-
Notifications
You must be signed in to change notification settings - Fork 70
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
partition, type analysis mismatch #880
Comments
per the recent jll bump, just to confirm, what error do you get? |
On Julia 1.9.0 and Enzyme 9487eb8 I get:
|
Possibly a related issue given the similar message and the fact that On Julia 1.9.0 and Enzyme 9487eb8: using Enzyme
function f(x)
y = [1.0, 2.0, 3.0]
return sum(sort(x) .* y) # Works if sort is removed
end
x = [5.0, 6.0, 7.0]
dx = zero(x)
autodiff(Reverse, f, Active, Duplicated(x, dx))
|
using Enzyme, Statistics
Enzyme.API.printall!(true)
function f(y)
first(sort!(y))
end
autodiff(Reverse, f, Active, Duplicated([1.0], [0.0])) |
This no longer gives the above error, but does have a type analysis mismatch, presumably due to the inner radix sort. Nevertheless, @jgreener64 would you consider contributing a custom rule for |
median
gradient errors on Julia 1.9
Happy to try. |
@jgreener64 I tried the code again, and it looks like this still would need a rule for [presumably this: https://github.com/JuliaLang/julia/blob/604609a3fc1bf76470f4bca2333628cb08419f81/base/sort.jl#L1070 ] |
Yes I can try. |
I think a rule for |
I am on Julia 1.9.0 and Enzyme 16818fd. The following errors, though it works on Julia 1.8.5 (with
warning: didn't implement memmove, using memcpy as fallback which can result in errors
).Version info:
The text was updated successfully, but these errors were encountered: