tqchen commented on PR #19750: URL: https://github.com/apache/tvm/pull/19750#issuecomment-4753551348
Looking more broadly, it appears that we have a series of fixes of nan behavior all via explicit nan checking steps, i think they are not the directions we want to go toward, e.g. paying extra computing cost for a corner case that may only appear in fuzzer setting. Instead, i think it is better to think about ways to define the right most efficient way in implementing relax.relu -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
