DenseNetConfig¶

class lucid.models.DenseNetConfig(block_config: tuple[int, int, int, int] | list[int], growth_rate: int = 32, num_init_features: int = 64, num_classes: int = 1000, in_channels: int = 3, bottleneck: int = 4, compression: float = 0.5)¶

DenseNetConfig stores the architectural settings used by lucid.models.DenseNet. It defines the dense block depths, growth rate, initial stem width, classifier size, and optional DenseNet-specific bottleneck and transition-compression hyperparameters.

Class Signature¶

@dataclass
class DenseNetConfig:
    block_config: tuple[int, int, int, int] | list[int]
    growth_rate: int = 32
    num_init_features: int = 64
    num_classes: int = 1000
    in_channels: int = 3
    bottleneck: int = 4
    compression: float = 0.5

Parameters¶

block_config (tuple[int, int, int, int] | list[int]): Number of dense layers in each of the four dense blocks.
growth_rate (int): Number of output channels added by each dense layer.
num_init_features (int): Number of output channels produced by the initial convolution stem.
num_classes (int): Number of output classes.
in_channels (int): Number of channels in the input image tensor.
bottleneck (int): Expansion factor used by the 1x1 bottleneck convolution inside each dense layer.
compression (float): Transition-layer compression ratio applied after each dense block except the last.

Validation¶

block_config must contain exactly four positive integers.
growth_rate, num_init_features, num_classes, in_channels, and bottleneck must be greater than 0.
compression must be in the range (0, 1].

Usage¶

import lucid.models as models

config = models.DenseNetConfig(
    block_config=[6, 12, 32, 32],
    growth_rate=32,
    num_init_features=64,
    num_classes=10,
    in_channels=1,
    compression=0.5,
)

model = models.DenseNet(config)