Hista and Numca: Estimate State Value Effectively for LLM Reinforcement Learning 事件

Hista and Numca: Estimate State Value Effectively for LLM Reinforcement Learning arXiv:2605.29782v1 Announce Type: cross Abstract: Reinforcement learning (RL) refines large language models (LLMs) by directly optimizing model behavior through reward signals. While accurate state value estimation is critical for stable training in classical RL, it remains an underexplored challenge in LLM post-training. In this work, we introduce the State Value Estimation Benchmark (SVEB) to assess state estimat