{
    "model_type": "vit",
    "architectures": ["ViTForImageClassification"],
    "hidden_size": 768,
    "num_attention_heads": 12,
    "image_size": 1024,
    "patch_size": 16
}