Please follow the steps below:
- Python Installation: Install Python from source. We used Python 3.11.6 throughout this project.
- Dependency Installation: Install necessary dependencies:
pip install -r requirements.txt
Place your dataset PDF files in accessible paths in the Data folder with a subfolder per entity.
Modify the configuration .yaml file. Set a variable documents_dir and the required topics. Modify any other hyperparameters, like model or document split sizes as wished.
Run the compare_batch.py script from the command line by specifying the path to your MIMIC ground truth data:
python3 compare_batch.py
You could also overwrite command line arguments here.
Configurations are loaded from the same .yaml file. If you wish to ask GPT-4 other questions, modify the topics list.
python3 naive_gpt4.py
Set the .json paths inside the evaluations.py file and run:
python3 evaluations.py