-
Pasted here for ease.
What I'm using locally for some testing @vchuravy
```
module ReverseRules
using Enzyme
using Enzyme: EnzymeRules
using LinearAlgebra
using Test
f(x) = x^2
function f_ip(x)
…
-
I am on Enzyme main (30b6b2d93d8ef1bdfb9f628e8c111d123cc4595e), CUDA master (7ff012) and Julia 1.10.6.
```julia
using Enzyme, CUDA
function f(a, b)
c = a .+ b
Array(c)[1]
end
a = CuArray(rand(…
-
When I try to differentiate OMEinsum, the following error were thrown. The code is shown bellow, where the variable `code` is a mutable, callable object. Since it does not carry gradient, I suppose it…
-
```julia
using Reactant, Enzyme, Lux, Random, LinearAlgebra
const xdev = reactant_device()
const cdev = cpu_device()
model = Dense(5 => 5, gelu);
ps, st = Lux.setup(Random.default_rng(), model) |> x…
-
It is not passed to the shuffled first-order operators
-
Now that direct adjoints are starting to work with Enzyme over OrdinaryDiffEq.jl, it would make sense to add this to the SciMLSensitivity.jl system.
```julia
using Enzyme, OrdinaryDiffEq, StaticA…
-
The following code (using Julia 1.10.6 and Enzyme 13.17) does simple Nested AD. However this results in an error.
```
using Enzyme, Lux, Random, ComponentArrays, LinearAlgebra
n = 1
x_batch = randn(…
-
[out.log](https://github.com/user-attachments/files/17638519/out.log)
The code below segfaults on Julia 1.10 and Enzyme v0.13.13
```julia
using Enzyme
import .EnzymeRules: augmented_primal, reverse, …
-
I've run into an issue where I want to compute the gradient of an implicit function that itself depends on another implicit function. I can do the operation successfully with `FowardDiff`, however I …
-
MWE:
```julia
module MWE
using Roots: Roots
using Enzyme: Enzyme
function find_zeta(y::T) where {T