News for hackers. Stuff that matters.
Re:Tensorflow 2.0 by akhilcacharya on Monday September 30, 2019 @02:31PM (#21118298)
Has ONNX not caught fire on mobile yet?
Re:Tensorflow 2.0 by sairahul82 on Monday September 30, 2019 @02:37PM (#21118368)
It has but not as good as tensorflow support. This is one thing i miss truly in pytorch. Otherwise pytorch is wonderful.
Re:Tensorflow 2.0 by m0zg on Monday September 30, 2019 @02:50PM (#21118488)
It has? Where? Who uses ONNX for anything? I doubt it even can work, period. The moment you do anything other than a bare bones classifier (which nobody really runs on devices - you need more complex models to solve real world problems) you run into ops unsupported by your inference framework, and that's if ONNX is supported by its tooling in the first place. In fact you could also run into unsupported ops during export as well: that is, it is somewhat likely that you won't even be able to export your model unless it consists entirely of the ops ONNX standard implements. The rest can be exported as opaque ops, but your inference tooling will not know what to do with those for sure.
Re:Tensorflow 2.0 by tsbinz on Monday September 30, 2019 @05:33PM (#21119862)
It can work for more than a bare bones classifier. It's certainly not painless and sometimes you need some manual work to translate your model but work it does ...
Re:Tensorflow 2.0 by throwawaybbqed on Monday September 30, 2019 @08:21PM (#21120885)
For mobile, the situation seems worse than a year ago. TF Mobile mostly worked. The current situation with TF lite is a joke .. tried a real-world Pytorch -> ONYX -> TF/TF lite and it has been weeks of misery.
Re:Tensorflow 2.0 by sairahul82 on Monday September 30, 2019 @04:40PM (#21119465)
I agree with you only basic models works !!
Re:Tensorflow 2.0 by applecrazy on Monday September 30, 2019 @02:40PM (#21118389)
Doesn't CoreML abstract 90% of the actual model format/training source away? Last time I played with it in Xcode, it was painless to pull a pretrained TF model and use as-is.
Re:Tensorflow 2.0 by Q6T46nT668w6i3m on Monday September 30, 2019 @03:27PM (#21118827)
The number of operations supported by CoreML or Onyx is limited.
Re:Tensorflow 2.0 by ghop02 on Monday September 30, 2019 @04:30PM (#21119387)
Core ML 3 adds a ton of new operations, including control flow support.

https://heartbeat.fritz.ai/whats-new-in-core-ml-3-d108d352e5...

Re:Tensorflow 2.0 by throwawaybbqed on Monday September 30, 2019 @08:02PM (#21120808)
What about Android?
Re:Tensorflow 2.0 by pjmlp on Tuesday October 01, 2019 @02:22AM (#21122529)
Android has its own incompatible world with NN and NDK fun.

https://developer.android.com/ndk/reference/group/neural-net...

Or you get to use Tensorflow Lite and target both platforms, also with a smaller feature set than its big brother.

https://www.tensorflow.org/lite/guide/get_started

Re:Tensorflow 2.0 by m0zg on Tuesday October 01, 2019 @03:39AM (#21122912)
But, crucially, larger feature set than the likes of CoreML etc. You don't get access to the NPU that way though, at least not on iOS. The only acceleration option there seems to be Metal. Which isn't bad, but also not the most power efficient thing the hardware supports.

Still though, it's the only game in town if you don't want to have insane un-debuggable headaches everywhere you deploy to device. Plus it also supports embedded Linux boards, and pretty much all current TPU-like things available there.

Re:Tensorflow 2.0 by akhilcacharya on Monday September 30, 2019 @03:32PM (#21118871)
What's an example of a missing operator?