THIS. OH MY GOSH. THIS. The only issue now is that pytorch has their own "shapes" solution[0], and last I checked, were kinda reserved about participating in t

THIS. OH MY GOSH. THIS. The only issue now is that pytorch has their own "shapes" solution[0], and last I checked, were kinda reserved about participating in the standardization of variadic generics because they don't expect to use it. I really really hope that the ML community comes together to use variadic generics because I believe it will save researchers and devs so many debug cycles (as well as compute resources, tbh). [0] https://pytorch.org/docs/stable/named_tensor.html https://pytorch.org/docs/stable/named_tensor.html