DiffEqFlux.jl
DiffEqFlux.jl copied to clipboard
SciML
Pre-built implicit layer architectures with O(1) backprop, GPUs, and stiff+non-stiff DE solvers, demonstrating scientific machine learning (SciML) and physics-informed machine learning methods
Hello. I was training a simple UODE and am encountering this error when running `sciml_train()`. The error seems to be somewhere in the interface between `DiffEqFlux` and `GalacticOptim`. I have...
All the collocation methods only work when the timescale is correct. This is easily fixed by artificially "normalising" the time-scale. But when the data has multiple dimensions, with multiple orders...
I'm not certain if this is a DiffEqFlux error or a Zygote error, considering it triggers depending on some seemingly harmless changes to the source code. MWE: ``` julia> using...
I am using a combination of packages to use ANNs within coupled ordinary differential equations. When coupling less than 20 ODEs everything runs smoothly, whereas for more than 20 ODEs...
This will improve load times and is easy to do.
I need to calculate the Laplacian of the densities modelled by a normalizing flow w.r.t. to the inputs. On CPU, I can e.g. use the following code (which works, but...
Considering the amount of code rewritten, it makes more sense to define it as a struct or purely define one ff layer that encapsulates a fourier feature encoding and dense...
The noise term may be weighted incorrectly for SGLD and pSGLD. I am just learning this topic and apologize if I have misread the theory. For SGLD, on page https://diffeqflux.sciml.ai/stable/examples/BayesianNODE_SGLD/,...
When I run this code: ```julia using DiffEqFlux, DifferentialEquations, GalacticOptim, Distributions nn = Chain( Dense(1, 1, tanh), ) |> f32 tspan = (0.0f0, 1.0f0) sensealg = InterpolatingAdjoint(autojacvec=false) ffjord_mdl = FFJORD(nn,...
I'd like to send to gpu a NeuralODE object (embedded neural network included), instead of creating a NeuralODE out of a model already on gpu. Below an example: ```julia julia>...
