This is the course final project for DS-UA 203 Machine Learning for Language Understanding, taught by Professor Sam Bowman at Spring 2022.
Tracy Zhu
Yuchen Zhu
To download the whole directory, enter
git clone https://github.com/yk803/DS203_Project.git
in terminal.
To run the code, first run
bash env.sh
to get the necessary packages and download the raw dataset (if there isn't one).
For selecting words and sentences, follow the insturctions in generate_dataset.ipynb to build the dataset.
Next, call python add_fake_words.py data_dir mode to generate the task dataset. It has two arguments: data_dir, representing the path name of the dataset we just built, and mode, determining how we want our fake new words to be generated.
Finally, call python main.py to send requests to access OpenAI GPT-3, with your unique API key. The key may rotate sometimes and may thus need updates.