文章预览
LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 1、[CV] LLMs can see and hear without any training 2、[CL] Beyond Turn-taking:Introducing Text-based Overlap into Human-LLM Interactions 3、[LG] Joint Learning of Energy-based Models and their Partition Function 4、[LG] Diverse Preference Optimization 5、[LG] Think Smarter not Harder:Adaptive Reasoning with Inference Aware Optimization 摘要:大语言模型无需任何训练就能看会听、在人与大模型交互中引入文本重叠、基于能量的模型及其配分函数的联合学习、多样化偏好优化、基于推算感知优化的自适应推理 1、[CV] LLMs can see and hear without any training K Ashutosh, Y Gandelsman, X Chen, I Misra… [Meta AI] 大语言模型无需任何训练就能看会听 要点: 多模态迭代LLM求解器 (MILS): 论文提出 MILS,一种新的免训练方法,赋予大型语言模型 (LLMs) 多模态能力(视觉和听觉)。这无需任何特定任务的训练
………………………………