diff --git a/python/llm/dev/benchmark/harness/bigdl-llm.patch b/python/llm/dev/benchmark/harness/bigdl-llm.patch
index 01ce06f1..b84ea631 100644
--- a/python/llm/dev/benchmark/harness/bigdl-llm.patch
+++ b/python/llm/dev/benchmark/harness/bigdl-llm.patch
@@ -95,7 +95,7 @@ index 00000000..74010da3
 +        self._device = device
 +        self.model = model.to(device)
 +
-+        self.tokenizer = LlamaTokenizer.from_pretrained(pretrained, trust_remote_code=True)
++        self.tokenizer = AutoTokenizer.from_pretrained(pretrained, trust_remote_code=True)
 +
 +        # setup for automatic batch size detection
 +        if batch_size == 'auto':