GPT-2 Small Portuguese Lyrics

Pretrained model from lyrics dataset in Portuguese.

Model description

The model was trained from a Kaggle Dataset, “Song lyrics from 6 musical genres”, with around 66,000 songs in portuguese.

The model was fine-tuned from GPorTuguese-2 on Colab Pro+ enviroment.

<!---

Intended uses & limitations

How to use

Limitations and bias

Training data

Training procedure

Preprocessing

BibTeX entry and citation info

-->

Evaluation results

Loss Perplexity Training Duration
3,301 27,15 06:45:09