Skip to content

[Bugfix][PyTorch] Support use_input_stats in instance_norm #14963

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
masahi merged 1 commit into from
May 27, 2023
Merged

[Bugfix][PyTorch] Support use_input_stats in instance_norm #14963

Show file tree
Hide file tree
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
17 changes: 17 additions & 0 deletions python/tvm/relay/frontend/pytorch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1385,6 +1385,9 @@ def instance_norm(self, inputs, input_types):
data = inputs[0]
data_type = input_types[0]
channels = self.infer_shape(data)
running_mean = inputs[3]
Copy link

@shingjan shingjan May 26, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this a most recent api change from pytorch or a long ignored params on our side? The former will need us to think about BC.

Copy link
Contributor Author

@liquanfeng liquanfeng May 27, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The latter I think. The "track_running_stats" has been ignored in the unit test.
By the way, there will be another PR later about BC, also on a similar issue.

running_var = inputs[4]
use_input_stats = inputs[5]

if isinstance(inputs[1], _expr.Expr) and isinstance(inputs[2], _expr.Expr):
scale = center = True
Expand All @@ -1401,6 +1404,20 @@ def instance_norm(self, inputs, input_types):
beta = _create_typed_const(np.zeros([int(channels[1])]), data_type)

epsilon = float(inputs[7])

if not use_input_stats:
return _op.nn.batch_norm(
data,
gamma,
beta,
running_mean,
running_var,
axis=1,
epsilon=epsilon,
center=center,
scale=scale,
)[0]

return _op.nn.instance_norm(
data, gamma, beta, axis=1, epsilon=epsilon, center=center, scale=scale
)
Expand Down
2 changes: 2 additions & 0 deletions tests/python/frontend/pytorch/test_forward.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1414,6 +1414,8 @@ def test_forward_instancenorm():
for ins_norm, inp in [
(torch.nn.InstanceNorm2d(16), inp_2d),
(torch.nn.InstanceNorm3d(16), inp_3d),
(torch.nn.InstanceNorm2d(16, track_running_stats=True), inp_2d),
(torch.nn.InstanceNorm3d(16, track_running_stats=True), inp_3d),
]:
verify_model(ins_norm.eval(), input_data=inp)

Expand Down