inception_next_base_cls(pretrained: bool | str = False, weights: InceptionNeXtBaseWeights | None = None, overrides: object = {})InceptionNeXt-Base image classifier (Yu et al., 2024).
Combines the inception_next_base backbone with the
reference MLP classifier head (Linear → GELU → LayerNorm → Linear). Default output is num_classes=1000 (ImageNet-1k).
Approximately 87M parameters.
Model Size
Parameters
pretrainedbool or str= FalsePretrained-weight selector.
False → random init; True
→ the DEFAULT tag (InceptionNeXtBaseWeights.SAIL_IN1K);
a tag string → that specific checkpoint. Mutually exclusive
with weights (which wins if both given).Explicit weights enum member. Takes precedence over
pretrained.**overridesobject= {}Keyword overrides on top of the canonical InceptionNeXt-B config.
Returns
InceptionNeXtForImageClassificationClassifier returning ImageClassificationOutput whose
logits has shape (B, num_classes).
Notes
InceptionNeXt-B reaches 84.0% top-1 on ImageNet-1k (Yu et al.,
2024, Table 2). Pretrained weights are converted from timm's
inception_next_base.sail_in1k and hosted on the Hugging Face Hub
under lucid-dl/inception-next-base.
Examples
>>> import lucid
>>> from lucid.models.vision.inception_next import inception_next_base_cls
>>> model = inception_next_base_cls(num_classes=1000)
>>> x = lucid.randn(1, 3, 224, 224)
>>> model(x).logits.shape
(1, 1000)