Hugging Face's logo

Spaces:
cvachet
/
pdf-chatbot
Running

App Files Community
18
pdf-chatbot / README.md
Clement Vachet
Add local execution section
e8f7645
preview code
|
raw
history blame
1.53 kB
metadata
title: PDF Chatbot
emoji: 🌍
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.16.1
app_file: app.py
pinned: true

code style linting: pylint

Aim: PDF-based AI chatbot with retrieval augmented generation

Architecture / Tech stack:

  • Front-end:
    • user interface via Gradio library
  • Back-end:
    • HuggingFace embeddings
    • HuggingFace Inference API for open-source LLMs
    • Chromadb vector database
    • LangChain conversational retrieval chain

Description:

  • This AI assistant, using Langchain and open-source LLMs, performs retrieval-augmented generation (RAG) from your PDF documents. The user interface explicitely shows multiple steps to help understand the RAG workflow. This chatbot takes past questions into account when generating answers (via conversational memory), and includes document references for clarity purposes. It leverages small LLM models to run directly on CPU hardware.

Available open-source LLMs:

  • Meta Llama series
  • Alibaba Qwen2.5 series
  • Mistral AI models
  • Microsoft Phi-3.5 series
  • Google Gemma models
  • HuggingFace zephyr and SmolLM series

You can try out the deployed Hugging Face Space!