Skip to content

fix FastSAC evaluate_policy missing set_is_evaluating call#53

Open
maedmatt wants to merge 1 commit intoamazon-far:mainfrom
maedmatt:fix/fast-sac-eval
Open

fix FastSAC evaluate_policy missing set_is_evaluating call#53
maedmatt wants to merge 1 commit intoamazon-far:mainfrom
maedmatt:fix/fast-sac-eval

Conversation

@maedmatt
Copy link

@maedmatt maedmatt commented Feb 13, 2026

Description of changes:

FastSAC's evaluate_policy() was missing a set_is_evaluating() call, unlike PPO which has it in _pre_evaluate_policy(). Without this the environment during evaluation:

  • Keeps resampling random velocity commands (command manager doesn't zero them)
  • Keeps applying push perturbations (randomization doesn't skip them)

This makes the robot appear to receive random actions and get pushed around during eval, rather than standing still awaiting keyboard/joystick input.

--
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

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