/DocSage

A document analyzer backed by local LLM.

Primary LanguageJupyter NotebookMIT LicenseMIT

DocSage

A document analyzer backed by local Large Language Models (LLMs).

How to use

  • Install the backend, Ollama. On Mac, it is recommended to use homebrew to install Ollama:
brew install ollama
  • Make sure to start the Ollam server:
brew service start ollama

  • Create a new virtual environment and install the required packages with environment.yml

  • Install DocSage as a package:

pip install -e .
  • Start the webapp with the following command:
streamlit run ollama_rag_streamlit.py

Additional Information

To learn more about the retrieval-augmented generation (RAG) model used here, please go through the notebooks in the notebooks directory.

License

See the LICENSE file for license rights and limitations (MIT).