{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"
\n

[Pytorch] ArcFace Starter

\n
","metadata":{}},{"cell_type":"markdown","source":"
\n\nIt is based on the work of Vlad Vaduva https://www.kaggle.com/vladvdv/pytorch-train-notebook-arcface-gem-pooling\n\n### I don't know how to use wandb with tpu. If you know how to use it with tpu, please comment!\n\n* v1: quick save\n* v2: add code for tpu with huggingface accelerate\n* v3: trying to fix error ...\n* v4: quick save\n* v5: trying to fix error ...\n* v6: check 2 epoch train time\n* v7: trying to fix error ...\n* v8: add pad_across_processes and increase batch size\n* v9: pad_index=-100\n* v10: trying to fix error ...\n* v11: trying to fix error ...\n* v12: trying to fix error ...\n* v13 ~ v48: trial and error\n* v49: revert to v12","metadata":{}},{"cell_type":"markdown","source":"#

Install Required Libraries

","metadata":{}},{"cell_type":"code","source":"!curl https://raw.githubusercontent.com/pytorch/xla/master/contrib/scripts/env-setup.py -o pytorch-xla-env-setup.py\n!python pytorch-xla-env-setup.py --version 1.8.1 --apt-packages libomp5 libopenblas-dev","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2022-03-01T13:07:57.874113Z","iopub.execute_input":"2022-03-01T13:07:57.874700Z","iopub.status.idle":"2022-03-01T13:09:28.091111Z","shell.execute_reply.started":"2022-03-01T13:07:57.874609Z","shell.execute_reply":"2022-03-01T13:09:28.090031Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install timm\n!pip install --upgrade wandb\n!pip install accelerate","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2022-03-01T13:09:28.092868Z","iopub.execute_input":"2022-03-01T13:09:28.093125Z","iopub.status.idle":"2022-03-01T13:09:58.593815Z","shell.execute_reply.started":"2022-03-01T13:09:28.093094Z","shell.execute_reply":"2022-03-01T13:09:58.592730Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Import Required Libraries 📚

","metadata":{}},{"cell_type":"code","source":"import os\nimport gc\nimport cv2\nimport math\nimport copy\nimport time\nimport random\n\n# For data manipulation\nimport numpy as np\nimport pandas as pd\n\n# Pytorch Imports\nimport torch\nimport torch.nn as nn\nimport torch.optim as optim\nimport torch.nn.functional as F\nfrom torch.optim import lr_scheduler\nfrom torch.utils.data import Dataset, DataLoader\n\n# Utils\nimport joblib\nfrom tqdm.auto import tqdm\nfrom collections import defaultdict\n\n# Sklearn Imports\nfrom sklearn.preprocessing import LabelEncoder\nfrom sklearn.model_selection import StratifiedKFold\n\n# For Image Models\nimport timm\n\n# Albumentations for augmentations\nimport albumentations as A\nfrom albumentations.pytorch import ToTensorV2\n\n# For colored terminal text\nfrom colorama import Fore, Back, Style\nb_ = Fore.BLUE\nsr_ = Style.RESET_ALL\n\n# For TPU Training\nfrom accelerate import Accelerator, notebook_launcher\n\nos.environ['XLA_TENSOR_ALLOCATOR_MAXSIZE'] = '100000000'\n\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\n# For descriptive error messages\nos.environ['CUDA_LAUNCH_BLOCKING'] = \"1\"","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:09:58.595892Z","iopub.execute_input":"2022-03-01T13:09:58.596243Z","iopub.status.idle":"2022-03-01T13:10:44.868299Z","shell.execute_reply.started":"2022-03-01T13:09:58.596198Z","shell.execute_reply":"2022-03-01T13:10:44.866551Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"\"Weights\n\n Weights & Biases (W&B) is a set of machine learning tools that helps you build better models faster. Kaggle competitions require fast-paced model development and evaluation. There are a lot of components: exploring the training data, training different models, combining trained models in different combinations (ensembling), and so on.\n\n> ⏳ Lots of components = Lots of places to go wrong = Lots of time spent debugging\n\nW&B can be useful for Kaggle competition with it's lightweight and interoperable tools:\n\n* Quickly track experiments,
\n* Version and iterate on datasets,
\n* Evaluate model performance,
\n* Reproduce models,
\n* Visualize results and spot regressions,
\n* Share findings with colleagues.\n\nTo learn more about Weights and Biases check out this kernel.","metadata":{}},{"cell_type":"code","source":"# import wandb\n\n# try:\n# from kaggle_secrets import UserSecretsClient\n# user_secrets = UserSecretsClient()\n# api_key = user_secrets.get_secret(\"wandb_api\")\n# wandb.login(key=api_key)\n# anony = None\n# except:\n# anony = \"must\"\n# print('If you want to use your W&B account, go to Add-ons -> Secrets and provide your W&B access token. Use the Label name as wandb_api. \\nGet your W&B access token from here: https://wandb.ai/authorize')","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:44.871765Z","iopub.execute_input":"2022-03-01T13:10:44.872475Z","iopub.status.idle":"2022-03-01T13:10:44.877722Z","shell.execute_reply.started":"2022-03-01T13:10:44.872429Z","shell.execute_reply":"2022-03-01T13:10:44.876836Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Training Configuration ⚙️

","metadata":{}},{"cell_type":"code","source":"CONFIG = {\"seed\": 2022,\n \"epochs\": 24,\n \"img_size\": 768,\n \"model_name\": \"tf_efficientnet_b4\",\n \"num_classes\": 15587,\n \"train_batch_size\": 16,\n \"valid_batch_size\": 8, # if valid_batch_size is bigger, \"Failed to meet rendezvous 'accelerate.utils.gather': Received message larger than max\" error occurred\n \"learning_rate\": 0.0001,\n \"scheduler\": 'OneCycleLR',\n \"min_lr\": 1e-6,\n \"T_max\": 500,\n \"weight_decay\": 1e-6,\n \"n_fold\": 5,\n \"fold_to_run\": 0,\n \"n_accumulate\": 1,\n \"test_mode\":True, # enable for testing pipeline, changes epochs to 2 and uses just 400 training samples\n \"progress_update_freq\": 20,\n # ArcFace Hyperparameters\n \"s\": 30.0, \n \"m\": 0.30,\n \"ls_eps\": 0.0,\n \"easy_margin\": False\n }","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:44.879270Z","iopub.execute_input":"2022-03-01T13:10:44.879521Z","iopub.status.idle":"2022-03-01T13:10:44.893391Z","shell.execute_reply.started":"2022-03-01T13:10:44.879480Z","shell.execute_reply":"2022-03-01T13:10:44.892187Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Set Seed for Reproducibility

","metadata":{}},{"cell_type":"code","source":"def set_seed(seed=42):\n '''Sets the seed of the entire notebook so results are the same every time we run.\n This is for REPRODUCIBILITY.'''\n np.random.seed(seed)\n torch.manual_seed(seed)\n torch.cuda.manual_seed(seed)\n # When running on the CuDNN backend, two further options must be set\n torch.backends.cudnn.deterministic = True\n torch.backends.cudnn.benchmark = False\n # Set a fixed value for the hash seed\n os.environ['PYTHONHASHSEED'] = str(seed)\n \nset_seed(CONFIG['seed'])","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:44.894305Z","iopub.execute_input":"2022-03-01T13:10:44.894559Z","iopub.status.idle":"2022-03-01T13:10:44.912382Z","shell.execute_reply.started":"2022-03-01T13:10:44.894524Z","shell.execute_reply":"2022-03-01T13:10:44.911367Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"ROOT_DIR = '../input/happy-whale-and-dolphin'\nTRAIN_DIR = '../input/happy-whale-and-dolphin/train_images'\nTEST_DIR = '../input/happy-whale-and-dolphin/test_images'","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:44.913651Z","iopub.execute_input":"2022-03-01T13:10:44.914340Z","iopub.status.idle":"2022-03-01T13:10:44.926085Z","shell.execute_reply.started":"2022-03-01T13:10:44.914298Z","shell.execute_reply":"2022-03-01T13:10:44.925156Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_train_file_path(id):\n return f\"{TRAIN_DIR}/{id}\"","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:44.928130Z","iopub.execute_input":"2022-03-01T13:10:44.929272Z","iopub.status.idle":"2022-03-01T13:10:44.941354Z","shell.execute_reply.started":"2022-03-01T13:10:44.929217Z","shell.execute_reply":"2022-03-01T13:10:44.940193Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Read the Data 📖

","metadata":{}},{"cell_type":"code","source":"df = pd.read_csv(f\"{ROOT_DIR}/train.csv\")\ndf['file_path'] = df['image'].apply(get_train_file_path)\ndf.head()\n\nif CONFIG[\"test_mode\"]==True:\n df=df[:400]\n CONFIG[\"epochs\"] = 2\n CONFIG[\"n_fold\"] = 3\nencoder = LabelEncoder()\ndf['individual_id'] = encoder.fit_transform(df['individual_id'])\n\nwith open(\"le.pkl\", \"wb\") as fp:\n joblib.dump(encoder, fp)","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:44.943045Z","iopub.execute_input":"2022-03-01T13:10:44.943362Z","iopub.status.idle":"2022-03-01T13:10:45.087331Z","shell.execute_reply.started":"2022-03-01T13:10:44.943309Z","shell.execute_reply":"2022-03-01T13:10:45.086299Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Create Folds

","metadata":{}},{"cell_type":"code","source":"skf = StratifiedKFold(n_splits=CONFIG['n_fold'])\n\nfor fold, ( _, val_) in enumerate(skf.split(X=df, y=df.individual_id)):\n df.loc[val_ , \"kfold\"] = fold","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.090468Z","iopub.execute_input":"2022-03-01T13:10:45.090845Z","iopub.status.idle":"2022-03-01T13:10:45.106051Z","shell.execute_reply.started":"2022-03-01T13:10:45.090803Z","shell.execute_reply":"2022-03-01T13:10:45.105082Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Dataset Class

","metadata":{}},{"cell_type":"code","source":"class HappyWhaleDataset(Dataset):\n def __init__(self, df, transforms=None):\n self.df = df\n self.file_names = df['file_path'].values\n self.labels = df['individual_id'].values\n self.transforms = transforms\n \n def __len__(self):\n return len(self.df)\n \n def __getitem__(self, index):\n img_path = self.file_names[index]\n img = cv2.imread(img_path)\n img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n label = self.labels[index]\n \n if self.transforms:\n img = self.transforms(image=img)[\"image\"]\n \n return {\n 'image': img,\n 'label': torch.tensor(label, dtype=torch.long)\n }","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.107899Z","iopub.execute_input":"2022-03-01T13:10:45.108343Z","iopub.status.idle":"2022-03-01T13:10:45.117021Z","shell.execute_reply.started":"2022-03-01T13:10:45.108298Z","shell.execute_reply":"2022-03-01T13:10:45.116208Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Augmentations

","metadata":{}},{"cell_type":"code","source":"data_transforms = {\n \"train\": A.Compose([\n A.Resize(CONFIG['img_size'], CONFIG['img_size']),\n A.HorizontalFlip(p=0.5),\n A.Normalize(\n mean=[0.485, 0.456, 0.406], \n std=[0.229, 0.224, 0.225], \n max_pixel_value=255.0, \n p=1.0\n ),\n ToTensorV2()], p=1.),\n \n \"valid\": A.Compose([\n A.Resize(CONFIG['img_size'], CONFIG['img_size']),\n A.Normalize(\n mean=[0.485, 0.456, 0.406], \n std=[0.229, 0.224, 0.225], \n max_pixel_value=255.0, \n p=1.0\n ),\n ToTensorV2()], p=1.)\n}","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.118092Z","iopub.execute_input":"2022-03-01T13:10:45.118314Z","iopub.status.idle":"2022-03-01T13:10:45.131619Z","shell.execute_reply.started":"2022-03-01T13:10:45.118288Z","shell.execute_reply":"2022-03-01T13:10:45.130715Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

GeM Pooling

\n\nCode taken from GeM Pooling Explained\n\n![](https://i.imgur.com/thTgYWG.jpg)","metadata":{}},{"cell_type":"code","source":"class GeM(nn.Module):\n def __init__(self, p=3, eps=1e-6):\n super(GeM, self).__init__()\n self.p = nn.Parameter(torch.ones(1)*p)\n self.eps = eps\n\n def forward(self, x):\n return self.gem(x, p=self.p, eps=self.eps)\n \n def gem(self, x, p=3, eps=1e-6):\n return F.avg_pool2d(x.clamp(min=eps).pow(p), (x.size(-2), x.size(-1))).pow(1./p)\n \n def __repr__(self):\n return self.__class__.__name__ + \\\n '(' + 'p=' + '{:.4f}'.format(self.p.data.tolist()[0]) + \\\n ', ' + 'eps=' + str(self.eps) + ')'","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.133355Z","iopub.execute_input":"2022-03-01T13:10:45.134147Z","iopub.status.idle":"2022-03-01T13:10:45.144856Z","shell.execute_reply.started":"2022-03-01T13:10:45.134106Z","shell.execute_reply":"2022-03-01T13:10:45.144138Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

ArcFace

\n\nCode taken from Landmark2019-1st-and-3rd-Place-Solution","metadata":{}},{"cell_type":"code","source":"class ArcMarginProduct(nn.Module):\n r\"\"\"Implement of large margin arc distance: :\n Args:\n in_features: size of each input sample\n out_features: size of each output sample\n s: norm of input feature\n m: margin\n cos(theta + m)\n \"\"\"\n def __init__(self, in_features, out_features, accelerator,s=30.0, \n m=0.50, easy_margin=False, ls_eps=0.0):\n super(ArcMarginProduct, self).__init__()\n self.in_features = in_features\n self.out_features = out_features\n self.s = s\n self.m = m\n self.ls_eps = ls_eps # label smoothing\n self.weight = nn.Parameter(torch.FloatTensor(out_features, in_features))\n nn.init.xavier_uniform_(self.weight)\n\n self.easy_margin = easy_margin\n self.cos_m = math.cos(m)\n self.sin_m = math.sin(m)\n self.th = math.cos(math.pi - m)\n self.mm = math.sin(math.pi - m) * m\n self.accelerator = accelerator\n\n def forward(self, input, label):\n # --------------------------- cos(theta) & phi(theta) ---------------------\n cosine = F.linear(F.normalize(input), F.normalize(self.weight))\n sine = torch.sqrt(1.0 - torch.pow(cosine, 2))\n phi = cosine * self.cos_m - sine * self.sin_m\n if self.easy_margin:\n phi = torch.where(cosine > 0, phi, cosine)\n else:\n phi = torch.where(cosine > self.th, phi, cosine - self.mm)\n # --------------------------- convert label to one-hot ---------------------\n # one_hot = torch.zeros(cosine.size(), requires_grad=True, device='cuda')\n one_hot = torch.zeros(cosine.size(), device=self.accelerator.device)\n one_hot.scatter_(1, label.view(-1, 1).long(), 1)\n if self.ls_eps > 0:\n one_hot = (1 - self.ls_eps) * one_hot + self.ls_eps / self.out_features\n # -------------torch.where(out_i = {x_i if condition_i else y_i) ------------\n output = (one_hot * phi) + ((1.0 - one_hot) * cosine)\n output *= self.s\n\n return output","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.146012Z","iopub.execute_input":"2022-03-01T13:10:45.146394Z","iopub.status.idle":"2022-03-01T13:10:45.164078Z","shell.execute_reply.started":"2022-03-01T13:10:45.146352Z","shell.execute_reply":"2022-03-01T13:10:45.163112Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Create Model

","metadata":{}},{"cell_type":"code","source":"class HappyWhaleModel(nn.Module):\n def __init__(self, model_name, accelerator,pretrained=True):\n super(HappyWhaleModel, self).__init__()\n self.model = timm.create_model(model_name, pretrained=pretrained)\n in_features = self.model.classifier.in_features\n self.model.classifier = nn.Identity()\n self.model.global_pool = nn.Identity()\n self.pooling = GeM()\n self.drop = nn.Dropout(p=0.2, inplace=False)\n self.fc = nn.Linear(in_features,512)\n self.arc = ArcMarginProduct(512, \n CONFIG[\"num_classes\"],\n accelerator,\n s=CONFIG[\"s\"], \n m=CONFIG[\"m\"], \n easy_margin=CONFIG[\"ls_eps\"], \n ls_eps=CONFIG[\"ls_eps\"])\n def forward(self, images, labels):\n features = self.model(images)\n pooled_features = self.pooling(features).flatten(1)\n pooled_drop = self.drop(pooled_features)\n emb = self.fc(pooled_drop)\n output = self.arc(emb,labels)\n return output,emb\n","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.165444Z","iopub.execute_input":"2022-03-01T13:10:45.165964Z","iopub.status.idle":"2022-03-01T13:10:45.181616Z","shell.execute_reply.started":"2022-03-01T13:10:45.165923Z","shell.execute_reply":"2022-03-01T13:10:45.180840Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Loss Function

","metadata":{}},{"cell_type":"code","source":"def criterion(outputs, labels):\n return nn.CrossEntropyLoss()(outputs, labels)","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.182691Z","iopub.execute_input":"2022-03-01T13:10:45.183055Z","iopub.status.idle":"2022-03-01T13:10:45.195324Z","shell.execute_reply.started":"2022-03-01T13:10:45.183026Z","shell.execute_reply":"2022-03-01T13:10:45.194531Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Training Function

","metadata":{}},{"cell_type":"code","source":"def train_one_epoch(model, optimizer, scheduler, dataloader, accelerator, epoch):\n model.train()\n \n dataset_size = 0\n running_loss = 0.0\n \n bar = tqdm(enumerate(dataloader), total=len(dataloader), disable=not accelerator.is_local_main_process)\n for step, data in bar:\n images = data['image']\n labels = data['label']\n \n batch_size = images.size(0)\n \n outputs, emb = model(images, labels)\n loss = criterion(outputs, labels)\n loss = loss / CONFIG['n_accumulate']\n \n accelerator.backward(loss)\n \n if (step + 1) % CONFIG['n_accumulate'] == 0:\n optimizer.step()\n\n # zero the parameter gradients\n optimizer.zero_grad()\n\n if scheduler is not None:\n scheduler.step()\n \n running_loss += loss * batch_size\n dataset_size += batch_size\n \n if step % CONFIG['progress_update_freq'] == 0:\n \n bar.set_postfix(Epoch=epoch, Train_Loss=running_loss / dataset_size,\n LR=optimizer.param_groups[0]['lr'])\n \n gc.collect()\n \n return running_loss.item() / dataset_size","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.196876Z","iopub.execute_input":"2022-03-01T13:10:45.197147Z","iopub.status.idle":"2022-03-01T13:10:45.211775Z","shell.execute_reply.started":"2022-03-01T13:10:45.197115Z","shell.execute_reply":"2022-03-01T13:10:45.210775Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Validation Function

","metadata":{}},{"cell_type":"code","source":"@torch.no_grad()\ndef valid_one_epoch(model,optimizer, dataloader, accelerator, epoch):\n model.eval()\n \n dataset_size = 0\n running_loss = 0.0\n \n bar = tqdm(enumerate(dataloader), total=len(dataloader), disable=not accelerator.is_local_main_process)\n for step, data in bar: \n images = data['image']\n labels = data['label']\n \n batch_size = images.size(0)\n\n outputs, emb = model(images, labels)\n \n labels = accelerator.pad_across_processes(labels, pad_index=-100)\n outputs = accelerator.pad_across_processes(outputs, pad_index=-100)\n \n all_outputs = accelerator.gather(outputs)\n all_labels = accelerator.gather(labels)\n \n loss = criterion(all_outputs, all_labels)\n \n running_loss += loss * batch_size\n dataset_size += batch_size\n\n bar.set_postfix(Epoch=epoch, Valid_Loss=running_loss / dataset_size,\n LR=optimizer.param_groups[0]['lr'])\n \n gc.collect()\n \n return running_loss.item() / dataset_size","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.213761Z","iopub.execute_input":"2022-03-01T13:10:45.214102Z","iopub.status.idle":"2022-03-01T13:10:45.230450Z","shell.execute_reply.started":"2022-03-01T13:10:45.214058Z","shell.execute_reply":"2022-03-01T13:10:45.229812Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

Run Training

","metadata":{}},{"cell_type":"code","source":"def run_training(num_epochs):\n accelerator = Accelerator()\n \n model = HappyWhaleModel(CONFIG['model_name'], accelerator)\n optimizer = optim.Adam(model.parameters(), lr=CONFIG['learning_rate'], \n weight_decay=CONFIG['weight_decay'])\n \n train_loader, valid_loader = prepare_loaders(df, fold=CONFIG['fold_to_run'])\n \n model, optimizer, train_loader, valid_loader = accelerator.prepare(model, optimizer, train_loader, valid_loader)\n \n scheduler = fetch_scheduler(optimizer, train_loader)\n \n# run = wandb.init(project='happywhale', \n# config=CONFIG,\n# job_type='Train',\n# tags=['arcface', 'gem-pooling', 'effnet-b4', '768'],\n# anonymous='must',\n# group='DDP')\n\n# # To automatically log gradients\n# wandb.watch(model, log_freq=100)\n \n# if torch.cuda.is_available():\n# print(\"[INFO] Using GPU: {}\\n\".format(torch.cuda.get_device_name()))\n \n start = time.time()\n best_epoch_loss = np.inf\n history = defaultdict(list)\n \n for epoch in range(1, num_epochs + 1): \n train_start = time.time()\n train_epoch_loss = train_one_epoch(model, optimizer, scheduler, \n dataloader=train_loader, \n accelerator=accelerator, epoch=epoch)\n \n train_end = time.time()\n time_elapsed = train_end - train_start\n accelerator.print('Epoch {} Training complete in {:.0f}h {:.0f}m {:.0f}s'.format(\n epoch, time_elapsed // 3600, (time_elapsed % 3600) // 60, (time_elapsed % 3600) % 60))\n \n \n val_start = time.time()\n val_epoch_loss = valid_one_epoch(model, optimizer, valid_loader, accelerator=accelerator, \n epoch=epoch)\n \n val_end = time.time()\n time_elapsed = val_end - val_start\n accelerator.print('Epoch {} Validation complete in {:.0f}h {:.0f}m {:.0f}s'.format(\n epoch, time_elapsed // 3600, (time_elapsed % 3600) // 60, (time_elapsed % 3600) % 60))\n \n \n history['Train Loss'].append(train_epoch_loss)\n history['Valid Loss'].append(val_epoch_loss)\n\n# # Log the metrics\n# wandb.log({\"Train Loss\": train_epoch_loss})\n# wandb.log({\"Valid Loss\": val_epoch_loss})\n# wandb.log({\"LR\": optimizer.param_groups[0]['lr']})\n \n # save model\n accelerator.print(f\"{b_}Validation Loss: {val_epoch_loss}\")\n# run.summary[\"Best Loss\"] = best_epoch_loss\n PATH = \"Loss{:.4f}_epoch{:.0f}.bin\".format(val_epoch_loss, epoch)\n accelerator.wait_for_everyone()\n unwrapped_model = accelerator.unwrap_model(model)\n accelerator.save(unwrapped_model.state_dict(), PATH)\n # Save a model file from the current directory\n accelerator.print(f\"Model Saved{sr_}\")\n \n del train_epoch_loss, val_epoch_loss, train_start, train_end, time_elapsed, val_start, val_end\n gc.collect()\n \n accelerator.print()\n \n end = time.time()\n time_elapsed = end - start\n accelerator.print('Training complete in {:.0f}h {:.0f}m {:.0f}s'.format(\n time_elapsed // 3600, (time_elapsed % 3600) // 60, (time_elapsed % 3600) % 60))\n# run.finish()","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:11:08.038010Z","iopub.execute_input":"2022-03-01T13:11:08.038334Z","iopub.status.idle":"2022-03-01T13:11:08.052799Z","shell.execute_reply.started":"2022-03-01T13:11:08.038297Z","shell.execute_reply":"2022-03-01T13:11:08.051805Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def fetch_scheduler(optimizer, train_loader):\n if CONFIG['scheduler'] == 'CosineAnnealingLR':\n scheduler = lr_scheduler.CosineAnnealingLR(optimizer,T_max=CONFIG['T_max'], \n eta_min=CONFIG['min_lr'])\n elif CONFIG['scheduler'] == 'CosineAnnealingWarmRestarts':\n scheduler = lr_scheduler.CosineAnnealingWarmRestarts(optimizer,T_0=CONFIG['T_0'], \n eta_min=CONFIG['min_lr'])\n \n elif CONFIG['scheduler'] == 'OneCycleLR':\n scheduler = lr_scheduler.OneCycleLR(optimizer,max_lr =CONFIG['learning_rate'],total_steps = CONFIG['epochs'] * len(train_loader))\n \n elif CONFIG['scheduler'] == None:\n return None\n \n return scheduler","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.251248Z","iopub.execute_input":"2022-03-01T13:10:45.251554Z","iopub.status.idle":"2022-03-01T13:10:45.266966Z","shell.execute_reply.started":"2022-03-01T13:10:45.251488Z","shell.execute_reply":"2022-03-01T13:10:45.265871Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def prepare_loaders(df, fold):\n df_train = df[df.kfold != fold].reset_index(drop=True)\n df_valid = df[df.kfold == fold].reset_index(drop=True)\n \n train_dataset = HappyWhaleDataset(df_train, transforms=data_transforms[\"train\"])\n valid_dataset = HappyWhaleDataset(df_valid, transforms=data_transforms[\"valid\"])\n\n train_loader = DataLoader(train_dataset, batch_size=CONFIG['train_batch_size'], \n num_workers=2, shuffle=True, drop_last=True)\n valid_loader = DataLoader(valid_dataset, batch_size=CONFIG['valid_batch_size'], \n num_workers=2, shuffle=False)\n \n return train_loader, valid_loader","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:10:45.268339Z","iopub.execute_input":"2022-03-01T13:10:45.268788Z","iopub.status.idle":"2022-03-01T13:10:45.283045Z","shell.execute_reply.started":"2022-03-01T13:10:45.268751Z","shell.execute_reply":"2022-03-01T13:10:45.281923Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Start Training","metadata":{}},{"cell_type":"code","source":"def launch():\n run_training(num_epochs=CONFIG['epochs'])\n \nnotebook_launcher(launch)","metadata":{"execution":{"iopub.status.busy":"2022-03-01T13:11:10.292591Z","iopub.execute_input":"2022-03-01T13:11:10.293346Z","iopub.status.idle":"2022-03-01T13:11:45.659222Z","shell.execute_reply.started":"2022-03-01T13:11:10.293302Z","shell.execute_reply":"2022-03-01T13:11:45.657359Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"![Upvote!](https://img.shields.io/badge/Upvote-If%20you%20like%20my%20work-07b3c8?style=for-the-badge&logo=kaggle)","metadata":{}}]}