jondeuce
jondeuce
I just watched your fantastic JuliaCon talk and would love to try out CompTime.jl in some of my packages. Do you have plans to register CompTime.jl in the near future?...
There have been several issues/PRs related to freezing model parameters: 1. https://github.com/FluxML/Flux.jl/issues/1022 2. https://github.com/FluxML/Flux.jl/issues/1001 3. https://github.com/FluxML/Flux.jl/pull/1101 4. https://github.com/FluxML/Zygote.jl/pull/505 5. https://github.com/FluxML/Optimisers.jl/pull/49 Right now, the recommendation made in the documentation is to...
Simple 1 character change
The following errors for `cholesky` factorizations: ```julia julia> using LinearAlgebra, SparseArrays julia> A = cholesky(sparse(Matrix(I,10,10))); # sparse cholesky factorization with real elements julia> z = rand(ComplexF64, 10); # complex RHS...
Seems to occur for broadcast expressions with more than two arguments within a `getindex`. Possibly related to #537. MWE: ```julia julia> format_text("a[1 .+ 1]") # okay "a[1 .+ 1]" julia>...
This adds a couple small changes on top of [this draft PR](https://github.com/FluxML/Optimisers.jl/pull/63) in order to fix #62: 1. Wrap offset indices in a dummy struct `Offset` to fix the issue...
For example, ```julia using MLUtils buffer = zeros(3,2) data = BatchView(rand(3,100); batchsize = 2) getobs!(buffer, data, 1) buffer == zeros(3,2) # true ``` Looks like it just falls back to...
MWE: ```julia julia> using LinearAlgebra, Zygote, CUDA julia> Zygote.gradient(x -> tr(x), CUDA.zeros(2,2))[1] # works 2×2 Diagonal{Float32, FillArrays.Fill{Float32, 1, Tuple{Base.OneTo{Int64}}}}: 1.0 ⋅ ⋅ 1.0 julia> Zygote.gradient(x -> sum(abs2, x) - tr(x),...