videoauto-r1-adaptive-video-reasoning

【效率与协作】Reduce inference latency and token cost in video question-answering by selectively invoking chain-of-thought reasoning. VideoAuto-R1 uses confidence-based early-exit to determine when explicit reasoning is necessary, achieving 3.3× response length reduction while maintaining state-of-the-art accuracy.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准