xlm-multi-finetuned-xquadv1

XLM (multilingual version) fine-tuned for multilingual Q&A

Released from Facebook together with the paper Cross-lingual Language Model Pretraining by Guillaume Lample and Alexis Conneau and fine-tuned on XQuAD for multilingual (11 different languages) Q&A downstream task.

Details of the language model('xlm-mlm-100-1280')

Language model

It includes the following languages:

<details> en-es-fr-de-zh-ru-pt-it-ar-ja-id-tr-nl-pl-simple-fa-vi-sv-ko-he-ro-no-hi-uk-cs-fi-hu-th-da-ca-el-bg-sr-ms-bn-hr-sl-zh_yue-az-sk-eo-ta-sh-lt-et-ml-la-bs-sq-arz-af-ka-mr-eu-tl-ang-gl-nn-ur-kk-be-hy-te-lv-mk-zh_classical-als-is-wuu-my-sco-mn-ceb-ast-cy-kn-br-an-gu-bar-uz-lb-ne-si-war-jv-ga-zh_min_nan-oc-ku-sw-nds-ckb-ia-yi-fy-scn-gan-tt-am </details>

Details of the downstream task (multilingual Q&A) - Dataset

Deepmind XQuAD

Languages covered:

• Arabic: ar
• German: de
• Greek: el
• English: en
• Spanish: es
• Hindi: hi
• Russian: ru
• Thai: th
• Turkish: tr
• Vietnamese: vi
• Chinese: zh

As the dataset is based on SQuAD v1.1, there are no unanswerable questions in the data. We chose this setting so that models can focus on cross-lingual transfer.

We show the average number of tokens per paragraph, question, and answer for each language in the table below. The statistics were obtained using Jieba for Chinese and the Moses tokenizer for the other languages.

Citation:

<details>

@article{Artetxe:etal:2019,
      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},
      title     = {On the cross-lingual transferability of monolingual representations},
      journal   = {CoRR},
      volume    = {abs/1910.11856},
      year      = {2019},
      archivePrefix = {arXiv},
      eprint    = {1910.11856}
}

</details>

As XQuAD is just an evaluation dataset, I used Data augmentation techniques (scraping, neural machine translation, etc) to obtain more samples and split the dataset in order to have a train and test set. The test set was created in a way that contains the same number of samples for each language. Finally, I got:

Model training

The model was trained on a Tesla P100 GPU and 25GB of RAM. The script for fine tuning can be found here

Model in action

Fast usage with pipelines:

from transformers import pipeline

qa_pipeline = pipeline(
    "question-answering",
    model="mrm8488/xlm-multi-finetuned-xquadv1",
    tokenizer="mrm8488/xlm-multi-finetuned-xquadv1"
)

# English
qa_pipeline({
    'context': "Manuel Romero has been working hardly in the repository hugginface/transformers lately",
    'question': "Who has been working hard for hugginface/transformers lately?"
})

#Output: {'answer': 'Manuel', 'end': 6, 'score': 8.531880747878265e-05, 'start': 0}

# Russian
qa_pipeline({
    'context': "Мануэль Ромеро в последнее время почти не работал в репозитории hugginface / transformers",
    'question': "Кто в последнее время усердно работал над обнимашками / трансформерами?"
    
})

#Output: {'answer': 'работал в репозитории hugginface /','end': 76, 'score': 0.00012340750456964894, 'start': 42}

Try it on a Colab (Do not forget to change the model and tokenizer path in the Colab if necessary):

<a href="https://colab.research.google.com/github/mrm8488/shared_colab_notebooks/blob/master/Try_mrm8488_xquad_finetuned_uncased_model.ipynb" target="_parent"><img src="https://camo.githubusercontent.com/52feade06f2fecbf006889a904d221e6a730c194/68747470733a2f2f636f6c61622e72657365617263682e676f6f676c652e636f6d2f6173736574732f636f6c61622d62616467652e737667" alt="Open In Colab" data-canonical-src="https://colab.research.google.com/assets/colab-badge.svg"></a>

Created by Manuel Romero/@mrm8488

Made with <span style="color: #e25555;">♥</span> in Spain