msp.rl_algorithm package

Module contents

The mps.rl_algorithm module defines reinforcement learning algorithm to train the model.

class msp.rl_algorithm.ReinforceAlgorithm(objective_func, optimizer, train_metric, val_metric, tol=0.001)

Bases: object

apply_gradient(model_train, model_baseline, batch_train)

Apply the gradients to the trainable model weights.

Parameters

• model_train – model for the training

• model_baseline – model baseline for comparison as per REINFORCE ALOGRITHM

• batch_train – input mini batch data

evaluate(model_train, model_baseline, val_dataset, verbose=1, callbacks=None)

Perform model evaluation.

Parameters

• model_train – model for the training

• model_baseline – model baseline for comparison as per REINFORCE ALOGRITHM

• val_dataset – dataset for validation

• verbose – Verbosity mode. 0 = silent, 1 = progress bar.

• callbacks – List of callbacks to apply during evaluation.

run(model_train, model_baseline, train_dataset, val_dataset, epochs=1, verbose=1, callbacks=None, epoch_when_last_save=0)

Run the REINFORCE Algorithm on model_train and baseline.

Parameters

• model_train – model for the training

• model_baseline – model baseline for comparison as per REINFORCE ALOGRITHM

• train_dataset – dataset for training

• val_dataset – dataset for validation

• epochs – number of epochs

• verbose – Verbosity mode. 0 = silent, 1 = progress bar.

• callbacks – List of callbacks to apply during training.

• epoch_when_last_save – Used when retraining a model

train_model_for_one_epoch(model_train, model_baseline, train_dataset, callbacks, verbose=1)

Computes the loss then updates the weights and metrics for one epoch.

Parameters

• model_train – model for the training

• model_baseline – model baseline for comparison as per REINFORCE ALOGRITHM

• train_dataset – dataset for training

• callbacks – List of callbacks to apply during training.

• verbose – Verbosity mode. 0 = silent, 1 = progress bar.