You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@singa.apache.org by "wangwei (JIRA)" <ji...@apache.org> on 2015/09/11 06:07:46 UTC
[jira] [Resolved] (SINGA-66) Fix bugs in Worker::RunOneBatch
function and ClusterProto
[ https://issues.apache.org/jira/browse/SINGA-66?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
wangwei resolved SINGA-66.
--------------------------
Resolution: Fixed
Th next release needs to update the test code to consider the case where the neural net is partitioned among workers.
> Fix bugs in Worker::RunOneBatch function and ClusterProto
> ---------------------------------------------------------
>
> Key: SINGA-66
> URL: https://issues.apache.org/jira/browse/SINGA-66
> Project: Singa
> Issue Type: Bug
> Reporter: wangwei
>
> For synchronous training with multiple workers, only the first worker will do the test. The condition check in Worker::RunOneBatch() should check the test_net_ and valid_net_ (nullptr) before conducting the test.
> The nworker_groups and nserver_groups should have default values.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)