Michael Abbott

MWE with only Zygote: ```julia julia> using Zygote julia> let x = rand(Bool, 12) w = rand(Float32, 4, 3) gradient(w -> sum(w * reshape(x, 3, 4)), w) end ERROR: MethodError:...

I think `x.^2` and `x.^3` should go here... is `x .^ (p - 1)` producing Float64 somehow? https://github.com/FluxML/Zygote.jl/blob/59e7ec1b32c4d4571120e143d9c40e48b370f22b/src/lib/broadcast.jl#L106-L109

Dup of #599 I think. There's a PR which could be revived.

Thanks I hadn't seen that. Got a wrong answer in this test on CI (tiny arrays though) but didn't investigate further: https://github.com/FluxML/NNlib.jl/pull/614/files#diff-df0d2a37225f09d22727651479dc1cd59f2b8358f4eb1e2be98c9b04e215be86R31-R34

Wait that's a very confusing answer. `setindex!` isn't wrong here, but the user didn't ask for any `Symmetric` array to be mutated. Something in Plots.jl has made a new matrix...

I have much the same error, with Julia 1.11.7 (although it's not new). This issue https://github.com/JuliaLang/Pkg.jl/issues/3404 says it may be something to do with versions of std. lib. packages not...

Thanks I'm glad if it's useful! And sorry I missed this earlier. The short answer is that you don't. `@tullio` is really not very smart, the macro always makes exactly...

I'm glad if it's useful! Unfortunately this is a known problem. `@tullio` defines various functions, and `@generated` does not allow that. See for example https://github.com/mcabbott/Tullio.jl/issues/11 and https://github.com/mcabbott/Tullio.jl/pull/4#issuecomment-643822313 . This was...

I can reproduce this, but don't know why it's failing. The next thing to check is probably whether the same error occurs with KernelAbstractions alone. Expanding the macro, or reading...

Ah, a possible culprit is this: https://github.com/mcabbott/Tullio.jl/blob/master/ext/TullioCUDAExt.jl The CPU path has recursive tiling before calling `act!`, but the GPU path is meant to skip that. But `tile_halves` in stacktrace above...