[Spec issue] Rounding or truncation behavior for Cast operator is not specified in the spec #3876
Labels
bug
operator
Issues related to ONNX operators
spec clarification
Clarification of the ONNX spec needed
Cast
operator is used from f32 to bfloat16.Cast
operator on documentation page.Cast
operator performs truncation, not rounding.Suggesting to revise documentation:
Cast
operator for floating-point conversions.Cast
from f32 to bfloat16.Repro for rounding
rounding.zip
The text was updated successfully, but these errors were encountered: