New pruning method balances quality and diversity for reasoning models
Original title: Direct Pass@$k$ Optimization for Thought Pruning: Balancing Quality and Diversity in Test-Time Reasoning
Direct pass@k optimization selects partial solutions to maximize correct final answers while penalizing redundancy in test-time scaling.
Paper presents a pruning procedure that optimizes pass@k over partial solutions generated during reasoning. It simultaneously rewards high-quality prefixes and diversity of trajectories, directly addressing the exploration-exploitation trade-off. On eight open-weight reasoning models the method outperforms quality-based and diversity-based pruning, with largest gains when retaining a moderate number of candidates for further generation.
- Optimizes pass@k directly on partial solution pools
- Balances quality and diversity without separate heuristics
- Outperforms existing pruning on eight open-weight models
- Particularly effective at moderate candidate counts
From the original article
1 min read at arxiv.orgTest-time scaling has emerged as a key inference strategy for reasoning models, where a model generates a large number of candidate solutions in the hope of finding at least one correct solution. Running all candidate solutions to completion can be computationally demanding, so recent work prunes the candidate pool after generating only a short prefix, committing further computation only to the surviving candidates.
By Mahdi Salmani · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.