← 研踪 PaperTrailarXiv:2608.27065 · v2
Video-OPSD: Exploiting Privileged Visual Evidence for On-Policy Self-Distillation in Video Large Language Models
首次提交 2026-08-27 20:51
版本更新 2026-11-18 14:14
雷达观测 2026-11-19 19:08
摘要
On-policy self-distillation (OPSD) has recently emerged as an effective post-training paradigm that provides dense token-level supervision from a privileged self-teacher, yet it remains underexplored for Video Large Language Models (Video-LLMs). Existing methods typically construct privileged teachers by augmenting their context with additional information while keeping the primary input unchanged for both teacher and student. In video reasoning, however, privileged information can also be found within the primary input itself: videos contain substantial temporal redundancy, and only a few frames carry the evidence needed to answer a question. Building on this observation, we present , an OPSD framework that exploits privileged visual evidence for both self-teacher construction and knowledge transfer. First, our Evidence-Grounded Self-Teacher conditions the teacher exclusively on annotated evidence frames, while the student reasons over a broader set of frames that contains them. This focused visual context enables more targeted supervision without introducing visual content unavailable to the student. Second, our Evidence-Guided Token Optimization adaptively weights token-level distillation according to an estimate of each reasoning token's reliance on privileged visual evidence, thereby emphasizing perceptually grounded reasoning. Across five video benchmarks and three Qwen-VL backbones, consistently outperforms Standard OPSD by 1.8-2.3 percentage points on average, establishing an effective and efficient post-training approach for Video-LLMs.
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。
v2 · 2026-11-18 14:14 · 最新观测版本
Video-OPSD: Exploiting Privileged Visual Evidence for On-Policy Self-Distillation in Video Large Language Models
On-policy self-distillation (OPSD) has recently emerged as an effective post-training paradigm that provides dense token-level supervision from a privileged self-teacher, yet it remains underexplored for Video Large Language Models (Video-LLMs). Existing methods typically construct privileged teachers by augmenting their context with additional information while keeping the primary input unchanged for both teacher and student. In video reasoning, however, privileged information can also be found within the primary input itself: videos contain substantial temporal redundancy, and only a few frames carry the evidence needed to answer a question. Building on this observation, we present , an OPSD framework that exploits privileged visual evidence for both self-teacher construction and knowledge transfer. First, our Evidence-Grounded Self-Teacher conditions the teacher exclusively on annotated evidence frames, while the student reasons over a broader set of frames that contains them. This focused visual context enables more targeted supervision without introducing visual content unavailable to the student. Second, our Evidence-Guided Token Optimization adaptively weights token-level distillation according to an estimate of each reasoning token's reliance on privileged visual evidence, thereby emphasizing perceptually grounded reasoning. Across five video benchmarks and three Qwen-VL backbones, consistently outperforms Standard OPSD by 1.8-2.3 percentage points on average, establishing an effective and efficient post-training approach for Video-LLMs.
查看此版本解读 →查看此版本原文 ↗