aha-moment-vlm-verification

【设计与创意】Reveals that inference-time scaling techniques for LLMs don't transfer to VLMs: majority voting beats verification, self-correction happens in <10% of cases, and models verify better without images. Use insights to design VLM evaluation methods that work rather than assuming LLM techniques apply directly.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准