Monty.Experiment.RecognitionStrategy.MaximumSteps (ExTbpMonty.Core v0.1.0)

Copy Markdown View Source

Done once the matching steps reach the limit for the current mode.

Mirrors tbp/monty/experiment/recognition_strategy.py::MaximumSteps. Two things have to hold: the model is not exploring — an exploring step is training, not matching — and model.matching_steps has reached max_train_steps or max_eval_steps, chosen by mode.

Summary

Functions

call(model, counter, state)

Done when the mode's limit is reached, unless the model is exploring.

new(options)

A policy with a limit for each mode.

Functions

call(model, counter, state)

@spec call(map(), Monty.Experiment.RecognitionStrategy.RecognitionCounter.t(), map()) ::
  Monty.Experiment.RecognitionStrategy.RecognitionResult.t()

Done when the mode's limit is reached, unless the model is exploring.

new(options)

@spec new(keyword()) :: {:ok, {module(), map()}} | {:error, String.t()}

A policy with a limit for each mode.

Monty raises ValueError for a limit that is not positive; here that is {:error, message} with Monty's own words.