model_builder module¶
- class model_builder.Identity(features)¶
Bases:
torch.nn.modules.module.Module- Identity network
This network is used to convert last layer of pre-trained network to the identity if feature_extractor flag is set.
- Parameters
nn ([type]) – [description]
- forward(y)¶
Defines the computation performed at every call.
Should be overridden by all subclasses.
Note
Although the recipe for forward pass needs to be defined within this function, one should call the
Moduleinstance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.
- training: bool¶
- class model_builder.build_models(model_name, out_classes, pretrained, requires_grad, in_channels, custom_pretrained=None, feature_extractor: bool = False)¶
Bases:
object- build_densenet()¶
Generates densenet according the parameters initiliazed
- Returns
Densenet model
- Return type
torch.model
- build_efficientnet()¶
Generates efficientnet according the parameters initiliazed
- Returns
efficientnet model
- Return type
torch.model
- build_resnet()¶
Generates resnet according the parameters initiliazed
- Returns
resnet model
- Return type
torch.model
- build_vgg()¶
Builds vgg according the parameters initiliazed
- Returns
vgg model
- Return type
torch.model
- class model_builder.classifier(in_features, out_features)¶
Bases:
torch.nn.modules.module.Module- forward(x, features)¶
Defines the computation performed at every call.
Should be overridden by all subclasses.
Note
Although the recipe for forward pass needs to be defined within this function, one should call the
Moduleinstance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.
- training: bool¶