update model movement

app.py

@@ -47,11 +47,7 @@ def generate_embeddings(prompts_text):
         # Load model if not already loaded
         pipe = load_model()
         
-        # Move pipeline to CUDA if available
-        if torch.cuda.is_available():
-            device = torch.device("cuda")
-            if hasattr(pipe, 'text_encoder') and pipe.text_encoder is not None:
-                pipe.text_encoder = pipe.text_encoder.to(device)
+        # Note: 8-bit models are already on the correct device, no need to move them
         
         # Parse prompts (one per line)
         prompts = [p.strip() for p in prompts_text.strip().split('\n') if p.strip()]