Condensr

Condensr is an AI-powered text and dialogue summarization web application built with FastAPI and the HuggingFace T5-small transformer model.

Features

• Privacy First: The model runs entirely locally. No external APIs, no data sent to third parties.
• Custom Dialogue Parsing: Specifically engineered to parse unformatted dialogue texts, extracting each speaker's voice to dramatically improve multi-character summarization context.
• Hybrid Inference: Automatically detects whether the input is prose or a conversation, tokenizing conversations into per-speaker batches using Beam Search decoding.
• Modern UI: A beautiful, responsive, zero-dependency Vanilla JS/CSS frontend styled with glassmorphism.

Installation

1. Clone this repository:

   git clone https://github.com/yourusername/condensr.git
   cd condensr

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   python -m uvicorn app:app --reload

4. Open your browser and navigate to http://127.0.0.1:8000/.

Note: On your first run, the app will automatically download the t5-small model (~240MB) from HuggingFace.

License

This project is licensed under the MIT License - see the LICENSE file for details.