@aipack-ai/eval 1.1.6 → 1.1.7
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.
- package/dist/index.d.ts +6 -1
- package/dist/index.js +5 -1
- package/dist/index.js.map +1 -1
- package/dist/src/cli.d.ts +5 -3
- package/dist/src/cli.js +163 -27
- package/dist/src/cli.js.map +1 -1
- package/dist/src/core/history.d.ts +39 -0
- package/dist/src/core/history.js +134 -0
- package/dist/src/core/history.js.map +1 -0
- package/dist/src/core/live.d.ts +77 -0
- package/dist/src/core/live.js +142 -0
- package/dist/src/core/live.js.map +1 -0
- package/dist/src/core/report.d.ts +9 -2
- package/dist/src/core/report.js +30 -5
- package/dist/src/core/report.js.map +1 -1
- package/dist/src/core/runner.d.ts +10 -5
- package/dist/src/core/runner.js +129 -40
- package/dist/src/core/runner.js.map +1 -1
- package/dist/src/core/types.d.ts +46 -4
- package/dist/src/core/validate.js +6 -0
- package/dist/src/core/validate.js.map +1 -1
- package/package.json +4 -3
package/dist/index.d.ts
CHANGED
|
@@ -20,5 +20,10 @@ export type { CaseScore } from './src/core/scorer';
|
|
|
20
20
|
export { validateEvalCase } from './src/core/validate';
|
|
21
21
|
export { loadCases, defaultCasesDir, packageRoot, } from './src/core/loader';
|
|
22
22
|
export type { LoadCasesResult } from './src/core/loader';
|
|
23
|
-
export { runEval } from './src/core/runner';
|
|
23
|
+
export { runEval, skipReason } from './src/core/runner';
|
|
24
24
|
export { renderMarkdown, writeReport, writeBaseline, readBaseline, reportToBaseline, compareBaseline, finalizeReport, } from './src/core/report';
|
|
25
|
+
export type { RenderOptions } from './src/core/report';
|
|
26
|
+
export { LIVE_ENV_KEYS, parseModelSpec, resolveLiveSpec, createLiveLlm, resolveLiveLlm, describeLiveLlm, assertLiveReady, } from './src/core/live';
|
|
27
|
+
export type { LiveLlmSpec, LiveLlm, ResolveLiveOptions } from './src/core/live';
|
|
28
|
+
export { appendHistory, readHistory, renderHistoryTrend, reportToHistoryEntry, sparkline, } from './src/core/history';
|
|
29
|
+
export type { HistoryEntry } from './src/core/history';
|
package/dist/index.js
CHANGED
|
@@ -21,6 +21,10 @@ export { expectedToScorers, resolveScorers, scoreTrace, RULE_SCORER_TYPES, runRu
|
|
|
21
21
|
export { validateEvalCase } from './src/core/validate.js';
|
|
22
22
|
export { loadCases, defaultCasesDir, packageRoot, } from './src/core/loader.js';
|
|
23
23
|
// ─── Runner 与报告 ────────────────────────────────────────────────
|
|
24
|
-
export { runEval } from './src/core/runner.js';
|
|
24
|
+
export { runEval, skipReason } from './src/core/runner.js';
|
|
25
25
|
export { renderMarkdown, writeReport, writeBaseline, readBaseline, reportToBaseline, compareBaseline, finalizeReport, } from './src/core/report.js';
|
|
26
|
+
// ─── live 模式(真实 LLM,M3)─────────────────────────────────────
|
|
27
|
+
export { LIVE_ENV_KEYS, parseModelSpec, resolveLiveSpec, createLiveLlm, resolveLiveLlm, describeLiveLlm, assertLiveReady, } from './src/core/live.js';
|
|
28
|
+
// ─── 历史趋势(日环比可视化,M3)───────────────────────────────────
|
|
29
|
+
export { appendHistory, readHistory, renderHistoryTrend, reportToHistoryEntry, sparkline, } from './src/core/history.js';
|
|
26
30
|
//# sourceMappingURL=index.js.map
|
package/dist/index.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AA0BH,kEAAkE;AAClE,OAAO,EAAE,kBAAkB,EAAE,MAAM,wBAAwB,CAAC;AAC5D,OAAO,EACL,eAAe,EACf,mBAAmB,GACpB,MAAM,uBAAuB,CAAC;AAG/B,iEAAiE;AACjE,OAAO,EAAE,iBAAiB,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAC;AAEtE,kEAAkE;AAClE,OAAO,EACL,iBAAiB,EACjB,cAAc,EACd,UAAU,EACV,iBAAiB,EACjB,aAAa,EACb,YAAY,GACb,MAAM,mBAAmB,CAAC;AAG3B,iEAAiE;AACjE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qBAAqB,CAAC;AACvD,OAAO,EACL,SAAS,EACT,eAAe,EACf,WAAW,GACZ,MAAM,mBAAmB,CAAC;AAG3B,kEAAkE;AAClE,OAAO,EAAE,OAAO,EAAE,MAAM,mBAAmB,CAAC;
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AA0BH,kEAAkE;AAClE,OAAO,EAAE,kBAAkB,EAAE,MAAM,wBAAwB,CAAC;AAC5D,OAAO,EACL,eAAe,EACf,mBAAmB,GACpB,MAAM,uBAAuB,CAAC;AAG/B,iEAAiE;AACjE,OAAO,EAAE,iBAAiB,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAC;AAEtE,kEAAkE;AAClE,OAAO,EACL,iBAAiB,EACjB,cAAc,EACd,UAAU,EACV,iBAAiB,EACjB,aAAa,EACb,YAAY,GACb,MAAM,mBAAmB,CAAC;AAG3B,iEAAiE;AACjE,OAAO,EAAE,gBAAgB,EAAE,MAAM,qBAAqB,CAAC;AACvD,OAAO,EACL,SAAS,EACT,eAAe,EACf,WAAW,GACZ,MAAM,mBAAmB,CAAC;AAG3B,kEAAkE;AAClE,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AACxD,OAAO,EACL,cAAc,EACd,WAAW,EACX,aAAa,EACb,YAAY,EACZ,gBAAgB,EAChB,eAAe,EACf,cAAc,GACf,MAAM,mBAAmB,CAAC;AAG3B,8DAA8D;AAC9D,OAAO,EACL,aAAa,EACb,cAAc,EACd,eAAe,EACf,aAAa,EACb,cAAc,EACd,eAAe,EACf,eAAe,GAChB,MAAM,iBAAiB,CAAC;AAGzB,yDAAyD;AACzD,OAAO,EACL,aAAa,EACb,WAAW,EACX,kBAAkB,EAClB,oBAAoB,EACpB,SAAS,GACV,MAAM,oBAAoB,CAAC"}
|
package/dist/src/cli.d.ts
CHANGED
|
@@ -3,9 +3,11 @@
|
|
|
3
3
|
* packages/eval/src/cli.ts - aipack-eval 命令行入口
|
|
4
4
|
*
|
|
5
5
|
* 用法:
|
|
6
|
-
* aipack-eval run [--suite <name>]... [--
|
|
7
|
-
* [--
|
|
8
|
-
* [--
|
|
6
|
+
* aipack-eval run [--suite <name>]... [--mode mock|live] [--model provider/id]
|
|
7
|
+
* [--api-key <key>] [--base-url <url>] [--temperature <0>]
|
|
8
|
+
* [--repeats <n>] [--cases-dir <dir>] [--report-dir <dir>]
|
|
9
|
+
* [--history <path>] [--baseline <path>] [--update-baseline]
|
|
10
|
+
* aipack-eval history [--path <file>] [--limit <n>] [--json]
|
|
9
11
|
*
|
|
10
12
|
* 退出码:0 = 全过且无回归;1 = 有失败或回归;2 = 用法错误。
|
|
11
13
|
*/
|
package/dist/src/cli.js
CHANGED
|
@@ -3,29 +3,55 @@
|
|
|
3
3
|
* packages/eval/src/cli.ts - aipack-eval 命令行入口
|
|
4
4
|
*
|
|
5
5
|
* 用法:
|
|
6
|
-
* aipack-eval run [--suite <name>]... [--
|
|
7
|
-
* [--
|
|
8
|
-
* [--
|
|
6
|
+
* aipack-eval run [--suite <name>]... [--mode mock|live] [--model provider/id]
|
|
7
|
+
* [--api-key <key>] [--base-url <url>] [--temperature <0>]
|
|
8
|
+
* [--repeats <n>] [--cases-dir <dir>] [--report-dir <dir>]
|
|
9
|
+
* [--history <path>] [--baseline <path>] [--update-baseline]
|
|
10
|
+
* aipack-eval history [--path <file>] [--limit <n>] [--json]
|
|
9
11
|
*
|
|
10
12
|
* 退出码:0 = 全过且无回归;1 = 有失败或回归;2 = 用法错误。
|
|
11
13
|
*/
|
|
12
14
|
import { loadCases } from './core/loader.js';
|
|
13
15
|
import { runEval } from './core/runner.js';
|
|
14
16
|
import { finalizeReport, renderMarkdown } from './core/report.js';
|
|
15
|
-
|
|
17
|
+
import { readHistory, renderHistoryTrend } from './core/history.js';
|
|
18
|
+
const USAGE = `用法: aipack-eval <run|history> [options]
|
|
16
19
|
|
|
17
|
-
选项:
|
|
20
|
+
run 选项:
|
|
18
21
|
--suite <name> 只跑指定套件(可多次)
|
|
22
|
+
--mode <mock|live> 运行模式(缺省 mock;live = 真实 LLM)
|
|
23
|
+
--model <provider/id> live 模型,如 deepseek/deepseek-chat(或 AIPACK_EVAL_MODEL)
|
|
24
|
+
--api-key <key> live API Key(或 <PROVIDER>_API_KEY)
|
|
25
|
+
--base-url <url> live 端点覆盖(代理 / 兼容网关)
|
|
26
|
+
--temperature <x> live 采样温度(缺省 0)
|
|
27
|
+
--repeats <n> 每用例重复次数(live 缺省 3,pass@k 判定)
|
|
28
|
+
--concurrency <n> 并发用例数(缺省 8)
|
|
29
|
+
--timeout <ms> 单用例墙钟超时(缺省 30000)
|
|
30
|
+
--max-tokens <n> 全局 token 预算,超限熔断剩余用例
|
|
19
31
|
--cases-dir <dir> 用例目录(缺省 <包根>/eval/cases)
|
|
20
|
-
--report-dir <dir>
|
|
32
|
+
--report-dir <dir> 报告输出目录(缺省不写盘;'none' 不写盘)
|
|
33
|
+
--history <path> 历史趋势 JSONL(缺省 <report-dir>/history.jsonl)
|
|
21
34
|
--baseline <path> baseline 文件(门禁对比)
|
|
22
35
|
--update-baseline 用本次结果更新 baseline
|
|
23
36
|
--threshold <x> 通过率回归阈值(缺省 0.02)
|
|
24
|
-
|
|
37
|
+
|
|
38
|
+
history 选项:
|
|
39
|
+
--path <file> 历史文件(缺省 ./eval-results/history.jsonl)
|
|
40
|
+
--limit <n> 展示最近 N 次(缺省 15)
|
|
41
|
+
--json 输出原始 JSONL 条目
|
|
42
|
+
|
|
43
|
+
公共:
|
|
25
44
|
-h, --help 显示帮助`;
|
|
26
|
-
function
|
|
45
|
+
function num(raw, label) {
|
|
46
|
+
const v = Number(raw);
|
|
47
|
+
if (!Number.isFinite(v))
|
|
48
|
+
throw new Error(`${label} 必须为数字`);
|
|
49
|
+
return v;
|
|
50
|
+
}
|
|
51
|
+
function parseRunArgs(argv) {
|
|
27
52
|
const args = {
|
|
28
53
|
suites: [],
|
|
54
|
+
mode: 'mock',
|
|
29
55
|
updateBaseline: false,
|
|
30
56
|
threshold: 0.02,
|
|
31
57
|
};
|
|
@@ -40,18 +66,65 @@ function parseArgs(argv) {
|
|
|
40
66
|
};
|
|
41
67
|
try {
|
|
42
68
|
switch (a) {
|
|
43
|
-
case 'run':
|
|
44
|
-
args.command = 'run';
|
|
45
|
-
break;
|
|
46
69
|
case '--suite':
|
|
47
70
|
args.suites.push(next());
|
|
48
71
|
break;
|
|
72
|
+
case '--mode': {
|
|
73
|
+
const v = next();
|
|
74
|
+
if (v !== 'mock' && v !== 'live')
|
|
75
|
+
throw new Error('--mode 只能为 mock 或 live');
|
|
76
|
+
args.mode = v;
|
|
77
|
+
break;
|
|
78
|
+
}
|
|
79
|
+
case '--model':
|
|
80
|
+
args.model = next();
|
|
81
|
+
break;
|
|
82
|
+
case '--api-key':
|
|
83
|
+
args.apiKey = next();
|
|
84
|
+
break;
|
|
85
|
+
case '--base-url':
|
|
86
|
+
args.baseUrl = next();
|
|
87
|
+
break;
|
|
88
|
+
case '--temperature':
|
|
89
|
+
args.temperature = num(next(), '--temperature');
|
|
90
|
+
break;
|
|
91
|
+
case '--repeats': {
|
|
92
|
+
const v = Number(next());
|
|
93
|
+
if (!Number.isInteger(v) || v < 1)
|
|
94
|
+
throw new Error('repeats 须为正整数');
|
|
95
|
+
args.repeats = v;
|
|
96
|
+
break;
|
|
97
|
+
}
|
|
98
|
+
case '--concurrency': {
|
|
99
|
+
const v = Number(next());
|
|
100
|
+
if (!Number.isInteger(v) || v < 1)
|
|
101
|
+
throw new Error('concurrency 须为正整数');
|
|
102
|
+
args.concurrency = v;
|
|
103
|
+
break;
|
|
104
|
+
}
|
|
105
|
+
case '--timeout': {
|
|
106
|
+
const v = Number(next());
|
|
107
|
+
if (!Number.isFinite(v) || v <= 0)
|
|
108
|
+
throw new Error('timeout 须为正数(ms)');
|
|
109
|
+
args.timeoutMs = v;
|
|
110
|
+
break;
|
|
111
|
+
}
|
|
112
|
+
case '--max-tokens': {
|
|
113
|
+
const v = Number(next());
|
|
114
|
+
if (!Number.isFinite(v) || v <= 0)
|
|
115
|
+
throw new Error('max-tokens 须为正数');
|
|
116
|
+
args.maxTotalTokens = v;
|
|
117
|
+
break;
|
|
118
|
+
}
|
|
49
119
|
case '--cases-dir':
|
|
50
120
|
args.casesDir = next();
|
|
51
121
|
break;
|
|
52
122
|
case '--report-dir':
|
|
53
123
|
args.reportDir = next();
|
|
54
124
|
break;
|
|
125
|
+
case '--history':
|
|
126
|
+
args.historyPath = next();
|
|
127
|
+
break;
|
|
55
128
|
case '--baseline':
|
|
56
129
|
args.baselinePath = next();
|
|
57
130
|
break;
|
|
@@ -65,13 +138,6 @@ function parseArgs(argv) {
|
|
|
65
138
|
args.threshold = v;
|
|
66
139
|
break;
|
|
67
140
|
}
|
|
68
|
-
case '--repeats': {
|
|
69
|
-
const v = Number(next());
|
|
70
|
-
if (!Number.isInteger(v) || v < 1)
|
|
71
|
-
throw new Error('repeats 须为正整数');
|
|
72
|
-
args.repeats = v;
|
|
73
|
-
break;
|
|
74
|
-
}
|
|
75
141
|
case '-h':
|
|
76
142
|
case '--help':
|
|
77
143
|
return { error: 'HELP' };
|
|
@@ -84,12 +150,10 @@ function parseArgs(argv) {
|
|
|
84
150
|
}
|
|
85
151
|
i += 1;
|
|
86
152
|
}
|
|
87
|
-
if (args.command !== 'run')
|
|
88
|
-
return { error: `缺少子命令 'run'\n${USAGE}` };
|
|
89
153
|
return args;
|
|
90
154
|
}
|
|
91
|
-
async function
|
|
92
|
-
const parsed =
|
|
155
|
+
async function cmdRun(argv) {
|
|
156
|
+
const parsed = parseRunArgs(argv);
|
|
93
157
|
if ('error' in parsed) {
|
|
94
158
|
if (parsed.error === 'HELP') {
|
|
95
159
|
console.log(USAGE);
|
|
@@ -109,22 +173,45 @@ async function main() {
|
|
|
109
173
|
console.error('没有可运行的用例');
|
|
110
174
|
return 1;
|
|
111
175
|
}
|
|
176
|
+
const reportDir = parsed.reportDir === 'none' ? undefined : parsed.reportDir;
|
|
177
|
+
const historyPath = parsed.historyPath ?? (reportDir ? `${reportDir.replace(/\/$/, '')}/history.jsonl` : undefined);
|
|
112
178
|
const config = {
|
|
113
|
-
mode:
|
|
179
|
+
mode: parsed.mode,
|
|
180
|
+
model: parsed.model,
|
|
181
|
+
apiKey: parsed.apiKey,
|
|
182
|
+
baseUrl: parsed.baseUrl,
|
|
183
|
+
temperature: parsed.temperature,
|
|
114
184
|
suites: parsed.suites,
|
|
115
185
|
repeats: parsed.repeats,
|
|
186
|
+
concurrency: parsed.concurrency,
|
|
187
|
+
timeoutMs: parsed.timeoutMs,
|
|
188
|
+
maxTotalTokens: parsed.maxTotalTokens,
|
|
116
189
|
casesDir: parsed.casesDir,
|
|
117
|
-
reportDir
|
|
190
|
+
reportDir,
|
|
191
|
+
historyPath,
|
|
118
192
|
baselinePath: parsed.baselinePath,
|
|
119
193
|
updateBaseline: parsed.updateBaseline,
|
|
120
194
|
regressionThreshold: parsed.threshold,
|
|
121
195
|
};
|
|
122
|
-
console.error(`加载 ${cases.length} 个用例${parsed.suites.length ? `(套件: ${parsed.suites.join(', ')})` : ''},mode
|
|
123
|
-
|
|
196
|
+
console.error(`加载 ${cases.length} 个用例${parsed.suites.length ? `(套件: ${parsed.suites.join(', ')})` : ''},mode=${parsed.mode}`);
|
|
197
|
+
let report;
|
|
198
|
+
try {
|
|
199
|
+
report = await runEval(cases, config);
|
|
200
|
+
}
|
|
201
|
+
catch (e) {
|
|
202
|
+
// live 装配失败(缺模型 / 缺 Key)属于使用错误
|
|
203
|
+
console.error(`live 模式配置错误: ${e.message}`);
|
|
204
|
+
return 2;
|
|
205
|
+
}
|
|
124
206
|
const finalized = await finalizeReport(report, config);
|
|
125
|
-
console.log(renderMarkdown(report));
|
|
207
|
+
console.log(renderMarkdown(report, { history: finalized.history }));
|
|
208
|
+
if (report.totals.cases === 0) {
|
|
209
|
+
console.error(`\n⚠️ 当前 mode=${parsed.mode} 下没有可运行的用例(${report.totals.skipped} 个被跳过),检查 --suite 与 --mode 是否匹配`);
|
|
210
|
+
}
|
|
126
211
|
if (finalized.jsonPath)
|
|
127
212
|
console.error(`\n报告已写入: ${finalized.jsonPath}`);
|
|
213
|
+
if (finalized.historyPath)
|
|
214
|
+
console.error(`历史已追加: ${finalized.historyPath}`);
|
|
128
215
|
if (config.updateBaseline && finalized.baseline === undefined) {
|
|
129
216
|
console.error(`baseline 已更新: ${config.baselinePath}`);
|
|
130
217
|
}
|
|
@@ -142,6 +229,55 @@ async function main() {
|
|
|
142
229
|
}
|
|
143
230
|
return exitCode;
|
|
144
231
|
}
|
|
232
|
+
async function cmdHistory(argv) {
|
|
233
|
+
let path = 'eval-results/history.jsonl';
|
|
234
|
+
let limit = 15;
|
|
235
|
+
let asJson = false;
|
|
236
|
+
for (let i = 0; i < argv.length; i++) {
|
|
237
|
+
const a = argv[i];
|
|
238
|
+
if (a === '--json') {
|
|
239
|
+
asJson = true;
|
|
240
|
+
}
|
|
241
|
+
else if (a === '--path') {
|
|
242
|
+
path = argv[++i] ?? path;
|
|
243
|
+
}
|
|
244
|
+
else if (a === '--limit') {
|
|
245
|
+
limit = Number(argv[++i] ?? limit);
|
|
246
|
+
}
|
|
247
|
+
else if (a === '-h' || a === '--help') {
|
|
248
|
+
console.log(USAGE);
|
|
249
|
+
return 0;
|
|
250
|
+
}
|
|
251
|
+
else {
|
|
252
|
+
console.error(`未知参数: ${a}\n${USAGE}`);
|
|
253
|
+
return 2;
|
|
254
|
+
}
|
|
255
|
+
}
|
|
256
|
+
const entries = await readHistory(path, limit);
|
|
257
|
+
if (entries.length === 0) {
|
|
258
|
+
console.error(`历史文件为空或不存在: ${path}`);
|
|
259
|
+
return 1;
|
|
260
|
+
}
|
|
261
|
+
if (asJson) {
|
|
262
|
+
console.log(entries.map((e) => JSON.stringify(e)).join('\n'));
|
|
263
|
+
return 0;
|
|
264
|
+
}
|
|
265
|
+
console.log(renderHistoryTrend(entries, limit));
|
|
266
|
+
return 0;
|
|
267
|
+
}
|
|
268
|
+
async function main() {
|
|
269
|
+
const [command, ...rest] = process.argv.slice(2);
|
|
270
|
+
if (command === undefined || command === '-h' || command === '--help') {
|
|
271
|
+
console.log(USAGE);
|
|
272
|
+
return command === undefined ? 2 : 0;
|
|
273
|
+
}
|
|
274
|
+
if (command === 'run')
|
|
275
|
+
return cmdRun(rest);
|
|
276
|
+
if (command === 'history')
|
|
277
|
+
return cmdHistory(rest);
|
|
278
|
+
console.error(`未知子命令: ${command}\n${USAGE}`);
|
|
279
|
+
return 2;
|
|
280
|
+
}
|
|
145
281
|
main().then((code) => process.exit(code), (e) => {
|
|
146
282
|
console.error(e);
|
|
147
283
|
process.exit(1);
|
package/dist/src/cli.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"cli.js","sourceRoot":"","sources":["../../src/cli.ts"],"names":[],"mappings":";AACA
|
|
1
|
+
{"version":3,"file":"cli.js","sourceRoot":"","sources":["../../src/cli.ts"],"names":[],"mappings":";AACA;;;;;;;;;;;GAWG;AAEH,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AAC1C,OAAO,EAAE,OAAO,EAAE,MAAM,eAAe,CAAC;AACxC,OAAO,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,eAAe,CAAC;AAC/D,OAAO,EAAE,WAAW,EAAE,kBAAkB,EAAE,MAAM,gBAAgB,CAAC;AAwBjE,MAAM,KAAK,GAAG;;;;;;;;;;;;;;;;;;;;;;;;;;8BA0BgB,CAAC;AAE/B,SAAS,GAAG,CAAC,GAAW,EAAE,KAAa;IACrC,MAAM,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC;IACtB,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,QAAQ,CAAC,CAAC;IAC3D,OAAO,CAAC,CAAC;AACX,CAAC;AAED,SAAS,YAAY,CAAC,IAAc;IAClC,MAAM,IAAI,GAAY;QACpB,MAAM,EAAE,EAAE;QACV,IAAI,EAAE,MAAM;QACZ,cAAc,EAAE,KAAK;QACrB,SAAS,EAAE,IAAI;KAChB,CAAC;IACF,IAAI,CAAC,GAAG,CAAC,CAAC;IACV,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;QACvB,MAAM,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QAClB,MAAM,IAAI,GAAG,GAAW,EAAE;YACxB,MAAM,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;YACpB,IAAI,CAAC,KAAK,SAAS;gBAAE,MAAM,IAAI,KAAK,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;YACtD,OAAO,CAAC,CAAC;QACX,CAAC,CAAC;QACF,IAAI,CAAC;YACH,QAAQ,CAAC,EAAE,CAAC;gBACV,KAAK,SAAS;oBACZ,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;oBACzB,MAAM;gBACR,KAAK,QAAQ,CAAC,CAAC,CAAC;oBACd,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC;oBACjB,IAAI,CAAC,KAAK,MAAM,IAAI,CAAC,KAAK,MAAM;wBAAE,MAAM,IAAI,KAAK,CAAC,wBAAwB,CAAC,CAAC;oBAC5E,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC;oBACd,MAAM;gBACR,CAAC;gBACD,KAAK,SAAS;oBACZ,IAAI,CAAC,KAAK,GAAG,IAAI,EAAE,CAAC;oBACpB,MAAM;gBACR,KAAK,WAAW;oBACd,IAAI,CAAC,MAAM,GAAG,IAAI,EAAE,CAAC;oBACrB,MAAM;gBACR,KAAK,YAAY;oBACf,IAAI,CAAC,OAAO,GAAG,IAAI,EAAE,CAAC;oBACtB,MAAM;gBACR,KAAK,eAAe;oBAClB,IAAI,CAAC,WAAW,GAAG,GAAG,CAAC,IAAI,EAAE,EAAE,eAAe,CAAC,CAAC;oBAChD,MAAM;gBACR,KAAK,WAAW,CAAC,CAAC,CAAC;oBACjB,MAAM,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC;oBACzB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC;wBAAE,MAAM,IAAI,KAAK,CAAC,eAAe,CAAC,CAAC;oBACpE,IAAI,CAAC,OAAO,GAAG,CAAC,CAAC;oBACjB,MAAM;gBACR,CAAC;gBACD,KAAK,eAAe,CAAC,CAAC,CAAC;oBACrB,MAAM,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC;oBACzB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC;wBAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;oBACxE,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC;oBACrB,MAAM;gBACR,CAAC;gBACD,KAAK,WAAW,CAAC,CAAC,CAAC;oBACjB,MAAM,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC;oBACzB,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;wBAAE,MAAM,IAAI,KAAK,CAAC,kBAAkB,CAAC,CAAC;oBACvE,IAAI,CAAC,SAAS,GAAG,CAAC,CAAC;oBACnB,MAAM;gBACR,CAAC;gBACD,KAAK,cAAc,CAAC,CAAC,CAAC;oBACpB,MAAM,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC;oBACzB,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;wBAAE,MAAM,IAAI,KAAK,CAAC,iBAAiB,CAAC,CAAC;oBACtE,IAAI,CAAC,cAAc,GAAG,CAAC,CAAC;oBACxB,MAAM;gBACR,CAAC;gBACD,KAAK,aAAa;oBAChB,IAAI,CAAC,QAAQ,GAAG,IAAI,EAAE,CAAC;oBACvB,MAAM;gBACR,KAAK,cAAc;oBACjB,IAAI,CAAC,SAAS,GAAG,IAAI,EAAE,CAAC;oBACxB,MAAM;gBACR,KAAK,WAAW;oBACd,IAAI,CAAC,WAAW,GAAG,IAAI,EAAE,CAAC;oBAC1B,MAAM;gBACR,KAAK,YAAY;oBACf,IAAI,CAAC,YAAY,GAAG,IAAI,EAAE,CAAC;oBAC3B,MAAM;gBACR,KAAK,mBAAmB;oBACtB,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC;oBAC3B,MAAM;gBACR,KAAK,aAAa,CAAC,CAAC,CAAC;oBACnB,MAAM,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC;oBACzB,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC;wBAAE,MAAM,IAAI,KAAK,CAAC,UAAU,CAAC,CAAC;oBACvE,IAAI,CAAC,SAAS,GAAG,CAAC,CAAC;oBACnB,MAAM;gBACR,CAAC;gBACD,KAAK,IAAI,CAAC;gBACV,KAAK,QAAQ;oBACX,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;gBAC3B;oBACE,OAAO,EAAE,KAAK,EAAE,SAAS,CAAC,KAAK,KAAK,EAAE,EAAE,CAAC;YAC7C,CAAC;QACH,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,OAAO,EAAE,KAAK,EAAE,GAAI,CAAW,CAAC,OAAO,KAAK,KAAK,EAAE,EAAE,CAAC;QACxD,CAAC;QACD,CAAC,IAAI,CAAC,CAAC;IACT,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,KAAK,UAAU,MAAM,CAAC,IAAc;IAClC,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAClC,IAAI,OAAO,IAAI,MAAM,EAAE,CAAC;QACtB,IAAI,MAAM,CAAC,KAAK,KAAK,MAAM,EAAE,CAAC;YAC5B,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;YACnB,OAAO,CAAC,CAAC;QACX,CAAC;QACD,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QAC5B,OAAO,CAAC,CAAC;IACX,CAAC;IAED,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM,SAAS,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;IAC3D,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACtB,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;QACzB,KAAK,MAAM,CAAC,IAAI,MAAM;YAAE,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;QAClD,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvB,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;QAC1B,OAAO,CAAC,CAAC;IACX,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,KAAK,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC;IAC7E,MAAM,WAAW,GACf,MAAM,CAAC,WAAW,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,gBAAgB,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;IAElG,MAAM,MAAM,GAAc;QACxB,IAAI,EAAE,MAAM,CAAC,IAAI;QACjB,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,MAAM,EAAE,MAAM,CAAC,MAAM;QACrB,OAAO,EAAE,MAAM,CAAC,OAAO;QACvB,WAAW,EAAE,MAAM,CAAC,WAAW;QAC/B,MAAM,EAAE,MAAM,CAAC,MAAM;QACrB,OAAO,EAAE,MAAM,CAAC,OAAO;QACvB,WAAW,EAAE,MAAM,CAAC,WAAW;QAC/B,SAAS,EAAE,MAAM,CAAC,SAAS;QAC3B,cAAc,EAAE,MAAM,CAAC,cAAc;QACrC,QAAQ,EAAE,MAAM,CAAC,QAAQ;QACzB,SAAS;QACT,WAAW;QACX,YAAY,EAAE,MAAM,CAAC,YAAY;QACjC,cAAc,EAAE,MAAM,CAAC,cAAc;QACrC,mBAAmB,EAAE,MAAM,CAAC,SAAS;KACtC,CAAC;IAEF,OAAO,CAAC,KAAK,CACX,MAAM,KAAK,CAAC,MAAM,OAAO,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,SAAS,MAAM,CAAC,IAAI,EAAE,CAC/G,CAAC;IAEF,IAAI,MAAkB,CAAC;IACvB,IAAI,CAAC;QACH,MAAM,GAAG,MAAM,OAAO,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IACxC,CAAC;IAAC,OAAO,CAAC,EAAE,CAAC;QACX,+BAA+B;QAC/B,OAAO,CAAC,KAAK,CAAC,gBAAiB,CAAW,CAAC,OAAO,EAAE,CAAC,CAAC;QACtD,OAAO,CAAC,CAAC;IACX,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACvD,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,MAAM,EAAE,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;IAEpE,IAAI,MAAM,CAAC,MAAM,CAAC,KAAK,KAAK,CAAC,EAAE,CAAC;QAC9B,OAAO,CAAC,KAAK,CACX,iBAAiB,MAAM,CAAC,IAAI,cAAc,MAAM,CAAC,MAAM,CAAC,OAAO,iCAAiC,CACjG,CAAC;IACJ,CAAC;IAED,IAAI,SAAS,CAAC,QAAQ;QAAE,OAAO,CAAC,KAAK,CAAC,YAAY,SAAS,CAAC,QAAQ,EAAE,CAAC,CAAC;IACxE,IAAI,SAAS,CAAC,WAAW;QAAE,OAAO,CAAC,KAAK,CAAC,UAAU,SAAS,CAAC,WAAW,EAAE,CAAC,CAAC;IAC5E,IAAI,MAAM,CAAC,cAAc,IAAI,SAAS,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;QAC9D,OAAO,CAAC,KAAK,CAAC,iBAAiB,MAAM,CAAC,YAAY,EAAE,CAAC,CAAC;IACxD,CAAC;IAED,IAAI,QAAQ,GAAG,MAAM,CAAC,MAAM,CAAC,MAAM,KAAK,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAEpE,IAAI,SAAS,CAAC,UAAU,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,EAAE,EAAE,CAAC;QACrD,QAAQ,GAAG,CAAC,CAAC;QACb,MAAM,EAAE,YAAY,EAAE,WAAW,EAAE,GAAG,SAAS,CAAC,UAAU,CAAC;QAC3D,OAAO,CAAC,KAAK,CACX,wBAAwB,YAAY,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,YAAY,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAC5F,CAAC;QACF,KAAK,MAAM,CAAC,IAAI,WAAW,EAAE,CAAC;YAC5B,OAAO,CAAC,KAAK,CACX,OAAO,CAAC,CAAC,KAAK,KAAK,CAAC,CAAC,CAAC,QAAQ,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CACvH,CAAC;QACJ,CAAC;IACH,CAAC;SAAM,IAAI,SAAS,CAAC,UAAU,EAAE,CAAC;QAChC,OAAO,CAAC,KAAK,CAAC,mBAAmB,CAAC,SAAS,CAAC,UAAU,CAAC,YAAY,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IAC7F,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,KAAK,UAAU,UAAU,CAAC,IAAc;IACtC,IAAI,IAAI,GAAG,4BAA4B,CAAC;IACxC,IAAI,KAAK,GAAG,EAAE,CAAC;IACf,IAAI,MAAM,GAAG,KAAK,CAAC;IACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACrC,MAAM,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QAClB,IAAI,CAAC,KAAK,QAAQ,EAAE,CAAC;YACnB,MAAM,GAAG,IAAI,CAAC;QAChB,CAAC;aAAM,IAAI,CAAC,KAAK,QAAQ,EAAE,CAAC;YAC1B,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,IAAI,IAAI,CAAC;QAC3B,CAAC;aAAM,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;YAC3B,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC;QACrC,CAAC;aAAM,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,QAAQ,EAAE,CAAC;YACxC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;YACnB,OAAO,CAAC,CAAC;QACX,CAAC;aAAM,CAAC;YACN,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,KAAK,EAAE,CAAC,CAAC;YACtC,OAAO,CAAC,CAAC;QACX,CAAC;IACH,CAAC;IAED,MAAM,OAAO,GAAG,MAAM,WAAW,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;IAC/C,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACzB,OAAO,CAAC,KAAK,CAAC,eAAe,IAAI,EAAE,CAAC,CAAC;QACrC,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,MAAM,EAAE,CAAC;QACX,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;QAC9D,OAAO,CAAC,CAAC;IACX,CAAC;IACD,OAAO,CAAC,GAAG,CAAC,kBAAkB,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC;IAChD,OAAO,CAAC,CAAC;AACX,CAAC;AAED,KAAK,UAAU,IAAI;IACjB,MAAM,CAAC,OAAO,EAAE,GAAG,IAAI,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACjD,IAAI,OAAO,KAAK,SAAS,IAAI,OAAO,KAAK,IAAI,IAAI,OAAO,KAAK,QAAQ,EAAE,CAAC;QACtE,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;QACnB,OAAO,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACvC,CAAC;IACD,IAAI,OAAO,KAAK,KAAK;QAAE,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,OAAO,KAAK,SAAS;QAAE,OAAO,UAAU,CAAC,IAAI,CAAC,CAAC;IACnD,OAAO,CAAC,KAAK,CAAC,UAAU,OAAO,KAAK,KAAK,EAAE,CAAC,CAAC;IAC7C,OAAO,CAAC,CAAC;AACX,CAAC;AAED,IAAI,EAAE,CAAC,IAAI,CACT,CAAC,IAAI,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAC5B,CAAC,CAAC,EAAE,EAAE;IACJ,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACjB,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AAClB,CAAC,CACF,CAAC"}
|
|
@@ -0,0 +1,39 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* packages/eval/src/core/history.ts - 历史趋势(日环比可视化)
|
|
3
|
+
*
|
|
4
|
+
* 每次运行向 JSONL 追加一行(append-only,天然按时间序),
|
|
5
|
+
* 读回后渲染 Markdown:趋势 sparkline + 逐次通过率 + 环比 Δ。
|
|
6
|
+
*
|
|
7
|
+
* 为什么用 JSONL 而不是单文件 JSON:cron 场景下并发/追加写更安全,
|
|
8
|
+
* 且 diff 友好(一行一次运行)。
|
|
9
|
+
*/
|
|
10
|
+
import type { EvalReport, SuiteSummary } from './types';
|
|
11
|
+
/** 每次运行落一行(JSONL) */
|
|
12
|
+
export interface HistoryEntry {
|
|
13
|
+
runId: string;
|
|
14
|
+
/** ISO 时间 */
|
|
15
|
+
at: string;
|
|
16
|
+
mode: 'mock' | 'live';
|
|
17
|
+
model?: string;
|
|
18
|
+
cases: number;
|
|
19
|
+
passed: number;
|
|
20
|
+
skipped: number;
|
|
21
|
+
passRate: number;
|
|
22
|
+
avgScore: number;
|
|
23
|
+
durationMs: number;
|
|
24
|
+
usageTotal: number;
|
|
25
|
+
bySuite: Record<string, Pick<SuiteSummary, 'total' | 'passed' | 'passRate'>>;
|
|
26
|
+
}
|
|
27
|
+
/** 报告 → 历史条目 */
|
|
28
|
+
export declare function reportToHistoryEntry(report: EvalReport): HistoryEntry;
|
|
29
|
+
/** 追加一次运行记录(JSONL,一行一条) */
|
|
30
|
+
export declare function appendHistory(report: EvalReport, historyPath: string): Promise<string>;
|
|
31
|
+
/** 读取历史(最新在最后);文件不存在返回空数组 */
|
|
32
|
+
export declare function readHistory(historyPath: string, limit?: number): Promise<HistoryEntry[]>;
|
|
33
|
+
/** 0~1 序列 → sparkline 字符串 */
|
|
34
|
+
export declare function sparkline(values: number[], width?: number): string;
|
|
35
|
+
/**
|
|
36
|
+
* 渲染通过率趋势(Markdown)。
|
|
37
|
+
* 环比定义:与上一次运行对比(cron 每日跑 → 即日环比)。
|
|
38
|
+
*/
|
|
39
|
+
export declare function renderHistoryTrend(entries: HistoryEntry[], limit?: number): string;
|
|
@@ -0,0 +1,134 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* packages/eval/src/core/history.ts - 历史趋势(日环比可视化)
|
|
3
|
+
*
|
|
4
|
+
* 每次运行向 JSONL 追加一行(append-only,天然按时间序),
|
|
5
|
+
* 读回后渲染 Markdown:趋势 sparkline + 逐次通过率 + 环比 Δ。
|
|
6
|
+
*
|
|
7
|
+
* 为什么用 JSONL 而不是单文件 JSON:cron 场景下并发/追加写更安全,
|
|
8
|
+
* 且 diff 友好(一行一次运行)。
|
|
9
|
+
*/
|
|
10
|
+
import { appendFile, mkdir, readFile } from 'node:fs/promises';
|
|
11
|
+
import { dirname, resolve } from 'node:path';
|
|
12
|
+
/** 报告 → 历史条目 */
|
|
13
|
+
export function reportToHistoryEntry(report) {
|
|
14
|
+
return {
|
|
15
|
+
runId: report.runId,
|
|
16
|
+
at: report.startedAt,
|
|
17
|
+
mode: report.mode,
|
|
18
|
+
model: report.model,
|
|
19
|
+
cases: report.totals.cases,
|
|
20
|
+
passed: report.totals.passed,
|
|
21
|
+
skipped: report.totals.skipped,
|
|
22
|
+
passRate: report.totals.passRate,
|
|
23
|
+
avgScore: report.totals.avgScore,
|
|
24
|
+
durationMs: report.durationMs,
|
|
25
|
+
usageTotal: report.totals.usageTotal,
|
|
26
|
+
bySuite: Object.fromEntries(Object.entries(report.bySuite).map(([k, s]) => [
|
|
27
|
+
k,
|
|
28
|
+
{ total: s.total, passed: s.passed, passRate: s.passRate },
|
|
29
|
+
])),
|
|
30
|
+
};
|
|
31
|
+
}
|
|
32
|
+
/** 追加一次运行记录(JSONL,一行一条) */
|
|
33
|
+
export async function appendHistory(report, historyPath) {
|
|
34
|
+
const path = resolve(historyPath);
|
|
35
|
+
await mkdir(dirname(path), { recursive: true });
|
|
36
|
+
await appendFile(path, `${JSON.stringify(reportToHistoryEntry(report))}\n`, 'utf-8');
|
|
37
|
+
return path;
|
|
38
|
+
}
|
|
39
|
+
/** 读取历史(最新在最后);文件不存在返回空数组 */
|
|
40
|
+
export async function readHistory(historyPath, limit = 30) {
|
|
41
|
+
const path = resolve(historyPath);
|
|
42
|
+
let raw;
|
|
43
|
+
try {
|
|
44
|
+
raw = await readFile(path, 'utf-8');
|
|
45
|
+
}
|
|
46
|
+
catch {
|
|
47
|
+
return [];
|
|
48
|
+
}
|
|
49
|
+
const entries = [];
|
|
50
|
+
for (const line of raw.split('\n')) {
|
|
51
|
+
const s = line.trim();
|
|
52
|
+
if (!s)
|
|
53
|
+
continue;
|
|
54
|
+
try {
|
|
55
|
+
entries.push(JSON.parse(s));
|
|
56
|
+
}
|
|
57
|
+
catch {
|
|
58
|
+
// 坏行跳过(不因一行损坏丢掉整份历史)
|
|
59
|
+
}
|
|
60
|
+
}
|
|
61
|
+
return limit > 0 ? entries.slice(-limit) : entries;
|
|
62
|
+
}
|
|
63
|
+
// ─── 可视化 ───────────────────────────────────────────────────────
|
|
64
|
+
const SPARK_CHARS = '▁▂▃▄▅▆▇█';
|
|
65
|
+
/** 0~1 序列 → sparkline 字符串 */
|
|
66
|
+
export function sparkline(values, width = 20) {
|
|
67
|
+
if (values.length === 0)
|
|
68
|
+
return '';
|
|
69
|
+
const seq = values.slice(-width);
|
|
70
|
+
return seq
|
|
71
|
+
.map((v) => {
|
|
72
|
+
const clamped = Math.min(1, Math.max(0, Number.isFinite(v) ? v : 0));
|
|
73
|
+
const idx = Math.round(clamped * (SPARK_CHARS.length - 1));
|
|
74
|
+
return SPARK_CHARS[idx];
|
|
75
|
+
})
|
|
76
|
+
.join('');
|
|
77
|
+
}
|
|
78
|
+
function pct(x) {
|
|
79
|
+
return `${(x * 100).toFixed(1)}%`;
|
|
80
|
+
}
|
|
81
|
+
function deltaCell(cur, prev) {
|
|
82
|
+
if (prev === undefined)
|
|
83
|
+
return '—';
|
|
84
|
+
const d = (cur - prev) * 100;
|
|
85
|
+
if (Math.abs(d) < 0.05)
|
|
86
|
+
return '±0.0%';
|
|
87
|
+
const sign = d > 0 ? '+' : '';
|
|
88
|
+
const icon = d < 0 ? '🔻' : '🔺';
|
|
89
|
+
return `${icon}${sign}${d.toFixed(1)}%`;
|
|
90
|
+
}
|
|
91
|
+
function shortAt(iso) {
|
|
92
|
+
// 2026-09-29T03:12:44.123Z → 09-29 03:12
|
|
93
|
+
return `${iso.slice(5, 10)} ${iso.slice(11, 16)}`;
|
|
94
|
+
}
|
|
95
|
+
/**
|
|
96
|
+
* 渲染通过率趋势(Markdown)。
|
|
97
|
+
* 环比定义:与上一次运行对比(cron 每日跑 → 即日环比)。
|
|
98
|
+
*/
|
|
99
|
+
export function renderHistoryTrend(entries, limit = 15) {
|
|
100
|
+
if (entries.length === 0)
|
|
101
|
+
return '(暂无历史记录)';
|
|
102
|
+
const rows = entries.slice(-limit);
|
|
103
|
+
const lines = [];
|
|
104
|
+
const rates = rows.map((e) => e.passRate);
|
|
105
|
+
const latest = rows[rows.length - 1];
|
|
106
|
+
const prev = rows.length > 1 ? rows[rows.length - 2] : undefined;
|
|
107
|
+
lines.push(`\`${sparkline(rates)}\` 最近 ${rows.length} 次:${pct(latest.passRate)}(环比 ${deltaCell(latest.passRate, prev?.passRate)})`);
|
|
108
|
+
lines.push('');
|
|
109
|
+
lines.push(`| 时间(UTC) | 模式 | 模型 | 通过 | 通过率 | 环比 Δ | 平均分 | 耗时 |`);
|
|
110
|
+
lines.push(`|---|---|---|---|---|---|---|---|`);
|
|
111
|
+
rows.forEach((e, i) => {
|
|
112
|
+
lines.push(`| ${shortAt(e.at)} | ${e.mode} | ${e.model ?? '—'} | ${e.passed}/${e.cases} | ${pct(e.passRate)} | ${deltaCell(e.passRate, i > 0 ? rows[i - 1].passRate : undefined)} | ${e.avgScore.toFixed(2)} | ${(e.durationMs / 1000).toFixed(1)}s |`);
|
|
113
|
+
});
|
|
114
|
+
// 分套件环比:定位是哪一组掉了
|
|
115
|
+
const suiteDeltas = [];
|
|
116
|
+
if (prev) {
|
|
117
|
+
for (const [suite, cur] of Object.entries(latest.bySuite)) {
|
|
118
|
+
const before = prev.bySuite[suite];
|
|
119
|
+
if (!before)
|
|
120
|
+
continue;
|
|
121
|
+
const d = (cur.passRate - before.passRate) * 100;
|
|
122
|
+
if (Math.abs(d) >= 0.05) {
|
|
123
|
+
suiteDeltas.push(` - ${suite}: ${pct(before.passRate)} → ${pct(cur.passRate)}(${d > 0 ? '+' : ''}${d.toFixed(1)}%)`);
|
|
124
|
+
}
|
|
125
|
+
}
|
|
126
|
+
}
|
|
127
|
+
if (suiteDeltas.length > 0) {
|
|
128
|
+
lines.push('');
|
|
129
|
+
lines.push(`分套件环比:`);
|
|
130
|
+
lines.push(...suiteDeltas);
|
|
131
|
+
}
|
|
132
|
+
return lines.join('\n');
|
|
133
|
+
}
|
|
134
|
+
//# sourceMappingURL=history.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"history.js","sourceRoot":"","sources":["../../../src/core/history.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,MAAM,kBAAkB,CAAC;AAC/D,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAoB7C,gBAAgB;AAChB,MAAM,UAAU,oBAAoB,CAAC,MAAkB;IACrD,OAAO;QACL,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,EAAE,EAAE,MAAM,CAAC,SAAS;QACpB,IAAI,EAAE,MAAM,CAAC,IAAI;QACjB,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,KAAK,EAAE,MAAM,CAAC,MAAM,CAAC,KAAK;QAC1B,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,MAAM;QAC5B,OAAO,EAAE,MAAM,CAAC,MAAM,CAAC,OAAO;QAC9B,QAAQ,EAAE,MAAM,CAAC,MAAM,CAAC,QAAQ;QAChC,QAAQ,EAAE,MAAM,CAAC,MAAM,CAAC,QAAQ;QAChC,UAAU,EAAE,MAAM,CAAC,UAAU;QAC7B,UAAU,EAAE,MAAM,CAAC,MAAM,CAAC,UAAU;QACpC,OAAO,EAAE,MAAM,CAAC,WAAW,CACzB,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;YAC7C,CAAC;YACD,EAAE,KAAK,EAAE,CAAC,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE;SAC3D,CAAC,CACH;KACF,CAAC;AACJ,CAAC;AAED,2BAA2B;AAC3B,MAAM,CAAC,KAAK,UAAU,aAAa,CACjC,MAAkB,EAClB,WAAmB;IAEnB,MAAM,IAAI,GAAG,OAAO,CAAC,WAAW,CAAC,CAAC;IAClC,MAAM,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IAChD,MAAM,UAAU,CAAC,IAAI,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,oBAAoB,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IACrF,OAAO,IAAI,CAAC;AACd,CAAC;AAED,6BAA6B;AAC7B,MAAM,CAAC,KAAK,UAAU,WAAW,CAC/B,WAAmB,EACnB,KAAK,GAAG,EAAE;IAEV,MAAM,IAAI,GAAG,OAAO,CAAC,WAAW,CAAC,CAAC;IAClC,IAAI,GAAW,CAAC;IAChB,IAAI,CAAC;QACH,GAAG,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,MAAM,OAAO,GAAmB,EAAE,CAAC;IACnC,KAAK,MAAM,IAAI,IAAI,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QACnC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QACtB,IAAI,CAAC,CAAC;YAAE,SAAS;QACjB,IAAI,CAAC;YACH,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAiB,CAAC,CAAC;QAC9C,CAAC;QAAC,MAAM,CAAC;YACP,qBAAqB;QACvB,CAAC;IACH,CAAC;IACD,OAAO,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC;AACrD,CAAC;AAED,kEAAkE;AAElE,MAAM,WAAW,GAAG,UAAU,CAAC;AAE/B,6BAA6B;AAC7B,MAAM,UAAU,SAAS,CAAC,MAAgB,EAAE,KAAK,GAAG,EAAE;IACpD,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IACnC,MAAM,GAAG,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC;IACjC,OAAO,GAAG;SACP,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE;QACT,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACrE,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,GAAG,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;QAC3D,OAAO,WAAW,CAAC,GAAG,CAAC,CAAC;IAC1B,CAAC,CAAC;SACD,IAAI,CAAC,EAAE,CAAC,CAAC;AACd,CAAC;AAED,SAAS,GAAG,CAAC,CAAS;IACpB,OAAO,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC;AACpC,CAAC;AAED,SAAS,SAAS,CAAC,GAAW,EAAE,IAAwB;IACtD,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,GAAG,CAAC;IACnC,MAAM,CAAC,GAAG,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG,GAAG,CAAC;IAC7B,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI;QAAE,OAAO,OAAO,CAAC;IACvC,MAAM,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;IAC9B,MAAM,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;IACjC,OAAO,GAAG,IAAI,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC;AAC1C,CAAC;AAED,SAAS,OAAO,CAAC,GAAW;IAC1B,yCAAyC;IACzC,OAAO,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,GAAG,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC;AACpD,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,kBAAkB,CAAC,OAAuB,EAAE,KAAK,GAAG,EAAE;IACpE,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,UAAU,CAAC;IAE5C,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC;IACnC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC;IAC1C,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IACrC,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAEjE,KAAK,CAAC,IAAI,CAAC,KAAK,SAAS,CAAC,KAAK,CAAC,UAAU,IAAI,CAAC,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,OAAO,SAAS,CAAC,MAAM,CAAC,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,GAAG,CAAC,CAAC;IACrI,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACf,KAAK,CAAC,IAAI,CAAC,oDAAoD,CAAC,CAAC;IACjE,KAAK,CAAC,IAAI,CAAC,mCAAmC,CAAC,CAAC;IAChD,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;QACpB,KAAK,CAAC,IAAI,CACR,KAAK,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC,IAAI,MAAM,CAAC,CAAC,KAAK,IAAI,GAAG,MAAM,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,KAAK,MAAM,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,SAAS,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,UAAU,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAC5O,CAAC;IACJ,CAAC,CAAC,CAAC;IAEH,iBAAiB;IACjB,MAAM,WAAW,GAAa,EAAE,CAAC;IACjC,IAAI,IAAI,EAAE,CAAC;QACT,KAAK,MAAM,CAAC,KAAK,EAAE,GAAG,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE,CAAC;YAC1D,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;YACnC,IAAI,CAAC,MAAM;gBAAE,SAAS;YACtB,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,GAAG,CAAC;YACjD,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,IAAI,EAAE,CAAC;gBACxB,WAAW,CAAC,IAAI,CAAC,OAAO,KAAK,KAAK,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;YACxH,CAAC;QACH,CAAC;IACH,CAAC;IACD,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC3B,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACf,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACrB,KAAK,CAAC,IAAI,CAAC,GAAG,WAAW,CAAC,CAAC;IAC7B,CAAC;IAED,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC1B,CAAC"}
|
|
@@ -0,0 +1,77 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* packages/eval/src/core/live.ts - live 模式(真实 LLM)装配
|
|
3
|
+
*
|
|
4
|
+
* 职责:把「provider/modelId + apiKey + baseUrl」装配成 Runtime 可直接消费的
|
|
5
|
+
* `{ model, streamFn }`,与 mock 模式的 fixture replay 对称。
|
|
6
|
+
*
|
|
7
|
+
* 消随机性(EVAL_PLAN.md 4.3):
|
|
8
|
+
* - temperature 缺省 0(provider 支持即生效)
|
|
9
|
+
* - 本仓库适配层不支持 seed → 靠 repeats + pass@k 消噪,见 runner.ts
|
|
10
|
+
*
|
|
11
|
+
* 装配路径:getBuiltinModel(内置目录)→ 兜底构造 custom model(代理 / 兼容网关)
|
|
12
|
+
* → adaptAiModel(框架 Model)+ createStreamFnFromAi(StreamFn)
|
|
13
|
+
*/
|
|
14
|
+
import type { Model as AiModel } from '@aipack-ai/agent/ai';
|
|
15
|
+
import type { Model, StreamFn } from '@aipack-ai/agent';
|
|
16
|
+
export interface LiveLlmSpec {
|
|
17
|
+
/** 提供商 id:deepseek / openai / anthropic / ... */
|
|
18
|
+
provider: string;
|
|
19
|
+
/** 模型 id:deepseek-chat / gpt-4o-mini / ...(不在内置目录时按 custom 兜底) */
|
|
20
|
+
modelId: string;
|
|
21
|
+
/** API Key;缺省取 <PROVIDER>_API_KEY 环境变量 */
|
|
22
|
+
apiKey?: string;
|
|
23
|
+
/** 覆盖内置端点(代理 / 兼容网关) */
|
|
24
|
+
baseUrl?: string;
|
|
25
|
+
/** 缺省 0(消随机性) */
|
|
26
|
+
temperature?: number;
|
|
27
|
+
/** 单次请求超时 ms */
|
|
28
|
+
timeoutMs?: number;
|
|
29
|
+
}
|
|
30
|
+
export interface LiveLlm {
|
|
31
|
+
spec: LiveLlmSpec;
|
|
32
|
+
/** AI 层模型(含 api / baseUrl) */
|
|
33
|
+
aiModel: AiModel;
|
|
34
|
+
/** 框架层 Model,放进 RuntimeOptions.model */
|
|
35
|
+
model: Model;
|
|
36
|
+
/** 真实流式函数,放进 RuntimeOptions.streamFn */
|
|
37
|
+
streamFn: StreamFn;
|
|
38
|
+
}
|
|
39
|
+
/** env 覆盖键(优先级低于显式参数,便于 CI 用 secrets 注入) */
|
|
40
|
+
export declare const LIVE_ENV_KEYS: {
|
|
41
|
+
readonly model: "AIPACK_EVAL_MODEL";
|
|
42
|
+
readonly provider: "AIPACK_EVAL_PROVIDER";
|
|
43
|
+
readonly apiKey: "AIPACK_EVAL_API_KEY";
|
|
44
|
+
readonly baseUrl: "AIPACK_EVAL_BASE_URL";
|
|
45
|
+
readonly temperature: "AIPACK_EVAL_TEMPERATURE";
|
|
46
|
+
};
|
|
47
|
+
/**
|
|
48
|
+
* 解析 'provider/modelId' 或 'modelId'。
|
|
49
|
+
* 仅给 modelId 时:优先 AIPACK_EVAL_PROVIDER,否则探测第一个已配置 Key 的 provider。
|
|
50
|
+
*/
|
|
51
|
+
export declare function parseModelSpec(spec: string): {
|
|
52
|
+
provider: string;
|
|
53
|
+
modelId: string;
|
|
54
|
+
};
|
|
55
|
+
export interface ResolveLiveOptions {
|
|
56
|
+
/** 'provider/modelId' 或 'modelId';缺省读 AIPACK_EVAL_MODEL */
|
|
57
|
+
model?: string;
|
|
58
|
+
apiKey?: string;
|
|
59
|
+
baseUrl?: string;
|
|
60
|
+
temperature?: number;
|
|
61
|
+
timeoutMs?: number;
|
|
62
|
+
}
|
|
63
|
+
/** 合并显式参数与环境变量,得到最终生效的 LiveLlmSpec */
|
|
64
|
+
export declare function resolveLiveSpec(opts?: ResolveLiveOptions): LiveLlmSpec;
|
|
65
|
+
/**
|
|
66
|
+
* 装配真实 LLM。不发起任何网络请求(只是闭包),可安全单测。
|
|
67
|
+
*
|
|
68
|
+
* 注意:createStreamFnFromAi 优先使用 runtime 传入的 model,
|
|
69
|
+
* 因此必须把返回的 model 一起交给 createRuntime,否则 api 分派会不一致。
|
|
70
|
+
*/
|
|
71
|
+
export declare function createLiveLlm(spec: LiveLlmSpec): LiveLlm;
|
|
72
|
+
/** 从 CLI/库参数一步到位:解析 spec → 装配 LLM */
|
|
73
|
+
export declare function resolveLiveLlm(opts?: ResolveLiveOptions): LiveLlm;
|
|
74
|
+
/** 报告 / 日志用的模型标识('deepseek/deepseek-chat') */
|
|
75
|
+
export declare function describeLiveLlm(spec: LiveLlmSpec): string;
|
|
76
|
+
/** live 模式前置检查:缺 Key 时给出可操作的错误信息 */
|
|
77
|
+
export declare function assertLiveReady(spec: LiveLlmSpec): void;
|