clip_vit_base_32_zero_shot

→CLIPForZeroShotImageClassification

clip_vit_base_32_zero_shot(pretrained: bool = False, overrides: object = {})

ViT-B/32 CLIP posed as an open-vocabulary classifier.

pretrainedbool= False

Load the OpenAI WIT-400M checkpoint from the Lucid hub.

**overridesobject= {}

Optional CLIPConfig field overrides.

CLIPForZeroShotImageClassification

Scores images against tokenised prompts.

Reference: Radford et al., ICML 2021 (arXiv:2103.00020), §3.1.

>>> from lucid.models import clip_vit_base_32_zero_shot
>>> model = clip_vit_base_32_zero_shot()
>>> model.config.patch_size
32