metadata
license: afl-3.0
language:
- pt
pipeline_tag: text2text-generation
Model Card for Model ID
This model is intended to be used generating questions and answers from brazilian portuguese text passages, so you can finetune another BERT model into your generated triples (context-question-answer) for extractive question answering without supervision or labeled data.
It was trained using unicamp-dl/ptt5-base-t5-portuguese-vocab base model and Squad 1.1 portuguese version dataset to generante question and answers from text passages.
Model Description
- Developed by: Vitor Alcantara Batista ([email protected])
- Model type: T5 base
- Language(s) (NLP): Brazilian Portuguese
- License: Academic Free License v. 3.0
- Finetuned from model : unicamp-dl/ptt5-base-t5-vocab
Model Sources [optional]
- Repository: This model used code from this github repo https://github.com/patil-suraj/question_generation/
Usage
How to use it (after cloning the github repo above):
from pipelines import pipeline
nlp = pipeline("multitask-qa-qg", model='vabatista/question-generation-t5-pt-br', tokenizer='vabatista/question-generation-t5-pt-br')
text = """ PUT YOUR TEXT PASSAGE HERE """
nlp(text)
Sample usage/results:
Training Details
TODO
Model Card Authors
Vitor Alcantara Batista