-
The ML model training lifecycle has standardized around three pieces:
1) Forward pass. This is generally invoked via `model(x)`, although in this repository I think you're more likely to use `mode…
-
When I attempt to add and precompile Flux, I receive the error `LoadError: UndefVarError: allequal not defined`. This problem persists when creating a new environment and adding just Flux, so it seems…
-
This doesn't work:
```julia
julia> using ChainRules, MeasureBase, InverseFunctions
julia> f(x) = 3x + 1
f (generic function with 1 method)
julia> finv(y) = (y - 1) / 3
finv (generic function…
-
Would be nice to have gradients for e.g. `QRPackedQ`, similar to what we have for Cholesky.
-
Quite a few packages define a function to tell whether they are inside AD, as of https://github.com/FluxML/Flux.jl/pull/1863/files#r806287154 Flux has:
```
istraining() = false
# @adjoint istrainin…
-
This was mentioned as an aside in https://github.com/JuliaLang/julia/issues/35667 but it deserves it's own issue I think.
Consider for a input of real `zeros`
```julia
julia> eigen(zeros(3,3))
E…
-
最近主要在了解NiLang和ChainRules~所以先提个关于ChainRules的问题。
项目要求里提到
>We will port the generated backward rules to ChainRules.jl as an extension
这里写个代码示例
``` julia
using ChainRules
# 定义函数foo
function foo(…
-
This only happens when the (co)tangent is 0.
```julia
julia> using ChainRules
julia> ChainRules.frule((ChainRules.ZeroTangent(), 0.0), sqrt, 0.0)
(0.0, NaN)
julia> ChainRules.rrule(sqrt, 0.0)…
-
Trying to compute gradients of the `rand` function wrt to parameters for certain distributions will produce incorrect results, because some of these functions use branching or iterated algorithms and …
-
Hi @b-fg! Just opening an issue to chat if you have any more questions or bugs related to DifferentiationInterface.jl. I didn't really understand from https://github.com/gdalle/DifferentiationInterfac…