cconvey commented on a change in pull request #9186:
URL: https://github.com/apache/tvm/pull/9186#discussion_r722481055
##########
File path: tests/python/frontend/onnx/test_forward.py
##########
@@ -1282,6 +1282,47 @@ def verify_batch_matmul(a_shape, b_shape, out_shape,
convert_config=None):
)
[email protected]_targets
+def test_matmulinteger16(target, dev):
+ def verify_matmulinteger16(a_shape, b_shape, out_shape):
+ a_dtype = "int16"
+ b_dtype = "int16"
+ low = -10
+ high = 10
Review comment:
Or maybe use `np.iinfo(np.int16).min` and `...max` as the bounds?
If the goal is to increase the random discovery of latent bugs, and we don't
know the _specific_ values at which corner-cases happen, perhaps it makes sense
to at least range over all possible values.
I'm not sure what the best strategy is here, but I'm wondering if this is a
step in the right direction.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
