From dev-return-4186-archive-asf-public=cust-asf.ponee.io@singa.apache.org  Wed Feb 12 11:54:32 2020
Return-Path: <dev-return-4186-archive-asf-public=cust-asf.ponee.io@singa.apache.org>
X-Original-To: archive-asf-public@cust-asf.ponee.io
Delivered-To: archive-asf-public@cust-asf.ponee.io
Received: from mail.apache.org (hermes.apache.org [207.244.88.153])
	by mx-eu-01.ponee.io (Postfix) with SMTP id BCFB3180657
	for <archive-asf-public@cust-asf.ponee.io>; Wed, 12 Feb 2020 12:54:31 +0100 (CET)
Received: (qmail 1480 invoked by uid 500); 12 Feb 2020 11:54:31 -0000
Mailing-List: contact dev-help@singa.apache.org; run by ezmlm
Precedence: bulk
List-Help: <mailto:dev-help@singa.apache.org>
List-Unsubscribe: <mailto:dev-unsubscribe@singa.apache.org>
List-Post: <mailto:dev@singa.apache.org>
List-Id: <dev.singa.apache.org>
Reply-To: dev@singa.apache.org
Delivered-To: mailing list dev@singa.apache.org
Received: (qmail 1468 invoked by uid 99); 12 Feb 2020 11:54:31 -0000
Received: from ec2-52-202-80-70.compute-1.amazonaws.com (HELO gitbox.apache.org) (52.202.80.70)
    by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 12 Feb 2020 11:54:31 +0000
From: GitBox <git@apache.org>
To: dev@singa.apache.org
Subject: [GitHub] [singa] chrishkchris commented on issue #598: fix softmax nan
 problem
Message-ID: <158150847105.19080.5156166260905272102.gitbox@gitbox.apache.org>
References: <infra.598.MDExOlB1bGxSZXF1ZXN0Mzc0MjYyNTU0.gitbox@gitbox.apache.org>
In-Reply-To: <infra.598.MDExOlB1bGxSZXF1ZXN0Mzc0MjYyNTU0.gitbox@gitbox.apache.org>
Date: Wed, 12 Feb 2020 11:54:31 -0000
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit

chrishkchris commented on issue #598: fix softmax nan problem
URL: https://github.com/apache/singa/pull/598#issuecomment-585170944
 
 
   I strongly support this change for two reasons:
   1. The current setting fails some network structure/dataset, which make the functions not universal for all applications.
   2. SoftMax forward spent only very short time within the CNN training, e.g. resnet50 has many conv layers but only one SoftMax layer, so the time spent in SoftMax is not the bottleneck. 
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services