Nice, but what if we want to use them with @PyTorch ? Are they still easy to use?
People tell me, "TPU are too hard to use". They *were*. We've made considerable usability improvements in TF 2.0 and 2.1. Check it out.https://twitter.com/fchollet/status/1203056199479881729 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I am organising a road show in europe with google to teach people how to sue TPUs. This will definitely make it! Thanks a lot.
-
Of course I meant *use* not sue...
End of conversation
New conversation -
-
-
The old namespace API for variable definition was great if you knew how to use it though.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
How about using them with other frameworks?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
That's awesome, but for the love of God keep backward compatibility...
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@martin_gorner released an amazing notebook when TF 2.1 rc wasn't released. He showed how to use TPU with Keras. I tried that out and the entire thing took 15 Mins max.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Is TPU available TF 2.0 on google colab? I switched my runtime to try use it, didn’t expect much without some “changes”, was surprised epoch dropped from ~1000s to ~200s. Did I just use TPU? Wohoo... just to be cautious, I still need to sanity check things.
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.