vit_h_14

torchvision.models.vit_h_14(*, weights: Optional[ViT_H_14_Weights] = None, progress: bool = True, **kwargs: Any) → VisionTransformer

Constructs a vit_h_14 architecture from An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale.

Parameters:

• weights (ViT_H_14_Weights, optional) – The pretrained weights to use. See ViT_H_14_Weights below for more details and possible values. By default, no pre-trained weights are used.

• progress (bool, optional) – If True, displays a progress bar of the download to stderr. Default is True.

• **kwargs – parameters passed to the torchvision.models.vision_transformer.VisionTransformer base class. Please refer to the source code for more details about this class.

class torchvision.models.ViT_H_14_Weights(value)

The model builder above accepts the following values as the weights parameter. ViT_H_14_Weights.DEFAULT is equivalent to ViT_H_14_Weights.IMAGENET1K_SWAG_E2E_V1. You can also use strings, e.g. weights='DEFAULT' or weights='IMAGENET1K_SWAG_E2E_V1'.

ViT_H_14_Weights.IMAGENET1K_SWAG_E2E_V1:

These weights are learnt via transfer learning by end-to-end fine-tuning the original SWAG weights on ImageNet-1K data. Also available as ViT_H_14_Weights.DEFAULT.

acc@1 (on ImageNet-1K)	88.552
acc@5 (on ImageNet-1K)	98.694
categories	tench, goldfish, great white shark, … (997 omitted)
recipe	link
license	link
num_params	633470440
min_size	height=518, width=518
GFLOPS	1016.72
File size	2416.6 MB

The inference transforms are available at ViT_H_14_Weights.IMAGENET1K_SWAG_E2E_V1.transforms and perform the following preprocessing operations: Accepts PIL.Image, batched (B, C, H, W) and single (C, H, W) image torch.Tensor objects. The images are resized to resize_size=[518] using interpolation=InterpolationMode.BICUBIC, followed by a central crop of crop_size=[518]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.485, 0.456, 0.406] and std=[0.229, 0.224, 0.225].

ViT_H_14_Weights.IMAGENET1K_SWAG_LINEAR_V1:

These weights are composed of the original frozen SWAG trunk weights and a linear classifier learnt on top of them trained on ImageNet-1K data.

acc@1 (on ImageNet-1K)	85.708
acc@5 (on ImageNet-1K)	97.73
categories	tench, goldfish, great white shark, … (997 omitted)
recipe	link
license	link
num_params	632045800
min_size	height=224, width=224
GFLOPS	167.29
File size	2411.2 MB

The inference transforms are available at ViT_H_14_Weights.IMAGENET1K_SWAG_LINEAR_V1.transforms and perform the following preprocessing operations: Accepts PIL.Image, batched (B, C, H, W) and single (C, H, W) image torch.Tensor objects. The images are resized to resize_size=[224] using interpolation=InterpolationMode.BICUBIC, followed by a central crop of crop_size=[224]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.485, 0.456, 0.406] and std=[0.229, 0.224, 0.225].