milmor's picture
Update README.md
c596492 verified
metadata
license: apache-2.0
language:
  - es
  - nah
tags:
  - translation
widget:
  - text: 'translate Spanish to Nahuatl: muchas flores son blancas'

t5-small-spanish-nahuatl

Model description

This model is a T5 Transformer (t5-small) fine-tuned on 29,007 spanish and nahuatl sentences using 12,890 samples collected from the web and 16,117 samples from the Axolotl dataset.

The dataset is normalized using 'sep' normalization from py-elotl.

Usage

from transformers import AutoModelForSeq2SeqLM
from transformers import AutoTokenizer

model = AutoModelForSeq2SeqLM.from_pretrained('milmor/t5-small-spanish-nahuatl')
tokenizer = AutoTokenizer.from_pretrained('milmor/t5-small-spanish-nahuatl')

model.eval()
sentence = 'muchas flores son blancas'
input_ids = tokenizer('translate Spanish to Nahuatl: ' + sentence, return_tensors='pt').input_ids
outputs = model.generate(input_ids)
# outputs = miak xochitl istak
outputs = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]

Evaluation results

The model is evaluated on 400 validation sentences.

  • Validation loss: 1.36

Note: Since the Axolotl corpus contains multiple misalignments, the real Validation loss is slightly better. These misalignments also introduce noise into the training.

References

  • Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2019. Exploring the limits of transfer learning with a unified Text-to-Text transformer.

  • Ximena Gutierrez-Vasques, Gerardo Sierra, and Hernandez Isaac. 2016. Axolotl: a web accessible parallel corpus for Spanish-Nahuatl. In International Conference on Language Resources and Evaluation (LREC).

Created by Emilio Morales.