masahi commented on issue #12524: URL: https://github.com/apache/tvm/issues/12524#issuecomment-1224757199
hmm maybe argmax got fixed, but the raw outputs should still be different. First, please run the test in https://github.com/apache/tvm/blob/f64a3bda253f4220d66eeb3348f93f486392cb8e/tests/python/frontend/pytorch/test_forward.py#L809-L814 to make sure that the existing test passes Second, try to find an input that causes the numerical issue and use that in the new test case. The new implementation should pass the test for such input. If it doesn't, you need to figure out what is still off from the PT `log_sigmoid`. > how does onnx frontend handle it? I think PT's ONNX exporter decomposes `log_sigmoid(x)` into `log(sigmoid(x))`, so there is nothing we can do about it. -- 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: commits-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
