sk_resnet_18_cls

→SKNetForImageClassification

sk_resnet_18_cls(pretrained: bool | str = False, weights: SKResNet18Weights | None = None, overrides: object = {})

source edit

SK-ResNet-18 image classifier (backbone + GAP + linear head).

Builds an SKNetForImageClassification with the SK-ResNet-18 backbone (basic blocks stacked [2, 2, 2, 2]; the first $3 \times 3$ of each block is a Selective Kernel unit, the second a plain conv) followed by global average pooling and a linear projection to config.num_classes. Approximately 11.96M parameters.

Model Size

Parameters

pretrainedbool or str= False

Pretrained-weight selector. False → random init; True → the DEFAULT tag (SKResNet18Weights.RA_IN1K); a tag string (e.g. "RA_IN1K") → that specific checkpoint. Mutually exclusive with weights (which wins if both are given).

weights(SKResNet18Weights, optional, keyword - only)= None

Explicit weights enum member, e.g. SKResNet18Weights.RA_IN1K. Takes precedence over pretrained.

**overridesobject= {}

Keyword overrides forwarded into SKNetConfig.

Returns

SKNetForImageClassification

Classifier with the SK-ResNet-18 configuration applied (or with overrides merged on top of it), optionally initialised from pretrained weights.

Notes

See Li et al., "Selective Kernel Networks", CVPR 2019 (arXiv:1903.06586). Pretrained weights are converted from timm's skresnet18.ra_in1k and hosted on the Hugging Face Hub under lucid-dl/sk-resnet-18.

Examples

>>> import lucid
>>> from lucid.models.vision.sknet import sk_resnet_18_cls
>>> model = sk_resnet_18_cls(num_classes=10)
>>> x = lucid.randn(2, 3, 224, 224)
>>> out = model(x)
>>> out.logits.shape
(2, 10)
Load ImageNet-pretrained weights:
>>> model = sk_resnet_18_cls(pretrained=True)            # DEFAULT tag
>>> from lucid.models.vision.sknet import SKResNet18Weights
>>> model = sk_resnet_18_cls(weights=SKResNet18Weights.RA_IN1K)

Used by 1

lucid.models.vision.sknet

>>> import lucid >>> from lucid.models.vision.sknet import sk_resnet_18_cls >>> model = sk_resnet_18_cls(num_classes=10) >>> x = lucid.randn(2, 3, 224, 224) >>> out = model(x) >>> out.logits.shape (2, 10) Load ImageNet-pretrained weights: >>> model = sk_resnet_18_cls(pretrained=True) # DEFAULT tag >>> from lucid.models.vision.sknet import SKResNet18Weights >>> model = sk_resnet_18_cls(weights=SKResNet18Weights.RA_IN1K)