Avg. Loss: 0.029538885205984116\n",
      "Epoch: 616, Batch: 500, Avg. Loss: 0.026890962421894073\n",
      "Epoch: 616, Batch: 600, Avg. Loss: 0.022912297695875168\n",
      "Epoch: 616, Batch: 700, Avg. Loss: 0.02736638329923153\n",
      "Epoch: 616, Batch: 800, Avg. Loss: 0.028414441719651223\n",
      "Epoch: 616, Batch: 900, Avg. Loss: 0.027054904326796532\n",
      "Epoch: 617, Batch: 100, Avg. Loss: 0.02607742205262184\n",
      "Epoch: 617, Batch: 200, Avg. Loss: 0.026584959626197814\n",
      "Epoch: 617, Batch: 300, Avg. Loss: 0.029575267359614372\n",
      "Epoch: 617, Batch: 400, Avg. Loss: 0.025610632002353667\n",
      "Epoch: 617, Batch: 500, Avg. Loss: 0.024659272208809854\n",
      "Epoch: 617, Batch: 600, Avg. Loss: 0.028166565746068954\n",
      "Epoch: 617, Batch: 700, Avg. Loss: 0.02999515160918236\n",
      "Epoch: 617, Batch: 800, Avg. Loss: 0.030558800622820854\n",
      "Epoch: 617, Batch: 900, Avg. Loss: 0.026548254862427713\n",
      "Epoch: 618, Batch: 100, Avg. Loss: 0.0255817099660635\n",
      "Epoch: 618, Batch: 200, Avg. Loss: 0.030608993619680405\n",
      "Epoch: 618, Batch: 300, Avg. Loss: 0.024403051733970643\n",
      "Epoch: 618, Batch: 400, Avg. Loss: 0.03251614607870579\n",
      "Epoch: 618, Batch: 500, Avg. Loss: 0.01772721767425537\n",
      "Epoch: 618, Batch: 600, Avg. Loss: 0.023575875088572503\n",
      "Epoch: 618, Batch: 700, Avg. Loss: 0.023557915315032006\n",
      "Epoch: 618, Batch: 800, Avg. Loss: 0.03015803836286068\n",
      "Epoch: 618, Batch: 900, Avg. Loss: 0.027935076132416727\n",
      "Epoch: 619, Batch: 100, Avg. Loss: 0.031112350150942804\n",
      "Epoch: 619, Batch: 200, Avg. Loss: 0.024106613025069237\n",
      "Epoch: 619, Batch: 300, Avg. Loss: 0.03144814513623714\n",
      "Epoch: 619, Batch: 400, Avg. Loss: 0.03147916242480278\n",
      "Epoch: 619, Batch: 500, Avg. Loss: 0.02308506742119789\n",
      "Epoch: 619, Batch: 600, Avg. Loss: 0.026034668758511542\n",
      "Epoch: 619, Batch: 700, Avg. Loss: 0.03498272925615311\n",
      "Epoch: 619, Batch: 800, Avg. Loss: 0.027428659200668334\n",
      "Epoch: 619, Batch: 900, Avg. Loss: 0.021195952594280244\n",
      "Epoch: 620, Batch: 100, Avg. Loss: 0.025485649332404138\n",
      "Epoch: 620, Batch: 200, Avg. Loss: 0.03306408330798149\n",
      "Epoch: 620, Batch: 300, Avg. Loss: 0.024577915593981745\n",
      "Epoch: 620, Batch: 400, Avg. Loss: 0.020847924202680588\n",
      "Epoch: 620, Batch: 500, Avg. Loss: 0.03192204006016255\n",
      "Epoch: 620, Batch: 600, Avg. Loss: 0.027493482828140257\n",
      "Epoch: 620, Batch: 700, Avg. Loss: 0.029836153462529182\n",
      "Epoch: 620, Batch: 800, Avg. Loss: 0.03291245825588703\n",
      "Epoch: 620, Batch: 900, Avg. Loss: 0.028630847707390786\n",
      "Epoch: 621, Batch: 100, Avg. Loss: 0.03033129446208477\n",
      "Epoch: 621, Batch: 200, Avg. Loss: 0.021083140969276427\n",
      "Epoch: 621, Batch: 300, Avg. Loss: 0.0244053003937006\n",
      "Epoch: 621, Batch: 400, Avg. Loss: 0.024832784980535507\n",
      "Epoch: 621, Batch: 500, Avg. Loss: 0.02816882364451885\n",
      "Epoch: 621, Batch: 600, Avg. Loss: 0.027482334673404694\n",
      "Epoch: 621, Batch: 700, Avg. Loss: 0.026007295101881028\n",
      "Epoch: 621, Batch: 800, Avg. Loss: 0.02588342010974884\n",
      "Epoch: 621, Batch: 900, Avg. Loss: 0.023517208471894264\n",
      "Epoch: 622, Batch: 100, Avg. Loss: 0.03216198295354843\n",
      "Epoch: 622, Batch: 200, Avg. Loss: 0.029519217386841774\n",
      "Epoch: 622, Batch: 300, Avg. Loss: 0.02854544185101986\n",
      "Epoch: 622, Batch: 400, Avg. Loss: 0.028260094448924064\n",
      "Epoch: 622, Batch: 500, Avg. Loss: 0.02699075862765312\n",
      "Epoch: 622, Batch: 600, Avg. Loss: 0.02439112588763237\n",
      "Epoch: 622, Batch: 700, Avg. Loss: 0.02870983026921749\n",
      "Epoch: 622, Batch: 800, Avg. Loss: 0.028253879100084305\n",
      "Epoch: 622, Batch: 900, Avg. Loss: 0.0257497189193964\n",
      "Epoch: 623, Batch: 100, Avg. Loss: 0.028215891569852828\n",
      "Epoch: 623, Batch: 200, Avg. Loss: 0.02018530398607254\n",
      "Epoch: 623, Batch: 300, Avg. Loss: 0.02951985478401184\n",
      "Epoch: 623, Batch: 400, Avg. Loss: 0.026482779532670975\n",
      "Epoch: 623, Batch: 500, Avg. Loss: 0.02652601197361946\n",
      "Epoch: 623, Batch: 600, Avg. Loss: 0.028266552165150644\n",
      "Epoch: 623, Batch: 700, Avg. Loss: 0.031194243133068084\n",
      "Epoch: 623, Batch: 800, Avg. Loss: 0.03289268121123314\n",
      "Epoch: 623, Batch: 900, Avg. Loss: 0.026193927228450774\n",
      "Epoch: 624, Batch: 100, Avg. Loss: 0.026707442849874495\n",
      "Epoch: 624, Batch: 200, Avg. Loss: 0.02223054476082325\n",
      "Epoch: 624, Batch: 300, Avg. Loss: 0.03284687541425228\n",
      "Epoch: 624, Batch: 400, Avg. Loss: 0.026921792551875115\n",
      "Epoch: 624, Batch: 500, Avg. Loss: 0.021658583655953408\n",
      "Epoch: 624, Batch: 600, Avg. Loss: 0.028791968300938605\n",
      "Epoch: 624, Batch: 700, Avg. Loss: 0.02737943798303604\n",
      "Epoch: 624, Batch: 800, Avg. Loss: 0.028850417360663416\n",
      "Epoch: 624, Batch: 900, Avg. Loss: 0.021860000491142274\n",
      "Epoch: 625, Batch: 100, Avg. Loss: 0.029697530940175056\n",
      "Epoch: 625, Batch: 200, Avg. Loss: 0.0315509844571352\n",
      "Epoch: 625, Batch: 300, Avg. Loss: 0.027711728140711784\n",
      "Epoch: 625, Batch: 400, Avg. Loss: 0.021117969527840615\n",
      "Epoch: 625, Batch: 500, Avg. Loss: 0.026599749326705932\n",
      "Epoch: 625, Batch: 600, Avg. Loss: 0.0267802956700325\n",
      "Epoch: 625, Batch: 700, Avg. Loss: 0.022590250968933107\n",
      "Epoch: 625, Batch: 800, Avg. Loss: 0.02007731169462204\n",
      "Epoch: 625, Batch: 900, Avg. Loss: 0.030450524315238\n",
      "Epoch: 626, Batch: 100, Avg. Loss: 0.025670821368694304\n",
      "Epoch: 626, Batch: 200, Avg. Loss: 0.01981472074985504\n",
      "Epoch: 626, Batch: 300, Avg. Loss: 0.0324191677570343\n",
      "Epoch: 626, Batch: 400, Avg. Loss: 0.034477443173527716\n",
      "Epoch: 626, Batch: 500, Avg. Loss: 0.024634883627295493\n",
      "Epoch: 626, Batch: 600, Avg. Loss: 0.028035776466131212\n",
      "Epoch: 626, Batch: 700, Avg. Loss: 0.025820604488253594\n",
      "Epoch: 626, Batch: 800, Avg. Loss: 0.023981526494026184\n",
      "Epoch: 626, Batch: 900, Avg. Loss: 0.025001362487673758\n",
      "Epoch: 627, Batch: 100, Avg. Loss: 0.03212670296430588\n",
      "Epoch: 627, Batch: 200, Avg. Loss: 0.027614753544330597\n",
      "Epoch: 627, Batch: 300, Avg. Loss: 0.029637902975082397\n",
      "Epoch: 627, Batch: 400, Avg. Loss: 0.02845651052892208\n",
      "Epoch: 627, Batch: 500, Avg. Loss: 0.025388979464769364\n",
      "Epoch: 627, Batch: 600, Avg. Loss: 0.028945960253477097\n",
      "Epoch: 627, Batch: 700, Avg. Loss: 0.028937004432082178\n",
      "Epoch: 627, Batch: 800, Avg. Loss: 0.02487448535859585\n",
      "Epoch: 627, Batch: 900, Avg. Loss: 0.029540897831320762\n",
      "Epoch: 628, Batch: 100, Avg. Loss: 0.02702718213200569\n",
      "Epoch: 628, Batch: 200, Avg. Loss: 0.035120216086506846\n",
      "Epoch: 628, Batch: 300, Avg. Loss: 0.027133211418986322\n",
      "Epoch: 628, Batch: 400, Avg. Loss: 0.02427327185869217\n",
      "Epoch: 628, Batch: 500, Avg. Loss: 0.03128934070467949\n",
      "Epoch: 628, Batch: 600, Avg. Loss: 0.03060633733868599\n",
      "Epoch: 628, Batch: 700, Avg. Loss: 0.02790574572980404\n",
      "Epoch: 628, Batch: 800, Avg. Loss: 0.02265461169183254\n",
      "Epoch: 628, Batch: 900, Avg. Loss: 0.026480114683508874\n",
      "Epoch: 629, Batch: 100, Avg. Loss: 0.020910696983337404\n",
      "Epoch: 629, Batch: 200, Avg. Loss: 0.02777896173298359\n",
      "Epoch: 629, Batch: 300, Avg. Loss: 0.026910333558917046\n",
      "Epoch: 629, Batch: 400, Avg. Loss: 0.026107994988560677\n",
      "Epoch: 629, Batch: 500, Avg. Loss: 0.031592746749520305\n",
      "Epoch: 629, Batch: 600, Avg. Loss: 0.028937380239367486\n",
      "Epoch: 629, Batch: 700, Avg. Loss: 0.029216049537062645\n",
      "Epoch: 629, Batch: 800, Avg. Loss: 0.025830725207924843\n",
      "Epoch: 629, Batch: 900, Avg. Loss: 0.02792051397264004\n",
      "Epoch: 630, Batch: 100, Avg. Loss: 0.023349208906292915\n",
      "Epoch: 630, Batch: 200, Avg. Loss: 0.03122978761792183\n",
      "Epoch: 630, Batch: 300, Avg. Loss: 0.023176162764430047\n",
      "Epoch: 630, Batch: 400, Avg. Loss: 0.029245654568076134\n",
      "Epoch: 630, Batch: 500, Avg. Loss: 0.021286086887121202\n",
      "Epoch: 630, Batch: 600, Avg. Loss: 0.027806039527058603\n",
      "Epoch: 630, Batch: 700, Avg. Loss: 0.02371081933379173\n",
      "Epoch: 630, Batch: 800, Avg. Loss: 0.030317734628915786\n",
      "Epoch: 630, Batch: 900, Avg. Loss: 0.02383848436176777\n",
      "Epoch: 631, Batch: 100, Avg. Loss: 0.030058903843164442\n",
      "Epoch: 631, Batch: 200, Avg. Loss: 0.032494087368249894\n",
      "Epoch: 631, Batch: 300, Avg. Loss: 0.03336049228906632\n",
      "Epoch: 631, Batch: 400, Avg. Loss: 0.031398880258202554\n",
      "Epoch: 631, Batch: 500, Avg. Loss: 0.02295957237482071\n",
      "Epoch: 631, Batch: 600, Avg. Loss: 0.029404551684856416\n",
      "Epoch: 631, Batch: 700, Avg. Loss: 0.023768966645002367\n",
      "Epoch: 631, Batch: 800, Avg. Loss: 0.025339993238449095\n",
      "Epoch: 631, Batch: 900, Avg. Loss: 0.026274649649858473\n",
      "Epoch: 632, Batch: 100, Avg. Loss: 0.02529973439872265\n",
      "Epoch: 632, Batch: 200, Avg. Loss: 0.02630303017795086\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 632, Batch: 300, Avg. Loss: 0.025766348391771318\n",
      "Epoch: 632, Batch: 400, Avg. Loss: 0.032083984613418576\n",
      "Epoch: 632, Batch: 500, Avg. Loss: 0.025013999789953233\n",
      "Epoch: 632, Batch: 600, Avg. Loss: 0.033782048746943474\n",
      "Epoch: 632, Batch: 700, Avg. Loss: 0.030376971662044526\n",
      "Epoch: 632, Batch: 800, Avg. Loss: 0.02780155047774315\n",
      "Epoch: 632, Batch: 900, Avg. Loss: 0.027407198920845987\n",
      "Epoch: 633, Batch: 100, Avg. Loss: 0.030963357090950012\n",
      "Epoch: 633, Batch: 200, Avg. Loss: 0.02488440915942192\n",
      "Epoch: 633, Batch: 300, Avg. Loss: 0.027265112102031707\n",
      "Epoch: 633, Batch: 400, Avg. Loss: 0.028803986459970475\n",
      "Epoch: 633, Batch: 500, Avg. Loss: 0.027182670906186104\n",
      "Epoch: 633, Batch: 600, Avg. Loss: 0.01989070899784565\n",
      "Epoch: 633, Batch: 700, Avg. Loss: 0.029382904022932053\n",
      "Epoch: 633, Batch: 800, Avg. Loss: 0.029401098042726517\n",
      "Epoch: 633, Batch: 900, Avg. Loss: 0.029113216400146483\n",
      "Epoch: 634, Batch: 100, Avg. Loss: 0.024417023211717605\n",
      "Epoch: 634, Batch: 200, Avg. Loss: 0.024219587817788123\n",
      "Epoch: 634, Batch: 300, Avg. Loss: 0.028249483555555344\n",
      "Epoch: 634, Batch: 400, Avg. Loss: 0.02706141471862793\n",
      "Epoch: 634, Batch: 500, Avg. Loss: 0.021391047462821007\n",
      "Epoch: 634, Batch: 600, Avg. Loss: 0.026063152849674226\n",
      "Epoch: 634, Batch: 700, Avg. Loss: 0.026964729502797127\n",
      "Epoch: 634, Batch: 800, Avg. Loss: 0.03105187214910984\n",
      "Epoch: 634, Batch: 900, Avg. Loss: 0.03337916031479835\n",
      "Epoch: 635, Batch: 100, Avg. Loss: 0.025657086819410323\n",
      "Epoch: 635, Batch: 200, Avg. Loss: 0.02586477801203728\n",
      "Epoch: 635, Batch: 300, Avg. Loss: 0.02550323501229286\n",
      "Epoch: 635, Batch: 400, Avg. Loss: 0.019296917766332626\n",
      "Epoch: 635, Batch: 500, Avg. Loss: 0.023569442629814148\n",
      "Epoch: 635, Batch: 600, Avg. Loss: 0.028241533190011978\n",
      "Epoch: 635, Batch: 700, Avg. Loss: 0.022699891924858093\n",
      "Epoch: 635, Batch: 800, Avg. Loss: 0.02631638579070568\n",
      "Epoch: 635, Batch: 900, Avg. Loss: 0.02898659698665142\n",
      "Epoch: 636, Batch: 100, Avg. Loss: 0.026492148637771606\n",
      "Epoch: 636, Batch: 200, Avg. Loss: 0.02926372900605202\n",
      "Epoch: 636, Batch: 300, Avg. Loss: 0.029231848418712614\n",
      "Epoch: 636, Batch: 400, Avg. Loss: 0.026138069555163384\n",
      "Epoch: 636, Batch: 500, Avg. Loss: 0.030831628516316413\n",
      "Epoch: 636, Batch: 600, Avg. Loss: 0.033716404512524605\n",
      "Epoch: 636, Batch: 700, Avg. Loss: 0.024101669937372207\n",
      "Epoch: 636, Batch: 800, Avg. Loss: 0.022465581148862837\n",
      "Epoch: 636, Batch: 900, Avg. Loss: 0.022959219440817833\n",
      "Epoch: 637, Batch: 100, Avg. Loss: 0.0189237754791975\n",
      "Epoch: 637, Batch: 200, Avg. Loss: 0.02474369779229164\n",
      "Epoch: 637, Batch: 300, Avg. Loss: 0.024394412413239478\n",
      "Epoch: 637, Batch: 400, Avg. Loss: 0.02925286702811718\n",
      "Epoch: 637, Batch: 500, Avg. Loss: 0.021876528039574624\n",
      "Epoch: 637, Batch: 600, Avg. Loss: 0.02178673379123211\n",
      "Epoch: 637, Batch: 700, Avg. Loss: 0.03337780110538006\n",
      "Epoch: 637, Batch: 800, Avg. Loss: 0.026628243029117583\n",
      "Epoch: 637, Batch: 900, Avg. Loss: 0.023841309770941736\n",
      "Epoch: 638, Batch: 100, Avg. Loss: 0.024102553352713586\n",
      "Epoch: 638, Batch: 200, Avg. Loss: 0.026987262666225434\n",
      "Epoch: 638, Batch: 300, Avg. Loss: 0.03043118752539158\n",
      "Epoch: 638, Batch: 400, Avg. Loss: 0.02350166633725166\n",
      "Epoch: 638, Batch: 500, Avg. Loss: 0.029484969079494477\n",
      "Epoch: 638, Batch: 600, Avg. Loss: 0.02749176263809204\n",
      "Epoch: 638, Batch: 700, Avg. Loss: 0.026514938548207283\n",
      "Epoch: 638, Batch: 800, Avg. Loss: 0.025235311686992647\n",
      "Epoch: 638, Batch: 900, Avg. Loss: 0.025661265030503273\n",
      "Epoch: 639, Batch: 100, Avg. Loss: 0.022150296717882156\n",
      "Epoch: 639, Batch: 200, Avg. Loss: 0.026370649486780168\n",
      "Epoch: 639, Batch: 300, Avg. Loss: 0.02489735245704651\n",
      "Epoch: 639, Batch: 400, Avg. Loss: 0.025814466401934622\n",
      "Epoch: 639, Batch: 500, Avg. Loss: 0.026647182032465935\n",
      "Epoch: 639, Batch: 600, Avg. Loss: 0.025471862778067587\n",
      "Epoch: 639, Batch: 700, Avg. Loss: 0.02716910161077976\n",
      "Epoch: 639, Batch: 800, Avg. Loss: 0.024881828501820566\n",
      "Epoch: 639, Batch: 900, Avg. Loss: 0.029868170469999313\n",
      "Epoch: 640, Batch: 100, Avg. Loss: 0.0275517800450325\n",
      "Epoch: 640, Batch: 200, Avg. Loss: 0.02830693542957306\n",
      "Epoch: 640, Batch: 300, Avg. Loss: 0.030847697034478188\n",
      "Epoch: 640, Batch: 400, Avg. Loss: 0.02164252385497093\n",
      "Epoch: 640, Batch: 500, Avg. Loss: 0.024933815374970435\n",
      "Epoch: 640, Batch: 600, Avg. Loss: 0.029194114431738853\n",
      "Epoch: 640, Batch: 700, Avg. Loss: 0.02335516668856144\n",
      "Epoch: 640, Batch: 800, Avg. Loss: 0.03327245332300663\n",
      "Epoch: 640, Batch: 900, Avg. Loss: 0.02472022354602814\n",
      "Epoch: 641, Batch: 100, Avg. Loss: 0.029203558415174483\n",
      "Epoch: 641, Batch: 200, Avg. Loss: 0.032865115106105805\n",
      "Epoch: 641, Batch: 300, Avg. Loss: 0.027224379554390908\n",
      "Epoch: 641, Batch: 400, Avg. Loss: 0.02390338458120823\n",
      "Epoch: 641, Batch: 500, Avg. Loss: 0.026603591069579123\n",
      "Epoch: 641, Batch: 600, Avg. Loss: 0.021995594054460527\n",
      "Epoch: 641, Batch: 700, Avg. Loss: 0.02777522034943104\n",
      "Epoch: 641, Batch: 800, Avg. Loss: 0.025554498881101607\n",
      "Epoch: 641, Batch: 900, Avg. Loss: 0.02467106580734253\n",
      "Epoch: 642, Batch: 100, Avg. Loss: 0.031483508944511414\n",
      "Epoch: 642, Batch: 200, Avg. Loss: 0.026799520850181578\n",
      "Epoch: 642, Batch: 300, Avg. Loss: 0.025949408411979676\n",
      "Epoch: 642, Batch: 400, Avg. Loss: 0.02100601777434349\n",
      "Epoch: 642, Batch: 500, Avg. Loss: 0.026550009027123452\n",
      "Epoch: 642, Batch: 600, Avg. Loss: 0.027966158613562584\n",
      "Epoch: 642, Batch: 700, Avg. Loss: 0.02549702934920788\n",
      "Epoch: 642, Batch: 800, Avg. Loss: 0.0284287279099226\n",
      "Epoch: 642, Batch: 900, Avg. Loss: 0.02304526314139366\n",
      "Epoch: 643, Batch: 100, Avg. Loss: 0.025400773137807846\n",
      "Epoch: 643, Batch: 200, Avg. Loss: 0.024990447536110877\n",
      "Epoch: 643, Batch: 300, Avg. Loss: 0.026499992460012434\n",
      "Epoch: 643, Batch: 400, Avg. Loss: 0.022462586611509322\n",
      "Epoch: 643, Batch: 500, Avg. Loss: 0.02753636248409748\n",
      "Epoch: 643, Batch: 600, Avg. Loss: 0.025273825600743295\n",
      "Epoch: 643, Batch: 700, Avg. Loss: 0.02606905058026314\n",
      "Epoch: 643, Batch: 800, Avg. Loss: 0.02548053801059723\n",
      "Epoch: 643, Batch: 900, Avg. Loss: 0.022523748129606246\n",
      "Epoch: 644, Batch: 100, Avg. Loss: 0.023880693912506103\n",
      "Epoch: 644, Batch: 200, Avg. Loss: 0.027877611741423607\n",
      "Epoch: 644, Batch: 300, Avg. Loss: 0.0308281784504652\n",
      "Epoch: 644, Batch: 400, Avg. Loss: 0.031500013694167134\n",
      "Epoch: 644, Batch: 500, Avg. Loss: 0.022551574781537055\n",
      "Epoch: 644, Batch: 600, Avg. Loss: 0.026495252922177314\n",
      "Epoch: 644, Batch: 700, Avg. Loss: 0.022835295572876932\n",
      "Epoch: 644, Batch: 800, Avg. Loss: 0.02353290729224682\n",
      "Epoch: 644, Batch: 900, Avg. Loss: 0.026980381906032563\n",
      "Epoch: 645, Batch: 100, Avg. Loss: 0.026616910099983217\n",
      "Epoch: 645, Batch: 200, Avg. Loss: 0.023029367849230768\n",
      "Epoch: 645, Batch: 300, Avg. Loss: 0.03245782032608986\n",
      "Epoch: 645, Batch: 400, Avg. Loss: 0.023080098778009414\n",
      "Epoch: 645, Batch: 500, Avg. Loss: 0.0257495629042387\n",
      "Epoch: 645, Batch: 600, Avg. Loss: 0.02983134999871254\n",
      "Epoch: 645, Batch: 700, Avg. Loss: 0.02502017542719841\n",
      "Epoch: 645, Batch: 800, Avg. Loss: 0.025374042391777037\n",
      "Epoch: 645, Batch: 900, Avg. Loss: 0.03068834938108921\n",
      "Epoch: 646, Batch: 100, Avg. Loss: 0.022563521042466162\n",
      "Epoch: 646, Batch: 200, Avg. Loss: 0.02249116562306881\n",
      "Epoch: 646, Batch: 300, Avg. Loss: 0.024433679431676864\n",
      "Epoch: 646, Batch: 400, Avg. Loss: 0.02491523787379265\n",
      "Epoch: 646, Batch: 500, Avg. Loss: 0.029086376428604125\n",
      "Epoch: 646, Batch: 600, Avg. Loss: 0.025871062725782393\n",
      "Epoch: 646, Batch: 700, Avg. Loss: 0.027460660189390182\n",
      "Epoch: 646, Batch: 800, Avg. Loss: 0.03456309996545315\n",
      "Epoch: 646, Batch: 900, Avg. Loss: 0.022335956022143364\n",
      "Epoch: 647, Batch: 100, Avg. Loss: 0.02288514479994774\n",
      "Epoch: 647, Batch: 200, Avg. Loss: 0.030799338296055793\n",
      "Epoch: 647, Batch: 300, Avg. Loss: 0.026693560034036636\n",
      "Epoch: 647, Batch: 400, Avg. Loss: 0.022391255497932434\n",
      "Epoch: 647, Batch: 500, Avg. Loss: 0.02851049244403839\n",
      "Epoch: 647, Batch: 600, Avg. Loss: 0.025711251720786095\n",
      "Epoch: 647, Batch: 700, Avg. Loss: 0.027139083407819273\n",
      "Epoch: 647, Batch: 800, Avg. Loss: 0.030778588205575944\n",
      "Epoch: 647, Batch: 900, Avg. Loss: 0.02526620663702488\n",
      "Epoch: 648, Batch: 100, Avg. Loss: 0.02340764008462429\n",
      "Epoch: 648, Batch: 200, Avg. Loss: 0.02994488537311554\n",
      "Epoch: 648, Batch: 300, Avg. Loss: 0.023976678252220152\n",
      "Epoch: 648, Batch: 400, Avg. Loss: 0.022522559836506844\n",
      "Epoch: 648, Batch: 500, Avg. Loss: 0.032657437324523926\n",
      "Epoch: 648, Batch: 600, Avg. Loss: 0.038134303018450735\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 648, Batch: 700, Avg. Loss: 0.023837864622473715\n",
      "Epoch: 648, Batch: 800, Avg. Loss: 0.027599300742149353\n",
      "Epoch: 648, Batch: 900, Avg. Loss: 0.021990585401654245\n",
      "Epoch: 649, Batch: 100, Avg. Loss: 0.025808074250817298\n",
      "Epoch: 649, Batch: 200, Avg. Loss: 0.026811155751347543\n",
      "Epoch: 649, Batch: 300, Avg. Loss: 0.024552328586578367\n",
      "Epoch: 649, Batch: 400, Avg. Loss: 0.026223218813538552\n",
      "Epoch: 649, Batch: 500, Avg. Loss: 0.02527888648211956\n",
      "Epoch: 649, Batch: 600, Avg. Loss: 0.020117087513208388\n",
      "Epoch: 649, Batch: 700, Avg. Loss: 0.028810950517654418\n",
      "Epoch: 649, Batch: 800, Avg. Loss: 0.025909959524869918\n",
      "Epoch: 649, Batch: 900, Avg. Loss: 0.03183477483689785\n",
      "Epoch: 650, Batch: 100, Avg. Loss: 0.031323141008615496\n",
      "Epoch: 650, Batch: 200, Avg. Loss: 0.025608238652348517\n",
      "Epoch: 650, Batch: 300, Avg. Loss: 0.026578460931777955\n",
      "Epoch: 650, Batch: 400, Avg. Loss: 0.025737779513001442\n",
      "Epoch: 650, Batch: 500, Avg. Loss: 0.023828845396637918\n",
      "Epoch: 650, Batch: 600, Avg. Loss: 0.027742114290595053\n",
      "Epoch: 650, Batch: 700, Avg. Loss: 0.02599794737994671\n",
      "Epoch: 650, Batch: 800, Avg. Loss: 0.028869897499680518\n",
      "Epoch: 650, Batch: 900, Avg. Loss: 0.025041544064879417\n",
      "Epoch: 651, Batch: 100, Avg. Loss: 0.02815155506134033\n",
      "Epoch: 651, Batch: 200, Avg. Loss: 0.02464042529463768\n",
      "Epoch: 651, Batch: 300, Avg. Loss: 0.025146198719739915\n",
      "Epoch: 651, Batch: 400, Avg. Loss: 0.0232229732722044\n",
      "Epoch: 651, Batch: 500, Avg. Loss: 0.02641231134533882\n",
      "Epoch: 651, Batch: 600, Avg. Loss: 0.02715377852320671\n",
      "Epoch: 651, Batch: 700, Avg. Loss: 0.03224737539887428\n",
      "Epoch: 651, Batch: 800, Avg. Loss: 0.024614128693938257\n",
      "Epoch: 651, Batch: 900, Avg. Loss: 0.02441770501434803\n",
      "Epoch: 652, Batch: 100, Avg. Loss: 0.02805599056184292\n",
      "Epoch: 652, Batch: 200, Avg. Loss: 0.028628739714622497\n",
      "Epoch: 652, Batch: 300, Avg. Loss: 0.028438953906297682\n",
      "Epoch: 652, Batch: 400, Avg. Loss: 0.02878532513976097\n",
      "Epoch: 652, Batch: 500, Avg. Loss: 0.023556646332144736\n",
      "Epoch: 652, Batch: 600, Avg. Loss: 0.02958934284746647\n",
      "Epoch: 652, Batch: 700, Avg. Loss: 0.026859201267361642\n",
      "Epoch: 652, Batch: 800, Avg. Loss: 0.026174262762069703\n",
      "Epoch: 652, Batch: 900, Avg. Loss: 0.029844221621751786\n",
      "Epoch: 653, Batch: 100, Avg. Loss: 0.02985086776316166\n",
      "Epoch: 653, Batch: 200, Avg. Loss: 0.026434117406606675\n",
      "Epoch: 653, Batch: 300, Avg. Loss: 0.024857423081994057\n",
      "Epoch: 653, Batch: 400, Avg. Loss: 0.028403124064207075\n",
      "Epoch: 653, Batch: 500, Avg. Loss: 0.028837514221668244\n",
      "Epoch: 653, Batch: 600, Avg. Loss: 0.02386870078742504\n",
      "Epoch: 653, Batch: 700, Avg. Loss: 0.023555608615279198\n",
      "Epoch: 653, Batch: 800, Avg. Loss: 0.02722449742257595\n",
      "Epoch: 653, Batch: 900, Avg. Loss: 0.022104099839925766\n",
      "Epoch: 654, Batch: 100, Avg. Loss: 0.02322113364934921\n",
      "Epoch: 654, Batch: 200, Avg. Loss: 0.026610418483614923\n",
      "Epoch: 654, Batch: 300, Avg. Loss: 0.02465078055858612\n",
      "Epoch: 654, Batch: 400, Avg. Loss: 0.026352261900901796\n",
      "Epoch: 654, Batch: 500, Avg. Loss: 0.027551848664879797\n",
      "Epoch: 654, Batch: 600, Avg. Loss: 0.022524767592549323\n",
      "Epoch: 654, Batch: 700, Avg. Loss: 0.023409415706992148\n",
      "Epoch: 654, Batch: 800, Avg. Loss: 0.026998924762010573\n",
      "Epoch: 654, Batch: 900, Avg. Loss: 0.021816778257489206\n",
      "Epoch: 655, Batch: 100, Avg. Loss: 0.0236642587184906\n",
      "Epoch: 655, Batch: 200, Avg. Loss: 0.02778027705848217\n",
      "Epoch: 655, Batch: 300, Avg. Loss: 0.027835747227072716\n",
      "Epoch: 655, Batch: 400, Avg. Loss: 0.030005476474761962\n",
      "Epoch: 655, Batch: 500, Avg. Loss: 0.029648753926157952\n",
      "Epoch: 655, Batch: 600, Avg. Loss: 0.023664964959025384\n",
      "Epoch: 655, Batch: 700, Avg. Loss: 0.03614574357867241\n",
      "Epoch: 655, Batch: 800, Avg. Loss: 0.022319697216153143\n",
      "Epoch: 655, Batch: 900, Avg. Loss: 0.028159278631210326\n",
      "Epoch: 656, Batch: 100, Avg. Loss: 0.024095937013626098\n",
      "Epoch: 656, Batch: 200, Avg. Loss: 0.02739305630326271\n",
      "Epoch: 656, Batch: 300, Avg. Loss: 0.026375777572393417\n",
      "Epoch: 656, Batch: 400, Avg. Loss: 0.026475282311439516\n",
      "Epoch: 656, Batch: 500, Avg. Loss: 0.02334020793437958\n",
      "Epoch: 656, Batch: 600, Avg. Loss: 0.024394283443689345\n",
      "Epoch: 656, Batch: 700, Avg. Loss: 0.030600211918354034\n",
      "Epoch: 656, Batch: 800, Avg. Loss: 0.02392185665667057\n",
      "Epoch: 656, Batch: 900, Avg. Loss: 0.026291917935013773\n",
      "Epoch: 657, Batch: 100, Avg. Loss: 0.033051461726427075\n",
      "Epoch: 657, Batch: 200, Avg. Loss: 0.028434285894036293\n",
      "Epoch: 657, Batch: 300, Avg. Loss: 0.02414589822292328\n",
      "Epoch: 657, Batch: 400, Avg. Loss: 0.03125741228461266\n",
      "Epoch: 657, Batch: 500, Avg. Loss: 0.024881891384720804\n",
      "Epoch: 657, Batch: 600, Avg. Loss: 0.02541155643761158\n",
      "Epoch: 657, Batch: 700, Avg. Loss: 0.023146134689450263\n",
      "Epoch: 657, Batch: 800, Avg. Loss: 0.028116578161716463\n",
      "Epoch: 657, Batch: 900, Avg. Loss: 0.025897122025489806\n",
      "Epoch: 658, Batch: 100, Avg. Loss: 0.030007056891918182\n",
      "Epoch: 658, Batch: 200, Avg. Loss: 0.02730788439512253\n",
      "Epoch: 658, Batch: 300, Avg. Loss: 0.03192068204283714\n",
      "Epoch: 658, Batch: 400, Avg. Loss: 0.024000832289457322\n",
      "Epoch: 658, Batch: 500, Avg. Loss: 0.022461322844028474\n",
      "Epoch: 658, Batch: 600, Avg. Loss: 0.026904080286622046\n",
      "Epoch: 658, Batch: 700, Avg. Loss: 0.019373844414949416\n",
      "Epoch: 658, Batch: 800, Avg. Loss: 0.02834515109658241\n",
      "Epoch: 658, Batch: 900, Avg. Loss: 0.030759676098823546\n",
      "Epoch: 659, Batch: 100, Avg. Loss: 0.023824896216392517\n",
      "Epoch: 659, Batch: 200, Avg. Loss: 0.028129548579454423\n",
      "Epoch: 659, Batch: 300, Avg. Loss: 0.023551265224814415\n",
      "Epoch: 659, Batch: 400, Avg. Loss: 0.02255709417164326\n",
      "Epoch: 659, Batch: 500, Avg. Loss: 0.028530950248241423\n",
      "Epoch: 659, Batch: 600, Avg. Loss: 0.029513564705848695\n",
      "Epoch: 659, Batch: 700, Avg. Loss: 0.026291630007326604\n",
      "Epoch: 659, Batch: 800, Avg. Loss: 0.02613538734614849\n",
      "Epoch: 659, Batch: 900, Avg. Loss: 0.028896868973970414\n",
      "Epoch: 660, Batch: 100, Avg. Loss: 0.026291972622275353\n",
      "Epoch: 660, Batch: 200, Avg. Loss: 0.0266705409437418\n",
      "Epoch: 660, Batch: 300, Avg. Loss: 0.024134547635912895\n",
      "Epoch: 660, Batch: 400, Avg. Loss: 0.026559703573584555\n",
      "Epoch: 660, Batch: 500, Avg. Loss: 0.026202402859926224\n",
      "Epoch: 660, Batch: 600, Avg. Loss: 0.02517413042485714\n",
      "Epoch: 660, Batch: 700, Avg. Loss: 0.029169014617800713\n",
      "Epoch: 660, Batch: 800, Avg. Loss: 0.028817925304174423\n",
      "Epoch: 660, Batch: 900, Avg. Loss: 0.029254143089056016\n",
      "Epoch: 661, Batch: 100, Avg. Loss: 0.020147605761885644\n",
      "Epoch: 661, Batch: 200, Avg. Loss: 0.027494501248002054\n",
      "Epoch: 661, Batch: 300, Avg. Loss: 0.02749916344881058\n",
      "Epoch: 661, Batch: 400, Avg. Loss: 0.027426482737064363\n",
      "Epoch: 661, Batch: 500, Avg. Loss: 0.024155417755246163\n",
      "Epoch: 661, Batch: 600, Avg. Loss: 0.029402056187391282\n",
      "Epoch: 661, Batch: 700, Avg. Loss: 0.026207565069198607\n",
      "Epoch: 661, Batch: 800, Avg. Loss: 0.03325360104441643\n",
      "Epoch: 661, Batch: 900, Avg. Loss: 0.02565589062869549\n",
      "Epoch: 662, Batch: 100, Avg. Loss: 0.026300661265850067\n",
      "Epoch: 662, Batch: 200, Avg. Loss: 0.027786933928728105\n",
      "Epoch: 662, Batch: 300, Avg. Loss: 0.023798018768429754\n",
      "Epoch: 662, Batch: 400, Avg. Loss: 0.02854891411960125\n",
      "Epoch: 662, Batch: 500, Avg. Loss: 0.022567533999681474\n",
      "Epoch: 662, Batch: 600, Avg. Loss: 0.018614285811781883\n",
      "Epoch: 662, Batch: 700, Avg. Loss: 0.024118232280015944\n",
      "Epoch: 662, Batch: 800, Avg. Loss: 0.024830236211419107\n",
      "Epoch: 662, Batch: 900, Avg. Loss: 0.027047649323940277\n",
      "Epoch: 663, Batch: 100, Avg. Loss: 0.026910841166973114\n",
      "Epoch: 663, Batch: 200, Avg. Loss: 0.021437639817595482\n",
      "Epoch: 663, Batch: 300, Avg. Loss: 0.020733631178736688\n",
      "Epoch: 663, Batch: 400, Avg. Loss: 0.025753360092639923\n",
      "Epoch: 663, Batch: 500, Avg. Loss: 0.02727675102651119\n",
      "Epoch: 663, Batch: 600, Avg. Loss: 0.028194860368967057\n",
      "Epoch: 663, Batch: 700, Avg. Loss: 0.03016999363899231\n",
      "Epoch: 663, Batch: 800, Avg. Loss: 0.027035056874156\n",
      "Epoch: 663, Batch: 900, Avg. Loss: 0.026297794431447984\n",
      "Epoch: 664, Batch: 100, Avg. Loss: 0.028591791316866875\n",
      "Epoch: 664, Batch: 200, Avg. Loss: 0.027902086526155473\n",
      "Epoch: 664, Batch: 300, Avg. Loss: 0.0279580669850111\n",
      "Epoch: 664, Batch: 400, Avg. Loss: 0.02625427931547165\n",
      "Epoch: 664, Batch: 500, Avg. Loss: 0.022653576880693437\n",
      "Epoch: 664, Batch: 600, Avg. Loss: 0.02759973779320717\n",
      "Epoch: 664, Batch: 700, Avg. Loss: 0.019496286436915398\n",
      "Epoch: 664, Batch: 800, Avg. Loss: 0.02364652819931507\n",
      "Epoch: 664, Batch: 900, Avg. Loss: 0.024265113323926925\n",
      "Epoch: 665, Batch: 100, Avg. Loss: 0.02256575331091881\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 665, Batch: 200, Avg. Loss: 0.030703022107481957\n",
      "Epoch: 665, Batch: 300, Avg. Loss: 0.029263936653733252\n",
      "Epoch: 665, Batch: 400, Avg. Loss: 0.020399455353617667\n",
      "Epoch: 665, Batch: 500, Avg. Loss: 0.03282058693468571\n",
      "Epoch: 665, Batch: 600, Avg. Loss: 0.02438843935728073\n",
      "Epoch: 665, Batch: 700, Avg. Loss: 0.028378662914037706\n",
      "Epoch: 665, Batch: 800, Avg. Loss: 0.024998705238103866\n",
      "Epoch: 665, Batch: 900, Avg. Loss: 0.026172567754983903\n",
      "Epoch: 666, Batch: 100, Avg. Loss: 0.0317560239136219\n",
      "Epoch: 666, Batch: 200, Avg. Loss: 0.03195554882287979\n",
      "Epoch: 666, Batch: 300, Avg. Loss: 0.029961284026503564\n",
      "Epoch: 666, Batch: 400, Avg. Loss: 0.026669780761003493\n",
      "Epoch: 666, Batch: 500, Avg. Loss: 0.026835046783089636\n",
      "Epoch: 666, Batch: 600, Avg. Loss: 0.028868011608719827\n",
      "Epoch: 666, Batch: 700, Avg. Loss: 0.02347200207412243\n",
      "Epoch: 666, Batch: 800, Avg. Loss: 0.027656094953417778\n",
      "Epoch: 666, Batch: 900, Avg. Loss: 0.022600257769227028\n",
      "Epoch: 667, Batch: 100, Avg. Loss: 0.02473806284368038\n",
      "Epoch: 667, Batch: 200, Avg. Loss: 0.02987977124750614\n",
      "Epoch: 667, Batch: 300, Avg. Loss: 0.025160844400525094\n",
      "Epoch: 667, Batch: 400, Avg. Loss: 0.020639014318585395\n",
      "Epoch: 667, Batch: 500, Avg. Loss: 0.02757743149995804\n",
      "Epoch: 667, Batch: 600, Avg. Loss: 0.02149274829775095\n",
      "Epoch: 667, Batch: 700, Avg. Loss: 0.02901284895837307\n",
      "Epoch: 667, Batch: 800, Avg. Loss: 0.02696895256638527\n",
      "Epoch: 667, Batch: 900, Avg. Loss: 0.02632126897573471\n",
      "Epoch: 668, Batch: 100, Avg. Loss: 0.027798812836408615\n",
      "Epoch: 668, Batch: 200, Avg. Loss: 0.021926720291376114\n",
      "Epoch: 668, Batch: 300, Avg. Loss: 0.02329683057963848\n",
      "Epoch: 668, Batch: 400, Avg. Loss: 0.029828150123357773\n",
      "Epoch: 668, Batch: 500, Avg. Loss: 0.027238404899835585\n",
      "Epoch: 668, Batch: 600, Avg. Loss: 0.02533662550151348\n",
      "Epoch: 668, Batch: 700, Avg. Loss: 0.02331750825047493\n",
      "Epoch: 668, Batch: 800, Avg. Loss: 0.025331332460045815\n",
      "Epoch: 668, Batch: 900, Avg. Loss: 0.026707415543496607\n",
      "Epoch: 669, Batch: 100, Avg. Loss: 0.02023924097418785\n",
      "Epoch: 669, Batch: 200, Avg. Loss: 0.027626202926039695\n",
      "Epoch: 669, Batch: 300, Avg. Loss: 0.024870843663811683\n",
      "Epoch: 669, Batch: 400, Avg. Loss: 0.029396111145615578\n",
      "Epoch: 669, Batch: 500, Avg. Loss: 0.029379257187247276\n",
      "Epoch: 669, Batch: 600, Avg. Loss: 0.02746537983417511\n",
      "Epoch: 669, Batch: 700, Avg. Loss: 0.02692521095275879\n",
      "Epoch: 669, Batch: 800, Avg. Loss: 0.023989795222878457\n",
      "Epoch: 669, Batch: 900, Avg. Loss: 0.024793688654899598\n",
      "Epoch: 670, Batch: 100, Avg. Loss: 0.026073388010263442\n",
      "Epoch: 670, Batch: 200, Avg. Loss: 0.02249513804912567\n",
      "Epoch: 670, Batch: 300, Avg. Loss: 0.028663278147578238\n",
      "Epoch: 670, Batch: 400, Avg. Loss: 0.027268274277448653\n",
      "Epoch: 670, Batch: 500, Avg. Loss: 0.0284610016644001\n",
      "Epoch: 670, Batch: 600, Avg. Loss: 0.022842348366975785\n",
      "Epoch: 670, Batch: 700, Avg. Loss: 0.02395495794713497\n",
      "Epoch: 670, Batch: 800, Avg. Loss: 0.027698899060487746\n",
      "Epoch: 670, Batch: 900, Avg. Loss: 0.022557895183563232\n",
      "Epoch: 671, Batch: 100, Avg. Loss: 0.033060749396681785\n",
      "Epoch: 671, Batch: 200, Avg. Loss: 0.022653964161872864\n",
      "Epoch: 671, Batch: 300, Avg. Loss: 0.024936653450131418\n",
      "Epoch: 671, Batch: 400, Avg. Loss: 0.026196350082755088\n",
      "Epoch: 671, Batch: 500, Avg. Loss: 0.024430977925658227\n",
      "Epoch: 671, Batch: 600, Avg. Loss: 0.030863642767071722\n",
      "Epoch: 671, Batch: 700, Avg. Loss: 0.031595822125673294\n",
      "Epoch: 671, Batch: 800, Avg. Loss: 0.029836974367499353\n",
      "Epoch: 671, Batch: 900, Avg. Loss: 0.023954217284917832\n",
      "Epoch: 672, Batch: 100, Avg. Loss: 0.02704350631684065\n",
      "Epoch: 672, Batch: 200, Avg. Loss: 0.023155934885144235\n",
      "Epoch: 672, Batch: 300, Avg. Loss: 0.02910391107201576\n",
      "Epoch: 672, Batch: 400, Avg. Loss: 0.025961723402142523\n",
      "Epoch: 672, Batch: 500, Avg. Loss: 0.027224391400814056\n",
      "Epoch: 672, Batch: 600, Avg. Loss: 0.025941423326730727\n",
      "Epoch: 672, Batch: 700, Avg. Loss: 0.026653673946857453\n",
      "Epoch: 672, Batch: 800, Avg. Loss: 0.023809861317276956\n",
      "Epoch: 672, Batch: 900, Avg. Loss: 0.028497091233730315\n",
      "Epoch: 673, Batch: 100, Avg. Loss: 0.024032820388674735\n",
      "Epoch: 673, Batch: 200, Avg. Loss: 0.02241347812116146\n",
      "Epoch: 673, Batch: 300, Avg. Loss: 0.02637449897825718\n",
      "Epoch: 673, Batch: 400, Avg. Loss: 0.027855853289365767\n",
      "Epoch: 673, Batch: 500, Avg. Loss: 0.027952995672821997\n",
      "Epoch: 673, Batch: 600, Avg. Loss: 0.028854452297091483\n",
      "Epoch: 673, Batch: 700, Avg. Loss: 0.0209952349960804\n",
      "Epoch: 673, Batch: 800, Avg. Loss: 0.027659740298986435\n",
      "Epoch: 673, Batch: 900, Avg. Loss: 0.03365798875689507\n",
      "Epoch: 674, Batch: 100, Avg. Loss: 0.027414146438241006\n",
      "Epoch: 674, Batch: 200, Avg. Loss: 0.0257360914349556\n",
      "Epoch: 674, Batch: 300, Avg. Loss: 0.030247754231095315\n",
      "Epoch: 674, Batch: 400, Avg. Loss: 0.024522143453359603\n",
      "Epoch: 674, Batch: 500, Avg. Loss: 0.020711053535342215\n",
      "Epoch: 674, Batch: 600, Avg. Loss: 0.03394897617399693\n",
      "Epoch: 674, Batch: 700, Avg. Loss: 0.024302195608615875\n",
      "Epoch: 674, Batch: 800, Avg. Loss: 0.019898513853549956\n",
      "Epoch: 674, Batch: 900, Avg. Loss: 0.02416920803487301\n",
      "Epoch: 675, Batch: 100, Avg. Loss: 0.02631732791662216\n",
      "Epoch: 675, Batch: 200, Avg. Loss: 0.023924961015582086\n",
      "Epoch: 675, Batch: 300, Avg. Loss: 0.029416990578174592\n",
      "Epoch: 675, Batch: 400, Avg. Loss: 0.025637049153447153\n",
      "Epoch: 675, Batch: 500, Avg. Loss: 0.025612064227461814\n",
      "Epoch: 675, Batch: 600, Avg. Loss: 0.020583874583244323\n",
      "Epoch: 675, Batch: 700, Avg. Loss: 0.023909399583935738\n",
      "Epoch: 675, Batch: 800, Avg. Loss: 0.022314895763993262\n",
      "Epoch: 675, Batch: 900, Avg. Loss: 0.023387355133891105\n",
      "Epoch: 676, Batch: 100, Avg. Loss: 0.026549342721700668\n",
      "Epoch: 676, Batch: 200, Avg. Loss: 0.02253501519560814\n",
      "Epoch: 676, Batch: 300, Avg. Loss: 0.02663913793861866\n",
      "Epoch: 676, Batch: 400, Avg. Loss: 0.026896864622831345\n",
      "Epoch: 676, Batch: 500, Avg. Loss: 0.025018764808773995\n",
      "Epoch: 676, Batch: 600, Avg. Loss: 0.02782162144780159\n",
      "Epoch: 676, Batch: 700, Avg. Loss: 0.01973288454115391\n",
      "Epoch: 676, Batch: 800, Avg. Loss: 0.025418906137347222\n",
      "Epoch: 676, Batch: 900, Avg. Loss: 0.02743427813053131\n",
      "Epoch: 677, Batch: 100, Avg. Loss: 0.02825016848742962\n",
      "Epoch: 677, Batch: 200, Avg. Loss: 0.02431798353791237\n",
      "Epoch: 677, Batch: 300, Avg. Loss: 0.023080217316746712\n",
      "Epoch: 677, Batch: 400, Avg. Loss: 0.029719621390104294\n",
      "Epoch: 677, Batch: 500, Avg. Loss: 0.02401185892522335\n",
      "Epoch: 677, Batch: 600, Avg. Loss: 0.029430652633309365\n",
      "Epoch: 677, Batch: 700, Avg. Loss: 0.027991186082363128\n",
      "Epoch: 677, Batch: 800, Avg. Loss: 0.027300004959106446\n",
      "Epoch: 677, Batch: 900, Avg. Loss: 0.025767516121268274\n",
      "Epoch: 678, Batch: 100, Avg. Loss: 0.022353619262576105\n",
      "Epoch: 678, Batch: 200, Avg. Loss: 0.024461166858673097\n",
      "Epoch: 678, Batch: 300, Avg. Loss: 0.027620031982660293\n",
      "Epoch: 678, Batch: 400, Avg. Loss: 0.030548986941576005\n",
      "Epoch: 678, Batch: 500, Avg. Loss: 0.031707392632961275\n",
      "Epoch: 678, Batch: 600, Avg. Loss: 0.025818263217806815\n",
      "Epoch: 678, Batch: 700, Avg. Loss: 0.023346035480499266\n",
      "Epoch: 678, Batch: 800, Avg. Loss: 0.025229148790240286\n",
      "Epoch: 678, Batch: 900, Avg. Loss: 0.03185616835951805\n",
      "Epoch: 679, Batch: 100, Avg. Loss: 0.02527627632021904\n",
      "Epoch: 679, Batch: 200, Avg. Loss: 0.02541418343782425\n",
      "Epoch: 679, Batch: 300, Avg. Loss: 0.027110659554600715\n",
      "Epoch: 679, Batch: 400, Avg. Loss: 0.02349365971982479\n",
      "Epoch: 679, Batch: 500, Avg. Loss: 0.025045761168003083\n",
      "Epoch: 679, Batch: 600, Avg. Loss: 0.028684445768594743\n",
      "Epoch: 679, Batch: 700, Avg. Loss: 0.02502327464520931\n",
      "Epoch: 679, Batch: 800, Avg. Loss: 0.02529663808643818\n",
      "Epoch: 679, Batch: 900, Avg. Loss: 0.027085490971803666\n",
      "Epoch: 680, Batch: 100, Avg. Loss: 0.02688268683850765\n",
      "Epoch: 680, Batch: 200, Avg. Loss: 0.02780046418309212\n",
      "Epoch: 680, Batch: 300, Avg. Loss: 0.022337944954633714\n",
      "Epoch: 680, Batch: 400, Avg. Loss: 0.02542115531861782\n",
      "Epoch: 680, Batch: 500, Avg. Loss: 0.023964553475379943\n",
      "Epoch: 680, Batch: 600, Avg. Loss: 0.02472859412431717\n",
      "Epoch: 680, Batch: 700, Avg. Loss: 0.02346606157720089\n",
      "Epoch: 680, Batch: 800, Avg. Loss: 0.026112997457385063\n",
      "Epoch: 680, Batch: 900, Avg. Loss: 0.02443148650228977\n",
      "Epoch: 681, Batch: 100, Avg. Loss: 0.023563585355877877\n",
      "Epoch: 681, Batch: 200, Avg. Loss: 0.03205924771726131\n",
      "Epoch: 681, Batch: 300, Avg. Loss: 0.022527835816144943\n",
      "Epoch: 681, Batch: 400, Avg. Loss: 0.02420505717396736\n",
      "Epoch: 681, Batch: 500, Avg. Loss: 0.020802183598279952\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 681, Batch: 600, Avg. Loss: 0.02761325165629387\n",
      "Epoch: 681, Batch: 700, Avg. Loss: 0.025979884043335913\n",
      "Epoch: 681, Batch: 800, Avg. Loss: 0.031486608386039734\n",
      "Epoch: 681, Batch: 900, Avg. Loss: 0.019979290664196014\n",
      "Epoch: 682, Batch: 100, Avg. Loss: 0.026860900744795798\n",
      "Epoch: 682, Batch: 200, Avg. Loss: 0.02539033517241478\n",
      "Epoch: 682, Batch: 300, Avg. Loss: 0.028194721788167953\n",
      "Epoch: 682, Batch: 400, Avg. Loss: 0.021731611043214798\n",
      "Epoch: 682, Batch: 500, Avg. Loss: 0.026726693883538245\n",
      "Epoch: 682, Batch: 600, Avg. Loss: 0.0270285215228796\n",
      "Epoch: 682, Batch: 700, Avg. Loss: 0.025735619962215423\n",
      "Epoch: 682, Batch: 800, Avg. Loss: 0.029462830200791358\n",
      "Epoch: 682, Batch: 900, Avg. Loss: 0.02423720695078373\n",
      "Epoch: 683, Batch: 100, Avg. Loss: 0.02423801876604557\n",
      "Epoch: 683, Batch: 200, Avg. Loss: 0.03135396234691143\n",
      "Epoch: 683, Batch: 300, Avg. Loss: 0.025274061113595963\n",
      "Epoch: 683, Batch: 400, Avg. Loss: 0.02420413672924042\n",
      "Epoch: 683, Batch: 500, Avg. Loss: 0.02575484737753868\n",
      "Epoch: 683, Batch: 600, Avg. Loss: 0.0239493016153574\n",
      "Epoch: 683, Batch: 700, Avg. Loss: 0.024778854250907898\n",
      "Epoch: 683, Batch: 800, Avg. Loss: 0.024575016275048257\n",
      "Epoch: 683, Batch: 900, Avg. Loss: 0.030530858039855956\n",
      "Epoch: 684, Batch: 100, Avg. Loss: 0.02896762825548649\n",
      "Epoch: 684, Batch: 200, Avg. Loss: 0.022664177007973196\n",
      "Epoch: 684, Batch: 300, Avg. Loss: 0.026733445078134535\n",
      "Epoch: 684, Batch: 400, Avg. Loss: 0.020440168604254724\n",
      "Epoch: 684, Batch: 500, Avg. Loss: 0.02360116221010685\n",
      "Epoch: 684, Batch: 600, Avg. Loss: 0.0281761460006237\n",
      "Epoch: 684, Batch: 700, Avg. Loss: 0.028014761134982108\n",
      "Epoch: 684, Batch: 800, Avg. Loss: 0.022463099136948585\n",
      "Epoch: 684, Batch: 900, Avg. Loss: 0.023810972273349763\n",
      "Epoch: 685, Batch: 100, Avg. Loss: 0.02610801689326763\n",
      "Epoch: 685, Batch: 200, Avg. Loss: 0.027022962272167207\n",
      "Epoch: 685, Batch: 300, Avg. Loss: 0.026467779725790022\n",
      "Epoch: 685, Batch: 400, Avg. Loss: 0.02006019689142704\n",
      "Epoch: 685, Batch: 500, Avg. Loss: 0.02383177787065506\n",
      "Epoch: 685, Batch: 600, Avg. Loss: 0.022901310846209527\n",
      "Epoch: 685, Batch: 700, Avg. Loss: 0.0301272364705801\n",
      "Epoch: 685, Batch: 800, Avg. Loss: 0.025780510678887366\n",
      "Epoch: 685, Batch: 900, Avg. Loss: 0.027386751472949982\n",
      "Epoch: 686, Batch: 100, Avg. Loss: 0.027069748863577842\n",
      "Epoch: 686, Batch: 200, Avg. Loss: 0.025916623175144194\n",
      "Epoch: 686, Batch: 300, Avg. Loss: 0.03019460029900074\n",
      "Epoch: 686, Batch: 400, Avg. Loss: 0.024651077538728715\n",
      "Epoch: 686, Batch: 500, Avg. Loss: 0.023137286752462388\n",
      "Epoch: 686, Batch: 600, Avg. Loss: 0.02113888628780842\n",
      "Epoch: 686, Batch: 700, Avg. Loss: 0.021747816130518915\n",
      "Epoch: 686, Batch: 800, Avg. Loss: 0.023684024661779404\n",
      "Epoch: 686, Batch: 900, Avg. Loss: 0.027527363002300263\n",
      "Epoch: 687, Batch: 100, Avg. Loss: 0.023362230435013773\n",
      "Epoch: 687, Batch: 200, Avg. Loss: 0.027434245347976685\n",
      "Epoch: 687, Batch: 300, Avg. Loss: 0.024406003877520562\n",
      "Epoch: 687, Batch: 400, Avg. Loss: 0.02460271406918764\n",
      "Epoch: 687, Batch: 500, Avg. Loss: 0.025209313109517097\n",
      "Epoch: 687, Batch: 600, Avg. Loss: 0.02750242218375206\n",
      "Epoch: 687, Batch: 700, Avg. Loss: 0.024239282086491586\n",
      "Epoch: 687, Batch: 800, Avg. Loss: 0.02023968830704689\n",
      "Epoch: 687, Batch: 900, Avg. Loss: 0.024220613315701484\n",
      "Epoch: 688, Batch: 100, Avg. Loss: 0.026020984128117562\n",
      "Epoch: 688, Batch: 200, Avg. Loss: 0.026287162452936174\n",
      "Epoch: 688, Batch: 300, Avg. Loss: 0.024541025161743165\n",
      "Epoch: 688, Batch: 400, Avg. Loss: 0.021640565916895865\n",
      "Epoch: 688, Batch: 500, Avg. Loss: 0.02794499322772026\n",
      "Epoch: 688, Batch: 600, Avg. Loss: 0.02258002191781998\n",
      "Epoch: 688, Batch: 700, Avg. Loss: 0.027412826344370842\n",
      "Epoch: 688, Batch: 800, Avg. Loss: 0.020728369876742363\n",
      "Epoch: 688, Batch: 900, Avg. Loss: 0.026244380846619605\n",
      "Epoch: 689, Batch: 100, Avg. Loss: 0.02363409213721752\n",
      "Epoch: 689, Batch: 200, Avg. Loss: 0.023542146012187005\n",
      "Epoch: 689, Batch: 300, Avg. Loss: 0.024973232746124268\n",
      "Epoch: 689, Batch: 400, Avg. Loss: 0.02596886832267046\n",
      "Epoch: 689, Batch: 500, Avg. Loss: 0.02557495519518852\n",
      "Epoch: 689, Batch: 600, Avg. Loss: 0.026550759598612785\n",
      "Epoch: 689, Batch: 700, Avg. Loss: 0.020845638513565065\n",
      "Epoch: 689, Batch: 800, Avg. Loss: 0.02681581139564514\n",
      "Epoch: 689, Batch: 900, Avg. Loss: 0.02764580078423023\n",
      "Epoch: 690, Batch: 100, Avg. Loss: 0.02433539889752865\n",
      "Epoch: 690, Batch: 200, Avg. Loss: 0.022625757530331612\n",
      "Epoch: 690, Batch: 300, Avg. Loss: 0.021587147489190103\n",
      "Epoch: 690, Batch: 400, Avg. Loss: 0.023255741074681282\n",
      "Epoch: 690, Batch: 500, Avg. Loss: 0.02165147140622139\n",
      "Epoch: 690, Batch: 600, Avg. Loss: 0.02466857507824898\n",
      "Epoch: 690, Batch: 700, Avg. Loss: 0.02658441536128521\n",
      "Epoch: 690, Batch: 800, Avg. Loss: 0.02960312768816948\n",
      "Epoch: 690, Batch: 900, Avg. Loss: 0.026972547993063926\n",
      "Epoch: 691, Batch: 100, Avg. Loss: 0.026053158417344093\n",
      "Epoch: 691, Batch: 200, Avg. Loss: 0.025702864304184912\n",
      "Epoch: 691, Batch: 300, Avg. Loss: 0.026311856508255006\n",
      "Epoch: 691, Batch: 400, Avg. Loss: 0.027680624574422837\n",
      "Epoch: 691, Batch: 500, Avg. Loss: 0.02509434886276722\n",
      "Epoch: 691, Batch: 600, Avg. Loss: 0.019961531311273574\n",
      "Epoch: 691, Batch: 700, Avg. Loss: 0.024968506395816804\n",
      "Epoch: 691, Batch: 800, Avg. Loss: 0.03133143842220307\n",
      "Epoch: 691, Batch: 900, Avg. Loss: 0.02339960552752018\n",
      "Epoch: 692, Batch: 100, Avg. Loss: 0.030535756796598434\n",
      "Epoch: 692, Batch: 200, Avg. Loss: 0.03266629107296467\n",
      "Epoch: 692, Batch: 300, Avg. Loss: 0.02645335055887699\n",
      "Epoch: 692, Batch: 400, Avg. Loss: 0.02470429427921772\n",
      "Epoch: 692, Batch: 500, Avg. Loss: 0.024687861055135728\n",
      "Epoch: 692, Batch: 600, Avg. Loss: 0.01774133898317814\n",
      "Epoch: 692, Batch: 700, Avg. Loss: 0.026499689519405366\n",
      "Epoch: 692, Batch: 800, Avg. Loss: 0.03048083171248436\n",
      "Epoch: 692, Batch: 900, Avg. Loss: 0.024967968091368677\n",
      "Epoch: 693, Batch: 100, Avg. Loss: 0.026991728991270066\n",
      "Epoch: 693, Batch: 200, Avg. Loss: 0.023532024919986724\n",
      "Epoch: 693, Batch: 300, Avg. Loss: 0.029036346822977066\n",
      "Epoch: 693, Batch: 400, Avg. Loss: 0.025823156014084817\n",
      "Epoch: 693, Batch: 500, Avg. Loss: 0.0188476937264204\n",
      "Epoch: 693, Batch: 600, Avg. Loss: 0.026239654198288918\n",
      "Epoch: 693, Batch: 700, Avg. Loss: 0.024933172091841697\n",
      "Epoch: 693, Batch: 800, Avg. Loss: 0.029150085374712944\n",
      "Epoch: 693, Batch: 900, Avg. Loss: 0.027549568265676498\n",
      "Epoch: 694, Batch: 100, Avg. Loss: 0.017308759689331054\n",
      "Epoch: 694, Batch: 200, Avg. Loss: 0.025882724747061728\n",
      "Epoch: 694, Batch: 300, Avg. Loss: 0.026995002031326293\n",
      "Epoch: 694, Batch: 400, Avg. Loss: 0.02809867113828659\n",
      "Epoch: 694, Batch: 500, Avg. Loss: 0.02338536851108074\n",
      "Epoch: 694, Batch: 600, Avg. Loss: 0.029107281714677812\n",
      "Epoch: 694, Batch: 700, Avg. Loss: 0.021363881826400758\n",
      "Epoch: 694, Batch: 800, Avg. Loss: 0.023769406005740165\n",
      "Epoch: 694, Batch: 900, Avg. Loss: 0.029390456452965737\n",
      "Epoch: 695, Batch: 100, Avg. Loss: 0.021961772963404656\n",
      "Epoch: 695, Batch: 200, Avg. Loss: 0.023014022037386894\n",
      "Epoch: 695, Batch: 300, Avg. Loss: 0.02422676034271717\n",
      "Epoch: 695, Batch: 400, Avg. Loss: 0.022732409834861755\n",
      "Epoch: 695, Batch: 500, Avg. Loss: 0.019306796714663507\n",
      "Epoch: 695, Batch: 600, Avg. Loss: 0.022112731635570527\n",
      "Epoch: 695, Batch: 700, Avg. Loss: 0.02496057726442814\n",
      "Epoch: 695, Batch: 800, Avg. Loss: 0.03002984181046486\n",
      "Epoch: 695, Batch: 900, Avg. Loss: 0.02612274743616581\n",
      "Epoch: 696, Batch: 100, Avg. Loss: 0.026505698263645173\n",
      "Epoch: 696, Batch: 200, Avg. Loss: 0.022065220028162004\n",
      "Epoch: 696, Batch: 300, Avg. Loss: 0.024287209436297415\n",
      "Epoch: 696, Batch: 400, Avg. Loss: 0.0197712067514658\n",
      "Epoch: 696, Batch: 500, Avg. Loss: 0.028725570365786553\n",
      "Epoch: 696, Batch: 600, Avg. Loss: 0.02739263355731964\n",
      "Epoch: 696, Batch: 700, Avg. Loss: 0.029535657092928887\n",
      "Epoch: 696, Batch: 800, Avg. Loss: 0.024259372875094415\n",
      "Epoch: 696, Batch: 900, Avg. Loss: 0.026967024952173232\n",
      "Epoch: 697, Batch: 100, Avg. Loss: 0.029030826315283774\n",
      "Epoch: 697, Batch: 200, Avg. Loss: 0.026317111477255822\n",
      "Epoch: 697, Batch: 300, Avg. Loss: 0.027031281590461732\n",
      "Epoch: 697, Batch: 400, Avg. Loss: 0.027206764444708823\n",
      "Epoch: 697, Batch: 500, Avg. Loss: 0.0271354465931654\n",
      "Epoch: 697, Batch: 600, Avg. Loss: 0.026279646307229995\n",
      "Epoch: 697, Batch: 700, Avg. Loss: 0.023774553164839743\n",
      "Epoch: 697, Batch: 800, Avg. Loss: 0.02563969373703003\n",
      "Epoch: 697, Batch: 900, Avg. Loss: 0.019254097640514375\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 698, Batch: 100, Avg. Loss: 0.026362474635243417\n",
      "Epoch: 698, Batch: 200, Avg. Loss: 0.027206643149256707\n",
      "Epoch: 698, Batch: 300, Avg. Loss: 0.025580749437212943\n",
      "Epoch: 698, Batch: 400, Avg. Loss: 0.02811017066240311\n",
      "Epoch: 698, Batch: 500, Avg. Loss: 0.01732903964817524\n",
      "Epoch: 698, Batch: 600, Avg. Loss: 0.0265799917280674\n",
      "Epoch: 698, Batch: 700, Avg. Loss: 0.029362602680921553\n",
      "Epoch: 698, Batch: 800, Avg. Loss: 0.02426609434187412\n",
      "Epoch: 698, Batch: 900, Avg. Loss: 0.03371778022497893\n",
      "Epoch: 699, Batch: 100, Avg. Loss: 0.02117431066930294\n",
      "Epoch: 699, Batch: 200, Avg. Loss: 0.033267521038651465\n",
      "Epoch: 699, Batch: 300, Avg. Loss: 0.024533756971359254\n",
      "Epoch: 699, Batch: 400, Avg. Loss: 0.027887798175215722\n",
      "Epoch: 699, Batch: 500, Avg. Loss: 0.02449849098920822\n",
      "Epoch: 699, Batch: 600, Avg. Loss: 0.023418596535921096\n",
      "Epoch: 699, Batch: 700, Avg. Loss: 0.020412741526961327\n",
      "Epoch: 699, Batch: 800, Avg. Loss: 0.02573482409119606\n",
      "Epoch: 699, Batch: 900, Avg. Loss: 0.02181174799799919\n",
      "Epoch: 700, Batch: 100, Avg. Loss: 0.02912060484290123\n",
      "Epoch: 700, Batch: 200, Avg. Loss: 0.024284722954034804\n",
      "Epoch: 700, Batch: 300, Avg. Loss: 0.022551596462726593\n",
      "Epoch: 700, Batch: 400, Avg. Loss: 0.022629510909318924\n",
      "Epoch: 700, Batch: 500, Avg. Loss: 0.022696602940559387\n",
      "Epoch: 700, Batch: 600, Avg. Loss: 0.02626586638391018\n",
      "Epoch: 700, Batch: 700, Avg. Loss: 0.026613288074731827\n",
      "Epoch: 700, Batch: 800, Avg. Loss: 0.02434626244008541\n",
      "Epoch: 700, Batch: 900, Avg. Loss: 0.023848631232976914\n",
      "Epoch: 701, Batch: 100, Avg. Loss: 0.02748238533735275\n",
      "Epoch: 701, Batch: 200, Avg. Loss: 0.02423655316233635\n",
      "Epoch: 701, Batch: 300, Avg. Loss: 0.023018264025449754\n",
      "Epoch: 701, Batch: 400, Avg. Loss: 0.02624611385166645\n",
      "Epoch: 701, Batch: 500, Avg. Loss: 0.023389085084199904\n",
      "Epoch: 701, Batch: 600, Avg. Loss: 0.023010076507925987\n",
      "Epoch: 701, Batch: 700, Avg. Loss: 0.02522396683692932\n",
      "Epoch: 701, Batch: 800, Avg. Loss: 0.022803600430488586\n",
      "Epoch: 701, Batch: 900, Avg. Loss: 0.028480227664113046\n",
      "Epoch: 702, Batch: 100, Avg. Loss: 0.02014233998954296\n",
      "Epoch: 702, Batch: 200, Avg. Loss: 0.026456974521279333\n",
      "Epoch: 702, Batch: 300, Avg. Loss: 0.022456253618001937\n",
      "Epoch: 702, Batch: 400, Avg. Loss: 0.030757340639829635\n",
      "Epoch: 702, Batch: 500, Avg. Loss: 0.02556388020515442\n",
      "Epoch: 702, Batch: 600, Avg. Loss: 0.02178231827914715\n",
      "Epoch: 702, Batch: 700, Avg. Loss: 0.028278639167547227\n",
      "Epoch: 702, Batch: 800, Avg. Loss: 0.02956051968038082\n",
      "Epoch: 702, Batch: 900, Avg. Loss: 0.024088508114218712\n",
      "Epoch: 703, Batch: 100, Avg. Loss: 0.021990040093660356\n",
      "Epoch: 703, Batch: 200, Avg. Loss: 0.024767626076936722\n",
      "Epoch: 703, Batch: 300, Avg. Loss: 0.026148532927036287\n",
      "Epoch: 703, Batch: 400, Avg. Loss: 0.028449116870760918\n",
      "Epoch: 703, Batch: 500, Avg. Loss: 0.01992521956562996\n",
      "Epoch: 703, Batch: 600, Avg. Loss: 0.026952930316329\n",
      "Epoch: 703, Batch: 700, Avg. Loss: 0.02944123089313507\n",
      "Epoch: 703, Batch: 800, Avg. Loss: 0.031327779069542885\n",
      "Epoch: 703, Batch: 900, Avg. Loss: 0.025708759352564812\n",
      "Epoch: 704, Batch: 100, Avg. Loss: 0.022951794639229774\n",
      "Epoch: 704, Batch: 200, Avg. Loss: 0.0243282712996006\n",
      "Epoch: 704, Batch: 300, Avg. Loss: 0.024764981865882874\n",
      "Epoch: 704, Batch: 400, Avg. Loss: 0.0247162526845932\n",
      "Epoch: 704, Batch: 500, Avg. Loss: 0.016368024572730063\n",
      "Epoch: 704, Batch: 600, Avg. Loss: 0.027849128097295762\n",
      "Epoch: 704, Batch: 700, Avg. Loss: 0.02348280981183052\n",
      "Epoch: 704, Batch: 800, Avg. Loss: 0.031215789914131164\n",
      "Epoch: 704, Batch: 900, Avg. Loss: 0.028759321197867393\n",
      "Epoch: 705, Batch: 100, Avg. Loss: 0.021439282223582268\n",
      "Epoch: 705, Batch: 200, Avg. Loss: 0.022743170857429506\n",
      "Epoch: 705, Batch: 300, Avg. Loss: 0.022058737203478812\n",
      "Epoch: 705, Batch: 400, Avg. Loss: 0.022952631041407587\n",
      "Epoch: 705, Batch: 500, Avg. Loss: 0.0238923516869545\n",
      "Epoch: 705, Batch: 600, Avg. Loss: 0.023666132912039756\n",
      "Epoch: 705, Batch: 700, Avg. Loss: 0.023654863387346268\n",
      "Epoch: 705, Batch: 800, Avg. Loss: 0.02588922142982483\n",
      "Epoch: 705, Batch: 900, Avg. Loss: 0.026962333619594575\n",
      "Epoch: 706, Batch: 100, Avg. Loss: 0.023976496458053588\n",
      "Epoch: 706, Batch: 200, Avg. Loss: 0.022816746085882186\n",
      "Epoch: 706, Batch: 300, Avg. Loss: 0.024036253839731216\n",
      "Epoch: 706, Batch: 400, Avg. Loss: 0.02941167488694191\n",
      "Epoch: 706, Batch: 500, Avg. Loss: 0.02782133899629116\n",
      "Epoch: 706, Batch: 600, Avg. Loss: 0.022636485248804093\n",
      "Epoch: 706, Batch: 700, Avg. Loss: 0.028547947853803636\n",
      "Epoch: 706, Batch: 800, Avg. Loss: 0.024373371601104737\n",
      "Epoch: 706, Batch: 900, Avg. Loss: 0.02449976809322834\n",
      "Epoch: 707, Batch: 100, Avg. Loss: 0.02523082382977009\n",
      "Epoch: 707, Batch: 200, Avg. Loss: 0.02645212061703205\n",
      "Epoch: 707, Batch: 300, Avg. Loss: 0.023396292701363564\n",
      "Epoch: 707, Batch: 400, Avg. Loss: 0.023924548551440238\n",
      "Epoch: 707, Batch: 500, Avg. Loss: 0.023274962902069093\n",
      "Epoch: 707, Batch: 600, Avg. Loss: 0.025723690316081047\n",
      "Epoch: 707, Batch: 700, Avg. Loss: 0.028402932211756705\n",
      "Epoch: 707, Batch: 800, Avg. Loss: 0.026619877889752386\n",
      "Epoch: 707, Batch: 900, Avg. Loss: 0.027286645472049714\n",
      "Epoch: 708, Batch: 100, Avg. Loss: 0.028135115578770638\n",
      "Epoch: 708, Batch: 200, Avg. Loss: 0.02547390162944794\n",
      "Epoch: 708, Batch: 300, Avg. Loss: 0.02651470437645912\n",
      "Epoch: 708, Batch: 400, Avg. Loss: 0.019427078142762184\n",
      "Epoch: 708, Batch: 500, Avg. Loss: 0.0254552511125803\n",
      "Epoch: 708, Batch: 600, Avg. Loss: 0.02662364736199379\n",
      "Epoch: 708, Batch: 700, Avg. Loss: 0.025771342664957047\n",
      "Epoch: 708, Batch: 800, Avg. Loss: 0.02546007849276066\n",
      "Epoch: 708, Batch: 900, Avg. Loss: 0.02232395574450493\n",
      "Epoch: 709, Batch: 100, Avg. Loss: 0.020817466303706168\n",
      "Epoch: 709, Batch: 200, Avg. Loss: 0.02213647246360779\n",
      "Epoch: 709, Batch: 300, Avg. Loss: 0.026544753164052963\n",
      "Epoch: 709, Batch: 400, Avg. Loss: 0.02459713362157345\n",
      "Epoch: 709, Batch: 500, Avg. Loss: 0.025285839214921\n",
      "Epoch: 709, Batch: 600, Avg. Loss: 0.026178435236215592\n",
      "Epoch: 709, Batch: 700, Avg. Loss: 0.02382872574031353\n",
      "Epoch: 709, Batch: 800, Avg. Loss: 0.028717742636799812\n",
      "Epoch: 709, Batch: 900, Avg. Loss: 0.02407224714756012\n",
      "Epoch: 710, Batch: 100, Avg. Loss: 0.022263468354940415\n",
      "Epoch: 710, Batch: 200, Avg. Loss: 0.024584796279668808\n",
      "Epoch: 710, Batch: 300, Avg. Loss: 0.02135540999472141\n",
      "Epoch: 710, Batch: 400, Avg. Loss: 0.025008802637457847\n",
      "Epoch: 710, Batch: 500, Avg. Loss: 0.021957393139600753\n",
      "Epoch: 710, Batch: 600, Avg. Loss: 0.026449433416128158\n",
      "Epoch: 710, Batch: 700, Avg. Loss: 0.020794044807553293\n",
      "Epoch: 710, Batch: 800, Avg. Loss: 0.025465074107050897\n",
      "Epoch: 710, Batch: 900, Avg. Loss: 0.024304341077804565\n",
      "Epoch: 711, Batch: 100, Avg. Loss: 0.024685885831713675\n",
      "Epoch: 711, Batch: 200, Avg. Loss: 0.02099826715886593\n",
      "Epoch: 711, Batch: 300, Avg. Loss: 0.03197019025683403\n",
      "Epoch: 711, Batch: 400, Avg. Loss: 0.023804179653525354\n",
      "Epoch: 711, Batch: 500, Avg. Loss: 0.025077565312385558\n",
      "Epoch: 711, Batch: 600, Avg. Loss: 0.02096377417445183\n",
      "Epoch: 711, Batch: 700, Avg. Loss: 0.02320976950228214\n",
      "Epoch: 711, Batch: 800, Avg. Loss: 0.022467301338911057\n",
      "Epoch: 711, Batch: 900, Avg. Loss: 0.02626748099923134\n",
      "Epoch: 712, Batch: 100, Avg. Loss: 0.025618699789047242\n",
      "Epoch: 712, Batch: 200, Avg. Loss: 0.023686074167490006\n",
      "Epoch: 712, Batch: 300, Avg. Loss: 0.029730990007519723\n",
      "Epoch: 712, Batch: 400, Avg. Loss: 0.02368283599615097\n",
      "Epoch: 712, Batch: 500, Avg. Loss: 0.025351500436663628\n",
      "Epoch: 712, Batch: 600, Avg. Loss: 0.029626793786883356\n",
      "Epoch: 712, Batch: 700, Avg. Loss: 0.029896497428417206\n",
      "Epoch: 712, Batch: 800, Avg. Loss: 0.02406656287610531\n",
      "Epoch: 712, Batch: 900, Avg. Loss: 0.025565849691629408\n",
      "Epoch: 713, Batch: 100, Avg. Loss: 0.02686512127518654\n",
      "Epoch: 713, Batch: 200, Avg. Loss: 0.02475785993039608\n",
      "Epoch: 713, Batch: 300, Avg. Loss: 0.029349475130438805\n",
      "Epoch: 713, Batch: 400, Avg. Loss: 0.028493582531809808\n",
      "Epoch: 713, Batch: 500, Avg. Loss: 0.024897800609469414\n",
      "Epoch: 713, Batch: 600, Avg. Loss: 0.025080533996224404\n",
      "Epoch: 713, Batch: 700, Avg. Loss: 0.019915264695882798\n",
      "Epoch: 713, Batch: 800, Avg. Loss: 0.022383535280823708\n",
      "Epoch: 713, Batch: 900, Avg. Loss: 0.02698144771158695\n",
      "Epoch: 714, Batch: 100, Avg. Loss: 0.02644558809697628\n",
      "Epoch: 714, Batch: 200, Avg. Loss: 0.02330460786819458\n",
      "Epoch: 714, Batch: 300, Avg. Loss: 0.029740679040551186\n",
      "Epoch: 714, Batch: 400, Avg. Loss: 0.019667316973209382\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 714, Batch: 500, Avg. Loss: 0.023390218019485473\n",
      "Epoch: 714, Batch: 600, Avg. Loss: 0.022499689683318138\n",
      "Epoch: 714, Batch: 700, Avg. Loss: 0.02309700518846512\n",
      "Epoch: 714, Batch: 800, Avg. Loss: 0.028154638558626176\n",
      "Epoch: 714, Batch: 900, Avg. Loss: 0.022363667413592338\n",
      "Epoch: 715, Batch: 100, Avg. Loss: 0.030482729822397234\n",
      "Epoch: 715, Batch: 200, Avg. Loss: 0.026445802450180054\n",
      "Epoch: 715, Batch: 300, Avg. Loss: 0.022060585021972657\n",
      "Epoch: 715, Batch: 400, Avg. Loss: 0.022993866726756094\n",
      "Epoch: 715, Batch: 500, Avg. Loss: 0.020986426919698715\n",
      "Epoch: 715, Batch: 600, Avg. Loss: 0.02326537176966667\n",
      "Epoch: 715, Batch: 700, Avg. Loss: 0.02272956021130085\n",
      "Epoch: 715, Batch: 800, Avg. Loss: 0.02372099108994007\n",
      "Epoch: 715, Batch: 900, Avg. Loss: 0.0216026521474123\n",
      "Epoch: 716, Batch: 100, Avg. Loss: 0.02282706491649151\n",
      "Epoch: 716, Batch: 200, Avg. Loss: 0.027451403364539148\n",
      "Epoch: 716, Batch: 300, Avg. Loss: 0.02307221941649914\n",
      "Epoch: 716, Batch: 400, Avg. Loss: 0.024978736639022826\n",
      "Epoch: 716, Batch: 500, Avg. Loss: 0.024147948175668718\n",
      "Epoch: 716, Batch: 600, Avg. Loss: 0.023457803800702094\n",
      "Epoch: 716, Batch: 700, Avg. Loss: 0.020423256009817124\n",
      "Epoch: 716, Batch: 800, Avg. Loss: 0.0228938889503479\n",
      "Epoch: 716, Batch: 900, Avg. Loss: 0.025832040160894396\n",
      "Epoch: 717, Batch: 100, Avg. Loss: 0.02719145782291889\n",
      "Epoch: 717, Batch: 200, Avg. Loss: 0.027336349859833717\n",
      "Epoch: 717, Batch: 300, Avg. Loss: 0.029434489831328393\n",
      "Epoch: 717, Batch: 400, Avg. Loss: 0.02585953541100025\n",
      "Epoch: 717, Batch: 500, Avg. Loss: 0.023125873655080797\n",
      "Epoch: 717, Batch: 600, Avg. Loss: 0.023778093308210374\n",
      "Epoch: 717, Batch: 700, Avg. Loss: 0.020848966985940933\n",
      "Epoch: 717, Batch: 800, Avg. Loss: 0.02399951122701168\n",
      "Epoch: 717, Batch: 900, Avg. Loss: 0.023077264428138733\n",
      "Epoch: 718, Batch: 100, Avg. Loss: 0.025040136426687242\n",
      "Epoch: 718, Batch: 200, Avg. Loss: 0.024502503611147405\n",
      "Epoch: 718, Batch: 300, Avg. Loss: 0.025922672525048256\n",
      "Epoch: 718, Batch: 400, Avg. Loss: 0.018384688645601274\n",
      "Epoch: 718, Batch: 500, Avg. Loss: 0.022192195653915406\n",
      "Epoch: 718, Batch: 600, Avg. Loss: 0.02215298131108284\n",
      "Epoch: 718, Batch: 700, Avg. Loss: 0.021595512703061103\n",
      "Epoch: 718, Batch: 800, Avg. Loss: 0.0273097562789917\n",
      "Epoch: 718, Batch: 900, Avg. Loss: 0.0213180647790432\n",
      "Epoch: 719, Batch: 100, Avg. Loss: 0.023394408226013182\n",
      "Epoch: 719, Batch: 200, Avg. Loss: 0.023222415447235106\n",
      "Epoch: 719, Batch: 300, Avg. Loss: 0.019573102965950967\n",
      "Epoch: 719, Batch: 400, Avg. Loss: 0.021202416941523552\n",
      "Epoch: 719, Batch: 500, Avg. Loss: 0.023014275953173638\n",
      "Epoch: 719, Batch: 600, Avg. Loss: 0.026303131729364396\n",
      "Epoch: 719, Batch: 700, Avg. Loss: 0.02497104287147522\n",
      "Epoch: 719, Batch: 800, Avg. Loss: 0.025210376605391504\n",
      "Epoch: 719, Batch: 900, Avg. Loss: 0.024814186990261076\n",
      "Epoch: 720, Batch: 100, Avg. Loss: 0.02102613590657711\n",
      "Epoch: 720, Batch: 200, Avg. Loss: 0.019805994257330893\n",
      "Epoch: 720, Batch: 300, Avg. Loss: 0.022952323034405708\n",
      "Epoch: 720, Batch: 400, Avg. Loss: 0.024037755727767944\n",
      "Epoch: 720, Batch: 500, Avg. Loss: 0.022621639147400854\n",
      "Epoch: 720, Batch: 600, Avg. Loss: 0.024950043484568595\n",
      "Epoch: 720, Batch: 700, Avg. Loss: 0.021799018681049345\n",
      "Epoch: 720, Batch: 800, Avg. Loss: 0.02006608374416828\n",
      "Epoch: 720, Batch: 900, Avg. Loss: 0.026342196837067604\n",
      "Epoch: 721, Batch: 100, Avg. Loss: 0.018106683641672134\n",
      "Epoch: 721, Batch: 200, Avg. Loss: 0.023243126273155213\n",
      "Epoch: 721, Batch: 300, Avg. Loss: 0.027237394079566\n",
      "Epoch: 721, Batch: 400, Avg. Loss: 0.023292887285351754\n",
      "Epoch: 721, Batch: 500, Avg. Loss: 0.027300402969121933\n",
      "Epoch: 721, Batch: 600, Avg. Loss: 0.02553554318845272\n",
      "Epoch: 721, Batch: 700, Avg. Loss: 0.022500638216733933\n",
      "Epoch: 721, Batch: 800, Avg. Loss: 0.027495447397232056\n",
      "Epoch: 721, Batch: 900, Avg. Loss: 0.025295643284916878\n",
      "Epoch: 722, Batch: 100, Avg. Loss: 0.02705729492008686\n",
      "Epoch: 722, Batch: 200, Avg. Loss: 0.02792523019015789\n",
      "Epoch: 722, Batch: 300, Avg. Loss: 0.023472653105854988\n",
      "Epoch: 722, Batch: 400, Avg. Loss: 0.024174011275172234\n",
      "Epoch: 722, Batch: 500, Avg. Loss: 0.021938759982585907\n",
      "Epoch: 722, Batch: 600, Avg. Loss: 0.020371729135513307\n",
      "Epoch: 722, Batch: 700, Avg. Loss: 0.0210586953163147\n",
      "Epoch: 722, Batch: 800, Avg. Loss: 0.022233114913105964\n",
      "Epoch: 722, Batch: 900, Avg. Loss: 0.030336899533867836\n",
      "Epoch: 723, Batch: 100, Avg. Loss: 0.026819786727428435\n",
      "Epoch: 723, Batch: 200, Avg. Loss: 0.02397671580314636\n",
      "Epoch: 723, Batch: 300, Avg. Loss: 0.02142200656235218\n",
      "Epoch: 723, Batch: 400, Avg. Loss: 0.026840775534510614\n",
      "Epoch: 723, Batch: 500, Avg. Loss: 0.022653830274939536\n",
      "Epoch: 723, Batch: 600, Avg. Loss: 0.02248672306537628\n",
      "Epoch: 723, Batch: 700, Avg. Loss: 0.028981321826577186\n",
      "Epoch: 723, Batch: 800, Avg. Loss: 0.022518057897686957\n",
      "Epoch: 723, Batch: 900, Avg. Loss: 0.0204914653301239\n",
      "Epoch: 724, Batch: 100, Avg. Loss: 0.02412217013537884\n",
      "Epoch: 724, Batch: 200, Avg. Loss: 0.022972820922732354\n",
      "Epoch: 724, Batch: 300, Avg. Loss: 0.017579951137304307\n",
      "Epoch: 724, Batch: 400, Avg. Loss: 0.025894918218255042\n",
      "Epoch: 724, Batch: 500, Avg. Loss: 0.021209246143698693\n",
      "Epoch: 724, Batch: 600, Avg. Loss: 0.021225794926285745\n",
      "Epoch: 724, Batch: 700, Avg. Loss: 0.026431156173348427\n",
      "Epoch: 724, Batch: 800, Avg. Loss: 0.022094327062368392\n",
      "Epoch: 724, Batch: 900, Avg. Loss: 0.024470831975340845\n",
      "Epoch: 725, Batch: 100, Avg. Loss: 0.020276522785425185\n",
      "Epoch: 725, Batch: 200, Avg. Loss: 0.02286182314157486\n",
      "Epoch: 725, Batch: 300, Avg. Loss: 0.02397972919046879\n",
      "Epoch: 725, Batch: 400, Avg. Loss: 0.026997575759887694\n",
      "Epoch: 725, Batch: 500, Avg. Loss: 0.023372772186994552\n",
      "Epoch: 725, Batch: 600, Avg. Loss: 0.024812002927064897\n",
      "Epoch: 725, Batch: 700, Avg. Loss: 0.023856233581900598\n",
      "Epoch: 725, Batch: 800, Avg. Loss: 0.018293915018439293\n",
      "Epoch: 725, Batch: 900, Avg. Loss: 0.026029576659202577\n",
      "Epoch: 726, Batch: 100, Avg. Loss: 0.02677790842950344\n",
      "Epoch: 726, Batch: 200, Avg. Loss: 0.023921078145503996\n",
      "Epoch: 726, Batch: 300, Avg. Loss: 0.022597895190119743\n",
      "Epoch: 726, Batch: 400, Avg. Loss: 0.020945022851228713\n",
      "Epoch: 726, Batch: 500, Avg. Loss: 0.02447781875729561\n",
      "Epoch: 726, Batch: 600, Avg. Loss: 0.03235308736562729\n",
      "Epoch: 726, Batch: 700, Avg. Loss: 0.024073485136032104\n",
      "Epoch: 726, Batch: 800, Avg. Loss: 0.024172295778989793\n",
      "Epoch: 726, Batch: 900, Avg. Loss: 0.02451442800462246\n",
      "Epoch: 727, Batch: 100, Avg. Loss: 0.028249043226242065\n",
      "Epoch: 727, Batch: 200, Avg. Loss: 0.02358373433351517\n",
      "Epoch: 727, Batch: 300, Avg. Loss: 0.019084899127483367\n",
      "Epoch: 727, Batch: 400, Avg. Loss: 0.03285961747169495\n",
      "Epoch: 727, Batch: 500, Avg. Loss: 0.026762053295969965\n",
      "Epoch: 727, Batch: 600, Avg. Loss: 0.02153210334479809\n",
      "Epoch: 727, Batch: 700, Avg. Loss: 0.027455558478832246\n",
      "Epoch: 727, Batch: 800, Avg. Loss: 0.024273437410593034\n",
      "Epoch: 727, Batch: 900, Avg. Loss: 0.02842559099197388\n",
      "Epoch: 728, Batch: 100, Avg. Loss: 0.024458103254437445\n",
      "Epoch: 728, Batch: 200, Avg. Loss: 0.021937030851840972\n",
      "Epoch: 728, Batch: 300, Avg. Loss: 0.02076647400856018\n",
      "Epoch: 728, Batch: 400, Avg. Loss: 0.02620384655892849\n",
      "Epoch: 728, Batch: 500, Avg. Loss: 0.021006610095500946\n",
      "Epoch: 728, Batch: 600, Avg. Loss: 0.024335831180214883\n",
      "Epoch: 728, Batch: 700, Avg. Loss: 0.025281591862440108\n",
      "Epoch: 728, Batch: 800, Avg. Loss: 0.02450494036078453\n",
      "Epoch: 728, Batch: 900, Avg. Loss: 0.016895006224513055\n",
      "Epoch: 729, Batch: 100, Avg. Loss: 0.02737685278058052\n",
      "Epoch: 729, Batch: 200, Avg. Loss: 0.025610109344124794\n",
      "Epoch: 729, Batch: 300, Avg. Loss: 0.02216181717813015\n",
      "Epoch: 729, Batch: 400, Avg. Loss: 0.023137570023536683\n",
      "Epoch: 729, Batch: 500, Avg. Loss: 0.02671648859977722\n",
      "Epoch: 729, Batch: 600, Avg. Loss: 0.01861286796629429\n",
      "Epoch: 729, Batch: 700, Avg. Loss: 0.018502501249313356\n",
      "Epoch: 729, Batch: 800, Avg. Loss: 0.02346815139055252\n",
      "Epoch: 729, Batch: 900, Avg. Loss: 0.02889996625483036\n",
      "Epoch: 730, Batch: 100, Avg. Loss: 0.02515194095671177\n",
      "Epoch: 730, Batch: 200, Avg. Loss: 0.026910753697156908\n",
      "Epoch: 730, Batch: 300, Avg. Loss: 0.025502628087997435\n",
      "Epoch: 730, Batch: 400, Avg. Loss: 0.020073341503739357\n",
      "Epoch: 730, Batch: 500, Avg. Loss: 0.021495866626501083\n",
      "Epoch: 730, Batch: 600, Avg. Loss: 0.02489257737994194\n",
      "Epoch: 730, Batch: 700, Avg. Loss: 0.03099933162331581\n",
      "Epoch: 730, Batch: 800, Avg. Loss: 0.023291181921958923\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 730, Batch: 900, Avg. Loss: 0.02279872577637434\n",
      "Epoch: 731, Batch: 100, Avg. Loss: 0.022704027593135834\n",
      "Epoch: 731, Batch: 200, Avg. Loss: 0.02337091512978077\n",
      "Epoch: 731, Batch: 300, Avg. Loss: 0.024433588907122612\n",
      "Epoch: 731, Batch: 400, Avg. Loss: 0.028712920993566513\n",
      "Epoch: 731, Batch: 500, Avg. Loss: 0.024692037105560304\n",
      "Epoch: 731, Batch: 600, Avg. Loss: 0.02203501507639885\n",
      "Epoch: 731, Batch: 700, Avg. Loss: 0.02470577672123909\n",
      "Epoch: 731, Batch: 800, Avg. Loss: 0.019395371973514555\n",
      "Epoch: 731, Batch: 900, Avg. Loss: 0.029215594232082368\n",
      "Epoch: 732, Batch: 100, Avg. Loss: 0.028821638226509093\n",
      "Epoch: 732, Batch: 200, Avg. Loss: 0.01804768607020378\n",
      "Epoch: 732, Batch: 300, Avg. Loss: 0.030805741176009177\n",
      "Epoch: 732, Batch: 400, Avg. Loss: 0.025521568655967712\n",
      "Epoch: 732, Batch: 500, Avg. Loss: 0.025638047009706497\n",
      "Epoch: 732, Batch: 600, Avg. Loss: 0.023608353957533836\n",
      "Epoch: 732, Batch: 700, Avg. Loss: 0.017952180728316308\n",
      "Epoch: 732, Batch: 800, Avg. Loss: 0.020413002967834472\n",
      "Epoch: 732, Batch: 900, Avg. Loss: 0.02447514086961746\n",
      "Epoch: 733, Batch: 100, Avg. Loss: 0.0277053701877594\n",
      "Epoch: 733, Batch: 200, Avg. Loss: 0.023047023862600328\n",
      "Epoch: 733, Batch: 300, Avg. Loss: 0.02435501664876938\n",
      "Epoch: 733, Batch: 400, Avg. Loss: 0.02447338357567787\n",
      "Epoch: 733, Batch: 500, Avg. Loss: 0.027062820345163344\n",
      "Epoch: 733, Batch: 600, Avg. Loss: 0.023711463138461113\n",
      "Epoch: 733, Batch: 700, Avg. Loss: 0.027131215035915375\n",
      "Epoch: 733, Batch: 800, Avg. Loss: 0.02179251119494438\n",
      "Epoch: 733, Batch: 900, Avg. Loss: 0.02694934085011482\n",
      "Epoch: 734, Batch: 100, Avg. Loss: 0.022089961245656015\n",
      "Epoch: 734, Batch: 200, Avg. Loss: 0.028443776965141297\n",
      "Epoch: 734, Batch: 300, Avg. Loss: 0.029454414770007133\n",
      "Epoch: 734, Batch: 400, Avg. Loss: 0.02568564847111702\n",
      "Epoch: 734, Batch: 500, Avg. Loss: 0.022908203825354575\n",
      "Epoch: 734, Batch: 600, Avg. Loss: 0.0237246385961771\n",
      "Epoch: 734, Batch: 700, Avg. Loss: 0.024716736152768135\n",
      "Epoch: 734, Batch: 800, Avg. Loss: 0.021364036425948144\n",
      "Epoch: 734, Batch: 900, Avg. Loss: 0.022649205550551416\n",
      "Epoch: 735, Batch: 100, Avg. Loss: 0.02448860228061676\n",
      "Epoch: 735, Batch: 200, Avg. Loss: 0.027266592979431153\n",
      "Epoch: 735, Batch: 300, Avg. Loss: 0.02291394256055355\n",
      "Epoch: 735, Batch: 400, Avg. Loss: 0.023144601732492446\n",
      "Epoch: 735, Batch: 500, Avg. Loss: 0.025752427875995635\n",
      "Epoch: 735, Batch: 600, Avg. Loss: 0.023539111539721488\n",
      "Epoch: 735, Batch: 700, Avg. Loss: 0.021820293739438057\n",
      "Epoch: 735, Batch: 800, Avg. Loss: 0.02480637699365616\n",
      "Epoch: 735, Batch: 900, Avg. Loss: 0.02347414031624794\n",
      "Epoch: 736, Batch: 100, Avg. Loss: 0.029058345779776572\n",
      "Epoch: 736, Batch: 200, Avg. Loss: 0.025897428169846536\n",
      "Epoch: 736, Batch: 300, Avg. Loss: 0.025550736710429192\n",
      "Epoch: 736, Batch: 400, Avg. Loss: 0.030397285521030427\n",
      "Epoch: 736, Batch: 500, Avg. Loss: 0.02158598355948925\n",
      "Epoch: 736, Batch: 600, Avg. Loss: 0.022828483879566194\n",
      "Epoch: 736, Batch: 700, Avg. Loss: 0.018771705701947212\n",
      "Epoch: 736, Batch: 800, Avg. Loss: 0.02106991708278656\n",
      "Epoch: 736, Batch: 900, Avg. Loss: 0.02127233162522316\n",
      "Epoch: 737, Batch: 100, Avg. Loss: 0.020693126171827315\n",
      "Epoch: 737, Batch: 200, Avg. Loss: 0.025806472301483155\n",
      "Epoch: 737, Batch: 300, Avg. Loss: 0.020606384575366975\n",
      "Epoch: 737, Batch: 400, Avg. Loss: 0.027174033969640733\n",
      "Epoch: 737, Batch: 500, Avg. Loss: 0.02850145794451237\n",
      "Epoch: 737, Batch: 600, Avg. Loss: 0.025584435388445854\n",
      "Epoch: 737, Batch: 700, Avg. Loss: 0.018655808493494987\n",
      "Epoch: 737, Batch: 800, Avg. Loss: 0.03067306138575077\n",
      "Epoch: 737, Batch: 900, Avg. Loss: 0.023970887362957\n",
      "Epoch: 738, Batch: 100, Avg. Loss: 0.029725074246525766\n",
      "Epoch: 738, Batch: 200, Avg. Loss: 0.023920482397079466\n",
      "Epoch: 738, Batch: 300, Avg. Loss: 0.022913142889738083\n",
      "Epoch: 738, Batch: 400, Avg. Loss: 0.021450350657105446\n",
      "Epoch: 738, Batch: 500, Avg. Loss: 0.02195949584245682\n",
      "Epoch: 738, Batch: 600, Avg. Loss: 0.022286816388368606\n",
      "Epoch: 738, Batch: 700, Avg. Loss: 0.022500586435198783\n",
      "Epoch: 738, Batch: 800, Avg. Loss: 0.022803092524409294\n",
      "Epoch: 738, Batch: 900, Avg. Loss: 0.02419551119208336\n",
      "Epoch: 739, Batch: 100, Avg. Loss: 0.03163218200206756\n",
      "Epoch: 739, Batch: 200, Avg. Loss: 0.021181999146938323\n",
      "Epoch: 739, Batch: 300, Avg. Loss: 0.027360595613718033\n",
      "Epoch: 739, Batch: 400, Avg. Loss: 0.02784748613834381\n",
      "Epoch: 739, Batch: 500, Avg. Loss: 0.02349011041224003\n",
      "Epoch: 739, Batch: 600, Avg. Loss: 0.021532723754644392\n",
      "Epoch: 739, Batch: 700, Avg. Loss: 0.022767093926668168\n",
      "Epoch: 739, Batch: 800, Avg. Loss: 0.021464604362845422\n",
      "Epoch: 739, Batch: 900, Avg. Loss: 0.02113654084503651\n",
      "Epoch: 740, Batch: 100, Avg. Loss: 0.028734447956085204\n",
      "Epoch: 740, Batch: 200, Avg. Loss: 0.026380237936973572\n",
      "Epoch: 740, Batch: 300, Avg. Loss: 0.01929664470255375\n",
      "Epoch: 740, Batch: 400, Avg. Loss: 0.024355145320296288\n",
      "Epoch: 740, Batch: 500, Avg. Loss: 0.023738220110535623\n",
      "Epoch: 740, Batch: 600, Avg. Loss: 0.023446846455335617\n",
      "Epoch: 740, Batch: 700, Avg. Loss: 0.02377235308289528\n",
      "Epoch: 740, Batch: 800, Avg. Loss: 0.02842508852481842\n",
      "Epoch: 740, Batch: 900, Avg. Loss: 0.021615038216114043\n",
      "Epoch: 741, Batch: 100, Avg. Loss: 0.02530617281794548\n",
      "Epoch: 741, Batch: 200, Avg. Loss: 0.018171475902199745\n",
      "Epoch: 741, Batch: 300, Avg. Loss: 0.03148310780525208\n",
      "Epoch: 741, Batch: 400, Avg. Loss: 0.022303638830780983\n",
      "Epoch: 741, Batch: 500, Avg. Loss: 0.020132966861128806\n",
      "Epoch: 741, Batch: 600, Avg. Loss: 0.030287208557128905\n",
      "Epoch: 741, Batch: 700, Avg. Loss: 0.0239798404276371\n",
      "Epoch: 741, Batch: 800, Avg. Loss: 0.027327084094285966\n",
      "Epoch: 741, Batch: 900, Avg. Loss: 0.01977355167269707\n",
      "Epoch: 742, Batch: 100, Avg. Loss: 0.022207584232091904\n",
      "Epoch: 742, Batch: 200, Avg. Loss: 0.022873121425509454\n",
      "Epoch: 742, Batch: 300, Avg. Loss: 0.02577715113759041\n",
      "Epoch: 742, Batch: 400, Avg. Loss: 0.01967753790318966\n",
      "Epoch: 742, Batch: 500, Avg. Loss: 0.026394912227988242\n",
      "Epoch: 742, Batch: 600, Avg. Loss: 0.023641664162278177\n",
      "Epoch: 742, Batch: 700, Avg. Loss: 0.024411671310663224\n",
      "Epoch: 742, Batch: 800, Avg. Loss: 0.031242891401052474\n",
      "Epoch: 742, Batch: 900, Avg. Loss: 0.027010554522275923\n",
      "Epoch: 743, Batch: 100, Avg. Loss: 0.02289590537548065\n",
      "Epoch: 743, Batch: 200, Avg. Loss: 0.03265636920928955\n",
      "Epoch: 743, Batch: 300, Avg. Loss: 0.02327298916876316\n",
      "Epoch: 743, Batch: 400, Avg. Loss: 0.024156836047768592\n",
      "Epoch: 743, Batch: 500, Avg. Loss: 0.02054592587053776\n",
      "Epoch: 743, Batch: 600, Avg. Loss: 0.028227936252951624\n",
      "Epoch: 743, Batch: 700, Avg. Loss: 0.02309686705470085\n",
      "Epoch: 743, Batch: 800, Avg. Loss: 0.02379237174987793\n",
      "Epoch: 743, Batch: 900, Avg. Loss: 0.023469428792595864\n",
      "Epoch: 744, Batch: 100, Avg. Loss: 0.024507273137569428\n",
      "Epoch: 744, Batch: 200, Avg. Loss: 0.022451663985848427\n",
      "Epoch: 744, Batch: 300, Avg. Loss: 0.024314840510487556\n",
      "Epoch: 744, Batch: 400, Avg. Loss: 0.01955091707408428\n",
      "Epoch: 744, Batch: 500, Avg. Loss: 0.020944374576210976\n",
      "Epoch: 744, Batch: 600, Avg. Loss: 0.022660470008850096\n",
      "Epoch: 744, Batch: 700, Avg. Loss: 0.02325574226677418\n",
      "Epoch: 744, Batch: 800, Avg. Loss: 0.021745976507663727\n",
      "Epoch: 744, Batch: 900, Avg. Loss: 0.02446043223142624\n",
      "Epoch: 745, Batch: 100, Avg. Loss: 0.030723353400826454\n",
      "Epoch: 745, Batch: 200, Avg. Loss: 0.02358446054160595\n",
      "Epoch: 745, Batch: 300, Avg. Loss: 0.028716169595718384\n",
      "Epoch: 745, Batch: 400, Avg. Loss: 0.021525699645280838\n",
      "Epoch: 745, Batch: 500, Avg. Loss: 0.0228520667552948\n",
      "Epoch: 745, Batch: 600, Avg. Loss: 0.020951932296156883\n",
      "Epoch: 745, Batch: 700, Avg. Loss: 0.02248753659427166\n",
      "Epoch: 745, Batch: 800, Avg. Loss: 0.024930725395679473\n",
      "Epoch: 745, Batch: 900, Avg. Loss: 0.026507290229201317\n",
      "Epoch: 746, Batch: 100, Avg. Loss: 0.026262294501066208\n",
      "Epoch: 746, Batch: 200, Avg. Loss: 0.019321144446730613\n",
      "Epoch: 746, Batch: 300, Avg. Loss: 0.02126164585351944\n",
      "Epoch: 746, Batch: 400, Avg. Loss: 0.017885540127754212\n",
      "Epoch: 746, Batch: 500, Avg. Loss: 0.022721332907676698\n",
      "Epoch: 746, Batch: 600, Avg. Loss: 0.023261492028832435\n",
      "Epoch: 746, Batch: 700, Avg. Loss: 0.028128945976495744\n",
      "Epoch: 746, Batch: 800, Avg. Loss: 0.025328318551182748\n",
      "Epoch: 746, Batch: 900, Avg. Loss: 0.019205790683627127\n",
      "Epoch: 747, Batch: 100, Avg. Loss: 0.021689368486404417\n",
      "Epoch: 747, Batch: 200, Avg. Loss: 0.024385034143924712\n",
      "Epoch: 747, Batch: 300, Avg. Loss: 0.025030920282006264\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 747, Batch: 400, Avg. Loss: 0.024787764251232147\n",
      "Epoch: 747, Batch: 500, Avg. Loss: 0.0306721580773592\n",
      "Epoch: 747, Batch: 600, Avg. Loss: 0.018237388655543328\n",
      "Epoch: 747, Batch: 700, Avg. Loss: 0.020589891746640207\n",
      "Epoch: 747, Batch: 800, Avg. Loss: 0.02160613439977169\n",
      "Epoch: 747, Batch: 900, Avg. Loss: 0.017205991372466086\n",
      "Epoch: 748, Batch: 100, Avg. Loss: 0.01974893592298031\n",
      "Epoch: 748, Batch: 200, Avg. Loss: 0.018977775126695632\n",
      "Epoch: 748, Batch: 300, Avg. Loss: 0.024209999814629554\n",
      "Epoch: 748, Batch: 400, Avg. Loss: 0.026566606685519217\n",
      "Epoch: 748, Batch: 500, Avg. Loss: 0.02487255208194256\n",
      "Epoch: 748, Batch: 600, Avg. Loss: 0.025292383655905724\n",
      "Epoch: 748, Batch: 700, Avg. Loss: 0.020921699665486813\n",
      "Epoch: 748, Batch: 800, Avg. Loss: 0.017952789068222046\n",
      "Epoch: 748, Batch: 900, Avg. Loss: 0.0313465141505003\n",
      "Epoch: 749, Batch: 100, Avg. Loss: 0.018703225776553156\n",
      "Epoch: 749, Batch: 200, Avg. Loss: 0.025408252105116845\n",
      "Epoch: 749, Batch: 300, Avg. Loss: 0.014741995856165886\n",
      "Epoch: 749, Batch: 400, Avg. Loss: 0.022942665591835977\n",
      "Epoch: 749, Batch: 500, Avg. Loss: 0.024769013971090315\n",
      "Epoch: 749, Batch: 600, Avg. Loss: 0.023948964774608613\n",
      "Epoch: 749, Batch: 700, Avg. Loss: 0.026468208730220794\n",
      "Epoch: 749, Batch: 800, Avg. Loss: 0.027246757820248605\n",
      "Epoch: 749, Batch: 900, Avg. Loss: 0.022699288353323938\n",
      "Epoch: 750, Batch: 100, Avg. Loss: 0.02588132902979851\n",
      "Epoch: 750, Batch: 200, Avg. Loss: 0.0224322472512722\n",
      "Epoch: 750, Batch: 300, Avg. Loss: 0.021917571648955345\n",
      "Epoch: 750, Batch: 400, Avg. Loss: 0.020675608590245247\n",
      "Epoch: 750, Batch: 500, Avg. Loss: 0.025918125435709954\n",
      "Epoch: 750, Batch: 600, Avg. Loss: 0.023398807942867278\n",
      "Epoch: 750, Batch: 700, Avg. Loss: 0.026301538050174714\n",
      "Epoch: 750, Batch: 800, Avg. Loss: 0.021636726632714272\n",
      "Epoch: 750, Batch: 900, Avg. Loss: 0.02008082963526249\n",
      "Epoch: 751, Batch: 100, Avg. Loss: 0.027380582094192505\n",
      "Epoch: 751, Batch: 200, Avg. Loss: 0.018640313521027563\n",
      "Epoch: 751, Batch: 300, Avg. Loss: 0.02552719235420227\n",
      "Epoch: 751, Batch: 400, Avg. Loss: 0.02748756282031536\n",
      "Epoch: 751, Batch: 500, Avg. Loss: 0.02202554926276207\n",
      "Epoch: 751, Batch: 600, Avg. Loss: 0.02580378070473671\n",
      "Epoch: 751, Batch: 700, Avg. Loss: 0.023601827174425126\n",
      "Epoch: 751, Batch: 800, Avg. Loss: 0.02561964750289917\n",
      "Epoch: 751, Batch: 900, Avg. Loss: 0.025970011055469512\n",
      "Epoch: 752, Batch: 100, Avg. Loss: 0.023338244259357453\n",
      "Epoch: 752, Batch: 200, Avg. Loss: 0.02521727256476879\n",
      "Epoch: 752, Batch: 300, Avg. Loss: 0.02671028859913349\n",
      "Epoch: 752, Batch: 400, Avg. Loss: 0.02532125011086464\n",
      "Epoch: 752, Batch: 500, Avg. Loss: 0.01981769658625126\n",
      "Epoch: 752, Batch: 600, Avg. Loss: 0.018914003521203995\n",
      "Epoch: 752, Batch: 700, Avg. Loss: 0.028259219601750374\n",
      "Epoch: 752, Batch: 800, Avg. Loss: 0.021716131418943404\n",
      "Epoch: 752, Batch: 900, Avg. Loss: 0.026266376078128813\n",
      "Epoch: 753, Batch: 100, Avg. Loss: 0.021173879131674768\n",
      "Epoch: 753, Batch: 200, Avg. Loss: 0.02486733369529247\n",
      "Epoch: 753, Batch: 300, Avg. Loss: 0.022905662283301355\n",
      "Epoch: 753, Batch: 400, Avg. Loss: 0.02166779085993767\n",
      "Epoch: 753, Batch: 500, Avg. Loss: 0.024768717736005783\n",
      "Epoch: 753, Batch: 600, Avg. Loss: 0.018215206936001777\n",
      "Epoch: 753, Batch: 700, Avg. Loss: 0.019779863581061364\n",
      "Epoch: 753, Batch: 800, Avg. Loss: 0.01835649162530899\n",
      "Epoch: 753, Batch: 900, Avg. Loss: 0.026557982936501502\n",
      "Epoch: 754, Batch: 100, Avg. Loss: 0.031765344217419625\n",
      "Epoch: 754, Batch: 200, Avg. Loss: 0.02467563025653362\n",
      "Epoch: 754, Batch: 300, Avg. Loss: 0.03040284715592861\n",
      "Epoch: 754, Batch: 400, Avg. Loss: 0.020331166759133337\n",
      "Epoch: 754, Batch: 500, Avg. Loss: 0.024974401146173476\n",
      "Epoch: 754, Batch: 600, Avg. Loss: 0.0196349336206913\n",
      "Epoch: 754, Batch: 700, Avg. Loss: 0.02537519931793213\n",
      "Epoch: 754, Batch: 800, Avg. Loss: 0.023736140727996825\n",
      "Epoch: 754, Batch: 900, Avg. Loss: 0.023716099709272385\n",
      "Epoch: 755, Batch: 100, Avg. Loss: 0.022004144191741945\n",
      "Epoch: 755, Batch: 200, Avg. Loss: 0.026137960255146028\n",
      "Epoch: 755, Batch: 300, Avg. Loss: 0.02428869605064392\n",
      "Epoch: 755, Batch: 400, Avg. Loss: 0.020915792360901832\n",
      "Epoch: 755, Batch: 500, Avg. Loss: 0.022414885237812997\n",
      "Epoch: 755, Batch: 600, Avg. Loss: 0.026023701131343842\n",
      "Epoch: 755, Batch: 700, Avg. Loss: 0.021444492414593695\n",
      "Epoch: 755, Batch: 800, Avg. Loss: 0.025159434191882612\n",
      "Epoch: 755, Batch: 900, Avg. Loss: 0.019118655771017075\n",
      "Epoch: 756, Batch: 100, Avg. Loss: 0.022229568511247636\n",
      "Epoch: 756, Batch: 200, Avg. Loss: 0.02322413183748722\n",
      "Epoch: 756, Batch: 300, Avg. Loss: 0.02684605784714222\n",
      "Epoch: 756, Batch: 400, Avg. Loss: 0.02265990577638149\n",
      "Epoch: 756, Batch: 500, Avg. Loss: 0.022457408532500268\n",
      "Epoch: 756, Batch: 600, Avg. Loss: 0.02840001605451107\n",
      "Epoch: 756, Batch: 700, Avg. Loss: 0.023265554904937744\n",
      "Epoch: 756, Batch: 800, Avg. Loss: 0.022576317712664603\n",
      "Epoch: 756, Batch: 900, Avg. Loss: 0.022042204812169074\n",
      "Epoch: 757, Batch: 100, Avg. Loss: 0.023882425501942636\n",
      "Epoch: 757, Batch: 200, Avg. Loss: 0.024199336916208267\n",
      "Epoch: 757, Batch: 300, Avg. Loss: 0.026068816035985945\n",
      "Epoch: 757, Batch: 400, Avg. Loss: 0.02441548280417919\n",
      "Epoch: 757, Batch: 500, Avg. Loss: 0.022668534740805626\n",
      "Epoch: 757, Batch: 600, Avg. Loss: 0.031493489667773246\n",
      "Epoch: 757, Batch: 700, Avg. Loss: 0.018491050079464912\n",
      "Epoch: 757, Batch: 800, Avg. Loss: 0.02607282280921936\n",
      "Epoch: 757, Batch: 900, Avg. Loss: 0.02203754596412182\n",
      "Epoch: 758, Batch: 100, Avg. Loss: 0.025104871094226835\n",
      "Epoch: 758, Batch: 200, Avg. Loss: 0.02021663598716259\n",
      "Epoch: 758, Batch: 300, Avg. Loss: 0.01905472733080387\n",
      "Epoch: 758, Batch: 400, Avg. Loss: 0.02187843732535839\n",
      "Epoch: 758, Batch: 500, Avg. Loss: 0.02732585147023201\n",
      "Epoch: 758, Batch: 600, Avg. Loss: 0.025886134058237076\n",
      "Epoch: 758, Batch: 700, Avg. Loss: 0.023646358102560043\n",
      "Epoch: 758, Batch: 800, Avg. Loss: 0.022436876371502875\n",
      "Epoch: 758, Batch: 900, Avg. Loss: 0.025821039080619813\n",
      "Epoch: 759, Batch: 100, Avg. Loss: 0.02314947709441185\n",
      "Epoch: 759, Batch: 200, Avg. Loss: 0.026134766936302185\n",
      "Epoch: 759, Batch: 300, Avg. Loss: 0.02125845119357109\n",
      "Epoch: 759, Batch: 400, Avg. Loss: 0.023462750613689423\n",
      "Epoch: 759, Batch: 500, Avg. Loss: 0.0208392371237278\n",
      "Epoch: 759, Batch: 600, Avg. Loss: 0.022288749814033507\n",
      "Epoch: 759, Batch: 700, Avg. Loss: 0.021330891847610473\n",
      "Epoch: 759, Batch: 800, Avg. Loss: 0.02422370880842209\n",
      "Epoch: 759, Batch: 900, Avg. Loss: 0.020759896412491798\n",
      "Epoch: 760, Batch: 100, Avg. Loss: 0.021783144995570183\n",
      "Epoch: 760, Batch: 200, Avg. Loss: 0.022608506605029107\n",
      "Epoch: 760, Batch: 300, Avg. Loss: 0.024483354538679124\n",
      "Epoch: 760, Batch: 400, Avg. Loss: 0.020650831758975984\n",
      "Epoch: 760, Batch: 500, Avg. Loss: 0.022764082700014114\n",
      "Epoch: 760, Batch: 600, Avg. Loss: 0.02682356961071491\n",
      "Epoch: 760, Batch: 700, Avg. Loss: 0.023662183731794358\n",
      "Epoch: 760, Batch: 800, Avg. Loss: 0.020123007297515868\n",
      "Epoch: 760, Batch: 900, Avg. Loss: 0.02559517666697502\n",
      "Epoch: 761, Batch: 100, Avg. Loss: 0.023624529987573625\n",
      "Epoch: 761, Batch: 200, Avg. Loss: 0.019552178457379343\n",
      "Epoch: 761, Batch: 300, Avg. Loss: 0.023421656414866446\n",
      "Epoch: 761, Batch: 400, Avg. Loss: 0.025195116996765136\n",
      "Epoch: 761, Batch: 500, Avg. Loss: 0.02151022545993328\n",
      "Epoch: 761, Batch: 600, Avg. Loss: 0.02196670599281788\n",
      "Epoch: 761, Batch: 700, Avg. Loss: 0.0250916638225317\n",
      "Epoch: 761, Batch: 800, Avg. Loss: 0.031214505061507226\n",
      "Epoch: 761, Batch: 900, Avg. Loss: 0.02334600269794464\n",
      "Epoch: 762, Batch: 100, Avg. Loss: 0.02051725745201111\n",
      "Epoch: 762, Batch: 200, Avg. Loss: 0.02116997107863426\n",
      "Epoch: 762, Batch: 300, Avg. Loss: 0.025261490792036056\n",
      "Epoch: 762, Batch: 400, Avg. Loss: 0.019277468398213385\n",
      "Epoch: 762, Batch: 500, Avg. Loss: 0.02799348771572113\n",
      "Epoch: 762, Batch: 600, Avg. Loss: 0.0211808443069458\n",
      "Epoch: 762, Batch: 700, Avg. Loss: 0.02272392950952053\n",
      "Epoch: 762, Batch: 800, Avg. Loss: 0.024686057940125464\n",
      "Epoch: 762, Batch: 900, Avg. Loss: 0.024074551314115525\n",
      "Epoch: 763, Batch: 100, Avg. Loss: 0.027558913603425026\n",
      "Epoch: 763, Batch: 200, Avg. Loss: 0.017845207080245017\n",
      "Epoch: 763, Batch: 300, Avg. Loss: 0.024028768837451933\n",
      "Epoch: 763, Batch: 400, Avg. Loss: 0.02554484561085701\n",
      "Epoch: 763, Batch: 500, Avg. Loss: 0.020693816766142844\n",
      "Epoch: 763, Batch: 600, Avg. Loss: 0.023539765402674676\n",
      "Epoch: 763, Batch: 700, Avg. Loss: 0.022190840393304826\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 763, Batch: 800, Avg. Loss: 0.022439413592219352\n",
      "Epoch: 763, Batch: 900, Avg. Loss: 0.020612038597464563\n",
      "Epoch: 764, Batch: 100, Avg. Loss: 0.030024368464946747\n",
      "Epoch: 764, Batch: 200, Avg. Loss: 0.02223421573638916\n",
      "Epoch: 764, Batch: 300, Avg. Loss: 0.02417781062424183\n",
      "Epoch: 764, Batch: 400, Avg. Loss: 0.026656787022948266\n",
      "Epoch: 764, Batch: 500, Avg. Loss: 0.0192979297041893\n",
      "Epoch: 764, Batch: 600, Avg. Loss: 0.028252330049872397\n",
      "Epoch: 764, Batch: 700, Avg. Loss: 0.0227632424980402\n",
      "Epoch: 764, Batch: 800, Avg. Loss: 0.023707000389695167\n",
      "Epoch: 764, Batch: 900, Avg. Loss: 0.02315661743283272\n",
      "Epoch: 765, Batch: 100, Avg. Loss: 0.02488306388258934\n",
      "Epoch: 765, Batch: 200, Avg. Loss: 0.02378480538725853\n",
      "Epoch: 765, Batch: 300, Avg. Loss: 0.024250283762812616\n",
      "Epoch: 765, Batch: 400, Avg. Loss: 0.021772546768188475\n",
      "Epoch: 765, Batch: 500, Avg. Loss: 0.024514163061976433\n",
      "Epoch: 765, Batch: 600, Avg. Loss: 0.024896619990468025\n",
      "Epoch: 765, Batch: 700, Avg. Loss: 0.02491988964378834\n",
      "Epoch: 765, Batch: 800, Avg. Loss: 0.02115785889327526\n",
      "Epoch: 765, Batch: 900, Avg. Loss: 0.022083222419023513\n",
      "Epoch: 766, Batch: 100, Avg. Loss: 0.026009032130241395\n",
      "Epoch: 766, Batch: 200, Avg. Loss: 0.021010317355394364\n",
      "Epoch: 766, Batch: 300, Avg. Loss: 0.025477355048060417\n",
      "Epoch: 766, Batch: 400, Avg. Loss: 0.020277439951896667\n",
      "Epoch: 766, Batch: 500, Avg. Loss: 0.023607199639081956\n",
      "Epoch: 766, Batch: 600, Avg. Loss: 0.020886156260967254\n",
      "Epoch: 766, Batch: 700, Avg. Loss: 0.02145952083170414\n",
      "Epoch: 766, Batch: 800, Avg. Loss: 0.023413165658712386\n",
      "Epoch: 766, Batch: 900, Avg. Loss: 0.024540928304195405\n",
      "Epoch: 767, Batch: 100, Avg. Loss: 0.022621422559022903\n",
      "Epoch: 767, Batch: 200, Avg. Loss: 0.020453064292669295\n",
      "Epoch: 767, Batch: 300, Avg. Loss: 0.020703575164079665\n",
      "Epoch: 767, Batch: 400, Avg. Loss: 0.02119955502450466\n",
      "Epoch: 767, Batch: 500, Avg. Loss: 0.02990319550037384\n",
      "Epoch: 767, Batch: 600, Avg. Loss: 0.030847263038158417\n",
      "Epoch: 767, Batch: 700, Avg. Loss: 0.02454328641295433\n",
      "Epoch: 767, Batch: 800, Avg. Loss: 0.02267842583358288\n",
      "Epoch: 767, Batch: 900, Avg. Loss: 0.02406113862991333\n",
      "Epoch: 768, Batch: 100, Avg. Loss: 0.029476610496640205\n",
      "Epoch: 768, Batch: 200, Avg. Loss: 0.024013408944010734\n",
      "Epoch: 768, Batch: 300, Avg. Loss: 0.021892657428979875\n",
      "Epoch: 768, Batch: 400, Avg. Loss: 0.025640299022197725\n",
      "Epoch: 768, Batch: 500, Avg. Loss: 0.02146860636770725\n",
      "Epoch: 768, Batch: 600, Avg. Loss: 0.021988687440752983\n",
      "Epoch: 768, Batch: 700, Avg. Loss: 0.023707874193787573\n",
      "Epoch: 768, Batch: 800, Avg. Loss: 0.023918129727244376\n",
      "Epoch: 768, Batch: 900, Avg. Loss: 0.01853568375110626\n",
      "Epoch: 769, Batch: 100, Avg. Loss: 0.02331713691353798\n",
      "Epoch: 769, Batch: 200, Avg. Loss: 0.016955013424158095\n",
      "Epoch: 769, Batch: 300, Avg. Loss: 0.02224334478378296\n",
      "Epoch: 769, Batch: 400, Avg. Loss: 0.025304476991295816\n",
      "Epoch: 769, Batch: 500, Avg. Loss: 0.02103532612323761\n",
      "Epoch: 769, Batch: 600, Avg. Loss: 0.027155807167291643\n",
      "Epoch: 769, Batch: 700, Avg. Loss: 0.023541409447789194\n",
      "Epoch: 769, Batch: 800, Avg. Loss: 0.019929556772112847\n",
      "Epoch: 769, Batch: 900, Avg. Loss: 0.020078416988253592\n",
      "Epoch: 770, Batch: 100, Avg. Loss: 0.0230535801500082\n",
      "Epoch: 770, Batch: 200, Avg. Loss: 0.020649472922086715\n",
      "Epoch: 770, Batch: 300, Avg. Loss: 0.026130736842751502\n",
      "Epoch: 770, Batch: 400, Avg. Loss: 0.024083028808236123\n",
      "Epoch: 770, Batch: 500, Avg. Loss: 0.022469720765948296\n",
      "Epoch: 770, Batch: 600, Avg. Loss: 0.024573097601532937\n",
      "Epoch: 770, Batch: 700, Avg. Loss: 0.026391114965081217\n",
      "Epoch: 770, Batch: 800, Avg. Loss: 0.019780077189207077\n",
      "Epoch: 770, Batch: 900, Avg. Loss: 0.027852586284279824\n",
      "Epoch: 771, Batch: 100, Avg. Loss: 0.020449583157896994\n",
      "Epoch: 771, Batch: 200, Avg. Loss: 0.02662356197834015\n",
      "Epoch: 771, Batch: 300, Avg. Loss: 0.020592399314045905\n",
      "Epoch: 771, Batch: 400, Avg. Loss: 0.03070578210055828\n",
      "Epoch: 771, Batch: 500, Avg. Loss: 0.0216678848862648\n",
      "Epoch: 771, Batch: 600, Avg. Loss: 0.02831333301961422\n",
      "Epoch: 771, Batch: 700, Avg. Loss: 0.019693049043416976\n",
      "Epoch: 771, Batch: 800, Avg. Loss: 0.023986986875534057\n",
      "Epoch: 771, Batch: 900, Avg. Loss: 0.018531231358647347\n",
      "Epoch: 772, Batch: 100, Avg. Loss: 0.025298362374305727\n",
      "Epoch: 772, Batch: 200, Avg. Loss: 0.02527751848101616\n",
      "Epoch: 772, Batch: 300, Avg. Loss: 0.01889298766851425\n",
      "Epoch: 772, Batch: 400, Avg. Loss: 0.025409924164414406\n",
      "Epoch: 772, Batch: 500, Avg. Loss: 0.02001934200525284\n",
      "Epoch: 772, Batch: 600, Avg. Loss: 0.024587162658572196\n",
      "Epoch: 772, Batch: 700, Avg. Loss: 0.01683383747935295\n",
      "Epoch: 772, Batch: 800, Avg. Loss: 0.01937858834862709\n",
      "Epoch: 772, Batch: 900, Avg. Loss: 0.025265525132417678\n",
      "Epoch: 773, Batch: 100, Avg. Loss: 0.02123636655509472\n",
      "Epoch: 773, Batch: 200, Avg. Loss: 0.025311372876167297\n",
      "Epoch: 773, Batch: 300, Avg. Loss: 0.02301734305918217\n",
      "Epoch: 773, Batch: 400, Avg. Loss: 0.03140288405120373\n",
      "Epoch: 773, Batch: 500, Avg. Loss: 0.02038807325065136\n",
      "Epoch: 773, Batch: 600, Avg. Loss: 0.019351906478404998\n",
      "Epoch: 773, Batch: 700, Avg. Loss: 0.0206781654804945\n",
      "Epoch: 773, Batch: 800, Avg. Loss: 0.022444324269890785\n",
      "Epoch: 773, Batch: 900, Avg. Loss: 0.023423037305474282\n",
      "Epoch: 774, Batch: 100, Avg. Loss: 0.02351865589618683\n",
      "Epoch: 774, Batch: 200, Avg. Loss: 0.025125503838062287\n",
      "Epoch: 774, Batch: 300, Avg. Loss: 0.021600635275244712\n",
      "Epoch: 774, Batch: 400, Avg. Loss: 0.018358474895358086\n",
      "Epoch: 774, Batch: 500, Avg. Loss: 0.03023479714989662\n",
      "Epoch: 774, Batch: 600, Avg. Loss: 0.023451180309057237\n",
      "Epoch: 774, Batch: 700, Avg. Loss: 0.02187018923461437\n",
      "Epoch: 774, Batch: 800, Avg. Loss: 0.02272139724344015\n",
      "Epoch: 774, Batch: 900, Avg. Loss: 0.024283490404486657\n",
      "Epoch: 775, Batch: 100, Avg. Loss: 0.022698796764016153\n",
      "Epoch: 775, Batch: 200, Avg. Loss: 0.026182532534003256\n",
      "Epoch: 775, Batch: 300, Avg. Loss: 0.02425201319158077\n",
      "Epoch: 775, Batch: 400, Avg. Loss: 0.02011206030845642\n",
      "Epoch: 775, Batch: 500, Avg. Loss: 0.021506688073277474\n",
      "Epoch: 775, Batch: 600, Avg. Loss: 0.021884176433086395\n",
      "Epoch: 775, Batch: 700, Avg. Loss: 0.02376102566719055\n",
      "Epoch: 775, Batch: 800, Avg. Loss: 0.024840072244405747\n",
      "Epoch: 775, Batch: 900, Avg. Loss: 0.024930443540215494\n",
      "Epoch: 776, Batch: 100, Avg. Loss: 0.022182398587465287\n",
      "Epoch: 776, Batch: 200, Avg. Loss: 0.0183687774091959\n",
      "Epoch: 776, Batch: 300, Avg. Loss: 0.026473606526851653\n",
      "Epoch: 776, Batch: 400, Avg. Loss: 0.027460438907146455\n",
      "Epoch: 776, Batch: 500, Avg. Loss: 0.02752498582005501\n",
      "Epoch: 776, Batch: 600, Avg. Loss: 0.018714576810598373\n",
      "Epoch: 776, Batch: 700, Avg. Loss: 0.026125453263521194\n",
      "Epoch: 776, Batch: 800, Avg. Loss: 0.01893978476524353\n",
      "Epoch: 776, Batch: 900, Avg. Loss: 0.01986499011516571\n",
      "Epoch: 777, Batch: 100, Avg. Loss: 0.018962558284401895\n",
      "Epoch: 777, Batch: 200, Avg. Loss: 0.023891166597604752\n",
      "Epoch: 777, Batch: 300, Avg. Loss: 0.020665422976017\n",
      "Epoch: 777, Batch: 400, Avg. Loss: 0.024301787987351418\n",
      "Epoch: 777, Batch: 500, Avg. Loss: 0.024287262335419654\n",
      "Epoch: 777, Batch: 600, Avg. Loss: 0.021826456040143966\n",
      "Epoch: 777, Batch: 700, Avg. Loss: 0.022890577539801598\n",
      "Epoch: 777, Batch: 800, Avg. Loss: 0.02594035793095827\n",
      "Epoch: 777, Batch: 900, Avg. Loss: 0.02551313653588295\n",
      "Epoch: 778, Batch: 100, Avg. Loss: 0.022250527143478395\n",
      "Epoch: 778, Batch: 200, Avg. Loss: 0.02574440099298954\n",
      "Epoch: 778, Batch: 300, Avg. Loss: 0.02098774567246437\n",
      "Epoch: 778, Batch: 400, Avg. Loss: 0.02366128794848919\n",
      "Epoch: 778, Batch: 500, Avg. Loss: 0.025625750347971918\n",
      "Epoch: 778, Batch: 600, Avg. Loss: 0.01927664816379547\n",
      "Epoch: 778, Batch: 700, Avg. Loss: 0.020173755139112473\n",
      "Epoch: 778, Batch: 800, Avg. Loss: 0.024429698213934897\n",
      "Epoch: 778, Batch: 900, Avg. Loss: 0.023274360597133635\n",
      "Epoch: 779, Batch: 100, Avg. Loss: 0.02129484198987484\n",
      "Epoch: 779, Batch: 200, Avg. Loss: 0.02511782541871071\n",
      "Epoch: 779, Batch: 300, Avg. Loss: 0.021728478595614432\n",
      "Epoch: 779, Batch: 400, Avg. Loss: 0.022759534344077112\n",
      "Epoch: 779, Batch: 500, Avg. Loss: 0.019833995178341866\n",
      "Epoch: 779, Batch: 600, Avg. Loss: 0.022887360602617264\n",
      "Epoch: 779, Batch: 700, Avg. Loss: 0.024204803109169008\n",
      "Epoch: 779, Batch: 800, Avg. Loss: 0.02154562659561634\n",
      "Epoch: 779, Batch: 900, Avg. Loss: 0.0256606288254261\n",
      "Epoch: 780, Batch: 100, Avg. Loss: 0.018308723121881487\n",
      "Epoch: 780, Batch: 200, Avg. Loss: 0.024224044010043144\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 780, Batch: 300, Avg. Loss: 0.020204313695430756\n",
      "Epoch: 780, Batch: 400, Avg. Loss: 0.02023339442908764\n",
      "Epoch: 780, Batch: 500, Avg. Loss: 0.017688119933009148\n",
      "Epoch: 780, Batch: 600, Avg. Loss: 0.020132598504424094\n",
      "Epoch: 780, Batch: 700, Avg. Loss: 0.023835707828402518\n",
      "Epoch: 780, Batch: 800, Avg. Loss: 0.02427508495748043\n",
      "Epoch: 780, Batch: 900, Avg. Loss: 0.02439406767487526\n",
      "Epoch: 781, Batch: 100, Avg. Loss: 0.021228761598467826\n",
      "Epoch: 781, Batch: 200, Avg. Loss: 0.02056445777416229\n",
      "Epoch: 781, Batch: 300, Avg. Loss: 0.021773060709238054\n",
      "Epoch: 781, Batch: 400, Avg. Loss: 0.021190073788166047\n",
      "Epoch: 781, Batch: 500, Avg. Loss: 0.02125948153436184\n",
      "Epoch: 781, Batch: 600, Avg. Loss: 0.025663337111473082\n",
      "Epoch: 781, Batch: 700, Avg. Loss: 0.02284769803285599\n",
      "Epoch: 781, Batch: 800, Avg. Loss: 0.023335403650999068\n",
      "Epoch: 781, Batch: 900, Avg. Loss: 0.019352425411343573\n",
      "Epoch: 782, Batch: 100, Avg. Loss: 0.02312276288866997\n",
      "Epoch: 782, Batch: 200, Avg. Loss: 0.020166637673974037\n",
      "Epoch: 782, Batch: 300, Avg. Loss: 0.026460166797041895\n",
      "Epoch: 782, Batch: 400, Avg. Loss: 0.02439926117658615\n",
      "Epoch: 782, Batch: 500, Avg. Loss: 0.022421737760305406\n",
      "Epoch: 782, Batch: 600, Avg. Loss: 0.020037859976291657\n",
      "Epoch: 782, Batch: 700, Avg. Loss: 0.02529444858431816\n",
      "Epoch: 782, Batch: 800, Avg. Loss: 0.022538503035902976\n",
      "Epoch: 782, Batch: 900, Avg. Loss: 0.019972778409719467\n",
      "Epoch: 783, Batch: 100, Avg. Loss: 0.028956362009048463\n",
      "Epoch: 783, Batch: 200, Avg. Loss: 0.022497958317399026\n",
      "Epoch: 783, Batch: 300, Avg. Loss: 0.023398785591125487\n",
      "Epoch: 783, Batch: 400, Avg. Loss: 0.02751743920147419\n",
      "Epoch: 783, Batch: 500, Avg. Loss: 0.02492608666419983\n",
      "Epoch: 783, Batch: 600, Avg. Loss: 0.025625649243593215\n",
      "Epoch: 783, Batch: 700, Avg. Loss: 0.01817078948020935\n",
      "Epoch: 783, Batch: 800, Avg. Loss: 0.021003327816724777\n",
      "Epoch: 783, Batch: 900, Avg. Loss: 0.02489979110658169\n",
      "Epoch: 784, Batch: 100, Avg. Loss: 0.021995017230510713\n",
      "Epoch: 784, Batch: 200, Avg. Loss: 0.02140047825872898\n",
      "Epoch: 784, Batch: 300, Avg. Loss: 0.022969833984971048\n",
      "Epoch: 784, Batch: 400, Avg. Loss: 0.02817055568099022\n",
      "Epoch: 784, Batch: 500, Avg. Loss: 0.024533698484301567\n",
      "Epoch: 784, Batch: 600, Avg. Loss: 0.024456941708922385\n",
      "Epoch: 784, Batch: 700, Avg. Loss: 0.021172429174184798\n",
      "Epoch: 784, Batch: 800, Avg. Loss: 0.02723384998738766\n",
      "Epoch: 784, Batch: 900, Avg. Loss: 0.02258587710559368\n",
      "Epoch: 785, Batch: 100, Avg. Loss: 0.0222809898853302\n",
      "Epoch: 785, Batch: 200, Avg. Loss: 0.02370570883154869\n",
      "Epoch: 785, Batch: 300, Avg. Loss: 0.027923511415719984\n",
      "Epoch: 785, Batch: 400, Avg. Loss: 0.01965928740799427\n",
      "Epoch: 785, Batch: 500, Avg. Loss: 0.019478555992245673\n",
      "Epoch: 785, Batch: 600, Avg. Loss: 0.02489076852798462\n",
      "Epoch: 785, Batch: 700, Avg. Loss: 0.024650933817029\n",
      "Epoch: 785, Batch: 800, Avg. Loss: 0.02451444186270237\n",
      "Epoch: 785, Batch: 900, Avg. Loss: 0.0260116308927536\n",
      "Epoch: 786, Batch: 100, Avg. Loss: 0.018769573718309403\n",
      "Epoch: 786, Batch: 200, Avg. Loss: 0.022365705743432045\n",
      "Epoch: 786, Batch: 300, Avg. Loss: 0.024361803829669952\n",
      "Epoch: 786, Batch: 400, Avg. Loss: 0.019117643013596534\n",
      "Epoch: 786, Batch: 500, Avg. Loss: 0.025871339589357376\n",
      "Epoch: 786, Batch: 600, Avg. Loss: 0.01970823958516121\n",
      "Epoch: 786, Batch: 700, Avg. Loss: 0.026992675215005875\n",
      "Epoch: 786, Batch: 800, Avg. Loss: 0.023486654460430145\n",
      "Epoch: 786, Batch: 900, Avg. Loss: 0.019954124614596367\n",
      "Epoch: 787, Batch: 100, Avg. Loss: 0.020925282165408135\n",
      "Epoch: 787, Batch: 200, Avg. Loss: 0.024963767230510713\n",
      "Epoch: 787, Batch: 300, Avg. Loss: 0.019905436709523203\n",
      "Epoch: 787, Batch: 400, Avg. Loss: 0.019224211797118185\n",
      "Epoch: 787, Batch: 500, Avg. Loss: 0.024780569896101952\n",
      "Epoch: 787, Batch: 600, Avg. Loss: 0.022233593836426736\n",
      "Epoch: 787, Batch: 700, Avg. Loss: 0.019127156287431717\n",
      "Epoch: 787, Batch: 800, Avg. Loss: 0.019239948987960817\n",
      "Epoch: 787, Batch: 900, Avg. Loss: 0.029408916532993316\n",
      "Epoch: 788, Batch: 100, Avg. Loss: 0.017310553565621375\n",
      "Epoch: 788, Batch: 200, Avg. Loss: 0.020442560017108917\n",
      "Epoch: 788, Batch: 300, Avg. Loss: 0.02165734201669693\n",
      "Epoch: 788, Batch: 400, Avg. Loss: 0.023470871895551682\n",
      "Epoch: 788, Batch: 500, Avg. Loss: 0.02291735589504242\n",
      "Epoch: 788, Batch: 600, Avg. Loss: 0.0200418072193861\n",
      "Epoch: 788, Batch: 700, Avg. Loss: 0.018986893594264986\n",
      "Epoch: 788, Batch: 800, Avg. Loss: 0.025800243243575095\n",
      "Epoch: 788, Batch: 900, Avg. Loss: 0.02609622575342655\n",
      "Epoch: 789, Batch: 100, Avg. Loss: 0.01977493330836296\n",
      "Epoch: 789, Batch: 200, Avg. Loss: 0.022680654674768447\n",
      "Epoch: 789, Batch: 300, Avg. Loss: 0.02376631088554859\n",
      "Epoch: 789, Batch: 400, Avg. Loss: 0.021422750130295754\n",
      "Epoch: 789, Batch: 500, Avg. Loss: 0.020900065377354622\n",
      "Epoch: 789, Batch: 600, Avg. Loss: 0.02222052849829197\n",
      "Epoch: 789, Batch: 700, Avg. Loss: 0.02414435923099518\n",
      "Epoch: 789, Batch: 800, Avg. Loss: 0.023733893260359764\n",
      "Epoch: 789, Batch: 900, Avg. Loss: 0.017040880769491194\n",
      "Epoch: 790, Batch: 100, Avg. Loss: 0.02215238980948925\n",
      "Epoch: 790, Batch: 200, Avg. Loss: 0.022336813658475875\n",
      "Epoch: 790, Batch: 300, Avg. Loss: 0.02104142226278782\n",
      "Epoch: 790, Batch: 400, Avg. Loss: 0.02286889009177685\n",
      "Epoch: 790, Batch: 500, Avg. Loss: 0.023717786222696303\n",
      "Epoch: 790, Batch: 600, Avg. Loss: 0.018761986792087557\n",
      "Epoch: 790, Batch: 700, Avg. Loss: 0.022825721353292465\n",
      "Epoch: 790, Batch: 800, Avg. Loss: 0.02070744164288044\n",
      "Epoch: 790, Batch: 900, Avg. Loss: 0.023031051978468896\n",
      "Epoch: 791, Batch: 100, Avg. Loss: 0.023337695226073266\n",
      "Epoch: 791, Batch: 200, Avg. Loss: 0.02261052846908569\n",
      "Epoch: 791, Batch: 300, Avg. Loss: 0.015937720164656638\n",
      "Epoch: 791, Batch: 400, Avg. Loss: 0.023906925171613694\n",
      "Epoch: 791, Batch: 500, Avg. Loss: 0.018999013975262644\n",
      "Epoch: 791, Batch: 600, Avg. Loss: 0.023712603300809862\n",
      "Epoch: 791, Batch: 700, Avg. Loss: 0.021233813241124153\n",
      "Epoch: 791, Batch: 800, Avg. Loss: 0.022558863759040832\n",
      "Epoch: 791, Batch: 900, Avg. Loss: 0.026275872811675073\n",
      "Epoch: 792, Batch: 100, Avg. Loss: 0.029557212069630624\n",
      "Epoch: 792, Batch: 200, Avg. Loss: 0.02193866275250912\n",
      "Epoch: 792, Batch: 300, Avg. Loss: 0.021017877385020256\n",
      "Epoch: 792, Batch: 400, Avg. Loss: 0.018507120311260224\n",
      "Epoch: 792, Batch: 500, Avg. Loss: 0.020401823371648788\n",
      "Epoch: 792, Batch: 600, Avg. Loss: 0.02428447298705578\n",
      "Epoch: 792, Batch: 700, Avg. Loss: 0.02065516099333763\n",
      "Epoch: 792, Batch: 800, Avg. Loss: 0.025462078005075453\n",
      "Epoch: 792, Batch: 900, Avg. Loss: 0.020165346041321756\n",
      "Epoch: 793, Batch: 100, Avg. Loss: 0.025943304523825646\n",
      "Epoch: 793, Batch: 200, Avg. Loss: 0.019737007543444634\n",
      "Epoch: 793, Batch: 300, Avg. Loss: 0.01883750356733799\n",
      "Epoch: 793, Batch: 400, Avg. Loss: 0.025343049094080927\n",
      "Epoch: 793, Batch: 500, Avg. Loss: 0.0257273268699646\n",
      "Epoch: 793, Batch: 600, Avg. Loss: 0.019862277358770372\n",
      "Epoch: 793, Batch: 700, Avg. Loss: 0.02104256346821785\n",
      "Epoch: 793, Batch: 800, Avg. Loss: 0.023216803036630155\n",
      "Epoch: 793, Batch: 900, Avg. Loss: 0.022976599782705307\n",
      "Epoch: 794, Batch: 100, Avg. Loss: 0.02254864163696766\n",
      "Epoch: 794, Batch: 200, Avg. Loss: 0.021934439167380332\n",
      "Epoch: 794, Batch: 300, Avg. Loss: 0.01743127912282944\n",
      "Epoch: 794, Batch: 400, Avg. Loss: 0.022124868631362916\n",
      "Epoch: 794, Batch: 500, Avg. Loss: 0.021544380262494087\n",
      "Epoch: 794, Batch: 600, Avg. Loss: 0.01952928140759468\n",
      "Epoch: 794, Batch: 700, Avg. Loss: 0.025903905034065245\n",
      "Epoch: 794, Batch: 800, Avg. Loss: 0.02483331210911274\n",
      "Epoch: 794, Batch: 900, Avg. Loss: 0.022323965057730676\n",
      "Epoch: 795, Batch: 100, Avg. Loss: 0.021076350957155227\n",
      "Epoch: 795, Batch: 200, Avg. Loss: 0.02768842615187168\n",
      "Epoch: 795, Batch: 300, Avg. Loss: 0.0234237639605999\n",
      "Epoch: 795, Batch: 400, Avg. Loss: 0.020153849199414254\n",
      "Epoch: 795, Batch: 500, Avg. Loss: 0.020512229204177855\n",
      "Epoch: 795, Batch: 600, Avg. Loss: 0.02340306594967842\n",
      "Epoch: 795, Batch: 700, Avg. Loss: 0.020873202309012413\n",
      "Epoch: 795, Batch: 800, Avg. Loss: 0.022413027510046958\n",
      "Epoch: 795, Batch: 900, Avg. Loss: 0.02445067510008812\n",
      "Epoch: 796, Batch: 100, Avg. Loss: 0.02103444702923298\n",
      "Epoch: 796, Batch: 200, Avg. Loss: 0.022314359545707704\n",
      "Epoch: 796, Batch: 300, Avg. Loss: 0.017415842562913893\n",
      "Epoch: 796, Batch: 400, Avg. Loss: 0.019724818244576454\n",
      "Epoch: 796, Batch: 500, Avg. Loss: 0.02219105564057827\n",
      "Epoch: 796, Batch: 600, Avg. Loss: 0.03180696569383144\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 796, Batch: 700, Avg. Loss: 0.017963137403130533\n",
      "Epoch: 796, Batch: 800, Avg. Loss: 0.028249157592654228\n",
      "Epoch: 796, Batch: 900, Avg. Loss: 0.019422310665249826\n",
      "Epoch: 797, Batch: 100, Avg. Loss: 0.018441743776202203\n",
      "Epoch: 797, Batch: 200, Avg. Loss: 0.02614381358027458\n",
      "Epoch: 797, Batch: 300, Avg. Loss: 0.022580027729272843\n",
      "Epoch: 797, Batch: 400, Avg. Loss: 0.02822003111243248\n",
      "Epoch: 797, Batch: 500, Avg. Loss: 0.027270831540226935\n",
      "Epoch: 797, Batch: 600, Avg. Loss: 0.018849817141890525\n",
      "Epoch: 797, Batch: 700, Avg. Loss: 0.022539758980274202\n",
      "Epoch: 797, Batch: 800, Avg. Loss: 0.031286423280835154\n",
      "Epoch: 797, Batch: 900, Avg. Loss: 0.02308042883872986\n",
      "Epoch: 798, Batch: 100, Avg. Loss: 0.024766593500971795\n",
      "Epoch: 798, Batch: 200, Avg. Loss: 0.021183686330914497\n",
      "Epoch: 798, Batch: 300, Avg. Loss: 0.020835923999547958\n",
      "Epoch: 798, Batch: 400, Avg. Loss: 0.022102388888597488\n",
      "Epoch: 798, Batch: 500, Avg. Loss: 0.025302854254841804\n",
      "Epoch: 798, Batch: 600, Avg. Loss: 0.016557396426796912\n",
      "Epoch: 798, Batch: 700, Avg. Loss: 0.022710879370570182\n",
      "Epoch: 798, Batch: 800, Avg. Loss: 0.01983834408223629\n",
      "Epoch: 798, Batch: 900, Avg. Loss: 0.021957379654049872\n",
      "Epoch: 799, Batch: 100, Avg. Loss: 0.027190678417682648\n",
      "Epoch: 799, Batch: 200, Avg. Loss: 0.026476111263036728\n",
      "Epoch: 799, Batch: 300, Avg. Loss: 0.02352308712899685\n",
      "Epoch: 799, Batch: 400, Avg. Loss: 0.020259609818458556\n",
      "Epoch: 799, Batch: 500, Avg. Loss: 0.022029145210981368\n",
      "Epoch: 799, Batch: 600, Avg. Loss: 0.023083654716610908\n",
      "Epoch: 799, Batch: 700, Avg. Loss: 0.02304152272641659\n",
      "Epoch: 799, Batch: 800, Avg. Loss: 0.022927728071808814\n",
      "Epoch: 799, Batch: 900, Avg. Loss: 0.024421532079577445\n",
      "Epoch: 800, Batch: 100, Avg. Loss: 0.02067713864147663\n",
      "Epoch: 800, Batch: 200, Avg. Loss: 0.024178813323378564\n",
      "Epoch: 800, Batch: 300, Avg. Loss: 0.016818852350115775\n",
      "Epoch: 800, Batch: 400, Avg. Loss: 0.01992765560746193\n",
      "Epoch: 800, Batch: 500, Avg. Loss: 0.021556359007954596\n",
      "Epoch: 800, Batch: 600, Avg. Loss: 0.019264243245124817\n",
      "Epoch: 800, Batch: 700, Avg. Loss: 0.024468479454517366\n",
      "Epoch: 800, Batch: 800, Avg. Loss: 0.02458831384778023\n",
      "Epoch: 800, Batch: 900, Avg. Loss: 0.019992823302745818\n",
      "Epoch: 801, Batch: 100, Avg. Loss: 0.02442376911640167\n",
      "Epoch: 801, Batch: 200, Avg. Loss: 0.023910007253289223\n",
      "Epoch: 801, Batch: 300, Avg. Loss: 0.02505873069167137\n",
      "Epoch: 801, Batch: 400, Avg. Loss: 0.019637469798326493\n",
      "Epoch: 801, Batch: 500, Avg. Loss: 0.02492177702486515\n",
      "Epoch: 801, Batch: 600, Avg. Loss: 0.0201938147097826\n",
      "Epoch: 801, Batch: 700, Avg. Loss: 0.02386606700718403\n",
      "Epoch: 801, Batch: 800, Avg. Loss: 0.022483886405825614\n",
      "Epoch: 801, Batch: 900, Avg. Loss: 0.02528891935944557\n",
      "Epoch: 802, Batch: 100, Avg. Loss: 0.02582287922501564\n",
      "Epoch: 802, Batch: 200, Avg. Loss: 0.021076443418860436\n",
      "Epoch: 802, Batch: 300, Avg. Loss: 0.02166561156511307\n",
      "Epoch: 802, Batch: 400, Avg. Loss: 0.025552203953266142\n",
      "Epoch: 802, Batch: 500, Avg. Loss: 0.021218839511275293\n",
      "Epoch: 802, Batch: 600, Avg. Loss: 0.02034223437309265\n",
      "Epoch: 802, Batch: 700, Avg. Loss: 0.02477125771343708\n",
      "Epoch: 802, Batch: 800, Avg. Loss: 0.021451383382081985\n",
      "Epoch: 802, Batch: 900, Avg. Loss: 0.022606816813349723\n",
      "Epoch: 803, Batch: 100, Avg. Loss: 0.020003219097852708\n",
      "Epoch: 803, Batch: 200, Avg. Loss: 0.026327935382723808\n",
      "Epoch: 803, Batch: 300, Avg. Loss: 0.024425130411982535\n",
      "Epoch: 803, Batch: 400, Avg. Loss: 0.023053335025906563\n",
      "Epoch: 803, Batch: 500, Avg. Loss: 0.02261421851813793\n",
      "Epoch: 803, Batch: 600, Avg. Loss: 0.027974403575062753\n",
      "Epoch: 803, Batch: 700, Avg. Loss: 0.02007599890232086\n",
      "Epoch: 803, Batch: 800, Avg. Loss: 0.0190547426789999\n",
      "Epoch: 803, Batch: 900, Avg. Loss: 0.021697884947061537\n",
      "Epoch: 804, Batch: 100, Avg. Loss: 0.02766260355710983\n",
      "Epoch: 804, Batch: 200, Avg. Loss: 0.0246798837184906\n",
      "Epoch: 804, Batch: 300, Avg. Loss: 0.025080953910946845\n",
      "Epoch: 804, Batch: 400, Avg. Loss: 0.018012490570545197\n",
      "Epoch: 804, Batch: 500, Avg. Loss: 0.023284938037395477\n",
      "Epoch: 804, Batch: 600, Avg. Loss: 0.022589311450719834\n",
      "Epoch: 804, Batch: 700, Avg. Loss: 0.0236933296918869\n",
      "Epoch: 804, Batch: 800, Avg. Loss: 0.020968720242381098\n",
      "Epoch: 804, Batch: 900, Avg. Loss: 0.02240706145763397\n",
      "Epoch: 805, Batch: 100, Avg. Loss: 0.019882729351520537\n",
      "Epoch: 805, Batch: 200, Avg. Loss: 0.028387683108448983\n",
      "Epoch: 805, Batch: 300, Avg. Loss: 0.019689121767878532\n",
      "Epoch: 805, Batch: 400, Avg. Loss: 0.024933314248919487\n",
      "Epoch: 805, Batch: 500, Avg. Loss: 0.02224348545074463\n",
      "Epoch: 805, Batch: 600, Avg. Loss: 0.022423379570245743\n",
      "Epoch: 805, Batch: 700, Avg. Loss: 0.02713377580046654\n",
      "Epoch: 805, Batch: 800, Avg. Loss: 0.020120626986026763\n",
      "Epoch: 805, Batch: 900, Avg. Loss: 0.02364320442080498\n",
      "Epoch: 806, Batch: 100, Avg. Loss: 0.021743605583906173\n",
      "Epoch: 806, Batch: 200, Avg. Loss: 0.024214785993099213\n",
      "Epoch: 806, Batch: 300, Avg. Loss: 0.018905741274356843\n",
      "Epoch: 806, Batch: 400, Avg. Loss: 0.022029090449213983\n",
      "Epoch: 806, Batch: 500, Avg. Loss: 0.019853935539722443\n",
      "Epoch: 806, Batch: 600, Avg. Loss: 0.030998827889561653\n",
      "Epoch: 806, Batch: 700, Avg. Loss: 0.023455874472856523\n",
      "Epoch: 806, Batch: 800, Avg. Loss: 0.022112258225679398\n",
      "Epoch: 806, Batch: 900, Avg. Loss: 0.02011999808251858\n",
      "Epoch: 807, Batch: 100, Avg. Loss: 0.021714004501700402\n",
      "Epoch: 807, Batch: 200, Avg. Loss: 0.02059282287955284\n",
      "Epoch: 807, Batch: 300, Avg. Loss: 0.019607649520039558\n",
      "Epoch: 807, Batch: 400, Avg. Loss: 0.024307738095521926\n",
      "Epoch: 807, Batch: 500, Avg. Loss: 0.019800784289836882\n",
      "Epoch: 807, Batch: 600, Avg. Loss: 0.028795862272381782\n",
      "Epoch: 807, Batch: 700, Avg. Loss: 0.02598857320845127\n",
      "Epoch: 807, Batch: 800, Avg. Loss: 0.021689334362745286\n",
      "Epoch: 807, Batch: 900, Avg. Loss: 0.01730752319097519\n",
      "Epoch: 808, Batch: 100, Avg. Loss: 0.02110030360519886\n",
      "Epoch: 808, Batch: 200, Avg. Loss: 0.019563556984066963\n",
      "Epoch: 808, Batch: 300, Avg. Loss: 0.020865519493818284\n",
      "Epoch: 808, Batch: 400, Avg. Loss: 0.02712469853460789\n",
      "Epoch: 808, Batch: 500, Avg. Loss: 0.023305682465434074\n",
      "Epoch: 808, Batch: 600, Avg. Loss: 0.023350549712777138\n",
      "Epoch: 808, Batch: 700, Avg. Loss: 0.021016713157296182\n",
      "Epoch: 808, Batch: 800, Avg. Loss: 0.02240758128464222\n",
      "Epoch: 808, Batch: 900, Avg. Loss: 0.021794553995132446\n",
      "Epoch: 809, Batch: 100, Avg. Loss: 0.026445204168558122\n",
      "Epoch: 809, Batch: 200, Avg. Loss: 0.01982241079211235\n",
      "Epoch: 809, Batch: 300, Avg. Loss: 0.023049604967236518\n",
      "Epoch: 809, Batch: 400, Avg. Loss: 0.022031465470790865\n",
      "Epoch: 809, Batch: 500, Avg. Loss: 0.02091198816895485\n",
      "Epoch: 809, Batch: 600, Avg. Loss: 0.029541490972042082\n",
      "Epoch: 809, Batch: 700, Avg. Loss: 0.0231370535492897\n",
      "Epoch: 809, Batch: 800, Avg. Loss: 0.020118904411792756\n",
      "Epoch: 809, Batch: 900, Avg. Loss: 0.020763411298394204\n",
      "Epoch: 810, Batch: 100, Avg. Loss: 0.023053724616765976\n",
      "Epoch: 810, Batch: 200, Avg. Loss: 0.01996721275150776\n",
      "Epoch: 810, Batch: 300, Avg. Loss: 0.027111406102776528\n",
      "Epoch: 810, Batch: 400, Avg. Loss: 0.01862434633076191\n",
      "Epoch: 810, Batch: 500, Avg. Loss: 0.022375839203596114\n",
      "Epoch: 810, Batch: 600, Avg. Loss: 0.018153648599982263\n",
      "Epoch: 810, Batch: 700, Avg. Loss: 0.023095596879720688\n",
      "Epoch: 810, Batch: 800, Avg. Loss: 0.025330060422420503\n",
      "Epoch: 810, Batch: 900, Avg. Loss: 0.0225449588149786\n",
      "Epoch: 811, Batch: 100, Avg. Loss: 0.02387983910739422\n",
      "Epoch: 811, Batch: 200, Avg. Loss: 0.02521116830408573\n",
      "Epoch: 811, Batch: 300, Avg. Loss: 0.02327907718718052\n",
      "Epoch: 811, Batch: 400, Avg. Loss: 0.02779000163078308\n",
      "Epoch: 811, Batch: 500, Avg. Loss: 0.02457584224641323\n",
      "Epoch: 811, Batch: 600, Avg. Loss: 0.024370453581213952\n",
      "Epoch: 811, Batch: 700, Avg. Loss: 0.019124692752957344\n",
      "Epoch: 811, Batch: 800, Avg. Loss: 0.022195075303316117\n",
      "Epoch: 811, Batch: 900, Avg. Loss: 0.019350368455052377\n",
      "Epoch: 812, Batch: 100, Avg. Loss: 0.021998389661312102\n",
      "Epoch: 812, Batch: 200, Avg. Loss: 0.026824956014752387\n",
      "Epoch: 812, Batch: 300, Avg. Loss: 0.03361514963209629\n",
      "Epoch: 812, Batch: 400, Avg. Loss: 0.01820862740278244\n",
      "Epoch: 812, Batch: 500, Avg. Loss: 0.017320888116955757\n",
      "Epoch: 812, Batch: 600, Avg. Loss: 0.02560656249523163\n",
      "Epoch: 812, Batch: 700, Avg. Loss: 0.02186767414212227\n",
      "Epoch: 812, Batch: 800, Avg. Loss: 0.018810254856944084\n",
      "Epoch: 812, Batch: 900, Avg. Loss: 0.023527024388313292\n",
      "Epoch: 813, Batch: 100, Avg. Loss: 0.01925892323255539\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 813, Batch: 200, Avg. Loss: 0.02947246514260769\n",
      "Epoch: 813, Batch: 300, Avg. Loss: 0.026352944150567055\n",
      "Epoch: 813, Batch: 400, Avg. Loss: 0.020303751826286316\n",
      "Epoch: 813, Batch: 500, Avg. Loss: 0.022899857088923455\n",
      "Epoch: 813, Batch: 600, Avg. Loss: 0.020746600851416587\n",
      "Epoch: 813, Batch: 700, Avg. Loss: 0.01924623526632786\n",
      "Epoch: 813, Batch: 800, Avg. Loss: 0.020164151787757875\n",
      "Epoch: 813, Batch: 900, Avg. Loss: 0.025351856872439384\n",
      "Epoch: 814, Batch: 100, Avg. Loss: 0.02205348163843155\n",
      "Epoch: 814, Batch: 200, Avg. Loss: 0.025371066331863402\n",
      "Epoch: 814, Batch: 300, Avg. Loss: 0.019662492796778678\n",
      "Epoch: 814, Batch: 400, Avg. Loss: 0.018023076355457305\n",
      "Epoch: 814, Batch: 500, Avg. Loss: 0.030261998996138573\n",
      "Epoch: 814, Batch: 600, Avg. Loss: 0.018186249434947968\n",
      "Epoch: 814, Batch: 700, Avg. Loss: 0.020938803255558015\n",
      "Epoch: 814, Batch: 800, Avg. Loss: 0.021491827741265297\n",
      "Epoch: 814, Batch: 900, Avg. Loss: 0.02565384604036808\n",
      "Epoch: 815, Batch: 100, Avg. Loss: 0.022390032708644866\n",
      "Epoch: 815, Batch: 200, Avg. Loss: 0.021707882061600684\n",
      "Epoch: 815, Batch: 300, Avg. Loss: 0.01952505648136139\n",
      "Epoch: 815, Batch: 400, Avg. Loss: 0.024030088409781457\n",
      "Epoch: 815, Batch: 500, Avg. Loss: 0.021555087193846703\n",
      "Epoch: 815, Batch: 600, Avg. Loss: 0.025602742582559585\n",
      "Epoch: 815, Batch: 700, Avg. Loss: 0.02837939016520977\n",
      "Epoch: 815, Batch: 800, Avg. Loss: 0.024106762930750848\n",
      "Epoch: 815, Batch: 900, Avg. Loss: 0.01745522439479828\n",
      "Epoch: 816, Batch: 100, Avg. Loss: 0.021189628839492797\n",
      "Epoch: 816, Batch: 200, Avg. Loss: 0.02358725421130657\n",
      "Epoch: 816, Batch: 300, Avg. Loss: 0.021858318001031874\n",
      "Epoch: 816, Batch: 400, Avg. Loss: 0.026796439290046693\n",
      "Epoch: 816, Batch: 500, Avg. Loss: 0.025710449442267417\n",
      "Epoch: 816, Batch: 600, Avg. Loss: 0.01987983949482441\n",
      "Epoch: 816, Batch: 700, Avg. Loss: 0.02128937393426895\n",
      "Epoch: 816, Batch: 800, Avg. Loss: 0.0256763955950737\n",
      "Epoch: 816, Batch: 900, Avg. Loss: 0.024331681579351425\n",
      "Epoch: 817, Batch: 100, Avg. Loss: 0.020019900649785996\n",
      "Epoch: 817, Batch: 200, Avg. Loss: 0.023066347315907478\n",
      "Epoch: 817, Batch: 300, Avg. Loss: 0.02761665903031826\n",
      "Epoch: 817, Batch: 400, Avg. Loss: 0.01960432879626751\n",
      "Epoch: 817, Batch: 500, Avg. Loss: 0.020487281903624536\n",
      "Epoch: 817, Batch: 600, Avg. Loss: 0.022634292989969253\n",
      "Epoch: 817, Batch: 700, Avg. Loss: 0.02232505165040493\n",
      "Epoch: 817, Batch: 800, Avg. Loss: 0.02225206308066845\n",
      "Epoch: 817, Batch: 900, Avg. Loss: 0.02435316063463688\n",
      "Epoch: 818, Batch: 100, Avg. Loss: 0.022801144048571587\n",
      "Epoch: 818, Batch: 200, Avg. Loss: 0.021905667185783385\n",
      "Epoch: 818, Batch: 300, Avg. Loss: 0.02486832432448864\n",
      "Epoch: 818, Batch: 400, Avg. Loss: 0.025644989237189292\n",
      "Epoch: 818, Batch: 500, Avg. Loss: 0.019821507185697557\n",
      "Epoch: 818, Batch: 600, Avg. Loss: 0.017446851953864098\n",
      "Epoch: 818, Batch: 700, Avg. Loss: 0.018528796583414078\n",
      "Epoch: 818, Batch: 800, Avg. Loss: 0.022840513214468956\n",
      "Epoch: 818, Batch: 900, Avg. Loss: 0.02328450933098793\n",
      "Epoch: 819, Batch: 100, Avg. Loss: 0.01772776447236538\n",
      "Epoch: 819, Batch: 200, Avg. Loss: 0.02781047947704792\n",
      "Epoch: 819, Batch: 300, Avg. Loss: 0.021205006539821623\n",
      "Epoch: 819, Batch: 400, Avg. Loss: 0.025499180033802984\n",
      "Epoch: 819, Batch: 500, Avg. Loss: 0.021346345394849777\n",
      "Epoch: 819, Batch: 600, Avg. Loss: 0.021617165952920913\n",
      "Epoch: 819, Batch: 700, Avg. Loss: 0.02069921225309372\n",
      "Epoch: 819, Batch: 800, Avg. Loss: 0.02424462862312794\n",
      "Epoch: 819, Batch: 900, Avg. Loss: 0.023540485724806784\n",
      "Epoch: 820, Batch: 100, Avg. Loss: 0.019645553603768347\n",
      "Epoch: 820, Batch: 200, Avg. Loss: 0.029794363155961037\n",
      "Epoch: 820, Batch: 300, Avg. Loss: 0.022662248015403747\n",
      "Epoch: 820, Batch: 400, Avg. Loss: 0.025265556648373603\n",
      "Epoch: 820, Batch: 500, Avg. Loss: 0.02159553609788418\n",
      "Epoch: 820, Batch: 600, Avg. Loss: 0.020732883438467978\n",
      "Epoch: 820, Batch: 700, Avg. Loss: 0.01677679128944874\n",
      "Epoch: 820, Batch: 800, Avg. Loss: 0.023985403254628182\n",
      "Epoch: 820, Batch: 900, Avg. Loss: 0.018611653745174407\n",
      "Epoch: 821, Batch: 100, Avg. Loss: 0.01889260932803154\n",
      "Epoch: 821, Batch: 200, Avg. Loss: 0.020441802889108657\n",
      "Epoch: 821, Batch: 300, Avg. Loss: 0.019733240380883216\n",
      "Epoch: 821, Batch: 400, Avg. Loss: 0.021376348063349725\n",
      "Epoch: 821, Batch: 500, Avg. Loss: 0.021182999312877655\n",
      "Epoch: 821, Batch: 600, Avg. Loss: 0.0224886467307806\n",
      "Epoch: 821, Batch: 700, Avg. Loss: 0.019934343099594118\n",
      "Epoch: 821, Batch: 800, Avg. Loss: 0.023145103231072425\n",
      "Epoch: 821, Batch: 900, Avg. Loss: 0.024794011786580087\n",
      "Epoch: 822, Batch: 100, Avg. Loss: 0.02286424197256565\n",
      "Epoch: 822, Batch: 200, Avg. Loss: 0.0231131312251091\n",
      "Epoch: 822, Batch: 300, Avg. Loss: 0.022966172620654104\n",
      "Epoch: 822, Batch: 400, Avg. Loss: 0.020664156377315522\n",
      "Epoch: 822, Batch: 500, Avg. Loss: 0.01966546632349491\n",
      "Epoch: 822, Batch: 600, Avg. Loss: 0.026111501902341842\n",
      "Epoch: 822, Batch: 700, Avg. Loss: 0.022902398258447646\n",
      "Epoch: 822, Batch: 800, Avg. Loss: 0.02071539416909218\n",
      "Epoch: 822, Batch: 900, Avg. Loss: 0.01963226906955242\n",
      "Epoch: 823, Batch: 100, Avg. Loss: 0.026528995260596277\n",
      "Epoch: 823, Batch: 200, Avg. Loss: 0.02290912337601185\n",
      "Epoch: 823, Batch: 300, Avg. Loss: 0.020177880227565767\n",
      "Epoch: 823, Batch: 400, Avg. Loss: 0.01837104894220829\n",
      "Epoch: 823, Batch: 500, Avg. Loss: 0.023120539784431456\n",
      "Epoch: 823, Batch: 600, Avg. Loss: 0.02328177563846111\n",
      "Epoch: 823, Batch: 700, Avg. Loss: 0.019424069151282312\n",
      "Epoch: 823, Batch: 800, Avg. Loss: 0.02407396212220192\n",
      "Epoch: 823, Batch: 900, Avg. Loss: 0.020127650201320648\n",
      "Epoch: 824, Batch: 100, Avg. Loss: 0.02121764801442623\n",
      "Epoch: 824, Batch: 200, Avg. Loss: 0.023308926001191138\n",
      "Epoch: 824, Batch: 300, Avg. Loss: 0.02667913481593132\n",
      "Epoch: 824, Batch: 400, Avg. Loss: 0.018531299382448196\n",
      "Epoch: 824, Batch: 500, Avg. Loss: 0.017094103321433067\n",
      "Epoch: 824, Batch: 600, Avg. Loss: 0.023539013266563415\n",
      "Epoch: 824, Batch: 700, Avg. Loss: 0.024304066747426988\n",
      "Epoch: 824, Batch: 800, Avg. Loss: 0.02004242181777954\n",
      "Epoch: 824, Batch: 900, Avg. Loss: 0.0213955744728446\n",
      "Epoch: 825, Batch: 100, Avg. Loss: 0.02407412923872471\n",
      "Epoch: 825, Batch: 200, Avg. Loss: 0.025490103140473366\n",
      "Epoch: 825, Batch: 300, Avg. Loss: 0.02591043770313263\n",
      "Epoch: 825, Batch: 400, Avg. Loss: 0.018154536485671998\n",
      "Epoch: 825, Batch: 500, Avg. Loss: 0.018833807557821273\n",
      "Epoch: 825, Batch: 600, Avg. Loss: 0.024219977557659148\n",
      "Epoch: 825, Batch: 700, Avg. Loss: 0.021681208983063697\n",
      "Epoch: 825, Batch: 800, Avg. Loss: 0.0285227832198143\n",
      "Epoch: 825, Batch: 900, Avg. Loss: 0.024299562573432923\n",
      "Epoch: 826, Batch: 100, Avg. Loss: 0.022899810522794724\n",
      "Epoch: 826, Batch: 200, Avg. Loss: 0.023289156407117845\n",
      "Epoch: 826, Batch: 300, Avg. Loss: 0.022542422860860823\n",
      "Epoch: 826, Batch: 400, Avg. Loss: 0.023510161116719247\n",
      "Epoch: 826, Batch: 500, Avg. Loss: 0.02352029249072075\n",
      "Epoch: 826, Batch: 600, Avg. Loss: 0.01948305021971464\n",
      "Epoch: 826, Batch: 700, Avg. Loss: 0.021566895693540575\n",
      "Epoch: 826, Batch: 800, Avg. Loss: 0.025622601583600045\n",
      "Epoch: 826, Batch: 900, Avg. Loss: 0.01776551179587841\n",
      "Epoch: 827, Batch: 100, Avg. Loss: 0.023290845602750777\n",
      "Epoch: 827, Batch: 200, Avg. Loss: 0.02682345524430275\n",
      "Epoch: 827, Batch: 300, Avg. Loss: 0.019331478402018546\n",
      "Epoch: 827, Batch: 400, Avg. Loss: 0.025620741993188857\n",
      "Epoch: 827, Batch: 500, Avg. Loss: 0.02180551692843437\n",
      "Epoch: 827, Batch: 600, Avg. Loss: 0.02481631964445114\n",
      "Epoch: 827, Batch: 700, Avg. Loss: 0.022642471268773078\n",
      "Epoch: 827, Batch: 800, Avg. Loss: 0.021098578795790673\n",
      "Epoch: 827, Batch: 900, Avg. Loss: 0.024476060718297957\n",
      "Epoch: 828, Batch: 100, Avg. Loss: 0.023400749564170837\n",
      "Epoch: 828, Batch: 200, Avg. Loss: 0.016596117317676545\n",
      "Epoch: 828, Batch: 300, Avg. Loss: 0.022885312139987946\n",
      "Epoch: 828, Batch: 400, Avg. Loss: 0.023860513269901275\n",
      "Epoch: 828, Batch: 500, Avg. Loss: 0.0200362665951252\n",
      "Epoch: 828, Batch: 600, Avg. Loss: 0.021313111335039138\n",
      "Epoch: 828, Batch: 700, Avg. Loss: 0.027426377832889558\n",
      "Epoch: 828, Batch: 800, Avg. Loss: 0.01742993600666523\n",
      "Epoch: 828, Batch: 900, Avg. Loss: 0.022273362651467322\n",
      "Epoch: 829, Batch: 100, Avg. Loss: 0.02233350768685341\n",
      "Epoch: 829, Batch: 200, Avg. Loss: 0.023903441205620764\n",
      "Epoch: 829, Batch: 300, Avg. Loss: 0.020259035602211952\n",
      "Epoch: 829, Batch: 400, Avg. Loss: 0.023871259391307832\n",
      "Epoch: 829, Batch: 500, Avg. Loss: 0.024258087277412414\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 829, Batch: 600, Avg. Loss: 0.022136725932359696\n",
      "Epoch: 829, Batch: 700, Avg. Loss: 0.024249968826770784\n",
      "Epoch: 829, Batch: 800, Avg. Loss: 0.021727796122431756\n",
      "Epoch: 829, Batch: 900, Avg. Loss: 0.02313661150634289\n",
      "Epoch: 830, Batch: 100, Avg. Loss: 0.0256593419611454\n",
      "Epoch: 830, Batch: 200, Avg. Loss: 0.02692878566682339\n",
      "Epoch: 830, Batch: 300, Avg. Loss: 0.018727963119745256\n",
      "Epoch: 830, Batch: 400, Avg. Loss: 0.018922305554151534\n",
      "Epoch: 830, Batch: 500, Avg. Loss: 0.01947392262518406\n",
      "Epoch: 830, Batch: 600, Avg. Loss: 0.0176166620105505\n",
      "Epoch: 830, Batch: 700, Avg. Loss: 0.021539283618330955\n",
      "Epoch: 830, Batch: 800, Avg. Loss: 0.01944467142224312\n",
      "Epoch: 830, Batch: 900, Avg. Loss: 0.01872113101184368\n",
      "Epoch: 831, Batch: 100, Avg. Loss: 0.017980887070298195\n",
      "Epoch: 831, Batch: 200, Avg. Loss: 0.023394369930028917\n",
      "Epoch: 831, Batch: 300, Avg. Loss: 0.01978951834142208\n",
      "Epoch: 831, Batch: 400, Avg. Loss: 0.027113724797964096\n",
      "Epoch: 831, Batch: 500, Avg. Loss: 0.021251542195677757\n",
      "Epoch: 831, Batch: 600, Avg. Loss: 0.02417745292186737\n",
      "Epoch: 831, Batch: 700, Avg. Loss: 0.02519367463886738\n",
      "Epoch: 831, Batch: 800, Avg. Loss: 0.02247852735221386\n",
      "Epoch: 831, Batch: 900, Avg. Loss: 0.021347533240914346\n",
      "Epoch: 832, Batch: 100, Avg. Loss: 0.021698052063584328\n",
      "Epoch: 832, Batch: 200, Avg. Loss: 0.02064346708357334\n",
      "Epoch: 832, Batch: 300, Avg. Loss: 0.021247546896338464\n",
      "Epoch: 832, Batch: 400, Avg. Loss: 0.02628105625510216\n",
      "Epoch: 832, Batch: 500, Avg. Loss: 0.026688173338770868\n",
      "Epoch: 832, Batch: 600, Avg. Loss: 0.022836157232522965\n",
      "Epoch: 832, Batch: 700, Avg. Loss: 0.01983265034854412\n",
      "Epoch: 832, Batch: 800, Avg. Loss: 0.024293853864073754\n",
      "Epoch: 832, Batch: 900, Avg. Loss: 0.021116055697202683\n",
      "Epoch: 833, Batch: 100, Avg. Loss: 0.024279320016503333\n",
      "Epoch: 833, Batch: 200, Avg. Loss: 0.022396124824881555\n",
      "Epoch: 833, Batch: 300, Avg. Loss: 0.026083538606762886\n",
      "Epoch: 833, Batch: 400, Avg. Loss: 0.01819964252412319\n",
      "Epoch: 833, Batch: 500, Avg. Loss: 0.020131315737962723\n",
      "Epoch: 833, Batch: 600, Avg. Loss: 0.021175426319241523\n",
      "Epoch: 833, Batch: 700, Avg. Loss: 0.018357638008892535\n",
      "Epoch: 833, Batch: 800, Avg. Loss: 0.01817039094865322\n",
      "Epoch: 833, Batch: 900, Avg. Loss: 0.026720514371991157\n",
      "Epoch: 834, Batch: 100, Avg. Loss: 0.023278939351439477\n",
      "Epoch: 834, Batch: 200, Avg. Loss: 0.025544707551598548\n",
      "Epoch: 834, Batch: 300, Avg. Loss: 0.018407029807567598\n",
      "Epoch: 834, Batch: 400, Avg. Loss: 0.020005990117788315\n",
      "Epoch: 834, Batch: 500, Avg. Loss: 0.016844336315989493\n",
      "Epoch: 834, Batch: 600, Avg. Loss: 0.021755223274230958\n",
      "Epoch: 834, Batch: 700, Avg. Loss: 0.024953852817416192\n",
      "Epoch: 834, Batch: 800, Avg. Loss: 0.02176692582666874\n",
      "Epoch: 834, Batch: 900, Avg. Loss: 0.01909635938704014\n",
      "Epoch: 835, Batch: 100, Avg. Loss: 0.017635938078165055\n",
      "Epoch: 835, Batch: 200, Avg. Loss: 0.020955957248806955\n",
      "Epoch: 835, Batch: 300, Avg. Loss: 0.022553592398762704\n",
      "Epoch: 835, Batch: 400, Avg. Loss: 0.021180074140429498\n",
      "Epoch: 835, Batch: 500, Avg. Loss: 0.018357503414154052\n",
      "Epoch: 835, Batch: 600, Avg. Loss: 0.021534569039940835\n",
      "Epoch: 835, Batch: 700, Avg. Loss: 0.023838738426566124\n",
      "Epoch: 835, Batch: 800, Avg. Loss: 0.02131512522697449\n",
      "Epoch: 835, Batch: 900, Avg. Loss: 0.020388848558068277\n",
      "Epoch: 836, Batch: 100, Avg. Loss: 0.02285924904048443\n",
      "Epoch: 836, Batch: 200, Avg. Loss: 0.022963239625096322\n",
      "Epoch: 836, Batch: 300, Avg. Loss: 0.023579068854451178\n",
      "Epoch: 836, Batch: 400, Avg. Loss: 0.021457718163728715\n",
      "Epoch: 836, Batch: 500, Avg. Loss: 0.017250980585813522\n",
      "Epoch: 836, Batch: 600, Avg. Loss: 0.02376689039170742\n",
      "Epoch: 836, Batch: 700, Avg. Loss: 0.023591099083423616\n",
      "Epoch: 836, Batch: 800, Avg. Loss: 0.024803256243467332\n",
      "Epoch: 836, Batch: 900, Avg. Loss: 0.022586362436413765\n",
      "Epoch: 837, Batch: 100, Avg. Loss: 0.02339239127933979\n",
      "Epoch: 837, Batch: 200, Avg. Loss: 0.02178397536277771\n",
      "Epoch: 837, Batch: 300, Avg. Loss: 0.019770029932260513\n",
      "Epoch: 837, Batch: 400, Avg. Loss: 0.01818540096282959\n",
      "Epoch: 837, Batch: 500, Avg. Loss: 0.019116306826472283\n",
      "Epoch: 837, Batch: 600, Avg. Loss: 0.018314220383763312\n",
      "Epoch: 837, Batch: 700, Avg. Loss: 0.019282082468271254\n",
      "Epoch: 837, Batch: 800, Avg. Loss: 0.02573861099779606\n",
      "Epoch: 837, Batch: 900, Avg. Loss: 0.018503371328115463\n",
      "Epoch: 838, Batch: 100, Avg. Loss: 0.025096892416477203\n",
      "Epoch: 838, Batch: 200, Avg. Loss: 0.022585905864834787\n",
      "Epoch: 838, Batch: 300, Avg. Loss: 0.02122581198811531\n",
      "Epoch: 838, Batch: 400, Avg. Loss: 0.020851225480437278\n",
      "Epoch: 838, Batch: 500, Avg. Loss: 0.01868330202996731\n",
      "Epoch: 838, Batch: 600, Avg. Loss: 0.019021513164043425\n",
      "Epoch: 838, Batch: 700, Avg. Loss: 0.02051320470869541\n",
      "Epoch: 838, Batch: 800, Avg. Loss: 0.024619112610816955\n",
      "Epoch: 838, Batch: 900, Avg. Loss: 0.019343605861067772\n",
      "Epoch: 839, Batch: 100, Avg. Loss: 0.02340503416955471\n",
      "Epoch: 839, Batch: 200, Avg. Loss: 0.020791682302951812\n",
      "Epoch: 839, Batch: 300, Avg. Loss: 0.02462635964155197\n",
      "Epoch: 839, Batch: 400, Avg. Loss: 0.020718745514750482\n",
      "Epoch: 839, Batch: 500, Avg. Loss: 0.019163613691926002\n",
      "Epoch: 839, Batch: 600, Avg. Loss: 0.017761876061558723\n",
      "Epoch: 839, Batch: 700, Avg. Loss: 0.022643743753433226\n",
      "Epoch: 839, Batch: 800, Avg. Loss: 0.022354023680090904\n",
      "Epoch: 839, Batch: 900, Avg. Loss: 0.019117910116910934\n",
      "Epoch: 840, Batch: 100, Avg. Loss: 0.027932153642177583\n",
      "Epoch: 840, Batch: 200, Avg. Loss: 0.014379335790872574\n",
      "Epoch: 840, Batch: 300, Avg. Loss: 0.032021067216992376\n",
      "Epoch: 840, Batch: 400, Avg. Loss: 0.022580356895923616\n",
      "Epoch: 840, Batch: 500, Avg. Loss: 0.018160734325647354\n",
      "Epoch: 840, Batch: 600, Avg. Loss: 0.021791421845555305\n",
      "Epoch: 840, Batch: 700, Avg. Loss: 0.024285591021180154\n",
      "Epoch: 840, Batch: 800, Avg. Loss: 0.01834057778120041\n",
      "Epoch: 840, Batch: 900, Avg. Loss: 0.02347194656729698\n",
      "Epoch: 841, Batch: 100, Avg. Loss: 0.022144163250923155\n",
      "Epoch: 841, Batch: 200, Avg. Loss: 0.02321007952094078\n",
      "Epoch: 841, Batch: 300, Avg. Loss: 0.025412486419081688\n",
      "Epoch: 841, Batch: 400, Avg. Loss: 0.017930159345269203\n",
      "Epoch: 841, Batch: 500, Avg. Loss: 0.020086317136883736\n",
      "Epoch: 841, Batch: 600, Avg. Loss: 0.026652981862425804\n",
      "Epoch: 841, Batch: 700, Avg. Loss: 0.024152878373861313\n",
      "Epoch: 841, Batch: 800, Avg. Loss: 0.020450348630547524\n",
      "Epoch: 841, Batch: 900, Avg. Loss: 0.019411107078194618\n",
      "Epoch: 842, Batch: 100, Avg. Loss: 0.021769371703267097\n",
      "Epoch: 842, Batch: 200, Avg. Loss: 0.023625846952199936\n",
      "Epoch: 842, Batch: 300, Avg. Loss: 0.022679741457104684\n",
      "Epoch: 842, Batch: 400, Avg. Loss: 0.022552696466445924\n",
      "Epoch: 842, Batch: 500, Avg. Loss: 0.019724288284778596\n",
      "Epoch: 842, Batch: 600, Avg. Loss: 0.017814600169658662\n",
      "Epoch: 842, Batch: 700, Avg. Loss: 0.019688713029026986\n",
      "Epoch: 842, Batch: 800, Avg. Loss: 0.020311824306845666\n",
      "Epoch: 842, Batch: 900, Avg. Loss: 0.018329908326268195\n",
      "Epoch: 843, Batch: 100, Avg. Loss: 0.021987638771533966\n",
      "Epoch: 843, Batch: 200, Avg. Loss: 0.0213910124450922\n",
      "Epoch: 843, Batch: 300, Avg. Loss: 0.027855434641242027\n",
      "Epoch: 843, Batch: 400, Avg. Loss: 0.017968213930726052\n",
      "Epoch: 843, Batch: 500, Avg. Loss: 0.02098345473408699\n",
      "Epoch: 843, Batch: 600, Avg. Loss: 0.022917468696832657\n",
      "Epoch: 843, Batch: 700, Avg. Loss: 0.02012673743069172\n",
      "Epoch: 843, Batch: 800, Avg. Loss: 0.022872217744588853\n",
      "Epoch: 843, Batch: 900, Avg. Loss: 0.02239926166832447\n",
      "Epoch: 844, Batch: 100, Avg. Loss: 0.028402615636587143\n",
      "Epoch: 844, Batch: 200, Avg. Loss: 0.024050684273242952\n",
      "Epoch: 844, Batch: 300, Avg. Loss: 0.02256827712059021\n",
      "Epoch: 844, Batch: 400, Avg. Loss: 0.018901077583432197\n",
      "Epoch: 844, Batch: 500, Avg. Loss: 0.020857061445713043\n",
      "Epoch: 844, Batch: 600, Avg. Loss: 0.02279622435569763\n",
      "Epoch: 844, Batch: 700, Avg. Loss: 0.01488777033984661\n",
      "Epoch: 844, Batch: 800, Avg. Loss: 0.020327402800321578\n",
      "Epoch: 844, Batch: 900, Avg. Loss: 0.01930727154016495\n",
      "Epoch: 845, Batch: 100, Avg. Loss: 0.021398593187332154\n",
      "Epoch: 845, Batch: 200, Avg. Loss: 0.029884165823459624\n",
      "Epoch: 845, Batch: 300, Avg. Loss: 0.021359952613711358\n",
      "Epoch: 845, Batch: 400, Avg. Loss: 0.020339411422610282\n",
      "Epoch: 845, Batch: 500, Avg. Loss: 0.0213240759819746\n",
      "Epoch: 845, Batch: 600, Avg. Loss: 0.022006852626800535\n",
      "Epoch: 845, Batch: 700, Avg. Loss: 0.0228809268027544\n",
      "Epoch: 845, Batch: 800, Avg. Loss: 0.01723046600818634\n",
      "Epoch: 845, Batch: 900, Avg. Loss: 0.02342923529446125\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 846, Batch: 100, Avg. Loss: 0.024281932711601256\n",
      "Epoch: 846, Batch: 200, Avg. Loss: 0.023224409893155096\n",
      "Epoch: 846, Batch: 300, Avg. Loss: 0.019117904976010323\n",
      "Epoch: 846, Batch: 400, Avg. Loss: 0.020141567289829253\n",
      "Epoch: 846, Batch: 500, Avg. Loss: 0.025785735100507735\n",
      "Epoch: 846, Batch: 600, Avg. Loss: 0.017421211525797845\n",
      "Epoch: 846, Batch: 700, Avg. Loss: 0.022670923843979836\n",
      "Epoch: 846, Batch: 800, Avg. Loss: 0.025211572647094727\n",
      "Epoch: 846, Batch: 900, Avg. Loss: 0.021638299375772475\n",
      "Epoch: 847, Batch: 100, Avg. Loss: 0.021679919734597205\n",
      "Epoch: 847, Batch: 200, Avg. Loss: 0.024855478033423424\n",
      "Epoch: 847, Batch: 300, Avg. Loss: 0.017375430017709734\n",
      "Epoch: 847, Batch: 400, Avg. Loss: 0.018725920543074606\n",
      "Epoch: 847, Batch: 500, Avg. Loss: 0.021033866181969643\n",
      "Epoch: 847, Batch: 600, Avg. Loss: 0.022508109733462334\n",
      "Epoch: 847, Batch: 700, Avg. Loss: 0.02047321081161499\n",
      "Epoch: 847, Batch: 800, Avg. Loss: 0.019096857905387878\n",
      "Epoch: 847, Batch: 900, Avg. Loss: 0.02045104920864105\n",
      "Epoch: 848, Batch: 100, Avg. Loss: 0.01879676468670368\n",
      "Epoch: 848, Batch: 200, Avg. Loss: 0.02506989262998104\n",
      "Epoch: 848, Batch: 300, Avg. Loss: 0.020251017734408377\n",
      "Epoch: 848, Batch: 400, Avg. Loss: 0.018470900282263756\n",
      "Epoch: 848, Batch: 500, Avg. Loss: 0.016335793063044547\n",
      "Epoch: 848, Batch: 600, Avg. Loss: 0.02378443121910095\n",
      "Epoch: 848, Batch: 700, Avg. Loss: 0.016757234781980514\n",
      "Epoch: 848, Batch: 800, Avg. Loss: 0.01621003158390522\n",
      "Epoch: 848, Batch: 900, Avg. Loss: 0.025046311095356943\n",
      "Epoch: 849, Batch: 100, Avg. Loss: 0.02141757756471634\n",
      "Epoch: 849, Batch: 200, Avg. Loss: 0.019024289324879647\n",
      "Epoch: 849, Batch: 300, Avg. Loss: 0.018291023671627046\n",
      "Epoch: 849, Batch: 400, Avg. Loss: 0.018952391892671585\n",
      "Epoch: 849, Batch: 500, Avg. Loss: 0.028174776136875152\n",
      "Epoch: 849, Batch: 600, Avg. Loss: 0.016408453956246377\n",
      "Epoch: 849, Batch: 700, Avg. Loss: 0.02519754573702812\n",
      "Epoch: 849, Batch: 800, Avg. Loss: 0.021754674538969995\n",
      "Epoch: 849, Batch: 900, Avg. Loss: 0.024501276314258576\n",
      "Epoch: 850, Batch: 100, Avg. Loss: 0.02191998399794102\n",
      "Epoch: 850, Batch: 200, Avg. Loss: 0.02241714432835579\n",
      "Epoch: 850, Batch: 300, Avg. Loss: 0.020564825162291526\n",
      "Epoch: 850, Batch: 400, Avg. Loss: 0.01886623293161392\n",
      "Epoch: 850, Batch: 500, Avg. Loss: 0.02343900054693222\n",
      "Epoch: 850, Batch: 600, Avg. Loss: 0.019289376735687255\n",
      "Epoch: 850, Batch: 700, Avg. Loss: 0.02494921013712883\n",
      "Epoch: 850, Batch: 800, Avg. Loss: 0.024665133580565452\n",
      "Epoch: 850, Batch: 900, Avg. Loss: 0.01946429654955864\n",
      "Epoch: 851, Batch: 100, Avg. Loss: 0.024309070110321046\n",
      "Epoch: 851, Batch: 200, Avg. Loss: 0.022098318114876746\n",
      "Epoch: 851, Batch: 300, Avg. Loss: 0.021230002269148826\n",
      "Epoch: 851, Batch: 400, Avg. Loss: 0.018376992419362068\n",
      "Epoch: 851, Batch: 500, Avg. Loss: 0.02358274720609188\n",
      "Epoch: 851, Batch: 600, Avg. Loss: 0.020529242232441902\n",
      "Epoch: 851, Batch: 700, Avg. Loss: 0.021973114684224127\n",
      "Epoch: 851, Batch: 800, Avg. Loss: 0.02464323863387108\n",
      "Epoch: 851, Batch: 900, Avg. Loss: 0.012606563642621041\n",
      "Epoch: 852, Batch: 100, Avg. Loss: 0.024734493866562842\n",
      "Epoch: 852, Batch: 200, Avg. Loss: 0.0185703594237566\n",
      "Epoch: 852, Batch: 300, Avg. Loss: 0.020538440495729445\n",
      "Epoch: 852, Batch: 400, Avg. Loss: 0.018255430087447167\n",
      "Epoch: 852, Batch: 500, Avg. Loss: 0.021137989833950998\n",
      "Epoch: 852, Batch: 600, Avg. Loss: 0.020538491010665894\n",
      "Epoch: 852, Batch: 700, Avg. Loss: 0.01782710202038288\n",
      "Epoch: 852, Batch: 800, Avg. Loss: 0.021015429198741914\n",
      "Epoch: 852, Batch: 900, Avg. Loss: 0.023683942779898645\n",
      "Epoch: 853, Batch: 100, Avg. Loss: 0.023222654461860656\n",
      "Epoch: 853, Batch: 200, Avg. Loss: 0.017548691183328628\n",
      "Epoch: 853, Batch: 300, Avg. Loss: 0.02171217814087868\n",
      "Epoch: 853, Batch: 400, Avg. Loss: 0.017980554774403572\n",
      "Epoch: 853, Batch: 500, Avg. Loss: 0.023800847977399828\n",
      "Epoch: 853, Batch: 600, Avg. Loss: 0.019514801204204558\n",
      "Epoch: 853, Batch: 700, Avg. Loss: 0.02002183601260185\n",
      "Epoch: 853, Batch: 800, Avg. Loss: 0.023948272690176964\n",
      "Epoch: 853, Batch: 900, Avg. Loss: 0.028585453033447266\n",
      "Epoch: 854, Batch: 100, Avg. Loss: 0.028324024453759193\n",
      "Epoch: 854, Batch: 200, Avg. Loss: 0.022317461147904397\n",
      "Epoch: 854, Batch: 300, Avg. Loss: 0.024488605558872223\n",
      "Epoch: 854, Batch: 400, Avg. Loss: 0.019662501960992812\n",
      "Epoch: 854, Batch: 500, Avg. Loss: 0.01887227803468704\n",
      "Epoch: 854, Batch: 600, Avg. Loss: 0.020203316509723665\n",
      "Epoch: 854, Batch: 700, Avg. Loss: 0.021197776198387145\n",
      "Epoch: 854, Batch: 800, Avg. Loss: 0.01776359349489212\n",
      "Epoch: 854, Batch: 900, Avg. Loss: 0.018041259422898293\n",
      "Epoch: 855, Batch: 100, Avg. Loss: 0.019634518846869467\n",
      "Epoch: 855, Batch: 200, Avg. Loss: 0.02184317335486412\n",
      "Epoch: 855, Batch: 300, Avg. Loss: 0.01996166706085205\n",
      "Epoch: 855, Batch: 400, Avg. Loss: 0.019968704581260682\n",
      "Epoch: 855, Batch: 500, Avg. Loss: 0.021013147979974747\n",
      "Epoch: 855, Batch: 600, Avg. Loss: 0.016089094281196595\n",
      "Epoch: 855, Batch: 700, Avg. Loss: 0.02263761229813099\n",
      "Epoch: 855, Batch: 800, Avg. Loss: 0.020881955251097678\n",
      "Epoch: 855, Batch: 900, Avg. Loss: 0.021578212603926658\n",
      "Epoch: 856, Batch: 100, Avg. Loss: 0.01928304247558117\n",
      "Epoch: 856, Batch: 200, Avg. Loss: 0.018055202066898347\n",
      "Epoch: 856, Batch: 300, Avg. Loss: 0.018935763761401176\n",
      "Epoch: 856, Batch: 400, Avg. Loss: 0.016991485878825187\n",
      "Epoch: 856, Batch: 500, Avg. Loss: 0.022765856757760047\n",
      "Epoch: 856, Batch: 600, Avg. Loss: 0.017098639011383057\n",
      "Epoch: 856, Batch: 700, Avg. Loss: 0.022631355226039887\n",
      "Epoch: 856, Batch: 800, Avg. Loss: 0.021045040115714075\n",
      "Epoch: 856, Batch: 900, Avg. Loss: 0.015408075749874114\n",
      "Epoch: 857, Batch: 100, Avg. Loss: 0.02018179923295975\n",
      "Epoch: 857, Batch: 200, Avg. Loss: 0.020425323694944382\n",
      "Epoch: 857, Batch: 300, Avg. Loss: 0.025261994004249572\n",
      "Epoch: 857, Batch: 400, Avg. Loss: 0.020238413959741592\n",
      "Epoch: 857, Batch: 500, Avg. Loss: 0.023045157939195635\n",
      "Epoch: 857, Batch: 600, Avg. Loss: 0.026932846009731292\n",
      "Epoch: 857, Batch: 700, Avg. Loss: 0.023178224191069602\n",
      "Epoch: 857, Batch: 800, Avg. Loss: 0.016531386226415635\n",
      "Epoch: 857, Batch: 900, Avg. Loss: 0.02148607812821865\n",
      "Epoch: 858, Batch: 100, Avg. Loss: 0.024655854627490044\n",
      "Epoch: 858, Batch: 200, Avg. Loss: 0.021913781687617304\n",
      "Epoch: 858, Batch: 300, Avg. Loss: 0.022788629531860352\n",
      "Epoch: 858, Batch: 400, Avg. Loss: 0.019687065035104753\n",
      "Epoch: 858, Batch: 500, Avg. Loss: 0.026981939151883124\n",
      "Epoch: 858, Batch: 600, Avg. Loss: 0.019358533918857574\n",
      "Epoch: 858, Batch: 700, Avg. Loss: 0.016710326746106147\n",
      "Epoch: 858, Batch: 800, Avg. Loss: 0.021828135401010515\n",
      "Epoch: 858, Batch: 900, Avg. Loss: 0.022247090414166452\n",
      "Epoch: 859, Batch: 100, Avg. Loss: 0.018325138315558435\n",
      "Epoch: 859, Batch: 200, Avg. Loss: 0.019310110211372376\n",
      "Epoch: 859, Batch: 300, Avg. Loss: 0.019666539654135704\n",
      "Epoch: 859, Batch: 400, Avg. Loss: 0.01785833440721035\n",
      "Epoch: 859, Batch: 500, Avg. Loss: 0.017299685180187225\n",
      "Epoch: 859, Batch: 600, Avg. Loss: 0.021064943373203276\n",
      "Epoch: 859, Batch: 700, Avg. Loss: 0.022201477512717248\n",
      "Epoch: 859, Batch: 800, Avg. Loss: 0.02061838872730732\n",
      "Epoch: 859, Batch: 900, Avg. Loss: 0.02276675708591938\n",
      "Epoch: 860, Batch: 100, Avg. Loss: 0.01957975335419178\n",
      "Epoch: 860, Batch: 200, Avg. Loss: 0.019771572425961493\n",
      "Epoch: 860, Batch: 300, Avg. Loss: 0.02254759833216667\n",
      "Epoch: 860, Batch: 400, Avg. Loss: 0.020272781029343604\n",
      "Epoch: 860, Batch: 500, Avg. Loss: 0.020918573513627053\n",
      "Epoch: 860, Batch: 600, Avg. Loss: 0.015350095704197883\n",
      "Epoch: 860, Batch: 700, Avg. Loss: 0.02744243070483208\n",
      "Epoch: 860, Batch: 800, Avg. Loss: 0.021613539308309556\n",
      "Epoch: 860, Batch: 900, Avg. Loss: 0.02280375309288502\n",
      "Epoch: 861, Batch: 100, Avg. Loss: 0.01801899164915085\n",
      "Epoch: 861, Batch: 200, Avg. Loss: 0.016248324513435365\n",
      "Epoch: 861, Batch: 300, Avg. Loss: 0.02626584567129612\n",
      "Epoch: 861, Batch: 400, Avg. Loss: 0.020610168427228927\n",
      "Epoch: 861, Batch: 500, Avg. Loss: 0.020220101848244666\n",
      "Epoch: 861, Batch: 600, Avg. Loss: 0.018747596591711043\n",
      "Epoch: 861, Batch: 700, Avg. Loss: 0.023998283743858338\n",
      "Epoch: 861, Batch: 800, Avg. Loss: 0.024284568578004838\n",
      "Epoch: 861, Batch: 900, Avg. Loss: 0.019235063418745993\n",
      "Epoch: 862, Batch: 100, Avg. Loss: 0.02423830598592758\n",
      "Epoch: 862, Batch: 200, Avg. Loss: 0.01545302852988243\n",
      "Epoch: 862, Batch: 300, Avg. Loss: 0.020697725713253023\n",
      "Epoch: 862, Batch: 400, Avg. Loss: 0.018313295021653176\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 862, Batch: 500, Avg. Loss: 0.026623247638344765\n",
      "Epoch: 862, Batch: 600, Avg. Loss: 0.022822428196668625\n",
      "Epoch: 862, Batch: 700, Avg. Loss: 0.022646662890911103\n",
      "Epoch: 862, Batch: 800, Avg. Loss: 0.020504784807562827\n",
      "Epoch: 862, Batch: 900, Avg. Loss: 0.021492275521159173\n",
      "Epoch: 863, Batch: 100, Avg. Loss: 0.017111981250345708\n",
      "Epoch: 863, Batch: 200, Avg. Loss: 0.02215523585677147\n",
      "Epoch: 863, Batch: 300, Avg. Loss: 0.018110616058111192\n",
      "Epoch: 863, Batch: 400, Avg. Loss: 0.02085579566657543\n",
      "Epoch: 863, Batch: 500, Avg. Loss: 0.02849612317979336\n",
      "Epoch: 863, Batch: 600, Avg. Loss: 0.018845807239413263\n",
      "Epoch: 863, Batch: 700, Avg. Loss: 0.02345175087451935\n",
      "Epoch: 863, Batch: 800, Avg. Loss: 0.023274259492754936\n",
      "Epoch: 863, Batch: 900, Avg. Loss: 0.020270300805568697\n",
      "Epoch: 864, Batch: 100, Avg. Loss: 0.020259130075573922\n",
      "Epoch: 864, Batch: 200, Avg. Loss: 0.020250038653612138\n",
      "Epoch: 864, Batch: 300, Avg. Loss: 0.020343219265341758\n",
      "Epoch: 864, Batch: 400, Avg. Loss: 0.01896032616496086\n",
      "Epoch: 864, Batch: 500, Avg. Loss: 0.01834522970020771\n",
      "Epoch: 864, Batch: 600, Avg. Loss: 0.022551321014761923\n",
      "Epoch: 864, Batch: 700, Avg. Loss: 0.018104394301772117\n",
      "Epoch: 864, Batch: 800, Avg. Loss: 0.022748077809810637\n",
      "Epoch: 864, Batch: 900, Avg. Loss: 0.02187854066491127\n",
      "Epoch: 865, Batch: 100, Avg. Loss: 0.02289310060441494\n",
      "Epoch: 865, Batch: 200, Avg. Loss: 0.024295162484049795\n",
      "Epoch: 865, Batch: 300, Avg. Loss: 0.016665262058377265\n",
      "Epoch: 865, Batch: 400, Avg. Loss: 0.0223657289147377\n",
      "Epoch: 865, Batch: 500, Avg. Loss: 0.019746291264891625\n",
      "Epoch: 865, Batch: 600, Avg. Loss: 0.018609593361616134\n",
      "Epoch: 865, Batch: 700, Avg. Loss: 0.017961397394537927\n",
      "Epoch: 865, Batch: 800, Avg. Loss: 0.0245009908080101\n",
      "Epoch: 865, Batch: 900, Avg. Loss: 0.020683440417051314\n",
      "Epoch: 866, Batch: 100, Avg. Loss: 0.019560289606451987\n",
      "Epoch: 866, Batch: 200, Avg. Loss: 0.02261965535581112\n",
      "Epoch: 866, Batch: 300, Avg. Loss: 0.0207144208997488\n",
      "Epoch: 866, Batch: 400, Avg. Loss: 0.020439105331897734\n",
      "Epoch: 866, Batch: 500, Avg. Loss: 0.023388373032212258\n",
      "Epoch: 866, Batch: 600, Avg. Loss: 0.016896080449223517\n",
      "Epoch: 866, Batch: 700, Avg. Loss: 0.019458185359835626\n",
      "Epoch: 866, Batch: 800, Avg. Loss: 0.01633686579763889\n",
      "Epoch: 866, Batch: 900, Avg. Loss: 0.017725987434387205\n",
      "Epoch: 867, Batch: 100, Avg. Loss: 0.017718407288193704\n",
      "Epoch: 867, Batch: 200, Avg. Loss: 0.016272651925683022\n",
      "Epoch: 867, Batch: 300, Avg. Loss: 0.01814931906759739\n",
      "Epoch: 867, Batch: 400, Avg. Loss: 0.02018924780189991\n",
      "Epoch: 867, Batch: 500, Avg. Loss: 0.023243574276566505\n",
      "Epoch: 867, Batch: 600, Avg. Loss: 0.02737362265586853\n",
      "Epoch: 867, Batch: 700, Avg. Loss: 0.016443881392478942\n",
      "Epoch: 867, Batch: 800, Avg. Loss: 0.0201417725533247\n",
      "Epoch: 867, Batch: 900, Avg. Loss: 0.02058531582355499\n",
      "Epoch: 868, Batch: 100, Avg. Loss: 0.02297047831118107\n",
      "Epoch: 868, Batch: 200, Avg. Loss: 0.015928487926721573\n",
      "Epoch: 868, Batch: 300, Avg. Loss: 0.022102441787719727\n",
      "Epoch: 868, Batch: 400, Avg. Loss: 0.0240798632055521\n",
      "Epoch: 868, Batch: 500, Avg. Loss: 0.019844983220100403\n",
      "Epoch: 868, Batch: 600, Avg. Loss: 0.01657067745923996\n",
      "Epoch: 868, Batch: 700, Avg. Loss: 0.02443226456642151\n",
      "Epoch: 868, Batch: 800, Avg. Loss: 0.018344548344612122\n",
      "Epoch: 868, Batch: 900, Avg. Loss: 0.024463612213730813\n",
      "Epoch: 869, Batch: 100, Avg. Loss: 0.0165884018689394\n",
      "Epoch: 869, Batch: 200, Avg. Loss: 0.020413623973727226\n",
      "Epoch: 869, Batch: 300, Avg. Loss: 0.01876459464430809\n",
      "Epoch: 869, Batch: 400, Avg. Loss: 0.02175439327955246\n",
      "Epoch: 869, Batch: 500, Avg. Loss: 0.018970772922039032\n",
      "Epoch: 869, Batch: 600, Avg. Loss: 0.01739027537405491\n",
      "Epoch: 869, Batch: 700, Avg. Loss: 0.022382125556468964\n",
      "Epoch: 869, Batch: 800, Avg. Loss: 0.024857019037008286\n",
      "Epoch: 869, Batch: 900, Avg. Loss: 0.02952795773744583\n",
      "Epoch: 870, Batch: 100, Avg. Loss: 0.022036311477422715\n",
      "Epoch: 870, Batch: 200, Avg. Loss: 0.025121533647179603\n",
      "Epoch: 870, Batch: 300, Avg. Loss: 0.02271736189723015\n",
      "Epoch: 870, Batch: 400, Avg. Loss: 0.017646652311086655\n",
      "Epoch: 870, Batch: 500, Avg. Loss: 0.016519511118531227\n",
      "Epoch: 870, Batch: 600, Avg. Loss: 0.019999839216470718\n",
      "Epoch: 870, Batch: 700, Avg. Loss: 0.021489982157945633\n",
      "Epoch: 870, Batch: 800, Avg. Loss: 0.01966987393796444\n",
      "Epoch: 870, Batch: 900, Avg. Loss: 0.019893680736422538\n",
      "Epoch: 871, Batch: 100, Avg. Loss: 0.02904810830950737\n",
      "Epoch: 871, Batch: 200, Avg. Loss: 0.020290352776646613\n",
      "Epoch: 871, Batch: 300, Avg. Loss: 0.01675807848572731\n",
      "Epoch: 871, Batch: 400, Avg. Loss: 0.01960001215338707\n",
      "Epoch: 871, Batch: 500, Avg. Loss: 0.023670256435871125\n",
      "Epoch: 871, Batch: 600, Avg. Loss: 0.017597615346312524\n",
      "Epoch: 871, Batch: 700, Avg. Loss: 0.0216509860008955\n",
      "Epoch: 871, Batch: 800, Avg. Loss: 0.021568943932652473\n",
      "Epoch: 871, Batch: 900, Avg. Loss: 0.018081168904900552\n",
      "Epoch: 872, Batch: 100, Avg. Loss: 0.024107467606663702\n",
      "Epoch: 872, Batch: 200, Avg. Loss: 0.021105538830161096\n",
      "Epoch: 872, Batch: 300, Avg. Loss: 0.020541856810450555\n",
      "Epoch: 872, Batch: 400, Avg. Loss: 0.021163398176431657\n",
      "Epoch: 872, Batch: 500, Avg. Loss: 0.022658295035362243\n",
      "Epoch: 872, Batch: 600, Avg. Loss: 0.01564733013510704\n",
      "Epoch: 872, Batch: 700, Avg. Loss: 0.021829915419220924\n",
      "Epoch: 872, Batch: 800, Avg. Loss: 0.02252971559762955\n",
      "Epoch: 872, Batch: 900, Avg. Loss: 0.02000072591006756\n",
      "Epoch: 873, Batch: 100, Avg. Loss: 0.024470310583710672\n",
      "Epoch: 873, Batch: 200, Avg. Loss: 0.025163505896925925\n",
      "Epoch: 873, Batch: 300, Avg. Loss: 0.02337021619081497\n",
      "Epoch: 873, Batch: 400, Avg. Loss: 0.018719017282128336\n",
      "Epoch: 873, Batch: 500, Avg. Loss: 0.018002798855304716\n",
      "Epoch: 873, Batch: 600, Avg. Loss: 0.021693224906921386\n",
      "Epoch: 873, Batch: 700, Avg. Loss: 0.01831069588661194\n",
      "Epoch: 873, Batch: 800, Avg. Loss: 0.016835022643208505\n",
      "Epoch: 873, Batch: 900, Avg. Loss: 0.023534757569432258\n",
      "Epoch: 874, Batch: 100, Avg. Loss: 0.023574457690119743\n",
      "Epoch: 874, Batch: 200, Avg. Loss: 0.023588353917002678\n",
      "Epoch: 874, Batch: 300, Avg. Loss: 0.02000344790518284\n",
      "Epoch: 874, Batch: 400, Avg. Loss: 0.020516373217105865\n",
      "Epoch: 874, Batch: 500, Avg. Loss: 0.013797001913189888\n",
      "Epoch: 874, Batch: 600, Avg. Loss: 0.02019913390278816\n",
      "Epoch: 874, Batch: 700, Avg. Loss: 0.01720045126974583\n",
      "Epoch: 874, Batch: 800, Avg. Loss: 0.02199845902621746\n",
      "Epoch: 874, Batch: 900, Avg. Loss: 0.01918515682220459\n",
      "Epoch: 875, Batch: 100, Avg. Loss: 0.020364776775240897\n",
      "Epoch: 875, Batch: 200, Avg. Loss: 0.018299707993865013\n",
      "Epoch: 875, Batch: 300, Avg. Loss: 0.02347167082130909\n",
      "Epoch: 875, Batch: 400, Avg. Loss: 0.019705553352832795\n",
      "Epoch: 875, Batch: 500, Avg. Loss: 0.02403823897242546\n",
      "Epoch: 875, Batch: 600, Avg. Loss: 0.018842140212655067\n",
      "Epoch: 875, Batch: 700, Avg. Loss: 0.01760754480957985\n",
      "Epoch: 875, Batch: 800, Avg. Loss: 0.018527094051241876\n",
      "Epoch: 875, Batch: 900, Avg. Loss: 0.02768009066581726\n",
      "Epoch: 876, Batch: 100, Avg. Loss: 0.021235257536172867\n",
      "Epoch: 876, Batch: 200, Avg. Loss: 0.022266349717974664\n",
      "Epoch: 876, Batch: 300, Avg. Loss: 0.024845593646168708\n",
      "Epoch: 876, Batch: 400, Avg. Loss: 0.0178896189853549\n",
      "Epoch: 876, Batch: 500, Avg. Loss: 0.018847273588180544\n",
      "Epoch: 876, Batch: 600, Avg. Loss: 0.017544721066951752\n",
      "Epoch: 876, Batch: 700, Avg. Loss: 0.02255999147891998\n",
      "Epoch: 876, Batch: 800, Avg. Loss: 0.02025124818086624\n",
      "Epoch: 876, Batch: 900, Avg. Loss: 0.018709645420312882\n",
      "Epoch: 877, Batch: 100, Avg. Loss: 0.022588563412427903\n",
      "Epoch: 877, Batch: 200, Avg. Loss: 0.019991664364933968\n",
      "Epoch: 877, Batch: 300, Avg. Loss: 0.01905648350715637\n",
      "Epoch: 877, Batch: 400, Avg. Loss: 0.021359723806381226\n",
      "Epoch: 877, Batch: 500, Avg. Loss: 0.022648384273052217\n",
      "Epoch: 877, Batch: 600, Avg. Loss: 0.01983981691300869\n",
      "Epoch: 877, Batch: 700, Avg. Loss: 0.018991019651293753\n",
      "Epoch: 877, Batch: 800, Avg. Loss: 0.018839052692055703\n",
      "Epoch: 877, Batch: 900, Avg. Loss: 0.021835412457585336\n",
      "Epoch: 878, Batch: 100, Avg. Loss: 0.019092750996351243\n",
      "Epoch: 878, Batch: 200, Avg. Loss: 0.016974329352378845\n",
      "Epoch: 878, Batch: 300, Avg. Loss: 0.02067861720919609\n",
      "Epoch: 878, Batch: 400, Avg. Loss: 0.019171153530478477\n",
      "Epoch: 878, Batch: 500, Avg. Loss: 0.02308570943772793\n",
      "Epoch: 878, Batch: 600, Avg. Loss: 0.016360688358545303\n",
      "Epoch: 878, Batch: 700, Avg. Loss: 0.01784328818321228\n",
      "Epoch: 878, Batch: 800, Avg. Loss: 0.026963311061263085\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 878, Batch: 900, Avg. Loss: 0.022837283760309218\n",
      "Epoch: 879, Batch: 100, Avg. Loss: 0.02047878459095955\n",
      "Epoch: 879, Batch: 200, Avg. Loss: 0.01616083823144436\n",
      "Epoch: 879, Batch: 300, Avg. Loss: 0.02186960734426975\n",
      "Epoch: 879, Batch: 400, Avg. Loss: 0.01955061823129654\n",
      "Epoch: 879, Batch: 500, Avg. Loss: 0.018602180257439614\n",
      "Epoch: 879, Batch: 600, Avg. Loss: 0.019123780131340026\n",
      "Epoch: 879, Batch: 700, Avg. Loss: 0.02517201229929924\n",
      "Epoch: 879, Batch: 800, Avg. Loss: 0.0202686108648777\n",
      "Epoch: 879, Batch: 900, Avg. Loss: 0.02834606237709522\n",
      "Epoch: 880, Batch: 100, Avg. Loss: 0.020979787185788155\n",
      "Epoch: 880, Batch: 200, Avg. Loss: 0.020725672766566276\n",
      "Epoch: 880, Batch: 300, Avg. Loss: 0.02156942889094353\n",
      "Epoch: 880, Batch: 400, Avg. Loss: 0.02100582145154476\n",
      "Epoch: 880, Batch: 500, Avg. Loss: 0.01639730788767338\n",
      "Epoch: 880, Batch: 600, Avg. Loss: 0.018001630678772928\n",
      "Epoch: 880, Batch: 700, Avg. Loss: 0.019026892706751824\n",
      "Epoch: 880, Batch: 800, Avg. Loss: 0.015444641262292861\n",
      "Epoch: 880, Batch: 900, Avg. Loss: 0.024959219992160796\n",
      "Epoch: 881, Batch: 100, Avg. Loss: 0.019518050402402877\n",
      "Epoch: 881, Batch: 200, Avg. Loss: 0.01870272435247898\n",
      "Epoch: 881, Batch: 300, Avg. Loss: 0.021723011806607247\n",
      "Epoch: 881, Batch: 400, Avg. Loss: 0.02080615796148777\n",
      "Epoch: 881, Batch: 500, Avg. Loss: 0.021289161816239358\n",
      "Epoch: 881, Batch: 600, Avg. Loss: 0.02344698317348957\n",
      "Epoch: 881, Batch: 700, Avg. Loss: 0.023864341229200364\n",
      "Epoch: 881, Batch: 800, Avg. Loss: 0.020146416276693346\n",
      "Epoch: 881, Batch: 900, Avg. Loss: 0.02261346869170666\n",
      "Epoch: 882, Batch: 100, Avg. Loss: 0.020415611639618872\n",
      "Epoch: 882, Batch: 200, Avg. Loss: 0.02094924099743366\n",
      "Epoch: 882, Batch: 300, Avg. Loss: 0.021470471993088724\n",
      "Epoch: 882, Batch: 400, Avg. Loss: 0.02564213179051876\n",
      "Epoch: 882, Batch: 500, Avg. Loss: 0.02006403900682926\n",
      "Epoch: 882, Batch: 600, Avg. Loss: 0.017397905215620995\n",
      "Epoch: 882, Batch: 700, Avg. Loss: 0.018081926554441453\n",
      "Epoch: 882, Batch: 800, Avg. Loss: 0.02013479806482792\n",
      "Epoch: 882, Batch: 900, Avg. Loss: 0.021014817282557486\n",
      "Epoch: 883, Batch: 100, Avg. Loss: 0.01640609160065651\n",
      "Epoch: 883, Batch: 200, Avg. Loss: 0.025161627680063248\n",
      "Epoch: 883, Batch: 300, Avg. Loss: 0.02342095263302326\n",
      "Epoch: 883, Batch: 400, Avg. Loss: 0.02201952263712883\n",
      "Epoch: 883, Batch: 500, Avg. Loss: 0.020454491749405862\n",
      "Epoch: 883, Batch: 600, Avg. Loss: 0.01559465542435646\n",
      "Epoch: 883, Batch: 700, Avg. Loss: 0.022155080065131186\n",
      "Epoch: 883, Batch: 800, Avg. Loss: 0.02069055363535881\n",
      "Epoch: 883, Batch: 900, Avg. Loss: 0.022267769426107406\n",
      "Epoch: 884, Batch: 100, Avg. Loss: 0.017277973741292953\n",
      "Epoch: 884, Batch: 200, Avg. Loss: 0.022675651162862777\n",
      "Epoch: 884, Batch: 300, Avg. Loss: 0.03178248330950737\n",
      "Epoch: 884, Batch: 400, Avg. Loss: 0.018919770941138267\n",
      "Epoch: 884, Batch: 500, Avg. Loss: 0.015907507762312888\n",
      "Epoch: 884, Batch: 600, Avg. Loss: 0.018750937804579735\n",
      "Epoch: 884, Batch: 700, Avg. Loss: 0.023376207947731018\n",
      "Epoch: 884, Batch: 800, Avg. Loss: 0.01627832494676113\n",
      "Epoch: 884, Batch: 900, Avg. Loss: 0.017137992158532143\n",
      "Epoch: 885, Batch: 100, Avg. Loss: 0.018137393593788145\n",
      "Epoch: 885, Batch: 200, Avg. Loss: 0.021361436173319818\n",
      "Epoch: 885, Batch: 300, Avg. Loss: 0.015454579219222069\n",
      "Epoch: 885, Batch: 400, Avg. Loss: 0.01968844309449196\n",
      "Epoch: 885, Batch: 500, Avg. Loss: 0.023915795907378197\n",
      "Epoch: 885, Batch: 600, Avg. Loss: 0.019765576273202898\n",
      "Epoch: 885, Batch: 700, Avg. Loss: 0.024342261627316474\n",
      "Epoch: 885, Batch: 800, Avg. Loss: 0.02247439056634903\n",
      "Epoch: 885, Batch: 900, Avg. Loss: 0.019711928516626356\n",
      "Epoch: 886, Batch: 100, Avg. Loss: 0.025254935920238496\n",
      "Epoch: 886, Batch: 200, Avg. Loss: 0.020477306768298148\n",
      "Epoch: 886, Batch: 300, Avg. Loss: 0.019680280685424805\n",
      "Epoch: 886, Batch: 400, Avg. Loss: 0.020799012184143068\n",
      "Epoch: 886, Batch: 500, Avg. Loss: 0.01958282120525837\n",
      "Epoch: 886, Batch: 600, Avg. Loss: 0.02132268399000168\n",
      "Epoch: 886, Batch: 700, Avg. Loss: 0.015754392370581627\n",
      "Epoch: 886, Batch: 800, Avg. Loss: 0.019397898763418197\n",
      "Epoch: 886, Batch: 900, Avg. Loss: 0.021965362578630448\n",
      "Epoch: 887, Batch: 100, Avg. Loss: 0.019504294395446778\n",
      "Epoch: 887, Batch: 200, Avg. Loss: 0.01930577300488949\n",
      "Epoch: 887, Batch: 300, Avg. Loss: 0.025462519377470016\n",
      "Epoch: 887, Batch: 400, Avg. Loss: 0.013335716724395753\n",
      "Epoch: 887, Batch: 500, Avg. Loss: 0.018204562216997146\n",
      "Epoch: 887, Batch: 600, Avg. Loss: 0.021015422642230986\n",
      "Epoch: 887, Batch: 700, Avg. Loss: 0.01817134812474251\n",
      "Epoch: 887, Batch: 800, Avg. Loss: 0.019646237567067147\n",
      "Epoch: 887, Batch: 900, Avg. Loss: 0.02045969642698765\n",
      "Epoch: 888, Batch: 100, Avg. Loss: 0.017050912752747536\n",
      "Epoch: 888, Batch: 200, Avg. Loss: 0.0200607381016016\n",
      "Epoch: 888, Batch: 300, Avg. Loss: 0.020751451775431633\n",
      "Epoch: 888, Batch: 400, Avg. Loss: 0.018453679010272027\n",
      "Epoch: 888, Batch: 500, Avg. Loss: 0.01598670944571495\n",
      "Epoch: 888, Batch: 600, Avg. Loss: 0.023499047309160234\n",
      "Epoch: 888, Batch: 700, Avg. Loss: 0.019403935223817826\n",
      "Epoch: 888, Batch: 800, Avg. Loss: 0.020271439626812934\n",
      "Epoch: 888, Batch: 900, Avg. Loss: 0.023578364551067352\n",
      "Epoch: 889, Batch: 100, Avg. Loss: 0.01915097288787365\n",
      "Epoch: 889, Batch: 200, Avg. Loss: 0.019777772277593614\n",
      "Epoch: 889, Batch: 300, Avg. Loss: 0.021859159618616106\n",
      "Epoch: 889, Batch: 400, Avg. Loss: 0.0181654904037714\n",
      "Epoch: 889, Batch: 500, Avg. Loss: 0.017583022266626357\n",
      "Epoch: 889, Batch: 600, Avg. Loss: 0.02106819733977318\n",
      "Epoch: 889, Batch: 700, Avg. Loss: 0.02167035885155201\n",
      "Epoch: 889, Batch: 800, Avg. Loss: 0.013586712554097175\n",
      "Epoch: 889, Batch: 900, Avg. Loss: 0.020588354244828223\n",
      "Epoch: 890, Batch: 100, Avg. Loss: 0.024305475950241087\n",
      "Epoch: 890, Batch: 200, Avg. Loss: 0.019040227755904197\n",
      "Epoch: 890, Batch: 300, Avg. Loss: 0.019010711759328842\n",
      "Epoch: 890, Batch: 400, Avg. Loss: 0.02114336997270584\n",
      "Epoch: 890, Batch: 500, Avg. Loss: 0.01488926388323307\n",
      "Epoch: 890, Batch: 600, Avg. Loss: 0.020188793540000916\n",
      "Epoch: 890, Batch: 700, Avg. Loss: 0.027406822592020035\n",
      "Epoch: 890, Batch: 800, Avg. Loss: 0.0173789032548666\n",
      "Epoch: 890, Batch: 900, Avg. Loss: 0.021596425622701646\n",
      "Epoch: 891, Batch: 100, Avg. Loss: 0.01963107369840145\n",
      "Epoch: 891, Batch: 200, Avg. Loss: 0.02271018013358116\n",
      "Epoch: 891, Batch: 300, Avg. Loss: 0.018457193076610565\n",
      "Epoch: 891, Batch: 400, Avg. Loss: 0.02176760010421276\n",
      "Epoch: 891, Batch: 500, Avg. Loss: 0.023476037904620172\n",
      "Epoch: 891, Batch: 600, Avg. Loss: 0.02034299969673157\n",
      "Epoch: 891, Batch: 700, Avg. Loss: 0.021930310279130935\n",
      "Epoch: 891, Batch: 800, Avg. Loss: 0.017953028678894044\n",
      "Epoch: 891, Batch: 900, Avg. Loss: 0.021236902251839638\n",
      "Epoch: 892, Batch: 100, Avg. Loss: 0.018033602759242057\n",
      "Epoch: 892, Batch: 200, Avg. Loss: 0.021594098657369613\n",
      "Epoch: 892, Batch: 300, Avg. Loss: 0.027681907713413237\n",
      "Epoch: 892, Batch: 400, Avg. Loss: 0.01972223527729511\n",
      "Epoch: 892, Batch: 500, Avg. Loss: 0.018226763978600503\n",
      "Epoch: 892, Batch: 600, Avg. Loss: 0.01577367514371872\n",
      "Epoch: 892, Batch: 700, Avg. Loss: 0.020982354879379272\n",
      "Epoch: 892, Batch: 800, Avg. Loss: 0.020201728120446204\n",
      "Epoch: 892, Batch: 900, Avg. Loss: 0.019279484748840333\n",
      "Epoch: 893, Batch: 100, Avg. Loss: 0.02322263479232788\n",
      "Epoch: 893, Batch: 200, Avg. Loss: 0.0173087228089571\n",
      "Epoch: 893, Batch: 300, Avg. Loss: 0.024666423499584197\n",
      "Epoch: 893, Batch: 400, Avg. Loss: 0.018082779720425605\n",
      "Epoch: 893, Batch: 500, Avg. Loss: 0.021099825948476793\n",
      "Epoch: 893, Batch: 600, Avg. Loss: 0.02347302161157131\n",
      "Epoch: 893, Batch: 700, Avg. Loss: 0.016943592131137847\n",
      "Epoch: 893, Batch: 800, Avg. Loss: 0.017316101789474486\n",
      "Epoch: 893, Batch: 900, Avg. Loss: 0.0236856971681118\n",
      "Epoch: 894, Batch: 100, Avg. Loss: 0.017976878955960274\n",
      "Epoch: 894, Batch: 200, Avg. Loss: 0.017875179946422577\n",
      "Epoch: 894, Batch: 300, Avg. Loss: 0.024580438435077668\n",
      "Epoch: 894, Batch: 400, Avg. Loss: 0.017838039472699165\n",
      "Epoch: 894, Batch: 500, Avg. Loss: 0.018359659388661385\n",
      "Epoch: 894, Batch: 600, Avg. Loss: 0.022514677494764327\n",
      "Epoch: 894, Batch: 700, Avg. Loss: 0.02390012688934803\n",
      "Epoch: 894, Batch: 800, Avg. Loss: 0.020946084037423134\n",
      "Epoch: 894, Batch: 900, Avg. Loss: 0.019875323623418806\n",
      "Epoch: 895, Batch: 100, Avg. Loss: 0.01670540288090706\n",
      "Epoch: 895, Batch: 200, Avg. Loss: 0.02218723438680172\n",
      "Epoch: 895, Batch: 300, Avg. Loss: 0.021026450023055077\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 895, Batch: 400, Avg. Loss: 0.020017127469182015\n",
      "Epoch: 895, Batch: 500, Avg. Loss: 0.025012613832950593\n",
      "Epoch: 895, Batch: 600, Avg. Loss: 0.019731187894940376\n",
      "Epoch: 895, Batch: 700, Avg. Loss: 0.01844244472682476\n",
      "Epoch: 895, Batch: 800, Avg. Loss: 0.01727038897573948\n",
      "Epoch: 895, Batch: 900, Avg. Loss: 0.01881115823984146\n",
      "Epoch: 896, Batch: 100, Avg. Loss: 0.020395245775580405\n",
      "Epoch: 896, Batch: 200, Avg. Loss: 0.01786045253276825\n",
      "Epoch: 896, Batch: 300, Avg. Loss: 0.020419780388474463\n",
      "Epoch: 896, Batch: 400, Avg. Loss: 0.022694299295544625\n",
      "Epoch: 896, Batch: 500, Avg. Loss: 0.0203960070759058\n",
      "Epoch: 896, Batch: 600, Avg. Loss: 0.017076102420687675\n",
      "Epoch: 896, Batch: 700, Avg. Loss: 0.018244484886527062\n",
      "Epoch: 896, Batch: 800, Avg. Loss: 0.019635929241776467\n",
      "Epoch: 896, Batch: 900, Avg. Loss: 0.020604601353406905\n",
      "Epoch: 897, Batch: 100, Avg. Loss: 0.0159156509488821\n",
      "Epoch: 897, Batch: 200, Avg. Loss: 0.01828557774424553\n",
      "Epoch: 897, Batch: 300, Avg. Loss: 0.024520707726478578\n",
      "Epoch: 897, Batch: 400, Avg. Loss: 0.01963043965399265\n",
      "Epoch: 897, Batch: 500, Avg. Loss: 0.02238004848361015\n",
      "Epoch: 897, Batch: 600, Avg. Loss: 0.02300034113228321\n",
      "Epoch: 897, Batch: 700, Avg. Loss: 0.017849329747259616\n",
      "Epoch: 897, Batch: 800, Avg. Loss: 0.02140253357589245\n",
      "Epoch: 897, Batch: 900, Avg. Loss: 0.01975686855614185\n",
      "Epoch: 898, Batch: 100, Avg. Loss: 0.018540306314826012\n",
      "Epoch: 898, Batch: 200, Avg. Loss: 0.020602150335907937\n",
      "Epoch: 898, Batch: 300, Avg. Loss: 0.01715181365609169\n",
      "Epoch: 898, Batch: 400, Avg. Loss: 0.02273820549249649\n",
      "Epoch: 898, Batch: 500, Avg. Loss: 0.02240362785756588\n",
      "Epoch: 898, Batch: 600, Avg. Loss: 0.01722930446267128\n",
      "Epoch: 898, Batch: 700, Avg. Loss: 0.021594206690788268\n",
      "Epoch: 898, Batch: 800, Avg. Loss: 0.019970807880163192\n",
      "Epoch: 898, Batch: 900, Avg. Loss: 0.0174184288084507\n",
      "Epoch: 899, Batch: 100, Avg. Loss: 0.02161603406071663\n",
      "Epoch: 899, Batch: 200, Avg. Loss: 0.02608794830739498\n",
      "Epoch: 899, Batch: 300, Avg. Loss: 0.018616070598363878\n",
      "Epoch: 899, Batch: 400, Avg. Loss: 0.020600408539175986\n",
      "Epoch: 899, Batch: 500, Avg. Loss: 0.01977932296693325\n",
      "Epoch: 899, Batch: 600, Avg. Loss: 0.025762609541416168\n",
      "Epoch: 899, Batch: 700, Avg. Loss: 0.019727589264512062\n",
      "Epoch: 899, Batch: 800, Avg. Loss: 0.020181612074375153\n",
      "Epoch: 899, Batch: 900, Avg. Loss: 0.02081219054758549\n",
      "Epoch: 900, Batch: 100, Avg. Loss: 0.016749523803591727\n",
      "Epoch: 900, Batch: 200, Avg. Loss: 0.01927998848259449\n",
      "Epoch: 900, Batch: 300, Avg. Loss: 0.017091370224952697\n",
      "Epoch: 900, Batch: 400, Avg. Loss: 0.0145354975014925\n",
      "Epoch: 900, Batch: 500, Avg. Loss: 0.020753381848335264\n",
      "Epoch: 900, Batch: 600, Avg. Loss: 0.024518323838710786\n",
      "Epoch: 900, Batch: 700, Avg. Loss: 0.022323605939745902\n",
      "Epoch: 900, Batch: 800, Avg. Loss: 0.024029012620449066\n",
      "Epoch: 900, Batch: 900, Avg. Loss: 0.02110140986740589\n",
      "Epoch: 901, Batch: 100, Avg. Loss: 0.02051344349980354\n",
      "Epoch: 901, Batch: 200, Avg. Loss: 0.017227404937148093\n",
      "Epoch: 901, Batch: 300, Avg. Loss: 0.01993381083011627\n",
      "Epoch: 901, Batch: 400, Avg. Loss: 0.027847626581788062\n",
      "Epoch: 901, Batch: 500, Avg. Loss: 0.02128538057208061\n",
      "Epoch: 901, Batch: 600, Avg. Loss: 0.0193367899954319\n",
      "Epoch: 901, Batch: 700, Avg. Loss: 0.01792436771094799\n",
      "Epoch: 901, Batch: 800, Avg. Loss: 0.02293258734047413\n",
      "Epoch: 901, Batch: 900, Avg. Loss: 0.021676329225301744\n",
      "Epoch: 902, Batch: 100, Avg. Loss: 0.021293143182992934\n",
      "Epoch: 902, Batch: 200, Avg. Loss: 0.01915366768836975\n",
      "Epoch: 902, Batch: 300, Avg. Loss: 0.02057406686246395\n",
      "Epoch: 902, Batch: 400, Avg. Loss: 0.020381339266896246\n",
      "Epoch: 902, Batch: 500, Avg. Loss: 0.022686650082468986\n",
      "Epoch: 902, Batch: 600, Avg. Loss: 0.018873994797468187\n",
      "Epoch: 902, Batch: 700, Avg. Loss: 0.024950919821858406\n",
      "Epoch: 902, Batch: 800, Avg. Loss: 0.015356605276465415\n",
      "Epoch: 902, Batch: 900, Avg. Loss: 0.020800668373703958\n",
      "Epoch: 903, Batch: 100, Avg. Loss: 0.017792239785194397\n",
      "Epoch: 903, Batch: 200, Avg. Loss: 0.020512424781918524\n",
      "Epoch: 903, Batch: 300, Avg. Loss: 0.015605389922857284\n",
      "Epoch: 903, Batch: 400, Avg. Loss: 0.02001462534070015\n",
      "Epoch: 903, Batch: 500, Avg. Loss: 0.02091730624437332\n",
      "Epoch: 903, Batch: 600, Avg. Loss: 0.026813054531812666\n",
      "Epoch: 903, Batch: 700, Avg. Loss: 0.020880520343780518\n",
      "Epoch: 903, Batch: 800, Avg. Loss: 0.020205331444740297\n",
      "Epoch: 903, Batch: 900, Avg. Loss: 0.017592633739113808\n",
      "Epoch: 904, Batch: 100, Avg. Loss: 0.022149649411439896\n",
      "Epoch: 904, Batch: 200, Avg. Loss: 0.01702900432050228\n",
      "Epoch: 904, Batch: 300, Avg. Loss: 0.02313629038631916\n",
      "Epoch: 904, Batch: 400, Avg. Loss: 0.021165911108255386\n",
      "Epoch: 904, Batch: 500, Avg. Loss: 0.022534493580460547\n",
      "Epoch: 904, Batch: 600, Avg. Loss: 0.020554560422897338\n",
      "Epoch: 904, Batch: 700, Avg. Loss: 0.015264238044619561\n",
      "Epoch: 904, Batch: 800, Avg. Loss: 0.021600615233182907\n",
      "Epoch: 904, Batch: 900, Avg. Loss: 0.01915497936308384\n",
      "Epoch: 905, Batch: 100, Avg. Loss: 0.01872128814458847\n",
      "Epoch: 905, Batch: 200, Avg. Loss: 0.014803243279457092\n",
      "Epoch: 905, Batch: 300, Avg. Loss: 0.021573334485292434\n",
      "Epoch: 905, Batch: 400, Avg. Loss: 0.021991559565067292\n",
      "Epoch: 905, Batch: 500, Avg. Loss: 0.021445616781711578\n",
      "Epoch: 905, Batch: 600, Avg. Loss: 0.02094325102865696\n",
      "Epoch: 905, Batch: 700, Avg. Loss: 0.019169511348009108\n",
      "Epoch: 905, Batch: 800, Avg. Loss: 0.018947327062487602\n",
      "Epoch: 905, Batch: 900, Avg. Loss: 0.015858924388885497\n",
      "Epoch: 906, Batch: 100, Avg. Loss: 0.02246761493384838\n",
      "Epoch: 906, Batch: 200, Avg. Loss: 0.021520407348871232\n",
      "Epoch: 906, Batch: 300, Avg. Loss: 0.01745028719305992\n",
      "Epoch: 906, Batch: 400, Avg. Loss: 0.021514060348272322\n",
      "Epoch: 906, Batch: 500, Avg. Loss: 0.023072278201580046\n",
      "Epoch: 906, Batch: 600, Avg. Loss: 0.020408793464303018\n",
      "Epoch: 906, Batch: 700, Avg. Loss: 0.016350567117333414\n",
      "Epoch: 906, Batch: 800, Avg. Loss: 0.019271623864769934\n",
      "Epoch: 906, Batch: 900, Avg. Loss: 0.019409169629216193\n",
      "Epoch: 907, Batch: 100, Avg. Loss: 0.01705814018845558\n",
      "Epoch: 907, Batch: 200, Avg. Loss: 0.016916735172271727\n",
      "Epoch: 907, Batch: 300, Avg. Loss: 0.022920734360814096\n",
      "Epoch: 907, Batch: 400, Avg. Loss: 0.021410772129893303\n",
      "Epoch: 907, Batch: 500, Avg. Loss: 0.021495947763323784\n",
      "Epoch: 907, Batch: 600, Avg. Loss: 0.021369367986917496\n",
      "Epoch: 907, Batch: 700, Avg. Loss: 0.016544924527406694\n",
      "Epoch: 907, Batch: 800, Avg. Loss: 0.019642313793301582\n",
      "Epoch: 907, Batch: 900, Avg. Loss: 0.020873311460018158\n",
      "Epoch: 908, Batch: 100, Avg. Loss: 0.01772385597229004\n",
      "Epoch: 908, Batch: 200, Avg. Loss: 0.01951372981071472\n",
      "Epoch: 908, Batch: 300, Avg. Loss: 0.019749108254909515\n",
      "Epoch: 908, Batch: 400, Avg. Loss: 0.017667853236198426\n",
      "Epoch: 908, Batch: 500, Avg. Loss: 0.024928052499890326\n",
      "Epoch: 908, Batch: 600, Avg. Loss: 0.018211869597434996\n",
      "Epoch: 908, Batch: 700, Avg. Loss: 0.017721810638904573\n",
      "Epoch: 908, Batch: 800, Avg. Loss: 0.021603784188628197\n",
      "Epoch: 908, Batch: 900, Avg. Loss: 0.018929606676101683\n",
      "Epoch: 909, Batch: 100, Avg. Loss: 0.019344548657536507\n",
      "Epoch: 909, Batch: 200, Avg. Loss: 0.02473255082964897\n",
      "Epoch: 909, Batch: 300, Avg. Loss: 0.019394353851675986\n",
      "Epoch: 909, Batch: 400, Avg. Loss: 0.020844405069947242\n",
      "Epoch: 909, Batch: 500, Avg. Loss: 0.015329059138894081\n",
      "Epoch: 909, Batch: 600, Avg. Loss: 0.025356677547097205\n",
      "Epoch: 909, Batch: 700, Avg. Loss: 0.01967949904501438\n",
      "Epoch: 909, Batch: 800, Avg. Loss: 0.02233735740184784\n",
      "Epoch: 909, Batch: 900, Avg. Loss: 0.017833757475018502\n",
      "Epoch: 910, Batch: 100, Avg. Loss: 0.019840942174196245\n",
      "Epoch: 910, Batch: 200, Avg. Loss: 0.018346361219882965\n",
      "Epoch: 910, Batch: 300, Avg. Loss: 0.016092484220862388\n",
      "Epoch: 910, Batch: 400, Avg. Loss: 0.017590874657034874\n",
      "Epoch: 910, Batch: 500, Avg. Loss: 0.023567092046141623\n",
      "Epoch: 910, Batch: 600, Avg. Loss: 0.02294609971344471\n",
      "Epoch: 910, Batch: 700, Avg. Loss: 0.016163348630070688\n",
      "Epoch: 910, Batch: 800, Avg. Loss: 0.024847273305058478\n",
      "Epoch: 910, Batch: 900, Avg. Loss: 0.016064387857913972\n",
      "Epoch: 911, Batch: 100, Avg. Loss: 0.01990901455283165\n",
      "Epoch: 911, Batch: 200, Avg. Loss: 0.019457411468029023\n",
      "Epoch: 911, Batch: 300, Avg. Loss: 0.01740145489573479\n",
      "Epoch: 911, Batch: 400, Avg. Loss: 0.021425303444266318\n",
      "Epoch: 911, Batch: 500, Avg. Loss: 0.022263096198439598\n",
      "Epoch: 911, Batch: 600, Avg. Loss: 0.017827313020825387\n",
      "Epoch: 911, Batch: 700, Avg. Loss: 0.017044285461306572\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 911, Batch: 800, Avg. Loss: 0.01948010928928852\n",
      "Epoch: 911, Batch: 900, Avg. Loss: 0.020437993854284287\n",
      "Epoch: 912, Batch: 100, Avg. Loss: 0.01845417231321335\n",
      "Epoch: 912, Batch: 200, Avg. Loss: 0.018207742124795912\n",
      "Epoch: 912, Batch: 300, Avg. Loss: 0.02604652389883995\n",
      "Epoch: 912, Batch: 400, Avg. Loss: 0.020885208621621132\n",
      "Epoch: 912, Batch: 500, Avg. Loss: 0.024072247818112374\n",
      "Epoch: 912, Batch: 600, Avg. Loss: 0.016234031319618224\n",
      "Epoch: 912, Batch: 700, Avg. Loss: 0.01642429880797863\n",
      "Epoch: 912, Batch: 800, Avg. Loss: 0.021962937116622925\n",
      "Epoch: 912, Batch: 900, Avg. Loss: 0.018028604090213774\n",
      "Epoch: 913, Batch: 100, Avg. Loss: 0.023292988911271094\n",
      "Epoch: 913, Batch: 200, Avg. Loss: 0.0213146460801363\n",
      "Epoch: 913, Batch: 300, Avg. Loss: 0.021280434727668763\n",
      "Epoch: 913, Batch: 400, Avg. Loss: 0.018022118508815764\n",
      "Epoch: 913, Batch: 500, Avg. Loss: 0.017401907593011856\n",
      "Epoch: 913, Batch: 600, Avg. Loss: 0.018248206600546837\n",
      "Epoch: 913, Batch: 700, Avg. Loss: 0.01905434384942055\n",
      "Epoch: 913, Batch: 800, Avg. Loss: 0.018011229187250136\n",
      "Epoch: 913, Batch: 900, Avg. Loss: 0.02133529670536518\n",
      "Epoch: 914, Batch: 100, Avg. Loss: 0.020521329939365385\n",
      "Epoch: 914, Batch: 200, Avg. Loss: 0.021308539733290673\n",
      "Epoch: 914, Batch: 300, Avg. Loss: 0.014976854473352431\n",
      "Epoch: 914, Batch: 400, Avg. Loss: 0.022604209035634995\n",
      "Epoch: 914, Batch: 500, Avg. Loss: 0.02545414596796036\n",
      "Epoch: 914, Batch: 600, Avg. Loss: 0.015554028674960136\n",
      "Epoch: 914, Batch: 700, Avg. Loss: 0.02067277893424034\n",
      "Epoch: 914, Batch: 800, Avg. Loss: 0.020600293800234794\n",
      "Epoch: 914, Batch: 900, Avg. Loss: 0.024536880925297738\n",
      "Epoch: 915, Batch: 100, Avg. Loss: 0.024033157005906105\n",
      "Epoch: 915, Batch: 200, Avg. Loss: 0.022803207337856294\n",
      "Epoch: 915, Batch: 300, Avg. Loss: 0.019621280133724214\n",
      "Epoch: 915, Batch: 400, Avg. Loss: 0.021467264890670777\n",
      "Epoch: 915, Batch: 500, Avg. Loss: 0.016727547124028207\n",
      "Epoch: 915, Batch: 600, Avg. Loss: 0.017979309111833573\n",
      "Epoch: 915, Batch: 700, Avg. Loss: 0.02251102201640606\n",
      "Epoch: 915, Batch: 800, Avg. Loss: 0.019776287525892257\n",
      "Epoch: 915, Batch: 900, Avg. Loss: 0.02240016482770443\n",
      "Epoch: 916, Batch: 100, Avg. Loss: 0.015262908041477204\n",
      "Epoch: 916, Batch: 200, Avg. Loss: 0.022545866668224335\n",
      "Epoch: 916, Batch: 300, Avg. Loss: 0.019435197487473486\n",
      "Epoch: 916, Batch: 400, Avg. Loss: 0.019412533640861512\n",
      "Epoch: 916, Batch: 500, Avg. Loss: 0.019026537388563157\n",
      "Epoch: 916, Batch: 600, Avg. Loss: 0.020952498316764833\n",
      "Epoch: 916, Batch: 700, Avg. Loss: 0.02157415121793747\n",
      "Epoch: 916, Batch: 800, Avg. Loss: 0.019061304852366446\n",
      "Epoch: 916, Batch: 900, Avg. Loss: 0.02067274235188961\n",
      "Epoch: 917, Batch: 100, Avg. Loss: 0.02521149270236492\n",
      "Epoch: 917, Batch: 200, Avg. Loss: 0.015280278101563453\n",
      "Epoch: 917, Batch: 300, Avg. Loss: 0.019447320550680162\n",
      "Epoch: 917, Batch: 400, Avg. Loss: 0.021978871896862984\n",
      "Epoch: 917, Batch: 500, Avg. Loss: 0.020716782212257385\n",
      "Epoch: 917, Batch: 600, Avg. Loss: 0.019260440170764923\n",
      "Epoch: 917, Batch: 700, Avg. Loss: 0.021531586349010468\n",
      "Epoch: 917, Batch: 800, Avg. Loss: 0.023232221975922583\n",
      "Epoch: 917, Batch: 900, Avg. Loss: 0.018879724219441413\n",
      "Epoch: 918, Batch: 100, Avg. Loss: 0.01611030824482441\n",
      "Epoch: 918, Batch: 200, Avg. Loss: 0.023687038496136667\n",
      "Epoch: 918, Batch: 300, Avg. Loss: 0.022273278906941415\n",
      "Epoch: 918, Batch: 400, Avg. Loss: 0.016525664627552034\n",
      "Epoch: 918, Batch: 500, Avg. Loss: 0.018828401044011116\n",
      "Epoch: 918, Batch: 600, Avg. Loss: 0.01859372019767761\n",
      "Epoch: 918, Batch: 700, Avg. Loss: 0.018158151507377623\n",
      "Epoch: 918, Batch: 800, Avg. Loss: 0.01543685346841812\n",
      "Epoch: 918, Batch: 900, Avg. Loss: 0.021779442951083184\n",
      "Epoch: 919, Batch: 100, Avg. Loss: 0.018928094506263732\n",
      "Epoch: 919, Batch: 200, Avg. Loss: 0.01827133111655712\n",
      "Epoch: 919, Batch: 300, Avg. Loss: 0.014819392710924148\n",
      "Epoch: 919, Batch: 400, Avg. Loss: 0.014487004056572914\n",
      "Epoch: 919, Batch: 500, Avg. Loss: 0.022060655802488328\n",
      "Epoch: 919, Batch: 600, Avg. Loss: 0.019608048647642137\n",
      "Epoch: 919, Batch: 700, Avg. Loss: 0.020937833413481714\n",
      "Epoch: 919, Batch: 800, Avg. Loss: 0.02199920527637005\n",
      "Epoch: 919, Batch: 900, Avg. Loss: 0.018987169787287712\n",
      "Epoch: 920, Batch: 100, Avg. Loss: 0.020606567934155464\n",
      "Epoch: 920, Batch: 200, Avg. Loss: 0.017529076486825942\n",
      "Epoch: 920, Batch: 300, Avg. Loss: 0.021508302465081217\n",
      "Epoch: 920, Batch: 400, Avg. Loss: 0.01846304051578045\n",
      "Epoch: 920, Batch: 500, Avg. Loss: 0.01802357278764248\n",
      "Epoch: 920, Batch: 600, Avg. Loss: 0.02237366609275341\n",
      "Epoch: 920, Batch: 700, Avg. Loss: 0.019516141340136527\n",
      "Epoch: 920, Batch: 800, Avg. Loss: 0.01815998710691929\n",
      "Epoch: 920, Batch: 900, Avg. Loss: 0.026217744201421738\n",
      "Epoch: 921, Batch: 100, Avg. Loss: 0.020188561156392098\n",
      "Epoch: 921, Batch: 200, Avg. Loss: 0.01874430276453495\n",
      "Epoch: 921, Batch: 300, Avg. Loss: 0.019703699424862862\n",
      "Epoch: 921, Batch: 400, Avg. Loss: 0.028663368225097657\n",
      "Epoch: 921, Batch: 500, Avg. Loss: 0.018526632636785507\n",
      "Epoch: 921, Batch: 600, Avg. Loss: 0.01757318541407585\n",
      "Epoch: 921, Batch: 700, Avg. Loss: 0.020526355504989623\n",
      "Epoch: 921, Batch: 800, Avg. Loss: 0.019734733253717423\n",
      "Epoch: 921, Batch: 900, Avg. Loss: 0.01628660164773464\n",
      "Epoch: 922, Batch: 100, Avg. Loss: 0.022777794972062112\n",
      "Epoch: 922, Batch: 200, Avg. Loss: 0.015678000003099442\n",
      "Epoch: 922, Batch: 300, Avg. Loss: 0.019873855337500572\n",
      "Epoch: 922, Batch: 400, Avg. Loss: 0.018969659470021726\n",
      "Epoch: 922, Batch: 500, Avg. Loss: 0.020822039693593978\n",
      "Epoch: 922, Batch: 600, Avg. Loss: 0.016670400574803353\n",
      "Epoch: 922, Batch: 700, Avg. Loss: 0.01814457982778549\n",
      "Epoch: 922, Batch: 800, Avg. Loss: 0.018846402987837792\n",
      "Epoch: 922, Batch: 900, Avg. Loss: 0.016297892704606055\n",
      "Epoch: 923, Batch: 100, Avg. Loss: 0.01574415624141693\n",
      "Epoch: 923, Batch: 200, Avg. Loss: 0.01852736383676529\n",
      "Epoch: 923, Batch: 300, Avg. Loss: 0.020245237052440645\n",
      "Epoch: 923, Batch: 400, Avg. Loss: 0.019855407476425172\n",
      "Epoch: 923, Batch: 500, Avg. Loss: 0.0245259290933609\n",
      "Epoch: 923, Batch: 600, Avg. Loss: 0.0262115803360939\n",
      "Epoch: 923, Batch: 700, Avg. Loss: 0.015416666343808174\n",
      "Epoch: 923, Batch: 800, Avg. Loss: 0.022720181345939637\n",
      "Epoch: 923, Batch: 900, Avg. Loss: 0.017797956243157387\n",
      "Epoch: 924, Batch: 100, Avg. Loss: 0.01922278232872486\n",
      "Epoch: 924, Batch: 200, Avg. Loss: 0.0221492587774992\n",
      "Epoch: 924, Batch: 300, Avg. Loss: 0.019464715272188186\n",
      "Epoch: 924, Batch: 400, Avg. Loss: 0.021894965842366217\n",
      "Epoch: 924, Batch: 500, Avg. Loss: 0.019225765243172644\n",
      "Epoch: 924, Batch: 600, Avg. Loss: 0.026468818858265877\n",
      "Epoch: 924, Batch: 700, Avg. Loss: 0.017566834911704063\n",
      "Epoch: 924, Batch: 800, Avg. Loss: 0.0239597287774086\n",
      "Epoch: 924, Batch: 900, Avg. Loss: 0.02417202979326248\n",
      "Epoch: 925, Batch: 100, Avg. Loss: 0.02069594316184521\n",
      "Epoch: 925, Batch: 200, Avg. Loss: 0.01859367474913597\n",
      "Epoch: 925, Batch: 300, Avg. Loss: 0.0197357264906168\n",
      "Epoch: 925, Batch: 400, Avg. Loss: 0.019958153665065766\n",
      "Epoch: 925, Batch: 500, Avg. Loss: 0.019669727683067323\n",
      "Epoch: 925, Batch: 600, Avg. Loss: 0.01828882783651352\n",
      "Epoch: 925, Batch: 700, Avg. Loss: 0.01861036367714405\n",
      "Epoch: 925, Batch: 800, Avg. Loss: 0.01821835748851299\n",
      "Epoch: 925, Batch: 900, Avg. Loss: 0.01701666824519634\n",
      "Epoch: 926, Batch: 100, Avg. Loss: 0.026571234092116355\n",
      "Epoch: 926, Batch: 200, Avg. Loss: 0.017857332676649094\n",
      "Epoch: 926, Batch: 300, Avg. Loss: 0.02176506206393242\n",
      "Epoch: 926, Batch: 400, Avg. Loss: 0.020088164284825325\n",
      "Epoch: 926, Batch: 500, Avg. Loss: 0.02001196675002575\n",
      "Epoch: 926, Batch: 600, Avg. Loss: 0.016372359469532968\n",
      "Epoch: 926, Batch: 700, Avg. Loss: 0.016902076080441475\n",
      "Epoch: 926, Batch: 800, Avg. Loss: 0.01624928042292595\n",
      "Epoch: 926, Batch: 900, Avg. Loss: 0.021671339571475982\n",
      "Epoch: 927, Batch: 100, Avg. Loss: 0.022390538156032563\n",
      "Epoch: 927, Batch: 200, Avg. Loss: 0.022776820659637452\n",
      "Epoch: 927, Batch: 300, Avg. Loss: 0.014827235639095306\n",
      "Epoch: 927, Batch: 400, Avg. Loss: 0.014880644604563713\n",
      "Epoch: 927, Batch: 500, Avg. Loss: 0.016998954489827155\n",
      "Epoch: 927, Batch: 600, Avg. Loss: 0.02337743230164051\n",
      "Epoch: 927, Batch: 700, Avg. Loss: 0.023214397951960564\n",
      "Epoch: 927, Batch: 800, Avg. Loss: 0.01949158824980259\n",
      "Epoch: 927, Batch: 900, Avg. Loss: 0.017535741552710534\n",
      "Epoch: 928, Batch: 100, Avg. Loss: 0.018617140799760817\n",
      "Epoch: 928, Batch: 200, Avg. Loss: 0.015684503316879272\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 928, Batch: 300, Avg. Loss: 0.020450354367494584\n",
      "Epoch: 928, Batch: 400, Avg. Loss: 0.022370588779449464\n",
      "Epoch: 928, Batch: 500, Avg. Loss: 0.0185375814139843\n",
      "Epoch: 928, Batch: 600, Avg. Loss: 0.017370249181985854\n",
      "Epoch: 928, Batch: 700, Avg. Loss: 0.020004004538059235\n",
      "Epoch: 928, Batch: 800, Avg. Loss: 0.020418824553489687\n",
      "Epoch: 928, Batch: 900, Avg. Loss: 0.01893457807600498\n",
      "Epoch: 929, Batch: 100, Avg. Loss: 0.018047422021627426\n",
      "Epoch: 929, Batch: 200, Avg. Loss: 0.013860682621598244\n",
      "Epoch: 929, Batch: 300, Avg. Loss: 0.026552625447511673\n",
      "Epoch: 929, Batch: 400, Avg. Loss: 0.02105523243546486\n",
      "Epoch: 929, Batch: 500, Avg. Loss: 0.01768032096326351\n",
      "Epoch: 929, Batch: 600, Avg. Loss: 0.02387509424239397\n",
      "Epoch: 929, Batch: 700, Avg. Loss: 0.022538037672638893\n",
      "Epoch: 929, Batch: 800, Avg. Loss: 0.01751884274184704\n",
      "Epoch: 929, Batch: 900, Avg. Loss: 0.019589430168271065\n",
      "Epoch: 930, Batch: 100, Avg. Loss: 0.022427477836608888\n",
      "Epoch: 930, Batch: 200, Avg. Loss: 0.02052658297121525\n",
      "Epoch: 930, Batch: 300, Avg. Loss: 0.01685973085463047\n",
      "Epoch: 930, Batch: 400, Avg. Loss: 0.01843847893178463\n",
      "Epoch: 930, Batch: 500, Avg. Loss: 0.022692295983433725\n",
      "Epoch: 930, Batch: 600, Avg. Loss: 0.01837137769907713\n",
      "Epoch: 930, Batch: 700, Avg. Loss: 0.016260112971067428\n",
      "Epoch: 930, Batch: 800, Avg. Loss: 0.01743529997766018\n",
      "Epoch: 930, Batch: 900, Avg. Loss: 0.018582631796598435\n",
      "Epoch: 931, Batch: 100, Avg. Loss: 0.017149509489536287\n",
      "Epoch: 931, Batch: 200, Avg. Loss: 0.01694471389055252\n",
      "Epoch: 931, Batch: 300, Avg. Loss: 0.021862111389636993\n",
      "Epoch: 931, Batch: 400, Avg. Loss: 0.02146810628473759\n",
      "Epoch: 931, Batch: 500, Avg. Loss: 0.015403981953859329\n",
      "Epoch: 931, Batch: 600, Avg. Loss: 0.016720146611332892\n",
      "Epoch: 931, Batch: 700, Avg. Loss: 0.021472817733883858\n",
      "Epoch: 931, Batch: 800, Avg. Loss: 0.019183860868215562\n",
      "Epoch: 931, Batch: 900, Avg. Loss: 0.020399340093135834\n",
      "Epoch: 932, Batch: 100, Avg. Loss: 0.02042515590786934\n",
      "Epoch: 932, Batch: 200, Avg. Loss: 0.019615287035703657\n",
      "Epoch: 932, Batch: 300, Avg. Loss: 0.01942532368004322\n",
      "Epoch: 932, Batch: 400, Avg. Loss: 0.017404765188694\n",
      "Epoch: 932, Batch: 500, Avg. Loss: 0.02086331255733967\n",
      "Epoch: 932, Batch: 600, Avg. Loss: 0.019679245576262473\n",
      "Epoch: 932, Batch: 700, Avg. Loss: 0.0225107366591692\n",
      "Epoch: 932, Batch: 800, Avg. Loss: 0.02013414919376373\n",
      "Epoch: 932, Batch: 900, Avg. Loss: 0.01858546130359173\n",
      "Epoch: 933, Batch: 100, Avg. Loss: 0.01583618849515915\n",
      "Epoch: 933, Batch: 200, Avg. Loss: 0.017774989530444145\n",
      "Epoch: 933, Batch: 300, Avg. Loss: 0.01628115974366665\n",
      "Epoch: 933, Batch: 400, Avg. Loss: 0.022762363627552986\n",
      "Epoch: 933, Batch: 500, Avg. Loss: 0.01869284823536873\n",
      "Epoch: 933, Batch: 600, Avg. Loss: 0.020964203104376793\n",
      "Epoch: 933, Batch: 700, Avg. Loss: 0.017284172102808953\n",
      "Epoch: 933, Batch: 800, Avg. Loss: 0.01791635662317276\n",
      "Epoch: 933, Batch: 900, Avg. Loss: 0.019379512816667558\n",
      "Epoch: 934, Batch: 100, Avg. Loss: 0.01831139236688614\n",
      "Epoch: 934, Batch: 200, Avg. Loss: 0.02201729603111744\n",
      "Epoch: 934, Batch: 300, Avg. Loss: 0.018113539069890976\n",
      "Epoch: 934, Batch: 400, Avg. Loss: 0.02396179400384426\n",
      "Epoch: 934, Batch: 500, Avg. Loss: 0.021150591522455214\n",
      "Epoch: 934, Batch: 600, Avg. Loss: 0.022369813248515127\n",
      "Epoch: 934, Batch: 700, Avg. Loss: 0.02627487599849701\n",
      "Epoch: 934, Batch: 800, Avg. Loss: 0.021895422115921975\n",
      "Epoch: 934, Batch: 900, Avg. Loss: 0.014876340702176093\n",
      "Epoch: 935, Batch: 100, Avg. Loss: 0.018347925171256066\n",
      "Epoch: 935, Batch: 200, Avg. Loss: 0.023864996209740637\n",
      "Epoch: 935, Batch: 300, Avg. Loss: 0.014381666481494904\n",
      "Epoch: 935, Batch: 400, Avg. Loss: 0.023196747228503227\n",
      "Epoch: 935, Batch: 500, Avg. Loss: 0.022348467186093332\n",
      "Epoch: 935, Batch: 600, Avg. Loss: 0.024578730165958403\n",
      "Epoch: 935, Batch: 700, Avg. Loss: 0.019099352508783342\n",
      "Epoch: 935, Batch: 800, Avg. Loss: 0.02171566404402256\n",
      "Epoch: 935, Batch: 900, Avg. Loss: 0.018500407263636588\n",
      "Epoch: 936, Batch: 100, Avg. Loss: 0.017071285247802735\n",
      "Epoch: 936, Batch: 200, Avg. Loss: 0.019240343794226646\n",
      "Epoch: 936, Batch: 300, Avg. Loss: 0.02041391223669052\n",
      "Epoch: 936, Batch: 400, Avg. Loss: 0.021346075013279914\n",
      "Epoch: 936, Batch: 500, Avg. Loss: 0.019334445968270303\n",
      "Epoch: 936, Batch: 600, Avg. Loss: 0.021652635112404825\n",
      "Epoch: 936, Batch: 700, Avg. Loss: 0.023324009552598\n",
      "Epoch: 936, Batch: 800, Avg. Loss: 0.026055936887860297\n",
      "Epoch: 936, Batch: 900, Avg. Loss: 0.01496952049434185\n",
      "Epoch: 937, Batch: 100, Avg. Loss: 0.01617700934410095\n",
      "Epoch: 937, Batch: 200, Avg. Loss: 0.024576421976089478\n",
      "Epoch: 937, Batch: 300, Avg. Loss: 0.01980872355401516\n",
      "Epoch: 937, Batch: 400, Avg. Loss: 0.01875035248696804\n",
      "Epoch: 937, Batch: 500, Avg. Loss: 0.01945588640868664\n",
      "Epoch: 937, Batch: 600, Avg. Loss: 0.021237308382987975\n",
      "Epoch: 937, Batch: 700, Avg. Loss: 0.021727324202656745\n",
      "Epoch: 937, Batch: 800, Avg. Loss: 0.020245266631245613\n",
      "Epoch: 937, Batch: 900, Avg. Loss: 0.021394370943307876\n",
      "Epoch: 938, Batch: 100, Avg. Loss: 0.016121156066656112\n",
      "Epoch: 938, Batch: 200, Avg. Loss: 0.018890420645475386\n",
      "Epoch: 938, Batch: 300, Avg. Loss: 0.018049101606011392\n",
      "Epoch: 938, Batch: 400, Avg. Loss: 0.021194879487156867\n",
      "Epoch: 938, Batch: 500, Avg. Loss: 0.017487048655748367\n",
      "Epoch: 938, Batch: 600, Avg. Loss: 0.021920739486813546\n",
      "Epoch: 938, Batch: 700, Avg. Loss: 0.018113455697894097\n",
      "Epoch: 938, Batch: 800, Avg. Loss: 0.01996972516179085\n",
      "Epoch: 938, Batch: 900, Avg. Loss: 0.02065000243484974\n",
      "Epoch: 939, Batch: 100, Avg. Loss: 0.01768206261098385\n",
      "Epoch: 939, Batch: 200, Avg. Loss: 0.017200467996299266\n",
      "Epoch: 939, Batch: 300, Avg. Loss: 0.022174898870289325\n",
      "Epoch: 939, Batch: 400, Avg. Loss: 0.017281779572367668\n",
      "Epoch: 939, Batch: 500, Avg. Loss: 0.021326106265187262\n",
      "Epoch: 939, Batch: 600, Avg. Loss: 0.01794586181640625\n",
      "Epoch: 939, Batch: 700, Avg. Loss: 0.01830557219684124\n",
      "Epoch: 939, Batch: 800, Avg. Loss: 0.01936967097222805\n",
      "Epoch: 939, Batch: 900, Avg. Loss: 0.022174664288759232\n",
      "Epoch: 940, Batch: 100, Avg. Loss: 0.018739666901528836\n",
      "Epoch: 940, Batch: 200, Avg. Loss: 0.015945247486233712\n",
      "Epoch: 940, Batch: 300, Avg. Loss: 0.022082458063960076\n",
      "Epoch: 940, Batch: 400, Avg. Loss: 0.021647544801235197\n",
      "Epoch: 940, Batch: 500, Avg. Loss: 0.01806009329855442\n",
      "Epoch: 940, Batch: 600, Avg. Loss: 0.024054035395383835\n",
      "Epoch: 940, Batch: 700, Avg. Loss: 0.01875376082956791\n",
      "Epoch: 940, Batch: 800, Avg. Loss: 0.020995881408452988\n",
      "Epoch: 940, Batch: 900, Avg. Loss: 0.01932235449552536\n",
      "Epoch: 941, Batch: 100, Avg. Loss: 0.017427842020988463\n",
      "Epoch: 941, Batch: 200, Avg. Loss: 0.019714013412594794\n",
      "Epoch: 941, Batch: 300, Avg. Loss: 0.01860587775707245\n",
      "Epoch: 941, Batch: 400, Avg. Loss: 0.0164325612783432\n",
      "Epoch: 941, Batch: 500, Avg. Loss: 0.021265295296907426\n",
      "Epoch: 941, Batch: 600, Avg. Loss: 0.017662319019436837\n",
      "Epoch: 941, Batch: 700, Avg. Loss: 0.019784082621335984\n",
      "Epoch: 941, Batch: 800, Avg. Loss: 0.01959523156285286\n",
      "Epoch: 941, Batch: 900, Avg. Loss: 0.02133646883070469\n",
      "Epoch: 942, Batch: 100, Avg. Loss: 0.018350211977958678\n",
      "Epoch: 942, Batch: 200, Avg. Loss: 0.020817050859332085\n",
      "Epoch: 942, Batch: 300, Avg. Loss: 0.018402974084019662\n",
      "Epoch: 942, Batch: 400, Avg. Loss: 0.01379418969154358\n",
      "Epoch: 942, Batch: 500, Avg. Loss: 0.02121881678700447\n",
      "Epoch: 942, Batch: 600, Avg. Loss: 0.017121948897838593\n",
      "Epoch: 942, Batch: 700, Avg. Loss: 0.019242637529969214\n",
      "Epoch: 942, Batch: 800, Avg. Loss: 0.020235908031463624\n",
      "Epoch: 942, Batch: 900, Avg. Loss: 0.01780146114528179\n",
      "Epoch: 943, Batch: 100, Avg. Loss: 0.021671739518642427\n",
      "Epoch: 943, Batch: 200, Avg. Loss: 0.02189657486975193\n",
      "Epoch: 943, Batch: 300, Avg. Loss: 0.016843928769230843\n",
      "Epoch: 943, Batch: 400, Avg. Loss: 0.014187714904546738\n",
      "Epoch: 943, Batch: 500, Avg. Loss: 0.016801535934209823\n",
      "Epoch: 943, Batch: 600, Avg. Loss: 0.022281334698200227\n",
      "Epoch: 943, Batch: 700, Avg. Loss: 0.02127546951174736\n",
      "Epoch: 943, Batch: 800, Avg. Loss: 0.019206782579422\n",
      "Epoch: 943, Batch: 900, Avg. Loss: 0.017236119061708452\n",
      "Epoch: 944, Batch: 100, Avg. Loss: 0.020053674206137656\n",
      "Epoch: 944, Batch: 200, Avg. Loss: 0.01921821989119053\n",
      "Epoch: 944, Batch: 300, Avg. Loss: 0.019560373947024347\n",
      "Epoch: 944, Batch: 400, Avg. Loss: 0.018619750514626502\n",
      "Epoch: 944, Batch: 500, Avg. Loss: 0.015829293057322502\n",
      "Epoch: 944, Batch: 600, Avg. Loss: 0.026111396104097365\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 944, Batch: 700, Avg. Loss: 0.01865643210709095\n",
      "Epoch: 944, Batch: 800, Avg. Loss: 0.019258482754230498\n",
      "Epoch: 944, Batch: 900, Avg. Loss: 0.020712050944566726\n",
      "Epoch: 945, Batch: 100, Avg. Loss: 0.018409117087721825\n",
      "Epoch: 945, Batch: 200, Avg. Loss: 0.01981715828180313\n",
      "Epoch: 945, Batch: 300, Avg. Loss: 0.017318799421191217\n",
      "Epoch: 945, Batch: 400, Avg. Loss: 0.019745902419090272\n",
      "Epoch: 945, Batch: 500, Avg. Loss: 0.019676816537976264\n",
      "Epoch: 945, Batch: 600, Avg. Loss: 0.019186920151114464\n",
      "Epoch: 945, Batch: 700, Avg. Loss: 0.020481125563383103\n",
      "Epoch: 945, Batch: 800, Avg. Loss: 0.01640361487865448\n",
      "Epoch: 945, Batch: 900, Avg. Loss: 0.02428879849612713\n",
      "Epoch: 946, Batch: 100, Avg. Loss: 0.018738025799393654\n",
      "Epoch: 946, Batch: 200, Avg. Loss: 0.017889830023050307\n",
      "Epoch: 946, Batch: 300, Avg. Loss: 0.019787745401263237\n",
      "Epoch: 946, Batch: 400, Avg. Loss: 0.01894675500690937\n",
      "Epoch: 946, Batch: 500, Avg. Loss: 0.018279389813542365\n",
      "Epoch: 946, Batch: 600, Avg. Loss: 0.018519540801644325\n",
      "Epoch: 946, Batch: 700, Avg. Loss: 0.017105231285095213\n",
      "Epoch: 946, Batch: 800, Avg. Loss: 0.01725102126598358\n",
      "Epoch: 946, Batch: 900, Avg. Loss: 0.01987241320312023\n",
      "Epoch: 947, Batch: 100, Avg. Loss: 0.02044025108218193\n",
      "Epoch: 947, Batch: 200, Avg. Loss: 0.021734990626573563\n",
      "Epoch: 947, Batch: 300, Avg. Loss: 0.016318723261356353\n",
      "Epoch: 947, Batch: 400, Avg. Loss: 0.018541096523404122\n",
      "Epoch: 947, Batch: 500, Avg. Loss: 0.01959207661449909\n",
      "Epoch: 947, Batch: 600, Avg. Loss: 0.018079885989427568\n",
      "Epoch: 947, Batch: 700, Avg. Loss: 0.017843567579984666\n",
      "Epoch: 947, Batch: 800, Avg. Loss: 0.022793853580951692\n",
      "Epoch: 947, Batch: 900, Avg. Loss: 0.021839636117219924\n",
      "Epoch: 948, Batch: 100, Avg. Loss: 0.01860363334417343\n",
      "Epoch: 948, Batch: 200, Avg. Loss: 0.01871232010424137\n",
      "Epoch: 948, Batch: 300, Avg. Loss: 0.016631036102771758\n",
      "Epoch: 948, Batch: 400, Avg. Loss: 0.015504354536533356\n",
      "Epoch: 948, Batch: 500, Avg. Loss: 0.018376059606671335\n",
      "Epoch: 948, Batch: 600, Avg. Loss: 0.017554691657423974\n",
      "Epoch: 948, Batch: 700, Avg. Loss: 0.017401315495371817\n",
      "Epoch: 948, Batch: 800, Avg. Loss: 0.024416805133223533\n",
      "Epoch: 948, Batch: 900, Avg. Loss: 0.014160517677664757\n",
      "Epoch: 949, Batch: 100, Avg. Loss: 0.01919645644724369\n",
      "Epoch: 949, Batch: 200, Avg. Loss: 0.02124137319624424\n",
      "Epoch: 949, Batch: 300, Avg. Loss: 0.017692066207528116\n",
      "Epoch: 949, Batch: 400, Avg. Loss: 0.025093551576137543\n",
      "Epoch: 949, Batch: 500, Avg. Loss: 0.022503597512841225\n",
      "Epoch: 949, Batch: 600, Avg. Loss: 0.017272157594561577\n",
      "Epoch: 949, Batch: 700, Avg. Loss: 0.018774900883436203\n",
      "Epoch: 949, Batch: 800, Avg. Loss: 0.014732850790023804\n",
      "Epoch: 949, Batch: 900, Avg. Loss: 0.02038650281727314\n",
      "Epoch: 950, Batch: 100, Avg. Loss: 0.023400758877396583\n",
      "Epoch: 950, Batch: 200, Avg. Loss: 0.019774306043982507\n",
      "Epoch: 950, Batch: 300, Avg. Loss: 0.02588819555938244\n",
      "Epoch: 950, Batch: 400, Avg. Loss: 0.02207210749387741\n",
      "Epoch: 950, Batch: 500, Avg. Loss: 0.01736852988600731\n",
      "Epoch: 950, Batch: 600, Avg. Loss: 0.023928517028689384\n",
      "Epoch: 950, Batch: 700, Avg. Loss: 0.016027419045567513\n",
      "Epoch: 950, Batch: 800, Avg. Loss: 0.018696266561746597\n",
      "Epoch: 950, Batch: 900, Avg. Loss: 0.016606359630823134\n",
      "Epoch: 951, Batch: 100, Avg. Loss: 0.023585931658744813\n",
      "Epoch: 951, Batch: 200, Avg. Loss: 0.018906288594007493\n",
      "Epoch: 951, Batch: 300, Avg. Loss: 0.01849122956395149\n",
      "Epoch: 951, Batch: 400, Avg. Loss: 0.014488125294446945\n",
      "Epoch: 951, Batch: 500, Avg. Loss: 0.018747887685894967\n",
      "Epoch: 951, Batch: 600, Avg. Loss: 0.019753958731889724\n",
      "Epoch: 951, Batch: 700, Avg. Loss: 0.01784982331097126\n",
      "Epoch: 951, Batch: 800, Avg. Loss: 0.014291985780000687\n",
      "Epoch: 951, Batch: 900, Avg. Loss: 0.01990462064743042\n",
      "Epoch: 952, Batch: 100, Avg. Loss: 0.022850177735090255\n",
      "Epoch: 952, Batch: 200, Avg. Loss: 0.01946709804236889\n",
      "Epoch: 952, Batch: 300, Avg. Loss: 0.01619929477572441\n",
      "Epoch: 952, Batch: 400, Avg. Loss: 0.022880623638629912\n",
      "Epoch: 952, Batch: 500, Avg. Loss: 0.022383570447564127\n",
      "Epoch: 952, Batch: 600, Avg. Loss: 0.02464648850262165\n",
      "Epoch: 952, Batch: 700, Avg. Loss: 0.017952514439821245\n",
      "Epoch: 952, Batch: 800, Avg. Loss: 0.020909374728798866\n",
      "Epoch: 952, Batch: 900, Avg. Loss: 0.02209279529750347\n",
      "Epoch: 953, Batch: 100, Avg. Loss: 0.019826002568006516\n",
      "Epoch: 953, Batch: 200, Avg. Loss: 0.021206847727298736\n",
      "Epoch: 953, Batch: 300, Avg. Loss: 0.018910378813743592\n",
      "Epoch: 953, Batch: 400, Avg. Loss: 0.01749472863972187\n",
      "Epoch: 953, Batch: 500, Avg. Loss: 0.016052917167544364\n",
      "Epoch: 953, Batch: 600, Avg. Loss: 0.018520442843437193\n",
      "Epoch: 953, Batch: 700, Avg. Loss: 0.017156883701682092\n",
      "Epoch: 953, Batch: 800, Avg. Loss: 0.020836033299565314\n",
      "Epoch: 953, Batch: 900, Avg. Loss: 0.013093575239181518\n",
      "Epoch: 954, Batch: 100, Avg. Loss: 0.016144018843770026\n",
      "Epoch: 954, Batch: 200, Avg. Loss: 0.015027190148830414\n",
      "Epoch: 954, Batch: 300, Avg. Loss: 0.019167251735925674\n",
      "Epoch: 954, Batch: 400, Avg. Loss: 0.0156511253118515\n",
      "Epoch: 954, Batch: 500, Avg. Loss: 0.0182949198782444\n",
      "Epoch: 954, Batch: 600, Avg. Loss: 0.020484518855810166\n",
      "Epoch: 954, Batch: 700, Avg. Loss: 0.017573984041810035\n",
      "Epoch: 954, Batch: 800, Avg. Loss: 0.02058522753417492\n",
      "Epoch: 954, Batch: 900, Avg. Loss: 0.018922256156802177\n",
      "Epoch: 955, Batch: 100, Avg. Loss: 0.018319099172949792\n",
      "Epoch: 955, Batch: 200, Avg. Loss: 0.018246178552508353\n",
      "Epoch: 955, Batch: 300, Avg. Loss: 0.019840999245643615\n",
      "Epoch: 955, Batch: 400, Avg. Loss: 0.016767381876707076\n",
      "Epoch: 955, Batch: 500, Avg. Loss: 0.021345838159322738\n",
      "Epoch: 955, Batch: 600, Avg. Loss: 0.01865643784403801\n",
      "Epoch: 955, Batch: 700, Avg. Loss: 0.017615966200828552\n",
      "Epoch: 955, Batch: 800, Avg. Loss: 0.025652417689561845\n",
      "Epoch: 955, Batch: 900, Avg. Loss: 0.02048251383006573\n",
      "Epoch: 956, Batch: 100, Avg. Loss: 0.013896168991923332\n",
      "Epoch: 956, Batch: 200, Avg. Loss: 0.02490239642560482\n",
      "Epoch: 956, Batch: 300, Avg. Loss: 0.017785888016223907\n",
      "Epoch: 956, Batch: 400, Avg. Loss: 0.023191768154501916\n",
      "Epoch: 956, Batch: 500, Avg. Loss: 0.018309259861707686\n",
      "Epoch: 956, Batch: 600, Avg. Loss: 0.018893709853291512\n",
      "Epoch: 956, Batch: 700, Avg. Loss: 0.015766895115375518\n",
      "Epoch: 956, Batch: 800, Avg. Loss: 0.020931746140122415\n",
      "Epoch: 956, Batch: 900, Avg. Loss: 0.016737127453088762\n",
      "Epoch: 957, Batch: 100, Avg. Loss: 0.02100224085152149\n",
      "Epoch: 957, Batch: 200, Avg. Loss: 0.02180262692272663\n",
      "Epoch: 957, Batch: 300, Avg. Loss: 0.015366728901863098\n",
      "Epoch: 957, Batch: 400, Avg. Loss: 0.01592811070382595\n",
      "Epoch: 957, Batch: 500, Avg. Loss: 0.022430281043052672\n",
      "Epoch: 957, Batch: 600, Avg. Loss: 0.016667936220765112\n",
      "Epoch: 957, Batch: 700, Avg. Loss: 0.018143261149525643\n",
      "Epoch: 957, Batch: 800, Avg. Loss: 0.015559056773781776\n",
      "Epoch: 957, Batch: 900, Avg. Loss: 0.021020200550556183\n",
      "Epoch: 958, Batch: 100, Avg. Loss: 0.023652521073818208\n",
      "Epoch: 958, Batch: 200, Avg. Loss: 0.020308721140027047\n",
      "Epoch: 958, Batch: 300, Avg. Loss: 0.022319322675466536\n",
      "Epoch: 958, Batch: 400, Avg. Loss: 0.021160996928811075\n",
      "Epoch: 958, Batch: 500, Avg. Loss: 0.022173208072781562\n",
      "Epoch: 958, Batch: 600, Avg. Loss: 0.01739131338894367\n",
      "Epoch: 958, Batch: 700, Avg. Loss: 0.022719528824090958\n",
      "Epoch: 958, Batch: 800, Avg. Loss: 0.018252439200878143\n",
      "Epoch: 958, Batch: 900, Avg. Loss: 0.017718560248613357\n",
      "Epoch: 959, Batch: 100, Avg. Loss: 0.018958204314112662\n",
      "Epoch: 959, Batch: 200, Avg. Loss: 0.017449781969189643\n",
      "Epoch: 959, Batch: 300, Avg. Loss: 0.014856637716293334\n",
      "Epoch: 959, Batch: 400, Avg. Loss: 0.016809071376919747\n",
      "Epoch: 959, Batch: 500, Avg. Loss: 0.017014999166131018\n",
      "Epoch: 959, Batch: 600, Avg. Loss: 0.0235930784791708\n",
      "Epoch: 959, Batch: 700, Avg. Loss: 0.021660499572753907\n",
      "Epoch: 959, Batch: 800, Avg. Loss: 0.024313420802354813\n",
      "Epoch: 959, Batch: 900, Avg. Loss: 0.018255109041929243\n",
      "Epoch: 960, Batch: 100, Avg. Loss: 0.01955851346254349\n",
      "Epoch: 960, Batch: 200, Avg. Loss: 0.01690445676445961\n",
      "Epoch: 960, Batch: 300, Avg. Loss: 0.020702672228217124\n",
      "Epoch: 960, Batch: 400, Avg. Loss: 0.019834757447242737\n",
      "Epoch: 960, Batch: 500, Avg. Loss: 0.01709196634590626\n",
      "Epoch: 960, Batch: 600, Avg. Loss: 0.019534580260515213\n",
      "Epoch: 960, Batch: 700, Avg. Loss: 0.017300484851002693\n",
      "Epoch: 960, Batch: 800, Avg. Loss: 0.019881154373288155\n",
      "Epoch: 960, Batch: 900, Avg. Loss: 0.021125528514385223\n",
      "Epoch: 961, Batch: 100, Avg. Loss: 0.01613410286605358\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 961, Batch: 200, Avg. Loss: 0.014650461748242378\n",
      "Epoch: 961, Batch: 300, Avg. Loss: 0.021100963801145553\n",
      "Epoch: 961, Batch: 400, Avg. Loss: 0.018451673835515978\n",
      "Epoch: 961, Batch: 500, Avg. Loss: 0.022087303772568703\n",
      "Epoch: 961, Batch: 600, Avg. Loss: 0.021085231602191924\n",
      "Epoch: 961, Batch: 700, Avg. Loss: 0.01929569013416767\n",
      "Epoch: 961, Batch: 800, Avg. Loss: 0.023522054851055144\n",
      "Epoch: 961, Batch: 900, Avg. Loss: 0.02017075605690479\n",
      "Epoch: 962, Batch: 100, Avg. Loss: 0.020827485769987105\n",
      "Epoch: 962, Batch: 200, Avg. Loss: 0.018500097393989563\n",
      "Epoch: 962, Batch: 300, Avg. Loss: 0.017174892798066138\n",
      "Epoch: 962, Batch: 400, Avg. Loss: 0.02071679286658764\n",
      "Epoch: 962, Batch: 500, Avg. Loss: 0.020903636366128922\n",
      "Epoch: 962, Batch: 600, Avg. Loss: 0.021227344498038293\n",
      "Epoch: 962, Batch: 700, Avg. Loss: 0.020576022416353226\n",
      "Epoch: 962, Batch: 800, Avg. Loss: 0.021330427527427673\n",
      "Epoch: 962, Batch: 900, Avg. Loss: 0.02287614732980728\n",
      "Epoch: 963, Batch: 100, Avg. Loss: 0.021057844534516333\n",
      "Epoch: 963, Batch: 200, Avg. Loss: 0.018085587322711944\n",
      "Epoch: 963, Batch: 300, Avg. Loss: 0.01772640384733677\n",
      "Epoch: 963, Batch: 400, Avg. Loss: 0.020872900262475014\n",
      "Epoch: 963, Batch: 500, Avg. Loss: 0.016842442005872725\n",
      "Epoch: 963, Batch: 600, Avg. Loss: 0.024926865696907042\n",
      "Epoch: 963, Batch: 700, Avg. Loss: 0.01686030440032482\n",
      "Epoch: 963, Batch: 800, Avg. Loss: 0.019729399159550666\n",
      "Epoch: 963, Batch: 900, Avg. Loss: 0.020039269849658013\n",
      "Epoch: 964, Batch: 100, Avg. Loss: 0.0170548839867115\n",
      "Epoch: 964, Batch: 200, Avg. Loss: 0.01595612220466137\n",
      "Epoch: 964, Batch: 300, Avg. Loss: 0.018583037853240968\n",
      "Epoch: 964, Batch: 400, Avg. Loss: 0.017750061228871346\n",
      "Epoch: 964, Batch: 500, Avg. Loss: 0.0207076670974493\n",
      "Epoch: 964, Batch: 600, Avg. Loss: 0.014545349553227425\n",
      "Epoch: 964, Batch: 700, Avg. Loss: 0.014486254379153251\n",
      "Epoch: 964, Batch: 800, Avg. Loss: 0.02745055139064789\n",
      "Epoch: 964, Batch: 900, Avg. Loss: 0.01855172961950302\n",
      "Epoch: 965, Batch: 100, Avg. Loss: 0.020849146917462348\n",
      "Epoch: 965, Batch: 200, Avg. Loss: 0.019033476486802103\n",
      "Epoch: 965, Batch: 300, Avg. Loss: 0.017698163539171217\n",
      "Epoch: 965, Batch: 400, Avg. Loss: 0.01659314699470997\n",
      "Epoch: 965, Batch: 500, Avg. Loss: 0.0151084403693676\n",
      "Epoch: 965, Batch: 600, Avg. Loss: 0.02227805748581886\n",
      "Epoch: 965, Batch: 700, Avg. Loss: 0.019424950182437898\n",
      "Epoch: 965, Batch: 800, Avg. Loss: 0.01522398941218853\n",
      "Epoch: 965, Batch: 900, Avg. Loss: 0.016497868523001672\n",
      "Epoch: 966, Batch: 100, Avg. Loss: 0.017946555763483047\n",
      "Epoch: 966, Batch: 200, Avg. Loss: 0.0205443761497736\n",
      "Epoch: 966, Batch: 300, Avg. Loss: 0.017239001989364625\n",
      "Epoch: 966, Batch: 400, Avg. Loss: 0.018859352096915246\n",
      "Epoch: 966, Batch: 500, Avg. Loss: 0.020572046637535094\n",
      "Epoch: 966, Batch: 600, Avg. Loss: 0.01718398556113243\n",
      "Epoch: 966, Batch: 700, Avg. Loss: 0.014750480577349663\n",
      "Epoch: 966, Batch: 800, Avg. Loss: 0.016880433559417724\n",
      "Epoch: 966, Batch: 900, Avg. Loss: 0.019890106320381164\n",
      "Epoch: 967, Batch: 100, Avg. Loss: 0.015369462817907333\n",
      "Epoch: 967, Batch: 200, Avg. Loss: 0.015288601517677308\n",
      "Epoch: 967, Batch: 300, Avg. Loss: 0.022222820222377777\n",
      "Epoch: 967, Batch: 400, Avg. Loss: 0.020224051326513292\n",
      "Epoch: 967, Batch: 500, Avg. Loss: 0.015536667704582214\n",
      "Epoch: 967, Batch: 600, Avg. Loss: 0.02076542377471924\n",
      "Epoch: 967, Batch: 700, Avg. Loss: 0.024492133110761642\n",
      "Epoch: 967, Batch: 800, Avg. Loss: 0.015033155009150504\n",
      "Epoch: 967, Batch: 900, Avg. Loss: 0.02262586072087288\n",
      "Epoch: 968, Batch: 100, Avg. Loss: 0.021779253780841827\n",
      "Epoch: 968, Batch: 200, Avg. Loss: 0.01756280355155468\n",
      "Epoch: 968, Batch: 300, Avg. Loss: 0.01886009119451046\n",
      "Epoch: 968, Batch: 400, Avg. Loss: 0.01514623560011387\n",
      "Epoch: 968, Batch: 500, Avg. Loss: 0.02049375355243683\n",
      "Epoch: 968, Batch: 600, Avg. Loss: 0.016194086372852325\n",
      "Epoch: 968, Batch: 700, Avg. Loss: 0.019234470650553705\n",
      "Epoch: 968, Batch: 800, Avg. Loss: 0.02374896451830864\n",
      "Epoch: 968, Batch: 900, Avg. Loss: 0.018090077117085455\n",
      "Epoch: 969, Batch: 100, Avg. Loss: 0.01685049258172512\n",
      "Epoch: 969, Batch: 200, Avg. Loss: 0.02547152906656265\n",
      "Epoch: 969, Batch: 300, Avg. Loss: 0.023289235830307006\n",
      "Epoch: 969, Batch: 400, Avg. Loss: 0.021420604288578032\n",
      "Epoch: 969, Batch: 500, Avg. Loss: 0.016208467110991477\n",
      "Epoch: 969, Batch: 600, Avg. Loss: 0.018786024823784827\n",
      "Epoch: 969, Batch: 700, Avg. Loss: 0.01791569985449314\n",
      "Epoch: 969, Batch: 800, Avg. Loss: 0.01550091803073883\n",
      "Epoch: 969, Batch: 900, Avg. Loss: 0.021292134001851083\n",
      "Epoch: 970, Batch: 100, Avg. Loss: 0.016656623408198356\n",
      "Epoch: 970, Batch: 200, Avg. Loss: 0.020325143188238144\n",
      "Epoch: 970, Batch: 300, Avg. Loss: 0.018096599727869034\n",
      "Epoch: 970, Batch: 400, Avg. Loss: 0.019942094534635545\n",
      "Epoch: 970, Batch: 500, Avg. Loss: 0.014874729067087174\n",
      "Epoch: 970, Batch: 600, Avg. Loss: 0.016300218477845192\n",
      "Epoch: 970, Batch: 700, Avg. Loss: 0.014598047658801079\n",
      "Epoch: 970, Batch: 800, Avg. Loss: 0.02138190597295761\n",
      "Epoch: 970, Batch: 900, Avg. Loss: 0.01798106133937836\n",
      "Epoch: 971, Batch: 100, Avg. Loss: 0.021471655145287514\n",
      "Epoch: 971, Batch: 200, Avg. Loss: 0.02216554567217827\n",
      "Epoch: 971, Batch: 300, Avg. Loss: 0.018549719452857973\n",
      "Epoch: 971, Batch: 400, Avg. Loss: 0.01610650636255741\n",
      "Epoch: 971, Batch: 500, Avg. Loss: 0.01700827419757843\n",
      "Epoch: 971, Batch: 600, Avg. Loss: 0.01924110971391201\n",
      "Epoch: 971, Batch: 700, Avg. Loss: 0.016311956346035005\n",
      "Epoch: 971, Batch: 800, Avg. Loss: 0.0241293191164732\n",
      "Epoch: 971, Batch: 900, Avg. Loss: 0.019459222629666328\n",
      "Epoch: 972, Batch: 100, Avg. Loss: 0.02265521854162216\n",
      "Epoch: 972, Batch: 200, Avg. Loss: 0.020847900360822676\n",
      "Epoch: 972, Batch: 300, Avg. Loss: 0.018435443937778472\n",
      "Epoch: 972, Batch: 400, Avg. Loss: 0.020747105404734612\n",
      "Epoch: 972, Batch: 500, Avg. Loss: 0.019175353795289993\n",
      "Epoch: 972, Batch: 600, Avg. Loss: 0.018885691836476325\n",
      "Epoch: 972, Batch: 700, Avg. Loss: 0.016099560409784316\n",
      "Epoch: 972, Batch: 800, Avg. Loss: 0.014488974958658219\n",
      "Epoch: 972, Batch: 900, Avg. Loss: 0.01956014670431614\n",
      "Epoch: 973, Batch: 100, Avg. Loss: 0.014411008879542352\n",
      "Epoch: 973, Batch: 200, Avg. Loss: 0.01853687286376953\n",
      "Epoch: 973, Batch: 300, Avg. Loss: 0.017287416905164717\n",
      "Epoch: 973, Batch: 400, Avg. Loss: 0.015260752812027932\n",
      "Epoch: 973, Batch: 500, Avg. Loss: 0.026632238104939462\n",
      "Epoch: 973, Batch: 600, Avg. Loss: 0.020191276744008065\n",
      "Epoch: 973, Batch: 700, Avg. Loss: 0.01797335810959339\n",
      "Epoch: 973, Batch: 800, Avg. Loss: 0.02102088391780853\n",
      "Epoch: 973, Batch: 900, Avg. Loss: 0.018129622116684913\n",
      "Epoch: 974, Batch: 100, Avg. Loss: 0.02022799551486969\n",
      "Epoch: 974, Batch: 200, Avg. Loss: 0.019527307972311975\n",
      "Epoch: 974, Batch: 300, Avg. Loss: 0.01389931820333004\n",
      "Epoch: 974, Batch: 400, Avg. Loss: 0.015458853170275687\n",
      "Epoch: 974, Batch: 500, Avg. Loss: 0.018088507503271102\n",
      "Epoch: 974, Batch: 600, Avg. Loss: 0.01671181097626686\n",
      "Epoch: 974, Batch: 700, Avg. Loss: 0.015462894812226296\n",
      "Epoch: 974, Batch: 800, Avg. Loss: 0.019309485852718352\n",
      "Epoch: 974, Batch: 900, Avg. Loss: 0.0163009212911129\n",
      "Epoch: 975, Batch: 100, Avg. Loss: 0.021259819567203523\n",
      "Epoch: 975, Batch: 200, Avg. Loss: 0.02121258556842804\n",
      "Epoch: 975, Batch: 300, Avg. Loss: 0.024688755348324775\n",
      "Epoch: 975, Batch: 400, Avg. Loss: 0.01790699116885662\n",
      "Epoch: 975, Batch: 500, Avg. Loss: 0.015773887634277343\n",
      "Epoch: 975, Batch: 600, Avg. Loss: 0.015587541386485099\n",
      "Epoch: 975, Batch: 700, Avg. Loss: 0.015783865600824357\n",
      "Epoch: 975, Batch: 800, Avg. Loss: 0.022790522426366807\n",
      "Epoch: 975, Batch: 900, Avg. Loss: 0.02207262009382248\n",
      "Epoch: 976, Batch: 100, Avg. Loss: 0.021014263331890107\n",
      "Epoch: 976, Batch: 200, Avg. Loss: 0.020866046100854872\n",
      "Epoch: 976, Batch: 300, Avg. Loss: 0.016473389938473703\n",
      "Epoch: 976, Batch: 400, Avg. Loss: 0.01641899175941944\n",
      "Epoch: 976, Batch: 500, Avg. Loss: 0.020589131191372873\n",
      "Epoch: 976, Batch: 600, Avg. Loss: 0.01978422187268734\n",
      "Epoch: 976, Batch: 700, Avg. Loss: 0.01991798147559166\n",
      "Epoch: 976, Batch: 800, Avg. Loss: 0.015951390415430068\n",
      "Epoch: 976, Batch: 900, Avg. Loss: 0.02223202109336853\n",
      "Epoch: 977, Batch: 100, Avg. Loss: 0.01647478148341179\n",
      "Epoch: 977, Batch: 200, Avg. Loss: 0.016806254610419274\n",
      "Epoch: 977, Batch: 300, Avg. Loss: 0.02159273348748684\n",
      "Epoch: 977, Batch: 400, Avg. Loss: 0.013551560714840889\n",
      "Epoch: 977, Batch: 500, Avg. Loss: 0.015706917345523833\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 977, Batch: 600, Avg. Loss: 0.02188482791185379\n",
      "Epoch: 977, Batch: 700, Avg. Loss: 0.021635554805397986\n",
      "Epoch: 977, Batch: 800, Avg. Loss: 0.024476888105273247\n",
      "Epoch: 977, Batch: 900, Avg. Loss: 0.018096009492874144\n",
      "Epoch: 978, Batch: 100, Avg. Loss: 0.018278135284781456\n",
      "Epoch: 978, Batch: 200, Avg. Loss: 0.020080217719078065\n",
      "Epoch: 978, Batch: 300, Avg. Loss: 0.021304000839591025\n",
      "Epoch: 978, Batch: 400, Avg. Loss: 0.019864272102713586\n",
      "Epoch: 978, Batch: 500, Avg. Loss: 0.019786065444350243\n",
      "Epoch: 978, Batch: 600, Avg. Loss: 0.012805707454681396\n",
      "Epoch: 978, Batch: 700, Avg. Loss: 0.020028958693146706\n",
      "Epoch: 978, Batch: 800, Avg. Loss: 0.017842644527554512\n",
      "Epoch: 978, Batch: 900, Avg. Loss: 0.021212111786007883\n",
      "Epoch: 979, Batch: 100, Avg. Loss: 0.02153428107500076\n",
      "Epoch: 979, Batch: 200, Avg. Loss: 0.017606933191418647\n",
      "Epoch: 979, Batch: 300, Avg. Loss: 0.020439085140824317\n",
      "Epoch: 979, Batch: 400, Avg. Loss: 0.021779038682579993\n",
      "Epoch: 979, Batch: 500, Avg. Loss: 0.0158326318860054\n",
      "Epoch: 979, Batch: 600, Avg. Loss: 0.01457848310470581\n",
      "Epoch: 979, Batch: 700, Avg. Loss: 0.019799689278006554\n",
      "Epoch: 979, Batch: 800, Avg. Loss: 0.011322500854730607\n",
      "Epoch: 979, Batch: 900, Avg. Loss: 0.018870936036109925\n",
      "Epoch: 980, Batch: 100, Avg. Loss: 0.020761550441384315\n",
      "Epoch: 980, Batch: 200, Avg. Loss: 0.018730934336781503\n",
      "Epoch: 980, Batch: 300, Avg. Loss: 0.017930009067058564\n",
      "Epoch: 980, Batch: 400, Avg. Loss: 0.01920504465699196\n",
      "Epoch: 980, Batch: 500, Avg. Loss: 0.016819012388586997\n",
      "Epoch: 980, Batch: 600, Avg. Loss: 0.01804115891456604\n",
      "Epoch: 980, Batch: 700, Avg. Loss: 0.014781981632113457\n",
      "Epoch: 980, Batch: 800, Avg. Loss: 0.017369834780693053\n",
      "Epoch: 980, Batch: 900, Avg. Loss: 0.018943629190325738\n",
      "Epoch: 981, Batch: 100, Avg. Loss: 0.018562763929367065\n",
      "Epoch: 981, Batch: 200, Avg. Loss: 0.017993801087141038\n",
      "Epoch: 981, Batch: 300, Avg. Loss: 0.02062764011323452\n",
      "Epoch: 981, Batch: 400, Avg. Loss: 0.017226234897971152\n",
      "Epoch: 981, Batch: 500, Avg. Loss: 0.0174140265583992\n",
      "Epoch: 981, Batch: 600, Avg. Loss: 0.021624391674995424\n",
      "Epoch: 981, Batch: 700, Avg. Loss: 0.017673382088541984\n",
      "Epoch: 981, Batch: 800, Avg. Loss: 0.019119147807359696\n",
      "Epoch: 981, Batch: 900, Avg. Loss: 0.017461934722959996\n",
      "Epoch: 982, Batch: 100, Avg. Loss: 0.01949323169887066\n",
      "Epoch: 982, Batch: 200, Avg. Loss: 0.017564483880996705\n",
      "Epoch: 982, Batch: 300, Avg. Loss: 0.018950964510440826\n",
      "Epoch: 982, Batch: 400, Avg. Loss: 0.01560963161289692\n",
      "Epoch: 982, Batch: 500, Avg. Loss: 0.014312903061509133\n",
      "Epoch: 982, Batch: 600, Avg. Loss: 0.014537662491202355\n",
      "Epoch: 982, Batch: 700, Avg. Loss: 0.015444322749972343\n",
      "Epoch: 982, Batch: 800, Avg. Loss: 0.019656029865145683\n",
      "Epoch: 982, Batch: 900, Avg. Loss: 0.01942267835140228\n",
      "Epoch: 983, Batch: 100, Avg. Loss: 0.019568397775292398\n",
      "Epoch: 983, Batch: 200, Avg. Loss: 0.016835185512900353\n",
      "Epoch: 983, Batch: 300, Avg. Loss: 0.01836422711610794\n",
      "Epoch: 983, Batch: 400, Avg. Loss: 0.017846177220344542\n",
      "Epoch: 983, Batch: 500, Avg. Loss: 0.015222146809101104\n",
      "Epoch: 983, Batch: 600, Avg. Loss: 0.020392772927880286\n",
      "Epoch: 983, Batch: 700, Avg. Loss: 0.02263104647397995\n",
      "Epoch: 983, Batch: 800, Avg. Loss: 0.01807823583483696\n",
      "Epoch: 983, Batch: 900, Avg. Loss: 0.013906264156103134\n",
      "Epoch: 984, Batch: 100, Avg. Loss: 0.01825884960591793\n",
      "Epoch: 984, Batch: 200, Avg. Loss: 0.015211359038949013\n",
      "Epoch: 984, Batch: 300, Avg. Loss: 0.020717324540019035\n",
      "Epoch: 984, Batch: 400, Avg. Loss: 0.017023793980479242\n",
      "Epoch: 984, Batch: 500, Avg. Loss: 0.014807131066918373\n",
      "Epoch: 984, Batch: 600, Avg. Loss: 0.019558308497071266\n",
      "Epoch: 984, Batch: 700, Avg. Loss: 0.020973905101418496\n",
      "Epoch: 984, Batch: 800, Avg. Loss: 0.015009049773216248\n",
      "Epoch: 984, Batch: 900, Avg. Loss: 0.019091774150729178\n",
      "Epoch: 985, Batch: 100, Avg. Loss: 0.015526537522673607\n",
      "Epoch: 985, Batch: 200, Avg. Loss: 0.01827666863799095\n",
      "Epoch: 985, Batch: 300, Avg. Loss: 0.018432475924491883\n",
      "Epoch: 985, Batch: 400, Avg. Loss: 0.014987836852669715\n",
      "Epoch: 985, Batch: 500, Avg. Loss: 0.017324200123548506\n",
      "Epoch: 985, Batch: 600, Avg. Loss: 0.018564159721136092\n",
      "Epoch: 985, Batch: 700, Avg. Loss: 0.021402398869395257\n",
      "Epoch: 985, Batch: 800, Avg. Loss: 0.021648779436945917\n",
      "Epoch: 985, Batch: 900, Avg. Loss: 0.01871074751019478\n",
      "Epoch: 986, Batch: 100, Avg. Loss: 0.0182548538595438\n",
      "Epoch: 986, Batch: 200, Avg. Loss: 0.018723310828208925\n",
      "Epoch: 986, Batch: 300, Avg. Loss: 0.01436494916677475\n",
      "Epoch: 986, Batch: 400, Avg. Loss: 0.02414792440831661\n",
      "Epoch: 986, Batch: 500, Avg. Loss: 0.0148891232162714\n",
      "Epoch: 986, Batch: 600, Avg. Loss: 0.01965205080807209\n",
      "Epoch: 986, Batch: 700, Avg. Loss: 0.01576821617782116\n",
      "Epoch: 986, Batch: 800, Avg. Loss: 0.018955951556563377\n",
      "Epoch: 986, Batch: 900, Avg. Loss: 0.015987652987241745\n",
      "Epoch: 987, Batch: 100, Avg. Loss: 0.02095115028321743\n",
      "Epoch: 987, Batch: 200, Avg. Loss: 0.01847204379737377\n",
      "Epoch: 987, Batch: 300, Avg. Loss: 0.01808243937790394\n",
      "Epoch: 987, Batch: 400, Avg. Loss: 0.01918328456580639\n",
      "Epoch: 987, Batch: 500, Avg. Loss: 0.019714903831481934\n",
      "Epoch: 987, Batch: 600, Avg. Loss: 0.019534406363964082\n",
      "Epoch: 987, Batch: 700, Avg. Loss: 0.019881457909941672\n",
      "Epoch: 987, Batch: 800, Avg. Loss: 0.019927028343081476\n",
      "Epoch: 987, Batch: 900, Avg. Loss: 0.019461584240198136\n",
      "Epoch: 988, Batch: 100, Avg. Loss: 0.0216411492228508\n",
      "Epoch: 988, Batch: 200, Avg. Loss: 0.015607139319181442\n",
      "Epoch: 988, Batch: 300, Avg. Loss: 0.026249927952885627\n",
      "Epoch: 988, Batch: 400, Avg. Loss: 0.01961986243724823\n",
      "Epoch: 988, Batch: 500, Avg. Loss: 0.016003170982003212\n",
      "Epoch: 988, Batch: 600, Avg. Loss: 0.017747230976819992\n",
      "Epoch: 988, Batch: 700, Avg. Loss: 0.01616001509130001\n",
      "Epoch: 988, Batch: 800, Avg. Loss: 0.017569060549139977\n",
      "Epoch: 988, Batch: 900, Avg. Loss: 0.021215601786971093\n",
      "Epoch: 989, Batch: 100, Avg. Loss: 0.01544246383011341\n",
      "Epoch: 989, Batch: 200, Avg. Loss: 0.01789149783551693\n",
      "Epoch: 989, Batch: 300, Avg. Loss: 0.02038549929857254\n",
      "Epoch: 989, Batch: 400, Avg. Loss: 0.01988627791404724\n",
      "Epoch: 989, Batch: 500, Avg. Loss: 0.023385385572910308\n",
      "Epoch: 989, Batch: 600, Avg. Loss: 0.021642664447426796\n",
      "Epoch: 989, Batch: 700, Avg. Loss: 0.024156990572810174\n",
      "Epoch: 989, Batch: 800, Avg. Loss: 0.014837819784879684\n",
      "Epoch: 989, Batch: 900, Avg. Loss: 0.017771091163158417\n",
      "Epoch: 990, Batch: 100, Avg. Loss: 0.01626918625086546\n",
      "Epoch: 990, Batch: 200, Avg. Loss: 0.015066927075386047\n",
      "Epoch: 990, Batch: 300, Avg. Loss: 0.02579905904829502\n",
      "Epoch: 990, Batch: 400, Avg. Loss: 0.020920078679919244\n",
      "Epoch: 990, Batch: 500, Avg. Loss: 0.0209251057356596\n",
      "Epoch: 990, Batch: 600, Avg. Loss: 0.02061445973813534\n",
      "Epoch: 990, Batch: 700, Avg. Loss: 0.015269876196980477\n",
      "Epoch: 990, Batch: 800, Avg. Loss: 0.01692445881664753\n",
      "Epoch: 990, Batch: 900, Avg. Loss: 0.01836573041975498\n",
      "Epoch: 991, Batch: 100, Avg. Loss: 0.017405890971422196\n",
      "Epoch: 991, Batch: 200, Avg. Loss: 0.013469283431768418\n",
      "Epoch: 991, Batch: 300, Avg. Loss: 0.014173430800437927\n",
      "Epoch: 991, Batch: 400, Avg. Loss: 0.021101192757487296\n",
      "Epoch: 991, Batch: 500, Avg. Loss: 0.016308542490005493\n",
      "Epoch: 991, Batch: 600, Avg. Loss: 0.02089592307806015\n",
      "Epoch: 991, Batch: 700, Avg. Loss: 0.014792043715715408\n",
      "Epoch: 991, Batch: 800, Avg. Loss: 0.019338604137301443\n",
      "Epoch: 991, Batch: 900, Avg. Loss: 0.017726400271058082\n",
      "Epoch: 992, Batch: 100, Avg. Loss: 0.015942167192697525\n",
      "Epoch: 992, Batch: 200, Avg. Loss: 0.021510871052742006\n",
      "Epoch: 992, Batch: 300, Avg. Loss: 0.018709241449832915\n",
      "Epoch: 992, Batch: 400, Avg. Loss: 0.022054336220026016\n",
      "Epoch: 992, Batch: 500, Avg. Loss: 0.01879699245095253\n",
      "Epoch: 992, Batch: 600, Avg. Loss: 0.02148452565073967\n",
      "Epoch: 992, Batch: 700, Avg. Loss: 0.016768838688731195\n",
      "Epoch: 992, Batch: 800, Avg. Loss: 0.0177778859436512\n",
      "Epoch: 992, Batch: 900, Avg. Loss: 0.018379166275262832\n",
      "Epoch: 993, Batch: 100, Avg. Loss: 0.016066461876034735\n",
      "Epoch: 993, Batch: 200, Avg. Loss: 0.015412019193172454\n",
      "Epoch: 993, Batch: 300, Avg. Loss: 0.020420849695801734\n",
      "Epoch: 993, Batch: 400, Avg. Loss: 0.021514841914176942\n",
      "Epoch: 993, Batch: 500, Avg. Loss: 0.013264863714575768\n",
      "Epoch: 993, Batch: 600, Avg. Loss: 0.022010405212640763\n",
      "Epoch: 993, Batch: 700, Avg. Loss: 0.01523052379488945\n",
      "Epoch: 993, Batch: 800, Avg. Loss: 0.019866676926612856\n",
      "Epoch: 993, Batch: 900, Avg. Loss: 0.02225212372839451\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 994, Batch: 100, Avg. Loss: 0.016430007442831993\n",
      "Epoch: 994, Batch: 200, Avg. Loss: 0.018818060010671614\n",
      "Epoch: 994, Batch: 300, Avg. Loss: 0.01585436090826988\n",
      "Epoch: 994, Batch: 400, Avg. Loss: 0.019191576465964318\n",
      "Epoch: 994, Batch: 500, Avg. Loss: 0.018015542179346086\n",
      "Epoch: 994, Batch: 600, Avg. Loss: 0.016467298716306686\n",
      "Epoch: 994, Batch: 700, Avg. Loss: 0.019940569177269937\n",
      "Epoch: 994, Batch: 800, Avg. Loss: 0.016545674726366998\n",
      "Epoch: 994, Batch: 900, Avg. Loss: 0.027107697948813438\n",
      "Epoch: 995, Batch: 100, Avg. Loss: 0.017542347311973572\n",
      "Epoch: 995, Batch: 200, Avg. Loss: 0.021197694316506387\n",
      "Epoch: 995, Batch: 300, Avg. Loss: 0.018295625299215315\n",
      "Epoch: 995, Batch: 400, Avg. Loss: 0.01831632561981678\n",
      "Epoch: 995, Batch: 500, Avg. Loss: 0.017908887043595313\n",
      "Epoch: 995, Batch: 600, Avg. Loss: 0.01925144501030445\n",
      "Epoch: 995, Batch: 700, Avg. Loss: 0.01481009930372238\n",
      "Epoch: 995, Batch: 800, Avg. Loss: 0.017341116070747377\n",
      "Epoch: 995, Batch: 900, Avg. Loss: 0.01919193282723427\n",
      "Epoch: 996, Batch: 100, Avg. Loss: 0.01982574485242367\n",
      "Epoch: 996, Batch: 200, Avg. Loss: 0.015761243999004363\n",
      "Epoch: 996, Batch: 300, Avg. Loss: 0.014763665944337845\n",
      "Epoch: 996, Batch: 400, Avg. Loss: 0.018018732741475106\n",
      "Epoch: 996, Batch: 500, Avg. Loss: 0.022383580431342125\n",
      "Epoch: 996, Batch: 600, Avg. Loss: 0.01671122565865517\n",
      "Epoch: 996, Batch: 700, Avg. Loss: 0.018626348078250886\n",
      "Epoch: 996, Batch: 800, Avg. Loss: 0.01905367985367775\n",
      "Epoch: 996, Batch: 900, Avg. Loss: 0.01711924433708191\n",
      "Epoch: 997, Batch: 100, Avg. Loss: 0.018392977342009544\n",
      "Epoch: 997, Batch: 200, Avg. Loss: 0.023963188156485557\n",
      "Epoch: 997, Batch: 300, Avg. Loss: 0.017066799253225327\n",
      "Epoch: 997, Batch: 400, Avg. Loss: 0.020928155928850174\n",
      "Epoch: 997, Batch: 500, Avg. Loss: 0.019846746996045113\n",
      "Epoch: 997, Batch: 600, Avg. Loss: 0.016831684336066247\n",
      "Epoch: 997, Batch: 700, Avg. Loss: 0.01829312339425087\n",
      "Epoch: 997, Batch: 800, Avg. Loss: 0.017866581156849862\n",
      "Epoch: 997, Batch: 900, Avg. Loss: 0.017977014258503912\n",
      "Epoch: 998, Batch: 100, Avg. Loss: 0.023904751464724542\n",
      "Epoch: 998, Batch: 200, Avg. Loss: 0.01937735877931118\n",
      "Epoch: 998, Batch: 300, Avg. Loss: 0.01668604403734207\n",
      "Epoch: 998, Batch: 400, Avg. Loss: 0.023485462069511413\n",
      "Epoch: 998, Batch: 500, Avg. Loss: 0.022933509349822998\n",
      "Epoch: 998, Batch: 600, Avg. Loss: 0.018398312702775\n",
      "Epoch: 998, Batch: 700, Avg. Loss: 0.014985188022255897\n",
      "Epoch: 998, Batch: 800, Avg. Loss: 0.018691561073064803\n",
      "Epoch: 998, Batch: 900, Avg. Loss: 0.016056624986231328\n",
      "Epoch: 999, Batch: 100, Avg. Loss: 0.02086560659110546\n",
      "Epoch: 999, Batch: 200, Avg. Loss: 0.014814963266253471\n",
      "Epoch: 999, Batch: 300, Avg. Loss: 0.01878454618155956\n",
      "Epoch: 999, Batch: 400, Avg. Loss: 0.019969548434019088\n",
      "Epoch: 999, Batch: 500, Avg. Loss: 0.015435951575636864\n",
      "Epoch: 999, Batch: 600, Avg. Loss: 0.015539233386516572\n",
      "Epoch: 999, Batch: 700, Avg. Loss: 0.01970336139202118\n",
      "Epoch: 999, Batch: 800, Avg. Loss: 0.01765830397605896\n",
      "Epoch: 999, Batch: 900, Avg. Loss: 0.0188543813675642\n",
      "Epoch: 1000, Batch: 100, Avg. Loss: 0.016927150413393974\n",
      "Epoch: 1000, Batch: 200, Avg. Loss: 0.01600743316113949\n",
      "Epoch: 1000, Batch: 300, Avg. Loss: 0.018590649515390398\n",
      "Epoch: 1000, Batch: 400, Avg. Loss: 0.015354905650019646\n",
      "Epoch: 1000, Batch: 500, Avg. Loss: 0.021694218963384627\n",
      "Epoch: 1000, Batch: 600, Avg. Loss: 0.01634027473628521\n",
      "Epoch: 1000, Batch: 700, Avg. Loss: 0.020157045871019363\n",
      "Epoch: 1000, Batch: 800, Avg. Loss: 0.018040825426578522\n",
      "Epoch: 1000, Batch: 900, Avg. Loss: 0.025583872348070146\n",
      "Finished Training\n"
     ]
    }
   ],
   "source": [
    "# define the number of epochs to train for\n",
    "n_epochs = 1000 # start small to see if your model works, initially\n",
    "\n",
    "#net.load_state_dict(torch.load('saved_models/<name>.pt'))\n",
    "\n",
    "# call train and record the loss over time\n",
    "training_loss = train(n_epochs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize loss\n",
    "\n",
    "This shows the loss changing over time and can help visualize when the loss flattens."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualize the loss as the network trained\n",
    "plt.plot(training_loss)\n",
    "plt.rc('xtick', labelsize=12) \n",
    "plt.rc('ytick', labelsize=12)\n",
    "#plt.rc('xlabel') \n",
    "#plt.rc('ylabel', fontsize=12)\n",
    "plt.xlabel('100\\'s of batches', fontsize=12)\n",
    "plt.ylabel('loss', fontsize=12)\n",
    "plt.ylim(0, 2.5) # consistent scale\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test loss and accuracy\n",
    "\n",
    "This calculates the loss for each batch to find the total loss. If this number is significantly worse than the training data, it is a sign of overfitting. \n",
    "\n",
    "It also finds the accuracy by keeping track of the number of correct labels for each class of data to find each class's accuracy and the total accuracy. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Loss: 0.018882\n",
      "\n",
      "Test Accuracy of  zero: 99% (978/980)\n",
      "Test Accuracy of   one: 99% (1133/1135)\n",
      "Test Accuracy of   two: 99% (1028/1032)\n",
      "Test Accuracy of three: 99% (1004/1010)\n",
      "Test Accuracy of  four: 99% (979/982)\n",
      "Test Accuracy of  five: 99% (887/892)\n",
      "Test Accuracy of   six: 98% (945/958)\n",
      "Test Accuracy of seven: 99% (1023/1028)\n",
      "Test Accuracy of eight: 99% (969/974)\n",
      "Test Accuracy of  nine: 98% (997/1009)\n",
      "\n",
      "Test Accuracy (Overall): 99% (9943/10000)\n"
     ]
    }
   ],
   "source": [
    "# initialize tensor and lists to monitor test loss and accuracy\n",
    "test_loss = torch.zeros(1).cuda()\n",
    "class_correct = list(0. for i in range(len(classes)))\n",
    "class_total = list(0. for i in range(len(classes)))\n",
    "\n",
    "# set the module to evaluation mode\n",
    "# used to turn off layers that are only useful for training\n",
    "# like dropout and batch_norm\n",
    "net.eval()\n",
    "\n",
    "for batch_i, data in enumerate(test_loader):\n",
    "    \n",
    "    # get the input images and their corresponding labels\n",
    "    inputs, labels = data\n",
    "    inputs, labels = inputs.cuda(), labels.cuda()\n",
    "    \n",
    "    # forward pass to get outputs\n",
    "    outputs = net(inputs)\n",
    "\n",
    "    # calculate the loss\n",
    "    loss = criterion(outputs, labels)\n",
    "            \n",
    "    # update average test loss \n",
    "    test_loss = test_loss + ((torch.ones(1).cuda() / (batch_i + 1)) * (loss.data - test_loss))\n",
    "    \n",
    "    # get the predicted class from the maximum value in the output-list of class scores\n",
    "    _, predicted = torch.max(outputs.data, 1)\n",
    "    \n",
    "    # compare predictions to true label\n",
    "    # this creates a `correct` Tensor that holds the number of correctly classified images in a batch\n",
    "    correct = np.squeeze(predicted.eq(labels.data.view_as(predicted)))\n",
    "    \n",
    "    # calculate test accuracy for *each* object class\n",
    "    # we get the scalar value of correct items for a class, by calling `correct[i].item()`\n",
    "    for l, c in zip(labels.data, correct):\n",
    "        class_correct[l] += c.item()\n",
    "        class_total[l] += 1\n",
    "\n",
    "print('Test Loss: {:.6f}\\n'.format(test_loss.cpu().numpy()[0]))\n",
    "\n",
    "for i in range(len(classes)):\n",
    "    if class_total[i] > 0:\n",
    "        print('Test Accuracy of %5s: %2d%% (%2d/%2d)' % (\n",
    "            classes[i], 100 * class_correct[i] / class_total[i],\n",
    "            np.sum(class_correct[i]), np.sum(class_total[i])))\n",
    "    else:\n",
    "        print('Test Accuracy of %5s: N/A (no training examples)' % (classes[i]))\n",
    "\n",
    "        \n",
    "print('\\nTest Accuracy (Overall): %2d%% (%2d/%2d)' % (\n",
    "    100. * np.sum(class_correct) / np.sum(class_total),\n",
    "    np.sum(class_correct), np.sum(class_total)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize test data\n",
    "\n",
    "This cell iterates over the test dataset, loading a random batch of image/label data, using `dataiter.next()`. It then uses the trained model to infer the predicted label, and plots the batch of images and predicted labels in a `batch_size/8 x 8` grid.\n",
    "\n",
    "If the label is wrong, it is marked in red and the correct label is indicated in parentheses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x1536 with 64 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize Sample Results\n",
    "# obtain one batch of test images\n",
    "dataiter = iter(test_loader)\n",
    "images, labels = dataiter.next()\n",
    "images, labels = images.cuda(), labels.cuda()\n",
    "# get predictions\n",
    "preds = np.squeeze(net(images).data.max(1, keepdim=True)[1].cpu().numpy())\n",
    "images = images.cpu().numpy()\n",
    "\n",
    "# plot the images in the batch, along with predicted and true labels\n",
    "fig = plt.figure(figsize=(25, batch_size/3))\n",
    "for idx in np.arange(batch_size):\n",
    "    ax = fig.add_subplot(batch_size/8, 8, idx+1, xticks=[], yticks=[])\n",
    "    ax.imshow(np.squeeze(images[idx]), cmap='gray')\n",
    "    if preds[idx]==labels[idx]:\n",
    "        ax.set_title(\"{}\".format(classes[preds[idx]], classes[labels[idx]]), color=\"green\")\n",
    "    else:\n",
    "        ax.set_title(\"{} ({})\".format(classes[preds[idx]], classes[labels[idx]]), color=\"red\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save the model\n",
    "\n",
    "This cell saved the state of the model to a file that can be loaded in later. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save state_dict\n",
    "# Saving the model\n",
    "model_dir = 'saved_models/'\n",
    "model_name = 'Net2_1000.pt'\n",
    "\n",
    "# after training, save your model parameters in the dir 'saved_models'\n",
    "torch.save(net.state_dict(), model_dir+model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}