argmin and argmax functions of algorithm.reduction are removed in new version. I don’t see anything about this on change log. Are there alternatives to these in new version? Implementing optimized argmin/argmax functions need some effort; So, I think having them would be nice as they had been implemented in previous versions.
It was removed due to inefficient implementation but was missed in the changelog. We’ll update the changelog and more will land soon that will include tensor type revamp etc.
Ah, ok. Thank you for letting me know!