RAG-Application-Lesson-Docs-QA

Description

In this project, we explore the use of Retrieval Augmented Generation (RAG) to improve the quality of responses generated by Large Language Models (LLMs). RAG enhances LLMs by incorporating retrieved information from a document database, making the generated content more accurate and relevant.

• Input: A document file and a related question.
• Output: An answer derived from the document content.

Pipeline

Demo

How to use

I. Run without UI

You will run the project in the file RAG_Without_UI.ipynb to test an example.

2. Run with UI using chainlit

You will run the file RAG_With_UI.ipynb and follow the step-by-step instructions in the file, but there will be some notes when deploying with ngrok. Alternatively, you can use Colab to clone this project and run it for simplicity.

Next, I will provide step-by-step instructions to run the project on google colab.

2.1 Clone this repository

!git clone https://github.com/zoanhy20/RAG-Application-Lesson-Docs-QA.git
!cd RAG-Application-Lesson-Docs-QA

2.2 Install all reuired packages

!pip install -q transformers==4.41.2
!pip install -q bitsandbytes==0.43.1
!pip install -q accelerate==0.31.0
!pip install -q langchain==0.2.5
!pip install -q langchainhub==0.1.20
!pip install -q langchain-chroma==0.1.1
!pip install -q langchain-community==0.2.5
!pip install -q langchain-openai==0.1.9
!pip install -q langchain_huggingface==0.0.3
!pip install -q chainlit==1.1.304
!pip install -q python-dotenv==1.0.1
!pip install -q pypdf==4.2.0
!npm install -g localtunnel
!pip install -q numpy==1.24.4

2.3 Install and config ngrok

!pip install -q ngrok

Add your token in ngrok

from ngrok import ngrok
!ngrok config add-authtoken <YOUR_TOKEN_IN_NGROK>

Show the public network for connecting after building the project.

public_url = ngrok.connect(8000).public_url
print(f"Public URL: {public_url}")

2.4 Build project with chainlit

!chainlit run app.py