Importing Libraries
","metadata":{}},{"cell_type":"code","source":"import sys\n\nsys.path.append('../input/timm-pytorch-image-models/pytorch-image-models-master')\n! pip install ../input/einops-030/einops-0.3.0-py2.py3-none-any.whl","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:54:57.739211Z","iopub.execute_input":"2023-01-04T15:54:57.739586Z","iopub.status.idle":"2023-01-04T15:55:08.977802Z","shell.execute_reply.started":"2023-01-04T15:54:57.739554Z","shell.execute_reply":"2023-01-04T15:55:08.976611Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"! pip install timm\n! pip install einops\n\n! pip install python-gdcm -q\n! pip install pylibjpeg -q\n\n! pip install efficientnet_pytorch -q","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:56:38.795142Z","iopub.status.idle":"2023-01-04T15:56:38.796022Z","shell.execute_reply.started":"2023-01-04T15:56:38.795757Z","shell.execute_reply":"2023-01-04T15:56:38.795781Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import cv2\nimport os\nimport glob\nimport timm\nimport torch\nimport wandb\nimport pydicom\nimport numpy as np\nimport pandas as pd\nimport seaborn as sns\nimport torch.nn as nn\nfrom tqdm import tqdm\nimport albumentations\nimport albumentations as A\nfrom einops import rearrange\nfrom torchvision import transforms\nfrom joblib import Parallel, delayed\nfrom matplotlib import pyplot as plt\nfrom kaggle_secrets import UserSecretsClient\nfrom mpl_toolkits.axes_grid1 import ImageGrid\nfrom albumentations.pytorch import ToTensorV2\nfrom efficientnet_pytorch import EfficientNet\nfrom torch.utils.data import Dataset, DataLoader\nfrom accelerate import Accelerator, notebook_launcher\n\nfrom albumentations.augmentations.dropout.coarse_dropout import CoarseDropout ","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:48.182476Z","iopub.execute_input":"2023-01-04T15:55:48.183733Z","iopub.status.idle":"2023-01-04T15:55:54.605958Z","shell.execute_reply.started":"2023-01-04T15:55:48.183686Z","shell.execute_reply":"2023-01-04T15:55:54.604896Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Config
","metadata":{}},{"cell_type":"code","source":"class CFG:\n class data:\n fold=0\n batch_size=32\n image_size=(224, 224)\n path_to_train=\"../input/split-folds-rsna/5_folds_data.csv\"\n path_to_dcms=\"../input/rsna-breast-cancer-detection/train_images\"\n path_to_train_images=\"/kaggle/input/rsna-bcd-roi-1024x-png-dataset/train_images\"\n path_to_dcm_images = \"/kaggle/input/rsna-breast-cancer-detection/train_images/*/*.dcm\"\n \n class monitoring:\n accelerator=None\n \n class model:\n pretrained_name='vit_base_patch16_224'\n learning_rate=1e-4\n auto_break_n=3\n n_epochs=6\n n_labels=1","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:54.608971Z","iopub.execute_input":"2023-01-04T15:55:54.60978Z","iopub.status.idle":"2023-01-04T15:55:54.616316Z","shell.execute_reply.started":"2023-01-04T15:55:54.609738Z","shell.execute_reply":"2023-01-04T15:55:54.614968Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Monitoring Settings
","metadata":{}},{"cell_type":"code","source":"user_secrets = UserSecretsClient()\n\nwb_key = user_secrets.get_secret(\"WANDB_API_KEY\")\nwandb.login(key=wb_key)\n\nCFG.monitoring.accelerator = Accelerator(mixed_precision='fp16', log_with='wandb')\nCFG.monitoring.accelerator = Accelerator()\n\nCFG.monitoring.accelerator.init_trackers(\"rsna_mammography_pytorch\")","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:54.617879Z","iopub.execute_input":"2023-01-04T15:55:54.618604Z","iopub.status.idle":"2023-01-04T15:55:56.055927Z","shell.execute_reply.started":"2023-01-04T15:55:54.618565Z","shell.execute_reply":"2023-01-04T15:55:56.054847Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Data Loading
","metadata":{}},{"cell_type":"code","source":"train_df = pd.read_csv(CFG.data.path_to_train)\ntrain_df['image_path'] = f'{CFG.data.path_to_train_images}'\\\n + '/' + train_df.patient_id.astype(str)\\\n + '/' + train_df.image_id.astype(str)\\\n + '.png'\n\nprint(f\"train.shape = {train_df.shape}\")\ntrain_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.05803Z","iopub.execute_input":"2023-01-04T15:55:56.058658Z","iopub.status.idle":"2023-01-04T15:55:56.302006Z","shell.execute_reply.started":"2023-01-04T15:55:56.058617Z","shell.execute_reply":"2023-01-04T15:55:56.300839Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Label Preprocessing (Encoding) \n","metadata":{}},{"cell_type":"code","source":"from sklearn.preprocessing import LabelEncoder, normalize\n\n# Keep only columns in test + target variable\n\ntrain_df = train_df[[\"patient_id\", \"image_id\", \"laterality\", \"view\", \"age\", \"implant\", \"image_path\", \"cancer\", \"fold\"]]\ntrain_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.304317Z","iopub.execute_input":"2023-01-04T15:55:56.305513Z","iopub.status.idle":"2023-01-04T15:55:56.331216Z","shell.execute_reply.started":"2023-01-04T15:55:56.305461Z","shell.execute_reply":"2023-01-04T15:55:56.330068Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Encode categorical variables\n\nlabel_laterality = LabelEncoder()\nlabel_view = LabelEncoder()\n\ntrain_df.isnull().sum()[train_df.isnull().sum() > 0]","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.332605Z","iopub.execute_input":"2023-01-04T15:55:56.332999Z","iopub.status.idle":"2023-01-04T15:55:56.360382Z","shell.execute_reply.started":"2023-01-04T15:55:56.332962Z","shell.execute_reply":"2023-01-04T15:55:56.359198Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df['age'] = train_df['age'].fillna(58)\n\ntrain_df[\"laterality\"] = label_laterality.fit_transform(train_df.laterality)\ntrain_df[\"view\"] = label_view.fit_transform(train_df.view)\n\ntrain_df.to_csv(\"train_path.csv\", index=False)\ntrain_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.36288Z","iopub.execute_input":"2023-01-04T15:55:56.36406Z","iopub.status.idle":"2023-01-04T15:55:56.601331Z","shell.execute_reply.started":"2023-01-04T15:55:56.364018Z","shell.execute_reply":"2023-01-04T15:55:56.600182Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Train/Validatio
","metadata":{}},{"cell_type":"code","source":"train = train_df.query(f'fold != {CFG.data.fold}').reset_index(drop=True)\nvalid = train_df.query(f'fold == {CFG.data.fold}').reset_index(drop=True)\n\ntrain.shape, valid.shape","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.607103Z","iopub.execute_input":"2023-01-04T15:55:56.608069Z","iopub.status.idle":"2023-01-04T15:55:56.634892Z","shell.execute_reply.started":"2023-01-04T15:55:56.608023Z","shell.execute_reply":"2023-01-04T15:55:56.633739Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Transforms & Data Generator
","metadata":{}},{"cell_type":"code","source":"class RSNAData(Dataset):\n def __init__(self, df, img_folder, transform=None, is_test=False):\n self.df = df\n self.is_test = is_test\n self.transform = transform\n self.img_folder = img_folder\n \n self.csv_columns = ['laterality', 'view', 'age', 'implant']\n\n def __getitem__(self, idx):\n img_path = self.df['image_path'][idx]\n img = cv2.imread(img_path)\n\n if self.transform:\n img = self.transform(image=img)[\"image\"]\n \n# img = rearrange(img, 'h w c -> c h w')\n img = torch.tensor(img, dtype=torch.float)\n \n meta_data = np.array(\n self.df.iloc[idx][self.csv_columns].values, \n dtype=np.float32\n )\n \n if not self.is_test:\n target = self.df['cancer'][idx]\n target = torch.tensor(target, dtype=torch.float)\n \n return {\n \"image\": img,\n \"meta\": meta_data,\n \"target\": target,\n }\n \n return {\n \"image\": img,\n \"meta\": meta_data,\n }\n \n def __len__(self):\n return len(self.df)","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.636659Z","iopub.execute_input":"2023-01-04T15:55:56.637267Z","iopub.status.idle":"2023-01-04T15:55:56.649472Z","shell.execute_reply.started":"2023-01-04T15:55:56.637229Z","shell.execute_reply":"2023-01-04T15:55:56.648293Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def augmantation(image_size, is_train=True):\n vertical_flip = 0.5\n horizontal_flip = 0.5\n \n if is_train:\n transforms = [\n# A.RandomResizedCrop(height=224, width=224),\n A.Resize(height=224, width=224), \n A.ShiftScaleRotate(rotate_limit=90, scale_limit = [0.8, 1.2]),\n A.HorizontalFlip(p = horizontal_flip),\n A.VerticalFlip(p = vertical_flip) \n ]\n \n else:\n transforms = [\n A.Resize(height=224, width=224), \n ]\n \n transforms.extend([\n ToTensorV2()\n ])\n \n manipulation = A.Compose(transforms, p=1)\n \n return manipulation","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.651104Z","iopub.execute_input":"2023-01-04T15:55:56.65157Z","iopub.status.idle":"2023-01-04T15:55:56.662673Z","shell.execute_reply.started":"2023-01-04T15:55:56.65153Z","shell.execute_reply":"2023-01-04T15:55:56.661663Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_dataset = RSNAData(df=train, img_folder=CFG.data.path_to_train_images, transform=augmantation(CFG.data.image_size, True))\ntrain_loader = DataLoader(train_dataset, batch_size=CFG.data.batch_size, shuffle=True)\n\nvalid_dataset = RSNAData(df=valid, img_folder=CFG.data.path_to_train_images, transform=augmantation(CFG.data.image_size, False))\nvalid_loader = DataLoader(valid_dataset, batch_size=CFG.data.batch_size, shuffle=False)","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.664075Z","iopub.execute_input":"2023-01-04T15:55:56.664659Z","iopub.status.idle":"2023-01-04T15:55:56.674207Z","shell.execute_reply.started":"2023-01-04T15:55:56.664622Z","shell.execute_reply":"2023-01-04T15:55:56.673266Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"next(iter(train_loader)).keys()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:56.67548Z","iopub.execute_input":"2023-01-04T15:55:56.676045Z","iopub.status.idle":"2023-01-04T15:55:57.425859Z","shell.execute_reply.started":"2023-01-04T15:55:56.676007Z","shell.execute_reply":"2023-01-04T15:55:57.424743Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Example image samples
","metadata":{}},{"cell_type":"code","source":"batch_sample = next(iter(train_loader))\nimg = batch_sample[\"image\"][0]\nimg.size()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:57.428333Z","iopub.execute_input":"2023-01-04T15:55:57.429069Z","iopub.status.idle":"2023-01-04T15:55:58.183315Z","shell.execute_reply.started":"2023-01-04T15:55:57.429027Z","shell.execute_reply":"2023-01-04T15:55:58.182103Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"fig = plt.figure(figsize=(30, 20))\ngrid = ImageGrid(fig, 111,\n nrows_ncols=(4, 8),\n axes_pad=0.25\n)\n\nfor ax, img in zip(grid, batch_sample[\"image\"]):\n ax.imshow(img.type(torch.int).permute(1, 2, 0))\n\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:55:58.18501Z","iopub.execute_input":"2023-01-04T15:55:58.185467Z","iopub.status.idle":"2023-01-04T15:56:02.49879Z","shell.execute_reply.started":"2023-01-04T15:55:58.185423Z","shell.execute_reply":"2023-01-04T15:56:02.497345Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Model
","metadata":{}},{"cell_type":"code","source":"class EffNetNetwork(nn.Module):\n def __init__(self, output_size, no_columns):\n super().__init__()\n self.no_columns, self.output_size = no_columns, output_size\n \n # Define Feature part (IMAGE)\n self.features = EfficientNet.from_pretrained('efficientnet-b2')\n \n # (CSV)\n self.csv = nn.Sequential(\n nn.Linear(self.no_columns, 250),\n nn.BatchNorm1d(250),\n nn.ReLU(),\n nn.Dropout(p=0.2),\n\n nn.Linear(250, 250),\n nn.BatchNorm1d(250),\n nn.ReLU(),\n nn.Dropout(p=0.2)\n )\n \n # Define Classification part\n self.classification = nn.Sequential(nn.Linear(1408 + 250, self.output_size))\n \n \n def forward(self, image, meta, prints=False): \n \n if prints: print('Input Image shape:', image.shape, '\\n'+\n 'Input metadata shape:', meta.shape)\n \n # Image CNN\n image = self.features.extract_features(image)\n image = nn.functional.avg_pool2d(image, image.size()[2:]).reshape(-1, 1408)\n if prints: print('Features Image shape:', image.shape)\n \n # CSV FNN\n meta = self.csv(meta)\n if prints: print('Meta Data:', meta.shape)\n \n # Concatenate layers from image with layers from csv_data\n image_meta_data = torch.cat((image, meta), dim=1)\n if prints: print('Concatenated Data:', image_meta_data.shape)\n \n # CLASSIF\n out = self.classification(image_meta_data)\n if prints: print('Out shape:', out.shape)\n \n return out","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:56:38.797489Z","iopub.status.idle":"2023-01-04T15:56:38.798366Z","shell.execute_reply.started":"2023-01-04T15:56:38.798078Z","shell.execute_reply":"2023-01-04T15:56:38.798102Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"output_size = 1\ncsv_columns = ['laterality', 'view', 'age', 'implant']\nno_columns = len(csv_columns)","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:57:43.37225Z","iopub.execute_input":"2023-01-04T15:57:43.372678Z","iopub.status.idle":"2023-01-04T15:57:43.377958Z","shell.execute_reply.started":"2023-01-04T15:57:43.372644Z","shell.execute_reply":"2023-01-04T15:57:43.376884Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"with CFG.monitoring.accelerator.main_process_first():\n model = EffNetNetwork(output_size=output_size, no_columns=no_columns)","metadata":{"execution":{"iopub.status.busy":"2023-01-04T16:00:47.173152Z","iopub.execute_input":"2023-01-04T16:00:47.173763Z","iopub.status.idle":"2023-01-04T16:00:53.130082Z","shell.execute_reply.started":"2023-01-04T16:00:47.173725Z","shell.execute_reply":"2023-01-04T16:00:53.128785Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"out = model(batch_sample[\"image\"], batch_sample[\"meta\"], prints=True)","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"optimizer = torch.optim.Adam(params=model.parameters(), lr=CFG.model.learning_rate)\ncriterion = nn.BCEWithLogitsLoss()\n\nmodel, optimizer, train_loader, valid_loader = CFG.monitoring.accelerator.prepare(\n model, optimizer, train_loader, valid_loader\n)","metadata":{"execution":{"iopub.status.busy":"2023-01-04T16:04:03.111453Z","iopub.execute_input":"2023-01-04T16:04:03.111856Z","iopub.status.idle":"2023-01-04T16:04:03.334875Z","shell.execute_reply.started":"2023-01-04T16:04:03.111824Z","shell.execute_reply":"2023-01-04T16:04:03.333532Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Class Trainer
","metadata":{}},{"cell_type":"code","source":"class Trainer:\n def __init__(self, model, accelerator, criterion, optimizer, epochs, auto_break_n, monitoring=True):\n self.model = model\n self.accelerator = accelerator\n self.criterion = criterion\n self.optimizer = optimizer\n self.epochs = epochs\n \n self.history_train_loss = []\n self.history_train_score = []\n \n self.history_val_loss = []\n self.history_val_score = []\n \n self.auto_break_n = auto_break_n\n self.monitoring = monitoring\n \n self.best_val_score = 0 \n self.no_improvement_epoch = 0\n\n\n def fit(self, train_loader, valid_loader):\n \n for epoch in range(self.epochs):\n self.accelerator.print(f\"{'=' * 20} Epoch: {epoch + 1} {'=' * 20}\")\n\n self.model.train()\n \n avg_loss = 0\n all_outputs, all_targets = [], []\n \n pbar_train = tqdm(train_loader)\n for index, batch in enumerate(pbar_train, start=1):\n\n images, meta, targets = batch[\"image\"], batch[\"meta\"], batch[\"target\"]\n \n # forward()\n outputs = self.model(images, meta).view(-1)\n loss = self.criterion(outputs, targets)\n \n avg_loss += loss.item()\n all_outputs.extend(torch.sigmoid(outputs).cpu().detach().tolist())\n all_targets.extend(targets.cpu().detach().tolist())\n \n pbar_train.set_postfix({\n 'batch': index,\n 'train_loss': loss.item(),\n })\n \n self.accelerator.backward(loss)\n self.optimizer.step()\n self.optimizer.zero_grad(set_to_none=True)\n\n \n avg_loss /= len(train_loader)\n prob_f1_score = self.probabilistic_f1(all_targets, all_outputs, beta=0.5)\n \n self.history_train_loss.append(avg_loss)\n self.history_train_score.append(prob_f1_score)\n\n self.accelerator.print(f\"\\nEpoch: {epoch + 1} / {self.epochs} | Training Loss: {avg_loss:.4f} | F1 Score: {prob_f1_score:.4f}\\n\")\n \n if self.monitoring:\n self.accelerator.log({\n 'train_loss': avg_loss,\n 'train_prob_f1': prob_f1_score,\n }\n )\n \n if valid_loader:\n prob_f1_score, avg_loss = self.valid_score(valid_loader)\n \n self.history_val_loss.append(avg_loss)\n self.history_val_score.append(prob_f1_score)\n \n self.accelerator.print(f\"\\nEpoch: {epoch + 1} / {self.epochs} | Validation Loss: {avg_loss:.4f} | F1 Score: {prob_f1_score:.4f}\\n\")\n \n if self.monitoring:\n self.accelerator.log({\n 'valid_loss': avg_loss,\n 'valid_prob_f1': prob_f1_score,\n }\n )\n \n \n if prob_f1_score > self.best_val_score:\n self.no_improvement_epoch = 0\n self.best_val_score = prob_f1_score\n \n # Save the model\n self.accelerator.wait_for_everyone() \n model = self.accelerator.unwrap_model(self.model)\n self.accelerator.save({\n \"model\": self.model.state_dict(),\n \"optimizer\": self.optimizer.optimizer.state_dict()\n }, f\"epoch_{epoch + 1}_model.pth\")\n \n else: \n self.no_improvement_epoch += 1\n \n self.accelerator.print(f\"no improvement iter = {self.no_improvement_epoch}\")\n\n if self.no_improvement_epoch == self.auto_break_n:\n self.accelerator.print('Auto_break !!!')\n\n if self.monitoring:\n wandb.finish()\n break\n \n def valid_score(self, valid_loader):\n self.model.eval() # switch for some specific layers/parts\n \n avg_loss = 0\n all_outputs, all_targets = [], []\n \n with torch.no_grad():\n \n pbar_valid = tqdm(valid_loader)\n for index, batch in enumerate(pbar_valid, start=1):\n\n images, meta, targets = batch[\"image\"], batch[\"meta\"], batch[\"target\"]\n \n outputs = self.model(images, meta).view(-1)\n loss = self.criterion(outputs, targets)\n \n avg_loss += loss.item()\n outputs, targets = self.accelerator.gather_for_metrics((\n outputs, targets\n ))\n \n all_outputs.extend(torch.sigmoid(outputs).cpu().detach().tolist())\n all_targets.extend(targets.cpu().detach().tolist())\n \n pbar_valid.set_postfix({\n 'batch': index,\n 'valid_loss': loss.item(),\n })\n \n prob_f1_score = self.probabilistic_f1(all_targets, all_outputs, beta=0.5)\n avg_loss /= len(valid_loader)\n \n return prob_f1_score, avg_loss\n \n \n @staticmethod\n def probabilistic_f1(labels, predictions, beta=0.5):\n y_true_count = 0\n ctp = 0\n cfp = 0\n\n for idx in range(len(labels)):\n prediction = min(max(predictions[idx], 0), 1)\n if (labels[idx]):\n y_true_count += 1\n ctp += prediction\n cfp += 1 - prediction\n else:\n cfp += prediction\n\n beta_squared = beta * beta\n c_precision = ctp / (ctp + cfp)\n c_recall = ctp / y_true_count\n if (c_precision > 0 and c_recall > 0):\n result = (1 + beta_squared) * (c_precision * c_recall) / (beta_squared * c_precision + c_recall)\n return result\n else:\n return 0","metadata":{"execution":{"iopub.status.busy":"2023-01-04T16:04:06.50651Z","iopub.execute_input":"2023-01-04T16:04:06.506916Z","iopub.status.idle":"2023-01-04T16:04:06.533289Z","shell.execute_reply.started":"2023-01-04T16:04:06.506883Z","shell.execute_reply":"2023-01-04T16:04:06.531729Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Fit
","metadata":{}},{"cell_type":"code","source":"my_model = Trainer(model, CFG.monitoring.accelerator, criterion, optimizer, CFG.model.n_epochs, CFG.model.auto_break_n)\nnotebook_launcher(my_model.fit, args=(train_loader, valid_loader), num_processes=2)","metadata":{"execution":{"iopub.status.busy":"2023-01-04T16:04:07.980908Z","iopub.execute_input":"2023-01-04T16:04:07.981321Z","iopub.status.idle":"2023-01-04T16:05:24.52558Z","shell.execute_reply.started":"2023-01-04T16:04:07.981285Z","shell.execute_reply":"2023-01-04T16:05:24.523868Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\n#Plot probabilistic F1/Loss
","metadata":{}},{"cell_type":"code","source":"fig, ax = plt.subplots(1, 2, figsize=(30, 10))\n\nax[0].plot(my_model.history_train_score, '-o')\nax[0].plot(my_model.history_val_score, '-o')\n\nax[1].plot(my_model.history_train_loss, '-o')\nax[1].plot(my_model.history_val_loss, '-o')\n\nax[0].set_xlabel(\"epoch\")\nax[1].set_xlabel(\"epoch\")\n\nax[0].set_ylabel(\"probabilistic f1\")\nax[1].set_ylabel(\"loss\")\n\nax[0].legend(['Train','Valid'])\nax[0].set_title('Train vs Valid')\nax[1].legend(['Train','Valid'])\nax[1].set_title('Train vs Valid')\n\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2023-01-04T15:56:38.792778Z","iopub.status.idle":"2023-01-04T15:56:38.793142Z","shell.execute_reply.started":"2023-01-04T15:56:38.792959Z","shell.execute_reply":"2023-01-04T15:56:38.792976Z"},"trusted":true},"execution_count":null,"outputs":[]}]}