sk_resnet_50

→SKNet

sk_resnet_50(pretrained: bool = False, overrides: object = {})

source edit

SK-ResNet-50 feature-extracting backbone (no classification head).

Builds an SKNet with ResNet-50 bottleneck topology (_SelectiveKernelBottleneck blocks stacked [3, 4, 6, 3]). The central $3 \times 3$ of every bottleneck is replaced by a two-branch Selective Kernel unit with split_input=True (each branch receives half the channels — matching timm's skresnet50 layout). Approximately 25.8M parameters.

Model Size

Parameters

pretrainedbool= False

Reserved for future pretrained-weight loading. Currently ignored.

**overridesobject= {}

Keyword overrides forwarded into SKNetConfig.

Returns

SKNet

Backbone with the SK-ResNet-50 configuration applied (or with overrides merged on top of it).

Notes

See Li et al., "Selective Kernel Networks", CVPR 2019 (arXiv:1903.06586), Table 1. Final-stage output is 2048 channels.

Examples

>>> import lucid
>>> from lucid.models.vision.sknet import sk_resnet_50
>>> model = sk_resnet_50()
>>> x = lucid.randn(1, 3, 224, 224)
>>> out = model(x)
>>> out.last_hidden_state.shape
(1, 2048, 7, 7)

Used by 1

lucid.models.vision.sknet

>>> import lucid >>> from lucid.models.vision.sknet import sk_resnet_50 >>> model = sk_resnet_50() >>> x = lucid.randn(1, 3, 224, 224) >>> out = model(x) >>> out.last_hidden_state.shape (1, 2048, 7, 7)