You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@singa.apache.org by "wentong (JIRA)" <ji...@apache.org> on 2018/05/14 06:49:00 UTC

[jira] [Created] (SINGA-363) Add DenseNet for ImageNet classification

wentong created SINGA-363:
-----------------------------

             Summary: Add DenseNet for ImageNet classification
                 Key: SINGA-363
                 URL: https://issues.apache.org/jira/browse/SINGA-363
             Project: Singa
          Issue Type: New Feature
            Reporter: wentong


The DenseNet[ |http://arxiv.org/abs/1602.07261]models are a series of deep CNN models with high performance.
We convert the parameters pre-trained from Pytorch into pickle dictionary and load them into the net created using SINGA FeedForwardNet.

In the meantime, some other minor bugs in "examples" folder would be fixed in order to run in Python 3 environment.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)