search

JuliaGaussianProcesses / AbstractGPs.jl

Abstract types and methods for Gaussian Processes.
https://juliagaussianprocesses.github.io/AbstractGPs.jl/dev
Other
220 stars 21 forks source link

Opt out of `rrule` for `mean` #323

Closed devmotion closed 2 years ago

devmotion commented 2 years ago

https://github.com/JuliaDiff/ChainRules.jl/pull/615 broke AD since it captures also mean(::AbstractGP, ::AbstractVector). Opting out of the specific rule seems to fix test errors locally.

cc @simsurace

codecov[bot] commented 2 years ago

Codecov Report

Merging #323 (388a5eb) into master (ad9eb28) will not change coverage. The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #323   +/-   ##
=======================================
  Coverage   97.62%   97.62%           
=======================================
  Files          10       10           
  Lines         379      379           
=======================================
  Hits          370      370           
  Misses          9        9
Impacted Files Coverage Δ
src/abstract_gp.jl 100.00% <ø> (ø)

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update ad9eb28...388a5eb. Read the comment docs.

simsurace commented 2 years ago

Awesome! This fixes my issue at least.

oxinabox commented 2 years ago

Can we add AbstractGPs as a reverse dependency test for ChainRules.jl to avoid this in future?