coatnet_3

lucid.models.coatnet_3(num_classes: int = 1000, **kwargs) CoAtNet

The coatnet_3 function constructs the CoAtNet-3 preset. This preset uses the default CoAtNetConfig stage layout for the 3 variant.

Total Parameters: 157,790,656

Function Signature

@register_model
def coatnet_3(num_classes: int = 1000, **kwargs) -> CoAtNet

Parameters

  • num_classes (int, optional): Number of output classes for classification. Default is 1000.

  • kwargs (dict, optional): Additional keyword arguments forwarded to CoAtNetConfig, excluding the preset img_size, in_channels, num_blocks, and channels fields.

Returns

  • CoAtNet: A CoAtNet model instance constructed from the 3 preset config.