ElectroGPT: Querying Datasheets Made Easy!

Introduction

ElectroGPT is chatbot designed specifically to be used to query datasheets for electrical and electronic components. Designed to accept multiple PDF datasheet documents as input, ElectroGPT can quickly go through the datasheets and extract relevant information reliably. User-friendly and simple UI enables seamless operation.

Features

✅ Upload multiple PDF files through PyPDF2.
✅ Split raw text into text chunks using Langchain Text-Splitter.
✅ Create vectorDB and populate with text embeddings using Langchain FAISS.
✅ Both OpenAI and Hugging Face (Instructor-XL) model implemented for embeddings.
✅ Maintain conversation chain using Langchain.
✅ LLM chat support from OpenAI or Hugging Face models.
✅ Streamlit user interface.

Installation

We recommend using Conda, but you can use any virtual environment of your choice. If you use Conda, create a new environment with:

conda create -n electrogptenv python=3.11
conda activate electrogptenv
pip install -r requirements.txt

Usage

Run the following command to start the ElectroGPT UI in a browser:

streamlit run app.py

Upload the required PDF datasheets and click the process button once, once processing is done, ask your questions in the text bar provided.

Note: Don't forget to add OpenAI and Hugging Face API keys in .env file.