masahi commented on issue #8679: URL: https://github.com/apache/tvm/issues/8679#issuecomment-894568511
PyTorch resnet 18 is tested on every CI job, https://github.com/apache/tvm/blob/874ea7a81d91857a2495892c598a8b3b87a6da64/tests/python/frontend/pytorch/test_forward.py#L2030. So I don't expect any accuracy difference. Can you try evaluating the model that is not serialized to disk? When PyTorch jit models are serialized, PyTorch erase all type information. This issue has caused some problems for us in the past. -- 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]
