Running_MixText_v3.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Running MixText v3.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/abhinivesh-s/Semi-Supervised-Text/blob/main/Running_MixText_v3.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Kmu-qOQP7ziG",
        "outputId": "74abe989-8870-40c7-f97f-c0062af29fce"
      },
      "source": [
        "# memory footprint support libraries/code\n",
        "!ln -sf /opt/bin/nvidia-smi /usr/bin/nvidia-smi\n",
        "!pip install gputil\n",
        "!pip install psutil\n",
        "!pip install humanize\n",
        "import psutil\n",
        "import humanize\n",
        "import os\n",
        "import GPUtil as GPU\n",
        "GPUs = GPU.getGPUs()\n",
        "# XXX: only one GPU on Colab and isn’t guaranteed\n",
        "gpu = GPUs[0]\n",
        "def printm():\n",
        " process = psutil.Process(os.getpid())\n",
        " print(\"Gen RAM Free: \" + humanize.naturalsize( psutil.virtual_memory().available ), \" | Proc size: \" + humanize.naturalsize( process.memory_info().rss))\n",
        " print(\"GPU RAM Free: {0:.0f}MB | Used: {1:.0f}MB | Util {2:3.0f}% | Total {3:.0f}MB\".format(gpu.memoryFree, gpu.memoryUsed, gpu.memoryUtil*100, gpu.memoryTotal))\n",
        "printm()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting gputil\n",
            "  Downloading https://files.pythonhosted.org/packages/ed/0e/5c61eedde9f6c87713e89d794f01e378cfd9565847d4576fa627d758c554/GPUtil-1.4.0.tar.gz\n",
            "Building wheels for collected packages: gputil\n",
            "  Building wheel for gputil (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for gputil: filename=GPUtil-1.4.0-cp37-none-any.whl size=7411 sha256=c6a0cd28a8e1ffef2fbaeadae22562164492323abd37b8bcc06eb7b6c147d79a\n",
            "  Stored in directory: /root/.cache/pip/wheels/3d/77/07/80562de4bb0786e5ea186911a2c831fdd0018bda69beab71fd\n",
            "Successfully built gputil\n",
            "Installing collected packages: gputil\n",
            "Successfully installed gputil-1.4.0\n",
            "Requirement already satisfied: psutil in /usr/local/lib/python3.7/dist-packages (5.4.8)\n",
            "Requirement already satisfied: humanize in /usr/local/lib/python3.7/dist-packages (0.5.1)\n",
            "Gen RAM Free: 26.3 GB  | Proc size: 118.1 MB\n",
            "GPU RAM Free: 16160MB | Used: 0MB | Util   0% | Total 16160MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 36
        },
        "id": "Ib7r9edVYwkD",
        "outputId": "945fdacb-d39d-4a0f-8406-13e7a530ba35"
      },
      "source": [
        "import torch\n",
        "torch.cuda.get_device_name()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Tesla V100-SXM2-16GB'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3r92HqyYznae",
        "outputId": "73a1a794-8eb6-4a55-e4cf-2be0b734fd0c"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UvE2i9nrz1V4",
        "outputId": "81ca40cb-34fb-4a13-cbe2-9dc22355ed64"
      },
      "source": [
        "%cd \"/content/drive/My Drive/Colab Notebooks/MixText-master\""
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/My Drive/Colab Notebooks/MixText-master\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zskiPUVM0Tck",
        "outputId": "91874f2b-b13e-4b35-f03d-776357ce09da"
      },
      "source": [
        "!ls"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            " code   data   LICENSE\t README.md  'Running MixText v2.ipynb'\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gextFx2T0ZKH",
        "outputId": "5b397176-7e70-49cb-b437-b1186feb0751"
      },
      "source": [
        "!pip install pytorch-transformers"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting pytorch-transformers\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/a3/b7/d3d18008a67e0b968d1ab93ad444fc05699403fa662f634b2f2c318a508b/pytorch_transformers-1.2.0-py3-none-any.whl (176kB)\n",
            "\u001b[K     |████████████████████████████████| 184kB 5.8MB/s \n",
            "\u001b[?25hRequirement already satisfied: torch>=1.0.0 in /usr/local/lib/python3.7/dist-packages (from pytorch-transformers) (1.8.1+cu101)\n",
            "Requirement already satisfied: regex in /usr/local/lib/python3.7/dist-packages (from pytorch-transformers) (2019.12.20)\n",
            "Collecting boto3\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/f1/99/43e5571005c792284276986eabd956699fac65d283df409b1482ca8722d8/boto3-1.17.67-py2.py3-none-any.whl (131kB)\n",
            "\u001b[K     |████████████████████████████████| 133kB 38.8MB/s \n",
            "\u001b[?25hRequirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from pytorch-transformers) (2.23.0)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from pytorch-transformers) (4.41.1)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from pytorch-transformers) (1.19.5)\n",
            "Collecting sacremoses\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/75/ee/67241dc87f266093c533a2d4d3d69438e57d7a90abb216fa076e7d475d4a/sacremoses-0.0.45-py3-none-any.whl (895kB)\n",
            "\u001b[K     |████████████████████████████████| 901kB 41.4MB/s \n",
            "\u001b[?25hCollecting sentencepiece\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/f5/99/e0808cb947ba10f575839c43e8fafc9cc44e4a7a2c8f79c60db48220a577/sentencepiece-0.1.95-cp37-cp37m-manylinux2014_x86_64.whl (1.2MB)\n",
            "\u001b[K     |████████████████████████████████| 1.2MB 58.6MB/s \n",
            "\u001b[?25hRequirement already satisfied: typing-extensions in /usr/local/lib/python3.7/dist-packages (from torch>=1.0.0->pytorch-transformers) (3.7.4.3)\n",
            "Collecting s3transfer<0.5.0,>=0.4.0\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/63/d0/693477c688348654ddc21dcdce0817653a294aa43f41771084c25e7ff9c7/s3transfer-0.4.2-py2.py3-none-any.whl (79kB)\n",
            "\u001b[K     |████████████████████████████████| 81kB 9.6MB/s \n",
            "\u001b[?25hCollecting botocore<1.21.0,>=1.20.67\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/82/2b/6a23d63e1b9593919fbe622596fe92e02e3abcec7d2f91594443ee6e4ef9/botocore-1.20.67-py2.py3-none-any.whl (7.5MB)\n",
            "\u001b[K     |████████████████████████████████| 7.5MB 53.8MB/s \n",
            "\u001b[?25hCollecting jmespath<1.0.0,>=0.7.1\n",
            "  Downloading https://files.pythonhosted.org/packages/07/cb/5f001272b6faeb23c1c9e0acc04d48eaaf5c862c17709d20e3469c6e0139/jmespath-0.10.0-py2.py3-none-any.whl\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->pytorch-transformers) (2.10)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->pytorch-transformers) (3.0.4)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests->pytorch-transformers) (2020.12.5)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests->pytorch-transformers) (1.24.3)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from sacremoses->pytorch-transformers) (1.15.0)\n",
            "Requirement already satisfied: click in /usr/local/lib/python3.7/dist-packages (from sacremoses->pytorch-transformers) (7.1.2)\n",
            "Requirement already satisfied: joblib in /usr/local/lib/python3.7/dist-packages (from sacremoses->pytorch-transformers) (1.0.1)\n",
            "Requirement already satisfied: python-dateutil<3.0.0,>=2.1 in /usr/local/lib/python3.7/dist-packages (from botocore<1.21.0,>=1.20.67->boto3->pytorch-transformers) (2.8.1)\n",
            "\u001b[31mERROR: botocore 1.20.67 has requirement urllib3<1.27,>=1.25.4, but you'll have urllib3 1.24.3 which is incompatible.\u001b[0m\n",
            "Installing collected packages: jmespath, botocore, s3transfer, boto3, sacremoses, sentencepiece, pytorch-transformers\n",
            "Successfully installed boto3-1.17.67 botocore-1.20.67 jmespath-0.10.0 pytorch-transformers-1.2.0 s3transfer-0.4.2 sacremoses-0.0.45 sentencepiece-0.1.95\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JwOlQcRjzSTF"
      },
      "source": [
        "# !python ./code/normal_train.py --gpu 0,1 --n-labeled 10 --data-path ./data/imdb_data/ \\\n",
        "# --batch-size 8 --epochs 20 "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5D3KgYLZzwH0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "30985286-b747-4f6f-ae9c-59546a49d33a"
      },
      "source": [
        "#n_unlabeled = 2000\n",
        "\n",
        "!python ./code/train.py --gpu 0,1,2,3 --n-labeled 200 \\\n",
        "--data-path ./data/imdb_data/ --batch-size 1 --batch-size-u 1 --epochs 20 --val-iteration 100 \\\n",
        "--lambda-u 1 --T 1 --alpha 16 --mix-layers-set 7 9 12 \\\n",
        "--lrmain 0.000005 --lrlast 0.0005"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2021-05-04 12:45:22.360529: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0\n",
            "GPU num:  1\n",
            "Whether mix:  True\n",
            "Mix layers sets:  [7, 9, 12]\n",
            "We entered get_data\n",
            "Read data successful (37500, 2) (12500, 2) n_labeled =  200\n",
            "                                                   0  1\n",
            "0  I figure this to be an \"alternate reality\" tee...  0\n",
            "1  This is the kind of movie that wants to be goo...  0\n",
            "2  This was by far the worst movie I've ever seen...  0\n",
            "3  Awful, awful, awful...<br /><br />I loved the ...  0\n",
            "4  Fragile Carne, just before his great period. A...  1\n",
            "tcmalloc: large alloc 2055602176 bytes == 0x564c30662000 @  0x7f6224d38001 0x7f622212354f 0x7f6222173b58 0x7f6222176e83 0x7f622217707b 0x7f6222218761 0x564c1e00e050 0x564c1e00dde0 0x564c1e082244 0x564c1e07cb0e 0x564c1e00f77a 0x564c1e07e86a 0x564c1e00f69a 0x564c1e07da45 0x564c1e07cb0e 0x564c1e07c813 0x564c1e146592 0x564c1e14690d 0x564c1e1467b6 0x564c1e11e103 0x564c1e11ddac 0x7f6223b20bf7 0x564c1e11dc8a\n",
            "[0 0 0]\n",
            "['I figure this to be an \"alternate reality\" teen flick...More precisely a Ferris Bueller type character as the leader of a cheat ring . Yeah, I know it\\'s not meant to compared to Ferris Bueller, at least not in a \"oranges-to-oranges\" way, but it will none-the-less.<br /><br />Bottom-line: It\\'s galaxies away from even being even a minor classic. It is watchable, though only if you\\'re not expecting very much. That said, the main character has some charm, but the premise wears thin because the writing just isn\\'t clever. The movie just did not deliver enough laughs, twists, or tension to keep my interest. <br /><br />To be honest I did continue watching...Watching with hopes to see if anything suddenly clicked. It didn\\'t. So, stylish as it is, I wouldn\\'t recommend this movie. BTW, it seems odd to see Mary Tyler Moore as the principal. She\\'s truly miscast, I hope the paycheck was inordinately big.'\n",
            " \"This is the kind of movie that wants to be good but sucks. First thing, what the hell are those punk trying to do with the school? I think the kids doesn't seem to realize the gravity of the situation. Deker guy say to the girl that they under his responsibility when she ask why he wants to go back for them but right after this he gives a gun to the wheel chair dude and wants him to go alone repair the phone line. Where is the responsibility there? I understand poor actors must pay their food but why not just give them the money that takes to make a stupid movie like that or give that money to a charity. Oh yea and none of them knows how to aim. The stupid punk guy shoots in the cafeteria nowhere like a crazy. They all want to look professional but they all suck. One more thing I don't believe that there's no emergency exit in the school the kids are trying several doors but they all locked. What happens if there's a fire and the dumass security guard is dead? It is illegal to not have an emergency exit in school. Anyway there's a lot more to say but it would be too long. I spent some time of my life to watch a crap.\"\n",
            " \"This was by far the worst movie I've ever seen. And thats compared to Alexander, Fortress 2 and The new world.<br /><br />I should go back to blockbuster and ask for my money back along with compensation as it was a truly traumatic experience. For the first ten minutes i was changing the zoom on my widescreen TV because the actors seemed to be out of screen. I didn't think it was possible to make such a bad film in this day and age, i was wrong. While typing this message, I've thought of a good reason to buy this movie. A joke present at Xmas. I'm blaming the Mrs for this one as she picked it, thanks babe.<br /><br />Be warned.......A true shocker all round!!!!!!\"]\n",
            "#labels 2\n",
            "#Labeled: 400, Unlabeled 2000, Val 4000, Test 12500\n",
            "/usr/local/lib/python3.7/dist-packages/pytorch_transformers/optimization.py:166: UserWarning: This overload of add_ is deprecated:\n",
            "\tadd_(Number alpha, Tensor other)\n",
            "Consider using one of the following signatures instead:\n",
            "\tadd_(Tensor other, *, Number alpha) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:1005.)\n",
            "  exp_avg.mul_(beta1).add_(1.0 - beta1, grad)\n",
            "epoch 0, step 0, loss 0.7396106719970703, Lx 0.7396106719970703, Lu 0.00014463099068962038, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, val acc 0.57, val_loss 0.6853376774787903\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, test acc 0.5904,test loss 0.6814659536552429\n",
            "Epoch:  0\n",
            "Best acc:\n",
            "0.57\n",
            "Test acc:\n",
            "[0.5904]\n",
            "epoch 1, step 0, loss 0.6997432708740234, Lx 0.6976981163024902, Lu 0.04090310260653496, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 1, val acc 0.595, val_loss 0.6668491749763489\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 1, test acc 0.58704,test loss 0.6671121945381164\n",
            "Epoch:  1\n",
            "Best acc:\n",
            "0.595\n",
            "Test acc:\n",
            "[0.5904, 0.58704]\n",
            "epoch 2, step 0, loss 0.8122017979621887, Lx 0.8121722936630249, Lu 0.00029488373547792435, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 2, val acc 0.77075, val_loss 0.6520356574058532\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 2, test acc 0.77256,test loss 0.6506742690467835\n",
            "Epoch:  2\n",
            "Best acc:\n",
            "0.77075\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256]\n",
            "epoch 3, step 0, loss 0.6803975701332092, Lx 0.6775100231170654, Lu 0.019250448793172836, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 3, val acc 0.604, val_loss 0.6354194402694702\n",
            "Epoch:  3\n",
            "Best acc:\n",
            "0.77075\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256]\n",
            "epoch 4, step 0, loss 0.5208114981651306, Lx 0.5204668045043945, Lu 0.0017233811086043715, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 4, val acc 0.8215, val_loss 0.5574319248199463\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 4, test acc 0.8068,test loss 0.5584695840072632\n",
            "Epoch:  4\n",
            "Best acc:\n",
            "0.8215\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068]\n",
            "epoch 5, step 0, loss 0.7445650696754456, Lx 0.7393462061882019, Lu 0.02087537758052349, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 5, val acc 0.791, val_loss 0.49828960180282594\n",
            "Epoch:  5\n",
            "Best acc:\n",
            "0.8215\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068]\n",
            "epoch 6, step 0, loss 0.5570533275604248, Lx 0.54173743724823, Lu 0.051053017377853394, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 6, val acc 0.79725, val_loss 0.4884404957294464\n",
            "Epoch:  6\n",
            "Best acc:\n",
            "0.8215\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068]\n",
            "epoch 7, step 0, loss 0.6021679043769836, Lx 0.5897341370582581, Lu 0.03552503138780594, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 7, val acc 0.79425, val_loss 0.47059365582466123\n",
            "Epoch:  7\n",
            "Best acc:\n",
            "0.8215\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068]\n",
            "epoch 8, step 0, loss 0.29115843772888184, Lx 0.2829664945602417, Lu 0.020479891449213028, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 8, val acc 0.779, val_loss 0.467824214220047\n",
            "Epoch:  8\n",
            "Best acc:\n",
            "0.8215\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068]\n",
            "epoch 9, step 0, loss 0.18605868518352509, Lx 0.14928248524665833, Lu 0.08172489702701569, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 9, val acc 0.81375, val_loss 0.4120035512447357\n",
            "Epoch:  9\n",
            "Best acc:\n",
            "0.8215\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068]\n",
            "epoch 10, step 0, loss 0.41334760189056396, Lx 0.3873244524002075, Lu 0.0520462803542614, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 10, val acc 0.8415, val_loss 0.3921748993396759\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 10, test acc 0.84664,test loss 0.3916944317340851\n",
            "Epoch:  10\n",
            "Best acc:\n",
            "0.8415\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664]\n",
            "epoch 11, step 0, loss 0.18430617451667786, Lx 0.1641652137041092, Lu 0.036619920283555984, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 11, val acc 0.81275, val_loss 0.3981356403827667\n",
            "Epoch:  11\n",
            "Best acc:\n",
            "0.8415\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664]\n",
            "epoch 12, step 0, loss 0.15184679627418518, Lx 0.13520804047584534, Lu 0.02773125469684601, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, val acc 0.859, val_loss 0.33300032091140747\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, test acc 0.86568,test loss 0.33609743103027345\n",
            "Epoch:  12\n",
            "Best acc:\n",
            "0.859\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568]\n",
            "epoch 13, step 0, loss 0.5672394633293152, Lx 0.5331591367721558, Lu 0.05243131145834923, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 13, val acc 0.87625, val_loss 0.3141206321716309\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 13, test acc 0.87112,test loss 0.32120098769187927\n",
            "Epoch:  13\n",
            "Best acc:\n",
            "0.87625\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112]\n",
            "epoch 14, step 0, loss 0.4192795157432556, Lx 0.4074949026107788, Lu 0.016835171729326248, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 14, val acc 0.87925, val_loss 0.30982738184928893\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 14, test acc 0.87208,test loss 0.317490104265213\n",
            "Epoch:  14\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n",
            "epoch 15, step 0, loss 0.670700192451477, Lx 0.6518872380256653, Lu 0.02508392743766308, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 15, val acc 0.84625, val_loss 0.3533463141918182\n",
            "Epoch:  15\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n",
            "epoch 16, step 0, loss 0.6315488219261169, Lx 0.6229164600372314, Lu 0.010790441185235977, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 16, val acc 0.873, val_loss 0.35247458672523496\n",
            "Epoch:  16\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n",
            "epoch 17, step 0, loss 0.2947189509868622, Lx 0.270882785320282, Lu 0.028042536228895187, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 17, val acc 0.87325, val_loss 0.30839163160324096\n",
            "Epoch:  17\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n",
            "epoch 18, step 0, loss 0.7917821407318115, Lx 0.598315417766571, Lu 0.21496304869651794, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 18, val acc 0.865, val_loss 0.31791316890716553\n",
            "Epoch:  18\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n",
            "epoch 19, step 0, loss 0.18486429750919342, Lx 0.15717072784900665, Lu 0.029151126742362976, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 19, val acc 0.87525, val_loss 0.32266535902023313\n",
            "Epoch:  19\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n",
            "Finished training!\n",
            "Best acc:\n",
            "0.87925\n",
            "Test acc:\n",
            "[0.5904, 0.58704, 0.77256, 0.8068, 0.84664, 0.86568, 0.87112, 0.87208]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vsep-UTbV2NU",
        "outputId": "f0f940f6-87d7-472f-b8ba-90e3e61e205a"
      },
      "source": [
        "#n_unlabeled = 20\n",
        "\n",
        "!python ./code/train.py --gpu 0,1,2,3 --n-labeled 200 \\\n",
        "--data-path ./data/imdb_data/ --batch-size 1 --batch-size-u 1 --epochs 20 --val-iteration 100 \\\n",
        "--lambda-u 1 --T 1 --alpha 16 --mix-layers-set 7 9 12 \\\n",
        "--lrmain 0.000005 --lrlast 0.0005"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2021-05-06 11:02:36.613973: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0\n",
            "GPU num:  1\n",
            "Whether mix:  True\n",
            "Mix layers sets:  [7, 9, 12]\n",
            "We entered get_data\n",
            "100% 231508/231508 [00:00<00:00, 314455.58B/s]\n",
            "Read data successful (37500, 2) (12500, 2) n_labeled =  200\n",
            "                                                   0  1\n",
            "0  I figure this to be an \"alternate reality\" tee...  0\n",
            "1  This is the kind of movie that wants to be goo...  0\n",
            "2  This was by far the worst movie I've ever seen...  0\n",
            "3  Awful, awful, awful...<br /><br />I loved the ...  0\n",
            "4  Fragile Carne, just before his great period. A...  1\n",
            "tcmalloc: large alloc 2055602176 bytes == 0x5620b2ede000 @  0x7fb89f771001 0x7fb89cb5c54f 0x7fb89cbacb58 0x7fb89cbafe83 0x7fb89cbb007b 0x7fb89cc51761 0x5620a145c050 0x5620a145bde0 0x5620a14d0244 0x5620a14cab0e 0x5620a145d77a 0x5620a14cc86a 0x5620a145d69a 0x5620a14cba45 0x5620a14cab0e 0x5620a14ca813 0x5620a1594592 0x5620a159490d 0x5620a15947b6 0x5620a156c103 0x5620a156bdac 0x7fb89e559bf7 0x5620a156bc8a\n",
            "[0 0 0]\n",
            "['I figure this to be an \"alternate reality\" teen flick...More precisely a Ferris Bueller type character as the leader of a cheat ring . Yeah, I know it\\'s not meant to compared to Ferris Bueller, at least not in a \"oranges-to-oranges\" way, but it will none-the-less.<br /><br />Bottom-line: It\\'s galaxies away from even being even a minor classic. It is watchable, though only if you\\'re not expecting very much. That said, the main character has some charm, but the premise wears thin because the writing just isn\\'t clever. The movie just did not deliver enough laughs, twists, or tension to keep my interest. <br /><br />To be honest I did continue watching...Watching with hopes to see if anything suddenly clicked. It didn\\'t. So, stylish as it is, I wouldn\\'t recommend this movie. BTW, it seems odd to see Mary Tyler Moore as the principal. She\\'s truly miscast, I hope the paycheck was inordinately big.'\n",
            " \"This is the kind of movie that wants to be good but sucks. First thing, what the hell are those punk trying to do with the school? I think the kids doesn't seem to realize the gravity of the situation. Deker guy say to the girl that they under his responsibility when she ask why he wants to go back for them but right after this he gives a gun to the wheel chair dude and wants him to go alone repair the phone line. Where is the responsibility there? I understand poor actors must pay their food but why not just give them the money that takes to make a stupid movie like that or give that money to a charity. Oh yea and none of them knows how to aim. The stupid punk guy shoots in the cafeteria nowhere like a crazy. They all want to look professional but they all suck. One more thing I don't believe that there's no emergency exit in the school the kids are trying several doors but they all locked. What happens if there's a fire and the dumass security guard is dead? It is illegal to not have an emergency exit in school. Anyway there's a lot more to say but it would be too long. I spent some time of my life to watch a crap.\"\n",
            " \"This was by far the worst movie I've ever seen. And thats compared to Alexander, Fortress 2 and The new world.<br /><br />I should go back to blockbuster and ask for my money back along with compensation as it was a truly traumatic experience. For the first ten minutes i was changing the zoom on my widescreen TV because the actors seemed to be out of screen. I didn't think it was possible to make such a bad film in this day and age, i was wrong. While typing this message, I've thought of a good reason to buy this movie. A joke present at Xmas. I'm blaming the Mrs for this one as she picked it, thanks babe.<br /><br />Be warned.......A true shocker all round!!!!!!\"]\n",
            "#labels 2\n",
            "#Labeled: 400, Unlabeled 20, Val 4000, Test 12500\n",
            "100% 433/433 [00:00<00:00, 345930.22B/s]\n",
            "100% 440473133/440473133 [00:36<00:00, 12145034.41B/s]\n",
            "/usr/local/lib/python3.7/dist-packages/pytorch_transformers/optimization.py:166: UserWarning: This overload of add_ is deprecated:\n",
            "\tadd_(Number alpha, Tensor other)\n",
            "Consider using one of the following signatures instead:\n",
            "\tadd_(Tensor other, *, Number alpha) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:1005.)\n",
            "  exp_avg.mul_(beta1).add_(1.0 - beta1, grad)\n",
            "epoch 0, step 0, loss 0.650644838809967, Lx 0.650644838809967, Lu 0.024831295013427734, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 1, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, val acc 0.5255, val_loss 0.6960830965042114\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, test acc 0.514,test loss 0.6981466621780396\n",
            "Epoch:  0\n",
            "Best acc:\n",
            "0.5255\n",
            "Test acc:\n",
            "[0.514]\n",
            "epoch 1, step 0, loss 0.702397882938385, Lx 0.6994996070861816, Lu 0.05796527490019798, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 1, val acc 0.57675, val_loss 0.6784839615821838\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 1, test acc 0.58752,test loss 0.6765265465354919\n",
            "Epoch:  1\n",
            "Best acc:\n",
            "0.57675\n",
            "Test acc:\n",
            "[0.514, 0.58752]\n",
            "epoch 2, step 0, loss 0.5686669945716858, Lx 0.564593493938446, Lu 0.040734726935625076, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 2, val acc 0.52425, val_loss 0.6884443445205688\n",
            "Epoch:  2\n",
            "Best acc:\n",
            "0.57675\n",
            "Test acc:\n",
            "[0.514, 0.58752]\n",
            "epoch 3, step 0, loss 0.5085157752037048, Lx 0.5066030621528625, Lu 0.012751532718539238, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 3, val acc 0.507, val_loss 0.6762183394432068\n",
            "Epoch:  3\n",
            "Best acc:\n",
            "0.57675\n",
            "Test acc:\n",
            "[0.514, 0.58752]\n",
            "epoch 4, step 0, loss 0.5860884785652161, Lx 0.5761222839355469, Lu 0.049831002950668335, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 4, val acc 0.5345, val_loss 0.6701716613769532\n",
            "Epoch:  4\n",
            "Best acc:\n",
            "0.57675\n",
            "Test acc:\n",
            "[0.514, 0.58752]\n",
            "epoch 5, step 0, loss 0.6090359091758728, Lx 0.6026454567909241, Lu 0.025561809539794922, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 5, val acc 0.66525, val_loss 0.6137947525978088\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 5, test acc 0.66456,test loss 0.6118102260398864\n",
            "Epoch:  5\n",
            "Best acc:\n",
            "0.66525\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456]\n",
            "epoch 6, step 0, loss 0.374682754278183, Lx 0.3724271059036255, Lu 0.007518833037465811, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 6, val acc 0.673, val_loss 0.6052378997802734\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 6, test acc 0.6728,test loss 0.6062600430679321\n",
            "Epoch:  6\n",
            "Best acc:\n",
            "0.673\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728]\n",
            "epoch 7, step 0, loss 0.3570529818534851, Lx 0.3421803414821625, Lu 0.042493294924497604, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 7, val acc 0.77925, val_loss 0.5170664167404175\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 7, test acc 0.77368,test loss 0.5162439514541626\n",
            "Epoch:  7\n",
            "Best acc:\n",
            "0.77925\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368]\n",
            "epoch 8, step 0, loss 0.8064042925834656, Lx 0.7810063362121582, Lu 0.06349486112594604, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 8, val acc 0.841, val_loss 0.4160185399055481\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 8, test acc 0.8408,test loss 0.41214957021713255\n",
            "Epoch:  8\n",
            "Best acc:\n",
            "0.841\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408]\n",
            "epoch 9, step 0, loss 0.2584199011325836, Lx 0.24370095133781433, Lu 0.03270876407623291, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 9, val acc 0.7645, val_loss 0.5403406667709351\n",
            "Epoch:  9\n",
            "Best acc:\n",
            "0.841\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408]\n",
            "epoch 10, step 0, loss 0.9352860450744629, Lx 0.7557734251022339, Lu 0.35902518033981323, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 10, val acc 0.86225, val_loss 0.3577397601604462\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 10, test acc 0.85744,test loss 0.3621852727794647\n",
            "Epoch:  10\n",
            "Best acc:\n",
            "0.86225\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744]\n",
            "epoch 11, step 0, loss 0.6824837923049927, Lx 0.6592611074447632, Lu 0.04222303628921509, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 11, val acc 0.8045, val_loss 0.4404148101806641\n",
            "Epoch:  11\n",
            "Best acc:\n",
            "0.86225\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744]\n",
            "epoch 12, step 0, loss 0.1523970067501068, Lx 0.13287682831287384, Lu 0.03253363072872162, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, val acc 0.86475, val_loss 0.3674861674308777\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, test acc 0.86256,test loss 0.36603427421569823\n",
            "Epoch:  12\n",
            "Best acc:\n",
            "0.86475\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256]\n",
            "epoch 13, step 0, loss 0.1363743096590042, Lx 0.1322396844625473, Lu 0.006360962986946106, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 13, val acc 0.8625, val_loss 0.3483399276733398\n",
            "Epoch:  13\n",
            "Best acc:\n",
            "0.86475\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256]\n",
            "epoch 14, step 0, loss 0.6580904126167297, Lx 0.6537113785743713, Lu 0.006255753338336945, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 1, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 14, val acc 0.86975, val_loss 0.3202381811141968\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 14, test acc 0.86392,test loss 0.32847700792312623\n",
            "Epoch:  14\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n",
            "epoch 15, step 0, loss 0.14649009704589844, Lx 0.14276185631752014, Lu 0.0049709840677678585, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 15, val acc 0.8275, val_loss 0.38719457340240476\n",
            "Epoch:  15\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n",
            "epoch 16, step 0, loss 0.7567310333251953, Lx 0.6963838338851929, Lu 0.07543402910232544, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 16, val acc 0.86275, val_loss 0.31855640292167664\n",
            "Epoch:  16\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n",
            "epoch 17, step 0, loss 0.13508932292461395, Lx 0.1304309070110321, Lu 0.005480486899614334, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 17, val acc 0.86775, val_loss 0.32064328670501707\n",
            "Epoch:  17\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n",
            "epoch 18, step 0, loss 0.15728886425495148, Lx 0.14038224518299103, Lu 0.018785135820508003, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 18, val acc 0.8655, val_loss 0.3241582999229431\n",
            "Epoch:  18\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n",
            "epoch 19, step 0, loss 0.9362813830375671, Lx 0.8861849904060364, Lu 0.05273304134607315, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 19, val acc 0.8425, val_loss 0.36075281429290773\n",
            "Epoch:  19\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n",
            "Finished training!\n",
            "Best acc:\n",
            "0.86975\n",
            "Test acc:\n",
            "[0.514, 0.58752, 0.66456, 0.6728, 0.77368, 0.8408, 0.85744, 0.86256, 0.86392]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tUZbRIS-WLIm",
        "outputId": "6a44cdf7-132b-41ce-a82d-fe2ec248a2a1"
      },
      "source": [
        "#n_unlabeled = 200\n",
        "\n",
        "!python ./code/train.py --gpu 0,1,2,3 --n-labeled 200 \\\n",
        "--data-path ./data/imdb_data/ --batch-size 1 --batch-size-u 1 --epochs 20 --val-iteration 100 \\\n",
        "--lambda-u 1 --T 1 --alpha 16 --mix-layers-set 7 9 12 \\\n",
        "--lrmain 0.000005 --lrlast 0.0005"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2021-05-06 11:41:05.632726: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0\n",
            "GPU num:  1\n",
            "Whether mix:  True\n",
            "Mix layers sets:  [7, 9, 12]\n",
            "We entered get_data\n",
            "Read data successful (37500, 2) (12500, 2) n_labeled =  200\n",
            "                                                   0  1\n",
            "0  I figure this to be an \"alternate reality\" tee...  0\n",
            "1  This is the kind of movie that wants to be goo...  0\n",
            "2  This was by far the worst movie I've ever seen...  0\n",
            "3  Awful, awful, awful...<br /><br />I loved the ...  0\n",
            "4  Fragile Carne, just before his great period. A...  1\n",
            "tcmalloc: large alloc 2055602176 bytes == 0x5608fb8fa000 @  0x7f2f5e4b7001 0x7f2f5b8a254f 0x7f2f5b8f2b58 0x7f2f5b8f5e83 0x7f2f5b8f607b 0x7f2f5b997761 0x5608eaabd050 0x5608eaabcde0 0x5608eab31244 0x5608eab2bb0e 0x5608eaabe77a 0x5608eab2d86a 0x5608eaabe69a 0x5608eab2ca45 0x5608eab2bb0e 0x5608eab2b813 0x5608eabf5592 0x5608eabf590d 0x5608eabf57b6 0x5608eabcd103 0x5608eabccdac 0x7f2f5d29fbf7 0x5608eabccc8a\n",
            "[0 0 0]\n",
            "['I figure this to be an \"alternate reality\" teen flick...More precisely a Ferris Bueller type character as the leader of a cheat ring . Yeah, I know it\\'s not meant to compared to Ferris Bueller, at least not in a \"oranges-to-oranges\" way, but it will none-the-less.<br /><br />Bottom-line: It\\'s galaxies away from even being even a minor classic. It is watchable, though only if you\\'re not expecting very much. That said, the main character has some charm, but the premise wears thin because the writing just isn\\'t clever. The movie just did not deliver enough laughs, twists, or tension to keep my interest. <br /><br />To be honest I did continue watching...Watching with hopes to see if anything suddenly clicked. It didn\\'t. So, stylish as it is, I wouldn\\'t recommend this movie. BTW, it seems odd to see Mary Tyler Moore as the principal. She\\'s truly miscast, I hope the paycheck was inordinately big.'\n",
            " \"This is the kind of movie that wants to be good but sucks. First thing, what the hell are those punk trying to do with the school? I think the kids doesn't seem to realize the gravity of the situation. Deker guy say to the girl that they under his responsibility when she ask why he wants to go back for them but right after this he gives a gun to the wheel chair dude and wants him to go alone repair the phone line. Where is the responsibility there? I understand poor actors must pay their food but why not just give them the money that takes to make a stupid movie like that or give that money to a charity. Oh yea and none of them knows how to aim. The stupid punk guy shoots in the cafeteria nowhere like a crazy. They all want to look professional but they all suck. One more thing I don't believe that there's no emergency exit in the school the kids are trying several doors but they all locked. What happens if there's a fire and the dumass security guard is dead? It is illegal to not have an emergency exit in school. Anyway there's a lot more to say but it would be too long. I spent some time of my life to watch a crap.\"\n",
            " \"This was by far the worst movie I've ever seen. And thats compared to Alexander, Fortress 2 and The new world.<br /><br />I should go back to blockbuster and ask for my money back along with compensation as it was a truly traumatic experience. For the first ten minutes i was changing the zoom on my widescreen TV because the actors seemed to be out of screen. I didn't think it was possible to make such a bad film in this day and age, i was wrong. While typing this message, I've thought of a good reason to buy this movie. A joke present at Xmas. I'm blaming the Mrs for this one as she picked it, thanks babe.<br /><br />Be warned.......A true shocker all round!!!!!!\"]\n",
            "#labels 2\n",
            "#Labeled: 400, Unlabeled 200, Val 4000, Test 12500\n",
            "/usr/local/lib/python3.7/dist-packages/pytorch_transformers/optimization.py:166: UserWarning: This overload of add_ is deprecated:\n",
            "\tadd_(Number alpha, Tensor other)\n",
            "Consider using one of the following signatures instead:\n",
            "\tadd_(Tensor other, *, Number alpha) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:1005.)\n",
            "  exp_avg.mul_(beta1).add_(1.0 - beta1, grad)\n",
            "epoch 0, step 0, loss 0.7574734091758728, Lx 0.7574734091758728, Lu 0.043063461780548096, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 0, val acc 0.5, val_loss 0.9327751135826111\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, test acc 0.49256,test loss 0.9427609080696105\n",
            "Epoch:  0\n",
            "Best acc:\n",
            "0.5\n",
            "Test acc:\n",
            "[0.49256]\n",
            "epoch 1, step 0, loss 0.3770219087600708, Lx 0.37658122181892395, Lu 0.008813553489744663, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 1, val acc 0.53175, val_loss 0.6881238470077514\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 1, test acc 0.54304,test loss 0.6860962712287902\n",
            "Epoch:  1\n",
            "Best acc:\n",
            "0.53175\n",
            "Test acc:\n",
            "[0.49256, 0.54304]\n",
            "epoch 2, step 0, loss 0.7471444010734558, Lx 0.7415210008621216, Lu 0.05623414367437363, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 2, val acc 0.5875, val_loss 0.6773163042068482\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 1, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 2, test acc 0.59712,test loss 0.6755100682640076\n",
            "Epoch:  2\n",
            "Best acc:\n",
            "0.5875\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712]\n",
            "epoch 3, step 0, loss 0.8042441606521606, Lx 0.7940496206283569, Lu 0.06796364486217499, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 3, val acc 0.5725, val_loss 0.6644533243179321\n",
            "Epoch:  3\n",
            "Best acc:\n",
            "0.5875\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712]\n",
            "epoch 4, step 0, loss 0.6566349267959595, Lx 0.6514391899108887, Lu 0.025978561490774155, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 4, val acc 0.58375, val_loss 0.6595965142250061\n",
            "Epoch:  4\n",
            "Best acc:\n",
            "0.5875\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712]\n",
            "epoch 5, step 0, loss 0.8051663637161255, Lx 0.7890709638595581, Lu 0.06438156962394714, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 5, val acc 0.5055, val_loss 0.7734551095962524\n",
            "Epoch:  5\n",
            "Best acc:\n",
            "0.5875\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712]\n",
            "epoch 6, step 0, loss 1.0574569702148438, Lx 1.0351848602294922, Lu 0.0742405503988266, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 6, val acc 0.60675, val_loss 0.604145510673523\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 6, test acc 0.61384,test loss 0.5966447203826905\n",
            "Epoch:  6\n",
            "Best acc:\n",
            "0.60675\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384]\n",
            "epoch 7, step 0, loss 0.8146408796310425, Lx 0.7956076264381409, Lu 0.05438077449798584, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 7, val acc 0.746, val_loss 0.5278854079246521\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 7, test acc 0.7344,test loss 0.5360772274398804\n",
            "Epoch:  7\n",
            "Best acc:\n",
            "0.746\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344]\n",
            "epoch 8, step 0, loss 0.489507794380188, Lx 0.4603242576122284, Lu 0.07295883446931839, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 8, val acc 0.852, val_loss 0.3839909029006958\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 8, test acc 0.85376,test loss 0.3848819563961029\n",
            "Epoch:  8\n",
            "Best acc:\n",
            "0.852\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376]\n",
            "epoch 9, step 0, loss 0.5492282509803772, Lx 0.538707435131073, Lu 0.02337961457669735, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 9, val acc 0.77525, val_loss 0.47900087881088255\n",
            "Epoch:  9\n",
            "Best acc:\n",
            "0.852\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376]\n",
            "epoch 10, step 0, loss 0.20643101632595062, Lx 0.18565088510513306, Lu 0.04156026616692543, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 10, val acc 0.8245, val_loss 0.408937004327774\n",
            "Epoch:  10\n",
            "Best acc:\n",
            "0.852\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376]\n",
            "epoch 11, step 0, loss 1.000005841255188, Lx 0.9513038396835327, Lu 0.08854909241199493, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 11, val acc 0.792, val_loss 0.4640701143741608\n",
            "Epoch:  11\n",
            "Best acc:\n",
            "0.852\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376]\n",
            "epoch 12, step 0, loss 0.6199882626533508, Lx 0.5792830586433411, Lu 0.06784196943044662, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 12, val acc 0.86525, val_loss 0.32732873392105105\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, test acc 0.86472,test loss 0.32881740372657775\n",
            "Epoch:  12\n",
            "Best acc:\n",
            "0.86525\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472]\n",
            "epoch 13, step 0, loss 0.23079022765159607, Lx 0.17041832208633423, Lu 0.09287986159324646, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 13, val acc 0.869, val_loss 0.3236159331798553\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 13, test acc 0.8664,test loss 0.3263912340068817\n",
            "Epoch:  13\n",
            "Best acc:\n",
            "0.869\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664]\n",
            "epoch 14, step 0, loss 0.5090233683586121, Lx 0.4519437551498413, Lu 0.08154232800006866, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 14, val acc 0.86625, val_loss 0.3297935223579407\n",
            "Epoch:  14\n",
            "Best acc:\n",
            "0.869\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664]\n",
            "epoch 15, step 0, loss 0.7053481340408325, Lx 0.7032783031463623, Lu 0.0027597867883741856, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 15, val acc 0.83825, val_loss 0.38427788186073303\n",
            "Epoch:  15\n",
            "Best acc:\n",
            "0.869\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664]\n",
            "epoch 16, step 0, loss 0.1015184074640274, Lx 0.03963209316134453, Lu 0.07735789567232132, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 16, val acc 0.86925, val_loss 0.3190321319103241\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 16, test acc 0.86408,test loss 0.3323489342594147\n",
            "Epoch:  16\n",
            "Best acc:\n",
            "0.86925\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664, 0.86408]\n",
            "epoch 17, step 0, loss 0.4768367409706116, Lx 0.46487393975257874, Lu 0.014073878526687622, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 17, val acc 0.8785, val_loss 0.324767746925354\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 17, test acc 0.8672,test loss 0.33721817273139953\n",
            "Epoch:  17\n",
            "Best acc:\n",
            "0.8785\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664, 0.86408, 0.8672]\n",
            "epoch 18, step 0, loss 0.21785350143909454, Lx 0.20381329953670502, Lu 0.015600228682160378, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 18, val acc 0.8725, val_loss 0.3222604763507843\n",
            "Epoch:  18\n",
            "Best acc:\n",
            "0.8785\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664, 0.86408, 0.8672]\n",
            "epoch 19, step 0, loss 0.1544860452413559, Lx 0.11031848937273026, Lu 0.046492159366607666, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "epoch 19, val acc 0.87275, val_loss 0.3147008948326111\n",
            "Epoch:  19\n",
            "Best acc:\n",
            "0.8785\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664, 0.86408, 0.8672]\n",
            "Finished training!\n",
            "Best acc:\n",
            "0.8785\n",
            "Test acc:\n",
            "[0.49256, 0.54304, 0.59712, 0.61384, 0.7344, 0.85376, 0.86472, 0.8664, 0.86408, 0.8672]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BrwYoX47WL7i",
        "outputId": "0762d52c-5740-4e83-9579-a3c9f5970267"
      },
      "source": [
        "#n_unlabeled = 1000\n",
        "\n",
        "!python ./code/train.py --gpu 0,1,2,3 --n-labeled 200 \\\n",
        "--data-path ./data/imdb_data/ --batch-size 1 --batch-size-u 1 --epochs 20 --val-iteration 100 \\\n",
        "--lambda-u 1 --T 1 --alpha 16 --mix-layers-set 7 9 12 \\\n",
        "--lrmain 0.000005 --lrlast 0.0005"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2021-05-06 12:30:38.107901: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0\n",
            "GPU num:  1\n",
            "Whether mix:  True\n",
            "Mix layers sets:  [7, 9, 12]\n",
            "We entered get_data\n",
            "Read data successful (37500, 2) (12500, 2) n_labeled =  200\n",
            "                                                   0  1\n",
            "0  I figure this to be an \"alternate reality\" tee...  0\n",
            "1  This is the kind of movie that wants to be goo...  0\n",
            "2  This was by far the worst movie I've ever seen...  0\n",
            "3  Awful, awful, awful...<br /><br />I loved the ...  0\n",
            "4  Fragile Carne, just before his great period. A...  1\n",
            "tcmalloc: large alloc 2055602176 bytes == 0x55b5e6592000 @  0x7f1bdff75001 0x7f1bdd36054f 0x7f1bdd3b0b58 0x7f1bdd3b3e83 0x7f1bdd3b407b 0x7f1bdd455761 0x55b5d4f6e050 0x55b5d4f6dde0 0x55b5d4fe2244 0x55b5d4fdcb0e 0x55b5d4f6f77a 0x55b5d4fde86a 0x55b5d4f6f69a 0x55b5d4fdda45 0x55b5d4fdcb0e 0x55b5d4fdc813 0x55b5d50a6592 0x55b5d50a690d 0x55b5d50a67b6 0x55b5d507e103 0x55b5d507ddac 0x7f1bded5dbf7 0x55b5d507dc8a\n",
            "[0 0 0]\n",
            "['I figure this to be an \"alternate reality\" teen flick...More precisely a Ferris Bueller type character as the leader of a cheat ring . Yeah, I know it\\'s not meant to compared to Ferris Bueller, at least not in a \"oranges-to-oranges\" way, but it will none-the-less.<br /><br />Bottom-line: It\\'s galaxies away from even being even a minor classic. It is watchable, though only if you\\'re not expecting very much. That said, the main character has some charm, but the premise wears thin because the writing just isn\\'t clever. The movie just did not deliver enough laughs, twists, or tension to keep my interest. <br /><br />To be honest I did continue watching...Watching with hopes to see if anything suddenly clicked. It didn\\'t. So, stylish as it is, I wouldn\\'t recommend this movie. BTW, it seems odd to see Mary Tyler Moore as the principal. She\\'s truly miscast, I hope the paycheck was inordinately big.'\n",
            " \"This is the kind of movie that wants to be good but sucks. First thing, what the hell are those punk trying to do with the school? I think the kids doesn't seem to realize the gravity of the situation. Deker guy say to the girl that they under his responsibility when she ask why he wants to go back for them but right after this he gives a gun to the wheel chair dude and wants him to go alone repair the phone line. Where is the responsibility there? I understand poor actors must pay their food but why not just give them the money that takes to make a stupid movie like that or give that money to a charity. Oh yea and none of them knows how to aim. The stupid punk guy shoots in the cafeteria nowhere like a crazy. They all want to look professional but they all suck. One more thing I don't believe that there's no emergency exit in the school the kids are trying several doors but they all locked. What happens if there's a fire and the dumass security guard is dead? It is illegal to not have an emergency exit in school. Anyway there's a lot more to say but it would be too long. I spent some time of my life to watch a crap.\"\n",
            " \"This was by far the worst movie I've ever seen. And thats compared to Alexander, Fortress 2 and The new world.<br /><br />I should go back to blockbuster and ask for my money back along with compensation as it was a truly traumatic experience. For the first ten minutes i was changing the zoom on my widescreen TV because the actors seemed to be out of screen. I didn't think it was possible to make such a bad film in this day and age, i was wrong. While typing this message, I've thought of a good reason to buy this movie. A joke present at Xmas. I'm blaming the Mrs for this one as she picked it, thanks babe.<br /><br />Be warned.......A true shocker all round!!!!!!\"]\n",
            "#labels 2\n",
            "#Labeled: 400, Unlabeled 1000, Val 4000, Test 12500\n",
            "/usr/local/lib/python3.7/dist-packages/pytorch_transformers/optimization.py:166: UserWarning: This overload of add_ is deprecated:\n",
            "\tadd_(Number alpha, Tensor other)\n",
            "Consider using one of the following signatures instead:\n",
            "\tadd_(Tensor other, *, Number alpha) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:1005.)\n",
            "  exp_avg.mul_(beta1).add_(1.0 - beta1, grad)\n",
            "epoch 0, step 0, loss 0.7336769700050354, Lx 0.7336769700050354, Lu 0.049696292728185654, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, val acc 0.575, val_loss 0.6877691512107849\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 0, test acc 0.5672,test loss 0.6903820365715027\n",
            "Epoch:  0\n",
            "Best acc:\n",
            "0.575\n",
            "Test acc:\n",
            "[0.5672]\n",
            "epoch 1, step 0, loss 0.6311439871788025, Lx 0.6291066408157349, Lu 0.04074747860431671, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 1, val acc 0.5, val_loss 0.702558961391449\n",
            "Epoch:  1\n",
            "Best acc:\n",
            "0.575\n",
            "Test acc:\n",
            "[0.5672]\n",
            "epoch 2, step 0, loss 0.43648841977119446, Lx 0.4364425837993622, Lu 0.00045849260641261935, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 2, val acc 0.5, val_loss 0.6925452165603637\n",
            "Epoch:  2\n",
            "Best acc:\n",
            "0.575\n",
            "Test acc:\n",
            "[0.5672]\n",
            "epoch 3, step 0, loss 0.5067897439002991, Lx 0.5057917833328247, Lu 0.006652909331023693, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 3, val acc 0.69375, val_loss 0.6507252497673035\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 3, test acc 0.70584,test loss 0.6490635489082336\n",
            "Epoch:  3\n",
            "Best acc:\n",
            "0.69375\n",
            "Test acc:\n",
            "[0.5672, 0.70584]\n",
            "epoch 4, step 0, loss 0.7762799859046936, Lx 0.7655279040336609, Lu 0.05376027524471283, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 4, val acc 0.688, val_loss 0.6105033531188965\n",
            "Epoch:  4\n",
            "Best acc:\n",
            "0.69375\n",
            "Test acc:\n",
            "[0.5672, 0.70584]\n",
            "epoch 5, step 0, loss 0.6711828708648682, Lx 0.6538444757461548, Lu 0.06935358047485352, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 5, val acc 0.83825, val_loss 0.40972396516799925\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 5, test acc 0.84024,test loss 0.41251777837753295\n",
            "Epoch:  5\n",
            "Best acc:\n",
            "0.83825\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024]\n",
            "epoch 6, step 0, loss 0.8934648633003235, Lx 0.8837721347808838, Lu 0.03230912610888481, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 6, val acc 0.7535, val_loss 0.49393098187446594\n",
            "Epoch:  6\n",
            "Best acc:\n",
            "0.83825\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024]\n",
            "epoch 7, step 0, loss 0.8456793427467346, Lx 0.8120992183685303, Lu 0.09594319760799408, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 7, val acc 0.8055, val_loss 0.433753630399704\n",
            "Epoch:  7\n",
            "Best acc:\n",
            "0.83825\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024]\n",
            "epoch 8, step 0, loss 0.6944429278373718, Lx 0.6645926833152771, Lu 0.0746256560087204, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 8, val acc 0.85275, val_loss 0.3878583300113678\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 8, test acc 0.84384,test loss 0.3922086692619324\n",
            "Epoch:  8\n",
            "Best acc:\n",
            "0.85275\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384]\n",
            "epoch 9, step 0, loss 0.39766913652420044, Lx 0.39526110887527466, Lu 0.005351146683096886, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 9, val acc 0.84125, val_loss 0.4073146097660065\n",
            "Epoch:  9\n",
            "Best acc:\n",
            "0.85275\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384]\n",
            "epoch 10, step 0, loss 0.3784376084804535, Lx 0.3686269521713257, Lu 0.019621338695287704, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 10, val acc 0.74375, val_loss 0.5682485265731811\n",
            "Epoch:  10\n",
            "Best acc:\n",
            "0.85275\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384]\n",
            "epoch 11, step 0, loss 0.05547560751438141, Lx 0.052526529878377914, Lu 0.005361958406865597, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 11, val acc 0.813, val_loss 0.40592108488082884\n",
            "Epoch:  11\n",
            "Best acc:\n",
            "0.85275\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384]\n",
            "epoch 12, step 0, loss 0.3315160572528839, Lx 0.3299200236797333, Lu 0.002660031896084547, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, val acc 0.866, val_loss 0.34697330021858214\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 12, test acc 0.86304,test loss 0.34970808387756347\n",
            "Epoch:  12\n",
            "Best acc:\n",
            "0.866\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304]\n",
            "epoch 13, step 0, loss 0.39651602506637573, Lx 0.3732106387615204, Lu 0.035854458808898926, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 13, val acc 0.866, val_loss 0.3372043137550354\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 13, test acc 0.86288,test loss 0.3408915389728546\n",
            "Epoch:  13\n",
            "Best acc:\n",
            "0.866\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288]\n",
            "epoch 14, step 0, loss 0.2423585206270218, Lx 0.22482329607009888, Lu 0.0250503271818161, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 14, val acc 0.84825, val_loss 0.3759905173778534\n",
            "Epoch:  14\n",
            "Best acc:\n",
            "0.866\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288]\n",
            "epoch 15, step 0, loss 0.3446080982685089, Lx 0.305366575717926, Lu 0.05232202261686325, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 15, val acc 0.878, val_loss 0.3180456516742706\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "epoch 15, test acc 0.86784,test loss 0.32762994766235354\n",
            "Epoch:  15\n",
            "Best acc:\n",
            "0.878\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288, 0.86784]\n",
            "epoch 16, step 0, loss 0.8238041400909424, Lx 0.7985482215881348, Lu 0.03156987577676773, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 16, val acc 0.85875, val_loss 0.3352411606311798\n",
            "Epoch:  16\n",
            "Best acc:\n",
            "0.878\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288, 0.86784]\n",
            "epoch 17, step 0, loss 0.7803709506988525, Lx 0.7381078004837036, Lu 0.04972133785486221, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 17, val acc 0.874, val_loss 0.31306718015670776\n",
            "Epoch:  17\n",
            "Best acc:\n",
            "0.878\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288, 0.86784]\n",
            "epoch 18, step 0, loss 0.4352932870388031, Lx 0.13755470514297485, Lu 0.33082064986228943, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 18, val acc 0.8765, val_loss 0.34186310720443724\n",
            "Epoch:  18\n",
            "Best acc:\n",
            "0.878\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288, 0.86784]\n",
            "epoch 19, step 0, loss 0.6125064492225647, Lx 0.5939635038375854, Lu 0.01951889507472515, Lu2 0.0\n",
            "Sample some true labeles and predicted labels\n",
            "tensor([0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1],\n",
            "       device='cuda:0')\n",
            "tensor([0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1],\n",
            "       device='cuda:0')\n",
            "epoch 19, val acc 0.842, val_loss 0.3759227991104126\n",
            "Epoch:  19\n",
            "Best acc:\n",
            "0.878\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288, 0.86784]\n",
            "Finished training!\n",
            "Best acc:\n",
            "0.878\n",
            "Test acc:\n",
            "[0.5672, 0.70584, 0.84024, 0.84384, 0.86304, 0.86288, 0.86784]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G0e04b0SZjYn"
      },
      "source": [
        "# python ./code/train.py --gpu 0,1 --n-labeled 10 --data-path ./data/imdb_data/ \\\n",
        "# --batch-size 1 --batch-size-u 1 --epochs 50 --val-iteration 20 \\\n",
        "# --lambda-u 0 --T 1 --alpha 16 --mix-layers-set 7 9 12 --separate-mix True "
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}