michaelthwan's picture
empty commit
22d3c40
|
raw
history blame
2.43 kB
metadata
title: Digest Everything Gpt
emoji: 💻
colorFrom: green
colorTo: green
sdk: gradio
sdk_version: 3.24.1
app_file: main.py
pinned: true
license: mit

DigestEverythingGPT

DigestEverythingGPT provides world-class content summarization/query tool that leverages ChatGPT/LLMs to help users quickly understand essential information from various forms of content, such as podcasts, YouTube videos, and PDF documents.

The prompt engineering is chained and tuned so that is result is of high quality and fast. It is not a simple single query and response tool.

Live website

https://huggingface.co/spaces/michaelthwan/digest-everything-gpt

Showcases

Example of summary

final_full_summary

DigestEverythingGPT's final output will adopt to video type.

n_things_example

LLM Loading in progress screen - chained prompt engineering, batched inference, etc.

in_process

Support for multiple languages regardless of video language

multi_language

Features

  • Content Summarization:
    • Automatically generate concise summaries of various types of content, allowing users to save time and make informed decisions for in-depth engagement.
    • Chained/Batched/Advanced prompt engineering for great quality/faster results.
  • Cross-Medium Support:
    • DigestEverythingGPT is designed to work with a wide range of content mediums.
    • Currently, the tool supports
      • YouTube videos [beta]
      • podcasts (in progress)
      • PDF documents (in progress)
  • Interactive "Ask" Feature (in progress):
    • Users can pose questions to the tool and receive answers extracted from specific sections within the full content.

Installation

Use python 3.10+ (tested in 3.10.8). Install using requirement.txt then launch gradio UI using main.py

pip install -r requirements.txt
python main.py

License

DigestEverything-GPT is licensed under the MIT License.

Acknowledgements

  • chatgpt_academic for gradio code framework