Update ChatHaruhi/Qwen118k2GPT.py

ChatHaruhi/Qwen118k2GPT.py

@@ -16,10 +16,10 @@ def initialize_Qwen2LORA(model):
     if model_qwen is None:
         model_qwen = AutoModelForCausalLM.from_pretrained(
             model,
-            # torch_dtype=torch.float16,
+            torch_dtype=torch.float16,
             device_map="auto",
             trust_remote_code=True
-        ).half()
+        )
         model_qwen = model_qwen.eval()
         # model_qwen = PeftModel.from_pretrained(
         #     model_qwen,