Options
All
  • Public
  • Public/Protected
  • All
Menu

Some metrics about how a model is performing during training or at validation.

Hierarchy

  • Metrics

Index

Properties

accuracy

accuracy: number

Accuracy of the model over the samples.

Accuracy = proportion of correctly predicted samples.

Optional averageConfidence

averageConfidence: number

Average confidence of the model in its prediction. Ideally, this value should be approximatively equal to the model's accuracy.

Only defined for validation metrics.

Optional epoch

epoch: number

Epoch of the training.

Not defined for validation metrics.

failingSamples

failingSamples: string[]

The array of names of the failling samples (under 0.999 accuracy).

Optional loss

loss: number

Model's average loss.

Only defined on training metrics.

precision

precision: number

Precision of the model over the samples.

Precision = (number of correctly assigned samples to a label) / (number of samples assigned to a label)

recall

recall: number

Recall of the model over the samples.

Recall = (number of correctly assigned samples to a label) / (number of samples that belong to a label)

Legend

  • Function
  • Type alias
  • Interface
  • Property
  • Class
  • Class with type parameter
  • Enumeration

Generated using TypeDoc