← 研踪 PaperTrailarXiv:2610.03792 · v3
What Do Verifiable Rewards Teach Video-Language Models About Time? A Controlled Multi-Model Study
首次提交 2026-10-01 05:42
版本更新 2026-11-04 05:11
雷达观测 2026-11-05 10:08
摘要
Reinforcement learning from verifiable rewards (RLVR) has produced large reasoning gains in language models, and verifiable video benchmarks make the recipe directly applicable to causal-temporal video question answering. We present a controlled study of what RLVR actually teaches video-language models about temporal structure. We fine-tune four open models with group relative policy optimization (GRPO) under three data recipes using annotated synthetic questions, self-supervised real-video pretext tasks, and a one-to-one mixture, plus a targeted real-video augmentation arm. We evaluate the model-recipe grid on 105 CLEVRER questions with one correct option and on two real-video transfer sets. The 8B ordered gain is 4.8 points (95% video-bootstrap CI [-5.6, 15.1]), while blind accuracy increases 10.5 points ([0.9, 20.6]). The 4B model gains an accuracy advantage for ordered over shuffled frames. On the real-video sets, the 8B synthetic-only checkpoint loses 26.7 and 25.2 points, with smaller regressions under mixed and augmented recipes. In a separate scene-disjoint, single-gold Qwen3-VL-8B experiment, supervised fine-tuning (SFT) and GRPO each complete 1,200 updates at three fixed seeds. GRPO minus SFT is -3.89 points in ordered accuracy (nominal 97.5% video-bootstrap CI [-7.36, -0.45]) and -2.78 points in ordered-minus-shuffled accuracy ([-6.60, 0.87]). Transfer differences are sensitive to parseability. Intervals quantify test-video uncertainty conditional on the fitted runs. The results show why benchmark accuracy, transfer, and presented-order sensitivity need to be evaluated separately.
作者备注：8 pages, 2 figures. Accepted to WACV 2027
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v3 · 2026-11-04 05:11 · 最新观测版本
What Do Verifiable Rewards Teach Video-Language Models About Time? A Controlled Multi-Model Study
Reinforcement learning from verifiable rewards (RLVR) has produced large reasoning gains in language models, and verifiable video benchmarks make the recipe directly applicable to causal-temporal video question answering. We present a controlled study of what RLVR actually teaches video-language models about temporal structure. We fine-tune four open models with group relative policy optimization (GRPO) under three data recipes using annotated synthetic questions, self-supervised real-video pretext tasks, and a one-to-one mixture, plus a targeted real-video augmentation arm. We evaluate the model-recipe grid on 105 CLEVRER questions with one correct option and on two real-video transfer sets. The 8B ordered gain is 4.8 points (95% video-bootstrap CI [-5.6, 15.1]), while blind accuracy increases 10.5 points ([0.9, 20.6]). The 4B model gains an accuracy advantage for ordered over shuffled frames. On the real-video sets, the 8B synthetic-only checkpoint loses 26.7 and 25.2 points, with smaller regressions under mixed and augmented recipes. In a separate scene-disjoint, single-gold Qwen3-VL-8B experiment, supervised fine-tuning (SFT) and GRPO each complete 1,200 updates at three fixed seeds. GRPO minus SFT is -3.89 points in ordered accuracy (nominal 97.5% video-bootstrap CI [-7.36, -0.45]) and -2.78 points in ordered-minus-shuffled accuracy ([-6.60, 0.87]). Transfer differences are sensitive to parseability. Intervals quantify test-video uncertainty conditional on the fitted runs. The results show why benchmark accuracy, transfer, and presented-order sensitivity need to be evaluated separately.
查看此版本解读 →查看此版本原文 ↗