The most modern way to implement a GAN in Keras is actually simply this.pic.twitter.com/7XQqQxZnvz
Show this thread
-
You can call `compile()` & `fit()` on this bad boy. It handles callbacks, it has built-in distribution support, etc. It does everything `fit()` usually does. But with your own low-level training algorithm.pic.twitter.com/m2mmbRtcH2
Show this thread
This is *progressive disclosure of complexity* at work: exactly the amount of low-level control over the details that you need, together with the highest achievable amount of high-level convenience & performance optimization. That's exactly the reason why to use a framework.
-
-
This Tweet is unavailable.
-
This is actually super elegant.
End of conversation
-
-
-
And that’s about what it looks like in jax, too. I think at the right level of control/convenience, a sane framework just brings you here.
-
I’d love to see a screenshot of the Jax version - I’ve been meaning to see what that framework is about.
- Show replies
New conversation -
-
-
That's why I call tf.keras a beauty!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
amazing description of what we should expect from a good framework: enough flexibility to play and understand the fundamentals together with the ease-of-use and speed of implementation (somehow it is what I also find in
#scikitlearn)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.