Tag: Variance Minimization

All the papers with the tag "Variance Minimization".

• Optimizing Chain-of-Thought Reasoners via Gradient Variance Minimization in Rejection Sampling and RL

  grok-3-latest

  Score: 0.58

  Published:2025年5月5日 at  at 06:26

  #LLM, #Chain of Thought, #Sampling, #Variance Minimization, #Reinforcement Learning

  本文提出GVM-RAFT方法，通过动态采样分配策略最小化梯度方差，显著提升大型语言模型在链式思维推理任务中的训练效率和性能。