transport_classification_singletask.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Transport Mode Classification using BiLSTM (RNN)\n",
    "To better understand the code, check out `./prep_files/biLSTM_transport_classification.ipynb`. That notebook explains the thought process behind this code on a sample dataset whilst this goes straight to the point. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchsummary import summary\n",
    "\n",
    "from src.engine import biLSTM_engine\n",
    "from src.model_transportMode import BiLSTMNetwork\n",
    "from src.plot import plot_history, plot_endgame_history\n",
    "from src.dataset import TransportModeDataset\n",
    "\n",
    "from sklearn.utils.class_weight import compute_class_weight\n",
    "from torch.utils.data import DataLoader, random_split"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Computing class weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classes: [0. 1. 2. 3. 4. 5. 6. 7.]\n",
      "Class weights: tensor([0.6685, 0.7598, 6.0714, 1.4663, 0.8401, 1.3729, 0.8195, 0.8317],\n",
      "       device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "y_train = pd.read_csv('./data/lstm_features_labels/train/metadata.csv')\n",
    "y_train = y_train.iloc[:,1].values\n",
    "print('Classes:', np.unique(y_train))\n",
    "\n",
    "class_weights = torch.tensor(compute_class_weight('balanced', classes=np.unique(y_train), y=y_train), dtype=torch.float32)\n",
    "class_weights = class_weights.to(torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\"))\n",
    "\n",
    "print('Class weights:', class_weights)\n",
    "del y_train"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Original coarse labels are shown below. During preprocessing, the Null class (`0.0`), was dropped and with the remaining, I substracted `1` from the labels so that it would start from `0`.\n",
    "\n",
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_dir = './data/lstm_features_labels/'\n",
    "\n",
    "train_datasets = TransportModeDataset(base_dir, mode='train')\n",
    "val_test_datasets = TransportModeDataset(base_dir, mode='valid')\n",
    "\n",
    "np.random.seed(42)\n",
    "valid_datasets, test_datasets = random_split(val_test_datasets, [0.5, 0.5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(46240, 5782, 5781)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_datasets), len(valid_datasets), len(test_datasets)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1st Run with 1st Config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "Fst_config= {'optimizer': 'adamw', 'lr': 0.007696965501070433, 'scheduler': 'exp', \n",
    "             'gamma': 0.9318166309092948, 'patience': 10, 'epochs': 1200, \n",
    "             'batch_size': 64, 'weight_decay': 0.07563306484196267, 'hidden_size': 300, \n",
    "             'num_layers': 1, 'dropout': 0.494696861183782}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 64\n",
    "train_dl = DataLoader(train_datasets, batch_size, shuffle=True, num_workers=4)\n",
    "valid_dl = DataLoader(valid_datasets, batch_size, shuffle=True, num_workers=4)\n",
    "test_dl = DataLoader(test_datasets, batch_size, shuffle=False, num_workers=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss(weight=class_weights)\n",
    "model = BiLSTMNetwork(input_size=6, hidden_size=300, num_layers=1, dropout=0.494697) #drop is technically == 0 since n_layers == 1\n",
    "optimizer = optim.AdamW(model.parameters(), lr=0.00769697, weight_decay=0.0756331)\n",
    "scheduler = optim.lr_scheduler.ExponentialLR(optimizer, gamma=0.931817)\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "model = model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "engine = biLSTM_engine(model, optimizer, scheduler, criterion, device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   Epoch [1/100], Train Loss: 1.3762, Train Accuracy: 42.13%, Val Loss: 1.2007, Val Accuracy: 49.90%\n",
      "   Epoch [2/100], Train Loss: 1.1691, Train Accuracy: 51.23%, Val Loss: 1.1444, Val Accuracy: 52.18%\n",
      "   Epoch [3/100], Train Loss: 1.1178, Train Accuracy: 53.51%, Val Loss: 1.0328, Val Accuracy: 56.90%\n",
      "   Epoch [4/100], Train Loss: 1.1338, Train Accuracy: 53.39%, Val Loss: 1.2138, Val Accuracy: 48.56%\n",
      "   Epoch [5/100], Train Loss: 1.0899, Train Accuracy: 54.73%, Val Loss: 1.0583, Val Accuracy: 56.71%\n",
      "   Epoch [6/100], Train Loss: 1.0507, Train Accuracy: 56.39%, Val Loss: 1.0358, Val Accuracy: 57.90%\n",
      "   Epoch [7/100], Train Loss: 0.9951, Train Accuracy: 58.28%, Val Loss: 0.9348, Val Accuracy: 60.29%\n",
      "   Epoch [8/100], Train Loss: 0.8853, Train Accuracy: 62.47%, Val Loss: 0.8691, Val Accuracy: 62.90%\n",
      "   Epoch [9/100], Train Loss: 0.8302, Train Accuracy: 64.79%, Val Loss: 0.9988, Val Accuracy: 55.66%\n",
      "   Epoch [10/100], Train Loss: 0.8115, Train Accuracy: 65.80%, Val Loss: 0.8005, Val Accuracy: 66.07%\n",
      "   Epoch [11/100], Train Loss: 0.7638, Train Accuracy: 67.75%, Val Loss: 0.7994, Val Accuracy: 67.36%\n",
      "   Epoch [12/100], Train Loss: 0.7475, Train Accuracy: 68.48%, Val Loss: 0.7256, Val Accuracy: 69.02%\n",
      "   Epoch [13/100], Train Loss: 0.7273, Train Accuracy: 69.19%, Val Loss: 0.7288, Val Accuracy: 69.44%\n",
      "   Epoch [14/100], Train Loss: 0.7077, Train Accuracy: 69.91%, Val Loss: 0.7332, Val Accuracy: 67.93%\n",
      "   Epoch [15/100], Train Loss: 0.6935, Train Accuracy: 70.40%, Val Loss: 0.6858, Val Accuracy: 71.93%\n",
      "   Epoch [16/100], Train Loss: 0.6659, Train Accuracy: 71.71%, Val Loss: 0.6739, Val Accuracy: 72.45%\n",
      "   Epoch [17/100], Train Loss: 0.6491, Train Accuracy: 72.38%, Val Loss: 0.6478, Val Accuracy: 72.59%\n",
      "   Epoch [18/100], Train Loss: 0.6281, Train Accuracy: 73.30%, Val Loss: 0.6427, Val Accuracy: 72.69%\n",
      "   Epoch [19/100], Train Loss: 0.6105, Train Accuracy: 74.08%, Val Loss: 0.6263, Val Accuracy: 73.97%\n",
      "   Epoch [20/100], Train Loss: 0.6021, Train Accuracy: 74.47%, Val Loss: 0.6281, Val Accuracy: 73.63%\n",
      "   Epoch [21/100], Train Loss: 0.5959, Train Accuracy: 74.38%, Val Loss: 0.6143, Val Accuracy: 74.26%\n",
      "   Epoch [22/100], Train Loss: 0.5787, Train Accuracy: 75.15%, Val Loss: 0.5928, Val Accuracy: 74.85%\n",
      "   Epoch [23/100], Train Loss: 0.5560, Train Accuracy: 76.17%, Val Loss: 0.5995, Val Accuracy: 73.99%\n",
      "   Epoch [24/100], Train Loss: 0.5338, Train Accuracy: 77.33%, Val Loss: 0.6016, Val Accuracy: 75.13%\n",
      "   Epoch [25/100], Train Loss: 0.5189, Train Accuracy: 77.65%, Val Loss: 0.5769, Val Accuracy: 76.34%\n",
      "   Epoch [26/100], Train Loss: 0.5121, Train Accuracy: 77.74%, Val Loss: 0.5840, Val Accuracy: 75.79%\n",
      "   Epoch [27/100], Train Loss: 0.4945, Train Accuracy: 78.74%, Val Loss: 0.5758, Val Accuracy: 76.58%\n",
      "   Epoch [28/100], Train Loss: 0.4963, Train Accuracy: 78.74%, Val Loss: 0.5473, Val Accuracy: 77.14%\n",
      "   Epoch [29/100], Train Loss: 0.4620, Train Accuracy: 80.13%, Val Loss: 0.5526, Val Accuracy: 76.43%\n",
      "   Epoch [30/100], Train Loss: 0.4594, Train Accuracy: 80.15%, Val Loss: 0.5450, Val Accuracy: 77.60%\n",
      "   Epoch [31/100], Train Loss: 0.4429, Train Accuracy: 80.85%, Val Loss: 0.5519, Val Accuracy: 77.40%\n",
      "   Epoch [32/100], Train Loss: 0.4267, Train Accuracy: 81.62%, Val Loss: 0.5343, Val Accuracy: 77.78%\n",
      "   Epoch [33/100], Train Loss: 0.4169, Train Accuracy: 81.99%, Val Loss: 0.5238, Val Accuracy: 78.21%\n",
      "   Epoch [34/100], Train Loss: 0.4026, Train Accuracy: 82.67%, Val Loss: 0.5214, Val Accuracy: 78.29%\n",
      "   Epoch [35/100], Train Loss: 0.3934, Train Accuracy: 82.94%, Val Loss: 0.5104, Val Accuracy: 78.78%\n",
      "   Epoch [36/100], Train Loss: 0.3837, Train Accuracy: 83.55%, Val Loss: 0.5175, Val Accuracy: 79.28%\n",
      "   Epoch [37/100], Train Loss: 0.3751, Train Accuracy: 83.92%, Val Loss: 0.5044, Val Accuracy: 79.63%\n",
      "   Epoch [38/100], Train Loss: 0.3635, Train Accuracy: 84.33%, Val Loss: 0.4999, Val Accuracy: 79.71%\n",
      "   Epoch [39/100], Train Loss: 0.3498, Train Accuracy: 84.65%, Val Loss: 0.5043, Val Accuracy: 79.61%\n",
      "   Epoch [40/100], Train Loss: 0.3419, Train Accuracy: 85.18%, Val Loss: 0.5163, Val Accuracy: 79.38%\n",
      "   Epoch [41/100], Train Loss: 0.3408, Train Accuracy: 85.45%, Val Loss: 0.4924, Val Accuracy: 80.04%\n",
      "   Epoch [42/100], Train Loss: 0.3303, Train Accuracy: 85.90%, Val Loss: 0.5159, Val Accuracy: 80.11%\n",
      "   Epoch [43/100], Train Loss: 0.3131, Train Accuracy: 86.59%, Val Loss: 0.5076, Val Accuracy: 80.63%\n",
      "   Epoch [44/100], Train Loss: 0.3061, Train Accuracy: 86.90%, Val Loss: 0.4902, Val Accuracy: 80.54%\n",
      "   Epoch [45/100], Train Loss: 0.2962, Train Accuracy: 87.36%, Val Loss: 0.5003, Val Accuracy: 81.20%\n",
      "   Epoch [46/100], Train Loss: 0.2888, Train Accuracy: 87.61%, Val Loss: 0.4802, Val Accuracy: 81.65%\n",
      "   Epoch [47/100], Train Loss: 0.2788, Train Accuracy: 87.94%, Val Loss: 0.4865, Val Accuracy: 81.82%\n",
      "   Epoch [48/100], Train Loss: 0.2732, Train Accuracy: 88.23%, Val Loss: 0.4949, Val Accuracy: 82.22%\n",
      "   Epoch [49/100], Train Loss: 0.2657, Train Accuracy: 88.69%, Val Loss: 0.4865, Val Accuracy: 81.91%\n",
      "   Epoch [50/100], Train Loss: 0.2592, Train Accuracy: 88.98%, Val Loss: 0.4970, Val Accuracy: 81.34%\n",
      "   Epoch [51/100], Train Loss: 0.2481, Train Accuracy: 89.55%, Val Loss: 0.4918, Val Accuracy: 82.12%\n",
      "   Epoch [52/100], Train Loss: 0.2422, Train Accuracy: 89.82%, Val Loss: 0.5046, Val Accuracy: 81.44%\n",
      "   Epoch [53/100], Train Loss: 0.2376, Train Accuracy: 89.97%, Val Loss: 0.4944, Val Accuracy: 82.03%\n",
      "   Epoch [54/100], Train Loss: 0.2313, Train Accuracy: 90.24%, Val Loss: 0.5087, Val Accuracy: 81.81%\n",
      "   Epoch [55/100], Train Loss: 0.2204, Train Accuracy: 90.64%, Val Loss: 0.5149, Val Accuracy: 81.70%\n",
      "   Epoch [56/100], Train Loss: 0.2160, Train Accuracy: 90.87%, Val Loss: 0.5231, Val Accuracy: 82.13%\n",
      "   Epoch [57/100], Train Loss: 0.2149, Train Accuracy: 91.01%, Val Loss: 0.5124, Val Accuracy: 82.31%\n",
      "   Epoch [58/100], Train Loss: 0.2040, Train Accuracy: 91.48%, Val Loss: 0.5211, Val Accuracy: 82.10%\n",
      "   Epoch [59/100], Train Loss: 0.1996, Train Accuracy: 91.67%, Val Loss: 0.5372, Val Accuracy: 81.89%\n",
      "   Epoch [60/100], Train Loss: 0.1956, Train Accuracy: 91.92%, Val Loss: 0.5364, Val Accuracy: 81.84%\n",
      "   Epoch [61/100], Train Loss: 0.1909, Train Accuracy: 92.01%, Val Loss: 0.5264, Val Accuracy: 82.12%\n",
      "   Epoch [62/100], Train Loss: 0.1856, Train Accuracy: 92.23%, Val Loss: 0.5415, Val Accuracy: 82.34%\n",
      "   Epoch [63/100], Train Loss: 0.1813, Train Accuracy: 92.53%, Val Loss: 0.5395, Val Accuracy: 81.96%\n",
      "   Epoch [64/100], Train Loss: 0.1762, Train Accuracy: 92.69%, Val Loss: 0.5426, Val Accuracy: 81.91%\n",
      "   Epoch [65/100], Train Loss: 0.1708, Train Accuracy: 92.89%, Val Loss: 0.5480, Val Accuracy: 82.31%\n",
      "   Epoch [66/100], Train Loss: 0.1681, Train Accuracy: 93.07%, Val Loss: 0.5473, Val Accuracy: 82.55%\n",
      "   Epoch [67/100], Train Loss: 0.1643, Train Accuracy: 93.16%, Val Loss: 0.5612, Val Accuracy: 82.34%\n",
      "   Epoch [68/100], Train Loss: 0.1596, Train Accuracy: 93.52%, Val Loss: 0.5557, Val Accuracy: 82.43%\n",
      "   Epoch [69/100], Train Loss: 0.1586, Train Accuracy: 93.57%, Val Loss: 0.5604, Val Accuracy: 82.51%\n",
      "   Epoch [70/100], Train Loss: 0.1528, Train Accuracy: 93.86%, Val Loss: 0.5645, Val Accuracy: 82.22%\n",
      "   Epoch [71/100], Train Loss: 0.1491, Train Accuracy: 93.88%, Val Loss: 0.5688, Val Accuracy: 82.55%\n",
      "   Epoch [72/100], Train Loss: 0.1494, Train Accuracy: 93.90%, Val Loss: 0.5740, Val Accuracy: 82.31%\n",
      "   Epoch [73/100], Train Loss: 0.1449, Train Accuracy: 94.10%, Val Loss: 0.5854, Val Accuracy: 82.51%\n",
      "   Epoch [74/100], Train Loss: 0.1421, Train Accuracy: 94.24%, Val Loss: 0.5884, Val Accuracy: 82.39%\n",
      "   Epoch [75/100], Train Loss: 0.1380, Train Accuracy: 94.40%, Val Loss: 0.5906, Val Accuracy: 82.39%\n",
      "   Epoch [76/100], Train Loss: 0.1356, Train Accuracy: 94.51%, Val Loss: 0.6010, Val Accuracy: 82.31%\n",
      "   Epoch [77/100], Train Loss: 0.1337, Train Accuracy: 94.69%, Val Loss: 0.6007, Val Accuracy: 82.26%\n",
      "   Epoch [78/100], Train Loss: 0.1320, Train Accuracy: 94.73%, Val Loss: 0.5994, Val Accuracy: 82.46%\n",
      "   Epoch [79/100], Train Loss: 0.1298, Train Accuracy: 94.79%, Val Loss: 0.6036, Val Accuracy: 82.26%\n",
      "   Epoch [80/100], Train Loss: 0.1277, Train Accuracy: 94.86%, Val Loss: 0.6135, Val Accuracy: 82.46%\n",
      "   Epoch [81/100], Train Loss: 0.1250, Train Accuracy: 95.00%, Val Loss: 0.6187, Val Accuracy: 82.22%\n",
      "   Epoch [82/100], Train Loss: 0.1235, Train Accuracy: 95.07%, Val Loss: 0.6222, Val Accuracy: 82.41%\n",
      "   Epoch [83/100], Train Loss: 0.1209, Train Accuracy: 95.17%, Val Loss: 0.6324, Val Accuracy: 81.98%\n",
      "   Epoch [84/100], Train Loss: 0.1192, Train Accuracy: 95.22%, Val Loss: 0.6226, Val Accuracy: 82.39%\n",
      "   Epoch [85/100], Train Loss: 0.1185, Train Accuracy: 95.30%, Val Loss: 0.6371, Val Accuracy: 82.48%\n",
      "   Epoch [86/100], Train Loss: 0.1174, Train Accuracy: 95.34%, Val Loss: 0.6388, Val Accuracy: 82.10%\n",
      "   Epoch [87/100], Train Loss: 0.1153, Train Accuracy: 95.46%, Val Loss: 0.6329, Val Accuracy: 82.19%\n",
      "   Epoch [88/100], Train Loss: 0.1142, Train Accuracy: 95.53%, Val Loss: 0.6422, Val Accuracy: 82.41%\n",
      "   Epoch [89/100], Train Loss: 0.1126, Train Accuracy: 95.58%, Val Loss: 0.6417, Val Accuracy: 82.00%\n"
     ]
    },
    {
     "ename": "RuntimeError",
     "evalue": "CUDA error: unspecified launch failure\nCUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1.\nCompile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.\n",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "Cell \u001b[1;32mIn[9], line 1\u001b[0m\n\u001b[1;32m----> 1\u001b[0m firstRun_hist \u001b[38;5;241m=\u001b[39m \u001b[43mengine\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtrain_validation\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtrain_dl\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mvalid_dl\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mepochs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m100\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msave_path\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43m./model_checkpoint/run1_transpMode_BiLSTM_chkpt.pth\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[0;32m      3\u001b[0m np\u001b[38;5;241m.\u001b[39msave(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m./model_checkpoint/run1_transpMode_BiLSTM_history.npy\u001b[39m\u001b[38;5;124m'\u001b[39m, np\u001b[38;5;241m.\u001b[39marray(firstRun_hist))\n",
      "File \u001b[1;32mc:\\Users\\LEGION\\Documents\\Msc Data Science - Uni of Exeter\\ECMM451 - Data Science Research Project (2023)\\workspace\\src\\engine.py:293\u001b[0m, in \u001b[0;36mbiLSTM_engine.train_validation\u001b[1;34m(self, train_loader, val_loader, epochs, save_path)\u001b[0m\n\u001b[0;32m    289\u001b[0m val_accuracies \u001b[38;5;241m=\u001b[39m []\n\u001b[0;32m    291\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m epoch \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mrange\u001b[39m(epochs):\n\u001b[0;32m    292\u001b[0m       \u001b[38;5;66;03m# training\u001b[39;00m\n\u001b[1;32m--> 293\u001b[0m       train_loss, train_accuracy \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtrain\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtrain_loader\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m    294\u001b[0m       train_losses\u001b[38;5;241m.\u001b[39mappend(train_loss)\n\u001b[0;32m    295\u001b[0m       train_accuracies\u001b[38;5;241m.\u001b[39mappend(train_accuracy)\n",
      "File \u001b[1;32mc:\\Users\\LEGION\\Documents\\Msc Data Science - Uni of Exeter\\ECMM451 - Data Science Research Project (2023)\\workspace\\src\\engine.py:254\u001b[0m, in \u001b[0;36mbiLSTM_engine.train\u001b[1;34m(self, train_loader)\u001b[0m\n\u001b[0;32m    251\u001b[0m loss\u001b[38;5;241m.\u001b[39mbackward()\n\u001b[0;32m    252\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39moptimizer\u001b[38;5;241m.\u001b[39mstep()\n\u001b[1;32m--> 254\u001b[0m running_loss \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m \u001b[43mloss\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mitem\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m    255\u001b[0m _, predicted \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39mmax(outputs, \u001b[38;5;241m1\u001b[39m)\n\u001b[0;32m    256\u001b[0m total \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m labels\u001b[38;5;241m.\u001b[39msize(\u001b[38;5;241m0\u001b[39m)\n",
      "\u001b[1;31mRuntimeError\u001b[0m: CUDA error: unspecified launch failure\nCUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1.\nCompile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.\n"
     ]
    }
   ],
   "source": [
    "firstRun_hist = engine.train_validation(train_dl, valid_dl, epochs=100, save_path='./model_checkpoint/run1_transpMode_BiLSTM_chkpt.pth')\n",
    "\n",
    "np.save('./model_checkpoint/run1_transpMode_BiLSTM_history.npy', np.array(firstRun_hist))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plot_history(firstRun_hist)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2nd Run with 1st Config\n",
    "Uses normalized class_weights and gradient clipping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classes: [0. 1. 2. 3. 4. 5. 6. 7.]\n",
      "Class weights: tensor([0.0521, 0.0592, 0.4732, 0.1143, 0.0655, 0.1070, 0.0639, 0.0648],\n",
      "       device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "y_train = pd.read_csv('./data/lstm_features_labels/train/metadata.csv')\n",
    "y_train = y_train.iloc[:,1].values\n",
    "print('Classes:', np.unique(y_train))\n",
    "\n",
    "class_weights = torch.tensor(compute_class_weight('balanced', classes=np.unique(y_train), y=y_train), dtype=torch.float32)\n",
    "class_weights = class_weights / class_weights.sum()\n",
    "class_weights = class_weights.to(torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\"))\n",
    "\n",
    "print('Class weights:', class_weights)\n",
    "del y_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "Fst_config= {'optimizer': 'adamw', 'lr': 0.007696965501070433, 'scheduler': 'exp', \n",
    "             'gamma': 0.9318166309092948, 'patience': 10, 'epochs': 1200, \n",
    "             'batch_size': 64, 'weight_decay': 0.07563306484196267, 'hidden_size': 300, \n",
    "             'num_layers': 1, 'dropout': 0.494696861183782}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 64\n",
    "train_dl = DataLoader(train_datasets, batch_size, shuffle=True, num_workers=4)\n",
    "valid_dl = DataLoader(valid_datasets, batch_size, shuffle=True, num_workers=4)\n",
    "test_dl = DataLoader(test_datasets, batch_size, shuffle=False, num_workers=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss(weight=class_weights)\n",
    "model = BiLSTMNetwork(input_size=6, hidden_size=300, num_layers=1, dropout=0.494697) #drop is technically == 0 since n_layers == 1\n",
    "optimizer = optim.AdamW(model.parameters(), lr=0.00769697, weight_decay=0.0756331)\n",
    "scheduler = optim.lr_scheduler.ExponentialLR(optimizer, gamma=0.931817)\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "model = model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "engine = biLSTM_engine(model, optimizer, scheduler, criterion, device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   Epoch [1/100], Train Loss: 1.3652, Train Accuracy: 42.31%, Val Loss: 1.1483, Val Accuracy: 51.25%\n",
      "   Epoch [2/100], Train Loss: 1.1060, Train Accuracy: 53.37%, Val Loss: 1.0636, Val Accuracy: 55.00%\n",
      "   Epoch [3/100], Train Loss: 0.9890, Train Accuracy: 57.84%, Val Loss: 0.9931, Val Accuracy: 58.68%\n",
      "   Epoch [4/100], Train Loss: 0.8897, Train Accuracy: 62.20%, Val Loss: 0.8287, Val Accuracy: 64.23%\n",
      "   Epoch [5/100], Train Loss: 0.8073, Train Accuracy: 66.04%, Val Loss: 0.7883, Val Accuracy: 66.14%\n",
      "   Epoch [6/100], Train Loss: 0.7710, Train Accuracy: 67.58%, Val Loss: 0.7750, Val Accuracy: 68.14%\n",
      "   Epoch [7/100], Train Loss: 0.7358, Train Accuracy: 69.01%, Val Loss: 0.7709, Val Accuracy: 68.54%\n",
      "   Epoch [8/100], Train Loss: 0.7114, Train Accuracy: 69.87%, Val Loss: 0.7323, Val Accuracy: 70.17%\n",
      "   Epoch [9/100], Train Loss: 0.6775, Train Accuracy: 71.34%, Val Loss: 0.7032, Val Accuracy: 71.60%\n",
      "   Epoch [10/100], Train Loss: 0.6554, Train Accuracy: 72.29%, Val Loss: 0.6770, Val Accuracy: 71.95%\n",
      "   Epoch [11/100], Train Loss: 0.6320, Train Accuracy: 73.21%, Val Loss: 0.6462, Val Accuracy: 72.03%\n",
      "   Epoch [12/100], Train Loss: 0.6104, Train Accuracy: 74.17%, Val Loss: 0.6389, Val Accuracy: 73.16%\n",
      "   Epoch [13/100], Train Loss: 0.5933, Train Accuracy: 74.67%, Val Loss: 0.6063, Val Accuracy: 74.70%\n",
      "   Epoch [14/100], Train Loss: 0.5764, Train Accuracy: 75.60%, Val Loss: 0.6306, Val Accuracy: 73.88%\n",
      "   Epoch [15/100], Train Loss: 0.5552, Train Accuracy: 76.30%, Val Loss: 0.6250, Val Accuracy: 73.38%\n",
      "   Epoch [16/100], Train Loss: 0.5397, Train Accuracy: 77.08%, Val Loss: 0.5836, Val Accuracy: 76.48%\n",
      "   Epoch [17/100], Train Loss: 0.5270, Train Accuracy: 77.67%, Val Loss: 0.5876, Val Accuracy: 76.24%\n",
      "   Epoch [18/100], Train Loss: 0.5091, Train Accuracy: 78.37%, Val Loss: 0.5694, Val Accuracy: 76.63%\n",
      "   Epoch [19/100], Train Loss: 0.4939, Train Accuracy: 79.10%, Val Loss: 0.5782, Val Accuracy: 76.96%\n",
      "   Epoch [20/100], Train Loss: 0.4820, Train Accuracy: 79.53%, Val Loss: 0.5643, Val Accuracy: 76.48%\n",
      "   Epoch [21/100], Train Loss: 0.4636, Train Accuracy: 80.26%, Val Loss: 0.5569, Val Accuracy: 77.71%\n",
      "   Epoch [22/100], Train Loss: 0.4457, Train Accuracy: 81.12%, Val Loss: 0.5478, Val Accuracy: 77.62%\n",
      "   Epoch [23/100], Train Loss: 0.4344, Train Accuracy: 81.54%, Val Loss: 0.5472, Val Accuracy: 78.19%\n",
      "   Epoch [24/100], Train Loss: 0.4182, Train Accuracy: 82.28%, Val Loss: 0.5515, Val Accuracy: 78.55%\n",
      "   Epoch [25/100], Train Loss: 0.4063, Train Accuracy: 82.56%, Val Loss: 0.5341, Val Accuracy: 78.55%\n",
      "   Epoch [26/100], Train Loss: 0.3936, Train Accuracy: 83.28%, Val Loss: 0.5241, Val Accuracy: 79.44%\n",
      "   Epoch [27/100], Train Loss: 0.3835, Train Accuracy: 83.79%, Val Loss: 0.5361, Val Accuracy: 79.11%\n",
      "   Epoch [28/100], Train Loss: 0.3712, Train Accuracy: 84.09%, Val Loss: 0.4924, Val Accuracy: 79.92%\n",
      "   Epoch [29/100], Train Loss: 0.3570, Train Accuracy: 84.85%, Val Loss: 0.5147, Val Accuracy: 79.94%\n",
      "   Epoch [30/100], Train Loss: 0.3486, Train Accuracy: 85.17%, Val Loss: 0.5014, Val Accuracy: 80.61%\n",
      "   Epoch [31/100], Train Loss: 0.3380, Train Accuracy: 85.64%, Val Loss: 0.5157, Val Accuracy: 79.49%\n",
      "   Epoch [32/100], Train Loss: 0.3289, Train Accuracy: 86.03%, Val Loss: 0.5088, Val Accuracy: 80.13%\n",
      "   Epoch [33/100], Train Loss: 0.3163, Train Accuracy: 86.61%, Val Loss: 0.4962, Val Accuracy: 81.23%\n",
      "   Epoch [34/100], Train Loss: 0.3058, Train Accuracy: 86.88%, Val Loss: 0.5022, Val Accuracy: 81.18%\n",
      "   Epoch [35/100], Train Loss: 0.2977, Train Accuracy: 87.43%, Val Loss: 0.4925, Val Accuracy: 81.94%\n",
      "   Epoch [36/100], Train Loss: 0.2855, Train Accuracy: 87.87%, Val Loss: 0.5076, Val Accuracy: 80.75%\n",
      "   Epoch [37/100], Train Loss: 0.2747, Train Accuracy: 88.37%, Val Loss: 0.4997, Val Accuracy: 82.39%\n",
      "   Epoch [38/100], Train Loss: 0.2689, Train Accuracy: 88.70%, Val Loss: 0.4983, Val Accuracy: 82.41%\n",
      "   Epoch [39/100], Train Loss: 0.2594, Train Accuracy: 88.98%, Val Loss: 0.5239, Val Accuracy: 81.94%\n",
      "   Epoch [40/100], Train Loss: 0.2518, Train Accuracy: 89.33%, Val Loss: 0.5065, Val Accuracy: 82.55%\n",
      "   Epoch [41/100], Train Loss: 0.2425, Train Accuracy: 89.74%, Val Loss: 0.5173, Val Accuracy: 82.32%\n",
      "   Epoch [42/100], Train Loss: 0.2345, Train Accuracy: 90.02%, Val Loss: 0.5200, Val Accuracy: 82.48%\n",
      "   Epoch [43/100], Train Loss: 0.2276, Train Accuracy: 90.39%, Val Loss: 0.5159, Val Accuracy: 83.17%\n",
      "   Epoch [44/100], Train Loss: 0.2183, Train Accuracy: 90.75%, Val Loss: 0.5339, Val Accuracy: 83.14%\n",
      "   Epoch [45/100], Train Loss: 0.2126, Train Accuracy: 91.06%, Val Loss: 0.5222, Val Accuracy: 82.50%\n",
      "   Epoch [46/100], Train Loss: 0.2017, Train Accuracy: 91.43%, Val Loss: 0.5460, Val Accuracy: 82.84%\n",
      "   Epoch [47/100], Train Loss: 0.1976, Train Accuracy: 91.62%, Val Loss: 0.5331, Val Accuracy: 83.43%\n",
      "   Epoch [48/100], Train Loss: 0.1908, Train Accuracy: 91.96%, Val Loss: 0.5378, Val Accuracy: 83.59%\n",
      "   Epoch [49/100], Train Loss: 0.1841, Train Accuracy: 92.26%, Val Loss: 0.5674, Val Accuracy: 82.65%\n",
      "   Epoch [50/100], Train Loss: 0.1770, Train Accuracy: 92.49%, Val Loss: 0.5345, Val Accuracy: 83.15%\n",
      "   Epoch [51/100], Train Loss: 0.1715, Train Accuracy: 92.88%, Val Loss: 0.5452, Val Accuracy: 83.64%\n",
      "   Epoch [52/100], Train Loss: 0.1662, Train Accuracy: 93.05%, Val Loss: 0.5745, Val Accuracy: 83.00%\n",
      "   Epoch [53/100], Train Loss: 0.1604, Train Accuracy: 93.29%, Val Loss: 0.5505, Val Accuracy: 83.59%\n",
      "   Epoch [54/100], Train Loss: 0.1541, Train Accuracy: 93.56%, Val Loss: 0.5782, Val Accuracy: 83.17%\n",
      "   Epoch [55/100], Train Loss: 0.1481, Train Accuracy: 93.90%, Val Loss: 0.5758, Val Accuracy: 83.28%\n",
      "   Epoch [56/100], Train Loss: 0.1429, Train Accuracy: 93.95%, Val Loss: 0.5772, Val Accuracy: 83.90%\n",
      "   Epoch [57/100], Train Loss: 0.1380, Train Accuracy: 94.30%, Val Loss: 0.6172, Val Accuracy: 83.59%\n",
      "   Epoch [58/100], Train Loss: 0.1319, Train Accuracy: 94.45%, Val Loss: 0.6303, Val Accuracy: 83.15%\n",
      "   Epoch [59/100], Train Loss: 0.1264, Train Accuracy: 94.81%, Val Loss: 0.6129, Val Accuracy: 83.45%\n",
      "   Epoch [60/100], Train Loss: 0.1224, Train Accuracy: 94.97%, Val Loss: 0.6175, Val Accuracy: 83.54%\n",
      "   Epoch [61/100], Train Loss: 0.1197, Train Accuracy: 95.11%, Val Loss: 0.6294, Val Accuracy: 83.85%\n",
      "   Epoch [62/100], Train Loss: 0.1149, Train Accuracy: 95.32%, Val Loss: 0.6542, Val Accuracy: 83.76%\n",
      "   Epoch [63/100], Train Loss: 0.1096, Train Accuracy: 95.53%, Val Loss: 0.6475, Val Accuracy: 83.69%\n",
      "   Epoch [64/100], Train Loss: 0.1061, Train Accuracy: 95.59%, Val Loss: 0.6414, Val Accuracy: 83.95%\n",
      "   Epoch [65/100], Train Loss: 0.1017, Train Accuracy: 95.87%, Val Loss: 0.6582, Val Accuracy: 84.19%\n",
      "   Epoch [66/100], Train Loss: 0.0985, Train Accuracy: 96.01%, Val Loss: 0.6768, Val Accuracy: 83.88%\n",
      "   Epoch [67/100], Train Loss: 0.0944, Train Accuracy: 96.08%, Val Loss: 0.6867, Val Accuracy: 83.73%\n",
      "   Epoch [68/100], Train Loss: 0.0915, Train Accuracy: 96.30%, Val Loss: 0.6801, Val Accuracy: 83.60%\n",
      "   Epoch [69/100], Train Loss: 0.0882, Train Accuracy: 96.41%, Val Loss: 0.6959, Val Accuracy: 83.81%\n",
      "   Epoch [70/100], Train Loss: 0.0849, Train Accuracy: 96.57%, Val Loss: 0.7139, Val Accuracy: 83.78%\n",
      "   Epoch [71/100], Train Loss: 0.0820, Train Accuracy: 96.71%, Val Loss: 0.7193, Val Accuracy: 83.50%\n",
      "   Epoch [72/100], Train Loss: 0.0791, Train Accuracy: 96.82%, Val Loss: 0.7312, Val Accuracy: 83.86%\n",
      "   Epoch [73/100], Train Loss: 0.0764, Train Accuracy: 96.98%, Val Loss: 0.7310, Val Accuracy: 83.79%\n",
      "   Epoch [74/100], Train Loss: 0.0746, Train Accuracy: 97.08%, Val Loss: 0.7534, Val Accuracy: 83.67%\n",
      "   Epoch [75/100], Train Loss: 0.0721, Train Accuracy: 97.13%, Val Loss: 0.7532, Val Accuracy: 83.59%\n",
      "   Epoch [76/100], Train Loss: 0.0699, Train Accuracy: 97.25%, Val Loss: 0.7594, Val Accuracy: 83.78%\n",
      "   Epoch [77/100], Train Loss: 0.0671, Train Accuracy: 97.36%, Val Loss: 0.7612, Val Accuracy: 83.88%\n",
      "   Epoch [78/100], Train Loss: 0.0659, Train Accuracy: 97.43%, Val Loss: 0.7609, Val Accuracy: 83.90%\n",
      "   Epoch [79/100], Train Loss: 0.0639, Train Accuracy: 97.49%, Val Loss: 0.7700, Val Accuracy: 83.78%\n",
      "   Epoch [80/100], Train Loss: 0.0620, Train Accuracy: 97.56%, Val Loss: 0.7820, Val Accuracy: 83.97%\n",
      "   Epoch [81/100], Train Loss: 0.0604, Train Accuracy: 97.71%, Val Loss: 0.7968, Val Accuracy: 83.47%\n",
      "   Epoch [82/100], Train Loss: 0.0589, Train Accuracy: 97.74%, Val Loss: 0.7967, Val Accuracy: 83.73%\n",
      "   Epoch [83/100], Train Loss: 0.0570, Train Accuracy: 97.77%, Val Loss: 0.7916, Val Accuracy: 83.50%\n",
      "   Epoch [84/100], Train Loss: 0.0562, Train Accuracy: 97.88%, Val Loss: 0.8059, Val Accuracy: 83.74%\n",
      "   Epoch [85/100], Train Loss: 0.0556, Train Accuracy: 97.88%, Val Loss: 0.8116, Val Accuracy: 83.73%\n",
      "   Epoch [86/100], Train Loss: 0.0534, Train Accuracy: 97.99%, Val Loss: 0.8164, Val Accuracy: 83.62%\n",
      "   Epoch [87/100], Train Loss: 0.0525, Train Accuracy: 97.98%, Val Loss: 0.8148, Val Accuracy: 83.52%\n",
      "   Epoch [88/100], Train Loss: 0.0508, Train Accuracy: 98.11%, Val Loss: 0.8132, Val Accuracy: 83.73%\n",
      "   Epoch [89/100], Train Loss: 0.0502, Train Accuracy: 98.14%, Val Loss: 0.8198, Val Accuracy: 83.71%\n",
      "   Epoch [90/100], Train Loss: 0.0493, Train Accuracy: 98.18%, Val Loss: 0.8346, Val Accuracy: 83.62%\n",
      "   Epoch [91/100], Train Loss: 0.0480, Train Accuracy: 98.20%, Val Loss: 0.8461, Val Accuracy: 83.55%\n",
      "   Epoch [92/100], Train Loss: 0.0476, Train Accuracy: 98.23%, Val Loss: 0.8440, Val Accuracy: 83.81%\n",
      "   Epoch [93/100], Train Loss: 0.0469, Train Accuracy: 98.28%, Val Loss: 0.8431, Val Accuracy: 83.45%\n",
      "   Epoch [94/100], Train Loss: 0.0460, Train Accuracy: 98.31%, Val Loss: 0.8534, Val Accuracy: 83.79%\n",
      "   Epoch [95/100], Train Loss: 0.0455, Train Accuracy: 98.34%, Val Loss: 0.8472, Val Accuracy: 83.41%\n",
      "   Epoch [96/100], Train Loss: 0.0441, Train Accuracy: 98.36%, Val Loss: 0.8573, Val Accuracy: 83.60%\n",
      "   Epoch [97/100], Train Loss: 0.0434, Train Accuracy: 98.44%, Val Loss: 0.8572, Val Accuracy: 83.81%\n",
      "   Epoch [98/100], Train Loss: 0.0433, Train Accuracy: 98.43%, Val Loss: 0.8538, Val Accuracy: 83.52%\n",
      "   Epoch [99/100], Train Loss: 0.0426, Train Accuracy: 98.46%, Val Loss: 0.8689, Val Accuracy: 83.54%\n",
      "   Epoch [100/100], Train Loss: 0.0424, Train Accuracy: 98.47%, Val Loss: 0.8616, Val Accuracy: 83.47%\n"
     ]
    }
   ],
   "source": [
    "secondRun_hist = engine.train_validation(train_dl, valid_dl, epochs=100, save_path='./model_checkpoint/run2_transpMode_BiLSTM_chkpt.pth')\n",
    "\n",
    "np.save('./model_checkpoint/run2_transpMode_BiLSTM_history.npy', np.array(secondRun_hist))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history(secondRun_hist)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I believe the model at epoch 40 will give a good fit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "tempHist = secondRun_hist[0][:45], secondRun_hist[1][:45], secondRun_hist[2][:45], secondRun_hist[3][:45]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history(tempHist)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Although this looks good, the 1st run had a slightly lesser loss at around the 40th epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "checkpoint = torch.load('./model_checkpoint/run2_transpMode_BiLSTM_chkpt.pth_epoch_40.pt')\n",
    "\n",
    "model.load_state_dict(checkpoint['model_state_dict'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Loss: 0.4858, Test Accuracy: 81.92%\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.48576374843225373, 81.92354263968171)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "engine = biLSTM_engine(model, optimizer, scheduler, criterion, device)\n",
    "\n",
    "engine.test(test_dl)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3rd Run: 2nd Config (1st Run)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "Snd_config= {'optimizer': 'adamw', 'lr': 0.005198989965043782, 'scheduler': 'ReduceLROnPlateau', \n",
    "             'gamma': 0.9383597426554285, 'patience': 5, 'epochs': 50, 'batch_size': 64, \n",
    "             'weight_decay': 0.008375293519355285, 'hidden_size': 300, 'num_layers': 1, \n",
    "             'dropout': 0.618753235646235}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 64\n",
    "train_dl = DataLoader(train_datasets, batch_size, shuffle=True, num_workers=4)\n",
    "valid_dl = DataLoader(valid_datasets, batch_size, shuffle=True, num_workers=4)\n",
    "test_dl = DataLoader(test_datasets, batch_size, shuffle=False, num_workers=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss(weight=class_weights)\n",
    "model = BiLSTMNetwork(input_size=6, hidden_size=300, num_layers=1, dropout=0.618753235646235) #drop is technically == 0 since n_layers == 1\n",
    "optimizer = optim.AdamW(model.parameters(), lr=0.005198989965043782, weight_decay=0.008375293519355285)\n",
    "scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, patience=5,factor=0.9383597426554285)\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "model = model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "engine = biLSTM_engine(model, optimizer, scheduler, criterion, device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "thirdRun_hist = engine.train_validation(train_dl, valid_dl, epochs=80, save_path='./model_checkpoint/run3_transpMode_BiLSTM_chkpt.pth')\n",
    "\n",
    "np.save('./model_checkpoint/run23_transpMode_BiLSTM_history.npy', np.array(thirdRun_hist))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plot_history(thirdRun_hist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "torch-gpu",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}