785 lines
23 KiB
JSON
785 lines
23 KiB
JSON
{
|
|
"version": "1.0",
|
|
"timestamp": "2026-08-25T17:44:25.622293",
|
|
"run_id": "20260825-174425",
|
|
"environment": {
|
|
"python_version": "3.13.11",
|
|
"platform": "linux",
|
|
"model": "qwen3.6:35b-a3b"
|
|
},
|
|
"config": {
|
|
"cases_requested": [
|
|
"case_01_generation",
|
|
"case_02_simple_tool",
|
|
"case_03_file_analysis",
|
|
"case_04_parallel_tool",
|
|
"case_05_long_context",
|
|
"case_06_reasoning",
|
|
"case_07_multiturn"
|
|
],
|
|
"repeats": 2,
|
|
"skip_heavy": false,
|
|
"total_cases": 7
|
|
},
|
|
"summary": {
|
|
"total_cases": 7,
|
|
"passed": 7,
|
|
"failed": 0,
|
|
"stopped": false,
|
|
"total_elapsed_seconds": 728.71
|
|
},
|
|
"results": [
|
|
{
|
|
"case_id": "case_01_generation",
|
|
"case_name": "纯文本生成基准",
|
|
"difficulty": "简单",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_01_generation",
|
|
"name": "纯文本生成基准",
|
|
"description": "基础生成速度基准测试",
|
|
"difficulty": "简单",
|
|
"estimated_seconds": 5,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 14715.57,
|
|
"prefill_ms": 328.38,
|
|
"decode_speed_tok_s": 80.65,
|
|
"total_tokens": 331,
|
|
"prompt_tokens": 31,
|
|
"completion_tokens": 300,
|
|
"e2e_ms": 18438.39,
|
|
"elapsed_ms": 18438.39,
|
|
"response_length": 912
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 13780.79,
|
|
"prefill_ms": 338.35,
|
|
"decode_speed_tok_s": 79.88,
|
|
"total_tokens": 331,
|
|
"prompt_tokens": 31,
|
|
"completion_tokens": 300,
|
|
"e2e_ms": 17539.77,
|
|
"elapsed_ms": 17539.77,
|
|
"response_length": 988
|
|
}
|
|
],
|
|
"prompt_used": "请用200字左右概述人工智能从诞生至今的关键发展阶段，每段不超过50字。"
|
|
},
|
|
"statistics": {
|
|
"count": 2,
|
|
"mean_ms": 17989.08,
|
|
"std_ms": 635.42,
|
|
"min_ms": 17539.77,
|
|
"max_ms": 18438.39,
|
|
"p50_ms": 17989.08,
|
|
"p95_ms": 18393.46,
|
|
"p99_ms": 18429.4
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 14248.18,
|
|
"ttft_ms_p95": 14668.83,
|
|
"prefill_ms_mean": 333.37,
|
|
"prefill_ms_p95": 337.85,
|
|
"decode_speed_tok_s_mean": 80.27,
|
|
"decode_speed_tok_s_p95": 80.61,
|
|
"e2e_ms_mean": 17989.08,
|
|
"e2e_ms_p95": 18393.46,
|
|
"total_tokens_mean": 331.0,
|
|
"total_tokens_p95": 331.0
|
|
},
|
|
"elapsed_seconds": 35.98
|
|
},
|
|
{
|
|
"case_id": "case_02_simple_tool",
|
|
"case_name": "简单工具调用延迟",
|
|
"difficulty": "简单",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_02_simple_tool",
|
|
"name": "简单工具调用延迟",
|
|
"description": "测量简单工具调用的额外延迟",
|
|
"difficulty": "简单",
|
|
"estimated_seconds": 3,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 16196.0,
|
|
"prefill_ms": 282.79,
|
|
"decode_speed_tok_s": 84.08,
|
|
"total_tokens": 83,
|
|
"prompt_tokens": 61,
|
|
"completion_tokens": 22,
|
|
"e2e_ms": 16459.21,
|
|
"elapsed_ms": 16459.21,
|
|
"response_length": 60,
|
|
"tool": "get_date",
|
|
"matched": false
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 14408.31,
|
|
"prefill_ms": 289.76,
|
|
"decode_speed_tok_s": 81.91,
|
|
"total_tokens": 122,
|
|
"prompt_tokens": 62,
|
|
"completion_tokens": 60,
|
|
"e2e_ms": 15142.91,
|
|
"elapsed_ms": 15142.91,
|
|
"response_length": 117,
|
|
"tool": "get_cpu_usage",
|
|
"matched": false
|
|
}
|
|
]
|
|
},
|
|
"statistics": {
|
|
"count": 2,
|
|
"mean_ms": 15801.06,
|
|
"std_ms": 930.76,
|
|
"min_ms": 15142.91,
|
|
"max_ms": 16459.21,
|
|
"p50_ms": 15801.06,
|
|
"p95_ms": 16393.4,
|
|
"p99_ms": 16446.05
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 15302.15,
|
|
"ttft_ms_p95": 16106.62,
|
|
"prefill_ms_mean": 286.27,
|
|
"prefill_ms_p95": 289.41,
|
|
"decode_speed_tok_s_mean": 83.0,
|
|
"decode_speed_tok_s_p95": 83.97,
|
|
"e2e_ms_mean": 15801.06,
|
|
"e2e_ms_p95": 16393.4,
|
|
"total_tokens_mean": 102.5,
|
|
"total_tokens_p95": 120.05
|
|
},
|
|
"elapsed_seconds": 31.6
|
|
},
|
|
{
|
|
"case_id": "case_03_file_analysis",
|
|
"case_name": "文件读取+分析",
|
|
"difficulty": "中等",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_03_file_analysis",
|
|
"name": "文件读取+分析",
|
|
"description": "文件读取与分析完整链路计时",
|
|
"difficulty": "中等",
|
|
"estimated_seconds": 2,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 4497.2,
|
|
"prefill_ms": 96.54,
|
|
"decode_speed_tok_s": 81.86,
|
|
"total_tokens": 485,
|
|
"prompt_tokens": 185,
|
|
"completion_tokens": 300,
|
|
"e2e_ms": 8166.59,
|
|
"elapsed_ms": 8166.59,
|
|
"response_length": 1122,
|
|
"file_chars": 485
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 30717.41,
|
|
"prefill_ms": 103.84,
|
|
"decode_speed_tok_s": 79.84,
|
|
"total_tokens": 485,
|
|
"prompt_tokens": 185,
|
|
"completion_tokens": 300,
|
|
"e2e_ms": 34478.78,
|
|
"elapsed_ms": 34478.78,
|
|
"response_length": 1075,
|
|
"file_chars": 485
|
|
}
|
|
]
|
|
},
|
|
"statistics": {
|
|
"count": 2,
|
|
"mean_ms": 21322.68,
|
|
"std_ms": 18605.53,
|
|
"min_ms": 8166.59,
|
|
"max_ms": 34478.78,
|
|
"p50_ms": 21322.68,
|
|
"p95_ms": 33163.17,
|
|
"p99_ms": 34215.66
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 17607.31,
|
|
"ttft_ms_p95": 29406.4,
|
|
"prefill_ms_mean": 100.19,
|
|
"prefill_ms_p95": 103.47,
|
|
"decode_speed_tok_s_mean": 80.85,
|
|
"decode_speed_tok_s_p95": 81.76,
|
|
"e2e_ms_mean": 21322.68,
|
|
"e2e_ms_p95": 33163.17,
|
|
"total_tokens_mean": 485.0,
|
|
"total_tokens_p95": 485.0
|
|
},
|
|
"elapsed_seconds": 42.65
|
|
},
|
|
{
|
|
"case_id": "case_04_parallel_tool",
|
|
"case_name": "并行工具调用",
|
|
"difficulty": "中等",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_04_parallel_tool",
|
|
"name": "并行工具调用",
|
|
"description": "对比串行 vs 并行工具调用的延迟差异",
|
|
"difficulty": "中等",
|
|
"estimated_seconds": 5,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"parallel_tasks": 3,
|
|
"serial_ms": 37309.93,
|
|
"parallel_ms": 45353.04,
|
|
"speedup_ratio": 0.82,
|
|
"elapsed_ms": 45353.04,
|
|
"avg_ttft_ms": 43165.98,
|
|
"avg_decode_speed_tok_s": 83.01,
|
|
"ttft_ms": 43165.98,
|
|
"decode_speed_tok_s": 83.01
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"parallel_tasks": 3,
|
|
"serial_ms": 30577.28,
|
|
"parallel_ms": 10193.83,
|
|
"speedup_ratio": 3.0,
|
|
"elapsed_ms": 10193.83,
|
|
"avg_ttft_ms": 7917.49,
|
|
"avg_decode_speed_tok_s": 81.22,
|
|
"ttft_ms": 7917.49,
|
|
"decode_speed_tok_s": 81.22
|
|
}
|
|
]
|
|
},
|
|
"statistics": {
|
|
"count": 2,
|
|
"mean_ms": 27773.44,
|
|
"std_ms": 24861.32,
|
|
"min_ms": 10193.83,
|
|
"max_ms": 45353.04,
|
|
"p50_ms": 27773.44,
|
|
"p95_ms": 43595.08,
|
|
"p99_ms": 45001.45
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 25541.74,
|
|
"ttft_ms_p95": 41403.56,
|
|
"decode_speed_tok_s_mean": 82.12,
|
|
"decode_speed_tok_s_p95": 82.92
|
|
},
|
|
"elapsed_seconds": 123.43
|
|
},
|
|
{
|
|
"case_id": "case_05_long_context",
|
|
"case_name": "长上下文处理",
|
|
"difficulty": "耗时",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_05_long_context",
|
|
"name": "长上下文处理",
|
|
"description": "不同上下文长度对延迟的影响",
|
|
"difficulty": "耗时",
|
|
"estimated_seconds": 10,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 5375.1,
|
|
"prefill_ms": 842.82,
|
|
"decode_speed_tok_s": 80.26,
|
|
"total_tokens": 728,
|
|
"prompt_tokens": 648,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 6378.84,
|
|
"elapsed_ms": 6378.84,
|
|
"response_length": 319,
|
|
"context_tokens_target": 500,
|
|
"context_chars": 1254
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 57883.93,
|
|
"prefill_ms": 95.78,
|
|
"decode_speed_tok_s": 83.77,
|
|
"total_tokens": 728,
|
|
"prompt_tokens": 648,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 58844.25,
|
|
"elapsed_ms": 58844.25,
|
|
"response_length": 319,
|
|
"context_tokens_target": 500,
|
|
"context_chars": 1254
|
|
},
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 14493.28,
|
|
"prefill_ms": 1041.01,
|
|
"decode_speed_tok_s": 81.17,
|
|
"total_tokens": 1355,
|
|
"prompt_tokens": 1275,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 15491.6,
|
|
"elapsed_ms": 15491.6,
|
|
"response_length": 315,
|
|
"context_tokens_target": 1000,
|
|
"context_chars": 2508
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 88553.55,
|
|
"prefill_ms": 102.99,
|
|
"decode_speed_tok_s": 82.75,
|
|
"total_tokens": 1355,
|
|
"prompt_tokens": 1275,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 89532.55,
|
|
"elapsed_ms": 89532.55,
|
|
"response_length": 319,
|
|
"context_tokens_target": 1000,
|
|
"context_chars": 2508
|
|
},
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 19499.27,
|
|
"prefill_ms": 1513.23,
|
|
"decode_speed_tok_s": 80.77,
|
|
"total_tokens": 2628,
|
|
"prompt_tokens": 2548,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 20512.64,
|
|
"elapsed_ms": 20512.64,
|
|
"response_length": 331,
|
|
"context_tokens_target": 2000,
|
|
"context_chars": 5054
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 11929.32,
|
|
"prefill_ms": 114.78,
|
|
"decode_speed_tok_s": 77.52,
|
|
"total_tokens": 2628,
|
|
"prompt_tokens": 2548,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 12983.72,
|
|
"elapsed_ms": 12983.72,
|
|
"response_length": 329,
|
|
"context_tokens_target": 2000,
|
|
"context_chars": 5054
|
|
},
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 14999.33,
|
|
"prefill_ms": 3185.49,
|
|
"decode_speed_tok_s": 79.84,
|
|
"total_tokens": 5155,
|
|
"prompt_tokens": 5075,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 16042.1,
|
|
"elapsed_ms": 16042.1,
|
|
"response_length": 329,
|
|
"context_tokens_target": 4000,
|
|
"context_chars": 10108
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 9751.38,
|
|
"prefill_ms": 116.54,
|
|
"decode_speed_tok_s": 83.8,
|
|
"total_tokens": 5155,
|
|
"prompt_tokens": 5075,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 10748.62,
|
|
"elapsed_ms": 10748.62,
|
|
"response_length": 331,
|
|
"context_tokens_target": 4000,
|
|
"context_chars": 10108
|
|
}
|
|
],
|
|
"context_sizes_tested": [
|
|
500,
|
|
1000,
|
|
2000,
|
|
4000
|
|
]
|
|
},
|
|
"statistics": {
|
|
"count": 8,
|
|
"mean_ms": 28816.79,
|
|
"std_ms": 29467.59,
|
|
"min_ms": 6378.84,
|
|
"max_ms": 89532.55,
|
|
"p50_ms": 15766.85,
|
|
"p95_ms": 78791.64,
|
|
"p99_ms": 87384.37
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 27810.65,
|
|
"ttft_ms_p95": 77819.18,
|
|
"prefill_ms_mean": 876.58,
|
|
"prefill_ms_p95": 2600.2,
|
|
"decode_speed_tok_s_mean": 81.23,
|
|
"decode_speed_tok_s_p95": 83.79,
|
|
"e2e_ms_mean": 28816.79,
|
|
"e2e_ms_p95": 78791.64,
|
|
"total_tokens_mean": 2466.5,
|
|
"total_tokens_p95": 5155.0
|
|
},
|
|
"elapsed_seconds": 230.54
|
|
},
|
|
{
|
|
"case_id": "case_06_reasoning",
|
|
"case_name": "复杂推理任务",
|
|
"difficulty": "耗时",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_06_reasoning",
|
|
"name": "复杂推理任务",
|
|
"description": "不同推理复杂度下的延迟对比",
|
|
"difficulty": "耗时",
|
|
"estimated_seconds": 15,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 17139.94,
|
|
"prefill_ms": 284.82,
|
|
"decode_speed_tok_s": 81.61,
|
|
"total_tokens": 93,
|
|
"prompt_tokens": 13,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 18122.0,
|
|
"elapsed_ms": 18122.0,
|
|
"response_length": 314,
|
|
"reasoning_level": "简单问答"
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 47602.11,
|
|
"prefill_ms": 285.31,
|
|
"decode_speed_tok_s": 80.04,
|
|
"total_tokens": 93,
|
|
"prompt_tokens": 13,
|
|
"completion_tokens": 80,
|
|
"e2e_ms": 48603.85,
|
|
"elapsed_ms": 48603.85,
|
|
"response_length": 313,
|
|
"reasoning_level": "简单问答"
|
|
},
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 10822.25,
|
|
"prefill_ms": 315.46,
|
|
"decode_speed_tok_s": 82.22,
|
|
"total_tokens": 220,
|
|
"prompt_tokens": 20,
|
|
"completion_tokens": 200,
|
|
"e2e_ms": 13257.81,
|
|
"elapsed_ms": 13257.81,
|
|
"response_length": 811,
|
|
"reasoning_level": "中等推理"
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 20657.36,
|
|
"prefill_ms": 314.18,
|
|
"decode_speed_tok_s": 80.9,
|
|
"total_tokens": 220,
|
|
"prompt_tokens": 20,
|
|
"completion_tokens": 200,
|
|
"e2e_ms": 23132.11,
|
|
"elapsed_ms": 23132.11,
|
|
"response_length": 745,
|
|
"reasoning_level": "中等推理"
|
|
},
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 8117.27,
|
|
"prefill_ms": 335.93,
|
|
"decode_speed_tok_s": 82.11,
|
|
"total_tokens": 427,
|
|
"prompt_tokens": 27,
|
|
"completion_tokens": 400,
|
|
"e2e_ms": 12993.07,
|
|
"elapsed_ms": 12993.07,
|
|
"response_length": 1762,
|
|
"reasoning_level": "复杂推理"
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 6073.91,
|
|
"prefill_ms": 333.3,
|
|
"decode_speed_tok_s": 81.79,
|
|
"total_tokens": 427,
|
|
"prompt_tokens": 27,
|
|
"completion_tokens": 400,
|
|
"e2e_ms": 10969.79,
|
|
"elapsed_ms": 10969.79,
|
|
"response_length": 1809,
|
|
"reasoning_level": "复杂推理"
|
|
}
|
|
]
|
|
},
|
|
"statistics": {
|
|
"count": 6,
|
|
"mean_ms": 21179.77,
|
|
"std_ms": 14136.94,
|
|
"min_ms": 10969.79,
|
|
"max_ms": 48603.85,
|
|
"p50_ms": 15689.9,
|
|
"p95_ms": 42235.91,
|
|
"p99_ms": 47330.26
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 18402.14,
|
|
"ttft_ms_p95": 40865.92,
|
|
"prefill_ms_mean": 311.5,
|
|
"prefill_ms_p95": 335.27,
|
|
"decode_speed_tok_s_mean": 81.45,
|
|
"decode_speed_tok_s_p95": 82.19,
|
|
"e2e_ms_mean": 21179.77,
|
|
"e2e_ms_p95": 42235.91,
|
|
"total_tokens_mean": 246.67,
|
|
"total_tokens_p95": 427.0
|
|
},
|
|
"elapsed_seconds": 127.08
|
|
},
|
|
{
|
|
"case_id": "case_07_multiturn",
|
|
"case_name": "多轮对话累积",
|
|
"difficulty": "耗时",
|
|
"status": "passed",
|
|
"raw_data": {
|
|
"case_id": "case_07_multiturn",
|
|
"name": "多轮对话累积",
|
|
"description": "追踪连续多轮对话中延迟随上下文累积的变化",
|
|
"difficulty": "耗时",
|
|
"estimated_seconds": 10,
|
|
"results": [
|
|
{
|
|
"iteration": 1,
|
|
"ttft_ms": 352.84,
|
|
"prefill_ms": 352.84,
|
|
"decode_speed_tok_s": 81.88,
|
|
"total_tokens": 153,
|
|
"prompt_tokens": 33,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 7458.81,
|
|
"elapsed_ms": 7458.81,
|
|
"response_length": 0,
|
|
"round": 1,
|
|
"repeat": 1,
|
|
"scenario": "修复Python中的索引越界错误",
|
|
"messages_in_context": 2
|
|
},
|
|
{
|
|
"iteration": 2,
|
|
"ttft_ms": 570.91,
|
|
"prefill_ms": 570.91,
|
|
"decode_speed_tok_s": 83.44,
|
|
"total_tokens": 186,
|
|
"prompt_tokens": 66,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 9924.72,
|
|
"elapsed_ms": 9924.72,
|
|
"response_length": 0,
|
|
"round": 2,
|
|
"repeat": 1,
|
|
"scenario": "修复JavaScript闭包变量捕获问题",
|
|
"messages_in_context": 4
|
|
},
|
|
{
|
|
"iteration": 3,
|
|
"ttft_ms": 592.78,
|
|
"prefill_ms": 592.78,
|
|
"decode_speed_tok_s": 80.86,
|
|
"total_tokens": 216,
|
|
"prompt_tokens": 96,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 8754.91,
|
|
"elapsed_ms": 8754.91,
|
|
"response_length": 0,
|
|
"round": 3,
|
|
"repeat": 1,
|
|
"scenario": "修复SQL注入漏洞",
|
|
"messages_in_context": 6
|
|
},
|
|
{
|
|
"iteration": 4,
|
|
"ttft_ms": 468.43,
|
|
"prefill_ms": 468.43,
|
|
"decode_speed_tok_s": 83.91,
|
|
"total_tokens": 247,
|
|
"prompt_tokens": 127,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 8867.78,
|
|
"elapsed_ms": 8867.78,
|
|
"response_length": 0,
|
|
"round": 4,
|
|
"repeat": 1,
|
|
"scenario": "修复多线程竞态条件",
|
|
"messages_in_context": 8
|
|
},
|
|
{
|
|
"iteration": 5,
|
|
"ttft_ms": 472.24,
|
|
"prefill_ms": 472.24,
|
|
"decode_speed_tok_s": 81.92,
|
|
"total_tokens": 277,
|
|
"prompt_tokens": 157,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 9069.2,
|
|
"elapsed_ms": 9069.2,
|
|
"response_length": 0,
|
|
"round": 5,
|
|
"repeat": 1,
|
|
"scenario": "修复递归栈溢出",
|
|
"messages_in_context": 10
|
|
},
|
|
{
|
|
"iteration": 6,
|
|
"ttft_ms": 496.68,
|
|
"prefill_ms": 496.68,
|
|
"decode_speed_tok_s": 83.06,
|
|
"total_tokens": 308,
|
|
"prompt_tokens": 188,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 9110.29,
|
|
"elapsed_ms": 9110.29,
|
|
"response_length": 0,
|
|
"round": 6,
|
|
"repeat": 1,
|
|
"scenario": "修复正则表达式灾难回溯",
|
|
"messages_in_context": 12
|
|
},
|
|
{
|
|
"iteration": 7,
|
|
"ttft_ms": 356.02,
|
|
"prefill_ms": 356.02,
|
|
"decode_speed_tok_s": 82.34,
|
|
"total_tokens": 153,
|
|
"prompt_tokens": 33,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 9302.89,
|
|
"elapsed_ms": 9302.89,
|
|
"response_length": 0,
|
|
"round": 1,
|
|
"repeat": 2,
|
|
"scenario": "修复Python中的索引越界错误",
|
|
"messages_in_context": 2
|
|
},
|
|
{
|
|
"iteration": 8,
|
|
"ttft_ms": 105.61,
|
|
"prefill_ms": 105.61,
|
|
"decode_speed_tok_s": 80.39,
|
|
"total_tokens": 186,
|
|
"prompt_tokens": 66,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 9569.96,
|
|
"elapsed_ms": 9569.96,
|
|
"response_length": 0,
|
|
"round": 2,
|
|
"repeat": 2,
|
|
"scenario": "修复JavaScript闭包变量捕获问题",
|
|
"messages_in_context": 4
|
|
},
|
|
{
|
|
"iteration": 9,
|
|
"ttft_ms": 596.24,
|
|
"prefill_ms": 596.24,
|
|
"decode_speed_tok_s": 78.49,
|
|
"total_tokens": 216,
|
|
"prompt_tokens": 96,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 9446.57,
|
|
"elapsed_ms": 9446.57,
|
|
"response_length": 0,
|
|
"round": 3,
|
|
"repeat": 2,
|
|
"scenario": "修复SQL注入漏洞",
|
|
"messages_in_context": 6
|
|
},
|
|
{
|
|
"iteration": 10,
|
|
"ttft_ms": 468.0,
|
|
"prefill_ms": 468.0,
|
|
"decode_speed_tok_s": 81.77,
|
|
"total_tokens": 247,
|
|
"prompt_tokens": 127,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 7437.1,
|
|
"elapsed_ms": 7437.1,
|
|
"response_length": 0,
|
|
"round": 4,
|
|
"repeat": 2,
|
|
"scenario": "修复多线程竞态条件",
|
|
"messages_in_context": 8
|
|
},
|
|
{
|
|
"iteration": 11,
|
|
"ttft_ms": 470.8,
|
|
"prefill_ms": 470.8,
|
|
"decode_speed_tok_s": 81.38,
|
|
"total_tokens": 277,
|
|
"prompt_tokens": 157,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 22114.83,
|
|
"elapsed_ms": 22114.83,
|
|
"response_length": 0,
|
|
"round": 5,
|
|
"repeat": 2,
|
|
"scenario": "修复递归栈溢出",
|
|
"messages_in_context": 10
|
|
},
|
|
{
|
|
"iteration": 12,
|
|
"ttft_ms": 503.16,
|
|
"prefill_ms": 503.16,
|
|
"decode_speed_tok_s": 81.86,
|
|
"total_tokens": 308,
|
|
"prompt_tokens": 188,
|
|
"completion_tokens": 120,
|
|
"e2e_ms": 26371.35,
|
|
"elapsed_ms": 26371.35,
|
|
"response_length": 0,
|
|
"round": 6,
|
|
"repeat": 2,
|
|
"scenario": "修复正则表达式灾难回溯",
|
|
"messages_in_context": 12
|
|
}
|
|
],
|
|
"total_rounds": 12
|
|
},
|
|
"statistics": {
|
|
"count": 12,
|
|
"mean_ms": 11452.37,
|
|
"std_ms": 6090.06,
|
|
"min_ms": 7437.1,
|
|
"max_ms": 26371.35,
|
|
"p50_ms": 9206.59,
|
|
"p95_ms": 24030.26,
|
|
"p99_ms": 25903.13
|
|
},
|
|
"inference_metrics": {
|
|
"ttft_ms_mean": 454.48,
|
|
"ttft_ms_p95": 594.34,
|
|
"prefill_ms_mean": 454.48,
|
|
"prefill_ms_p95": 594.34,
|
|
"decode_speed_tok_s_mean": 81.77,
|
|
"decode_speed_tok_s_p95": 83.65,
|
|
"e2e_ms_mean": 11452.37,
|
|
"e2e_ms_p95": 24030.26,
|
|
"total_tokens_mean": 231.17,
|
|
"total_tokens_p95": 308.0
|
|
},
|
|
"elapsed_seconds": 137.43
|
|
}
|
|
]
|
|
} |