Skip to content

fix tests when scalar is bfloat16, half

Reference issue

What does this implement/fix?

Our funny 16 bit float types don't support constexpr.

Additional information

Edited by Charles Schlosser

Merge request reports

Loading