lfengad commented on a change in pull request #4990: [TF][Relay] BatchNorm
support with run-time mean and variance calculation
URL: https://github.com/apache/incubator-tvm/pull/4990#discussion_r388831831
##########
File path: python/tvm/relay/frontend/tensorflow.py
##########
@@ -887,7 +887,14 @@ def _impl(inputs, attr, params):
if 'U' in attr:
need_cast = True
inputs[0] = _op.cast(inputs[0], dtype=attr['U'].name)
-
+ # Check if mean and variance are empty
+ # If so, replace them with Mean and Variance Ops
+ # For run-time calculation
+ moving_mean_shape = [int(n) for n in inputs[3].type_annotation.shape]
Review comment:
Thank you for the detailed review comments! I have corrected the related
codes in the newest commit. But for the `add CHECK of len(inputs)` issue, does
it mean that to check the number of nodes in `inputs` and to see whether this
number is 5? Also, it seems in python there is no `CHECK` method?
Thank you so much for the help!
----------------------------------------------------------------
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]
With regards,
Apache Git Services
