Functions — github.com/YU1ut/imet-6th-soltuion

CodeHub MCPcopy Create free account

hub / github.com/YU1ut/imet-6th-soltuion / functions

Functions49 in github.com/YU1ut/imet-6th-soltuion

⨍ Functions 49 ◇ Types & classes 11

↓ 3 callersFunction_reduce_loss

↓ 3 callersFunctionload_model

(model: nn.Module, path: Path)

↓ 3 callersFunctionmake_loader

(df: pd.DataFrame, image_transform)

↓ 3 callersFunctiontrain

(args, model: nn.Module, criterion, *, params, train_loader, valid_loader, init_optimizer, use_cuda,

↓ 3 callersFunctionwrite_event

(log, step: int, **data)

↓ 2 callersFunction_make_mask

(argsorted, top_n: int)

↓ 2 callersFunction_smooth

↓ 2 callersFunctionbinarize_prediction

Return matrix of 0/1 predictions, same shape as probabilities.

↓ 2 callersFunctioncreate_net

(net_cls, pretrained: bool)

↓ 2 callersFunctionget_ids

↓ 2 callersFunctionload_image

(item, root: Path)

↓ 2 callersFunctionmain

↓ 2 callersFunctionmean_df

(df: pd.DataFrame)

↓ 2 callersFunctionpredict

(model, root: Path, df: pd.DataFrame, out_path: Path, batch_size: int, tta: int, workers: int, use

↓ 2 callersFunctionvalidation

( model: nn.Module, criterion, valid_loader, use_cuda, )

↓ 1 callersMethod__init__

(self, num_classes, pretrained=False, net_cls=M.resnet50, dropout=False)

↓ 1 callersMethod__init__

(self, root: Path, df: pd.DataFrame, image_transform: Callable, tta: int)

↓ 1 callersMethod_get_item

↓ 1 callersFunctioncosine_rampdown

Cosine rampdown from https://arxiv.org/abs/1608.03983

↓ 1 callersFunctionget_score

↓ 1 callersFunctionload_transform_image

( item, root: Path, image_transform: Callable, debug: bool = False)

↓ 1 callersFunctionmain

make_folds.py:32

↓ 1 callersFunctionmake_folds

(n_folds: int, seed: int)

make_folds.py:11

↓ 1 callersFunctiontrain_load_transform_image

( item, root: Path, image_transform: Callable, debug: bool = False)

transforms.py:28

transforms.py:62

Method__getitem__

(self, idx: int)

Method__getitem__

(self, size, interpolation=Image.BILINEAR, min_aspect=4/5, max_aspect=5/4, m

transforms.py:18

(self, EPSILON = 0.5, sl = 0.02, sh = 0.4, r1 = 0.3, mean=[0.4914, 0.4822, 0.4465])

transforms.py:55

(self, num_classes, pretrained=False, net_cls=M.densenet121)

(self, root: Path, df: pd.DataFrame, image_transform: Callable, debug: bool = True)

(self, gamma=2)

Method_get_item

Functionadjust_learning_rate

if epoch <= 1: lr = np.interp([epoch], [0, 1], [0, max_lr])[0] else:

(self, logit, target)

Methodfresh_params

Methodfresh_params

Functionget_classes

make_submission.py:33

Functiongmean_df

(df: pd.DataFrame)

make_submission.py:10

Use in the notebook like this:: %matplotlib inline from imet.utils import plot plot('./runs/oc2', './runs/oc1', 'loss',