cvt_w24_cls(pretrained: bool | str = False, weights: CvTW24Weights | None = None, overrides: object = {})CvT-W24 wide image classifier (Wu et al., 2021).
Combines the cvt_w24 backbone (dims=(192, 768, 1024),
depths=(2, 2, 20)) with a mean pool + LayerNorm + linear
classification head. Approximately 277.2M parameters — the
largest CvT variant in the paper, typically used only with
ImageNet-22k pretraining.
Model Size
Parameters
pretrainedbool= FalseIf
True, loads ImageNet-22k pretrained weights when
available. Defaults to False.Explicit weights enum member; takes precedence over
pretrained.**overridesobject= {}Keyword overrides on top of the canonical CvT-W24 config.
Returns
CvTForImageClassificationClassifier whose logits has shape (B, num_classes).
Notes
CvT-W24 reaches 87.7% top-1 on ImageNet-1k at 384x384 fine-tune resolution after ImageNet-22k pretraining (Wu et al., 2021, Table 4) — the headline result of the paper.
Examples
>>> import lucid
>>> from lucid.models.vision.cvt import cvt_w24_cls
>>> model = cvt_w24_cls(num_classes=1000)
>>> x = lucid.randn(1, 3, 224, 224)
>>> model(x).logits.shape
(1, 1000)