open-vision-reasoner

【设计与创意】Transfer reasoning behaviors learned in language models to visual domains through two-stage training: cold-start linguistic fine-tuning followed by multimodal RL. Open-Vision-Reasoner achieves 95.3% on MATH500 and 54.6% on MathVerse by learning visual analogs of backtracking, verification, and subgoal decomposition using rule-based rewards.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准