resnext_50_32x4d

lucid.models.resnext_50_32x4d(num_classes: int = 1000, **kwargs) ResNeXt

The resnext_50_32x4d function constructs a ResNeXt-50 model with 32 groups and a base width of 4.

Total Parameters: 25,028,904

Function Signature

@register_model
def resnext_50_32x4d(num_classes: int = 1000, **kwargs) -> ResNeXt

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 ResNeXtConfig, excluding the preset layers, cardinality, and base_width fields.

Returns

  • ResNeXt: A ResNeXt-50 model instance with the preset (layers=[3, 4, 6, 3], cardinality=32, base_width=4).