convnext_small_cls(pretrained: bool | str = False, weights: ConvNeXtSmallWeights | None = None, overrides: object = {})ConvNeXt-Small image classifier (Liu et al., 2022).
Combines the convnext_small backbone (depths=(3, 3, 27, 3),
dims=(96, 192, 384, 768)) with a global average pool +
LayerNorm + linear classification head. ~50M parameters.
Model Size
Parameters
pretrainedbool= FalseIf
True, loads ImageNet-1k pretrained weights when
available. Defaults to False.Explicit weights enum member; takes precedence over
pretrained.**overridesobject= {}Keyword overrides on top of the canonical ConvNeXt-S config.
Returns
ConvNeXtForImageClassificationClassifier whose logits has shape (B, num_classes).
Notes
ConvNeXt-S reaches 83.1% top-1 on ImageNet-1k (Liu et al., 2022, Table 1).
Examples
>>> import lucid
>>> from lucid.models.vision.convnext import convnext_small_cls
>>> model = convnext_small_cls(num_classes=1000)
>>> x = lucid.randn(1, 3, 224, 224)
>>> model(x).logits.shape
(1, 1000)
Load ImageNet-pretrained weights:
>>> model = convnext_small_cls(pretrained=True)
>>> from lucid.models.weights import ConvNeXtSmallWeights
>>> model = convnext_small_cls(weights=ConvNeXtSmallWeights.IMAGENET1K_V1)