fix: merge experience when missing ID#209
Merged
YZ-Cai merged 2 commits intoDec 5, 2025
Merged
Conversation
Collaborator
|
It appears you are currently using the preview version of the Training-Free GRPO implementation, which has been archived as noted in the current README. We have released a stable version of Training-Free GRPO. To other unstable issues present in the preview version, please use Agent Practice Module. |
YZ-Cai
approved these changes
Dec 5, 2025
YZ-Cai
merged commit Dec 5, 2025
d90f266
into
TencentCloudADP:training_free_GRPO
1 of 2 checks passed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
During batch updates, the LLM sometimes hallucinates non-existent IDs for merging, which previously crashed the pipeline. This fix changes the behavior to ignore invalid IDs while still processing valid ones. This is crucial because completely aborting the merge would discard valuable new insights and fail to remove redundant entries, leading to wasted critiques and an increasingly bloated context window.