openpond-sdk 0.2.4 → 0.2.5

package/dist/types/packages/sdk/src/model-batch-review-preparation.d.ts

@@ -0,0 +1,351 @@
+import { type LearningTextAsset } from "@openpond/evals/learning";
+import { type RewardBinding, type RewardRelease } from "@openpond/evals/rewards";
+import type { TasksetPackage } from "./taskset-package-contracts.js";
+import type { ModelBatchReviewRequest } from "./model-batch-review-contracts.js";
+/** Hosts supply the authorized package and selected Reward graph. Only this
+ * new namespace becomes editable; parent sources and decisions are not copied. */
+export declare function prepareModelBatchReview(input: {
+    scope: string;
+    request: ModelBatchReviewRequest;
+    package: TasksetPackage;
+    now: string;
+    binding: RewardBinding;
+    rewards: RewardRelease[];
+    assets: LearningTextAsset[];
+}): {
+    source: {
+        schemaVersion: "openpond.learningSource.v1";
+        id: string;
+        revision: number;
+        name: string;
+        kind: "work" | "provider" | "direct" | "benchmark" | "opentelemetry";
+        taskDefinition: {
+            id: string;
+            contentHash: string;
+            revision: number;
+        };
+        enabled: boolean;
+        allowedSplits: ("train" | "validation" | "test" | "frozen_eval")[];
+        mapping: {
+            schemaVersion: "openpond.taskSourceMapping.v1";
+            exampleId: string;
+            attemptId: string;
+            occurredAt: string;
+            input: string;
+            observedOutput: string | null;
+            expected: string | null;
+            evaluatorContext: string | null;
+            familyKey: string;
+            split: "train" | "validation" | "test" | "frozen_eval";
+        } | null;
+        adapterVersion: string | null;
+        contentHash: string;
+        reviewOrigin?: {
+            modelId: string;
+            packageHash: string;
+            taskset: {
+                id: string;
+                contentHash: string;
+                revision: number;
+            };
+            batch: {
+                id: string;
+                contentHash: string;
+                revision: number;
+            };
+            rewardBinding: {
+                id: string;
+                contentHash: string;
+                revision: number;
+            };
+        } | undefined;
+    };
+    definition: {
+        schemaVersion: "openpond.taskDefinition.v1";
+        id: string;
+        revision: number;
+        name: string;
+        description: string;
+        instructions: string;
+        category: "custom" | "structured" | "question_answering" | "coding" | "tool_workflow";
+        familyNamespace: string;
+        inputSchema: Record<string, unknown>;
+        outputSchema: Record<string, unknown>;
+        rewardBinding: {
+            id: string;
+            contentHash: string;
+            revision: number;
+        };
+        harness: {
+            id: string;
+            contentHash: string;
+        } | null;
+        execution: {
+            policy: {
+                policyVisibleFields: string[];
+                privilegedFields: string[];
+                hiddenGraderRefs: string[];
+                connectedAppScopes: string[];
+            };
+            environment: {
+                protocolVersion: "openpond.environment.v1";
+                kind: "text" | "work" | "agent" | "custom_program";
+                entrypoint: string;
+                stateful: boolean;
+                deterministicSeeds: boolean;
+                lifecycle: ("create" | "reset" | "step" | "collect" | "destroy")[];
+                networkPolicy: "none" | "declared_read_only" | "declared_scoped";
+                defaultTimeoutMs: number;
+            };
+            tools: {
+                name: string;
+                description: string;
+                inputSchema: Record<string, unknown>;
+                inputSchemaHash: string;
+                sideEffect: "read" | "write";
+                timeoutMs: number;
+            }[];
+            capabilities: {
+                id: string;
+                required: boolean;
+                scopes: string[];
+                portability: "portable" | "host_adapter" | "local_only" | "hosted_only";
+            }[];
+            environmentRelease?: {
+                id: string;
+                contentHash: string;
+            } | undefined;
+            verifierSetRelease?: {
+                id: string;
+                contentHash: string;
+            } | undefined;
+        };
+        contentHash: string;
+    };
+    binding: {
+        schemaVersion: "openpond.rewardBinding.v1";
+        id: string;
+        revision: number;
+        sources: {
+            graderId: string;
+            reward: {
+                id: string;
+                contentHash: string;
+                revision: number;
+            };
+            role: "evaluation" | "training";
+            normalization: {
+                kind: "identity";
+            } | {
+                kind: "linear";
+                minimum: number;
+                maximum: number;
+                direction: "higher" | "lower";
+            };
+            weight: number;
+            required: boolean;
+            hardGate: boolean;
+            privileged: boolean;
+            fixtureRefs: {
+                id: string;
+                contentHash: string;
+            }[];
+        }[];
+        aggregation: "weighted_mean";
+        unscorable: "exclude_optional_require_all_required";
+        contentHash: string;
+        name?: string | undefined;
+        description?: string | undefined;
+        recipeRef?: {
+            id: string;
+            contentHash: string;
+            revision: number;
+        } | undefined;
+    };
+    rewards: {
+        schemaVersion: "openpond.rewardRelease.v1";
+        id: string;
+        revision: number;
+        name: string;
+        description: string;
+        implementation: {
+            kind: "state" | "content" | "schema" | "artifact" | "runtime_event";
+            config: Record<string, unknown>;
+        } | {
+            kind: "model_judge";
+            rubricRef: {
+                id: string;
+                path: string;
+                contentHash: string;
+                sizeBytes: number;
+                mediaType: string;
+                visibility: "policy" | "verifier" | "host_private";
+            };
+            calibrationStatus: "failed" | "pending" | "passed";
+            model?: {
+                providerId: string;
+                modelId: string;
+                revision: string | null;
+            } | undefined;
+            temperature?: number | undefined;
+        } | {
+            kind: "custom_verifier";
+            networkPolicy: "none";
+            verifierRef: {
+                id: string;
+                path: string;
+                contentHash: string;
+                sizeBytes: number;
+                mediaType: string;
+                visibility: "policy" | "verifier" | "host_private";
+            };
+            timeoutMs: number;
+            exportName?: string | undefined;
+        } | {
+            kind: "human";
+            rubricRef: {
+                id: string;
+                path: string;
+                contentHash: string;
+                sizeBytes: number;
+                mediaType: string;
+                visibility: "policy" | "verifier" | "host_private";
+            };
+            reviewerRole: string;
+        } | {
+            kind: "learned_model";
+            modelVersion: {
+                id: string;
+                contentHash: string;
+            };
+            inputContract: {
+                id: string;
+                path: string;
+                contentHash: string;
+                sizeBytes: number;
+                mediaType: string;
+                visibility: "policy" | "verifier" | "host_private";
+            };
+        };
+        rawScore: {
+            minimum: number;
+            maximum: number;
+        };
+        assets: {
+            id: string;
+            path: string;
+            contentHash: string;
+            sizeBytes: number;
+            mediaType: string;
+            visibility: "policy" | "verifier" | "host_private";
+        }[];
+        contentHash: string;
+        fixtureSetRef?: {
+            id: string;
+            path: string;
+            contentHash: string;
+            sizeBytes: number;
+            mediaType: string;
+            visibility: "policy" | "verifier" | "host_private";
+        } | undefined;
+    }[];
+    assets: {
+        schemaVersion: "openpond.learningTextAsset.v1";
+        id: string;
+        revision: 1;
+        asset: {
+            id: string;
+            path: string;
+            contentHash: string;
+            sizeBytes: number;
+            mediaType: string;
+            visibility: "policy" | "verifier" | "host_private";
+        };
+        text: string;
+        contentHash: string;
+    }[];
+    evidence: {
+        schemaVersion: "openpond.taskEvidence.v1";
+        id: string;
+        revision: number;
+        source: {
+            id: string;
+            contentHash: string;
+            revision: number;
+        };
+        submission: {
+            schemaVersion: "openpond.taskExample.v1";
+            sourceId: string;
+            idempotencyKey: string;
+            taskDefinition: {
+                id: string;
+                contentHash: string;
+                revision: number;
+            };
+            exampleId: string;
+            attemptId: string;
+            occurredAt: string;
+            familyKey: string | null;
+            split: "train" | "validation" | "test" | "frozen_eval";
+            input: Record<string, import("zod").JSONType>;
+            observedOutput: Record<string, import("zod").JSONType> | null;
+            expected: Record<string, import("zod").JSONType> | null;
+            evaluatorContext: Record<string, import("zod").JSONType> | null;
+            assets: {
+                id: string;
+                path: string;
+                contentHash: string;
+                sizeBytes: number;
+                mediaType: string;
+                visibility: "policy" | "verifier" | "host_private";
+            }[];
+            provenance: {
+                sourceRecordRef: string | null;
+                mappingHash: string | null;
+            };
+        };
+        supersedes: {
+            id: string;
+            contentHash: string;
+            revision: number;
+        } | null;
+        correctionFeedbackId: string | null;
+        receivedAt: string;
+        contentHash: string;
+    }[];
+    proposals: {
+        schemaVersion: "openpond.taskFeedbackRecord.v1";
+        id: string;
+        submission: {
+            schemaVersion: "openpond.taskFeedback.v1";
+            sourceId: string;
+            idempotencyKey: string;
+            exampleId: string;
+            attemptId: string;
+            expectedEvidenceHash: string | null;
+            occurredAt: string;
+            kind: "outcome" | "target_correction" | "ground_truth_correction" | "input_correction" | "family_resolution";
+            value: Record<string, import("zod").JSONType>;
+            note: string;
+        };
+        status: "rejected" | "applied" | "superseded" | "pending_example" | "pending_review";
+        evidence: {
+            id: string;
+            contentHash: string;
+            revision: number;
+        } | null;
+        createdAt: string;
+        revision: number;
+        review: {
+            actorId: string;
+            decision: {
+                id: string;
+                contentHash: string;
+                revision: number;
+            } | null;
+            note: string;
+            resolvedAt: string;
+        } | null;
+    }[];
+};
+//# sourceMappingURL=model-batch-review-preparation.d.ts.map

package/dist/types/packages/sdk/src/model-batch-review-preparation.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"model-batch-review-preparation.d.ts","sourceRoot":"","sources":["../../../../../src/model-batch-review-preparation.ts"],"names":[],"mappings":"AACA,OAAO,EAE8C,KAAK,iBAAiB,EAAE,MAAM,0BAA0B,CAAC;AAC9G,OAAO,EAAsF,KAAK,aAAa,EAAE,KAAK,aAAa,EAAE,MAAM,yBAAyB,CAAC;AACrK,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,gCAAgC,CAAC;AACrE,OAAO,KAAK,EAAE,uBAAuB,EAAE,MAAM,mCAAmC,CAAC;AAEjF;kFACkF;AAClF,wBAAgB,uBAAuB,CAAC,KAAK,EAAE;IAC7C,KAAK,EAAE,MAAM,CAAC;IAAC,OAAO,EAAE,uBAAuB,CAAC;IAAC,OAAO,EAAE,cAAc,CAAC;IAAC,GAAG,EAAE,MAAM,CAAC;IACtF,OAAO,EAAE,aAAa,CAAC;IAAC,OAAO,EAAE,aAAa,EAAE,CAAC;IAAC,MAAM,EAAE,iBAAiB,EAAE,CAAC;CAC/E;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAsFA"}