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

Posted to commits@mxnet.apache.org by GitBox <gi...@apache.org> on 2018/08/26 11:41:09 UTC

[GitHub] juliusshufan opened a new pull request #12362: Removing the re-size for validation data, which breaking the validation accuracy of CIFAR training

juliusshufan opened a new pull request #12362: Removing the re-size for validation data, which breaking the validation accuracy of CIFAR training
URL: https://github.com/apache/incubator-mxnet/pull/12362
 
 
   ## Description ##
   The dataloader for example/image-classification resizes the imagesize of validation dataset to 256 by default, and which breaks the validation accuracy of CIFAR10, this PR trying to fix it.
   Below firgure is the validation accuracy trends without this fix (model is ResNet-50).
   ![image](https://user-images.githubusercontent.com/33112206/44627860-f744b180-a967-11e8-8429-95a49c4f94c0.png)
   
   ## Checklist ##
   ### Changes ###
   example/image-classification/common/data.py
   
   ## Comments ##
   - If this change is a backward incompatible change, why must this change be made.
   - Interesting edge cases to note here
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on 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


With regards,
Apache Git Services