allenact.algorithms.onpolicy_sync.storage

[view_source]

RolloutStorage

class RolloutStorage(object)

[view_source]

Class for storing rollout information for RL trainers.

RolloutStorage.narrow

 | narrow()

[view_source]

This function is used by the training engine (in decentralized distributed settings) to temporarily narrow the step dimension in the storage.

The reverse operation, unnarrow, is automatically called by after_update.