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 2018/06/04 11:14:40 UTC

[GitHub] lynneChan opened a new issue #11137: memory cost problem for different label width when using ssd example

lynneChan opened a new issue #11137: memory cost problem for different label width when using ssd example
URL: https://github.com/apache/incubator-mxnet/issues/11137
 
 
    I am working on training face detectors using the code in example/ssd on the [WIDER](http://mmlab.ie.cuhk.edu.hk/projects/WIDERFace/) dataset, and i found some strange things.
   If i use the all the labels and images for training(label width = 4784), the memory cost will be about 10367MB, but if i filter images and labels which number of labels is bigger than 100(label width = 602), the memory cost will be about 6991MB. 
   
   Does anyone knows why mxnet works this way ? Should this be normal or some unknow issues ?
   
   I run my code on Ubuntu 14.04 with GPU Tesla K40c(11439MB), data_shape = 640, network = mobilenet, batch_size = 12.

----------------------------------------------------------------
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