3 ms·
Would be interesting to get more info on the neural engine. On one hand, I find it fascinating that major manufacturers are now putting neural architectures int
by t_mann 5y ago
Would be interesting to get more info on the neural engine. On one hand, I find it fascinating that major manufacturers are now putting neural architectures into mainstream hardware.
On the other hand I wonder what exactly it can do. To what degree are you tied into a specific neural architecture (eg recurrent vs convolutional), what APIs are available for training it, if it's even meant to be used that way (not just by Apple-provided featues lke FaceID)?
- slimsag 5y agoThey're doing quite a lot of work here: https://developer.apple.com/machine-learning/ https://developer.apple.com/machine-learning/ https://developer.apple.com/machine-learning/create-ml/ https://developer.apple.com/machine-learning/create-ml/ https://developer.apple.com/documentation/createml https://developer.apple.com/documentation/createml
- zozbot234 5y agoTypical "neural engines" are intended for real-time network inference, not training. Training is highly parallel and benefits more from GPU-like vector processing.
- mlajtos 5y agoApple is pushing for training & fine-tuning on the devices too. https://developer.apple.com/documentation/coreml/model_customization/personalizing_a_model_with_on-device_updates https://developer.apple.com/documentation/coreml/model_custo...
- zitterbewegung 5y agoIt's a general purpose accelerator. You have coremltools[1] to convert your trained model into a format or you can make your own using CreateML[2]. [1] https://coremltools.readme.io/docs https://coremltools.readme.io/docs [2] https://developer.apple.com/machine-learning/create-ml/ https://developer.apple.com/machine-learning/create-ml/