https://developers.google.com/ml-kit/custom-models
replyJust see out of touch with reality this section is compared to the insanse community work on Apple Silicon across whisper.cpp/mflux/llama.cpp/MLX/Exo & way more
What about it? Google's investment in Tensorflow Lite has paid off a lot better than MLX or CoreML did.
replyAndroid phones didn't need "insane community work" because (unlike Apple hardware) they support the same Vulkan compute shaders that PC GPUs use.