-
Notifications
You must be signed in to change notification settings - Fork 82
Open
Description
When I use float16 to train model, it says:
RuntimeError: Expected a.scalar_type() == torch::kBFloat16 to be true, but got false..
Seems this is related to
grouped_gemm/csrc/grouped_gemm.cu
Line 364 in f4c08bc
| TORCH_CHECK(a.scalar_type() == torch::kBFloat16); |
Look forward to your reply sincerely.
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels