comaniac commented on a change in pull request #6109:
URL: https://github.com/apache/incubator-tvm/pull/6109#discussion_r459616949
##########
File path: tests/python/contrib/test_arm_compute_lib/test_conv2d.py
##########
@@ -127,51 +241,31 @@ def test_conv2d():
device = Device()
np.random.seed(0)
+ r = random.Random(0)
Review comment:
Although it is not the best practice neither to randomly generate input
tensors as you pointed out, it is relatively acceptable in general. The reasons
are:
1. The module we are testing does not highly depend on the input values.
2. It is tedious to put a constant large tensor (e.g., 1, 3, 224, 224 = 150k
numbers) in the code base.
3. Sometimes people use `np.ones` or `np.zeros` to be inputs for the above
two reasons, but these tensors are too far away from real world applications
and may not fit our testing purposes.
----------------------------------------------------------------
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]
