mbrookhart commented on pull request #7487: URL: https://github.com/apache/tvm/pull/7487#issuecomment-785376392
My main concern with this, and please correct me if I'm wrong, is concatenating something like shape [1, 3] and shape [1, ?] on axis 1. In that case, I believe this will return [2, 3], and we don't do any runtime checks on the second argument. If the model is wrong, and, say, the input comes in as [1, 2] instead of [1, 3], we'll end up with a runtime segfault in generated code, which is very difficult to debug. ---------------------------------------------------------------- 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]
