Update evaluation to support MT1 and fix metalearning evaluation issue - #577
Open
rainx0r wants to merge 4 commits into
Open
Update evaluation to support MT1 and fix metalearning evaluation issue#577rainx0r wants to merge 4 commits into
rainx0r wants to merge 4 commits into
Conversation
axis and avoid misreported success rate in metalearning evaluation (rainx0r/metaworld-algorithms#10)
rainx0r
changed the title
rainx0r
changed the title
reginald-mclean
marked this pull request as ready for review
June 24, 2026 22:55
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A user of my metaworld-algorithms repository recently reported a bug with the metalearning evaluation procedure, primarily seeming to affect ML1 but likely also affects ML10 and ML45. This was discussed in this issue
This PR addresses this issue by decoupling how we track successes per task sampled in the meta batch from the task_name.
Draft status
What remains to be done is run some meta-learning experiments for ML10 and ML45 under this fix and compare