Pretrained weights for lucid.models.mobilenet_v4_hybrid_large_cls.
Qin et al. ECCV 2024 Hybrid-Large (37.8 M params, top-1 84.00%).
Attributes
IX_E600_R384_IN1KWeightEntryImageNet-1k checkpoint trained 600 epochs at 384x384 (top-1
83.996% / top-5 96.714% at 384, per timm's
results-imagenet.csv for this exact tag), sourced from
timm/mobilenetv4_hybrid_large.ix_e600_r384_in1k.DEFAULTWeightEntryAlias for
IX_E600_R384_IN1K.Notes
Reference: Qin et al., "MobileNetV4: Universal Models for the Mobile Ecosystem", ECCV 2024 (arXiv:2404.10518).
Examples
>>> from lucid.models.weights import MobileNetV4HybridLargeWeights
>>> tf = MobileNetV4HybridLargeWeights.DEFAULT.transforms()
>>> tf.crop_size, tf.resize_size, tf.interpolation
(384, 404, 'bicubic')