Use deterministic policies when evaluating PCN #75
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
I'm letting Lucas review this since he's more knowledgeable about this algorithm. Makes sense to me but surely impacts performance. |
|
Hi @vaidas-sl, thanks for the PR! Although in the original PCN code (https://github.com/mathieu-reymond/pareto-conditioned-networks/blob/main/pcn/pcn.py#L186C69-L186C69) the evaluation during training uses the stochastic policy, they evaluate using a deterministic policy after training. I think it makes sense that the evaluation results that we report are obtained using the deterministic policy, as it is the policy that matters at the end of training. |
LucasAlegre
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
According to PCN paper, "at execution time — i.e., after the training process — we use a deterministic policy by systematically selecting the action with the highest confidence" however current implementation always uses sampled actions