Update vision_encoder.py

vision_encoder.py

@@ -18,7 +18,7 @@ class VisionEncoder(nn.Module):
         ])
 
     def forward(self, image, device):
-        processed_image = self.image_transform(image).unsqueeze(0).to(device)
+        processed_images = torch.stack([self.image_transform(image) for image in images]).to(device)        
         with torch.no_grad():
             pixel_values = self.vision_model(processed_image)
             image_features = pixel_values.last_hidden_state