[
https://issues.apache.org/jira/browse/SINGA-231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15413775#comment-15413775
]
ASF subversion and git services commented on SINGA-231:
-------------------------------------------------------
Commit bc3b74b3662230f867c42344f0600498368f4785 in incubator-singa's branch
refs/heads/dev from WANG Ji
[ https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git;h=bc3b74b ]
SINGA-231 Batchnormlized VGG model for cifar-10
In this ticket, we implemented a batch normalized VGG model for cifar10
dataset (refer to http://torch.ch/blog/2015/07/30/cifar.html).
* +vgg-parallel.cc for parallel training
* +vgg.py using python language
* fix a bug in ResetLike() method in tensor.h, which before did not
reset shape.
* fix a bug in local_updater.cc, which may cause race condition when
multi-threads try to initialize mutexes concurrently.
* revise batch nomalization layer to support 2D tensor input
> Batchnormlized VGG model for cifar-10
> -------------------------------------
>
> Key: SINGA-231
> URL: https://issues.apache.org/jira/browse/SINGA-231
> Project: Singa
> Issue Type: New Feature
> Reporter: Wang Ji
> Assignee: Wang Ji
> Priority: Minor
>
> In this ticket, we implemented a batch normalized VGG model for cifar10
> dataset (refer to http://torch.ch/blog/2015/07/30/cifar.html).
> * +vgg-parallel.cc for parallel training
> * +vgg.py using python language
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
