I have had so much trouble with GradientTape for the so called Physics-Informed ML. Can we easily take differentiation of Keras layer outputs wrt inputs using GT?
-
This Tweet is unavailable.
-
-
Replying to @EhsanHaghighat
Sure, you just need to make sure the inputs are tracked by the tape
1 reply 0 retweets 0 likes -
Replying to @fchollet
Is this supposed to work? If so, can you please point me to where I am doing wrong? I know that evaluating the network on tensor data works fine, but it is not a good setup for PINNs.pic.twitter.com/dgrtpgz5xb
1 reply 0 retweets 0 likes -
Replying to @fchollet @EhsanHaghighat
François Chollet Retweeted François Chollet
Note that `Input()` is for Functional model construction only. The Functional API is a graph-building API. See e.g.https://twitter.com/fchollet/status/1414665804570796038 …
François Chollet added,
François CholletVerified account @fchollet
Tweetorial: the Functional API in Keras.
Deep learning models are basically graphs of layers. Therefore, an intuitive API for defining deep learning models should be a *graph-definition API*.
That's what the Functional API is: a Python-based DSL for graphs. It looks like this: pic.twitter.com/rthCY2Jqhz
Show this thread
8:08 PM - 6 Aug 2021
0 replies
0 retweets
1 like
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.