torcheval.metrics.functional.multiclass_accuracy

torcheval.metrics.functional.multiclass_accuracy(input: Tensor, target: Tensor, *, average: str | None = 'micro', num_classes: int | None = None, k: int = 1) → Tensor

Compute accuracy score, which is the frequency of input matching target. Its class version is torcheval.metrics.MultiClassAccuracy.

Parameters:

• input (Tensor) – Tensor of label predictions It could be the predicted labels, with shape of (n_sample, ). It could also be probabilities or logits with shape of (n_sample, n_class). torch.argmax will be used to convert input into predicted labels.

• target (Tensor) – Tensor of ground truth labels with shape of (n_sample, ).

• average –

  • 'micro' [default]:

    Calculate the metrics globally.

  • 'macro':

    Calculate metrics for each class separately, and return their unweighted mean. Classes with 0 true instances are ignored.

  • None:

    Calculate the metric for each class separately, and return the metric for every class. NaN is returned if a class has no sample in target.

• num_classes – Number of classes. Required for 'macro' and None average methods.

• k – Number of top probabilities to be considered. K should be an integer greater than or equal to 1. If k >1, the input tensor must contain probabilities or logits for every class.

Examples:

>>> import torch
>>> from torcheval.metrics.functional import multiclass_accuracy
>>> input = torch.tensor([0, 2, 1, 3])
>>> target = torch.tensor([0, 1, 2, 3])
>>> multiclass_accuracy(input, target)
tensor(0.5)
>>> multiclass_accuracy(input, target, average=None, num_classes=4)
tensor([1., 0., 0., 1.])
>>> multiclass_accuracy(input, target, average="macro", num_classes=4)
tensor(0.5)
>>> input = torch.tensor([[0.9, 0.1, 0, 0], [0.1, 0.2, 0.4, 0,3], [0, 1.0, 0, 0], [0, 0, 0.2, 0.8]])
>>> multiclass_accuracy(input, target)
tensor(0.5)