面壁智能提出 Trident：解决多模态检索文本偏好问题
🔱 Why can irrelevant text outrank an image that actually contains the answer?
面壁智能（OpenBMB）提出 Trident，用 Multi-Positive View InfoNCE 联合训练 Text/Image/Fused 三视图，解决多模态检索中无关文本得分高于含答案图像的“Chaos in the Text”现象。在 6 个视觉文档检索基准上，Trident-Qwen3VL-2B 将 Mix Avg NDCG@10 从 40.12 提升至 73.90。分析显示，多模态检索器在纯文本或纯图像语料上表现良好，但图文混合时性能明显下降，原因是文本常获得比语义对齐图像更高的相似度分数。
正文
当前显示原文，中文全文译文尚未就绪。
🔱 Why can irrelevant text outrank an image that actually contains the answer?
We find that multimodal retrievers can perform well on text-only and image-only corpora, yet degrade sharply when text and images are mixed.
We call this phenomenon "Chaos in the Text."
Our analysis reveals a systematic Modality Preference: text often receives higher similarity scores than semantically aligned images. We introduce Trident, which jointly trains Text / Image / Fused views with Multi-Positive View InfoNCE.
On 6 visual-document retrieval benchmarks, Trident-Qwen3VL-2B improves Mix Avg NDCG@10 from 40.12 → 73.90.