[GitHub] [incubator-mxnet] ptrendx commented on issue #17164: net.Cast("float16") doesn't work: Check failed: (*in_type)[i] == dtype_param (2 vs. 0) : This layer requires uniform type. Expected 'float32' v.s. given 'float16' at 'gamma'

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

ptrendx commented on issue #17164: net.Cast("float16") doesn't work: Check failed: (*in_type)[i] == dtype_param (2 vs. 0) : This layer requires uniform type. Expected 'float32' v.s. given 'float16' at 'gamma'
URL: https://github.com/apache/incubator-mxnet/issues/17164#issuecomment-570295684
 
 
   This error actually seems to come from the fact that the `SymbolBlock` does not know really what symbols are inside it, and so it casts all of the parameters in the `net.cast('float16')` call. However, BatchNorm layer is slightly special in that it actually requires its parameters `gamma` and `beta` to be `float32` even if input is `float16` in order to not lose precision. So the BatchNorm layer expects `gamma` to be `float32`, but the parameter given to it is in `float16`.
   
   @Rainweic I would recommend looking into AMP for training: https://mxnet.apache.org/api/python/docs/tutorials/performance/backend/amp.html
   
   @zhreshold FYI

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