slyubomirsky opened a new pull request #7006:
URL: https://github.com/apache/tvm/pull/7006


   This PR is an attempt to support the 
[`TimeDistributed`](https://keras.io/api/layers/recurrent_layers/time_distributed/)
 layer in the Keras frontend. Because `TimeDistributed` takes a layer as an 
argument, I had to modify the existing conversion functions to take parameters 
instead of making assumptions about the Keras model or `etab` data structure. 
Perhaps it might be better not to attempt to reuse the existing conversion 
functions and special-case the different possible arguments to 
`TimeDistributed` -- I would be glad to take any suggestions on this.
   
   I would also be interested to know what you might advise as to further test 
cases for `TimeDistributed`; I went by the example from the documentation and 
another use-case I found that used `TimeDistributed` with `Dense`.
   
   @jroesch I am not entirely sure who would be the best person to review a 
change to the Keras frontend and would appreciate any advice


----------------------------------------------------------------
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:
[email protected]


Reply via email to