Description
ResearchGPT is a project that offers an LLM-based research assistant, enabling users to engage in conversational interactions with research papers. The application is built as a simple FastAPI app, providing a clean interface for users to upload their own PDF documents or provide links to PDFs hosted online.
Upon receiving a PDF, ResearchGPT processes it by extracting the text content. This extracted text is then used to create embeddings, which are numerical representations of the text's meaning. These embeddings are stored locally using Redis. The application then leverages the OpenAI API to process user queries. When a question is asked about the research paper, ResearchGPT uses the embeddings to find relevant sections of the text and generates a response based on that information.
A key feature of ResearchGPT is its ability to provide source attribution. For each part of the generated response, it returns the specific source text used and the corresponding page number from the original PDF. This ensures transparency and allows users to easily verify the information. The project was developed by mukulpatnaik and is hosted on GitHub.
While the original repository is archived, a more advanced version is available for free at chat.portal.so, with plans for it to be open-sourced. The project requires Python, FastAPI, and Redis for local embedding storage. Users need an OpenAI API key to run the application, which can be set as an environment variable. Installation involves cloning the repository, installing dependencies via pip, and setting up Redis.
This tool is particularly useful for researchers, students, and anyone who needs to quickly extract and understand information from multiple research papers without reading them in their entirety. It streamlines the process of literature review and information retrieval by allowing direct, conversational access to the content of academic documents.
ResearchGPT's Core Features
LLM-based research assistant
Conversational interface for research papers
PDF text extraction
Text embedding generation
OpenAI API integration
Source citation and page number retrieval
FastAPI application framework
Local embedding storage with Redis
Online PDF link support
User-uploaded PDF support
Getting Started with ResearchGPT
Clone: git clone https://github.com/mukulpatnaik/researchgpt.git
Install dependencies: cd researchgpt && pip install -r requirements.txt
Set up Redis: Install Redis and start the server (e.g., 'redis-server' on MacOS)
Configure API key: Set OPENAI_API_KEY environment variable
Execute: uvicorn main:app --reload
ResearchGPT's Use Cases
- Research Paper Analysis
- Literature Review
- Information Extraction
- Academic Q&A
- Document Summarization
