ju-resplande / dlb_absapt2022

DLB at ABSAPT 2022.

ABSAPT 2022

Deep Learning Brasil group submission at ABSAPT 2022.

Submission files are available on DeepLearningBrasil_task1.csv and DeepLearningBrasil_task2.csv. Here is the presentation.

Installation

pip install -r requirements.txt

All experiments were made on V100 GPU (32GB).

How-to

Task 1 - ATE

1. Train ensemble. Run Notebooks in order:

   1. huggingface-roberta.ipynb
   2. huggingface-multilingual.ipynb
   3. huggingface-futher-training.ipynb

2. Generate submission file eval.ipynb

Task 2 - SOE

1. Train ensemble

bash SOE/train_ensemble.sh

2. Predict ensemble

bash SOE/predict_ensemble.sh

Citation

@inproceedings{gomes2022deep,
  title={Deep learning brasil at absapt 2022: Portuguese transformer ensemble approaches},
  author={Gomes, JRS and Rodrigues, RC and Garcia, EAS and Junior, AFB and Silva, Diogo Fernandes Costa and Maia, Dyonnatan Ferreira},
  booktitle={Proceedings of the Iberian Languages Evaluation F{\'o}rum (IberLEF 2022), co-located with the 38th Conference of the Spanish Society for Natural Language Processing (SEPLN 2022), Online. CEUR. org},
  year={2022}
}

License

This work is licensed under a Creative Commons Attribution 4.0 International License. See LICENSE.txt for more information.

Acknowledgments

This work has been supported by the AI Center of Excellence (Centro de Excelência em Inteligência Artificial – CEIA) of the Institute of Informatics at the Federal University of Goiás (INF-UFG).