Update app.py

app.py

@@ -76,6 +76,7 @@ coreocr_model = Qwen2VLForConditionalGeneration.from_pretrained(
 
 # Main Inference Function
 @spaces.GPU
+@torch.no_grad()
 def model_inference(message, history, use_coreocr):
     text = message["text"].strip()
     files = message.get("files", [])