使用vllm进行qwen3-vl模型推理时，vllm==0.11.0, qwen_vl_utils=0.0.14,transformers==4.57.0,出现RuntimeError: Numpy is not available，numpy版本是2.2.6

[Nobelvictory]

Traceback (most recent call last):
File "/checkpoint/binary/train_package/swift/cli/infer.py", line 5, in
infer_main()
File "/checkpoint/binary/train_package/swift/pipelines/infer/infer.py", line 308, in infer_main
return SwiftInfer(args).main()
File "/checkpoint/binary/train_package/swift/pipelines/infer/infer.py", line 40, in init
self.infer_engine = self.get_infer_engine(args, self.template)
File "/checkpoint/binary/train_package/swift/pipelines/infer/infer.py", line 89, in get_infer_engine
return infer_engine_cls(**kwargs)
File "/checkpoint/binary/train_package/swift/infer_engine/vllm_engine.py", line 153, in init
self._prepare_engine()
File "/checkpoint/binary/train_package/swift/infer_engine/vllm_engine.py", line 175, in _prepare_engine
engine = llm_engine_cls.from_engine_args(self.engine_args)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/engine/llm_engine.py", line 177, in from_engine_args
return cls(vllm_config=vllm_config,
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/engine/llm_engine.py", line 114, in init
self.engine_core = EngineCoreClient.make_client(
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/engine/core_client.py", line 82, in make_client
return InprocClient(vllm_config, executor_class, log_stats)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/engine/core_client.py", line 245, in init
self.engine_core = EngineCore(*args, **kwargs)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 123, in init
self.model_executor = executor_class(vllm_config)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/executor/executor_base.py", line 54, in init
self._init_executor()
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/executor/uniproc_executor.py", line 144, in _init_executor
super()._init_executor()
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/executor/uniproc_executor.py", line 54, in _init_executor
self.collective_rpc("init_device")
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/executor/uniproc_executor.py", line 83, in collective_rpc
return [run_method(self.driver_worker, method, args, kwargs)]
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/utils/init.py", line 3139, in run_method
return func(*args, **kwargs)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 259, in init_device
self.worker.init_device() # type: ignore
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/worker/gpu_worker.py", line 169, in init_device
init_worker_distributed_environment(self.vllm_config, self.rank,
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/v1/worker/gpu_worker.py", line 705, in init_worker_distributed_environment
ensure_model_parallel_initialized(
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/distributed/parallel_state.py", line 1228, in ensure_model_parallel_initialized
initialize_model_parallel(tensor_model_parallel_size,
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/distributed/parallel_state.py", line 1152, in initialize_model_parallel
_TP = init_model_parallel_group(group_ranks,
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/distributed/parallel_state.py", line 924, in init_model_parallel_group
return GroupCoordinator(
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/distributed/parallel_state.py", line 266, in init
self.mq_broadcaster = MessageQueue.create_from_process_group(
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/vllm/distributed/device_communicators/shm_broadcast.py", line 581, in create_from_process_group
dist.broadcast_object_list(recv,
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper
return func(*args, **kwargs)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3653, in broadcast_object_list
object_list[i] = _tensor_to_object(obj_view, obj_size, group)
File "/opt/conda/envs/python3.10.13/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3089, in _tensor_to_object
buf = tensor.numpy().tobytes()[:tensor_size]
RuntimeError: Numpy is not available

我看之前issue中numpy为2.2.6是可以的，而且我用qwen2.5-vl推理时使用numpy为2.2.6是可以的，请问这个是什么原因？

[JJJYmmm]

尝试回推到 numpy < 2吧；感觉是torch和numpy的冲突