Skip to content

Add synchronous on-policy rollout mode#76

Draft
nightlessbaron wants to merge 1 commit into
prodfrom
agent/disable-async-rollout
Draft

Add synchronous on-policy rollout mode#76
nightlessbaron wants to merge 1 commit into
prodfrom
agent/disable-async-rollout

Conversation

@nightlessbaron

Copy link
Copy Markdown

Summary

  • add --disable-async-rollout
  • generate each rollout inline with current weights when enabled
  • refresh actor weights after every training step in synchronous mode
  • preserve the existing asynchronous path by default

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant