[
https://issues.apache.org/jira/browse/SINGA-371?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16533231#comment-16533231
]
ASF subversion and git services commented on SINGA-371:
-------------------------------------------------------
Commit 82ef41799ead7c6fd13746f4b155c1d98c14e6ae in incubator-singa's branch
refs/heads/master from xuewanqi
[ https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git;h=82ef417 ]
SINGA-371 Implement functional operations in c++ for autograd
- make convolution operation stateless
- implement new conv2d layer linear layer by calling corresponding operations
- new developed layers have passed test in
network(/example/autograd/mnist_cnn.py)
> Implement functional operations in c++ for autograd
> ---------------------------------------------------
>
> Key: SINGA-371
> URL: https://issues.apache.org/jira/browse/SINGA-371
> Project: Singa
> Issue Type: New Feature
> Reporter: Xue Wanqi
> Priority: Major
>
> In former implementation, some autograd operations do calculation by invoking
> layer, making the codes of these operation complex.
> To simplified this, corresponding functional operations need to be realized.
> Then these PyOperations can do calculation by calling these function instead
> of layer.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
