4_augmentation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:05.655061625Z",
     "start_time": "2023-05-24T18:10:57.793903721Z"
    }
   },
   "outputs": [],
   "source": [
    "import random\n",
    "import sys\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import torch\n",
    "\n",
    "sys.path.append(\"src\")\n",
    "\n",
    "import utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [
    {
     "data": {
      "text/plain": "<torch._C.Generator at 0x7fae4d0e69f0>"
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random.seed(42)\n",
    "torch.manual_seed(42)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:05.661194227Z",
     "start_time": "2023-05-24T18:11:05.656630464Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:18.570965248Z",
     "start_time": "2023-05-24T18:11:05.661820462Z"
    }
   },
   "outputs": [],
   "source": [
    "train = utils.read_data(\"train\")\n",
    "test = utils.read_data(\"test\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "outputs": [
    {
     "data": {
      "text/plain": "468"
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:18.574562436Z",
     "start_time": "2023-05-24T18:11:18.572572327Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:18.603181147Z",
     "start_time": "2023-05-24T18:11:18.576425992Z"
    }
   },
   "outputs": [],
   "source": [
    "train_sequences = [(train[key]['ts'].to_numpy(), 1 if train[key]['class'] == \"a\" else 0) for key in train]\n",
    "test_sequences = [(test[key]['ts'].to_numpy(), 1 if test[key]['class'] == \"a\" else 0) for key in test]\n",
    "val_sequences = train_sequences[200:]\n",
    "train_sequences = train_sequences[:200]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:18.926674902Z",
     "start_time": "2023-05-24T18:11:18.603021983Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from tsaug import Crop, Quantize, Drift, AddNoise\n",
    "\n",
    "augmentor = (\n",
    "        Crop(size=256) + Quantize(n_levels=[10, 20, 30]) + Drift(max_drift=(0.05, 0.2)) @ 0.8 + AddNoise(scale=0.1))\n",
    "x1 = np.array(train_sequences[0][0]).T\n",
    "synthetic_x1 = augmentor.augment(x1)\n",
    "\n",
    "plt.plot(synthetic_x1[0], label=\"synthesised\")\n",
    "plt.plot(x1[0], label=\"original\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch.utils.data import Dataset\n",
    "\n",
    "\n",
    "class EEGDataset(Dataset):\n",
    "    def __init__(self, sequences, synthetic: bool = False):\n",
    "        self.sequences = sequences\n",
    "        self.synthetic = synthetic\n",
    "        if self.synthetic:\n",
    "            self.augmentor = (Crop(size=256) + Quantize(n_levels=[10, 20, 30])\n",
    "                              + Drift(max_drift=(0.05, 0.2)) @ 0.8 + AddNoise(scale=0.1))\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.sequences) * 5\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        sequence, label = self.sequences[idx // 5]\n",
    "        if self.synthetic:\n",
    "            sequence = self.augmentor.augment(sequence.T).T\n",
    "        return dict(\n",
    "            sequence=torch.Tensor(sequence),\n",
    "            label=torch.tensor(label).long()\n",
    "        )"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:18.974004090Z",
     "start_time": "2023-05-24T18:11:18.929072350Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "outputs": [],
   "source": [
    "import pytorch_lightning as pl\n",
    "from torch.utils.data import DataLoader\n",
    "\n",
    "\n",
    "class EEGDataModule(pl.LightningDataModule):\n",
    "    def __init__(self, train_sequences, val_sequences, test_sequences, batch_size):\n",
    "        super().__init__()\n",
    "        self.train_sequences = train_sequences\n",
    "        self.val_sequences = val_sequences\n",
    "        self.test_sequences = test_sequences\n",
    "        self.batch_size = batch_size\n",
    "\n",
    "    def setup(self, stage=None):\n",
    "        self.train_dataset = EEGDataset(self.train_sequences, True)\n",
    "        self.val_dataset = EEGDataset(self.val_sequences)\n",
    "        self.test_dataset = EEGDataset(self.test_sequences)\n",
    "\n",
    "    def train_dataloader(self):\n",
    "        return DataLoader(\n",
    "            self.train_dataset,\n",
    "            batch_size=self.batch_size,\n",
    "            shuffle=True,\n",
    "            num_workers=6\n",
    "        )\n",
    "\n",
    "    def val_dataloader(self):\n",
    "        return DataLoader(\n",
    "            self.val_dataset,\n",
    "            batch_size=self.batch_size,\n",
    "            shuffle=False,\n",
    "            num_workers=6\n",
    "        )\n",
    "\n",
    "    def test_dataloader(self):\n",
    "        return DataLoader(\n",
    "            self.test_dataset,\n",
    "            batch_size=self.batch_size,\n",
    "            shuffle=False,\n",
    "            num_workers=6\n",
    "        )"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.262044742Z",
     "start_time": "2023-05-24T18:11:18.974198725Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "from typing import Literal\n",
    "\n",
    "\n",
    "class EEGModel(nn.Module):\n",
    "    def __init__(self, n_features: int = 64, n_hidden: int = 256, n_layers: int = 3,\n",
    "                 rnn: Literal[\"lstm\", \"gru\"] = \"lstm\", dropout: float = 0.3):\n",
    "        super().__init__()\n",
    "\n",
    "        match rnn:\n",
    "            case \"lstm\":\n",
    "                self.rnn_class = nn.LSTM\n",
    "            case \"gru\":\n",
    "                self.rnn_class = nn.GRU\n",
    "            case _:\n",
    "                raise ValueError(\"Invalid rnn architecture\")\n",
    "\n",
    "        self.rnn = self.rnn_class(\n",
    "            input_size=n_features,\n",
    "            hidden_size=n_hidden,\n",
    "            num_layers=n_layers,\n",
    "            batch_first=True,\n",
    "            bidirectional=True,\n",
    "            dropout=dropout\n",
    "        )\n",
    "        self.classifier = nn.Linear(2 * n_hidden, 1)\n",
    "        self.sigmoid = nn.Sigmoid()\n",
    "\n",
    "    def forward(self, x):\n",
    "        self.rnn.to(x.device)\n",
    "        self.classifier.to(x.device)\n",
    "\n",
    "        out, _ = self.rnn(x)\n",
    "        out = out.max(-2).values\n",
    "        out = self.classifier(out)\n",
    "        out = self.sigmoid(out)\n",
    "        return out"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.266145972Z",
     "start_time": "2023-05-24T18:11:19.264935823Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "from torchmetrics import Accuracy\n",
    "\n",
    "\n",
    "class EEGPredictior(pl.LightningModule):\n",
    "    def __init__(self, n_features, *args, **kwargs):\n",
    "        super().__init__()\n",
    "        self.model = EEGModel(n_features, *args, **kwargs)\n",
    "        self.criterion = nn.BCELoss()\n",
    "        self.acc = Accuracy(task=\"binary\")\n",
    "\n",
    "    def forward(self, x, labels=None):\n",
    "        output = self.model(x)\n",
    "        loss = 0\n",
    "        if labels is not None:\n",
    "            loss = self.criterion(output, labels.float().unsqueeze(-1))\n",
    "        return loss, output\n",
    "\n",
    "    def predict(self, sequence):\n",
    "        loss, output = self(torch.Tensor(sequence))\n",
    "        predictions = (output > 0.5).float().squeeze(-1)\n",
    "        return predictions\n",
    "\n",
    "    def training_step(self, batch, batch_idx):\n",
    "        sequences = batch[\"sequence\"]\n",
    "        labels = batch[\"label\"]\n",
    "        loss, outputs = self(sequences, labels)\n",
    "        predictions = (outputs > 0.5).float().squeeze(-1)\n",
    "        step_accuracy = self.acc(predictions, labels)\n",
    "\n",
    "        self.log(\"train_loss\", loss, prog_bar=True, logger=True)\n",
    "        self.log(\"train_accuracy\", step_accuracy, prog_bar=True, logger=True)\n",
    "        return {\"loss\": loss, \"accuracy\": step_accuracy}\n",
    "\n",
    "    def validation_step(self, batch, batch_idx):\n",
    "        sequences = batch[\"sequence\"]\n",
    "        labels = batch[\"label\"]\n",
    "        loss, outputs = self(sequences, labels)\n",
    "        predictions = (outputs > 0.5).float().squeeze(-1)\n",
    "        step_accuracy = self.acc(predictions, labels)\n",
    "\n",
    "        self.log(\"val_loss\", loss, prog_bar=True, logger=True)\n",
    "        self.log(\"val_accuracy\", step_accuracy, prog_bar=True, logger=True)\n",
    "        return {\"loss\": loss, \"accuracy\": step_accuracy}\n",
    "\n",
    "    def test_step(self, batch, batch_idx):\n",
    "        sequences = batch[\"sequence\"]\n",
    "        labels = batch[\"label\"]\n",
    "        loss, outputs = self(sequences, labels)\n",
    "        predictions = (outputs > 0.5).float().squeeze(-1)\n",
    "        step_accuracy = self.acc(predictions, labels)\n",
    "\n",
    "        self.log(\"test_loss\", loss, prog_bar=True, logger=True)\n",
    "        self.log(\"test_accuracy\", step_accuracy, prog_bar=True, logger=True)\n",
    "        return {\"loss\": loss, \"accuracy\": step_accuracy}\n",
    "\n",
    "    def configure_optimizers(self):\n",
    "        return optim.Adam(self.parameters(), lr=0.001)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.292137698Z",
     "start_time": "2023-05-24T18:11:19.269930195Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "outputs": [],
   "source": [
    "model = EEGPredictior(n_features=64, n_hidden=256, rnn=\"lstm\", n_layers=2, dropout=0.3)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.319098816Z",
     "start_time": "2023-05-24T18:11:19.293359408Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "outputs": [
    {
     "data": {
      "text/plain": "Launching TensorBoard..."
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%reload_ext tensorboard\n",
    "%tensorboard --logdir./lightning_logs"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.826200704Z",
     "start_time": "2023-05-24T18:11:19.318429610Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "outputs": [],
   "source": [
    "data_module = EEGDataModule(train_sequences, val_sequences, test_sequences, 32)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.828226584Z",
     "start_time": "2023-05-24T18:11:19.826374988Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "GPU available: True (cuda), used: True\n",
      "TPU available: False, using: 0 TPU cores\n",
      "IPU available: False, using: 0 IPUs\n",
      "HPU available: False, using: 0 HPUs\n"
     ]
    }
   ],
   "source": [
    "from pytorch_lightning.loggers import TensorBoardLogger\n",
    "from pytorch_lightning.callbacks import ModelCheckpoint\n",
    "\n",
    "checkpoint_callback = ModelCheckpoint(\n",
    "    dirpath=\"checkpoints\",\n",
    "    filename=\"best-checkpoint\",\n",
    "    save_top_k=1,\n",
    "    verbose=True,\n",
    "    monitor=\"val_loss\",\n",
    "    mode=\"min\"\n",
    ")\n",
    "\n",
    "logger = TensorBoardLogger(\"lightning_logs\", name=\"EEG\")\n",
    "\n",
    "trainer = pl.Trainer(\n",
    "    logger=logger,\n",
    "    callbacks=checkpoint_callback,\n",
    "    max_epochs=10,\n",
    ")"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:19.902545690Z",
     "start_time": "2023-05-24T18:11:19.830421157Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/jkarolczak/miniconda3/envs/eeg/lib/python3.10/site-packages/pytorch_lightning/callbacks/model_checkpoint.py:612: UserWarning: Checkpoint directory /home/jkarolczak/Projects/eeg-alcoholics/checkpoints exists and is not empty.\n",
      "  rank_zero_warn(f\"Checkpoint directory {dirpath} exists and is not empty.\")\n",
      "LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]\n",
      "\n",
      "  | Name      | Type           | Params\n",
      "---------------------------------------------\n",
      "0 | model     | EEGModel       | 2.2 M \n",
      "1 | criterion | BCELoss        | 0     \n",
      "2 | acc       | BinaryAccuracy | 0     \n",
      "---------------------------------------------\n",
      "2.2 M     Trainable params\n",
      "0         Non-trainable params\n",
      "2.2 M     Total params\n",
      "8.948     Total estimated model params size (MB)\n"
     ]
    },
    {
     "data": {
      "text/plain": "Sanity Checking: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "c0d0428d7ca943d5acaff12c07314d9f"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/jkarolczak/miniconda3/envs/eeg/lib/python3.10/site-packages/pytorch_lightning/loops/fit_loop.py:280: PossibleUserWarning: The number of training batches (32) is smaller than the logging interval Trainer(log_every_n_steps=50). Set a lower value for log_every_n_steps if you want to see logs for the training epoch.\n",
      "  rank_zero_warn(\n"
     ]
    },
    {
     "data": {
      "text/plain": "Training: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "b86e063e347f4309b881415071c9a8b0"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "148cae9c50a444959d55541822bf4ad6"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 0, global step 32: 'val_loss' reached 0.49399 (best 0.49399), saving model to '/home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt' as top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "0377a23b8ec84e8da6a4a5b58f198c9b"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 1, global step 64: 'val_loss' reached 0.41529 (best 0.41529), saving model to '/home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt' as top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "f3844f5e373141fda7cde037ab544f4b"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 2, global step 96: 'val_loss' reached 0.31873 (best 0.31873), saving model to '/home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt' as top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "84cff8fa8a094db289dded13bf58a2a8"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 3, global step 128: 'val_loss' reached 0.31768 (best 0.31768), saving model to '/home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt' as top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "244fed6b53e747b98e4006deb0567944"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 4, global step 160: 'val_loss' was not in top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "d8324c5cf45a440a84d31211ce45526f"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 5, global step 192: 'val_loss' was not in top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "454592e40d434392aea6113dc44a1398"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 6, global step 224: 'val_loss' was not in top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "4a19b1f09bd74c26bac2c25cc63276de"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 7, global step 256: 'val_loss' was not in top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "84324f84bbba4cb484cb3d502ac1ab66"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 8, global step 288: 'val_loss' was not in top 1\n"
     ]
    },
    {
     "data": {
      "text/plain": "Validation: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "f9419d47fb94453688e356e9f70795a1"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 9, global step 320: 'val_loss' reached 0.29948 (best 0.29948), saving model to '/home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt' as top 1\n",
      "`Trainer.fit` stopped: `max_epochs=10` reached.\n"
     ]
    }
   ],
   "source": [
    "trainer.fit(model, data_module)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:42.645470197Z",
     "start_time": "2023-05-24T18:11:19.903678898Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/jkarolczak/miniconda3/envs/eeg/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/checkpoint_connector.py:148: UserWarning: `.test(ckpt_path=None)` was called without a model. The best model of the previous `fit` call will be used. You can pass `.test(ckpt_path='best')` to use the best model or `.test(ckpt_path='last')` to use the last model. If you pass a value, this warning will be silenced.\n",
      "  rank_zero_warn(\n",
      "Restoring states from the checkpoint path at /home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt\n",
      "LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]\n",
      "Loaded model weights from the checkpoint at /home/jkarolczak/Projects/eeg-alcoholics/checkpoints/best-checkpoint-v47.ckpt\n"
     ]
    },
    {
     "data": {
      "text/plain": "Testing: 0it [00:00, ?it/s]",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "2cf119ad61ad46f488305b7fc0555e78"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "       Test metric             DataLoader 0\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "      test_accuracy          0.918749988079071\n",
      "        test_loss           0.22188563644886017\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n"
     ]
    },
    {
     "data": {
      "text/plain": "[{'test_loss': 0.22188563644886017, 'test_accuracy': 0.918749988079071}]"
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.test(dataloaders=data_module.test_dataloader())"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:44.126176709Z",
     "start_time": "2023-05-24T18:11:42.646991203Z"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "outputs": [],
   "source": [],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-05-24T18:11:44.126313492Z",
     "start_time": "2023-05-24T18:11:44.126052686Z"
    }
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py311",
   "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": "3.11.3"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}