Update app.py

app.py

@@ -417,8 +417,8 @@ def run_flux_generation(instruction_text, source_img, width, height):
 def run_vlm_evaluation(messages, loaded_images):
     """Run VLM model on GPU to evaluate images and stream output text."""
     # Ensure model is on CUDA/device for evaluation
-    if not (VLM_MAX_MEMORY or VLM_QUANTIZATION_4BIT or VLM_QUANTIZATION_8BIT):
-        model.to(device_vlm)
+    # if not (VLM_MAX_MEMORY or VLM_QUANTIZATION_4BIT or VLM_QUANTIZATION_8BIT):
+    model.to(device_vlm)
 
     # Generate and stream text
     prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)