Update app.py

app.py

@@ -31,7 +31,7 @@ DEFAULT_MODEL_ID = os.environ.get("QTSPLUS_MODEL_ID", "AlpachinoNLP/QTSplus-3B")
 DEFAULT_QUESTION = "What is happening in the video?"
 
 DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-DTYPE = torch.bfloat16 if DEVICE.type == "cuda" else torch.float32
+DTYPE = torch.bfloat16 if DEVICE.type == "cuda" else torch.float16
 
 _MODEL: Optional[AutoModelForCausalLM] = None
 _PROCESSOR: Optional[AutoProcessor] = None