upd

README.md

@@ -23,11 +23,11 @@ from transformers import AutoModel, CLIPImageProcessor
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 model = AutoModel.from_pretrained(
-    'jienengchen/ViTamin-XL-384px',
+    'jienengchen/ViTamin-XL-256px',
     trust_remote_code=True).to(device).eval()
 
 image = Image.open('./image.png').convert('RGB')
-image_processor = CLIPImageProcessor.from_pretrained('jienengchen/ViTamin-XL-384px')
+image_processor = CLIPImageProcessor.from_pretrained('jienengchen/ViTamin-XL-256px')
 
 pixel_values = image_processor(images=image, return_tensors='pt').pixel_values
 pixel_values = pixel_values.to(torch.bfloat16).cuda()

vitamin.py

@@ -765,7 +765,7 @@ def vitamin_large_384(pretrained=False, **kwargs) -> VisionTransformer:
         head_type='1d',
     ), 
     )
-    model_args = dict(img_size=384, embed_dim=1024, depth=31, num_heads=16, mlp_layer=GeGluMlp, mlp_ratio=2., class_token=False, is_pos_embed=False, global_pool='avg')
+    model_args = dict(img_size=384, embed_dim=1024, depth=31, num_heads=16, mlp_layer=GeGluMlp, mlp_ratio=2., class_token=False, global_pool='avg')
     model = _create_vision_transformer_hybrid(
         'vitamin_large_384', backbone=backbone, pretrained=pretrained, **dict(model_args, **kwargs))
     return model