arXiv: cs.AI preprintsInternational11 November 2026
Margin-Guided Token Credit for Reinforcement Learning with Verifiable Rewards
This is an official announcement record
Firsthand records what arXiv: cs.AI preprints announced and links to the original. The wording below is theirs, not ours.
arXiv:2611.14015v1 Announce Type: cross Abstract: In reinforcement learning with verifiable rewards, a language model is rewarded for a correct final answer, and GRPO passes that reward to every token of the solution in equal measure. We instead place the credit where the model hesitated, in a method we call MCPO (Margin-Credit Policy Optimization). We measure hesitation by the top-2 margin $M=p_1-p_2$ of the policy's next-token distribution, map it to a score with two fixed thresholds, and spread each solution's advantage over its tokens so that the mean stays unchanged. The margin is read fr
Read the official announcement
Opens arxiv.org
More from arXiv: cs.AI preprints
- Auditable Human-Agent Teams for Autonomous X-ray Experimentation11 November 2026
- EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery11 November 2026
- Train on prefill, monitor at decode: Using sparse autoencoders for topic monitoring11 November 2026
- LandslideFM: A Foundation Model Towards Unified Landslide Perception and Understanding11 November 2026
- Tracing Psychometric Inference in Large Language Models11 November 2026
This content is for informational purposes only and is not professional advice. Specifications, prices, plan tiers, and features change frequently and may differ from what is shown here; verify current details on the manufacturer's or company's official page before purchasing. Ratings are based on analysis of published documentation, not independent lab testing.