I suspect that less than 10 years from now, all of the DL training/architecture tricks that came from the arXiv firehose over 2015-2019 will have been entirely superseded by automated search techniques. The future: no alchemy, just clean APIs, and quite a bit of compute.
-
-
This is already the case
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I feel TensorFlow is high level, after trying to reimplement some layers in Apple accelerate API in its raw form, maybe I am several years behind the curve.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@fastdotai has already started this trend by encapsulating the latest techniques ... takes 3 lines of code to train a network with SoTA resultsThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
We have to do them now though to make that future possible. I'm already excited by the advancements in optimization being abstracted i.e. GPipe and am looking forward to the future.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This ismy dream
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.