Skip to content
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

Fixes backward definition in triton tutorial #3282

Merged
merged 2 commits into from
Mar 7, 2025
Merged

Fixes backward definition in triton tutorial #3282

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c7c94aa
Fixes backward definition in triton tutorial
thomasjpfan Mar 2, 2025
667a1c1
Merge branch 'main' into fix_backword_definition
svekars Mar 7, 2025
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
8 changes: 4 additions & 4 deletions recipes_source/torch_compile_user_defined_triton_kernel_tutorial.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -257,9 +257,9 @@ def sin_triton(x):
# Prefer this to using ``torch.autograd.Function`` (which has various composability footguns
# with ``torch.compile``).

def backward(ctx, grad_output):
def backward(ctx, grad):
x, = ctx.saved_tensors
return grad_input * x.cos()
return grad * x.cos()

def setup_context(ctx, inputs, output):
x, = inputs
Expand Down Expand Up @@ -293,9 +293,9 @@ def mycos(x: torch.Tensor) -> torch.Tensor:
wrap_triton(cos_kernel)[(n_elements,)](x, out, n_elements, BLOCK_SIZE=4)
return out

def backward(ctx, grad_output):
def backward(ctx, grad):
x, = ctx.saved_tensors
return grad_input * mycos(x)
return grad * mycos(x)

def setup_context(ctx, inputs, output):
x, = inputs
Expand Down
Loading