swin_small

lucid.models.swin_small(img_size: int = 224, num_classes: int = 1000, **kwargs) → SwinTransformer

The swin_small function constructs the Swin-Small preset. This preset uses the default SwinTransformerConfig stage layout for the small variant.

Total Parameters: 49,606,258

Function Signature

@register_model
def swin_small(img_size: int = 224, num_classes: int = 1000, **kwargs) -> SwinTransformer

Parameters

  • img_size (int, optional): Input image size. Default is 224.

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

  • kwargs (dict, optional): Additional keyword arguments forwarded to SwinTransformerConfig, excluding the preset embed_dim, depths, and num_heads fields.

Returns

  • SwinTransformer: A Swin Transformer model instance constructed from the small preset config.