Closed forFudan closed 1 month ago
@forFudan We could do that, I don't see any problem with it. Initially, I set it as DType.float32
to match PyTorch.
Is there any potential performance cost on this change?
Is there any potential performance cost on this change?
The Float64 takes double memory space for higher precision.
Currently, the default data type of a
NDArray
isDType.float32
. This is somehow different from numpy which uses float64 as default type. For example, numpy.ones().I am wondering whether it would be nice to change the default data type of
NDArray
fromfloat32
tofloat64
also for numojo, so that it is aligned with numpy and is of higher precision.I do not think it is a top priority but we can have an open discussion on this.