Spaces:
Sleeping
Sleeping
metadata
title: Multi Beam Text Streamer
emoji: π
colorFrom: green
colorTo: pink
sdk: docker
pinned: false
short_description: A generic text streamer that supports beam search and manage
Beam Search Generator with MultiBeamTextStreamer
This Hugging Face Space hosts a Beam Search Generator web application, powered by a Flask backend and integrated with Hugging Face Transformers. The application provides real-time visualization of beam search generation, offering insights into how language models explore multiple text completion possibilities simultaneously.
π Features
- Real-time visualization of active and completed beams.
- Adjustable generation parameters:
- Number of beams.
- Maximum tokens.
- Generation speed (using delay sliders).
- Seamless interaction with Hugging Face Transformers.
- Powered by the custom
MultiBeamTextStreamer
from the Transformers library.
π How It Works
- Input a prompt in the text box.
- Configure generation settings:
- Number of beams.
- Maximum token count.
- Delay speed.
- Click Generate to visualize the beam search process in real-time.
π οΈ Setup Instructions
Running Locally
- Clone the repository:
git clone https://huggingface.co/spaces/<your-username>/<your-space-name> cd <your-space-name>