Archive — US — techcrunch
Pruna AI open sources its AI model optimization framework
Pruna AI, a European startup, open-sources its AI model optimization framework, allowing developers to employ various efficiency methods like caching, pruning, quantization, and distillation on AI models. This new framework enables easy combination and evaluation of these methods, similar to what Hugging Face has done for transformers. Pruna AI has raised $6.5 million in seed funding and is focusing on image and video generation models.
Key points
- Pruna AI is releasing an open-source optimization framework for AI models on Thursday.
- The framework incorporates various methods like caching, pruning, quantization, and distillation.
- It allows for simple evaluation of model quality after compression and performance improvements.
- The approach parallels Hugging Face’s efforts to standardize the use of transformers in AI.
- Existing users include Scenario and PhotoRoom, indicating immediate industry interest.
- Pruna AI has raised $6.5 million in seed funding, showing financial backing for its technology.
Why it matters
- The trend of open-sourcing AI frameworks aligns with the broader IT shift towards collaboration and accessibility in software development.
- Historical reliance on in-house AI model optimization may see changes with more open source solutions emerging.
- Optimizing AI models is increasingly critical with growing business demands for efficient algorithms and cloud computing needs, especially by 2025.
Pruna AI's open-source framework represents a significant advancement in AI model optimization, facilitating broader access to cutting-edge efficiency techniques while fostering innovation in the AI industry.