resnest_269_cls(pretrained: bool | str = False, weights: ResNeSt269Weights | None = None, overrides: object = {})ResNeSt-269 image classifier (backbone + GAP + linear head).
Builds a ResNeStForImageClassification with the
deepest paper-cited backbone ([3, 30, 48, 8],
stem_width = 64) followed by global average pooling and a
linear classifier. Approximately 110.9M parameters and 84.5%
ImageNet-1k top-1 accuracy in Zhang et al., 2022 (Table 4) at
416×416 input resolution.
Model Size
Parameters
pretrainedbool or str= FalsePretrained-weight selector.
False → random init; True
→ the DEFAULT tag (ResNeSt269Weights.IN1K); a tag
string → that specific checkpoint. Mutually exclusive with
weights (which wins if both are given).Explicit weights enum member. Takes precedence over
pretrained.**overridesobject= {}Keyword overrides forwarded into
ResNeStConfig.Returns
ResNeStForImageClassificationClassifier with the ResNeSt-269 configuration applied (or
with overrides merged on top of it).
Notes
See Zhang et al., "ResNeSt: Split-Attention Networks", CVPR
Workshops 2022 (arXiv:2004.08955), Table 4. Memory footprint
is substantial. Pretrained weights are converted from timm's
resnest269e.in1k and hosted under lucid-dl/resnest-269.
Examples
>>> import lucid
>>> from lucid.models.vision.resnest import resnest_269_cls
>>> model = resnest_269_cls()
>>> x = lucid.randn(1, 3, 224, 224)
>>> out = model(x)
>>> out.logits.shape
(1, 1000)