maxvit_tiny¶

lucid.models.maxvit_tiny(in_channels: int = 3, num_classes: int = 1000, **kwargs) → MaxViT¶

The maxvit_tiny function constructs the tiny MaxViT preset.

Function Signature¶

@register_model
def maxvit_tiny(in_channels: int = 3, num_classes: int = 1000, **kwargs) -> MaxViT

Parameters¶

in_channels (int, optional): Number of input channels. Default is 3.
num_classes (int, optional): Number of output classes for classification. Default is 1000.
kwargs (dict, optional): Additional keyword arguments forwarded to MaxViTConfig, excluding the preset depths, channels, and embed_dim fields.

Returns¶

MaxViT: A MaxViT instance constructed from the tiny preset config.

Preset Config¶

depths: (2, 2, 5, 2)
channels: (64, 128, 256, 512)
embed_dim: 64