benvansleen / LLModelica

LLModelica

Running the project

1. Clone the repo
2. Add a file LLModel/.env containing OPENAI_API_KEY = '<API_KEY>'
3. Grab the Modelica Standard Library source so it can be embedded (cd LLModel && mkdir data && cd data && git clone [e-mail]:modelica/ModelicaStandardLibrary.git)
4. Build the container (docker build -t om-python .)
5. Run the container (./run)
6. Run the LLM chain within the container (./run-llm <PROMPT>)

Upon first execution, a HuggingFace transformer will be downloaded and used to embed the Modelica Standard Library. The resulting embedding vectors will be persisted in a Chroma vector database at data/om_embeddings. This step took ~1 hour on my machine; YMMV.