From a10caf7c2360cb18c788c19dc73ec7a1e055866a Mon Sep 17 00:00:00 2001 From: Yibing Liu Date: Tue, 30 Jan 2018 04:41:50 -0800 Subject: [PATCH] Fix typos in label_smooth_op --- paddle/operators/label_smooth_op.cc | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/paddle/operators/label_smooth_op.cc b/paddle/operators/label_smooth_op.cc index 432d4c7d01a..c89082f44b3 100644 --- a/paddle/operators/label_smooth_op.cc +++ b/paddle/operators/label_smooth_op.cc @@ -79,8 +79,7 @@ it is not guaranteed to generalize. Second, it encourages the differences between the largest logit and all others to become large, reducing the ability of the model to adapt. Label smoothing is proposed to encourage the model to be less confident, which replaces the ground-truth label $y$ with the weighted -sum of itselft and some fixed distribution $\mu$, -i.e. +sum of itself and some fixed distribution $\mu$, i.e. $$ \tilde{y} = (1 - \epsilon) * y + \epsilon * \mu, -- GitLab