Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.11297 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908959944736768 |
|---|---|
| author | Liu, Yang Wang, Enxi Gao, Yufei Zhang, Weixin Wang, Bo Zeng, Zhiyuan Zhang, Yikai Zheng, Yining Qiu, Xipeng |
| author_facet | Liu, Yang Wang, Enxi Gao, Yufei Zhang, Weixin Wang, Bo Zeng, Zhiyuan Zhang, Yikai Zheng, Yining Qiu, Xipeng |
| contents | Despite the success of reinforcement learning for large language models, a common failure mode is reduced sampling diversity, where the policy repeatedly generates similar erroneous behaviors. Classical entropy regularization encourages randomness under the current policy, but does not explicitly discourage recurrent failure patterns across rollouts. We propose MEDS, a Memory-Enhanced Dynamic reward Shaping framework that incorporates historical behavioral signals into reward design. By storing and leveraging intermediate model representations, we capture features of past rollouts and use density-based clustering to identify frequently recurring error patterns. Rollouts assigned to more prevalent error clusters are penalized more heavily, encouraging broader exploration while reducing repeated mistakes. Across five datasets and three base models, MEDS consistently improves average performance over existing baselines, achieving gains of up to 4.13 pass@1 points and 4.37 pass@128 points. Additional analyses using both LLM-based annotations and quantitative diversity metrics show that MEDS increases behavioral diversity during sampling. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_11297 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The Past Is Not Past: Memory-Enhanced Dynamic Reward Shaping Liu, Yang Wang, Enxi Gao, Yufei Zhang, Weixin Wang, Bo Zeng, Zhiyuan Zhang, Yikai Zheng, Yining Qiu, Xipeng Machine Learning Artificial Intelligence Computation and Language Despite the success of reinforcement learning for large language models, a common failure mode is reduced sampling diversity, where the policy repeatedly generates similar erroneous behaviors. Classical entropy regularization encourages randomness under the current policy, but does not explicitly discourage recurrent failure patterns across rollouts. We propose MEDS, a Memory-Enhanced Dynamic reward Shaping framework that incorporates historical behavioral signals into reward design. By storing and leveraging intermediate model representations, we capture features of past rollouts and use density-based clustering to identify frequently recurring error patterns. Rollouts assigned to more prevalent error clusters are penalized more heavily, encouraging broader exploration while reducing repeated mistakes. Across five datasets and three base models, MEDS consistently improves average performance over existing baselines, achieving gains of up to 4.13 pass@1 points and 4.37 pass@128 points. Additional analyses using both LLM-based annotations and quantitative diversity metrics show that MEDS increases behavioral diversity during sampling. |
| title | The Past Is Not Past: Memory-Enhanced Dynamic Reward Shaping |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2604.11297 |