You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@singa.apache.org by GitBox <gi...@apache.org> on 2021/05/27 11:21:26 UTC

[GitHub] [singa] lzjpaul opened a new issue #856: Improve CNN training efficiency by moving augmentation outside of training iterations

lzjpaul opened a new issue #856:
URL: https://github.com/apache/singa/issues/856


   For benchmark datasets, e.g., CIFAR-10 and ImageNet, the augmentation method is pre-defined before the model training, so the augmentation does not need to be carried out mini-batch wise. 
   
   Instead, the augmentation can be done once before the training begins, in order to save time.


-- 
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] [singa] lzjpaul closed issue #856: Improve CNN training efficiency by moving augmentation outside of training iterations

Posted by GitBox <gi...@apache.org>.
lzjpaul closed issue #856:
URL: https://github.com/apache/singa/issues/856


   


-- 
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] [singa] lzjpaul commented on issue #856: Improve CNN training efficiency by moving augmentation outside of training iterations

Posted by GitBox <gi...@apache.org>.
lzjpaul commented on issue #856:
URL: https://github.com/apache/singa/issues/856#issuecomment-850766008


   In examples/cifar_distributed_cnn, the data augmentation is moved outside of training iterations for CIFAR-10 dataset


-- 
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] [singa] lzjpaul commented on issue #856: Improve CNN training efficiency by moving augmentation outside of training iterations

Posted by GitBox <gi...@apache.org>.
lzjpaul commented on issue #856:
URL: https://github.com/apache/singa/issues/856#issuecomment-850765968


   In examples/cifar_distributed_cnn, the data augmentation is moved outside of training iterations for CIFAR-10 dataset


-- 
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