pvt_medium

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

The pvt_medium function constructs the PVT-Medium preset. This preset uses the default PVTConfig stage layout for the medium variant.

Total Parameters: 41,492,648

Function Signature

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

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 PVTConfig, excluding the preset patch_size, embed_dims, num_heads, mlp_ratios, qkv_bias, norm_layer, depths, and sr_ratios fields.

Returns

  • PVT: A PVT model instance constructed from the medium preset config.