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 2020/05/20 06:06:44 UTC

[GitHub] [singa] joddiy commented on issue #706: Refactor SONNX

joddiy commented on issue #706:
URL: https://github.com/apache/singa/issues/706#issuecomment-631257280


   1. can you extract the values for the shape from the tensor and pass them to init Reshape?
   Yes, I'm going to do this.
   
   2. running_mean and running_var are not params (not updated via sgd). They are state variables.
   Got it, but I canot see the set_states yet at Layer class.
   
   3. out_channel is required. Rename it to nb_kernels.
   Got it, I think I can parse the out_channels from the ONNX's weight.
   
   4. With SONNXModel, I think we do not need metaclass anymore. You just need to consider the aux_output
   Yes, this part has been done.


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