@fchollet If I write a network in Keras/tf.keras which includes a Lambda layer, and given the fact that backend is tf, will that layer be supported by TFLlite?
-
The next big thing we can do is to make sure that every Keras op can run seamlessly on TFLite. This would encourage people using Keras to go for on-device ML inference instead of putting the model on a server.
Yes, ideally we will get to a world where any model (even with custom layers) can be exported to any device, mobile, browser, embedded. And even retrained/fine-tuned on-device. We're already pretty close to that world.
-
-
I put a lot of models to mobile and embedded devices(mostly RPi) and TBH we aren't there yet. Every new thing generates so many errors but TFLite is still much better than Tensorflow Mobile.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.