tree 4cddda41f294d2ca26736bf128e9426973d961ff
parent 2fe90e4d47551d0b886166983b57d82bfc17d0de
author Aaron Gokaslan <aaronGokaslan@gmail.com> 1704685024 +0000
committer PyTorch MergeBot <pytorchmergebot@users.noreply.github.com> 1704685027 +0000

Add float16 support to CUDA logaddexp2 (#116948)

float16 is already supported on CPU for this op and on gpu for `logaddexp` so let's expand support to the function with the base2 variant as well.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/116948
Approved by: https://github.com/lezcano