Skip to content

Latest commit

 

History

History

databricks_DBRX_website_bot

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Hogwarts chatbot with Open source RAG using DBRX, LanceDB, and LLama-index with Huggingface Embeddings

This application is a website chatbot that uses the Open source RAG model with DBRX, LanceDB, and LLama-index with Hugginface Embeddings.

Steps to Run the Application

  1. Install Dependencies
pip install -r requirements.txt
  1. Setup Databricks Serving Endpoint and token as environment variables for using databricks serving endpoint. You can also use the dbrx model locally as it is open source.
export DATABRICKS_TOKEN=<your api key>
DATABRICKS_SERVING_ENDPOINT=<your api serving endpoint>
  1. Run the application in CLI mode
python main.py

Accepted arguments:

  • url: URL of the document to be indexed. Default is the Hogwarts School of Witchcraft and Wizardry Wikipedia page.
  • embed_model: Huggingface model to use for embeddings. Default is mixedbread-ai/mxbai-embed-large-v1.
  • uri: URI of the vector store. Default is ~/tmp/lancedb_hogwarts.
  • force_create_embeddings: Whether to force create embeddings. Default is False.
  • illustrate: Whether to illustrate the responses. Default is True.
  1. Run the application in GUI mode
streamlit run gui.py

MLX SDXL

The MLX SDXL implementation is taken from MLX examples repo. The implementation is modified a bit to make it work faster with the current application.