How is the fluency scorer being fine-tuned?

PrithivirajDamodaran / Parrot_Paraphraser

A practical and feature-rich paraphrasing framework to augment human intents in text form to build robust NLU models for conversational engines. Created by Prithiviraj Damodaran. Open to pull requests and other forms of collaboration.

Apache License 2.0

867 stars 143 forks source link

How is the fluency scorer being fine-tuned? #9

Closed zjiayao closed 2 years ago

zjiayao commented 2 years ago

Thanks for the great work Prithiviraj. I am curious what dataset and procedure did you use to fine-tune the fluency scorer (https://huggingface.co/prithivida/parrot_fluency_on_BERT)?