Skip to content

Don't set requires_grad in Arm-backend fold_qdq-pass #8027

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

Merged
zingo merged 1 commit into from
Jan 29, 2025
Merged

Don't set requires_grad in Arm-backend fold_qdq-pass #8027

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
15 changes: 0 additions & 15 deletions backends/arm/_passes/fold_qdq_with_annotated_qparams_pass.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,21 +105,6 @@ def fold_and_annotate_arg(
for arg in arg_list:
if not isinstance(arg, Node):
return
"""
Make sure arg has requires_grad set to False
For parameters that are not quantized, sometimes (i.e. convolution)
the Parameter(FakeTensor(...)) has requires_grad set to True, which
causes the retracing of the graph to fail with:
E RuntimeError: isDifferentiableType(variable.scalar_type()) INTERNAL ASSERT FAILED at "/Users/runner/work/pytorch/pytorch/pytorch/torch/csrc/autograd/functions/utils.h":74, please report a bug to PyTorch.
E
E While executing %aten_convolution_default : [num_users=1] = call_function[target=executorch.exir.dialects.edge._ops.aten.convolution.default](args = (%quantized_decomposed_quantize_per_tensor_default, %b__frozen_param0, %p__param_constant1, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
E Original traceback:
E File "/Users/perast01/src/executorch/backends/arm/test/ops/test_conv2d.py", line 110, in forward
E x = conv(x)
"""
if arg.op == "placeholder":
arg.meta["val"].requires_grad = False

arg_quant_params = None
if arg.target == dq_op:
Expand Down
Loading