@openpond/evals 0.2.0 → 0.4.0
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/CONTRACT.md +24 -5
- package/README.md +27 -8
- package/RELEASING.md +13 -45
- package/dist/compatibility.js +45 -0
- package/dist/conformance.js +7 -11
- package/dist/evidence/authoring.js +1 -1
- package/dist/evidence/conformance.js +1 -1
- package/dist/evidence/contracts.js +1 -2
- package/dist/evidence/eligibility.js +1 -1
- package/dist/graders.js +1 -1
- package/dist/harness.js +33 -103
- package/dist/index.js +3 -1
- package/dist/model-improvement-qualification.js +125 -0
- package/dist/review-conformance.js +209 -0
- package/dist/runs.js +33 -13
- package/dist/tasksets.js +3 -17
- package/dist/types/compatibility.d.ts +11 -0
- package/dist/types/compatibility.d.ts.map +1 -0
- package/dist/types/conformance.d.ts +10 -48
- package/dist/types/conformance.d.ts.map +1 -1
- package/dist/types/evidence/conformance.d.ts +8 -8
- package/dist/types/evidence/contracts.d.ts +8 -8
- package/dist/types/harness.d.ts +3 -471
- package/dist/types/harness.d.ts.map +1 -1
- package/dist/types/index.d.ts +3 -1
- package/dist/types/index.d.ts.map +1 -1
- package/dist/types/model-improvement-qualification.d.ts +222 -0
- package/dist/types/model-improvement-qualification.d.ts.map +1 -0
- package/dist/types/review-conformance.d.ts +607 -0
- package/dist/types/review-conformance.d.ts.map +1 -0
- package/dist/types/runs.d.ts +47 -10
- package/dist/types/runs.d.ts.map +1 -1
- package/dist/types/tasksets.d.ts +4 -35
- package/dist/types/tasksets.d.ts.map +1 -1
- package/package.json +14 -2
- package/dist/common.js +0 -67
- package/dist/sha256.js +0 -91
- package/dist/types/common.d.ts +0 -51
- package/dist/types/common.d.ts.map +0 -1
- package/dist/types/sha256.d.ts +0 -2
- package/dist/types/sha256.d.ts.map +0 -1
|
@@ -0,0 +1,607 @@
|
|
|
1
|
+
export declare const harnessEvaluationReviewConformance: {
|
|
2
|
+
noAction: {
|
|
3
|
+
schemaVersion: "openpond.harnessEvaluationReviewReceipt.v1";
|
|
4
|
+
id: string;
|
|
5
|
+
ownerScope: {
|
|
6
|
+
kind: "personal" | "team";
|
|
7
|
+
id: string;
|
|
8
|
+
};
|
|
9
|
+
workspaceRef: string;
|
|
10
|
+
harnessRelease: {
|
|
11
|
+
id: string;
|
|
12
|
+
contentHash: string;
|
|
13
|
+
};
|
|
14
|
+
previousWatermark: {
|
|
15
|
+
cursor: string;
|
|
16
|
+
throughCreatedAt: string;
|
|
17
|
+
} | null;
|
|
18
|
+
nextWatermark: {
|
|
19
|
+
cursor: string;
|
|
20
|
+
throughCreatedAt: string;
|
|
21
|
+
};
|
|
22
|
+
selectedEvidence: {
|
|
23
|
+
evidence: {
|
|
24
|
+
id: string;
|
|
25
|
+
contentHash: string;
|
|
26
|
+
};
|
|
27
|
+
kind: "validation" | "observation" | "taskset" | "trigger" | "route_decision" | "refiner_outcome" | "proposal" | "apply_receipt" | "harness_advance" | "rollback" | "work_outcome" | "evaluation" | "training_qualification" | "model_candidate";
|
|
28
|
+
sourceRef: string;
|
|
29
|
+
sourcePolicy: {
|
|
30
|
+
policy: {
|
|
31
|
+
id: string;
|
|
32
|
+
contentHash: string;
|
|
33
|
+
};
|
|
34
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
35
|
+
checkedAt: string;
|
|
36
|
+
};
|
|
37
|
+
occurrenceKey: string;
|
|
38
|
+
occurredAt: string;
|
|
39
|
+
}[];
|
|
40
|
+
excludedEvidence: {
|
|
41
|
+
evidence: {
|
|
42
|
+
id: string;
|
|
43
|
+
contentHash: string;
|
|
44
|
+
};
|
|
45
|
+
sourcePolicy: {
|
|
46
|
+
policy: {
|
|
47
|
+
id: string;
|
|
48
|
+
contentHash: string;
|
|
49
|
+
};
|
|
50
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
51
|
+
checkedAt: string;
|
|
52
|
+
} | null;
|
|
53
|
+
reason: "revoked" | "deleted" | "expired" | "outside_scope" | "before_watermark" | "duplicate" | "resolved" | "sensitive" | "unverified" | "budget";
|
|
54
|
+
}[];
|
|
55
|
+
claim: {
|
|
56
|
+
fingerprint: string;
|
|
57
|
+
recurrenceFamily: string;
|
|
58
|
+
statement: string;
|
|
59
|
+
independentOccurrences: number;
|
|
60
|
+
unresolvedOccurrences: number;
|
|
61
|
+
} | null;
|
|
62
|
+
classification: "runtime" | "taskset" | "no_action" | "harness_maintenance" | "product" | "model_improvement";
|
|
63
|
+
triage: {
|
|
64
|
+
layer: "model" | "tools" | "runtime" | "evaluation" | "product" | "harness" | "retrieval";
|
|
65
|
+
status: "resolved" | "not_applicable" | "unresolved" | "blocked";
|
|
66
|
+
reason: string;
|
|
67
|
+
evidenceRefs: {
|
|
68
|
+
id: string;
|
|
69
|
+
contentHash: string;
|
|
70
|
+
}[];
|
|
71
|
+
}[];
|
|
72
|
+
reason: string;
|
|
73
|
+
nextAuthority: "none" | "runtime_service" | "product_team" | "human_review" | "evaluation_system" | "training_system";
|
|
74
|
+
maxEstimatedCostUsd: number;
|
|
75
|
+
tasksetProposal: {
|
|
76
|
+
id: string;
|
|
77
|
+
contentHash: string;
|
|
78
|
+
} | null;
|
|
79
|
+
evaluation: {
|
|
80
|
+
id: string;
|
|
81
|
+
contentHash: string;
|
|
82
|
+
} | null;
|
|
83
|
+
trainingQualification: {
|
|
84
|
+
id: string;
|
|
85
|
+
contentHash: string;
|
|
86
|
+
} | null;
|
|
87
|
+
policyVersion: string;
|
|
88
|
+
createdAt: string;
|
|
89
|
+
metadata: Record<string, unknown>;
|
|
90
|
+
contentHash: string;
|
|
91
|
+
};
|
|
92
|
+
runtime: {
|
|
93
|
+
schemaVersion: "openpond.harnessEvaluationReviewReceipt.v1";
|
|
94
|
+
id: string;
|
|
95
|
+
ownerScope: {
|
|
96
|
+
kind: "personal" | "team";
|
|
97
|
+
id: string;
|
|
98
|
+
};
|
|
99
|
+
workspaceRef: string;
|
|
100
|
+
harnessRelease: {
|
|
101
|
+
id: string;
|
|
102
|
+
contentHash: string;
|
|
103
|
+
};
|
|
104
|
+
previousWatermark: {
|
|
105
|
+
cursor: string;
|
|
106
|
+
throughCreatedAt: string;
|
|
107
|
+
} | null;
|
|
108
|
+
nextWatermark: {
|
|
109
|
+
cursor: string;
|
|
110
|
+
throughCreatedAt: string;
|
|
111
|
+
};
|
|
112
|
+
selectedEvidence: {
|
|
113
|
+
evidence: {
|
|
114
|
+
id: string;
|
|
115
|
+
contentHash: string;
|
|
116
|
+
};
|
|
117
|
+
kind: "validation" | "observation" | "taskset" | "trigger" | "route_decision" | "refiner_outcome" | "proposal" | "apply_receipt" | "harness_advance" | "rollback" | "work_outcome" | "evaluation" | "training_qualification" | "model_candidate";
|
|
118
|
+
sourceRef: string;
|
|
119
|
+
sourcePolicy: {
|
|
120
|
+
policy: {
|
|
121
|
+
id: string;
|
|
122
|
+
contentHash: string;
|
|
123
|
+
};
|
|
124
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
125
|
+
checkedAt: string;
|
|
126
|
+
};
|
|
127
|
+
occurrenceKey: string;
|
|
128
|
+
occurredAt: string;
|
|
129
|
+
}[];
|
|
130
|
+
excludedEvidence: {
|
|
131
|
+
evidence: {
|
|
132
|
+
id: string;
|
|
133
|
+
contentHash: string;
|
|
134
|
+
};
|
|
135
|
+
sourcePolicy: {
|
|
136
|
+
policy: {
|
|
137
|
+
id: string;
|
|
138
|
+
contentHash: string;
|
|
139
|
+
};
|
|
140
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
141
|
+
checkedAt: string;
|
|
142
|
+
} | null;
|
|
143
|
+
reason: "revoked" | "deleted" | "expired" | "outside_scope" | "before_watermark" | "duplicate" | "resolved" | "sensitive" | "unverified" | "budget";
|
|
144
|
+
}[];
|
|
145
|
+
claim: {
|
|
146
|
+
fingerprint: string;
|
|
147
|
+
recurrenceFamily: string;
|
|
148
|
+
statement: string;
|
|
149
|
+
independentOccurrences: number;
|
|
150
|
+
unresolvedOccurrences: number;
|
|
151
|
+
} | null;
|
|
152
|
+
classification: "runtime" | "taskset" | "no_action" | "harness_maintenance" | "product" | "model_improvement";
|
|
153
|
+
triage: {
|
|
154
|
+
layer: "model" | "tools" | "runtime" | "evaluation" | "product" | "harness" | "retrieval";
|
|
155
|
+
status: "resolved" | "not_applicable" | "unresolved" | "blocked";
|
|
156
|
+
reason: string;
|
|
157
|
+
evidenceRefs: {
|
|
158
|
+
id: string;
|
|
159
|
+
contentHash: string;
|
|
160
|
+
}[];
|
|
161
|
+
}[];
|
|
162
|
+
reason: string;
|
|
163
|
+
nextAuthority: "none" | "runtime_service" | "product_team" | "human_review" | "evaluation_system" | "training_system";
|
|
164
|
+
maxEstimatedCostUsd: number;
|
|
165
|
+
tasksetProposal: {
|
|
166
|
+
id: string;
|
|
167
|
+
contentHash: string;
|
|
168
|
+
} | null;
|
|
169
|
+
evaluation: {
|
|
170
|
+
id: string;
|
|
171
|
+
contentHash: string;
|
|
172
|
+
} | null;
|
|
173
|
+
trainingQualification: {
|
|
174
|
+
id: string;
|
|
175
|
+
contentHash: string;
|
|
176
|
+
} | null;
|
|
177
|
+
policyVersion: string;
|
|
178
|
+
createdAt: string;
|
|
179
|
+
metadata: Record<string, unknown>;
|
|
180
|
+
contentHash: string;
|
|
181
|
+
};
|
|
182
|
+
product: {
|
|
183
|
+
schemaVersion: "openpond.harnessEvaluationReviewReceipt.v1";
|
|
184
|
+
id: string;
|
|
185
|
+
ownerScope: {
|
|
186
|
+
kind: "personal" | "team";
|
|
187
|
+
id: string;
|
|
188
|
+
};
|
|
189
|
+
workspaceRef: string;
|
|
190
|
+
harnessRelease: {
|
|
191
|
+
id: string;
|
|
192
|
+
contentHash: string;
|
|
193
|
+
};
|
|
194
|
+
previousWatermark: {
|
|
195
|
+
cursor: string;
|
|
196
|
+
throughCreatedAt: string;
|
|
197
|
+
} | null;
|
|
198
|
+
nextWatermark: {
|
|
199
|
+
cursor: string;
|
|
200
|
+
throughCreatedAt: string;
|
|
201
|
+
};
|
|
202
|
+
selectedEvidence: {
|
|
203
|
+
evidence: {
|
|
204
|
+
id: string;
|
|
205
|
+
contentHash: string;
|
|
206
|
+
};
|
|
207
|
+
kind: "validation" | "observation" | "taskset" | "trigger" | "route_decision" | "refiner_outcome" | "proposal" | "apply_receipt" | "harness_advance" | "rollback" | "work_outcome" | "evaluation" | "training_qualification" | "model_candidate";
|
|
208
|
+
sourceRef: string;
|
|
209
|
+
sourcePolicy: {
|
|
210
|
+
policy: {
|
|
211
|
+
id: string;
|
|
212
|
+
contentHash: string;
|
|
213
|
+
};
|
|
214
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
215
|
+
checkedAt: string;
|
|
216
|
+
};
|
|
217
|
+
occurrenceKey: string;
|
|
218
|
+
occurredAt: string;
|
|
219
|
+
}[];
|
|
220
|
+
excludedEvidence: {
|
|
221
|
+
evidence: {
|
|
222
|
+
id: string;
|
|
223
|
+
contentHash: string;
|
|
224
|
+
};
|
|
225
|
+
sourcePolicy: {
|
|
226
|
+
policy: {
|
|
227
|
+
id: string;
|
|
228
|
+
contentHash: string;
|
|
229
|
+
};
|
|
230
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
231
|
+
checkedAt: string;
|
|
232
|
+
} | null;
|
|
233
|
+
reason: "revoked" | "deleted" | "expired" | "outside_scope" | "before_watermark" | "duplicate" | "resolved" | "sensitive" | "unverified" | "budget";
|
|
234
|
+
}[];
|
|
235
|
+
claim: {
|
|
236
|
+
fingerprint: string;
|
|
237
|
+
recurrenceFamily: string;
|
|
238
|
+
statement: string;
|
|
239
|
+
independentOccurrences: number;
|
|
240
|
+
unresolvedOccurrences: number;
|
|
241
|
+
} | null;
|
|
242
|
+
classification: "runtime" | "taskset" | "no_action" | "harness_maintenance" | "product" | "model_improvement";
|
|
243
|
+
triage: {
|
|
244
|
+
layer: "model" | "tools" | "runtime" | "evaluation" | "product" | "harness" | "retrieval";
|
|
245
|
+
status: "resolved" | "not_applicable" | "unresolved" | "blocked";
|
|
246
|
+
reason: string;
|
|
247
|
+
evidenceRefs: {
|
|
248
|
+
id: string;
|
|
249
|
+
contentHash: string;
|
|
250
|
+
}[];
|
|
251
|
+
}[];
|
|
252
|
+
reason: string;
|
|
253
|
+
nextAuthority: "none" | "runtime_service" | "product_team" | "human_review" | "evaluation_system" | "training_system";
|
|
254
|
+
maxEstimatedCostUsd: number;
|
|
255
|
+
tasksetProposal: {
|
|
256
|
+
id: string;
|
|
257
|
+
contentHash: string;
|
|
258
|
+
} | null;
|
|
259
|
+
evaluation: {
|
|
260
|
+
id: string;
|
|
261
|
+
contentHash: string;
|
|
262
|
+
} | null;
|
|
263
|
+
trainingQualification: {
|
|
264
|
+
id: string;
|
|
265
|
+
contentHash: string;
|
|
266
|
+
} | null;
|
|
267
|
+
policyVersion: string;
|
|
268
|
+
createdAt: string;
|
|
269
|
+
metadata: Record<string, unknown>;
|
|
270
|
+
contentHash: string;
|
|
271
|
+
};
|
|
272
|
+
taskset: {
|
|
273
|
+
schemaVersion: "openpond.harnessEvaluationReviewReceipt.v1";
|
|
274
|
+
id: string;
|
|
275
|
+
ownerScope: {
|
|
276
|
+
kind: "personal" | "team";
|
|
277
|
+
id: string;
|
|
278
|
+
};
|
|
279
|
+
workspaceRef: string;
|
|
280
|
+
harnessRelease: {
|
|
281
|
+
id: string;
|
|
282
|
+
contentHash: string;
|
|
283
|
+
};
|
|
284
|
+
previousWatermark: {
|
|
285
|
+
cursor: string;
|
|
286
|
+
throughCreatedAt: string;
|
|
287
|
+
} | null;
|
|
288
|
+
nextWatermark: {
|
|
289
|
+
cursor: string;
|
|
290
|
+
throughCreatedAt: string;
|
|
291
|
+
};
|
|
292
|
+
selectedEvidence: {
|
|
293
|
+
evidence: {
|
|
294
|
+
id: string;
|
|
295
|
+
contentHash: string;
|
|
296
|
+
};
|
|
297
|
+
kind: "validation" | "observation" | "taskset" | "trigger" | "route_decision" | "refiner_outcome" | "proposal" | "apply_receipt" | "harness_advance" | "rollback" | "work_outcome" | "evaluation" | "training_qualification" | "model_candidate";
|
|
298
|
+
sourceRef: string;
|
|
299
|
+
sourcePolicy: {
|
|
300
|
+
policy: {
|
|
301
|
+
id: string;
|
|
302
|
+
contentHash: string;
|
|
303
|
+
};
|
|
304
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
305
|
+
checkedAt: string;
|
|
306
|
+
};
|
|
307
|
+
occurrenceKey: string;
|
|
308
|
+
occurredAt: string;
|
|
309
|
+
}[];
|
|
310
|
+
excludedEvidence: {
|
|
311
|
+
evidence: {
|
|
312
|
+
id: string;
|
|
313
|
+
contentHash: string;
|
|
314
|
+
};
|
|
315
|
+
sourcePolicy: {
|
|
316
|
+
policy: {
|
|
317
|
+
id: string;
|
|
318
|
+
contentHash: string;
|
|
319
|
+
};
|
|
320
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
321
|
+
checkedAt: string;
|
|
322
|
+
} | null;
|
|
323
|
+
reason: "revoked" | "deleted" | "expired" | "outside_scope" | "before_watermark" | "duplicate" | "resolved" | "sensitive" | "unverified" | "budget";
|
|
324
|
+
}[];
|
|
325
|
+
claim: {
|
|
326
|
+
fingerprint: string;
|
|
327
|
+
recurrenceFamily: string;
|
|
328
|
+
statement: string;
|
|
329
|
+
independentOccurrences: number;
|
|
330
|
+
unresolvedOccurrences: number;
|
|
331
|
+
} | null;
|
|
332
|
+
classification: "runtime" | "taskset" | "no_action" | "harness_maintenance" | "product" | "model_improvement";
|
|
333
|
+
triage: {
|
|
334
|
+
layer: "model" | "tools" | "runtime" | "evaluation" | "product" | "harness" | "retrieval";
|
|
335
|
+
status: "resolved" | "not_applicable" | "unresolved" | "blocked";
|
|
336
|
+
reason: string;
|
|
337
|
+
evidenceRefs: {
|
|
338
|
+
id: string;
|
|
339
|
+
contentHash: string;
|
|
340
|
+
}[];
|
|
341
|
+
}[];
|
|
342
|
+
reason: string;
|
|
343
|
+
nextAuthority: "none" | "runtime_service" | "product_team" | "human_review" | "evaluation_system" | "training_system";
|
|
344
|
+
maxEstimatedCostUsd: number;
|
|
345
|
+
tasksetProposal: {
|
|
346
|
+
id: string;
|
|
347
|
+
contentHash: string;
|
|
348
|
+
} | null;
|
|
349
|
+
evaluation: {
|
|
350
|
+
id: string;
|
|
351
|
+
contentHash: string;
|
|
352
|
+
} | null;
|
|
353
|
+
trainingQualification: {
|
|
354
|
+
id: string;
|
|
355
|
+
contentHash: string;
|
|
356
|
+
} | null;
|
|
357
|
+
policyVersion: string;
|
|
358
|
+
createdAt: string;
|
|
359
|
+
metadata: Record<string, unknown>;
|
|
360
|
+
contentHash: string;
|
|
361
|
+
};
|
|
362
|
+
blockedRl: {
|
|
363
|
+
schemaVersion: "openpond.modelImprovementQualificationReceipt.v1";
|
|
364
|
+
id: string;
|
|
365
|
+
review: {
|
|
366
|
+
id: string;
|
|
367
|
+
contentHash: string;
|
|
368
|
+
};
|
|
369
|
+
harnessRelease: {
|
|
370
|
+
id: string;
|
|
371
|
+
contentHash: string;
|
|
372
|
+
};
|
|
373
|
+
tasksetRelease: {
|
|
374
|
+
id: string;
|
|
375
|
+
contentHash: string;
|
|
376
|
+
} | null;
|
|
377
|
+
baselineEvaluation: {
|
|
378
|
+
id: string;
|
|
379
|
+
contentHash: string;
|
|
380
|
+
} | null;
|
|
381
|
+
model: {
|
|
382
|
+
provider: string;
|
|
383
|
+
model: string;
|
|
384
|
+
revision: string | null;
|
|
385
|
+
artifactHash: string | null;
|
|
386
|
+
tokenizerRevision: string | null;
|
|
387
|
+
chatTemplateHash: string | null;
|
|
388
|
+
};
|
|
389
|
+
environmentHash: string | null;
|
|
390
|
+
toolContractHash: string | null;
|
|
391
|
+
permissionContractHash: string | null;
|
|
392
|
+
policyHash: string | null;
|
|
393
|
+
verifierRef: {
|
|
394
|
+
id: string;
|
|
395
|
+
contentHash: string;
|
|
396
|
+
} | null;
|
|
397
|
+
sourcePolicies: {
|
|
398
|
+
policy: {
|
|
399
|
+
id: string;
|
|
400
|
+
contentHash: string;
|
|
401
|
+
};
|
|
402
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
403
|
+
checkedAt: string;
|
|
404
|
+
}[];
|
|
405
|
+
trainingEvidenceRefs: {
|
|
406
|
+
id: string;
|
|
407
|
+
contentHash: string;
|
|
408
|
+
}[];
|
|
409
|
+
frozenEvaluationEvidenceRefs: {
|
|
410
|
+
id: string;
|
|
411
|
+
contentHash: string;
|
|
412
|
+
}[];
|
|
413
|
+
privacyApproval: {
|
|
414
|
+
id: string;
|
|
415
|
+
contentHash: string;
|
|
416
|
+
} | null;
|
|
417
|
+
budgetApproval: {
|
|
418
|
+
id: string;
|
|
419
|
+
contentHash: string;
|
|
420
|
+
} | null;
|
|
421
|
+
maximumCostUsd: number;
|
|
422
|
+
signal: {
|
|
423
|
+
kind: "none" | "demonstrations" | "chosen_rejected" | "scalar_reward";
|
|
424
|
+
strength: "absent" | "weak" | "usable";
|
|
425
|
+
calibrated: boolean;
|
|
426
|
+
confounded: boolean;
|
|
427
|
+
variance: number | null;
|
|
428
|
+
evidenceRefs: {
|
|
429
|
+
id: string;
|
|
430
|
+
contentHash: string;
|
|
431
|
+
}[];
|
|
432
|
+
};
|
|
433
|
+
decision: "no_training" | "sft" | "preference" | "rl";
|
|
434
|
+
reasons: string[];
|
|
435
|
+
createdAt: string;
|
|
436
|
+
metadata: Record<string, unknown>;
|
|
437
|
+
contentHash: string;
|
|
438
|
+
};
|
|
439
|
+
qualifiedRl: {
|
|
440
|
+
schemaVersion: "openpond.modelImprovementQualificationReceipt.v1";
|
|
441
|
+
id: string;
|
|
442
|
+
review: {
|
|
443
|
+
id: string;
|
|
444
|
+
contentHash: string;
|
|
445
|
+
};
|
|
446
|
+
harnessRelease: {
|
|
447
|
+
id: string;
|
|
448
|
+
contentHash: string;
|
|
449
|
+
};
|
|
450
|
+
tasksetRelease: {
|
|
451
|
+
id: string;
|
|
452
|
+
contentHash: string;
|
|
453
|
+
} | null;
|
|
454
|
+
baselineEvaluation: {
|
|
455
|
+
id: string;
|
|
456
|
+
contentHash: string;
|
|
457
|
+
} | null;
|
|
458
|
+
model: {
|
|
459
|
+
provider: string;
|
|
460
|
+
model: string;
|
|
461
|
+
revision: string | null;
|
|
462
|
+
artifactHash: string | null;
|
|
463
|
+
tokenizerRevision: string | null;
|
|
464
|
+
chatTemplateHash: string | null;
|
|
465
|
+
};
|
|
466
|
+
environmentHash: string | null;
|
|
467
|
+
toolContractHash: string | null;
|
|
468
|
+
permissionContractHash: string | null;
|
|
469
|
+
policyHash: string | null;
|
|
470
|
+
verifierRef: {
|
|
471
|
+
id: string;
|
|
472
|
+
contentHash: string;
|
|
473
|
+
} | null;
|
|
474
|
+
sourcePolicies: {
|
|
475
|
+
policy: {
|
|
476
|
+
id: string;
|
|
477
|
+
contentHash: string;
|
|
478
|
+
};
|
|
479
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
480
|
+
checkedAt: string;
|
|
481
|
+
}[];
|
|
482
|
+
trainingEvidenceRefs: {
|
|
483
|
+
id: string;
|
|
484
|
+
contentHash: string;
|
|
485
|
+
}[];
|
|
486
|
+
frozenEvaluationEvidenceRefs: {
|
|
487
|
+
id: string;
|
|
488
|
+
contentHash: string;
|
|
489
|
+
}[];
|
|
490
|
+
privacyApproval: {
|
|
491
|
+
id: string;
|
|
492
|
+
contentHash: string;
|
|
493
|
+
} | null;
|
|
494
|
+
budgetApproval: {
|
|
495
|
+
id: string;
|
|
496
|
+
contentHash: string;
|
|
497
|
+
} | null;
|
|
498
|
+
maximumCostUsd: number;
|
|
499
|
+
signal: {
|
|
500
|
+
kind: "none" | "demonstrations" | "chosen_rejected" | "scalar_reward";
|
|
501
|
+
strength: "absent" | "weak" | "usable";
|
|
502
|
+
calibrated: boolean;
|
|
503
|
+
confounded: boolean;
|
|
504
|
+
variance: number | null;
|
|
505
|
+
evidenceRefs: {
|
|
506
|
+
id: string;
|
|
507
|
+
contentHash: string;
|
|
508
|
+
}[];
|
|
509
|
+
};
|
|
510
|
+
decision: "no_training" | "sft" | "preference" | "rl";
|
|
511
|
+
reasons: string[];
|
|
512
|
+
createdAt: string;
|
|
513
|
+
metadata: Record<string, unknown>;
|
|
514
|
+
contentHash: string;
|
|
515
|
+
};
|
|
516
|
+
modelImprovement: {
|
|
517
|
+
schemaVersion: "openpond.harnessEvaluationReviewReceipt.v1";
|
|
518
|
+
id: string;
|
|
519
|
+
ownerScope: {
|
|
520
|
+
kind: "personal" | "team";
|
|
521
|
+
id: string;
|
|
522
|
+
};
|
|
523
|
+
workspaceRef: string;
|
|
524
|
+
harnessRelease: {
|
|
525
|
+
id: string;
|
|
526
|
+
contentHash: string;
|
|
527
|
+
};
|
|
528
|
+
previousWatermark: {
|
|
529
|
+
cursor: string;
|
|
530
|
+
throughCreatedAt: string;
|
|
531
|
+
} | null;
|
|
532
|
+
nextWatermark: {
|
|
533
|
+
cursor: string;
|
|
534
|
+
throughCreatedAt: string;
|
|
535
|
+
};
|
|
536
|
+
selectedEvidence: {
|
|
537
|
+
evidence: {
|
|
538
|
+
id: string;
|
|
539
|
+
contentHash: string;
|
|
540
|
+
};
|
|
541
|
+
kind: "validation" | "observation" | "taskset" | "trigger" | "route_decision" | "refiner_outcome" | "proposal" | "apply_receipt" | "harness_advance" | "rollback" | "work_outcome" | "evaluation" | "training_qualification" | "model_candidate";
|
|
542
|
+
sourceRef: string;
|
|
543
|
+
sourcePolicy: {
|
|
544
|
+
policy: {
|
|
545
|
+
id: string;
|
|
546
|
+
contentHash: string;
|
|
547
|
+
};
|
|
548
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
549
|
+
checkedAt: string;
|
|
550
|
+
};
|
|
551
|
+
occurrenceKey: string;
|
|
552
|
+
occurredAt: string;
|
|
553
|
+
}[];
|
|
554
|
+
excludedEvidence: {
|
|
555
|
+
evidence: {
|
|
556
|
+
id: string;
|
|
557
|
+
contentHash: string;
|
|
558
|
+
};
|
|
559
|
+
sourcePolicy: {
|
|
560
|
+
policy: {
|
|
561
|
+
id: string;
|
|
562
|
+
contentHash: string;
|
|
563
|
+
};
|
|
564
|
+
state: "revoked" | "deleted" | "expired" | "authorized";
|
|
565
|
+
checkedAt: string;
|
|
566
|
+
} | null;
|
|
567
|
+
reason: "revoked" | "deleted" | "expired" | "outside_scope" | "before_watermark" | "duplicate" | "resolved" | "sensitive" | "unverified" | "budget";
|
|
568
|
+
}[];
|
|
569
|
+
claim: {
|
|
570
|
+
fingerprint: string;
|
|
571
|
+
recurrenceFamily: string;
|
|
572
|
+
statement: string;
|
|
573
|
+
independentOccurrences: number;
|
|
574
|
+
unresolvedOccurrences: number;
|
|
575
|
+
} | null;
|
|
576
|
+
classification: "runtime" | "taskset" | "no_action" | "harness_maintenance" | "product" | "model_improvement";
|
|
577
|
+
triage: {
|
|
578
|
+
layer: "model" | "tools" | "runtime" | "evaluation" | "product" | "harness" | "retrieval";
|
|
579
|
+
status: "resolved" | "not_applicable" | "unresolved" | "blocked";
|
|
580
|
+
reason: string;
|
|
581
|
+
evidenceRefs: {
|
|
582
|
+
id: string;
|
|
583
|
+
contentHash: string;
|
|
584
|
+
}[];
|
|
585
|
+
}[];
|
|
586
|
+
reason: string;
|
|
587
|
+
nextAuthority: "none" | "runtime_service" | "product_team" | "human_review" | "evaluation_system" | "training_system";
|
|
588
|
+
maxEstimatedCostUsd: number;
|
|
589
|
+
tasksetProposal: {
|
|
590
|
+
id: string;
|
|
591
|
+
contentHash: string;
|
|
592
|
+
} | null;
|
|
593
|
+
evaluation: {
|
|
594
|
+
id: string;
|
|
595
|
+
contentHash: string;
|
|
596
|
+
} | null;
|
|
597
|
+
trainingQualification: {
|
|
598
|
+
id: string;
|
|
599
|
+
contentHash: string;
|
|
600
|
+
} | null;
|
|
601
|
+
policyVersion: string;
|
|
602
|
+
createdAt: string;
|
|
603
|
+
metadata: Record<string, unknown>;
|
|
604
|
+
contentHash: string;
|
|
605
|
+
};
|
|
606
|
+
};
|
|
607
|
+
//# sourceMappingURL=review-conformance.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"review-conformance.d.ts","sourceRoot":"","sources":["../../../../src/review-conformance.ts"],"names":[],"mappings":"AAoMA,eAAO,MAAM,kCAAkC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAQ9C,CAAC"}
|
package/dist/types/runs.d.ts
CHANGED
|
@@ -1,12 +1,4 @@
|
|
|
1
1
|
import { z } from "zod";
|
|
2
|
-
export declare const ModelRefSchema: z.ZodObject<{
|
|
3
|
-
provider: z.ZodString;
|
|
4
|
-
model: z.ZodString;
|
|
5
|
-
revision: z.ZodDefault<z.ZodNullable<z.ZodString>>;
|
|
6
|
-
artifactHash: z.ZodDefault<z.ZodNullable<z.ZodString>>;
|
|
7
|
-
tokenizerRevision: z.ZodDefault<z.ZodNullable<z.ZodString>>;
|
|
8
|
-
chatTemplateHash: z.ZodDefault<z.ZodNullable<z.ZodString>>;
|
|
9
|
-
}, z.core.$strict>;
|
|
10
2
|
export declare const RuntimeTargetBindingSchema: z.ZodObject<{
|
|
11
3
|
adapterId: z.ZodString;
|
|
12
4
|
placement: z.ZodEnum<{
|
|
@@ -273,6 +265,49 @@ export declare const EvaluationResultSchema: z.ZodObject<{
|
|
|
273
265
|
metadata: z.ZodDefault<z.ZodRecord<z.ZodString, z.ZodUnknown>>;
|
|
274
266
|
contentHash: z.ZodString;
|
|
275
267
|
}, z.core.$strict>;
|
|
268
|
+
export declare const HarnessCompatibilityReceiptContentSchema: z.ZodObject<{
|
|
269
|
+
schemaVersion: z.ZodLiteral<"openpond.harnessCompatibility.v1">;
|
|
270
|
+
id: z.ZodString;
|
|
271
|
+
baseHarnessRelease: z.ZodObject<{
|
|
272
|
+
id: z.ZodString;
|
|
273
|
+
contentHash: z.ZodString;
|
|
274
|
+
}, z.core.$strict>;
|
|
275
|
+
candidateHarnessRelease: z.ZodObject<{
|
|
276
|
+
id: z.ZodString;
|
|
277
|
+
contentHash: z.ZodString;
|
|
278
|
+
}, z.core.$strict>;
|
|
279
|
+
tasksetRelease: z.ZodObject<{
|
|
280
|
+
id: z.ZodString;
|
|
281
|
+
contentHash: z.ZodString;
|
|
282
|
+
}, z.core.$strict>;
|
|
283
|
+
environmentHash: z.ZodString;
|
|
284
|
+
toolContractHash: z.ZodString;
|
|
285
|
+
policyHash: z.ZodString;
|
|
286
|
+
graderInterfaceHash: z.ZodString;
|
|
287
|
+
metadata: z.ZodDefault<z.ZodRecord<z.ZodString, z.ZodUnknown>>;
|
|
288
|
+
}, z.core.$strict>;
|
|
289
|
+
export declare const HarnessCompatibilityReceiptSchema: z.ZodObject<{
|
|
290
|
+
schemaVersion: z.ZodLiteral<"openpond.harnessCompatibility.v1">;
|
|
291
|
+
id: z.ZodString;
|
|
292
|
+
baseHarnessRelease: z.ZodObject<{
|
|
293
|
+
id: z.ZodString;
|
|
294
|
+
contentHash: z.ZodString;
|
|
295
|
+
}, z.core.$strict>;
|
|
296
|
+
candidateHarnessRelease: z.ZodObject<{
|
|
297
|
+
id: z.ZodString;
|
|
298
|
+
contentHash: z.ZodString;
|
|
299
|
+
}, z.core.$strict>;
|
|
300
|
+
tasksetRelease: z.ZodObject<{
|
|
301
|
+
id: z.ZodString;
|
|
302
|
+
contentHash: z.ZodString;
|
|
303
|
+
}, z.core.$strict>;
|
|
304
|
+
environmentHash: z.ZodString;
|
|
305
|
+
toolContractHash: z.ZodString;
|
|
306
|
+
policyHash: z.ZodString;
|
|
307
|
+
graderInterfaceHash: z.ZodString;
|
|
308
|
+
metadata: z.ZodDefault<z.ZodRecord<z.ZodString, z.ZodUnknown>>;
|
|
309
|
+
contentHash: z.ZodString;
|
|
310
|
+
}, z.core.$strict>;
|
|
276
311
|
export declare function createRunManifest(input: z.input<typeof RunManifestContentSchema>): RunManifest;
|
|
277
312
|
export declare function createAttemptReceipt(input: z.input<typeof AttemptReceiptContentSchema>): AttemptReceipt;
|
|
278
313
|
export declare function verifyAttemptReceipt(receipt: AttemptReceipt): boolean;
|
|
@@ -282,12 +317,14 @@ export declare function aggregateEvaluationReceipts(input: {
|
|
|
282
317
|
receipts: AttemptReceipt[];
|
|
283
318
|
metadata?: Record<string, unknown>;
|
|
284
319
|
}): EvaluationResult;
|
|
285
|
-
export declare function
|
|
320
|
+
export declare function createHarnessCompatibilityReceipt(input: z.input<typeof HarnessCompatibilityReceiptContentSchema>): z.infer<typeof HarnessCompatibilityReceiptSchema>;
|
|
321
|
+
export declare function assertComparableRunManifests(base: RunManifest, candidate: RunManifest, compatibility?: z.input<typeof HarnessCompatibilityReceiptSchema>): void;
|
|
286
322
|
export declare function rewardEligibleReceipts(receipts: AttemptReceipt[]): AttemptReceipt[];
|
|
287
|
-
export type ModelRef
|
|
323
|
+
export type { ModelRef } from "@openpond/harness";
|
|
288
324
|
export type RuntimeTargetBinding = z.infer<typeof RuntimeTargetBindingSchema>;
|
|
289
325
|
export type RunLimits = z.infer<typeof RunLimitsSchema>;
|
|
290
326
|
export type RunManifest = z.infer<typeof RunManifestSchema>;
|
|
291
327
|
export type AttemptReceipt = z.infer<typeof AttemptReceiptSchema>;
|
|
292
328
|
export type EvaluationResult = z.infer<typeof EvaluationResultSchema>;
|
|
329
|
+
export type HarnessCompatibilityReceipt = z.infer<typeof HarnessCompatibilityReceiptSchema>;
|
|
293
330
|
//# sourceMappingURL=runs.d.ts.map
|