[formalsystem]

The issue here is memory in PyTorch is byte addressable and that's a limitation we can't solve without making a lot more changes to PyTorch. But in your specific case, if you'd like to pack more data into `values` you can use a combination of clever bit shifting, torch.cat and other bit twiddling pytorch like ops to pack more data. It's a trick we use quite heavily in torchao

[Evidlo]

Arent int8s byte-aligned though? I thought this restriction was originally motivated by maintenance overhead of having to support more dtypes.