sk_resnet_50¶
The sk_resnet_50 function constructs a Selective Kernel ResNet-50 model. This preset uses SKNetConfig(block=”bottleneck”, layers=[3, 4, 6, 3]).
Total Parameters: 57,073,368
Function Signature¶
@register_model
def sk_resnet_50(num_classes: int = 1000, **kwargs) -> SKNet
Parameters¶
num_classes (int, optional): Number of output classes for the final fully connected layer. Default is 1000.
kwargs (dict, optional): Additional keyword arguments forwarded to SKNetConfig, excluding the preset block and layers fields.
Returns¶
SKNet: A Selective Kernel ResNet-50 model instance with the preset (block=”bottleneck”, layers=[3, 4, 6, 3]).