masahi commented on pull request #7441: URL: https://github.com/apache/tvm/pull/7441#issuecomment-782546478
I can do the GPU version. It will likely require ir builder. But let me know if you *want* to do GPU as well, you can certainly do it. The idea is identical with CPU version, just using different parallelization. If `unique_with_counts` can be supported by adding another option to `unique`, that sounds good. We shouldn't add `relay.unique_with_counts` or `topi.unique_with_counts`. ---------------------------------------------------------------- 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]
