xlmv_base

XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models

converted checkpoint of XLM-V from fairseq to huggingface

Fairseq

if original model is needed, please check, model checkpoint:

https://dl.fbaipublicfiles.com/fairseq/xlmv/xlmv.base.tar.gz

and how to use it

https://github.com/facebookresearch/fairseq/blob/main/examples/xlmr/README.md

Note: please use official checkpoints, if they will be added to transformers (this repo is for personal usage/experiments)

Citation

@misc{https://doi.org/10.48550/arxiv.2301.10472,
  doi = {10.48550/ARXIV.2301.10472},
  url = {https://arxiv.org/abs/2301.10472},
  author = {Liang, Davis and Gonen, Hila and Mao, Yuning and Hou, Rui and Goyal, Naman and Ghazvininejad, Marjan and Zettlemoyer, Luke and Khabsa, Madian},
  keywords = {Computation and Language (cs.CL), Machine Learning (cs.LG), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models},
  publisher = {arXiv},
  year = {2023},
  copyright = {Creative Commons Attribution Share Alike 4.0 International}
}