torchtraining.metrics.classification.multiclass¶

class torchtraining.metrics.classification.multiclass.Accuracy(reduction=<built-in method mean of type object>)[source]¶

Calculate accuracy score between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable(torch.Tensor) -> Any, optional) – One argument callable getting torch.Tensor and returning argument after specified reduction. Default: torch.mean (mean across batch, user can use torchtraining.savers.Mean to get mean across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default mean is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.BalancedAccuracy[source]¶

Critical success index between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.ConfusionMatrix[source]¶

Confusion matrix between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable, optional) – One argument callable getting torch.Tensor and returning torch.Tensor. Default: torch.sum (sum of all elements, user can use torchtraining.savers.Sum to get sum across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default sum is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.CriticalSuccessIndex[source]¶

Critical success index between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.F1[source]¶

F1 score between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.FBeta(beta: float)[source]¶

Get f-beta score between outputs and targets.

Works for both logits and probabilities of output out of the box.

Parameters

beta (float) – Beta coefficient of f-beta score.
data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$

Returns

Scalar tensor

Return type

torch.Tensor

forward(data)[source]¶

class torchtraining.metrics.classification.multiclass.FalseDiscoveryRate[source]¶

False discovery rate between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.FalseNegative(reduction=<built-in method sum of type object>)[source]¶

Number of false negatives between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable, optional) – One argument callable getting torch.Tensor and returning torch.Tensor. Default: torch.sum (sum of all elements, user can use torchtraining.savers.Sum to get sum across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default sum is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.FalseNegativeRate[source]¶

False negative rate between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.FalseOmissionRate[source]¶

False omission rate between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.FalsePositive(reduction=<built-in method sum of type object>)[source]¶

Number of false positives between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable, optional) – One argument callable getting torch.Tensor and returning torch.Tensor. Default: torch.sum (sum of all elements, user can use torchtraining.savers.Sum to get sum across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default sum is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.FalsePositiveRate[source]¶

False positive rate between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.Jaccard(reduction=<built-in method mean of type object>)[source]¶

Calculate Jaccard score between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable(torch.Tensor) -> Any, optional) – One argument callable getting torch.Tensor and returning argument after specified reduction. Default: torch.mean (mean across batch, user can use torchtraining.savers.Mean to get mean across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default mean is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.MatthewsCorrelationCoefficient[source]¶

Matthews correlation coefficient between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.NegativePredictiveValue[source]¶

Negative predictive value between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.Precision[source]¶

Precision between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.Recall[source]¶

Recall between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.Specificity[source]¶

Specificity between output and target.

Works for both logits and probabilities of output out of the box.

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: Scalar tensor
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.TopK(k: int, reduction=<built-in method mean of type object>)[source]¶

Get top-k accuracy score between outputs and targets.

Works for both logits and probabilities of output out of the box.

Parameters

k (int) – How many top results should be chosen.
reduction (Callable, optional) – One argument callable getting torch.Tensor and returning torch.Tensor. Default: torch.sum (sum of all elements, user can use torchtraining.savers.Sum to get sum across iterations/epochs).
data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$

Returns

If reduction is left as default mean is taken and single value returned. Otherwise whatever reduction returns.

Return type

torch.Tensor

forward(data)[source]¶

class torchtraining.metrics.classification.multiclass.TrueNegative(reduction=<built-in method sum of type object>)[source]¶

Number of true negatives between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable, optional) – One argument callable getting torch.Tensor and returning torch.Tensor. Default: torch.sum (sum of all elements, user can use torchtraining.savers.Sum to get sum across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default sum is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor

class torchtraining.metrics.classification.multiclass.TruePositive(reduction=<built-in method sum of type object>)[source]¶

Number of false positives between output and target.

Works for both logits and probabilities of output out of the box.

Parameters: reduction (Callable, optional) – One argument callable getting torch.Tensor and returning torch.Tensor. Default: torch.sum (sum of all elements, user can use torchtraining.savers.Sum to get sum across iterations/epochs).

forward(data)[source]¶

Parameters: data (Tuple[torch.Tensor, torch.Tensor]) – Tuple containing outputs from neural network and targets (ground truths). outputs should be of shape $(N, *, C-1)$ , where $C$ is the number of classes. Should contain logits (unnormalized probabilities) or probabilities after softmax activation or similar. targets should be of shape $(N, *)$ and contain integers in the range $[0, C-1]$
Returns: If reduction is left as default sum is taken and single value returned. Otherwise whatever reduction returns.
Return type: torch.Tensor