{
  "version": "v2.0",
  "motor_deploy_config": {
    "p_instances_num": 1,
    "d_instances_num": 1,
    "single_p_instance_pod_num": 1,
    "single_d_instance_pod_num": 1,
    "p_pod_npu_num": 16,
    "d_pod_npu_num": 16,
    "image_name": "",
    "job_id": "mindie-motor",
    "hardware_type": "800I_A3",
    "weight_mount_path": "/mnt/weight/"
  },
  "motor_controller_config": {},
  "motor_coordinator_config": {},
  "motor_engine_prefill_config": {
    "engine_type": "vllm",
    "motor_nodemanger_config": {},
    "engine_config": {
      "served_model_name": "qwen3-8B",
      "model": "/mnt/weight/qwen3_8B",
      "gpu_memory_utilization": 0.9,
      "data_parallel_size": 1,
      "tensor_parallel_size": 2,
      "pipeline_parallel_size": 1,
      "data_parallel_rpc_port": 9000,
      "enable_expert_parallel": false,
      "enforce-eager": true,
      "max_model_len": 2048,
      "kv_transfer_config": {
        "kv_connector": "MooncakeLayerwiseConnector",
        "kv_buffer_device": "npu",
        "kv_role": "kv_producer",
        "kv_parallel_size": 1,
        "kv_port": "30001",
        "engine_id": "0",
        "kv_rank": 0,
        "kv_connector_extra_config": {}
      }
    }
  },
  "motor_engine_decode_config": {
    "engine_type": "vllm",
    "motor_nodemanger_config": {},
    "engine_config": {
      "served_model_name": "qwen3-8B",
      "model": "/mnt/weight/qwen3_8B",
      "gpu_memory_utilization": 0.9,
      "data_parallel_size": 1,
      "tensor_parallel_size": 2,
      "pipeline_parallel_size": 1,
      "data_parallel_rpc_port": 9000,
      "enable_expert_parallel": false,
      "max_model_len": 2048,
      "kv_transfer_config": {
        "kv_connector": "MooncakeLayerwiseConnector",
        "kv_buffer_device": "npu",
        "kv_role": "kv_consumer",
        "kv_parallel_size": 1,
        "kv_port": "30001",
        "engine_id": "0",
        "kv_rank": 0,
        "kv_connector_extra_config": {}
      }
    }
  }
}