视觉变换器¶

模型构建器¶

以下模型构建器可用于实例化VisionTransformer模型，无论是否使用预训练权重。所有模型构建器内部都依赖于torchvision.models.vision_transformer.VisionTransformer基类。有关此类的更多详细信息，请参阅源代码。

`vit_b_16`(*[, 权重, 进度])	从An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale构建一个vit_b_16架构。
`vit_b_32`(*[, 权重, 进度])	从An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale构建了一个vit_b_32架构。
`vit_l_16`(*[, weights, progress])	从An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale构建了一个vit_l_16架构。
`vit_l_32`(*[, 权重, 进度])	从An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale构建一个vit_l_32架构。
`vit_h_14`(*[, weights, progress])	从An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale构建了一个vit_h_14架构。