mt5-simplification-spanish

Model description

This is a fine-tuned mt5-small model for generating simple text from complex text.

This model was created with the IXA Group research group of the University of the Basque Country, the model has been evaluated with the Sari, Bleu and Fklg metrics; it was trained and tested using the Simplext corpus.

Dataset

Simplext

Model Evaluation

Bleu: 13,186

Sari: 42,203

Fklg: 10,284

Authors

Oscar M. Cumbicus-Pineda, Itziar Gonzalez-Dios, Aitor Soroa, November 2021

Code

https://github.com/oskrmiguel/mt5-simplification