Data-Efficient French Language Modeling with CamemBERTa - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2023

Data-Efficient French Language Modeling with CamemBERTa

Abstract

Recent advances in NLP have significantly improved the performance of language models on a variety of tasks. While these advances are largely driven by the availability of large amounts of data and compute, they also benefit from the development of better training methods and architectures. In this paper, we introduce CAMEMBERTA, a French DeBERTa model that builds upon the DeBERTaV3 architecture and training objective. We evaluate our model's performance on a variety of French downstream tasks and datasets, including question answering, part-of-speech tagging, dependency parsing, named entity recognition, and the FLUE benchmark, and compare against CamemBERT, the state-of-the-art monolingual model for French. Our results show that, given the same amount of training tokens, our model outperforms BERT-based models trained with MLM on most tasks. Furthermore, our new model reaches similar or superior performance on downstream tasks compared to Camem-BERT, despite being trained on only 30% of its total number of input tokens. In addition to our experimental results, we also publicly release the weights and code implementation of CAMEMBERTA, making it the first publicly available DeBERTaV3 model outside of the original paper and the first openly available implementation of a DeBERTaV3 training objective.
Fichier principal
Vignette du fichier
French_DeBERTa___ACL_2023 to be uploaded.pdf (210.13 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03963729 , version 1 (30-01-2023)

Licence

Attribution

Identifiers

Cite

Wissam Antoun, Benoît Sagot, Djamé Seddah. Data-Efficient French Language Modeling with CamemBERTa. Findings of the Association for Computational Linguistics: ACL 2023, Jul 2023, Toronto, Canada. pp.5174-5185, ⟨10.18653/v1/2023.findings-acl.320⟩. ⟨hal-03963729v1⟩
276 View
84 Download

Altmetric

Share

Gmail Facebook X LinkedIn More