爱可可 AI 前沿推介(10.19)

文章预览

LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 AS - 音频与语音 RO - 机器人 1、[CL] Merge to Learn：Efficiently Adding Skills to Language Models with Model Merging 2、[LG] Cliqueformer：Model-Based Optimization with Structured Transformers 3、[LG] Artificial Kuramoto Oscillatory Neurons 4、[CL] Improving Instruction-Following in Language Models through Activation Steering 5、[CV] Janus：Decoupling Visual Encoding for Unified Multimodal Understanding and Generation 摘要：通过将模型合并为语言模型有效添加技能、基于模型的结构化Transformer优化、人工库拉莫托振荡神经元、通过激活引导改进语言模型的指令遵循、视觉编码解耦以实现统一的多模态理解和生成 1、[CL] Merge to Learn: Efficiently Adding Skills to Language Models with Model Merging J Morrison, N A. Smith, H Hajishirzi, P W Koh... [Allen Institute for AI] 通过合并学习：通过将模型合并为语言模型有效添加技能 要 ………………………………