Pretrained weights for lucid.models.pvt_v2_b3_cls.
Single ImageNet-1k checkpoint converted from timm's
pvt_v2_b3.in1k (Wang et al., 2022; ~45.2M params, 83.1% top-1).
Examples
>>> from lucid.models.vision.pvt._weights import PVTv2B3Weights
>>> list(PVTv2B3Weights.__members__)
['IN1K', 'DEFAULT']
DEFAULT is an alias rather than a fourth entry, so a bare
pretrained=True and the tag it resolves to cannot drift apart.
>>> PVTv2B3Weights.DEFAULT is PVTv2B3Weights.IN1K
True
>>> PVTv2B3Weights.IN1K.num_classes
1000