Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

[BUGFIX] Add check to make sure num_group is non-zero #20186

Merged
merged 2 commits into from
Apr 19, 2021
Merged

[BUGFIX] Add check to make sure num_group is non-zero #20186

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
fcf37f9
add check for group not equal zero
Adnios Apr 17, 2021
2824daa
num_group in convolution must be positive
Adnios Apr 19, 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
6 changes: 6 additions & 0 deletions src/operator/nn/convolution.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,6 +99,8 @@ static bool ConvolutionShape(const nnvm::NodeAttrs& attrs,
// 1d conv
CHECK_EQ(dshp.ndim(), 3U) << "Input data should be 3D in batch-num_filter-x";
Shape<3> dshape = ConvertLayout(dshp.get<3>(), param_.layout.value(), kNCW);
CHECK_GT(param_.num_group, 0U) \
<< "Range only supports num_group > 0, received " << param_.num_group;
Shape<3> wshape = Shape3(param_.num_filter / param_.num_group,
mxnet::dim_size_is_known(dshape, 1) ? dshape[1] / param_.num_group : -1,
param_.kernel[0]);
Expand Down Expand Up @@ -149,6 +151,8 @@ static bool ConvolutionShape(const nnvm::NodeAttrs& attrs,
CHECK_EQ(dshp.ndim(), 4U) \
<< "Input data should be 4D in batch-num_filter-y-x";
Shape<4> dshape = ConvertLayout(dshp.get<4>(), param_.layout.value(), kNCHW);
CHECK_GT(param_.num_group, 0U) \
<< "Range only supports num_group > 0, received " << param_.num_group;
Shape<4> wshape = Shape4(param_.num_filter / param_.num_group,
mxnet::dim_size_is_known(dshape, 1) ? dshape[1] / param_.num_group : -1,
param_.kernel[0], param_.kernel[1]);
Expand Down Expand Up @@ -208,6 +212,8 @@ static bool ConvolutionShape(const nnvm::NodeAttrs& attrs,
CHECK_EQ(dshp.ndim(), 5U) \
<< "Input data should be 5D in batch-num_filter-depth-y-x";
Shape<5> dshape = ConvertLayout(dshp.get<5>(), param_.layout.value(), kNCDHW);
CHECK_GT(param_.num_group, 0U) \
<< "Range only supports num_group > 0, received " << param_.num_group;
Shape<5> wshape = Shape5(param_.num_filter / param_.num_group,
mxnet::dim_size_is_known(dshape, 1) ? dshape[1] / param_.num_group : -1,
param_.kernel[0], param_.kernel[1], param_.kernel[2]);
Expand Down