access2rohit commented on pull request #19064: URL: https://github.com/apache/incubator-mxnet/pull/19064#issuecomment-758336793
> Do we also need to change the logic of TopKType? > > https://github.com/apache/incubator-mxnet/blob/124d8417984ed9205f972eb1c2cadbf028b94eb3/src/operator/tensor/ordering_op-inl.h#L948-L964 @sxjscience @samskalicky This PR doesn't change output index types but only ensures that internal workspace used for them is int32 for smaller inputs. They are casted to their respective types while returning. Also currently we don't have means to check the input size at type inference pass. Let me know what you guys think ? ---------------------------------------------------------------- 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]
