This template is licensed under Apache 2.0 and contains the following components:
A Retrieval-Augmented Generation (RAG) application is a type of AI-powered system that combines the capabilities of retrieval-based models and generation-based models. The main goal of a RAG application is to enhance the quality and accuracy of generated content by incorporating relevant information retrieved from a vast knowledge base. This approach leverages the strengths of both retrieval and generation, providing more coherent and contextually appropriate outputs.
This project provides a template for a streamlit application which is then hosted on Domino. We are using an OpenAI embedding model to create the emebeddings of our documents and a gpt model (GPT-4, GPT-4o, etc.) as the base model.
The important files in this template are:
- .env: Our environment file, a simple text file used to set environment variables.
- utils.py: This file contains utility functions that are used throughout the codebase.
- base_agent.py: This file defines a BaseAgent class that initializes a chat model and embeddings using Azure's OpenAI services, configured with environment variables for deployment and API access.
- UseCases.py: The page on the streamlit application for use case management.
- pages/Chatbot.py: The chatbot interface on the application where users can select their use case and speak with BioRAG.
- app.sh: The script needed to run the app.
Step 1 Ensure you have the following external services set up with the correct credentials provided in the .env file
- Azure blob storage - Credentials tutorial
- Qdrant Vector Database - Tutorial
- Azure OpenAI Chat Model Resource - Deployment Types (See below for Azure OpenAI Keys and Endpoints)
- Azure OpenAI Embeddings Resource - Deployment Types (See below for Azure OpenAI Keys and Endpoints)
Should you require assistance setting these services up, please reach out to Gabriele Oliva from BIP, [email protected]
The necessary packages and versions can be found in the requirements.txt file. Ensure these packages are installed in a custom Domino Environment. Please find the docker instructions below:
Step 2 Create a custom environment on Domino with a base environment of Domino Standard Environment
Step 3 Under dockerfile instructions use the instructions provided below to install the python packages into the environment.
RUN pip install langchain==0.2.0 --user
RUN pip install langchain_community==0.2.0 --user
RUN pip install langchain_openai==0.1.7 --user
RUN pip install langchain_text_splitters==0.2.0 --user
RUN pip install pandas==2.2.2 --user
RUN pip install pdfplumber==0.11.0 --user
RUN pip install python-dotenv==1.0.1 --user
RUN pip install qdrant_client==1.9.1 --user
RUN pip install streamlit==1.33.0 --user
RUN pip install azure-storage-blob==12.20.0 --user
RUN pip install python-docx==1.1.1 --user
Step 4 Utilize small hardware tier requirements
In your domino project, you will want to add the following environment variables in your project's settings:
| Environment Variable Name | Your value |
|---|---|
| QDRANT_URL | "YOUR QDRANT URL" |
| QDRANT_KEY | "YOUR QDRANT KEY" |
| AZURE_EMBEDDINGS_DEPLOYMENT_NAME | "YOUR AZURE EMBEDDINGS DEPLOYMENT NAME" |
| AZURE_EMBEDDINGS_API_KEY | "YOUR AZURE EMBEDDINGS API KEY" |
| AZURE_EMBEDDINGS_MODEL_NAME | "text-embedding-ada-002" |
| AZURE_EMBEDDINGS_ENDPOINT | "YOUR AZURE EMBEDDINGS ENDPOINT" |
| OPENAI_API_VERSION | "2023-05-15" |
| OPEN_AI_TYPE | "azure" |
| AZURE_CHAT_ENDPOINT | "YOUR AZURE CHAT ENDPOINT" |
| AZURE_CHAT_DEPLOYMENT_NAME | "YOUR AZURE CHAT DEPLOYMENT NAME" |
| AZURE_CHAT_API_KEY | "YOUR AZURE CHAT API KEY" |
| AZURE_CHAT_MODEL | "YOUR AZURE CHAT MODEL" |
| AZURE_BLOB_CONTAINER_NAME | "YOUR AZURE BLOB CONTAINER NAME" |
| AZURE_BLOB_CONNECTION_STRING | "YOUR AZURE BLOB CONNECTION STRING" |
Once all are saved, you are ready!