Views
No views yet
vit_base_patch16_224.orig_in21k Vision Transformer checkpoint.1from urllib.request import urlopen
2
3import jax
4from PIL import Image
5
6import jaxnn
7
8img = Image.open(urlopen(
9 "https://huggingface.co/datasets/huggingface/cats-image/resolve/main/cats_image.jpeg"
10))
11
12model = jaxnn.create_model("vit_base_patch16_224.orig_in21k", pretrained=True)
13model.eval()
14
15data_config = jaxnn.data.resolve_model_data_config(model)
16transforms = jaxnn.data.create_transform(**data_config, is_training=False)
17
18x = jax.numpy.expand_dims(transforms(img), 0)
19output = model(x, deterministic=True)
20
21top5_probabilities, top5_class_indices = jax.lax.top_k(
22 jax.nn.softmax(output, axis=-1) * 100,
23 k=5,
24)1from urllib.request import urlopen
2
3import jax
4from PIL import Image
5
6import jaxnn
7
8img = Image.open(urlopen(
9 "https://huggingface.co/datasets/huggingface/cats-image/resolve/main/cats_image.jpeg"
10))
11
12model = jaxnn.create_model(
13 "vit_base_patch16_224.orig_in21k",
14 pretrained=True,
15 num_classes=0,
16)
17model.eval()
18
19data_config = jaxnn.data.resolve_model_data_config(model)
20transforms = jaxnn.data.create_transform(**data_config, is_training=False)
21
22x = jax.numpy.expand_dims(transforms(img), 0)
23output = model(x, deterministic=True)1@article{dosovitskiy2020vit,
2 title={An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale},
3 author={Dosovitskiy, Alexey and Beyer, Lucas and Kolesnikov, Alexander and Weissenborn, Dirk and Zhai, Xiaohua and Unterthiner, Thomas and Dehghani, Mostafa and Minderer, Matthias and Heigold, Georg and Gelly, Sylvain and Uszkoreit, Jakob and Houlsby, Neil},
4 journal={ICLR},
5 year={2021}
6}1@misc{rw2019timm,
2 author = {Ross Wightman},
3 title = {PyTorch Image Models},
4 year = {2019},
5 publisher = {GitHub},
6 journal = {GitHub repository},
7 doi = {10.5281/zenodo.4414861},
8 howpublished = {\url{https://github.com/huggingface/pytorch-image-models}}
9}