@openpond/evals 0.10.9 → 0.10.10
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/LEARNING.md +15 -0
- package/dist/graders.js +19 -4
- package/dist/learning/budgeted-reward-fixtures.js +34 -0
- package/dist/learning/contracts.js +3 -0
- package/dist/learning/grade-judge-budget.js +26 -0
- package/dist/learning/index.js +6 -0
- package/dist/learning/intake-history.js +2 -1
- package/dist/learning/intake-text.js +14 -0
- package/dist/learning/intake.js +5 -2
- package/dist/learning/judge-budget.js +59 -0
- package/dist/learning/model-judge.js +49 -0
- package/dist/learning/operations.js +3 -1
- package/dist/learning/reward-authoring.js +4 -0
- package/dist/learning/reward-calibration.js +66 -0
- package/dist/learning/reward-check-budget.js +27 -0
- package/dist/learning/reward-check-service.js +2 -2
- package/dist/learning/reward-check-worker.js +10 -2
- package/dist/learning/reward-checks.js +4 -1
- package/dist/learning/reward-fixture-executor.js +3 -2
- package/dist/learning/service.js +30 -5
- package/dist/rewards.js +15 -3
- package/dist/rollouts.js +2 -2
- package/dist/types/graders.d.ts +51 -0
- package/dist/types/graders.d.ts.map +1 -1
- package/dist/types/learning/admission.d.ts +5 -0
- package/dist/types/learning/admission.d.ts.map +1 -1
- package/dist/types/learning/budgeted-reward-fixtures.d.ts +27 -0
- package/dist/types/learning/budgeted-reward-fixtures.d.ts.map +1 -0
- package/dist/types/learning/contracts.d.ts +301 -0
- package/dist/types/learning/contracts.d.ts.map +1 -1
- package/dist/types/learning/grade-judge-budget.d.ts +10 -0
- package/dist/types/learning/grade-judge-budget.d.ts.map +1 -0
- package/dist/types/learning/index.d.ts +6 -0
- package/dist/types/learning/index.d.ts.map +1 -1
- package/dist/types/learning/intake-history.d.ts.map +1 -1
- package/dist/types/learning/intake-text.d.ts +6 -0
- package/dist/types/learning/intake-text.d.ts.map +1 -0
- package/dist/types/learning/intake.d.ts +1 -0
- package/dist/types/learning/intake.d.ts.map +1 -1
- package/dist/types/learning/iteration-dispatch-contracts.d.ts +1 -1
- package/dist/types/learning/iteration-reservation-contracts.d.ts +1 -1
- package/dist/types/learning/judge-budget.d.ts +43 -0
- package/dist/types/learning/judge-budget.d.ts.map +1 -0
- package/dist/types/learning/model-judge.d.ts +29 -0
- package/dist/types/learning/model-judge.d.ts.map +1 -0
- package/dist/types/learning/operations.d.ts +41 -0
- package/dist/types/learning/operations.d.ts.map +1 -1
- package/dist/types/learning/repository.d.ts +177 -4
- package/dist/types/learning/repository.d.ts.map +1 -1
- package/dist/types/learning/reward-authoring.d.ts.map +1 -1
- package/dist/types/learning/reward-calibration.d.ts +129 -0
- package/dist/types/learning/reward-calibration.d.ts.map +1 -0
- package/dist/types/learning/reward-check-budget.d.ts +12 -0
- package/dist/types/learning/reward-check-budget.d.ts.map +1 -0
- package/dist/types/learning/reward-check-service.d.ts +1 -1
- package/dist/types/learning/reward-check-service.d.ts.map +1 -1
- package/dist/types/learning/reward-check-worker.d.ts.map +1 -1
- package/dist/types/learning/reward-checks.d.ts +84 -0
- package/dist/types/learning/reward-checks.d.ts.map +1 -1
- package/dist/types/learning/reward-fixture-executor.d.ts +3 -0
- package/dist/types/learning/reward-fixture-executor.d.ts.map +1 -1
- package/dist/types/learning/schedule-contracts.d.ts +1 -1
- package/dist/types/learning/service.d.ts.map +1 -1
- package/dist/types/learning/transport.d.ts +198 -5
- package/dist/types/learning/transport.d.ts.map +1 -1
- package/dist/types/rewards.d.ts +107 -0
- package/dist/types/rewards.d.ts.map +1 -1
- package/dist/types/rollouts.d.ts.map +1 -1
- package/package.json +1 -1
- package/schemas/learning/v1/command-request.schema.json +225 -169
- package/schemas/learning/v1/decision.schema.json +192 -0
- package/schemas/learning/v1/grade.schema.json +335 -3
- package/schemas/learning/v1/reward.schema.json +22 -0
- package/schemas/learning/v1/reward_check.schema.json +335 -3
|
@@ -708,6 +708,38 @@ export declare const TaskAdmissionDecisionContentSchema: z.ZodObject<{
|
|
|
708
708
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
709
709
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
710
710
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
711
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
712
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
713
|
+
graderId: z.ZodString;
|
|
714
|
+
graderVersion: z.ZodString;
|
|
715
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
716
|
+
passed: z.ZodBoolean;
|
|
717
|
+
rewardEligible: z.ZodBoolean;
|
|
718
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
719
|
+
policy_failure: "policy_failure";
|
|
720
|
+
grader_failure: "grader_failure";
|
|
721
|
+
environment_failure: "environment_failure";
|
|
722
|
+
infrastructure_failure: "infrastructure_failure";
|
|
723
|
+
timeout: "timeout";
|
|
724
|
+
cancelled: "cancelled";
|
|
725
|
+
}>>;
|
|
726
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
727
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
728
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
729
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
730
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
731
|
+
providerId: z.ZodString;
|
|
732
|
+
modelId: z.ZodString;
|
|
733
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
734
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
735
|
+
requestHash: z.ZodString;
|
|
736
|
+
responseHash: z.ZodString;
|
|
737
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
738
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
739
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
740
|
+
}, z.core.$strict>>;
|
|
741
|
+
contentHash: z.ZodString;
|
|
742
|
+
}, z.core.$strict>>;
|
|
711
743
|
message: z.ZodNullable<z.ZodString>;
|
|
712
744
|
}, z.core.$strict>>;
|
|
713
745
|
training: z.ZodObject<{
|
|
@@ -760,6 +792,38 @@ export declare const TaskAdmissionDecisionContentSchema: z.ZodObject<{
|
|
|
760
792
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
761
793
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
762
794
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
795
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
796
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
797
|
+
graderId: z.ZodString;
|
|
798
|
+
graderVersion: z.ZodString;
|
|
799
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
800
|
+
passed: z.ZodBoolean;
|
|
801
|
+
rewardEligible: z.ZodBoolean;
|
|
802
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
803
|
+
policy_failure: "policy_failure";
|
|
804
|
+
grader_failure: "grader_failure";
|
|
805
|
+
environment_failure: "environment_failure";
|
|
806
|
+
infrastructure_failure: "infrastructure_failure";
|
|
807
|
+
timeout: "timeout";
|
|
808
|
+
cancelled: "cancelled";
|
|
809
|
+
}>>;
|
|
810
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
811
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
812
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
813
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
814
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
815
|
+
providerId: z.ZodString;
|
|
816
|
+
modelId: z.ZodString;
|
|
817
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
818
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
819
|
+
requestHash: z.ZodString;
|
|
820
|
+
responseHash: z.ZodString;
|
|
821
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
822
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
823
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
824
|
+
}, z.core.$strict>>;
|
|
825
|
+
contentHash: z.ZodString;
|
|
826
|
+
}, z.core.$strict>>;
|
|
763
827
|
message: z.ZodNullable<z.ZodString>;
|
|
764
828
|
}, z.core.$strict>>;
|
|
765
829
|
training: z.ZodObject<{
|
|
@@ -864,6 +928,38 @@ export declare const TaskAdmissionDecisionSchema: z.ZodObject<{
|
|
|
864
928
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
865
929
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
866
930
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
931
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
932
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
933
|
+
graderId: z.ZodString;
|
|
934
|
+
graderVersion: z.ZodString;
|
|
935
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
936
|
+
passed: z.ZodBoolean;
|
|
937
|
+
rewardEligible: z.ZodBoolean;
|
|
938
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
939
|
+
policy_failure: "policy_failure";
|
|
940
|
+
grader_failure: "grader_failure";
|
|
941
|
+
environment_failure: "environment_failure";
|
|
942
|
+
infrastructure_failure: "infrastructure_failure";
|
|
943
|
+
timeout: "timeout";
|
|
944
|
+
cancelled: "cancelled";
|
|
945
|
+
}>>;
|
|
946
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
947
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
948
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
949
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
950
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
951
|
+
providerId: z.ZodString;
|
|
952
|
+
modelId: z.ZodString;
|
|
953
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
954
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
955
|
+
requestHash: z.ZodString;
|
|
956
|
+
responseHash: z.ZodString;
|
|
957
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
958
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
959
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
960
|
+
}, z.core.$strict>>;
|
|
961
|
+
contentHash: z.ZodString;
|
|
962
|
+
}, z.core.$strict>>;
|
|
867
963
|
message: z.ZodNullable<z.ZodString>;
|
|
868
964
|
}, z.core.$strict>>;
|
|
869
965
|
training: z.ZodObject<{
|
|
@@ -916,6 +1012,38 @@ export declare const TaskAdmissionDecisionSchema: z.ZodObject<{
|
|
|
916
1012
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
917
1013
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
918
1014
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
1015
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
1016
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
1017
|
+
graderId: z.ZodString;
|
|
1018
|
+
graderVersion: z.ZodString;
|
|
1019
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
1020
|
+
passed: z.ZodBoolean;
|
|
1021
|
+
rewardEligible: z.ZodBoolean;
|
|
1022
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
1023
|
+
policy_failure: "policy_failure";
|
|
1024
|
+
grader_failure: "grader_failure";
|
|
1025
|
+
environment_failure: "environment_failure";
|
|
1026
|
+
infrastructure_failure: "infrastructure_failure";
|
|
1027
|
+
timeout: "timeout";
|
|
1028
|
+
cancelled: "cancelled";
|
|
1029
|
+
}>>;
|
|
1030
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
1031
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
1032
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
1033
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
1034
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
1035
|
+
providerId: z.ZodString;
|
|
1036
|
+
modelId: z.ZodString;
|
|
1037
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
1038
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
1039
|
+
requestHash: z.ZodString;
|
|
1040
|
+
responseHash: z.ZodString;
|
|
1041
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
1042
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
1043
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
1044
|
+
}, z.core.$strict>>;
|
|
1045
|
+
contentHash: z.ZodString;
|
|
1046
|
+
}, z.core.$strict>>;
|
|
919
1047
|
message: z.ZodNullable<z.ZodString>;
|
|
920
1048
|
}, z.core.$strict>>;
|
|
921
1049
|
training: z.ZodObject<{
|
|
@@ -1407,6 +1535,38 @@ export declare const TaskGradeRunSchema: z.ZodObject<{
|
|
|
1407
1535
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
1408
1536
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
1409
1537
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
1538
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
1539
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
1540
|
+
graderId: z.ZodString;
|
|
1541
|
+
graderVersion: z.ZodString;
|
|
1542
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
1543
|
+
passed: z.ZodBoolean;
|
|
1544
|
+
rewardEligible: z.ZodBoolean;
|
|
1545
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
1546
|
+
policy_failure: "policy_failure";
|
|
1547
|
+
grader_failure: "grader_failure";
|
|
1548
|
+
environment_failure: "environment_failure";
|
|
1549
|
+
infrastructure_failure: "infrastructure_failure";
|
|
1550
|
+
timeout: "timeout";
|
|
1551
|
+
cancelled: "cancelled";
|
|
1552
|
+
}>>;
|
|
1553
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
1554
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
1555
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
1556
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
1557
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
1558
|
+
providerId: z.ZodString;
|
|
1559
|
+
modelId: z.ZodString;
|
|
1560
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
1561
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
1562
|
+
requestHash: z.ZodString;
|
|
1563
|
+
responseHash: z.ZodString;
|
|
1564
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
1565
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
1566
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
1567
|
+
}, z.core.$strict>>;
|
|
1568
|
+
contentHash: z.ZodString;
|
|
1569
|
+
}, z.core.$strict>>;
|
|
1410
1570
|
message: z.ZodNullable<z.ZodString>;
|
|
1411
1571
|
}, z.core.$strict>>;
|
|
1412
1572
|
training: z.ZodObject<{
|
|
@@ -1434,6 +1594,26 @@ export declare const TaskGradeRunSchema: z.ZodObject<{
|
|
|
1434
1594
|
attemptCount: z.ZodNumber;
|
|
1435
1595
|
timeoutMs: z.ZodNumber;
|
|
1436
1596
|
maximumSpendUsd: z.ZodNumber;
|
|
1597
|
+
judgeCalls: z.ZodOptional<z.ZodArray<z.ZodObject<{
|
|
1598
|
+
id: z.ZodString;
|
|
1599
|
+
requestHash: z.ZodString;
|
|
1600
|
+
reservedUsd: z.ZodNumber;
|
|
1601
|
+
status: z.ZodEnum<{
|
|
1602
|
+
reserved: "reserved";
|
|
1603
|
+
settled: "settled";
|
|
1604
|
+
not_dispatched: "not_dispatched";
|
|
1605
|
+
}>;
|
|
1606
|
+
response: z.ZodNullable<z.ZodObject<{
|
|
1607
|
+
text: z.ZodString;
|
|
1608
|
+
modelId: z.ZodString;
|
|
1609
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
1610
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
1611
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
1612
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
1613
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
1614
|
+
}, z.core.$strict>>;
|
|
1615
|
+
}, z.core.$strict>>>;
|
|
1616
|
+
requestedBy: z.ZodOptional<z.ZodString>;
|
|
1437
1617
|
failure: z.ZodNullable<z.ZodString>;
|
|
1438
1618
|
createdAt: z.ZodString;
|
|
1439
1619
|
updatedAt: z.ZodString;
|
|
@@ -1666,6 +1846,11 @@ export declare const LearningResourceSchema: z.ZodUnion<readonly [z.ZodObject<{
|
|
|
1666
1846
|
host_private: "host_private";
|
|
1667
1847
|
}>;
|
|
1668
1848
|
}, z.core.$strict>>;
|
|
1849
|
+
calibrationCheckRef: z.ZodOptional<z.ZodObject<{
|
|
1850
|
+
id: z.ZodString;
|
|
1851
|
+
contentHash: z.ZodString;
|
|
1852
|
+
revision: z.ZodNumber;
|
|
1853
|
+
}, z.core.$strict>>;
|
|
1669
1854
|
contentHash: z.ZodString;
|
|
1670
1855
|
}, z.core.$strict>, z.ZodObject<{
|
|
1671
1856
|
schemaVersion: z.ZodLiteral<"openpond.rewardBinding.v1">;
|
|
@@ -1964,6 +2149,38 @@ export declare const LearningResourceSchema: z.ZodUnion<readonly [z.ZodObject<{
|
|
|
1964
2149
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
1965
2150
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
1966
2151
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
2152
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
2153
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
2154
|
+
graderId: z.ZodString;
|
|
2155
|
+
graderVersion: z.ZodString;
|
|
2156
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
2157
|
+
passed: z.ZodBoolean;
|
|
2158
|
+
rewardEligible: z.ZodBoolean;
|
|
2159
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
2160
|
+
policy_failure: "policy_failure";
|
|
2161
|
+
grader_failure: "grader_failure";
|
|
2162
|
+
environment_failure: "environment_failure";
|
|
2163
|
+
infrastructure_failure: "infrastructure_failure";
|
|
2164
|
+
timeout: "timeout";
|
|
2165
|
+
cancelled: "cancelled";
|
|
2166
|
+
}>>;
|
|
2167
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
2168
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
2169
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
2170
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
2171
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
2172
|
+
providerId: z.ZodString;
|
|
2173
|
+
modelId: z.ZodString;
|
|
2174
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
2175
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
2176
|
+
requestHash: z.ZodString;
|
|
2177
|
+
responseHash: z.ZodString;
|
|
2178
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2179
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2180
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
2181
|
+
}, z.core.$strict>>;
|
|
2182
|
+
contentHash: z.ZodString;
|
|
2183
|
+
}, z.core.$strict>>;
|
|
1967
2184
|
message: z.ZodNullable<z.ZodString>;
|
|
1968
2185
|
}, z.core.$strict>>;
|
|
1969
2186
|
training: z.ZodObject<{
|
|
@@ -2016,6 +2233,38 @@ export declare const LearningResourceSchema: z.ZodUnion<readonly [z.ZodObject<{
|
|
|
2016
2233
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
2017
2234
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
2018
2235
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
2236
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
2237
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
2238
|
+
graderId: z.ZodString;
|
|
2239
|
+
graderVersion: z.ZodString;
|
|
2240
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
2241
|
+
passed: z.ZodBoolean;
|
|
2242
|
+
rewardEligible: z.ZodBoolean;
|
|
2243
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
2244
|
+
policy_failure: "policy_failure";
|
|
2245
|
+
grader_failure: "grader_failure";
|
|
2246
|
+
environment_failure: "environment_failure";
|
|
2247
|
+
infrastructure_failure: "infrastructure_failure";
|
|
2248
|
+
timeout: "timeout";
|
|
2249
|
+
cancelled: "cancelled";
|
|
2250
|
+
}>>;
|
|
2251
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
2252
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
2253
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
2254
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
2255
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
2256
|
+
providerId: z.ZodString;
|
|
2257
|
+
modelId: z.ZodString;
|
|
2258
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
2259
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
2260
|
+
requestHash: z.ZodString;
|
|
2261
|
+
responseHash: z.ZodString;
|
|
2262
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2263
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2264
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
2265
|
+
}, z.core.$strict>>;
|
|
2266
|
+
contentHash: z.ZodString;
|
|
2267
|
+
}, z.core.$strict>>;
|
|
2019
2268
|
message: z.ZodNullable<z.ZodString>;
|
|
2020
2269
|
}, z.core.$strict>>;
|
|
2021
2270
|
training: z.ZodObject<{
|
|
@@ -2328,6 +2577,38 @@ export declare const LearningResourceSchema: z.ZodUnion<readonly [z.ZodObject<{
|
|
|
2328
2577
|
normalizedScore: z.ZodNullable<z.ZodNumber>;
|
|
2329
2578
|
passed: z.ZodNullable<z.ZodBoolean>;
|
|
2330
2579
|
evidenceHashes: z.ZodArray<z.ZodString>;
|
|
2580
|
+
graderEvidence: z.ZodOptional<z.ZodObject<{
|
|
2581
|
+
schemaVersion: z.ZodLiteral<"openpond.graderEvidence.v1">;
|
|
2582
|
+
graderId: z.ZodString;
|
|
2583
|
+
graderVersion: z.ZodString;
|
|
2584
|
+
score: z.ZodNullable<z.ZodNumber>;
|
|
2585
|
+
passed: z.ZodBoolean;
|
|
2586
|
+
rewardEligible: z.ZodBoolean;
|
|
2587
|
+
failureClass: z.ZodNullable<z.ZodEnum<{
|
|
2588
|
+
policy_failure: "policy_failure";
|
|
2589
|
+
grader_failure: "grader_failure";
|
|
2590
|
+
environment_failure: "environment_failure";
|
|
2591
|
+
infrastructure_failure: "infrastructure_failure";
|
|
2592
|
+
timeout: "timeout";
|
|
2593
|
+
cancelled: "cancelled";
|
|
2594
|
+
}>>;
|
|
2595
|
+
feedback: z.ZodArray<z.ZodString>;
|
|
2596
|
+
visibleEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
2597
|
+
privilegedEvidenceRefs: z.ZodArray<z.ZodString>;
|
|
2598
|
+
modelJudgeReceipt: z.ZodOptional<z.ZodObject<{
|
|
2599
|
+
schemaVersion: z.ZodLiteral<"openpond.modelJudgeReceipt.v1">;
|
|
2600
|
+
providerId: z.ZodString;
|
|
2601
|
+
modelId: z.ZodString;
|
|
2602
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
2603
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
2604
|
+
requestHash: z.ZodString;
|
|
2605
|
+
responseHash: z.ZodString;
|
|
2606
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2607
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2608
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
2609
|
+
}, z.core.$strict>>;
|
|
2610
|
+
contentHash: z.ZodString;
|
|
2611
|
+
}, z.core.$strict>>;
|
|
2331
2612
|
message: z.ZodNullable<z.ZodString>;
|
|
2332
2613
|
}, z.core.$strict>>;
|
|
2333
2614
|
training: z.ZodObject<{
|
|
@@ -2355,6 +2636,26 @@ export declare const LearningResourceSchema: z.ZodUnion<readonly [z.ZodObject<{
|
|
|
2355
2636
|
attemptCount: z.ZodNumber;
|
|
2356
2637
|
timeoutMs: z.ZodNumber;
|
|
2357
2638
|
maximumSpendUsd: z.ZodNumber;
|
|
2639
|
+
judgeCalls: z.ZodOptional<z.ZodArray<z.ZodObject<{
|
|
2640
|
+
id: z.ZodString;
|
|
2641
|
+
requestHash: z.ZodString;
|
|
2642
|
+
reservedUsd: z.ZodNumber;
|
|
2643
|
+
status: z.ZodEnum<{
|
|
2644
|
+
reserved: "reserved";
|
|
2645
|
+
settled: "settled";
|
|
2646
|
+
not_dispatched: "not_dispatched";
|
|
2647
|
+
}>;
|
|
2648
|
+
response: z.ZodNullable<z.ZodObject<{
|
|
2649
|
+
text: z.ZodString;
|
|
2650
|
+
modelId: z.ZodString;
|
|
2651
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
2652
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
2653
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2654
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
2655
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
2656
|
+
}, z.core.$strict>>;
|
|
2657
|
+
}, z.core.$strict>>>;
|
|
2658
|
+
requestedBy: z.ZodOptional<z.ZodString>;
|
|
2358
2659
|
failure: z.ZodNullable<z.ZodString>;
|
|
2359
2660
|
createdAt: z.ZodString;
|
|
2360
2661
|
updatedAt: z.ZodString;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"contracts.d.ts","sourceRoot":"","sources":["../../../../../src/learning/contracts.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;
|
|
1
|
+
{"version":3,"file":"contracts.d.ts","sourceRoot":"","sources":["../../../../../src/learning/contracts.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAQxB,eAAO,MAAM,wBAAwB,2CAAiC,CAAC;AACvE,eAAO,MAAM,yBAAyB;;;;kBAAyB,CAAC;AAChE,eAAO,MAAM,yBAAyB,aAAiJ,CAAC;AAExL,eAAO,MAAM,2BAA2B;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAwBtC,CAAC;AACH,eAAO,MAAM,oBAAoB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAAsF,CAAC;AAExH,eAAO,MAAM,uBAAuB;;;;;;;;;;;;;;;;kBAWzB,CAAC;AACZ;0EAC0E;AAC1E,eAAO,MAAM,0BAA0B;;;;;;;;;;;;;;;;;;kBAM5B,CAAC;AACZ,eAAO,MAAM,2BAA2B;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAgBtC,CAAC;AACH,eAAO,MAAM,oBAAoB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAAsF,CAAC;AAExH,eAAO,MAAM,2BAA2B;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAgB7B,CAAC;AAEZ,eAAO,MAAM,yBAAyB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAS3B,CAAC;AACZ,eAAO,MAAM,kBAAkB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAAgF,CAAC;AAEhH,+EAA+E;AAC/E,eAAO,MAAM,gBAAgB;;;;;;;;;;kBAU3B,CAAC;AACH,MAAM,MAAM,UAAU,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,gBAAgB,CAAC,CAAC;AAE1D,eAAO,MAAM,4BAA4B;;;;;;;;;;;;;;;;;kBAiBvC,CAAC;AACH,eAAO,MAAM,kBAAkB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAUpB,CAAC;AAEZ,eAAO,MAAM,kCAAkC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAqB7C,CAAC;AACH,eAAO,MAAM,2BAA2B;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAA6F,CAAC;AAEtI,eAAO,MAAM,sBAAsB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAoBjC,CAAC;AACH,eAAO,MAAM,eAAe;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAAiF,CAAC;AAE9G,eAAO,MAAM,2BAA2B;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBA8BtC,CAAC;AACH,eAAO,MAAM,oBAAoB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAAsF,CAAC;AACxH,eAAO,MAAM,4BAA4B;;;;;;;;;;;;;kBAM9B,CAAC;AACZ,eAAO,MAAM,6BAA6B;;;;;;;;;;;;;;;EAAgO,CAAC;AAC3Q,eAAO,MAAM,uBAAuB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAwBzB,CAAC;AAEZ,eAAO,MAAM,kBAAkB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;kBAoBpB,CAAC;AAEZ,eAAO,MAAM,sBAAsB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;oBAA2P,CAAC;AAE/R,MAAM,MAAM,cAAc,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,oBAAoB,CAAC,CAAC;AAClE,MAAM,MAAM,cAAc,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,oBAAoB,CAAC,CAAC;AAClE,MAAM,MAAM,qBAAqB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,2BAA2B,CAAC,CAAC;AAChF,MAAM,MAAM,YAAY,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,kBAAkB,CAAC,CAAC;AAC9D,MAAM,MAAM,sBAAsB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,4BAA4B,CAAC,CAAC;AAClF,MAAM,MAAM,YAAY,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,kBAAkB,CAAC,CAAC;AAC9D,MAAM,MAAM,qBAAqB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,2BAA2B,CAAC,CAAC;AAChF,MAAM,MAAM,SAAS,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,eAAe,CAAC,CAAC;AACxD,MAAM,MAAM,cAAc,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,oBAAoB,CAAC,CAAC;AAClE,MAAM,MAAM,iBAAiB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,uBAAuB,CAAC,CAAC;AACxE,MAAM,MAAM,YAAY,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,kBAAkB,CAAC,CAAC;AAC9D,MAAM,MAAM,gBAAgB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,sBAAsB,CAAC,CAAC;AACtE,MAAM,MAAM,mBAAmB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,yBAAyB,CAAC,CAAC;AAE5E,wBAAgB,WAAW,CAAC,QAAQ,EAAE;IAAE,EAAE,EAAE,MAAM,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAC;IAAC,WAAW,EAAE,MAAM,CAAA;CAAE,GAAG,mBAAmB,CAEhH;AACD,wBAAgB,eAAe,CAAC,IAAI,EAAE,mBAAmB,EAAE,KAAK,EAAE,mBAAmB,GAAG,OAAO,CAE9F;AACD,wBAAgB,mBAAmB,CAAC,CAAC,SAAS,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,OAAO,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,WAAW,EAAE,MAAM,CAAA;CAAE,CAG9G;AACD,wBAAgB,yBAAyB,CAAC,QAAQ,EAAE;IAAE,WAAW,EAAE,MAAM,CAAA;CAAE,GAAG,IAAI,CAIjF"}
|
|
@@ -0,0 +1,10 @@
|
|
|
1
|
+
import type { JudgeBudgetStore } from "./judge-budget.js";
|
|
2
|
+
import { type LearningRepository } from "./repository.js";
|
|
3
|
+
import { type TaskGradeRun } from "./contracts.js";
|
|
4
|
+
export declare function createTaskGradeJudgeBudgetStore(input: {
|
|
5
|
+
repository: LearningRepository;
|
|
6
|
+
scope: string;
|
|
7
|
+
run: TaskGradeRun;
|
|
8
|
+
now?: () => string;
|
|
9
|
+
}): JudgeBudgetStore;
|
|
10
|
+
//# sourceMappingURL=grade-judge-budget.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"grade-judge-budget.d.ts","sourceRoot":"","sources":["../../../../../src/learning/grade-judge-budget.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AAC1D,OAAO,EAA2B,KAAK,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AACnF,OAAO,EAAsB,KAAK,YAAY,EAAE,MAAM,gBAAgB,CAAC;AAEvE,wBAAgB,+BAA+B,CAAC,KAAK,EAAE;IACrD,UAAU,EAAE,kBAAkB,CAAC;IAAC,KAAK,EAAE,MAAM,CAAC;IAAC,GAAG,EAAE,YAAY,CAAC;IAAC,GAAG,CAAC,EAAE,MAAM,MAAM,CAAC;CACtF,GAAG,gBAAgB,CAoBnB"}
|
|
@@ -12,7 +12,13 @@ export * from "./reward-authoring.js";
|
|
|
12
12
|
export * from "./rating-reward-fixture.js";
|
|
13
13
|
export * from "./intake-contracts.js";
|
|
14
14
|
export * from "./intake.js";
|
|
15
|
+
export * from "./model-judge.js";
|
|
16
|
+
export * from "./judge-budget.js";
|
|
17
|
+
export * from "./reward-check-budget.js";
|
|
18
|
+
export * from "./grade-judge-budget.js";
|
|
19
|
+
export * from "./budgeted-reward-fixtures.js";
|
|
15
20
|
export * from "./reward-checks.js";
|
|
21
|
+
export * from "./reward-calibration.js";
|
|
16
22
|
export * from "./reward-check-worker.js";
|
|
17
23
|
export * from "./reward-fixture-executor.js";
|
|
18
24
|
export * from "./iteration-reservation-contracts.js";
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../../src/learning/index.ts"],"names":[],"mappings":"AAAA,cAAc,gBAAgB,CAAC;AAC/B,cAAc,aAAa,CAAC;AAC5B,cAAc,gBAAgB,CAAC;AAC/B,cAAc,iBAAiB,CAAC;AAChC,cAAc,iBAAiB,CAAC;AAChC,cAAc,cAAc,CAAC;AAC7B,cAAc,mBAAmB,CAAC;AAClC,cAAc,gBAAgB,CAAC;AAC/B,cAAc,aAAa,CAAC;AAC5B,cAAc,gBAAgB,CAAC;AAC/B,cAAc,uBAAuB,CAAC;AACtC,cAAc,4BAA4B,CAAC;AAC3C,cAAc,uBAAuB,CAAC;AACtC,cAAc,aAAa,CAAC;AAC5B,cAAc,oBAAoB,CAAC;AACnC,cAAc,0BAA0B,CAAC;AACzC,cAAc,8BAA8B,CAAC;AAC7C,cAAc,sCAAsC,CAAC;AACrD,cAAc,mCAAmC,CAAC;AAClD,cAAc,gCAAgC,CAAC;AAC/C,cAAc,mCAAmC,CAAC;AAClD,OAAO,EAAE,6BAA6B,EAAE,MAAM,gCAAgC,CAAC;AAC/E,cAAc,yBAAyB,CAAC;AACxC,cAAc,uBAAuB,CAAC;AACtC,cAAc,sBAAsB,CAAC;AACrC,cAAc,wBAAwB,CAAC"}
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../../src/learning/index.ts"],"names":[],"mappings":"AAAA,cAAc,gBAAgB,CAAC;AAC/B,cAAc,aAAa,CAAC;AAC5B,cAAc,gBAAgB,CAAC;AAC/B,cAAc,iBAAiB,CAAC;AAChC,cAAc,iBAAiB,CAAC;AAChC,cAAc,cAAc,CAAC;AAC7B,cAAc,mBAAmB,CAAC;AAClC,cAAc,gBAAgB,CAAC;AAC/B,cAAc,aAAa,CAAC;AAC5B,cAAc,gBAAgB,CAAC;AAC/B,cAAc,uBAAuB,CAAC;AACtC,cAAc,4BAA4B,CAAC;AAC3C,cAAc,uBAAuB,CAAC;AACtC,cAAc,aAAa,CAAC;AAC5B,cAAc,kBAAkB,CAAC;AACjC,cAAc,mBAAmB,CAAC;AAClC,cAAc,0BAA0B,CAAC;AACzC,cAAc,yBAAyB,CAAC;AACxC,cAAc,+BAA+B,CAAC;AAC9C,cAAc,oBAAoB,CAAC;AACnC,cAAc,yBAAyB,CAAC;AACxC,cAAc,0BAA0B,CAAC;AACzC,cAAc,8BAA8B,CAAC;AAC7C,cAAc,sCAAsC,CAAC;AACrD,cAAc,mCAAmC,CAAC;AAClD,cAAc,gCAAgC,CAAC;AAC/C,cAAc,mCAAmC,CAAC;AAClD,OAAO,EAAE,6BAA6B,EAAE,MAAM,gCAAgC,CAAC;AAC/E,cAAc,yBAAyB,CAAC;AACxC,cAAc,uBAAuB,CAAC;AACtC,cAAc,sBAAsB,CAAC;AACrC,cAAc,wBAAwB,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"intake-history.d.ts","sourceRoot":"","sources":["../../../../../src/learning/intake-history.ts"],"names":[],"mappings":"AAIA,OAAO,EAAoE,KAAK,cAAc,EAAE,KAAK,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;
|
|
1
|
+
{"version":3,"file":"intake-history.d.ts","sourceRoot":"","sources":["../../../../../src/learning/intake-history.ts"],"names":[],"mappings":"AAIA,OAAO,EAAoE,KAAK,cAAc,EAAE,KAAK,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAQrJ,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,OAAO,GAAG,gBAAgB,EAAE,CAQrE;AAWD,4EAA4E;AAC5E,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,cAAc,EAAE,GAAG,gBAAgB,EAAE,CA0B/E"}
|
|
@@ -0,0 +1,6 @@
|
|
|
1
|
+
/** Preserve the original UTF-8 bytes, including a leading BOM. Binary or invalid
|
|
2
|
+
* UTF-8 input must not silently turn into replacement characters in evidence. */
|
|
3
|
+
export declare function decodeTaskIntakeText(bytes: Uint8Array): string;
|
|
4
|
+
/** A BOM is a transport marker only at the start of a structured text file. */
|
|
5
|
+
export declare function structuredIntakeText(text: string): string;
|
|
6
|
+
//# sourceMappingURL=intake-text.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"intake-text.d.ts","sourceRoot":"","sources":["../../../../../src/learning/intake-text.ts"],"names":[],"mappings":"AAAA;iFACiF;AACjF,wBAAgB,oBAAoB,CAAC,KAAK,EAAE,UAAU,GAAG,MAAM,CAG9D;AAED,+EAA+E;AAC/E,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEzD"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"intake.d.ts","sourceRoot":"","sources":["../../../../../src/learning/intake.ts"],"names":[],"mappings":"AAOA,OAAO,EACoB,KAAK,cAAc,EAAE,KAAK,gBAAgB,EAAE,KAAK,iBAAiB,EAAyB,MAAM,uBAAuB,CAAC;AAapJ,wBAAgB,iBAAiB,CAAC,KAAK,EAAE;IAAE,MAAM,EAAE,gBAAgB,CAAC;IAAC,KAAK,EAAE,cAAc,EAAE,CAAA;CAAE,GAAG,iBAAiB,
|
|
1
|
+
{"version":3,"file":"intake.d.ts","sourceRoot":"","sources":["../../../../../src/learning/intake.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,oBAAoB,EAAE,MAAM,kBAAkB,CAAC;AAExD,OAAO,EACoB,KAAK,cAAc,EAAE,KAAK,gBAAgB,EAAE,KAAK,iBAAiB,EAAyB,MAAM,uBAAuB,CAAC;AAapJ,wBAAgB,iBAAiB,CAAC,KAAK,EAAE;IAAE,MAAM,EAAE,gBAAgB,CAAC;IAAC,KAAK,EAAE,cAAc,EAAE,CAAA;CAAE,GAAG,iBAAiB,CAqDjH"}
|
|
@@ -24,10 +24,10 @@ export declare const LearningIterationDispatchSchema: z.ZodObject<{
|
|
|
24
24
|
}, z.core.$strict>;
|
|
25
25
|
state: z.ZodEnum<{
|
|
26
26
|
blocked: "blocked";
|
|
27
|
+
settled: "settled";
|
|
27
28
|
preparing: "preparing";
|
|
28
29
|
prepared: "prepared";
|
|
29
30
|
submitted: "submitted";
|
|
30
|
-
settled: "settled";
|
|
31
31
|
}>;
|
|
32
32
|
submission: z.ZodNullable<z.ZodObject<{
|
|
33
33
|
provider: z.ZodString;
|
|
@@ -78,9 +78,9 @@ export declare const LearningIterationReservationSchema: z.ZodObject<{
|
|
|
78
78
|
}, z.core.$strict>], "kind">;
|
|
79
79
|
requestHash: z.ZodString;
|
|
80
80
|
outcome: z.ZodEnum<{
|
|
81
|
+
reserved: "reserved";
|
|
81
82
|
waiting_for_data: "waiting_for_data";
|
|
82
83
|
waiting_for_review: "waiting_for_review";
|
|
83
|
-
reserved: "reserved";
|
|
84
84
|
}>;
|
|
85
85
|
counts: z.ZodObject<{
|
|
86
86
|
eligible: z.ZodNumber;
|
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
import { z } from "zod";
|
|
2
|
+
import { type BoundJudgeRequest, type BoundJudgeResponse } from "./model-judge.js";
|
|
3
|
+
export declare const JudgeCallReservationSchema: z.ZodObject<{
|
|
4
|
+
id: z.ZodString;
|
|
5
|
+
requestHash: z.ZodString;
|
|
6
|
+
reservedUsd: z.ZodNumber;
|
|
7
|
+
status: z.ZodEnum<{
|
|
8
|
+
reserved: "reserved";
|
|
9
|
+
settled: "settled";
|
|
10
|
+
not_dispatched: "not_dispatched";
|
|
11
|
+
}>;
|
|
12
|
+
response: z.ZodNullable<z.ZodObject<{
|
|
13
|
+
text: z.ZodString;
|
|
14
|
+
modelId: z.ZodString;
|
|
15
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
16
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
17
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
18
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
19
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
20
|
+
}, z.core.$strict>>;
|
|
21
|
+
}, z.core.$strict>;
|
|
22
|
+
export type JudgeCallReservation = z.infer<typeof JudgeCallReservationSchema>;
|
|
23
|
+
export interface JudgeBudgetState {
|
|
24
|
+
maximumSpendUsd: number;
|
|
25
|
+
calls: JudgeCallReservation[];
|
|
26
|
+
}
|
|
27
|
+
export interface JudgeBudgetStore {
|
|
28
|
+
/** Atomic and durable. A dispatch transaction must also enforce the owner's
|
|
29
|
+
* current lease and running state. Settlement remains writable after cancel. */
|
|
30
|
+
transaction<T>(intent: "dispatch" | "settle", update: (state: JudgeBudgetState) => {
|
|
31
|
+
calls: JudgeCallReservation[];
|
|
32
|
+
result: T;
|
|
33
|
+
}): Promise<T>;
|
|
34
|
+
}
|
|
35
|
+
export declare function judgeCommittedSpend(calls: JudgeCallReservation[]): number;
|
|
36
|
+
/** A reservation with no response is an uncertain charge, never permission to
|
|
37
|
+
* retry. Replays return the retained response without another provider call. */
|
|
38
|
+
export declare function createBudgetedJudgeExecutor(options: {
|
|
39
|
+
store: JudgeBudgetStore;
|
|
40
|
+
maximumCharge: (request: BoundJudgeRequest) => number;
|
|
41
|
+
dispatch: (request: BoundJudgeRequest, signal?: AbortSignal) => Promise<BoundJudgeResponse>;
|
|
42
|
+
}): (callId: string, request: BoundJudgeRequest, signal?: AbortSignal) => Promise<BoundJudgeResponse>;
|
|
43
|
+
//# sourceMappingURL=judge-budget.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"judge-budget.d.ts","sourceRoot":"","sources":["../../../../../src/learning/judge-budget.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,OAAO,EAA4B,KAAK,iBAAiB,EAAE,KAAK,kBAAkB,EAAE,MAAM,kBAAkB,CAAC;AAE7G,eAAO,MAAM,0BAA0B;;;;;;;;;;;;;;;;;;kBAKrC,CAAC;AACH,MAAM,MAAM,oBAAoB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,0BAA0B,CAAC,CAAC;AAC9E,MAAM,WAAW,gBAAgB;IAAG,eAAe,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,oBAAoB,EAAE,CAAA;CAAE;AAC5F,MAAM,WAAW,gBAAgB;IAC/B;oFACgF;IAChF,WAAW,CAAC,CAAC,EAAE,MAAM,EAAE,UAAU,GAAG,QAAQ,EAAE,MAAM,EAAE,CAAC,KAAK,EAAE,gBAAgB,KAAK;QAAE,KAAK,EAAE,oBAAoB,EAAE,CAAC;QAAC,MAAM,EAAE,CAAC,CAAA;KAAE,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;CAC9I;AAED,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,oBAAoB,EAAE,GAAG,MAAM,CAEzE;AAED;gFACgF;AAChF,wBAAgB,2BAA2B,CAAC,OAAO,EAAE;IACnD,KAAK,EAAE,gBAAgB,CAAC;IACxB,aAAa,EAAE,CAAC,OAAO,EAAE,iBAAiB,KAAK,MAAM,CAAC;IACtD,QAAQ,EAAE,CAAC,OAAO,EAAE,iBAAiB,EAAE,MAAM,CAAC,EAAE,WAAW,KAAK,OAAO,CAAC,kBAAkB,CAAC,CAAC;CAC7F,IACe,QAAQ,MAAM,EAAE,SAAS,iBAAiB,EAAE,SAAS,WAAW,KAAG,OAAO,CAAC,kBAAkB,CAAC,CAmC7G"}
|
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
import { z } from "zod";
|
|
2
|
+
import { type ImmutableAssetRef } from "@openpond/harness";
|
|
3
|
+
import { type ModelJudgeRunner } from "../graders.js";
|
|
4
|
+
export interface BoundJudgeRequest {
|
|
5
|
+
providerId: string;
|
|
6
|
+
modelId: string;
|
|
7
|
+
revision: string | null;
|
|
8
|
+
temperature: number;
|
|
9
|
+
system: string;
|
|
10
|
+
data: string;
|
|
11
|
+
}
|
|
12
|
+
export declare const BoundJudgeResponseSchema: z.ZodObject<{
|
|
13
|
+
text: z.ZodString;
|
|
14
|
+
modelId: z.ZodString;
|
|
15
|
+
modelRevision: z.ZodNullable<z.ZodString>;
|
|
16
|
+
responseId: z.ZodNullable<z.ZodString>;
|
|
17
|
+
inputTokens: z.ZodNullable<z.ZodNumber>;
|
|
18
|
+
outputTokens: z.ZodNullable<z.ZodNumber>;
|
|
19
|
+
costUsd: z.ZodNullable<z.ZodNumber>;
|
|
20
|
+
}, z.core.$strict>;
|
|
21
|
+
export type BoundJudgeResponse = z.infer<typeof BoundJudgeResponseSchema>;
|
|
22
|
+
/** Hosts resolve private assets and own durable authorization before dispatch.
|
|
23
|
+
* executeBudgeted must reserve the maximum charge before a provider call and
|
|
24
|
+
* retain its outcome even when the caller cancels or the response is malformed. */
|
|
25
|
+
export declare function createBoundModelJudgeRunner(options: {
|
|
26
|
+
readRubric: (reference: ImmutableAssetRef) => Promise<string>;
|
|
27
|
+
executeBudgeted: (request: BoundJudgeRequest, signal?: AbortSignal) => Promise<BoundJudgeResponse>;
|
|
28
|
+
}): ModelJudgeRunner;
|
|
29
|
+
//# sourceMappingURL=model-judge.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"model-judge.d.ts","sourceRoot":"","sources":["../../../../../src/learning/model-judge.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,EAAuB,KAAK,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AAChF,OAAO,EAA2B,KAAK,gBAAgB,EAAE,MAAM,eAAe,CAAC;AAI/E,MAAM,WAAW,iBAAiB;IAChC,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,MAAM,CAAC;IAChB,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,WAAW,EAAE,MAAM,CAAC;IACpB,MAAM,EAAE,MAAM,CAAC;IACf,IAAI,EAAE,MAAM,CAAC;CACd;AACD,eAAO,MAAM,wBAAwB;;;;;;;;kBAK1B,CAAC;AACZ,MAAM,MAAM,kBAAkB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,wBAAwB,CAAC,CAAC;AAE1E;;mFAEmF;AACnF,wBAAgB,2BAA2B,CAAC,OAAO,EAAE;IACnD,UAAU,EAAE,CAAC,SAAS,EAAE,iBAAiB,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IAC9D,eAAe,EAAE,CAAC,OAAO,EAAE,iBAAiB,EAAE,MAAM,CAAC,EAAE,WAAW,KAAK,OAAO,CAAC,kBAAkB,CAAC,CAAC;CACpG,GAAG,gBAAgB,CA2BnB"}
|