sk_resnext_50_32x4d

lucid.models.sk_resnext_50_32x4d(num_classes: int = 1000, **kwargs) → SKNet

The sk_resnext_50_32x4d function constructs an SK-ResNeXt-50 (32x4d) model. This preset uses SKNetConfig(block=”bottleneck”, layers=[3, 4, 6, 3], cardinality=32, base_width=4).

Total Parameters: 29,274,760

Function Signature

@register_model
def sk_resnext_50_32x4d(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, layers, cardinality, and base_width fields.

Returns

  • SKNet: An SK-ResNeXt-50 (32x4d) model instance with the preset (block=”bottleneck”, layers=[3, 4, 6, 3], cardinality=32, base_width=4).