Related
This is related to the cleanup of the datatype module in #2092.
Feature functionality
Torch has some functions such as torch.is_floating_point that are lacking a heat equivalent, or in this case have a funnily named equivalent (heat.heat_type_is_realfloating).
Similar in numpy, etc.
There are also inconsistencies. For instance, torch.is_complex tests if the datatype is complex, whereas heat.iscomplex is also False if the datatype is complex with zero imaginary part.
The first step is to collect all of these functions. Please post your favorite one below and when somebody gets around to addressing this issue, they will be added. Please post a link to the function that you need to we can make sure the implementation is doing what you want.
For now, I can think of:
Related
This is related to the cleanup of the datatype module in #2092.
Feature functionality
Torch has some functions such as
torch.is_floating_pointthat are lacking a heat equivalent, or in this case have a funnily named equivalent (heat.heat_type_is_realfloating).Similar in numpy, etc.
There are also inconsistencies. For instance,
torch.is_complextests if the datatype is complex, whereasheat.iscomplexis alsoFalseif the datatype is complex with zero imaginary part.The first step is to collect all of these functions. Please post your favorite one below and when somebody gets around to addressing this issue, they will be added. Please post a link to the function that you need to we can make sure the implementation is doing what you want.
For now, I can think of:
heat.is_floating_point: alias toheat.heat_type_is_realfloatingheat.is_complex: decide how to reconcile with torch version first.heat.isscalaras innumpy.isscalar