metadata
title: README
emoji: 🌍
colorFrom: gray
colorTo: purple
sdk: static
pinned: false
Join the Pruna AI community!
Simply make AI models faster, cheaper, smaller, greener!
Pruna AI makes AI models faster, cheaper, smaller, greener with the pruna package.
- It supports various models including CV, NLP, audio, graphs for predictive and generative AI.
- It supports various hardware including GPU, CPU, Edge.
- It supports various compression algortihms including quantization, pruning, distillation, caching, recovery, compilation that can be combined together.
- You can either play on your own with smash/compression configurations or let the smashing/compressing agent find the optimal configuration [Pro].
- You can evaluate reliable quality and efficiency metrics of your base vs smashed/compressed models. You can set it up in minutes and compress your first models in few lines of code!
How to get started?
You can smash your own models by installing pruna with:
pip install pruna[gpu]==0.1.3 --extra-index-url https://prunaai.pythonanywhere.com/
For more details about installation and tutorials, you can check the Pruna AI documentation.
| Use Case | Free Notebooks |
|---|---|
| 3x Faster Stable Diffusion Models | ▶️ Smash for free |
| Turbocharge Stable Diffusion Video Generation | ▶️ Smash for free |
| Making your LLMs 4x smaller | ▶️ Smash for free |
| Blazingly fast Computer Vision Models | ▶️ Smash for free |
| Smash your model with a CPU only | ▶️ Smash for free |
| Transcribe 2 hours of audio in less than 2 minutes with Whisper | ▶️ Smash for free |
| 100% faster Whisper Transcription | ▶️ Smash for free |
| Flux generation in a heartbeat, literally | ▶️ Smash for free |
| Run your Flux model without an A100 | ▶️ Smash for free |