SEASON：通过自诊断对比解码缓解视频大语言模型中的时序幻觉
SEASON: Mitigating Temporal Hallucination in Video Large Language Models via Self-Diagnostic Contrastive Decoding
提出免训练的对比解码框架，逐 token 诊断幻觉倾向并构造时序/空间负样本做自适应对比，可直接套用到任意 VideoLLM 的推理阶段以抑制时序不一致描述，无需重训。
原文依据 we propose Self-Diagnostic Contrastive Decoding (SEASON), a training-free method that adaptively enhances temporal and spatial faithfulness for each output token
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视频大语言模型（VideoLLM）在视频理解方面已取得显著进展。然而，这些模型在响应用户查询时，仍难以有效感知和利用视频中丰富的时序信息。因此，它们常常生成时序不一致或因果上不合理的的事件描述，造成严重的幻觉问题。以往多数研究关注空间幻觉（例如物体不匹配），而视频理解中的时序推理仍相对缺乏探索。为解决该问题，我们提出自诊断对比解码（SEASON），一种免训练方法，能够针对每个输出 token 自适应地增强时序与空间忠实度。该方法通过动态诊断每个 token 的幻觉倾向，并对其对应的时序负样本与空间负样本施加自适应对比解码来实现。大量实验表明，SEASON 在三个幻觉评测基准上优于所有现有的免训练幻觉缓解方法，同时在四个通用视频理解基准上进一步提升了 VideoLLM 的性能。项目主页：https://chriswu018.github.io/season/
英文原文摘要 1223 字符
Video Large Language Models (VideoLLMs) have shown remarkable progress in video understanding. However, these models still struggle to effectively perceive and exploit rich temporal information in videos when responding to user queries. Therefore, they often generate descriptions of events that are temporal inconsistent or causally implausible, causing severe hallucination issues. While most prior studies have focused on spatial hallucinations (e.g. object mismatches), temporal reasoning in video understanding remains relatively underexplored. To address this issue, we propose Self-Diagnostic Contrastive Decoding (SEASON), a training-free method that adaptively enhances temporal and spatial faithfulness for each output token. It achieves this by dynamically diagnosing each token's hallucination tendency and applying adaptive contrastive decoding against its corresponding temporal and spatial negatives. Extensive experiments demonstrate that SEASON outperforms all existing training-free hallucination mitigation approaches on three hallucination examination benchmarks, while further improves VideoLLMs across four general video understanding benchmarks. Our project page: https://chriswu018.github.io/season/