[GitHub] [incubator-mxnet] yifeim commented on issue #14888: Memory usage estimator

[GitBox <gi...@apache.org>]

yifeim commented on issue #14888: Memory usage estimator
URL: https://github.com/apache/incubator-mxnet/issues/14888#issuecomment-490365117
 
 
   @TaoLv This feature is more geared towards federated training, when some of the model shapes are inferred from the user-provided dataset. In this case, trying to utilize memory more efficiently, we have to understand how to set other parameters, e.g., batch_size, accordingly. Essentially, we aim to "try a larger input size" inside a script.
   
   An example would be to fit [word_language_model](https://github.com/apache/incubator-mxnet/tree/master/example/gluon/word_language_model) with different vocabulary sizes, where filtering by word frequency is not an option.
   
   Additionally, it would be useful to consider memory planning for sparse data. In this case, we would provide the 'max_num_numzeros' parameter to reduce the uncertainties.

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


With regards,
Apache Git Services