Tag: Cross-Modal Integration

All the papers with the tag "Cross-Modal Integration".

• EchoInk-R1: Exploring Audio-Visual Reasoning in Multimodal LLMs via Reinforcement Learning

  grok-3-latest

  Score: 0.60

  Published:2025年5月7日 at  at 17:59

  #Multimodal LLM, #Reinforcement Learning, #Audio-Visual Reasoning, #Cross-Modal Integration

  本文提出 EchoInk-R1 框架，通过 Group Relative Policy Optimization 强化学习显著提升多模态大语言模型在音频-视觉推理任务上的性能，首次实现音频、视觉和文本模态的统一开放世界推理。