Saved in:
Bibliographic Details
Main Authors: Faysse, Manuel, Fernandes, Patrick, Guerreiro, Nuno M., Loison, António, Alves, Duarte M., Corro, Caio, Boizard, Nicolas, Alves, João, Rei, Ricardo, Martins, Pedro H., Casademunt, Antoni Bigata, Yvon, François, Martins, André F. T., Viaud, Gautier, Hudelot, Céline, Colombo, Pierre
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.00786
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910907124154368
author Faysse, Manuel
Fernandes, Patrick
Guerreiro, Nuno M.
Loison, António
Alves, Duarte M.
Corro, Caio
Boizard, Nicolas
Alves, João
Rei, Ricardo
Martins, Pedro H.
Casademunt, Antoni Bigata
Yvon, François
Martins, André F. T.
Viaud, Gautier
Hudelot, Céline
Colombo, Pierre
author_facet Faysse, Manuel
Fernandes, Patrick
Guerreiro, Nuno M.
Loison, António
Alves, Duarte M.
Corro, Caio
Boizard, Nicolas
Alves, João
Rei, Ricardo
Martins, Pedro H.
Casademunt, Antoni Bigata
Yvon, François
Martins, André F. T.
Viaud, Gautier
Hudelot, Céline
Colombo, Pierre
contents We introduce CroissantLLM, a 1.3B language model pretrained on a set of 3T English and French tokens, to bring to the research and industrial community a high-performance, fully open-sourced bilingual model that runs swiftly on consumer-grade local hardware. To that end, we pioneer the approach of training an intrinsically bilingual model with a 1:1 English-to-French pretraining data ratio, a custom tokenizer, and bilingual finetuning datasets. We release the training dataset, notably containing a French split with manually curated, high-quality, and varied data sources. To assess performance outside of English, we craft a novel benchmark, FrenchBench, consisting of an array of classification and generation tasks, covering various orthogonal aspects of model performance in the French Language. Additionally, rooted in transparency and to foster further Large Language Model research, we release codebases, and dozens of checkpoints across various model sizes, training data distributions, and training steps, as well as fine-tuned Chat models, and strong translation models. We evaluate our model through the FMTI framework, and validate 81 % of the transparency criteria, far beyond the scores of even most open initiatives. This work enriches the NLP landscape, breaking away from previous English-centric work in order to strengthen our understanding of multilinguality in language models.
format Preprint
id arxiv_https___arxiv_org_abs_2402_00786
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CroissantLLM: A Truly Bilingual French-English Language Model
Faysse, Manuel
Fernandes, Patrick
Guerreiro, Nuno M.
Loison, António
Alves, Duarte M.
Corro, Caio
Boizard, Nicolas
Alves, João
Rei, Ricardo
Martins, Pedro H.
Casademunt, Antoni Bigata
Yvon, François
Martins, André F. T.
Viaud, Gautier
Hudelot, Céline
Colombo, Pierre
Computation and Language
Machine Learning
We introduce CroissantLLM, a 1.3B language model pretrained on a set of 3T English and French tokens, to bring to the research and industrial community a high-performance, fully open-sourced bilingual model that runs swiftly on consumer-grade local hardware. To that end, we pioneer the approach of training an intrinsically bilingual model with a 1:1 English-to-French pretraining data ratio, a custom tokenizer, and bilingual finetuning datasets. We release the training dataset, notably containing a French split with manually curated, high-quality, and varied data sources. To assess performance outside of English, we craft a novel benchmark, FrenchBench, consisting of an array of classification and generation tasks, covering various orthogonal aspects of model performance in the French Language. Additionally, rooted in transparency and to foster further Large Language Model research, we release codebases, and dozens of checkpoints across various model sizes, training data distributions, and training steps, as well as fine-tuned Chat models, and strong translation models. We evaluate our model through the FMTI framework, and validate 81 % of the transparency criteria, far beyond the scores of even most open initiatives. This work enriches the NLP landscape, breaking away from previous English-centric work in order to strengthen our understanding of multilinguality in language models.
title CroissantLLM: A Truly Bilingual French-English Language Model
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2402.00786