fit_scores#

scio.eval.fit_scores(scores_and_layers, net, calib_data, calib_labels, *, show_progress=True)[source]#

Fit confidence scores given net and calibration data.

Parameters:

scores_and_layers (Collection[ScoreClassifAndLayers]) –
Instanciated classification score algorithms with their associated layers to record. Example:
```
scores_and_layers = [
    (IsoMax(), []),
    (KNN(k=6), [(1, 1)]),
]
```
net (nn.Module) – Neural net.
calib_data (Tensor) – Calibration data (InD).
calib_labels (Tensor) – Calibration labels.
show_progress (bool) – Whether to show progress bar. Defaults to True.

Returns:

scores_fit (list[BaseScoreClassif]) – Fit score functions.