Package ai.djl.modality.rl

Interface ReplayBuffer

All Known Implementing Classes:
LruReplayBuffer
public interface ReplayBuffer
Records RlEnv.Steps so that they can be trained on.

Using a replay buffer ensures that a variety of states are trained on for every training batch making the training more stable.

  • Method Summary

    Modifier and Type
    Method
    Description
    void
    addStep(RlEnv.Step step)
    Adds a new step to the buffer.
    RlEnv.Step[]
    getBatch()
    Returns a batch of steps from this buffer.

  • Method Details

    • getBatch

      RlEnv.Step[] getBatch()
      Returns a batch of steps from this buffer.
      Returns:
      a batch of steps from this buffer

    • addStep

      void addStep(RlEnv.Step step)
      Adds a new step to the buffer.
      Parameters:
      step - the step to add