mirrors/pytorch-image-models
1
0
Fork 0
mirror of https://github.com/huggingface/pytorch-image-models.git synced 2025-06-03 15:01:08 +08:00
Code Issues Projects Releases Wiki Activity

add rdnet

Browse Source
This commit is contained in:
dong-hyun 2024-08-01 14:54:29 +09:00
parent 025259024d
commit 8248122f82
1 changed files with 0 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
timm/models/rdnet.py
View File

@ -143,7 +143,6 @@ class RDNet(nn.Module):
num_init_features: int = 64, num_init_features: int = 64,
head_init_scale: float = 1., head_init_scale: float = 1.,
head_norm_first: bool = False, head_norm_first: bool = False,
head_hidden_size: Optional[int] = None,
conv_bias: bool = True, conv_bias: bool = True,
act_layer: Union[str, Callable] = 'gelu', act_layer: Union[str, Callable] = 'gelu',
norm_layer: str = "layernorm2d", norm_layer: str = "layernorm2d",
@ -167,7 +166,6 @@ class RDNet(nn.Module):
num_init_features: Number of features of stem. num_init_features: Number of features of stem.
head_init_scale: Init scaling value for classifier weights and biases. head_init_scale: Init scaling value for classifier weights and biases.
head_norm_first: Apply normalization before global pool + head. head_norm_first: Apply normalization before global pool + head.
head_hidden_size: Size of MLP hidden layer in head if not None and head_norm_first == False.
conv_bias: Use bias layers w/ all convolutions. conv_bias: Use bias layers w/ all convolutions.
act_layer: Activation layer type. act_layer: Activation layer type.
norm_layer: Normalization layer type. norm_layer: Normalization layer type.
@ -257,7 +255,6 @@ class RDNet(nn.Module):
# if head_norm_first == true, norm -> global pool -> fc ordering, like most other nets # if head_norm_first == true, norm -> global pool -> fc ordering, like most other nets
# otherwise pool -> norm -> fc, the default RDNet ordering (pretrained NV weights) # otherwise pool -> norm -> fc, the default RDNet ordering (pretrained NV weights)
if head_norm_first: if head_norm_first:
assert not head_hidden_size
self.norm_pre = norm_layer(self.num_features) self.norm_pre = norm_layer(self.num_features)
self.head = ClassifierHead( self.head = ClassifierHead(
self.num_features, self.num_features,
@ -274,7 +271,6 @@ class RDNet(nn.Module):
drop_rate=self.drop_rate, drop_rate=self.drop_rate,
norm_layer=norm_layer, norm_layer=norm_layer,
) )
self.head_hidden_size = self.head.num_features
named_apply(partial(_init_weights, head_init_scale=head_init_scale), self) named_apply(partial(_init_weights, head_init_scale=head_init_scale), self)