VA-Judger: Reward Modeling from Human Preference Feedback for Joint Video-Audio Generation

  • 类型：arxiv
  • 标识：2608.18607
  • 链接：https://arxiv.org/abs/2608.18607
  • 主分类：multimodal
  • 形态：method
  • 被引：4
  • 被引来源：Semantic Scholar
  • S2被引：4
  • OpenAlex被引：0
  • 影响力被引：0
  • TLDR：VA-Judger is proposed, a chain-of-thought omni-reward model that learns from pairs with clear quality gaps to establish structured output and coarse preference discrimination, then distills reliable preference explanations for harder near-quality comparisons via rejection sampling verified against human annotations, and finally performs dimension-wise reinforcement learning that decomposes human feedback into individual quality dimensions for denser reward signals.
  • OpenAlex ID：W7203936654
  • OpenAlex DOI：10.48550/arxiv.2608.18607
  • DOI：10.48550/arxiv.2608.18607
  • DOI来源：OpenAlex
  • 开放获取：green
  • 开放获取链接：https://arxiv.org/pdf/2608.18607
  • OpenAlex更新：2026-08-28
  • 待LLM分类：否
  • 标题中文：VA-Judger：基于人类偏好反馈的联合音视频生成奖励建模
  • TLDR中文：提出 VA-Judger，一种链式思考的通用奖励模型：从质量差距明显的样本对中学习以建立结构化输出和粗粒度偏好判别，再通过拒绝采样（以人类标注验证）蒸馏出可靠的偏好解释用于更难的质量相近样本比较，最后执行维度级强化学习，将人类反馈分解到各独立质量维度以获得更稠密的奖励信号。
  • 来源文件：
  • /inbox/tom/_candidates/2026-08-21-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • [OpenAlex backfill]