I was hoping to keep track of loss values while running training with Flux.jl. However, so far I have been unable to both evaluate my loss gradient and evaluate my loss at the same time. I have to run them separately: ℓ = get_loss(stuff) ∇ = gradient(() -> get_loss(stuff), params) However, I assum…

Flux Get Forward Pass Results when Taking Gradient

JLDC October 16, 2022, 4:24pm 2

You should be able to do so by using Zygote.pullback() as per the very last code snippet on this page: Training · Flux

1 Like

Topic		Replies	Views
Get intertermediate results from gradient computation General Usage flux , zygote	0	320	October 12, 2020
Zygote pullback vs gradient New to Julia flux , zygote	2	2207	December 25, 2020
Is it possible perform reverse mode differentiation (Flux.jl with Zygote.jl) of a forward mode differentiation result (e.g. ForwardDiff)? Machine Learning question , flux	3	1466	March 10, 2020
Zygote: how to get intermediate values without evaluating function twice? General Usage differentiation , zygote	4	766	October 11, 2021
Gradient of gradient Machine Learning	9	1304	November 6, 2020