Repo for Orion project @ HackGT 12
Video: https://www.youtube.com/watch?v=ueU99vmpKck
Devpost: https://devpost.com/software/orion-owa3ry?ref_content=my-projects-tab&ref_feature=my_projects
You must have Gemini API key in your .env. Additionally, if you wish to use Slack and Github features, add the relevant keys to .env.
- Install ollama for embeddings
-
ollama pull nomic-embed-text -
git clone https://github.com/josephp1005/orion.git -
pip install -r requirements.txt -
cd local_scripts -
chmod +x local_kill.sh -
./local_kill.sh -
chmod +x local_setup.sh -
./local_setup.sh
Your current terminal is now running ollama, and will show any requests made to ollama. To continue with the next steps, keep this terminal running and open a new terminal window. Navigate to the directory where the cloned repository is located.
- Upload pdfs to the data directory.
-
python dense_embeddings.py -
python sparse_embedding.py
If all of the above was done properly, you can now run:
python answer.py <query>
where <query> is your question.
Now run Flask app on separate terminal.
-
python app.py
Finally run the React app on a separate terminal
-
cd ui -
npm install -
npm run dev
- Run orion-slack.py to poll slack and update dense
- Currently would need to manually update sparse
- You can do it from any directory, but need to run the script oterm/oterm.sh
- Once done capturing commands, do exit; will update dense
- Again would need to manually update sparse
- Simlar to Slack, run ./run_fetch_prs.sh
- If you delete the ChromaDB, make sure you also delete sparse embeddings pickle file.
- Need to run Flask App and UI for things to work