When trying to replicate a paper that uses GANs with unconventional loss calculations, I’m having problems with null gradients. The basic steps are the following: genOutput = gen(genInput) MSE(genOut, label) MAE(genOut |> volFrac, label |> volFrac) discOutFake = cat( genInput, condition, genOutput…

Null gradients with GANs

LucasMSpereira December 2, 2022, 12:41am 2

The gradients aren’t null. LinearAlgebra.norm() can’t take nothing as an argument. Using LinearAlgebra.norm(::Nothing, p::Real = 2) = false, the norms of the gradients are calculated as expected.

Topic		Replies	Views
Take gradient of parameters not working Machine Learning question , flux	0	356	January 12, 2021
How to add norm of gradient to a loss function? Machine Learning flux , zygote	16	2275	November 17, 2021
Flux.gradient returns dict of param and Nothing Machine Learning flux	3	794	September 22, 2021
Different loss with Zygote when taking gradients vs. without Machine Learning flux	0	402	December 20, 2019
Autodiff on image convolution Machine Learning question	6	891	June 24, 2019

Null gradients with GANs

Related topics