Meta Chinese Researcher Releases ATLAS for Generalizable Visual Reasoning with One Word
Meta华人发布ATLAS,一个词搞定可泛化的视觉推理!
Meta AI and the Chinese University of Hong Kong proposed ATLAS, a visual reasoning method that uses one Functional Token to connect Agentic and Latent Visual Reasoning, with ATLAS-178K, a two-stage SFT+RL pipeline, and LA-GRPO to train sparse visual-operation tokens.
Why it matters: Meta AI and CUHK propose ATLAS, adding function words to a vision-language model so one inference pass can switch between explicit tool calls and internal latent computation.