zELO：受ELO启发的重排序器与嵌入模型训练方法

zELO: ELO-inspired Training Method for Rerankers and Embedding Models

September 16, 2025

作者: Nicholas Pipitone, Ghita Houir Alami, Advaith Avadhanam, Anton Kaminskyi, Ashley Khoo

cs.AI

摘要

我們提出了一種名為zELO的新穎訓練方法，該方法通過分析排名任務在靜態上等同於瑟斯頓模型來優化檢索性能。基於zELO方法，我們利用無監督數據訓練了一套最先進的開放權重重排序模型：zerank-1和zerank-1-small。這些模型在多個領域，包括金融、法律、代碼和STEM中，達到了最高的檢索分數，在NDCG@10和召回率上均超越了閉源專有重排序器。這些模型還展現了極強的適應性，在域外和私人客戶數據集上保持了零樣本性能。訓練數據包括112,000個查詢，每個查詢對應100份文檔，並在不到10,000個H100小時內，從未標註的查詢和文檔中進行了端到端的訓練。

English

We introduce a novel training methodology named zELO, which optimizes retrieval performance via the analysis that ranking tasks are statically equivalent to a Thurstone model. Based on the zELO method, we use unsupervised data in order train a suite of state-of-the-art open-weight reranker models: zerank-1 and zerank-1-small. These models achieve the highest retrieval scores in multiple domains, including finance, legal, code, and STEM, outperforming closed-source proprietary rerankers on both NDCG@10 and Recall. These models also demonstrate great versatility, maintaining their 0-shot performance on out-of-domain and private customer datasets. The training data included 112,000 queries and 100 documents per query, and was trained end-to-end from unannotated queries and documents in less than 10,000 H100-hours.