sitorchtools 0.0.1

A small package

sitorchtools

Support function for train dataset using Pytorch

Features

Early Stopping based on validation loss
Folder loader based on pytorch DataLoader
Imblanaced image data handling
Spliting Image on Folder to train and test dataset

Usage

1. EarlyStopping

from sitorchtools import EarlyStopping, folder_loader, img_folder_split
    early_stopping = EarlyStopping(
        patience=7,
        verbose=True,
        delta=0,
        path="best_model.pth",
        trace_func=print,
        model_class=None
        )

    early_stopping(model, train_loss, valid_loss, y_true, y_pred, plot=False)

2. Data Loader

train_set, train_loader = folder_loader.loader(
    your_train_path,
    transform=your_train_transform,
    batch_size=your_bs,
    imbalance=True)

3. image folder split

img_folder_split.split_folder(path_to_train_data, path_to_test_data, train_ratio)