This project was done as homework for the CSE256: Natural Language Processing course and is heavily inspired by minGPT. It consists of a speech classifier and a language model, allowing you to specify which part of the model to run by passing the --part argument. Additionally, you can optionally provide a custom sentence for which attention maps will be generated using the --sentence argument.
To run the code, use the following command:
python main.py --part <part_number> --sentence "<your_sentence>"