LangChain Question Answering Web Application

This web application uses ideas and code from the LangChain - Develop LLM powered applications with LangChain Udemy course, with some adjustments made to the code. It answers questions about your PDFs, and uses Pinecone as a vectorstore. Additionally, a Dockerfile has been created to simplify the deployment process.

The application is Dockerized and ready to be deployed on any system that supports Docker.

Data Directory

The data directory contains the following PDF files:

• Essential drugs Practical guide intended for physicians, pharmacists, nurses and medical auxiliaries

  • Guideline 339 (PDF)

• Clinical guidelines - Diagnosis and treatment manual For curative programmes in hospitals and dispensaries Guidance for prescribing

  • Guideline 170 (PDF)

If you want to perform question answering over different PDF files, simply delete the existing PDFs in the data directory and replace them with your own PDF files.

Prerequisites

Before running the application, make sure you have Docker installed on your system. If not, you can install Docker by following the instructions on the Docker website.

Setup

1. Clone the project:

   git clone https://github.com/SinaRampe/multiple_pdf_chatbot.git
   

2. Customization:

You can customize the chat model in backend/retrieval_qa_chain.py by changing the following code:

chat = ChatOpenAI(
       temperature=0,
       model_name="gpt-4"
 )

The available OpenAI models can be found in the OpenAI API documentation. If no OpenAI GPT-4 API access is given, the user should change model_name to "gpt-3.5-turbo". GPT-3.5 is faster in generating responses and doesn't come with the hourly prompt restrictions GPT-4 does. GPT-4 is smarter, can handle longer prompts and conversations, and doesn't make as many factual errors.

The temperature parameter controls the randomness of the responses. Higher values result in more random and surprising responses. Other parameters that may be changed include max_tokens (maximum length of the response). More available parameters and what they do can be read in the OpenAI API reference for completions.

3. Go to the project directory:

cd multiple_pdf_chatbot

4. Create a .env file in the project directory with the following environment variables:

OPENAI_API_KEY=<your_openai_api_key>
PINECONE_API_KEY=<your_pinecone_api_key>
PINECONE_ENVIRONMENT=<your_pinecone_environment>

5. Build the Docker image:

docker build -t msf_chatbot .

6. To run the application, execute the following command:

docker run -p 8501:8501 msf_chatbot

Then navigate to http://localhost:8501 in your browser to use the application.

TBD (Backlog)

• The PyPDFLoader class from document_loaders can load PDFs and split them into a list of documents, where each document contains the page content and metadata with the page number. Add functionality to display the page number in the source.
• improve frontend, example: defidocbot
• latest question should be inserted at the top, history appears at the bottom -> see Chat With Your Healthcare Documents: Build a Chatbot With ChatGPT and LangChain
• current chain returns only 4 text blocks, which is suboptimal for some UCs, e.g. searching and Q&A of receipts
• further readme testing