This repository contains the code and products produced by the MIT team as part of the DAPRA Automating Scientific Knowledge Extraction and Modeling (ASKEM) project.
The MIT team consists of (in alphabetical order):
- Michael Cafarella
- Peter Baile Chen
- Wenjia He
- Chunwei Liu
- Markos Markakis
- Oscar Moll
- Theo Olausson
- Anna Zeng
pip install -e .
For clients: (eg notebooks)
Set up environment, specifically OPENAI_API_KEY, for client scripts/notebooks.
For server (no key needed on server side) From within mitaskm root:
uvicorn --reload --reload-dir=./mitaskem/ mitaskem.api.server:app
The above will hot-reload the project as you change files.
Ensure a server is running on a different tab. The will run the different notebooks as a client, ensuring things are not totally broken.
pytest
to re-generate the notebooks with the latest output:
pytest --nbmake --overwrite ./demos/2023-07/*ipynb
docker compose up --build
Our functionality is provided via a public API available here. Many of the calls also require you to provide a GPT key, which you can obtain from OpenAI.
For examples of usage, you can refer to our most recent demo here.