You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@systemds.apache.org by GitBox <gi...@apache.org> on 2021/01/28 12:37:27 UTC
[GitHub] [systemds] sebwrede opened a new pull request #1174: [SYSTEMDS-2812] Conv1D
sebwrede opened a new pull request #1174:
URL: https://github.com/apache/systemds/pull/1174
Initial implementation of Conv1D using the existing Conv2D implementation.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [systemds] asfgit closed pull request #1174: [SYSTEMDS-2812] Conv1D
Posted by GitBox <gi...@apache.org>.
asfgit closed pull request #1174:
URL: https://github.com/apache/systemds/pull/1174
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [systemds] sebwrede edited a comment on pull request #1174: [SYSTEMDS-2812] Conv1D
Posted by GitBox <gi...@apache.org>.
sebwrede edited a comment on pull request #1174:
URL: https://github.com/apache/systemds/pull/1174#issuecomment-769676594
> do you think that a dedicated internal operation for this would be good / maybe even going to a ConvND ?
Yes, this could be an additional feature to implement. The goal of this PR is to have a basic implementation of Conv1D. Later we can make a builtin version of it and see how efficient that is compared to the DML-version.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [systemds] Baunsgaard commented on pull request #1174: [SYSTEMDS-2812] Conv1D
Posted by GitBox <gi...@apache.org>.
Baunsgaard commented on pull request #1174:
URL: https://github.com/apache/systemds/pull/1174#issuecomment-769671586
LGTM,
do you think that a dedicated internal operation for this would be good / maybe even going to a ConvND ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [systemds] mboehm7 commented on pull request #1174: [SYSTEMDS-2812] Conv1D
Posted by GitBox <gi...@apache.org>.
mboehm7 commented on pull request #1174:
URL: https://github.com/apache/systemds/pull/1174#issuecomment-770297542
LGTM - thanks for the extension @sebwrede, this is great. I think it's the right step to first call the existing conv2d (for which internal Java/C++/GPU operators with dedicated implementations for combinations of dense/sparse inputs exist). If we measure substantial overhead of conv1d via conv2d then we could go ahead and add the different kernels.
Btw, I see that the conv2d / pooling layers might be confusing as _builtin refers to wrappers around natively supported operators, while in scripts/builtin we have dml-bodied functions. Perhaps we should remove the script-based conv2d and pooling layers altogether.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [systemds] sebwrede commented on pull request #1174: [SYSTEMDS-2812] Conv1D
Posted by GitBox <gi...@apache.org>.
sebwrede commented on pull request #1174:
URL: https://github.com/apache/systemds/pull/1174#issuecomment-769676594
> do you think that a dedicated internal operation for this would be good / maybe even going to a ConvND ?
Yes, this could be an additional feature to implement. The goal of this PR is to have a basic implementation of Conv1D. Later we can make a builtin version of it and see how efficient that is compared to the DML-version.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org