OnPolicyReplay

Code: slm_lab/agent/memory/onpolicy.py​

OnPolicyReplay differs from the plain Replay in that when memory.sample()is called in a training step, it returns all of its stored elements, then flushes its storage clean. By default, on-policy training happens at the end of every episode.

Suitable for on-policy algorithms.

Source Documentation

Refer to the class documentation and example memory spec from the source: slm_lab/agent/memory/onpolicy.py#L14-L35​

Example Memory Spec

The spec for this memory has no parameters, since it automatically flushes at the end of an episode after training.

{
    ...
    "agent": [{
      "memory": {
        "name": "OnPolicyReplay"
      }
    }],
    ...
}

For more concrete examples of memory spec specific to algorithms, refer to the existing spec files.