IT5 Cased Small Efficient EL32 for Question Generation 💭 🇮🇹

Shout-out to Stefan Schweter for contributing the pre-trained efficient model!

This repository contains the checkpoint for the IT5 Cased Small Efficient EL32 model fine-tuned on question generation on the SQuAD-IT corpus as part of the experiments of the paper IT5: Large-scale Text-to-text Pretraining for Italian Language Understanding and Generation by Gabriele Sarti and Malvina Nissim.

Efficient IT5 models differ from the standard ones by adopting a different vocabulary that enables cased text generation and an optimized model architecture to improve performances while reducing parameter count. The Small-EL32 replaces the original encoder from the T5 Small architecture with a 32-layer deep encoder, showing improved performances over the base model.

A comprehensive overview of other released materials is provided in the gsarti/it5 repository. Refer to the paper for additional details concerning the reported scores and the evaluation approach.

Using the model

Model checkpoints are available for usage in Tensorflow, Pytorch and JAX. They can be used directly with pipelines as:

from transformers import pipelines

qg = pipeline("text2text-generation", model='it5/it5-efficient-small-el32-question-generation')
qg("Le conoscenze mediche erano stagnanti durante il Medioevo. Il resoconto più autorevole di allora è venuto dalla facoltà di medicina di Parigi in un rapporto al re di Francia che ha incolpato i cieli, sotto forma di una congiunzione di tre pianeti nel 1345 che causò una "grande pestilenza nell\' aria". Questa relazione è diventata la prima e più diffusa di una serie di casi di peste che cercava di dare consigli ai malati. Che la peste fosse causata dalla cattiva aria divenne la teoria più accettata. Oggi, questo è conosciuto come la teoria di Miasma. La parola "peste" non aveva un significato particolare in questo momento, e solo la ricorrenza dei focolai durante il Medioevo gli diede il nome che è diventato il termine medico. Risposta: re di Francia")
>>> [{"generated_text": "Per chi è stato redatto il referto medico?"}]

or loaded using autoclasses:

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("it5/it5-efficient-small-el32-question-generation")
model = AutoModelForSeq2SeqLM.from_pretrained("it5/it5-efficient-small-el32-question-generation")

If you use this model in your research, please cite our work as:

@article{sarti-nissim-2022-it5,
    title={{IT5}: Large-scale Text-to-text Pretraining for Italian Language Understanding and Generation},
    author={Sarti, Gabriele and Nissim, Malvina},
    journal={ArXiv preprint 2203.03759},
    url={https://arxiv.org/abs/2203.03759},
    year={2022},
    month={mar}
}

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0003
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 7.0

Framework versions

  • Transformers 4.15.0
  • Pytorch 1.10.0+cu102
  • Datasets 1.17.0
  • Tokenizers 0.10.3
Downloads last month
726
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Dataset used to train gsarti/it5-efficient-small-el32-question-generation

Collection including gsarti/it5-efficient-small-el32-question-generation

Evaluation results