Skip to content

Fixup casting issues for weights/bias in fp32 norm layers #2539

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
merged 1 commit into from
Jul 7, 2025
Merged

Fixup casting issues for weights/bias in fp32 norm layers #2539

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
20 changes: 14 additions & 6 deletions timm/layers/norm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,7 +104,9 @@ def __init__(
super().__init__(num_channels, eps=eps, elementwise_affine=affine, **kwargs)

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = F.layer_norm(x.float(), self.normalized_shape, self.weight, self.bias, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
bias = self.bias.float() if self.bias is not None else None
x = F.layer_norm(x.float(), self.normalized_shape, weight, bias, self.eps).to(x.dtype)
return x


Expand Down Expand Up @@ -146,7 +148,9 @@ def __init__(

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = x.permute(0, 2, 3, 1)
x = F.layer_norm(x.float(), self.normalized_shape, self.weight, self.bias, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
bias = self.bias.float() if self.bias is not None else None
x = F.layer_norm(x.float(), self.normalized_shape, weight, bias, self.eps).to(x.dtype)
x = x.permute(0, 3, 1, 2)
return x

Expand Down Expand Up @@ -282,7 +286,8 @@ def reset_parameters(self) -> None:
nn.init.ones_(self.weight)

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = rms_norm(x.float(), self.normalized_shape, self.weight, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
x = rms_norm(x.float(), self.normalized_shape, weight, self.eps).to(x.dtype)
return x


Expand Down Expand Up @@ -381,7 +386,8 @@ def reset_parameters(self) -> None:
nn.init.ones_(self.weight)

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = rms_norm2d(x.float(), self.normalized_shape, self.weight, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
x = rms_norm2d(x.float(), self.normalized_shape, weight, self.eps).to(x.dtype)
return x


Expand Down Expand Up @@ -470,7 +476,8 @@ def reset_parameters(self) -> None:
nn.init.ones_(self.weight)

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = simple_norm(x.float(), self.normalized_shape, self.weight, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
x = simple_norm(x.float(), self.normalized_shape, weight, self.eps).to(x.dtype)
return x


Expand Down Expand Up @@ -562,6 +569,7 @@ def reset_parameters(self) -> None:

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = x.permute(0, 2, 3, 1)
x = simple_norm(x.float(), self.normalized_shape, self.weight, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
x = simple_norm(x.float(), self.normalized_shape, weight, self.eps).to(x.dtype)
x = x.permute(0, 3, 1, 2)
return x
14 changes: 10 additions & 4 deletions timm/layers/norm_act.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -482,7 +482,9 @@ def __init__(
self.act = _create_act(act_layer, act_kwargs=act_kwargs, inplace=inplace, apply_act=apply_act)

def forward(self, x):
x = F.layer_norm(x.float(), self.normalized_shape, self.weight, self.bias, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
bias = self.bias.float() if self.bias is not None else None
x = F.layer_norm(x.float(), self.normalized_shape, weight, bias, self.eps).to(x.dtype)
x = self.drop(x)
x = self.act(x)
return x
Expand Down Expand Up @@ -540,7 +542,9 @@ def __init__(

def forward(self, x):
x = x.permute(0, 2, 3, 1)
x = F.layer_norm(x.float(), self.normalized_shape, self.weight, self.bias, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
bias = self.bias.float() if self.bias is not None else None
x = F.layer_norm(x.float(), self.normalized_shape, weight, bias, self.eps).to(x.dtype)
x = x.permute(0, 3, 1, 2)
x = self.drop(x)
x = self.act(x)
Expand Down Expand Up @@ -605,7 +609,8 @@ def __init__(
self.act = _create_act(act_layer, act_kwargs=act_kwargs, inplace=inplace, apply_act=apply_act)

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = rms_norm(x.float(), self.normalized_shape, self.weight, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
x = rms_norm(x.float(), self.normalized_shape, weight, self.eps).to(x.dtype)
x = self.drop(x)
x = self.act(x)
return x
Expand Down Expand Up @@ -667,7 +672,8 @@ def __init__(
self.act = _create_act(act_layer, act_kwargs=act_kwargs, inplace=inplace, apply_act=apply_act)

def forward(self, x: torch.Tensor) -> torch.Tensor:
x = rms_norm2d(x.float(), self.normalized_shape, self.weight, self.eps).to(x.dtype)
weight = self.weight.float() if self.weight is not None else None
x = rms_norm2d(x.float(), self.normalized_shape, weight, self.eps).to(x.dtype)
x = self.drop(x)
x = self.act(x)
return x
19 changes: 16 additions & 3 deletions timm/models/efficientnet.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2881,22 +2881,35 @@ def test_efficientnet(pretrained=False, **kwargs) -> EfficientNet:

@register_model
def test_efficientnet_gn(pretrained=False, **kwargs) -> EfficientNet:

model = _gen_test_efficientnet(
'test_efficientnet_gn', pretrained=pretrained, norm_layer=partial(GroupNormAct, group_size=8), **kwargs)
'test_efficientnet_gn',
pretrained=pretrained,
norm_layer=kwargs.pop('norm_layer', partial(GroupNormAct, group_size=8)),
**kwargs
)
return model


@register_model
def test_efficientnet_ln(pretrained=False, **kwargs) -> EfficientNet:
model = _gen_test_efficientnet(
'test_efficientnet_ln', pretrained=pretrained, norm_layer=LayerNormAct2d, **kwargs)
'test_efficientnet_ln',
pretrained=pretrained,
norm_layer=kwargs.pop('norm_layer', LayerNormAct2d),
**kwargs
)
return model


@register_model
def test_efficientnet_evos(pretrained=False, **kwargs) -> EfficientNet:
model = _gen_test_efficientnet(
'test_efficientnet_evos', pretrained=pretrained, norm_layer=partial(EvoNorm2dS0, group_size=8), **kwargs)
'test_efficientnet_evos',
pretrained=pretrained,
norm_layer=kwargs.pop('norm_layer', partial(EvoNorm2dS0, group_size=8)),
**kwargs
)
return model


Expand Down