| 1 | { |
| 2 | "defaults": { |
| 3 | "qps_list": [ |
| 4 | "inf" |
| 5 | ], |
| 6 | "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], |
| 7 | "server_parameters": { |
| 8 | "model": "meta-llama/Llama-3.1-8B-Instruct", |
| 9 | "tensor_parallel_size": 1, |
| 10 | "dtype": "bfloat16" |
| 11 | }, |
| 12 | "client_parameters": { |
| 13 | "model": "meta-llama/Llama-3.1-8B-Instruct", |
| 14 | "backend": "vllm", |
| 15 | "ignore-eos": "", |
| 16 | "temperature": 0, |
| 17 | "num_prompts": 200 |
| 18 | } |
| 19 | }, |
| 20 | "tests": [ |
| 21 | { |
| 22 | "test_name": "serving_llama8B_tp1_sharegpt", |
| 23 | "server_parameters": { |
| 24 | "tensor_parallel_size": 1 |
| 25 | }, |
| 26 | "client_parameters": { |
| 27 | "dataset_name": "sharegpt", |
| 28 | "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json" |
| 29 | } |
| 30 | }, |
| 31 | { |
| 32 | "test_name": "serving_llama8B_tp1_random_128_128", |
| 33 | "server_parameters": { |
| 34 | "tensor_parallel_size": 1 |
| 35 | }, |
| 36 | "client_parameters": { |
| 37 | "dataset_name": "random", |
| 38 | "random-input-len": 128, |
| 39 | "random-output-len": 128 |
| 40 | } |
| 41 | }, |
| 42 | { |
| 43 | "test_name": "serving_llama8B_tp1_random_128_2048", |
| 44 | "server_parameters": { |
| 45 | "tensor_parallel_size": 1 |
| 46 | }, |
| 47 | "client_parameters": { |
| 48 | "dataset_name": "random", |
| 49 | "random-input-len": 128, |
| 50 | "random-output-len": 2048 |
| 51 | } |
| 52 | }, |
| 53 | { |
| 54 | "test_name": "serving_llama8B_tp1_random_2048_128", |
| 55 | "server_parameters": { |
| 56 | "tensor_parallel_size": 1 |
| 57 | }, |
| 58 | "client_parameters": { |
| 59 | "dataset_name": "random", |
| 60 | "random-input-len": 2048, |
| 61 | "random-output-len": 128 |
| 62 | } |
| 63 | }, |
| 64 | { |
| 65 | "test_name": "serving_llama8B_tp1_random_2048_2048", |
| 66 | "server_parameters": { |
| 67 | "tensor_parallel_size": 1 |
| 68 | }, |
| 69 | "client_parameters": { |
| 70 | "dataset_name": "random", |
| 71 | "random-input-len": 2048, |
| 72 | "random-output-len": 2048 |
| 73 | } |
| 74 | }, |
| 75 | { |
| 76 | "test_name": "serving_llama70B_tp4_random_128_128", |
| 77 | "server_parameters": { |
| 78 | "model": "meta-llama/Llama-3.3-70B-Instruct", |
| 79 | "tensor_parallel_size": 4, |
| 80 | "async_scheduling": "", |
| 81 | "no_enable_prefix_caching": "", |
| 82 | "max_num_batched_tokens": 8192 |
| 83 | }, |
| 84 | "client_parameters": { |
| 85 | "model": "meta-llama/Llama-3.3-70B-Instruct", |
| 86 | "dataset_name": "random", |
| 87 | "random-input-len": 128, |
| 88 | "random-output-len": 128 |
| 89 | } |
| 90 | }, |
| 91 | { |
| 92 | "test_name": "serving_gemma4-e4b_tp1_random_128_128", |
| 93 | "server_parameters": { |
| 94 | "model": "google/gemma-4-E4B-it", |
| 95 | "enable_auto_tool_choice": "", |
| 96 | "tool_call_parser": "gemma4", |
| 97 | "chat_template": "examples/tool_chat_template_gemma4.jinja", |
| 98 | "reasoning_parser": "gemma4" |
| 99 | }, |
| 100 | "client_parameters": { |
| 101 | "model": "google/gemma-4-E4B-it", |
| 102 | "dataset_name": "random", |
| 103 | "random-input-len": 128, |
| 104 | "random-output-len": 128 |
| 105 | } |
| 106 | } |
| 107 | ] |
| 108 | } |