README.md

metadata

language: pt
license: mit
tags:
  - msmarco
  - t5
  - pytorch
  - tensorflow
  - pt
  - pt-br
datasets:
  - msmarco
widget:
  - text: Texto de exemplo em português
inference: false

PTT5-base Reranker finetuned on both English and Portuguese MS MARCO

Introduction

ptt5-base-msmarco-en-pt-10k is a T5-based model pretrained in the BrWac corpus, finetuned on both English and Portuguese translated version of MS MARCO passage dataset. This model was finetuned for 10k steps. Further information about the dataset or the translation method can be found on our Cross-Lingual repository.

Usage

from transformers import T5Tokenizer, T5ForConditionalGeneration

model_name = 'unicamp-dl/ptt5-base-msmarco-en-pt-10k'
tokenizer  = T5Tokenizer.from_pretrained(model_name)
model      = T5ForConditionalGeneration.from_pretrained(model_name)

Citation

If you use ptt5-base-msmarco-en-pt-10k, please cite:

@article{rosa2021cost,
  title={A cost-benefit analysis of cross-lingual transfer methods},
  author={Rosa, Guilherme Moraes and Bonifacio, Luiz Henrique and de Souza, Leandro Rodrigues and Lotufo, Roberto and Nogueira, Rodrigo},
  journal={arXiv preprint arXiv:2105.06813},
  year={2021}
}