Fritz AI Heartbeat

Core ML is going to be your best bet. Most training is still done server side using frameworks like TensorFlow, Keras, and PyTorch. Once you've trained your model, you can convert it to Core ML with coremltools or export it to Core ML directly if the platform supports it.

Apple has a couple tools, Turi Create and Create ML, to train ML models specifically for mobile use, but their not nearly as fully featured or widely used.

If you're interested specifically in mobile ML, check out https://heartbeat.fritz.ai. We've got a bunch of resources for mobile machine learning. If you're looking for ready-to-use models or tools to manage them in your app, check out Fritz (https://fritz.ai). Disclaimer, I'm a founder at Fritz which sponsors Heartbeat. Happy to answer any questions!

Both OpenGL shader code and on-device ML can be done in React Native. However, you likely won't reach full native performance today.

https://www.youtube.com/watch?v=1HYnrdKEVMU
https://www.youtube.com/watch?v=LbcuLfrt8XA
https://heartbeat.fritz.ai/image-classification-on-react-native-with-tensorflow-js-and-mobilenet-48a39185717c

You're right there will always be a very high demand for native engineers to build the underlying bridge code to back these capabilities. As cross-platform tools gain new bridged capabilities and become faster (https://engineering.fb.com/android/hermes/), as they have been slowly, I think the use-cases where using something like React Native or Flutter is possible and makes sense will grow.

You are right, I would not say that to a person in meatspace and I should not say things like that to people on the internet either.

But:

Looking at their other articles gives me the impression of a content creator that just pollutes the internet for keywords, potentially paid-for.

Just look at https://heartbeat.fritz.ai/ , which this random assortment of code snippets was compiled for, it's all just keyword stuffing as promotion for fritz.ai, plain and simple.

And oh look, nicely delayed submissions of article after article by OP: https://news.ycombinator.com/submitted?id=salma-ghoneim

As a developer, you can. The APIs exist (https://developer.apple.com/documentation/vision/vnrecognizetextrequest), seem simple to use (https://www.hackingwithswift.com/example-code/vision/how-to-use-vnrecognizetextrequests-optical-character-recognition-to-detect-text-in-an-image) and appear to perform decently (https://heartbeat.fritz.ai/comparing-apples-and-google-s-on-device-ocr-technologies-fc5c7becf9f0; compares with Google’s on-device OCR)

So, finally, they don’t step on the toes of third-party developers, and you still complain? ;-)

From 2019: https://heartbeat.fritz.ai/deep-learning-has-a-size-problem-ea601304cd8

> Earlier this year, researchers at NVIDIA announced MegatronLM, a massive transformer model with 8.3 billion parameters (24 times larger than BERT)

> The parameters alone weigh in at just over 33 GB on disk. Training the final model took 512 V100 GPUs running continuously for 9.2 days.

Running this model on a "regular" machine at some useful rate is probably not possible at this time.