Pneumonia_Detection_Using_Transfer_Learning.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U_LWl3CCWkz1"
      },
      "source": [
        "Initialisation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6AoLLVfxsy7X",
        "outputId": "86260a58-2433-4fc5-9819-73f630ceddda"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting pennylane\n",
            "  Downloading PennyLane-0.24.0-py3-none-any.whl (957 kB)\n",
            "\u001b[K     |████████████████████████████████| 957 kB 7.8 MB/s \n",
            "\u001b[?25hRequirement already satisfied: toml in /usr/local/lib/python3.7/dist-packages (from pennylane) (0.10.2)\n",
            "Requirement already satisfied: appdirs in /usr/local/lib/python3.7/dist-packages (from pennylane) (1.4.4)\n",
            "Collecting semantic-version==2.6\n",
            "  Downloading semantic_version-2.6.0-py3-none-any.whl (14 kB)\n",
            "Requirement already satisfied: networkx in /usr/local/lib/python3.7/dist-packages (from pennylane) (2.6.3)\n",
            "Collecting autoray>=0.3.1\n",
            "  Downloading autoray-0.3.2-py3-none-any.whl (36 kB)\n",
            "Collecting pennylane-lightning>=0.24\n",
            "  Downloading PennyLane_Lightning-0.24.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (9.3 MB)\n",
            "\u001b[K     |████████████████████████████████| 9.3 MB 32.0 MB/s \n",
            "\u001b[?25hCollecting retworkx\n",
            "  Downloading retworkx-0.11.0-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_12_x86_64.manylinux2010_x86_64.whl (1.6 MB)\n",
            "\u001b[K     |████████████████████████████████| 1.6 MB 58.4 MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from pennylane) (1.21.6)\n",
            "Requirement already satisfied: cachetools in /usr/local/lib/python3.7/dist-packages (from pennylane) (4.2.4)\n",
            "Requirement already satisfied: autograd in /usr/local/lib/python3.7/dist-packages (from pennylane) (1.4)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from pennylane) (1.7.3)\n",
            "Collecting ninja\n",
            "  Downloading ninja-1.10.2.3-py2.py3-none-manylinux_2_5_x86_64.manylinux1_x86_64.whl (108 kB)\n",
            "\u001b[K     |████████████████████████████████| 108 kB 60.2 MB/s \n",
            "\u001b[?25hRequirement already satisfied: future>=0.15.2 in /usr/local/lib/python3.7/dist-packages (from autograd->pennylane) (0.16.0)\n",
            "Installing collected packages: ninja, semantic-version, retworkx, pennylane-lightning, autoray, pennylane\n",
            "Successfully installed autoray-0.3.2 ninja-1.10.2.3 pennylane-0.24.0 pennylane-lightning-0.24.0 retworkx-0.11.0 semantic-version-2.6.0\n"
          ]
        }
      ],
      "source": [
        "!pip install pennylane"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "GHu2SFtXQJbw"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torchvision \n",
        "from torchvision import datasets, models, transforms\n",
        "\n",
        "import numpy as np\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "import pandas as pd\n",
        "import cv2\n",
        "import os\n",
        "import glob \n",
        "from skimage import io \n",
        "import random \n",
        "%matplotlib inline\n",
        "\n",
        "from PIL import Image\n",
        "from numpy import asarray\n",
        "\n",
        "import time\n",
        "import pickle\n",
        "\n",
        "import pennylane as qml\n",
        "import copy\n",
        "import torch.optim as optim\n",
        "from torch.optim import lr_scheduler\n",
        "\n",
        "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "CPftFC9paUzl"
      },
      "outputs": [],
      "source": [
        "\n",
        "n_qubits = 4                # Number of qubits\n",
        "step = 0.0004               # Learning rate\n",
        "batch_size = 4               # Number of samples for each training step\n",
        "num_epochs = 3              # Number of training epochs\n",
        "q_depth = 6                 # Depth of the quantum circuit (number of variational layers)\n",
        "gamma_lr_scheduler = 0.1    # Learning rate reduction applied every 10 epochs.\n",
        "q_delta = 0.01              # Initial spread of random quantum weights\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5m28VD8mWncZ"
      },
      "source": [
        "Pre-processing\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hi_e6NQjwKW6",
        "outputId": "df53a6ab-30c0-4a51-fffa-db53c8435020"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ef1XItx3ps2e"
      },
      "outputs": [],
      "source": [
        "data_transforms =transforms.Compose([\n",
        "        transforms.Resize(256),\n",
        "        transforms.CenterCrop(224),\n",
        "        transforms.ToTensor(),\n",
        "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
        "    ])\n",
        "\n",
        "data_dir = '/content/drive/MyDrive/Projects/qBITS/Pneumonia/chest_xray'\n",
        "\n",
        "image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),data_transforms) for x in ['train', 'val','test']}\n",
        "dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val','test']}\n",
        "class_names = image_datasets['train'].classes\n",
        "dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], \n",
        "                  batch_size=batch_size, shuffle=True, drop_last=True) for x in ['train', 'val','test']}\n",
        "\n",
        "def imshow(inp, title=None):\n",
        "    inp = inp.numpy().transpose((1, 2, 0))  \n",
        "    inp = np.clip(inp, 0, 1)\n",
        "    plt.imshow(inp)\n",
        "    if title is not None:\n",
        "        plt.title(title)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GsbwlprzaCgM"
      },
      "source": [
        "Visualization"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 148
        },
        "id": "D48tP_w6wFRa",
        "outputId": "191199b0-e0a8-40df-85e2-6a60ed56cfba"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "#Get a batch of training data\n",
        "inputs, classes = next(iter(dataloaders['train']))\n",
        "\n",
        "#Make a grid from the batch\n",
        "out = torchvision.utils.make_grid(inputs)\n",
        "\n",
        "imshow(out, title=[class_names[x] for x in classes])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3QXF4cl_xF0m"
      },
      "source": [
        "Variational Circuit"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "j5OrvaDcasOB"
      },
      "outputs": [],
      "source": [
        "dev = qml.device('default.qubit', wires=4)\n",
        "def H_layer(nqubits):\n",
        "    for idx in range(nqubits):\n",
        "        qml.Hadamard(wires=idx)\n",
        "\n",
        "\n",
        "def RY_layer(w):\n",
        "    for idx, element in enumerate(w):\n",
        "        qml.RY(element, wires=idx)\n",
        "\n",
        "\n",
        "def entangling_layer(nqubits):\n",
        "    for i in range(0, nqubits - 1, 2):  \n",
        "        qml.CNOT(wires=[i, i + 1])\n",
        "    for i in range(1, nqubits - 1, 2):  \n",
        "        qml.CNOT(wires=[i, i + 1])\n",
        "\n",
        "def U_gate(w):\n",
        "\n",
        "  qml.RX(w[0],wires=0)\n",
        "  qml.RX(w[2],wires=2)\n",
        "  qml.RX(w[3],wires=3)\n",
        "\n",
        "  qml.CNOT(wires=[1,0])\n",
        "\n",
        "  qml.RY(w[1],1)\n",
        "\n",
        "  qml.CNOT(wires=[3,0])\n",
        "\n",
        "\n",
        "def encoding_layer(f,w,nqubits):\n",
        "\n",
        "  for i in range(nqubits):\n",
        "    qml.RY(f[i],i)\n",
        "    qml.Barrier(wires=[0,1,2,3])\n",
        "    U_gate(w)\n",
        "    qml.Barrier(wires=[0,1,2,3])\n",
        "        \n",
        "\n",
        "@qml.qnode(dev, interface=\"torch\")\n",
        "def quantum_net(q_input_features, q_weights_flat):\n",
        "    \"\"\"\n",
        "    The variational quantum circuit.\n",
        "    \"\"\"\n",
        "\n",
        "    # Reshape weights\n",
        "    q_weights = q_weights_flat.reshape(q_depth+1, n_qubits)\n",
        "\n",
        "    # Start from state |+> , unbiased w.r.t. |0> and |1>\n",
        "    H_layer(n_qubits)\n",
        "\n",
        "    qml.Barrier(wires=range(n_qubits))\n",
        "    # Embed features in the quantum node\n",
        "    encoding_layer(q_input_features,q_weights[0],n_qubits)\n",
        "\n",
        "\n",
        "    # Sequence of trainable variational layers\n",
        "    for k in range(q_depth):\n",
        "        qml.Barrier(wires=range(n_qubits))\n",
        "        entangling_layer(n_qubits)\n",
        "        qml.Barrier(wires=range(n_qubits))\n",
        "        encoding_layer(q_input_features,q_weights[k+1],n_qubits)\n",
        "\n",
        "    # Expectation values in the Z basis\n",
        "    qml.Barrier(wires=range(n_qubits))\n",
        "    exp_vals = [qml.expval(qml.PauliZ(position)) for position in range(n_qubits)]\n",
        "    return tuple(exp_vals)\n",
        "class DressedQuantumNet(nn.Module):\n",
        "    \"\"\"\n",
        "    Torch module implementing the *dressed* quantum net.\n",
        "    \"\"\"\n",
        "\n",
        "    def __init__(self):\n",
        "        \"\"\"\n",
        "        Definition of the *dressed* layout.\n",
        "        \"\"\"\n",
        "\n",
        "        super().__init__()\n",
        "        self.pre_net = nn.Linear(512, n_qubits)\n",
        "        self.q_params = nn.Parameter(q_delta * torch.randn((q_depth+1) * n_qubits))\n",
        "        self.post_net = nn.Linear(n_qubits, 2)\n",
        "\n",
        "    def forward(self, input_features):\n",
        "        \"\"\"\n",
        "        Defining how tensors are supposed to move through the *dressed* quantum\n",
        "        net.\n",
        "        \"\"\"\n",
        "\n",
        "        # obtain the input features for the quantum circuit\n",
        "        # by reducing the feature dimension from 512 to 4\n",
        "        pre_out = self.pre_net(input_features)\n",
        "        q_in = torch.tanh(pre_out) * np.pi / 2.0\n",
        "\n",
        "        # Apply the quantum circuit to each element of the batch and append to q_out\n",
        "        q_out = torch.Tensor(0, n_qubits)\n",
        "        q_out = q_out.to(device)\n",
        "        for elem in q_in:\n",
        "            q_out_elem = quantum_net(elem, self.q_params).float().unsqueeze(0)\n",
        "            q_out = torch.cat((q_out, q_out_elem))\n",
        "        \n",
        "        # return the two-dimensional prediction from the postprocessing layer\n",
        "        return self.post_net(q_out)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KBDXUTtoyXbm"
      },
      "outputs": [],
      "source": [
        "model_hybrid = torchvision.models.resnet18(weights='ResNet18_Weights.DEFAULT')\n",
        "for param in model_hybrid.parameters():\n",
        "    param.requires_grad = False\n",
        "# Notice that model_hybrid.fc is the last layer of ResNet18\n",
        "model_hybrid.fc = DressedQuantumNet()\n",
        "\n",
        "model_hybrid = model_hybrid.to(device)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "![download (2).png](data:image/png;base64,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)"
      ],
      "metadata": {
        "id": "JoCfgbSHZQGu"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3Ifzdv-1zY4l",
        "outputId": "5ef24c66-676e-492c-b59f-79c5f75d3d99"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Training started:\n",
            "Phase: train Epoch: 1/3 Iter: 1/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 1/3 Iter: 2/1305 Batch time: 1.1069\n",
            "Phase: train Epoch: 1/3 Iter: 3/1305 Batch time: 1.1160\n",
            "Phase: train Epoch: 1/3 Iter: 4/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 1/3 Iter: 5/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 1/3 Iter: 6/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 1/3 Iter: 7/1305 Batch time: 1.1331\n",
            "Phase: train Epoch: 1/3 Iter: 8/1305 Batch time: 1.1314\n",
            "Phase: train Epoch: 1/3 Iter: 9/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 1/3 Iter: 10/1305 Batch time: 1.1348\n",
            "Phase: train Epoch: 1/3 Iter: 11/1305 Batch time: 1.1419\n",
            "Phase: train Epoch: 1/3 Iter: 12/1305 Batch time: 1.1210\n",
            "Phase: train Epoch: 1/3 Iter: 13/1305 Batch time: 1.6640\n",
            "Phase: train Epoch: 1/3 Iter: 14/1305 Batch time: 1.6106\n",
            "Phase: train Epoch: 1/3 Iter: 15/1305 Batch time: 1.1232\n",
            "Phase: train Epoch: 1/3 Iter: 16/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 1/3 Iter: 17/1305 Batch time: 1.1370\n",
            "Phase: train Epoch: 1/3 Iter: 18/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 1/3 Iter: 19/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 20/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 1/3 Iter: 21/1305 Batch time: 1.1256\n",
            "Phase: train Epoch: 1/3 Iter: 22/1305 Batch time: 1.1325\n",
            "Phase: train Epoch: 1/3 Iter: 23/1305 Batch time: 1.6124\n",
            "Phase: train Epoch: 1/3 Iter: 24/1305 Batch time: 1.1335\n",
            "Phase: train Epoch: 1/3 Iter: 25/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 1/3 Iter: 26/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 27/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 1/3 Iter: 28/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 1/3 Iter: 29/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 1/3 Iter: 30/1305 Batch time: 1.1196\n",
            "Phase: train Epoch: 1/3 Iter: 31/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 1/3 Iter: 32/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 1/3 Iter: 33/1305 Batch time: 1.1357\n",
            "Phase: train Epoch: 1/3 Iter: 34/1305 Batch time: 2.6474\n",
            "Phase: train Epoch: 1/3 Iter: 35/1305 Batch time: 1.1404\n",
            "Phase: train Epoch: 1/3 Iter: 36/1305 Batch time: 1.1732\n",
            "Phase: train Epoch: 1/3 Iter: 37/1305 Batch time: 1.1416\n",
            "Phase: train Epoch: 1/3 Iter: 38/1305 Batch time: 1.1414\n",
            "Phase: train Epoch: 1/3 Iter: 39/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 1/3 Iter: 40/1305 Batch time: 1.1456\n",
            "Phase: train Epoch: 1/3 Iter: 41/1305 Batch time: 1.1161\n",
            "Phase: train Epoch: 1/3 Iter: 42/1305 Batch time: 3.1602\n",
            "Phase: train Epoch: 1/3 Iter: 43/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 1/3 Iter: 44/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 1/3 Iter: 45/1305 Batch time: 1.1412\n",
            "Phase: train Epoch: 1/3 Iter: 46/1305 Batch time: 1.1366\n",
            "Phase: train Epoch: 1/3 Iter: 47/1305 Batch time: 1.1246\n",
            "Phase: train Epoch: 1/3 Iter: 48/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 1/3 Iter: 49/1305 Batch time: 1.1215\n",
            "Phase: train Epoch: 1/3 Iter: 50/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 1/3 Iter: 51/1305 Batch time: 1.1370\n",
            "Phase: train Epoch: 1/3 Iter: 52/1305 Batch time: 1.1455\n",
            "Phase: train Epoch: 1/3 Iter: 53/1305 Batch time: 1.1319\n",
            "Phase: train Epoch: 1/3 Iter: 54/1305 Batch time: 1.1381\n",
            "Phase: train Epoch: 1/3 Iter: 55/1305 Batch time: 1.1580\n",
            "Phase: train Epoch: 1/3 Iter: 56/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 1/3 Iter: 57/1305 Batch time: 1.1317\n",
            "Phase: train Epoch: 1/3 Iter: 58/1305 Batch time: 1.1424\n",
            "Phase: train Epoch: 1/3 Iter: 59/1305 Batch time: 1.2538\n",
            "Phase: train Epoch: 1/3 Iter: 60/1305 Batch time: 1.1378\n",
            "Phase: train Epoch: 1/3 Iter: 61/1305 Batch time: 1.1453\n",
            "Phase: train Epoch: 1/3 Iter: 62/1305 Batch time: 1.1445\n",
            "Phase: train Epoch: 1/3 Iter: 63/1305 Batch time: 1.1201\n",
            "Phase: train Epoch: 1/3 Iter: 64/1305 Batch time: 1.1293\n",
            "Phase: train Epoch: 1/3 Iter: 65/1305 Batch time: 1.1282\n",
            "Phase: train Epoch: 1/3 Iter: 66/1305 Batch time: 1.1337\n",
            "Phase: train Epoch: 1/3 Iter: 67/1305 Batch time: 1.1227\n",
            "Phase: train Epoch: 1/3 Iter: 68/1305 Batch time: 1.1423\n",
            "Phase: train Epoch: 1/3 Iter: 69/1305 Batch time: 1.1413\n",
            "Phase: train Epoch: 1/3 Iter: 70/1305 Batch time: 1.1416\n",
            "Phase: train Epoch: 1/3 Iter: 71/1305 Batch time: 1.1318\n",
            "Phase: train Epoch: 1/3 Iter: 72/1305 Batch time: 1.5705\n",
            "Phase: train Epoch: 1/3 Iter: 73/1305 Batch time: 1.1417\n",
            "Phase: train Epoch: 1/3 Iter: 74/1305 Batch time: 1.1304\n",
            "Phase: train Epoch: 1/3 Iter: 75/1305 Batch time: 1.2397\n",
            "Phase: train Epoch: 1/3 Iter: 76/1305 Batch time: 1.1250\n",
            "Phase: train Epoch: 1/3 Iter: 77/1305 Batch time: 1.1274\n",
            "Phase: train Epoch: 1/3 Iter: 78/1305 Batch time: 1.1401\n",
            "Phase: train Epoch: 1/3 Iter: 79/1305 Batch time: 1.1304\n",
            "Phase: train Epoch: 1/3 Iter: 80/1305 Batch time: 1.1252\n",
            "Phase: train Epoch: 1/3 Iter: 81/1305 Batch time: 1.1372\n",
            "Phase: train Epoch: 1/3 Iter: 82/1305 Batch time: 1.1231\n",
            "Phase: train Epoch: 1/3 Iter: 83/1305 Batch time: 1.1476\n",
            "Phase: train Epoch: 1/3 Iter: 84/1305 Batch time: 1.1306\n",
            "Phase: train Epoch: 1/3 Iter: 85/1305 Batch time: 1.1591\n",
            "Phase: train Epoch: 1/3 Iter: 86/1305 Batch time: 1.2557\n",
            "Phase: train Epoch: 1/3 Iter: 87/1305 Batch time: 1.1449\n",
            "Phase: train Epoch: 1/3 Iter: 88/1305 Batch time: 1.1365\n",
            "Phase: train Epoch: 1/3 Iter: 89/1305 Batch time: 1.1523\n",
            "Phase: train Epoch: 1/3 Iter: 90/1305 Batch time: 1.1492\n",
            "Phase: train Epoch: 1/3 Iter: 91/1305 Batch time: 1.1401\n",
            "Phase: train Epoch: 1/3 Iter: 92/1305 Batch time: 1.1409\n",
            "Phase: train Epoch: 1/3 Iter: 93/1305 Batch time: 1.1315\n",
            "Phase: train Epoch: 1/3 Iter: 94/1305 Batch time: 1.1338\n",
            "Phase: train Epoch: 1/3 Iter: 95/1305 Batch time: 1.1418\n",
            "Phase: train Epoch: 1/3 Iter: 96/1305 Batch time: 1.2530\n",
            "Phase: train Epoch: 1/3 Iter: 97/1305 Batch time: 1.1429\n",
            "Phase: train Epoch: 1/3 Iter: 98/1305 Batch time: 1.1519\n",
            "Phase: train Epoch: 1/3 Iter: 99/1305 Batch time: 1.1515\n",
            "Phase: train Epoch: 1/3 Iter: 100/1305 Batch time: 1.1572\n",
            "Phase: train Epoch: 1/3 Iter: 101/1305 Batch time: 1.1551\n",
            "Phase: train Epoch: 1/3 Iter: 102/1305 Batch time: 1.1294\n",
            "Phase: train Epoch: 1/3 Iter: 103/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 1/3 Iter: 104/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 1/3 Iter: 105/1305 Batch time: 1.0932\n",
            "Phase: train Epoch: 1/3 Iter: 106/1305 Batch time: 1.1801\n",
            "Phase: train Epoch: 1/3 Iter: 107/1305 Batch time: 1.1456\n",
            "Phase: train Epoch: 1/3 Iter: 108/1305 Batch time: 1.2195\n",
            "Phase: train Epoch: 1/3 Iter: 109/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 1/3 Iter: 110/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 1/3 Iter: 111/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 1/3 Iter: 112/1305 Batch time: 1.0871\n",
            "Phase: train Epoch: 1/3 Iter: 113/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 1/3 Iter: 114/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 1/3 Iter: 115/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 1/3 Iter: 116/1305 Batch time: 1.1054\n",
            "Phase: train Epoch: 1/3 Iter: 117/1305 Batch time: 1.1229\n",
            "Phase: train Epoch: 1/3 Iter: 118/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 1/3 Iter: 119/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 1/3 Iter: 120/1305 Batch time: 1.2063\n",
            "Phase: train Epoch: 1/3 Iter: 121/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 1/3 Iter: 122/1305 Batch time: 1.3496\n",
            "Phase: train Epoch: 1/3 Iter: 123/1305 Batch time: 1.4057\n",
            "Phase: train Epoch: 1/3 Iter: 124/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 1/3 Iter: 125/1305 Batch time: 1.0920\n",
            "Phase: train Epoch: 1/3 Iter: 126/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 1/3 Iter: 127/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 1/3 Iter: 128/1305 Batch time: 1.1324\n",
            "Phase: train Epoch: 1/3 Iter: 129/1305 Batch time: 1.1299\n",
            "Phase: train Epoch: 1/3 Iter: 130/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 1/3 Iter: 131/1305 Batch time: 1.1238\n",
            "Phase: train Epoch: 1/3 Iter: 132/1305 Batch time: 1.1196\n",
            "Phase: train Epoch: 1/3 Iter: 133/1305 Batch time: 1.2249\n",
            "Phase: train Epoch: 1/3 Iter: 134/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 1/3 Iter: 135/1305 Batch time: 1.1312\n",
            "Phase: train Epoch: 1/3 Iter: 136/1305 Batch time: 1.1150\n",
            "Phase: train Epoch: 1/3 Iter: 137/1305 Batch time: 1.1318\n",
            "Phase: train Epoch: 1/3 Iter: 138/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 1/3 Iter: 139/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 1/3 Iter: 140/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 1/3 Iter: 141/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 1/3 Iter: 142/1305 Batch time: 1.0910\n",
            "Phase: train Epoch: 1/3 Iter: 143/1305 Batch time: 1.2241\n",
            "Phase: train Epoch: 1/3 Iter: 144/1305 Batch time: 1.1183\n",
            "Phase: train Epoch: 1/3 Iter: 145/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 1/3 Iter: 146/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 1/3 Iter: 147/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 1/3 Iter: 148/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 1/3 Iter: 149/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 1/3 Iter: 150/1305 Batch time: 1.0960\n",
            "Phase: train Epoch: 1/3 Iter: 151/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 1/3 Iter: 152/1305 Batch time: 1.1022\n",
            "Phase: train Epoch: 1/3 Iter: 153/1305 Batch time: 1.2302\n",
            "Phase: train Epoch: 1/3 Iter: 154/1305 Batch time: 1.0882\n",
            "Phase: train Epoch: 1/3 Iter: 155/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 1/3 Iter: 156/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 1/3 Iter: 157/1305 Batch time: 1.0891\n",
            "Phase: train Epoch: 1/3 Iter: 158/1305 Batch time: 1.0847\n",
            "Phase: train Epoch: 1/3 Iter: 159/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 1/3 Iter: 160/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 1/3 Iter: 161/1305 Batch time: 1.2326\n",
            "Phase: train Epoch: 1/3 Iter: 162/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 1/3 Iter: 163/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 1/3 Iter: 164/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 1/3 Iter: 165/1305 Batch time: 1.1102\n",
            "Phase: train Epoch: 1/3 Iter: 166/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 167/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 1/3 Iter: 168/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 1/3 Iter: 169/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 1/3 Iter: 170/1305 Batch time: 1.2120\n",
            "Phase: train Epoch: 1/3 Iter: 171/1305 Batch time: 1.0920\n",
            "Phase: train Epoch: 1/3 Iter: 172/1305 Batch time: 1.1142\n",
            "Phase: train Epoch: 1/3 Iter: 173/1305 Batch time: 1.0881\n",
            "Phase: train Epoch: 1/3 Iter: 174/1305 Batch time: 1.4670\n",
            "Phase: train Epoch: 1/3 Iter: 175/1305 Batch time: 1.1400\n",
            "Phase: train Epoch: 1/3 Iter: 176/1305 Batch time: 1.0875\n",
            "Phase: train Epoch: 1/3 Iter: 177/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 1/3 Iter: 178/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 1/3 Iter: 179/1305 Batch time: 1.1155\n",
            "Phase: train Epoch: 1/3 Iter: 180/1305 Batch time: 1.1256\n",
            "Phase: train Epoch: 1/3 Iter: 181/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 1/3 Iter: 182/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 1/3 Iter: 183/1305 Batch time: 1.0982\n",
            "Phase: train Epoch: 1/3 Iter: 184/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 1/3 Iter: 185/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 1/3 Iter: 186/1305 Batch time: 1.1020\n",
            "Phase: train Epoch: 1/3 Iter: 187/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 1/3 Iter: 188/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 1/3 Iter: 189/1305 Batch time: 1.2198\n",
            "Phase: train Epoch: 1/3 Iter: 190/1305 Batch time: 1.0835\n",
            "Phase: train Epoch: 1/3 Iter: 191/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 192/1305 Batch time: 1.1218\n",
            "Phase: train Epoch: 1/3 Iter: 193/1305 Batch time: 1.0957\n",
            "Phase: train Epoch: 1/3 Iter: 194/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 1/3 Iter: 195/1305 Batch time: 1.1343\n",
            "Phase: train Epoch: 1/3 Iter: 196/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 1/3 Iter: 197/1305 Batch time: 1.0914\n",
            "Phase: train Epoch: 1/3 Iter: 198/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 1/3 Iter: 199/1305 Batch time: 1.1035\n",
            "Phase: train Epoch: 1/3 Iter: 200/1305 Batch time: 1.1325\n",
            "Phase: train Epoch: 1/3 Iter: 201/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 1/3 Iter: 202/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 1/3 Iter: 203/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 1/3 Iter: 204/1305 Batch time: 1.2104\n",
            "Phase: train Epoch: 1/3 Iter: 205/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 1/3 Iter: 206/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 1/3 Iter: 207/1305 Batch time: 1.1206\n",
            "Phase: train Epoch: 1/3 Iter: 208/1305 Batch time: 1.0868\n",
            "Phase: train Epoch: 1/3 Iter: 209/1305 Batch time: 1.1224\n",
            "Phase: train Epoch: 1/3 Iter: 210/1305 Batch time: 1.1023\n",
            "Phase: train Epoch: 1/3 Iter: 211/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 1/3 Iter: 212/1305 Batch time: 1.0920\n",
            "Phase: train Epoch: 1/3 Iter: 213/1305 Batch time: 1.1161\n",
            "Phase: train Epoch: 1/3 Iter: 214/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 1/3 Iter: 215/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 1/3 Iter: 216/1305 Batch time: 1.2297\n",
            "Phase: train Epoch: 1/3 Iter: 217/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 1/3 Iter: 218/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 1/3 Iter: 219/1305 Batch time: 1.1219\n",
            "Phase: train Epoch: 1/3 Iter: 220/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 1/3 Iter: 221/1305 Batch time: 1.0801\n",
            "Phase: train Epoch: 1/3 Iter: 222/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 1/3 Iter: 223/1305 Batch time: 1.2430\n",
            "Phase: train Epoch: 1/3 Iter: 224/1305 Batch time: 1.4713\n",
            "Phase: train Epoch: 1/3 Iter: 225/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 1/3 Iter: 226/1305 Batch time: 1.1221\n",
            "Phase: train Epoch: 1/3 Iter: 227/1305 Batch time: 1.1249\n",
            "Phase: train Epoch: 1/3 Iter: 228/1305 Batch time: 1.2031\n",
            "Phase: train Epoch: 1/3 Iter: 229/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 1/3 Iter: 230/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 1/3 Iter: 231/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 1/3 Iter: 232/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 233/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 1/3 Iter: 234/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 1/3 Iter: 235/1305 Batch time: 1.0938\n",
            "Phase: train Epoch: 1/3 Iter: 236/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 1/3 Iter: 237/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 1/3 Iter: 238/1305 Batch time: 1.2011\n",
            "Phase: train Epoch: 1/3 Iter: 239/1305 Batch time: 1.0978\n",
            "Phase: train Epoch: 1/3 Iter: 240/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 241/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 1/3 Iter: 242/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 243/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 1/3 Iter: 244/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 1/3 Iter: 245/1305 Batch time: 1.1065\n",
            "Phase: train Epoch: 1/3 Iter: 246/1305 Batch time: 1.1283\n",
            "Phase: train Epoch: 1/3 Iter: 247/1305 Batch time: 1.2164\n",
            "Phase: train Epoch: 1/3 Iter: 248/1305 Batch time: 1.0879\n",
            "Phase: train Epoch: 1/3 Iter: 249/1305 Batch time: 1.0961\n",
            "Phase: train Epoch: 1/3 Iter: 250/1305 Batch time: 1.1023\n",
            "Phase: train Epoch: 1/3 Iter: 251/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 1/3 Iter: 252/1305 Batch time: 1.0932\n",
            "Phase: train Epoch: 1/3 Iter: 253/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 1/3 Iter: 254/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 1/3 Iter: 255/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 1/3 Iter: 256/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 1/3 Iter: 257/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 1/3 Iter: 258/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 1/3 Iter: 259/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 1/3 Iter: 260/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 1/3 Iter: 261/1305 Batch time: 1.2017\n",
            "Phase: train Epoch: 1/3 Iter: 262/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 1/3 Iter: 263/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 1/3 Iter: 264/1305 Batch time: 1.1183\n",
            "Phase: train Epoch: 1/3 Iter: 265/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 1/3 Iter: 266/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 1/3 Iter: 267/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 1/3 Iter: 268/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 1/3 Iter: 269/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 1/3 Iter: 270/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 1/3 Iter: 271/1305 Batch time: 1.0858\n",
            "Phase: train Epoch: 1/3 Iter: 272/1305 Batch time: 1.0882\n",
            "Phase: train Epoch: 1/3 Iter: 273/1305 Batch time: 1.2049\n",
            "Phase: train Epoch: 1/3 Iter: 274/1305 Batch time: 1.4252\n",
            "Phase: train Epoch: 1/3 Iter: 275/1305 Batch time: 1.4761\n",
            "Phase: train Epoch: 1/3 Iter: 276/1305 Batch time: 1.1197\n",
            "Phase: train Epoch: 1/3 Iter: 277/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 1/3 Iter: 278/1305 Batch time: 1.1192\n",
            "Phase: train Epoch: 1/3 Iter: 279/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 1/3 Iter: 280/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 281/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 1/3 Iter: 282/1305 Batch time: 1.1023\n",
            "Phase: train Epoch: 1/3 Iter: 283/1305 Batch time: 1.2173\n",
            "Phase: train Epoch: 1/3 Iter: 284/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 285/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 1/3 Iter: 286/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 1/3 Iter: 287/1305 Batch time: 1.1359\n",
            "Phase: train Epoch: 1/3 Iter: 288/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 1/3 Iter: 289/1305 Batch time: 1.1154\n",
            "Phase: train Epoch: 1/3 Iter: 290/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 1/3 Iter: 291/1305 Batch time: 1.2103\n",
            "Phase: train Epoch: 1/3 Iter: 292/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 293/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 1/3 Iter: 294/1305 Batch time: 1.0898\n",
            "Phase: train Epoch: 1/3 Iter: 295/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 1/3 Iter: 296/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 1/3 Iter: 297/1305 Batch time: 1.0908\n",
            "Phase: train Epoch: 1/3 Iter: 298/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 1/3 Iter: 299/1305 Batch time: 1.0888\n",
            "Phase: train Epoch: 1/3 Iter: 300/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 1/3 Iter: 301/1305 Batch time: 1.1367\n",
            "Phase: train Epoch: 1/3 Iter: 302/1305 Batch time: 1.0967\n",
            "Phase: train Epoch: 1/3 Iter: 303/1305 Batch time: 1.0932\n",
            "Phase: train Epoch: 1/3 Iter: 304/1305 Batch time: 1.1230\n",
            "Phase: train Epoch: 1/3 Iter: 305/1305 Batch time: 1.2226\n",
            "Phase: train Epoch: 1/3 Iter: 306/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 1/3 Iter: 307/1305 Batch time: 1.1052\n",
            "Phase: train Epoch: 1/3 Iter: 308/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 309/1305 Batch time: 1.0950\n",
            "Phase: train Epoch: 1/3 Iter: 310/1305 Batch time: 1.1183\n",
            "Phase: train Epoch: 1/3 Iter: 311/1305 Batch time: 1.1294\n",
            "Phase: train Epoch: 1/3 Iter: 312/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 1/3 Iter: 313/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 314/1305 Batch time: 1.0933\n",
            "Phase: train Epoch: 1/3 Iter: 315/1305 Batch time: 1.0944\n",
            "Phase: train Epoch: 1/3 Iter: 316/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 1/3 Iter: 317/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 1/3 Iter: 318/1305 Batch time: 1.1979\n",
            "Phase: train Epoch: 1/3 Iter: 319/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 1/3 Iter: 320/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 1/3 Iter: 321/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 1/3 Iter: 322/1305 Batch time: 1.1166\n",
            "Phase: train Epoch: 1/3 Iter: 323/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 1/3 Iter: 324/1305 Batch time: 1.0963\n",
            "Phase: train Epoch: 1/3 Iter: 325/1305 Batch time: 1.4234\n",
            "Phase: train Epoch: 1/3 Iter: 326/1305 Batch time: 1.3297\n",
            "Phase: train Epoch: 1/3 Iter: 327/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 1/3 Iter: 328/1305 Batch time: 1.1969\n",
            "Phase: train Epoch: 1/3 Iter: 329/1305 Batch time: 1.0891\n",
            "Phase: train Epoch: 1/3 Iter: 330/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 1/3 Iter: 331/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 332/1305 Batch time: 1.0881\n",
            "Phase: train Epoch: 1/3 Iter: 333/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 1/3 Iter: 334/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 1/3 Iter: 335/1305 Batch time: 1.0909\n",
            "Phase: train Epoch: 1/3 Iter: 336/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 1/3 Iter: 337/1305 Batch time: 1.0924\n",
            "Phase: train Epoch: 1/3 Iter: 338/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 1/3 Iter: 339/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 1/3 Iter: 340/1305 Batch time: 1.0885\n",
            "Phase: train Epoch: 1/3 Iter: 341/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 1/3 Iter: 342/1305 Batch time: 1.2199\n",
            "Phase: train Epoch: 1/3 Iter: 343/1305 Batch time: 1.0790\n",
            "Phase: train Epoch: 1/3 Iter: 344/1305 Batch time: 1.0794\n",
            "Phase: train Epoch: 1/3 Iter: 345/1305 Batch time: 1.0934\n",
            "Phase: train Epoch: 1/3 Iter: 346/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 1/3 Iter: 347/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 1/3 Iter: 348/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 1/3 Iter: 349/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 1/3 Iter: 350/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 1/3 Iter: 351/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 1/3 Iter: 352/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 1/3 Iter: 353/1305 Batch time: 1.2009\n",
            "Phase: train Epoch: 1/3 Iter: 354/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 1/3 Iter: 355/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 1/3 Iter: 356/1305 Batch time: 1.0921\n",
            "Phase: train Epoch: 1/3 Iter: 357/1305 Batch time: 1.1134\n",
            "Phase: train Epoch: 1/3 Iter: 358/1305 Batch time: 1.1142\n",
            "Phase: train Epoch: 1/3 Iter: 359/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 1/3 Iter: 360/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 1/3 Iter: 361/1305 Batch time: 1.0886\n",
            "Phase: train Epoch: 1/3 Iter: 362/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 1/3 Iter: 363/1305 Batch time: 1.2037\n",
            "Phase: train Epoch: 1/3 Iter: 364/1305 Batch time: 1.0885\n",
            "Phase: train Epoch: 1/3 Iter: 365/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 366/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 1/3 Iter: 367/1305 Batch time: 1.0905\n",
            "Phase: train Epoch: 1/3 Iter: 368/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 1/3 Iter: 369/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 370/1305 Batch time: 1.1073\n",
            "Phase: train Epoch: 1/3 Iter: 371/1305 Batch time: 1.2138\n",
            "Phase: train Epoch: 1/3 Iter: 372/1305 Batch time: 1.0835\n",
            "Phase: train Epoch: 1/3 Iter: 373/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 1/3 Iter: 374/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 1/3 Iter: 375/1305 Batch time: 1.5044\n",
            "Phase: train Epoch: 1/3 Iter: 376/1305 Batch time: 1.3421\n",
            "Phase: train Epoch: 1/3 Iter: 377/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 1/3 Iter: 378/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 1/3 Iter: 379/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 1/3 Iter: 380/1305 Batch time: 1.2315\n",
            "Phase: train Epoch: 1/3 Iter: 381/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 1/3 Iter: 382/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 1/3 Iter: 383/1305 Batch time: 1.0975\n",
            "Phase: train Epoch: 1/3 Iter: 384/1305 Batch time: 1.0978\n",
            "Phase: train Epoch: 1/3 Iter: 385/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 1/3 Iter: 386/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 1/3 Iter: 387/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 1/3 Iter: 388/1305 Batch time: 1.0828\n",
            "Phase: train Epoch: 1/3 Iter: 389/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 1/3 Iter: 390/1305 Batch time: 1.1253\n",
            "Phase: train Epoch: 1/3 Iter: 391/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 392/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 393/1305 Batch time: 1.1240\n",
            "Phase: train Epoch: 1/3 Iter: 394/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 1/3 Iter: 395/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 1/3 Iter: 396/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 1/3 Iter: 397/1305 Batch time: 1.0936\n",
            "Phase: train Epoch: 1/3 Iter: 398/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 1/3 Iter: 399/1305 Batch time: 1.2078\n",
            "Phase: train Epoch: 1/3 Iter: 400/1305 Batch time: 1.0889\n",
            "Phase: train Epoch: 1/3 Iter: 401/1305 Batch time: 1.0859\n",
            "Phase: train Epoch: 1/3 Iter: 402/1305 Batch time: 1.0922\n",
            "Phase: train Epoch: 1/3 Iter: 403/1305 Batch time: 1.1007\n",
            "Phase: train Epoch: 1/3 Iter: 404/1305 Batch time: 1.0910\n",
            "Phase: train Epoch: 1/3 Iter: 405/1305 Batch time: 1.0995\n",
            "Phase: train Epoch: 1/3 Iter: 406/1305 Batch time: 1.0737\n",
            "Phase: train Epoch: 1/3 Iter: 407/1305 Batch time: 1.0956\n",
            "Phase: train Epoch: 1/3 Iter: 408/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 1/3 Iter: 409/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 1/3 Iter: 410/1305 Batch time: 1.0847\n",
            "Phase: train Epoch: 1/3 Iter: 411/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 1/3 Iter: 412/1305 Batch time: 1.1032\n",
            "Phase: train Epoch: 1/3 Iter: 413/1305 Batch time: 1.0852\n",
            "Phase: train Epoch: 1/3 Iter: 414/1305 Batch time: 1.2127\n",
            "Phase: train Epoch: 1/3 Iter: 415/1305 Batch time: 1.0806\n",
            "Phase: train Epoch: 1/3 Iter: 416/1305 Batch time: 1.1006\n",
            "Phase: train Epoch: 1/3 Iter: 417/1305 Batch time: 1.0848\n",
            "Phase: train Epoch: 1/3 Iter: 418/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 1/3 Iter: 419/1305 Batch time: 1.0880\n",
            "Phase: train Epoch: 1/3 Iter: 420/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 421/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 1/3 Iter: 422/1305 Batch time: 1.1052\n",
            "Phase: train Epoch: 1/3 Iter: 423/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 1/3 Iter: 424/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 1/3 Iter: 425/1305 Batch time: 1.5515\n",
            "Phase: train Epoch: 1/3 Iter: 426/1305 Batch time: 1.3494\n",
            "Phase: train Epoch: 1/3 Iter: 427/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 1/3 Iter: 428/1305 Batch time: 1.0865\n",
            "Phase: train Epoch: 1/3 Iter: 429/1305 Batch time: 1.0942\n",
            "Phase: train Epoch: 1/3 Iter: 430/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 1/3 Iter: 431/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 1/3 Iter: 432/1305 Batch time: 1.0918\n",
            "Phase: train Epoch: 1/3 Iter: 433/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 1/3 Iter: 434/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 1/3 Iter: 435/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 1/3 Iter: 436/1305 Batch time: 1.2155\n",
            "Phase: train Epoch: 1/3 Iter: 437/1305 Batch time: 1.0957\n",
            "Phase: train Epoch: 1/3 Iter: 438/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 1/3 Iter: 439/1305 Batch time: 1.1261\n",
            "Phase: train Epoch: 1/3 Iter: 440/1305 Batch time: 1.0857\n",
            "Phase: train Epoch: 1/3 Iter: 441/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 1/3 Iter: 442/1305 Batch time: 1.1018\n",
            "Phase: train Epoch: 1/3 Iter: 443/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 1/3 Iter: 444/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 1/3 Iter: 445/1305 Batch time: 1.0771\n",
            "Phase: train Epoch: 1/3 Iter: 446/1305 Batch time: 1.2315\n",
            "Phase: train Epoch: 1/3 Iter: 447/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 448/1305 Batch time: 1.0926\n",
            "Phase: train Epoch: 1/3 Iter: 449/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 450/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 1/3 Iter: 451/1305 Batch time: 1.1352\n",
            "Phase: train Epoch: 1/3 Iter: 452/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 1/3 Iter: 453/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 1/3 Iter: 454/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 1/3 Iter: 455/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 1/3 Iter: 456/1305 Batch time: 1.0789\n",
            "Phase: train Epoch: 1/3 Iter: 457/1305 Batch time: 1.1299\n",
            "Phase: train Epoch: 1/3 Iter: 458/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 1/3 Iter: 459/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 1/3 Iter: 460/1305 Batch time: 1.1916\n",
            "Phase: train Epoch: 1/3 Iter: 461/1305 Batch time: 1.1220\n",
            "Phase: train Epoch: 1/3 Iter: 462/1305 Batch time: 1.1160\n",
            "Phase: train Epoch: 1/3 Iter: 463/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 1/3 Iter: 464/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 1/3 Iter: 465/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 1/3 Iter: 466/1305 Batch time: 1.1154\n",
            "Phase: train Epoch: 1/3 Iter: 467/1305 Batch time: 1.1148\n",
            "Phase: train Epoch: 1/3 Iter: 468/1305 Batch time: 1.0917\n",
            "Phase: train Epoch: 1/3 Iter: 469/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 1/3 Iter: 470/1305 Batch time: 1.0970\n",
            "Phase: train Epoch: 1/3 Iter: 471/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 1/3 Iter: 472/1305 Batch time: 1.1189\n",
            "Phase: train Epoch: 1/3 Iter: 473/1305 Batch time: 1.2159\n",
            "Phase: train Epoch: 1/3 Iter: 474/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 1/3 Iter: 475/1305 Batch time: 1.4919\n",
            "Phase: train Epoch: 1/3 Iter: 476/1305 Batch time: 1.2805\n",
            "Phase: train Epoch: 1/3 Iter: 477/1305 Batch time: 1.0934\n",
            "Phase: train Epoch: 1/3 Iter: 478/1305 Batch time: 1.0958\n",
            "Phase: train Epoch: 1/3 Iter: 479/1305 Batch time: 1.0644\n",
            "Phase: train Epoch: 1/3 Iter: 480/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 481/1305 Batch time: 1.0950\n",
            "Phase: train Epoch: 1/3 Iter: 482/1305 Batch time: 1.0898\n",
            "Phase: train Epoch: 1/3 Iter: 483/1305 Batch time: 1.2126\n",
            "Phase: train Epoch: 1/3 Iter: 484/1305 Batch time: 1.1065\n",
            "Phase: train Epoch: 1/3 Iter: 485/1305 Batch time: 1.0989\n",
            "Phase: train Epoch: 1/3 Iter: 486/1305 Batch time: 1.0993\n",
            "Phase: train Epoch: 1/3 Iter: 487/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 1/3 Iter: 488/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 1/3 Iter: 489/1305 Batch time: 1.0823\n",
            "Phase: train Epoch: 1/3 Iter: 490/1305 Batch time: 1.1073\n",
            "Phase: train Epoch: 1/3 Iter: 491/1305 Batch time: 1.0943\n",
            "Phase: train Epoch: 1/3 Iter: 492/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 1/3 Iter: 493/1305 Batch time: 1.2258\n",
            "Phase: train Epoch: 1/3 Iter: 494/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 1/3 Iter: 495/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 1/3 Iter: 496/1305 Batch time: 1.0861\n",
            "Phase: train Epoch: 1/3 Iter: 497/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 1/3 Iter: 498/1305 Batch time: 1.0800\n",
            "Phase: train Epoch: 1/3 Iter: 499/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 1/3 Iter: 500/1305 Batch time: 1.0800\n",
            "Phase: train Epoch: 1/3 Iter: 501/1305 Batch time: 1.2213\n",
            "Phase: train Epoch: 1/3 Iter: 502/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 1/3 Iter: 503/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 504/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 1/3 Iter: 505/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 506/1305 Batch time: 1.1004\n",
            "Phase: train Epoch: 1/3 Iter: 507/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 508/1305 Batch time: 1.0844\n",
            "Phase: train Epoch: 1/3 Iter: 509/1305 Batch time: 1.1187\n",
            "Phase: train Epoch: 1/3 Iter: 510/1305 Batch time: 1.2149\n",
            "Phase: train Epoch: 1/3 Iter: 511/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 512/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 1/3 Iter: 513/1305 Batch time: 1.1144\n",
            "Phase: train Epoch: 1/3 Iter: 514/1305 Batch time: 1.1255\n",
            "Phase: train Epoch: 1/3 Iter: 515/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 1/3 Iter: 516/1305 Batch time: 1.1005\n",
            "Phase: train Epoch: 1/3 Iter: 517/1305 Batch time: 1.0937\n",
            "Phase: train Epoch: 1/3 Iter: 518/1305 Batch time: 1.1135\n",
            "Phase: train Epoch: 1/3 Iter: 519/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 1/3 Iter: 520/1305 Batch time: 1.1132\n",
            "Phase: train Epoch: 1/3 Iter: 521/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 1/3 Iter: 522/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 1/3 Iter: 523/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 1/3 Iter: 524/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 1/3 Iter: 525/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 1/3 Iter: 526/1305 Batch time: 1.0960\n",
            "Phase: train Epoch: 1/3 Iter: 527/1305 Batch time: 1.5344\n",
            "Phase: train Epoch: 1/3 Iter: 528/1305 Batch time: 1.2135\n",
            "Phase: train Epoch: 1/3 Iter: 529/1305 Batch time: 1.1941\n",
            "Phase: train Epoch: 1/3 Iter: 530/1305 Batch time: 1.0853\n",
            "Phase: train Epoch: 1/3 Iter: 531/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 1/3 Iter: 532/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 1/3 Iter: 533/1305 Batch time: 1.0777\n",
            "Phase: train Epoch: 1/3 Iter: 534/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 1/3 Iter: 535/1305 Batch time: 1.0836\n",
            "Phase: train Epoch: 1/3 Iter: 536/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 1/3 Iter: 537/1305 Batch time: 1.0973\n",
            "Phase: train Epoch: 1/3 Iter: 538/1305 Batch time: 1.1102\n",
            "Phase: train Epoch: 1/3 Iter: 539/1305 Batch time: 1.1051\n",
            "Phase: train Epoch: 1/3 Iter: 540/1305 Batch time: 1.0903\n",
            "Phase: train Epoch: 1/3 Iter: 541/1305 Batch time: 1.0855\n",
            "Phase: train Epoch: 1/3 Iter: 542/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 1/3 Iter: 543/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 1/3 Iter: 544/1305 Batch time: 1.2266\n",
            "Phase: train Epoch: 1/3 Iter: 545/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 1/3 Iter: 546/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 1/3 Iter: 547/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 1/3 Iter: 548/1305 Batch time: 1.0933\n",
            "Phase: train Epoch: 1/3 Iter: 549/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 1/3 Iter: 550/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 1/3 Iter: 551/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 1/3 Iter: 552/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 1/3 Iter: 553/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 1/3 Iter: 554/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 1/3 Iter: 555/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 1/3 Iter: 556/1305 Batch time: 1.2181\n",
            "Phase: train Epoch: 1/3 Iter: 557/1305 Batch time: 1.1068\n",
            "Phase: train Epoch: 1/3 Iter: 558/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 1/3 Iter: 559/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 1/3 Iter: 560/1305 Batch time: 1.1275\n",
            "Phase: train Epoch: 1/3 Iter: 561/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 1/3 Iter: 562/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 1/3 Iter: 563/1305 Batch time: 1.1027\n",
            "Phase: train Epoch: 1/3 Iter: 564/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 1/3 Iter: 565/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 1/3 Iter: 566/1305 Batch time: 1.0914\n",
            "Phase: train Epoch: 1/3 Iter: 567/1305 Batch time: 1.0873\n",
            "Phase: train Epoch: 1/3 Iter: 568/1305 Batch time: 1.2067\n",
            "Phase: train Epoch: 1/3 Iter: 569/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 1/3 Iter: 570/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 1/3 Iter: 571/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 1/3 Iter: 572/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 1/3 Iter: 573/1305 Batch time: 1.0930\n",
            "Phase: train Epoch: 1/3 Iter: 574/1305 Batch time: 1.0948\n",
            "Phase: train Epoch: 1/3 Iter: 575/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 1/3 Iter: 576/1305 Batch time: 1.0780\n",
            "Phase: train Epoch: 1/3 Iter: 577/1305 Batch time: 1.5376\n",
            "Phase: train Epoch: 1/3 Iter: 578/1305 Batch time: 1.2511\n",
            "Phase: train Epoch: 1/3 Iter: 579/1305 Batch time: 1.0963\n",
            "Phase: train Epoch: 1/3 Iter: 580/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 1/3 Iter: 581/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 582/1305 Batch time: 1.0921\n",
            "Phase: train Epoch: 1/3 Iter: 583/1305 Batch time: 1.1252\n",
            "Phase: train Epoch: 1/3 Iter: 584/1305 Batch time: 1.0840\n",
            "Phase: train Epoch: 1/3 Iter: 585/1305 Batch time: 1.0944\n",
            "Phase: train Epoch: 1/3 Iter: 586/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 1/3 Iter: 587/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 1/3 Iter: 588/1305 Batch time: 1.0996\n",
            "Phase: train Epoch: 1/3 Iter: 589/1305 Batch time: 1.0776\n",
            "Phase: train Epoch: 1/3 Iter: 590/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 1/3 Iter: 591/1305 Batch time: 1.2208\n",
            "Phase: train Epoch: 1/3 Iter: 592/1305 Batch time: 1.0745\n",
            "Phase: train Epoch: 1/3 Iter: 593/1305 Batch time: 1.0900\n",
            "Phase: train Epoch: 1/3 Iter: 594/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 1/3 Iter: 595/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 1/3 Iter: 596/1305 Batch time: 1.0925\n",
            "Phase: train Epoch: 1/3 Iter: 597/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 1/3 Iter: 598/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 1/3 Iter: 599/1305 Batch time: 1.1207\n",
            "Phase: train Epoch: 1/3 Iter: 600/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 1/3 Iter: 601/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 1/3 Iter: 602/1305 Batch time: 1.2194\n",
            "Phase: train Epoch: 1/3 Iter: 603/1305 Batch time: 1.0897\n",
            "Phase: train Epoch: 1/3 Iter: 604/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 1/3 Iter: 605/1305 Batch time: 1.0922\n",
            "Phase: train Epoch: 1/3 Iter: 606/1305 Batch time: 1.0917\n",
            "Phase: train Epoch: 1/3 Iter: 607/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 608/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 609/1305 Batch time: 1.1009\n",
            "Phase: train Epoch: 1/3 Iter: 610/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 1/3 Iter: 611/1305 Batch time: 1.2235\n",
            "Phase: train Epoch: 1/3 Iter: 612/1305 Batch time: 1.0844\n",
            "Phase: train Epoch: 1/3 Iter: 613/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 1/3 Iter: 614/1305 Batch time: 1.0904\n",
            "Phase: train Epoch: 1/3 Iter: 615/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 616/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 617/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 1/3 Iter: 618/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 619/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 1/3 Iter: 620/1305 Batch time: 1.2028\n",
            "Phase: train Epoch: 1/3 Iter: 621/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 1/3 Iter: 622/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 1/3 Iter: 623/1305 Batch time: 1.5609\n",
            "Phase: train Epoch: 1/3 Iter: 624/1305 Batch time: 1.0901\n",
            "Phase: train Epoch: 1/3 Iter: 625/1305 Batch time: 1.1166\n",
            "Phase: train Epoch: 1/3 Iter: 626/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 1/3 Iter: 627/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 628/1305 Batch time: 1.0938\n",
            "Phase: train Epoch: 1/3 Iter: 629/1305 Batch time: 1.2445\n",
            "Phase: train Epoch: 1/3 Iter: 630/1305 Batch time: 1.0868\n",
            "Phase: train Epoch: 1/3 Iter: 631/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 632/1305 Batch time: 1.0797\n",
            "Phase: train Epoch: 1/3 Iter: 633/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 634/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 1/3 Iter: 635/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 1/3 Iter: 636/1305 Batch time: 1.0917\n",
            "Phase: train Epoch: 1/3 Iter: 637/1305 Batch time: 1.0933\n",
            "Phase: train Epoch: 1/3 Iter: 638/1305 Batch time: 1.0874\n",
            "Phase: train Epoch: 1/3 Iter: 639/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 1/3 Iter: 640/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 1/3 Iter: 641/1305 Batch time: 1.0817\n",
            "Phase: train Epoch: 1/3 Iter: 642/1305 Batch time: 1.0828\n",
            "Phase: train Epoch: 1/3 Iter: 643/1305 Batch time: 1.0975\n",
            "Phase: train Epoch: 1/3 Iter: 644/1305 Batch time: 1.2006\n",
            "Phase: train Epoch: 1/3 Iter: 645/1305 Batch time: 1.0854\n",
            "Phase: train Epoch: 1/3 Iter: 646/1305 Batch time: 1.0857\n",
            "Phase: train Epoch: 1/3 Iter: 647/1305 Batch time: 1.0944\n",
            "Phase: train Epoch: 1/3 Iter: 648/1305 Batch time: 1.0880\n",
            "Phase: train Epoch: 1/3 Iter: 649/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 1/3 Iter: 650/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 1/3 Iter: 651/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 1/3 Iter: 652/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 1/3 Iter: 653/1305 Batch time: 1.0995\n",
            "Phase: train Epoch: 1/3 Iter: 654/1305 Batch time: 1.0878\n",
            "Phase: train Epoch: 1/3 Iter: 655/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 1/3 Iter: 656/1305 Batch time: 1.2173\n",
            "Phase: train Epoch: 1/3 Iter: 657/1305 Batch time: 1.0835\n",
            "Phase: train Epoch: 1/3 Iter: 658/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 1/3 Iter: 659/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 660/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 1/3 Iter: 661/1305 Batch time: 1.0961\n",
            "Phase: train Epoch: 1/3 Iter: 662/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 663/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 664/1305 Batch time: 1.0922\n",
            "Phase: train Epoch: 1/3 Iter: 665/1305 Batch time: 1.0955\n",
            "Phase: train Epoch: 1/3 Iter: 666/1305 Batch time: 1.0749\n",
            "Phase: train Epoch: 1/3 Iter: 667/1305 Batch time: 1.2067\n",
            "Phase: train Epoch: 1/3 Iter: 668/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 669/1305 Batch time: 1.0759\n",
            "Phase: train Epoch: 1/3 Iter: 670/1305 Batch time: 1.0973\n",
            "Phase: train Epoch: 1/3 Iter: 671/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 1/3 Iter: 672/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 1/3 Iter: 673/1305 Batch time: 1.5540\n",
            "Phase: train Epoch: 1/3 Iter: 674/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 1/3 Iter: 675/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 1/3 Iter: 676/1305 Batch time: 1.0861\n",
            "Phase: train Epoch: 1/3 Iter: 677/1305 Batch time: 1.2046\n",
            "Phase: train Epoch: 1/3 Iter: 678/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 679/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 680/1305 Batch time: 1.1007\n",
            "Phase: train Epoch: 1/3 Iter: 681/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 1/3 Iter: 682/1305 Batch time: 1.1024\n",
            "Phase: train Epoch: 1/3 Iter: 683/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 1/3 Iter: 684/1305 Batch time: 1.0858\n",
            "Phase: train Epoch: 1/3 Iter: 685/1305 Batch time: 1.0934\n",
            "Phase: train Epoch: 1/3 Iter: 686/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 1/3 Iter: 687/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 1/3 Iter: 688/1305 Batch time: 1.0857\n",
            "Phase: train Epoch: 1/3 Iter: 689/1305 Batch time: 1.1968\n",
            "Phase: train Epoch: 1/3 Iter: 690/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 691/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 692/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 1/3 Iter: 693/1305 Batch time: 1.0881\n",
            "Phase: train Epoch: 1/3 Iter: 694/1305 Batch time: 1.0932\n",
            "Phase: train Epoch: 1/3 Iter: 695/1305 Batch time: 1.0779\n",
            "Phase: train Epoch: 1/3 Iter: 696/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 1/3 Iter: 697/1305 Batch time: 1.0964\n",
            "Phase: train Epoch: 1/3 Iter: 698/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 1/3 Iter: 699/1305 Batch time: 1.0909\n",
            "Phase: train Epoch: 1/3 Iter: 700/1305 Batch time: 1.1962\n",
            "Phase: train Epoch: 1/3 Iter: 701/1305 Batch time: 1.0884\n",
            "Phase: train Epoch: 1/3 Iter: 702/1305 Batch time: 1.1101\n",
            "Phase: train Epoch: 1/3 Iter: 703/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 1/3 Iter: 704/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 1/3 Iter: 705/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 1/3 Iter: 706/1305 Batch time: 1.0837\n",
            "Phase: train Epoch: 1/3 Iter: 707/1305 Batch time: 1.1020\n",
            "Phase: train Epoch: 1/3 Iter: 708/1305 Batch time: 1.0769\n",
            "Phase: train Epoch: 1/3 Iter: 709/1305 Batch time: 1.2021\n",
            "Phase: train Epoch: 1/3 Iter: 710/1305 Batch time: 1.1083\n",
            "Phase: train Epoch: 1/3 Iter: 711/1305 Batch time: 1.1069\n",
            "Phase: train Epoch: 1/3 Iter: 712/1305 Batch time: 1.0970\n",
            "Phase: train Epoch: 1/3 Iter: 713/1305 Batch time: 1.1277\n",
            "Phase: train Epoch: 1/3 Iter: 714/1305 Batch time: 1.1127\n",
            "Phase: train Epoch: 1/3 Iter: 715/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 1/3 Iter: 716/1305 Batch time: 1.1221\n",
            "Phase: train Epoch: 1/3 Iter: 717/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 1/3 Iter: 718/1305 Batch time: 1.2331\n",
            "Phase: train Epoch: 1/3 Iter: 719/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 1/3 Iter: 720/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 1/3 Iter: 721/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 1/3 Iter: 722/1305 Batch time: 1.1304\n",
            "Phase: train Epoch: 1/3 Iter: 723/1305 Batch time: 1.4717\n",
            "Phase: train Epoch: 1/3 Iter: 724/1305 Batch time: 1.5769\n",
            "Phase: train Epoch: 1/3 Iter: 725/1305 Batch time: 1.1155\n",
            "Phase: train Epoch: 1/3 Iter: 726/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 1/3 Iter: 727/1305 Batch time: 1.2195\n",
            "Phase: train Epoch: 1/3 Iter: 728/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 1/3 Iter: 729/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 1/3 Iter: 730/1305 Batch time: 1.1004\n",
            "Phase: train Epoch: 1/3 Iter: 731/1305 Batch time: 1.1237\n",
            "Phase: train Epoch: 1/3 Iter: 732/1305 Batch time: 1.0967\n",
            "Phase: train Epoch: 1/3 Iter: 733/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 1/3 Iter: 734/1305 Batch time: 1.1027\n",
            "Phase: train Epoch: 1/3 Iter: 735/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 736/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 1/3 Iter: 737/1305 Batch time: 1.0897\n",
            "Phase: train Epoch: 1/3 Iter: 738/1305 Batch time: 1.1013\n",
            "Phase: train Epoch: 1/3 Iter: 739/1305 Batch time: 1.1160\n",
            "Phase: train Epoch: 1/3 Iter: 740/1305 Batch time: 1.0891\n",
            "Phase: train Epoch: 1/3 Iter: 741/1305 Batch time: 1.1029\n",
            "Phase: train Epoch: 1/3 Iter: 742/1305 Batch time: 1.0882\n",
            "Phase: train Epoch: 1/3 Iter: 743/1305 Batch time: 1.0823\n",
            "Phase: train Epoch: 1/3 Iter: 744/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 1/3 Iter: 745/1305 Batch time: 1.1995\n",
            "Phase: train Epoch: 1/3 Iter: 746/1305 Batch time: 1.0901\n",
            "Phase: train Epoch: 1/3 Iter: 747/1305 Batch time: 1.0823\n",
            "Phase: train Epoch: 1/3 Iter: 748/1305 Batch time: 1.0708\n",
            "Phase: train Epoch: 1/3 Iter: 749/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 1/3 Iter: 750/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 1/3 Iter: 751/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 1/3 Iter: 752/1305 Batch time: 1.1196\n",
            "Phase: train Epoch: 1/3 Iter: 753/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 1/3 Iter: 754/1305 Batch time: 1.1051\n",
            "Phase: train Epoch: 1/3 Iter: 755/1305 Batch time: 1.1038\n",
            "Phase: train Epoch: 1/3 Iter: 756/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 1/3 Iter: 757/1305 Batch time: 1.0772\n",
            "Phase: train Epoch: 1/3 Iter: 758/1305 Batch time: 1.1917\n",
            "Phase: train Epoch: 1/3 Iter: 759/1305 Batch time: 1.0885\n",
            "Phase: train Epoch: 1/3 Iter: 760/1305 Batch time: 1.0828\n",
            "Phase: train Epoch: 1/3 Iter: 761/1305 Batch time: 1.0812\n",
            "Phase: train Epoch: 1/3 Iter: 762/1305 Batch time: 1.0912\n",
            "Phase: train Epoch: 1/3 Iter: 763/1305 Batch time: 1.0982\n",
            "Phase: train Epoch: 1/3 Iter: 764/1305 Batch time: 1.0956\n",
            "Phase: train Epoch: 1/3 Iter: 765/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 1/3 Iter: 766/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 767/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 1/3 Iter: 768/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 1/3 Iter: 769/1305 Batch time: 1.1799\n",
            "Phase: train Epoch: 1/3 Iter: 770/1305 Batch time: 1.0689\n",
            "Phase: train Epoch: 1/3 Iter: 771/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 1/3 Iter: 772/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 1/3 Iter: 773/1305 Batch time: 1.0963\n",
            "Phase: train Epoch: 1/3 Iter: 774/1305 Batch time: 1.4874\n",
            "Phase: train Epoch: 1/3 Iter: 775/1305 Batch time: 1.1818\n",
            "Phase: train Epoch: 1/3 Iter: 776/1305 Batch time: 1.0796\n",
            "Phase: train Epoch: 1/3 Iter: 777/1305 Batch time: 1.1230\n",
            "Phase: train Epoch: 1/3 Iter: 778/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 1/3 Iter: 779/1305 Batch time: 1.1905\n",
            "Phase: train Epoch: 1/3 Iter: 780/1305 Batch time: 1.1052\n",
            "Phase: train Epoch: 1/3 Iter: 781/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 782/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 1/3 Iter: 783/1305 Batch time: 1.1153\n",
            "Phase: train Epoch: 1/3 Iter: 784/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 1/3 Iter: 785/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 1/3 Iter: 786/1305 Batch time: 1.0915\n",
            "Phase: train Epoch: 1/3 Iter: 787/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 1/3 Iter: 788/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 1/3 Iter: 789/1305 Batch time: 1.2082\n",
            "Phase: train Epoch: 1/3 Iter: 790/1305 Batch time: 1.1038\n",
            "Phase: train Epoch: 1/3 Iter: 791/1305 Batch time: 1.0865\n",
            "Phase: train Epoch: 1/3 Iter: 792/1305 Batch time: 1.0861\n",
            "Phase: train Epoch: 1/3 Iter: 793/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 1/3 Iter: 794/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 1/3 Iter: 795/1305 Batch time: 1.0868\n",
            "Phase: train Epoch: 1/3 Iter: 796/1305 Batch time: 1.0845\n",
            "Phase: train Epoch: 1/3 Iter: 797/1305 Batch time: 1.1238\n",
            "Phase: train Epoch: 1/3 Iter: 798/1305 Batch time: 1.0860\n",
            "Phase: train Epoch: 1/3 Iter: 799/1305 Batch time: 1.1063\n",
            "Phase: train Epoch: 1/3 Iter: 800/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 1/3 Iter: 801/1305 Batch time: 1.0877\n",
            "Phase: train Epoch: 1/3 Iter: 802/1305 Batch time: 1.0930\n",
            "Phase: train Epoch: 1/3 Iter: 803/1305 Batch time: 1.2032\n",
            "Phase: train Epoch: 1/3 Iter: 804/1305 Batch time: 1.0824\n",
            "Phase: train Epoch: 1/3 Iter: 805/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 1/3 Iter: 806/1305 Batch time: 1.0948\n",
            "Phase: train Epoch: 1/3 Iter: 807/1305 Batch time: 1.0852\n",
            "Phase: train Epoch: 1/3 Iter: 808/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 1/3 Iter: 809/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 1/3 Iter: 810/1305 Batch time: 1.1004\n",
            "Phase: train Epoch: 1/3 Iter: 811/1305 Batch time: 1.0990\n",
            "Phase: train Epoch: 1/3 Iter: 812/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 813/1305 Batch time: 1.0808\n",
            "Phase: train Epoch: 1/3 Iter: 814/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 815/1305 Batch time: 1.1197\n",
            "Phase: train Epoch: 1/3 Iter: 816/1305 Batch time: 1.1913\n",
            "Phase: train Epoch: 1/3 Iter: 817/1305 Batch time: 1.0828\n",
            "Phase: train Epoch: 1/3 Iter: 818/1305 Batch time: 1.0894\n",
            "Phase: train Epoch: 1/3 Iter: 819/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 1/3 Iter: 820/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 1/3 Iter: 821/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 822/1305 Batch time: 1.0857\n",
            "Phase: train Epoch: 1/3 Iter: 823/1305 Batch time: 1.0892\n",
            "Phase: train Epoch: 1/3 Iter: 824/1305 Batch time: 1.3272\n",
            "Phase: train Epoch: 1/3 Iter: 825/1305 Batch time: 1.4653\n",
            "Phase: train Epoch: 1/3 Iter: 826/1305 Batch time: 1.0861\n",
            "Phase: train Epoch: 1/3 Iter: 827/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 1/3 Iter: 828/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 1/3 Iter: 829/1305 Batch time: 1.0836\n",
            "Phase: train Epoch: 1/3 Iter: 830/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 1/3 Iter: 831/1305 Batch time: 1.0780\n",
            "Phase: train Epoch: 1/3 Iter: 832/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 1/3 Iter: 833/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 1/3 Iter: 834/1305 Batch time: 1.1954\n",
            "Phase: train Epoch: 1/3 Iter: 835/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 1/3 Iter: 836/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 837/1305 Batch time: 1.1013\n",
            "Phase: train Epoch: 1/3 Iter: 838/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 1/3 Iter: 839/1305 Batch time: 1.0835\n",
            "Phase: train Epoch: 1/3 Iter: 840/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 1/3 Iter: 841/1305 Batch time: 1.1023\n",
            "Phase: train Epoch: 1/3 Iter: 842/1305 Batch time: 1.0793\n",
            "Phase: train Epoch: 1/3 Iter: 843/1305 Batch time: 1.1776\n",
            "Phase: train Epoch: 1/3 Iter: 844/1305 Batch time: 1.1446\n",
            "Phase: train Epoch: 1/3 Iter: 845/1305 Batch time: 1.0777\n",
            "Phase: train Epoch: 1/3 Iter: 846/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 1/3 Iter: 847/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 1/3 Iter: 848/1305 Batch time: 1.0804\n",
            "Phase: train Epoch: 1/3 Iter: 849/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 1/3 Iter: 850/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 1/3 Iter: 851/1305 Batch time: 1.0924\n",
            "Phase: train Epoch: 1/3 Iter: 852/1305 Batch time: 1.0867\n",
            "Phase: train Epoch: 1/3 Iter: 853/1305 Batch time: 1.1027\n",
            "Phase: train Epoch: 1/3 Iter: 854/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 1/3 Iter: 855/1305 Batch time: 1.1022\n",
            "Phase: train Epoch: 1/3 Iter: 856/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 1/3 Iter: 857/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 1/3 Iter: 858/1305 Batch time: 1.2128\n",
            "Phase: train Epoch: 1/3 Iter: 859/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 1/3 Iter: 860/1305 Batch time: 1.1222\n",
            "Phase: train Epoch: 1/3 Iter: 861/1305 Batch time: 1.0865\n",
            "Phase: train Epoch: 1/3 Iter: 862/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 1/3 Iter: 863/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 1/3 Iter: 864/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 1/3 Iter: 865/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 1/3 Iter: 866/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 1/3 Iter: 867/1305 Batch time: 1.1289\n",
            "Phase: train Epoch: 1/3 Iter: 868/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 1/3 Iter: 869/1305 Batch time: 1.1947\n",
            "Phase: train Epoch: 1/3 Iter: 870/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 1/3 Iter: 871/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 1/3 Iter: 872/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 873/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 1/3 Iter: 874/1305 Batch time: 1.0890\n",
            "Phase: train Epoch: 1/3 Iter: 875/1305 Batch time: 1.4631\n",
            "Phase: train Epoch: 1/3 Iter: 876/1305 Batch time: 1.2822\n",
            "Phase: train Epoch: 1/3 Iter: 877/1305 Batch time: 1.1209\n",
            "Phase: train Epoch: 1/3 Iter: 878/1305 Batch time: 1.1268\n",
            "Phase: train Epoch: 1/3 Iter: 879/1305 Batch time: 1.2115\n",
            "Phase: train Epoch: 1/3 Iter: 880/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 1/3 Iter: 881/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 882/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 1/3 Iter: 883/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 1/3 Iter: 884/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 1/3 Iter: 885/1305 Batch time: 1.0840\n",
            "Phase: train Epoch: 1/3 Iter: 886/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 1/3 Iter: 887/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 1/3 Iter: 888/1305 Batch time: 1.0886\n",
            "Phase: train Epoch: 1/3 Iter: 889/1305 Batch time: 1.0915\n",
            "Phase: train Epoch: 1/3 Iter: 890/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 1/3 Iter: 891/1305 Batch time: 1.2456\n",
            "Phase: train Epoch: 1/3 Iter: 892/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 893/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 1/3 Iter: 894/1305 Batch time: 1.0794\n",
            "Phase: train Epoch: 1/3 Iter: 895/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 1/3 Iter: 896/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 1/3 Iter: 897/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 898/1305 Batch time: 1.0871\n",
            "Phase: train Epoch: 1/3 Iter: 899/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 1/3 Iter: 900/1305 Batch time: 1.1135\n",
            "Phase: train Epoch: 1/3 Iter: 901/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 1/3 Iter: 902/1305 Batch time: 1.0800\n",
            "Phase: train Epoch: 1/3 Iter: 903/1305 Batch time: 1.0869\n",
            "Phase: train Epoch: 1/3 Iter: 904/1305 Batch time: 1.2627\n",
            "Phase: train Epoch: 1/3 Iter: 905/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 1/3 Iter: 906/1305 Batch time: 1.0908\n",
            "Phase: train Epoch: 1/3 Iter: 907/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 1/3 Iter: 908/1305 Batch time: 1.0921\n",
            "Phase: train Epoch: 1/3 Iter: 909/1305 Batch time: 1.1072\n",
            "Phase: train Epoch: 1/3 Iter: 910/1305 Batch time: 1.1376\n",
            "Phase: train Epoch: 1/3 Iter: 911/1305 Batch time: 1.1280\n",
            "Phase: train Epoch: 1/3 Iter: 912/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 913/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 914/1305 Batch time: 1.2141\n",
            "Phase: train Epoch: 1/3 Iter: 915/1305 Batch time: 1.1447\n",
            "Phase: train Epoch: 1/3 Iter: 916/1305 Batch time: 1.1272\n",
            "Phase: train Epoch: 1/3 Iter: 917/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 1/3 Iter: 918/1305 Batch time: 1.1103\n",
            "Phase: train Epoch: 1/3 Iter: 919/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 1/3 Iter: 920/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 1/3 Iter: 921/1305 Batch time: 1.1206\n",
            "Phase: train Epoch: 1/3 Iter: 922/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 1/3 Iter: 923/1305 Batch time: 1.1035\n",
            "Phase: train Epoch: 1/3 Iter: 924/1305 Batch time: 1.1907\n",
            "Phase: train Epoch: 1/3 Iter: 925/1305 Batch time: 1.4506\n",
            "Phase: train Epoch: 1/3 Iter: 926/1305 Batch time: 1.4166\n",
            "Phase: train Epoch: 1/3 Iter: 927/1305 Batch time: 1.1074\n",
            "Phase: train Epoch: 1/3 Iter: 928/1305 Batch time: 1.1032\n",
            "Phase: train Epoch: 1/3 Iter: 929/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 1/3 Iter: 930/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 1/3 Iter: 931/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 1/3 Iter: 932/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 933/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 934/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 1/3 Iter: 935/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 1/3 Iter: 936/1305 Batch time: 1.0962\n",
            "Phase: train Epoch: 1/3 Iter: 937/1305 Batch time: 1.1323\n",
            "Phase: train Epoch: 1/3 Iter: 938/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 1/3 Iter: 939/1305 Batch time: 1.0917\n",
            "Phase: train Epoch: 1/3 Iter: 940/1305 Batch time: 1.1225\n",
            "Phase: train Epoch: 1/3 Iter: 941/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 1/3 Iter: 942/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 1/3 Iter: 943/1305 Batch time: 1.2357\n",
            "Phase: train Epoch: 1/3 Iter: 944/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 945/1305 Batch time: 1.0895\n",
            "Phase: train Epoch: 1/3 Iter: 946/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 1/3 Iter: 947/1305 Batch time: 1.0993\n",
            "Phase: train Epoch: 1/3 Iter: 948/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 1/3 Iter: 949/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 1/3 Iter: 950/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 1/3 Iter: 951/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 1/3 Iter: 952/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 1/3 Iter: 953/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 1/3 Iter: 954/1305 Batch time: 1.1229\n",
            "Phase: train Epoch: 1/3 Iter: 955/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 1/3 Iter: 956/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 1/3 Iter: 957/1305 Batch time: 1.2324\n",
            "Phase: train Epoch: 1/3 Iter: 958/1305 Batch time: 1.1204\n",
            "Phase: train Epoch: 1/3 Iter: 959/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 1/3 Iter: 960/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 1/3 Iter: 961/1305 Batch time: 1.0926\n",
            "Phase: train Epoch: 1/3 Iter: 962/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 1/3 Iter: 963/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 1/3 Iter: 964/1305 Batch time: 1.0915\n",
            "Phase: train Epoch: 1/3 Iter: 965/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 1/3 Iter: 966/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 1/3 Iter: 967/1305 Batch time: 1.1416\n",
            "Phase: train Epoch: 1/3 Iter: 968/1305 Batch time: 1.2382\n",
            "Phase: train Epoch: 1/3 Iter: 969/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 1/3 Iter: 970/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 1/3 Iter: 971/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 1/3 Iter: 972/1305 Batch time: 1.1065\n",
            "Phase: train Epoch: 1/3 Iter: 973/1305 Batch time: 1.1288\n",
            "Phase: train Epoch: 1/3 Iter: 974/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 1/3 Iter: 975/1305 Batch time: 1.2684\n",
            "Phase: train Epoch: 1/3 Iter: 976/1305 Batch time: 1.5341\n",
            "Phase: train Epoch: 1/3 Iter: 977/1305 Batch time: 1.1013\n",
            "Phase: train Epoch: 1/3 Iter: 978/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 1/3 Iter: 979/1305 Batch time: 1.0967\n",
            "Phase: train Epoch: 1/3 Iter: 980/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 1/3 Iter: 981/1305 Batch time: 1.2324\n",
            "Phase: train Epoch: 1/3 Iter: 982/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 1/3 Iter: 983/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 1/3 Iter: 984/1305 Batch time: 1.1484\n",
            "Phase: train Epoch: 1/3 Iter: 985/1305 Batch time: 1.1260\n",
            "Phase: train Epoch: 1/3 Iter: 986/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 1/3 Iter: 987/1305 Batch time: 1.1262\n",
            "Phase: train Epoch: 1/3 Iter: 988/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 1/3 Iter: 989/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 1/3 Iter: 990/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 1/3 Iter: 991/1305 Batch time: 1.2165\n",
            "Phase: train Epoch: 1/3 Iter: 992/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 1/3 Iter: 993/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 1/3 Iter: 994/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 1/3 Iter: 995/1305 Batch time: 1.1222\n",
            "Phase: train Epoch: 1/3 Iter: 996/1305 Batch time: 1.1023\n",
            "Phase: train Epoch: 1/3 Iter: 997/1305 Batch time: 1.1043\n",
            "Phase: train Epoch: 1/3 Iter: 998/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 1/3 Iter: 999/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 1/3 Iter: 1000/1305 Batch time: 1.2297\n",
            "Phase: train Epoch: 1/3 Iter: 1001/1305 Batch time: 1.1102\n",
            "Phase: train Epoch: 1/3 Iter: 1002/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 1/3 Iter: 1003/1305 Batch time: 1.0807\n",
            "Phase: train Epoch: 1/3 Iter: 1004/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 1/3 Iter: 1005/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 1/3 Iter: 1006/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 1/3 Iter: 1007/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 1/3 Iter: 1008/1305 Batch time: 1.1221\n",
            "Phase: train Epoch: 1/3 Iter: 1009/1305 Batch time: 1.2301\n",
            "Phase: train Epoch: 1/3 Iter: 1010/1305 Batch time: 1.1219\n",
            "Phase: train Epoch: 1/3 Iter: 1011/1305 Batch time: 1.0973\n",
            "Phase: train Epoch: 1/3 Iter: 1012/1305 Batch time: 1.0982\n",
            "Phase: train Epoch: 1/3 Iter: 1013/1305 Batch time: 1.1027\n",
            "Phase: train Epoch: 1/3 Iter: 1014/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 1/3 Iter: 1015/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 1/3 Iter: 1016/1305 Batch time: 1.1187\n",
            "Phase: train Epoch: 1/3 Iter: 1017/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 1/3 Iter: 1018/1305 Batch time: 1.1267\n",
            "Phase: train Epoch: 1/3 Iter: 1019/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 1/3 Iter: 1020/1305 Batch time: 1.1054\n",
            "Phase: train Epoch: 1/3 Iter: 1021/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 1/3 Iter: 1022/1305 Batch time: 1.1330\n",
            "Phase: train Epoch: 1/3 Iter: 1023/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 1/3 Iter: 1024/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 1/3 Iter: 1025/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 1/3 Iter: 1026/1305 Batch time: 1.4383\n",
            "Phase: train Epoch: 1/3 Iter: 1027/1305 Batch time: 1.3310\n",
            "Phase: train Epoch: 1/3 Iter: 1028/1305 Batch time: 1.2343\n",
            "Phase: train Epoch: 1/3 Iter: 1029/1305 Batch time: 1.1499\n",
            "Phase: train Epoch: 1/3 Iter: 1030/1305 Batch time: 1.1158\n",
            "Phase: train Epoch: 1/3 Iter: 1031/1305 Batch time: 1.1319\n",
            "Phase: train Epoch: 1/3 Iter: 1032/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 1033/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 1/3 Iter: 1034/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 1/3 Iter: 1035/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 1/3 Iter: 1036/1305 Batch time: 1.1312\n",
            "Phase: train Epoch: 1/3 Iter: 1037/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 1/3 Iter: 1038/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 1/3 Iter: 1039/1305 Batch time: 1.1186\n",
            "Phase: train Epoch: 1/3 Iter: 1040/1305 Batch time: 1.2149\n",
            "Phase: train Epoch: 1/3 Iter: 1041/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 1/3 Iter: 1042/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 1043/1305 Batch time: 1.1210\n",
            "Phase: train Epoch: 1/3 Iter: 1044/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 1/3 Iter: 1045/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 1/3 Iter: 1046/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 1/3 Iter: 1047/1305 Batch time: 1.1355\n",
            "Phase: train Epoch: 1/3 Iter: 1048/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 1/3 Iter: 1049/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 1/3 Iter: 1050/1305 Batch time: 1.1296\n",
            "Phase: train Epoch: 1/3 Iter: 1051/1305 Batch time: 1.2219\n",
            "Phase: train Epoch: 1/3 Iter: 1052/1305 Batch time: 1.1308\n",
            "Phase: train Epoch: 1/3 Iter: 1053/1305 Batch time: 1.1107\n",
            "Phase: train Epoch: 1/3 Iter: 1054/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 1055/1305 Batch time: 1.1279\n",
            "Phase: train Epoch: 1/3 Iter: 1056/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 1/3 Iter: 1057/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 1/3 Iter: 1058/1305 Batch time: 1.1201\n",
            "Phase: train Epoch: 1/3 Iter: 1059/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 1/3 Iter: 1060/1305 Batch time: 1.1209\n",
            "Phase: train Epoch: 1/3 Iter: 1061/1305 Batch time: 1.2254\n",
            "Phase: train Epoch: 1/3 Iter: 1062/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 1/3 Iter: 1063/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 1064/1305 Batch time: 1.1064\n",
            "Phase: train Epoch: 1/3 Iter: 1065/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 1/3 Iter: 1066/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 1067/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 1/3 Iter: 1068/1305 Batch time: 1.1323\n",
            "Phase: train Epoch: 1/3 Iter: 1069/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 1/3 Iter: 1070/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 1/3 Iter: 1071/1305 Batch time: 1.2021\n",
            "Phase: train Epoch: 1/3 Iter: 1072/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 1/3 Iter: 1073/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 1/3 Iter: 1074/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 1/3 Iter: 1075/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 1/3 Iter: 1076/1305 Batch time: 1.5700\n",
            "Phase: train Epoch: 1/3 Iter: 1077/1305 Batch time: 1.2375\n",
            "Phase: train Epoch: 1/3 Iter: 1078/1305 Batch time: 1.1454\n",
            "Phase: train Epoch: 1/3 Iter: 1079/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 1/3 Iter: 1080/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 1/3 Iter: 1081/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 1/3 Iter: 1082/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 1/3 Iter: 1083/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 1/3 Iter: 1084/1305 Batch time: 1.1239\n",
            "Phase: train Epoch: 1/3 Iter: 1085/1305 Batch time: 1.2184\n",
            "Phase: train Epoch: 1/3 Iter: 1086/1305 Batch time: 1.0842\n",
            "Phase: train Epoch: 1/3 Iter: 1087/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 1088/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 1089/1305 Batch time: 1.1020\n",
            "Phase: train Epoch: 1/3 Iter: 1090/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 1091/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 1/3 Iter: 1092/1305 Batch time: 1.1426\n",
            "Phase: train Epoch: 1/3 Iter: 1093/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 1/3 Iter: 1094/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 1095/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 1/3 Iter: 1096/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 1/3 Iter: 1097/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 1/3 Iter: 1098/1305 Batch time: 1.2040\n",
            "Phase: train Epoch: 1/3 Iter: 1099/1305 Batch time: 1.1233\n",
            "Phase: train Epoch: 1/3 Iter: 1100/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 1/3 Iter: 1101/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 1/3 Iter: 1102/1305 Batch time: 1.0909\n",
            "Phase: train Epoch: 1/3 Iter: 1103/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 1/3 Iter: 1104/1305 Batch time: 1.1051\n",
            "Phase: train Epoch: 1/3 Iter: 1105/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 1/3 Iter: 1106/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 1/3 Iter: 1107/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 1/3 Iter: 1108/1305 Batch time: 1.2186\n",
            "Phase: train Epoch: 1/3 Iter: 1109/1305 Batch time: 1.1102\n",
            "Phase: train Epoch: 1/3 Iter: 1110/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 1111/1305 Batch time: 1.0777\n",
            "Phase: train Epoch: 1/3 Iter: 1112/1305 Batch time: 1.0918\n",
            "Phase: train Epoch: 1/3 Iter: 1113/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 1/3 Iter: 1114/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 1/3 Iter: 1115/1305 Batch time: 1.0888\n",
            "Phase: train Epoch: 1/3 Iter: 1116/1305 Batch time: 1.0966\n",
            "Phase: train Epoch: 1/3 Iter: 1117/1305 Batch time: 1.2068\n",
            "Phase: train Epoch: 1/3 Iter: 1118/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 1/3 Iter: 1119/1305 Batch time: 1.0884\n",
            "Phase: train Epoch: 1/3 Iter: 1120/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 1/3 Iter: 1121/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 1122/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 1/3 Iter: 1123/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 1/3 Iter: 1124/1305 Batch time: 1.0873\n",
            "Phase: train Epoch: 1/3 Iter: 1125/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 1/3 Iter: 1126/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 1127/1305 Batch time: 1.0880\n",
            "Phase: train Epoch: 1/3 Iter: 1128/1305 Batch time: 1.3881\n",
            "Phase: train Epoch: 1/3 Iter: 1129/1305 Batch time: 1.2357\n",
            "Phase: train Epoch: 1/3 Iter: 1130/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 1/3 Iter: 1131/1305 Batch time: 1.0821\n",
            "Phase: train Epoch: 1/3 Iter: 1132/1305 Batch time: 1.1022\n",
            "Phase: train Epoch: 1/3 Iter: 1133/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 1/3 Iter: 1134/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 1/3 Iter: 1135/1305 Batch time: 1.0788\n",
            "Phase: train Epoch: 1/3 Iter: 1136/1305 Batch time: 1.0781\n",
            "Phase: train Epoch: 1/3 Iter: 1137/1305 Batch time: 1.2019\n",
            "Phase: train Epoch: 1/3 Iter: 1138/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 1/3 Iter: 1139/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 1/3 Iter: 1140/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 1/3 Iter: 1141/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 1/3 Iter: 1142/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 1/3 Iter: 1143/1305 Batch time: 1.0878\n",
            "Phase: train Epoch: 1/3 Iter: 1144/1305 Batch time: 1.0942\n",
            "Phase: train Epoch: 1/3 Iter: 1145/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 1/3 Iter: 1146/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 1147/1305 Batch time: 1.0926\n",
            "Phase: train Epoch: 1/3 Iter: 1148/1305 Batch time: 1.0816\n",
            "Phase: train Epoch: 1/3 Iter: 1149/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 1/3 Iter: 1150/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 1/3 Iter: 1151/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 1/3 Iter: 1152/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 1/3 Iter: 1153/1305 Batch time: 1.1859\n",
            "Phase: train Epoch: 1/3 Iter: 1154/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 1/3 Iter: 1155/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 1/3 Iter: 1156/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 1/3 Iter: 1157/1305 Batch time: 1.1005\n",
            "Phase: train Epoch: 1/3 Iter: 1158/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 1/3 Iter: 1159/1305 Batch time: 1.1029\n",
            "Phase: train Epoch: 1/3 Iter: 1160/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 1161/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 1/3 Iter: 1162/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 1/3 Iter: 1163/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 1/3 Iter: 1164/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 1/3 Iter: 1165/1305 Batch time: 1.2201\n",
            "Phase: train Epoch: 1/3 Iter: 1166/1305 Batch time: 1.1183\n",
            "Phase: train Epoch: 1/3 Iter: 1167/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 1/3 Iter: 1168/1305 Batch time: 1.0966\n",
            "Phase: train Epoch: 1/3 Iter: 1169/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 1/3 Iter: 1170/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 1/3 Iter: 1171/1305 Batch time: 1.1222\n",
            "Phase: train Epoch: 1/3 Iter: 1172/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 1/3 Iter: 1173/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 1/3 Iter: 1174/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 1/3 Iter: 1175/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 1/3 Iter: 1176/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 1177/1305 Batch time: 1.2196\n",
            "Phase: train Epoch: 1/3 Iter: 1178/1305 Batch time: 1.4679\n",
            "Phase: train Epoch: 1/3 Iter: 1179/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 1/3 Iter: 1180/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 1/3 Iter: 1181/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 1/3 Iter: 1182/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 1/3 Iter: 1183/1305 Batch time: 1.1160\n",
            "Phase: train Epoch: 1/3 Iter: 1184/1305 Batch time: 1.0996\n",
            "Phase: train Epoch: 1/3 Iter: 1185/1305 Batch time: 1.1179\n",
            "Phase: train Epoch: 1/3 Iter: 1186/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 1/3 Iter: 1187/1305 Batch time: 1.2539\n",
            "Phase: train Epoch: 1/3 Iter: 1188/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 1/3 Iter: 1189/1305 Batch time: 1.1234\n",
            "Phase: train Epoch: 1/3 Iter: 1190/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 1/3 Iter: 1191/1305 Batch time: 1.1227\n",
            "Phase: train Epoch: 1/3 Iter: 1192/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 1/3 Iter: 1193/1305 Batch time: 1.1346\n",
            "Phase: train Epoch: 1/3 Iter: 1194/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 1/3 Iter: 1195/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 1/3 Iter: 1196/1305 Batch time: 1.2144\n",
            "Phase: train Epoch: 1/3 Iter: 1197/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 1/3 Iter: 1198/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 1/3 Iter: 1199/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 1/3 Iter: 1200/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 1/3 Iter: 1201/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 1/3 Iter: 1202/1305 Batch time: 1.0943\n",
            "Phase: train Epoch: 1/3 Iter: 1203/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 1/3 Iter: 1204/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 1/3 Iter: 1205/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 1/3 Iter: 1206/1305 Batch time: 1.0885\n",
            "Phase: train Epoch: 1/3 Iter: 1207/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 1/3 Iter: 1208/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 1/3 Iter: 1209/1305 Batch time: 1.0963\n",
            "Phase: train Epoch: 1/3 Iter: 1210/1305 Batch time: 1.1127\n",
            "Phase: train Epoch: 1/3 Iter: 1211/1305 Batch time: 1.2113\n",
            "Phase: train Epoch: 1/3 Iter: 1212/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 1/3 Iter: 1213/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 1/3 Iter: 1214/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 1/3 Iter: 1215/1305 Batch time: 1.1256\n",
            "Phase: train Epoch: 1/3 Iter: 1216/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 1/3 Iter: 1217/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 1/3 Iter: 1218/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 1/3 Iter: 1219/1305 Batch time: 1.1481\n",
            "Phase: train Epoch: 1/3 Iter: 1220/1305 Batch time: 1.1018\n",
            "Phase: train Epoch: 1/3 Iter: 1221/1305 Batch time: 1.1521\n",
            "Phase: train Epoch: 1/3 Iter: 1222/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 1/3 Iter: 1223/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 1/3 Iter: 1224/1305 Batch time: 1.2268\n",
            "Phase: train Epoch: 1/3 Iter: 1225/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 1/3 Iter: 1226/1305 Batch time: 1.1238\n",
            "Phase: train Epoch: 1/3 Iter: 1227/1305 Batch time: 1.1262\n",
            "Phase: train Epoch: 1/3 Iter: 1228/1305 Batch time: 1.1052\n",
            "Phase: train Epoch: 1/3 Iter: 1229/1305 Batch time: 1.1005\n",
            "Phase: train Epoch: 1/3 Iter: 1230/1305 Batch time: 1.5961\n",
            "Phase: train Epoch: 1/3 Iter: 1231/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 1/3 Iter: 1232/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 1/3 Iter: 1233/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 1/3 Iter: 1234/1305 Batch time: 1.2081\n",
            "Phase: train Epoch: 1/3 Iter: 1235/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 1/3 Iter: 1236/1305 Batch time: 1.1123\n",
            "Phase: train Epoch: 1/3 Iter: 1237/1305 Batch time: 1.1358\n",
            "Phase: train Epoch: 1/3 Iter: 1238/1305 Batch time: 1.1023\n",
            "Phase: train Epoch: 1/3 Iter: 1239/1305 Batch time: 1.0947\n",
            "Phase: train Epoch: 1/3 Iter: 1240/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 1/3 Iter: 1241/1305 Batch time: 1.1150\n",
            "Phase: train Epoch: 1/3 Iter: 1242/1305 Batch time: 1.1150\n",
            "Phase: train Epoch: 1/3 Iter: 1243/1305 Batch time: 1.0975\n",
            "Phase: train Epoch: 1/3 Iter: 1244/1305 Batch time: 1.1964\n",
            "Phase: train Epoch: 1/3 Iter: 1245/1305 Batch time: 1.0916\n",
            "Phase: train Epoch: 1/3 Iter: 1246/1305 Batch time: 1.1179\n",
            "Phase: train Epoch: 1/3 Iter: 1247/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 1/3 Iter: 1248/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 1/3 Iter: 1249/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 1/3 Iter: 1250/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 1/3 Iter: 1251/1305 Batch time: 1.1207\n",
            "Phase: train Epoch: 1/3 Iter: 1252/1305 Batch time: 1.2050\n",
            "Phase: train Epoch: 1/3 Iter: 1253/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 1/3 Iter: 1254/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 1/3 Iter: 1255/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 1/3 Iter: 1256/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 1/3 Iter: 1257/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 1/3 Iter: 1258/1305 Batch time: 1.1005\n",
            "Phase: train Epoch: 1/3 Iter: 1259/1305 Batch time: 1.0914\n",
            "Phase: train Epoch: 1/3 Iter: 1260/1305 Batch time: 1.1127\n",
            "Phase: train Epoch: 1/3 Iter: 1261/1305 Batch time: 1.2128\n",
            "Phase: train Epoch: 1/3 Iter: 1262/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 1/3 Iter: 1263/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 1/3 Iter: 1264/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 1/3 Iter: 1265/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 1/3 Iter: 1266/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 1/3 Iter: 1267/1305 Batch time: 1.1310\n",
            "Phase: train Epoch: 1/3 Iter: 1268/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 1/3 Iter: 1269/1305 Batch time: 1.1027\n",
            "Phase: train Epoch: 1/3 Iter: 1270/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 1/3 Iter: 1271/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 1/3 Iter: 1272/1305 Batch time: 1.1102\n",
            "Phase: train Epoch: 1/3 Iter: 1273/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 1/3 Iter: 1274/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 1/3 Iter: 1275/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 1/3 Iter: 1276/1305 Batch time: 1.0957\n",
            "Phase: train Epoch: 1/3 Iter: 1277/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 1/3 Iter: 1278/1305 Batch time: 1.0940\n",
            "Phase: train Epoch: 1/3 Iter: 1279/1305 Batch time: 1.1192\n",
            "Phase: train Epoch: 1/3 Iter: 1280/1305 Batch time: 1.6940\n",
            "Phase: train Epoch: 1/3 Iter: 1281/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 1/3 Iter: 1282/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 1/3 Iter: 1283/1305 Batch time: 1.1232\n",
            "Phase: train Epoch: 1/3 Iter: 1284/1305 Batch time: 1.0943\n",
            "Phase: train Epoch: 1/3 Iter: 1285/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 1/3 Iter: 1286/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 1/3 Iter: 1287/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 1/3 Iter: 1288/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 1/3 Iter: 1289/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 1/3 Iter: 1290/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 1/3 Iter: 1291/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 1/3 Iter: 1292/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 1/3 Iter: 1293/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 1/3 Iter: 1294/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 1/3 Iter: 1295/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 1/3 Iter: 1296/1305 Batch time: 1.2234\n",
            "Phase: train Epoch: 1/3 Iter: 1297/1305 Batch time: 1.1263\n",
            "Phase: train Epoch: 1/3 Iter: 1298/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 1/3 Iter: 1299/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 1/3 Iter: 1300/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 1/3 Iter: 1301/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 1/3 Iter: 1302/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 1/3 Iter: 1303/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 1/3 Iter: 1304/1305 Batch time: 1.0918\n",
            "Phase: train Epoch: 1/3 Loss: 0.4882 Acc: 0.7972        \n",
            "Phase: val Epoch: 1/3 Iter: 1/5 Batch time: 0.7290\n",
            "Phase: val Epoch: 1/3 Iter: 2/5 Batch time: 0.6976\n",
            "Phase: val Epoch: 1/3 Iter: 3/5 Batch time: 0.6840\n",
            "Phase: val Epoch: 1/3 Iter: 4/5 Batch time: 0.7131\n",
            "Phase: val Epoch: 1/3 Loss: 0.6345 Acc: 0.5000        \n",
            "Phase: train Epoch: 2/3 Iter: 1/1305 Batch time: 1.0944\n",
            "Phase: train Epoch: 2/3 Iter: 2/1305 Batch time: 1.2320\n",
            "Phase: train Epoch: 2/3 Iter: 3/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 2/3 Iter: 4/1305 Batch time: 1.0990\n",
            "Phase: train Epoch: 2/3 Iter: 5/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 2/3 Iter: 6/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 2/3 Iter: 7/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 2/3 Iter: 8/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 2/3 Iter: 9/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 2/3 Iter: 10/1305 Batch time: 1.2387\n",
            "Phase: train Epoch: 2/3 Iter: 11/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 12/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 2/3 Iter: 13/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 2/3 Iter: 14/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 2/3 Iter: 15/1305 Batch time: 1.1186\n",
            "Phase: train Epoch: 2/3 Iter: 16/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 2/3 Iter: 17/1305 Batch time: 1.1024\n",
            "Phase: train Epoch: 2/3 Iter: 18/1305 Batch time: 1.1221\n",
            "Phase: train Epoch: 2/3 Iter: 19/1305 Batch time: 1.2075\n",
            "Phase: train Epoch: 2/3 Iter: 20/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 2/3 Iter: 21/1305 Batch time: 1.1251\n",
            "Phase: train Epoch: 2/3 Iter: 22/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 2/3 Iter: 23/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 2/3 Iter: 24/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 2/3 Iter: 25/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 2/3 Iter: 26/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 2/3 Iter: 27/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 2/3 Iter: 28/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 2/3 Iter: 29/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 2/3 Iter: 30/1305 Batch time: 1.1247\n",
            "Phase: train Epoch: 2/3 Iter: 31/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 2/3 Iter: 32/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 2/3 Iter: 33/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 2/3 Iter: 34/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 2/3 Iter: 35/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 2/3 Iter: 36/1305 Batch time: 1.3044\n",
            "Phase: train Epoch: 2/3 Iter: 37/1305 Batch time: 1.5678\n",
            "Phase: train Epoch: 2/3 Iter: 38/1305 Batch time: 1.2460\n",
            "Phase: train Epoch: 2/3 Iter: 39/1305 Batch time: 1.1187\n",
            "Phase: train Epoch: 2/3 Iter: 40/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 2/3 Iter: 41/1305 Batch time: 1.1148\n",
            "Phase: train Epoch: 2/3 Iter: 42/1305 Batch time: 1.1187\n",
            "Phase: train Epoch: 2/3 Iter: 43/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 2/3 Iter: 44/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 2/3 Iter: 45/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 2/3 Iter: 46/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 2/3 Iter: 47/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 2/3 Iter: 48/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 2/3 Iter: 49/1305 Batch time: 1.1165\n",
            "Phase: train Epoch: 2/3 Iter: 50/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 2/3 Iter: 51/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 2/3 Iter: 52/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 2/3 Iter: 53/1305 Batch time: 1.2041\n",
            "Phase: train Epoch: 2/3 Iter: 54/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 2/3 Iter: 55/1305 Batch time: 1.1177\n",
            "Phase: train Epoch: 2/3 Iter: 56/1305 Batch time: 1.1322\n",
            "Phase: train Epoch: 2/3 Iter: 57/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 2/3 Iter: 58/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 2/3 Iter: 59/1305 Batch time: 1.1260\n",
            "Phase: train Epoch: 2/3 Iter: 60/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 2/3 Iter: 61/1305 Batch time: 1.1153\n",
            "Phase: train Epoch: 2/3 Iter: 62/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 2/3 Iter: 63/1305 Batch time: 1.1320\n",
            "Phase: train Epoch: 2/3 Iter: 64/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 2/3 Iter: 65/1305 Batch time: 1.2166\n",
            "Phase: train Epoch: 2/3 Iter: 66/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 2/3 Iter: 67/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 2/3 Iter: 68/1305 Batch time: 1.1222\n",
            "Phase: train Epoch: 2/3 Iter: 69/1305 Batch time: 1.1282\n",
            "Phase: train Epoch: 2/3 Iter: 70/1305 Batch time: 1.1271\n",
            "Phase: train Epoch: 2/3 Iter: 71/1305 Batch time: 1.1294\n",
            "Phase: train Epoch: 2/3 Iter: 72/1305 Batch time: 1.1377\n",
            "Phase: train Epoch: 2/3 Iter: 73/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 2/3 Iter: 74/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 2/3 Iter: 75/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 2/3 Iter: 76/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 2/3 Iter: 77/1305 Batch time: 1.2509\n",
            "Phase: train Epoch: 2/3 Iter: 78/1305 Batch time: 1.1166\n",
            "Phase: train Epoch: 2/3 Iter: 79/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 2/3 Iter: 80/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 2/3 Iter: 81/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 2/3 Iter: 82/1305 Batch time: 1.0897\n",
            "Phase: train Epoch: 2/3 Iter: 83/1305 Batch time: 1.1381\n",
            "Phase: train Epoch: 2/3 Iter: 84/1305 Batch time: 1.1370\n",
            "Phase: train Epoch: 2/3 Iter: 85/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 2/3 Iter: 86/1305 Batch time: 1.1206\n",
            "Phase: train Epoch: 2/3 Iter: 87/1305 Batch time: 1.2021\n",
            "Phase: train Epoch: 2/3 Iter: 88/1305 Batch time: 1.0920\n",
            "Phase: train Epoch: 2/3 Iter: 89/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 2/3 Iter: 90/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 2/3 Iter: 91/1305 Batch time: 1.1038\n",
            "Phase: train Epoch: 2/3 Iter: 92/1305 Batch time: 1.1154\n",
            "Phase: train Epoch: 2/3 Iter: 93/1305 Batch time: 1.1326\n",
            "Phase: train Epoch: 2/3 Iter: 94/1305 Batch time: 1.1177\n",
            "Phase: train Epoch: 2/3 Iter: 95/1305 Batch time: 1.1278\n",
            "Phase: train Epoch: 2/3 Iter: 96/1305 Batch time: 1.2208\n",
            "Phase: train Epoch: 2/3 Iter: 97/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 2/3 Iter: 98/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 2/3 Iter: 99/1305 Batch time: 1.1144\n",
            "Phase: train Epoch: 2/3 Iter: 100/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 2/3 Iter: 101/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 2/3 Iter: 102/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 2/3 Iter: 103/1305 Batch time: 1.1009\n",
            "Phase: train Epoch: 2/3 Iter: 104/1305 Batch time: 1.1269\n",
            "Phase: train Epoch: 2/3 Iter: 105/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 2/3 Iter: 106/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 2/3 Iter: 107/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 2/3 Iter: 108/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 2/3 Iter: 109/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 2/3 Iter: 110/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 2/3 Iter: 111/1305 Batch time: 1.2030\n",
            "Phase: train Epoch: 2/3 Iter: 112/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 2/3 Iter: 113/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 2/3 Iter: 114/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 2/3 Iter: 115/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 2/3 Iter: 116/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 2/3 Iter: 117/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 2/3 Iter: 118/1305 Batch time: 1.5198\n",
            "Phase: train Epoch: 2/3 Iter: 119/1305 Batch time: 1.4039\n",
            "Phase: train Epoch: 2/3 Iter: 120/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 2/3 Iter: 121/1305 Batch time: 1.0963\n",
            "Phase: train Epoch: 2/3 Iter: 122/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 2/3 Iter: 123/1305 Batch time: 1.1385\n",
            "Phase: train Epoch: 2/3 Iter: 124/1305 Batch time: 1.2032\n",
            "Phase: train Epoch: 2/3 Iter: 125/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 2/3 Iter: 126/1305 Batch time: 1.1166\n",
            "Phase: train Epoch: 2/3 Iter: 127/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 2/3 Iter: 128/1305 Batch time: 1.1345\n",
            "Phase: train Epoch: 2/3 Iter: 129/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 2/3 Iter: 130/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 2/3 Iter: 131/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 2/3 Iter: 132/1305 Batch time: 1.1166\n",
            "Phase: train Epoch: 2/3 Iter: 133/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 2/3 Iter: 134/1305 Batch time: 1.2308\n",
            "Phase: train Epoch: 2/3 Iter: 135/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 2/3 Iter: 136/1305 Batch time: 1.1136\n",
            "Phase: train Epoch: 2/3 Iter: 137/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 2/3 Iter: 138/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 2/3 Iter: 139/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 2/3 Iter: 140/1305 Batch time: 1.1357\n",
            "Phase: train Epoch: 2/3 Iter: 141/1305 Batch time: 1.1103\n",
            "Phase: train Epoch: 2/3 Iter: 142/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 2/3 Iter: 143/1305 Batch time: 1.1259\n",
            "Phase: train Epoch: 2/3 Iter: 144/1305 Batch time: 1.2098\n",
            "Phase: train Epoch: 2/3 Iter: 145/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 2/3 Iter: 146/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 2/3 Iter: 147/1305 Batch time: 1.0978\n",
            "Phase: train Epoch: 2/3 Iter: 148/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 2/3 Iter: 149/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 2/3 Iter: 150/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 2/3 Iter: 151/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 2/3 Iter: 152/1305 Batch time: 1.2353\n",
            "Phase: train Epoch: 2/3 Iter: 153/1305 Batch time: 1.0962\n",
            "Phase: train Epoch: 2/3 Iter: 154/1305 Batch time: 1.1338\n",
            "Phase: train Epoch: 2/3 Iter: 155/1305 Batch time: 1.1007\n",
            "Phase: train Epoch: 2/3 Iter: 156/1305 Batch time: 1.1024\n",
            "Phase: train Epoch: 2/3 Iter: 157/1305 Batch time: 1.1155\n",
            "Phase: train Epoch: 2/3 Iter: 158/1305 Batch time: 1.1257\n",
            "Phase: train Epoch: 2/3 Iter: 159/1305 Batch time: 1.0943\n",
            "Phase: train Epoch: 2/3 Iter: 160/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 2/3 Iter: 161/1305 Batch time: 1.2118\n",
            "Phase: train Epoch: 2/3 Iter: 162/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 2/3 Iter: 163/1305 Batch time: 1.1224\n",
            "Phase: train Epoch: 2/3 Iter: 164/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 2/3 Iter: 165/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 2/3 Iter: 166/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 2/3 Iter: 167/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 2/3 Iter: 168/1305 Batch time: 1.0841\n",
            "Phase: train Epoch: 2/3 Iter: 169/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 2/3 Iter: 170/1305 Batch time: 1.0972\n",
            "Phase: train Epoch: 2/3 Iter: 171/1305 Batch time: 1.1043\n",
            "Phase: train Epoch: 2/3 Iter: 172/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 2/3 Iter: 173/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 2/3 Iter: 174/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 2/3 Iter: 175/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 176/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 2/3 Iter: 177/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 2/3 Iter: 178/1305 Batch time: 1.1278\n",
            "Phase: train Epoch: 2/3 Iter: 179/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 2/3 Iter: 180/1305 Batch time: 1.2184\n",
            "Phase: train Epoch: 2/3 Iter: 181/1305 Batch time: 1.1132\n",
            "Phase: train Epoch: 2/3 Iter: 182/1305 Batch time: 1.1136\n",
            "Phase: train Epoch: 2/3 Iter: 183/1305 Batch time: 1.1272\n",
            "Phase: train Epoch: 2/3 Iter: 184/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 2/3 Iter: 185/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 2/3 Iter: 186/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 2/3 Iter: 187/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 2/3 Iter: 188/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 2/3 Iter: 189/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 2/3 Iter: 190/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 2/3 Iter: 191/1305 Batch time: 1.1035\n",
            "Phase: train Epoch: 2/3 Iter: 192/1305 Batch time: 1.1278\n",
            "Phase: train Epoch: 2/3 Iter: 193/1305 Batch time: 1.1246\n",
            "Phase: train Epoch: 2/3 Iter: 194/1305 Batch time: 1.1009\n",
            "Phase: train Epoch: 2/3 Iter: 195/1305 Batch time: 1.2106\n",
            "Phase: train Epoch: 2/3 Iter: 196/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 2/3 Iter: 197/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 2/3 Iter: 198/1305 Batch time: 1.4973\n",
            "Phase: train Epoch: 2/3 Iter: 199/1305 Batch time: 1.4336\n",
            "Phase: train Epoch: 2/3 Iter: 200/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 2/3 Iter: 201/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 2/3 Iter: 202/1305 Batch time: 1.1334\n",
            "Phase: train Epoch: 2/3 Iter: 203/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 2/3 Iter: 204/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 2/3 Iter: 205/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 2/3 Iter: 206/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 2/3 Iter: 207/1305 Batch time: 1.2254\n",
            "Phase: train Epoch: 2/3 Iter: 208/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 2/3 Iter: 209/1305 Batch time: 1.0914\n",
            "Phase: train Epoch: 2/3 Iter: 210/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 2/3 Iter: 211/1305 Batch time: 1.1420\n",
            "Phase: train Epoch: 2/3 Iter: 212/1305 Batch time: 1.0853\n",
            "Phase: train Epoch: 2/3 Iter: 213/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 2/3 Iter: 214/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 2/3 Iter: 215/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 2/3 Iter: 216/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 2/3 Iter: 217/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 2/3 Iter: 218/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 2/3 Iter: 219/1305 Batch time: 1.1989\n",
            "Phase: train Epoch: 2/3 Iter: 220/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 2/3 Iter: 221/1305 Batch time: 1.1052\n",
            "Phase: train Epoch: 2/3 Iter: 222/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 2/3 Iter: 223/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 2/3 Iter: 224/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 2/3 Iter: 225/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 2/3 Iter: 226/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 227/1305 Batch time: 1.1216\n",
            "Phase: train Epoch: 2/3 Iter: 228/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 2/3 Iter: 229/1305 Batch time: 1.2214\n",
            "Phase: train Epoch: 2/3 Iter: 230/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 2/3 Iter: 231/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 2/3 Iter: 232/1305 Batch time: 1.1362\n",
            "Phase: train Epoch: 2/3 Iter: 233/1305 Batch time: 1.1280\n",
            "Phase: train Epoch: 2/3 Iter: 234/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 2/3 Iter: 235/1305 Batch time: 1.1312\n",
            "Phase: train Epoch: 2/3 Iter: 236/1305 Batch time: 1.1161\n",
            "Phase: train Epoch: 2/3 Iter: 237/1305 Batch time: 1.1328\n",
            "Phase: train Epoch: 2/3 Iter: 238/1305 Batch time: 1.2421\n",
            "Phase: train Epoch: 2/3 Iter: 239/1305 Batch time: 1.1255\n",
            "Phase: train Epoch: 2/3 Iter: 240/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 2/3 Iter: 241/1305 Batch time: 1.1142\n",
            "Phase: train Epoch: 2/3 Iter: 242/1305 Batch time: 1.0964\n",
            "Phase: train Epoch: 2/3 Iter: 243/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 2/3 Iter: 244/1305 Batch time: 1.1063\n",
            "Phase: train Epoch: 2/3 Iter: 245/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 2/3 Iter: 246/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 2/3 Iter: 247/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 2/3 Iter: 248/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 2/3 Iter: 249/1305 Batch time: 1.1038\n",
            "Phase: train Epoch: 2/3 Iter: 250/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 2/3 Iter: 251/1305 Batch time: 1.0936\n",
            "Phase: train Epoch: 2/3 Iter: 252/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 2/3 Iter: 253/1305 Batch time: 1.2260\n",
            "Phase: train Epoch: 2/3 Iter: 254/1305 Batch time: 1.1154\n",
            "Phase: train Epoch: 2/3 Iter: 255/1305 Batch time: 1.1101\n",
            "Phase: train Epoch: 2/3 Iter: 256/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 2/3 Iter: 257/1305 Batch time: 1.0973\n",
            "Phase: train Epoch: 2/3 Iter: 258/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 2/3 Iter: 259/1305 Batch time: 1.0936\n",
            "Phase: train Epoch: 2/3 Iter: 260/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 2/3 Iter: 261/1305 Batch time: 1.0933\n",
            "Phase: train Epoch: 2/3 Iter: 262/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 2/3 Iter: 263/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 2/3 Iter: 264/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 2/3 Iter: 265/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 2/3 Iter: 266/1305 Batch time: 1.1911\n",
            "Phase: train Epoch: 2/3 Iter: 267/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 2/3 Iter: 268/1305 Batch time: 1.0975\n",
            "Phase: train Epoch: 2/3 Iter: 269/1305 Batch time: 1.0897\n",
            "Phase: train Epoch: 2/3 Iter: 270/1305 Batch time: 1.0940\n",
            "Phase: train Epoch: 2/3 Iter: 271/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 2/3 Iter: 272/1305 Batch time: 1.1345\n",
            "Phase: train Epoch: 2/3 Iter: 273/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 2/3 Iter: 274/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 2/3 Iter: 275/1305 Batch time: 1.0901\n",
            "Phase: train Epoch: 2/3 Iter: 276/1305 Batch time: 1.2178\n",
            "Phase: train Epoch: 2/3 Iter: 277/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 2/3 Iter: 278/1305 Batch time: 1.1073\n",
            "Phase: train Epoch: 2/3 Iter: 279/1305 Batch time: 1.3962\n",
            "Phase: train Epoch: 2/3 Iter: 280/1305 Batch time: 1.4932\n",
            "Phase: train Epoch: 2/3 Iter: 281/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 2/3 Iter: 282/1305 Batch time: 1.1211\n",
            "Phase: train Epoch: 2/3 Iter: 283/1305 Batch time: 1.0890\n",
            "Phase: train Epoch: 2/3 Iter: 284/1305 Batch time: 1.0865\n",
            "Phase: train Epoch: 2/3 Iter: 285/1305 Batch time: 1.0964\n",
            "Phase: train Epoch: 2/3 Iter: 286/1305 Batch time: 1.2099\n",
            "Phase: train Epoch: 2/3 Iter: 287/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 2/3 Iter: 288/1305 Batch time: 1.0915\n",
            "Phase: train Epoch: 2/3 Iter: 289/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 2/3 Iter: 290/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 2/3 Iter: 291/1305 Batch time: 1.1259\n",
            "Phase: train Epoch: 2/3 Iter: 292/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 2/3 Iter: 293/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 2/3 Iter: 294/1305 Batch time: 1.2143\n",
            "Phase: train Epoch: 2/3 Iter: 295/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 2/3 Iter: 296/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 2/3 Iter: 297/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 2/3 Iter: 298/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 2/3 Iter: 299/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 2/3 Iter: 300/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 2/3 Iter: 301/1305 Batch time: 1.0828\n",
            "Phase: train Epoch: 2/3 Iter: 302/1305 Batch time: 1.1183\n",
            "Phase: train Epoch: 2/3 Iter: 303/1305 Batch time: 1.2194\n",
            "Phase: train Epoch: 2/3 Iter: 304/1305 Batch time: 1.1245\n",
            "Phase: train Epoch: 2/3 Iter: 305/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 2/3 Iter: 306/1305 Batch time: 1.1269\n",
            "Phase: train Epoch: 2/3 Iter: 307/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 2/3 Iter: 308/1305 Batch time: 1.1158\n",
            "Phase: train Epoch: 2/3 Iter: 309/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 2/3 Iter: 310/1305 Batch time: 1.1257\n",
            "Phase: train Epoch: 2/3 Iter: 311/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 312/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 2/3 Iter: 313/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 2/3 Iter: 314/1305 Batch time: 1.1135\n",
            "Phase: train Epoch: 2/3 Iter: 315/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 2/3 Iter: 316/1305 Batch time: 1.0926\n",
            "Phase: train Epoch: 2/3 Iter: 317/1305 Batch time: 1.1054\n",
            "Phase: train Epoch: 2/3 Iter: 318/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 2/3 Iter: 319/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 2/3 Iter: 320/1305 Batch time: 1.1249\n",
            "Phase: train Epoch: 2/3 Iter: 321/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 2/3 Iter: 322/1305 Batch time: 1.2014\n",
            "Phase: train Epoch: 2/3 Iter: 323/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 2/3 Iter: 324/1305 Batch time: 1.1038\n",
            "Phase: train Epoch: 2/3 Iter: 325/1305 Batch time: 1.1373\n",
            "Phase: train Epoch: 2/3 Iter: 326/1305 Batch time: 1.1248\n",
            "Phase: train Epoch: 2/3 Iter: 327/1305 Batch time: 1.1072\n",
            "Phase: train Epoch: 2/3 Iter: 328/1305 Batch time: 1.1274\n",
            "Phase: train Epoch: 2/3 Iter: 329/1305 Batch time: 1.1361\n",
            "Phase: train Epoch: 2/3 Iter: 330/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 2/3 Iter: 331/1305 Batch time: 1.1146\n",
            "Phase: train Epoch: 2/3 Iter: 332/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 2/3 Iter: 333/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 2/3 Iter: 334/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 2/3 Iter: 335/1305 Batch time: 1.1123\n",
            "Phase: train Epoch: 2/3 Iter: 336/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 2/3 Iter: 337/1305 Batch time: 1.2312\n",
            "Phase: train Epoch: 2/3 Iter: 338/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 2/3 Iter: 339/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 2/3 Iter: 340/1305 Batch time: 1.1314\n",
            "Phase: train Epoch: 2/3 Iter: 341/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 2/3 Iter: 342/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 2/3 Iter: 343/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 2/3 Iter: 344/1305 Batch time: 1.1073\n",
            "Phase: train Epoch: 2/3 Iter: 345/1305 Batch time: 1.0948\n",
            "Phase: train Epoch: 2/3 Iter: 346/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 2/3 Iter: 347/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 2/3 Iter: 348/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 2/3 Iter: 349/1305 Batch time: 1.2160\n",
            "Phase: train Epoch: 2/3 Iter: 350/1305 Batch time: 1.0972\n",
            "Phase: train Epoch: 2/3 Iter: 351/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 2/3 Iter: 352/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 2/3 Iter: 353/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 2/3 Iter: 354/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 2/3 Iter: 355/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 2/3 Iter: 356/1305 Batch time: 1.1316\n",
            "Phase: train Epoch: 2/3 Iter: 357/1305 Batch time: 1.1022\n",
            "Phase: train Epoch: 2/3 Iter: 358/1305 Batch time: 1.2210\n",
            "Phase: train Epoch: 2/3 Iter: 359/1305 Batch time: 1.5762\n",
            "Phase: train Epoch: 2/3 Iter: 360/1305 Batch time: 1.2276\n",
            "Phase: train Epoch: 2/3 Iter: 361/1305 Batch time: 1.2397\n",
            "Phase: train Epoch: 2/3 Iter: 362/1305 Batch time: 1.0964\n",
            "Phase: train Epoch: 2/3 Iter: 363/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 2/3 Iter: 364/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 2/3 Iter: 365/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 2/3 Iter: 366/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 2/3 Iter: 367/1305 Batch time: 1.0963\n",
            "Phase: train Epoch: 2/3 Iter: 368/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 2/3 Iter: 369/1305 Batch time: 1.0943\n",
            "Phase: train Epoch: 2/3 Iter: 370/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 2/3 Iter: 371/1305 Batch time: 1.2245\n",
            "Phase: train Epoch: 2/3 Iter: 372/1305 Batch time: 1.0990\n",
            "Phase: train Epoch: 2/3 Iter: 373/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 2/3 Iter: 374/1305 Batch time: 1.0909\n",
            "Phase: train Epoch: 2/3 Iter: 375/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 2/3 Iter: 376/1305 Batch time: 1.1357\n",
            "Phase: train Epoch: 2/3 Iter: 377/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 2/3 Iter: 378/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 2/3 Iter: 379/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 2/3 Iter: 380/1305 Batch time: 1.2443\n",
            "Phase: train Epoch: 2/3 Iter: 381/1305 Batch time: 1.0906\n",
            "Phase: train Epoch: 2/3 Iter: 382/1305 Batch time: 1.1142\n",
            "Phase: train Epoch: 2/3 Iter: 383/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 2/3 Iter: 384/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 2/3 Iter: 385/1305 Batch time: 1.1264\n",
            "Phase: train Epoch: 2/3 Iter: 386/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 2/3 Iter: 387/1305 Batch time: 1.0882\n",
            "Phase: train Epoch: 2/3 Iter: 388/1305 Batch time: 1.1289\n",
            "Phase: train Epoch: 2/3 Iter: 389/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 2/3 Iter: 390/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 2/3 Iter: 391/1305 Batch time: 1.1158\n",
            "Phase: train Epoch: 2/3 Iter: 392/1305 Batch time: 1.1135\n",
            "Phase: train Epoch: 2/3 Iter: 393/1305 Batch time: 1.0908\n",
            "Phase: train Epoch: 2/3 Iter: 394/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 2/3 Iter: 395/1305 Batch time: 1.2274\n",
            "Phase: train Epoch: 2/3 Iter: 396/1305 Batch time: 1.0896\n",
            "Phase: train Epoch: 2/3 Iter: 397/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 2/3 Iter: 398/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 2/3 Iter: 399/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 2/3 Iter: 400/1305 Batch time: 1.1065\n",
            "Phase: train Epoch: 2/3 Iter: 401/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 2/3 Iter: 402/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 2/3 Iter: 403/1305 Batch time: 1.1188\n",
            "Phase: train Epoch: 2/3 Iter: 404/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 2/3 Iter: 405/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 2/3 Iter: 406/1305 Batch time: 1.1320\n",
            "Phase: train Epoch: 2/3 Iter: 407/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 2/3 Iter: 408/1305 Batch time: 1.2303\n",
            "Phase: train Epoch: 2/3 Iter: 409/1305 Batch time: 1.1298\n",
            "Phase: train Epoch: 2/3 Iter: 410/1305 Batch time: 1.0944\n",
            "Phase: train Epoch: 2/3 Iter: 411/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 2/3 Iter: 412/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 2/3 Iter: 413/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 2/3 Iter: 414/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 2/3 Iter: 415/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 2/3 Iter: 416/1305 Batch time: 1.1222\n",
            "Phase: train Epoch: 2/3 Iter: 417/1305 Batch time: 1.0918\n",
            "Phase: train Epoch: 2/3 Iter: 418/1305 Batch time: 1.2336\n",
            "Phase: train Epoch: 2/3 Iter: 419/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 2/3 Iter: 420/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 2/3 Iter: 421/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 2/3 Iter: 422/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 2/3 Iter: 423/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 2/3 Iter: 424/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 2/3 Iter: 425/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 2/3 Iter: 426/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 2/3 Iter: 427/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 2/3 Iter: 428/1305 Batch time: 1.2288\n",
            "Phase: train Epoch: 2/3 Iter: 429/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 2/3 Iter: 430/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 2/3 Iter: 431/1305 Batch time: 1.1310\n",
            "Phase: train Epoch: 2/3 Iter: 432/1305 Batch time: 1.0760\n",
            "Phase: train Epoch: 2/3 Iter: 433/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 2/3 Iter: 434/1305 Batch time: 1.1252\n",
            "Phase: train Epoch: 2/3 Iter: 435/1305 Batch time: 1.1107\n",
            "Phase: train Epoch: 2/3 Iter: 436/1305 Batch time: 1.2147\n",
            "Phase: train Epoch: 2/3 Iter: 437/1305 Batch time: 1.1200\n",
            "Phase: train Epoch: 2/3 Iter: 438/1305 Batch time: 1.1009\n",
            "Phase: train Epoch: 2/3 Iter: 439/1305 Batch time: 1.1725\n",
            "Phase: train Epoch: 2/3 Iter: 440/1305 Batch time: 1.5264\n",
            "Phase: train Epoch: 2/3 Iter: 441/1305 Batch time: 1.2825\n",
            "Phase: train Epoch: 2/3 Iter: 442/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 2/3 Iter: 443/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 2/3 Iter: 444/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 2/3 Iter: 445/1305 Batch time: 1.2105\n",
            "Phase: train Epoch: 2/3 Iter: 446/1305 Batch time: 1.1247\n",
            "Phase: train Epoch: 2/3 Iter: 447/1305 Batch time: 1.1051\n",
            "Phase: train Epoch: 2/3 Iter: 448/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 2/3 Iter: 449/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 2/3 Iter: 450/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 2/3 Iter: 451/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 2/3 Iter: 452/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 2/3 Iter: 453/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 2/3 Iter: 454/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 2/3 Iter: 455/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 2/3 Iter: 456/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 2/3 Iter: 457/1305 Batch time: 1.0850\n",
            "Phase: train Epoch: 2/3 Iter: 458/1305 Batch time: 1.1436\n",
            "Phase: train Epoch: 2/3 Iter: 459/1305 Batch time: 1.1251\n",
            "Phase: train Epoch: 2/3 Iter: 460/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 2/3 Iter: 461/1305 Batch time: 1.1210\n",
            "Phase: train Epoch: 2/3 Iter: 462/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 2/3 Iter: 463/1305 Batch time: 1.1265\n",
            "Phase: train Epoch: 2/3 Iter: 464/1305 Batch time: 1.2442\n",
            "Phase: train Epoch: 2/3 Iter: 465/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 2/3 Iter: 466/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 2/3 Iter: 467/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 2/3 Iter: 468/1305 Batch time: 1.0947\n",
            "Phase: train Epoch: 2/3 Iter: 469/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 2/3 Iter: 470/1305 Batch time: 1.1253\n",
            "Phase: train Epoch: 2/3 Iter: 471/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 2/3 Iter: 472/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 2/3 Iter: 473/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 2/3 Iter: 474/1305 Batch time: 1.1144\n",
            "Phase: train Epoch: 2/3 Iter: 475/1305 Batch time: 1.1165\n",
            "Phase: train Epoch: 2/3 Iter: 476/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 2/3 Iter: 477/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 2/3 Iter: 478/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 2/3 Iter: 479/1305 Batch time: 1.2204\n",
            "Phase: train Epoch: 2/3 Iter: 480/1305 Batch time: 1.1189\n",
            "Phase: train Epoch: 2/3 Iter: 481/1305 Batch time: 1.1177\n",
            "Phase: train Epoch: 2/3 Iter: 482/1305 Batch time: 1.1340\n",
            "Phase: train Epoch: 2/3 Iter: 483/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 2/3 Iter: 484/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 2/3 Iter: 485/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 2/3 Iter: 486/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 2/3 Iter: 487/1305 Batch time: 1.1225\n",
            "Phase: train Epoch: 2/3 Iter: 488/1305 Batch time: 1.1336\n",
            "Phase: train Epoch: 2/3 Iter: 489/1305 Batch time: 1.1068\n",
            "Phase: train Epoch: 2/3 Iter: 490/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 2/3 Iter: 491/1305 Batch time: 1.2239\n",
            "Phase: train Epoch: 2/3 Iter: 492/1305 Batch time: 1.1020\n",
            "Phase: train Epoch: 2/3 Iter: 493/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 2/3 Iter: 494/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 2/3 Iter: 495/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 2/3 Iter: 496/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 2/3 Iter: 497/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 2/3 Iter: 498/1305 Batch time: 1.0859\n",
            "Phase: train Epoch: 2/3 Iter: 499/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 2/3 Iter: 500/1305 Batch time: 1.1189\n",
            "Phase: train Epoch: 2/3 Iter: 501/1305 Batch time: 1.0912\n",
            "Phase: train Epoch: 2/3 Iter: 502/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 2/3 Iter: 503/1305 Batch time: 1.2549\n",
            "Phase: train Epoch: 2/3 Iter: 504/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 2/3 Iter: 505/1305 Batch time: 1.1263\n",
            "Phase: train Epoch: 2/3 Iter: 506/1305 Batch time: 1.1134\n",
            "Phase: train Epoch: 2/3 Iter: 507/1305 Batch time: 1.1137\n",
            "Phase: train Epoch: 2/3 Iter: 508/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 2/3 Iter: 509/1305 Batch time: 1.1384\n",
            "Phase: train Epoch: 2/3 Iter: 510/1305 Batch time: 1.1103\n",
            "Phase: train Epoch: 2/3 Iter: 511/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 2/3 Iter: 512/1305 Batch time: 1.0879\n",
            "Phase: train Epoch: 2/3 Iter: 513/1305 Batch time: 1.2226\n",
            "Phase: train Epoch: 2/3 Iter: 514/1305 Batch time: 1.0978\n",
            "Phase: train Epoch: 2/3 Iter: 515/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 2/3 Iter: 516/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 2/3 Iter: 517/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 2/3 Iter: 518/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 2/3 Iter: 519/1305 Batch time: 1.4224\n",
            "Phase: train Epoch: 2/3 Iter: 520/1305 Batch time: 1.4749\n",
            "Phase: train Epoch: 2/3 Iter: 521/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 2/3 Iter: 522/1305 Batch time: 1.2281\n",
            "Phase: train Epoch: 2/3 Iter: 523/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 2/3 Iter: 524/1305 Batch time: 1.1242\n",
            "Phase: train Epoch: 2/3 Iter: 525/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 2/3 Iter: 526/1305 Batch time: 1.0835\n",
            "Phase: train Epoch: 2/3 Iter: 527/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 2/3 Iter: 528/1305 Batch time: 1.1196\n",
            "Phase: train Epoch: 2/3 Iter: 529/1305 Batch time: 1.1274\n",
            "Phase: train Epoch: 2/3 Iter: 530/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 2/3 Iter: 531/1305 Batch time: 1.0993\n",
            "Phase: train Epoch: 2/3 Iter: 532/1305 Batch time: 1.0982\n",
            "Phase: train Epoch: 2/3 Iter: 533/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 534/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 2/3 Iter: 535/1305 Batch time: 1.0940\n",
            "Phase: train Epoch: 2/3 Iter: 536/1305 Batch time: 1.1153\n",
            "Phase: train Epoch: 2/3 Iter: 537/1305 Batch time: 1.2366\n",
            "Phase: train Epoch: 2/3 Iter: 538/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 2/3 Iter: 539/1305 Batch time: 1.1144\n",
            "Phase: train Epoch: 2/3 Iter: 540/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 2/3 Iter: 541/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 2/3 Iter: 542/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 2/3 Iter: 543/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 2/3 Iter: 544/1305 Batch time: 1.1065\n",
            "Phase: train Epoch: 2/3 Iter: 545/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 2/3 Iter: 546/1305 Batch time: 1.1186\n",
            "Phase: train Epoch: 2/3 Iter: 547/1305 Batch time: 1.1007\n",
            "Phase: train Epoch: 2/3 Iter: 548/1305 Batch time: 1.0891\n",
            "Phase: train Epoch: 2/3 Iter: 549/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 2/3 Iter: 550/1305 Batch time: 1.2207\n",
            "Phase: train Epoch: 2/3 Iter: 551/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 2/3 Iter: 552/1305 Batch time: 1.1347\n",
            "Phase: train Epoch: 2/3 Iter: 553/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 2/3 Iter: 554/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 2/3 Iter: 555/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 2/3 Iter: 556/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 2/3 Iter: 557/1305 Batch time: 1.1294\n",
            "Phase: train Epoch: 2/3 Iter: 558/1305 Batch time: 1.1240\n",
            "Phase: train Epoch: 2/3 Iter: 559/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 2/3 Iter: 560/1305 Batch time: 1.2062\n",
            "Phase: train Epoch: 2/3 Iter: 561/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 2/3 Iter: 562/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 2/3 Iter: 563/1305 Batch time: 1.1132\n",
            "Phase: train Epoch: 2/3 Iter: 564/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 2/3 Iter: 565/1305 Batch time: 1.0903\n",
            "Phase: train Epoch: 2/3 Iter: 566/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 2/3 Iter: 567/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 2/3 Iter: 568/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 2/3 Iter: 569/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 2/3 Iter: 570/1305 Batch time: 1.2022\n",
            "Phase: train Epoch: 2/3 Iter: 571/1305 Batch time: 1.1006\n",
            "Phase: train Epoch: 2/3 Iter: 572/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 2/3 Iter: 573/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 2/3 Iter: 574/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 2/3 Iter: 575/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 2/3 Iter: 576/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 2/3 Iter: 577/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 2/3 Iter: 578/1305 Batch time: 1.2318\n",
            "Phase: train Epoch: 2/3 Iter: 579/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 2/3 Iter: 580/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 581/1305 Batch time: 1.1242\n",
            "Phase: train Epoch: 2/3 Iter: 582/1305 Batch time: 1.1297\n",
            "Phase: train Epoch: 2/3 Iter: 583/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 2/3 Iter: 584/1305 Batch time: 1.1354\n",
            "Phase: train Epoch: 2/3 Iter: 585/1305 Batch time: 1.1359\n",
            "Phase: train Epoch: 2/3 Iter: 586/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 2/3 Iter: 587/1305 Batch time: 1.2359\n",
            "Phase: train Epoch: 2/3 Iter: 588/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 2/3 Iter: 589/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 2/3 Iter: 590/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 2/3 Iter: 591/1305 Batch time: 1.1083\n",
            "Phase: train Epoch: 2/3 Iter: 592/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 2/3 Iter: 593/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 2/3 Iter: 594/1305 Batch time: 1.0955\n",
            "Phase: train Epoch: 2/3 Iter: 595/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 2/3 Iter: 596/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 2/3 Iter: 597/1305 Batch time: 1.0872\n",
            "Phase: train Epoch: 2/3 Iter: 598/1305 Batch time: 1.0951\n",
            "Phase: train Epoch: 2/3 Iter: 599/1305 Batch time: 1.3170\n",
            "Phase: train Epoch: 2/3 Iter: 600/1305 Batch time: 1.5513\n",
            "Phase: train Epoch: 2/3 Iter: 601/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 2/3 Iter: 602/1305 Batch time: 1.1177\n",
            "Phase: train Epoch: 2/3 Iter: 603/1305 Batch time: 1.1253\n",
            "Phase: train Epoch: 2/3 Iter: 604/1305 Batch time: 1.1136\n",
            "Phase: train Epoch: 2/3 Iter: 605/1305 Batch time: 1.0909\n",
            "Phase: train Epoch: 2/3 Iter: 606/1305 Batch time: 1.2184\n",
            "Phase: train Epoch: 2/3 Iter: 607/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 2/3 Iter: 608/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 2/3 Iter: 609/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 2/3 Iter: 610/1305 Batch time: 1.0901\n",
            "Phase: train Epoch: 2/3 Iter: 611/1305 Batch time: 1.0964\n",
            "Phase: train Epoch: 2/3 Iter: 612/1305 Batch time: 1.1035\n",
            "Phase: train Epoch: 2/3 Iter: 613/1305 Batch time: 1.1068\n",
            "Phase: train Epoch: 2/3 Iter: 614/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 2/3 Iter: 615/1305 Batch time: 1.1103\n",
            "Phase: train Epoch: 2/3 Iter: 616/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 2/3 Iter: 617/1305 Batch time: 1.0874\n",
            "Phase: train Epoch: 2/3 Iter: 618/1305 Batch time: 1.1038\n",
            "Phase: train Epoch: 2/3 Iter: 619/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 2/3 Iter: 620/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 2/3 Iter: 621/1305 Batch time: 1.2120\n",
            "Phase: train Epoch: 2/3 Iter: 622/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 2/3 Iter: 623/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 2/3 Iter: 624/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 2/3 Iter: 625/1305 Batch time: 1.0865\n",
            "Phase: train Epoch: 2/3 Iter: 626/1305 Batch time: 1.0922\n",
            "Phase: train Epoch: 2/3 Iter: 627/1305 Batch time: 1.0943\n",
            "Phase: train Epoch: 2/3 Iter: 628/1305 Batch time: 1.0832\n",
            "Phase: train Epoch: 2/3 Iter: 629/1305 Batch time: 1.0804\n",
            "Phase: train Epoch: 2/3 Iter: 630/1305 Batch time: 1.0993\n",
            "Phase: train Epoch: 2/3 Iter: 631/1305 Batch time: 1.0775\n",
            "Phase: train Epoch: 2/3 Iter: 632/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 2/3 Iter: 633/1305 Batch time: 1.2155\n",
            "Phase: train Epoch: 2/3 Iter: 634/1305 Batch time: 1.0789\n",
            "Phase: train Epoch: 2/3 Iter: 635/1305 Batch time: 1.0852\n",
            "Phase: train Epoch: 2/3 Iter: 636/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 2/3 Iter: 637/1305 Batch time: 1.1107\n",
            "Phase: train Epoch: 2/3 Iter: 638/1305 Batch time: 1.0858\n",
            "Phase: train Epoch: 2/3 Iter: 639/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 2/3 Iter: 640/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 2/3 Iter: 641/1305 Batch time: 1.0896\n",
            "Phase: train Epoch: 2/3 Iter: 642/1305 Batch time: 1.0854\n",
            "Phase: train Epoch: 2/3 Iter: 643/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 2/3 Iter: 644/1305 Batch time: 1.1165\n",
            "Phase: train Epoch: 2/3 Iter: 645/1305 Batch time: 1.2055\n",
            "Phase: train Epoch: 2/3 Iter: 646/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 2/3 Iter: 647/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 2/3 Iter: 648/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 2/3 Iter: 649/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 2/3 Iter: 650/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 2/3 Iter: 651/1305 Batch time: 1.0930\n",
            "Phase: train Epoch: 2/3 Iter: 652/1305 Batch time: 1.0903\n",
            "Phase: train Epoch: 2/3 Iter: 653/1305 Batch time: 1.0769\n",
            "Phase: train Epoch: 2/3 Iter: 654/1305 Batch time: 1.0808\n",
            "Phase: train Epoch: 2/3 Iter: 655/1305 Batch time: 1.1891\n",
            "Phase: train Epoch: 2/3 Iter: 656/1305 Batch time: 1.0781\n",
            "Phase: train Epoch: 2/3 Iter: 657/1305 Batch time: 1.0740\n",
            "Phase: train Epoch: 2/3 Iter: 658/1305 Batch time: 1.1286\n",
            "Phase: train Epoch: 2/3 Iter: 659/1305 Batch time: 1.0938\n",
            "Phase: train Epoch: 2/3 Iter: 660/1305 Batch time: 1.0712\n",
            "Phase: train Epoch: 2/3 Iter: 661/1305 Batch time: 1.0633\n",
            "Phase: train Epoch: 2/3 Iter: 662/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 2/3 Iter: 663/1305 Batch time: 1.0749\n",
            "Phase: train Epoch: 2/3 Iter: 664/1305 Batch time: 1.1769\n",
            "Phase: train Epoch: 2/3 Iter: 665/1305 Batch time: 1.1511\n",
            "Phase: train Epoch: 2/3 Iter: 666/1305 Batch time: 1.2158\n",
            "Phase: train Epoch: 2/3 Iter: 667/1305 Batch time: 1.1726\n",
            "Phase: train Epoch: 2/3 Iter: 668/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 2/3 Iter: 669/1305 Batch time: 1.1473\n",
            "Phase: train Epoch: 2/3 Iter: 670/1305 Batch time: 1.1409\n",
            "Phase: train Epoch: 2/3 Iter: 671/1305 Batch time: 1.1539\n",
            "Phase: train Epoch: 2/3 Iter: 672/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 2/3 Iter: 673/1305 Batch time: 1.1331\n",
            "Phase: train Epoch: 2/3 Iter: 674/1305 Batch time: 1.1407\n",
            "Phase: train Epoch: 2/3 Iter: 675/1305 Batch time: 1.1319\n",
            "Phase: train Epoch: 2/3 Iter: 676/1305 Batch time: 1.1281\n",
            "Phase: train Epoch: 2/3 Iter: 677/1305 Batch time: 1.1415\n",
            "Phase: train Epoch: 2/3 Iter: 678/1305 Batch time: 1.1588\n",
            "Phase: train Epoch: 2/3 Iter: 679/1305 Batch time: 1.2555\n",
            "Phase: train Epoch: 2/3 Iter: 680/1305 Batch time: 1.3756\n",
            "Phase: train Epoch: 2/3 Iter: 681/1305 Batch time: 1.5815\n",
            "Phase: train Epoch: 2/3 Iter: 682/1305 Batch time: 1.1644\n",
            "Phase: train Epoch: 2/3 Iter: 683/1305 Batch time: 1.1441\n",
            "Phase: train Epoch: 2/3 Iter: 684/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 2/3 Iter: 685/1305 Batch time: 1.1296\n",
            "Phase: train Epoch: 2/3 Iter: 686/1305 Batch time: 1.1500\n",
            "Phase: train Epoch: 2/3 Iter: 687/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 2/3 Iter: 688/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 2/3 Iter: 689/1305 Batch time: 1.1430\n",
            "Phase: train Epoch: 2/3 Iter: 690/1305 Batch time: 1.1343\n",
            "Phase: train Epoch: 2/3 Iter: 691/1305 Batch time: 1.1300\n",
            "Phase: train Epoch: 2/3 Iter: 692/1305 Batch time: 1.2332\n",
            "Phase: train Epoch: 2/3 Iter: 693/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 2/3 Iter: 694/1305 Batch time: 1.1288\n",
            "Phase: train Epoch: 2/3 Iter: 695/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 2/3 Iter: 696/1305 Batch time: 1.1243\n",
            "Phase: train Epoch: 2/3 Iter: 697/1305 Batch time: 1.1362\n",
            "Phase: train Epoch: 2/3 Iter: 698/1305 Batch time: 1.1357\n",
            "Phase: train Epoch: 2/3 Iter: 699/1305 Batch time: 1.1366\n",
            "Phase: train Epoch: 2/3 Iter: 700/1305 Batch time: 1.1326\n",
            "Phase: train Epoch: 2/3 Iter: 701/1305 Batch time: 1.1379\n",
            "Phase: train Epoch: 2/3 Iter: 702/1305 Batch time: 1.2644\n",
            "Phase: train Epoch: 2/3 Iter: 703/1305 Batch time: 1.1322\n",
            "Phase: train Epoch: 2/3 Iter: 704/1305 Batch time: 1.1464\n",
            "Phase: train Epoch: 2/3 Iter: 705/1305 Batch time: 1.1411\n",
            "Phase: train Epoch: 2/3 Iter: 706/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 2/3 Iter: 707/1305 Batch time: 1.1691\n",
            "Phase: train Epoch: 2/3 Iter: 708/1305 Batch time: 1.1434\n",
            "Phase: train Epoch: 2/3 Iter: 709/1305 Batch time: 1.1357\n",
            "Phase: train Epoch: 2/3 Iter: 710/1305 Batch time: 1.1357\n",
            "Phase: train Epoch: 2/3 Iter: 711/1305 Batch time: 1.1404\n",
            "Phase: train Epoch: 2/3 Iter: 712/1305 Batch time: 1.2544\n",
            "Phase: train Epoch: 2/3 Iter: 713/1305 Batch time: 1.1345\n",
            "Phase: train Epoch: 2/3 Iter: 714/1305 Batch time: 1.1371\n",
            "Phase: train Epoch: 2/3 Iter: 715/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 2/3 Iter: 716/1305 Batch time: 1.1296\n",
            "Phase: train Epoch: 2/3 Iter: 717/1305 Batch time: 1.1229\n",
            "Phase: train Epoch: 2/3 Iter: 718/1305 Batch time: 1.1217\n",
            "Phase: train Epoch: 2/3 Iter: 719/1305 Batch time: 1.1363\n",
            "Phase: train Epoch: 2/3 Iter: 720/1305 Batch time: 1.2360\n",
            "Phase: train Epoch: 2/3 Iter: 721/1305 Batch time: 1.1424\n",
            "Phase: train Epoch: 2/3 Iter: 722/1305 Batch time: 1.1563\n",
            "Phase: train Epoch: 2/3 Iter: 723/1305 Batch time: 1.1272\n",
            "Phase: train Epoch: 2/3 Iter: 724/1305 Batch time: 1.1429\n",
            "Phase: train Epoch: 2/3 Iter: 725/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 2/3 Iter: 726/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 2/3 Iter: 727/1305 Batch time: 1.1442\n",
            "Phase: train Epoch: 2/3 Iter: 728/1305 Batch time: 1.1355\n",
            "Phase: train Epoch: 2/3 Iter: 729/1305 Batch time: 1.2236\n",
            "Phase: train Epoch: 2/3 Iter: 730/1305 Batch time: 1.1170\n",
            "Phase: train Epoch: 2/3 Iter: 731/1305 Batch time: 1.1478\n",
            "Phase: train Epoch: 2/3 Iter: 732/1305 Batch time: 1.1433\n",
            "Phase: train Epoch: 2/3 Iter: 733/1305 Batch time: 1.1312\n",
            "Phase: train Epoch: 2/3 Iter: 734/1305 Batch time: 1.1333\n",
            "Phase: train Epoch: 2/3 Iter: 735/1305 Batch time: 1.1502\n",
            "Phase: train Epoch: 2/3 Iter: 736/1305 Batch time: 1.1282\n",
            "Phase: train Epoch: 2/3 Iter: 737/1305 Batch time: 1.1217\n",
            "Phase: train Epoch: 2/3 Iter: 738/1305 Batch time: 1.1299\n",
            "Phase: train Epoch: 2/3 Iter: 739/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 2/3 Iter: 740/1305 Batch time: 1.1336\n",
            "Phase: train Epoch: 2/3 Iter: 741/1305 Batch time: 1.1298\n",
            "Phase: train Epoch: 2/3 Iter: 742/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 2/3 Iter: 743/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 2/3 Iter: 744/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 2/3 Iter: 745/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 2/3 Iter: 746/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 2/3 Iter: 747/1305 Batch time: 1.1358\n",
            "Phase: train Epoch: 2/3 Iter: 748/1305 Batch time: 1.2266\n",
            "Phase: train Epoch: 2/3 Iter: 749/1305 Batch time: 1.1351\n",
            "Phase: train Epoch: 2/3 Iter: 750/1305 Batch time: 1.1274\n",
            "Phase: train Epoch: 2/3 Iter: 751/1305 Batch time: 1.1376\n",
            "Phase: train Epoch: 2/3 Iter: 752/1305 Batch time: 1.1355\n",
            "Phase: train Epoch: 2/3 Iter: 753/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 2/3 Iter: 754/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 2/3 Iter: 755/1305 Batch time: 1.1421\n",
            "Phase: train Epoch: 2/3 Iter: 756/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 2/3 Iter: 757/1305 Batch time: 1.1415\n",
            "Phase: train Epoch: 2/3 Iter: 758/1305 Batch time: 1.1367\n",
            "Phase: train Epoch: 2/3 Iter: 759/1305 Batch time: 1.3657\n",
            "Phase: train Epoch: 2/3 Iter: 760/1305 Batch time: 1.5817\n",
            "Phase: train Epoch: 2/3 Iter: 761/1305 Batch time: 1.1232\n",
            "Phase: train Epoch: 2/3 Iter: 762/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 763/1305 Batch time: 1.2626\n",
            "Phase: train Epoch: 2/3 Iter: 764/1305 Batch time: 1.1324\n",
            "Phase: train Epoch: 2/3 Iter: 765/1305 Batch time: 1.1380\n",
            "Phase: train Epoch: 2/3 Iter: 766/1305 Batch time: 1.1413\n",
            "Phase: train Epoch: 2/3 Iter: 767/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 2/3 Iter: 768/1305 Batch time: 1.1487\n",
            "Phase: train Epoch: 2/3 Iter: 769/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 770/1305 Batch time: 1.1345\n",
            "Phase: train Epoch: 2/3 Iter: 771/1305 Batch time: 1.1372\n",
            "Phase: train Epoch: 2/3 Iter: 772/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 2/3 Iter: 773/1305 Batch time: 1.1364\n",
            "Phase: train Epoch: 2/3 Iter: 774/1305 Batch time: 1.1386\n",
            "Phase: train Epoch: 2/3 Iter: 775/1305 Batch time: 1.2435\n",
            "Phase: train Epoch: 2/3 Iter: 776/1305 Batch time: 1.1376\n",
            "Phase: train Epoch: 2/3 Iter: 777/1305 Batch time: 1.1315\n",
            "Phase: train Epoch: 2/3 Iter: 778/1305 Batch time: 1.1224\n",
            "Phase: train Epoch: 2/3 Iter: 779/1305 Batch time: 1.1309\n",
            "Phase: train Epoch: 2/3 Iter: 780/1305 Batch time: 1.1319\n",
            "Phase: train Epoch: 2/3 Iter: 781/1305 Batch time: 1.1502\n",
            "Phase: train Epoch: 2/3 Iter: 782/1305 Batch time: 1.1404\n",
            "Phase: train Epoch: 2/3 Iter: 783/1305 Batch time: 1.1274\n",
            "Phase: train Epoch: 2/3 Iter: 784/1305 Batch time: 1.1331\n",
            "Phase: train Epoch: 2/3 Iter: 785/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 2/3 Iter: 786/1305 Batch time: 1.1405\n",
            "Phase: train Epoch: 2/3 Iter: 787/1305 Batch time: 1.2549\n",
            "Phase: train Epoch: 2/3 Iter: 788/1305 Batch time: 1.1249\n",
            "Phase: train Epoch: 2/3 Iter: 789/1305 Batch time: 1.1462\n",
            "Phase: train Epoch: 2/3 Iter: 790/1305 Batch time: 1.1275\n",
            "Phase: train Epoch: 2/3 Iter: 791/1305 Batch time: 1.1301\n",
            "Phase: train Epoch: 2/3 Iter: 792/1305 Batch time: 1.1374\n",
            "Phase: train Epoch: 2/3 Iter: 793/1305 Batch time: 1.1571\n",
            "Phase: train Epoch: 2/3 Iter: 794/1305 Batch time: 1.1633\n",
            "Phase: train Epoch: 2/3 Iter: 795/1305 Batch time: 1.1338\n",
            "Phase: train Epoch: 2/3 Iter: 796/1305 Batch time: 1.1455\n",
            "Phase: train Epoch: 2/3 Iter: 797/1305 Batch time: 1.2568\n",
            "Phase: train Epoch: 2/3 Iter: 798/1305 Batch time: 1.1354\n",
            "Phase: train Epoch: 2/3 Iter: 799/1305 Batch time: 1.1510\n",
            "Phase: train Epoch: 2/3 Iter: 800/1305 Batch time: 1.1497\n",
            "Phase: train Epoch: 2/3 Iter: 801/1305 Batch time: 1.1528\n",
            "Phase: train Epoch: 2/3 Iter: 802/1305 Batch time: 1.1604\n",
            "Phase: train Epoch: 2/3 Iter: 803/1305 Batch time: 1.1674\n",
            "Phase: train Epoch: 2/3 Iter: 804/1305 Batch time: 1.1689\n",
            "Phase: train Epoch: 2/3 Iter: 805/1305 Batch time: 1.1712\n",
            "Phase: train Epoch: 2/3 Iter: 806/1305 Batch time: 1.2763\n",
            "Phase: train Epoch: 2/3 Iter: 807/1305 Batch time: 1.1485\n",
            "Phase: train Epoch: 2/3 Iter: 808/1305 Batch time: 1.1494\n",
            "Phase: train Epoch: 2/3 Iter: 809/1305 Batch time: 1.1429\n",
            "Phase: train Epoch: 2/3 Iter: 810/1305 Batch time: 1.1397\n",
            "Phase: train Epoch: 2/3 Iter: 811/1305 Batch time: 1.1583\n",
            "Phase: train Epoch: 2/3 Iter: 812/1305 Batch time: 1.1542\n",
            "Phase: train Epoch: 2/3 Iter: 813/1305 Batch time: 1.1380\n",
            "Phase: train Epoch: 2/3 Iter: 814/1305 Batch time: 1.1448\n",
            "Phase: train Epoch: 2/3 Iter: 815/1305 Batch time: 1.1482\n",
            "Phase: train Epoch: 2/3 Iter: 816/1305 Batch time: 1.1405\n",
            "Phase: train Epoch: 2/3 Iter: 817/1305 Batch time: 1.1440\n",
            "Phase: train Epoch: 2/3 Iter: 818/1305 Batch time: 1.1565\n",
            "Phase: train Epoch: 2/3 Iter: 819/1305 Batch time: 1.1587\n",
            "Phase: train Epoch: 2/3 Iter: 820/1305 Batch time: 1.1441\n",
            "Phase: train Epoch: 2/3 Iter: 821/1305 Batch time: 1.2592\n",
            "Phase: train Epoch: 2/3 Iter: 822/1305 Batch time: 1.1512\n",
            "Phase: train Epoch: 2/3 Iter: 823/1305 Batch time: 1.1683\n",
            "Phase: train Epoch: 2/3 Iter: 824/1305 Batch time: 1.1418\n",
            "Phase: train Epoch: 2/3 Iter: 825/1305 Batch time: 1.1311\n",
            "Phase: train Epoch: 2/3 Iter: 826/1305 Batch time: 1.1390\n",
            "Phase: train Epoch: 2/3 Iter: 827/1305 Batch time: 1.1483\n",
            "Phase: train Epoch: 2/3 Iter: 828/1305 Batch time: 1.1510\n",
            "Phase: train Epoch: 2/3 Iter: 829/1305 Batch time: 1.1410\n",
            "Phase: train Epoch: 2/3 Iter: 830/1305 Batch time: 1.1664\n",
            "Phase: train Epoch: 2/3 Iter: 831/1305 Batch time: 1.1530\n",
            "Phase: train Epoch: 2/3 Iter: 832/1305 Batch time: 1.1568\n",
            "Phase: train Epoch: 2/3 Iter: 833/1305 Batch time: 1.1588\n",
            "Phase: train Epoch: 2/3 Iter: 834/1305 Batch time: 1.2467\n",
            "Phase: train Epoch: 2/3 Iter: 835/1305 Batch time: 1.1547\n",
            "Phase: train Epoch: 2/3 Iter: 836/1305 Batch time: 1.2326\n",
            "Phase: train Epoch: 2/3 Iter: 837/1305 Batch time: 1.6192\n",
            "Phase: train Epoch: 2/3 Iter: 838/1305 Batch time: 1.2953\n",
            "Phase: train Epoch: 2/3 Iter: 839/1305 Batch time: 1.1465\n",
            "Phase: train Epoch: 2/3 Iter: 840/1305 Batch time: 1.1617\n",
            "Phase: train Epoch: 2/3 Iter: 841/1305 Batch time: 1.1443\n",
            "Phase: train Epoch: 2/3 Iter: 842/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 2/3 Iter: 843/1305 Batch time: 1.1451\n",
            "Phase: train Epoch: 2/3 Iter: 844/1305 Batch time: 1.2606\n",
            "Phase: train Epoch: 2/3 Iter: 845/1305 Batch time: 1.1484\n",
            "Phase: train Epoch: 2/3 Iter: 846/1305 Batch time: 1.1552\n",
            "Phase: train Epoch: 2/3 Iter: 847/1305 Batch time: 1.1574\n",
            "Phase: train Epoch: 2/3 Iter: 848/1305 Batch time: 1.1465\n",
            "Phase: train Epoch: 2/3 Iter: 849/1305 Batch time: 1.1643\n",
            "Phase: train Epoch: 2/3 Iter: 850/1305 Batch time: 1.1578\n",
            "Phase: train Epoch: 2/3 Iter: 851/1305 Batch time: 1.1609\n",
            "Phase: train Epoch: 2/3 Iter: 852/1305 Batch time: 1.1397\n",
            "Phase: train Epoch: 2/3 Iter: 853/1305 Batch time: 1.1526\n",
            "Phase: train Epoch: 2/3 Iter: 854/1305 Batch time: 1.2667\n",
            "Phase: train Epoch: 2/3 Iter: 855/1305 Batch time: 1.1664\n",
            "Phase: train Epoch: 2/3 Iter: 856/1305 Batch time: 1.1554\n",
            "Phase: train Epoch: 2/3 Iter: 857/1305 Batch time: 1.1510\n",
            "Phase: train Epoch: 2/3 Iter: 858/1305 Batch time: 1.1490\n",
            "Phase: train Epoch: 2/3 Iter: 859/1305 Batch time: 1.1498\n",
            "Phase: train Epoch: 2/3 Iter: 860/1305 Batch time: 1.1470\n",
            "Phase: train Epoch: 2/3 Iter: 861/1305 Batch time: 1.1719\n",
            "Phase: train Epoch: 2/3 Iter: 862/1305 Batch time: 1.2737\n",
            "Phase: train Epoch: 2/3 Iter: 863/1305 Batch time: 1.1526\n",
            "Phase: train Epoch: 2/3 Iter: 864/1305 Batch time: 1.1586\n",
            "Phase: train Epoch: 2/3 Iter: 865/1305 Batch time: 1.1473\n",
            "Phase: train Epoch: 2/3 Iter: 866/1305 Batch time: 1.1694\n",
            "Phase: train Epoch: 2/3 Iter: 867/1305 Batch time: 1.1587\n",
            "Phase: train Epoch: 2/3 Iter: 868/1305 Batch time: 1.1655\n",
            "Phase: train Epoch: 2/3 Iter: 869/1305 Batch time: 1.1605\n",
            "Phase: train Epoch: 2/3 Iter: 870/1305 Batch time: 1.1693\n",
            "Phase: train Epoch: 2/3 Iter: 871/1305 Batch time: 1.2752\n",
            "Phase: train Epoch: 2/3 Iter: 872/1305 Batch time: 1.1480\n",
            "Phase: train Epoch: 2/3 Iter: 873/1305 Batch time: 1.1469\n",
            "Phase: train Epoch: 2/3 Iter: 874/1305 Batch time: 1.1619\n",
            "Phase: train Epoch: 2/3 Iter: 875/1305 Batch time: 1.1621\n",
            "Phase: train Epoch: 2/3 Iter: 876/1305 Batch time: 1.1440\n",
            "Phase: train Epoch: 2/3 Iter: 877/1305 Batch time: 1.1637\n",
            "Phase: train Epoch: 2/3 Iter: 878/1305 Batch time: 1.1721\n",
            "Phase: train Epoch: 2/3 Iter: 879/1305 Batch time: 1.1606\n",
            "Phase: train Epoch: 2/3 Iter: 880/1305 Batch time: 1.1607\n",
            "Phase: train Epoch: 2/3 Iter: 881/1305 Batch time: 1.1310\n",
            "Phase: train Epoch: 2/3 Iter: 882/1305 Batch time: 1.1586\n",
            "Phase: train Epoch: 2/3 Iter: 883/1305 Batch time: 1.1683\n",
            "Phase: train Epoch: 2/3 Iter: 884/1305 Batch time: 1.1543\n",
            "Phase: train Epoch: 2/3 Iter: 885/1305 Batch time: 1.1592\n",
            "Phase: train Epoch: 2/3 Iter: 886/1305 Batch time: 1.1608\n",
            "Phase: train Epoch: 2/3 Iter: 887/1305 Batch time: 1.1521\n",
            "Phase: train Epoch: 2/3 Iter: 888/1305 Batch time: 1.1452\n",
            "Phase: train Epoch: 2/3 Iter: 889/1305 Batch time: 1.1441\n",
            "Phase: train Epoch: 2/3 Iter: 890/1305 Batch time: 1.2538\n",
            "Phase: train Epoch: 2/3 Iter: 891/1305 Batch time: 1.1503\n",
            "Phase: train Epoch: 2/3 Iter: 892/1305 Batch time: 1.1591\n",
            "Phase: train Epoch: 2/3 Iter: 893/1305 Batch time: 1.1449\n",
            "Phase: train Epoch: 2/3 Iter: 894/1305 Batch time: 1.1603\n",
            "Phase: train Epoch: 2/3 Iter: 895/1305 Batch time: 1.1689\n",
            "Phase: train Epoch: 2/3 Iter: 896/1305 Batch time: 1.1562\n",
            "Phase: train Epoch: 2/3 Iter: 897/1305 Batch time: 1.1641\n",
            "Phase: train Epoch: 2/3 Iter: 898/1305 Batch time: 1.1355\n",
            "Phase: train Epoch: 2/3 Iter: 899/1305 Batch time: 1.1706\n",
            "Phase: train Epoch: 2/3 Iter: 900/1305 Batch time: 1.1638\n",
            "Phase: train Epoch: 2/3 Iter: 901/1305 Batch time: 1.1485\n",
            "Phase: train Epoch: 2/3 Iter: 902/1305 Batch time: 1.1577\n",
            "Phase: train Epoch: 2/3 Iter: 903/1305 Batch time: 1.1638\n",
            "Phase: train Epoch: 2/3 Iter: 904/1305 Batch time: 1.1500\n",
            "Phase: train Epoch: 2/3 Iter: 905/1305 Batch time: 1.2716\n",
            "Phase: train Epoch: 2/3 Iter: 906/1305 Batch time: 1.1529\n",
            "Phase: train Epoch: 2/3 Iter: 907/1305 Batch time: 1.1711\n",
            "Phase: train Epoch: 2/3 Iter: 908/1305 Batch time: 1.1704\n",
            "Phase: train Epoch: 2/3 Iter: 909/1305 Batch time: 1.1530\n",
            "Phase: train Epoch: 2/3 Iter: 910/1305 Batch time: 1.1634\n",
            "Phase: train Epoch: 2/3 Iter: 911/1305 Batch time: 1.1472\n",
            "Phase: train Epoch: 2/3 Iter: 912/1305 Batch time: 1.1564\n",
            "Phase: train Epoch: 2/3 Iter: 913/1305 Batch time: 1.2872\n",
            "Phase: train Epoch: 2/3 Iter: 914/1305 Batch time: 1.6301\n",
            "Phase: train Epoch: 2/3 Iter: 915/1305 Batch time: 1.2928\n",
            "Phase: train Epoch: 2/3 Iter: 916/1305 Batch time: 1.1550\n",
            "Phase: train Epoch: 2/3 Iter: 917/1305 Batch time: 1.2589\n",
            "Phase: train Epoch: 2/3 Iter: 918/1305 Batch time: 1.1634\n",
            "Phase: train Epoch: 2/3 Iter: 919/1305 Batch time: 1.1595\n",
            "Phase: train Epoch: 2/3 Iter: 920/1305 Batch time: 1.1567\n",
            "Phase: train Epoch: 2/3 Iter: 921/1305 Batch time: 1.1530\n",
            "Phase: train Epoch: 2/3 Iter: 922/1305 Batch time: 1.1353\n",
            "Phase: train Epoch: 2/3 Iter: 923/1305 Batch time: 1.1479\n",
            "Phase: train Epoch: 2/3 Iter: 924/1305 Batch time: 1.1441\n",
            "Phase: train Epoch: 2/3 Iter: 925/1305 Batch time: 1.1652\n",
            "Phase: train Epoch: 2/3 Iter: 926/1305 Batch time: 1.1658\n",
            "Phase: train Epoch: 2/3 Iter: 927/1305 Batch time: 1.1673\n",
            "Phase: train Epoch: 2/3 Iter: 928/1305 Batch time: 1.1464\n",
            "Phase: train Epoch: 2/3 Iter: 929/1305 Batch time: 1.2626\n",
            "Phase: train Epoch: 2/3 Iter: 930/1305 Batch time: 1.1517\n",
            "Phase: train Epoch: 2/3 Iter: 931/1305 Batch time: 1.1571\n",
            "Phase: train Epoch: 2/3 Iter: 932/1305 Batch time: 1.1632\n",
            "Phase: train Epoch: 2/3 Iter: 933/1305 Batch time: 1.1546\n",
            "Phase: train Epoch: 2/3 Iter: 934/1305 Batch time: 1.1506\n",
            "Phase: train Epoch: 2/3 Iter: 935/1305 Batch time: 1.1364\n",
            "Phase: train Epoch: 2/3 Iter: 936/1305 Batch time: 1.1437\n",
            "Phase: train Epoch: 2/3 Iter: 937/1305 Batch time: 1.1481\n",
            "Phase: train Epoch: 2/3 Iter: 938/1305 Batch time: 1.1510\n",
            "Phase: train Epoch: 2/3 Iter: 939/1305 Batch time: 1.2679\n",
            "Phase: train Epoch: 2/3 Iter: 940/1305 Batch time: 1.1487\n",
            "Phase: train Epoch: 2/3 Iter: 941/1305 Batch time: 1.1649\n",
            "Phase: train Epoch: 2/3 Iter: 942/1305 Batch time: 1.1536\n",
            "Phase: train Epoch: 2/3 Iter: 943/1305 Batch time: 1.1494\n",
            "Phase: train Epoch: 2/3 Iter: 944/1305 Batch time: 1.1648\n",
            "Phase: train Epoch: 2/3 Iter: 945/1305 Batch time: 1.1452\n",
            "Phase: train Epoch: 2/3 Iter: 946/1305 Batch time: 1.1761\n",
            "Phase: train Epoch: 2/3 Iter: 947/1305 Batch time: 1.1565\n",
            "Phase: train Epoch: 2/3 Iter: 948/1305 Batch time: 1.2665\n",
            "Phase: train Epoch: 2/3 Iter: 949/1305 Batch time: 1.1700\n",
            "Phase: train Epoch: 2/3 Iter: 950/1305 Batch time: 1.1656\n",
            "Phase: train Epoch: 2/3 Iter: 951/1305 Batch time: 1.1610\n",
            "Phase: train Epoch: 2/3 Iter: 952/1305 Batch time: 1.1408\n",
            "Phase: train Epoch: 2/3 Iter: 953/1305 Batch time: 1.1343\n",
            "Phase: train Epoch: 2/3 Iter: 954/1305 Batch time: 1.1417\n",
            "Phase: train Epoch: 2/3 Iter: 955/1305 Batch time: 1.1541\n",
            "Phase: train Epoch: 2/3 Iter: 956/1305 Batch time: 1.1455\n",
            "Phase: train Epoch: 2/3 Iter: 957/1305 Batch time: 1.1509\n",
            "Phase: train Epoch: 2/3 Iter: 958/1305 Batch time: 1.1559\n",
            "Phase: train Epoch: 2/3 Iter: 959/1305 Batch time: 1.1393\n",
            "Phase: train Epoch: 2/3 Iter: 960/1305 Batch time: 1.1503\n",
            "Phase: train Epoch: 2/3 Iter: 961/1305 Batch time: 1.1373\n",
            "Phase: train Epoch: 2/3 Iter: 962/1305 Batch time: 1.1527\n",
            "Phase: train Epoch: 2/3 Iter: 963/1305 Batch time: 1.2390\n",
            "Phase: train Epoch: 2/3 Iter: 964/1305 Batch time: 1.1467\n",
            "Phase: train Epoch: 2/3 Iter: 965/1305 Batch time: 1.1375\n",
            "Phase: train Epoch: 2/3 Iter: 966/1305 Batch time: 1.1688\n",
            "Phase: train Epoch: 2/3 Iter: 967/1305 Batch time: 1.1465\n",
            "Phase: train Epoch: 2/3 Iter: 968/1305 Batch time: 1.1568\n",
            "Phase: train Epoch: 2/3 Iter: 969/1305 Batch time: 1.1505\n",
            "Phase: train Epoch: 2/3 Iter: 970/1305 Batch time: 1.1749\n",
            "Phase: train Epoch: 2/3 Iter: 971/1305 Batch time: 1.1535\n",
            "Phase: train Epoch: 2/3 Iter: 972/1305 Batch time: 1.1437\n",
            "Phase: train Epoch: 2/3 Iter: 973/1305 Batch time: 1.1637\n",
            "Phase: train Epoch: 2/3 Iter: 974/1305 Batch time: 1.1429\n",
            "Phase: train Epoch: 2/3 Iter: 975/1305 Batch time: 1.1650\n",
            "Phase: train Epoch: 2/3 Iter: 976/1305 Batch time: 1.2524\n",
            "Phase: train Epoch: 2/3 Iter: 977/1305 Batch time: 1.1483\n",
            "Phase: train Epoch: 2/3 Iter: 978/1305 Batch time: 1.1600\n",
            "Phase: train Epoch: 2/3 Iter: 979/1305 Batch time: 1.1487\n",
            "Phase: train Epoch: 2/3 Iter: 980/1305 Batch time: 1.1420\n",
            "Phase: train Epoch: 2/3 Iter: 981/1305 Batch time: 1.1709\n",
            "Phase: train Epoch: 2/3 Iter: 982/1305 Batch time: 1.1607\n",
            "Phase: train Epoch: 2/3 Iter: 983/1305 Batch time: 1.1440\n",
            "Phase: train Epoch: 2/3 Iter: 984/1305 Batch time: 1.1439\n",
            "Phase: train Epoch: 2/3 Iter: 985/1305 Batch time: 1.1578\n",
            "Phase: train Epoch: 2/3 Iter: 986/1305 Batch time: 1.2706\n",
            "Phase: train Epoch: 2/3 Iter: 987/1305 Batch time: 1.1470\n",
            "Phase: train Epoch: 2/3 Iter: 988/1305 Batch time: 1.1449\n",
            "Phase: train Epoch: 2/3 Iter: 989/1305 Batch time: 1.1482\n",
            "Phase: train Epoch: 2/3 Iter: 990/1305 Batch time: 1.1393\n",
            "Phase: train Epoch: 2/3 Iter: 991/1305 Batch time: 1.3051\n",
            "Phase: train Epoch: 2/3 Iter: 992/1305 Batch time: 1.6002\n",
            "Phase: train Epoch: 2/3 Iter: 993/1305 Batch time: 1.2606\n",
            "Phase: train Epoch: 2/3 Iter: 994/1305 Batch time: 1.1631\n",
            "Phase: train Epoch: 2/3 Iter: 995/1305 Batch time: 1.1533\n",
            "Phase: train Epoch: 2/3 Iter: 996/1305 Batch time: 1.2713\n",
            "Phase: train Epoch: 2/3 Iter: 997/1305 Batch time: 1.1499\n",
            "Phase: train Epoch: 2/3 Iter: 998/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 2/3 Iter: 999/1305 Batch time: 1.1483\n",
            "Phase: train Epoch: 2/3 Iter: 1000/1305 Batch time: 1.1608\n",
            "Phase: train Epoch: 2/3 Iter: 1001/1305 Batch time: 1.1435\n",
            "Phase: train Epoch: 2/3 Iter: 1002/1305 Batch time: 1.1545\n",
            "Phase: train Epoch: 2/3 Iter: 1003/1305 Batch time: 1.1518\n",
            "Phase: train Epoch: 2/3 Iter: 1004/1305 Batch time: 1.2697\n",
            "Phase: train Epoch: 2/3 Iter: 1005/1305 Batch time: 1.1497\n",
            "Phase: train Epoch: 2/3 Iter: 1006/1305 Batch time: 1.1611\n",
            "Phase: train Epoch: 2/3 Iter: 1007/1305 Batch time: 1.1571\n",
            "Phase: train Epoch: 2/3 Iter: 1008/1305 Batch time: 1.1436\n",
            "Phase: train Epoch: 2/3 Iter: 1009/1305 Batch time: 1.1365\n",
            "Phase: train Epoch: 2/3 Iter: 1010/1305 Batch time: 1.1498\n",
            "Phase: train Epoch: 2/3 Iter: 1011/1305 Batch time: 1.1599\n",
            "Phase: train Epoch: 2/3 Iter: 1012/1305 Batch time: 1.1344\n",
            "Phase: train Epoch: 2/3 Iter: 1013/1305 Batch time: 1.2607\n",
            "Phase: train Epoch: 2/3 Iter: 1014/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 2/3 Iter: 1015/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 2/3 Iter: 1016/1305 Batch time: 1.0875\n",
            "Phase: train Epoch: 2/3 Iter: 1017/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 2/3 Iter: 1018/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 2/3 Iter: 1019/1305 Batch time: 1.1340\n",
            "Phase: train Epoch: 2/3 Iter: 1020/1305 Batch time: 1.1656\n",
            "Phase: train Epoch: 2/3 Iter: 1021/1305 Batch time: 1.2123\n",
            "Phase: train Epoch: 2/3 Iter: 1022/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 2/3 Iter: 1023/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 2/3 Iter: 1024/1305 Batch time: 1.1264\n",
            "Phase: train Epoch: 2/3 Iter: 1025/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 2/3 Iter: 1026/1305 Batch time: 1.1204\n",
            "Phase: train Epoch: 2/3 Iter: 1027/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 2/3 Iter: 1028/1305 Batch time: 1.1177\n",
            "Phase: train Epoch: 2/3 Iter: 1029/1305 Batch time: 1.1234\n",
            "Phase: train Epoch: 2/3 Iter: 1030/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 2/3 Iter: 1031/1305 Batch time: 1.1308\n",
            "Phase: train Epoch: 2/3 Iter: 1032/1305 Batch time: 1.2201\n",
            "Phase: train Epoch: 2/3 Iter: 1033/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 2/3 Iter: 1034/1305 Batch time: 1.1101\n",
            "Phase: train Epoch: 2/3 Iter: 1035/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 2/3 Iter: 1036/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 2/3 Iter: 1037/1305 Batch time: 1.1161\n",
            "Phase: train Epoch: 2/3 Iter: 1038/1305 Batch time: 1.1359\n",
            "Phase: train Epoch: 2/3 Iter: 1039/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 2/3 Iter: 1040/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 2/3 Iter: 1041/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 2/3 Iter: 1042/1305 Batch time: 1.1216\n",
            "Phase: train Epoch: 2/3 Iter: 1043/1305 Batch time: 1.1426\n",
            "Phase: train Epoch: 2/3 Iter: 1044/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 2/3 Iter: 1045/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 2/3 Iter: 1046/1305 Batch time: 1.1202\n",
            "Phase: train Epoch: 2/3 Iter: 1047/1305 Batch time: 1.2162\n",
            "Phase: train Epoch: 2/3 Iter: 1048/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 2/3 Iter: 1049/1305 Batch time: 1.1230\n",
            "Phase: train Epoch: 2/3 Iter: 1050/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 2/3 Iter: 1051/1305 Batch time: 1.1227\n",
            "Phase: train Epoch: 2/3 Iter: 1052/1305 Batch time: 1.1228\n",
            "Phase: train Epoch: 2/3 Iter: 1053/1305 Batch time: 1.0982\n",
            "Phase: train Epoch: 2/3 Iter: 1054/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 2/3 Iter: 1055/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 2/3 Iter: 1056/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 2/3 Iter: 1057/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 2/3 Iter: 1058/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 2/3 Iter: 1059/1305 Batch time: 1.2205\n",
            "Phase: train Epoch: 2/3 Iter: 1060/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 2/3 Iter: 1061/1305 Batch time: 1.1266\n",
            "Phase: train Epoch: 2/3 Iter: 1062/1305 Batch time: 1.1301\n",
            "Phase: train Epoch: 2/3 Iter: 1063/1305 Batch time: 1.1207\n",
            "Phase: train Epoch: 2/3 Iter: 1064/1305 Batch time: 1.1303\n",
            "Phase: train Epoch: 2/3 Iter: 1065/1305 Batch time: 1.1390\n",
            "Phase: train Epoch: 2/3 Iter: 1066/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 2/3 Iter: 1067/1305 Batch time: 1.1297\n",
            "Phase: train Epoch: 2/3 Iter: 1068/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 1069/1305 Batch time: 1.1201\n",
            "Phase: train Epoch: 2/3 Iter: 1070/1305 Batch time: 1.1346\n",
            "Phase: train Epoch: 2/3 Iter: 1071/1305 Batch time: 1.7128\n",
            "Phase: train Epoch: 2/3 Iter: 1072/1305 Batch time: 1.3538\n",
            "Phase: train Epoch: 2/3 Iter: 1073/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 2/3 Iter: 1074/1305 Batch time: 1.1333\n",
            "Phase: train Epoch: 2/3 Iter: 1075/1305 Batch time: 1.1464\n",
            "Phase: train Epoch: 2/3 Iter: 1076/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 2/3 Iter: 1077/1305 Batch time: 1.1429\n",
            "Phase: train Epoch: 2/3 Iter: 1078/1305 Batch time: 1.1328\n",
            "Phase: train Epoch: 2/3 Iter: 1079/1305 Batch time: 1.1333\n",
            "Phase: train Epoch: 2/3 Iter: 1080/1305 Batch time: 1.1327\n",
            "Phase: train Epoch: 2/3 Iter: 1081/1305 Batch time: 1.2424\n",
            "Phase: train Epoch: 2/3 Iter: 1082/1305 Batch time: 1.1299\n",
            "Phase: train Epoch: 2/3 Iter: 1083/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 2/3 Iter: 1084/1305 Batch time: 1.1318\n",
            "Phase: train Epoch: 2/3 Iter: 1085/1305 Batch time: 1.1429\n",
            "Phase: train Epoch: 2/3 Iter: 1086/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 2/3 Iter: 1087/1305 Batch time: 1.1334\n",
            "Phase: train Epoch: 2/3 Iter: 1088/1305 Batch time: 1.1228\n",
            "Phase: train Epoch: 2/3 Iter: 1089/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 2/3 Iter: 1090/1305 Batch time: 1.2499\n",
            "Phase: train Epoch: 2/3 Iter: 1091/1305 Batch time: 1.1296\n",
            "Phase: train Epoch: 2/3 Iter: 1092/1305 Batch time: 1.1262\n",
            "Phase: train Epoch: 2/3 Iter: 1093/1305 Batch time: 1.1228\n",
            "Phase: train Epoch: 2/3 Iter: 1094/1305 Batch time: 1.1182\n",
            "Phase: train Epoch: 2/3 Iter: 1095/1305 Batch time: 1.1298\n",
            "Phase: train Epoch: 2/3 Iter: 1096/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 2/3 Iter: 1097/1305 Batch time: 1.1332\n",
            "Phase: train Epoch: 2/3 Iter: 1098/1305 Batch time: 1.1411\n",
            "Phase: train Epoch: 2/3 Iter: 1099/1305 Batch time: 1.1286\n",
            "Phase: train Epoch: 2/3 Iter: 1100/1305 Batch time: 1.1275\n",
            "Phase: train Epoch: 2/3 Iter: 1101/1305 Batch time: 1.1281\n",
            "Phase: train Epoch: 2/3 Iter: 1102/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 2/3 Iter: 1103/1305 Batch time: 1.1264\n",
            "Phase: train Epoch: 2/3 Iter: 1104/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 1105/1305 Batch time: 1.2336\n",
            "Phase: train Epoch: 2/3 Iter: 1106/1305 Batch time: 1.1236\n",
            "Phase: train Epoch: 2/3 Iter: 1107/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 2/3 Iter: 1108/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 2/3 Iter: 1109/1305 Batch time: 1.1277\n",
            "Phase: train Epoch: 2/3 Iter: 1110/1305 Batch time: 1.1185\n",
            "Phase: train Epoch: 2/3 Iter: 1111/1305 Batch time: 1.1556\n",
            "Phase: train Epoch: 2/3 Iter: 1112/1305 Batch time: 1.1374\n",
            "Phase: train Epoch: 2/3 Iter: 1113/1305 Batch time: 1.1240\n",
            "Phase: train Epoch: 2/3 Iter: 1114/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 2/3 Iter: 1115/1305 Batch time: 1.1404\n",
            "Phase: train Epoch: 2/3 Iter: 1116/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 2/3 Iter: 1117/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 2/3 Iter: 1118/1305 Batch time: 1.2151\n",
            "Phase: train Epoch: 2/3 Iter: 1119/1305 Batch time: 1.1358\n",
            "Phase: train Epoch: 2/3 Iter: 1120/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 2/3 Iter: 1121/1305 Batch time: 1.1261\n",
            "Phase: train Epoch: 2/3 Iter: 1122/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 2/3 Iter: 1123/1305 Batch time: 1.1321\n",
            "Phase: train Epoch: 2/3 Iter: 1124/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 2/3 Iter: 1125/1305 Batch time: 1.1110\n",
            "Phase: train Epoch: 2/3 Iter: 1126/1305 Batch time: 1.1319\n",
            "Phase: train Epoch: 2/3 Iter: 1127/1305 Batch time: 1.0942\n",
            "Phase: train Epoch: 2/3 Iter: 1128/1305 Batch time: 1.2301\n",
            "Phase: train Epoch: 2/3 Iter: 1129/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 2/3 Iter: 1130/1305 Batch time: 1.1077\n",
            "Phase: train Epoch: 2/3 Iter: 1131/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 2/3 Iter: 1132/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 2/3 Iter: 1133/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 2/3 Iter: 1134/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 2/3 Iter: 1135/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 2/3 Iter: 1136/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 2/3 Iter: 1137/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 2/3 Iter: 1138/1305 Batch time: 1.2193\n",
            "Phase: train Epoch: 2/3 Iter: 1139/1305 Batch time: 1.1013\n",
            "Phase: train Epoch: 2/3 Iter: 1140/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 2/3 Iter: 1141/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 2/3 Iter: 1142/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 1143/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 2/3 Iter: 1144/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 2/3 Iter: 1145/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 2/3 Iter: 1146/1305 Batch time: 1.2318\n",
            "Phase: train Epoch: 2/3 Iter: 1147/1305 Batch time: 1.1301\n",
            "Phase: train Epoch: 2/3 Iter: 1148/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 2/3 Iter: 1149/1305 Batch time: 1.1243\n",
            "Phase: train Epoch: 2/3 Iter: 1150/1305 Batch time: 1.4193\n",
            "Phase: train Epoch: 2/3 Iter: 1151/1305 Batch time: 1.5022\n",
            "Phase: train Epoch: 2/3 Iter: 1152/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 2/3 Iter: 1153/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 2/3 Iter: 1154/1305 Batch time: 1.1369\n",
            "Phase: train Epoch: 2/3 Iter: 1155/1305 Batch time: 1.2086\n",
            "Phase: train Epoch: 2/3 Iter: 1156/1305 Batch time: 1.1269\n",
            "Phase: train Epoch: 2/3 Iter: 1157/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 2/3 Iter: 1158/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 2/3 Iter: 1159/1305 Batch time: 1.1219\n",
            "Phase: train Epoch: 2/3 Iter: 1160/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 2/3 Iter: 1161/1305 Batch time: 1.1309\n",
            "Phase: train Epoch: 2/3 Iter: 1162/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 2/3 Iter: 1163/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 2/3 Iter: 1164/1305 Batch time: 1.1179\n",
            "Phase: train Epoch: 2/3 Iter: 1165/1305 Batch time: 1.1311\n",
            "Phase: train Epoch: 2/3 Iter: 1166/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 2/3 Iter: 1167/1305 Batch time: 1.1134\n",
            "Phase: train Epoch: 2/3 Iter: 1168/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 2/3 Iter: 1169/1305 Batch time: 1.0932\n",
            "Phase: train Epoch: 2/3 Iter: 1170/1305 Batch time: 1.1182\n",
            "Phase: train Epoch: 2/3 Iter: 1171/1305 Batch time: 1.1221\n",
            "Phase: train Epoch: 2/3 Iter: 1172/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 2/3 Iter: 1173/1305 Batch time: 1.1248\n",
            "Phase: train Epoch: 2/3 Iter: 1174/1305 Batch time: 1.2274\n",
            "Phase: train Epoch: 2/3 Iter: 1175/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 2/3 Iter: 1176/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 2/3 Iter: 1177/1305 Batch time: 1.1461\n",
            "Phase: train Epoch: 2/3 Iter: 1178/1305 Batch time: 1.1312\n",
            "Phase: train Epoch: 2/3 Iter: 1179/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 1180/1305 Batch time: 1.1259\n",
            "Phase: train Epoch: 2/3 Iter: 1181/1305 Batch time: 1.1188\n",
            "Phase: train Epoch: 2/3 Iter: 1182/1305 Batch time: 1.1186\n",
            "Phase: train Epoch: 2/3 Iter: 1183/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 2/3 Iter: 1184/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 2/3 Iter: 1185/1305 Batch time: 1.1419\n",
            "Phase: train Epoch: 2/3 Iter: 1186/1305 Batch time: 1.1416\n",
            "Phase: train Epoch: 2/3 Iter: 1187/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 2/3 Iter: 1188/1305 Batch time: 1.1218\n",
            "Phase: train Epoch: 2/3 Iter: 1189/1305 Batch time: 1.2424\n",
            "Phase: train Epoch: 2/3 Iter: 1190/1305 Batch time: 1.1217\n",
            "Phase: train Epoch: 2/3 Iter: 1191/1305 Batch time: 1.1293\n",
            "Phase: train Epoch: 2/3 Iter: 1192/1305 Batch time: 1.1307\n",
            "Phase: train Epoch: 2/3 Iter: 1193/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 2/3 Iter: 1194/1305 Batch time: 1.1265\n",
            "Phase: train Epoch: 2/3 Iter: 1195/1305 Batch time: 1.1197\n",
            "Phase: train Epoch: 2/3 Iter: 1196/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 2/3 Iter: 1197/1305 Batch time: 1.1271\n",
            "Phase: train Epoch: 2/3 Iter: 1198/1305 Batch time: 1.1260\n",
            "Phase: train Epoch: 2/3 Iter: 1199/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 2/3 Iter: 1200/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 2/3 Iter: 1201/1305 Batch time: 1.2143\n",
            "Phase: train Epoch: 2/3 Iter: 1202/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 2/3 Iter: 1203/1305 Batch time: 1.1218\n",
            "Phase: train Epoch: 2/3 Iter: 1204/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 2/3 Iter: 1205/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 2/3 Iter: 1206/1305 Batch time: 1.1308\n",
            "Phase: train Epoch: 2/3 Iter: 1207/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 2/3 Iter: 1208/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 2/3 Iter: 1209/1305 Batch time: 1.1143\n",
            "Phase: train Epoch: 2/3 Iter: 1210/1305 Batch time: 1.1143\n",
            "Phase: train Epoch: 2/3 Iter: 1211/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 2/3 Iter: 1212/1305 Batch time: 1.1267\n",
            "Phase: train Epoch: 2/3 Iter: 1213/1305 Batch time: 1.2110\n",
            "Phase: train Epoch: 2/3 Iter: 1214/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 2/3 Iter: 1215/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 2/3 Iter: 1216/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 2/3 Iter: 1217/1305 Batch time: 1.1257\n",
            "Phase: train Epoch: 2/3 Iter: 1218/1305 Batch time: 1.1397\n",
            "Phase: train Epoch: 2/3 Iter: 1219/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 2/3 Iter: 1220/1305 Batch time: 1.1325\n",
            "Phase: train Epoch: 2/3 Iter: 1221/1305 Batch time: 1.1148\n",
            "Phase: train Epoch: 2/3 Iter: 1222/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 2/3 Iter: 1223/1305 Batch time: 1.2605\n",
            "Phase: train Epoch: 2/3 Iter: 1224/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 2/3 Iter: 1225/1305 Batch time: 1.1239\n",
            "Phase: train Epoch: 2/3 Iter: 1226/1305 Batch time: 1.1223\n",
            "Phase: train Epoch: 2/3 Iter: 1227/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 2/3 Iter: 1228/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 1229/1305 Batch time: 1.1240\n",
            "Phase: train Epoch: 2/3 Iter: 1230/1305 Batch time: 1.2381\n",
            "Phase: train Epoch: 2/3 Iter: 1231/1305 Batch time: 1.5555\n",
            "Phase: train Epoch: 2/3 Iter: 1232/1305 Batch time: 1.2953\n",
            "Phase: train Epoch: 2/3 Iter: 1233/1305 Batch time: 1.1185\n",
            "Phase: train Epoch: 2/3 Iter: 1234/1305 Batch time: 1.0970\n",
            "Phase: train Epoch: 2/3 Iter: 1235/1305 Batch time: 1.1234\n",
            "Phase: train Epoch: 2/3 Iter: 1236/1305 Batch time: 1.0978\n",
            "Phase: train Epoch: 2/3 Iter: 1237/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 2/3 Iter: 1238/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 2/3 Iter: 1239/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 2/3 Iter: 1240/1305 Batch time: 1.1215\n",
            "Phase: train Epoch: 2/3 Iter: 1241/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 2/3 Iter: 1242/1305 Batch time: 1.1009\n",
            "Phase: train Epoch: 2/3 Iter: 1243/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 2/3 Iter: 1244/1305 Batch time: 1.1293\n",
            "Phase: train Epoch: 2/3 Iter: 1245/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 1246/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 2/3 Iter: 1247/1305 Batch time: 1.2255\n",
            "Phase: train Epoch: 2/3 Iter: 1248/1305 Batch time: 1.1247\n",
            "Phase: train Epoch: 2/3 Iter: 1249/1305 Batch time: 1.1182\n",
            "Phase: train Epoch: 2/3 Iter: 1250/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 2/3 Iter: 1251/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 2/3 Iter: 1252/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 2/3 Iter: 1253/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 2/3 Iter: 1254/1305 Batch time: 1.1153\n",
            "Phase: train Epoch: 2/3 Iter: 1255/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 2/3 Iter: 1256/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 2/3 Iter: 1257/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 2/3 Iter: 1258/1305 Batch time: 1.1370\n",
            "Phase: train Epoch: 2/3 Iter: 1259/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 2/3 Iter: 1260/1305 Batch time: 1.2363\n",
            "Phase: train Epoch: 2/3 Iter: 1261/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 2/3 Iter: 1262/1305 Batch time: 1.1264\n",
            "Phase: train Epoch: 2/3 Iter: 1263/1305 Batch time: 1.1263\n",
            "Phase: train Epoch: 2/3 Iter: 1264/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 2/3 Iter: 1265/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 2/3 Iter: 1266/1305 Batch time: 1.1356\n",
            "Phase: train Epoch: 2/3 Iter: 1267/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 2/3 Iter: 1268/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 2/3 Iter: 1269/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 2/3 Iter: 1270/1305 Batch time: 1.2344\n",
            "Phase: train Epoch: 2/3 Iter: 1271/1305 Batch time: 1.1262\n",
            "Phase: train Epoch: 2/3 Iter: 1272/1305 Batch time: 1.1201\n",
            "Phase: train Epoch: 2/3 Iter: 1273/1305 Batch time: 1.1231\n",
            "Phase: train Epoch: 2/3 Iter: 1274/1305 Batch time: 1.1179\n",
            "Phase: train Epoch: 2/3 Iter: 1275/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 2/3 Iter: 1276/1305 Batch time: 1.1206\n",
            "Phase: train Epoch: 2/3 Iter: 1277/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 2/3 Iter: 1278/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 2/3 Iter: 1279/1305 Batch time: 1.1149\n",
            "Phase: train Epoch: 2/3 Iter: 1280/1305 Batch time: 1.2299\n",
            "Phase: train Epoch: 2/3 Iter: 1281/1305 Batch time: 1.1224\n",
            "Phase: train Epoch: 2/3 Iter: 1282/1305 Batch time: 1.1123\n",
            "Phase: train Epoch: 2/3 Iter: 1283/1305 Batch time: 1.1402\n",
            "Phase: train Epoch: 2/3 Iter: 1284/1305 Batch time: 1.1223\n",
            "Phase: train Epoch: 2/3 Iter: 1285/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 2/3 Iter: 1286/1305 Batch time: 1.1224\n",
            "Phase: train Epoch: 2/3 Iter: 1287/1305 Batch time: 1.1317\n",
            "Phase: train Epoch: 2/3 Iter: 1288/1305 Batch time: 1.2141\n",
            "Phase: train Epoch: 2/3 Iter: 1289/1305 Batch time: 1.1107\n",
            "Phase: train Epoch: 2/3 Iter: 1290/1305 Batch time: 1.1306\n",
            "Phase: train Epoch: 2/3 Iter: 1291/1305 Batch time: 1.1171\n",
            "Phase: train Epoch: 2/3 Iter: 1292/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 2/3 Iter: 1293/1305 Batch time: 1.1143\n",
            "Phase: train Epoch: 2/3 Iter: 1294/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 2/3 Iter: 1295/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 2/3 Iter: 1296/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 2/3 Iter: 1297/1305 Batch time: 1.2217\n",
            "Phase: train Epoch: 2/3 Iter: 1298/1305 Batch time: 1.1485\n",
            "Phase: train Epoch: 2/3 Iter: 1299/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 2/3 Iter: 1300/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 2/3 Iter: 1301/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 2/3 Iter: 1302/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 2/3 Iter: 1303/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 2/3 Iter: 1304/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 2/3 Loss: 0.3323 Acc: 0.8731        \n",
            "Phase: val Epoch: 2/3 Iter: 1/5 Batch time: 0.7249\n",
            "Phase: val Epoch: 2/3 Iter: 2/5 Batch time: 0.7234\n",
            "Phase: val Epoch: 2/3 Iter: 3/5 Batch time: 0.7191\n",
            "Phase: val Epoch: 2/3 Iter: 4/5 Batch time: 0.7155\n",
            "Phase: val Epoch: 2/3 Loss: 1.0902 Acc: 0.5625        \n",
            "Phase: train Epoch: 3/3 Iter: 1/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 3/3 Iter: 2/1305 Batch time: 1.1377\n",
            "Phase: train Epoch: 3/3 Iter: 3/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 4/1305 Batch time: 1.2010\n",
            "Phase: train Epoch: 3/3 Iter: 5/1305 Batch time: 1.5328\n",
            "Phase: train Epoch: 3/3 Iter: 6/1305 Batch time: 1.2757\n",
            "Phase: train Epoch: 3/3 Iter: 7/1305 Batch time: 1.2294\n",
            "Phase: train Epoch: 3/3 Iter: 8/1305 Batch time: 1.1250\n",
            "Phase: train Epoch: 3/3 Iter: 9/1305 Batch time: 1.1327\n",
            "Phase: train Epoch: 3/3 Iter: 10/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 3/3 Iter: 11/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 3/3 Iter: 12/1305 Batch time: 1.1160\n",
            "Phase: train Epoch: 3/3 Iter: 13/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 3/3 Iter: 14/1305 Batch time: 1.1299\n",
            "Phase: train Epoch: 3/3 Iter: 15/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 3/3 Iter: 16/1305 Batch time: 1.1494\n",
            "Phase: train Epoch: 3/3 Iter: 17/1305 Batch time: 1.2340\n",
            "Phase: train Epoch: 3/3 Iter: 18/1305 Batch time: 1.1472\n",
            "Phase: train Epoch: 3/3 Iter: 19/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 3/3 Iter: 20/1305 Batch time: 1.1277\n",
            "Phase: train Epoch: 3/3 Iter: 21/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 3/3 Iter: 22/1305 Batch time: 1.1232\n",
            "Phase: train Epoch: 3/3 Iter: 23/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 3/3 Iter: 24/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 3/3 Iter: 25/1305 Batch time: 1.1541\n",
            "Phase: train Epoch: 3/3 Iter: 26/1305 Batch time: 1.1438\n",
            "Phase: train Epoch: 3/3 Iter: 27/1305 Batch time: 1.2426\n",
            "Phase: train Epoch: 3/3 Iter: 28/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 3/3 Iter: 29/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 3/3 Iter: 30/1305 Batch time: 1.1539\n",
            "Phase: train Epoch: 3/3 Iter: 31/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 3/3 Iter: 32/1305 Batch time: 1.1469\n",
            "Phase: train Epoch: 3/3 Iter: 33/1305 Batch time: 1.1311\n",
            "Phase: train Epoch: 3/3 Iter: 34/1305 Batch time: 1.1239\n",
            "Phase: train Epoch: 3/3 Iter: 35/1305 Batch time: 1.2461\n",
            "Phase: train Epoch: 3/3 Iter: 36/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 3/3 Iter: 37/1305 Batch time: 1.1187\n",
            "Phase: train Epoch: 3/3 Iter: 38/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 3/3 Iter: 39/1305 Batch time: 1.1303\n",
            "Phase: train Epoch: 3/3 Iter: 40/1305 Batch time: 1.1337\n",
            "Phase: train Epoch: 3/3 Iter: 41/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 3/3 Iter: 42/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 3/3 Iter: 43/1305 Batch time: 1.1248\n",
            "Phase: train Epoch: 3/3 Iter: 44/1305 Batch time: 1.2305\n",
            "Phase: train Epoch: 3/3 Iter: 45/1305 Batch time: 1.1238\n",
            "Phase: train Epoch: 3/3 Iter: 46/1305 Batch time: 1.1365\n",
            "Phase: train Epoch: 3/3 Iter: 47/1305 Batch time: 1.1252\n",
            "Phase: train Epoch: 3/3 Iter: 48/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 3/3 Iter: 49/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 3/3 Iter: 50/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 51/1305 Batch time: 1.1343\n",
            "Phase: train Epoch: 3/3 Iter: 52/1305 Batch time: 1.1238\n",
            "Phase: train Epoch: 3/3 Iter: 53/1305 Batch time: 1.1188\n",
            "Phase: train Epoch: 3/3 Iter: 54/1305 Batch time: 1.1170\n",
            "Phase: train Epoch: 3/3 Iter: 55/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 3/3 Iter: 56/1305 Batch time: 1.1069\n",
            "Phase: train Epoch: 3/3 Iter: 57/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 3/3 Iter: 58/1305 Batch time: 1.1291\n",
            "Phase: train Epoch: 3/3 Iter: 59/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 3/3 Iter: 60/1305 Batch time: 1.1177\n",
            "Phase: train Epoch: 3/3 Iter: 61/1305 Batch time: 1.1249\n",
            "Phase: train Epoch: 3/3 Iter: 62/1305 Batch time: 1.1337\n",
            "Phase: train Epoch: 3/3 Iter: 63/1305 Batch time: 1.2090\n",
            "Phase: train Epoch: 3/3 Iter: 64/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 3/3 Iter: 65/1305 Batch time: 1.1255\n",
            "Phase: train Epoch: 3/3 Iter: 66/1305 Batch time: 1.1186\n",
            "Phase: train Epoch: 3/3 Iter: 67/1305 Batch time: 1.1339\n",
            "Phase: train Epoch: 3/3 Iter: 68/1305 Batch time: 1.1393\n",
            "Phase: train Epoch: 3/3 Iter: 69/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 3/3 Iter: 70/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 3/3 Iter: 71/1305 Batch time: 1.1242\n",
            "Phase: train Epoch: 3/3 Iter: 72/1305 Batch time: 1.1063\n",
            "Phase: train Epoch: 3/3 Iter: 73/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 3/3 Iter: 74/1305 Batch time: 1.1271\n",
            "Phase: train Epoch: 3/3 Iter: 75/1305 Batch time: 1.1250\n",
            "Phase: train Epoch: 3/3 Iter: 76/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 3/3 Iter: 77/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 3/3 Iter: 78/1305 Batch time: 1.2115\n",
            "Phase: train Epoch: 3/3 Iter: 79/1305 Batch time: 1.1356\n",
            "Phase: train Epoch: 3/3 Iter: 80/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 3/3 Iter: 81/1305 Batch time: 1.1192\n",
            "Phase: train Epoch: 3/3 Iter: 82/1305 Batch time: 1.1356\n",
            "Phase: train Epoch: 3/3 Iter: 83/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 3/3 Iter: 84/1305 Batch time: 1.3505\n",
            "Phase: train Epoch: 3/3 Iter: 85/1305 Batch time: 1.5645\n",
            "Phase: train Epoch: 3/3 Iter: 86/1305 Batch time: 1.1230\n",
            "Phase: train Epoch: 3/3 Iter: 87/1305 Batch time: 1.1072\n",
            "Phase: train Epoch: 3/3 Iter: 88/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 3/3 Iter: 89/1305 Batch time: 1.1225\n",
            "Phase: train Epoch: 3/3 Iter: 90/1305 Batch time: 1.2355\n",
            "Phase: train Epoch: 3/3 Iter: 91/1305 Batch time: 1.1253\n",
            "Phase: train Epoch: 3/3 Iter: 92/1305 Batch time: 1.1241\n",
            "Phase: train Epoch: 3/3 Iter: 93/1305 Batch time: 1.1314\n",
            "Phase: train Epoch: 3/3 Iter: 94/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 3/3 Iter: 95/1305 Batch time: 1.1334\n",
            "Phase: train Epoch: 3/3 Iter: 96/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 3/3 Iter: 97/1305 Batch time: 1.1260\n",
            "Phase: train Epoch: 3/3 Iter: 98/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 3/3 Iter: 99/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 3/3 Iter: 100/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 3/3 Iter: 101/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 3/3 Iter: 102/1305 Batch time: 1.2251\n",
            "Phase: train Epoch: 3/3 Iter: 103/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 3/3 Iter: 104/1305 Batch time: 1.1069\n",
            "Phase: train Epoch: 3/3 Iter: 105/1305 Batch time: 1.1269\n",
            "Phase: train Epoch: 3/3 Iter: 106/1305 Batch time: 1.1204\n",
            "Phase: train Epoch: 3/3 Iter: 107/1305 Batch time: 1.1207\n",
            "Phase: train Epoch: 3/3 Iter: 108/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 3/3 Iter: 109/1305 Batch time: 1.1239\n",
            "Phase: train Epoch: 3/3 Iter: 110/1305 Batch time: 1.1205\n",
            "Phase: train Epoch: 3/3 Iter: 111/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 3/3 Iter: 112/1305 Batch time: 1.2127\n",
            "Phase: train Epoch: 3/3 Iter: 113/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 3/3 Iter: 114/1305 Batch time: 1.1345\n",
            "Phase: train Epoch: 3/3 Iter: 115/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 3/3 Iter: 116/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 3/3 Iter: 117/1305 Batch time: 1.1182\n",
            "Phase: train Epoch: 3/3 Iter: 118/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 3/3 Iter: 119/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 120/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 3/3 Iter: 121/1305 Batch time: 1.2310\n",
            "Phase: train Epoch: 3/3 Iter: 122/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 3/3 Iter: 123/1305 Batch time: 1.1332\n",
            "Phase: train Epoch: 3/3 Iter: 124/1305 Batch time: 1.1219\n",
            "Phase: train Epoch: 3/3 Iter: 125/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 3/3 Iter: 126/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 3/3 Iter: 127/1305 Batch time: 1.1132\n",
            "Phase: train Epoch: 3/3 Iter: 128/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 3/3 Iter: 129/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 3/3 Iter: 130/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 3/3 Iter: 131/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 3/3 Iter: 132/1305 Batch time: 1.1166\n",
            "Phase: train Epoch: 3/3 Iter: 133/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 3/3 Iter: 134/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 3/3 Iter: 135/1305 Batch time: 1.1181\n",
            "Phase: train Epoch: 3/3 Iter: 136/1305 Batch time: 1.2125\n",
            "Phase: train Epoch: 3/3 Iter: 137/1305 Batch time: 1.1350\n",
            "Phase: train Epoch: 3/3 Iter: 138/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 139/1305 Batch time: 1.1185\n",
            "Phase: train Epoch: 3/3 Iter: 140/1305 Batch time: 1.1288\n",
            "Phase: train Epoch: 3/3 Iter: 141/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 3/3 Iter: 142/1305 Batch time: 1.1304\n",
            "Phase: train Epoch: 3/3 Iter: 143/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 3/3 Iter: 144/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 3/3 Iter: 145/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 3/3 Iter: 146/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 3/3 Iter: 147/1305 Batch time: 1.1285\n",
            "Phase: train Epoch: 3/3 Iter: 148/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 3/3 Iter: 149/1305 Batch time: 1.2399\n",
            "Phase: train Epoch: 3/3 Iter: 150/1305 Batch time: 1.1172\n",
            "Phase: train Epoch: 3/3 Iter: 151/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 3/3 Iter: 152/1305 Batch time: 1.1148\n",
            "Phase: train Epoch: 3/3 Iter: 153/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 3/3 Iter: 154/1305 Batch time: 1.1155\n",
            "Phase: train Epoch: 3/3 Iter: 155/1305 Batch time: 1.1431\n",
            "Phase: train Epoch: 3/3 Iter: 156/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 3/3 Iter: 157/1305 Batch time: 1.1345\n",
            "Phase: train Epoch: 3/3 Iter: 158/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 3/3 Iter: 159/1305 Batch time: 1.2346\n",
            "Phase: train Epoch: 3/3 Iter: 160/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 3/3 Iter: 161/1305 Batch time: 1.1277\n",
            "Phase: train Epoch: 3/3 Iter: 162/1305 Batch time: 1.0874\n",
            "Phase: train Epoch: 3/3 Iter: 163/1305 Batch time: 1.1269\n",
            "Phase: train Epoch: 3/3 Iter: 164/1305 Batch time: 1.3527\n",
            "Phase: train Epoch: 3/3 Iter: 165/1305 Batch time: 1.5759\n",
            "Phase: train Epoch: 3/3 Iter: 166/1305 Batch time: 1.1135\n",
            "Phase: train Epoch: 3/3 Iter: 167/1305 Batch time: 1.1424\n",
            "Phase: train Epoch: 3/3 Iter: 168/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 3/3 Iter: 169/1305 Batch time: 1.2172\n",
            "Phase: train Epoch: 3/3 Iter: 170/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 3/3 Iter: 171/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 3/3 Iter: 172/1305 Batch time: 1.1281\n",
            "Phase: train Epoch: 3/3 Iter: 173/1305 Batch time: 1.1370\n",
            "Phase: train Epoch: 3/3 Iter: 174/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 3/3 Iter: 175/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 3/3 Iter: 176/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 3/3 Iter: 177/1305 Batch time: 1.2229\n",
            "Phase: train Epoch: 3/3 Iter: 178/1305 Batch time: 1.1204\n",
            "Phase: train Epoch: 3/3 Iter: 179/1305 Batch time: 1.1099\n",
            "Phase: train Epoch: 3/3 Iter: 180/1305 Batch time: 1.1197\n",
            "Phase: train Epoch: 3/3 Iter: 181/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 3/3 Iter: 182/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 3/3 Iter: 183/1305 Batch time: 1.1280\n",
            "Phase: train Epoch: 3/3 Iter: 184/1305 Batch time: 1.1322\n",
            "Phase: train Epoch: 3/3 Iter: 185/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 3/3 Iter: 186/1305 Batch time: 1.2275\n",
            "Phase: train Epoch: 3/3 Iter: 187/1305 Batch time: 1.1271\n",
            "Phase: train Epoch: 3/3 Iter: 188/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 3/3 Iter: 189/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 3/3 Iter: 190/1305 Batch time: 1.1160\n",
            "Phase: train Epoch: 3/3 Iter: 191/1305 Batch time: 1.1225\n",
            "Phase: train Epoch: 3/3 Iter: 192/1305 Batch time: 1.1287\n",
            "Phase: train Epoch: 3/3 Iter: 193/1305 Batch time: 1.1210\n",
            "Phase: train Epoch: 3/3 Iter: 194/1305 Batch time: 1.1064\n",
            "Phase: train Epoch: 3/3 Iter: 195/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 196/1305 Batch time: 1.1248\n",
            "Phase: train Epoch: 3/3 Iter: 197/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 3/3 Iter: 198/1305 Batch time: 1.1124\n",
            "Phase: train Epoch: 3/3 Iter: 199/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 3/3 Iter: 200/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 3/3 Iter: 201/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 3/3 Iter: 202/1305 Batch time: 1.1332\n",
            "Phase: train Epoch: 3/3 Iter: 203/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 3/3 Iter: 204/1305 Batch time: 1.1230\n",
            "Phase: train Epoch: 3/3 Iter: 205/1305 Batch time: 1.2323\n",
            "Phase: train Epoch: 3/3 Iter: 206/1305 Batch time: 1.1063\n",
            "Phase: train Epoch: 3/3 Iter: 207/1305 Batch time: 1.1322\n",
            "Phase: train Epoch: 3/3 Iter: 208/1305 Batch time: 1.1431\n",
            "Phase: train Epoch: 3/3 Iter: 209/1305 Batch time: 1.1275\n",
            "Phase: train Epoch: 3/3 Iter: 210/1305 Batch time: 1.1420\n",
            "Phase: train Epoch: 3/3 Iter: 211/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 3/3 Iter: 212/1305 Batch time: 1.1404\n",
            "Phase: train Epoch: 3/3 Iter: 213/1305 Batch time: 1.1255\n",
            "Phase: train Epoch: 3/3 Iter: 214/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 3/3 Iter: 215/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 3/3 Iter: 216/1305 Batch time: 1.1373\n",
            "Phase: train Epoch: 3/3 Iter: 217/1305 Batch time: 1.1264\n",
            "Phase: train Epoch: 3/3 Iter: 218/1305 Batch time: 1.1335\n",
            "Phase: train Epoch: 3/3 Iter: 219/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 3/3 Iter: 220/1305 Batch time: 1.2210\n",
            "Phase: train Epoch: 3/3 Iter: 221/1305 Batch time: 1.1032\n",
            "Phase: train Epoch: 3/3 Iter: 222/1305 Batch time: 1.1236\n",
            "Phase: train Epoch: 3/3 Iter: 223/1305 Batch time: 1.1351\n",
            "Phase: train Epoch: 3/3 Iter: 224/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 3/3 Iter: 225/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 3/3 Iter: 226/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 3/3 Iter: 227/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 3/3 Iter: 228/1305 Batch time: 1.1233\n",
            "Phase: train Epoch: 3/3 Iter: 229/1305 Batch time: 1.1170\n",
            "Phase: train Epoch: 3/3 Iter: 230/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 3/3 Iter: 231/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 3/3 Iter: 232/1305 Batch time: 1.2288\n",
            "Phase: train Epoch: 3/3 Iter: 233/1305 Batch time: 1.1458\n",
            "Phase: train Epoch: 3/3 Iter: 234/1305 Batch time: 1.1245\n",
            "Phase: train Epoch: 3/3 Iter: 235/1305 Batch time: 1.1107\n",
            "Phase: train Epoch: 3/3 Iter: 236/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 3/3 Iter: 237/1305 Batch time: 1.1299\n",
            "Phase: train Epoch: 3/3 Iter: 238/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 3/3 Iter: 239/1305 Batch time: 1.1196\n",
            "Phase: train Epoch: 3/3 Iter: 240/1305 Batch time: 1.1254\n",
            "Phase: train Epoch: 3/3 Iter: 241/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 3/3 Iter: 242/1305 Batch time: 1.1490\n",
            "Phase: train Epoch: 3/3 Iter: 243/1305 Batch time: 1.1259\n",
            "Phase: train Epoch: 3/3 Iter: 244/1305 Batch time: 1.6268\n",
            "Phase: train Epoch: 3/3 Iter: 245/1305 Batch time: 1.4204\n",
            "Phase: train Epoch: 3/3 Iter: 246/1305 Batch time: 1.1266\n",
            "Phase: train Epoch: 3/3 Iter: 247/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 3/3 Iter: 248/1305 Batch time: 1.1051\n",
            "Phase: train Epoch: 3/3 Iter: 249/1305 Batch time: 1.1225\n",
            "Phase: train Epoch: 3/3 Iter: 250/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 3/3 Iter: 251/1305 Batch time: 1.1209\n",
            "Phase: train Epoch: 3/3 Iter: 252/1305 Batch time: 1.1138\n",
            "Phase: train Epoch: 3/3 Iter: 253/1305 Batch time: 1.1150\n",
            "Phase: train Epoch: 3/3 Iter: 254/1305 Batch time: 1.2415\n",
            "Phase: train Epoch: 3/3 Iter: 255/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 3/3 Iter: 256/1305 Batch time: 1.1136\n",
            "Phase: train Epoch: 3/3 Iter: 257/1305 Batch time: 1.1289\n",
            "Phase: train Epoch: 3/3 Iter: 258/1305 Batch time: 1.1197\n",
            "Phase: train Epoch: 3/3 Iter: 259/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 3/3 Iter: 260/1305 Batch time: 1.1265\n",
            "Phase: train Epoch: 3/3 Iter: 261/1305 Batch time: 1.1337\n",
            "Phase: train Epoch: 3/3 Iter: 262/1305 Batch time: 1.1329\n",
            "Phase: train Epoch: 3/3 Iter: 263/1305 Batch time: 1.2223\n",
            "Phase: train Epoch: 3/3 Iter: 264/1305 Batch time: 1.1242\n",
            "Phase: train Epoch: 3/3 Iter: 265/1305 Batch time: 1.1402\n",
            "Phase: train Epoch: 3/3 Iter: 266/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 3/3 Iter: 267/1305 Batch time: 1.1220\n",
            "Phase: train Epoch: 3/3 Iter: 268/1305 Batch time: 1.1257\n",
            "Phase: train Epoch: 3/3 Iter: 269/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 3/3 Iter: 270/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 3/3 Iter: 271/1305 Batch time: 1.1403\n",
            "Phase: train Epoch: 3/3 Iter: 272/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 3/3 Iter: 273/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 3/3 Iter: 274/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 3/3 Iter: 275/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 276/1305 Batch time: 1.1216\n",
            "Phase: train Epoch: 3/3 Iter: 277/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 3/3 Iter: 278/1305 Batch time: 1.2415\n",
            "Phase: train Epoch: 3/3 Iter: 279/1305 Batch time: 1.1287\n",
            "Phase: train Epoch: 3/3 Iter: 280/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 3/3 Iter: 281/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 282/1305 Batch time: 1.1586\n",
            "Phase: train Epoch: 3/3 Iter: 283/1305 Batch time: 1.1234\n",
            "Phase: train Epoch: 3/3 Iter: 284/1305 Batch time: 1.1158\n",
            "Phase: train Epoch: 3/3 Iter: 285/1305 Batch time: 1.1244\n",
            "Phase: train Epoch: 3/3 Iter: 286/1305 Batch time: 1.1470\n",
            "Phase: train Epoch: 3/3 Iter: 287/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 3/3 Iter: 288/1305 Batch time: 1.1157\n",
            "Phase: train Epoch: 3/3 Iter: 289/1305 Batch time: 1.1073\n",
            "Phase: train Epoch: 3/3 Iter: 290/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 3/3 Iter: 291/1305 Batch time: 1.2265\n",
            "Phase: train Epoch: 3/3 Iter: 292/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 3/3 Iter: 293/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 294/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 3/3 Iter: 295/1305 Batch time: 1.1122\n",
            "Phase: train Epoch: 3/3 Iter: 296/1305 Batch time: 1.1245\n",
            "Phase: train Epoch: 3/3 Iter: 297/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 3/3 Iter: 298/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 3/3 Iter: 299/1305 Batch time: 1.0938\n",
            "Phase: train Epoch: 3/3 Iter: 300/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 3/3 Iter: 301/1305 Batch time: 1.2081\n",
            "Phase: train Epoch: 3/3 Iter: 302/1305 Batch time: 1.1197\n",
            "Phase: train Epoch: 3/3 Iter: 303/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 3/3 Iter: 304/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 3/3 Iter: 305/1305 Batch time: 1.1020\n",
            "Phase: train Epoch: 3/3 Iter: 306/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 3/3 Iter: 307/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 3/3 Iter: 308/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 3/3 Iter: 309/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 3/3 Iter: 310/1305 Batch time: 1.1211\n",
            "Phase: train Epoch: 3/3 Iter: 311/1305 Batch time: 1.2287\n",
            "Phase: train Epoch: 3/3 Iter: 312/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 313/1305 Batch time: 1.1185\n",
            "Phase: train Epoch: 3/3 Iter: 314/1305 Batch time: 1.1466\n",
            "Phase: train Epoch: 3/3 Iter: 315/1305 Batch time: 1.1180\n",
            "Phase: train Epoch: 3/3 Iter: 316/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 3/3 Iter: 317/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 3/3 Iter: 318/1305 Batch time: 1.1314\n",
            "Phase: train Epoch: 3/3 Iter: 319/1305 Batch time: 1.2175\n",
            "Phase: train Epoch: 3/3 Iter: 320/1305 Batch time: 1.1327\n",
            "Phase: train Epoch: 3/3 Iter: 321/1305 Batch time: 1.1356\n",
            "Phase: train Epoch: 3/3 Iter: 322/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 3/3 Iter: 323/1305 Batch time: 1.1201\n",
            "Phase: train Epoch: 3/3 Iter: 324/1305 Batch time: 1.4708\n",
            "Phase: train Epoch: 3/3 Iter: 325/1305 Batch time: 1.4733\n",
            "Phase: train Epoch: 3/3 Iter: 326/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 327/1305 Batch time: 1.1394\n",
            "Phase: train Epoch: 3/3 Iter: 328/1305 Batch time: 1.2253\n",
            "Phase: train Epoch: 3/3 Iter: 329/1305 Batch time: 1.1360\n",
            "Phase: train Epoch: 3/3 Iter: 330/1305 Batch time: 1.1155\n",
            "Phase: train Epoch: 3/3 Iter: 331/1305 Batch time: 1.1305\n",
            "Phase: train Epoch: 3/3 Iter: 332/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 333/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 334/1305 Batch time: 1.1019\n",
            "Phase: train Epoch: 3/3 Iter: 335/1305 Batch time: 1.1271\n",
            "Phase: train Epoch: 3/3 Iter: 336/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 3/3 Iter: 337/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 3/3 Iter: 338/1305 Batch time: 1.1354\n",
            "Phase: train Epoch: 3/3 Iter: 339/1305 Batch time: 1.1311\n",
            "Phase: train Epoch: 3/3 Iter: 340/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 3/3 Iter: 341/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 3/3 Iter: 342/1305 Batch time: 1.1278\n",
            "Phase: train Epoch: 3/3 Iter: 343/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 3/3 Iter: 344/1305 Batch time: 1.1273\n",
            "Phase: train Epoch: 3/3 Iter: 345/1305 Batch time: 1.1226\n",
            "Phase: train Epoch: 3/3 Iter: 346/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 3/3 Iter: 347/1305 Batch time: 1.2293\n",
            "Phase: train Epoch: 3/3 Iter: 348/1305 Batch time: 1.1347\n",
            "Phase: train Epoch: 3/3 Iter: 349/1305 Batch time: 1.1126\n",
            "Phase: train Epoch: 3/3 Iter: 350/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 3/3 Iter: 351/1305 Batch time: 1.1098\n",
            "Phase: train Epoch: 3/3 Iter: 352/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 3/3 Iter: 353/1305 Batch time: 1.1165\n",
            "Phase: train Epoch: 3/3 Iter: 354/1305 Batch time: 1.1101\n",
            "Phase: train Epoch: 3/3 Iter: 355/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 3/3 Iter: 356/1305 Batch time: 1.1261\n",
            "Phase: train Epoch: 3/3 Iter: 357/1305 Batch time: 1.1189\n",
            "Phase: train Epoch: 3/3 Iter: 358/1305 Batch time: 1.1043\n",
            "Phase: train Epoch: 3/3 Iter: 359/1305 Batch time: 1.1258\n",
            "Phase: train Epoch: 3/3 Iter: 360/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 3/3 Iter: 361/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 3/3 Iter: 362/1305 Batch time: 1.2150\n",
            "Phase: train Epoch: 3/3 Iter: 363/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 3/3 Iter: 364/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 365/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 366/1305 Batch time: 1.1125\n",
            "Phase: train Epoch: 3/3 Iter: 367/1305 Batch time: 1.1355\n",
            "Phase: train Epoch: 3/3 Iter: 368/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 369/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 3/3 Iter: 370/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 371/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 3/3 Iter: 372/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 3/3 Iter: 373/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 3/3 Iter: 374/1305 Batch time: 1.2281\n",
            "Phase: train Epoch: 3/3 Iter: 375/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 3/3 Iter: 376/1305 Batch time: 1.1251\n",
            "Phase: train Epoch: 3/3 Iter: 377/1305 Batch time: 1.1279\n",
            "Phase: train Epoch: 3/3 Iter: 378/1305 Batch time: 1.1006\n",
            "Phase: train Epoch: 3/3 Iter: 379/1305 Batch time: 1.1188\n",
            "Phase: train Epoch: 3/3 Iter: 380/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 3/3 Iter: 381/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 3/3 Iter: 382/1305 Batch time: 1.1209\n",
            "Phase: train Epoch: 3/3 Iter: 383/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 3/3 Iter: 384/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 3/3 Iter: 385/1305 Batch time: 1.1212\n",
            "Phase: train Epoch: 3/3 Iter: 386/1305 Batch time: 1.2238\n",
            "Phase: train Epoch: 3/3 Iter: 387/1305 Batch time: 1.1085\n",
            "Phase: train Epoch: 3/3 Iter: 388/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 389/1305 Batch time: 1.1220\n",
            "Phase: train Epoch: 3/3 Iter: 390/1305 Batch time: 1.1195\n",
            "Phase: train Epoch: 3/3 Iter: 391/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 3/3 Iter: 392/1305 Batch time: 1.1332\n",
            "Phase: train Epoch: 3/3 Iter: 393/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 3/3 Iter: 394/1305 Batch time: 1.1243\n",
            "Phase: train Epoch: 3/3 Iter: 395/1305 Batch time: 1.1256\n",
            "Phase: train Epoch: 3/3 Iter: 396/1305 Batch time: 1.2222\n",
            "Phase: train Epoch: 3/3 Iter: 397/1305 Batch time: 1.1379\n",
            "Phase: train Epoch: 3/3 Iter: 398/1305 Batch time: 1.1223\n",
            "Phase: train Epoch: 3/3 Iter: 399/1305 Batch time: 1.1311\n",
            "Phase: train Epoch: 3/3 Iter: 400/1305 Batch time: 1.1214\n",
            "Phase: train Epoch: 3/3 Iter: 401/1305 Batch time: 1.1017\n",
            "Phase: train Epoch: 3/3 Iter: 402/1305 Batch time: 1.1189\n",
            "Phase: train Epoch: 3/3 Iter: 403/1305 Batch time: 1.4456\n",
            "Phase: train Epoch: 3/3 Iter: 404/1305 Batch time: 1.4608\n",
            "Phase: train Epoch: 3/3 Iter: 405/1305 Batch time: 1.2113\n",
            "Phase: train Epoch: 3/3 Iter: 406/1305 Batch time: 1.1277\n",
            "Phase: train Epoch: 3/3 Iter: 407/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 408/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 3/3 Iter: 409/1305 Batch time: 1.1065\n",
            "Phase: train Epoch: 3/3 Iter: 410/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 3/3 Iter: 411/1305 Batch time: 1.1173\n",
            "Phase: train Epoch: 3/3 Iter: 412/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 413/1305 Batch time: 1.1131\n",
            "Phase: train Epoch: 3/3 Iter: 414/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 3/3 Iter: 415/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 3/3 Iter: 416/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 3/3 Iter: 417/1305 Batch time: 1.1176\n",
            "Phase: train Epoch: 3/3 Iter: 418/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 3/3 Iter: 419/1305 Batch time: 1.1284\n",
            "Phase: train Epoch: 3/3 Iter: 420/1305 Batch time: 1.2205\n",
            "Phase: train Epoch: 3/3 Iter: 421/1305 Batch time: 1.1140\n",
            "Phase: train Epoch: 3/3 Iter: 422/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 3/3 Iter: 423/1305 Batch time: 1.1190\n",
            "Phase: train Epoch: 3/3 Iter: 424/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 3/3 Iter: 425/1305 Batch time: 1.1142\n",
            "Phase: train Epoch: 3/3 Iter: 426/1305 Batch time: 1.1187\n",
            "Phase: train Epoch: 3/3 Iter: 427/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 3/3 Iter: 428/1305 Batch time: 1.0930\n",
            "Phase: train Epoch: 3/3 Iter: 429/1305 Batch time: 1.0845\n",
            "Phase: train Epoch: 3/3 Iter: 430/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 431/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 3/3 Iter: 432/1305 Batch time: 1.0906\n",
            "Phase: train Epoch: 3/3 Iter: 433/1305 Batch time: 1.2140\n",
            "Phase: train Epoch: 3/3 Iter: 434/1305 Batch time: 1.1018\n",
            "Phase: train Epoch: 3/3 Iter: 435/1305 Batch time: 1.1022\n",
            "Phase: train Epoch: 3/3 Iter: 436/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 3/3 Iter: 437/1305 Batch time: 1.0922\n",
            "Phase: train Epoch: 3/3 Iter: 438/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 3/3 Iter: 439/1305 Batch time: 1.0890\n",
            "Phase: train Epoch: 3/3 Iter: 440/1305 Batch time: 1.0900\n",
            "Phase: train Epoch: 3/3 Iter: 441/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 442/1305 Batch time: 1.1083\n",
            "Phase: train Epoch: 3/3 Iter: 443/1305 Batch time: 1.2138\n",
            "Phase: train Epoch: 3/3 Iter: 444/1305 Batch time: 1.1054\n",
            "Phase: train Epoch: 3/3 Iter: 445/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 3/3 Iter: 446/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 3/3 Iter: 447/1305 Batch time: 1.0826\n",
            "Phase: train Epoch: 3/3 Iter: 448/1305 Batch time: 1.0950\n",
            "Phase: train Epoch: 3/3 Iter: 449/1305 Batch time: 1.1113\n",
            "Phase: train Epoch: 3/3 Iter: 450/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 3/3 Iter: 451/1305 Batch time: 1.0961\n",
            "Phase: train Epoch: 3/3 Iter: 452/1305 Batch time: 1.1141\n",
            "Phase: train Epoch: 3/3 Iter: 453/1305 Batch time: 1.2035\n",
            "Phase: train Epoch: 3/3 Iter: 454/1305 Batch time: 1.0899\n",
            "Phase: train Epoch: 3/3 Iter: 455/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 3/3 Iter: 456/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 457/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 3/3 Iter: 458/1305 Batch time: 1.1071\n",
            "Phase: train Epoch: 3/3 Iter: 459/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 3/3 Iter: 460/1305 Batch time: 1.0969\n",
            "Phase: train Epoch: 3/3 Iter: 461/1305 Batch time: 1.2230\n",
            "Phase: train Epoch: 3/3 Iter: 462/1305 Batch time: 1.0784\n",
            "Phase: train Epoch: 3/3 Iter: 463/1305 Batch time: 1.0810\n",
            "Phase: train Epoch: 3/3 Iter: 464/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 3/3 Iter: 465/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 3/3 Iter: 466/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 3/3 Iter: 467/1305 Batch time: 1.1108\n",
            "Phase: train Epoch: 3/3 Iter: 468/1305 Batch time: 1.0814\n",
            "Phase: train Epoch: 3/3 Iter: 469/1305 Batch time: 1.1235\n",
            "Phase: train Epoch: 3/3 Iter: 470/1305 Batch time: 1.2102\n",
            "Phase: train Epoch: 3/3 Iter: 471/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 3/3 Iter: 472/1305 Batch time: 1.0896\n",
            "Phase: train Epoch: 3/3 Iter: 473/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 474/1305 Batch time: 1.0990\n",
            "Phase: train Epoch: 3/3 Iter: 475/1305 Batch time: 1.0880\n",
            "Phase: train Epoch: 3/3 Iter: 476/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 3/3 Iter: 477/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 3/3 Iter: 478/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 3/3 Iter: 479/1305 Batch time: 1.0875\n",
            "Phase: train Epoch: 3/3 Iter: 480/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 3/3 Iter: 481/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 3/3 Iter: 482/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 483/1305 Batch time: 1.3994\n",
            "Phase: train Epoch: 3/3 Iter: 484/1305 Batch time: 1.5286\n",
            "Phase: train Epoch: 3/3 Iter: 485/1305 Batch time: 1.0947\n",
            "Phase: train Epoch: 3/3 Iter: 486/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 3/3 Iter: 487/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 3/3 Iter: 488/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 3/3 Iter: 489/1305 Batch time: 1.2224\n",
            "Phase: train Epoch: 3/3 Iter: 490/1305 Batch time: 1.0890\n",
            "Phase: train Epoch: 3/3 Iter: 491/1305 Batch time: 1.0921\n",
            "Phase: train Epoch: 3/3 Iter: 492/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 3/3 Iter: 493/1305 Batch time: 1.0874\n",
            "Phase: train Epoch: 3/3 Iter: 494/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 3/3 Iter: 495/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 3/3 Iter: 496/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 3/3 Iter: 497/1305 Batch time: 1.0730\n",
            "Phase: train Epoch: 3/3 Iter: 498/1305 Batch time: 1.1192\n",
            "Phase: train Epoch: 3/3 Iter: 499/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 3/3 Iter: 500/1305 Batch time: 1.0942\n",
            "Phase: train Epoch: 3/3 Iter: 501/1305 Batch time: 1.0849\n",
            "Phase: train Epoch: 3/3 Iter: 502/1305 Batch time: 1.1074\n",
            "Phase: train Epoch: 3/3 Iter: 503/1305 Batch time: 1.0754\n",
            "Phase: train Epoch: 3/3 Iter: 504/1305 Batch time: 1.1926\n",
            "Phase: train Epoch: 3/3 Iter: 505/1305 Batch time: 1.1083\n",
            "Phase: train Epoch: 3/3 Iter: 506/1305 Batch time: 1.0970\n",
            "Phase: train Epoch: 3/3 Iter: 507/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 3/3 Iter: 508/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 3/3 Iter: 509/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 3/3 Iter: 510/1305 Batch time: 1.0682\n",
            "Phase: train Epoch: 3/3 Iter: 511/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 3/3 Iter: 512/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 3/3 Iter: 513/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 3/3 Iter: 514/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 3/3 Iter: 515/1305 Batch time: 1.0966\n",
            "Phase: train Epoch: 3/3 Iter: 516/1305 Batch time: 1.1843\n",
            "Phase: train Epoch: 3/3 Iter: 517/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 518/1305 Batch time: 1.0954\n",
            "Phase: train Epoch: 3/3 Iter: 519/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 3/3 Iter: 520/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 3/3 Iter: 521/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 522/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 3/3 Iter: 523/1305 Batch time: 1.1060\n",
            "Phase: train Epoch: 3/3 Iter: 524/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 3/3 Iter: 525/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 3/3 Iter: 526/1305 Batch time: 1.0806\n",
            "Phase: train Epoch: 3/3 Iter: 527/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 3/3 Iter: 528/1305 Batch time: 1.2119\n",
            "Phase: train Epoch: 3/3 Iter: 529/1305 Batch time: 1.0925\n",
            "Phase: train Epoch: 3/3 Iter: 530/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 3/3 Iter: 531/1305 Batch time: 1.1175\n",
            "Phase: train Epoch: 3/3 Iter: 532/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 3/3 Iter: 533/1305 Batch time: 1.1139\n",
            "Phase: train Epoch: 3/3 Iter: 534/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 3/3 Iter: 535/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 3/3 Iter: 536/1305 Batch time: 1.1025\n",
            "Phase: train Epoch: 3/3 Iter: 537/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 538/1305 Batch time: 1.2088\n",
            "Phase: train Epoch: 3/3 Iter: 539/1305 Batch time: 1.1158\n",
            "Phase: train Epoch: 3/3 Iter: 540/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 3/3 Iter: 541/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 3/3 Iter: 542/1305 Batch time: 1.1218\n",
            "Phase: train Epoch: 3/3 Iter: 543/1305 Batch time: 1.1333\n",
            "Phase: train Epoch: 3/3 Iter: 544/1305 Batch time: 1.1330\n",
            "Phase: train Epoch: 3/3 Iter: 545/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 3/3 Iter: 546/1305 Batch time: 1.1120\n",
            "Phase: train Epoch: 3/3 Iter: 547/1305 Batch time: 1.2373\n",
            "Phase: train Epoch: 3/3 Iter: 548/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 3/3 Iter: 549/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 3/3 Iter: 550/1305 Batch time: 1.0797\n",
            "Phase: train Epoch: 3/3 Iter: 551/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 3/3 Iter: 552/1305 Batch time: 1.0869\n",
            "Phase: train Epoch: 3/3 Iter: 553/1305 Batch time: 1.0720\n",
            "Phase: train Epoch: 3/3 Iter: 554/1305 Batch time: 1.0900\n",
            "Phase: train Epoch: 3/3 Iter: 555/1305 Batch time: 1.0940\n",
            "Phase: train Epoch: 3/3 Iter: 556/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 3/3 Iter: 557/1305 Batch time: 1.0953\n",
            "Phase: train Epoch: 3/3 Iter: 558/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 3/3 Iter: 559/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 3/3 Iter: 560/1305 Batch time: 1.0895\n",
            "Phase: train Epoch: 3/3 Iter: 561/1305 Batch time: 1.0980\n",
            "Phase: train Epoch: 3/3 Iter: 562/1305 Batch time: 1.1896\n",
            "Phase: train Epoch: 3/3 Iter: 563/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 3/3 Iter: 564/1305 Batch time: 1.3932\n",
            "Phase: train Epoch: 3/3 Iter: 565/1305 Batch time: 1.4958\n",
            "Phase: train Epoch: 3/3 Iter: 566/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 3/3 Iter: 567/1305 Batch time: 1.0905\n",
            "Phase: train Epoch: 3/3 Iter: 568/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 3/3 Iter: 569/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 3/3 Iter: 570/1305 Batch time: 1.0864\n",
            "Phase: train Epoch: 3/3 Iter: 571/1305 Batch time: 1.1286\n",
            "Phase: train Epoch: 3/3 Iter: 572/1305 Batch time: 1.0848\n",
            "Phase: train Epoch: 3/3 Iter: 573/1305 Batch time: 1.0950\n",
            "Phase: train Epoch: 3/3 Iter: 574/1305 Batch time: 1.0940\n",
            "Phase: train Epoch: 3/3 Iter: 575/1305 Batch time: 1.2008\n",
            "Phase: train Epoch: 3/3 Iter: 576/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 3/3 Iter: 577/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 3/3 Iter: 578/1305 Batch time: 1.0831\n",
            "Phase: train Epoch: 3/3 Iter: 579/1305 Batch time: 1.0786\n",
            "Phase: train Epoch: 3/3 Iter: 580/1305 Batch time: 1.0937\n",
            "Phase: train Epoch: 3/3 Iter: 581/1305 Batch time: 1.0962\n",
            "Phase: train Epoch: 3/3 Iter: 582/1305 Batch time: 1.0861\n",
            "Phase: train Epoch: 3/3 Iter: 583/1305 Batch time: 1.0907\n",
            "Phase: train Epoch: 3/3 Iter: 584/1305 Batch time: 1.1101\n",
            "Phase: train Epoch: 3/3 Iter: 585/1305 Batch time: 1.1870\n",
            "Phase: train Epoch: 3/3 Iter: 586/1305 Batch time: 1.0995\n",
            "Phase: train Epoch: 3/3 Iter: 587/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 3/3 Iter: 588/1305 Batch time: 1.0942\n",
            "Phase: train Epoch: 3/3 Iter: 589/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 3/3 Iter: 590/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 3/3 Iter: 591/1305 Batch time: 1.0940\n",
            "Phase: train Epoch: 3/3 Iter: 592/1305 Batch time: 1.1067\n",
            "Phase: train Epoch: 3/3 Iter: 593/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 3/3 Iter: 594/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 3/3 Iter: 595/1305 Batch time: 1.2207\n",
            "Phase: train Epoch: 3/3 Iter: 596/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 3/3 Iter: 597/1305 Batch time: 1.1159\n",
            "Phase: train Epoch: 3/3 Iter: 598/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 3/3 Iter: 599/1305 Batch time: 1.1356\n",
            "Phase: train Epoch: 3/3 Iter: 600/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 3/3 Iter: 601/1305 Batch time: 1.0970\n",
            "Phase: train Epoch: 3/3 Iter: 602/1305 Batch time: 1.0895\n",
            "Phase: train Epoch: 3/3 Iter: 603/1305 Batch time: 1.2049\n",
            "Phase: train Epoch: 3/3 Iter: 604/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 605/1305 Batch time: 1.0993\n",
            "Phase: train Epoch: 3/3 Iter: 606/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 607/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 3/3 Iter: 608/1305 Batch time: 1.1018\n",
            "Phase: train Epoch: 3/3 Iter: 609/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 610/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 3/3 Iter: 611/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 3/3 Iter: 612/1305 Batch time: 1.2082\n",
            "Phase: train Epoch: 3/3 Iter: 613/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 614/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 3/3 Iter: 615/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 3/3 Iter: 616/1305 Batch time: 1.1077\n",
            "Phase: train Epoch: 3/3 Iter: 617/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 3/3 Iter: 618/1305 Batch time: 1.0825\n",
            "Phase: train Epoch: 3/3 Iter: 619/1305 Batch time: 1.1191\n",
            "Phase: train Epoch: 3/3 Iter: 620/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 3/3 Iter: 621/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 3/3 Iter: 622/1305 Batch time: 1.1106\n",
            "Phase: train Epoch: 3/3 Iter: 623/1305 Batch time: 1.0982\n",
            "Phase: train Epoch: 3/3 Iter: 624/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 3/3 Iter: 625/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 3/3 Iter: 626/1305 Batch time: 1.0694\n",
            "Phase: train Epoch: 3/3 Iter: 627/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 3/3 Iter: 628/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 3/3 Iter: 629/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 3/3 Iter: 630/1305 Batch time: 1.1021\n",
            "Phase: train Epoch: 3/3 Iter: 631/1305 Batch time: 1.2151\n",
            "Phase: train Epoch: 3/3 Iter: 632/1305 Batch time: 1.0922\n",
            "Phase: train Epoch: 3/3 Iter: 633/1305 Batch time: 1.0941\n",
            "Phase: train Epoch: 3/3 Iter: 634/1305 Batch time: 1.1135\n",
            "Phase: train Epoch: 3/3 Iter: 635/1305 Batch time: 1.0727\n",
            "Phase: train Epoch: 3/3 Iter: 636/1305 Batch time: 1.0994\n",
            "Phase: train Epoch: 3/3 Iter: 637/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 3/3 Iter: 638/1305 Batch time: 1.0875\n",
            "Phase: train Epoch: 3/3 Iter: 639/1305 Batch time: 1.0775\n",
            "Phase: train Epoch: 3/3 Iter: 640/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 3/3 Iter: 641/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 3/3 Iter: 642/1305 Batch time: 1.0880\n",
            "Phase: train Epoch: 3/3 Iter: 643/1305 Batch time: 1.0857\n",
            "Phase: train Epoch: 3/3 Iter: 644/1305 Batch time: 1.0981\n",
            "Phase: train Epoch: 3/3 Iter: 645/1305 Batch time: 1.5150\n",
            "Phase: train Epoch: 3/3 Iter: 646/1305 Batch time: 1.4381\n",
            "Phase: train Epoch: 3/3 Iter: 647/1305 Batch time: 1.0868\n",
            "Phase: train Epoch: 3/3 Iter: 648/1305 Batch time: 1.0975\n",
            "Phase: train Epoch: 3/3 Iter: 649/1305 Batch time: 1.0916\n",
            "Phase: train Epoch: 3/3 Iter: 650/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 3/3 Iter: 651/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 3/3 Iter: 652/1305 Batch time: 1.1014\n",
            "Phase: train Epoch: 3/3 Iter: 653/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 3/3 Iter: 654/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 3/3 Iter: 655/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 3/3 Iter: 656/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 657/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 3/3 Iter: 658/1305 Batch time: 1.2007\n",
            "Phase: train Epoch: 3/3 Iter: 659/1305 Batch time: 1.1237\n",
            "Phase: train Epoch: 3/3 Iter: 660/1305 Batch time: 1.0814\n",
            "Phase: train Epoch: 3/3 Iter: 661/1305 Batch time: 1.0726\n",
            "Phase: train Epoch: 3/3 Iter: 662/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 3/3 Iter: 663/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 3/3 Iter: 664/1305 Batch time: 1.0859\n",
            "Phase: train Epoch: 3/3 Iter: 665/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 3/3 Iter: 666/1305 Batch time: 1.1250\n",
            "Phase: train Epoch: 3/3 Iter: 667/1305 Batch time: 1.0903\n",
            "Phase: train Epoch: 3/3 Iter: 668/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 3/3 Iter: 669/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 3/3 Iter: 670/1305 Batch time: 1.2129\n",
            "Phase: train Epoch: 3/3 Iter: 671/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 672/1305 Batch time: 1.0809\n",
            "Phase: train Epoch: 3/3 Iter: 673/1305 Batch time: 1.1316\n",
            "Phase: train Epoch: 3/3 Iter: 674/1305 Batch time: 1.0881\n",
            "Phase: train Epoch: 3/3 Iter: 675/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 3/3 Iter: 676/1305 Batch time: 1.0753\n",
            "Phase: train Epoch: 3/3 Iter: 677/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 3/3 Iter: 678/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 3/3 Iter: 679/1305 Batch time: 1.0844\n",
            "Phase: train Epoch: 3/3 Iter: 680/1305 Batch time: 1.2138\n",
            "Phase: train Epoch: 3/3 Iter: 681/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 3/3 Iter: 682/1305 Batch time: 1.0956\n",
            "Phase: train Epoch: 3/3 Iter: 683/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 3/3 Iter: 684/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 3/3 Iter: 685/1305 Batch time: 1.0871\n",
            "Phase: train Epoch: 3/3 Iter: 686/1305 Batch time: 1.0853\n",
            "Phase: train Epoch: 3/3 Iter: 687/1305 Batch time: 1.0783\n",
            "Phase: train Epoch: 3/3 Iter: 688/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 3/3 Iter: 689/1305 Batch time: 1.2007\n",
            "Phase: train Epoch: 3/3 Iter: 690/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 3/3 Iter: 691/1305 Batch time: 1.0944\n",
            "Phase: train Epoch: 3/3 Iter: 692/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 3/3 Iter: 693/1305 Batch time: 1.0911\n",
            "Phase: train Epoch: 3/3 Iter: 694/1305 Batch time: 1.1111\n",
            "Phase: train Epoch: 3/3 Iter: 695/1305 Batch time: 1.0947\n",
            "Phase: train Epoch: 3/3 Iter: 696/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 3/3 Iter: 697/1305 Batch time: 1.0923\n",
            "Phase: train Epoch: 3/3 Iter: 698/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 3/3 Iter: 699/1305 Batch time: 1.1194\n",
            "Phase: train Epoch: 3/3 Iter: 700/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 701/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 3/3 Iter: 702/1305 Batch time: 1.1170\n",
            "Phase: train Epoch: 3/3 Iter: 703/1305 Batch time: 1.0937\n",
            "Phase: train Epoch: 3/3 Iter: 704/1305 Batch time: 1.2033\n",
            "Phase: train Epoch: 3/3 Iter: 705/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 3/3 Iter: 706/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 3/3 Iter: 707/1305 Batch time: 1.1093\n",
            "Phase: train Epoch: 3/3 Iter: 708/1305 Batch time: 1.1169\n",
            "Phase: train Epoch: 3/3 Iter: 709/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 710/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 3/3 Iter: 711/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 3/3 Iter: 712/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 3/3 Iter: 713/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 3/3 Iter: 714/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 3/3 Iter: 715/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 3/3 Iter: 716/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 717/1305 Batch time: 1.2224\n",
            "Phase: train Epoch: 3/3 Iter: 718/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 3/3 Iter: 719/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 3/3 Iter: 720/1305 Batch time: 1.1127\n",
            "Phase: train Epoch: 3/3 Iter: 721/1305 Batch time: 1.1193\n",
            "Phase: train Epoch: 3/3 Iter: 722/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 3/3 Iter: 723/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 3/3 Iter: 724/1305 Batch time: 1.1646\n",
            "Phase: train Epoch: 3/3 Iter: 725/1305 Batch time: 1.5344\n",
            "Phase: train Epoch: 3/3 Iter: 726/1305 Batch time: 1.2451\n",
            "Phase: train Epoch: 3/3 Iter: 727/1305 Batch time: 1.2216\n",
            "Phase: train Epoch: 3/3 Iter: 728/1305 Batch time: 1.1063\n",
            "Phase: train Epoch: 3/3 Iter: 729/1305 Batch time: 1.1105\n",
            "Phase: train Epoch: 3/3 Iter: 730/1305 Batch time: 1.1109\n",
            "Phase: train Epoch: 3/3 Iter: 731/1305 Batch time: 1.1057\n",
            "Phase: train Epoch: 3/3 Iter: 732/1305 Batch time: 1.1102\n",
            "Phase: train Epoch: 3/3 Iter: 733/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 3/3 Iter: 734/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 3/3 Iter: 735/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 3/3 Iter: 736/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 737/1305 Batch time: 1.2080\n",
            "Phase: train Epoch: 3/3 Iter: 738/1305 Batch time: 1.1068\n",
            "Phase: train Epoch: 3/3 Iter: 739/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 3/3 Iter: 740/1305 Batch time: 1.1119\n",
            "Phase: train Epoch: 3/3 Iter: 741/1305 Batch time: 1.0967\n",
            "Phase: train Epoch: 3/3 Iter: 742/1305 Batch time: 1.1257\n",
            "Phase: train Epoch: 3/3 Iter: 743/1305 Batch time: 1.1043\n",
            "Phase: train Epoch: 3/3 Iter: 744/1305 Batch time: 1.1156\n",
            "Phase: train Epoch: 3/3 Iter: 745/1305 Batch time: 1.2306\n",
            "Phase: train Epoch: 3/3 Iter: 746/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 3/3 Iter: 747/1305 Batch time: 1.0929\n",
            "Phase: train Epoch: 3/3 Iter: 748/1305 Batch time: 1.1203\n",
            "Phase: train Epoch: 3/3 Iter: 749/1305 Batch time: 1.1248\n",
            "Phase: train Epoch: 3/3 Iter: 750/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 3/3 Iter: 751/1305 Batch time: 1.1039\n",
            "Phase: train Epoch: 3/3 Iter: 752/1305 Batch time: 1.1002\n",
            "Phase: train Epoch: 3/3 Iter: 753/1305 Batch time: 1.0964\n",
            "Phase: train Epoch: 3/3 Iter: 754/1305 Batch time: 1.2080\n",
            "Phase: train Epoch: 3/3 Iter: 755/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 3/3 Iter: 756/1305 Batch time: 1.0937\n",
            "Phase: train Epoch: 3/3 Iter: 757/1305 Batch time: 1.0910\n",
            "Phase: train Epoch: 3/3 Iter: 758/1305 Batch time: 1.0863\n",
            "Phase: train Epoch: 3/3 Iter: 759/1305 Batch time: 1.0883\n",
            "Phase: train Epoch: 3/3 Iter: 760/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 3/3 Iter: 761/1305 Batch time: 1.1015\n",
            "Phase: train Epoch: 3/3 Iter: 762/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 3/3 Iter: 763/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 3/3 Iter: 764/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 3/3 Iter: 765/1305 Batch time: 1.0962\n",
            "Phase: train Epoch: 3/3 Iter: 766/1305 Batch time: 1.0931\n",
            "Phase: train Epoch: 3/3 Iter: 767/1305 Batch time: 1.0958\n",
            "Phase: train Epoch: 3/3 Iter: 768/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 3/3 Iter: 769/1305 Batch time: 1.1192\n",
            "Phase: train Epoch: 3/3 Iter: 770/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 3/3 Iter: 771/1305 Batch time: 1.1018\n",
            "Phase: train Epoch: 3/3 Iter: 772/1305 Batch time: 1.0912\n",
            "Phase: train Epoch: 3/3 Iter: 773/1305 Batch time: 1.2227\n",
            "Phase: train Epoch: 3/3 Iter: 774/1305 Batch time: 1.1012\n",
            "Phase: train Epoch: 3/3 Iter: 775/1305 Batch time: 1.0862\n",
            "Phase: train Epoch: 3/3 Iter: 776/1305 Batch time: 1.0855\n",
            "Phase: train Epoch: 3/3 Iter: 777/1305 Batch time: 1.0955\n",
            "Phase: train Epoch: 3/3 Iter: 778/1305 Batch time: 1.0876\n",
            "Phase: train Epoch: 3/3 Iter: 779/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 3/3 Iter: 780/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 3/3 Iter: 781/1305 Batch time: 1.0928\n",
            "Phase: train Epoch: 3/3 Iter: 782/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 3/3 Iter: 783/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 3/3 Iter: 784/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 3/3 Iter: 785/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 3/3 Iter: 786/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 3/3 Iter: 787/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 3/3 Iter: 788/1305 Batch time: 1.1908\n",
            "Phase: train Epoch: 3/3 Iter: 789/1305 Batch time: 1.0849\n",
            "Phase: train Epoch: 3/3 Iter: 790/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 3/3 Iter: 791/1305 Batch time: 1.0879\n",
            "Phase: train Epoch: 3/3 Iter: 792/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 3/3 Iter: 793/1305 Batch time: 1.1095\n",
            "Phase: train Epoch: 3/3 Iter: 794/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 3/3 Iter: 795/1305 Batch time: 1.0927\n",
            "Phase: train Epoch: 3/3 Iter: 796/1305 Batch time: 1.1064\n",
            "Phase: train Epoch: 3/3 Iter: 797/1305 Batch time: 1.1076\n",
            "Phase: train Epoch: 3/3 Iter: 798/1305 Batch time: 1.0828\n",
            "Phase: train Epoch: 3/3 Iter: 799/1305 Batch time: 1.1003\n",
            "Phase: train Epoch: 3/3 Iter: 800/1305 Batch time: 1.2406\n",
            "Phase: train Epoch: 3/3 Iter: 801/1305 Batch time: 1.1068\n",
            "Phase: train Epoch: 3/3 Iter: 802/1305 Batch time: 1.1362\n",
            "Phase: train Epoch: 3/3 Iter: 803/1305 Batch time: 1.1048\n",
            "Phase: train Epoch: 3/3 Iter: 804/1305 Batch time: 1.0868\n",
            "Phase: train Epoch: 3/3 Iter: 805/1305 Batch time: 1.1608\n",
            "Phase: train Epoch: 3/3 Iter: 806/1305 Batch time: 1.5661\n",
            "Phase: train Epoch: 3/3 Iter: 807/1305 Batch time: 1.2421\n",
            "Phase: train Epoch: 3/3 Iter: 808/1305 Batch time: 1.1004\n",
            "Phase: train Epoch: 3/3 Iter: 809/1305 Batch time: 1.1133\n",
            "Phase: train Epoch: 3/3 Iter: 810/1305 Batch time: 1.0952\n",
            "Phase: train Epoch: 3/3 Iter: 811/1305 Batch time: 1.1241\n",
            "Phase: train Epoch: 3/3 Iter: 812/1305 Batch time: 1.2103\n",
            "Phase: train Epoch: 3/3 Iter: 813/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 814/1305 Batch time: 1.1153\n",
            "Phase: train Epoch: 3/3 Iter: 815/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 3/3 Iter: 816/1305 Batch time: 1.1035\n",
            "Phase: train Epoch: 3/3 Iter: 817/1305 Batch time: 1.1088\n",
            "Phase: train Epoch: 3/3 Iter: 818/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 3/3 Iter: 819/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 3/3 Iter: 820/1305 Batch time: 1.1182\n",
            "Phase: train Epoch: 3/3 Iter: 821/1305 Batch time: 1.1001\n",
            "Phase: train Epoch: 3/3 Iter: 822/1305 Batch time: 1.1991\n",
            "Phase: train Epoch: 3/3 Iter: 823/1305 Batch time: 1.1252\n",
            "Phase: train Epoch: 3/3 Iter: 824/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 3/3 Iter: 825/1305 Batch time: 1.1022\n",
            "Phase: train Epoch: 3/3 Iter: 826/1305 Batch time: 1.0979\n",
            "Phase: train Epoch: 3/3 Iter: 827/1305 Batch time: 1.1118\n",
            "Phase: train Epoch: 3/3 Iter: 828/1305 Batch time: 1.0910\n",
            "Phase: train Epoch: 3/3 Iter: 829/1305 Batch time: 1.1029\n",
            "Phase: train Epoch: 3/3 Iter: 830/1305 Batch time: 1.0996\n",
            "Phase: train Epoch: 3/3 Iter: 831/1305 Batch time: 1.1875\n",
            "Phase: train Epoch: 3/3 Iter: 832/1305 Batch time: 1.1069\n",
            "Phase: train Epoch: 3/3 Iter: 833/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 3/3 Iter: 834/1305 Batch time: 1.0945\n",
            "Phase: train Epoch: 3/3 Iter: 835/1305 Batch time: 1.0941\n",
            "Phase: train Epoch: 3/3 Iter: 836/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 3/3 Iter: 837/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 3/3 Iter: 838/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 3/3 Iter: 839/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 3/3 Iter: 840/1305 Batch time: 1.0955\n",
            "Phase: train Epoch: 3/3 Iter: 841/1305 Batch time: 1.0995\n",
            "Phase: train Epoch: 3/3 Iter: 842/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 3/3 Iter: 843/1305 Batch time: 1.0722\n",
            "Phase: train Epoch: 3/3 Iter: 844/1305 Batch time: 1.1162\n",
            "Phase: train Epoch: 3/3 Iter: 845/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 3/3 Iter: 846/1305 Batch time: 1.1979\n",
            "Phase: train Epoch: 3/3 Iter: 847/1305 Batch time: 1.0955\n",
            "Phase: train Epoch: 3/3 Iter: 848/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 3/3 Iter: 849/1305 Batch time: 1.0925\n",
            "Phase: train Epoch: 3/3 Iter: 850/1305 Batch time: 1.0885\n",
            "Phase: train Epoch: 3/3 Iter: 851/1305 Batch time: 1.1081\n",
            "Phase: train Epoch: 3/3 Iter: 852/1305 Batch time: 1.1008\n",
            "Phase: train Epoch: 3/3 Iter: 853/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 3/3 Iter: 854/1305 Batch time: 1.0993\n",
            "Phase: train Epoch: 3/3 Iter: 855/1305 Batch time: 1.1074\n",
            "Phase: train Epoch: 3/3 Iter: 856/1305 Batch time: 1.0926\n",
            "Phase: train Epoch: 3/3 Iter: 857/1305 Batch time: 1.1136\n",
            "Phase: train Epoch: 3/3 Iter: 858/1305 Batch time: 1.0920\n",
            "Phase: train Epoch: 3/3 Iter: 859/1305 Batch time: 1.2060\n",
            "Phase: train Epoch: 3/3 Iter: 860/1305 Batch time: 1.0917\n",
            "Phase: train Epoch: 3/3 Iter: 861/1305 Batch time: 1.0839\n",
            "Phase: train Epoch: 3/3 Iter: 862/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 3/3 Iter: 863/1305 Batch time: 1.0962\n",
            "Phase: train Epoch: 3/3 Iter: 864/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 3/3 Iter: 865/1305 Batch time: 1.0956\n",
            "Phase: train Epoch: 3/3 Iter: 866/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 3/3 Iter: 867/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 3/3 Iter: 868/1305 Batch time: 1.0798\n",
            "Phase: train Epoch: 3/3 Iter: 869/1305 Batch time: 1.1962\n",
            "Phase: train Epoch: 3/3 Iter: 870/1305 Batch time: 1.1097\n",
            "Phase: train Epoch: 3/3 Iter: 871/1305 Batch time: 1.0890\n",
            "Phase: train Epoch: 3/3 Iter: 872/1305 Batch time: 1.0894\n",
            "Phase: train Epoch: 3/3 Iter: 873/1305 Batch time: 1.1058\n",
            "Phase: train Epoch: 3/3 Iter: 874/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 3/3 Iter: 875/1305 Batch time: 1.0948\n",
            "Phase: train Epoch: 3/3 Iter: 876/1305 Batch time: 1.1270\n",
            "Phase: train Epoch: 3/3 Iter: 877/1305 Batch time: 1.0823\n",
            "Phase: train Epoch: 3/3 Iter: 878/1305 Batch time: 1.0920\n",
            "Phase: train Epoch: 3/3 Iter: 879/1305 Batch time: 1.2072\n",
            "Phase: train Epoch: 3/3 Iter: 880/1305 Batch time: 1.0857\n",
            "Phase: train Epoch: 3/3 Iter: 881/1305 Batch time: 1.0951\n",
            "Phase: train Epoch: 3/3 Iter: 882/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 883/1305 Batch time: 1.1037\n",
            "Phase: train Epoch: 3/3 Iter: 884/1305 Batch time: 1.1168\n",
            "Phase: train Epoch: 3/3 Iter: 885/1305 Batch time: 1.3231\n",
            "Phase: train Epoch: 3/3 Iter: 886/1305 Batch time: 1.5453\n",
            "Phase: train Epoch: 3/3 Iter: 887/1305 Batch time: 1.1781\n",
            "Phase: train Epoch: 3/3 Iter: 888/1305 Batch time: 1.1069\n",
            "Phase: train Epoch: 3/3 Iter: 889/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 3/3 Iter: 890/1305 Batch time: 1.0909\n",
            "Phase: train Epoch: 3/3 Iter: 891/1305 Batch time: 1.1055\n",
            "Phase: train Epoch: 3/3 Iter: 892/1305 Batch time: 1.1032\n",
            "Phase: train Epoch: 3/3 Iter: 893/1305 Batch time: 1.0893\n",
            "Phase: train Epoch: 3/3 Iter: 894/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 3/3 Iter: 895/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 3/3 Iter: 896/1305 Batch time: 1.1946\n",
            "Phase: train Epoch: 3/3 Iter: 897/1305 Batch time: 1.0947\n",
            "Phase: train Epoch: 3/3 Iter: 898/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 3/3 Iter: 899/1305 Batch time: 1.0794\n",
            "Phase: train Epoch: 3/3 Iter: 900/1305 Batch time: 1.0739\n",
            "Phase: train Epoch: 3/3 Iter: 901/1305 Batch time: 1.1245\n",
            "Phase: train Epoch: 3/3 Iter: 902/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 3/3 Iter: 903/1305 Batch time: 1.0860\n",
            "Phase: train Epoch: 3/3 Iter: 904/1305 Batch time: 1.0921\n",
            "Phase: train Epoch: 3/3 Iter: 905/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 3/3 Iter: 906/1305 Batch time: 1.0916\n",
            "Phase: train Epoch: 3/3 Iter: 907/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 3/3 Iter: 908/1305 Batch time: 1.0968\n",
            "Phase: train Epoch: 3/3 Iter: 909/1305 Batch time: 1.0950\n",
            "Phase: train Epoch: 3/3 Iter: 910/1305 Batch time: 1.0958\n",
            "Phase: train Epoch: 3/3 Iter: 911/1305 Batch time: 1.1040\n",
            "Phase: train Epoch: 3/3 Iter: 912/1305 Batch time: 1.0792\n",
            "Phase: train Epoch: 3/3 Iter: 913/1305 Batch time: 1.0879\n",
            "Phase: train Epoch: 3/3 Iter: 914/1305 Batch time: 1.0735\n",
            "Phase: train Epoch: 3/3 Iter: 915/1305 Batch time: 1.1970\n",
            "Phase: train Epoch: 3/3 Iter: 916/1305 Batch time: 1.0869\n",
            "Phase: train Epoch: 3/3 Iter: 917/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 3/3 Iter: 918/1305 Batch time: 1.0866\n",
            "Phase: train Epoch: 3/3 Iter: 919/1305 Batch time: 1.1217\n",
            "Phase: train Epoch: 3/3 Iter: 920/1305 Batch time: 1.1248\n",
            "Phase: train Epoch: 3/3 Iter: 921/1305 Batch time: 1.0804\n",
            "Phase: train Epoch: 3/3 Iter: 922/1305 Batch time: 1.0965\n",
            "Phase: train Epoch: 3/3 Iter: 923/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 924/1305 Batch time: 1.0826\n",
            "Phase: train Epoch: 3/3 Iter: 925/1305 Batch time: 1.0860\n",
            "Phase: train Epoch: 3/3 Iter: 926/1305 Batch time: 1.0902\n",
            "Phase: train Epoch: 3/3 Iter: 927/1305 Batch time: 1.1207\n",
            "Phase: train Epoch: 3/3 Iter: 928/1305 Batch time: 1.1077\n",
            "Phase: train Epoch: 3/3 Iter: 929/1305 Batch time: 1.0949\n",
            "Phase: train Epoch: 3/3 Iter: 930/1305 Batch time: 1.1938\n",
            "Phase: train Epoch: 3/3 Iter: 931/1305 Batch time: 1.0841\n",
            "Phase: train Epoch: 3/3 Iter: 932/1305 Batch time: 1.0912\n",
            "Phase: train Epoch: 3/3 Iter: 933/1305 Batch time: 1.0851\n",
            "Phase: train Epoch: 3/3 Iter: 934/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 3/3 Iter: 935/1305 Batch time: 1.0887\n",
            "Phase: train Epoch: 3/3 Iter: 936/1305 Batch time: 1.0978\n",
            "Phase: train Epoch: 3/3 Iter: 937/1305 Batch time: 1.0990\n",
            "Phase: train Epoch: 3/3 Iter: 938/1305 Batch time: 1.0919\n",
            "Phase: train Epoch: 3/3 Iter: 939/1305 Batch time: 1.1255\n",
            "Phase: train Epoch: 3/3 Iter: 940/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 3/3 Iter: 941/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 3/3 Iter: 942/1305 Batch time: 1.2131\n",
            "Phase: train Epoch: 3/3 Iter: 943/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 3/3 Iter: 944/1305 Batch time: 1.1049\n",
            "Phase: train Epoch: 3/3 Iter: 945/1305 Batch time: 1.1130\n",
            "Phase: train Epoch: 3/3 Iter: 946/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 3/3 Iter: 947/1305 Batch time: 1.0984\n",
            "Phase: train Epoch: 3/3 Iter: 948/1305 Batch time: 1.1084\n",
            "Phase: train Epoch: 3/3 Iter: 949/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 3/3 Iter: 950/1305 Batch time: 1.0996\n",
            "Phase: train Epoch: 3/3 Iter: 951/1305 Batch time: 1.1199\n",
            "Phase: train Epoch: 3/3 Iter: 952/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 953/1305 Batch time: 1.1028\n",
            "Phase: train Epoch: 3/3 Iter: 954/1305 Batch time: 1.2056\n",
            "Phase: train Epoch: 3/3 Iter: 955/1305 Batch time: 1.0915\n",
            "Phase: train Epoch: 3/3 Iter: 956/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 3/3 Iter: 957/1305 Batch time: 1.1000\n",
            "Phase: train Epoch: 3/3 Iter: 958/1305 Batch time: 1.0841\n",
            "Phase: train Epoch: 3/3 Iter: 959/1305 Batch time: 1.1062\n",
            "Phase: train Epoch: 3/3 Iter: 960/1305 Batch time: 1.0851\n",
            "Phase: train Epoch: 3/3 Iter: 961/1305 Batch time: 1.0881\n",
            "Phase: train Epoch: 3/3 Iter: 962/1305 Batch time: 1.0976\n",
            "Phase: train Epoch: 3/3 Iter: 963/1305 Batch time: 1.1200\n",
            "Phase: train Epoch: 3/3 Iter: 964/1305 Batch time: 1.2000\n",
            "Phase: train Epoch: 3/3 Iter: 965/1305 Batch time: 1.1047\n",
            "Phase: train Epoch: 3/3 Iter: 966/1305 Batch time: 1.4648\n",
            "Phase: train Epoch: 3/3 Iter: 967/1305 Batch time: 1.4319\n",
            "Phase: train Epoch: 3/3 Iter: 968/1305 Batch time: 1.0845\n",
            "Phase: train Epoch: 3/3 Iter: 969/1305 Batch time: 1.1027\n",
            "Phase: train Epoch: 3/3 Iter: 970/1305 Batch time: 1.1225\n",
            "Phase: train Epoch: 3/3 Iter: 971/1305 Batch time: 1.0913\n",
            "Phase: train Epoch: 3/3 Iter: 972/1305 Batch time: 1.0932\n",
            "Phase: train Epoch: 3/3 Iter: 973/1305 Batch time: 1.2083\n",
            "Phase: train Epoch: 3/3 Iter: 974/1305 Batch time: 1.0783\n",
            "Phase: train Epoch: 3/3 Iter: 975/1305 Batch time: 1.1091\n",
            "Phase: train Epoch: 3/3 Iter: 976/1305 Batch time: 1.0999\n",
            "Phase: train Epoch: 3/3 Iter: 977/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 3/3 Iter: 978/1305 Batch time: 1.0936\n",
            "Phase: train Epoch: 3/3 Iter: 979/1305 Batch time: 1.0875\n",
            "Phase: train Epoch: 3/3 Iter: 980/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 3/3 Iter: 981/1305 Batch time: 1.1117\n",
            "Phase: train Epoch: 3/3 Iter: 982/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 3/3 Iter: 983/1305 Batch time: 1.0852\n",
            "Phase: train Epoch: 3/3 Iter: 984/1305 Batch time: 1.0901\n",
            "Phase: train Epoch: 3/3 Iter: 985/1305 Batch time: 1.1011\n",
            "Phase: train Epoch: 3/3 Iter: 986/1305 Batch time: 1.0898\n",
            "Phase: train Epoch: 3/3 Iter: 987/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 3/3 Iter: 988/1305 Batch time: 1.2169\n",
            "Phase: train Epoch: 3/3 Iter: 989/1305 Batch time: 1.0977\n",
            "Phase: train Epoch: 3/3 Iter: 990/1305 Batch time: 1.1148\n",
            "Phase: train Epoch: 3/3 Iter: 991/1305 Batch time: 1.0935\n",
            "Phase: train Epoch: 3/3 Iter: 992/1305 Batch time: 1.0883\n",
            "Phase: train Epoch: 3/3 Iter: 993/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 3/3 Iter: 994/1305 Batch time: 1.1240\n",
            "Phase: train Epoch: 3/3 Iter: 995/1305 Batch time: 1.0903\n",
            "Phase: train Epoch: 3/3 Iter: 996/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 3/3 Iter: 997/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 998/1305 Batch time: 1.1077\n",
            "Phase: train Epoch: 3/3 Iter: 999/1305 Batch time: 1.0873\n",
            "Phase: train Epoch: 3/3 Iter: 1000/1305 Batch time: 1.1034\n",
            "Phase: train Epoch: 3/3 Iter: 1001/1305 Batch time: 1.2096\n",
            "Phase: train Epoch: 3/3 Iter: 1002/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 3/3 Iter: 1003/1305 Batch time: 1.1070\n",
            "Phase: train Epoch: 3/3 Iter: 1004/1305 Batch time: 1.1053\n",
            "Phase: train Epoch: 3/3 Iter: 1005/1305 Batch time: 1.1033\n",
            "Phase: train Epoch: 3/3 Iter: 1006/1305 Batch time: 1.1129\n",
            "Phase: train Epoch: 3/3 Iter: 1007/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 3/3 Iter: 1008/1305 Batch time: 1.1030\n",
            "Phase: train Epoch: 3/3 Iter: 1009/1305 Batch time: 1.1151\n",
            "Phase: train Epoch: 3/3 Iter: 1010/1305 Batch time: 1.1016\n",
            "Phase: train Epoch: 3/3 Iter: 1011/1305 Batch time: 1.2323\n",
            "Phase: train Epoch: 3/3 Iter: 1012/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 3/3 Iter: 1013/1305 Batch time: 1.0959\n",
            "Phase: train Epoch: 3/3 Iter: 1014/1305 Batch time: 1.1077\n",
            "Phase: train Epoch: 3/3 Iter: 1015/1305 Batch time: 1.1090\n",
            "Phase: train Epoch: 3/3 Iter: 1016/1305 Batch time: 1.0930\n",
            "Phase: train Epoch: 3/3 Iter: 1017/1305 Batch time: 1.1262\n",
            "Phase: train Epoch: 3/3 Iter: 1018/1305 Batch time: 1.1188\n",
            "Phase: train Epoch: 3/3 Iter: 1019/1305 Batch time: 1.0974\n",
            "Phase: train Epoch: 3/3 Iter: 1020/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 3/3 Iter: 1021/1305 Batch time: 1.1925\n",
            "Phase: train Epoch: 3/3 Iter: 1022/1305 Batch time: 1.0985\n",
            "Phase: train Epoch: 3/3 Iter: 1023/1305 Batch time: 1.1114\n",
            "Phase: train Epoch: 3/3 Iter: 1024/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 3/3 Iter: 1025/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 3/3 Iter: 1026/1305 Batch time: 1.1032\n",
            "Phase: train Epoch: 3/3 Iter: 1027/1305 Batch time: 1.1208\n",
            "Phase: train Epoch: 3/3 Iter: 1028/1305 Batch time: 1.1036\n",
            "Phase: train Epoch: 3/3 Iter: 1029/1305 Batch time: 1.1981\n",
            "Phase: train Epoch: 3/3 Iter: 1030/1305 Batch time: 1.1238\n",
            "Phase: train Epoch: 3/3 Iter: 1031/1305 Batch time: 1.0983\n",
            "Phase: train Epoch: 3/3 Iter: 1032/1305 Batch time: 1.0996\n",
            "Phase: train Epoch: 3/3 Iter: 1033/1305 Batch time: 1.0946\n",
            "Phase: train Epoch: 3/3 Iter: 1034/1305 Batch time: 1.0917\n",
            "Phase: train Epoch: 3/3 Iter: 1035/1305 Batch time: 1.1052\n",
            "Phase: train Epoch: 3/3 Iter: 1036/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 3/3 Iter: 1037/1305 Batch time: 1.0889\n",
            "Phase: train Epoch: 3/3 Iter: 1038/1305 Batch time: 1.2294\n",
            "Phase: train Epoch: 3/3 Iter: 1039/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 3/3 Iter: 1040/1305 Batch time: 1.0972\n",
            "Phase: train Epoch: 3/3 Iter: 1041/1305 Batch time: 1.1174\n",
            "Phase: train Epoch: 3/3 Iter: 1042/1305 Batch time: 1.0854\n",
            "Phase: train Epoch: 3/3 Iter: 1043/1305 Batch time: 1.0867\n",
            "Phase: train Epoch: 3/3 Iter: 1044/1305 Batch time: 1.1100\n",
            "Phase: train Epoch: 3/3 Iter: 1045/1305 Batch time: 1.1161\n",
            "Phase: train Epoch: 3/3 Iter: 1046/1305 Batch time: 1.5680\n",
            "Phase: train Epoch: 3/3 Iter: 1047/1305 Batch time: 1.3543\n",
            "Phase: train Epoch: 3/3 Iter: 1048/1305 Batch time: 1.0992\n",
            "Phase: train Epoch: 3/3 Iter: 1049/1305 Batch time: 1.0951\n",
            "Phase: train Epoch: 3/3 Iter: 1050/1305 Batch time: 1.0986\n",
            "Phase: train Epoch: 3/3 Iter: 1051/1305 Batch time: 1.1312\n",
            "Phase: train Epoch: 3/3 Iter: 1052/1305 Batch time: 1.1349\n",
            "Phase: train Epoch: 3/3 Iter: 1053/1305 Batch time: 1.1112\n",
            "Phase: train Epoch: 3/3 Iter: 1054/1305 Batch time: 1.1031\n",
            "Phase: train Epoch: 3/3 Iter: 1055/1305 Batch time: 1.1331\n",
            "Phase: train Epoch: 3/3 Iter: 1056/1305 Batch time: 1.1046\n",
            "Phase: train Epoch: 3/3 Iter: 1057/1305 Batch time: 1.2495\n",
            "Phase: train Epoch: 3/3 Iter: 1058/1305 Batch time: 1.1167\n",
            "Phase: train Epoch: 3/3 Iter: 1059/1305 Batch time: 1.1409\n",
            "Phase: train Epoch: 3/3 Iter: 1060/1305 Batch time: 1.1026\n",
            "Phase: train Epoch: 3/3 Iter: 1061/1305 Batch time: 1.0942\n",
            "Phase: train Epoch: 3/3 Iter: 1062/1305 Batch time: 1.0939\n",
            "Phase: train Epoch: 3/3 Iter: 1063/1305 Batch time: 1.1198\n",
            "Phase: train Epoch: 3/3 Iter: 1064/1305 Batch time: 1.1082\n",
            "Phase: train Epoch: 3/3 Iter: 1065/1305 Batch time: 1.0997\n",
            "Phase: train Epoch: 3/3 Iter: 1066/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 3/3 Iter: 1067/1305 Batch time: 1.1143\n",
            "Phase: train Epoch: 3/3 Iter: 1068/1305 Batch time: 1.1196\n",
            "Phase: train Epoch: 3/3 Iter: 1069/1305 Batch time: 1.1343\n",
            "Phase: train Epoch: 3/3 Iter: 1070/1305 Batch time: 1.1078\n",
            "Phase: train Epoch: 3/3 Iter: 1071/1305 Batch time: 1.1010\n",
            "Phase: train Epoch: 3/3 Iter: 1072/1305 Batch time: 1.2387\n",
            "Phase: train Epoch: 3/3 Iter: 1073/1305 Batch time: 1.1115\n",
            "Phase: train Epoch: 3/3 Iter: 1074/1305 Batch time: 1.0848\n",
            "Phase: train Epoch: 3/3 Iter: 1075/1305 Batch time: 1.1094\n",
            "Phase: train Epoch: 3/3 Iter: 1076/1305 Batch time: 1.1089\n",
            "Phase: train Epoch: 3/3 Iter: 1077/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 3/3 Iter: 1078/1305 Batch time: 1.1121\n",
            "Phase: train Epoch: 3/3 Iter: 1079/1305 Batch time: 1.1006\n",
            "Phase: train Epoch: 3/3 Iter: 1080/1305 Batch time: 1.1230\n",
            "Phase: train Epoch: 3/3 Iter: 1081/1305 Batch time: 1.1101\n",
            "Phase: train Epoch: 3/3 Iter: 1082/1305 Batch time: 1.1246\n",
            "Phase: train Epoch: 3/3 Iter: 1083/1305 Batch time: 1.1253\n",
            "Phase: train Epoch: 3/3 Iter: 1084/1305 Batch time: 1.2347\n",
            "Phase: train Epoch: 3/3 Iter: 1085/1305 Batch time: 1.1272\n",
            "Phase: train Epoch: 3/3 Iter: 1086/1305 Batch time: 1.1080\n",
            "Phase: train Epoch: 3/3 Iter: 1087/1305 Batch time: 1.1209\n",
            "Phase: train Epoch: 3/3 Iter: 1088/1305 Batch time: 1.1128\n",
            "Phase: train Epoch: 3/3 Iter: 1089/1305 Batch time: 1.0991\n",
            "Phase: train Epoch: 3/3 Iter: 1090/1305 Batch time: 1.1116\n",
            "Phase: train Epoch: 3/3 Iter: 1091/1305 Batch time: 1.1092\n",
            "Phase: train Epoch: 3/3 Iter: 1092/1305 Batch time: 1.0971\n",
            "Phase: train Epoch: 3/3 Iter: 1093/1305 Batch time: 1.1004\n",
            "Phase: train Epoch: 3/3 Iter: 1094/1305 Batch time: 1.1145\n",
            "Phase: train Epoch: 3/3 Iter: 1095/1305 Batch time: 1.1056\n",
            "Phase: train Epoch: 3/3 Iter: 1096/1305 Batch time: 1.2118\n",
            "Phase: train Epoch: 3/3 Iter: 1097/1305 Batch time: 1.0998\n",
            "Phase: train Epoch: 3/3 Iter: 1098/1305 Batch time: 1.1044\n",
            "Phase: train Epoch: 3/3 Iter: 1099/1305 Batch time: 1.1075\n",
            "Phase: train Epoch: 3/3 Iter: 1100/1305 Batch time: 1.0988\n",
            "Phase: train Epoch: 3/3 Iter: 1101/1305 Batch time: 1.0903\n",
            "Phase: train Epoch: 3/3 Iter: 1102/1305 Batch time: 1.1213\n",
            "Phase: train Epoch: 3/3 Iter: 1103/1305 Batch time: 1.1066\n",
            "Phase: train Epoch: 3/3 Iter: 1104/1305 Batch time: 1.1050\n",
            "Phase: train Epoch: 3/3 Iter: 1105/1305 Batch time: 1.1096\n",
            "Phase: train Epoch: 3/3 Iter: 1106/1305 Batch time: 1.2057\n",
            "Phase: train Epoch: 3/3 Iter: 1107/1305 Batch time: 1.1079\n",
            "Phase: train Epoch: 3/3 Iter: 1108/1305 Batch time: 1.1061\n",
            "Phase: train Epoch: 3/3 Iter: 1109/1305 Batch time: 1.1147\n",
            "Phase: train Epoch: 3/3 Iter: 1110/1305 Batch time: 1.1045\n",
            "Phase: train Epoch: 3/3 Iter: 1111/1305 Batch time: 1.1087\n",
            "Phase: train Epoch: 3/3 Iter: 1112/1305 Batch time: 1.1059\n",
            "Phase: train Epoch: 3/3 Iter: 1113/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 3/3 Iter: 1114/1305 Batch time: 1.1152\n",
            "Phase: train Epoch: 3/3 Iter: 1115/1305 Batch time: 1.2112\n",
            "Phase: train Epoch: 3/3 Iter: 1116/1305 Batch time: 1.1042\n",
            "Phase: train Epoch: 3/3 Iter: 1117/1305 Batch time: 1.1261\n",
            "Phase: train Epoch: 3/3 Iter: 1118/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 3/3 Iter: 1119/1305 Batch time: 1.1300\n",
            "Phase: train Epoch: 3/3 Iter: 1120/1305 Batch time: 1.1186\n",
            "Phase: train Epoch: 3/3 Iter: 1121/1305 Batch time: 1.1029\n",
            "Phase: train Epoch: 3/3 Iter: 1122/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 3/3 Iter: 1123/1305 Batch time: 1.1184\n",
            "Phase: train Epoch: 3/3 Iter: 1124/1305 Batch time: 1.1041\n",
            "Phase: train Epoch: 3/3 Iter: 1125/1305 Batch time: 1.0987\n",
            "Phase: train Epoch: 3/3 Iter: 1126/1305 Batch time: 1.1549\n",
            "Phase: train Epoch: 3/3 Iter: 1127/1305 Batch time: 1.5370\n",
            "Phase: train Epoch: 3/3 Iter: 1128/1305 Batch time: 1.2908\n",
            "Phase: train Epoch: 3/3 Iter: 1129/1305 Batch time: 1.0877\n",
            "Phase: train Epoch: 3/3 Iter: 1130/1305 Batch time: 1.2412\n",
            "Phase: train Epoch: 3/3 Iter: 1131/1305 Batch time: 1.1086\n",
            "Phase: train Epoch: 3/3 Iter: 1132/1305 Batch time: 1.1276\n",
            "Phase: train Epoch: 3/3 Iter: 1133/1305 Batch time: 1.1489\n",
            "Phase: train Epoch: 3/3 Iter: 1134/1305 Batch time: 1.1304\n",
            "Phase: train Epoch: 3/3 Iter: 1135/1305 Batch time: 1.1250\n",
            "Phase: train Epoch: 3/3 Iter: 1136/1305 Batch time: 1.1448\n",
            "Phase: train Epoch: 3/3 Iter: 1137/1305 Batch time: 1.1269\n",
            "Phase: train Epoch: 3/3 Iter: 1138/1305 Batch time: 1.1442\n",
            "Phase: train Epoch: 3/3 Iter: 1139/1305 Batch time: 1.1302\n",
            "Phase: train Epoch: 3/3 Iter: 1140/1305 Batch time: 1.1217\n",
            "Phase: train Epoch: 3/3 Iter: 1141/1305 Batch time: 1.1376\n",
            "Phase: train Epoch: 3/3 Iter: 1142/1305 Batch time: 1.1473\n",
            "Phase: train Epoch: 3/3 Iter: 1143/1305 Batch time: 1.3287\n",
            "Phase: train Epoch: 3/3 Iter: 1144/1305 Batch time: 1.2227\n",
            "Phase: train Epoch: 3/3 Iter: 1145/1305 Batch time: 1.2253\n",
            "Phase: train Epoch: 3/3 Iter: 1146/1305 Batch time: 1.2128\n",
            "Phase: train Epoch: 3/3 Iter: 1147/1305 Batch time: 1.2372\n",
            "Phase: train Epoch: 3/3 Iter: 1148/1305 Batch time: 1.2406\n",
            "Phase: train Epoch: 3/3 Iter: 1149/1305 Batch time: 1.2203\n",
            "Phase: train Epoch: 3/3 Iter: 1150/1305 Batch time: 1.2021\n",
            "Phase: train Epoch: 3/3 Iter: 1151/1305 Batch time: 1.1482\n",
            "Phase: train Epoch: 3/3 Iter: 1152/1305 Batch time: 1.1385\n",
            "Phase: train Epoch: 3/3 Iter: 1153/1305 Batch time: 1.2571\n",
            "Phase: train Epoch: 3/3 Iter: 1154/1305 Batch time: 1.1668\n",
            "Phase: train Epoch: 3/3 Iter: 1155/1305 Batch time: 1.1579\n",
            "Phase: train Epoch: 3/3 Iter: 1156/1305 Batch time: 1.1409\n",
            "Phase: train Epoch: 3/3 Iter: 1157/1305 Batch time: 1.1344\n",
            "Phase: train Epoch: 3/3 Iter: 1158/1305 Batch time: 1.1340\n",
            "Phase: train Epoch: 3/3 Iter: 1159/1305 Batch time: 1.1321\n",
            "Phase: train Epoch: 3/3 Iter: 1160/1305 Batch time: 1.1104\n",
            "Phase: train Epoch: 3/3 Iter: 1161/1305 Batch time: 1.1472\n",
            "Phase: train Epoch: 3/3 Iter: 1162/1305 Batch time: 1.1489\n",
            "Phase: train Epoch: 3/3 Iter: 1163/1305 Batch time: 1.2615\n",
            "Phase: train Epoch: 3/3 Iter: 1164/1305 Batch time: 1.1354\n",
            "Phase: train Epoch: 3/3 Iter: 1165/1305 Batch time: 1.1532\n",
            "Phase: train Epoch: 3/3 Iter: 1166/1305 Batch time: 1.1579\n",
            "Phase: train Epoch: 3/3 Iter: 1167/1305 Batch time: 1.1534\n",
            "Phase: train Epoch: 3/3 Iter: 1168/1305 Batch time: 1.1452\n",
            "Phase: train Epoch: 3/3 Iter: 1169/1305 Batch time: 1.1326\n",
            "Phase: train Epoch: 3/3 Iter: 1170/1305 Batch time: 1.1570\n",
            "Phase: train Epoch: 3/3 Iter: 1171/1305 Batch time: 1.2579\n",
            "Phase: train Epoch: 3/3 Iter: 1172/1305 Batch time: 1.1624\n",
            "Phase: train Epoch: 3/3 Iter: 1173/1305 Batch time: 1.1331\n",
            "Phase: train Epoch: 3/3 Iter: 1174/1305 Batch time: 1.1431\n",
            "Phase: train Epoch: 3/3 Iter: 1175/1305 Batch time: 1.1508\n",
            "Phase: train Epoch: 3/3 Iter: 1176/1305 Batch time: 1.1295\n",
            "Phase: train Epoch: 3/3 Iter: 1177/1305 Batch time: 1.1326\n",
            "Phase: train Epoch: 3/3 Iter: 1178/1305 Batch time: 1.1356\n",
            "Phase: train Epoch: 3/3 Iter: 1179/1305 Batch time: 1.1660\n",
            "Phase: train Epoch: 3/3 Iter: 1180/1305 Batch time: 1.2562\n",
            "Phase: train Epoch: 3/3 Iter: 1181/1305 Batch time: 1.1515\n",
            "Phase: train Epoch: 3/3 Iter: 1182/1305 Batch time: 1.1494\n",
            "Phase: train Epoch: 3/3 Iter: 1183/1305 Batch time: 1.1462\n",
            "Phase: train Epoch: 3/3 Iter: 1184/1305 Batch time: 1.1450\n",
            "Phase: train Epoch: 3/3 Iter: 1185/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 3/3 Iter: 1186/1305 Batch time: 1.1333\n",
            "Phase: train Epoch: 3/3 Iter: 1187/1305 Batch time: 1.1401\n",
            "Phase: train Epoch: 3/3 Iter: 1188/1305 Batch time: 1.1376\n",
            "Phase: train Epoch: 3/3 Iter: 1189/1305 Batch time: 1.1185\n",
            "Phase: train Epoch: 3/3 Iter: 1190/1305 Batch time: 1.1416\n",
            "Phase: train Epoch: 3/3 Iter: 1191/1305 Batch time: 1.1410\n",
            "Phase: train Epoch: 3/3 Iter: 1192/1305 Batch time: 1.1441\n",
            "Phase: train Epoch: 3/3 Iter: 1193/1305 Batch time: 1.1290\n",
            "Phase: train Epoch: 3/3 Iter: 1194/1305 Batch time: 1.1517\n",
            "Phase: train Epoch: 3/3 Iter: 1195/1305 Batch time: 1.1770\n",
            "Phase: train Epoch: 3/3 Iter: 1196/1305 Batch time: 1.1488\n",
            "Phase: train Epoch: 3/3 Iter: 1197/1305 Batch time: 1.1339\n",
            "Phase: train Epoch: 3/3 Iter: 1198/1305 Batch time: 1.1415\n",
            "Phase: train Epoch: 3/3 Iter: 1199/1305 Batch time: 1.2631\n",
            "Phase: train Epoch: 3/3 Iter: 1200/1305 Batch time: 1.1467\n",
            "Phase: train Epoch: 3/3 Iter: 1201/1305 Batch time: 1.1471\n",
            "Phase: train Epoch: 3/3 Iter: 1202/1305 Batch time: 1.1543\n",
            "Phase: train Epoch: 3/3 Iter: 1203/1305 Batch time: 1.4972\n",
            "Phase: train Epoch: 3/3 Iter: 1204/1305 Batch time: 1.5707\n",
            "Phase: train Epoch: 3/3 Iter: 1205/1305 Batch time: 1.1515\n",
            "Phase: train Epoch: 3/3 Iter: 1206/1305 Batch time: 1.1497\n",
            "Phase: train Epoch: 3/3 Iter: 1207/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 3/3 Iter: 1208/1305 Batch time: 1.1501\n",
            "Phase: train Epoch: 3/3 Iter: 1209/1305 Batch time: 1.1487\n",
            "Phase: train Epoch: 3/3 Iter: 1210/1305 Batch time: 1.1388\n",
            "Phase: train Epoch: 3/3 Iter: 1211/1305 Batch time: 1.1488\n",
            "Phase: train Epoch: 3/3 Iter: 1212/1305 Batch time: 1.1164\n",
            "Phase: train Epoch: 3/3 Iter: 1213/1305 Batch time: 1.1635\n",
            "Phase: train Epoch: 3/3 Iter: 1214/1305 Batch time: 1.2854\n",
            "Phase: train Epoch: 3/3 Iter: 1215/1305 Batch time: 1.1401\n",
            "Phase: train Epoch: 3/3 Iter: 1216/1305 Batch time: 1.1430\n",
            "Phase: train Epoch: 3/3 Iter: 1217/1305 Batch time: 1.1495\n",
            "Phase: train Epoch: 3/3 Iter: 1218/1305 Batch time: 1.1625\n",
            "Phase: train Epoch: 3/3 Iter: 1219/1305 Batch time: 1.1480\n",
            "Phase: train Epoch: 3/3 Iter: 1220/1305 Batch time: 1.1465\n",
            "Phase: train Epoch: 3/3 Iter: 1221/1305 Batch time: 1.1484\n",
            "Phase: train Epoch: 3/3 Iter: 1222/1305 Batch time: 1.1508\n",
            "Phase: train Epoch: 3/3 Iter: 1223/1305 Batch time: 1.1408\n",
            "Phase: train Epoch: 3/3 Iter: 1224/1305 Batch time: 1.1266\n",
            "Phase: train Epoch: 3/3 Iter: 1225/1305 Batch time: 1.1383\n",
            "Phase: train Epoch: 3/3 Iter: 1226/1305 Batch time: 1.2739\n",
            "Phase: train Epoch: 3/3 Iter: 1227/1305 Batch time: 1.1842\n",
            "Phase: train Epoch: 3/3 Iter: 1228/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 3/3 Iter: 1229/1305 Batch time: 1.1652\n",
            "Phase: train Epoch: 3/3 Iter: 1230/1305 Batch time: 1.1435\n",
            "Phase: train Epoch: 3/3 Iter: 1231/1305 Batch time: 1.1485\n",
            "Phase: train Epoch: 3/3 Iter: 1232/1305 Batch time: 1.1518\n",
            "Phase: train Epoch: 3/3 Iter: 1233/1305 Batch time: 1.1350\n",
            "Phase: train Epoch: 3/3 Iter: 1234/1305 Batch time: 1.1434\n",
            "Phase: train Epoch: 3/3 Iter: 1235/1305 Batch time: 1.1396\n",
            "Phase: train Epoch: 3/3 Iter: 1236/1305 Batch time: 1.1413\n",
            "Phase: train Epoch: 3/3 Iter: 1237/1305 Batch time: 1.1502\n",
            "Phase: train Epoch: 3/3 Iter: 1238/1305 Batch time: 1.2650\n",
            "Phase: train Epoch: 3/3 Iter: 1239/1305 Batch time: 1.1623\n",
            "Phase: train Epoch: 3/3 Iter: 1240/1305 Batch time: 1.1481\n",
            "Phase: train Epoch: 3/3 Iter: 1241/1305 Batch time: 1.1406\n",
            "Phase: train Epoch: 3/3 Iter: 1242/1305 Batch time: 1.1608\n",
            "Phase: train Epoch: 3/3 Iter: 1243/1305 Batch time: 1.1347\n",
            "Phase: train Epoch: 3/3 Iter: 1244/1305 Batch time: 1.1178\n",
            "Phase: train Epoch: 3/3 Iter: 1245/1305 Batch time: 1.1507\n",
            "Phase: train Epoch: 3/3 Iter: 1246/1305 Batch time: 1.1377\n",
            "Phase: train Epoch: 3/3 Iter: 1247/1305 Batch time: 1.1537\n",
            "Phase: train Epoch: 3/3 Iter: 1248/1305 Batch time: 1.2462\n",
            "Phase: train Epoch: 3/3 Iter: 1249/1305 Batch time: 1.1688\n",
            "Phase: train Epoch: 3/3 Iter: 1250/1305 Batch time: 1.1605\n",
            "Phase: train Epoch: 3/3 Iter: 1251/1305 Batch time: 1.1742\n",
            "Phase: train Epoch: 3/3 Iter: 1252/1305 Batch time: 1.1612\n",
            "Phase: train Epoch: 3/3 Iter: 1253/1305 Batch time: 1.1529\n",
            "Phase: train Epoch: 3/3 Iter: 1254/1305 Batch time: 1.1382\n",
            "Phase: train Epoch: 3/3 Iter: 1255/1305 Batch time: 1.1561\n",
            "Phase: train Epoch: 3/3 Iter: 1256/1305 Batch time: 1.1659\n",
            "Phase: train Epoch: 3/3 Iter: 1257/1305 Batch time: 1.2501\n",
            "Phase: train Epoch: 3/3 Iter: 1258/1305 Batch time: 1.1569\n",
            "Phase: train Epoch: 3/3 Iter: 1259/1305 Batch time: 1.1547\n",
            "Phase: train Epoch: 3/3 Iter: 1260/1305 Batch time: 1.1496\n",
            "Phase: train Epoch: 3/3 Iter: 1261/1305 Batch time: 1.1363\n",
            "Phase: train Epoch: 3/3 Iter: 1262/1305 Batch time: 1.1492\n",
            "Phase: train Epoch: 3/3 Iter: 1263/1305 Batch time: 1.1685\n",
            "Phase: train Epoch: 3/3 Iter: 1264/1305 Batch time: 1.1416\n",
            "Phase: train Epoch: 3/3 Iter: 1265/1305 Batch time: 1.1395\n",
            "Phase: train Epoch: 3/3 Iter: 1266/1305 Batch time: 1.1623\n",
            "Phase: train Epoch: 3/3 Iter: 1267/1305 Batch time: 1.1582\n",
            "Phase: train Epoch: 3/3 Iter: 1268/1305 Batch time: 1.1564\n",
            "Phase: train Epoch: 3/3 Iter: 1269/1305 Batch time: 1.1344\n",
            "Phase: train Epoch: 3/3 Iter: 1270/1305 Batch time: 1.1634\n",
            "Phase: train Epoch: 3/3 Iter: 1271/1305 Batch time: 1.1593\n",
            "Phase: train Epoch: 3/3 Iter: 1272/1305 Batch time: 1.2758\n",
            "Phase: train Epoch: 3/3 Iter: 1273/1305 Batch time: 1.1606\n",
            "Phase: train Epoch: 3/3 Iter: 1274/1305 Batch time: 1.1673\n",
            "Phase: train Epoch: 3/3 Iter: 1275/1305 Batch time: 1.1777\n",
            "Phase: train Epoch: 3/3 Iter: 1276/1305 Batch time: 1.1825\n",
            "Phase: train Epoch: 3/3 Iter: 1277/1305 Batch time: 1.1487\n",
            "Phase: train Epoch: 3/3 Iter: 1278/1305 Batch time: 1.1506\n",
            "Phase: train Epoch: 3/3 Iter: 1279/1305 Batch time: 1.3846\n",
            "Phase: train Epoch: 3/3 Iter: 1280/1305 Batch time: 1.6172\n",
            "Phase: train Epoch: 3/3 Iter: 1281/1305 Batch time: 1.1510\n",
            "Phase: train Epoch: 3/3 Iter: 1282/1305 Batch time: 1.1519\n",
            "Phase: train Epoch: 3/3 Iter: 1283/1305 Batch time: 1.1257\n",
            "Phase: train Epoch: 3/3 Iter: 1284/1305 Batch time: 1.1544\n",
            "Phase: train Epoch: 3/3 Iter: 1285/1305 Batch time: 1.2838\n",
            "Phase: train Epoch: 3/3 Iter: 1286/1305 Batch time: 1.1633\n",
            "Phase: train Epoch: 3/3 Iter: 1287/1305 Batch time: 1.1457\n",
            "Phase: train Epoch: 3/3 Iter: 1288/1305 Batch time: 1.1496\n",
            "Phase: train Epoch: 3/3 Iter: 1289/1305 Batch time: 1.1603\n",
            "Phase: train Epoch: 3/3 Iter: 1290/1305 Batch time: 1.1539\n",
            "Phase: train Epoch: 3/3 Iter: 1291/1305 Batch time: 1.1631\n",
            "Phase: train Epoch: 3/3 Iter: 1292/1305 Batch time: 1.1451\n",
            "Phase: train Epoch: 3/3 Iter: 1293/1305 Batch time: 1.1620\n",
            "Phase: train Epoch: 3/3 Iter: 1294/1305 Batch time: 1.1489\n",
            "Phase: train Epoch: 3/3 Iter: 1295/1305 Batch time: 1.2712\n",
            "Phase: train Epoch: 3/3 Iter: 1296/1305 Batch time: 1.1522\n",
            "Phase: train Epoch: 3/3 Iter: 1297/1305 Batch time: 1.1595\n",
            "Phase: train Epoch: 3/3 Iter: 1298/1305 Batch time: 1.1488\n",
            "Phase: train Epoch: 3/3 Iter: 1299/1305 Batch time: 1.1579\n",
            "Phase: train Epoch: 3/3 Iter: 1300/1305 Batch time: 1.1663\n",
            "Phase: train Epoch: 3/3 Iter: 1301/1305 Batch time: 1.1554\n",
            "Phase: train Epoch: 3/3 Iter: 1302/1305 Batch time: 1.1519\n",
            "Phase: train Epoch: 3/3 Iter: 1303/1305 Batch time: 1.1728\n",
            "Phase: train Epoch: 3/3 Iter: 1304/1305 Batch time: 1.1538\n",
            "Phase: train Epoch: 3/3 Loss: 0.2751 Acc: 0.8955        \n",
            "Phase: val Epoch: 3/3 Iter: 1/5 Batch time: 0.7656\n",
            "Phase: val Epoch: 3/3 Iter: 2/5 Batch time: 0.7401\n",
            "Phase: val Epoch: 3/3 Iter: 3/5 Batch time: 0.8359\n",
            "Phase: val Epoch: 3/3 Iter: 4/5 Batch time: 0.7400\n",
            "Phase: val Epoch: 3/3 Loss: 0.6276 Acc: 0.6875        \n",
            "Training completed in 95m 17s\n",
            "Best test loss: 0.6276 | Best test accuracy: 0.6875\n"
          ]
        }
      ],
      "source": [
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer_hybrid = optim.Adam(model_hybrid.fc.parameters(), lr=step)\n",
        "exp_lr_scheduler = lr_scheduler.StepLR(\n",
        "    optimizer_hybrid, step_size=10, gamma=gamma_lr_scheduler\n",
        ")\n",
        "def train_model(model, criterion, optimizer, scheduler, num_epochs):\n",
        "    since = time.time()\n",
        "    best_model_wts = copy.deepcopy(model.state_dict())\n",
        "    best_acc = 0.0\n",
        "    best_loss = 10000.0  # Large arbitrary number\n",
        "    best_acc_train = 0.0\n",
        "    best_loss_train = 10000.0  # Large arbitrary number\n",
        "    print(\"Training started:\")\n",
        "\n",
        "    for epoch in range(num_epochs):\n",
        "\n",
        "        # Each epoch has a training and validation phase\n",
        "        for phase in ['train', 'val']:\n",
        "            if phase == 'train':\n",
        "                # Set model to training mode\n",
        "                model.train()\n",
        "            else:\n",
        "                # Set model to evaluate mode\n",
        "                model.eval()\n",
        "            running_loss = 0.0\n",
        "            running_corrects = 0\n",
        "\n",
        "            # Iterate over data.\n",
        "\n",
        "            n_batches = dataset_sizes[phase] // batch_size\n",
        "            it = 0\n",
        "            for inputs, labels in dataloaders[phase]:\n",
        "                \n",
        "                since_batch = time.time()\n",
        "                batch_size_ = len(inputs)\n",
        "                inputs = inputs.to(device)\n",
        "                labels = labels.to(device)\n",
        "                optimizer.zero_grad()\n",
        "                \n",
        "                # Track/compute gradient and make an optimization step only when training\n",
        "                with torch.set_grad_enabled(phase == 'train'):\n",
        "                    outputs = model(inputs)\n",
        "                    _, preds = torch.max(outputs, 1)\n",
        "                    \n",
        "                    loss = criterion(outputs, labels)\n",
        "                    if phase == 'train':\n",
        "                        loss.backward()\n",
        "                        optimizer.step()\n",
        "                \n",
        "                # Print iteration results\n",
        "                running_loss += loss.item() * batch_size_\n",
        "                batch_corrects = torch.sum(preds == labels.data).item()\n",
        "                running_corrects += batch_corrects\n",
        "                \n",
        "                print(f\"Phase: {phase} Epoch: {epoch + 1}/{num_epochs} Iter: {it + 1}/{n_batches + 1} Batch time: {time.time() - since_batch:.4f}\")\n",
        "                it += 1\n",
        "                \n",
        "            # Print epoch results\n",
        "            epoch_loss = running_loss / dataset_sizes[phase]\n",
        "            epoch_acc = running_corrects / dataset_sizes[phase]\n",
        "            print(\n",
        "                \"Phase: {} Epoch: {}/{} Loss: {:.4f} Acc: {:.4f}        \".format(\n",
        "                    'train' if phase == 'train' else 'val',\n",
        "                    epoch + 1,\n",
        "                    num_epochs,\n",
        "                    epoch_loss,\n",
        "                    epoch_acc,\n",
        "                )\n",
        "            )\n",
        "\n",
        "            # Check if this is the best model wrt previous epochs\n",
        "            if phase == 'val' and epoch_acc > best_acc:\n",
        "                best_acc = epoch_acc\n",
        "                best_model_wts = copy.deepcopy(model.state_dict())\n",
        "            if phase == 'val' and epoch_loss < best_loss:\n",
        "                best_loss = epoch_loss\n",
        "            if phase == 'train' and epoch_acc > best_acc_train:\n",
        "                best_acc_train = epoch_acc\n",
        "            if phase == 'train' and epoch_loss < best_loss_train:\n",
        "                best_loss_train = epoch_loss\n",
        "\n",
        "            # Update learning rate\n",
        "            if phase == 'train':\n",
        "                scheduler.step()\n",
        "\n",
        "    # Print final results\n",
        "    model.load_state_dict(best_model_wts)\n",
        "    time_elapsed = time.time() - since\n",
        "    print(\n",
        "        \"Training completed in {:.0f}m {:.0f}s\".format(time_elapsed // 60, time_elapsed % 60)\n",
        "    )\n",
        "    print(\"Best test loss: {:.4f} | Best test accuracy: {:.4f}\".format(best_loss, best_acc))\n",
        "    return model\n",
        "model_hybrid = train_model(\n",
        "    model_hybrid, criterion, optimizer_hybrid, exp_lr_scheduler, num_epochs=num_epochs\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "IZn5E4eX72hi"
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}