Causal discovery with GPT

content

• data folder with descriptions of causal discovery problems and eventually datasets
• gptci python module that implement CI testing via GPT queries
• .env.example example of needed environemntal variables

how to

tested with python3.10

• python -m venv env )
• source env/bin/activate
• pip install -r requirements.txt
• cp .env.example .env generate the .env file
• add your openai API key into the .env file (.env is already ignored in the .gitignore file but doubel check you are not pushing it)
• python example.py to run a simple example
• For running a specific CI test var1, var2| (var3, var4,...) in the context of a specific YAML_FILE python scripts/ci_test.py data/YAML_FILE var1 var2 [var3] [var4] ...
• For running evaluations on a specific data use python scripts/evaluate_cis.py data/FILE --random R --n N to run on R randomly sampled ci statements with N repeated answers from the model. [check python scripts/evaluate_cis.py --help for the full list of options]