AndrewZhaoLuo commented on a change in pull request #7928:
URL: https://github.com/apache/tvm/pull/7928#discussion_r624725442
##########
File path: tests/python/relay/test_op_level5.py
##########
@@ -671,10 +671,10 @@ def verify_roi_align(
print("test on", target)
intrp1 = relay.create_executor("graph", device=dev, target=target)
op_res1 = intrp1.evaluate(func)(np_data, np_rois)
- tvm.testing.assert_allclose(op_res1.asnumpy(), ref_res, rtol=1e-4)
+ tvm.testing.assert_allclose(op_res1.asnumpy(), ref_res, rtol=2e-4)
intrp2 = relay.create_executor("debug", device=dev, target=target)
op_res2 = intrp2.evaluate(func)(np_data, np_rois)
- tvm.testing.assert_allclose(op_res2.asnumpy(), ref_res, rtol=1e-4)
+ tvm.testing.assert_allclose(op_res2.asnumpy(), ref_res, rtol=2e-4)
Review comment:
Ok it passes still with the original tolerances, it was probably due to
another bug that was fixed a long time ago
--
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]
