文章预览
LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 1、[LG] Formal Theorem Proving by Rewarding LLMs to Decompose Proofs Hierarchically 2、[CL] Teaching Models to Improve on Tape 3、[LG] "Give Me BF16 or Give Me Death"? Accuracy-Performance Trade-Offs in LLM Quantization 4、[LG] Nash Equilibria via Stochastic Eigendecomposition 5、[LG] Thinking Forward and Backward：Effective Backward Planning with Large Language Models 摘要：奖励大语言模型分层分解证明的形式定理证明、利用纠正性反馈训练大型语言模型满足特定约束的强化学习框架、大语言模型量化的精度-性能权衡、基于随机特征分解的纳什均衡、大型语言模型的有效后向规划 1、[LG] Formal Theorem Proving by Rewarding LLMs to Decompose Proofs Hierarchically K Dong, A Mahankali, T Ma [Stanford University] 奖励大语言模型分层分解证明的形式定理证明 要点: 奖励机制：本文提出了一种强化学习方法，即使主要定
………………………………
