Hi Julia community! I’m trying to fit a custom, one-layer NN to an SDE by computing an MSE loss. Here is my code defining the model and computing the loss (noise and sample_features are cuda matrices defined globally) d = 4 l_list = ones((d,1)) samples_rff = rand(TDist(2*nu), (d, n_samples)); A = …

DiffEqGPU.jl with CUDA: Error computing gradients through SDE solver

MatthieuDarcy January 9, 2024, 5:25am 5

Unfortunately, this didn’t work, but thank you for the suggestion. To be clear I’m using

val, grads = Zygote.withgradient(objective, weights)

to compute the gradients.

Topic		Replies	Views
Zygote and SDE failure: cannot differentiate through Lux NN Machine Learning diffeq , flux , sde , zygote , autodiff	5	655	July 14, 2022
Correct sensealg for an ensemble problem of stochastic UODEs with diagonal noise Machine Learning diffeq , flux	17	982	November 23, 2020
Problems using BacksolveAdjoint for an SDE with DiffEqFlux Machine Learning sde	5	721	September 29, 2021
Reverse mode AD for SDEs Modelling & Simulations sde , sciml	0	72	July 22, 2024
DiffEqFlux: SOSRA2() algorithm + BacksolveAdjoint() gives an error Machine Learning diffeq , neural-network	2	558	May 23, 2021