experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Improve error message format when reporting batch dimension mismatch in conv2d.

Browse Source 
Before:
input and out_backprop must have the same batch sizeinput batch: 2outbackprop batch: 1 batch_dim: 0

After:
input and out_backprop must have the same batch size. Input batch: 2, outbackprop batch: 1 , batch_dim: 0
PiperOrigin-RevId: 324314629
Change-Id: I5e65341d545203209629933202d463374596b8e2
This commit is contained in:
Haitang Hu 2020-07-31 17:16:41 -07:00 committed by TensorFlower Gardener
parent d5baa96fed
commit 5313d56b1e
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
tensorflow/core/kernels/conv_grad_shape_utils.cc
View File

@ -115,10 +115,10 @@ Status ConvBackpropComputeDimensionsV2(
dims->batch_size = input_shape.dim_size(batch_dim);
if (dims->batch_size != out_backprop_shape.dim_size(batch_dim)) {
return errors::InvalidArgument(
label, ": input and out_backprop must have the same batch size",
"input batch: ", dims->batch_size,
"outbackprop batch: ", out_backprop_shape.dim_size(batch_dim),
" batch_dim: ", batch_dim);
label, ": input and out_backprop must have the same batch size.",
" Input batch: ", dims->batch_size,
", outbackprop batch: ", out_backprop_shape.dim_size(batch_dim),
", batch_dim: ", batch_dim);
}
int feature_dim = GetTensorFeatureDimIndex(num_dims, data_format);