xc-llm-ascend/tests/e2e/multi_node/config/config.json

[
    {
        "test_name": "test_deepseek_v3",
        "disaggregate_prefill": false,
        "enable_multithread_load": false,
        "num_nodes": 2,
        "server_parameters": {
            "leader_config": {
                "model": "vllm-ascend/DeepSeek-V3-W8A8",
                "quantization": "ascend",
                "additional_config": {
                    "ascend_scheduler_config": {
                        "enabled": true
                    },
                    "torchair_graph_config": {
                        "enabled": true
                    }
                }
            },
            "worker_config": {
                "model": "vllm-ascend/DeepSeek-V3-W8A8",
                "quantization": "ascend",
                "additional_config": {
                    "ascend_scheduler_config": {
                        "enabled": true
                    },
                    "torchair_graph_config": {
                        "enabled": true
                    }
                }
            }
        },
        "client_parameters": {
            "model": "vllm-ascend/DeepSeek-V3-W8A8",
            "backend": "vllm",
            "dataset_name": "sharegpt",
            "dataset_path": "/root/.cache/datasets/ShareGPT_V3_unfiltered_cleaned_split.json",
            "num_prompts": 200,
            "request_rate": 1
        },
        "accuracy_parameters": {}
    }
]