Saved in:
Bibliographic Details
Main Authors: Hu, Xinshuo, Shan, Zifei, Zhao, Xinping, Sun, Zetian, Liu, Zhenyu, Li, Dongfang, Ye, Shaolin, Wei, Xinyuan, Chen, Qian, Hu, Baotian, Wang, Haofen, Yu, Jun, Zhang, Min
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.01028
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912189041868800
author Hu, Xinshuo
Shan, Zifei
Zhao, Xinping
Sun, Zetian
Liu, Zhenyu
Li, Dongfang
Ye, Shaolin
Wei, Xinyuan
Chen, Qian
Hu, Baotian
Wang, Haofen
Yu, Jun
Zhang, Min
author_facet Hu, Xinshuo
Shan, Zifei
Zhao, Xinping
Sun, Zetian
Liu, Zhenyu
Li, Dongfang
Ye, Shaolin
Wei, Xinyuan
Chen, Qian
Hu, Baotian
Wang, Haofen
Yu, Jun
Zhang, Min
contents As retrieval-augmented generation prevails in large language models, embedding models are becoming increasingly crucial. Despite the growing number of general embedding models, prior work often overlooks the critical role of training data quality. In this work, we introduce KaLM-Embedding, a general multilingual embedding model that leverages a large quantity of cleaner, more diverse, and domain-specific training data. Our model has been trained with key techniques proven to enhance performance: (1) persona-based synthetic data to create diversified examples distilled from LLMs, (2) ranking consistency filtering to remove less informative samples, and (3) semi-homogeneous task batch sampling to improve training efficacy. Departing from traditional BERT-like architectures, we adopt Qwen2-0.5B as the pre-trained model, facilitating the adaptation of auto-regressive language models for general embedding tasks. Extensive evaluations of the MTEB benchmark across multiple languages show that our model outperforms others of comparable size, setting a new standard for multilingual embedding models with <1B parameters.
format Preprint
id arxiv_https___arxiv_org_abs_2501_01028
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle KaLM-Embedding: Superior Training Data Brings A Stronger Embedding Model
Hu, Xinshuo
Shan, Zifei
Zhao, Xinping
Sun, Zetian
Liu, Zhenyu
Li, Dongfang
Ye, Shaolin
Wei, Xinyuan
Chen, Qian
Hu, Baotian
Wang, Haofen
Yu, Jun
Zhang, Min
Computation and Language
As retrieval-augmented generation prevails in large language models, embedding models are becoming increasingly crucial. Despite the growing number of general embedding models, prior work often overlooks the critical role of training data quality. In this work, we introduce KaLM-Embedding, a general multilingual embedding model that leverages a large quantity of cleaner, more diverse, and domain-specific training data. Our model has been trained with key techniques proven to enhance performance: (1) persona-based synthetic data to create diversified examples distilled from LLMs, (2) ranking consistency filtering to remove less informative samples, and (3) semi-homogeneous task batch sampling to improve training efficacy. Departing from traditional BERT-like architectures, we adopt Qwen2-0.5B as the pre-trained model, facilitating the adaptation of auto-regressive language models for general embedding tasks. Extensive evaluations of the MTEB benchmark across multiple languages show that our model outperforms others of comparable size, setting a new standard for multilingual embedding models with <1B parameters.
title KaLM-Embedding: Superior Training Data Brings A Stronger Embedding Model
topic Computation and Language
url https://arxiv.org/abs/2501.01028