爱可可 AI 前沿推介(11.3)

文章预览

LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 1、[LG] Generator Matching：Generative modeling with arbitrary Markov processes 2、[LG] Trajectory Flow Matching with Applications to Clinical Time Series Modeling 3、[LG]Flow Matching for Posterior Inference with Simulator Feedback 4、[LG] Mind Your Step (by Step)：Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse 5、[CL] Scaling LLM Inference with Optimized Sample Compute Allocation 摘要：任意马尔可夫过程的生成式建模、轨迹流匹配在临床时间序列建模中的应用、基于模拟器反馈的后验推理流匹配、思考会让人变更糟而思维链也会降低任务的表现、用样本计算分配优化扩展LLM推理 1、[LG] Generator Matching: Generative modeling with arbitrary Markov processes P Holderrieth, M Havasi, J Yim, N Shaul… [MIT CSAIL & FAIR, Meta] 生成器匹配：任意马尔可夫过程的生成式建模 要点: 生成器匹配 (GM) 的 ………………………………