From e5d71579006320f88ce4406233e15caa029e02f8 Mon Sep 17 00:00:00 2001 From: Ubuntu Date: Fri, 1 Feb 2019 02:37:28 +0000 Subject: [PATCH] modifying SyncBN doc for FP16 use case --- python/mxnet/gluon/contrib/nn/basic_layers.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/python/mxnet/gluon/contrib/nn/basic_layers.py b/python/mxnet/gluon/contrib/nn/basic_layers.py index 28fea1592da4..56f0809b345f 100644 --- a/python/mxnet/gluon/contrib/nn/basic_layers.py +++ b/python/mxnet/gluon/contrib/nn/basic_layers.py @@ -165,7 +165,10 @@ class SyncBatchNorm(BatchNorm): Standard BN [1]_ implementation only normalize the data within each device. SyncBN normalizes the input within the whole mini-batch. - We follow the sync-onece implmentation described in the paper [2]_. + We follow the implementation described in the paper [2]_. + + Note: Current implementation of SyncBN does not support FP16 training. + For FP16 inference, use standard nn.BatchNorm instead of SyncBN. Parameters ----------