sfaira.estimators.EstimatorKerasEmbedding.evaluate

EstimatorKerasEmbedding.evaluate(batch_size: int = 128, max_steps: int = inf)

Evaluate the custom model on test data.

Defaults to run on full data if idx_test was not set before, ie. train() has not been called before.

Parameters

• batch_size – Batch size for evaluation.

• max_steps – Maximum steps before evaluation round is considered complete.

Returns

Dictionary of metric names and values.