{"slug": "answer-self-consistency-with-margin-triggered-question-re-arbitration-for-the", "title": "Answer Self-Consistency with Margin-Triggered Question Re-Arbitration for the CVPR 2026 VidLLMs Challenge", "summary": "Researchers proposed Answer Self-Consistency with Margin-Triggered Question Re-Arbitration (ASC-MQRA) for the CVPR 2026 VidLLMs Challenge Track 2, a training-free test-time reasoning framework that improves video relational reasoning through multiple stochastic question-answering runs. The ASC component achieved 81.16% average accuracy on the test set by aggregating answer choices across runs, while the MQRA module, designed to re-arbitrate low-confidence examples, showed validation improvements but slightly degraded test performance. The team submitted ASC without re-arbitration as their final solution, demonstrating that answer-level self-consistency substantially outperforms single-pass inference for multimodal reasoning in videos.", "body_md": "arXiv:2606.04323v1 Announce Type: new\nAbstract: In this report, we present our solution for Track 2 of the CVPR 2026 VidLLMs Challenge. This track evaluates visual relational reasoning in videos, where models must infer relations that are not always explicitly visible. We propose Answer Self-Consistency with Margin-Triggered Question Re-Arbitration (ASC-MQRA), a training-free test-time reasoning framework built on a multimodal reasoning model. The core ASC component performs multiple stochastic video question-answering runs and aggregates their answer choices through answer-level self-consistency. This substantially improves over single-pass inference and forms our final test submission. We further study MQRA, a conditional re-arbitration module for low-margin examples where the first-stage vote distribution indicates uncertainty. Our vote-margin analysis shows that low-margin examples often retain the ground-truth answer among the top candidates, motivating MQRA to narrow the candidate set and re-watch the video only over the retained candidates. On validation, MQRA further improves over ASC, indicating that low-margin vote distributions can provide a useful uncertainty signal. On test, however, MQRA slightly degrades performance relative to ASC, suggesting that re-arbitration is sensitive to the size and category distribution of the triggered subset. Our final test submission therefore uses ASC without re-arbitration, achieving 72.73 average accuracy and 78.34 category-wise macro average accuracy on validation, and 81.16 average accuracy and 80.91 category-wise macro average accuracy on test. This report details our prompting strategy, implementation setup, ablation studies, and diagnostic analyses. The code is available at https://github.com/data-analytics-labo/ASC-MQRA", "url": "https://wpnews.pro/news/answer-self-consistency-with-margin-triggered-question-re-arbitration-for-the", "canonical_source": "https://arxiv.org/abs/2606.04323", "published_at": "2026-06-04 04:00:00+00:00", "updated_at": "2026-06-04 04:20:17.972146+00:00", "lang": "en", "topics": ["computer-vision", "artificial-intelligence", "machine-learning", "large-language-models", "ai-research"], "entities": ["CVPR 2026 VidLLMs Challenge", "ASC-MQRA", "Answer Self-Consistency with Margin-Triggered Question Re-Arbitration"], "alternates": {"html": "https://wpnews.pro/news/answer-self-consistency-with-margin-triggered-question-re-arbitration-for-the", "markdown": "https://wpnews.pro/news/answer-self-consistency-with-margin-triggered-question-re-arbitration-for-the.md", "text": "https://wpnews.pro/news/answer-self-consistency-with-margin-triggered-question-re-arbitration-for-the.txt", "jsonld": "https://wpnews.pro/news/answer-self-consistency-with-margin-triggered-question-re-arbitration-for-the.jsonld"}}