Update app.py

app.py

@@ -43,7 +43,7 @@ def load_model():
             model = AutoModelForCausalLM.from_pretrained(
                 MODEL_NAME,
                 trust_remote_code=True,
-                torch_dtype=torch.float32,
+                torch_dtype=torch.float16,
                 low_cpu_mem_usage=True,
             )
             model = model.to(DEVICE)