port: 12312
served_model_name: mymodel
tensor_parallel_size: 2