MuRating: A High Quality Data Selecting Approach to Multilingual Large Language Model Pretraining

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Zhixun, Guo, Ping, Han, Wenhan, Zhang, Yifan, Liu, Binbin, Lin, Haobin, Liu, Fengze, Zhao, Yan, Zhang, Bingni, Wang, Taifeng, Zheng, Yin, Cohn, Trevor, Fang, Meng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908866306899968
author Chen, Zhixun
Guo, Ping
Han, Wenhan
Zhang, Yifan
Liu, Binbin
Lin, Haobin
Liu, Fengze
Zhao, Yan
Zhang, Bingni
Wang, Taifeng
Zheng, Yin
Cohn, Trevor
Fang, Meng
author_facet Chen, Zhixun
Guo, Ping
Han, Wenhan
Zhang, Yifan
Liu, Binbin
Lin, Haobin
Liu, Fengze
Zhao, Yan
Zhang, Bingni
Wang, Taifeng
Zheng, Yin
Cohn, Trevor
Fang, Meng
contents Data quality is a critical driver of large language model performance, yet existing model-based selection methods focus almost exclusively on English. We introduce MuRating, a scalable framework that transfers high-quality English data-quality signals into a single rater for 17 target languages. MuRating aggregates multiple English "raters" via pairwise comparisons to learn unified document-quality scores,then projects these judgments through translation to train a multilingual evaluator on monolingual, cross-lingual, and parallel text pairs. Applied to web data, MuRating selects balanced subsets of English and multilingual content to pretrain a 1.2 B-parameter LLaMA model. Compared to strong baselines, including QuRater, AskLLM, DCLM and so on, our approach boosts average accuracy on both English benchmarks and multilingual evaluations, with especially large gains on knowledge-intensive tasks. We further analyze translation fidelity, selection biases, and underrepresentation of narrative material, outlining directions for future work.
format Preprint
id arxiv_https___arxiv_org_abs_2507_01785
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MuRating: A High Quality Data Selecting Approach to Multilingual Large Language Model Pretraining
Chen, Zhixun
Guo, Ping
Han, Wenhan
Zhang, Yifan
Liu, Binbin
Lin, Haobin
Liu, Fengze
Zhao, Yan
Zhang, Bingni
Wang, Taifeng
Zheng, Yin
Cohn, Trevor
Fang, Meng
Computation and Language
Artificial Intelligence
Machine Learning
Data quality is a critical driver of large language model performance, yet existing model-based selection methods focus almost exclusively on English. We introduce MuRating, a scalable framework that transfers high-quality English data-quality signals into a single rater for 17 target languages. MuRating aggregates multiple English "raters" via pairwise comparisons to learn unified document-quality scores,then projects these judgments through translation to train a multilingual evaluator on monolingual, cross-lingual, and parallel text pairs. Applied to web data, MuRating selects balanced subsets of English and multilingual content to pretrain a 1.2 B-parameter LLaMA model. Compared to strong baselines, including QuRater, AskLLM, DCLM and so on, our approach boosts average accuracy on both English benchmarks and multilingual evaluations, with especially large gains on knowledge-intensive tasks. We further analyze translation fidelity, selection biases, and underrepresentation of narrative material, outlining directions for future work.
title MuRating: A High Quality Data Selecting Approach to Multilingual Large Language Model Pretraining
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2507.01785