Skip to content

Remove redundant runtime checks for activation recompute #3436

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
ghadiaravi13 wants to merge 1 commit into
base: main
Choose a base branch
from
+0 −12
Draft

Remove redundant runtime checks for activation recompute #3436

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 0 additions & 12 deletions transformer_engine/pytorch/ops/basic/activation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -402,12 +402,6 @@ def fuser_forward(
next_op_input_quantizer: Optional[Quantizer], # pylint: disable=unused-argument
basic_op_kwargs: list[dict[str, Any]], # pylint: disable=unused-argument
) -> tuple[torch.Tensor, Sequence[Sequence[torch.Tensor]]]:
if self.activation_recompute_in_mlp:
raise RuntimeError(
f"{self.__class__.__name__}(activation_recompute_in_mlp=True) requires the "
"fused grouped MLP path."
)

extra_input = basic_op_extra_inputs[0][0]

if torch.is_autocast_enabled():
Expand Down Expand Up @@ -445,12 +439,6 @@ def fuser_backward(
]:
del basic_op_grad_extra_outputs

if self.activation_recompute_in_mlp:
raise RuntimeError(
f"{self.__class__.__name__}(activation_recompute_in_mlp=True) requires the "
"fused grouped MLP path."
)

ctx = basic_op_ctxs[0]
x, scales = ctx.saved_tensors
x = maybe_dequantize(x.contiguous(), ctx.dtype)
Expand Down
Loading