Hugging Face's logo

Spaces:
ashrma
/
Chat-with-Docs
Runtime error

App Files Community
Chat-with-Docs / README.md
ashrma's picture
ashrma
updated sdk version
3b30092
preview code
|
raw
history blame
2.14 kB
metadata
title: Chat with Docs
emoji: 🦙
license: mit
sdk: streamlit
sdk_version: 1.21.0
python_version: 3.9
app_file: app.py
colorFrom: pink
colorTo: blue
pinned: false

Chat-with-Docs

image

Chat with your Docs and gain better insights. Powered by LlamaIndex and Streamlit is used for UI. Handles CSV/PDFs/Txt/Doc. CSV file is catered via PandasAI loader and rest of the docs are handled via GPTVectorStoreIndex.

Clone the repo or copy the .py file in your local machine.

Install required Dependencies 

pip install -r requirements.txt

Create a folder in the root dir and name it as documents

Run the application

streamlit run chat_with_docs.py

How to Contribute

Feel free to open any Issue or PR request. This small application can help anyone to interact with their docs more smartly in just 2-3 steps.

Roadmap

  • Add support for choosing in between GPT-3/GPT-3.5/GPT-4 or HuggingFace model for creating vectors and generating rich responses.
  • Blog explaining the entire application in detail.
  • Add Docker support.
  • Deploy the project on Streamlit or DataButton platform.
  • Add support to handle multiple files at once.

Snapshots

  • Upload a CSV file. Get better insights by just asking question, Render graphs based on the Data image image image

  • In Doc section, Upload PDFs/Txt/Docs to chat with your docs directly. No need to press CTRL+F to search for anything in the Docs image image