Unsupervised-Classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Unsupervised Classification.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNkjuoGB97kTLhfng8oXdbd"
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wTz2d3m86ngZ"
      },
      "source": [
        "# Setup"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z_4mjAIu8NNR",
        "outputId": "9dbf76ed-8cf0-499f-e6d0-26f18a4bb241"
      },
      "source": [
        "!pip install pytorch_metric_learning\n",
        "!pip install faiss-gpu\n",
        "# !pip install faiss-cpu"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting pytorch_metric_learning\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/8e/1b/d85136231f8d14fdadcd0a13359bbef02b565f0a0c954297f38c4997c61b/pytorch_metric_learning-0.9.99-py3-none-any.whl (105kB)\n",
            "\r\u001b[K     |███▏                            | 10kB 21.7MB/s eta 0:00:01\r\u001b[K     |██████▎                         | 20kB 20.5MB/s eta 0:00:01\r\u001b[K     |█████████▍                      | 30kB 15.1MB/s eta 0:00:01\r\u001b[K     |████████████▌                   | 40kB 13.0MB/s eta 0:00:01\r\u001b[K     |███████████████▋                | 51kB 7.6MB/s eta 0:00:01\r\u001b[K     |██████████████████▊             | 61kB 8.7MB/s eta 0:00:01\r\u001b[K     |█████████████████████▉          | 71kB 8.6MB/s eta 0:00:01\r\u001b[K     |█████████████████████████       | 81kB 9.3MB/s eta 0:00:01\r\u001b[K     |████████████████████████████    | 92kB 9.0MB/s eta 0:00:01\r\u001b[K     |███████████████████████████████▏| 102kB 7.4MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 112kB 7.4MB/s \n",
            "\u001b[?25hRequirement already satisfied: torchvision in /usr/local/lib/python3.7/dist-packages (from pytorch_metric_learning) (0.9.1+cu101)\n",
            "Requirement already satisfied: torch>=1.6.0 in /usr/local/lib/python3.7/dist-packages (from pytorch_metric_learning) (1.8.1+cu101)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from pytorch_metric_learning) (4.41.1)\n",
            "Requirement already satisfied: scikit-learn in /usr/local/lib/python3.7/dist-packages (from pytorch_metric_learning) (0.22.2.post1)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from pytorch_metric_learning) (1.19.5)\n",
            "Requirement already satisfied: pillow>=4.1.1 in /usr/local/lib/python3.7/dist-packages (from torchvision->pytorch_metric_learning) (7.1.2)\n",
            "Requirement already satisfied: typing-extensions in /usr/local/lib/python3.7/dist-packages (from torch>=1.6.0->pytorch_metric_learning) (3.7.4.3)\n",
            "Requirement already satisfied: scipy>=0.17.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn->pytorch_metric_learning) (1.4.1)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn->pytorch_metric_learning) (1.0.1)\n",
            "Installing collected packages: pytorch-metric-learning\n",
            "Successfully installed pytorch-metric-learning-0.9.99\n",
            "Collecting faiss-cpu\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/48/0c/efd43c4feac172867409f38f07949c36602355ec7196749d10f905d09228/faiss_cpu-1.7.0-cp37-cp37m-manylinux2014_x86_64.whl (8.1MB)\n",
            "\u001b[K     |████████████████████████████████| 8.2MB 7.1MB/s \n",
            "\u001b[?25hInstalling collected packages: faiss-cpu\n",
            "Successfully installed faiss-cpu-1.7.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WxZqKDGMCI-r"
      },
      "source": [
        "import torch\n",
        "import pickle\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import torch.nn as nn\n",
        "import matplotlib.pyplot as plt\n",
        "from tqdm import tqdm\n",
        "from google.colab import drive\n",
        "from sklearn.metrics import accuracy_score\n",
        "from torch.utils.data import Dataset, DataLoader, SequentialSampler, TensorDataset, random_split\n",
        "from pytorch_metric_learning import losses, miners, reducers, samplers, distances\n",
        "\n",
        "\n",
        "np.set_printoptions(suppress=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mxsY5IR_CTVo",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d56c3f14-3783-4ccc-e4ec-88422e700117"
      },
      "source": [
        "drive.mount('/content/gdrive/')\n",
        "path = 'gdrive/MyDrive/Programming Projects/Embeddings/data/'"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/gdrive/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7cPCsjG9CQnH"
      },
      "source": [
        "use_cuda = torch.cuda.is_available()\n",
        "\n",
        "device = torch.device(\"cuda:0\" if use_cuda else \"cpu\") "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qpCy7BRaCVFd"
      },
      "source": [
        "#HYPERPARAMETERS\n",
        "\n",
        "np.random.seed(1871)\n",
        "CHECKPOINT_PATH = path + \"model.pt\"\n",
        "LOAD_PICKLED_DATASET = True\n",
        "\n",
        "#Positive pairs generation\n",
        "SEQUENCE_LEN = 120\n",
        "SEQUENCES_PER_CLIENT = 10\n",
        "\n",
        "#Encoder\n",
        "EVENT_ENCODING_DIMENSIONS = 32\n",
        "RNN_HIDDEN_STATE_DIMENSIONS = 256\n",
        "EVENT_EMBEDDING_DIMENSIONS = 8\n",
        "\n",
        "EPOCHS = 40\n",
        "BATCH_SIZE = 128\n",
        "LEARNING_RATE = 5e-4\n",
        "\n",
        "#Classifier\n",
        "CLASSIFIER_LAYER1_DIM = 256\n",
        "CLASSIFIER_LAYER2_DIM = 256\n",
        "\n",
        "CLASSIFIER_EPOCHS = 100\n",
        "CLASSIFIER_BATCH_SIZE = 256\n",
        "CLASSIFIER_LEARNING_RATE = 1e-4\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "An6RYp5h6lrU"
      },
      "source": [
        "# Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kiTWpqgoE1kz"
      },
      "source": [
        "def subsequence_generator(sequence, pad_vector, to_generate=10, length=120):\n",
        "    start_idx = np.random.randint(0, len(sequence) - 1, size=to_generate)\n",
        "    ret = np.tile(pad_vector, (to_generate, length, 1))\n",
        "\n",
        "    for i, idx in enumerate(start_idx):\n",
        "        subsequence = sequence[idx:]\n",
        "        ret[i, 0:subsequence.shape[0]] = subsequence[:min(len(subsequence), length)]\n",
        "\n",
        "    return ret"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yQQDPjlZE2FU"
      },
      "source": [
        "class TransactionDataset(Dataset):\n",
        "    def __init__(self, data, num_subsequences=10, subsequence_len=120):\n",
        "        self.clients = data['client_id'].unique()\n",
        "        \n",
        "\n",
        "        data['small_group'] = data['small_group'] + 1 # allowing us to use \"0\" as padding\n",
        "\n",
        "        data['trans_date'] = pd.to_datetime(data['trans_date'], unit='d')\n",
        "        data['day_of_week'] = data['trans_date'].dt.dayofweek + 1\n",
        "        data['month'] = data['trans_date'].dt.month\n",
        "        data = data.drop(columns=['trans_date'])\n",
        "\n",
        "        data=data.reindex(columns=['client_id','small_group', 'day_of_week', 'month', 'amount_rur'])\n",
        "\n",
        "\n",
        "        pad_sequence = np.array([0,0,0,0.])\n",
        "\n",
        "        self.data = list()\n",
        "        for i, client in enumerate(self.clients):\n",
        "            sequence = data[data['client_id'] == client].drop(columns=['client_id']).to_numpy()\n",
        "            self.data.append(subsequence_generator(sequence, pad_sequence, num_subsequences, subsequence_len))\n",
        "            \n",
        "        self.data = np.array(self.data)\n",
        "        \n",
        "\n",
        "    def __len__(self):\n",
        "        return self.clients.shape[0]\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        return self.data[idx], self.clients[idx]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KI7P2DhZ6RKS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ca46ed34-6a61-4cf5-94b4-aff6d9345ec2"
      },
      "source": [
        "df = pd.read_csv(path + \"transactions_train.csv\")\n",
        "\n",
        "\n",
        "if LOAD_PICKLED_DATASET:\n",
        "    train_dataset = pickle.load(open(path + \"unsupervised_train_dataset.p\", \"rb\" ))\n",
        "    test_dataset = pickle.load(open(path + \"unsupervised_test_dataset.p\", \"rb\" ))\n",
        "else:\n",
        "    test_clients = np.random.choice(df.client_id.unique(), size=int(0.2 * df.client_id.nunique()), replace=False)\n",
        "    train_clients = np.array(list(set(df.client_id.unique()) - set(test_clients)))\n",
        "    train_idx = ~df.client_id.isin(test_clients)\n",
        "\n",
        "    df_train = df[train_idx].copy()\n",
        "    df_test = df[~train_idx].copy()\n",
        "    train_dataset = TransactionDataset(df_train, \n",
        "                                    num_subsequences=SEQUENCES_PER_CLIENT,\n",
        "                                    subsequence_len=SEQUENCE_LEN)\n",
        "    pickle.dump(train_dataset, open(path + \"unsupervised_train_dataset.p\", \"wb\" ) )\n",
        "\n",
        "    test_dataset = TransactionDataset(df_test, \n",
        "                                    num_subsequences=SEQUENCES_PER_CLIENT,\n",
        "                                    subsequence_len=SEQUENCE_LEN)\n",
        "    pickle.dump(test_dataset, open(path + \"unsupervised_test_dataset.p\", \"wb\" ) )\n",
        "\n",
        "\n",
        "train_dataloader = DataLoader(train_dataset, batch_size=BATCH_SIZE, shuffle=True)\n",
        "test_dataloader = DataLoader(test_dataset, batch_size=BATCH_SIZE, shuffle=True)\n",
        "train_sampler = samplers.MPerClassSampler(train_dataset.clients, 1, batch_size=BATCH_SIZE, length_before_new_iter=len(train_dataset))\n",
        "test_sampler = samplers.MPerClassSampler(test_dataset.clients, 1, batch_size=BATCH_SIZE, length_before_new_iter=len(test_dataset))\n",
        "\n",
        "print(len(train_dataset), len(test_dataset))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "24000 6000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vyg5sFIo6hwR"
      },
      "source": [
        "# Encoder\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B3k--jciE4CT"
      },
      "source": [
        "class EventEncoder(nn.Module):\n",
        "    def __init__(self, num_features, cat_features, cat_vocab_sizes, out_dim, emb_dim=32):\n",
        "        super(EventEncoder, self).__init__()\n",
        "\n",
        "        self.initial_norm = nn.BatchNorm1d(num_features, affine=False)\n",
        "\n",
        "        # len(nn.ModuleList) is a bit wacky, so we'll store the size just in case\n",
        "        self.embeddings = nn.ModuleList() \n",
        "        self.cat_features = cat_features \n",
        "        for i in range(cat_features):\n",
        "            self.embeddings.append(nn.Embedding(cat_vocab_sizes[i], emb_dim, padding_idx=0))\n",
        "\n",
        "        cur_size = emb_dim * cat_features + num_features\n",
        "        self.process = nn.Sequential(\n",
        "            nn.Linear(cur_size, out_dim),\n",
        "            nn.Sigmoid(),\n",
        "            nn.BatchNorm1d(out_dim),\n",
        "        )\n",
        "\n",
        "    def forward(self, numerical, categorical):\n",
        "        concat_input = self.initial_norm(numerical)\n",
        "\n",
        "        for i in range(self.cat_features):\n",
        "            embedded_cat = self.embeddings[i](categorical[:,i])\n",
        "            concat_input = torch.cat((concat_input, embedded_cat), 1)\n",
        "        out = self.process(concat_input)\n",
        "        \n",
        "        return out"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3iNS70SYE4yd"
      },
      "source": [
        "class SequenceEncoder(nn.Module):\n",
        "    def __init__(self, num_features, cat_features, cat_vocab_sizes, sequence_len, event_dim=128, hidden_size=128, event_emb_dim=32):\n",
        "\n",
        "        super(SequenceEncoder, self).__init__()\n",
        "\n",
        "        self.num_features = num_features\n",
        "        self.cat_features = cat_features\n",
        "        self.sequence_len = sequence_len\n",
        "        self.event_dim = event_dim\n",
        "\n",
        "        self.event_encoder = EventEncoder(num_features, cat_features, cat_vocab_sizes, event_dim, emb_dim=event_emb_dim)\n",
        "\n",
        "        self.rnn = nn.GRU(event_dim, hidden_size, batch_first=True)\n",
        "\n",
        "                     \n",
        "    def forward(self, numerical, categorical):\n",
        "        # numerical of size (batch_size, num_of_sequences, sequence_len, num_features)\n",
        "        # categorical of size (batch_size, num_of_sequences, sequence_len, cat_features)\n",
        "\n",
        "        numerical = numerical.view(-1, self.num_features)\n",
        "        categorical = categorical.view(-1, self.cat_features)\n",
        "        \n",
        "        # receiving batch_size x num_of_sequences x sequence_len x event_dim events\n",
        "        events = self.event_encoder(numerical, categorical)\n",
        "        events = events.view(-1, self.sequence_len, self.event_dim)\n",
        "\n",
        "        #taking the output of the last cell        \n",
        "        out = self.rnn(events)[0][:,-1,:]\n",
        "        \n",
        "        return out"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lvDtbwMcE7GH"
      },
      "source": [
        "encoder = SequenceEncoder(num_features=1, \n",
        "                          cat_features=3, \n",
        "                          cat_vocab_sizes=[df['small_group'].max() + 2, 8, 13], # +1 for padding \n",
        "                          sequence_len=SEQUENCE_LEN, \n",
        "                          event_dim=EVENT_ENCODING_DIMENSIONS, \n",
        "                          hidden_size=RNN_HIDDEN_STATE_DIMENSIONS,\n",
        "                          event_emb_dim=EVENT_EMBEDDING_DIMENSIONS)\n",
        "\n",
        "encoder.to(device)\n",
        "\n",
        "optimizer = torch.optim.AdamW(encoder.parameters(), lr=LEARNING_RATE)\n",
        "\n",
        "reduce_func = reducers.AvgNonZeroReducer()\n",
        "distance_func = distances.LpDistance()\n",
        "loss_func = losses.TripletMarginLoss(margin=0.3, distance = distance_func, reducer = reduce_func)\n",
        "miner_func = miners.TripletMarginMiner(margin=0.3, distance = distance_func, type_of_triplets=\"semihard\")\n",
        "\n",
        "categorical_mask = [True, True, True, False]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "kAoqrOzcJt-Y",
        "outputId": "56113edc-1a75-4ead-d760-7b0c373bde1c"
      },
      "source": [
        "train_loss_history = list()\n",
        "val_loss_history = list()\n",
        "\n",
        "\n",
        "for epoch in tqdm(range(EPOCHS)):\n",
        "    cur_losses = list()\n",
        "\n",
        "    encoder.train()\n",
        "    for batch_idx, (sequences, labels) in enumerate(train_dataloader):\n",
        "        \n",
        "        numerical = sequences[:,:,:,np.logical_not(categorical_mask)].to(device)\n",
        "        categorical = sequences[:,:,:,categorical_mask].to(device)\n",
        "        \n",
        "        labels = np.repeat(labels, SEQUENCES_PER_CLIENT).to(device)\n",
        "\n",
        "        embeddings = encoder(numerical.float(), categorical.int()).to(device)\n",
        "        mining = miner_func(embeddings, labels)\n",
        "        train_loss = loss_func(embeddings, labels, mining)\n",
        "\n",
        "        train_loss.backward()\n",
        "        optimizer.step()\n",
        "        optimizer.zero_grad()\n",
        "        cur_losses.append(train_loss.item())\n",
        "\n",
        "    train_loss_history.append(np.mean(cur_losses))\n",
        "\n",
        "    cur_losses = list()\n",
        "\n",
        "    encoder.eval()\n",
        "    for batch_idx, (sequences, labels) in enumerate(test_dataloader):\n",
        "\n",
        "        numerical = sequences[:,:,:,np.logical_not(categorical_mask)].to(device)\n",
        "        categorical = sequences[:,:,:,categorical_mask].to(device)\n",
        "\n",
        "        labels = np.repeat(labels, SEQUENCES_PER_CLIENT).to(device)\n",
        "\n",
        "        with torch.no_grad():\n",
        "            embeddings = encoder(numerical.float(), categorical.int()).to(device)\n",
        "            mining = miner_func(embeddings, labels)\n",
        "            val_loss = loss_func(embeddings, labels, mining)\n",
        "\n",
        "            cur_losses.append(val_loss.item())\n",
        "\n",
        "    val_loss_history.append(np.mean(cur_losses))\n",
        "\n",
        "    print(\"\\nEpoch {}; Train loss = {}; Val loss = {}\".format(epoch+1, train_loss_history[-1], val_loss_history[-1]))\n",
        "\n",
        "    torch.save({\n",
        "            'epoch': epoch,\n",
        "            'model_state_dict': encoder.state_dict(),\n",
        "            'optimizer_state_dict': optimizer.state_dict(),\n",
        "            'loss': train_loss_history[-1],\n",
        "            }, CHECKPOINT_PATH)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r  0%|          | 0/64 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 1; Train loss = 0.16625371337571043; Val loss = 0.16228689221625633\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "  3%|▎         | 2/64 [04:37<2:24:00, 139.36s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 2; Train loss = 0.15947269537347428; Val loss = 0.1565939761856769\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r  5%|▍         | 3/64 [06:54<2:20:51, 138.55s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 3; Train loss = 0.1548767070028376; Val loss = 0.1542909208764421\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r  6%|▋         | 4/64 [09:10<2:17:41, 137.70s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 4; Train loss = 0.15391254813430158; Val loss = 0.15332157148959788\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r  8%|▊         | 5/64 [11:25<2:14:49, 137.12s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 5; Train loss = 0.15258044947652108; Val loss = 0.15161607810791503\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r  9%|▉         | 6/64 [13:42<2:12:22, 136.93s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 6; Train loss = 0.148026168980497; Val loss = 0.14355749303990223\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 11%|█         | 7/64 [15:57<2:09:40, 136.50s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 7; Train loss = 0.14141962764428018; Val loss = 0.1404232588854242\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 12%|█▎        | 8/64 [18:12<2:07:01, 136.09s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 8; Train loss = 0.13931827167881297; Val loss = 0.13697667356501234\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 14%|█▍        | 9/64 [20:28<2:04:28, 135.80s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 9; Train loss = 0.13652995981751603; Val loss = 0.13562903505690554\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 16%|█▌        | 10/64 [22:42<2:01:44, 135.27s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 10; Train loss = 0.13565169235176228; Val loss = 0.13572527912068874\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 17%|█▋        | 11/64 [24:55<1:59:05, 134.81s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 11; Train loss = 0.13480601309144752; Val loss = 0.13378043282539287\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 19%|█▉        | 12/64 [27:09<1:56:28, 134.39s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 12; Train loss = 0.1321726016541745; Val loss = 0.1308947062872826\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 20%|██        | 13/64 [29:22<1:53:49, 133.91s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 13; Train loss = 0.13063627362568328; Val loss = 0.12973396067923687\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 22%|██▏       | 14/64 [31:35<1:51:24, 133.70s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 14; Train loss = 0.12830507798873364; Val loss = 0.12752684855714758\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 23%|██▎       | 15/64 [33:47<1:48:49, 133.25s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 15; Train loss = 0.127288481220603; Val loss = 0.12673995532888047\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 25%|██▌       | 16/64 [35:59<1:46:18, 132.88s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 16; Train loss = 0.1265690441223535; Val loss = 0.12575725259933065\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 27%|██▋       | 17/64 [38:11<1:43:58, 132.74s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 17; Train loss = 0.12559227570098766; Val loss = 0.12449126183352571\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 28%|██▊       | 18/64 [40:24<1:41:37, 132.56s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 18; Train loss = 0.12410635505109391; Val loss = 0.12326602336574108\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 30%|██▉       | 19/64 [42:35<1:39:17, 132.39s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 19; Train loss = 0.122698746264932; Val loss = 0.12175938915064995\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 31%|███▏      | 20/64 [44:47<1:36:55, 132.16s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 20; Train loss = 0.12190166242579197; Val loss = 0.1214027426978375\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 33%|███▎      | 21/64 [46:59<1:34:36, 132.02s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 21; Train loss = 0.12114436781786858; Val loss = 0.12089819175765869\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 34%|███▍      | 22/64 [49:10<1:32:17, 131.84s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 22; Train loss = 0.12060888682274108; Val loss = 0.12034695595502853\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 36%|███▌      | 23/64 [51:22<1:30:00, 131.72s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 23; Train loss = 0.1199673002863184; Val loss = 0.11997639752448873\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 38%|███▊      | 24/64 [53:33<1:27:39, 131.49s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 24; Train loss = 0.11939102411270142; Val loss = 0.11906284458459691\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 39%|███▉      | 25/64 [55:44<1:25:22, 131.36s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 25; Train loss = 0.11882944971798583; Val loss = 0.11826733737549883\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 41%|████      | 26/64 [57:55<1:23:09, 131.30s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 26; Train loss = 0.1184467797742245; Val loss = 0.11830399740249553\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 42%|████▏     | 27/64 [1:00:06<1:20:51, 131.13s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 27; Train loss = 0.1180110919348737; Val loss = 0.11795518721671815\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 44%|████▍     | 28/64 [1:02:16<1:18:35, 130.99s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 28; Train loss = 0.11774782792843402; Val loss = 0.11732610045595372\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 45%|████▌     | 29/64 [1:04:27<1:16:20, 130.86s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 29; Train loss = 0.11738292702493515; Val loss = 0.11753104595427817\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 47%|████▋     | 30/64 [1:06:37<1:14:06, 130.77s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 30; Train loss = 0.11710233645553285; Val loss = 0.11756765446130266\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 48%|████▊     | 31/64 [1:08:48<1:11:50, 130.63s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 31; Train loss = 0.11693774163722992; Val loss = 0.11706097852042381\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 50%|█████     | 32/64 [1:10:58<1:09:36, 130.50s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 32; Train loss = 0.11663260564525077; Val loss = 0.11718289839460495\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 52%|█████▏    | 33/64 [1:13:08<1:07:23, 130.42s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 33; Train loss = 0.11635251446290219; Val loss = 0.11643129904219444\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 53%|█████▎    | 34/64 [1:15:18<1:05:11, 130.38s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 34; Train loss = 0.11613090213467465; Val loss = 0.11643382383787886\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 55%|█████▍    | 35/64 [1:17:29<1:03:00, 130.35s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 35; Train loss = 0.11583422421616443; Val loss = 0.11630770817716071\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 56%|█████▋    | 36/64 [1:19:39<1:00:48, 130.31s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 36; Train loss = 0.11565923124076204; Val loss = 0.11551093516197611\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 58%|█████▊    | 37/64 [1:21:49<58:38, 130.30s/it]  "
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 37; Train loss = 0.11543807490391934; Val loss = 0.11567193491661802\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 59%|█████▉    | 38/64 [1:23:59<56:27, 130.28s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 38; Train loss = 0.11515078598514517; Val loss = 0.11567733151481506\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 61%|██████    | 39/64 [1:26:10<54:17, 130.29s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 39; Train loss = 0.11491943444026277; Val loss = 0.11592633546666896\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\r 62%|██████▎   | 40/64 [1:28:20<52:06, 130.26s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 40; Train loss = 0.11468583916095977; Val loss = 0.11552763540059963\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-12-58fd8675355c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     18\u001b[0m         \u001b[0mtrain_loss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mloss_func\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0membeddings\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmining\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     19\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 20\u001b[0;31m         \u001b[0mtrain_loss\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     21\u001b[0m         \u001b[0moptimizer\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     22\u001b[0m         \u001b[0moptimizer\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mzero_grad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/torch/tensor.py\u001b[0m in \u001b[0;36mbackward\u001b[0;34m(self, gradient, retain_graph, create_graph, inputs)\u001b[0m\n\u001b[1;32m    243\u001b[0m                 \u001b[0mcreate_graph\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcreate_graph\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    244\u001b[0m                 inputs=inputs)\n\u001b[0;32m--> 245\u001b[0;31m         \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mautograd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgradient\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mretain_graph\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcreate_graph\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minputs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    246\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    247\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mregister_hook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/torch/autograd/__init__.py\u001b[0m in \u001b[0;36mbackward\u001b[0;34m(tensors, grad_tensors, retain_graph, create_graph, grad_variables, inputs)\u001b[0m\n\u001b[1;32m    145\u001b[0m     Variable._execution_engine.run_backward(\n\u001b[1;32m    146\u001b[0m         \u001b[0mtensors\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgrad_tensors_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mretain_graph\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcreate_graph\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minputs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 147\u001b[0;31m         allow_unreachable=True, accumulate_grad=True)  # allow_unreachable flag\n\u001b[0m\u001b[1;32m    148\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    149\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BSL5YMnw_fiO",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 769
        },
        "outputId": "eb1724fb-8890-480b-8268-8c7c1d268218"
      },
      "source": [
        "plt.figure(figsize=(20,13))\n",
        "\n",
        "plt.plot(train_loss_history, label='Training loss')\n",
        "plt.plot(val_loss_history, label='Validation loss')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x936 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VLXbzqCZ_nni"
      },
      "source": [
        "pickle.dump(embeddings, open(path + \"embeddings.p\", \"wb\" ) )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WtjmBD9_Cw2E"
      },
      "source": [
        "# Classification"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "auSZsCcjiqOo"
      },
      "source": [
        "target = pd.read_csv(path + \"train_target.csv\")  "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q6E5HXGe-WOo"
      },
      "source": [
        "target = pd.read_csv(path + \"train_target.csv\")  \n",
        "client_target_dict = pd.Series(target.bins.values, index=target.client_id).to_dict()\n",
        "age_target = np.vectorize(client_target_dict.get)(train_dataset.clients)\n",
        "\n",
        "\n",
        "predict_dataloader = DataLoader(train_dataset, batch_size=BATCH_SIZE, sampler=SequentialSampler(train_dataset))\n",
        "\n",
        "pred = np.zeros(shape=(len(train_dataset), SEQUENCES_PER_CLIENT, RNN_HIDDEN_STATE_DIMENSIONS))\n",
        "client_targets = np.zeros(shape=len(train_dataset))\n",
        "age_targets = np.zeros(shape=len(train_dataset))\n",
        "\n",
        "encoder.eval()\n",
        "for batch_idx, (sequences, labels) in enumerate(predict_dataloader):\n",
        "    numerical = sequences[:,:,:,np.logical_not(categorical_mask)].to(device)\n",
        "    categorical = sequences[:,:,:,categorical_mask].to(device)\n",
        "    labels = labels.numpy()\n",
        "\n",
        "    with torch.no_grad():\n",
        "        embeddings = encoder(numerical.float(), categorical.int()).to(device)\n",
        "    embeddings = embeddings.view(-1, SEQUENCES_PER_CLIENT, RNN_HIDDEN_STATE_DIMENSIONS)\n",
        "\n",
        "    pred[batch_idx * BATCH_SIZE : (batch_idx + 1) * BATCH_SIZE, :, :] = embeddings.cpu().numpy()\n",
        "    client_targets[batch_idx * BATCH_SIZE : (batch_idx + 1) * BATCH_SIZE] = labels\n",
        "    age_targets[batch_idx * BATCH_SIZE : (batch_idx + 1) * BATCH_SIZE] = np.vectorize(client_target_dict.get)(labels)\n",
        "\n",
        "\n",
        "train_pred = pred.reshape(-1, RNN_HIDDEN_STATE_DIMENSIONS)\n",
        "train_client_targets = np.repeat(client_targets, SEQUENCES_PER_CLIENT)\n",
        "train_age_targets = np.repeat(age_targets, SEQUENCES_PER_CLIENT)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ph7hlDEVAY8k"
      },
      "source": [
        "age_target = np.vectorize(client_target_dict.get)(test_dataset.clients)\n",
        "\n",
        "predict_dataloader = DataLoader(test_dataset, batch_size=BATCH_SIZE, sampler=SequentialSampler(test_dataset))\n",
        "\n",
        "pred = np.zeros(shape=(len(test_dataset), SEQUENCES_PER_CLIENT, RNN_HIDDEN_STATE_DIMENSIONS))\n",
        "client_targets = np.zeros(shape=len(test_dataset))\n",
        "age_targets = np.zeros(shape=len(test_dataset))\n",
        "\n",
        "encoder.eval()\n",
        "for batch_idx, (sequences, labels) in enumerate(predict_dataloader):\n",
        "    numerical = sequences[:,:,:,np.logical_not(categorical_mask)].to(device)\n",
        "    categorical = sequences[:,:,:,categorical_mask].to(device)\n",
        "    labels = labels.numpy()\n",
        "\n",
        "    with torch.no_grad():\n",
        "        embeddings = encoder(numerical.float(), categorical.int()).to(device)\n",
        "    embeddings = embeddings.view(-1, SEQUENCES_PER_CLIENT, RNN_HIDDEN_STATE_DIMENSIONS)\n",
        "\n",
        "    pred[batch_idx * BATCH_SIZE : (batch_idx + 1) * BATCH_SIZE, :, :] = embeddings.cpu().numpy()\n",
        "    client_targets[batch_idx * BATCH_SIZE : (batch_idx + 1) * BATCH_SIZE] = labels\n",
        "    age_targets[batch_idx * BATCH_SIZE : (batch_idx + 1) * BATCH_SIZE] = np.vectorize(client_target_dict.get)(labels)\n",
        "\n",
        "\n",
        "test_pred = pred.reshape(-1, RNN_HIDDEN_STATE_DIMENSIONS)\n",
        "test_client_targets = np.repeat(client_targets, SEQUENCES_PER_CLIENT)\n",
        "test_age_targets = np.repeat(age_targets, SEQUENCES_PER_CLIENT)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ja_5lyQCBh_R"
      },
      "source": [
        "pred = np.concatenate([train_pred, test_pred])\n",
        "client_targets = np.concatenate([train_client_targets, test_client_targets])\n",
        "age_targets = np.concatenate([train_age_targets, test_age_targets])\n",
        "\n",
        "pickle.dump(pred, open(path + \"pred.p\", \"wb\" ) )\n",
        "pickle.dump(client_targets, open(path + \"client_targets.p\", \"wb\" ) )\n",
        "pickle.dump(age_targets, open(path + \"age_targets.p\", \"wb\" ) )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S2HCMtXFiVzV"
      },
      "source": [
        "pred = pickle.load(open(path + \"pred.p\", \"rb\" ))\n",
        "client_targets = pickle.load(open(path + \"client_targets.p\", \"rb\" ))\n",
        "age_targets = pickle.load(open(path + \"age_targets.p\", \"rb\" ))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0Xz50CYWC0na"
      },
      "source": [
        "X = torch.Tensor(pred)\n",
        "y = torch.Tensor(age_targets)\n",
        "\n",
        "classifier_dataset = TensorDataset(X, y)\n",
        "train_classifier_dataset, test_classifier_dataset = random_split(classifier_dataset, [int(len(classifier_dataset) * 0.8), len(classifier_dataset) - (int(len(classifier_dataset) * 0.8))])\n",
        "train_classifier_dataloader = DataLoader(train_classifier_dataset, batch_size=CLASSIFIER_BATCH_SIZE, shuffle=True)\n",
        "test_classifier_dataloader = DataLoader(test_classifier_dataset, batch_size=CLASSIFIER_BATCH_SIZE, shuffle=True)\n",
        "\n",
        "num_classes = target['bins'].nunique()\n",
        "\n",
        "classifier = nn.Sequential(\n",
        "    nn.Linear(RNN_HIDDEN_STATE_DIMENSIONS, CLASSIFIER_LAYER1_DIM),\n",
        "    nn.Sigmoid(),\n",
        "    nn.BatchNorm1d(CLASSIFIER_LAYER1_DIM),\n",
        "    nn.Linear(CLASSIFIER_LAYER1_DIM, CLASSIFIER_LAYER2_DIM),\n",
        "    nn.Sigmoid(),\n",
        "    nn.BatchNorm1d(CLASSIFIER_LAYER2_DIM),\n",
        "    nn.Linear(CLASSIFIER_LAYER2_DIM, num_classes),\n",
        ")\n",
        "classifier.to(device)\n",
        "\n",
        "optimizer = torch.optim.AdamW(classifier.parameters(), lr=CLASSIFIER_LEARNING_RATE)\n",
        "\n",
        "loss_func = nn.CrossEntropyLoss()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TV5jvDMrF2pY",
        "outputId": "7de30c03-dd0f-4e60-db63-2322458a3049"
      },
      "source": [
        "train_loss_history = list()\n",
        "val_loss_history = list()\n",
        "\n",
        "train_accuracy_history = list()\n",
        "val_accuracy_history = list()\n",
        "\n",
        "\n",
        "for epoch in tqdm(range(CLASSIFIER_EPOCHS)):\n",
        "\n",
        "    classifier.train()\n",
        "    cur_losses = list()\n",
        "    cur_accuracy = list()\n",
        "\n",
        "    for batch_idx, (sequences, labels) in enumerate(train_classifier_dataloader):\n",
        "\n",
        "        sequences = sequences.to(device)\n",
        "        labels = labels.long().to(device)\n",
        "        \n",
        "        predictions = classifier(sequences).to(device)\n",
        "        train_loss = loss_func(predictions, labels)\n",
        "\n",
        "        train_loss.backward()\n",
        "        optimizer.step()\n",
        "        optimizer.zero_grad()\n",
        "        cur_losses.append(train_loss.item())\n",
        "        cur_accuracy.append(accuracy_score(torch.argmax(predictions.cpu(), dim=1), labels.cpu()))\n",
        "\n",
        "    train_loss_history.append(np.mean(cur_losses))\n",
        "    train_accuracy_history.append(np.mean(cur_accuracy))\n",
        "\n",
        "    classifier.eval()\n",
        "    cur_losses = list()\n",
        "    cur_accuracy = list()\n",
        "\n",
        "    for batch_idx, (sequences, labels) in enumerate(test_classifier_dataloader):\n",
        " \n",
        "        sequences = sequences.to(device)\n",
        "        labels = labels.long().to(device)\n",
        "        with torch.no_grad():\n",
        "            predictions = classifier(sequences).to(device)\n",
        "            test_loss = loss_func(predictions, labels)\n",
        "\n",
        "            cur_losses.append(test_loss.item())\n",
        "            cur_accuracy.append(accuracy_score(torch.argmax(predictions.cpu(), dim=1), labels.cpu()))\n",
        "\n",
        "    \n",
        "    val_loss_history.append(np.mean(cur_losses))\n",
        "    val_accuracy_history.append(np.mean(cur_accuracy))\n",
        "\n",
        "    print(\"\\nEpoch {}\\n Train Loss = {}\\n Train Accuracy = {}\\n Validation Loss = {}\\n Validation Accuracy = {}\".format(\n",
        "        epoch, train_loss_history[-1], train_accuracy_history[-1], val_loss_history[-1], val_accuracy_history[-1]))\n",
        "\n",
        "    torch.save({\n",
        "            'epoch': epoch,\n",
        "            'model_state_dict': classifier.state_dict(),\n",
        "            'optimizer_state_dict': optimizer.state_dict(),\n",
        "            'loss': train_loss_history[-1],\n",
        "            }, CHECKPOINT_PATH)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  0%|          | 0/64 [00:00<?, ?it/s]\u001b[A\n",
            "  2%|▏         | 1/64 [00:04<04:25,  4.21s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 0\n",
            " Train Loss = 0.9970711440419846\n",
            " Train Accuracy = 0.5490946495202559\n",
            " Validation Loss = 0.9826903490310019\n",
            " Validation Accuracy = 0.5544824911347519\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  3%|▎         | 2/64 [00:08<04:18,  4.18s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 1\n",
            " Train Loss = 0.9783536301874148\n",
            " Train Accuracy = 0.5583522121535182\n",
            " Validation Loss = 0.9799205031800777\n",
            " Validation Accuracy = 0.5568484042553191\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  5%|▍         | 3/64 [00:12<04:14,  4.18s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 2\n",
            " Train Loss = 0.9735069977703379\n",
            " Train Accuracy = 0.5607051239339019\n",
            " Validation Loss = 0.9763268445400481\n",
            " Validation Accuracy = 0.5564605496453902\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  6%|▋         | 4/64 [00:16<04:08,  4.14s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 3\n",
            " Train Loss = 0.9705911586279554\n",
            " Train Accuracy = 0.5623709021855011\n",
            " Validation Loss = 0.9753141342325413\n",
            " Validation Accuracy = 0.558771054964539\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  8%|▊         | 5/64 [00:20<04:03,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 4\n",
            " Train Loss = 0.9683480039080069\n",
            " Train Accuracy = 0.5636618803304904\n",
            " Validation Loss = 0.9741492717824084\n",
            " Validation Accuracy = 0.5589926861702128\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "  9%|▉         | 6/64 [00:24<03:59,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 5\n",
            " Train Loss = 0.9666410654719705\n",
            " Train Accuracy = 0.5637326759061834\n",
            " Validation Loss = 0.971271393907831\n",
            " Validation Accuracy = 0.5610926418439716\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 11%|█         | 7/64 [00:28<03:54,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 6\n",
            " Train Loss = 0.964434119620557\n",
            " Train Accuracy = 0.566431236673774\n",
            " Validation Loss = 0.9701024717472969\n",
            " Validation Accuracy = 0.560832225177305\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 12%|█▎        | 8/64 [00:32<03:50,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 7\n",
            " Train Loss = 0.9622667404507269\n",
            " Train Accuracy = 0.5674640191897654\n",
            " Validation Loss = 0.9671857582761886\n",
            " Validation Accuracy = 0.5626717641843971\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 14%|█▍        | 9/64 [00:37<03:45,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 8\n",
            " Train Loss = 0.9606235435903708\n",
            " Train Accuracy = 0.5685551039445629\n",
            " Validation Loss = 0.9653246727395565\n",
            " Validation Accuracy = 0.5634807180851064\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 16%|█▌        | 10/64 [00:41<03:42,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 9\n",
            " Train Loss = 0.9590738432239622\n",
            " Train Accuracy = 0.5686217350746269\n",
            " Validation Loss = 0.9639250856764773\n",
            " Validation Accuracy = 0.5651540336879434\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 17%|█▋        | 11/64 [00:45<03:38,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 10\n",
            " Train Loss = 0.9576277146969777\n",
            " Train Accuracy = 0.5697794509594882\n",
            " Validation Loss = 0.9637178532620694\n",
            " Validation Accuracy = 0.5636303191489361\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 19%|█▉        | 12/64 [00:49<03:36,  4.16s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 11\n",
            " Train Loss = 0.9556339644928222\n",
            " Train Accuracy = 0.5713661047441365\n",
            " Validation Loss = 0.9622280661095964\n",
            " Validation Accuracy = 0.5647717198581561\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 20%|██        | 13/64 [00:53<03:32,  4.18s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 12\n",
            " Train Loss = 0.954130106833952\n",
            " Train Accuracy = 0.5718741671108742\n",
            " Validation Loss = 0.960937100014788\n",
            " Validation Accuracy = 0.5666666666666668\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 22%|██▏       | 14/64 [00:58<03:29,  4.20s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 13\n",
            " Train Loss = 0.9524477946478674\n",
            " Train Accuracy = 0.5728236607142857\n",
            " Validation Loss = 0.9605805303188081\n",
            " Validation Accuracy = 0.5672761524822694\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 23%|██▎       | 15/64 [01:02<03:24,  4.18s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 14\n",
            " Train Loss = 0.9506062783920435\n",
            " Train Accuracy = 0.5739355676972282\n",
            " Validation Loss = 0.9592494733790133\n",
            " Validation Accuracy = 0.5670378989361702\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 25%|██▌       | 16/64 [01:06<03:20,  4.17s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 15\n",
            " Train Loss = 0.9486521716311034\n",
            " Train Accuracy = 0.5753139992004265\n",
            " Validation Loss = 0.9570856211033273\n",
            " Validation Accuracy = 0.5691600177304965\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 27%|██▋       | 17/64 [01:10<03:15,  4.15s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 16\n",
            " Train Loss = 0.9470953051088207\n",
            " Train Accuracy = 0.5765050306503199\n",
            " Validation Loss = 0.9567012239009776\n",
            " Validation Accuracy = 0.5671708776595744\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 28%|██▊       | 18/64 [01:14<03:10,  4.14s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 17\n",
            " Train Loss = 0.9448609779765611\n",
            " Train Accuracy = 0.5777418710021321\n",
            " Validation Loss = 0.9555695122860848\n",
            " Validation Accuracy = 0.5687222960992907\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 30%|██▉       | 19/64 [01:18<03:06,  4.14s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 18\n",
            " Train Loss = 0.9435734826364497\n",
            " Train Accuracy = 0.5788246268656716\n",
            " Validation Loss = 0.9544270515441895\n",
            " Validation Accuracy = 0.5698747783687944\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 31%|███▏      | 20/64 [01:22<03:02,  4.14s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 19\n",
            " Train Loss = 0.9414249157219299\n",
            " Train Accuracy = 0.5795200892857143\n",
            " Validation Loss = 0.9516712315539096\n",
            " Validation Accuracy = 0.5720356826241135\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 33%|███▎      | 21/64 [01:26<02:57,  4.14s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 20\n",
            " Train Loss = 0.9388486995244585\n",
            " Train Accuracy = 0.5812649920042644\n",
            " Validation Loss = 0.9524404322847407\n",
            " Validation Accuracy = 0.5720578457446809\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 34%|███▍      | 22/64 [01:31<02:53,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 21\n",
            " Train Loss = 0.9368101805766255\n",
            " Train Accuracy = 0.5833138992537313\n",
            " Validation Loss = 0.950134358000248\n",
            " Validation Accuracy = 0.572905585106383\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 36%|███▌      | 23/64 [01:35<02:49,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 22\n",
            " Train Loss = 0.93454881695542\n",
            " Train Accuracy = 0.5845798907249466\n",
            " Validation Loss = 0.9506814401200477\n",
            " Validation Accuracy = 0.571985815602837\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 38%|███▊      | 24/64 [01:39<02:44,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 23\n",
            " Train Loss = 0.9324393163738983\n",
            " Train Accuracy = 0.5864872068230277\n",
            " Validation Loss = 0.9494346519733997\n",
            " Validation Accuracy = 0.5724567819148936\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 39%|███▉      | 25/64 [01:43<02:40,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 24\n",
            " Train Loss = 0.9302325678278388\n",
            " Train Accuracy = 0.5864289045842217\n",
            " Validation Loss = 0.9462512794961321\n",
            " Validation Accuracy = 0.5749058067375886\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 41%|████      | 26/64 [01:47<02:36,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 25\n",
            " Train Loss = 0.9273626506964027\n",
            " Train Accuracy = 0.5887693230277186\n",
            " Validation Loss = 0.9456034962167131\n",
            " Validation Accuracy = 0.5756316489361702\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 42%|████▏     | 27/64 [01:51<02:31,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 26\n",
            " Train Loss = 0.9252395230823997\n",
            " Train Accuracy = 0.5891399586886994\n",
            " Validation Loss = 0.9440203019913207\n",
            " Validation Accuracy = 0.5752327127659574\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 44%|████▍     | 28/64 [01:55<02:28,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 27\n",
            " Train Loss = 0.9230089279126003\n",
            " Train Accuracy = 0.5925964485607675\n",
            " Validation Loss = 0.9426072310894094\n",
            " Validation Accuracy = 0.5758366578014184\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 45%|████▌     | 29/64 [01:59<02:23,  4.09s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 28\n",
            " Train Loss = 0.9203497645443183\n",
            " Train Accuracy = 0.5935584355010661\n",
            " Validation Loss = 0.9416884883921197\n",
            " Validation Accuracy = 0.5775265957446809\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 47%|████▋     | 30/64 [02:03<02:19,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 29\n",
            " Train Loss = 0.917577016582367\n",
            " Train Accuracy = 0.5953491471215352\n",
            " Validation Loss = 0.9395575069366617\n",
            " Validation Accuracy = 0.5788674645390072\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 48%|████▊     | 31/64 [02:08<02:15,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 30\n",
            " Train Loss = 0.9152195289699253\n",
            " Train Accuracy = 0.5966068097014925\n",
            " Validation Loss = 0.9393933646222379\n",
            " Validation Accuracy = 0.5778091755319149\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 50%|█████     | 32/64 [02:12<02:11,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 31\n",
            " Train Loss = 0.9128011813931374\n",
            " Train Accuracy = 0.5983350546375267\n",
            " Validation Loss = 0.9364420162870529\n",
            " Validation Accuracy = 0.5810671542553192\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 52%|█████▏    | 33/64 [02:16<02:07,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 32\n",
            " Train Loss = 0.9098432382413828\n",
            " Train Accuracy = 0.5997384728144989\n",
            " Validation Loss = 0.9352703023464122\n",
            " Validation Accuracy = 0.5802360372340426\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 53%|█████▎    | 34/64 [02:20<02:03,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 33\n",
            " Train Loss = 0.9070446551608633\n",
            " Train Accuracy = 0.6005588686034116\n",
            " Validation Loss = 0.9347291061218749\n",
            " Validation Accuracy = 0.5834164450354609\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 55%|█████▍    | 35/64 [02:24<01:59,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 34\n",
            " Train Loss = 0.9045837956196718\n",
            " Train Accuracy = 0.6031907982409381\n",
            " Validation Loss = 0.9339395698080671\n",
            " Validation Accuracy = 0.5810449911347518\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 56%|█████▋    | 36/64 [02:28<01:55,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 35\n",
            " Train Loss = 0.9020774999279966\n",
            " Train Accuracy = 0.6051439232409381\n",
            " Validation Loss = 0.9341913831994889\n",
            " Validation Accuracy = 0.5826628989361702\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 58%|█████▊    | 37/64 [02:32<01:50,  4.09s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 36\n",
            " Train Loss = 0.8986579362771658\n",
            " Train Accuracy = 0.6067222481343284\n",
            " Validation Loss = 0.9312961806642248\n",
            " Validation Accuracy = 0.582673980496454\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 59%|█████▉    | 38/64 [02:36<01:46,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 37\n",
            " Train Loss = 0.8969790293082499\n",
            " Train Accuracy = 0.6080173907249466\n",
            " Validation Loss = 0.930990628485984\n",
            " Validation Accuracy = 0.584158909574468\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 61%|██████    | 39/64 [02:40<01:43,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 38\n",
            " Train Loss = 0.8935534012978519\n",
            " Train Accuracy = 0.609329191098081\n",
            " Validation Loss = 0.9296771757146145\n",
            " Validation Accuracy = 0.584513519503546\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 62%|██████▎   | 40/64 [02:45<01:38,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 39\n",
            " Train Loss = 0.8912722634227037\n",
            " Train Accuracy = 0.6117987073560768\n",
            " Validation Loss = 0.9288286424697714\n",
            " Validation Accuracy = 0.5858543882978723\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 64%|██████▍   | 41/64 [02:49<01:34,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 40\n",
            " Train Loss = 0.8880386041171515\n",
            " Train Accuracy = 0.6123275919509595\n",
            " Validation Loss = 0.9278252150150056\n",
            " Validation Accuracy = 0.5845910904255319\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 66%|██████▌   | 42/64 [02:53<01:30,  4.09s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 41\n",
            " Train Loss = 0.8855346735479481\n",
            " Train Accuracy = 0.6150844549573561\n",
            " Validation Loss = 0.9268950619596116\n",
            " Validation Accuracy = 0.5861979166666668\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 67%|██████▋   | 43/64 [02:57<01:26,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 42\n",
            " Train Loss = 0.8823390368586664\n",
            " Train Accuracy = 0.6162338419509595\n",
            " Validation Loss = 0.9270204361448897\n",
            " Validation Accuracy = 0.5864029255319149\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 69%|██████▉   | 44/64 [03:01<01:21,  4.07s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 43\n",
            " Train Loss = 0.879906302258404\n",
            " Train Accuracy = 0.6180828558102346\n",
            " Validation Loss = 0.924788676931503\n",
            " Validation Accuracy = 0.588530585106383\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 70%|███████   | 45/64 [03:05<01:17,  4.08s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 44\n",
            " Train Loss = 0.8773292934716638\n",
            " Train Accuracy = 0.6195820562366737\n",
            " Validation Loss = 0.9254269186486589\n",
            " Validation Accuracy = 0.5875554078014184\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 72%|███████▏  | 46/64 [03:09<01:13,  4.08s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 45\n",
            " Train Loss = 0.8740107171825254\n",
            " Train Accuracy = 0.6215143590085288\n",
            " Validation Loss = 0.9235157370567322\n",
            " Validation Accuracy = 0.5888076241134752\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 73%|███████▎  | 47/64 [03:13<01:09,  4.09s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 46\n",
            " Train Loss = 0.8715973326138088\n",
            " Train Accuracy = 0.6225596348614072\n",
            " Validation Loss = 0.9234865667972159\n",
            " Validation Accuracy = 0.5892176418439715\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 75%|███████▌  | 48/64 [03:17<01:05,  4.07s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 47\n",
            " Train Loss = 0.8683869513367285\n",
            " Train Accuracy = 0.6234799773454158\n",
            " Validation Loss = 0.9213203962813032\n",
            " Validation Accuracy = 0.5887799202127659\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 77%|███████▋  | 49/64 [03:21<01:01,  4.08s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 48\n",
            " Train Loss = 0.8660714117957077\n",
            " Train Accuracy = 0.6258120668976546\n",
            " Validation Loss = 0.9234385130253244\n",
            " Validation Accuracy = 0.5908909574468085\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 78%|███████▊  | 50/64 [03:25<00:57,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 49\n",
            " Train Loss = 0.8630637382266364\n",
            " Train Accuracy = 0.6272904450959488\n",
            " Validation Loss = 0.9191175422769912\n",
            " Validation Accuracy = 0.5939771719858157\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 80%|███████▉  | 51/64 [03:30<00:53,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 50\n",
            " Train Loss = 0.8605625279295419\n",
            " Train Accuracy = 0.6281982942430704\n",
            " Validation Loss = 0.9216390726414133\n",
            " Validation Accuracy = 0.5892952127659574\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 81%|████████▏ | 52/64 [03:34<00:49,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 51\n",
            " Train Loss = 0.8577965757867166\n",
            " Train Accuracy = 0.6307635927505331\n",
            " Validation Loss = 0.9202487818738248\n",
            " Validation Accuracy = 0.5910738031914894\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 83%|████████▎ | 53/64 [03:38<00:45,  4.09s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 52\n",
            " Train Loss = 0.8551635549012532\n",
            " Train Accuracy = 0.6320170908848614\n",
            " Validation Loss = 0.9168584630844441\n",
            " Validation Accuracy = 0.5941489361702128\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 84%|████████▍ | 54/64 [03:42<00:40,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 53\n",
            " Train Loss = 0.852899212636419\n",
            " Train Accuracy = 0.6325126599147122\n",
            " Validation Loss = 0.9215655291334112\n",
            " Validation Accuracy = 0.5919437056737588\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 86%|████████▌ | 55/64 [03:46<00:36,  4.08s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 54\n",
            " Train Loss = 0.8494320307840416\n",
            " Train Accuracy = 0.6350821228678039\n",
            " Validation Loss = 0.9194414029730127\n",
            " Validation Accuracy = 0.593074024822695\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 88%|████████▊ | 56/64 [03:50<00:32,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 55\n",
            " Train Loss = 0.8472020231457408\n",
            " Train Accuracy = 0.6361815365138592\n",
            " Validation Loss = 0.9167608918027675\n",
            " Validation Accuracy = 0.5949855939716311\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 89%|████████▉ | 57/64 [03:54<00:28,  4.09s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 56\n",
            " Train Loss = 0.8444760100546677\n",
            " Train Accuracy = 0.6377681902985075\n",
            " Validation Loss = 0.923562633230331\n",
            " Validation Accuracy = 0.5903313386524822\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 91%|█████████ | 58/64 [03:58<00:24,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 57\n",
            " Train Loss = 0.8419274121586448\n",
            " Train Accuracy = 0.6387676572494669\n",
            " Validation Loss = 0.9189922487482112\n",
            " Validation Accuracy = 0.5945921985815603\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 92%|█████████▏| 59/64 [04:02<00:20,  4.13s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 58\n",
            " Train Loss = 0.8388812059659694\n",
            " Train Accuracy = 0.6403085021321961\n",
            " Validation Loss = 0.9180890481522743\n",
            " Validation Accuracy = 0.5937444592198582\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 94%|█████████▍| 60/64 [04:06<00:16,  4.12s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 59\n",
            " Train Loss = 0.8369316645522615\n",
            " Train Accuracy = 0.6420867204157783\n",
            " Validation Loss = 0.9150845563158075\n",
            " Validation Accuracy = 0.5954399379432624\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 95%|█████████▌| 61/64 [04:11<00:12,  4.11s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 60\n",
            " Train Loss = 0.8339442406127702\n",
            " Train Accuracy = 0.6430278851279317\n",
            " Validation Loss = 0.9153036168281068\n",
            " Validation Accuracy = 0.5951906028368795\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 97%|█████████▋| 62/64 [04:15<00:08,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 61\n",
            " Train Loss = 0.8318605825209668\n",
            " Train Accuracy = 0.6440898187633263\n",
            " Validation Loss = 0.917477404563985\n",
            " Validation Accuracy = 0.5971021719858157\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            " 98%|█████████▊| 63/64 [04:19<00:04,  4.10s/it]\u001b[A"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 62\n",
            " Train Loss = 0.8297399488974736\n",
            " Train Accuracy = 0.6457014592217484\n",
            " Validation Loss = 0.91858265704297\n",
            " Validation Accuracy = 0.5930296985815604\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "100%|██████████| 64/64 [04:23<00:00,  4.11s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Epoch 63\n",
            " Train Loss = 0.8270446111017199\n",
            " Train Accuracy = 0.646659281716418\n",
            " Validation Loss = 0.9188440855513228\n",
            " Validation Accuracy = 0.5953457446808511\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "P5OUlQgxJeNF",
        "outputId": "d1570a35-31cd-4a73-ec86-b82b2b7e9afc"
      },
      "source": [
        "plt.figure(figsize=(20,13))\n",
        "\n",
        "plt.plot(train_loss_history, label='Training loss')\n",
        "plt.plot(val_loss_history, label='Validation loss')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()\n",
        "\n",
        "plt.figure(figsize=(20,13))\n",
        "\n",
        "plt.plot(train_accuracy_history, label='Training accuracy')\n",
        "plt.plot(val_accuracy_history, label='Validation accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x936 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x936 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I-qE3iwfNMo6"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}