arXiv cs.CL· Mayur Kulkarni·· 12 小时前AI 评分24
Reason-or-Recite：用 LLM 裁判协调两个互补专家的交通异常推理框架
Reason or Recite: Complementary Specialists with an LLM Judge for Traffic Anomaly Reasoning
AI 导读
Reason-or-Recite（RoR）在 2026 AI City Challenge 交通异常推理赛道（Track 3）排名第一，平均得分 0.6788。该框架将官方思维链标注改写为结构化、时间对齐的 schema，用 LoRA 微调 9B 视觉语言模型；另对 4-bit 量化的 27B 模型仅用答案微调以吸收标注风格，再由无需训练的 LLM 裁判逐题在两者间选择。
正文 · 原文
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.
| Comments: | 18 pages, 3 figures, 5 tables. Accepted at the AI City Challenge Workshop, ECCV 2026. 1st place in Track 3 (Traffic Anomaly Reasoning) |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) |
| ACM classes: | I.2.10; I.4.8 |
| Cite as: | arXiv:2610.34993 [cs.CV] |
| (or arXiv:2610.34993v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.34993 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mayur Kulkarni [view email]
[v1]
Mon, 26 Oct 2026 08:05:21 UTC (306 KB)
来源：arXiv cs.CL · arxiv.org