Cache AI models in the browser | AI on Chrome | Chrome for Developers
June 28, 2024
Generative AI models can be enormous. This is why a lot of AI usage today happens in the cloud. Increasingly, apps are running highly optimized models directly on-device. While demos of LLMs running in the browser exist, here are some production-grade examples of other models running in the browser:
- Adobe Photoshop runs a variant of the Conv2D model on-device for its intelligent object selection tool.
- Google Meet runs an optimized version of the MobileNetV3-small model for person segmentation for its background blur feature.
- Tokopedia runs the MediaPipeFaceDetector-TFJS model for real-time face detection to prevent invalid signups to its service.
- Google Colab allows users to use models from their hard disk in Colab notebooks.
Source: Cache AI models in the browser | AI on Chrome | Chrome for Developers
Running Large Language Models on device is something we’re already seeing, with much more promised. But what about running your own models in the browser on a user’s device? Well that too is happening already. Here Thomas Steiner looks at web apis for storing large amounts of data on device, in a use case like running LLMs locally.
Related Speakers
