google-deepmind / google-deepmind/acme
DQN eval_epsilon overwrite with epsilon on 0.0
Open
- Dominant language
- Python
- Stars
- 4.1k
- Forks
- 553
- PR merge metrics
- No merged PRs in 30d
Description
Hi Acme team, I think JAX DQN might set the evaluation epsilon to the exploration epsilon if deterministic evaluation is requested (eps=0.0, [here](https://github.com/deepmind/acme/blob/860dbab686042573569b84223d8da6d43d09c304/acme/agents/jax/dqn/builder.py#L183)). Replacing this with `self._config.eval_epsilon is not None` fixed it for me - does this also occur on your end?
Thanks so much for checking!
Contributor guide
Assessment
This issue has not been assessed yet.