Add adaptive turn budget and early stopping #33
Conversation
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
This implements a meta-controller that dynamically predicts the optimal number of turns per problem based on complexity signals, and enables confidence-based early stopping to improve efficiency. Key changes: - Created AdaptiveTurnPredictor class with heuristic complexity estimation using prompt length, token diversity, entropy, and historical performance - Integrated per-sample turn prediction into AgentHelper.run_llm_loop - Added confidence-based early stopping logic with convergence detection - Extended meta_info to track predicted turns, early stopping, and confidence scores - Added configuration section in simpletir_trainer.yaml for adaptive turns (disabled by default) - Scales to 50+ turns for hard problems while maintaining backward compatibility
MagellaX
force-pushed
the
feature/adaptive-turn-budget
branch
from
c97e061 to
74f3802
Compare
Author
|
Hey @ltzheng @AIDefender u can merge this, any thoughts from you side? |
Author
let me know your thoughts, will be deleting the branch later, so please merge this, it has been thoroughly tested... |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
This implements a meta-controller that dynamically predicts the optimal number of turns per problem based on complexity signals, and enables confidence-based early stopping to improve efficiency.
Key changes: