跳到正文
返回论文列表
cs.CV提交于 已译

推理还是复述:面向交通异常推理的互补专家与 LLM 裁判

Reason or Recite: Complementary Specialists with an LLM Judge for Traffic Anomaly Reasoning

Mayur Kulkarni

中文摘要

道路碰撞、违规和突发危险等异常不仅要求视频检测,还需验证事件经过、描述现场并解释成因。实验发现,视觉语言模型的两类能力存在冲突:封闭式问题依赖精确且有证据支撑的推理,开放式答案则按与参考文本的相似度评分,标注风格与推理正确性同样重要。面向 2026 AI City Challenge 的交通异常推理赛道,提出 Reason-or-Recite(RoR)双专家框架:将官方思维链标注改写为结构化、带时间依据的模式,并用 LoRA 微调 9B 视觉语言模型;同时仅用答案微调 4-bit 量化的 27B 模型,使其吸收标注风格;再通过免训练的 LLM 裁判按问题选择候选结果。RoR 在域内排行榜以平均分 0.6788 位居第一。

关键要点

  1. 01交通异常理解需同时完成事件验证、现场描述与成因解释,能力要求高于检测。
  2. 02封闭式推理强调证据精确,开放式生成则受参考答案标注风格显著影响。
  3. 03RoR 以 9B 推理专家和 4-bit 量化 27B 复述专家生成互补候选答案。
  4. 04免训练的 LLM 裁判根据具体问题在两类候选结果之间进行选择。
  5. 05RoR 在 2026 AI City Challenge 域内排行榜平均分 0.6788,排名第一。

解读

尚无解读。

原始英文摘要

arXiv:2610.34993v1 Announce Type: new Abstract: Road anomalies such as collisions, violations, and sudden hazards demand more from video understanding than detection: a practical system must verify what happened, describe the scene, and explain the cause. We find these competences pull in opposite directions for vision-language models: closed-form questions reward precise, evidence-grounded reasoning, while open-ended answers are scored by similarity to reference text, where matching the annotation style matters as much as reasoning correctly. We present Reason-or-Recite (RoR), a two-specialist framework for the Traffic Anomaly Reasoning track of the 2026 AI City Challenge. RoR (1) rewrites the official chain-of-thought annotations into a structured, temporally grounded schema and LoRA-tunes a 9B vision-language model on the enhanced traces; (2) fine-tunes a 4-bit-quantized 27B model on answers alone, absorbing the annotation style; and (3) selects between the two candidates per question with a training-free LLM judge. RoR ranks first on the in-domain leaderboard with mean score 0.6788.

同方向论文 · cs.CV

查看全部 →

反向发现 · 这篇论文相关的工具与提示词

去交叉发现 →

关联工具 · 1

AWS Machine Learning Blog68

Amazon SageMaker AI 推出训练会话 API 与训练 cookbook

Amazon SageMaker AI 推出 training session APIs,允许开发者在本地 Python 流程中编写训练循环,由服务托管 GPU、模型与 LoRA 加载、分布式执行、故障恢复和检查点。

推荐理由:文章拆解了训练会话 API、GRPO 循环和 Code RL 配方,并给出公开示例与 cookbook 入口,便于评估自定义训练循环的实现方式。