Update Visual-CoT demo - 2025-10-12 22:46:22

Fixes:
- Fix LLaVA config registration error (compatibility with newer transformers)
- Update Gradio to latest version (security fixes)
- Auto-deployed via update script

llava/mm_utils.py

@@ -28,13 +28,21 @@ def expand2square(pil_img, background_color):
 def process_images(images, image_processor, model_cfg):
     image_aspect_ratio = getattr(model_cfg, "image_aspect_ratio", None)
     new_images = []
+    
+    # Handle both single processor and list of processors (multi-scale vision)
+    if isinstance(image_processor, list):
+        # Multi-scale: use first processor for preprocessing
+        processor = image_processor[0]
+    else:
+        processor = image_processor
+    
     if image_aspect_ratio == 'pad':
         for image in images:
-            image = expand2square(image, tuple(int(x*255) for x in image_processor.image_mean))
-            image = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0]
+            image = expand2square(image, tuple(int(x*255) for x in processor.image_mean))
+            image = processor.preprocess(image, return_tensors='pt')['pixel_values'][0]
             new_images.append(image)
     else:
-        return image_processor(images, return_tensors='pt')['pixel_values']
+        return processor(images, return_tensors='pt')['pixel_values']
     if all(x.shape == new_images[0].shape for x in new_images):
         new_images = torch.stack(new_images, dim=0)
     return new_images