Germán Abrevaya

I the second case (remaking the problem with new `tspan`), if we don't use an `EnsembleProblem`, we get another error, which might be the cause of Error 2. MWE 3...

Hi! Thanks for your responses! Unfortunately, neither the workaround suggested by @avik-pal nor pining Zygote to v0.6.43 worked for any of the MWEs. @ChrisRackauckas, note that in contrast to the...

Hi, I'm also interested in setting different colors to edges and to node borders. I found [here](https://github.com/JuliaPlots/GraphRecipes.jl/pull/120) that at some point it was possible to do it. Setting different edges...

Here is a MWE in the context of LatentDiffEq.jl. Later I'll create a more general (minimal) MWE. ```julia using LatentDiffEq using Flux using OrdinaryDiffEq using DiffEqSensitivity using Random import LatentDiffEq.Decoder...

I reported the more general MWE in SciML/DiffEqSensitivity.jl#611.

Is there any advances on this issue?

Hi! After rewriting the loss in different ways I found that the problem is related to the broadcasting: ```julia using Flux h = [0.0] function m(x) y = Wx .*...

Hi! I tried to test this with something similar to [SciML/DiffEqFlux.jl#391 (comment)](https://github.com/SciML/DiffEqFlux.jl/issues/391#issuecomment-680127253) and found that there are some issues. ```julia using Flux using Random Random.seed!(1) ANN = RNN(1, 1); par,...

In any case, I'm also trying to understand the motivation of this modification. In general, don't you reset the state of a recurrent model after updating the parameters of the...

However, when using the latest version of Flux from master, GalacticOptim still throws an error when trying to use a recurrent model (even in the case when resetting the state...