When using a pictures as input to the model before and after compiling by TVM, the prediction results are slightly inconsistent. However, it did not cause a classification error
`np.testing.assert_allclose(frame_output, tvm_output, rtol=1e-3, atol=1e-3)` is used as the criterion. Can such inconsistencies indicate a bug? --- [Visit Topic](https://discuss.tvm.apache.org/t/is-the-inconsistency-which-doesnt-cause-the-classification-error-a-bug/9957/1) to respond. You are receiving this because you enabled mailing list mode. To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/0a127dd2511080cf494e79a375f2bbe27160f4c19aa51e0772a33bba08a2365e).
