Skip to content

Gradients for batched_adjoint etc. #272

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.

Sign up for GitHub

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

Merged
merged 3 commits into from
Jan 20, 2021
Merged

Gradients for batched_adjoint etc. #272

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9588541
gradients for batched_adjoint etc.
Jan 20, 2021
c8f99f4
fix finitediff problem
Jan 20, 2021
fc31036
import
Jan 20, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 9 additions & 0 deletions src/batched/batchedadjtrans.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,3 +91,12 @@ end
Base.unsafe_convert(::Type{Ptr{T}}, A::BatchedAdjOrTrans{T}) where {T} =
Base.unsafe_convert(Ptr{T}, parent(A))

# Gradients
function ChainRulesCore.rrule(::typeof(batched_transpose), A::AbstractArray{<:Any,3})
b_transpose_back(Δ) = (NO_FIELDS, batched_transpose(Δ))
batched_transpose(A), b_transpose_back
end
function ChainRulesCore.rrule(::typeof(batched_adjoint), A::AbstractArray{<:Any,3})
b_adjoint_back(Δ) = (NO_FIELDS, batched_adjoint(Δ))
batched_adjoint(A), b_adjoint_back
end
8 changes: 7 additions & 1 deletion test/batchedmul.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
using NNlib, Test, LinearAlgebra
using NNlib: storage_type, storage_typejoin, is_strided,
batched_mul!, _unbatch, _copy_if_faster, BatchedAdjoint
batched_mul!, _unbatch, _copy_if_faster,
BatchedAdjoint, BatchedTranspose

function bmm_test(a,b; transA = false, transB = false)
bs = size(a,3)
Expand Down Expand Up @@ -232,9 +233,14 @@ end

end

FiniteDifferences.to_vec(x::BatchedAdjoint) = FiniteDifferences.to_vec(collect(x))
FiniteDifferences.to_vec(x::BatchedTranspose) = FiniteDifferences.to_vec(collect(x))

@testset "AutoDiff" begin
M, P, Q = 13, 7, 11
B = 3
gradtest(batched_mul, randn(rng, M, P, B), randn(rng, P, Q, B))
gradtest(batched_mul, batched_adjoint(randn(rng, P, M, B)), randn(rng, P, Q, B))
gradtest(batched_mul, randn(rng, M, P, B), batched_transpose(randn(rng, Q, P, B)))
end