Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2512.05962 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915838936743936 |
|---|---|
| author | Kruszewski, Germán Erbacher, Pierre Rozen, Jos Dymetman, Marc |
| author_facet | Kruszewski, Germán Erbacher, Pierre Rozen, Jos Dymetman, Marc |
| contents | Reinforcement Learning (RL) has become the de facto standard for tuning LLMs to solve tasks involving reasoning. However, growing evidence shows that models trained in such way often suffer from a significant loss in diversity. We argue that this arises because RL implicitly optimizes the "mode-seeking" or "zero-forcing" Reverse KL to a target distribution causing the model to concentrate mass on certain high-probability regions of the target while neglecting others. In this work, we instead begin from an explicit target distribution, obtained by filtering out incorrect answers while preserving the relative probabilities of correct ones. Starting from a pre-trained LLM, we approximate this target distribution using the $α$-divergence family, which unifies prior approaches and enables direct control of the precision-diversity trade-off by interpolating between mode-seeking and mass-covering divergences. On a Lean theorem-proving benchmark, our method achieves state-of-the-art performance along the coverage-precision Pareto frontier, outperforming all prior methods on the coverage axis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_05962 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Whatever Remains Must Be True: Filtering Drives Reasoning in LLMs, Shaping Diversity Kruszewski, Germán Erbacher, Pierre Rozen, Jos Dymetman, Marc Machine Learning Artificial Intelligence Reinforcement Learning (RL) has become the de facto standard for tuning LLMs to solve tasks involving reasoning. However, growing evidence shows that models trained in such way often suffer from a significant loss in diversity. We argue that this arises because RL implicitly optimizes the "mode-seeking" or "zero-forcing" Reverse KL to a target distribution causing the model to concentrate mass on certain high-probability regions of the target while neglecting others. In this work, we instead begin from an explicit target distribution, obtained by filtering out incorrect answers while preserving the relative probabilities of correct ones. Starting from a pre-trained LLM, we approximate this target distribution using the $α$-divergence family, which unifies prior approaches and enables direct control of the precision-diversity trade-off by interpolating between mode-seeking and mass-covering divergences. On a Lean theorem-proving benchmark, our method achieves state-of-the-art performance along the coverage-precision Pareto frontier, outperforming all prior methods on the coverage axis. |
| title | Whatever Remains Must Be True: Filtering Drives Reasoning in LLMs, Shaping Diversity |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2512.05962 |