---
language: 
- multilingual
- af
- am
- ar
- as
- az
- be
- bg
- bn
- br
- bs
- ca
- cs
- cy
- da
- de
- el
- en
- eo
- es
- et
- eu
- fa
- fi
- fr
- fy
- ga
- gd
- gl
- gu
- ha
- he
- hi
- hr
- hu
- hy
- id
- is
- it
- ja
- jv
- ka
- kk
- km
- kn
- ko
- ku
- ky
- la
- lo
- lt
- lv
- mg
- mk
- ml
- mn
- mr
- ms
- my
- ne
- nl
- no
- om
- or
- pa
- pl
- ps
- pt
- ro
- ru
- sa
- sd
- si
- sk
- sl
- so
- sq
- sr
- su
- sv
- sw
- ta
- te
- th
- tl
- tr
- ug
- uk
- ur
- uz
- vi
- xh
- yi
- zh
license: mit
---

# XLM-V (Base-sized model)

XLM-V is multilingual language model with a one million token vocabulary trained on 2.5TB of data from Common Crawl (same as XLM-R).
It was introduced in the [XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models](https://arxiv.org/abs/2301.10472)
paper by Davis Liang, Hila Gonen, Yuning Mao, Rui Hou, Naman Goyal, Marjan Ghazvininejad, Luke Zettlemoyer and Madian Khabsa.

**Disclaimer**: The team releasing XLM-V did not write a model card for this model so this model card has been written by the Hugging Face team. [This repository](https://github.com/stefan-it/xlm-v-experiments) documents all necessary integeration steps.

## Model description

From the abstract of the XLM-V paper:

> Large multilingual language models typically rely on a single vocabulary shared across 100+ languages.
> As these models have increased in parameter count and depth, vocabulary size has remained largely unchanged.
> This vocabulary bottleneck limits the representational capabilities of multilingual models like XLM-R.
> In this paper, we introduce a new approach for scaling to very large multilingual vocabularies by
> de-emphasizing token sharing between languages with little lexical overlap and assigning vocabulary capacity
> to achieve sufficient coverage for each individual language. Tokenizations using our vocabulary are typically
> more semantically meaningful and shorter compared to XLM-R. Leveraging this improved vocabulary, we train XLM-V,
> a multilingual language model with a one million token vocabulary. XLM-V outperforms XLM-R on every task we
> tested on ranging from natural language inference (XNLI), question answering (MLQA, XQuAD, TyDiQA), and
> named entity recognition (WikiAnn) to low-resource tasks (Americas NLI, MasakhaNER).

## Usage

You can use this model directly with a pipeline for masked language modeling:

```python
>>> from transformers import pipeline
>>> unmasker = pipeline('fill-mask', model='facebook/xlm-v-base')
>>> unmasker("Paris is the <mask> of France.")

[{'score': 0.9286897778511047,
  'token': 133852,
  'token_str': 'capital',
  'sequence': 'Paris is the capital of France.'},
 {'score': 0.018073994666337967,
  'token': 46562,
  'token_str': 'Capital',
  'sequence': 'Paris is the Capital of France.'},
 {'score': 0.013238662853837013,
  'token': 8696,
  'token_str': 'centre',
  'sequence': 'Paris is the centre of France.'},
 {'score': 0.010450296103954315,
  'token': 550136,
  'token_str': 'heart',
  'sequence': 'Paris is the heart of France.'},
 {'score': 0.005028395913541317,
  'token': 60041,
  'token_str': 'center',
  'sequence': 'Paris is the center of France.'}]
```

## Bias, Risks, and Limitations

Please refer to the model card of [XLM-R](https://huggingface.co/xlm-roberta-base), because XLM-V has a similar architecture
and has been trained on similar training data.

### BibTeX entry and citation info

```bibtex
@ARTICLE{2023arXiv230110472L,
       author = {{Liang}, Davis and {Gonen}, Hila and {Mao}, Yuning and {Hou}, Rui and {Goyal}, Naman and {Ghazvininejad}, Marjan and {Zettlemoyer}, Luke and {Khabsa}, Madian},
        title = "{XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models}",
      journal = {arXiv e-prints},
     keywords = {Computer Science - Computation and Language, Computer Science - Machine Learning},
         year = 2023,
        month = jan,
          eid = {arXiv:2301.10472},
        pages = {arXiv:2301.10472},
          doi = {10.48550/arXiv.2301.10472},
archivePrefix = {arXiv},
       eprint = {2301.10472},
 primaryClass = {cs.CL},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2023arXiv230110472L},
      adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}
```