model-wrapper 0.1.0

Model wrapper for Pytorch, which can training, predict, evaluate, etc.

Usage Sample ''''''''''''

.. code:: python

    from model_wrapper import SplitClassModelWrapper

    classes = ['class1', 'class2', 'class3'...]
    X = [[...], [...],]
    y = [0, 0, 1, 2, 1...]

    model = ...
    wrapper = SplitClassModelWrapper(model, classes=classes)
    wrapper.train(X, y, val_size=0.2)

    X_test = [[...], [...],]
    y_test = [0, 1, 1, 2, 1...]
    result = wrapper.evaluate(X_test, y_test)
    # 0.953125

    result = wrapper.predict(X_test)
    # [0, 1]

    result = wrapper.predict_classes(X_test)
    # ['class1', 'class2']

    result = wrapper.predict_proba(X_test)
    # ([0, 1], array([0.99439645, 0.99190724], dtype=float32))

    result = wrapper.predict_classes_proba(X_test)
    # (['class1', 'class2'], array([0.99439645, 0.99190724], dtype=float32))