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

Posted to dev@singa.apache.org by "zhangzhaoqi (Jira)" <ji...@apache.org> on 2020/02/24 10:36:00 UTC

[jira] [Created] (SINGA-507) conv and pooling support same padding mode

zhangzhaoqi created SINGA-507:
---------------------------------

             Summary: conv and pooling support same padding mode
                 Key: SINGA-507
                 URL: https://issues.apache.org/jira/browse/SINGA-507
             Project: Singa
          Issue Type: New Feature
            Reporter: zhangzhaoqi


since onnx requests same padding mode, we need to optimize conv and pooling these 2 operators.

FYI:
{quote}auto_pad must be either NOTSET, SAME_UPPER, SAME_LOWER or VALID. Where default value is NOTSET, which means explicit padding is used. SAME_UPPER or SAME_LOWER mean pad the input so that the output spatial size match the input.In case of odd number add the extra padding at the end for SAME_UPPER and at the beginning for SAME_LOWER. VALID mean no padding.
{quote}
This JR will finish:
 # conv and pooling support SAME_UPPER, SAME_LOWER for both forward and backward, 2d and 1d.
 # For SAME_UPPER, SAME_LOWER, can do shape checking, and if not set shape, can compute the correct shape.
 # Test cases.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)