| title | Beyond Reasoning: Reinforcement Learning Unlocks Parametric Knowledge in LLMs |
|---|---|
| authors | Wanli Yang, Hongyu Zang, Junwei Zhang, Wenjie Shi, Du Su, Jingang Wang, Xueqi Cheng, Fei Sun |
| date | 2026-05-08 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.07153 |
| upvotes | 6 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Reinforcement learning (RL) has achieved remarkable success in LLM reasoning, but whether it can also improve direct recall of parametric knowledge remains an open question. We study this question in a controlled zero-shot, one-hop, closed-book QA setting with no chain-of-thought, training only on binary correctness rewards and applying fact-level train-test deduplication to ensure gains reflect improved recall rather than reasoning or memorization. Across three model families and multiple factual QA benchmarks, RL yields ~27% average relative gains, surpassing both training- and inference-time baselines alike. Mechanistically, RL primarily redistributes probability mass over existing knowledge rather than acquiring new facts, moving correct answers from the low-probability tail into reliable greedy generations. Our data-attribution study reveals that the hardest examples are the most informative: those whose answers never appear in 128 pre-RL samples (only ~18% of training data) drive ~83% of the gain, since rare correct rollouts still emerge during training and get reinforced. Together, these findings broaden the role of RL beyond reasoning, repositioning it as a tool for unlocking rather than acquiring latent parametric knowledge. |
cite
@misc{yang2026beyond,
title={Beyond Reasoning: Reinforcement Learning Unlocks Parametric Knowledge in LLMs},
author={Wanli Yang and Hongyu Zang and Junwei Zhang and Wenjie Shi and Du Su and Jingang Wang and Xueqi Cheng and Fei Sun},
year={2026},
eprint={2605.07153},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.07153}
}