diff --git a/cosyvoice/llm/llm_vllm.py b/cosyvoice/llm/llm_vllm.py
index 3fd7152..1e9bc28 100644
--- a/cosyvoice/llm/llm_vllm.py
+++ b/cosyvoice/llm/llm_vllm.py
@@ -103,7 +103,7 @@ class VllmQwen2LM(Qwen2LM):
     async def inference_processor(self, task_queue):
         while True:
             try:
-                print(f"inference_processor")
+                logging.debug(f"inference_processor")
                 out_queue, prompt_token_ids, request_id, stop_token_ids, max_tokens = task_queue.get()
                 sampling_params = SamplingParams(**SAMPLING_PARAMS)
                 sampling_params.stop_token_ids = stop_token_ids or [6561]
diff --git a/requirements_vllm.txt b/requirements_vllm.txt
new file mode 100644
index 0000000..f3dcb25
--- /dev/null
+++ b/requirements_vllm.txt
@@ -0,0 +1,40 @@
+vllm==0.7.3
+pydantic==2.10.6
+torch==2.5.1
+torchaudio==2.5.1
+
+conformer==0.3.2
+
+diffusers==0.32.2
+gdown==5.1.0
+grpcio==1.57.0
+grpcio-tools==1.57.0
+hydra-core==1.3.2
+HyperPyYAML==1.2.2
+inflect==7.3.1
+librosa==0.10.2
+
+lightning==2.5.0.post0
+matplotlib==3.7.5
+modelscope==1.15.0
+
+networkx==3.4.2
+omegaconf==2.3.0
+onnx==1.17.0
+
+onnxruntime-gpu==1.19.0; sys_platform == 'linux'
+
+#openai-whisper==20231117
+openai-whisper==20240930
+protobuf==4.25
+pyworld==0.3.4
+rich==13.7.1
+soundfile==0.12.1
+tensorboard==2.14.0
+wget==3.2
+WeTextProcessing==1.0.3
+
+# trt use
+tensorrt-cu12==10.0.1
+tensorrt-cu12-bindings==10.0.1
+tensorrt-cu12-libs==10.0.1
\ No newline at end of file