tesnorflow Batch Normalization

1.train或者从checkpoint restore后发现moving_mean和moving_variance都是0和1

bn1_mean = graph.get_tensor_by_name("bn1/moving_mean/read:0")

"bn1/moving_variance:0"

将updates_collections=None即可

net = slim.batch_norm(net,epsilon=0.001,updates_collections=None,scale=True,activation_fn=nn.relu,scope = 'bn1')

posted @ 2017-06-29 14:40  mlj318  阅读(310)  评论(0编辑  收藏  举报