LeNetConfig¶

class lucid.models.LeNetConfig(conv_layers: list[dict[str, int]], clf_layers: list[int], clf_in_features: int, base_activation: Type[lucid.nn.module.Module] = <class 'lucid.nn.modules.activation.Tanh'>)¶

LeNetConfig stores the architectural choices used by lucid.models.LeNet. It defines the two convolution stages, the classifier widths, the flattened input size for the first linear layer, and the activation module class used between layers.

Class Signature¶

@dataclass
class LeNetConfig:
    conv_layers: list[dict[str, int]]
    clf_layers: list[int]
    clf_in_features: int
    base_activation: Type[nn.Module] = nn.Tanh

Parameters¶

conv_layers (list[dict[str, int]]): Two convolution layer configs. Each entry must define out_channels.
clf_layers (list[int]): Output widths of the classifier linear layers.
clf_in_features (int): Flattened feature size consumed by the first classifier layer.
base_activation (Type[nn.Module]): Activation module class inserted after each convolution and between classifier layers.

Validation¶

conv_layers must contain exactly 2 entries.
Each convolution config must include out_channels.
clf_layers must not be empty.
clf_in_features must be greater than 0.

Usage¶

import lucid.models as models
import lucid.nn as nn

config = models.LeNetConfig(
    conv_layers=[
        {"out_channels": 6},
        {"out_channels": 16},
    ],
    clf_layers=[120, 84, 10],
    clf_in_features=16 * 5 * 5,
    base_activation=nn.Tanh,
)

model = models.LeNet(config)