A chatbot to answer documentations in spikeinterface.
This is a flask app that uses LangChain and OpenAI model for answering questions using the documentation of spikeinterface.
flask --app spikeinterface_chatbot.app run
Note that for running the app locally your OpenAI key should be an environment variable:
export OPENAI_API_KEY="your_key_goes_here"
export QDRANT_API_KEY="your_quadrant_api_key_goes_here"
docker build -t spikeinterface_chatbot_container .
Where the -t adds the name as a tag of the docker file so it can be referenced later.
docker run -d -p 5000:80 spikeinterface_chatbot_container
Where the -d is the detach command (so you get back your terminal) and the -p is mapping port 5000 in the local computer to port 80 on the container.
Note that for running the docker container locally you will need to pass your OpenAI key as an environment variable:
docker run -d -p 5000:80 -e OPENAI_API_KEY=your_key_goes_here -e QDRANT_API_KEY=your_quadrant_api_key_goes_here spikeinterface_chatbot_container
docker tag spikeinterface_chatbot_container ghcr.io/catalystneuro/spikeinterface_chatbot_container:latest
docker push ghcr.io/catalystneuro/spikeinterface_chatbot_container:latest