用大模型生成细粒度相关性标签，让检索模型更精准。
BiXSE: Improving Dense Retrieval via Probabilistic Graded Relevance Distillation
- 将大模型生成的相关性得分当作概率目标，单对样本即可训练。
- 在多个基准上超越传统对比学习方法，接近顶尖排序模型表现。
- 只需少量标注和计算，适合大规模检索模型训练。
密集检索的神经句子嵌入模型通常依赖二值相关性标签，将查询-文档对视为相关或不相关。然而真实场景中的相关性往往呈连续分布，而大语言模型（LLM）的发展使得生成细粒度分级相关性标签成为可能。本文提出BiXSE，一种简单有效的点式训练方法，在LLM生成的分级相关性得分上优化二元交叉熵（BCE）。BiXSE将这些得分解释为概率目标，使单个标注的查询-文档对即可提供精细监督。与需要多个标注比较的成对或列表损失不同，BiXSE通过利用批次内负样本，以更低的标注和计算成本实现优异性能。在句子嵌入（MMTEB）和检索基准（BEIR、TREC-DL）上的大量实验表明，BiXSE始终优于基于Softmax的对比学习（InfoNCE），在使用LLM监督数据训练时，达到甚至超过强成对排序基线。随着分级相关性监督日益普及，BiXSE为训练密集检索模型提供了稳健且可扩展的替代方案。
原文摘要 · Abstract (English)
Neural sentence embedding models for dense retrieval typically rely on binary relevance labels, treating query-document pairs as either relevant or irrelevant. However, real-world relevance often exists on a continuum, and recent advances in large language models (LLMs) have made it feasible to scale the generation of fine-grained graded relevance labels. In this work, we propose BiXSE, a simple and effective pointwise training method that optimizes binary cross-entropy (BCE) over LLM-generated graded relevance scores. BiXSE interprets these scores as probabilistic targets, enabling granular supervision from a single labeled query-document pair per query. Unlike pairwise or listwise losses that require multiple annotated comparisons per query, BiXSE achieves strong performance with reduced annotation and compute costs by leveraging in-batch negatives. Extensive experiments across sentence embedding (MMTEB) and retrieval benchmarks (BEIR, TREC-DL) show that BiXSE consistently outperforms softmax-based contrastive learning (InfoNCE), and matches or exceeds strong pairwise ranking baselines when trained on LLM-supervised data. BiXSE offers a robust, scalable alternative for training dense retrieval models as graded relevance supervision becomes increasingly accessible.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。