d-smirnov commented on a change in pull request #6018:
URL: https://github.com/apache/incubator-tvm/pull/6018#discussion_r466594422
##########
File path: python/tvm/relay/frontend/tflite.py
##########
@@ -1089,7 +1093,7 @@ def convert_square(self, op):
return out
- def _convert_elemwise(self, relay_op, op):
+ def _convert_elemwise(self, relay_op, op, use_real_qnn=True):
Review comment:
I might be not correct here, but the whole idea of using same qnn
parameters is about being able to re-use non-quantized version of the
operation. In case of Slice op #6217 there is only a check and the
non-quantized operation is always used. In case of maximum and minimum this is
not possible without changes either in the command operands (the qnn_params
should stripped off) or, alternatively changes in _convert_elemwise in order
explicitly prevent it going via quantized version.
----------------------------------------------------------------
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]
