class

LeNet

extendsPretrainedModelBackboneMixin

LeNet(config: LeNetConfig)

source edit

LeNet-5 feature extractor (no fully-connected head).

Implements the canonical convolutional trunk of the original LeNet-5 network from LeCun et al., "Gradient-Based Learning Applied to Document Recognition", Proc. IEEE 1998: three convolutional layers (C1, C3, C5) interleaved with two sub-sampling layers (S2, S4). The receptive field grows from a single $5\times5$ patch in C1 to the entire $32\times32$ input by C5 — so for the canonical input size the C5 output collapses to a $1\times1$ spatial map of 120 channels, acting as a fully-connected layer in conv form. Used as the canonical "Hello World" of convolutional networks and as the smallest paper-cited baseline in the model zoo.

Parameters

configLeNetConfig

Frozen architecture spec. Use lenet_5 for the paper-cited tanh + average-pool variant; pass a custom config to switch to modern activation="relu" / pooling="max" or to accept RGB input via in_channels=3.

Attributes

configLeNetConfig

Stored copy of the config that built this model.

featuresnn.Sequential

The C1-S2-C3-S4-C5 convolutional stack — see _build_features for the exact layer chain.

feature_infolist[FeatureInfo]

Per-stage descriptor (channels + reduction factor) exposed via BackboneMixin for downstream decoder modules.

Notes

From LeCun et al., "Gradient-Based Learning Applied to Document Recognition", Proc. IEEE 86(11):2278–2324, 1998, Figure 2. Each convolution computes

h_{i,j}^{(k)} = \phi\!\left( \sum_{c}\sum_{u,v} W_{u,v,c}^{(k)} \, x_{i+u,\,j+v,\,c} + b^{(k)} \right),

where $\phi$ is $\tanh$ in the paper. The original sub-sampling layer was a trainable $2\times2$ average pool with learnable scale and bias; this implementation uses standard parameter-free lucid.nn.AvgPool2d for compatibility with modern training recipes. Total parameter count is approximately 60 k — small enough to train to convergence on MNIST in minutes on a CPU.

Examples

Run a single forward pass on a batch of MNIST-shaped inputs:
>>> import lucid
>>> from lucid.models.vision.lenet import lenet_5
>>> backbone = lenet_5()
>>> x = lucid.randn(8, 1, 32, 32)
>>> out = backbone(x)
>>> out.last_hidden_state.shape   # (B, 120, 1, 1)
(8, 120, 1, 1)

Used by 2

Constructors

dunder

init

→None

__init__(config: LeNetConfig)

source edit

Configure a DataLoader; see the class docstring for parameter semantics.

Raises

ValueError

On any of the above mutual-exclusion / range violations.

Notes

sampler / batch_sampler / shuffle / batch_size / drop_last interact: passing batch_sampler precludes the other four; passing sampler precludes shuffle. When no sampler is supplied, a SequentialSampler (shuffle=False) or RandomSampler (shuffle=True) is constructed automatically. persistent_workers requires num_workers > 0.

Properties

prop

feature_info

→list[FeatureInfo]

feature_info: list[FeatureInfo]

source edit

Instance methods

forward

→BaseModelOutput

forward(x: Tensor)

source edit

forward_features

→Tensor

forward_features(x: Tensor)

source edit

Run a single forward pass on a batch of MNIST-shaped inputs: >>> import lucid >>> from lucid.models.vision.lenet import lenet_5 >>> backbone = lenet_5() >>> x = lucid.randn(8, 1, 32, 32) >>> out = backbone(x) >>> out.last_hidden_state.shape # (B, 120, 1, 1) (8, 120, 1, 1)