PeeteKeesel / PaperGPT

:microscope: Choose a academic research papers and get a TL;DR for it in seconds.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Paper Explainer

Python numpy huggingface_hub openai

Simply specify of the paper your want to understand and obtain a short summary (TL;DR) of it in seconds.

Installation

# -----------------------------------------------------------------------------
# Setting up the environment

# Create the PaperGPT environment
conda env create -f environment.yml

# (Optiona) Run the following to check if PaperGPT got successfully created
conda env list

# Activate the environment
conda activate PaperGPT


# -----------------------------------------------------------------------------
conda install -c huggingface transformers

pip install PyPDF2
conda install -c conda-forge pypdf2

pip install openai
conda install -c conda-forge openai -y

Action Points

  • Ease of use: ensure that library is easy to install. Clear documentation and examples.
  • Automatic processing: automatically extract text and relevant content from papers, eliminate need for users to manually input or format the content
  • Compatibility: should be able to handle various document formats (PDF, PNG, JPG), and possibly even accept URLs to academic papers hosted online
  • Multi-language support
  • Use cases: Provide real-world use cases and scenarios where the library can be valuable, such as research paper review, studying, or staying updated on the latest research.
  • Model Options: Provide option for the user to decide between different LLM models such as LLaMa, GPT, Bert
    • BART (Facebook AI) --
    • BERT (Google AI) -- bert-base-uncased
    • GPT-2 (OpenAI)

About

:microscope: Choose a academic research papers and get a TL;DR for it in seconds.

License:MIT License


Languages

Language:Jupyter Notebook 99.9%Language:Python 0.1%