Skip to content
This repository was archived by the owner on Jul 1, 2023. It is now read-only.

_vjpConv2DBackpropInput using shape instead of using filter size for … #331

Merged
merged 3 commits into from
Jul 3, 2019
Merged

_vjpConv2DBackpropInput using shape instead of using filter size for … #331

Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
697abcc
_vjpConv2DBackpropInput using shape instead of using filter size for …
Jul 3, 2019
61a5d30
Restore indentation
rxwei Jul 3, 2019
6f89871
Not capture shape because `filter` is already being captured.
rxwei Jul 3, 2019
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
4 changes: 2 additions & 2 deletions Sources/TensorFlow/Operators/NN.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -184,8 +184,8 @@ func _vjpConv2DBackpropInput<Scalar: TensorFlowFloatingPoint>(
) -> (Tensor<Scalar>, (Tensor<Scalar>) -> (Tensor<Scalar>, Tensor<Scalar>)) {
let value = conv2DBackpropInput(x, shape: shape, filter: filter,
strides: strides, padding: padding, dilations: dilations)
return (value, { v in
(conv2DBackpropFilter(x, input: v, filterSizes: shape, strides: strides,
return (value, { [filterShape = filter.shapeTensor] v in
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually this pullback is already capturing filter (on line 190), so I think your original implementation is better! I reverted it for you.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks!

(conv2DBackpropFilter(x, input: v, filterSizes: filterShape, strides: strides,
padding: padding, dilations: dilations),
conv2D(v, filter: filter, strides: strides, padding: padding, dilations: dilations))
})
Expand Down