It looks like the activations() function (in basic.jl) computes the forward pass to obtain the activations, which would mean that it is computed twice. I guess a fix for this would be to implement a custom alternative to Chain() that propagates both the prediction and a list of activations (or, …

Flux Activations() computes the forward pass again

CarloLucibello May 9, 2020, 5:53am 2

@jaynick could you open an issue on Flux?

Topic		Replies	Views
Flux.jl: Changing the value of activations from hidden nodes during backprop Machine Learning flux	0	525	March 30, 2019
Simple Flux model not learning Machine Learning flux	4	1105	October 21, 2019
Flux.jl RNN performance Machine Learning	11	3173	October 28, 2018
Flux.jl changes in api General Usage	2	228	March 17, 2023
Flux Custom Loss Function Not Working Properly Machine Learning flux , zygote	20	2335	April 2, 2021