[incubator-mxnet] branch master updated: Removing the re-size for validation data, which breaking the validation accuracy of CIFAR training (#12362)

[zh...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

zhreshold pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git


The following commit(s) were added to refs/heads/master by this push:
     new ceabcaa  Removing the re-size for validation data, which breaking the validation accuracy of CIFAR training (#12362)
ceabcaa is described below

commit ceabcaac77543d99246415b2fb2d8c973a830453
Author: Shufan <33...@users.noreply.github.com>
AuthorDate: Thu Sep 13 01:59:54 2018 +0800

    Removing the re-size for validation data, which breaking the validation accuracy of CIFAR training (#12362)
---
 example/image-classification/common/data.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/example/image-classification/common/data.py b/example/image-classification/common/data.py
index df1449b..78385a7 100755
--- a/example/image-classification/common/data.py
+++ b/example/image-classification/common/data.py
@@ -194,7 +194,6 @@ def get_rec_iter(args, kv=None):
         std_r               = rgb_std[0],
         std_g               = rgb_std[1],
         std_b               = rgb_std[2],
-        resize              = 256,
         data_name           = 'data',
         label_name          = 'softmax_label',
         batch_size          = args.batch_size,