The focus of this project is to fine-tune a DistilBERT model over 4 different classification tasks on the tweet-eval dataset.
To perform the experimental phase, 3 different helper classes were created. It is possible to find such classes in the helper folder.
- Service helper, to handle the login process in the used services
- Dataset helper, to load and work with different datasets
- Engine helper, to perform optimization and training on the chosed models
To further readings please refer to the report present in the home of the project.
| Package | README |
|---|---|
| Hugging Face | https://huggingface.co |
| Wandb | https://wandb.ai/site |