Temporal_Stream_Sampled.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "New_Data_Temporal_Stream_SG3I.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3rV0SqtlQhca"
      },
      "source": [
        "Use Colab Pro with GPU and High-RAM"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-tgqAhYHvNg2"
      },
      "source": [
        "# Check GPU"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PA580dbFbyEn",
        "outputId": "b1030282-3830-4e0b-f441-cfb11c1ff94a"
      },
      "source": [
        "gpu_info = !nvidia-smi\n",
        "gpu_info = '\\n'.join(gpu_info)\n",
        "if gpu_info.find('failed') >= 0:\n",
        "  print('Select the Runtime > \"Change runtime type\" menu to enable a GPU accelerator, ')\n",
        "  print('and then re-execute this cell.')\n",
        "else:\n",
        "  print(gpu_info)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sat Dec 12 18:50:14 2020       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 455.45.01    Driver Version: 418.67       CUDA Version: 10.1     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   45C    P0    29W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
            "|                               |                      |                 ERR! |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JOC44nKYDOn8"
      },
      "source": [
        "# Run this command once and restart the runtime"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MhZffaB6C7Z7",
        "outputId": "f9d525ac-6f08-43e6-e517-c8367cf26b92"
      },
      "source": [
        "!pip install av"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting av\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/9e/62/9a992be76f8e13ce0e3a24a838191b546805545116f9fc869bd11bd21b5f/av-8.0.2-cp36-cp36m-manylinux2010_x86_64.whl (36.9MB)\n",
            "\u001b[K     |████████████████████████████████| 36.9MB 146kB/s \n",
            "\u001b[?25hInstalling collected packages: av\n",
            "Successfully installed av-8.0.2\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iPTplTI-QZnu"
      },
      "source": [
        "# Video Frame Stacking (SG3I)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ziw00A6oYxrG"
      },
      "source": [
        "from google.colab.patches import cv2_imshow\n",
        "import cv2 as cv\n",
        "import numpy as np\n",
        "\n",
        "\n",
        "def getSG3I(videos):\n",
        "    bz = videos.size(0)\n",
        "\n",
        "    frame_list_batch = []\n",
        "\n",
        "    for b in range(bz):\n",
        "        images = videos[b]\n",
        "\n",
        "        w = images[0].size(0)\n",
        "        h = images[0].size(1)\n",
        "        num_frames = images.size(0)\n",
        "        frame_list = []\n",
        "\n",
        "        for i in range(0, num_frames):\n",
        "            frame_list.append(cv.cvtColor(images[i].numpy(), cv.COLOR_BGR2GRAY))\n",
        "\n",
        "        frame_list_batch.append(np.stack(frame_list, axis=-1))\n",
        "\n",
        "    SG3I = np.stack(frame_list_batch, axis=0)\n",
        "    #cv2_imshow(SG3I[0])\n",
        "\n",
        "    return torch.Tensor(SG3I)"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nwpGXLdtuq8i"
      },
      "source": [
        "# Mount drive\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "F6nsoL9Kuqil",
        "outputId": "fafe991f-62cb-4fea-8693-401c6189c47b"
      },
      "source": [
        "import os\n",
        "import io\n",
        "from google.colab import drive\n",
        "\n",
        "# Mount google drive\n",
        "DRIVE_MOUNT='/content/drive'\n",
        "\n",
        "drive.mount(DRIVE_MOUNT)\n",
        "\n",
        "\n",
        "# create folder to write data to\n",
        "DATA_FOLDER = os.path.join(DRIVE_MOUNT, 'Shared drives', 'CIS680 Final Project', 'data')"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Xn8iwL5jQuQw"
      },
      "source": [
        "# Load Dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RqT2Q4KEphJ4",
        "outputId": "f0f6b4d3-dc94-4ff8-8f17-23cf076961e7"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "train_size = 138\n",
        "test_size = 41\n",
        "\n",
        "train_path = DATA_FOLDER + \"/dataset_3/train/train_b\"\n",
        "test_path = DATA_FOLDER + \"/dataset_3/test/test_b\"\n",
        "\n",
        "train_loader = []\n",
        "test_loader = []\n",
        "\n",
        "for batch in range(train_size):\n",
        "    batch_path = train_path + str(batch) + \".pt\"\n",
        "    train_loader.append(torch.load(batch_path))\n",
        "    if batch % 10 == 0:\n",
        "        print(\"train: \" + str(batch) + \"/\" + str(train_size))\n",
        "\n",
        "for batch in range(test_size):\n",
        "    batch_path = test_path + str(batch) + \".pt\"\n",
        "    test_loader.append(torch.load(batch_path))\n",
        "    if batch % 10 == 0:\n",
        "        print(\"test: \" + str(batch) + \"/\" + str(test_size))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "train: 0/138\n",
            "train: 10/138\n",
            "train: 20/138\n",
            "train: 30/138\n",
            "train: 40/138\n",
            "train: 50/138\n",
            "train: 60/138\n",
            "train: 70/138\n",
            "train: 80/138\n",
            "train: 90/138\n",
            "train: 100/138\n",
            "train: 110/138\n",
            "train: 120/138\n",
            "train: 130/138\n",
            "test: 0/41\n",
            "test: 10/41\n",
            "test: 20/41\n",
            "test: 30/41\n",
            "test: 40/41\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vK7WkGhWuUUu"
      },
      "source": [
        "# Debugging Visualization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "id": "NW6nlYpXpoEk",
        "outputId": "f95cb4f6-d465-4947-ad84-56a15dd39705"
      },
      "source": [
        "import torchvision.transforms as tf\n",
        "\n",
        "\n",
        "transform = tf.RandomHorizontalFlip(p=0.5)\n",
        "\n",
        "for iter, data in enumerate(train_loader, 0):\n",
        "    videos = data[\"videos\"]\n",
        "    labels = data[\"labels\"]\n",
        "    indexes = data[\"indexes\"]\n",
        "\n",
        "    videos = transform(videos.permute(0,1,4,2,3)).permute(0,1,3,4,2)\n",
        "\n",
        "    plt.imshow(videos[0][0])\n",
        "\n",
        "    SG3I = getSG3I(videos)\n",
        "\n",
        "    if iter == 0:\n",
        "      break\n",
        "    \n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tXHuIb9JpBxE"
      },
      "source": [
        "# Temporal Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KWeRsuUXo7PE"
      },
      "source": [
        "import torch\n",
        "from torch.nn import functional as F\n",
        "from torchvision import transforms\n",
        "from torch import nn, Tensor\n",
        "import random\n",
        "import torchvision\n",
        "import torchvision.models as models\n",
        "import torch.optim as optim\n",
        "import torchvision.transforms as tf\n",
        "\n",
        "\n",
        "class TemporalStream(torch.nn.Module):\n",
        "\n",
        "    def __init__(self,\n",
        "                 device='cuda',\n",
        "                 num_classes=51,\n",
        "                 dropout_probability=0.5):\n",
        "\n",
        "        # Initialize the stream layers\n",
        "        super(TemporalStream, self).__init__()\n",
        "        self.device = device\n",
        "        self.num_classes = num_classes\n",
        "\n",
        "        # Backbone\n",
        "        self.temporal = models.resnet50(pretrained=True)\n",
        "        for param in self.temporal.parameters():\n",
        "            param.requires_grad = True\n",
        "\n",
        "        self.temporal.fc = nn.Sequential(nn.Linear(2048, 1024),\n",
        "                                nn.ReLU(),\n",
        "                                nn.Dropout(p=dropout_probability),\n",
        "                                nn.Linear(1024, self.num_classes),\n",
        "                                nn.ReLU(),\n",
        "                                nn.Dropout(p=dropout_probability),\n",
        "                                nn.Softmax())\n",
        "      \n",
        "    def forward(self, X):\n",
        "        X = self.temporal(X)\n",
        "        return X\n",
        "\n",
        "    def compute_loss(self, output, target):\n",
        "        criterion = nn.CrossEntropyLoss()\n",
        "        loss = criterion(output, target)\n",
        "        return loss\n"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2ORv0x1cpJnz"
      },
      "source": [
        "# Train & Test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8RmzH4-EpLax"
      },
      "source": [
        "def train(epoch):\n",
        "    temporal.train()\n",
        "    counter = 0\n",
        "    train_loss = 0\n",
        "    log_interval = 10\n",
        "    save_interval = 200\n",
        "\n",
        "    epoch_loss = []\n",
        "    log_int_loss = 0\n",
        "    for iter, data in enumerate(train_loader, 0):\n",
        "\n",
        "        # Unpack data\n",
        "        videos = data[\"videos\"]\n",
        "        labels = torch.tensor(data[\"labels\"])\n",
        "        indexes = data[\"indexes\"]\n",
        "\n",
        "        # Input Transforms\n",
        "        videos = videos.type(torch.FloatTensor)\n",
        "        videos = transform(videos.permute(0,1,4,2,3)).permute(0,1,3,4,2)\n",
        "\n",
        "        # Frame Stacking\n",
        "        SG3I = getSG3I(videos)\n",
        "        SG3I = SG3I.permute(0,3,1,2)\n",
        "        SG3I = SG3I.to(device)\n",
        "\n",
        "        # Output\n",
        "        optimizer.zero_grad()\n",
        "        output = temporal(SG3I)\n",
        "        output = output.to(device)\n",
        "        labels = labels.to(device)\n",
        "\n",
        "        # calculate losses\n",
        "        loss = temporal.compute_loss(output, labels)\n",
        "\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        train_loss += loss.item()\n",
        "\n",
        "        # Logging Interval\n",
        "        log_int_loss += loss.item()\n",
        "        epoch_loss.append(loss.item())\n",
        "\n",
        "        if counter == 0:\n",
        "            print('Epoch: ', epoch, ', Batch: ', iter, ', loss avg over log interval: ', log_int_loss)\n",
        "            train_loss_list.append(train_loss / batch_size)\n",
        "            train_counter.append((iter + 1) * batch_size + epoch * len(train_loader))\n",
        "            log_int_loss = 0\n",
        "        elif counter % log_interval == log_interval - 1:\n",
        "            print('Epoch: ', epoch, ', Batch: ', iter, ', loss avg over log interval: ', log_int_loss / log_interval)\n",
        "            train_loss_list.append(train_loss / batch_size)\n",
        "            train_counter.append((iter + 1) * batch_size + epoch * len(train_loader))\n",
        "            log_int_loss = 0\n",
        "\n",
        "        counter += 1\n",
        "\n",
        "    avg_loss = sum(epoch_loss) / len(epoch_loss)\n",
        "    epoch_loss_list.append(avg_loss)\n",
        "    print('Epoch: ', epoch, ', avg total loss: ', avg_loss)\n",
        "\n",
        "\n",
        "def test():\n",
        "    temporal.eval()\n",
        "    test_loss = 0\n",
        "    correct = 0\n",
        "\n",
        "    # Data Loop\n",
        "    with torch.no_grad():\n",
        "        for iter, data in enumerate(test_loader, 0):\n",
        "\n",
        "            # Unpack data\n",
        "            videos = data[\"videos\"]\n",
        "            labels = torch.tensor(data[\"labels\"])\n",
        "            indexes = data[\"indexes\"]\n",
        "\n",
        "            # Input\n",
        "            videos = videos.type(torch.FloatTensor)\n",
        "            labels = labels.to(device)\n",
        "            SG3I = getSG3I(videos)\n",
        "            SG3I = SG3I.permute(0,3,1,2)\n",
        "            SG3I = SG3I.to(device)\n",
        "\n",
        "            # Output\n",
        "            output = temporal(SG3I)\n",
        "            output = output.to(device)\n",
        "\n",
        "            # calculate losses\n",
        "            loss = temporal.compute_loss(output, labels)\n",
        "\n",
        "            test_loss += loss.item()\n",
        "\n",
        "            # calculate number of correct predictions in batch\n",
        "            correct += sum(torch.argmax(output,1) == labels).item()\n",
        "            if iter % 100 == 0:\n",
        "                print (\"iter  \", iter)\n",
        "                print(\"accuracy so far = \", correct / ((iter + 1) * len(labels)))\n",
        "    \n",
        "    # Log\n",
        "    test_loss_list.append(test_loss / len(test_loader))\n",
        "    accuracy = correct / (len(test_loader) * 128)\n",
        "    accuracy_list.append(accuracy)\n",
        "    print('Avg Validation Loss: ', test_loss / len(test_loader))\n",
        "    print('Accuracy: ', accuracy)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BiCgDsOOpRlK"
      },
      "source": [
        "# Train Loop"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "v06ikvOBpPlU",
        "outputId": "ccd01b63-e863-4e0d-c87c-350b50474937"
      },
      "source": [
        "EPOCH_SAVE_PREFIX = '/content/drive/Shared drives/CIS680 Final Project/models/temporal_new_data/'\n",
        "\n",
        "# Device\n",
        "device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')\n",
        "torch.random.manual_seed(1)\n",
        "\n",
        "# model\n",
        "learning_rate = 0.001\n",
        "temporal = TemporalStream()\n",
        "temporal.to(device)\n",
        "optimizer = optim.SGD(temporal.parameters(), lr=learning_rate, momentum=0.9)\n",
        "\n",
        "# Transforms\n",
        "transform = tf.RandomHorizontalFlip(p=0.5)\n",
        "\n",
        "# Epochs\n",
        "last_epoch = -1\n",
        "num_epochs = 20\n",
        "batch_size = 128\n",
        "\n",
        "# Logging setup: train\n",
        "train_loss_list = []\n",
        "epoch_loss_list = []\n",
        "train_counter = []\n",
        "\n",
        "# Logging setup: test\n",
        "test_loss_list = []\n",
        "accuracy_list = []\n",
        "epoch_list = np.arange(num_epochs)\n",
        "\n",
        "# epoch loop\n",
        "for epoch in range(num_epochs):\n",
        "\n",
        "    # Train & Validate\n",
        "    train(epoch)\n",
        "    test()\n",
        "\n",
        "    # Save Model Version\n",
        "    save_path = os.path.join(EPOCH_SAVE_PREFIX, 'temporal_epoch' + str(epoch))\n",
        "    torch.save({\n",
        "        'epoch': epoch,\n",
        "        'train_total_loss_list': train_loss_list,\n",
        "        'epoch_total_loss_list': epoch_loss_list,\n",
        "        'test_loss_list': test_loss_list,\n",
        "        'train_counter': train_counter,\n",
        "        'accuracy_lsit': accuracy_list,\n",
        "        'model_state_dict': temporal.state_dict(),\n",
        "        'optimizer_state_dict': optimizer.state_dict()\n",
        "    }, save_path)\n",
        "\n",
        "    print(\"Epoch %d/%d Completed\" % (epoch, num_epochs - 1))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/container.py:117: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
            "  input = module(input)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch:  0 , Batch:  0 , loss avg over log interval:  3.9317588806152344\n",
            "Epoch:  0 , Batch:  9 , loss avg over log interval:  3.5384957790374756\n",
            "Epoch:  0 , Batch:  19 , loss avg over log interval:  3.9317105531692507\n",
            "Epoch:  0 , Batch:  29 , loss avg over log interval:  3.931578540802002\n",
            "Epoch:  0 , Batch:  39 , loss avg over log interval:  3.931746768951416\n",
            "Epoch:  0 , Batch:  49 , loss avg over log interval:  3.931655025482178\n",
            "Epoch:  0 , Batch:  59 , loss avg over log interval:  3.931948113441467\n",
            "Epoch:  0 , Batch:  69 , loss avg over log interval:  3.931746220588684\n",
            "Epoch:  0 , Batch:  79 , loss avg over log interval:  3.9317609786987306\n",
            "Epoch:  0 , Batch:  89 , loss avg over log interval:  3.9317273616790773\n",
            "Epoch:  0 , Batch:  99 , loss avg over log interval:  3.9316843032836912\n",
            "Epoch:  0 , Batch:  109 , loss avg over log interval:  3.9316733360290526\n",
            "Epoch:  0 , Batch:  119 , loss avg over log interval:  3.9313134431838987\n",
            "Epoch:  0 , Batch:  129 , loss avg over log interval:  3.9312376737594605\n",
            "Epoch:  0 , avg total loss:  3.931638532790585\n",
            "iter   0\n",
            "accuracy so far =  0.859375\n",
            "Avg Validation Loss:  3.9315117452202775\n",
            "Accuracy:  6.341463414634147\n",
            "Epoch 0/19 Completed\n",
            "Epoch:  1 , Batch:  0 , loss avg over log interval:  3.9316000938415527\n",
            "Epoch:  1 , Batch:  9 , loss avg over log interval:  3.5378943920135497\n",
            "Epoch:  1 , Batch:  19 , loss avg over log interval:  3.931011509895325\n",
            "Epoch:  1 , Batch:  29 , loss avg over log interval:  3.931180238723755\n",
            "Epoch:  1 , Batch:  39 , loss avg over log interval:  3.9312328100204468\n",
            "Epoch:  1 , Batch:  49 , loss avg over log interval:  3.9311185121536254\n",
            "Epoch:  1 , Batch:  59 , loss avg over log interval:  3.931024694442749\n",
            "Epoch:  1 , Batch:  69 , loss avg over log interval:  3.9307768821716307\n",
            "Epoch:  1 , Batch:  79 , loss avg over log interval:  3.931040620803833\n",
            "Epoch:  1 , Batch:  89 , loss avg over log interval:  3.930967617034912\n",
            "Epoch:  1 , Batch:  99 , loss avg over log interval:  3.930264711380005\n",
            "Epoch:  1 , Batch:  109 , loss avg over log interval:  3.9303975343704223\n",
            "Epoch:  1 , Batch:  119 , loss avg over log interval:  3.929697561264038\n",
            "Epoch:  1 , Batch:  129 , loss avg over log interval:  3.930266094207764\n",
            "Epoch:  1 , avg total loss:  3.9307368872822197\n",
            "iter   0\n",
            "accuracy so far =  0.1015625\n",
            "Avg Validation Loss:  3.931294121393343\n",
            "Accuracy:  3.1707317073170733\n",
            "Epoch 1/19 Completed\n",
            "Epoch:  2 , Batch:  0 , loss avg over log interval:  3.9298675060272217\n",
            "Epoch:  2 , Batch:  9 , loss avg over log interval:  3.536261534690857\n",
            "Epoch:  2 , Batch:  19 , loss avg over log interval:  3.9290090799331665\n",
            "Epoch:  2 , Batch:  29 , loss avg over log interval:  3.929576563835144\n",
            "Epoch:  2 , Batch:  39 , loss avg over log interval:  3.9289532899856567\n",
            "Epoch:  2 , Batch:  49 , loss avg over log interval:  3.9286259412765503\n",
            "Epoch:  2 , Batch:  59 , loss avg over log interval:  3.927679109573364\n",
            "Epoch:  2 , Batch:  69 , loss avg over log interval:  3.9283517360687257\n",
            "Epoch:  2 , Batch:  79 , loss avg over log interval:  3.9270694971084597\n",
            "Epoch:  2 , Batch:  89 , loss avg over log interval:  3.9266263008117677\n",
            "Epoch:  2 , Batch:  99 , loss avg over log interval:  3.9241026401519776\n",
            "Epoch:  2 , Batch:  109 , loss avg over log interval:  3.922806739807129\n",
            "Epoch:  2 , Batch:  119 , loss avg over log interval:  3.921521282196045\n",
            "Epoch:  2 , Batch:  129 , loss avg over log interval:  3.9177518606185915\n",
            "Epoch:  2 , avg total loss:  3.925762302633645\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.9310520276790712\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 2/19 Completed\n",
            "Epoch:  3 , Batch:  0 , loss avg over log interval:  3.9161183834075928\n",
            "Epoch:  3 , Batch:  9 , loss avg over log interval:  3.5217512369155886\n",
            "Epoch:  3 , Batch:  19 , loss avg over log interval:  3.9059032917022707\n",
            "Epoch:  3 , Batch:  29 , loss avg over log interval:  3.910338354110718\n",
            "Epoch:  3 , Batch:  39 , loss avg over log interval:  3.9015070676803587\n",
            "Epoch:  3 , Batch:  49 , loss avg over log interval:  3.9012736797332765\n",
            "Epoch:  3 , Batch:  59 , loss avg over log interval:  3.90022189617157\n",
            "Epoch:  3 , Batch:  69 , loss avg over log interval:  3.8999860286712646\n",
            "Epoch:  3 , Batch:  79 , loss avg over log interval:  3.8960232734680176\n",
            "Epoch:  3 , Batch:  89 , loss avg over log interval:  3.9053107261657716\n",
            "Epoch:  3 , Batch:  99 , loss avg over log interval:  3.89215145111084\n",
            "Epoch:  3 , Batch:  109 , loss avg over log interval:  3.894016218185425\n",
            "Epoch:  3 , Batch:  119 , loss avg over log interval:  3.8945687294006346\n",
            "Epoch:  3 , Batch:  129 , loss avg over log interval:  3.8956902503967283\n",
            "Epoch:  3 , avg total loss:  3.9007457045541294\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.930793145807778\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 3/19 Completed\n",
            "Epoch:  4 , Batch:  0 , loss avg over log interval:  3.9079854488372803\n",
            "Epoch:  4 , Batch:  9 , loss avg over log interval:  3.508902978897095\n",
            "Epoch:  4 , Batch:  19 , loss avg over log interval:  3.8925887823104857\n",
            "Epoch:  4 , Batch:  29 , loss avg over log interval:  3.902876353263855\n",
            "Epoch:  4 , Batch:  39 , loss avg over log interval:  3.892074537277222\n",
            "Epoch:  4 , Batch:  49 , loss avg over log interval:  3.890381908416748\n",
            "Epoch:  4 , Batch:  59 , loss avg over log interval:  3.888941693305969\n",
            "Epoch:  4 , Batch:  69 , loss avg over log interval:  3.90338716506958\n",
            "Epoch:  4 , Batch:  79 , loss avg over log interval:  3.895809864997864\n",
            "Epoch:  4 , Batch:  89 , loss avg over log interval:  3.911143159866333\n",
            "Epoch:  4 , Batch:  99 , loss avg over log interval:  3.894421172142029\n",
            "Epoch:  4 , Batch:  109 , loss avg over log interval:  3.8964759588241575\n",
            "Epoch:  4 , Batch:  119 , loss avg over log interval:  3.886207914352417\n",
            "Epoch:  4 , Batch:  129 , loss avg over log interval:  3.887940502166748\n",
            "Epoch:  4 , avg total loss:  3.895314233890478\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.9304759676863505\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 4/19 Completed\n",
            "Epoch:  5 , Batch:  0 , loss avg over log interval:  3.8788058757781982\n",
            "Epoch:  5 , Batch:  9 , loss avg over log interval:  3.501752734184265\n",
            "Epoch:  5 , Batch:  19 , loss avg over log interval:  3.900517439842224\n",
            "Epoch:  5 , Batch:  29 , loss avg over log interval:  3.897551655769348\n",
            "Epoch:  5 , Batch:  39 , loss avg over log interval:  3.8992130041122435\n",
            "Epoch:  5 , Batch:  49 , loss avg over log interval:  3.8967994451522827\n",
            "Epoch:  5 , Batch:  59 , loss avg over log interval:  3.890970301628113\n",
            "Epoch:  5 , Batch:  69 , loss avg over log interval:  3.905723714828491\n",
            "Epoch:  5 , Batch:  79 , loss avg over log interval:  3.8987650156021116\n",
            "Epoch:  5 , Batch:  89 , loss avg over log interval:  3.9056368112564086\n",
            "Epoch:  5 , Batch:  99 , loss avg over log interval:  3.895115065574646\n",
            "Epoch:  5 , Batch:  109 , loss avg over log interval:  3.899850106239319\n",
            "Epoch:  5 , Batch:  119 , loss avg over log interval:  3.8948853254318236\n",
            "Epoch:  5 , Batch:  129 , loss avg over log interval:  3.88728404045105\n",
            "Epoch:  5 , avg total loss:  3.896820825079213\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.9299875468742558\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 5/19 Completed\n",
            "Epoch:  6 , Batch:  0 , loss avg over log interval:  3.906346559524536\n",
            "Epoch:  6 , Batch:  9 , loss avg over log interval:  3.5051311016082765\n",
            "Epoch:  6 , Batch:  19 , loss avg over log interval:  3.8967217445373534\n",
            "Epoch:  6 , Batch:  29 , loss avg over log interval:  3.8971872568130492\n",
            "Epoch:  6 , Batch:  39 , loss avg over log interval:  3.888490343093872\n",
            "Epoch:  6 , Batch:  49 , loss avg over log interval:  3.906425642967224\n",
            "Epoch:  6 , Batch:  59 , loss avg over log interval:  3.891489768028259\n",
            "Epoch:  6 , Batch:  69 , loss avg over log interval:  3.9002790451049805\n",
            "Epoch:  6 , Batch:  79 , loss avg over log interval:  3.8994627952575684\n",
            "Epoch:  6 , Batch:  89 , loss avg over log interval:  3.896449112892151\n",
            "Epoch:  6 , Batch:  99 , loss avg over log interval:  3.8956549406051635\n",
            "Epoch:  6 , Batch:  109 , loss avg over log interval:  3.8920382499694823\n",
            "Epoch:  6 , Batch:  119 , loss avg over log interval:  3.8850700855255127\n",
            "Epoch:  6 , Batch:  129 , loss avg over log interval:  3.8932132005691527\n",
            "Epoch:  6 , avg total loss:  3.8948961444523023\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.9294092655181885\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 6/19 Completed\n",
            "Epoch:  7 , Batch:  0 , loss avg over log interval:  3.8910410404205322\n",
            "Epoch:  7 , Batch:  9 , loss avg over log interval:  3.507898187637329\n",
            "Epoch:  7 , Batch:  19 , loss avg over log interval:  3.903319525718689\n",
            "Epoch:  7 , Batch:  29 , loss avg over log interval:  3.8958846807479857\n",
            "Epoch:  7 , Batch:  39 , loss avg over log interval:  3.89743390083313\n",
            "Epoch:  7 , Batch:  49 , loss avg over log interval:  3.895764684677124\n",
            "Epoch:  7 , Batch:  59 , loss avg over log interval:  3.8907598495483398\n",
            "Epoch:  7 , Batch:  69 , loss avg over log interval:  3.904040551185608\n",
            "Epoch:  7 , Batch:  79 , loss avg over log interval:  3.8943806409835817\n",
            "Epoch:  7 , Batch:  89 , loss avg over log interval:  3.9018872737884522\n",
            "Epoch:  7 , Batch:  99 , loss avg over log interval:  3.8965088367462157\n",
            "Epoch:  7 , Batch:  109 , loss avg over log interval:  3.8950980424880983\n",
            "Epoch:  7 , Batch:  119 , loss avg over log interval:  3.8919743061065675\n",
            "Epoch:  7 , Batch:  129 , loss avg over log interval:  3.889729452133179\n",
            "Epoch:  7 , avg total loss:  3.895736673603887\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.928300537714144\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 7/19 Completed\n",
            "Epoch:  8 , Batch:  0 , loss avg over log interval:  3.9125351905822754\n",
            "Epoch:  8 , Batch:  9 , loss avg over log interval:  3.5133581876754763\n",
            "Epoch:  8 , Batch:  19 , loss avg over log interval:  3.89762921333313\n",
            "Epoch:  8 , Batch:  29 , loss avg over log interval:  3.8989041566848757\n",
            "Epoch:  8 , Batch:  39 , loss avg over log interval:  3.900069808959961\n",
            "Epoch:  8 , Batch:  49 , loss avg over log interval:  3.9010962724685667\n",
            "Epoch:  8 , Batch:  59 , loss avg over log interval:  3.889864182472229\n",
            "Epoch:  8 , Batch:  69 , loss avg over log interval:  3.8984951972961426\n",
            "Epoch:  8 , Batch:  79 , loss avg over log interval:  3.8854339122772217\n",
            "Epoch:  8 , Batch:  89 , loss avg over log interval:  3.899350380897522\n",
            "Epoch:  8 , Batch:  99 , loss avg over log interval:  3.8942208290100098\n",
            "Epoch:  8 , Batch:  109 , loss avg over log interval:  3.893737816810608\n",
            "Epoch:  8 , Batch:  119 , loss avg over log interval:  3.88865921497345\n",
            "Epoch:  8 , Batch:  129 , loss avg over log interval:  3.8922980070114135\n",
            "Epoch:  8 , avg total loss:  3.8955586371214492\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.9274282222840844\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 8/19 Completed\n",
            "Epoch:  9 , Batch:  0 , loss avg over log interval:  3.914074420928955\n",
            "Epoch:  9 , Batch:  9 , loss avg over log interval:  3.5068641185760496\n",
            "Epoch:  9 , Batch:  19 , loss avg over log interval:  3.896405816078186\n",
            "Epoch:  9 , Batch:  29 , loss avg over log interval:  3.8953739404678345\n",
            "Epoch:  9 , Batch:  39 , loss avg over log interval:  3.895434856414795\n",
            "Epoch:  9 , Batch:  49 , loss avg over log interval:  3.8991020202636717\n",
            "Epoch:  9 , Batch:  59 , loss avg over log interval:  3.8844961643218996\n",
            "Epoch:  9 , Batch:  69 , loss avg over log interval:  3.8879480838775633\n",
            "Epoch:  9 , Batch:  79 , loss avg over log interval:  3.887878346443176\n",
            "Epoch:  9 , Batch:  89 , loss avg over log interval:  3.8959912300109862\n",
            "Epoch:  9 , Batch:  99 , loss avg over log interval:  3.893005609512329\n",
            "Epoch:  9 , Batch:  109 , loss avg over log interval:  3.8936743259429933\n",
            "Epoch:  9 , Batch:  119 , loss avg over log interval:  3.8902904748916627\n",
            "Epoch:  9 , Batch:  129 , loss avg over log interval:  3.8879038810729982\n",
            "Epoch:  9 , avg total loss:  3.8927961225095005\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.925970973038092\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 9/19 Completed\n",
            "Epoch:  10 , Batch:  0 , loss avg over log interval:  3.885383129119873\n",
            "Epoch:  10 , Batch:  9 , loss avg over log interval:  3.504518175125122\n",
            "Epoch:  10 , Batch:  19 , loss avg over log interval:  3.8890709400177004\n",
            "Epoch:  10 , Batch:  29 , loss avg over log interval:  3.8948859214782714\n",
            "Epoch:  10 , Batch:  39 , loss avg over log interval:  3.900315690040588\n",
            "Epoch:  10 , Batch:  49 , loss avg over log interval:  3.8840612888336183\n",
            "Epoch:  10 , Batch:  59 , loss avg over log interval:  3.8873981714248655\n",
            "Epoch:  10 , Batch:  69 , loss avg over log interval:  3.886412191390991\n",
            "Epoch:  10 , Batch:  79 , loss avg over log interval:  3.890975904464722\n",
            "Epoch:  10 , Batch:  89 , loss avg over log interval:  3.9043392658233644\n",
            "Epoch:  10 , Batch:  99 , loss avg over log interval:  3.8892335891723633\n",
            "Epoch:  10 , Batch:  109 , loss avg over log interval:  3.888150691986084\n",
            "Epoch:  10 , Batch:  119 , loss avg over log interval:  3.8864094257354735\n",
            "Epoch:  10 , Batch:  129 , loss avg over log interval:  3.887697768211365\n",
            "Epoch:  10 , avg total loss:  3.890152824097785\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.92437842997109\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 10/19 Completed\n",
            "Epoch:  11 , Batch:  0 , loss avg over log interval:  3.911081075668335\n",
            "Epoch:  11 , Batch:  9 , loss avg over log interval:  3.505476975440979\n",
            "Epoch:  11 , Batch:  19 , loss avg over log interval:  3.886572742462158\n",
            "Epoch:  11 , Batch:  29 , loss avg over log interval:  3.8984040021896362\n",
            "Epoch:  11 , Batch:  39 , loss avg over log interval:  3.8870191097259523\n",
            "Epoch:  11 , Batch:  49 , loss avg over log interval:  3.8877343416213987\n",
            "Epoch:  11 , Batch:  59 , loss avg over log interval:  3.891105389595032\n",
            "Epoch:  11 , Batch:  69 , loss avg over log interval:  3.898379588127136\n",
            "Epoch:  11 , Batch:  79 , loss avg over log interval:  3.8861510038375853\n",
            "Epoch:  11 , Batch:  89 , loss avg over log interval:  3.8875122547149656\n",
            "Epoch:  11 , Batch:  99 , loss avg over log interval:  3.901709699630737\n",
            "Epoch:  11 , Batch:  109 , loss avg over log interval:  3.8918405532836915\n",
            "Epoch:  11 , Batch:  119 , loss avg over log interval:  3.8914628267288207\n",
            "Epoch:  11 , Batch:  129 , loss avg over log interval:  3.8876873731613157\n",
            "Epoch:  11 , avg total loss:  3.8914814669152964\n",
            "iter   0\n",
            "accuracy so far =  0.0\n",
            "Avg Validation Loss:  3.9217584714656923\n",
            "Accuracy:  2.8048780487804876\n",
            "Epoch 11/19 Completed\n",
            "Epoch:  12 , Batch:  0 , loss avg over log interval:  3.9040350914001465\n",
            "Epoch:  12 , Batch:  9 , loss avg over log interval:  3.4982107162475584\n",
            "Epoch:  12 , Batch:  19 , loss avg over log interval:  3.8824658393859863\n",
            "Epoch:  12 , Batch:  29 , loss avg over log interval:  3.8942535877227784\n",
            "Epoch:  12 , Batch:  39 , loss avg over log interval:  3.8864696502685545\n",
            "Epoch:  12 , Batch:  49 , loss avg over log interval:  3.888178586959839\n",
            "Epoch:  12 , Batch:  59 , loss avg over log interval:  3.8786614656448366\n",
            "Epoch:  12 , Batch:  69 , loss avg over log interval:  3.890628981590271\n",
            "Epoch:  12 , Batch:  79 , loss avg over log interval:  3.8785306453704833\n",
            "Epoch:  12 , Batch:  89 , loss avg over log interval:  3.893293023109436\n",
            "Epoch:  12 , Batch:  99 , loss avg over log interval:  3.8805344343185424\n",
            "Epoch:  12 , Batch:  109 , loss avg over log interval:  3.8892974853515625\n",
            "Epoch:  12 , Batch:  119 , loss avg over log interval:  3.878446865081787\n",
            "Epoch:  12 , Batch:  129 , loss avg over log interval:  3.879699730873108\n",
            "Epoch:  12 , avg total loss:  3.885139577630637\n",
            "iter   0\n",
            "accuracy so far =  0.0234375\n",
            "Avg Validation Loss:  3.918896413430935\n",
            "Accuracy:  3.8048780487804876\n",
            "Epoch 12/19 Completed\n",
            "Epoch:  13 , Batch:  0 , loss avg over log interval:  3.9130098819732666\n",
            "Epoch:  13 , Batch:  9 , loss avg over log interval:  3.4967447996139525\n",
            "Epoch:  13 , Batch:  19 , loss avg over log interval:  3.8860710144042967\n",
            "Epoch:  13 , Batch:  29 , loss avg over log interval:  3.891244649887085\n",
            "Epoch:  13 , Batch:  39 , loss avg over log interval:  3.8820359230041506\n",
            "Epoch:  13 , Batch:  49 , loss avg over log interval:  3.8818006992340086\n",
            "Epoch:  13 , Batch:  59 , loss avg over log interval:  3.8890833139419554\n",
            "Epoch:  13 , Batch:  69 , loss avg over log interval:  3.8836084365844727\n",
            "Epoch:  13 , Batch:  79 , loss avg over log interval:  3.8792039155960083\n",
            "Epoch:  13 , Batch:  89 , loss avg over log interval:  3.8921865224838257\n",
            "Epoch:  13 , Batch:  99 , loss avg over log interval:  3.8842477083206175\n",
            "Epoch:  13 , Batch:  109 , loss avg over log interval:  3.8842225790023805\n",
            "Epoch:  13 , Batch:  119 , loss avg over log interval:  3.877263879776001\n",
            "Epoch:  13 , Batch:  129 , loss avg over log interval:  3.8741548538208006\n",
            "Epoch:  13 , avg total loss:  3.883687480636265\n",
            "iter   0\n",
            "accuracy so far =  0.2890625\n",
            "Avg Validation Loss:  3.916158280721525\n",
            "Accuracy:  5.829268292682927\n",
            "Epoch 13/19 Completed\n",
            "Epoch:  14 , Batch:  0 , loss avg over log interval:  3.899663209915161\n",
            "Epoch:  14 , Batch:  9 , loss avg over log interval:  3.4929866552352906\n",
            "Epoch:  14 , Batch:  19 , loss avg over log interval:  3.885475587844849\n",
            "Epoch:  14 , Batch:  29 , loss avg over log interval:  3.8866902112960817\n",
            "Epoch:  14 , Batch:  39 , loss avg over log interval:  3.8787101984024046\n",
            "Epoch:  14 , Batch:  49 , loss avg over log interval:  3.8919932365417482\n",
            "Epoch:  14 , Batch:  59 , loss avg over log interval:  3.875093364715576\n",
            "Epoch:  14 , Batch:  69 , loss avg over log interval:  3.8853904724121096\n",
            "Epoch:  14 , Batch:  79 , loss avg over log interval:  3.8830509901046755\n",
            "Epoch:  14 , Batch:  89 , loss avg over log interval:  3.8931920766830443\n",
            "Epoch:  14 , Batch:  99 , loss avg over log interval:  3.876305603981018\n",
            "Epoch:  14 , Batch:  109 , loss avg over log interval:  3.8791544437408447\n",
            "Epoch:  14 , Batch:  119 , loss avg over log interval:  3.882508730888367\n",
            "Epoch:  14 , Batch:  129 , loss avg over log interval:  3.8726315021514894\n",
            "Epoch:  14 , avg total loss:  3.8817819961603135\n",
            "iter   0\n",
            "accuracy so far =  0.609375\n",
            "Avg Validation Loss:  3.913370626728709\n",
            "Accuracy:  8.878048780487806\n",
            "Epoch 14/19 Completed\n",
            "Epoch:  15 , Batch:  0 , loss avg over log interval:  3.874415636062622\n",
            "Epoch:  15 , Batch:  9 , loss avg over log interval:  3.4935940742492675\n",
            "Epoch:  15 , Batch:  19 , loss avg over log interval:  3.8803814888000487\n",
            "Epoch:  15 , Batch:  29 , loss avg over log interval:  3.8806679248809814\n",
            "Epoch:  15 , Batch:  39 , loss avg over log interval:  3.885931611061096\n",
            "Epoch:  15 , Batch:  49 , loss avg over log interval:  3.8683592796325685\n",
            "Epoch:  15 , Batch:  59 , loss avg over log interval:  3.874373269081116\n",
            "Epoch:  15 , Batch:  69 , loss avg over log interval:  3.8829083681106566\n",
            "Epoch:  15 , Batch:  79 , loss avg over log interval:  3.877142071723938\n",
            "Epoch:  15 , Batch:  89 , loss avg over log interval:  3.879353666305542\n",
            "Epoch:  15 , Batch:  99 , loss avg over log interval:  3.880255603790283\n",
            "Epoch:  15 , Batch:  109 , loss avg over log interval:  3.8819183111190796\n",
            "Epoch:  15 , Batch:  119 , loss avg over log interval:  3.874650573730469\n",
            "Epoch:  15 , Batch:  129 , loss avg over log interval:  3.862854170799255\n",
            "Epoch:  15 , avg total loss:  3.8775284774061562\n",
            "iter   0\n",
            "accuracy so far =  0.765625\n",
            "Avg Validation Loss:  3.9114596087758136\n",
            "Accuracy:  10.78048780487805\n",
            "Epoch 15/19 Completed\n",
            "Epoch:  16 , Batch:  0 , loss avg over log interval:  3.886216163635254\n",
            "Epoch:  16 , Batch:  9 , loss avg over log interval:  3.493420934677124\n",
            "Epoch:  16 , Batch:  19 , loss avg over log interval:  3.8772780895233154\n",
            "Epoch:  16 , Batch:  29 , loss avg over log interval:  3.8857818603515626\n",
            "Epoch:  16 , Batch:  39 , loss avg over log interval:  3.868732452392578\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-27-4afbebecfd6c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     34\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     35\u001b[0m     \u001b[0;31m# Train & Validate\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 36\u001b[0;31m     \u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mepoch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     37\u001b[0m     \u001b[0mtest\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     38\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-26-9adc05807883>\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(epoch)\u001b[0m\n\u001b[1;32m     17\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m         \u001b[0;31m# Input Transforms\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 19\u001b[0;31m         \u001b[0mvideos\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mvideos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mFloatTensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     20\u001b[0m         \u001b[0mvideos\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtransform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvideos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpermute\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpermute\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aPT-gwaOIer1"
      },
      "source": [
        "# Resume Train Loop"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Zag-t87wIeaG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b4ae6d62-bbba-4fec-9138-567294d424c6"
      },
      "source": [
        "EPOCH_SAVE_PREFIX = '/content/drive/Shared drives/CIS680 Final Project/models/temporal_new_data/'\n",
        "\n",
        "# Device\n",
        "device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')\n",
        "torch.random.manual_seed(1)\n",
        "\n",
        "# model\n",
        "learning_rate = 0.001\n",
        "temporal = TemporalStream()\n",
        "temporal.to(device)\n",
        "optimizer = optim.SGD(temporal.parameters(), lr=learning_rate, momentum=0.9)\n",
        "\n",
        "# Load model\n",
        "last_epoch = 106\n",
        "network_path = EPOCH_SAVE_PREFIX + 'temporal_epoch' + str(last_epoch)\n",
        "checkpoint = torch.load(network_path)\n",
        "temporal.load_state_dict(checkpoint['model_state_dict'])\n",
        "optimizer.load_state_dict(checkpoint['optimizer_state_dict'])\n",
        "\n",
        "# Epochs\n",
        "num_epochs = 250\n",
        "batch_size = 128\n",
        "\n",
        "# Logging setup: train\n",
        "train_loss_list = checkpoint['train_total_loss_list']\n",
        "epoch_loss_list = checkpoint['epoch_total_loss_list']\n",
        "train_counter = checkpoint['train_counter']\n",
        "test_loss_list = checkpoint['test_loss_list']\n",
        "accuracy_list = checkpoint['accuracy_lsit']\n",
        "epoch_list = np.arange(num_epochs)\n",
        "\n",
        "# epoch loop\n",
        "for epoch in range(last_epoch + 1, num_epochs): \n",
        "\n",
        "    # Train & Validate\n",
        "    train(epoch)\n",
        "    test()\n",
        "\n",
        "    # Save Model Version\n",
        "    save_path = os.path.join(EPOCH_SAVE_PREFIX, 'temporal_epoch' + str(epoch))\n",
        "    torch.save({\n",
        "        'epoch': epoch,\n",
        "        'train_total_loss_list': train_loss_list,\n",
        "        'epoch_total_loss_list': epoch_loss_list,\n",
        "        'test_loss_list': test_loss_list,\n",
        "        'train_counter': train_counter,\n",
        "        'accuracy_lsit': accuracy_list,\n",
        "        'model_state_dict': temporal.state_dict(),\n",
        "        'optimizer_state_dict': optimizer.state_dict()\n",
        "    }, save_path)\n",
        "\n",
        "    print(\"Epoch %d/%d Completed\" % (epoch, num_epochs - 1))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/container.py:117: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
            "  input = module(input)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch:  16 , Batch:  0 , loss avg over log interval:  3.871645212173462\n",
            "Epoch:  16 , Batch:  9 , loss avg over log interval:  3.480390191078186\n",
            "Epoch:  16 , Batch:  19 , loss avg over log interval:  3.8734632253646852\n",
            "Epoch:  16 , Batch:  29 , loss avg over log interval:  3.8873563051223754\n",
            "Epoch:  16 , Batch:  39 , loss avg over log interval:  3.8774696350097657\n",
            "Epoch:  16 , Batch:  49 , loss avg over log interval:  3.884968876838684\n",
            "Epoch:  16 , Batch:  59 , loss avg over log interval:  3.8813621520996096\n",
            "Epoch:  16 , Batch:  69 , loss avg over log interval:  3.885278344154358\n",
            "Epoch:  16 , Batch:  79 , loss avg over log interval:  3.875833296775818\n",
            "Epoch:  16 , Batch:  89 , loss avg over log interval:  3.8785472869873048\n",
            "Epoch:  16 , Batch:  99 , loss avg over log interval:  3.8779523611068725\n",
            "Epoch:  16 , Batch:  109 , loss avg over log interval:  3.8699994802474977\n",
            "Epoch:  16 , Batch:  119 , loss avg over log interval:  3.86314742565155\n",
            "Epoch:  16 , Batch:  129 , loss avg over log interval:  3.866294264793396\n",
            "Epoch:  16 , avg total loss:  3.875273205231929\n",
            "iter   0\n",
            "accuracy so far =  0.8359375\n",
            "Avg Validation Loss:  3.907437353599362\n",
            "Accuracy:  11.121951219512194\n",
            "Epoch 16/149 Completed\n",
            "Epoch:  17 , Batch:  0 , loss avg over log interval:  3.881763458251953\n",
            "Epoch:  17 , Batch:  9 , loss avg over log interval:  3.476013970375061\n",
            "Epoch:  17 , Batch:  19 , loss avg over log interval:  3.8695806503295898\n",
            "Epoch:  17 , Batch:  29 , loss avg over log interval:  3.8818992137908936\n",
            "Epoch:  17 , Batch:  39 , loss avg over log interval:  3.8699925899505616\n",
            "Epoch:  17 , Batch:  49 , loss avg over log interval:  3.871875691413879\n",
            "Epoch:  17 , Batch:  59 , loss avg over log interval:  3.873051881790161\n",
            "Epoch:  17 , Batch:  69 , loss avg over log interval:  3.873553156852722\n",
            "Epoch:  17 , Batch:  79 , loss avg over log interval:  3.8778058767318724\n",
            "Epoch:  17 , Batch:  89 , loss avg over log interval:  3.881626296043396\n",
            "Epoch:  17 , Batch:  99 , loss avg over log interval:  3.8677008152008057\n",
            "Epoch:  17 , Batch:  109 , loss avg over log interval:  3.8697887659072876\n",
            "Epoch:  17 , Batch:  119 , loss avg over log interval:  3.8526422023773192\n",
            "Epoch:  17 , Batch:  129 , loss avg over log interval:  3.866188716888428\n",
            "Epoch:  17 , avg total loss:  3.8705362392508467\n",
            "iter   0\n",
            "accuracy so far =  0.84375\n",
            "Avg Validation Loss:  3.9061573133236025\n",
            "Accuracy:  11.439024390243903\n",
            "Epoch 17/149 Completed\n",
            "Epoch:  18 , Batch:  0 , loss avg over log interval:  3.877378463745117\n",
            "Epoch:  18 , Batch:  9 , loss avg over log interval:  3.47851402759552\n",
            "Epoch:  18 , Batch:  19 , loss avg over log interval:  3.85922749042511\n",
            "Epoch:  18 , Batch:  29 , loss avg over log interval:  3.8819118976593017\n",
            "Epoch:  18 , Batch:  39 , loss avg over log interval:  3.871298050880432\n",
            "Epoch:  18 , Batch:  49 , loss avg over log interval:  3.8734691381454467\n",
            "Epoch:  18 , Batch:  59 , loss avg over log interval:  3.8656700611114503\n",
            "Epoch:  18 , Batch:  69 , loss avg over log interval:  3.8727704524993896\n",
            "Epoch:  18 , Batch:  79 , loss avg over log interval:  3.8754078149795532\n",
            "Epoch:  18 , Batch:  89 , loss avg over log interval:  3.875534176826477\n",
            "Epoch:  18 , Batch:  99 , loss avg over log interval:  3.870075559616089\n",
            "Epoch:  18 , Batch:  109 , loss avg over log interval:  3.8696681022644044\n",
            "Epoch:  18 , Batch:  119 , loss avg over log interval:  3.866144227981567\n",
            "Epoch:  18 , Batch:  129 , loss avg over log interval:  3.860966920852661\n",
            "Epoch:  18 , avg total loss:  3.870145068652388\n",
            "iter   0\n",
            "accuracy so far =  0.8828125\n",
            "Avg Validation Loss:  3.901778732858053\n",
            "Accuracy:  11.463414634146341\n",
            "Epoch 18/149 Completed\n",
            "Epoch:  19 , Batch:  0 , loss avg over log interval:  3.8626468181610107\n",
            "Epoch:  19 , Batch:  9 , loss avg over log interval:  3.477889156341553\n",
            "Epoch:  19 , Batch:  19 , loss avg over log interval:  3.863555836677551\n",
            "Epoch:  19 , Batch:  29 , loss avg over log interval:  3.8792702198028564\n",
            "Epoch:  19 , Batch:  39 , loss avg over log interval:  3.861364960670471\n",
            "Epoch:  19 , Batch:  49 , loss avg over log interval:  3.8650150537490844\n",
            "Epoch:  19 , Batch:  59 , loss avg over log interval:  3.872761273384094\n",
            "Epoch:  19 , Batch:  69 , loss avg over log interval:  3.8660574436187742\n",
            "Epoch:  19 , Batch:  79 , loss avg over log interval:  3.8708277463912966\n",
            "Epoch:  19 , Batch:  89 , loss avg over log interval:  3.8762458324432374\n",
            "Epoch:  19 , Batch:  99 , loss avg over log interval:  3.866809034347534\n",
            "Epoch:  19 , Batch:  109 , loss avg over log interval:  3.8619773149490357\n",
            "Epoch:  19 , Batch:  119 , loss avg over log interval:  3.8582571506500245\n",
            "Epoch:  19 , Batch:  129 , loss avg over log interval:  3.8678534030914307\n",
            "Epoch:  19 , avg total loss:  3.8670632199964663\n",
            "iter   0\n",
            "accuracy so far =  0.8515625\n",
            "Avg Validation Loss:  3.9033064086262774\n",
            "Accuracy:  11.560975609756097\n",
            "Epoch 19/149 Completed\n",
            "Epoch:  20 , Batch:  0 , loss avg over log interval:  3.864253520965576\n",
            "Epoch:  20 , Batch:  9 , loss avg over log interval:  3.477495288848877\n",
            "Epoch:  20 , Batch:  19 , loss avg over log interval:  3.8596250772476197\n",
            "Epoch:  20 , Batch:  29 , loss avg over log interval:  3.8782334327697754\n",
            "Epoch:  20 , Batch:  39 , loss avg over log interval:  3.8638590812683105\n",
            "Epoch:  20 , Batch:  49 , loss avg over log interval:  3.860723376274109\n",
            "Epoch:  20 , Batch:  59 , loss avg over log interval:  3.858058786392212\n",
            "Epoch:  20 , Batch:  69 , loss avg over log interval:  3.8688227415084837\n",
            "Epoch:  20 , Batch:  79 , loss avg over log interval:  3.8672500133514403\n",
            "Epoch:  20 , Batch:  89 , loss avg over log interval:  3.885274600982666\n",
            "Epoch:  20 , Batch:  99 , loss avg over log interval:  3.8681362152099608\n",
            "Epoch:  20 , Batch:  109 , loss avg over log interval:  3.8655118703842164\n",
            "Epoch:  20 , Batch:  119 , loss avg over log interval:  3.849195957183838\n",
            "Epoch:  20 , Batch:  129 , loss avg over log interval:  3.853462648391724\n",
            "Epoch:  20 , avg total loss:  3.864817249602166\n",
            "iter   0\n",
            "accuracy so far =  0.90625\n",
            "Avg Validation Loss:  3.8974236918658747\n",
            "Accuracy:  11.585365853658537\n",
            "Epoch 20/149 Completed\n",
            "Epoch:  21 , Batch:  0 , loss avg over log interval:  3.855105400085449\n",
            "Epoch:  21 , Batch:  9 , loss avg over log interval:  3.469621753692627\n",
            "Epoch:  21 , Batch:  19 , loss avg over log interval:  3.8685948610305787\n",
            "Epoch:  21 , Batch:  29 , loss avg over log interval:  3.866541934013367\n",
            "Epoch:  21 , Batch:  39 , loss avg over log interval:  3.866713619232178\n",
            "Epoch:  21 , Batch:  49 , loss avg over log interval:  3.86209499835968\n",
            "Epoch:  21 , Batch:  59 , loss avg over log interval:  3.860790157318115\n",
            "Epoch:  21 , Batch:  69 , loss avg over log interval:  3.872280979156494\n",
            "Epoch:  21 , Batch:  79 , loss avg over log interval:  3.8689226388931273\n",
            "Epoch:  21 , Batch:  89 , loss avg over log interval:  3.872570610046387\n",
            "Epoch:  21 , Batch:  99 , loss avg over log interval:  3.8674548864364624\n",
            "Epoch:  21 , Batch:  109 , loss avg over log interval:  3.869739532470703\n",
            "Epoch:  21 , Batch:  119 , loss avg over log interval:  3.8593231201171876\n",
            "Epoch:  21 , Batch:  129 , loss avg over log interval:  3.8549344301223756\n",
            "Epoch:  21 , avg total loss:  3.865100411401279\n",
            "iter   0\n",
            "accuracy so far =  0.90625\n",
            "Avg Validation Loss:  3.89602298852874\n",
            "Accuracy:  11.951219512195122\n",
            "Epoch 21/149 Completed\n",
            "Epoch:  22 , Batch:  0 , loss avg over log interval:  3.8701894283294678\n",
            "Epoch:  22 , Batch:  9 , loss avg over log interval:  3.473877787590027\n",
            "Epoch:  22 , Batch:  19 , loss avg over log interval:  3.866495633125305\n",
            "Epoch:  22 , Batch:  29 , loss avg over log interval:  3.865134024620056\n",
            "Epoch:  22 , Batch:  39 , loss avg over log interval:  3.8491108417510986\n",
            "Epoch:  22 , Batch:  49 , loss avg over log interval:  3.8732099294662476\n",
            "Epoch:  22 , Batch:  59 , loss avg over log interval:  3.8570058345794678\n",
            "Epoch:  22 , Batch:  69 , loss avg over log interval:  3.8690940618515013\n",
            "Epoch:  22 , Batch:  79 , loss avg over log interval:  3.870982360839844\n",
            "Epoch:  22 , Batch:  89 , loss avg over log interval:  3.861359429359436\n",
            "Epoch:  22 , Batch:  99 , loss avg over log interval:  3.8679906606674193\n",
            "Epoch:  22 , Batch:  109 , loss avg over log interval:  3.8557677268981934\n",
            "Epoch:  22 , Batch:  119 , loss avg over log interval:  3.84869544506073\n",
            "Epoch:  22 , Batch:  129 , loss avg over log interval:  3.861981248855591\n",
            "Epoch:  22 , avg total loss:  3.861664725386578\n",
            "iter   0\n",
            "accuracy so far =  0.890625\n",
            "Avg Validation Loss:  3.894865094161615\n",
            "Accuracy:  11.878048780487806\n",
            "Epoch 22/149 Completed\n",
            "Epoch:  23 , Batch:  0 , loss avg over log interval:  3.8697571754455566\n",
            "Epoch:  23 , Batch:  9 , loss avg over log interval:  3.4701563358306884\n",
            "Epoch:  23 , Batch:  19 , loss avg over log interval:  3.868569278717041\n",
            "Epoch:  23 , Batch:  29 , loss avg over log interval:  3.8688552618026733\n",
            "Epoch:  23 , Batch:  39 , loss avg over log interval:  3.8566533088684083\n",
            "Epoch:  23 , Batch:  49 , loss avg over log interval:  3.863833260536194\n",
            "Epoch:  23 , Batch:  59 , loss avg over log interval:  3.8598856925964355\n",
            "Epoch:  23 , Batch:  69 , loss avg over log interval:  3.8732909202575683\n",
            "Epoch:  23 , Batch:  79 , loss avg over log interval:  3.864467644691467\n",
            "Epoch:  23 , Batch:  89 , loss avg over log interval:  3.869326376914978\n",
            "Epoch:  23 , Batch:  99 , loss avg over log interval:  3.8672985315322874\n",
            "Epoch:  23 , Batch:  109 , loss avg over log interval:  3.8557825565338133\n",
            "Epoch:  23 , Batch:  119 , loss avg over log interval:  3.8589733362197878\n",
            "Epoch:  23 , Batch:  129 , loss avg over log interval:  3.8577212572097777\n",
            "Epoch:  23 , avg total loss:  3.8624326543531557\n",
            "iter   0\n",
            "accuracy so far =  0.8828125\n",
            "Avg Validation Loss:  3.895831550039896\n",
            "Accuracy:  11.829268292682928\n",
            "Epoch 23/149 Completed\n",
            "Epoch:  24 , Batch:  0 , loss avg over log interval:  3.894202470779419\n",
            "Epoch:  24 , Batch:  9 , loss avg over log interval:  3.4791758060455322\n",
            "Epoch:  24 , Batch:  19 , loss avg over log interval:  3.864951181411743\n",
            "Epoch:  24 , Batch:  29 , loss avg over log interval:  3.8704264163970947\n",
            "Epoch:  24 , Batch:  39 , loss avg over log interval:  3.866337609291077\n",
            "Epoch:  24 , Batch:  49 , loss avg over log interval:  3.8638390779495237\n",
            "Epoch:  24 , Batch:  59 , loss avg over log interval:  3.8538100004196165\n",
            "Epoch:  24 , Batch:  69 , loss avg over log interval:  3.865137481689453\n",
            "Epoch:  24 , Batch:  79 , loss avg over log interval:  3.857503366470337\n",
            "Epoch:  24 , Batch:  89 , loss avg over log interval:  3.8703418254852293\n",
            "Epoch:  24 , Batch:  99 , loss avg over log interval:  3.863471007347107\n",
            "Epoch:  24 , Batch:  109 , loss avg over log interval:  3.8612425327301025\n",
            "Epoch:  24 , Batch:  119 , loss avg over log interval:  3.8520447492599486\n",
            "Epoch:  24 , Batch:  129 , loss avg over log interval:  3.8565144300460816\n",
            "Epoch:  24 , avg total loss:  3.8625481145969336\n",
            "iter   0\n",
            "accuracy so far =  0.890625\n",
            "Avg Validation Loss:  3.891735321137963\n",
            "Accuracy:  11.902439024390244\n",
            "Epoch 24/149 Completed\n",
            "Epoch:  25 , Batch:  0 , loss avg over log interval:  3.8932132720947266\n",
            "Epoch:  25 , Batch:  9 , loss avg over log interval:  3.4759836196899414\n",
            "Epoch:  25 , Batch:  19 , loss avg over log interval:  3.8635404109954834\n",
            "Epoch:  25 , Batch:  29 , loss avg over log interval:  3.8597625970840452\n",
            "Epoch:  25 , Batch:  39 , loss avg over log interval:  3.859493136405945\n",
            "Epoch:  25 , Batch:  49 , loss avg over log interval:  3.8588060617446898\n",
            "Epoch:  25 , Batch:  59 , loss avg over log interval:  3.84843909740448\n",
            "Epoch:  25 , Batch:  69 , loss avg over log interval:  3.849821758270264\n",
            "Epoch:  25 , Batch:  79 , loss avg over log interval:  3.8534518003463747\n",
            "Epoch:  25 , Batch:  89 , loss avg over log interval:  3.864882731437683\n",
            "Epoch:  25 , Batch:  99 , loss avg over log interval:  3.864483189582825\n",
            "Epoch:  25 , Batch:  109 , loss avg over log interval:  3.859971213340759\n",
            "Epoch:  25 , Batch:  119 , loss avg over log interval:  3.851610541343689\n",
            "Epoch:  25 , Batch:  129 , loss avg over log interval:  3.8535207748413085\n",
            "Epoch:  25 , avg total loss:  3.857598862786224\n",
            "iter   0\n",
            "accuracy so far =  0.8671875\n",
            "Avg Validation Loss:  3.890840105894135\n",
            "Accuracy:  11.853658536585366\n",
            "Epoch 25/149 Completed\n",
            "Epoch:  26 , Batch:  0 , loss avg over log interval:  3.87227463722229\n",
            "Epoch:  26 , Batch:  9 , loss avg over log interval:  3.470635461807251\n",
            "Epoch:  26 , Batch:  19 , loss avg over log interval:  3.853536510467529\n",
            "Epoch:  26 , Batch:  29 , loss avg over log interval:  3.8667061567306518\n",
            "Epoch:  26 , Batch:  39 , loss avg over log interval:  3.8659014463424684\n",
            "Epoch:  26 , Batch:  49 , loss avg over log interval:  3.8526907682418825\n",
            "Epoch:  26 , Batch:  59 , loss avg over log interval:  3.8566281318664553\n",
            "Epoch:  26 , Batch:  69 , loss avg over log interval:  3.85082802772522\n",
            "Epoch:  26 , Batch:  79 , loss avg over log interval:  3.8583287477493284\n",
            "Epoch:  26 , Batch:  89 , loss avg over log interval:  3.8773410320281982\n",
            "Epoch:  26 , Batch:  99 , loss avg over log interval:  3.860009503364563\n",
            "Epoch:  26 , Batch:  109 , loss avg over log interval:  3.851732063293457\n",
            "Epoch:  26 , Batch:  119 , loss avg over log interval:  3.852028799057007\n",
            "Epoch:  26 , Batch:  129 , loss avg over log interval:  3.8557385206222534\n",
            "Epoch:  26 , avg total loss:  3.8576895534128384\n",
            "iter   0\n",
            "accuracy so far =  0.8984375\n",
            "Avg Validation Loss:  3.8872472251333843\n",
            "Accuracy:  12.707317073170731\n",
            "Epoch 26/149 Completed\n",
            "Epoch:  27 , Batch:  0 , loss avg over log interval:  3.8573155403137207\n",
            "Epoch:  27 , Batch:  9 , loss avg over log interval:  3.4772865056991575\n",
            "Epoch:  27 , Batch:  19 , loss avg over log interval:  3.848975324630737\n",
            "Epoch:  27 , Batch:  29 , loss avg over log interval:  3.863061761856079\n",
            "Epoch:  27 , Batch:  39 , loss avg over log interval:  3.8553491830825806\n",
            "Epoch:  27 , Batch:  49 , loss avg over log interval:  3.853813886642456\n",
            "Epoch:  27 , Batch:  59 , loss avg over log interval:  3.85203218460083\n",
            "Epoch:  27 , Batch:  69 , loss avg over log interval:  3.865912437438965\n",
            "Epoch:  27 , Batch:  79 , loss avg over log interval:  3.856777477264404\n",
            "Epoch:  27 , Batch:  89 , loss avg over log interval:  3.852954077720642\n",
            "Epoch:  27 , Batch:  99 , loss avg over log interval:  3.867371344566345\n",
            "Epoch:  27 , Batch:  109 , loss avg over log interval:  3.8618906497955323\n",
            "Epoch:  27 , Batch:  119 , loss avg over log interval:  3.8600671768188475\n",
            "Epoch:  27 , Batch:  129 , loss avg over log interval:  3.850324034690857\n",
            "Epoch:  27 , avg total loss:  3.857695508694303\n",
            "iter   0\n",
            "accuracy so far =  0.8828125\n",
            "Avg Validation Loss:  3.8882061039529194\n",
            "Accuracy:  15.073170731707316\n",
            "Epoch 27/149 Completed\n",
            "Epoch:  28 , Batch:  0 , loss avg over log interval:  3.8779220581054688\n",
            "Epoch:  28 , Batch:  9 , loss avg over log interval:  3.4643603324890138\n",
            "Epoch:  28 , Batch:  19 , loss avg over log interval:  3.8471649169921873\n",
            "Epoch:  28 , Batch:  29 , loss avg over log interval:  3.8594555616378785\n",
            "Epoch:  28 , Batch:  39 , loss avg over log interval:  3.848612141609192\n",
            "Epoch:  28 , Batch:  49 , loss avg over log interval:  3.85438985824585\n",
            "Epoch:  28 , Batch:  59 , loss avg over log interval:  3.841610336303711\n",
            "Epoch:  28 , Batch:  69 , loss avg over log interval:  3.853539323806763\n",
            "Epoch:  28 , Batch:  79 , loss avg over log interval:  3.850259304046631\n",
            "Epoch:  28 , Batch:  89 , loss avg over log interval:  3.8597612142562867\n",
            "Epoch:  28 , Batch:  99 , loss avg over log interval:  3.8527161359786986\n",
            "Epoch:  28 , Batch:  109 , loss avg over log interval:  3.8551035165786742\n",
            "Epoch:  28 , Batch:  119 , loss avg over log interval:  3.836216950416565\n",
            "Epoch:  28 , Batch:  129 , loss avg over log interval:  3.8431525945663454\n",
            "Epoch:  28 , avg total loss:  3.8497238936631577\n",
            "iter   0\n",
            "accuracy so far =  0.8828125\n",
            "Avg Validation Loss:  3.88261228072934\n",
            "Accuracy:  15.634146341463415\n",
            "Epoch 28/149 Completed\n",
            "Epoch:  29 , Batch:  0 , loss avg over log interval:  3.890031576156616\n",
            "Epoch:  29 , Batch:  9 , loss avg over log interval:  3.4591655254364015\n",
            "Epoch:  29 , Batch:  19 , loss avg over log interval:  3.852224516868591\n",
            "Epoch:  29 , Batch:  29 , loss avg over log interval:  3.8568541049957275\n",
            "Epoch:  29 , Batch:  39 , loss avg over log interval:  3.8456756114959716\n",
            "Epoch:  29 , Batch:  49 , loss avg over log interval:  3.843146729469299\n",
            "Epoch:  29 , Batch:  59 , loss avg over log interval:  3.855753993988037\n",
            "Epoch:  29 , Batch:  69 , loss avg over log interval:  3.8477802991867067\n",
            "Epoch:  29 , Batch:  79 , loss avg over log interval:  3.8485177516937257\n",
            "Epoch:  29 , Batch:  89 , loss avg over log interval:  3.8586641788482665\n",
            "Epoch:  29 , Batch:  99 , loss avg over log interval:  3.853950548171997\n",
            "Epoch:  29 , Batch:  109 , loss avg over log interval:  3.84892201423645\n",
            "Epoch:  29 , Batch:  119 , loss avg over log interval:  3.841381788253784\n",
            "Epoch:  29 , Batch:  129 , loss avg over log interval:  3.842393088340759\n",
            "Epoch:  29 , avg total loss:  3.8491028339966484\n",
            "iter   0\n",
            "accuracy so far =  0.8515625\n",
            "Avg Validation Loss:  3.8835117119114573\n",
            "Accuracy:  15.853658536585366\n",
            "Epoch 29/149 Completed\n",
            "Epoch:  30 , Batch:  0 , loss avg over log interval:  3.868540048599243\n",
            "Epoch:  30 , Batch:  9 , loss avg over log interval:  3.4618932723999025\n",
            "Epoch:  30 , Batch:  19 , loss avg over log interval:  3.850557637214661\n",
            "Epoch:  30 , Batch:  29 , loss avg over log interval:  3.8572919368743896\n",
            "Epoch:  30 , Batch:  39 , loss avg over log interval:  3.848404335975647\n",
            "Epoch:  30 , Batch:  49 , loss avg over log interval:  3.860636019706726\n",
            "Epoch:  30 , Batch:  59 , loss avg over log interval:  3.842578625679016\n",
            "Epoch:  30 , Batch:  69 , loss avg over log interval:  3.8554961919784545\n",
            "Epoch:  30 , Batch:  79 , loss avg over log interval:  3.857632040977478\n",
            "Epoch:  30 , Batch:  89 , loss avg over log interval:  3.8566908121109007\n",
            "Epoch:  30 , Batch:  99 , loss avg over log interval:  3.849174475669861\n",
            "Epoch:  30 , Batch:  109 , loss avg over log interval:  3.8451207876205444\n",
            "Epoch:  30 , Batch:  119 , loss avg over log interval:  3.8503769874572753\n",
            "Epoch:  30 , Batch:  129 , loss avg over log interval:  3.845952105522156\n",
            "Epoch:  30 , avg total loss:  3.8507808205010234\n",
            "iter   0\n",
            "accuracy so far =  0.84375\n",
            "Avg Validation Loss:  3.881182484510468\n",
            "Accuracy:  16.73170731707317\n",
            "Epoch 30/149 Completed\n",
            "Epoch:  31 , Batch:  0 , loss avg over log interval:  3.847158908843994\n",
            "Epoch:  31 , Batch:  9 , loss avg over log interval:  3.4634307622909546\n",
            "Epoch:  31 , Batch:  19 , loss avg over log interval:  3.842671608924866\n",
            "Epoch:  31 , Batch:  29 , loss avg over log interval:  3.845839262008667\n",
            "Epoch:  31 , Batch:  39 , loss avg over log interval:  3.8524887323379517\n",
            "Epoch:  31 , Batch:  49 , loss avg over log interval:  3.8323256254196165\n",
            "Epoch:  31 , Batch:  59 , loss avg over log interval:  3.8413511991500853\n",
            "Epoch:  31 , Batch:  69 , loss avg over log interval:  3.85414514541626\n",
            "Epoch:  31 , Batch:  79 , loss avg over log interval:  3.8468749284744264\n",
            "Epoch:  31 , Batch:  89 , loss avg over log interval:  3.8493139505386353\n",
            "Epoch:  31 , Batch:  99 , loss avg over log interval:  3.855687069892883\n",
            "Epoch:  31 , Batch:  109 , loss avg over log interval:  3.850370264053345\n",
            "Epoch:  31 , Batch:  119 , loss avg over log interval:  3.8441314697265625\n",
            "Epoch:  31 , Batch:  129 , loss avg over log interval:  3.8323004722595213\n",
            "Epoch:  31 , avg total loss:  3.845660182013028\n",
            "iter   0\n",
            "accuracy so far =  0.8515625\n",
            "Avg Validation Loss:  3.8754223148997236\n",
            "Accuracy:  17.70731707317073\n",
            "Epoch 31/149 Completed\n",
            "Epoch:  32 , Batch:  0 , loss avg over log interval:  3.8741049766540527\n",
            "Epoch:  32 , Batch:  9 , loss avg over log interval:  3.466352343559265\n",
            "Epoch:  32 , Batch:  19 , loss avg over log interval:  3.844101309776306\n",
            "Epoch:  32 , Batch:  29 , loss avg over log interval:  3.8534013509750364\n",
            "Epoch:  32 , Batch:  39 , loss avg over log interval:  3.8296268701553347\n",
            "Epoch:  32 , Batch:  49 , loss avg over log interval:  3.8520346164703367\n",
            "Epoch:  32 , Batch:  59 , loss avg over log interval:  3.8261287689208983\n",
            "Epoch:  32 , Batch:  69 , loss avg over log interval:  3.844498348236084\n",
            "Epoch:  32 , Batch:  79 , loss avg over log interval:  3.8550130128860474\n",
            "Epoch:  32 , Batch:  89 , loss avg over log interval:  3.846511149406433\n",
            "Epoch:  32 , Batch:  99 , loss avg over log interval:  3.8412375926971434\n",
            "Epoch:  32 , Batch:  109 , loss avg over log interval:  3.85137677192688\n",
            "Epoch:  32 , Batch:  119 , loss avg over log interval:  3.827072501182556\n",
            "Epoch:  32 , Batch:  129 , loss avg over log interval:  3.8348211526870726\n",
            "Epoch:  32 , avg total loss:  3.8428909225740293\n",
            "iter   0\n",
            "accuracy so far =  0.8359375\n",
            "Avg Validation Loss:  3.8733676061397646\n",
            "Accuracy:  18.121951219512194\n",
            "Epoch 32/149 Completed\n",
            "Epoch:  33 , Batch:  0 , loss avg over log interval:  3.8377702236175537\n",
            "Epoch:  33 , Batch:  9 , loss avg over log interval:  3.4518903255462647\n",
            "Epoch:  33 , Batch:  19 , loss avg over log interval:  3.839626836776733\n",
            "Epoch:  33 , Batch:  29 , loss avg over log interval:  3.8409332036972046\n",
            "Epoch:  33 , Batch:  39 , loss avg over log interval:  3.8401021480560305\n",
            "Epoch:  33 , Batch:  49 , loss avg over log interval:  3.84035177230835\n",
            "Epoch:  33 , Batch:  59 , loss avg over log interval:  3.8445541143417357\n",
            "Epoch:  33 , Batch:  69 , loss avg over log interval:  3.8514461278915406\n",
            "Epoch:  33 , Batch:  79 , loss avg over log interval:  3.8342250108718874\n",
            "Epoch:  33 , Batch:  89 , loss avg over log interval:  3.849765086174011\n",
            "Epoch:  33 , Batch:  99 , loss avg over log interval:  3.85121603012085\n",
            "Epoch:  33 , Batch:  109 , loss avg over log interval:  3.8494100332260133\n",
            "Epoch:  33 , Batch:  119 , loss avg over log interval:  3.835096573829651\n",
            "Epoch:  33 , Batch:  129 , loss avg over log interval:  3.8321250438690186\n",
            "Epoch:  33 , avg total loss:  3.840854211130004\n",
            "iter   0\n",
            "accuracy so far =  0.8515625\n",
            "Avg Validation Loss:  3.8694080259741805\n",
            "Accuracy:  18.365853658536587\n",
            "Epoch 33/149 Completed\n",
            "Epoch:  34 , Batch:  0 , loss avg over log interval:  3.8150997161865234\n",
            "Epoch:  34 , Batch:  9 , loss avg over log interval:  3.455192971229553\n",
            "Epoch:  34 , Batch:  19 , loss avg over log interval:  3.831439423561096\n",
            "Epoch:  34 , Batch:  29 , loss avg over log interval:  3.844494891166687\n",
            "Epoch:  34 , Batch:  39 , loss avg over log interval:  3.8379045724868774\n",
            "Epoch:  34 , Batch:  49 , loss avg over log interval:  3.852388024330139\n",
            "Epoch:  34 , Batch:  59 , loss avg over log interval:  3.8384671449661254\n",
            "Epoch:  34 , Batch:  69 , loss avg over log interval:  3.8495456695556642\n",
            "Epoch:  34 , Batch:  79 , loss avg over log interval:  3.8404475688934325\n",
            "Epoch:  34 , Batch:  89 , loss avg over log interval:  3.8443869590759276\n",
            "Epoch:  34 , Batch:  99 , loss avg over log interval:  3.848030352592468\n",
            "Epoch:  34 , Batch:  109 , loss avg over log interval:  3.8461759090423584\n",
            "Epoch:  34 , Batch:  119 , loss avg over log interval:  3.8409813404083253\n",
            "Epoch:  34 , Batch:  129 , loss avg over log interval:  3.8296611309051514\n",
            "Epoch:  34 , avg total loss:  3.8414411009221836\n",
            "iter   0\n",
            "accuracy so far =  0.828125\n",
            "Avg Validation Loss:  3.870602537946003\n",
            "Accuracy:  18.48780487804878\n",
            "Epoch 34/149 Completed\n",
            "Epoch:  35 , Batch:  0 , loss avg over log interval:  3.8270344734191895\n",
            "Epoch:  35 , Batch:  9 , loss avg over log interval:  3.4628538846969605\n",
            "Epoch:  35 , Batch:  19 , loss avg over log interval:  3.8359440565109253\n",
            "Epoch:  35 , Batch:  29 , loss avg over log interval:  3.8401238918304443\n",
            "Epoch:  35 , Batch:  39 , loss avg over log interval:  3.829535436630249\n",
            "Epoch:  35 , Batch:  49 , loss avg over log interval:  3.8337830305099487\n",
            "Epoch:  35 , Batch:  59 , loss avg over log interval:  3.840006422996521\n",
            "Epoch:  35 , Batch:  69 , loss avg over log interval:  3.8434718370437624\n",
            "Epoch:  35 , Batch:  79 , loss avg over log interval:  3.8391271114349363\n",
            "Epoch:  35 , Batch:  89 , loss avg over log interval:  3.849713897705078\n",
            "Epoch:  35 , Batch:  99 , loss avg over log interval:  3.852213501930237\n",
            "Epoch:  35 , Batch:  109 , loss avg over log interval:  3.8414050817489622\n",
            "Epoch:  35 , Batch:  119 , loss avg over log interval:  3.832626295089722\n",
            "Epoch:  35 , Batch:  129 , loss avg over log interval:  3.8320876121520997\n",
            "Epoch:  35 , avg total loss:  3.839338727619337\n",
            "iter   0\n",
            "accuracy so far =  0.828125\n",
            "Avg Validation Loss:  3.868159840746624\n",
            "Accuracy:  18.951219512195124\n",
            "Epoch 35/149 Completed\n",
            "Epoch:  36 , Batch:  0 , loss avg over log interval:  3.8465828895568848\n",
            "Epoch:  36 , Batch:  9 , loss avg over log interval:  3.447793698310852\n",
            "Epoch:  36 , Batch:  19 , loss avg over log interval:  3.8375839233398437\n",
            "Epoch:  36 , Batch:  29 , loss avg over log interval:  3.8387975692749023\n",
            "Epoch:  36 , Batch:  39 , loss avg over log interval:  3.8435413599014283\n",
            "Epoch:  36 , Batch:  49 , loss avg over log interval:  3.841420292854309\n",
            "Epoch:  36 , Batch:  59 , loss avg over log interval:  3.8297746419906615\n",
            "Epoch:  36 , Batch:  69 , loss avg over log interval:  3.8400138139724733\n",
            "Epoch:  36 , Batch:  79 , loss avg over log interval:  3.840140700340271\n",
            "Epoch:  36 , Batch:  89 , loss avg over log interval:  3.845309686660767\n",
            "Epoch:  36 , Batch:  99 , loss avg over log interval:  3.8513397455215452\n",
            "Epoch:  36 , Batch:  109 , loss avg over log interval:  3.8273129940032957\n",
            "Epoch:  36 , Batch:  119 , loss avg over log interval:  3.828425669670105\n",
            "Epoch:  36 , Batch:  129 , loss avg over log interval:  3.820649838447571\n",
            "Epoch:  36 , avg total loss:  3.8368922713874043\n",
            "iter   0\n",
            "accuracy so far =  0.8046875\n",
            "Avg Validation Loss:  3.8682432174682617\n",
            "Accuracy:  18.4390243902439\n",
            "Epoch 36/149 Completed\n",
            "Epoch:  37 , Batch:  0 , loss avg over log interval:  3.857470989227295\n",
            "Epoch:  37 , Batch:  9 , loss avg over log interval:  3.46374135017395\n",
            "Epoch:  37 , Batch:  19 , loss avg over log interval:  3.8359028100967407\n",
            "Epoch:  37 , Batch:  29 , loss avg over log interval:  3.8370196342468263\n",
            "Epoch:  37 , Batch:  39 , loss avg over log interval:  3.8364935398101805\n",
            "Epoch:  37 , Batch:  49 , loss avg over log interval:  3.8301313400268553\n",
            "Epoch:  37 , Batch:  59 , loss avg over log interval:  3.837199401855469\n",
            "Epoch:  37 , Batch:  69 , loss avg over log interval:  3.8414488792419434\n",
            "Epoch:  37 , Batch:  79 , loss avg over log interval:  3.8485018253326415\n",
            "Epoch:  37 , Batch:  89 , loss avg over log interval:  3.8463247060775756\n",
            "Epoch:  37 , Batch:  99 , loss avg over log interval:  3.8347289085388185\n",
            "Epoch:  37 , Batch:  109 , loss avg over log interval:  3.834844779968262\n",
            "Epoch:  37 , Batch:  119 , loss avg over log interval:  3.8384231090545655\n",
            "Epoch:  37 , Batch:  129 , loss avg over log interval:  3.8196772575378417\n",
            "Epoch:  37 , avg total loss:  3.8375986227090806\n",
            "iter   0\n",
            "accuracy so far =  0.8515625\n",
            "Avg Validation Loss:  3.8618922059128926\n",
            "Accuracy:  18.902439024390244\n",
            "Epoch 37/149 Completed\n",
            "Epoch:  38 , Batch:  0 , loss avg over log interval:  3.855809211730957\n",
            "Epoch:  38 , Batch:  9 , loss avg over log interval:  3.4519244909286497\n",
            "Epoch:  38 , Batch:  19 , loss avg over log interval:  3.8390445709228516\n",
            "Epoch:  38 , Batch:  29 , loss avg over log interval:  3.8391349792480467\n",
            "Epoch:  38 , Batch:  39 , loss avg over log interval:  3.83922803401947\n",
            "Epoch:  38 , Batch:  49 , loss avg over log interval:  3.820225715637207\n",
            "Epoch:  38 , Batch:  59 , loss avg over log interval:  3.8422422885894774\n",
            "Epoch:  38 , Batch:  69 , loss avg over log interval:  3.8466301441192625\n",
            "Epoch:  38 , Batch:  79 , loss avg over log interval:  3.8286773681640627\n",
            "Epoch:  38 , Batch:  89 , loss avg over log interval:  3.846082091331482\n",
            "Epoch:  38 , Batch:  99 , loss avg over log interval:  3.836426615715027\n",
            "Epoch:  38 , Batch:  109 , loss avg over log interval:  3.828855872154236\n",
            "Epoch:  38 , Batch:  119 , loss avg over log interval:  3.828005886077881\n",
            "Epoch:  38 , Batch:  129 , loss avg over log interval:  3.8297345638275146\n",
            "Epoch:  38 , avg total loss:  3.834754066190858\n",
            "iter   0\n",
            "accuracy so far =  0.8203125\n",
            "Avg Validation Loss:  3.8627067193752382\n",
            "Accuracy:  19.390243902439025\n",
            "Epoch 38/149 Completed\n",
            "Epoch:  39 , Batch:  0 , loss avg over log interval:  3.823685646057129\n",
            "Epoch:  39 , Batch:  9 , loss avg over log interval:  3.4495890855789186\n",
            "Epoch:  39 , Batch:  19 , loss avg over log interval:  3.836097574234009\n",
            "Epoch:  39 , Batch:  29 , loss avg over log interval:  3.8432645082473753\n",
            "Epoch:  39 , Batch:  39 , loss avg over log interval:  3.8281341552734376\n",
            "Epoch:  39 , Batch:  49 , loss avg over log interval:  3.8284674644470216\n",
            "Epoch:  39 , Batch:  59 , loss avg over log interval:  3.823194694519043\n",
            "Epoch:  39 , Batch:  69 , loss avg over log interval:  3.82276656627655\n",
            "Epoch:  39 , Batch:  79 , loss avg over log interval:  3.8310982465744017\n",
            "Epoch:  39 , Batch:  89 , loss avg over log interval:  3.8386754274368284\n",
            "Epoch:  39 , Batch:  99 , loss avg over log interval:  3.8406954526901247\n",
            "Epoch:  39 , Batch:  109 , loss avg over log interval:  3.8392733573913573\n",
            "Epoch:  39 , Batch:  119 , loss avg over log interval:  3.8294876337051393\n",
            "Epoch:  39 , Batch:  129 , loss avg over log interval:  3.825176477432251\n",
            "Epoch:  39 , avg total loss:  3.8320661783218384\n",
            "iter   0\n",
            "accuracy so far =  0.8203125\n",
            "Avg Validation Loss:  3.8623024893970026\n",
            "Accuracy:  19.5609756097561\n",
            "Epoch 39/149 Completed\n",
            "Epoch:  40 , Batch:  0 , loss avg over log interval:  3.875897169113159\n",
            "Epoch:  40 , Batch:  9 , loss avg over log interval:  3.4438826560974123\n",
            "Epoch:  40 , Batch:  19 , loss avg over log interval:  3.8409308433532714\n",
            "Epoch:  40 , Batch:  29 , loss avg over log interval:  3.8414693117141723\n",
            "Epoch:  40 , Batch:  39 , loss avg over log interval:  3.8299432754516602\n",
            "Epoch:  40 , Batch:  49 , loss avg over log interval:  3.8310917139053347\n",
            "Epoch:  40 , Batch:  59 , loss avg over log interval:  3.8224810361862183\n",
            "Epoch:  40 , Batch:  69 , loss avg over log interval:  3.8364943265914917\n",
            "Epoch:  40 , Batch:  79 , loss avg over log interval:  3.8340367555618284\n",
            "Epoch:  40 , Batch:  89 , loss avg over log interval:  3.831669545173645\n",
            "Epoch:  40 , Batch:  99 , loss avg over log interval:  3.839314913749695\n",
            "Epoch:  40 , Batch:  109 , loss avg over log interval:  3.835272693634033\n",
            "Epoch:  40 , Batch:  119 , loss avg over log interval:  3.8290834188461305\n",
            "Epoch:  40 , Batch:  129 , loss avg over log interval:  3.8174609422683714\n",
            "Epoch:  40 , avg total loss:  3.831925098446832\n",
            "iter   0\n",
            "accuracy so far =  0.8203125\n",
            "Avg Validation Loss:  3.862891883384891\n",
            "Accuracy:  20.121951219512194\n",
            "Epoch 40/149 Completed\n",
            "Epoch:  41 , Batch:  0 , loss avg over log interval:  3.826216697692871\n",
            "Epoch:  41 , Batch:  9 , loss avg over log interval:  3.4410387516021728\n",
            "Epoch:  41 , Batch:  19 , loss avg over log interval:  3.8275388956069945\n",
            "Epoch:  41 , Batch:  29 , loss avg over log interval:  3.8376301288604737\n",
            "Epoch:  41 , Batch:  39 , loss avg over log interval:  3.834435987472534\n",
            "Epoch:  41 , Batch:  49 , loss avg over log interval:  3.8342682838439943\n",
            "Epoch:  41 , Batch:  59 , loss avg over log interval:  3.8258199214935305\n",
            "Epoch:  41 , Batch:  69 , loss avg over log interval:  3.842555856704712\n",
            "Epoch:  41 , Batch:  79 , loss avg over log interval:  3.8430840253829954\n",
            "Epoch:  41 , Batch:  89 , loss avg over log interval:  3.8223626613616943\n",
            "Epoch:  41 , Batch:  99 , loss avg over log interval:  3.8312485456466674\n",
            "Epoch:  41 , Batch:  109 , loss avg over log interval:  3.8238171339035034\n",
            "Epoch:  41 , Batch:  119 , loss avg over log interval:  3.819542980194092\n",
            "Epoch:  41 , Batch:  129 , loss avg over log interval:  3.823384952545166\n",
            "Epoch:  41 , avg total loss:  3.829391761102538\n",
            "iter   0\n",
            "accuracy so far =  0.8203125\n",
            "Avg Validation Loss:  3.859259570517191\n",
            "Accuracy:  19.51219512195122\n",
            "Epoch 41/149 Completed\n",
            "Epoch:  42 , Batch:  0 , loss avg over log interval:  3.853942632675171\n",
            "Epoch:  42 , Batch:  9 , loss avg over log interval:  3.4506487369537355\n",
            "Epoch:  42 , Batch:  19 , loss avg over log interval:  3.8243676662445067\n",
            "Epoch:  42 , Batch:  29 , loss avg over log interval:  3.8413265228271483\n",
            "Epoch:  42 , Batch:  39 , loss avg over log interval:  3.8337175846099854\n",
            "Epoch:  42 , Batch:  49 , loss avg over log interval:  3.833465242385864\n",
            "Epoch:  42 , Batch:  59 , loss avg over log interval:  3.8224843025207518\n",
            "Epoch:  42 , Batch:  69 , loss avg over log interval:  3.840099406242371\n",
            "Epoch:  42 , Batch:  79 , loss avg over log interval:  3.833517384529114\n",
            "Epoch:  42 , Batch:  89 , loss avg over log interval:  3.8279812812805174\n",
            "Epoch:  42 , Batch:  99 , loss avg over log interval:  3.824413871765137\n",
            "Epoch:  42 , Batch:  109 , loss avg over log interval:  3.8178128242492675\n",
            "Epoch:  42 , Batch:  119 , loss avg over log interval:  3.8179569244384766\n",
            "Epoch:  42 , Batch:  129 , loss avg over log interval:  3.825082778930664\n",
            "Epoch:  42 , avg total loss:  3.8294593735017637\n",
            "iter   0\n",
            "accuracy so far =  0.828125\n",
            "Avg Validation Loss:  3.8572397871715265\n",
            "Accuracy:  18.878048780487806\n",
            "Epoch 42/149 Completed\n",
            "Epoch:  43 , Batch:  0 , loss avg over log interval:  3.8454508781433105\n",
            "Epoch:  43 , Batch:  9 , loss avg over log interval:  3.4440494060516356\n",
            "Epoch:  43 , Batch:  19 , loss avg over log interval:  3.831296944618225\n",
            "Epoch:  43 , Batch:  29 , loss avg over log interval:  3.843601107597351\n",
            "Epoch:  43 , Batch:  39 , loss avg over log interval:  3.8229256868362427\n",
            "Epoch:  43 , Batch:  49 , loss avg over log interval:  3.8262893676757814\n",
            "Epoch:  43 , Batch:  59 , loss avg over log interval:  3.821937155723572\n",
            "Epoch:  43 , Batch:  69 , loss avg over log interval:  3.83033287525177\n",
            "Epoch:  43 , Batch:  79 , loss avg over log interval:  3.8257704734802247\n",
            "Epoch:  43 , Batch:  89 , loss avg over log interval:  3.812973427772522\n",
            "Epoch:  43 , Batch:  99 , loss avg over log interval:  3.833748722076416\n",
            "Epoch:  43 , Batch:  109 , loss avg over log interval:  3.8220985889434815\n",
            "Epoch:  43 , Batch:  119 , loss avg over log interval:  3.8099717140197753\n",
            "Epoch:  43 , Batch:  129 , loss avg over log interval:  3.820305609703064\n",
            "Epoch:  43 , avg total loss:  3.8257587370665176\n",
            "iter   0\n",
            "accuracy so far =  0.8046875\n",
            "Avg Validation Loss:  3.8585759197793355\n",
            "Accuracy:  18.804878048780488\n",
            "Epoch 43/149 Completed\n",
            "Epoch:  44 , Batch:  0 , loss avg over log interval:  3.8329761028289795\n",
            "Epoch:  44 , Batch:  9 , loss avg over log interval:  3.4388898611068726\n",
            "Epoch:  44 , Batch:  19 , loss avg over log interval:  3.8280550718307493\n",
            "Epoch:  44 , Batch:  29 , loss avg over log interval:  3.8216841220855713\n",
            "Epoch:  44 , Batch:  39 , loss avg over log interval:  3.8211730241775514\n",
            "Epoch:  44 , Batch:  49 , loss avg over log interval:  3.826818037033081\n",
            "Epoch:  44 , Batch:  59 , loss avg over log interval:  3.8159511566162108\n",
            "Epoch:  44 , Batch:  69 , loss avg over log interval:  3.8242566347122193\n",
            "Epoch:  44 , Batch:  79 , loss avg over log interval:  3.8320582628250124\n",
            "Epoch:  44 , Batch:  89 , loss avg over log interval:  3.820758008956909\n",
            "Epoch:  44 , Batch:  99 , loss avg over log interval:  3.8308632612228393\n",
            "Epoch:  44 , Batch:  109 , loss avg over log interval:  3.8283063173294067\n",
            "Epoch:  44 , Batch:  119 , loss avg over log interval:  3.821449375152588\n",
            "Epoch:  44 , Batch:  129 , loss avg over log interval:  3.819226050376892\n",
            "Epoch:  44 , avg total loss:  3.82368344673212\n",
            "iter   0\n",
            "accuracy so far =  0.796875\n",
            "Avg Validation Loss:  3.857454753503567\n",
            "Accuracy:  20.463414634146343\n",
            "Epoch 44/149 Completed\n",
            "Epoch:  45 , Batch:  0 , loss avg over log interval:  3.8152964115142822\n",
            "Epoch:  45 , Batch:  9 , loss avg over log interval:  3.4347466468811034\n",
            "Epoch:  45 , Batch:  19 , loss avg over log interval:  3.8179423809051514\n",
            "Epoch:  45 , Batch:  29 , loss avg over log interval:  3.8223666191101073\n",
            "Epoch:  45 , Batch:  39 , loss avg over log interval:  3.818082594871521\n",
            "Epoch:  45 , Batch:  49 , loss avg over log interval:  3.8114038467407227\n",
            "Epoch:  45 , Batch:  59 , loss avg over log interval:  3.8181013107299804\n",
            "Epoch:  45 , Batch:  69 , loss avg over log interval:  3.818143391609192\n",
            "Epoch:  45 , Batch:  79 , loss avg over log interval:  3.8216912031173704\n",
            "Epoch:  45 , Batch:  89 , loss avg over log interval:  3.8343470811843874\n",
            "Epoch:  45 , Batch:  99 , loss avg over log interval:  3.821155381202698\n",
            "Epoch:  45 , Batch:  109 , loss avg over log interval:  3.8136752605438233\n",
            "Epoch:  45 , Batch:  119 , loss avg over log interval:  3.815972995758057\n",
            "Epoch:  45 , Batch:  129 , loss avg over log interval:  3.8032978534698487\n",
            "Epoch:  45 , avg total loss:  3.817322663638903\n",
            "iter   0\n",
            "accuracy so far =  0.7890625\n",
            "Avg Validation Loss:  3.8568031787872314\n",
            "Accuracy:  22.634146341463413\n",
            "Epoch 45/149 Completed\n",
            "Epoch:  46 , Batch:  0 , loss avg over log interval:  3.823629379272461\n",
            "Epoch:  46 , Batch:  9 , loss avg over log interval:  3.430998945236206\n",
            "Epoch:  46 , Batch:  19 , loss avg over log interval:  3.807704758644104\n",
            "Epoch:  46 , Batch:  29 , loss avg over log interval:  3.8231308460235596\n",
            "Epoch:  46 , Batch:  39 , loss avg over log interval:  3.800442695617676\n",
            "Epoch:  46 , Batch:  49 , loss avg over log interval:  3.81709086894989\n",
            "Epoch:  46 , Batch:  59 , loss avg over log interval:  3.816004729270935\n",
            "Epoch:  46 , Batch:  69 , loss avg over log interval:  3.8243016719818117\n",
            "Epoch:  46 , Batch:  79 , loss avg over log interval:  3.815185523033142\n",
            "Epoch:  46 , Batch:  89 , loss avg over log interval:  3.8103238344192505\n",
            "Epoch:  46 , Batch:  99 , loss avg over log interval:  3.8062764406204224\n",
            "Epoch:  46 , Batch:  109 , loss avg over log interval:  3.8083109140396116\n",
            "Epoch:  46 , Batch:  119 , loss avg over log interval:  3.810096025466919\n",
            "Epoch:  46 , Batch:  129 , loss avg over log interval:  3.8125372886657716\n",
            "Epoch:  46 , avg total loss:  3.812256443327752\n",
            "iter   0\n",
            "accuracy so far =  0.71875\n",
            "Avg Validation Loss:  3.850618862524265\n",
            "Accuracy:  23.70731707317073\n",
            "Epoch 46/149 Completed\n",
            "Epoch:  47 , Batch:  0 , loss avg over log interval:  3.8094565868377686\n",
            "Epoch:  47 , Batch:  9 , loss avg over log interval:  3.4227588415145873\n",
            "Epoch:  47 , Batch:  19 , loss avg over log interval:  3.81143319606781\n",
            "Epoch:  47 , Batch:  29 , loss avg over log interval:  3.8183756828308106\n",
            "Epoch:  47 , Batch:  39 , loss avg over log interval:  3.8084576606750487\n",
            "Epoch:  47 , Batch:  49 , loss avg over log interval:  3.812272787094116\n",
            "Epoch:  47 , Batch:  59 , loss avg over log interval:  3.796667218208313\n",
            "Epoch:  47 , Batch:  69 , loss avg over log interval:  3.8150108814239503\n",
            "Epoch:  47 , Batch:  79 , loss avg over log interval:  3.8138503551483156\n",
            "Epoch:  47 , Batch:  89 , loss avg over log interval:  3.8109556913375853\n",
            "Epoch:  47 , Batch:  99 , loss avg over log interval:  3.817198944091797\n",
            "Epoch:  47 , Batch:  109 , loss avg over log interval:  3.81141517162323\n",
            "Epoch:  47 , Batch:  119 , loss avg over log interval:  3.809525561332703\n",
            "Epoch:  47 , Batch:  129 , loss avg over log interval:  3.805418109893799\n",
            "Epoch:  47 , avg total loss:  3.8105919551158296\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.8469892304118085\n",
            "Accuracy:  24.195121951219512\n",
            "Epoch 47/149 Completed\n",
            "Epoch:  48 , Batch:  0 , loss avg over log interval:  3.8029136657714844\n",
            "Epoch:  48 , Batch:  9 , loss avg over log interval:  3.427935171127319\n",
            "Epoch:  48 , Batch:  19 , loss avg over log interval:  3.797163200378418\n",
            "Epoch:  48 , Batch:  29 , loss avg over log interval:  3.8074137210845946\n",
            "Epoch:  48 , Batch:  39 , loss avg over log interval:  3.8195823192596436\n",
            "Epoch:  48 , Batch:  49 , loss avg over log interval:  3.8171860218048095\n",
            "Epoch:  48 , Batch:  59 , loss avg over log interval:  3.795421075820923\n",
            "Epoch:  48 , Batch:  69 , loss avg over log interval:  3.8157342433929444\n",
            "Epoch:  48 , Batch:  79 , loss avg over log interval:  3.803225374221802\n",
            "Epoch:  48 , Batch:  89 , loss avg over log interval:  3.7997761249542235\n",
            "Epoch:  48 , Batch:  99 , loss avg over log interval:  3.8070522785186767\n",
            "Epoch:  48 , Batch:  109 , loss avg over log interval:  3.81316397190094\n",
            "Epoch:  48 , Batch:  119 , loss avg over log interval:  3.7949075222015383\n",
            "Epoch:  48 , Batch:  129 , loss avg over log interval:  3.7967867136001585\n",
            "Epoch:  48 , avg total loss:  3.8057300612546396\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.8404215079982107\n",
            "Accuracy:  25.121951219512194\n",
            "Epoch 48/149 Completed\n",
            "Epoch:  49 , Batch:  0 , loss avg over log interval:  3.810645341873169\n",
            "Epoch:  49 , Batch:  9 , loss avg over log interval:  3.407415246963501\n",
            "Epoch:  49 , Batch:  19 , loss avg over log interval:  3.8064419746398928\n",
            "Epoch:  49 , Batch:  29 , loss avg over log interval:  3.8138449668884276\n",
            "Epoch:  49 , Batch:  39 , loss avg over log interval:  3.798801040649414\n",
            "Epoch:  49 , Batch:  49 , loss avg over log interval:  3.801833486557007\n",
            "Epoch:  49 , Batch:  59 , loss avg over log interval:  3.8027097940444947\n",
            "Epoch:  49 , Batch:  69 , loss avg over log interval:  3.8012677669525146\n",
            "Epoch:  49 , Batch:  79 , loss avg over log interval:  3.807025194168091\n",
            "Epoch:  49 , Batch:  89 , loss avg over log interval:  3.7898629188537596\n",
            "Epoch:  49 , Batch:  99 , loss avg over log interval:  3.806732726097107\n",
            "Epoch:  49 , Batch:  109 , loss avg over log interval:  3.8236403942108153\n",
            "Epoch:  49 , Batch:  119 , loss avg over log interval:  3.784639000892639\n",
            "Epoch:  49 , Batch:  129 , loss avg over log interval:  3.784941840171814\n",
            "Epoch:  49 , avg total loss:  3.8013278090435527\n",
            "iter   0\n",
            "accuracy so far =  0.7109375\n",
            "Avg Validation Loss:  3.835656503351723\n",
            "Accuracy:  25.463414634146343\n",
            "Epoch 49/149 Completed\n",
            "Epoch:  50 , Batch:  0 , loss avg over log interval:  3.7902228832244873\n",
            "Epoch:  50 , Batch:  9 , loss avg over log interval:  3.423662853240967\n",
            "Epoch:  50 , Batch:  19 , loss avg over log interval:  3.7872342348098753\n",
            "Epoch:  50 , Batch:  29 , loss avg over log interval:  3.814732575416565\n",
            "Epoch:  50 , Batch:  39 , loss avg over log interval:  3.8054025888442995\n",
            "Epoch:  50 , Batch:  49 , loss avg over log interval:  3.7929473638534548\n",
            "Epoch:  50 , Batch:  59 , loss avg over log interval:  3.7977297782897947\n",
            "Epoch:  50 , Batch:  69 , loss avg over log interval:  3.804219055175781\n",
            "Epoch:  50 , Batch:  79 , loss avg over log interval:  3.7904797792434692\n",
            "Epoch:  50 , Batch:  89 , loss avg over log interval:  3.8098339557647707\n",
            "Epoch:  50 , Batch:  99 , loss avg over log interval:  3.809191584587097\n",
            "Epoch:  50 , Batch:  109 , loss avg over log interval:  3.8004266738891603\n",
            "Epoch:  50 , Batch:  119 , loss avg over log interval:  3.790537786483765\n",
            "Epoch:  50 , Batch:  129 , loss avg over log interval:  3.7963885068893433\n",
            "Epoch:  50 , avg total loss:  3.800457159678141\n",
            "iter   0\n",
            "accuracy so far =  0.7578125\n",
            "Avg Validation Loss:  3.829072859229111\n",
            "Accuracy:  26.634146341463413\n",
            "Epoch 50/149 Completed\n",
            "Epoch:  51 , Batch:  0 , loss avg over log interval:  3.783449649810791\n",
            "Epoch:  51 , Batch:  9 , loss avg over log interval:  3.404695200920105\n",
            "Epoch:  51 , Batch:  19 , loss avg over log interval:  3.8055888652801513\n",
            "Epoch:  51 , Batch:  29 , loss avg over log interval:  3.8043501377105713\n",
            "Epoch:  51 , Batch:  39 , loss avg over log interval:  3.804208254814148\n",
            "Epoch:  51 , Batch:  49 , loss avg over log interval:  3.7941665410995484\n",
            "Epoch:  51 , Batch:  59 , loss avg over log interval:  3.788767170906067\n",
            "Epoch:  51 , Batch:  69 , loss avg over log interval:  3.8004639387130736\n",
            "Epoch:  51 , Batch:  79 , loss avg over log interval:  3.810342001914978\n",
            "Epoch:  51 , Batch:  89 , loss avg over log interval:  3.806048583984375\n",
            "Epoch:  51 , Batch:  99 , loss avg over log interval:  3.809291219711304\n",
            "Epoch:  51 , Batch:  109 , loss avg over log interval:  3.7971275568008425\n",
            "Epoch:  51 , Batch:  119 , loss avg over log interval:  3.7947102546691895\n",
            "Epoch:  51 , Batch:  129 , loss avg over log interval:  3.7887381315231323\n",
            "Epoch:  51 , avg total loss:  3.7986055906268135\n",
            "iter   0\n",
            "accuracy so far =  0.6484375\n",
            "Avg Validation Loss:  3.8336608700635955\n",
            "Accuracy:  26.4390243902439\n",
            "Epoch 51/149 Completed\n",
            "Epoch:  52 , Batch:  0 , loss avg over log interval:  3.8289647102355957\n",
            "Epoch:  52 , Batch:  9 , loss avg over log interval:  3.4098782300949098\n",
            "Epoch:  52 , Batch:  19 , loss avg over log interval:  3.793032932281494\n",
            "Epoch:  52 , Batch:  29 , loss avg over log interval:  3.79300000667572\n",
            "Epoch:  52 , Batch:  39 , loss avg over log interval:  3.794605183601379\n",
            "Epoch:  52 , Batch:  49 , loss avg over log interval:  3.7987602233886717\n",
            "Epoch:  52 , Batch:  59 , loss avg over log interval:  3.798860192298889\n",
            "Epoch:  52 , Batch:  69 , loss avg over log interval:  3.7867496728897097\n",
            "Epoch:  52 , Batch:  79 , loss avg over log interval:  3.7984963178634645\n",
            "Epoch:  52 , Batch:  89 , loss avg over log interval:  3.7893718481063843\n",
            "Epoch:  52 , Batch:  99 , loss avg over log interval:  3.800402617454529\n",
            "Epoch:  52 , Batch:  109 , loss avg over log interval:  3.799892210960388\n",
            "Epoch:  52 , Batch:  119 , loss avg over log interval:  3.798377823829651\n",
            "Epoch:  52 , Batch:  129 , loss avg over log interval:  3.780906343460083\n",
            "Epoch:  52 , avg total loss:  3.7942193470139434\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.830900831920345\n",
            "Accuracy:  27.73170731707317\n",
            "Epoch 52/149 Completed\n",
            "Epoch:  53 , Batch:  0 , loss avg over log interval:  3.797029733657837\n",
            "Epoch:  53 , Batch:  9 , loss avg over log interval:  3.4040761232376098\n",
            "Epoch:  53 , Batch:  19 , loss avg over log interval:  3.78423535823822\n",
            "Epoch:  53 , Batch:  29 , loss avg over log interval:  3.797934150695801\n",
            "Epoch:  53 , Batch:  39 , loss avg over log interval:  3.7929774284362794\n",
            "Epoch:  53 , Batch:  49 , loss avg over log interval:  3.793296790122986\n",
            "Epoch:  53 , Batch:  59 , loss avg over log interval:  3.8009429216384887\n",
            "Epoch:  53 , Batch:  69 , loss avg over log interval:  3.79642596244812\n",
            "Epoch:  53 , Batch:  79 , loss avg over log interval:  3.8044331550598143\n",
            "Epoch:  53 , Batch:  89 , loss avg over log interval:  3.7920074462890625\n",
            "Epoch:  53 , Batch:  99 , loss avg over log interval:  3.8026177883148193\n",
            "Epoch:  53 , Batch:  109 , loss avg over log interval:  3.7887520790100098\n",
            "Epoch:  53 , Batch:  119 , loss avg over log interval:  3.788795304298401\n",
            "Epoch:  53 , Batch:  129 , loss avg over log interval:  3.77504301071167\n",
            "Epoch:  53 , avg total loss:  3.792468271393707\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.831081849772756\n",
            "Accuracy:  28.146341463414632\n",
            "Epoch 53/149 Completed\n",
            "Epoch:  54 , Batch:  0 , loss avg over log interval:  3.8135054111480713\n",
            "Epoch:  54 , Batch:  9 , loss avg over log interval:  3.4085666656494142\n",
            "Epoch:  54 , Batch:  19 , loss avg over log interval:  3.792601990699768\n",
            "Epoch:  54 , Batch:  29 , loss avg over log interval:  3.7968727111816407\n",
            "Epoch:  54 , Batch:  39 , loss avg over log interval:  3.788496208190918\n",
            "Epoch:  54 , Batch:  49 , loss avg over log interval:  3.7894808769226076\n",
            "Epoch:  54 , Batch:  59 , loss avg over log interval:  3.7833666086196898\n",
            "Epoch:  54 , Batch:  69 , loss avg over log interval:  3.7932881355285644\n",
            "Epoch:  54 , Batch:  79 , loss avg over log interval:  3.7923869609832765\n",
            "Epoch:  54 , Batch:  89 , loss avg over log interval:  3.801116871833801\n",
            "Epoch:  54 , Batch:  99 , loss avg over log interval:  3.802905035018921\n",
            "Epoch:  54 , Batch:  109 , loss avg over log interval:  3.793819856643677\n",
            "Epoch:  54 , Batch:  119 , loss avg over log interval:  3.7803993701934813\n",
            "Epoch:  54 , Batch:  129 , loss avg over log interval:  3.7859153270721437\n",
            "Epoch:  54 , avg total loss:  3.791486683099166\n",
            "iter   0\n",
            "accuracy so far =  0.7578125\n",
            "Avg Validation Loss:  3.8203661790708217\n",
            "Accuracy:  29.121951219512194\n",
            "Epoch 54/149 Completed\n",
            "Epoch:  55 , Batch:  0 , loss avg over log interval:  3.832761526107788\n",
            "Epoch:  55 , Batch:  9 , loss avg over log interval:  3.3935627222061155\n",
            "Epoch:  55 , Batch:  19 , loss avg over log interval:  3.7831923246383665\n",
            "Epoch:  55 , Batch:  29 , loss avg over log interval:  3.808949661254883\n",
            "Epoch:  55 , Batch:  39 , loss avg over log interval:  3.7831151485443115\n",
            "Epoch:  55 , Batch:  49 , loss avg over log interval:  3.7798532247543335\n",
            "Epoch:  55 , Batch:  59 , loss avg over log interval:  3.7821158885955812\n",
            "Epoch:  55 , Batch:  69 , loss avg over log interval:  3.806193161010742\n",
            "Epoch:  55 , Batch:  79 , loss avg over log interval:  3.789427065849304\n",
            "Epoch:  55 , Batch:  89 , loss avg over log interval:  3.7772806882858276\n",
            "Epoch:  55 , Batch:  99 , loss avg over log interval:  3.797088289260864\n",
            "Epoch:  55 , Batch:  109 , loss avg over log interval:  3.7751696825027468\n",
            "Epoch:  55 , Batch:  119 , loss avg over log interval:  3.7737797260284425\n",
            "Epoch:  55 , Batch:  129 , loss avg over log interval:  3.7734559774398804\n",
            "Epoch:  55 , avg total loss:  3.7852908182835234\n",
            "iter   0\n",
            "accuracy so far =  0.8046875\n",
            "Avg Validation Loss:  3.816689909958258\n",
            "Accuracy:  29.097560975609756\n",
            "Epoch 55/149 Completed\n",
            "Epoch:  56 , Batch:  0 , loss avg over log interval:  3.7575321197509766\n",
            "Epoch:  56 , Batch:  9 , loss avg over log interval:  3.3997642278671263\n",
            "Epoch:  56 , Batch:  19 , loss avg over log interval:  3.7788779973983764\n",
            "Epoch:  56 , Batch:  29 , loss avg over log interval:  3.7946609973907472\n",
            "Epoch:  56 , Batch:  39 , loss avg over log interval:  3.7777966499328612\n",
            "Epoch:  56 , Batch:  49 , loss avg over log interval:  3.7942963600158692\n",
            "Epoch:  56 , Batch:  59 , loss avg over log interval:  3.7895055532455446\n",
            "Epoch:  56 , Batch:  69 , loss avg over log interval:  3.792677617073059\n",
            "Epoch:  56 , Batch:  79 , loss avg over log interval:  3.79865882396698\n",
            "Epoch:  56 , Batch:  89 , loss avg over log interval:  3.792730712890625\n",
            "Epoch:  56 , Batch:  99 , loss avg over log interval:  3.7977395057678223\n",
            "Epoch:  56 , Batch:  109 , loss avg over log interval:  3.7951138019561768\n",
            "Epoch:  56 , Batch:  119 , loss avg over log interval:  3.7742910623550414\n",
            "Epoch:  56 , Batch:  129 , loss avg over log interval:  3.785758709907532\n",
            "Epoch:  56 , avg total loss:  3.788089116414388\n",
            "iter   0\n",
            "accuracy so far =  0.7734375\n",
            "Avg Validation Loss:  3.815690401123791\n",
            "Accuracy:  29.170731707317074\n",
            "Epoch 56/149 Completed\n",
            "Epoch:  57 , Batch:  0 , loss avg over log interval:  3.8134145736694336\n",
            "Epoch:  57 , Batch:  9 , loss avg over log interval:  3.3998557806015013\n",
            "Epoch:  57 , Batch:  19 , loss avg over log interval:  3.7948375701904298\n",
            "Epoch:  57 , Batch:  29 , loss avg over log interval:  3.794742751121521\n",
            "Epoch:  57 , Batch:  39 , loss avg over log interval:  3.794716811180115\n",
            "Epoch:  57 , Batch:  49 , loss avg over log interval:  3.780900239944458\n",
            "Epoch:  57 , Batch:  59 , loss avg over log interval:  3.7877863883972167\n",
            "Epoch:  57 , Batch:  69 , loss avg over log interval:  3.7827226638793947\n",
            "Epoch:  57 , Batch:  79 , loss avg over log interval:  3.7944110870361327\n",
            "Epoch:  57 , Batch:  89 , loss avg over log interval:  3.7915900707244874\n",
            "Epoch:  57 , Batch:  99 , loss avg over log interval:  3.808055353164673\n",
            "Epoch:  57 , Batch:  109 , loss avg over log interval:  3.7780010223388674\n",
            "Epoch:  57 , Batch:  119 , loss avg over log interval:  3.7631022453308107\n",
            "Epoch:  57 , Batch:  129 , loss avg over log interval:  3.7682403087615968\n",
            "Epoch:  57 , avg total loss:  3.7862323557121167\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.816727737101113\n",
            "Accuracy:  28.951219512195124\n",
            "Epoch 57/149 Completed\n",
            "Epoch:  58 , Batch:  0 , loss avg over log interval:  3.810012102127075\n",
            "Epoch:  58 , Batch:  9 , loss avg over log interval:  3.3947513103485107\n",
            "Epoch:  58 , Batch:  19 , loss avg over log interval:  3.7762568473815916\n",
            "Epoch:  58 , Batch:  29 , loss avg over log interval:  3.795045590400696\n",
            "Epoch:  58 , Batch:  39 , loss avg over log interval:  3.791133499145508\n",
            "Epoch:  58 , Batch:  49 , loss avg over log interval:  3.77921986579895\n",
            "Epoch:  58 , Batch:  59 , loss avg over log interval:  3.788875961303711\n",
            "Epoch:  58 , Batch:  69 , loss avg over log interval:  3.7805517435073854\n",
            "Epoch:  58 , Batch:  79 , loss avg over log interval:  3.784524393081665\n",
            "Epoch:  58 , Batch:  89 , loss avg over log interval:  3.7840778112411497\n",
            "Epoch:  58 , Batch:  99 , loss avg over log interval:  3.786103343963623\n",
            "Epoch:  58 , Batch:  109 , loss avg over log interval:  3.7798344612121584\n",
            "Epoch:  58 , Batch:  119 , loss avg over log interval:  3.764656114578247\n",
            "Epoch:  58 , Batch:  129 , loss avg over log interval:  3.7675743818283083\n",
            "Epoch:  58 , avg total loss:  3.7813361503075864\n",
            "iter   0\n",
            "accuracy so far =  0.78125\n",
            "Avg Validation Loss:  3.810706045569443\n",
            "Accuracy:  29.902439024390244\n",
            "Epoch 58/149 Completed\n",
            "Epoch:  59 , Batch:  0 , loss avg over log interval:  3.7668981552124023\n",
            "Epoch:  59 , Batch:  9 , loss avg over log interval:  3.4142153739929197\n",
            "Epoch:  59 , Batch:  19 , loss avg over log interval:  3.7698545932769774\n",
            "Epoch:  59 , Batch:  29 , loss avg over log interval:  3.776628112792969\n",
            "Epoch:  59 , Batch:  39 , loss avg over log interval:  3.77407488822937\n",
            "Epoch:  59 , Batch:  49 , loss avg over log interval:  3.787139964103699\n",
            "Epoch:  59 , Batch:  59 , loss avg over log interval:  3.7732505083084105\n",
            "Epoch:  59 , Batch:  69 , loss avg over log interval:  3.7773280143737793\n",
            "Epoch:  59 , Batch:  79 , loss avg over log interval:  3.782122278213501\n",
            "Epoch:  59 , Batch:  89 , loss avg over log interval:  3.794293451309204\n",
            "Epoch:  59 , Batch:  99 , loss avg over log interval:  3.7852561712265014\n",
            "Epoch:  59 , Batch:  109 , loss avg over log interval:  3.797808122634888\n",
            "Epoch:  59 , Batch:  119 , loss avg over log interval:  3.7688494443893434\n",
            "Epoch:  59 , Batch:  129 , loss avg over log interval:  3.772357726097107\n",
            "Epoch:  59 , avg total loss:  3.780676829642144\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.8102288478758277\n",
            "Accuracy:  29.975609756097562\n",
            "Epoch 59/149 Completed\n",
            "Epoch:  60 , Batch:  0 , loss avg over log interval:  3.773937940597534\n",
            "Epoch:  60 , Batch:  9 , loss avg over log interval:  3.3979009866714476\n",
            "Epoch:  60 , Batch:  19 , loss avg over log interval:  3.771488904953003\n",
            "Epoch:  60 , Batch:  29 , loss avg over log interval:  3.7727068185806276\n",
            "Epoch:  60 , Batch:  39 , loss avg over log interval:  3.7906378746032714\n",
            "Epoch:  60 , Batch:  49 , loss avg over log interval:  3.7774550914764404\n",
            "Epoch:  60 , Batch:  59 , loss avg over log interval:  3.7745558500289915\n",
            "Epoch:  60 , Batch:  69 , loss avg over log interval:  3.783234429359436\n",
            "Epoch:  60 , Batch:  79 , loss avg over log interval:  3.77209849357605\n",
            "Epoch:  60 , Batch:  89 , loss avg over log interval:  3.785241389274597\n",
            "Epoch:  60 , Batch:  99 , loss avg over log interval:  3.781002569198608\n",
            "Epoch:  60 , Batch:  109 , loss avg over log interval:  3.7797368288040163\n",
            "Epoch:  60 , Batch:  119 , loss avg over log interval:  3.772861194610596\n",
            "Epoch:  60 , Batch:  129 , loss avg over log interval:  3.7616411447525024\n",
            "Epoch:  60 , avg total loss:  3.776057949964551\n",
            "iter   0\n",
            "accuracy so far =  0.671875\n",
            "Avg Validation Loss:  3.8103929438242097\n",
            "Accuracy:  30.26829268292683\n",
            "Epoch 60/149 Completed\n",
            "Epoch:  61 , Batch:  0 , loss avg over log interval:  3.761568784713745\n",
            "Epoch:  61 , Batch:  9 , loss avg over log interval:  3.3890222549438476\n",
            "Epoch:  61 , Batch:  19 , loss avg over log interval:  3.7762922048568726\n",
            "Epoch:  61 , Batch:  29 , loss avg over log interval:  3.7728445768356322\n",
            "Epoch:  61 , Batch:  39 , loss avg over log interval:  3.772877740859985\n",
            "Epoch:  61 , Batch:  49 , loss avg over log interval:  3.781950092315674\n",
            "Epoch:  61 , Batch:  59 , loss avg over log interval:  3.7738497495651244\n",
            "Epoch:  61 , Batch:  69 , loss avg over log interval:  3.779574680328369\n",
            "Epoch:  61 , Batch:  79 , loss avg over log interval:  3.7881535053253175\n",
            "Epoch:  61 , Batch:  89 , loss avg over log interval:  3.7788416385650634\n",
            "Epoch:  61 , Batch:  99 , loss avg over log interval:  3.787018632888794\n",
            "Epoch:  61 , Batch:  109 , loss avg over log interval:  3.7852322816848756\n",
            "Epoch:  61 , Batch:  119 , loss avg over log interval:  3.752941942214966\n",
            "Epoch:  61 , Batch:  129 , loss avg over log interval:  3.772085404396057\n",
            "Epoch:  61 , avg total loss:  3.775686122369075\n",
            "iter   0\n",
            "accuracy so far =  0.734375\n",
            "Avg Validation Loss:  3.80382470968293\n",
            "Accuracy:  31.24390243902439\n",
            "Epoch 61/149 Completed\n",
            "Epoch:  62 , Batch:  0 , loss avg over log interval:  3.8096225261688232\n",
            "Epoch:  62 , Batch:  9 , loss avg over log interval:  3.3917739152908326\n",
            "Epoch:  62 , Batch:  19 , loss avg over log interval:  3.7628790140151978\n",
            "Epoch:  62 , Batch:  29 , loss avg over log interval:  3.783306121826172\n",
            "Epoch:  62 , Batch:  39 , loss avg over log interval:  3.7607694387435915\n",
            "Epoch:  62 , Batch:  49 , loss avg over log interval:  3.7600208282470704\n",
            "Epoch:  62 , Batch:  59 , loss avg over log interval:  3.7819490909576414\n",
            "Epoch:  62 , Batch:  69 , loss avg over log interval:  3.758675980567932\n",
            "Epoch:  62 , Batch:  79 , loss avg over log interval:  3.7787473440170287\n",
            "Epoch:  62 , Batch:  89 , loss avg over log interval:  3.7754457712173464\n",
            "Epoch:  62 , Batch:  99 , loss avg over log interval:  3.785892701148987\n",
            "Epoch:  62 , Batch:  109 , loss avg over log interval:  3.7669031620025635\n",
            "Epoch:  62 , Batch:  119 , loss avg over log interval:  3.758151078224182\n",
            "Epoch:  62 , Batch:  129 , loss avg over log interval:  3.7594648361206056\n",
            "Epoch:  62 , avg total loss:  3.769287399623705\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.8023635643284495\n",
            "Accuracy:  31.585365853658537\n",
            "Epoch 62/149 Completed\n",
            "Epoch:  63 , Batch:  0 , loss avg over log interval:  3.7677078247070312\n",
            "Epoch:  63 , Batch:  9 , loss avg over log interval:  3.3777207851409914\n",
            "Epoch:  63 , Batch:  19 , loss avg over log interval:  3.766067123413086\n",
            "Epoch:  63 , Batch:  29 , loss avg over log interval:  3.7737658262252807\n",
            "Epoch:  63 , Batch:  39 , loss avg over log interval:  3.760746788978577\n",
            "Epoch:  63 , Batch:  49 , loss avg over log interval:  3.764096713066101\n",
            "Epoch:  63 , Batch:  59 , loss avg over log interval:  3.7653951168060305\n",
            "Epoch:  63 , Batch:  69 , loss avg over log interval:  3.7727164745330812\n",
            "Epoch:  63 , Batch:  79 , loss avg over log interval:  3.7894865036010743\n",
            "Epoch:  63 , Batch:  89 , loss avg over log interval:  3.7679028511047363\n",
            "Epoch:  63 , Batch:  99 , loss avg over log interval:  3.7753118515014648\n",
            "Epoch:  63 , Batch:  109 , loss avg over log interval:  3.785192275047302\n",
            "Epoch:  63 , Batch:  119 , loss avg over log interval:  3.7619102954864503\n",
            "Epoch:  63 , Batch:  129 , loss avg over log interval:  3.763383722305298\n",
            "Epoch:  63 , avg total loss:  3.7688797781432886\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.802199881251265\n",
            "Accuracy:  31.585365853658537\n",
            "Epoch 63/149 Completed\n",
            "Epoch:  64 , Batch:  0 , loss avg over log interval:  3.7985785007476807\n",
            "Epoch:  64 , Batch:  9 , loss avg over log interval:  3.3980011701583863\n",
            "Epoch:  64 , Batch:  19 , loss avg over log interval:  3.7503844022750856\n",
            "Epoch:  64 , Batch:  29 , loss avg over log interval:  3.7608341693878176\n",
            "Epoch:  64 , Batch:  39 , loss avg over log interval:  3.774882197380066\n",
            "Epoch:  64 , Batch:  49 , loss avg over log interval:  3.7706145286560058\n",
            "Epoch:  64 , Batch:  59 , loss avg over log interval:  3.7696821689605713\n",
            "Epoch:  64 , Batch:  69 , loss avg over log interval:  3.774600863456726\n",
            "Epoch:  64 , Batch:  79 , loss avg over log interval:  3.7878960609436034\n",
            "Epoch:  64 , Batch:  89 , loss avg over log interval:  3.7762479543685914\n",
            "Epoch:  64 , Batch:  99 , loss avg over log interval:  3.7811579942703246\n",
            "Epoch:  64 , Batch:  109 , loss avg over log interval:  3.775659132003784\n",
            "Epoch:  64 , Batch:  119 , loss avg over log interval:  3.7491430759429933\n",
            "Epoch:  64 , Batch:  129 , loss avg over log interval:  3.7555903434753417\n",
            "Epoch:  64 , avg total loss:  3.769140604613484\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.7998778645585225\n",
            "Accuracy:  31.121951219512194\n",
            "Epoch 64/149 Completed\n",
            "Epoch:  65 , Batch:  0 , loss avg over log interval:  3.7654035091400146\n",
            "Epoch:  65 , Batch:  9 , loss avg over log interval:  3.379447913169861\n",
            "Epoch:  65 , Batch:  19 , loss avg over log interval:  3.7606938600540163\n",
            "Epoch:  65 , Batch:  29 , loss avg over log interval:  3.7681746006011965\n",
            "Epoch:  65 , Batch:  39 , loss avg over log interval:  3.775985288619995\n",
            "Epoch:  65 , Batch:  49 , loss avg over log interval:  3.7547371864318846\n",
            "Epoch:  65 , Batch:  59 , loss avg over log interval:  3.774215078353882\n",
            "Epoch:  65 , Batch:  69 , loss avg over log interval:  3.7872684717178347\n",
            "Epoch:  65 , Batch:  79 , loss avg over log interval:  3.7838602542877195\n",
            "Epoch:  65 , Batch:  89 , loss avg over log interval:  3.7727218151092528\n",
            "Epoch:  65 , Batch:  99 , loss avg over log interval:  3.766280460357666\n",
            "Epoch:  65 , Batch:  109 , loss avg over log interval:  3.7788512229919435\n",
            "Epoch:  65 , Batch:  119 , loss avg over log interval:  3.762671637535095\n",
            "Epoch:  65 , Batch:  129 , loss avg over log interval:  3.7470817804336547\n",
            "Epoch:  65 , avg total loss:  3.7685754834741787\n",
            "iter   0\n",
            "accuracy so far =  0.6328125\n",
            "Avg Validation Loss:  3.8005875029214997\n",
            "Accuracy:  31.341463414634145\n",
            "Epoch 65/149 Completed\n",
            "Epoch:  66 , Batch:  0 , loss avg over log interval:  3.7707102298736572\n",
            "Epoch:  66 , Batch:  9 , loss avg over log interval:  3.3824668884277345\n",
            "Epoch:  66 , Batch:  19 , loss avg over log interval:  3.7649617195129395\n",
            "Epoch:  66 , Batch:  29 , loss avg over log interval:  3.7715540647506716\n",
            "Epoch:  66 , Batch:  39 , loss avg over log interval:  3.780393123626709\n",
            "Epoch:  66 , Batch:  49 , loss avg over log interval:  3.762703800201416\n",
            "Epoch:  66 , Batch:  59 , loss avg over log interval:  3.770591688156128\n",
            "Epoch:  66 , Batch:  69 , loss avg over log interval:  3.765868663787842\n",
            "Epoch:  66 , Batch:  79 , loss avg over log interval:  3.7585351705551147\n",
            "Epoch:  66 , Batch:  89 , loss avg over log interval:  3.7680707931518556\n",
            "Epoch:  66 , Batch:  99 , loss avg over log interval:  3.7754844665527343\n",
            "Epoch:  66 , Batch:  109 , loss avg over log interval:  3.7884278774261473\n",
            "Epoch:  66 , Batch:  119 , loss avg over log interval:  3.7459969758987426\n",
            "Epoch:  66 , Batch:  129 , loss avg over log interval:  3.7474494218826293\n",
            "Epoch:  66 , avg total loss:  3.7655761483786763\n",
            "iter   0\n",
            "accuracy so far =  0.6484375\n",
            "Avg Validation Loss:  3.797446948726003\n",
            "Accuracy:  31.829268292682926\n",
            "Epoch 66/149 Completed\n",
            "Epoch:  67 , Batch:  0 , loss avg over log interval:  3.7528064250946045\n",
            "Epoch:  67 , Batch:  9 , loss avg over log interval:  3.377817726135254\n",
            "Epoch:  67 , Batch:  19 , loss avg over log interval:  3.7592417001724243\n",
            "Epoch:  67 , Batch:  29 , loss avg over log interval:  3.7596943855285643\n",
            "Epoch:  67 , Batch:  39 , loss avg over log interval:  3.765670657157898\n",
            "Epoch:  67 , Batch:  49 , loss avg over log interval:  3.7632442712783813\n",
            "Epoch:  67 , Batch:  59 , loss avg over log interval:  3.7695739030838014\n",
            "Epoch:  67 , Batch:  69 , loss avg over log interval:  3.7623161554336546\n",
            "Epoch:  67 , Batch:  79 , loss avg over log interval:  3.7698353052139284\n",
            "Epoch:  67 , Batch:  89 , loss avg over log interval:  3.767516255378723\n",
            "Epoch:  67 , Batch:  99 , loss avg over log interval:  3.782942295074463\n",
            "Epoch:  67 , Batch:  109 , loss avg over log interval:  3.7656953811645506\n",
            "Epoch:  67 , Batch:  119 , loss avg over log interval:  3.747001314163208\n",
            "Epoch:  67 , Batch:  129 , loss avg over log interval:  3.752324652671814\n",
            "Epoch:  67 , avg total loss:  3.7628424996915073\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.792800438113329\n",
            "Accuracy:  32.19512195121951\n",
            "Epoch 67/149 Completed\n",
            "Epoch:  68 , Batch:  0 , loss avg over log interval:  3.7710015773773193\n",
            "Epoch:  68 , Batch:  9 , loss avg over log interval:  3.3787450790405273\n",
            "Epoch:  68 , Batch:  19 , loss avg over log interval:  3.749557542800903\n",
            "Epoch:  68 , Batch:  29 , loss avg over log interval:  3.745248031616211\n",
            "Epoch:  68 , Batch:  39 , loss avg over log interval:  3.7523290395736693\n",
            "Epoch:  68 , Batch:  49 , loss avg over log interval:  3.76190505027771\n",
            "Epoch:  68 , Batch:  59 , loss avg over log interval:  3.7676992177963258\n",
            "Epoch:  68 , Batch:  69 , loss avg over log interval:  3.7622775077819823\n",
            "Epoch:  68 , Batch:  79 , loss avg over log interval:  3.780786919593811\n",
            "Epoch:  68 , Batch:  89 , loss avg over log interval:  3.770861864089966\n",
            "Epoch:  68 , Batch:  99 , loss avg over log interval:  3.775348114967346\n",
            "Epoch:  68 , Batch:  109 , loss avg over log interval:  3.7739007472991943\n",
            "Epoch:  68 , Batch:  119 , loss avg over log interval:  3.7424761772155763\n",
            "Epoch:  68 , Batch:  129 , loss avg over log interval:  3.7609411478042603\n",
            "Epoch:  68 , avg total loss:  3.7622653090435527\n",
            "iter   0\n",
            "accuracy so far =  0.7265625\n",
            "Avg Validation Loss:  3.7918365408734576\n",
            "Accuracy:  32.8780487804878\n",
            "Epoch 68/149 Completed\n",
            "Epoch:  69 , Batch:  0 , loss avg over log interval:  3.754086494445801\n",
            "Epoch:  69 , Batch:  9 , loss avg over log interval:  3.3809967041015625\n",
            "Epoch:  69 , Batch:  19 , loss avg over log interval:  3.7509006023406983\n",
            "Epoch:  69 , Batch:  29 , loss avg over log interval:  3.759441447257996\n",
            "Epoch:  69 , Batch:  39 , loss avg over log interval:  3.761912202835083\n",
            "Epoch:  69 , Batch:  49 , loss avg over log interval:  3.7644720792770388\n",
            "Epoch:  69 , Batch:  59 , loss avg over log interval:  3.7681071758270264\n",
            "Epoch:  69 , Batch:  69 , loss avg over log interval:  3.7747433185577393\n",
            "Epoch:  69 , Batch:  79 , loss avg over log interval:  3.763318347930908\n",
            "Epoch:  69 , Batch:  89 , loss avg over log interval:  3.7772045135498047\n",
            "Epoch:  69 , Batch:  99 , loss avg over log interval:  3.7762260675430297\n",
            "Epoch:  69 , Batch:  109 , loss avg over log interval:  3.7699175834655763\n",
            "Epoch:  69 , Batch:  119 , loss avg over log interval:  3.7436697244644166\n",
            "Epoch:  69 , Batch:  129 , loss avg over log interval:  3.761302638053894\n",
            "Epoch:  69 , avg total loss:  3.7639795317166094\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.793037211022726\n",
            "Accuracy:  32.4390243902439\n",
            "Epoch 69/149 Completed\n",
            "Epoch:  70 , Batch:  0 , loss avg over log interval:  3.731224536895752\n",
            "Epoch:  70 , Batch:  9 , loss avg over log interval:  3.3852144479751587\n",
            "Epoch:  70 , Batch:  19 , loss avg over log interval:  3.7570715665817263\n",
            "Epoch:  70 , Batch:  29 , loss avg over log interval:  3.7670847177505493\n",
            "Epoch:  70 , Batch:  39 , loss avg over log interval:  3.768260431289673\n",
            "Epoch:  70 , Batch:  49 , loss avg over log interval:  3.745741868019104\n",
            "Epoch:  70 , Batch:  59 , loss avg over log interval:  3.7682029008865356\n",
            "Epoch:  70 , Batch:  69 , loss avg over log interval:  3.7634119510650637\n",
            "Epoch:  70 , Batch:  79 , loss avg over log interval:  3.7599940061569215\n",
            "Epoch:  70 , Batch:  89 , loss avg over log interval:  3.7588682889938356\n",
            "Epoch:  70 , Batch:  99 , loss avg over log interval:  3.7793077707290648\n",
            "Epoch:  70 , Batch:  109 , loss avg over log interval:  3.762038159370422\n",
            "Epoch:  70 , Batch:  119 , loss avg over log interval:  3.7316619396209716\n",
            "Epoch:  70 , Batch:  129 , loss avg over log interval:  3.75407497882843\n",
            "Epoch:  70 , avg total loss:  3.75982611939527\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.794693161801594\n",
            "Accuracy:  32.951219512195124\n",
            "Epoch 70/149 Completed\n",
            "Epoch:  71 , Batch:  0 , loss avg over log interval:  3.7750353813171387\n",
            "Epoch:  71 , Batch:  9 , loss avg over log interval:  3.3708482503890993\n",
            "Epoch:  71 , Batch:  19 , loss avg over log interval:  3.7447383880615233\n",
            "Epoch:  71 , Batch:  29 , loss avg over log interval:  3.7761967182159424\n",
            "Epoch:  71 , Batch:  39 , loss avg over log interval:  3.7475342750549316\n",
            "Epoch:  71 , Batch:  49 , loss avg over log interval:  3.745710349082947\n",
            "Epoch:  71 , Batch:  59 , loss avg over log interval:  3.7488223552703857\n",
            "Epoch:  71 , Batch:  69 , loss avg over log interval:  3.7761035203933715\n",
            "Epoch:  71 , Batch:  79 , loss avg over log interval:  3.767703151702881\n",
            "Epoch:  71 , Batch:  89 , loss avg over log interval:  3.762519931793213\n",
            "Epoch:  71 , Batch:  99 , loss avg over log interval:  3.772795057296753\n",
            "Epoch:  71 , Batch:  109 , loss avg over log interval:  3.763375401496887\n",
            "Epoch:  71 , Batch:  119 , loss avg over log interval:  3.7289725065231325\n",
            "Epoch:  71 , Batch:  129 , loss avg over log interval:  3.7263174057006836\n",
            "Epoch:  71 , avg total loss:  3.7555380852326103\n",
            "iter   0\n",
            "accuracy so far =  0.6875\n",
            "Avg Validation Loss:  3.7884155482780644\n",
            "Accuracy:  33.19512195121951\n",
            "Epoch 71/149 Completed\n",
            "Epoch:  72 , Batch:  0 , loss avg over log interval:  3.759833574295044\n",
            "Epoch:  72 , Batch:  9 , loss avg over log interval:  3.3722891330718996\n",
            "Epoch:  72 , Batch:  19 , loss avg over log interval:  3.741124677658081\n",
            "Epoch:  72 , Batch:  29 , loss avg over log interval:  3.757963228225708\n",
            "Epoch:  72 , Batch:  39 , loss avg over log interval:  3.751763844490051\n",
            "Epoch:  72 , Batch:  49 , loss avg over log interval:  3.742680287361145\n",
            "Epoch:  72 , Batch:  59 , loss avg over log interval:  3.7448375940322878\n",
            "Epoch:  72 , Batch:  69 , loss avg over log interval:  3.753369116783142\n",
            "Epoch:  72 , Batch:  79 , loss avg over log interval:  3.7625675439834594\n",
            "Epoch:  72 , Batch:  89 , loss avg over log interval:  3.761654090881348\n",
            "Epoch:  72 , Batch:  99 , loss avg over log interval:  3.7589829206466674\n",
            "Epoch:  72 , Batch:  109 , loss avg over log interval:  3.7581417560577393\n",
            "Epoch:  72 , Batch:  119 , loss avg over log interval:  3.732166075706482\n",
            "Epoch:  72 , Batch:  129 , loss avg over log interval:  3.7510029077529907\n",
            "Epoch:  72 , avg total loss:  3.7504158227340034\n",
            "iter   0\n",
            "accuracy so far =  0.65625\n",
            "Avg Validation Loss:  3.789374467803211\n",
            "Accuracy:  33.75609756097561\n",
            "Epoch 72/149 Completed\n",
            "Epoch:  73 , Batch:  0 , loss avg over log interval:  3.7929399013519287\n",
            "Epoch:  73 , Batch:  9 , loss avg over log interval:  3.3741504669189455\n",
            "Epoch:  73 , Batch:  19 , loss avg over log interval:  3.7485953092575075\n",
            "Epoch:  73 , Batch:  29 , loss avg over log interval:  3.759658932685852\n",
            "Epoch:  73 , Batch:  39 , loss avg over log interval:  3.745634388923645\n",
            "Epoch:  73 , Batch:  49 , loss avg over log interval:  3.7474233627319338\n",
            "Epoch:  73 , Batch:  59 , loss avg over log interval:  3.7462756872177123\n",
            "Epoch:  73 , Batch:  69 , loss avg over log interval:  3.7590073108673097\n",
            "Epoch:  73 , Batch:  79 , loss avg over log interval:  3.7674681186676025\n",
            "Epoch:  73 , Batch:  89 , loss avg over log interval:  3.757688021659851\n",
            "Epoch:  73 , Batch:  99 , loss avg over log interval:  3.7663783550262453\n",
            "Epoch:  73 , Batch:  109 , loss avg over log interval:  3.754123497009277\n",
            "Epoch:  73 , Batch:  119 , loss avg over log interval:  3.7490508794784545\n",
            "Epoch:  73 , Batch:  129 , loss avg over log interval:  3.7389899015426638\n",
            "Epoch:  73 , avg total loss:  3.7533467645230503\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.790885198407057\n",
            "Accuracy:  34.65853658536585\n",
            "Epoch 73/149 Completed\n",
            "Epoch:  74 , Batch:  0 , loss avg over log interval:  3.7296464443206787\n",
            "Epoch:  74 , Batch:  9 , loss avg over log interval:  3.3776855945587156\n",
            "Epoch:  74 , Batch:  19 , loss avg over log interval:  3.7393903970718383\n",
            "Epoch:  74 , Batch:  29 , loss avg over log interval:  3.747225284576416\n",
            "Epoch:  74 , Batch:  39 , loss avg over log interval:  3.776001524925232\n",
            "Epoch:  74 , Batch:  49 , loss avg over log interval:  3.7454241275787354\n",
            "Epoch:  74 , Batch:  59 , loss avg over log interval:  3.7553892850875856\n",
            "Epoch:  74 , Batch:  69 , loss avg over log interval:  3.7659725666046144\n",
            "Epoch:  74 , Batch:  79 , loss avg over log interval:  3.764124274253845\n",
            "Epoch:  74 , Batch:  89 , loss avg over log interval:  3.7495789527893066\n",
            "Epoch:  74 , Batch:  99 , loss avg over log interval:  3.7599087238311766\n",
            "Epoch:  74 , Batch:  109 , loss avg over log interval:  3.748155045509338\n",
            "Epoch:  74 , Batch:  119 , loss avg over log interval:  3.7368620157241823\n",
            "Epoch:  74 , Batch:  129 , loss avg over log interval:  3.750244450569153\n",
            "Epoch:  74 , avg total loss:  3.752273326334746\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.7846784882429167\n",
            "Accuracy:  35.46341463414634\n",
            "Epoch 74/149 Completed\n",
            "Epoch:  75 , Batch:  0 , loss avg over log interval:  3.721604108810425\n",
            "Epoch:  75 , Batch:  9 , loss avg over log interval:  3.3727971076965333\n",
            "Epoch:  75 , Batch:  19 , loss avg over log interval:  3.74133505821228\n",
            "Epoch:  75 , Batch:  29 , loss avg over log interval:  3.7525192975997923\n",
            "Epoch:  75 , Batch:  39 , loss avg over log interval:  3.7650543689727782\n",
            "Epoch:  75 , Batch:  49 , loss avg over log interval:  3.7328582763671876\n",
            "Epoch:  75 , Batch:  59 , loss avg over log interval:  3.757761025428772\n",
            "Epoch:  75 , Batch:  69 , loss avg over log interval:  3.745992660522461\n",
            "Epoch:  75 , Batch:  79 , loss avg over log interval:  3.7513394117355348\n",
            "Epoch:  75 , Batch:  89 , loss avg over log interval:  3.7529833793640135\n",
            "Epoch:  75 , Batch:  99 , loss avg over log interval:  3.769655227661133\n",
            "Epoch:  75 , Batch:  109 , loss avg over log interval:  3.735618543624878\n",
            "Epoch:  75 , Batch:  119 , loss avg over log interval:  3.7283223628997804\n",
            "Epoch:  75 , Batch:  129 , loss avg over log interval:  3.735664415359497\n",
            "Epoch:  75 , avg total loss:  3.7481298913126406\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.7831446077765487\n",
            "Accuracy:  35.68292682926829\n",
            "Epoch 75/149 Completed\n",
            "Epoch:  76 , Batch:  0 , loss avg over log interval:  3.77193021774292\n",
            "Epoch:  76 , Batch:  9 , loss avg over log interval:  3.379007649421692\n",
            "Epoch:  76 , Batch:  19 , loss avg over log interval:  3.717978096008301\n",
            "Epoch:  76 , Batch:  29 , loss avg over log interval:  3.759799838066101\n",
            "Epoch:  76 , Batch:  39 , loss avg over log interval:  3.7454952955245973\n",
            "Epoch:  76 , Batch:  49 , loss avg over log interval:  3.751454305648804\n",
            "Epoch:  76 , Batch:  59 , loss avg over log interval:  3.747192215919495\n",
            "Epoch:  76 , Batch:  69 , loss avg over log interval:  3.7491567134857178\n",
            "Epoch:  76 , Batch:  79 , loss avg over log interval:  3.766474962234497\n",
            "Epoch:  76 , Batch:  89 , loss avg over log interval:  3.7632941007614136\n",
            "Epoch:  76 , Batch:  99 , loss avg over log interval:  3.7628071308135986\n",
            "Epoch:  76 , Batch:  109 , loss avg over log interval:  3.7748830318450928\n",
            "Epoch:  76 , Batch:  119 , loss avg over log interval:  3.751156044006348\n",
            "Epoch:  76 , Batch:  129 , loss avg over log interval:  3.741475248336792\n",
            "Epoch:  76 , avg total loss:  3.7526697058608565\n",
            "iter   0\n",
            "accuracy so far =  0.671875\n",
            "Avg Validation Loss:  3.7817711364932176\n",
            "Accuracy:  35.75609756097561\n",
            "Epoch 76/149 Completed\n",
            "Epoch:  77 , Batch:  0 , loss avg over log interval:  3.7718281745910645\n",
            "Epoch:  77 , Batch:  9 , loss avg over log interval:  3.3661086559295654\n",
            "Epoch:  77 , Batch:  19 , loss avg over log interval:  3.7362172842025756\n",
            "Epoch:  77 , Batch:  29 , loss avg over log interval:  3.7452362298965456\n",
            "Epoch:  77 , Batch:  39 , loss avg over log interval:  3.7433987617492677\n",
            "Epoch:  77 , Batch:  49 , loss avg over log interval:  3.7317779064178467\n",
            "Epoch:  77 , Batch:  59 , loss avg over log interval:  3.7385751962661744\n",
            "Epoch:  77 , Batch:  69 , loss avg over log interval:  3.7496638536453246\n",
            "Epoch:  77 , Batch:  79 , loss avg over log interval:  3.7576794147491457\n",
            "Epoch:  77 , Batch:  89 , loss avg over log interval:  3.741800236701965\n",
            "Epoch:  77 , Batch:  99 , loss avg over log interval:  3.752247619628906\n",
            "Epoch:  77 , Batch:  109 , loss avg over log interval:  3.7448282718658445\n",
            "Epoch:  77 , Batch:  119 , loss avg over log interval:  3.7486809730529784\n",
            "Epoch:  77 , Batch:  129 , loss avg over log interval:  3.7331387758255006\n",
            "Epoch:  77 , avg total loss:  3.743414799372355\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.77660302999543\n",
            "Accuracy:  36.65853658536585\n",
            "Epoch 77/149 Completed\n",
            "Epoch:  78 , Batch:  0 , loss avg over log interval:  3.7961645126342773\n",
            "Epoch:  78 , Batch:  9 , loss avg over log interval:  3.3552868366241455\n",
            "Epoch:  78 , Batch:  19 , loss avg over log interval:  3.7341192245483397\n",
            "Epoch:  78 , Batch:  29 , loss avg over log interval:  3.7449406147003175\n",
            "Epoch:  78 , Batch:  39 , loss avg over log interval:  3.7420584917068482\n",
            "Epoch:  78 , Batch:  49 , loss avg over log interval:  3.7358707904815676\n",
            "Epoch:  78 , Batch:  59 , loss avg over log interval:  3.747430372238159\n",
            "Epoch:  78 , Batch:  69 , loss avg over log interval:  3.7534835815429686\n",
            "Epoch:  78 , Batch:  79 , loss avg over log interval:  3.748797082901001\n",
            "Epoch:  78 , Batch:  89 , loss avg over log interval:  3.7389848470687865\n",
            "Epoch:  78 , Batch:  99 , loss avg over log interval:  3.74363317489624\n",
            "Epoch:  78 , Batch:  109 , loss avg over log interval:  3.7448742389678955\n",
            "Epoch:  78 , Batch:  119 , loss avg over log interval:  3.734166216850281\n",
            "Epoch:  78 , Batch:  129 , loss avg over log interval:  3.733087754249573\n",
            "Epoch:  78 , avg total loss:  3.7413191657135454\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.7749185329530297\n",
            "Accuracy:  36.65853658536585\n",
            "Epoch 78/149 Completed\n",
            "Epoch:  79 , Batch:  0 , loss avg over log interval:  3.753018379211426\n",
            "Epoch:  79 , Batch:  9 , loss avg over log interval:  3.36992027759552\n",
            "Epoch:  79 , Batch:  19 , loss avg over log interval:  3.731022357940674\n",
            "Epoch:  79 , Batch:  29 , loss avg over log interval:  3.7300050497055053\n",
            "Epoch:  79 , Batch:  39 , loss avg over log interval:  3.723009705543518\n",
            "Epoch:  79 , Batch:  49 , loss avg over log interval:  3.7389618396759032\n",
            "Epoch:  79 , Batch:  59 , loss avg over log interval:  3.7401478767395018\n",
            "Epoch:  79 , Batch:  69 , loss avg over log interval:  3.734151577949524\n",
            "Epoch:  79 , Batch:  79 , loss avg over log interval:  3.7571870565414427\n",
            "Epoch:  79 , Batch:  89 , loss avg over log interval:  3.7369021654129027\n",
            "Epoch:  79 , Batch:  99 , loss avg over log interval:  3.7585883378982543\n",
            "Epoch:  79 , Batch:  109 , loss avg over log interval:  3.7429473876953123\n",
            "Epoch:  79 , Batch:  119 , loss avg over log interval:  3.729199981689453\n",
            "Epoch:  79 , Batch:  129 , loss avg over log interval:  3.743393087387085\n",
            "Epoch:  79 , avg total loss:  3.7406045112057007\n",
            "iter   0\n",
            "accuracy so far =  0.65625\n",
            "Avg Validation Loss:  3.774069640694595\n",
            "Accuracy:  36.926829268292686\n",
            "Epoch 79/149 Completed\n",
            "Epoch:  80 , Batch:  0 , loss avg over log interval:  3.768714189529419\n",
            "Epoch:  80 , Batch:  9 , loss avg over log interval:  3.370998740196228\n",
            "Epoch:  80 , Batch:  19 , loss avg over log interval:  3.7177248001098633\n",
            "Epoch:  80 , Batch:  29 , loss avg over log interval:  3.739620900154114\n",
            "Epoch:  80 , Batch:  39 , loss avg over log interval:  3.7315752506256104\n",
            "Epoch:  80 , Batch:  49 , loss avg over log interval:  3.7354889631271364\n",
            "Epoch:  80 , Batch:  59 , loss avg over log interval:  3.7381420612335203\n",
            "Epoch:  80 , Batch:  69 , loss avg over log interval:  3.7439387321472166\n",
            "Epoch:  80 , Batch:  79 , loss avg over log interval:  3.749696373939514\n",
            "Epoch:  80 , Batch:  89 , loss avg over log interval:  3.7362178802490233\n",
            "Epoch:  80 , Batch:  99 , loss avg over log interval:  3.752310585975647\n",
            "Epoch:  80 , Batch:  109 , loss avg over log interval:  3.7392109394073487\n",
            "Epoch:  80 , Batch:  119 , loss avg over log interval:  3.735635757446289\n",
            "Epoch:  80 , Batch:  129 , loss avg over log interval:  3.741901183128357\n",
            "Epoch:  80 , avg total loss:  3.739073179770207\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.772819606269278\n",
            "Accuracy:  36.97560975609756\n",
            "Epoch 80/149 Completed\n",
            "Epoch:  81 , Batch:  0 , loss avg over log interval:  3.7373087406158447\n",
            "Epoch:  81 , Batch:  9 , loss avg over log interval:  3.3774903059005736\n",
            "Epoch:  81 , Batch:  19 , loss avg over log interval:  3.7460062980651854\n",
            "Epoch:  81 , Batch:  29 , loss avg over log interval:  3.7594969272613525\n",
            "Epoch:  81 , Batch:  39 , loss avg over log interval:  3.738155221939087\n",
            "Epoch:  81 , Batch:  49 , loss avg over log interval:  3.7606327056884767\n",
            "Epoch:  81 , Batch:  59 , loss avg over log interval:  3.7542240619659424\n",
            "Epoch:  81 , Batch:  69 , loss avg over log interval:  3.7502424240112306\n",
            "Epoch:  81 , Batch:  79 , loss avg over log interval:  3.7560287714004517\n",
            "Epoch:  81 , Batch:  89 , loss avg over log interval:  3.746091890335083\n",
            "Epoch:  81 , Batch:  99 , loss avg over log interval:  3.7404726266860964\n",
            "Epoch:  81 , Batch:  109 , loss avg over log interval:  3.7331546783447265\n",
            "Epoch:  81 , Batch:  119 , loss avg over log interval:  3.7263278722763062\n",
            "Epoch:  81 , Batch:  129 , loss avg over log interval:  3.727403712272644\n",
            "Epoch:  81 , avg total loss:  3.744341193765834\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7701214755453716\n",
            "Accuracy:  37.170731707317074\n",
            "Epoch 81/149 Completed\n",
            "Epoch:  82 , Batch:  0 , loss avg over log interval:  3.739546060562134\n",
            "Epoch:  82 , Batch:  9 , loss avg over log interval:  3.3728482246398928\n",
            "Epoch:  82 , Batch:  19 , loss avg over log interval:  3.7445107460021974\n",
            "Epoch:  82 , Batch:  29 , loss avg over log interval:  3.7440440893173217\n",
            "Epoch:  82 , Batch:  39 , loss avg over log interval:  3.749992609024048\n",
            "Epoch:  82 , Batch:  49 , loss avg over log interval:  3.7416574001312255\n",
            "Epoch:  82 , Batch:  59 , loss avg over log interval:  3.745536208152771\n",
            "Epoch:  82 , Batch:  69 , loss avg over log interval:  3.7413227558135986\n",
            "Epoch:  82 , Batch:  79 , loss avg over log interval:  3.7348657846450806\n",
            "Epoch:  82 , Batch:  89 , loss avg over log interval:  3.752229857444763\n",
            "Epoch:  82 , Batch:  99 , loss avg over log interval:  3.7576324224472044\n",
            "Epoch:  82 , Batch:  109 , loss avg over log interval:  3.748702049255371\n",
            "Epoch:  82 , Batch:  119 , loss avg over log interval:  3.73289794921875\n",
            "Epoch:  82 , Batch:  129 , loss avg over log interval:  3.7319159269332887\n",
            "Epoch:  82 , avg total loss:  3.744283346162326\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.772980841194711\n",
            "Accuracy:  36.8780487804878\n",
            "Epoch 82/149 Completed\n",
            "Epoch:  83 , Batch:  0 , loss avg over log interval:  3.7403762340545654\n",
            "Epoch:  83 , Batch:  9 , loss avg over log interval:  3.3598928928375242\n",
            "Epoch:  83 , Batch:  19 , loss avg over log interval:  3.7550533056259154\n",
            "Epoch:  83 , Batch:  29 , loss avg over log interval:  3.7398722171783447\n",
            "Epoch:  83 , Batch:  39 , loss avg over log interval:  3.7315866708755494\n",
            "Epoch:  83 , Batch:  49 , loss avg over log interval:  3.748548746109009\n",
            "Epoch:  83 , Batch:  59 , loss avg over log interval:  3.7420129776000977\n",
            "Epoch:  83 , Batch:  69 , loss avg over log interval:  3.7505984783172606\n",
            "Epoch:  83 , Batch:  79 , loss avg over log interval:  3.7503804922103883\n",
            "Epoch:  83 , Batch:  89 , loss avg over log interval:  3.7504029273986816\n",
            "Epoch:  83 , Batch:  99 , loss avg over log interval:  3.7306976079940797\n",
            "Epoch:  83 , Batch:  109 , loss avg over log interval:  3.7247951507568358\n",
            "Epoch:  83 , Batch:  119 , loss avg over log interval:  3.7383752346038817\n",
            "Epoch:  83 , Batch:  129 , loss avg over log interval:  3.7375807762145996\n",
            "Epoch:  83 , avg total loss:  3.741284938826077\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7682933923674793\n",
            "Accuracy:  37.68292682926829\n",
            "Epoch 83/149 Completed\n",
            "Epoch:  84 , Batch:  0 , loss avg over log interval:  3.741123676300049\n",
            "Epoch:  84 , Batch:  9 , loss avg over log interval:  3.3667380809783936\n",
            "Epoch:  84 , Batch:  19 , loss avg over log interval:  3.728900599479675\n",
            "Epoch:  84 , Batch:  29 , loss avg over log interval:  3.7253960847854612\n",
            "Epoch:  84 , Batch:  39 , loss avg over log interval:  3.7280877828598022\n",
            "Epoch:  84 , Batch:  49 , loss avg over log interval:  3.745570659637451\n",
            "Epoch:  84 , Batch:  59 , loss avg over log interval:  3.7436371564865114\n",
            "Epoch:  84 , Batch:  69 , loss avg over log interval:  3.7464619636535645\n",
            "Epoch:  84 , Batch:  79 , loss avg over log interval:  3.7306769609451296\n",
            "Epoch:  84 , Batch:  89 , loss avg over log interval:  3.7492491245269775\n",
            "Epoch:  84 , Batch:  99 , loss avg over log interval:  3.7577658176422117\n",
            "Epoch:  84 , Batch:  109 , loss avg over log interval:  3.742403674125671\n",
            "Epoch:  84 , Batch:  119 , loss avg over log interval:  3.729783129692078\n",
            "Epoch:  84 , Batch:  129 , loss avg over log interval:  3.7167642593383787\n",
            "Epoch:  84 , avg total loss:  3.7375447594601177\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.767825574409671\n",
            "Accuracy:  37.4390243902439\n",
            "Epoch 84/149 Completed\n",
            "Epoch:  85 , Batch:  0 , loss avg over log interval:  3.758878469467163\n",
            "Epoch:  85 , Batch:  9 , loss avg over log interval:  3.357500672340393\n",
            "Epoch:  85 , Batch:  19 , loss avg over log interval:  3.731756496429443\n",
            "Epoch:  85 , Batch:  29 , loss avg over log interval:  3.7458277702331544\n",
            "Epoch:  85 , Batch:  39 , loss avg over log interval:  3.7260642051696777\n",
            "Epoch:  85 , Batch:  49 , loss avg over log interval:  3.748537707328796\n",
            "Epoch:  85 , Batch:  59 , loss avg over log interval:  3.7322208166122435\n",
            "Epoch:  85 , Batch:  69 , loss avg over log interval:  3.7412562370300293\n",
            "Epoch:  85 , Batch:  79 , loss avg over log interval:  3.7377283811569213\n",
            "Epoch:  85 , Batch:  89 , loss avg over log interval:  3.740398955345154\n",
            "Epoch:  85 , Batch:  99 , loss avg over log interval:  3.7556280374526976\n",
            "Epoch:  85 , Batch:  109 , loss avg over log interval:  3.7384921312332153\n",
            "Epoch:  85 , Batch:  119 , loss avg over log interval:  3.7192981958389284\n",
            "Epoch:  85 , Batch:  129 , loss avg over log interval:  3.7136213541030885\n",
            "Epoch:  85 , avg total loss:  3.7362480716428896\n",
            "iter   0\n",
            "accuracy so far =  0.6640625\n",
            "Avg Validation Loss:  3.7654942535772555\n",
            "Accuracy:  38.0\n",
            "Epoch 85/149 Completed\n",
            "Epoch:  86 , Batch:  0 , loss avg over log interval:  3.7172436714172363\n",
            "Epoch:  86 , Batch:  9 , loss avg over log interval:  3.3541262626647947\n",
            "Epoch:  86 , Batch:  19 , loss avg over log interval:  3.728506827354431\n",
            "Epoch:  86 , Batch:  29 , loss avg over log interval:  3.7489081621170044\n",
            "Epoch:  86 , Batch:  39 , loss avg over log interval:  3.7491069078445434\n",
            "Epoch:  86 , Batch:  49 , loss avg over log interval:  3.7394344568252564\n",
            "Epoch:  86 , Batch:  59 , loss avg over log interval:  3.755612874031067\n",
            "Epoch:  86 , Batch:  69 , loss avg over log interval:  3.745783233642578\n",
            "Epoch:  86 , Batch:  79 , loss avg over log interval:  3.7314682245254516\n",
            "Epoch:  86 , Batch:  89 , loss avg over log interval:  3.74407217502594\n",
            "Epoch:  86 , Batch:  99 , loss avg over log interval:  3.757362222671509\n",
            "Epoch:  86 , Batch:  109 , loss avg over log interval:  3.7442711114883425\n",
            "Epoch:  86 , Batch:  119 , loss avg over log interval:  3.734265112876892\n",
            "Epoch:  86 , Batch:  129 , loss avg over log interval:  3.7182836055755617\n",
            "Epoch:  86 , avg total loss:  3.7393525147783584\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7647885171378532\n",
            "Accuracy:  38.1219512195122\n",
            "Epoch 86/149 Completed\n",
            "Epoch:  87 , Batch:  0 , loss avg over log interval:  3.781651258468628\n",
            "Epoch:  87 , Batch:  9 , loss avg over log interval:  3.3785468339920044\n",
            "Epoch:  87 , Batch:  19 , loss avg over log interval:  3.7131297826766967\n",
            "Epoch:  87 , Batch:  29 , loss avg over log interval:  3.737310194969177\n",
            "Epoch:  87 , Batch:  39 , loss avg over log interval:  3.7467387676239015\n",
            "Epoch:  87 , Batch:  49 , loss avg over log interval:  3.7377079248428347\n",
            "Epoch:  87 , Batch:  59 , loss avg over log interval:  3.7238458395004272\n",
            "Epoch:  87 , Batch:  69 , loss avg over log interval:  3.7521151542663573\n",
            "Epoch:  87 , Batch:  79 , loss avg over log interval:  3.7426608085632322\n",
            "Epoch:  87 , Batch:  89 , loss avg over log interval:  3.7326488494873047\n",
            "Epoch:  87 , Batch:  99 , loss avg over log interval:  3.741986346244812\n",
            "Epoch:  87 , Batch:  109 , loss avg over log interval:  3.751552677154541\n",
            "Epoch:  87 , Batch:  119 , loss avg over log interval:  3.7120899200439452\n",
            "Epoch:  87 , Batch:  129 , loss avg over log interval:  3.7184093952178956\n",
            "Epoch:  87 , avg total loss:  3.7354810220607813\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.762558995223627\n",
            "Accuracy:  38.46341463414634\n",
            "Epoch 87/149 Completed\n",
            "Epoch:  88 , Batch:  0 , loss avg over log interval:  3.734050750732422\n",
            "Epoch:  88 , Batch:  9 , loss avg over log interval:  3.3687259912490846\n",
            "Epoch:  88 , Batch:  19 , loss avg over log interval:  3.7357073783874513\n",
            "Epoch:  88 , Batch:  29 , loss avg over log interval:  3.7278735876083373\n",
            "Epoch:  88 , Batch:  39 , loss avg over log interval:  3.7388154029846192\n",
            "Epoch:  88 , Batch:  49 , loss avg over log interval:  3.723313593864441\n",
            "Epoch:  88 , Batch:  59 , loss avg over log interval:  3.739048719406128\n",
            "Epoch:  88 , Batch:  69 , loss avg over log interval:  3.7289606094360352\n",
            "Epoch:  88 , Batch:  79 , loss avg over log interval:  3.720808219909668\n",
            "Epoch:  88 , Batch:  89 , loss avg over log interval:  3.7330161571502685\n",
            "Epoch:  88 , Batch:  99 , loss avg over log interval:  3.7427798986434935\n",
            "Epoch:  88 , Batch:  109 , loss avg over log interval:  3.7402827978134154\n",
            "Epoch:  88 , Batch:  119 , loss avg over log interval:  3.72741014957428\n",
            "Epoch:  88 , Batch:  129 , loss avg over log interval:  3.715074610710144\n",
            "Epoch:  88 , avg total loss:  3.73181530703669\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7620004909794504\n",
            "Accuracy:  38.31707317073171\n",
            "Epoch 88/149 Completed\n",
            "Epoch:  89 , Batch:  0 , loss avg over log interval:  3.702867269515991\n",
            "Epoch:  89 , Batch:  9 , loss avg over log interval:  3.3485345363616945\n",
            "Epoch:  89 , Batch:  19 , loss avg over log interval:  3.729920482635498\n",
            "Epoch:  89 , Batch:  29 , loss avg over log interval:  3.729825162887573\n",
            "Epoch:  89 , Batch:  39 , loss avg over log interval:  3.7311966180801392\n",
            "Epoch:  89 , Batch:  49 , loss avg over log interval:  3.7255719661712647\n",
            "Epoch:  89 , Batch:  59 , loss avg over log interval:  3.705433893203735\n",
            "Epoch:  89 , Batch:  69 , loss avg over log interval:  3.724013066291809\n",
            "Epoch:  89 , Batch:  79 , loss avg over log interval:  3.7305977582931518\n",
            "Epoch:  89 , Batch:  89 , loss avg over log interval:  3.743769097328186\n",
            "Epoch:  89 , Batch:  99 , loss avg over log interval:  3.7444180965423586\n",
            "Epoch:  89 , Batch:  109 , loss avg over log interval:  3.7308842420578\n",
            "Epoch:  89 , Batch:  119 , loss avg over log interval:  3.7344372272491455\n",
            "Epoch:  89 , Batch:  129 , loss avg over log interval:  3.7336448431015015\n",
            "Epoch:  89 , avg total loss:  3.7301614163578423\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7638164148098086\n",
            "Accuracy:  38.4390243902439\n",
            "Epoch 89/149 Completed\n",
            "Epoch:  90 , Batch:  0 , loss avg over log interval:  3.7375621795654297\n",
            "Epoch:  90 , Batch:  9 , loss avg over log interval:  3.361863136291504\n",
            "Epoch:  90 , Batch:  19 , loss avg over log interval:  3.715075969696045\n",
            "Epoch:  90 , Batch:  29 , loss avg over log interval:  3.717931056022644\n",
            "Epoch:  90 , Batch:  39 , loss avg over log interval:  3.72514226436615\n",
            "Epoch:  90 , Batch:  49 , loss avg over log interval:  3.7282846212387084\n",
            "Epoch:  90 , Batch:  59 , loss avg over log interval:  3.748710799217224\n",
            "Epoch:  90 , Batch:  69 , loss avg over log interval:  3.733181524276733\n",
            "Epoch:  90 , Batch:  79 , loss avg over log interval:  3.742901635169983\n",
            "Epoch:  90 , Batch:  89 , loss avg over log interval:  3.7164811372756956\n",
            "Epoch:  90 , Batch:  99 , loss avg over log interval:  3.7482627630233765\n",
            "Epoch:  90 , Batch:  109 , loss avg over log interval:  3.7408727169036866\n",
            "Epoch:  90 , Batch:  119 , loss avg over log interval:  3.714274001121521\n",
            "Epoch:  90 , Batch:  129 , loss avg over log interval:  3.7339462757110597\n",
            "Epoch:  90 , avg total loss:  3.731539351352747\n",
            "iter   0\n",
            "accuracy so far =  0.6796875\n",
            "Avg Validation Loss:  3.759745272194467\n",
            "Accuracy:  39.21951219512195\n",
            "Epoch 90/149 Completed\n",
            "Epoch:  91 , Batch:  0 , loss avg over log interval:  3.757723093032837\n",
            "Epoch:  91 , Batch:  9 , loss avg over log interval:  3.359694981575012\n",
            "Epoch:  91 , Batch:  19 , loss avg over log interval:  3.726175045967102\n",
            "Epoch:  91 , Batch:  29 , loss avg over log interval:  3.72558057308197\n",
            "Epoch:  91 , Batch:  39 , loss avg over log interval:  3.753958320617676\n",
            "Epoch:  91 , Batch:  49 , loss avg over log interval:  3.737650227546692\n",
            "Epoch:  91 , Batch:  59 , loss avg over log interval:  3.7345923662185667\n",
            "Epoch:  91 , Batch:  69 , loss avg over log interval:  3.7286255598068236\n",
            "Epoch:  91 , Batch:  79 , loss avg over log interval:  3.7300458192825316\n",
            "Epoch:  91 , Batch:  89 , loss avg over log interval:  3.7385799169540403\n",
            "Epoch:  91 , Batch:  99 , loss avg over log interval:  3.7255762100219725\n",
            "Epoch:  91 , Batch:  109 , loss avg over log interval:  3.723854637145996\n",
            "Epoch:  91 , Batch:  119 , loss avg over log interval:  3.719879722595215\n",
            "Epoch:  91 , Batch:  129 , loss avg over log interval:  3.7144301414489744\n",
            "Epoch:  91 , avg total loss:  3.7300097976905713\n",
            "iter   0\n",
            "accuracy so far =  0.6875\n",
            "Avg Validation Loss:  3.763756019313161\n",
            "Accuracy:  39.0\n",
            "Epoch 91/149 Completed\n",
            "Epoch:  92 , Batch:  0 , loss avg over log interval:  3.753298044204712\n",
            "Epoch:  92 , Batch:  9 , loss avg over log interval:  3.3559591293334963\n",
            "Epoch:  92 , Batch:  19 , loss avg over log interval:  3.7294262409210206\n",
            "Epoch:  92 , Batch:  29 , loss avg over log interval:  3.724009156227112\n",
            "Epoch:  92 , Batch:  39 , loss avg over log interval:  3.72687668800354\n",
            "Epoch:  92 , Batch:  49 , loss avg over log interval:  3.7304923295974732\n",
            "Epoch:  92 , Batch:  59 , loss avg over log interval:  3.7187326669692995\n",
            "Epoch:  92 , Batch:  69 , loss avg over log interval:  3.715515613555908\n",
            "Epoch:  92 , Batch:  79 , loss avg over log interval:  3.7140476226806642\n",
            "Epoch:  92 , Batch:  89 , loss avg over log interval:  3.7168784141540527\n",
            "Epoch:  92 , Batch:  99 , loss avg over log interval:  3.7253318309783934\n",
            "Epoch:  92 , Batch:  109 , loss avg over log interval:  3.7367019176483156\n",
            "Epoch:  92 , Batch:  119 , loss avg over log interval:  3.720962905883789\n",
            "Epoch:  92 , Batch:  129 , loss avg over log interval:  3.7220061302185057\n",
            "Epoch:  92 , avg total loss:  3.724501689275106\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.758135394352238\n",
            "Accuracy:  39.951219512195124\n",
            "Epoch 92/149 Completed\n",
            "Epoch:  93 , Batch:  0 , loss avg over log interval:  3.723769187927246\n",
            "Epoch:  93 , Batch:  9 , loss avg over log interval:  3.3620584487915037\n",
            "Epoch:  93 , Batch:  19 , loss avg over log interval:  3.71882107257843\n",
            "Epoch:  93 , Batch:  29 , loss avg over log interval:  3.731791377067566\n",
            "Epoch:  93 , Batch:  39 , loss avg over log interval:  3.731442189216614\n",
            "Epoch:  93 , Batch:  49 , loss avg over log interval:  3.744185471534729\n",
            "Epoch:  93 , Batch:  59 , loss avg over log interval:  3.709713935852051\n",
            "Epoch:  93 , Batch:  69 , loss avg over log interval:  3.7229880332946776\n",
            "Epoch:  93 , Batch:  79 , loss avg over log interval:  3.746056652069092\n",
            "Epoch:  93 , Batch:  89 , loss avg over log interval:  3.7422528505325316\n",
            "Epoch:  93 , Batch:  99 , loss avg over log interval:  3.742003870010376\n",
            "Epoch:  93 , Batch:  109 , loss avg over log interval:  3.738039422035217\n",
            "Epoch:  93 , Batch:  119 , loss avg over log interval:  3.716227149963379\n",
            "Epoch:  93 , Batch:  129 , loss avg over log interval:  3.7185843706130983\n",
            "Epoch:  93 , avg total loss:  3.7304653015689575\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.7562872374930034\n",
            "Accuracy:  40.19512195121951\n",
            "Epoch 93/149 Completed\n",
            "Epoch:  94 , Batch:  0 , loss avg over log interval:  3.7540204524993896\n",
            "Epoch:  94 , Batch:  9 , loss avg over log interval:  3.3557096481323243\n",
            "Epoch:  94 , Batch:  19 , loss avg over log interval:  3.7214680671691895\n",
            "Epoch:  94 , Batch:  29 , loss avg over log interval:  3.7201893806457518\n",
            "Epoch:  94 , Batch:  39 , loss avg over log interval:  3.739709663391113\n",
            "Epoch:  94 , Batch:  49 , loss avg over log interval:  3.7350576400756834\n",
            "Epoch:  94 , Batch:  59 , loss avg over log interval:  3.7088531255722046\n",
            "Epoch:  94 , Batch:  69 , loss avg over log interval:  3.7196096420288085\n",
            "Epoch:  94 , Batch:  79 , loss avg over log interval:  3.7321743965148926\n",
            "Epoch:  94 , Batch:  89 , loss avg over log interval:  3.7262748003005983\n",
            "Epoch:  94 , Batch:  99 , loss avg over log interval:  3.742790126800537\n",
            "Epoch:  94 , Batch:  109 , loss avg over log interval:  3.724111461639404\n",
            "Epoch:  94 , Batch:  119 , loss avg over log interval:  3.718160033226013\n",
            "Epoch:  94 , Batch:  129 , loss avg over log interval:  3.7046881198883055\n",
            "Epoch:  94 , avg total loss:  3.7245562784913657\n",
            "iter   0\n",
            "accuracy so far =  0.71875\n",
            "Avg Validation Loss:  3.7557936063626918\n",
            "Accuracy:  41.0\n",
            "Epoch 94/149 Completed\n",
            "Epoch:  95 , Batch:  0 , loss avg over log interval:  3.704599142074585\n",
            "Epoch:  95 , Batch:  9 , loss avg over log interval:  3.3615397453308105\n",
            "Epoch:  95 , Batch:  19 , loss avg over log interval:  3.7423035144805907\n",
            "Epoch:  95 , Batch:  29 , loss avg over log interval:  3.7186319828033447\n",
            "Epoch:  95 , Batch:  39 , loss avg over log interval:  3.7286883354187013\n",
            "Epoch:  95 , Batch:  49 , loss avg over log interval:  3.7213276863098144\n",
            "Epoch:  95 , Batch:  59 , loss avg over log interval:  3.70970938205719\n",
            "Epoch:  95 , Batch:  69 , loss avg over log interval:  3.728266787528992\n",
            "Epoch:  95 , Batch:  79 , loss avg over log interval:  3.7319864511489866\n",
            "Epoch:  95 , Batch:  89 , loss avg over log interval:  3.7309510946273803\n",
            "Epoch:  95 , Batch:  99 , loss avg over log interval:  3.7329543113708494\n",
            "Epoch:  95 , Batch:  109 , loss avg over log interval:  3.732804036140442\n",
            "Epoch:  95 , Batch:  119 , loss avg over log interval:  3.722312593460083\n",
            "Epoch:  95 , Batch:  129 , loss avg over log interval:  3.7153785467147826\n",
            "Epoch:  95 , avg total loss:  3.727108701415684\n",
            "iter   0\n",
            "accuracy so far =  0.7265625\n",
            "Avg Validation Loss:  3.7529496158041606\n",
            "Accuracy:  41.09756097560975\n",
            "Epoch 95/149 Completed\n",
            "Epoch:  96 , Batch:  0 , loss avg over log interval:  3.7601168155670166\n",
            "Epoch:  96 , Batch:  9 , loss avg over log interval:  3.3475505352020263\n",
            "Epoch:  96 , Batch:  19 , loss avg over log interval:  3.7148760557174683\n",
            "Epoch:  96 , Batch:  29 , loss avg over log interval:  3.7236764907836912\n",
            "Epoch:  96 , Batch:  39 , loss avg over log interval:  3.7205618143081667\n",
            "Epoch:  96 , Batch:  49 , loss avg over log interval:  3.727715492248535\n",
            "Epoch:  96 , Batch:  59 , loss avg over log interval:  3.7260032176971434\n",
            "Epoch:  96 , Batch:  69 , loss avg over log interval:  3.704438400268555\n",
            "Epoch:  96 , Batch:  79 , loss avg over log interval:  3.7261481046676637\n",
            "Epoch:  96 , Batch:  89 , loss avg over log interval:  3.730120873451233\n",
            "Epoch:  96 , Batch:  99 , loss avg over log interval:  3.720819926261902\n",
            "Epoch:  96 , Batch:  109 , loss avg over log interval:  3.7322969913482664\n",
            "Epoch:  96 , Batch:  119 , loss avg over log interval:  3.726870608329773\n",
            "Epoch:  96 , Batch:  129 , loss avg over log interval:  3.7280861139297485\n",
            "Epoch:  96 , avg total loss:  3.72341517607371\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.753274684999047\n",
            "Accuracy:  41.53658536585366\n",
            "Epoch 96/149 Completed\n",
            "Epoch:  97 , Batch:  0 , loss avg over log interval:  3.7791783809661865\n",
            "Epoch:  97 , Batch:  9 , loss avg over log interval:  3.353938603401184\n",
            "Epoch:  97 , Batch:  19 , loss avg over log interval:  3.7050647497177125\n",
            "Epoch:  97 , Batch:  29 , loss avg over log interval:  3.715049719810486\n",
            "Epoch:  97 , Batch:  39 , loss avg over log interval:  3.733944320678711\n",
            "Epoch:  97 , Batch:  49 , loss avg over log interval:  3.7271105289459228\n",
            "Epoch:  97 , Batch:  59 , loss avg over log interval:  3.712740159034729\n",
            "Epoch:  97 , Batch:  69 , loss avg over log interval:  3.724648594856262\n",
            "Epoch:  97 , Batch:  79 , loss avg over log interval:  3.7126105546951296\n",
            "Epoch:  97 , Batch:  89 , loss avg over log interval:  3.736767601966858\n",
            "Epoch:  97 , Batch:  99 , loss avg over log interval:  3.7211922883987425\n",
            "Epoch:  97 , Batch:  109 , loss avg over log interval:  3.7119455099105836\n",
            "Epoch:  97 , Batch:  119 , loss avg over log interval:  3.7151824235916138\n",
            "Epoch:  97 , Batch:  129 , loss avg over log interval:  3.7087013006210325\n",
            "Epoch:  97 , avg total loss:  3.7199049462442813\n",
            "iter   0\n",
            "accuracy so far =  0.734375\n",
            "Avg Validation Loss:  3.7516514673465635\n",
            "Accuracy:  41.926829268292686\n",
            "Epoch 97/149 Completed\n",
            "Epoch:  98 , Batch:  0 , loss avg over log interval:  3.7412757873535156\n",
            "Epoch:  98 , Batch:  9 , loss avg over log interval:  3.3408660173416136\n",
            "Epoch:  98 , Batch:  19 , loss avg over log interval:  3.7168753385543822\n",
            "Epoch:  98 , Batch:  29 , loss avg over log interval:  3.704618239402771\n",
            "Epoch:  98 , Batch:  39 , loss avg over log interval:  3.7169894933700562\n",
            "Epoch:  98 , Batch:  49 , loss avg over log interval:  3.704484462738037\n",
            "Epoch:  98 , Batch:  59 , loss avg over log interval:  3.718231129646301\n",
            "Epoch:  98 , Batch:  69 , loss avg over log interval:  3.723205828666687\n",
            "Epoch:  98 , Batch:  79 , loss avg over log interval:  3.716100978851318\n",
            "Epoch:  98 , Batch:  89 , loss avg over log interval:  3.716053009033203\n",
            "Epoch:  98 , Batch:  99 , loss avg over log interval:  3.7207510471343994\n",
            "Epoch:  98 , Batch:  109 , loss avg over log interval:  3.725519037246704\n",
            "Epoch:  98 , Batch:  119 , loss avg over log interval:  3.707433342933655\n",
            "Epoch:  98 , Batch:  129 , loss avg over log interval:  3.713325929641724\n",
            "Epoch:  98 , avg total loss:  3.714350620905558\n",
            "iter   0\n",
            "accuracy so far =  0.7109375\n",
            "Avg Validation Loss:  3.7520606052584764\n",
            "Accuracy:  41.65853658536585\n",
            "Epoch 98/149 Completed\n",
            "Epoch:  99 , Batch:  0 , loss avg over log interval:  3.7709522247314453\n",
            "Epoch:  99 , Batch:  9 , loss avg over log interval:  3.3566375732421876\n",
            "Epoch:  99 , Batch:  19 , loss avg over log interval:  3.728625273704529\n",
            "Epoch:  99 , Batch:  29 , loss avg over log interval:  3.702057719230652\n",
            "Epoch:  99 , Batch:  39 , loss avg over log interval:  3.710265302658081\n",
            "Epoch:  99 , Batch:  49 , loss avg over log interval:  3.720848250389099\n",
            "Epoch:  99 , Batch:  59 , loss avg over log interval:  3.7355891704559325\n",
            "Epoch:  99 , Batch:  69 , loss avg over log interval:  3.7270482540130616\n",
            "Epoch:  99 , Batch:  79 , loss avg over log interval:  3.7316294431686403\n",
            "Epoch:  99 , Batch:  89 , loss avg over log interval:  3.7399805307388307\n",
            "Epoch:  99 , Batch:  99 , loss avg over log interval:  3.726560115814209\n",
            "Epoch:  99 , Batch:  109 , loss avg over log interval:  3.716509771347046\n",
            "Epoch:  99 , Batch:  119 , loss avg over log interval:  3.7046064376831054\n",
            "Epoch:  99 , Batch:  129 , loss avg over log interval:  3.7072731971740724\n",
            "Epoch:  99 , avg total loss:  3.72129303821619\n",
            "iter   0\n",
            "accuracy so far =  0.7109375\n",
            "Avg Validation Loss:  3.755612809483598\n",
            "Accuracy:  41.292682926829265\n",
            "Epoch 99/149 Completed\n",
            "Epoch:  100 , Batch:  0 , loss avg over log interval:  3.7298550605773926\n",
            "Epoch:  100 , Batch:  9 , loss avg over log interval:  3.34042649269104\n",
            "Epoch:  100 , Batch:  19 , loss avg over log interval:  3.723447346687317\n",
            "Epoch:  100 , Batch:  29 , loss avg over log interval:  3.7066649436950683\n",
            "Epoch:  100 , Batch:  39 , loss avg over log interval:  3.725187134742737\n",
            "Epoch:  100 , Batch:  49 , loss avg over log interval:  3.723633146286011\n",
            "Epoch:  100 , Batch:  59 , loss avg over log interval:  3.7214396476745604\n",
            "Epoch:  100 , Batch:  69 , loss avg over log interval:  3.7221973180770873\n",
            "Epoch:  100 , Batch:  79 , loss avg over log interval:  3.7487045764923095\n",
            "Epoch:  100 , Batch:  89 , loss avg over log interval:  3.7102326154708862\n",
            "Epoch:  100 , Batch:  99 , loss avg over log interval:  3.7181030988693236\n",
            "Epoch:  100 , Batch:  109 , loss avg over log interval:  3.733666682243347\n",
            "Epoch:  100 , Batch:  119 , loss avg over log interval:  3.7080544233322144\n",
            "Epoch:  100 , Batch:  129 , loss avg over log interval:  3.709809494018555\n",
            "Epoch:  100 , avg total loss:  3.7204543628554414\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.7512463127694478\n",
            "Accuracy:  41.853658536585364\n",
            "Epoch 100/149 Completed\n",
            "Epoch:  101 , Batch:  0 , loss avg over log interval:  3.7477645874023438\n",
            "Epoch:  101 , Batch:  9 , loss avg over log interval:  3.3472357511520388\n",
            "Epoch:  101 , Batch:  19 , loss avg over log interval:  3.7145830392837524\n",
            "Epoch:  101 , Batch:  29 , loss avg over log interval:  3.7160001516342165\n",
            "Epoch:  101 , Batch:  39 , loss avg over log interval:  3.715047001838684\n",
            "Epoch:  101 , Batch:  49 , loss avg over log interval:  3.693864035606384\n",
            "Epoch:  101 , Batch:  59 , loss avg over log interval:  3.7148743629455567\n",
            "Epoch:  101 , Batch:  69 , loss avg over log interval:  3.729932999610901\n",
            "Epoch:  101 , Batch:  79 , loss avg over log interval:  3.7352516889572143\n",
            "Epoch:  101 , Batch:  89 , loss avg over log interval:  3.717996025085449\n",
            "Epoch:  101 , Batch:  99 , loss avg over log interval:  3.7152242422103883\n",
            "Epoch:  101 , Batch:  109 , loss avg over log interval:  3.714786648750305\n",
            "Epoch:  101 , Batch:  119 , loss avg over log interval:  3.702560043334961\n",
            "Epoch:  101 , Batch:  129 , loss avg over log interval:  3.7125521659851075\n",
            "Epoch:  101 , avg total loss:  3.7165878648343296\n",
            "iter   0\n",
            "accuracy so far =  0.6875\n",
            "Avg Validation Loss:  3.7511477819303187\n",
            "Accuracy:  41.78048780487805\n",
            "Epoch 101/149 Completed\n",
            "Epoch:  102 , Batch:  0 , loss avg over log interval:  3.7234046459198\n",
            "Epoch:  102 , Batch:  9 , loss avg over log interval:  3.347706413269043\n",
            "Epoch:  102 , Batch:  19 , loss avg over log interval:  3.7181491374969484\n",
            "Epoch:  102 , Batch:  29 , loss avg over log interval:  3.6989696502685545\n",
            "Epoch:  102 , Batch:  39 , loss avg over log interval:  3.736970901489258\n",
            "Epoch:  102 , Batch:  49 , loss avg over log interval:  3.70575590133667\n",
            "Epoch:  102 , Batch:  59 , loss avg over log interval:  3.72164990901947\n",
            "Epoch:  102 , Batch:  69 , loss avg over log interval:  3.702031636238098\n",
            "Epoch:  102 , Batch:  79 , loss avg over log interval:  3.73072144985199\n",
            "Epoch:  102 , Batch:  89 , loss avg over log interval:  3.7206146001815794\n",
            "Epoch:  102 , Batch:  99 , loss avg over log interval:  3.738537406921387\n",
            "Epoch:  102 , Batch:  109 , loss avg over log interval:  3.7264389991760254\n",
            "Epoch:  102 , Batch:  119 , loss avg over log interval:  3.7030526638031005\n",
            "Epoch:  102 , Batch:  129 , loss avg over log interval:  3.718307876586914\n",
            "Epoch:  102 , avg total loss:  3.7181197335754614\n",
            "iter   0\n",
            "accuracy so far =  0.7109375\n",
            "Avg Validation Loss:  3.7471228867042354\n",
            "Accuracy:  42.073170731707314\n",
            "Epoch 102/149 Completed\n",
            "Epoch:  103 , Batch:  0 , loss avg over log interval:  3.689465045928955\n",
            "Epoch:  103 , Batch:  9 , loss avg over log interval:  3.344827151298523\n",
            "Epoch:  103 , Batch:  19 , loss avg over log interval:  3.717809462547302\n",
            "Epoch:  103 , Batch:  29 , loss avg over log interval:  3.7104687213897707\n",
            "Epoch:  103 , Batch:  39 , loss avg over log interval:  3.721368336677551\n",
            "Epoch:  103 , Batch:  49 , loss avg over log interval:  3.7197970390319823\n",
            "Epoch:  103 , Batch:  59 , loss avg over log interval:  3.712009620666504\n",
            "Epoch:  103 , Batch:  69 , loss avg over log interval:  3.7202173233032227\n",
            "Epoch:  103 , Batch:  79 , loss avg over log interval:  3.713727521896362\n",
            "Epoch:  103 , Batch:  89 , loss avg over log interval:  3.7272361516952515\n",
            "Epoch:  103 , Batch:  99 , loss avg over log interval:  3.7084887981414796\n",
            "Epoch:  103 , Batch:  109 , loss avg over log interval:  3.726880502700806\n",
            "Epoch:  103 , Batch:  119 , loss avg over log interval:  3.7028905391693114\n",
            "Epoch:  103 , Batch:  129 , loss avg over log interval:  3.709595203399658\n",
            "Epoch:  103 , avg total loss:  3.716170767079229\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7484202094194368\n",
            "Accuracy:  42.1219512195122\n",
            "Epoch 103/149 Completed\n",
            "Epoch:  104 , Batch:  0 , loss avg over log interval:  3.6975080966949463\n",
            "Epoch:  104 , Batch:  9 , loss avg over log interval:  3.330761194229126\n",
            "Epoch:  104 , Batch:  19 , loss avg over log interval:  3.704532504081726\n",
            "Epoch:  104 , Batch:  29 , loss avg over log interval:  3.7207383632659914\n",
            "Epoch:  104 , Batch:  39 , loss avg over log interval:  3.709257388114929\n",
            "Epoch:  104 , Batch:  49 , loss avg over log interval:  3.7130120277404783\n",
            "Epoch:  104 , Batch:  59 , loss avg over log interval:  3.7207221031188964\n",
            "Epoch:  104 , Batch:  69 , loss avg over log interval:  3.723816704750061\n",
            "Epoch:  104 , Batch:  79 , loss avg over log interval:  3.7225281953811646\n",
            "Epoch:  104 , Batch:  89 , loss avg over log interval:  3.7114424228668215\n",
            "Epoch:  104 , Batch:  99 , loss avg over log interval:  3.72587673664093\n",
            "Epoch:  104 , Batch:  109 , loss avg over log interval:  3.7083664417266844\n",
            "Epoch:  104 , Batch:  119 , loss avg over log interval:  3.6896061420440676\n",
            "Epoch:  104 , Batch:  129 , loss avg over log interval:  3.716639447212219\n",
            "Epoch:  104 , avg total loss:  3.7132074988406636\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7468324463541913\n",
            "Accuracy:  42.292682926829265\n",
            "Epoch 104/149 Completed\n",
            "Epoch:  105 , Batch:  0 , loss avg over log interval:  3.707948923110962\n",
            "Epoch:  105 , Batch:  9 , loss avg over log interval:  3.3532066345214844\n",
            "Epoch:  105 , Batch:  19 , loss avg over log interval:  3.7011149883270265\n",
            "Epoch:  105 , Batch:  29 , loss avg over log interval:  3.7171914339065553\n",
            "Epoch:  105 , Batch:  39 , loss avg over log interval:  3.7103549003601075\n",
            "Epoch:  105 , Batch:  49 , loss avg over log interval:  3.731715226173401\n",
            "Epoch:  105 , Batch:  59 , loss avg over log interval:  3.70256085395813\n",
            "Epoch:  105 , Batch:  69 , loss avg over log interval:  3.718907117843628\n",
            "Epoch:  105 , Batch:  79 , loss avg over log interval:  3.7198936700820924\n",
            "Epoch:  105 , Batch:  89 , loss avg over log interval:  3.704936718940735\n",
            "Epoch:  105 , Batch:  99 , loss avg over log interval:  3.712235450744629\n",
            "Epoch:  105 , Batch:  109 , loss avg over log interval:  3.7214284181594848\n",
            "Epoch:  105 , Batch:  119 , loss avg over log interval:  3.695165920257568\n",
            "Epoch:  105 , Batch:  129 , loss avg over log interval:  3.6947351932525634\n",
            "Epoch:  105 , avg total loss:  3.712034759314164\n",
            "iter   0\n",
            "accuracy so far =  0.6953125\n",
            "Avg Validation Loss:  3.7472677928645437\n",
            "Accuracy:  42.65853658536585\n",
            "Epoch 105/149 Completed\n",
            "Epoch:  106 , Batch:  0 , loss avg over log interval:  3.6967835426330566\n",
            "Epoch:  106 , Batch:  9 , loss avg over log interval:  3.3363551616668703\n",
            "Epoch:  106 , Batch:  19 , loss avg over log interval:  3.7094706773757933\n",
            "Epoch:  106 , Batch:  29 , loss avg over log interval:  3.705409550666809\n",
            "Epoch:  106 , Batch:  39 , loss avg over log interval:  3.709196424484253\n",
            "Epoch:  106 , Batch:  49 , loss avg over log interval:  3.7366292476654053\n",
            "Epoch:  106 , Batch:  59 , loss avg over log interval:  3.698402667045593\n",
            "Epoch:  106 , Batch:  69 , loss avg over log interval:  3.72351131439209\n",
            "Epoch:  106 , Batch:  79 , loss avg over log interval:  3.7175190925598143\n",
            "Epoch:  106 , Batch:  89 , loss avg over log interval:  3.7273911476135253\n",
            "Epoch:  106 , Batch:  99 , loss avg over log interval:  3.728451299667358\n",
            "Epoch:  106 , Batch:  109 , loss avg over log interval:  3.707827997207642\n",
            "Epoch:  106 , Batch:  119 , loss avg over log interval:  3.691115880012512\n",
            "Epoch:  106 , Batch:  129 , loss avg over log interval:  3.7016757488250733\n",
            "Epoch:  106 , avg total loss:  3.7117175485776817\n",
            "iter   0\n",
            "accuracy so far =  0.703125\n",
            "Avg Validation Loss:  3.7462485011030986\n",
            "Accuracy:  42.75609756097561\n",
            "Epoch 106/149 Completed\n",
            "Epoch:  107 , Batch:  0 , loss avg over log interval:  3.726530075073242\n",
            "Epoch:  107 , Batch:  9 , loss avg over log interval:  3.353508162498474\n",
            "Epoch:  107 , Batch:  19 , loss avg over log interval:  3.7068041801452636\n",
            "Epoch:  107 , Batch:  29 , loss avg over log interval:  3.7104156255722045\n",
            "Epoch:  107 , Batch:  39 , loss avg over log interval:  3.7063127756118774\n",
            "Epoch:  107 , Batch:  49 , loss avg over log interval:  3.7177210330963133\n",
            "Epoch:  107 , Batch:  59 , loss avg over log interval:  3.7065434217453004\n",
            "Epoch:  107 , Batch:  69 , loss avg over log interval:  3.718528485298157\n",
            "Epoch:  107 , Batch:  79 , loss avg over log interval:  3.70647714138031\n",
            "Epoch:  107 , Batch:  89 , loss avg over log interval:  3.7162784814834593\n",
            "Epoch:  107 , Batch:  99 , loss avg over log interval:  3.7157461166381838\n",
            "Epoch:  107 , Batch:  109 , loss avg over log interval:  3.7256446838378907\n",
            "Epoch:  107 , Batch:  119 , loss avg over log interval:  3.678389310836792\n",
            "Epoch:  107 , Batch:  129 , loss avg over log interval:  3.7029241800308226\n",
            "Epoch:  107 , avg total loss:  3.710874398549398\n",
            "iter   0\n",
            "accuracy so far =  0.7109375\n",
            "Avg Validation Loss:  3.7471701110281597\n",
            "Accuracy:  42.5609756097561\n",
            "Epoch 107/149 Completed\n",
            "Epoch:  108 , Batch:  0 , loss avg over log interval:  3.7198548316955566\n",
            "Epoch:  108 , Batch:  9 , loss avg over log interval:  3.345798897743225\n",
            "Epoch:  108 , Batch:  19 , loss avg over log interval:  3.7103960275650025\n",
            "Epoch:  108 , Batch:  29 , loss avg over log interval:  3.70810980796814\n",
            "Epoch:  108 , Batch:  39 , loss avg over log interval:  3.7072704315185545\n",
            "Epoch:  108 , Batch:  49 , loss avg over log interval:  3.7007489442825316\n",
            "Epoch:  108 , Batch:  59 , loss avg over log interval:  3.7209650039672852\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dPNS9vGlAgYE"
      },
      "source": [
        "# Plotting"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OlPLWagHAYEz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 803
        },
        "outputId": "129662b9-6a9f-4e93-c1eb-f31ddfe7515e"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "\n",
        "EPOCH_SAVE_PREFIX = '/content/drive/Shared drives/CIS680 Final Project/models/temporal_new_data/'\n",
        "\n",
        "# Device\n",
        "device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')\n",
        "torch.random.manual_seed(1)\n",
        "\n",
        "# Load params\n",
        "last_epoch = 100\n",
        "\n",
        "# load data\n",
        "network_path = EPOCH_SAVE_PREFIX + 'temporal_epoch' + str(last_epoch)\n",
        "checkpoint = torch.load(network_path)\n",
        "train_loss_list = checkpoint['train_total_loss_list']\n",
        "epoch_loss_list = checkpoint['epoch_total_loss_list']\n",
        "train_counter = checkpoint['train_counter']\n",
        "test_loss_list = checkpoint['test_loss_list']\n",
        "accuracy_list = checkpoint['accuracy_lsit']\n",
        "epoch_list = np.arange(last_epoch+1)\n",
        "\n",
        "\n",
        "# plots\n",
        "fig = plt.figure()\n",
        "plt.plot(epoch_loss_list, color='blue')\n",
        "plt.legend(['Temporal Train Loss'], loc='upper right')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Total Loss')\n",
        "\n",
        "fig = plt.figure()\n",
        "plt.plot(test_loss_list, color='green')\n",
        "plt.legend(['Temporal Validation Loss'], loc='upper right')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Total Loss')\n",
        "\n",
        "fig = plt.figure()\n",
        "plt.plot(accuracy_list, color='red')\n",
        "plt.legend(['Temporal Validation Accuracy'], loc='lower right')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Accuracy')\n",
        "\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}