readpaper 🔬

AI-powered research paper summaries that capture technical depth and nuance. Powered by Claude 3.5 Sonnet.

Why readpaper?

readpaper leverages Claude 3.5's advanced capabilities to deliver comprehensive research paper analysis:

• 🔍 Deep PDF Analysis: Process papers up to 32MB with native PDF support (1)
• ⚡ Performance Optimized:
  • Token counting for efficient processing (2)
  • Prompt caching for faster repeated analyses (3)
• 🧠 Intelligent Processing:
  • Multi-stage sectional analysis for thorough understanding
  • Progressive context building across sections
  • Research scientist persona for academic-quality insights
  • Token-optimized prompts for maximum context utilization

How it Works

1. Sectional Analysis: Papers are processed in three key stages:

   • Initial overview and context gathering
   • Detailed technical analysis
   • Synthesis and implications

2. Progressive Understanding: Each analysis stage builds upon previous insights, creating a coherent and deep understanding of the paper

3. Research Scientist Approach: A specialized research persona guides the analysis style, ensuring academic rigor and technical depth

4. Optimized Processing:

   • Efficient PDF handling with Claude's beta capabilities
   • Smart caching for faster repeated analyses
   • Token usage monitoring for optimal context management

Who is this for?

• 📚 Researchers staying current with literature
• 💻 Engineers implementing papers
• 🔧 Teams doing literature reviews
• 📊 Anyone needing deep technical understanding

Sample Output

Check out sample summaries for BERT and the attention paper

🚀 Getting Started

Prerequisites

• Python 3.10+
• uv (recommended) or pip

Installation

1. Clone the repository:

git clone https://github.com/benthecoder/readpaper.git
cd readpaper

2. Create and activate virtual environment with uv:

uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

uv pip install -r requirements.txt

4. Set up environment variables:

cp .env.example .env
# Edit .env and add your ANTHROPIC_API_KEY

Usage

Web Interface

Run the Streamlit app:

streamlit run app.py

Then open http://localhost:8501 in your browser.

Command Line

Process papers directly from the command line:

python main.py path/to/your/paper.pdf

The summary will be saved in paper_summaries/ directory.