algobench-sdk 2.0.4__tar.gz → 2.0.6__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/PKG-INFO +1 -1
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/__init__.py +1 -1
- algobench_sdk-2.0.6/algobench/config/metric_keywords.py +516 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/optimization/knee_point.py +3 -3
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/parsers/csv_parser.py +68 -10
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/core.py +16 -6
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/qvalue.py +7 -1
- algobench_sdk-2.0.6/algobench/stats/sample_processing.py +150 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/tests/data_type.py +3 -1
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/tests/nonparametric.py +14 -1
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/tests/smart.py +1 -1
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench_sdk.egg-info/PKG-INFO +1 -1
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/pyproject.toml +1 -1
- algobench_sdk-2.0.4/algobench/config/metric_keywords.py +0 -231
- algobench_sdk-2.0.4/algobench/stats/sample_processing.py +0 -102
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/README.md +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/__main__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/api.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/cli.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/config/__init__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/config/analysis_criteria.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/config/business.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/config/metrics.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/config/thresholds.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/exceptions.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/models.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/optimization/__init__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/optimization/decision_tree.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/optimization/feature_importance.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/optimization/response_surface.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/parsers/__init__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/__init__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/diagnosis.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/tests/__init__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/tests/effect_size.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/stats/tests/t_tests.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/utils/__init__.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/utils/improvement.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/utils/math_utils.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench/utils/numbers.py +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench_sdk.egg-info/SOURCES.txt +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench_sdk.egg-info/dependency_links.txt +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench_sdk.egg-info/requires.txt +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/algobench_sdk.egg-info/top_level.txt +0 -0
- {algobench_sdk-2.0.4 → algobench_sdk-2.0.6}/setup.cfg +0 -0
|
@@ -0,0 +1,516 @@
|
|
|
1
|
+
"""指标关键词库，对应 JS 的 src/config/metricKeywords.js。
|
|
2
|
+
|
|
3
|
+
由 JS keyword 文件自动生成，确保与前端 DEFAULT_KEYWORD_LIBRARY 1:1 对齐。
|
|
4
|
+
包含 ML_KEYWORDS + FINANCE_KEYWORDS + EDA_KEYWORDS + GENERAL_KEYWORDS。
|
|
5
|
+
"""
|
|
6
|
+
|
|
7
|
+
from __future__ import annotations
|
|
8
|
+
|
|
9
|
+
import re
|
|
10
|
+
|
|
11
|
+
# 默认关键词库（与 JS DEFAULT_KEYWORD_LIBRARY 完全一致）
|
|
12
|
+
# 顺序：ML → Finance → EDA → General（与 JS metricKeywords.js 合并顺序一致）
|
|
13
|
+
DEFAULT_KEYWORD_LIBRARY: dict[str, dict] = {
|
|
14
|
+
|
|
15
|
+
# ===== 机器学习指标 (ML_KEYWORDS) =====
|
|
16
|
+
"accuracy": {"better": "higher", "description": "准确率"},
|
|
17
|
+
"precision": {"better": "higher", "description": "精确率"},
|
|
18
|
+
"recall": {"better": "higher", "description": "召回率"},
|
|
19
|
+
"f1": {"better": "higher", "description": "F1分数"},
|
|
20
|
+
"auc": {"better": "higher", "description": "AUC值"},
|
|
21
|
+
"map": {"better": "higher", "description": "平均精度均值"},
|
|
22
|
+
"iou": {"better": "higher", "description": "交并比"},
|
|
23
|
+
"dice": {"better": "higher", "description": "Dice系数"},
|
|
24
|
+
"bleu": {"better": "higher", "description": "BLEU分数"},
|
|
25
|
+
"rouge": {"better": "higher", "description": "ROUGE分数"},
|
|
26
|
+
"perplexity": {"better": "lower", "description": "困惑度"},
|
|
27
|
+
"success": {"better": "higher", "description": "成功"},
|
|
28
|
+
"hit": {"better": "higher", "description": "命中"},
|
|
29
|
+
"win": {"better": "higher", "description": "胜利"},
|
|
30
|
+
"pass": {"better": "higher", "description": "通过"},
|
|
31
|
+
"correct": {"better": "higher", "description": "正确"},
|
|
32
|
+
"valid": {"better": "higher", "description": "有效"},
|
|
33
|
+
"coverage": {"better": "higher", "description": "覆盖率"},
|
|
34
|
+
"yield": {"better": "higher", "description": "产出率"},
|
|
35
|
+
"utilization": {"better": "higher", "description": "利用率"},
|
|
36
|
+
"occupancy": {"better": "higher", "description": "占用率"},
|
|
37
|
+
"availability": {"better": "higher", "description": "可用性"},
|
|
38
|
+
"reliability": {"better": "higher", "description": "稳定性"},
|
|
39
|
+
"stability": {"better": "higher", "description": "稳定性"},
|
|
40
|
+
"loss": {"better": "lower", "description": "损失"},
|
|
41
|
+
"error": {"better": "lower", "description": "错误"},
|
|
42
|
+
"mistake": {"better": "lower", "description": "错误"},
|
|
43
|
+
"fail": {"better": "lower", "description": "失败"},
|
|
44
|
+
"failure": {"better": "lower", "description": "失败"},
|
|
45
|
+
"miss": {"better": "lower", "description": "未命中"},
|
|
46
|
+
"reject": {"better": "lower", "description": "拒绝"},
|
|
47
|
+
"drop": {"better": "lower", "description": "丢弃"},
|
|
48
|
+
"train_loss": {"better": "lower", "description": "训练损失"},
|
|
49
|
+
"val_loss": {"better": "lower", "description": "验证损失"},
|
|
50
|
+
"test_loss": {"better": "lower", "description": "测试损失"},
|
|
51
|
+
"train_error": {"better": "lower", "description": "训练错误率"},
|
|
52
|
+
"val_error": {"better": "lower", "description": "验证错误率"},
|
|
53
|
+
"test_error": {"better": "lower", "description": "测试错误率"},
|
|
54
|
+
"train_acc": {"better": "higher", "description": "训练准确率"},
|
|
55
|
+
"val_acc": {"better": "higher", "description": "验证准确率"},
|
|
56
|
+
"test_acc": {"better": "higher", "description": "测试准确率"},
|
|
57
|
+
"top1": {"better": "higher", "description": "Top-1 准确率"},
|
|
58
|
+
"top5": {"better": "higher", "description": "Top-5 准确率"},
|
|
59
|
+
"top1_acc": {"better": "higher", "description": "Top-1 准确率"},
|
|
60
|
+
"top5_acc": {"better": "higher", "description": "Top-5 准确率"},
|
|
61
|
+
"wer": {"better": "lower", "description": "词错误率"},
|
|
62
|
+
"cer": {"better": "lower", "description": "字符错误率"},
|
|
63
|
+
"ter": {"better": "lower", "description": "翻译错误率"},
|
|
64
|
+
"mer": {"better": "lower", "description": "匹配错误率"},
|
|
65
|
+
"token_error": {"better": "lower", "description": "Token 错误率"},
|
|
66
|
+
"inference_time": {"better": "lower", "description": "推理时间"},
|
|
67
|
+
"inferencetime": {"better": "lower", "description": "推理时间"},
|
|
68
|
+
"inference_latency": {"better": "lower", "description": "推理延迟"},
|
|
69
|
+
"tokens_per_second": {"better": "higher", "description": "每秒Token数"},
|
|
70
|
+
"tokenspersecond": {"better": "higher", "description": "每秒Token数"},
|
|
71
|
+
"tps": {"better": "higher", "description": "每秒Token数"},
|
|
72
|
+
"reward": {"better": "higher", "description": "奖励"},
|
|
73
|
+
"return": {"better": "higher", "description": "回报"},
|
|
74
|
+
"score": {"better": "higher", "description": "得分"},
|
|
75
|
+
"episode_reward": {"better": "higher", "description": "回合奖励"},
|
|
76
|
+
"cumulative_reward": {"better": "higher", "description": "累积奖励"},
|
|
77
|
+
|
|
78
|
+
# ===== 金融指标 (FINANCE_KEYWORDS) =====
|
|
79
|
+
"alpha_cagr_vs_benchmark": {"better": "higher", "unit": "%", "description": "相对基准年化超额收益", "impDisplayMode": "percent"},
|
|
80
|
+
"max_drawdown_duration_days": {"better": "lower", "description": "最大回撤持续天数", "impDisplayMode": "percent"},
|
|
81
|
+
"annual_volatility": {"better": "lower", "unit": "%", "description": "年化波动率", "impDisplayMode": "percent"},
|
|
82
|
+
"max_drawdown": {"better": "higher", "unit": "%", "description": "最大回撤（负数口径时越高越好）", "impDisplayMode": "percent"},
|
|
83
|
+
"monthly_win_rate": {"better": "higher", "unit": "%", "description": "月度胜率", "impDisplayMode": "percent"},
|
|
84
|
+
"information_ratio": {"better": "higher", "description": "信息比率", "impDisplayMode": "percent"},
|
|
85
|
+
"cost_ratio": {"better": "lower", "unit": "%", "description": "成本率", "impDisplayMode": "percent"},
|
|
86
|
+
"total_return": {"better": "higher", "unit": "%", "description": "总收益率", "impDisplayMode": "percent"},
|
|
87
|
+
"volatility": {"better": "lower", "unit": "%", "description": "波动率", "impDisplayMode": "percent"},
|
|
88
|
+
"drawdown": {"better": "lower", "description": "回撤（正数损失口径时越低越好）", "impDisplayMode": "percent"},
|
|
89
|
+
"turnover": {"better": "lower", "unit": "%", "description": "换手率", "impDisplayMode": "percent"},
|
|
90
|
+
"sortino": {"better": "higher", "description": "Sortino比率", "impDisplayMode": "percent"},
|
|
91
|
+
"sharpe": {"better": "higher", "description": "Sharpe比率", "impDisplayMode": "percent"},
|
|
92
|
+
"calmar": {"better": "higher", "description": "Calmar比率", "impDisplayMode": "percent"},
|
|
93
|
+
"alpha": {"better": "higher", "description": "超额收益", "impDisplayMode": "percent"},
|
|
94
|
+
"cagr": {"better": "higher", "unit": "%", "description": "年化收益率", "impDisplayMode": "percent"},
|
|
95
|
+
"exposure": {"better": "higher", "description": "暴露度", "impDisplayMode": "percent"},
|
|
96
|
+
"final_equity": {"better": "higher", "description": "最终权益", "impDisplayMode": "percent"},
|
|
97
|
+
|
|
98
|
+
# ===== EDA / 芯片设计指标 (EDA_KEYWORDS) =====
|
|
99
|
+
"congestion": {"better": "lower", "description": "拥塞度"},
|
|
100
|
+
"hpwl": {"better": "lower", "description": "半周长线长"},
|
|
101
|
+
"wirelength": {"better": "lower", "description": "线长"},
|
|
102
|
+
"wl": {"better": "lower", "description": "线长"},
|
|
103
|
+
"wire": {"better": "lower", "description": "线长"},
|
|
104
|
+
"drc": {"better": "lower", "description": "设计规则违规"},
|
|
105
|
+
"density": {"better": "lower", "description": "密度"},
|
|
106
|
+
"slack": {"better": "higher", "description": "时序裕量"},
|
|
107
|
+
"wns": {"better": "higher", "description": "最差负裕量"},
|
|
108
|
+
"tns": {"better": "higher", "description": "总负裕量"},
|
|
109
|
+
"d2d": {"better": "lower", "description": "芯片间互联数量"},
|
|
110
|
+
"d2dnum": {"better": "lower", "description": "芯片间互联数量"},
|
|
111
|
+
"hb": {"better": "lower", "description": "混合键合数量"},
|
|
112
|
+
"hybrid bonding": {"better": "lower", "description": "混合键合数量"},
|
|
113
|
+
"hybridbonding": {"better": "lower", "description": "混合键合数量"},
|
|
114
|
+
"gp_tns": {"better": "higher", "description": "全局布局后总负裕量"},
|
|
115
|
+
"gp_wns": {"better": "higher", "description": "全局布局后最差负裕量"},
|
|
116
|
+
"gp_slack": {"better": "higher", "description": "全局布局后时序裕量"},
|
|
117
|
+
"gp_power": {"better": "lower", "description": "全局布局后功耗"},
|
|
118
|
+
"gp_totalpower": {"better": "lower", "description": "全局布局后总功耗"},
|
|
119
|
+
"gp_dynamicpower": {"better": "lower", "description": "全局布局后动态功耗"},
|
|
120
|
+
"gp_leakagepower": {"better": "lower", "description": "全局布局后漏电功耗"},
|
|
121
|
+
"gp_area": {"better": "lower", "description": "全局布局后面积"},
|
|
122
|
+
"gp_utilization": {"better": "lower", "description": "全局布局后利用率"},
|
|
123
|
+
"gp_congestion": {"better": "lower", "description": "全局布局后拥塞度"},
|
|
124
|
+
"gp_overflow": {"better": "lower", "description": "全局布局后溢出"},
|
|
125
|
+
"gp_hpwl": {"better": "lower", "description": "全局布局后半周长线长"},
|
|
126
|
+
"gp_wirelength": {"better": "lower", "description": "全局布局后线长"},
|
|
127
|
+
"gp_density": {"better": "lower", "description": "全局布局后密度"},
|
|
128
|
+
"dp_tns": {"better": "higher", "description": "详细布局后总负裕量"},
|
|
129
|
+
"dp_wns": {"better": "higher", "description": "详细布局后最差负裕量"},
|
|
130
|
+
"dp_slack": {"better": "higher", "description": "详细布局后时序裕量"},
|
|
131
|
+
"dp_power": {"better": "lower", "description": "详细布局后功耗"},
|
|
132
|
+
"dp_totalpower": {"better": "lower", "description": "详细布局后总功耗"},
|
|
133
|
+
"dp_dynamicpower": {"better": "lower", "description": "详细布局后动态功耗"},
|
|
134
|
+
"dp_leakagepower": {"better": "lower", "description": "详细布局后漏电功耗"},
|
|
135
|
+
"dp_area": {"better": "lower", "description": "详细布局后面积"},
|
|
136
|
+
"dp_utilization": {"better": "lower", "description": "详细布局后利用率"},
|
|
137
|
+
"dp_congestion": {"better": "lower", "description": "详细布局后拥塞度"},
|
|
138
|
+
"dp_overflow": {"better": "lower", "description": "详细布局后溢出"},
|
|
139
|
+
"dp_hpwl": {"better": "lower", "description": "详细布局后半周长线长"},
|
|
140
|
+
"dp_wirelength": {"better": "lower", "description": "详细布局后线长"},
|
|
141
|
+
"dp_density": {"better": "lower", "description": "详细布局后密度"},
|
|
142
|
+
"cts_tns": {"better": "higher", "description": "时钟树综合后总负裕量"},
|
|
143
|
+
"cts_wns": {"better": "higher", "description": "时钟树综合后最差负裕量"},
|
|
144
|
+
"cts_slack": {"better": "higher", "description": "时钟树综合后时序裕量"},
|
|
145
|
+
"cts_power": {"better": "lower", "description": "时钟树综合后功耗"},
|
|
146
|
+
"cts_totalpower": {"better": "lower", "description": "时钟树综合后总功耗"},
|
|
147
|
+
"cts_dynamicpower": {"better": "lower", "description": "时钟树综合后动态功耗"},
|
|
148
|
+
"cts_leakagepower": {"better": "lower", "description": "时钟树综合后漏电功耗"},
|
|
149
|
+
"cts_area": {"better": "lower", "description": "时钟树综合后面积"},
|
|
150
|
+
"cts_utilization": {"better": "lower", "description": "时钟树综合后利用率"},
|
|
151
|
+
"cts_skew": {"better": "lower", "description": "时钟树综合后时钟偏斜"},
|
|
152
|
+
"cts_latency": {"better": "lower", "description": "时钟树综合后时钟延迟"},
|
|
153
|
+
"route_tns": {"better": "higher", "description": "绕线后总负裕量"},
|
|
154
|
+
"route_wns": {"better": "higher", "description": "绕线后最差负裕量"},
|
|
155
|
+
"route_slack": {"better": "higher", "description": "绕线后时序裕量"},
|
|
156
|
+
"route_power": {"better": "lower", "description": "绕线后功耗"},
|
|
157
|
+
"route_totalpower": {"better": "lower", "description": "绕线后总功耗"},
|
|
158
|
+
"route_dynamicpower": {"better": "lower", "description": "绕线后动态功耗"},
|
|
159
|
+
"route_leakagepower": {"better": "lower", "description": "绕线后漏电功耗"},
|
|
160
|
+
"route_area": {"better": "lower", "description": "绕线后面积"},
|
|
161
|
+
"route_utilization": {"better": "lower", "description": "绕线后利用率"},
|
|
162
|
+
"route_wirelength": {"better": "lower", "description": "绕线后线长"},
|
|
163
|
+
"route_hpwl": {"better": "lower", "description": "绕线后半周长线长"},
|
|
164
|
+
"route_drc": {"better": "lower", "description": "绕线后设计规则违规"},
|
|
165
|
+
"route_violation": {"better": "lower", "description": "绕线后违规数量"},
|
|
166
|
+
"route_short": {"better": "lower", "description": "绕线后短路数量"},
|
|
167
|
+
"route_open": {"better": "lower", "description": "绕线后开路数量"},
|
|
168
|
+
"postroute_tns": {"better": "higher", "description": "绕线优化后总负裕量"},
|
|
169
|
+
"postroute_wns": {"better": "higher", "description": "绕线优化后最差负裕量"},
|
|
170
|
+
"postroute_slack": {"better": "higher", "description": "绕线优化后时序裕量"},
|
|
171
|
+
"postroute_power": {"better": "lower", "description": "绕线优化后功耗"},
|
|
172
|
+
"postroute_totalpower": {"better": "lower", "description": "绕线优化后总功耗"},
|
|
173
|
+
"postroute_dynamicpower": {"better": "lower", "description": "绕线优化后动态功耗"},
|
|
174
|
+
"postroute_leakagepower": {"better": "lower", "description": "绕线优化后漏电功耗"},
|
|
175
|
+
"postroute_area": {"better": "lower", "description": "绕线优化后面积"},
|
|
176
|
+
"postroute_utilization": {"better": "lower", "description": "绕线优化后利用率"},
|
|
177
|
+
"postroute_wirelength": {"better": "lower", "description": "绕线优化后线长"},
|
|
178
|
+
"postroute_hpwl": {"better": "lower", "description": "绕线优化后半周长线长"},
|
|
179
|
+
"postroute_drc": {"better": "lower", "description": "绕线优化后设计规则违规"},
|
|
180
|
+
"postroute_violation": {"better": "lower", "description": "绕线优化后违规数量"},
|
|
181
|
+
"postroute_ir": {"better": "lower", "description": "绕线优化后电压降"},
|
|
182
|
+
"postroute_irdrop": {"better": "lower", "description": "绕线优化后电压降"},
|
|
183
|
+
"setup": {"better": "higher", "description": "建立时间裕量"},
|
|
184
|
+
"hold": {"better": "higher", "description": "保持时间裕量"},
|
|
185
|
+
"setup_time": {"better": "higher", "description": "建立时间裕量"},
|
|
186
|
+
"hold_time": {"better": "higher", "description": "保持时间裕量"},
|
|
187
|
+
"setuptime": {"better": "higher", "description": "建立时间裕量"},
|
|
188
|
+
"holdtime": {"better": "higher", "description": "保持时间裕量"},
|
|
189
|
+
"skew": {"better": "lower", "description": "时钟偏斜"},
|
|
190
|
+
"ir_drop": {"better": "lower", "description": "电压降"},
|
|
191
|
+
"irdrop": {"better": "lower", "description": "电压降"},
|
|
192
|
+
"ir": {"better": "lower", "description": "电压降"},
|
|
193
|
+
"signal integrity": {"better": "higher", "description": "信号完整性"},
|
|
194
|
+
"si": {"better": "higher", "description": "信号完整性"},
|
|
195
|
+
"frequency": {"better": "target", "description": "频率"},
|
|
196
|
+
"voltage": {"better": "target", "description": "电压"},
|
|
197
|
+
"target": {"better": "target", "description": "目标值"},
|
|
198
|
+
|
|
199
|
+
# ===== 通用指标 (GENERAL_KEYWORDS) =====
|
|
200
|
+
"score": {"better": "higher", "description": "得分"},
|
|
201
|
+
"rating": {"better": "higher", "description": "评分"},
|
|
202
|
+
"rank": {"better": "lower", "description": "排名"},
|
|
203
|
+
"grade": {"better": "higher", "description": "等级"},
|
|
204
|
+
"level": {"better": "higher", "description": "级别"},
|
|
205
|
+
"语文": {"better": "higher", "description": "语文成绩"},
|
|
206
|
+
"chinese": {"better": "higher", "description": "语文成绩"},
|
|
207
|
+
"数学": {"better": "higher", "description": "数学成绩"},
|
|
208
|
+
"math": {"better": "higher", "description": "数学成绩"},
|
|
209
|
+
"mathematics": {"better": "higher", "description": "数学成绩"},
|
|
210
|
+
"英语": {"better": "higher", "description": "英语成绩"},
|
|
211
|
+
"english": {"better": "higher", "description": "英语成绩"},
|
|
212
|
+
"物理": {"better": "higher", "description": "物理成绩"},
|
|
213
|
+
"physics": {"better": "higher", "description": "物理成绩"},
|
|
214
|
+
"化学": {"better": "higher", "description": "化学成绩"},
|
|
215
|
+
"chemistry": {"better": "higher", "description": "化学成绩"},
|
|
216
|
+
"生物": {"better": "higher", "description": "生物成绩"},
|
|
217
|
+
"biology": {"better": "higher", "description": "生物成绩"},
|
|
218
|
+
"历史": {"better": "higher", "description": "历史成绩"},
|
|
219
|
+
"history": {"better": "higher", "description": "历史成绩"},
|
|
220
|
+
"地理": {"better": "higher", "description": "地理成绩"},
|
|
221
|
+
"geography": {"better": "higher", "description": "地理成绩"},
|
|
222
|
+
"政治": {"better": "higher", "description": "政治成绩"},
|
|
223
|
+
"politics": {"better": "higher", "description": "政治成绩"},
|
|
224
|
+
"政治学": {"better": "higher", "description": "政治成绩"},
|
|
225
|
+
"science": {"better": "higher", "description": "科学成绩"},
|
|
226
|
+
"理科": {"better": "higher", "description": "理科成绩"},
|
|
227
|
+
"文科": {"better": "higher", "description": "文科成绩"},
|
|
228
|
+
"总分": {"better": "higher", "description": "总分"},
|
|
229
|
+
"total": {"better": "higher", "description": "总分"},
|
|
230
|
+
"综合": {"better": "higher", "description": "综合成绩"},
|
|
231
|
+
"文综": {"better": "higher", "description": "文科综合"},
|
|
232
|
+
"理综": {"better": "higher", "description": "理科综合"},
|
|
233
|
+
"体育": {"better": "higher", "description": "体育成绩"},
|
|
234
|
+
"音乐": {"better": "higher", "description": "音乐成绩"},
|
|
235
|
+
"music": {"better": "higher", "description": "音乐成绩"},
|
|
236
|
+
"美术": {"better": "higher", "description": "美术成绩"},
|
|
237
|
+
"计算机": {"better": "higher", "description": "计算机成绩"},
|
|
238
|
+
"computer": {"better": "higher", "description": "计算机成绩"},
|
|
239
|
+
"信息": {"better": "higher", "description": "信息技术成绩"},
|
|
240
|
+
"throughput": {"better": "higher", "description": "吞吐量"},
|
|
241
|
+
"fps": {"better": "higher", "description": "帧率"},
|
|
242
|
+
"qps": {"better": "higher", "description": "每秒查询数"},
|
|
243
|
+
"tps": {"better": "higher", "description": "每秒事务数"},
|
|
244
|
+
"rps": {"better": "higher", "description": "每秒请求数"},
|
|
245
|
+
"iops": {"better": "higher", "description": "每秒IO操作"},
|
|
246
|
+
"bandwidth": {"better": "higher", "description": "带宽"},
|
|
247
|
+
"speedup": {"better": "higher", "description": "加速比"},
|
|
248
|
+
"speed": {"better": "higher", "description": "速度"},
|
|
249
|
+
"rate": {"better": "higher", "description": "速率"},
|
|
250
|
+
"efficiency": {"better": "higher", "description": "效率"},
|
|
251
|
+
"quality": {"better": "higher", "description": "质量"},
|
|
252
|
+
"performance": {"better": "higher", "description": "性能"},
|
|
253
|
+
"improvement": {"better": "higher", "description": "改进"},
|
|
254
|
+
"optimization": {"better": "higher", "description": "优化"},
|
|
255
|
+
"gain": {"better": "higher", "description": "增益"},
|
|
256
|
+
"benefit": {"better": "higher", "description": "收益"},
|
|
257
|
+
"profit": {"better": "higher", "description": "利润"},
|
|
258
|
+
"reward": {"better": "higher", "description": "奖励"},
|
|
259
|
+
"value": {"better": "higher", "description": "价值"},
|
|
260
|
+
"cost": {"better": "lower", "description": "成本"},
|
|
261
|
+
"expense": {"better": "lower", "description": "费用"},
|
|
262
|
+
"price": {"better": "lower", "description": "价格"},
|
|
263
|
+
"overhead": {"better": "lower", "description": "开销"},
|
|
264
|
+
"penalty": {"better": "lower", "description": "惩罚"},
|
|
265
|
+
"waste": {"better": "lower", "description": "浪费"},
|
|
266
|
+
"consumption": {"better": "lower", "description": "消耗"},
|
|
267
|
+
"runtime": {"better": "lower", "description": "运行时间"},
|
|
268
|
+
"time": {"better": "lower", "description": "时间"},
|
|
269
|
+
"latency": {"better": "lower", "description": "延迟"},
|
|
270
|
+
"delay": {"better": "lower", "description": "延迟"},
|
|
271
|
+
"wait": {"better": "lower", "description": "等待"},
|
|
272
|
+
"duration": {"better": "lower", "description": "持续时间"},
|
|
273
|
+
"elapsed": {"better": "lower", "description": "耗时"},
|
|
274
|
+
"timeout": {"better": "lower", "description": "超时"},
|
|
275
|
+
"memory": {"better": "lower", "description": "内存"},
|
|
276
|
+
"cpu": {"better": "lower", "description": "CPU使用"},
|
|
277
|
+
"gpu": {"better": "lower", "description": "GPU使用"},
|
|
278
|
+
"disk": {"better": "lower", "description": "磁盘"},
|
|
279
|
+
"storage": {"better": "lower", "description": "存储"},
|
|
280
|
+
"cache": {"better": "lower", "description": "缓存"},
|
|
281
|
+
"buffer": {"better": "lower", "description": "缓冲"},
|
|
282
|
+
"heap": {"better": "lower", "description": "堆"},
|
|
283
|
+
"stack": {"better": "lower", "description": "栈"},
|
|
284
|
+
"power": {"better": "lower", "description": "功率"},
|
|
285
|
+
"energy": {"better": "lower", "description": "能量"},
|
|
286
|
+
"thermal": {"better": "lower", "description": "热量"},
|
|
287
|
+
"temperature": {"better": "lower", "description": "温度"},
|
|
288
|
+
"heat": {"better": "lower", "description": "热量"},
|
|
289
|
+
"distance": {"better": "lower", "description": "距离"},
|
|
290
|
+
"deviation": {"better": "lower", "description": "偏差"},
|
|
291
|
+
"variance": {"better": "lower", "description": "方差"},
|
|
292
|
+
"std": {"better": "lower", "description": "标准差"},
|
|
293
|
+
"rmse": {"better": "lower", "description": "均方根误差"},
|
|
294
|
+
"mse": {"better": "lower", "description": "均方误差"},
|
|
295
|
+
"mae": {"better": "lower", "description": "平均绝对误差"},
|
|
296
|
+
"mape": {"better": "lower", "description": "平均绝对百分比误差"},
|
|
297
|
+
"count": {"better": "lower", "description": "计数"},
|
|
298
|
+
"size": {"better": "lower", "description": "大小"},
|
|
299
|
+
"length": {"better": "lower", "description": "长度"},
|
|
300
|
+
"width": {"better": "lower", "description": "宽度"},
|
|
301
|
+
"height": {"better": "lower", "description": "高度"},
|
|
302
|
+
"area": {"better": "lower", "description": "面积"},
|
|
303
|
+
"volume": {"better": "lower", "description": "体积"},
|
|
304
|
+
"weight": {"better": "lower", "description": "权重"},
|
|
305
|
+
"load": {"better": "lower", "description": "负载"},
|
|
306
|
+
"usage": {"better": "lower", "description": "使用量"},
|
|
307
|
+
"traffic": {"better": "lower", "description": "流量"},
|
|
308
|
+
"request": {"better": "lower", "description": "请求"},
|
|
309
|
+
"connection": {"better": "lower", "description": "连接"},
|
|
310
|
+
"session": {"better": "lower", "description": "会话"},
|
|
311
|
+
"overflow": {"better": "lower", "description": "溢出"},
|
|
312
|
+
"violation": {"better": "lower", "description": "违规"},
|
|
313
|
+
"conflict": {"better": "lower", "description": "冲突"},
|
|
314
|
+
"collision": {"better": "lower", "description": "碰撞"},
|
|
315
|
+
"deadlock": {"better": "lower", "description": "死锁"},
|
|
316
|
+
"retry": {"better": "lower", "description": "重试"},
|
|
317
|
+
"rollback": {"better": "lower", "description": "回滚"},
|
|
318
|
+
"p50": {"better": "lower", "description": "P50 延迟"},
|
|
319
|
+
"p90": {"better": "lower", "description": "P90 延迟"},
|
|
320
|
+
"p95": {"better": "lower", "description": "P95 延迟"},
|
|
321
|
+
"p99": {"better": "lower", "description": "P99 延迟"},
|
|
322
|
+
"p999": {"better": "lower", "description": "P99.9 延迟"},
|
|
323
|
+
"percentile": {"better": "lower", "description": "百分位延迟"},
|
|
324
|
+
"tail_latency": {"better": "lower", "description": "尾延迟"},
|
|
325
|
+
"taillatency": {"better": "lower", "description": "尾延迟"},
|
|
326
|
+
"packet_loss": {"better": "lower", "description": "丢包率"},
|
|
327
|
+
"packetloss": {"better": "lower", "description": "丢包率"},
|
|
328
|
+
"jitter": {"better": "lower", "description": "抖动"},
|
|
329
|
+
"ping": {"better": "lower", "description": "Ping延迟"},
|
|
330
|
+
"rtt": {"better": "lower", "description": "往返时延"},
|
|
331
|
+
"retransmit": {"better": "lower", "description": "重传"},
|
|
332
|
+
"snr": {"better": "higher", "description": "信噪比"},
|
|
333
|
+
"psnr": {"better": "higher", "description": "峰值信噪比"},
|
|
334
|
+
"ssim": {"better": "higher", "description": "结构相似度"},
|
|
335
|
+
"nmi": {"better": "higher", "description": "归一化互信息"},
|
|
336
|
+
"correlation": {"better": "higher", "description": "相关性"},
|
|
337
|
+
"wps": {"better": "higher", "description": "每秒写操作"},
|
|
338
|
+
"ops": {"better": "higher", "description": "每秒操作数"},
|
|
339
|
+
"eps": {"better": "higher", "description": "每秒事件数"},
|
|
340
|
+
"lps": {"better": "higher", "description": "每秒行数"},
|
|
341
|
+
"fragmentation": {"better": "lower", "description": "碎片率"},
|
|
342
|
+
"swap": {"better": "lower", "description": "交换区使用"},
|
|
343
|
+
"gc": {"better": "lower", "description": "GC暂停"},
|
|
344
|
+
"gc_pause": {"better": "lower", "description": "GC暂停时间"},
|
|
345
|
+
"gc_count": {"better": "lower", "description": "GC次数"},
|
|
346
|
+
|
|
347
|
+
}
|
|
348
|
+
|
|
349
|
+
|
|
350
|
+
def _build_pattern(keyword: str) -> re.Pattern:
|
|
351
|
+
escaped = re.escape(keyword).replace(r"\?", ".?")
|
|
352
|
+
return re.compile(rf"({escaped})", re.IGNORECASE)
|
|
353
|
+
|
|
354
|
+
|
|
355
|
+
def _matches_with_boundary(name_lower: str, keyword: str) -> bool:
|
|
356
|
+
"""边界匹配：关键词前后必须是非字母数字字符或字符串边界。
|
|
357
|
+
|
|
358
|
+
对应 JS matchesWithBoundary：'area' 匹配 'chip_area' 但不匹配 'hardware'。
|
|
359
|
+
"""
|
|
360
|
+
if not keyword:
|
|
361
|
+
return False
|
|
362
|
+
pattern = re.compile(
|
|
363
|
+
r"(?<![a-z0-9])" + re.escape(keyword) + r"(?![a-z0-9])",
|
|
364
|
+
re.IGNORECASE,
|
|
365
|
+
)
|
|
366
|
+
return bool(pattern.search(name_lower))
|
|
367
|
+
|
|
368
|
+
|
|
369
|
+
def match_keyword(metric_name: str, library: dict | None = None) -> dict | None:
|
|
370
|
+
"""在关键词库中匹配指标名称，返回匹配到的配置或 None。
|
|
371
|
+
|
|
372
|
+
优先匹配最长关键词，避免短关键词抢先匹配。
|
|
373
|
+
使用边界匹配（与 JS matchesWithBoundary 一致），确保 'area' 不误匹配 'hardware'。
|
|
374
|
+
"""
|
|
375
|
+
if not metric_name or not isinstance(metric_name, str):
|
|
376
|
+
return None
|
|
377
|
+
lib = library or DEFAULT_KEYWORD_LIBRARY
|
|
378
|
+
name_lower = metric_name.lower()
|
|
379
|
+
# 按关键词长度降序排序，优先匹配最长关键词（与 JS matchKeyword 一致）
|
|
380
|
+
sorted_keywords = sorted(lib.items(), key=lambda x: len(x[0]), reverse=True)
|
|
381
|
+
for keyword, config in sorted_keywords:
|
|
382
|
+
if _matches_with_boundary(name_lower, keyword):
|
|
383
|
+
return {"keyword": keyword, **config}
|
|
384
|
+
return None
|
|
385
|
+
|
|
386
|
+
|
|
387
|
+
# 单位提取模式（按优先级排序）
|
|
388
|
+
_UNIT_PATTERNS = [
|
|
389
|
+
(re.compile(r"\(([^)]+)\)$"), lambda m: m.group(1)), # power(W) → W
|
|
390
|
+
(re.compile(r"\[([^\]]+)\]$"), lambda m: m.group(1)), # latency[ms] → ms
|
|
391
|
+
(re.compile(r"_([a-zμ²³°/·]+)$", re.IGNORECASE), lambda m: m.group(1)), # latency_ms → ms
|
|
392
|
+
(re.compile(r"\s*[-—]\s*([a-zμ²³°]+)$", re.IGNORECASE), lambda m: m.group(1)), # power - W → W
|
|
393
|
+
]
|
|
394
|
+
|
|
395
|
+
# 已知单位白名单
|
|
396
|
+
_UNIT_WHITELIST = re.compile(
|
|
397
|
+
r"^(ms|s|us|ns|w|kw|mw|v|a|db|hz|khz|mhz|ghz|b|kb|mb|gb|tb|%|fps|rpm|°c|um|um²|nm|mm|cm|m|j|kj|cal)$",
|
|
398
|
+
re.IGNORECASE,
|
|
399
|
+
)
|
|
400
|
+
|
|
401
|
+
|
|
402
|
+
def extract_unit_from_name(metric_name: str) -> str:
|
|
403
|
+
"""从指标名称中提取单位。
|
|
404
|
+
|
|
405
|
+
>>> extract_unit_from_name("power(W)")
|
|
406
|
+
'W'
|
|
407
|
+
>>> extract_unit_from_name("latency_ms")
|
|
408
|
+
'ms'
|
|
409
|
+
>>> extract_unit_from_name("area")
|
|
410
|
+
''
|
|
411
|
+
"""
|
|
412
|
+
if not metric_name or not isinstance(metric_name, str):
|
|
413
|
+
return ""
|
|
414
|
+
for pattern, extract in _UNIT_PATTERNS:
|
|
415
|
+
m = pattern.search(metric_name)
|
|
416
|
+
if m:
|
|
417
|
+
unit = extract(m).strip()
|
|
418
|
+
if _UNIT_WHITELIST.match(unit):
|
|
419
|
+
return unit
|
|
420
|
+
return ""
|
|
421
|
+
|
|
422
|
+
|
|
423
|
+
# 后缀方向推断规则（关键词库未匹配时的兜底）
|
|
424
|
+
_SUFFIX_DIRECTIONS = [
|
|
425
|
+
{
|
|
426
|
+
"better": "lower",
|
|
427
|
+
"suffixes": [
|
|
428
|
+
"time", "latency", "delay", "duration", "runtime", "elapsed",
|
|
429
|
+
"cost", "price", "expense",
|
|
430
|
+
"error", "err", "loss", "fail", "failure",
|
|
431
|
+
"power", "energy", "consumption",
|
|
432
|
+
"area", "size", "memory", "mem", "footprint",
|
|
433
|
+
"violations", "violation", "overflow", "congestion", "density",
|
|
434
|
+
"iterations", "iteration", "count",
|
|
435
|
+
],
|
|
436
|
+
},
|
|
437
|
+
{
|
|
438
|
+
"better": "higher",
|
|
439
|
+
"suffixes": [
|
|
440
|
+
"accuracy", "acc", "precision", "recall", "f1", "auc", "iou",
|
|
441
|
+
"score", "rating", "gain", "profit", "reward", "yield",
|
|
442
|
+
"throughput", "tps", "qps", "fps", "iops", "bandwidth",
|
|
443
|
+
"util", "utilization", "coverage", "efficiency", "quality",
|
|
444
|
+
"speedup", "speed", "improvement",
|
|
445
|
+
"slack", "wns", "tns",
|
|
446
|
+
],
|
|
447
|
+
},
|
|
448
|
+
]
|
|
449
|
+
|
|
450
|
+
|
|
451
|
+
def infer_direction_by_suffix(metric_name: str) -> dict | None:
|
|
452
|
+
"""基于列名后缀推断方向（关键词库未匹配时的兜底）。
|
|
453
|
+
|
|
454
|
+
返回 {"better": "lower"|"higher", "matched_suffix": str} 或 None。
|
|
455
|
+
"""
|
|
456
|
+
if not metric_name or not isinstance(metric_name, str):
|
|
457
|
+
return None
|
|
458
|
+
name_lower = metric_name.lower()
|
|
459
|
+
# 按长度降序匹配，优先匹配更长的后缀
|
|
460
|
+
all_suffixes = sorted(
|
|
461
|
+
[{"suffix": s, "better": g["better"]} for g in _SUFFIX_DIRECTIONS for s in g["suffixes"]],
|
|
462
|
+
key=lambda x: len(x["suffix"]),
|
|
463
|
+
reverse=True,
|
|
464
|
+
)
|
|
465
|
+
for item in all_suffixes:
|
|
466
|
+
suffix = item["suffix"]
|
|
467
|
+
idx = name_lower.rfind(suffix)
|
|
468
|
+
if idx == -1:
|
|
469
|
+
continue
|
|
470
|
+
if idx + len(suffix) != len(name_lower):
|
|
471
|
+
continue
|
|
472
|
+
# 前一个字符必须是分隔符或字符串开始
|
|
473
|
+
if idx == 0 or name_lower[idx - 1] in "_- /.":
|
|
474
|
+
return {"better": item["better"], "matched_suffix": suffix}
|
|
475
|
+
return None
|
|
476
|
+
|
|
477
|
+
|
|
478
|
+
def infer_better_direction(metric_name: str) -> dict:
|
|
479
|
+
"""根据指标名称推断优化方向。
|
|
480
|
+
|
|
481
|
+
优先匹配关键词库；未匹配时尝试从列名提取单位并基于后缀推断方向。
|
|
482
|
+
"""
|
|
483
|
+
if not metric_name or not isinstance(metric_name, str):
|
|
484
|
+
return {"better": None, "target": None, "description": "", "is_keyword_matched": False}
|
|
485
|
+
matched = match_keyword(metric_name)
|
|
486
|
+
if matched:
|
|
487
|
+
return {
|
|
488
|
+
"better": matched["better"],
|
|
489
|
+
"target": matched.get("target"),
|
|
490
|
+
"description": matched.get("description", ""),
|
|
491
|
+
"is_keyword_matched": True,
|
|
492
|
+
"matched_keyword": matched["keyword"],
|
|
493
|
+
}
|
|
494
|
+
# 兜底 1：从列名提取单位
|
|
495
|
+
fallback_unit = extract_unit_from_name(metric_name)
|
|
496
|
+
# 如果提取到单位，去除单位部分再做后缀推断
|
|
497
|
+
# 例如 'carbon_footprint_ms' → 'carbon_footprint' → 'footprint' 后缀匹配
|
|
498
|
+
name_for_suffix = metric_name
|
|
499
|
+
if fallback_unit:
|
|
500
|
+
name_for_suffix = re.sub(r"\s*\([^)]+\)\s*$", "", metric_name)
|
|
501
|
+
name_for_suffix = re.sub(r"\s*\[[^\]]+\]\s*$", "", name_for_suffix)
|
|
502
|
+
name_for_suffix = re.sub(r"[_\-\s/][a-zμ²³°]+$", "", name_for_suffix, flags=re.IGNORECASE)
|
|
503
|
+
name_for_suffix = name_for_suffix.strip()
|
|
504
|
+
# 兜底 2：基于后缀推断方向（在去除单位后的名称上）
|
|
505
|
+
suffix_dir = infer_direction_by_suffix(name_for_suffix)
|
|
506
|
+
return {
|
|
507
|
+
"better": suffix_dir["better"] if suffix_dir else None,
|
|
508
|
+
"target": None,
|
|
509
|
+
"description": "",
|
|
510
|
+
"is_keyword_matched": False,
|
|
511
|
+
"is_suffix_inferred": bool(suffix_dir),
|
|
512
|
+
"matched_suffix": suffix_dir["matched_suffix"] if suffix_dir else None,
|
|
513
|
+
"is_suffix_extracted": bool(fallback_unit),
|
|
514
|
+
"unit": fallback_unit,
|
|
515
|
+
"confidence": 0.5 if suffix_dir else 0,
|
|
516
|
+
}
|
|
@@ -61,18 +61,18 @@ def detect_knee_point(
|
|
|
61
61
|
frontier_points: 前沿点数组，每个点包含 { x, y, caseName, raw, algo }
|
|
62
62
|
options: 选项
|
|
63
63
|
- x_reversed: X 轴是否反转（higher is better）
|
|
64
|
-
- y_reversed: Y 轴是否反转（higher is better）
|
|
65
64
|
|
|
66
65
|
Returns:
|
|
67
66
|
拐点信息 { point, index, distance, extremes, sorted_frontier } 或 None
|
|
67
|
+
|
|
68
|
+
注：y_reversed 不参与拐点算法（最大距离法只按 X 排序，Y 方向不影响距离几何），
|
|
69
|
+
与 JS detectKneePoint 保持一致，不暴露该参数。
|
|
68
70
|
"""
|
|
69
71
|
if not frontier_points or len(frontier_points) < 3:
|
|
70
72
|
return None
|
|
71
73
|
|
|
72
74
|
opts = options or {}
|
|
73
75
|
x_reversed = opts.get("x_reversed", False)
|
|
74
|
-
# y_reversed 在 JS 中存在但未影响本算法逻辑，保持参数一致性
|
|
75
|
-
# y_reversed = opts.get("y_reversed", False)
|
|
76
76
|
|
|
77
77
|
# 按 X 排序（考虑方向：如果反转，则降序）
|
|
78
78
|
sorted_points = sorted(
|