bfloat16 grads are not supported

[kurumuz]

Is there any plans to support models/grads with bfloat16 type? Bfloat gained quite the popularity lately as every ampere GPU supports the type, and eliminates the need for loss scaling compared to float16.
This is what I get when I try to initialize bnb.AdamW with a bfloat16 casted model:
ValueError: Gradient+optimizer bit data type combination not supported: grad torch.bfloat16, optimizer torch.uint8