{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/flaubert-des-mod-eles-de-langue-contextualis","title":"FlauBERT : des mod\\`eles de langue contextualis\\'es pr\\'e-entra\\^\\in\\'es pour le fran\\ccais (FlauBERT : Unsupervised Language Model Pre-training for French)","arxiv_id":null,"date":"2020-06-01","proceeding":"JEPTALNRECITAL 2020 6","authors":["Hang Le","Lo{\\\"\\i}c Vial","Jibril Frej","Vincent Segonne","Maximin Coavoux","Benjamin Lecouteux","Alex Allauzen","re","Beno{\\^\\i}t Crabb{\\'e}","Laurent Besacier","Didier Schwab"],"abstract":"Les mod{\\`e}les de langue pr{\\'e}-entra{\\^\\i}n{\\'e}s sont d{\\'e}sormais indispensables pour obtenir des r{\\'e}sultats {\\`a} l{'}{\\'e}tat-de-l{'}art dans de nombreuses t{\\^a}ches du TALN. Tirant avantage de l{'}{\\'e}norme quantit{\\'e} de textes bruts disponibles, ils permettent d{'}extraire des repr{\\'e}sentations continues des mots, contextualis{\\'e}es au niveau de la phrase. L{'}efficacit{\\'e} de ces repr{\\'e}sentations pour r{\\'e}soudre plusieurs t{\\^a}ches de TALN a {\\'e}t{\\'e} d{\\'e}montr{\\'e}e r{\\'e}cemment pour l{'}anglais. Dans cet article, nous pr{\\'e}sentons et partageons FlauBERT, un ensemble de mod{\\`e}les appris sur un corpus fran{\\c{c}}ais h{\\'e}t{\\'e}rog{\\`e}ne et de taille importante. Des mod{\\`e}les de complexit{\\'e} diff{\\'e}rente sont entra{\\^\\i}n{\\'e}s {\\`a} l{'}aide du nouveau supercalculateur Jean Zay du CNRS. Nous {\\'e}valuons nos mod{\\`e}les de langue sur diverses t{\\^a}ches en fran{\\c{c}}ais (classification de textes, paraphrase, inf{\\'e}rence en langage naturel, analyse syntaxique, d{\\'e}sambigu{\\\"\\i}sation automatique) et montrons qu{'}ils surpassent souvent les autres approches sur le r{\\'e}f{\\'e}rentiel d{'}{\\'e}valuation FLUE {\\'e}galement pr{\\'e}sent{\\'e} ici.","url_abs":"https://aclanthology.org/2020.jeptalnrecital-taln.26","url_pdf":"https://aclanthology.org/2020.jeptalnrecital-taln.26.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"flaubert-des-mod-eles-de-langue-contextualis","repo_url":"https://github.com/MindCode-4/code-3/tree/main/flaubert","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"flue","task_name":"FLUE"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}