Developers · Updated 29 Oct, 03:59 am IST
Google adds faster GPU inference and smaller app sizes with LiteRT in Play services
Why it matters for readers: It means some phone apps will take up less space and run ML features faster without needing extra downloads.
- LiteRT in Google Play services now supports a new GPU inference stack that runs about 1.4x faster than the legacy GPU delegate.1
- The CompiledModel API is presented as LiteRT’s modern successor to the legacy Interpreter API for running custom models on-device.1
- By shipping the runtime once in Google Play services instead of bundling it in each app, Google says apps that switch are about 8 MB smaller in testing.1
- More than 100,000 Android apps currently bundle their own copy of LiteRT or TensorFlow Lite, so switching to the shared runtime could multiply storage and memory savings across devices.1
- Google showcases the change in the Google AI Edge Gallery app, where features like image segmentation run on LiteRT in Play services.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started