Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?
Yue et al., 2025, Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model? (arXiv preprint arXiv:2504.13837)
Subscriber deck — free during early access
1 cards in “4.3. Effects of Different RL Algorithms” · whole deck
4.3. Effects of Different RL Algorithms
- What is the sampling efficiency gap, and how did RL algorithms compare on it?\(\Delta_{SE}\) = the base model’s pass@256 minus the RL model’s pass@1: how far RL is from turning everything the base model can do into first-try success. PPO, GRPO, Reinforce++, RLOO, ReMax and DAPO all differed only slightly and stayed above 40 points.
- Yue et al., 2025, Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model? (arXiv preprint arXiv:2504.13837), 4.3. Effects of Different RL Algorithms
Source
- Yang Yue, Zhiqi Chen, Rui Lu, Andrew Zhao, Zhaokai Wang, Yang Yue, Shiji Song and Gao Huang. 2025. Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?. In arXiv preprint arXiv:2504.13837. https://arxiv.org/abs/2504.13837