anpord 0.1.21 → 0.1.22
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/api-context.d.cts +1 -1
- package/dist/api-context.d.mts +1 -1
- package/dist/{api-mocks-BenVx0e8.d.cts → api-mocks-B8a93t2t.d.cts} +8 -8
- package/dist/{api-mocks-BenVx0e8.d.mts → api-mocks-B8a93t2t.d.mts} +8 -8
- package/dist/api-runtime.d.cts +1 -1
- package/dist/api-runtime.d.mts +1 -1
- package/dist/api.d.cts +1 -1
- package/dist/api.d.mts +1 -1
- package/dist/bin.cjs +45 -35
- package/dist/bin.mjs +45 -35
- package/dist/{client-CsGxLUZq.d.cts → client-CE9VurDH.d.cts} +1190 -894
- package/dist/{client-BZDftC1x.cjs → client-CWVAt_wk.cjs} +1 -1
- package/dist/{client-D0l3kwWV.mjs → client-CfPf0hIc.mjs} +1 -1
- package/dist/{client-C7iEhiDm.d.mts → client-EdYxRHX0.d.mts} +1190 -894
- package/dist/{compiler-Cy85vNbp.cjs → compiler-B1UrHfCL.cjs} +18 -1
- package/dist/{compiler-DSQRHvXS.mjs → compiler-D9jidpZ0.mjs} +13 -2
- package/dist/config.cjs +1 -1
- package/dist/config.d.cts +1 -1
- package/dist/config.d.mts +1 -1
- package/dist/config.mjs +1 -1
- package/dist/{eval-turns-CTCoSuB6.d.cts → eval-turns-Uwe0G7PS.d.cts} +1 -1
- package/dist/{eval-turns-CTCoSuB6.d.mts → eval-turns-Uwe0G7PS.d.mts} +1 -1
- package/dist/eval.cjs +1 -1
- package/dist/eval.d.cts +1 -1
- package/dist/eval.d.mts +1 -1
- package/dist/eval.mjs +1 -1
- package/dist/{evals-Dw2Ek-mx.d.cts → evals-M2G4zvnS.d.cts} +36 -36
- package/dist/{evals-Dw2Ek-mx.d.mts → evals-M2G4zvnS.d.mts} +36 -36
- package/dist/{evals-api-CNoLlXL5.cjs → evals-api-DUiZWsyC.cjs} +54 -1
- package/dist/{evals-api-CK3F1KoI.mjs → evals-api-NLUUV9tM.mjs} +54 -1
- package/dist/index.cjs +3 -2
- package/dist/index.d.cts +5 -4
- package/dist/index.d.mts +5 -4
- package/dist/index.mjs +3 -3
- package/dist/source.d.cts +1 -1
- package/dist/source.d.mts +1 -1
- package/dist/{types-D-h67MVJ.d.mts → types-B1rRGWeL.d.mts} +27 -18
- package/dist/{types-Dvd_z5ge.d.cts → types-C1Np-yYM.d.cts} +27 -18
- package/dist/validator-runtime.d.cts +1 -1
- package/dist/validator-runtime.d.mts +1 -1
- package/dist/validators.d.cts +1 -1
- package/dist/validators.d.mts +1 -1
- package/package.json +1 -1
|
@@ -117,6 +117,8 @@ var ApiKeyAuthentication = class extends HttpApiMiddleware.Tag()("@anpord/schema
|
|
|
117
117
|
const PROMPT_LIMIT = 16384;
|
|
118
118
|
const VERIFY_LIMIT = 8192;
|
|
119
119
|
const NAME_LIMIT = 200;
|
|
120
|
+
const TAG_LIMIT = 40;
|
|
121
|
+
const TAGS_PER_CASE = 12;
|
|
120
122
|
const VARIABLE_VALUE_LIMIT = 2048;
|
|
121
123
|
const EvalPrompt = Schema.String.pipe(Schema.maxLength(PROMPT_LIMIT), Schema.annotations({
|
|
122
124
|
description: "What the agent is asked to do.",
|
|
@@ -133,6 +135,11 @@ const EvalCaseName = Schema.String.pipe(Schema.maxLength(NAME_LIMIT), Schema.ann
|
|
|
133
135
|
identifier: "EvalCaseName",
|
|
134
136
|
message: () => `A case name must be at most ${NAME_LIMIT} characters.`
|
|
135
137
|
}));
|
|
138
|
+
Schema.Array(Schema.String.pipe(Schema.minLength(1), Schema.maxLength(TAG_LIMIT))).pipe(Schema.maxItems(TAGS_PER_CASE), Schema.annotations({
|
|
139
|
+
description: "What a case is grouped under.",
|
|
140
|
+
identifier: "EvalCaseTags",
|
|
141
|
+
message: () => `A case may carry at most ${TAGS_PER_CASE} tags of ${TAG_LIMIT} characters.`
|
|
142
|
+
}));
|
|
136
143
|
const EvalVariableValue = Schema.String.pipe(Schema.maxLength(VARIABLE_VALUE_LIMIT), Schema.annotations({
|
|
137
144
|
identifier: "EvalVariableValue",
|
|
138
145
|
message: () => `A variable value must be at most ${VARIABLE_VALUE_LIMIT} characters, because it is substituted into the prompt before it is quoted.`
|
|
@@ -536,6 +543,28 @@ const EvalRunPage = Schema.Struct({
|
|
|
536
543
|
runs: Schema.Array(EvalRunSummary),
|
|
537
544
|
total: Schema.Int
|
|
538
545
|
});
|
|
546
|
+
const EvalCaseSummary = Schema.Struct({
|
|
547
|
+
cellKey: Schema.String,
|
|
548
|
+
distribution: EvalDistribution,
|
|
549
|
+
harness: Schema.String,
|
|
550
|
+
lastRunAtMillis: Schema.Int,
|
|
551
|
+
lastRunId: Schema.String,
|
|
552
|
+
model: Schema.String,
|
|
553
|
+
name: Schema.String,
|
|
554
|
+
runCount: Schema.Int,
|
|
555
|
+
suite: Schema.NullOr(Schema.String),
|
|
556
|
+
tags: Schema.Array(Schema.String)
|
|
557
|
+
}).annotations({
|
|
558
|
+
description: "A case as the list shows it, with its newest run.",
|
|
559
|
+
identifier: "EvalCaseSummary"
|
|
560
|
+
});
|
|
561
|
+
const EvalCasePage = Schema.Struct({
|
|
562
|
+
cases: Schema.Array(EvalCaseSummary),
|
|
563
|
+
tags: Schema.Array(Schema.String)
|
|
564
|
+
}).annotations({
|
|
565
|
+
description: "Cases and every tag they carry between them.",
|
|
566
|
+
identifier: "EvalCasePage"
|
|
567
|
+
});
|
|
539
568
|
const EvalCellHistoryEntry = Schema.Struct({
|
|
540
569
|
trigger: Schema.optionalWith(Schema.NullOr(EvalTrigger), { default: () => null }),
|
|
541
570
|
distribution: EvalDistribution,
|
|
@@ -727,6 +756,13 @@ const PublicEvalSandbox = Schema.Literal(...HOSTED_SANDBOXES).annotations({
|
|
|
727
756
|
description: "The hosted sandbox a task runs in.",
|
|
728
757
|
identifier: "PublicEvalSandbox"
|
|
729
758
|
});
|
|
759
|
+
const ListCasesRequest = Schema.Struct({
|
|
760
|
+
limit: Schema.optional(Schema.Int),
|
|
761
|
+
tag: Schema.optional(Schema.NullOr(Schema.String))
|
|
762
|
+
}).annotations({
|
|
763
|
+
description: "Which cases to read.",
|
|
764
|
+
identifier: "ListCasesRequest"
|
|
765
|
+
});
|
|
730
766
|
const ListEvalsRequest = Schema.Struct({
|
|
731
767
|
cursor: Schema.optional(Schema.NullOr(EvalPageCursor)),
|
|
732
768
|
limit: Schema.optional(Schema.Int)
|
|
@@ -768,6 +804,23 @@ const PublicStartEvalRequest = Schema.Struct({
|
|
|
768
804
|
description: `Start a grid with at most 100 total case, task, and trial combinations.`,
|
|
769
805
|
identifier: "StartEvalRequest"
|
|
770
806
|
});
|
|
807
|
+
const CredentialLeaseRequest = Schema.Struct({
|
|
808
|
+
harness: EvalHarness,
|
|
809
|
+
id: Schema.String
|
|
810
|
+
}).annotations({
|
|
811
|
+
description: "Ask for the credentials a run the caller is executing needs, for the harness it names.",
|
|
812
|
+
identifier: "CredentialLeaseRequest"
|
|
813
|
+
});
|
|
814
|
+
const CredentialLease = Schema.Struct({
|
|
815
|
+
expiresAt: Schema.DateTimeUtc,
|
|
816
|
+
values: Schema.Record({
|
|
817
|
+
key: Schema.String,
|
|
818
|
+
value: Schema.String
|
|
819
|
+
})
|
|
820
|
+
}).annotations({
|
|
821
|
+
description: "Credentials for one run, held in memory and never written down. Short-lived: start another run rather than keeping these.",
|
|
822
|
+
identifier: "CredentialLease"
|
|
823
|
+
});
|
|
771
824
|
const ReportedTrial = Schema.Struct({
|
|
772
825
|
caseName: EvalCaseName,
|
|
773
826
|
events: Schema.Array(HarnessEvent),
|
|
@@ -787,6 +840,6 @@ const ReportTrialRequest = Schema.Struct({
|
|
|
787
840
|
description: "A trial the caller ran, named by the run it belongs to.",
|
|
788
841
|
identifier: "ReportTrialRequest"
|
|
789
842
|
});
|
|
790
|
-
var PublicEvalsGroup = class extends HttpApiGroup.make("evals").add(HttpApiEndpoint.post("list", "/evals.list").setPayload(ListEvalsRequest).addSuccess(EvalRunPage).annotate(OpenApi.Summary, "List eval runs").annotate(OpenApi.Description, "Newest first. Pass the `next` cursor from a response to read the page after it; a null `next` means there are no more.")).add(HttpApiEndpoint.post("start", "/evals.start").setPayload(PublicStartEvalRequest).addSuccess(StartedEval).annotate(OpenApi.Summary, "Start an eval run").annotate(OpenApi.Description, `Starts the grid and returns its id while trials continue in the background. ${PROFILE_HARNESS_RULE}`)).add(HttpApiEndpoint.post("reportTrial", "/evals.reportTrial").setPayload(ReportTrialRequest).addSuccess(Schema.Void).annotate(OpenApi.Summary, "Report a trial run outside the platform").annotate(OpenApi.Description, "For a run started with executeLocally. The result is recorded as reported: it was produced somewhere the platform cannot inspect, so it is marked and kept out of baselines.")).add(HttpApiEndpoint.post("finishRun", "/evals.finishRun").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(OpenApi.Summary, "Settle a run whose trials the caller ran").annotate(OpenApi.Description, "Settles the run from the trials reported so far. A run left unsettled is swept like any other abandoned work.")).add(HttpApiEndpoint.post("artifact", "/evals.artifact").setPayload(EvalArtifactRequest).addSuccess(EvalArtifact)).add(HttpApiEndpoint.post("get", "/evals.get").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(OpenApi.Summary, "Get an eval run")).add(HttpApiEndpoint.post("subscription", "/evals.subscription").setPayload(EvalRunRequest).addSuccess(RunSubscription).annotate(OpenApi.Summary, "Watch an eval run as it moves").annotate(OpenApi.Description, "Returns a read-only token scoped to this run, for following it in real time. Short-lived: request another when it expires.")).add(HttpApiEndpoint.post("cellHistory", "/evals.cellHistory").setPayload(EvalCellRequest).addSuccess(Schema.Array(EvalCellHistoryEntry)).annotate(OpenApi.Summary, "List a cell's history").annotate(OpenApi.Description, "Returns the 20 most recent results.")).add(HttpApiEndpoint.post("rerunCell", "/evals.rerunCell").setPayload(Schema.extend(EvalRunRequest, Schema.extend(EvalCellRequest, RerunCellRequest))).addSuccess(StartedEval).annotate(OpenApi.Summary, "Rerun one cell")).add(HttpApiEndpoint.post("models", "/evals.models").setPayload(EvalModelsRequest).addSuccess(ModelCatalogue).annotate(OpenApi.Summary, "List models available to the harness").annotate(OpenApi.Description, "The command harness has no catalogue of its own, so its list is empty: the model is whatever the profile's run command reads from ANPORD_MODEL.")).addError(BadRequest).addError(Conflict).addError(Forbidden).addError(NotFound).middleware(ApiKeyAuthentication).annotate(OpenApi.Title, "Evals").annotate(OpenApi.Description, "Run cases across harness, model, and sandbox combinations and compare the results with their baselines.") {};
|
|
843
|
+
var PublicEvalsGroup = class extends HttpApiGroup.make("evals").add(HttpApiEndpoint.post("list", "/evals.list").setPayload(ListEvalsRequest).addSuccess(EvalRunPage).annotate(OpenApi.Summary, "List eval runs").annotate(OpenApi.Description, "Newest first. Pass the `next` cursor from a response to read the page after it; a null `next` means there are no more.")).add(HttpApiEndpoint.post("cases", "/evals.cases").setPayload(ListCasesRequest).addSuccess(EvalCasePage).annotate(OpenApi.Summary, "List cases").annotate(OpenApi.Description, "One row per case, carrying its newest run. Pass `tag` to read only the cases carrying it.")).add(HttpApiEndpoint.post("start", "/evals.start").setPayload(PublicStartEvalRequest).addSuccess(StartedEval).annotate(OpenApi.Summary, "Start an eval run").annotate(OpenApi.Description, `Starts the grid and returns its id while trials continue in the background. ${PROFILE_HARNESS_RULE}`)).add(HttpApiEndpoint.post("credentials", "/evals.credentials").setPayload(CredentialLeaseRequest).addSuccess(CredentialLease).annotate(OpenApi.Summary, "Lease the credentials a local run needs").annotate(OpenApi.Description, "For a run started with executeLocally, so the machine running it holds no credentials of its own. Returns the organization credential for one harness, expiring in minutes. Sandbox credentials are never leased: a local run opens no cloud sandbox.")).add(HttpApiEndpoint.post("reportTrial", "/evals.reportTrial").setPayload(ReportTrialRequest).addSuccess(Schema.Void).annotate(OpenApi.Summary, "Report a trial run outside the platform").annotate(OpenApi.Description, "For a run started with executeLocally. The result is recorded as reported: it was produced somewhere the platform cannot inspect, so it is marked and kept out of baselines.")).add(HttpApiEndpoint.post("finishRun", "/evals.finishRun").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(OpenApi.Summary, "Settle a run whose trials the caller ran").annotate(OpenApi.Description, "Settles the run from the trials reported so far. A run left unsettled is swept like any other abandoned work.")).add(HttpApiEndpoint.post("artifact", "/evals.artifact").setPayload(EvalArtifactRequest).addSuccess(EvalArtifact)).add(HttpApiEndpoint.post("get", "/evals.get").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(OpenApi.Summary, "Get an eval run")).add(HttpApiEndpoint.post("subscription", "/evals.subscription").setPayload(EvalRunRequest).addSuccess(RunSubscription).annotate(OpenApi.Summary, "Watch an eval run as it moves").annotate(OpenApi.Description, "Returns a read-only token scoped to this run, for following it in real time. Short-lived: request another when it expires.")).add(HttpApiEndpoint.post("cellHistory", "/evals.cellHistory").setPayload(EvalCellRequest).addSuccess(Schema.Array(EvalCellHistoryEntry)).annotate(OpenApi.Summary, "List a cell's history").annotate(OpenApi.Description, "Returns the 20 most recent results.")).add(HttpApiEndpoint.post("rerunCell", "/evals.rerunCell").setPayload(Schema.extend(EvalRunRequest, Schema.extend(EvalCellRequest, RerunCellRequest))).addSuccess(StartedEval).annotate(OpenApi.Summary, "Rerun one cell")).add(HttpApiEndpoint.post("models", "/evals.models").setPayload(EvalModelsRequest).addSuccess(ModelCatalogue).annotate(OpenApi.Summary, "List models available to the harness").annotate(OpenApi.Description, "The command harness has no catalogue of its own, so its list is empty: the model is whatever the profile's run command reads from ANPORD_MODEL.")).addError(BadRequest).addError(Conflict).addError(Forbidden).addError(NotFound).middleware(ApiKeyAuthentication).annotate(OpenApi.Title, "Evals").annotate(OpenApi.Description, "Run cases across harness, model, and sandbox combinations and compare the results with their baselines.") {};
|
|
791
844
|
//#endregion
|
|
792
845
|
export { CredentialValues as S, Forbidden as _, EvalSandbox as a, CredentialIntegration as b, HarnessProfile as c, profileFitsHarness as d, EvalSourceFiles as f, Conflict as g, BadRequest as h, EvalRun as i, PROFILE_LIMITS as l, ApiKeyAuthentication as m, PublicStartEvalRequest as n, EvalValidator as o, EvalTrigger as p, HarnessEvent as r, EnvName as s, PublicEvalsGroup as t, ProfilePath as u, NotFound as v, CredentialScope as x, CredentialConnection as y };
|
package/dist/index.cjs
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
|
|
2
|
-
const require_client = require("./client-
|
|
2
|
+
const require_client = require("./client-CWVAt_wk.cjs");
|
|
3
3
|
const require_errors = require("./errors-BX1wry8K.cjs");
|
|
4
|
-
const require_compiler = require("./compiler-
|
|
4
|
+
const require_compiler = require("./compiler-B1UrHfCL.cjs");
|
|
5
5
|
const require_source = require("./source-aW5nSo2h.cjs");
|
|
6
6
|
let effect = require("effect");
|
|
7
7
|
let _effect_platform = require("@effect/platform");
|
|
@@ -413,6 +413,7 @@ exports.EvalTimeout = EvalTimeout;
|
|
|
413
413
|
exports.McpCallSchema = McpCallSchema;
|
|
414
414
|
exports.MissingApiKey = require_errors.MissingApiKey;
|
|
415
415
|
exports.empty = require_source.empty;
|
|
416
|
+
exports.evalCase = require_compiler.evalCase;
|
|
416
417
|
exports.files = require_source.files;
|
|
417
418
|
exports.layer = require_client.layer;
|
|
418
419
|
exports.make = require_client.make;
|
package/dist/index.d.cts
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { a as layer, i as DEFAULT_BASE_URL, n as AnpordClient, o as make, r as ClientOptions, t as AnpordApi } from "./client-
|
|
2
|
-
import { a as EvalTaskDefinition, c as Prepare, d as ProfileRef, f as
|
|
3
|
-
import { _ as EvalVerdict, a as EvalDistribution, c as EvalRun, d as EvalSandbox, f as EvalSource, g as EvalUsage, h as EvalTrialStatus, i as EvalComparison, l as EvalRunStatus, m as EvalTrial, n as EvalCell, o as EvalHarness, p as EvalTask, r as EvalCellHistoryEntry, s as EvalJournalEntry, t as EvalArtifact, u as EvalRunSummary } from "./evals-
|
|
1
|
+
import { a as layer, i as DEFAULT_BASE_URL, n as AnpordClient, o as make, r as ClientOptions, t as AnpordApi } from "./client-CE9VurDH.cjs";
|
|
2
|
+
import { a as EvalTaskDefinition, c as Prepare, d as ProfileRef, f as SingleCaseDefinition, h as ValidatorResult, i as EvalDefinition, l as PrepareContext, m as ValidatorContext, n as CommandResult, o as ExecOptions, p as Validator, r as EvalCaseDefinition, s as HarnessRef, t as CaseCache, u as PrepareValue } from "./types-C1Np-yYM.cjs";
|
|
3
|
+
import { _ as EvalVerdict, a as EvalDistribution, c as EvalRun, d as EvalSandbox, f as EvalSource, g as EvalUsage, h as EvalTrialStatus, i as EvalComparison, l as EvalRunStatus, m as EvalTrial, n as EvalCell, o as EvalHarness, p as EvalTask, r as EvalCellHistoryEntry, s as EvalJournalEntry, t as EvalArtifact, u as EvalRunSummary } from "./evals-M2G4zvnS.cjs";
|
|
4
4
|
import { n as McpCallSchema, t as McpCall } from "./calls-DJGZJPrF.cjs";
|
|
5
5
|
import { n as CliCallSchema, t as CliCall } from "./calls-6Z1EoEa-.cjs";
|
|
6
6
|
import { empty, files, repo } from "./source.cjs";
|
|
@@ -348,10 +348,11 @@ declare class AnpordError extends Error {
|
|
|
348
348
|
//#endregion
|
|
349
349
|
//#region src/evals/define.d.ts
|
|
350
350
|
declare function suite<const Definition extends EvalDefinition>(definition: Definition): Definition;
|
|
351
|
+
declare function evalCase(definition: SingleCaseDefinition): EvalDefinition;
|
|
351
352
|
//#endregion
|
|
352
353
|
//#region src/index.d.ts
|
|
353
354
|
type StartEvalRequest = Parameters<EvalsSurface["start"]>[0];
|
|
354
355
|
type EvalCase = StartEvalRequest["cases"][number];
|
|
355
356
|
type EvalTaskRequest = StartEvalRequest["tasks"][number];
|
|
356
357
|
//#endregion
|
|
357
|
-
export { Anpord, AnpordApi, type AnpordClient, AnpordError, type AnpordOptions, type AnpordPromptVariables, type CacheOptions, type CaseCache, type CatalogueModel, type CliCall, CliCallSchema, type ClientOptions, type CommandResult, DEFAULT_BASE_URL, EvalAborted, type EvalArtifact, EvalCase, type EvalCaseDefinition, type EvalCell, type EvalCellHistoryEntry, type EvalComparison, type EvalDefinition, type EvalDistribution, type EvalHarness, type EvalJournalEntry, type EvalRun, type EvalRunStatus, type EvalRunSummary, type EvalSandbox, type EvalSource, type EvalTask, type EvalTaskDefinition, EvalTaskRequest, EvalTimeout, type EvalTrial, type EvalTrialStatus, type EvalTrigger, type EvalUsage, type EvalValidation, type EvalVerdict, type EvalsSurface, type ExecOptions, type GetPromptOptions, type HarnessRef, type McpCall, McpCallSchema, MissingApiKey, type ModelCatalogue, type Prepare, type PrepareContext, type PrepareValue, type ProfileRef, type PromptFallback, type PromptMetadata, type PromptResult, type PromptsSurface, type PublicPrompt, type PublicPromptSummary, type PublicPromptWithVersions, type PublicVersion, type RerunCellRequest, StartEvalRequest, type StartedEval, type ValidationCall, type ValidationValue, type Validator, type ValidatorContext, type ValidatorResult, type WaitOptions, empty, files, layer, make, repo, suite };
|
|
358
|
+
export { Anpord, AnpordApi, type AnpordClient, AnpordError, type AnpordOptions, type AnpordPromptVariables, type CacheOptions, type CaseCache, type CatalogueModel, type CliCall, CliCallSchema, type ClientOptions, type CommandResult, DEFAULT_BASE_URL, EvalAborted, type EvalArtifact, EvalCase, type EvalCaseDefinition, type EvalCell, type EvalCellHistoryEntry, type EvalComparison, type EvalDefinition, type EvalDistribution, type EvalHarness, type EvalJournalEntry, type EvalRun, type EvalRunStatus, type EvalRunSummary, type EvalSandbox, type EvalSource, type EvalTask, type EvalTaskDefinition, EvalTaskRequest, EvalTimeout, type EvalTrial, type EvalTrialStatus, type EvalTrigger, type EvalUsage, type EvalValidation, type EvalVerdict, type EvalsSurface, type ExecOptions, type GetPromptOptions, type HarnessRef, type McpCall, McpCallSchema, MissingApiKey, type ModelCatalogue, type Prepare, type PrepareContext, type PrepareValue, type ProfileRef, type PromptFallback, type PromptMetadata, type PromptResult, type PromptsSurface, type PublicPrompt, type PublicPromptSummary, type PublicPromptWithVersions, type PublicVersion, type RerunCellRequest, StartEvalRequest, type StartedEval, type ValidationCall, type ValidationValue, type Validator, type ValidatorContext, type ValidatorResult, type WaitOptions, empty, evalCase, files, layer, make, repo, suite };
|
package/dist/index.d.mts
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { a as layer, i as DEFAULT_BASE_URL, n as AnpordClient, o as make, r as ClientOptions, t as AnpordApi } from "./client-
|
|
2
|
-
import { a as EvalTaskDefinition, c as Prepare, d as ProfileRef, f as
|
|
3
|
-
import { _ as EvalVerdict, a as EvalDistribution, c as EvalRun, d as EvalSandbox, f as EvalSource, g as EvalUsage, h as EvalTrialStatus, i as EvalComparison, l as EvalRunStatus, m as EvalTrial, n as EvalCell, o as EvalHarness, p as EvalTask, r as EvalCellHistoryEntry, s as EvalJournalEntry, t as EvalArtifact, u as EvalRunSummary } from "./evals-
|
|
1
|
+
import { a as layer, i as DEFAULT_BASE_URL, n as AnpordClient, o as make, r as ClientOptions, t as AnpordApi } from "./client-EdYxRHX0.mjs";
|
|
2
|
+
import { a as EvalTaskDefinition, c as Prepare, d as ProfileRef, f as SingleCaseDefinition, h as ValidatorResult, i as EvalDefinition, l as PrepareContext, m as ValidatorContext, n as CommandResult, o as ExecOptions, p as Validator, r as EvalCaseDefinition, s as HarnessRef, t as CaseCache, u as PrepareValue } from "./types-B1rRGWeL.mjs";
|
|
3
|
+
import { _ as EvalVerdict, a as EvalDistribution, c as EvalRun, d as EvalSandbox, f as EvalSource, g as EvalUsage, h as EvalTrialStatus, i as EvalComparison, l as EvalRunStatus, m as EvalTrial, n as EvalCell, o as EvalHarness, p as EvalTask, r as EvalCellHistoryEntry, s as EvalJournalEntry, t as EvalArtifact, u as EvalRunSummary } from "./evals-M2G4zvnS.mjs";
|
|
4
4
|
import { n as McpCallSchema, t as McpCall } from "./calls-DJGZJPrF.mjs";
|
|
5
5
|
import { n as CliCallSchema, t as CliCall } from "./calls-6Z1EoEa-.mjs";
|
|
6
6
|
import { empty, files, repo } from "./source.mjs";
|
|
@@ -348,10 +348,11 @@ declare class AnpordError extends Error {
|
|
|
348
348
|
//#endregion
|
|
349
349
|
//#region src/evals/define.d.ts
|
|
350
350
|
declare function suite<const Definition extends EvalDefinition>(definition: Definition): Definition;
|
|
351
|
+
declare function evalCase(definition: SingleCaseDefinition): EvalDefinition;
|
|
351
352
|
//#endregion
|
|
352
353
|
//#region src/index.d.ts
|
|
353
354
|
type StartEvalRequest = Parameters<EvalsSurface["start"]>[0];
|
|
354
355
|
type EvalCase = StartEvalRequest["cases"][number];
|
|
355
356
|
type EvalTaskRequest = StartEvalRequest["tasks"][number];
|
|
356
357
|
//#endregion
|
|
357
|
-
export { Anpord, AnpordApi, type AnpordClient, AnpordError, type AnpordOptions, type AnpordPromptVariables, type CacheOptions, type CaseCache, type CatalogueModel, type CliCall, CliCallSchema, type ClientOptions, type CommandResult, DEFAULT_BASE_URL, EvalAborted, type EvalArtifact, EvalCase, type EvalCaseDefinition, type EvalCell, type EvalCellHistoryEntry, type EvalComparison, type EvalDefinition, type EvalDistribution, type EvalHarness, type EvalJournalEntry, type EvalRun, type EvalRunStatus, type EvalRunSummary, type EvalSandbox, type EvalSource, type EvalTask, type EvalTaskDefinition, EvalTaskRequest, EvalTimeout, type EvalTrial, type EvalTrialStatus, type EvalTrigger, type EvalUsage, type EvalValidation, type EvalVerdict, type EvalsSurface, type ExecOptions, type GetPromptOptions, type HarnessRef, type McpCall, McpCallSchema, MissingApiKey, type ModelCatalogue, type Prepare, type PrepareContext, type PrepareValue, type ProfileRef, type PromptFallback, type PromptMetadata, type PromptResult, type PromptsSurface, type PublicPrompt, type PublicPromptSummary, type PublicPromptWithVersions, type PublicVersion, type RerunCellRequest, StartEvalRequest, type StartedEval, type ValidationCall, type ValidationValue, type Validator, type ValidatorContext, type ValidatorResult, type WaitOptions, empty, files, layer, make, repo, suite };
|
|
358
|
+
export { Anpord, AnpordApi, type AnpordClient, AnpordError, type AnpordOptions, type AnpordPromptVariables, type CacheOptions, type CaseCache, type CatalogueModel, type CliCall, CliCallSchema, type ClientOptions, type CommandResult, DEFAULT_BASE_URL, EvalAborted, type EvalArtifact, EvalCase, type EvalCaseDefinition, type EvalCell, type EvalCellHistoryEntry, type EvalComparison, type EvalDefinition, type EvalDistribution, type EvalHarness, type EvalJournalEntry, type EvalRun, type EvalRunStatus, type EvalRunSummary, type EvalSandbox, type EvalSource, type EvalTask, type EvalTaskDefinition, EvalTaskRequest, EvalTimeout, type EvalTrial, type EvalTrialStatus, type EvalTrigger, type EvalUsage, type EvalValidation, type EvalVerdict, type EvalsSurface, type ExecOptions, type GetPromptOptions, type HarnessRef, type McpCall, McpCallSchema, MissingApiKey, type ModelCatalogue, type Prepare, type PrepareContext, type PrepareValue, type ProfileRef, type PromptFallback, type PromptMetadata, type PromptResult, type PromptsSurface, type PublicPrompt, type PublicPromptSummary, type PublicPromptWithVersions, type PublicVersion, type RerunCellRequest, StartEvalRequest, type StartedEval, type ValidationCall, type ValidationValue, type Validator, type ValidatorContext, type ValidatorResult, type WaitOptions, empty, evalCase, files, layer, make, repo, suite };
|
package/dist/index.mjs
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { i as make, n as DEFAULT_BASE_URL, r as layer, t as AnpordApi } from "./client-
|
|
1
|
+
import { i as make, n as DEFAULT_BASE_URL, r as layer, t as AnpordApi } from "./client-CfPf0hIc.mjs";
|
|
2
2
|
import { i as tokenMatcher, n as MissingApiKey, r as asAnpordError, t as AnpordError } from "./errors-B0YknR5V.mjs";
|
|
3
|
-
import { a as
|
|
3
|
+
import { a as sourceUrlOf, i as evalCase, o as suite, t as compileDefinition } from "./compiler-D9jidpZ0.mjs";
|
|
4
4
|
import { n as files, r as repo, t as empty } from "./source-BP11pzVV.mjs";
|
|
5
5
|
import { Cache, Cause, Clock, Context, Data, DateTime, Duration, Effect, Exit, Layer, ManagedRuntime, Option, Redacted, Schema } from "effect";
|
|
6
6
|
import { FetchHttpClient } from "@effect/platform";
|
|
@@ -402,4 +402,4 @@ const CliCallSchema = Schema.Struct({
|
|
|
402
402
|
output: Schema.optional(Schema.Unknown)
|
|
403
403
|
});
|
|
404
404
|
//#endregion
|
|
405
|
-
export { Anpord, AnpordApi, AnpordError, CliCallSchema, DEFAULT_BASE_URL, EvalAborted, EvalTimeout, McpCallSchema, MissingApiKey, empty, files, layer, make, repo, suite };
|
|
405
|
+
export { Anpord, AnpordApi, AnpordError, CliCallSchema, DEFAULT_BASE_URL, EvalAborted, EvalTimeout, McpCallSchema, MissingApiKey, empty, evalCase, files, layer, make, repo, suite };
|
package/dist/source.d.cts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { f as EvalSource } from "./evals-
|
|
1
|
+
import { f as EvalSource } from "./evals-M2G4zvnS.cjs";
|
|
2
2
|
//#region src/evals/source.d.ts
|
|
3
3
|
declare const repo: (spec: string) => EvalSource;
|
|
4
4
|
declare const files: (contents: Readonly<Record<string, string>>) => EvalSource;
|
package/dist/source.d.mts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { f as EvalSource } from "./evals-
|
|
1
|
+
import { f as EvalSource } from "./evals-M2G4zvnS.mjs";
|
|
2
2
|
//#region src/evals/source.d.ts
|
|
3
3
|
declare const repo: (spec: string) => EvalSource;
|
|
4
4
|
declare const files: (contents: Readonly<Record<string, string>>) => EvalSource;
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import "./errors-aGrd21jy.mjs";
|
|
2
|
-
import { t as ApiCall } from "./api-mocks-
|
|
3
|
-
import { a as EvalJudge, i as EvalUser, r as EvalTurn } from "./eval-turns-
|
|
4
|
-
import { f as EvalSource, o as EvalHarness } from "./evals-
|
|
2
|
+
import { t as ApiCall } from "./api-mocks-B8a93t2t.mjs";
|
|
3
|
+
import { a as EvalJudge, i as EvalUser, r as EvalTurn } from "./eval-turns-Uwe0G7PS.mjs";
|
|
4
|
+
import { f as EvalSource, o as EvalHarness } from "./evals-M2G4zvnS.mjs";
|
|
5
5
|
import { t as McpCall } from "./calls-DJGZJPrF.mjs";
|
|
6
6
|
import { n as McpServerDefinition } from "./define-HK0eeSNn.mjs";
|
|
7
7
|
import { t as ApiDefinition } from "./define-NZvmQuIv.mjs";
|
|
@@ -21,8 +21,8 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
21
21
|
readonly kind: "empty";
|
|
22
22
|
} | {
|
|
23
23
|
readonly kind: "repo";
|
|
24
|
-
readonly url: string;
|
|
25
24
|
readonly ref: string | null;
|
|
25
|
+
readonly url: string;
|
|
26
26
|
} | {
|
|
27
27
|
readonly kind: "files";
|
|
28
28
|
readonly files: {
|
|
@@ -39,8 +39,8 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
39
39
|
} | null | undefined;
|
|
40
40
|
readonly user?: {
|
|
41
41
|
readonly kind: "simulated";
|
|
42
|
-
readonly goal: string;
|
|
43
42
|
readonly prompt: string;
|
|
43
|
+
readonly goal: string;
|
|
44
44
|
} | {
|
|
45
45
|
readonly kind: "scripted";
|
|
46
46
|
readonly replies: readonly string[];
|
|
@@ -59,8 +59,8 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
59
59
|
readonly content: string;
|
|
60
60
|
}[] | undefined;
|
|
61
61
|
}) | {
|
|
62
|
-
readonly name: string;
|
|
63
62
|
readonly kind: "judged";
|
|
63
|
+
readonly name: string;
|
|
64
64
|
readonly capture?: boolean | undefined;
|
|
65
65
|
readonly sourceFiles?: readonly {
|
|
66
66
|
readonly path: string;
|
|
@@ -76,12 +76,12 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
76
76
|
}[] | undefined;
|
|
77
77
|
}[];
|
|
78
78
|
readonly judges: readonly ({
|
|
79
|
-
readonly name: string;
|
|
80
79
|
readonly kind: "judge";
|
|
80
|
+
readonly provider: "openai";
|
|
81
81
|
readonly harness?: undefined;
|
|
82
|
+
readonly name: string;
|
|
82
83
|
readonly model: string;
|
|
83
84
|
readonly prompt: string;
|
|
84
|
-
readonly provider: "openai";
|
|
85
85
|
readonly expected?: string | undefined;
|
|
86
86
|
readonly choices: {
|
|
87
87
|
readonly [x: string]: number;
|
|
@@ -89,12 +89,12 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
89
89
|
readonly threshold: number;
|
|
90
90
|
readonly timeoutMs: number;
|
|
91
91
|
} | {
|
|
92
|
-
readonly name: string;
|
|
93
92
|
readonly kind: "judge";
|
|
94
|
-
readonly
|
|
93
|
+
readonly provider?: undefined;
|
|
94
|
+
readonly harness: "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor";
|
|
95
|
+
readonly name: string;
|
|
95
96
|
readonly model: string;
|
|
96
97
|
readonly prompt: string;
|
|
97
|
-
readonly provider?: undefined;
|
|
98
98
|
readonly expected?: string | undefined;
|
|
99
99
|
readonly choices: {
|
|
100
100
|
readonly [x: string]: number;
|
|
@@ -178,7 +178,7 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
178
178
|
}>;
|
|
179
179
|
}>, Schema.Struct<{
|
|
180
180
|
provider: Schema.optional<typeof Schema.Never>;
|
|
181
|
-
harness: Schema.SchemaClass<"
|
|
181
|
+
harness: Schema.SchemaClass<"codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", never>;
|
|
182
182
|
kind: Schema.Literal<["judge"]>;
|
|
183
183
|
name: Schema.filter<Schema.filter<typeof Schema.String>>;
|
|
184
184
|
model: Schema.filter<Schema.filter<typeof Schema.String>>;
|
|
@@ -204,19 +204,19 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
204
204
|
name: Schema.optional<Schema.filter<Schema.filter<typeof Schema.String>>>;
|
|
205
205
|
prompt: Schema.refine<string, typeof Schema.String>;
|
|
206
206
|
tasks: Schema.filter<Schema.filter<Schema.Array$<Schema.refine<{
|
|
207
|
-
readonly harness: "
|
|
207
|
+
readonly harness: "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor" | "command";
|
|
208
208
|
readonly model: string;
|
|
209
209
|
readonly sandbox?: "daytona" | "e2b" | "upstash" | "modal" | "cloudflare" | "vercel" | undefined;
|
|
210
210
|
readonly profile?: {
|
|
211
|
+
readonly run?: string | undefined;
|
|
212
|
+
readonly files: {
|
|
213
|
+
readonly [x: string]: string;
|
|
214
|
+
};
|
|
211
215
|
readonly name: string;
|
|
212
216
|
readonly env?: {
|
|
213
217
|
readonly [x: string]: string;
|
|
214
218
|
} | undefined;
|
|
215
|
-
readonly files: {
|
|
216
|
-
readonly [x: string]: string;
|
|
217
|
-
};
|
|
218
219
|
readonly install?: string | undefined;
|
|
219
|
-
readonly run?: string | undefined;
|
|
220
220
|
readonly systemPrompt?: string | undefined;
|
|
221
221
|
} | undefined;
|
|
222
222
|
}, Schema.Struct<{
|
|
@@ -316,6 +316,9 @@ interface EvalCaseBase {
|
|
|
316
316
|
readonly name: string;
|
|
317
317
|
readonly prepare?: Prepare | null;
|
|
318
318
|
readonly source?: DeclaredSource;
|
|
319
|
+
/** How this case is grouped. A case may carry several, and retagging one
|
|
320
|
+
* keeps its baseline, because tags are not part of its identity. */
|
|
321
|
+
readonly tags?: readonly string[];
|
|
319
322
|
readonly user?: EvalUser;
|
|
320
323
|
readonly variables?: Readonly<Record<string, string>>;
|
|
321
324
|
}
|
|
@@ -326,6 +329,12 @@ type EvalCaseDefinition = EvalCaseBase & ({
|
|
|
326
329
|
readonly validate?: never;
|
|
327
330
|
readonly verify: string;
|
|
328
331
|
});
|
|
332
|
+
/** One case and what to run it on, for an eval that measures a single thing. */
|
|
333
|
+
type SingleCaseDefinition = EvalCaseDefinition & {
|
|
334
|
+
readonly prompt: string;
|
|
335
|
+
readonly tasks: readonly EvalTaskDefinition[];
|
|
336
|
+
readonly trials: number;
|
|
337
|
+
};
|
|
329
338
|
interface EvalDefinition {
|
|
330
339
|
readonly api?: readonly ApiDefinition[];
|
|
331
340
|
readonly captureSource?: boolean;
|
|
@@ -340,4 +349,4 @@ interface EvalDefinition {
|
|
|
340
349
|
readonly trials: number;
|
|
341
350
|
}
|
|
342
351
|
//#endregion
|
|
343
|
-
export { EvalTaskDefinition as a, Prepare as c, ProfileRef as d,
|
|
352
|
+
export { EvalTaskDefinition as a, Prepare as c, ProfileRef as d, SingleCaseDefinition as f, PublicStartEvalRequest as g, ValidatorResult as h, EvalDefinition as i, PrepareContext as l, ValidatorContext as m, CommandResult as n, ExecOptions as o, Validator as p, EvalCaseDefinition as r, HarnessRef as s, CaseCache as t, PrepareValue as u };
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import "./errors-aGrd21jy.cjs";
|
|
2
|
-
import { t as ApiCall } from "./api-mocks-
|
|
3
|
-
import { a as EvalJudge, i as EvalUser, r as EvalTurn } from "./eval-turns-
|
|
4
|
-
import { f as EvalSource, o as EvalHarness } from "./evals-
|
|
2
|
+
import { t as ApiCall } from "./api-mocks-B8a93t2t.cjs";
|
|
3
|
+
import { a as EvalJudge, i as EvalUser, r as EvalTurn } from "./eval-turns-Uwe0G7PS.cjs";
|
|
4
|
+
import { f as EvalSource, o as EvalHarness } from "./evals-M2G4zvnS.cjs";
|
|
5
5
|
import { t as McpCall } from "./calls-DJGZJPrF.cjs";
|
|
6
6
|
import { n as McpServerDefinition } from "./define-HK0eeSNn.cjs";
|
|
7
7
|
import { t as ApiDefinition } from "./define-NZvmQuIv.cjs";
|
|
@@ -21,8 +21,8 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
21
21
|
readonly kind: "empty";
|
|
22
22
|
} | {
|
|
23
23
|
readonly kind: "repo";
|
|
24
|
-
readonly url: string;
|
|
25
24
|
readonly ref: string | null;
|
|
25
|
+
readonly url: string;
|
|
26
26
|
} | {
|
|
27
27
|
readonly kind: "files";
|
|
28
28
|
readonly files: {
|
|
@@ -39,8 +39,8 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
39
39
|
} | null | undefined;
|
|
40
40
|
readonly user?: {
|
|
41
41
|
readonly kind: "simulated";
|
|
42
|
-
readonly goal: string;
|
|
43
42
|
readonly prompt: string;
|
|
43
|
+
readonly goal: string;
|
|
44
44
|
} | {
|
|
45
45
|
readonly kind: "scripted";
|
|
46
46
|
readonly replies: readonly string[];
|
|
@@ -59,8 +59,8 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
59
59
|
readonly content: string;
|
|
60
60
|
}[] | undefined;
|
|
61
61
|
}) | {
|
|
62
|
-
readonly name: string;
|
|
63
62
|
readonly kind: "judged";
|
|
63
|
+
readonly name: string;
|
|
64
64
|
readonly capture?: boolean | undefined;
|
|
65
65
|
readonly sourceFiles?: readonly {
|
|
66
66
|
readonly path: string;
|
|
@@ -76,12 +76,12 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
76
76
|
}[] | undefined;
|
|
77
77
|
}[];
|
|
78
78
|
readonly judges: readonly ({
|
|
79
|
-
readonly name: string;
|
|
80
79
|
readonly kind: "judge";
|
|
80
|
+
readonly provider: "openai";
|
|
81
81
|
readonly harness?: undefined;
|
|
82
|
+
readonly name: string;
|
|
82
83
|
readonly model: string;
|
|
83
84
|
readonly prompt: string;
|
|
84
|
-
readonly provider: "openai";
|
|
85
85
|
readonly expected?: string | undefined;
|
|
86
86
|
readonly choices: {
|
|
87
87
|
readonly [x: string]: number;
|
|
@@ -89,12 +89,12 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
89
89
|
readonly threshold: number;
|
|
90
90
|
readonly timeoutMs: number;
|
|
91
91
|
} | {
|
|
92
|
-
readonly name: string;
|
|
93
92
|
readonly kind: "judge";
|
|
94
|
-
readonly
|
|
93
|
+
readonly provider?: undefined;
|
|
94
|
+
readonly harness: "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor";
|
|
95
|
+
readonly name: string;
|
|
95
96
|
readonly model: string;
|
|
96
97
|
readonly prompt: string;
|
|
97
|
-
readonly provider?: undefined;
|
|
98
98
|
readonly expected?: string | undefined;
|
|
99
99
|
readonly choices: {
|
|
100
100
|
readonly [x: string]: number;
|
|
@@ -178,7 +178,7 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
178
178
|
}>;
|
|
179
179
|
}>, Schema.Struct<{
|
|
180
180
|
provider: Schema.optional<typeof Schema.Never>;
|
|
181
|
-
harness: Schema.SchemaClass<"
|
|
181
|
+
harness: Schema.SchemaClass<"codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", never>;
|
|
182
182
|
kind: Schema.Literal<["judge"]>;
|
|
183
183
|
name: Schema.filter<Schema.filter<typeof Schema.String>>;
|
|
184
184
|
model: Schema.filter<Schema.filter<typeof Schema.String>>;
|
|
@@ -204,19 +204,19 @@ declare const PublicStartEvalRequest: Schema.Struct<{
|
|
|
204
204
|
name: Schema.optional<Schema.filter<Schema.filter<typeof Schema.String>>>;
|
|
205
205
|
prompt: Schema.refine<string, typeof Schema.String>;
|
|
206
206
|
tasks: Schema.filter<Schema.filter<Schema.Array$<Schema.refine<{
|
|
207
|
-
readonly harness: "
|
|
207
|
+
readonly harness: "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor" | "command";
|
|
208
208
|
readonly model: string;
|
|
209
209
|
readonly sandbox?: "daytona" | "e2b" | "upstash" | "modal" | "cloudflare" | "vercel" | undefined;
|
|
210
210
|
readonly profile?: {
|
|
211
|
+
readonly run?: string | undefined;
|
|
212
|
+
readonly files: {
|
|
213
|
+
readonly [x: string]: string;
|
|
214
|
+
};
|
|
211
215
|
readonly name: string;
|
|
212
216
|
readonly env?: {
|
|
213
217
|
readonly [x: string]: string;
|
|
214
218
|
} | undefined;
|
|
215
|
-
readonly files: {
|
|
216
|
-
readonly [x: string]: string;
|
|
217
|
-
};
|
|
218
219
|
readonly install?: string | undefined;
|
|
219
|
-
readonly run?: string | undefined;
|
|
220
220
|
readonly systemPrompt?: string | undefined;
|
|
221
221
|
} | undefined;
|
|
222
222
|
}, Schema.Struct<{
|
|
@@ -316,6 +316,9 @@ interface EvalCaseBase {
|
|
|
316
316
|
readonly name: string;
|
|
317
317
|
readonly prepare?: Prepare | null;
|
|
318
318
|
readonly source?: DeclaredSource;
|
|
319
|
+
/** How this case is grouped. A case may carry several, and retagging one
|
|
320
|
+
* keeps its baseline, because tags are not part of its identity. */
|
|
321
|
+
readonly tags?: readonly string[];
|
|
319
322
|
readonly user?: EvalUser;
|
|
320
323
|
readonly variables?: Readonly<Record<string, string>>;
|
|
321
324
|
}
|
|
@@ -326,6 +329,12 @@ type EvalCaseDefinition = EvalCaseBase & ({
|
|
|
326
329
|
readonly validate?: never;
|
|
327
330
|
readonly verify: string;
|
|
328
331
|
});
|
|
332
|
+
/** One case and what to run it on, for an eval that measures a single thing. */
|
|
333
|
+
type SingleCaseDefinition = EvalCaseDefinition & {
|
|
334
|
+
readonly prompt: string;
|
|
335
|
+
readonly tasks: readonly EvalTaskDefinition[];
|
|
336
|
+
readonly trials: number;
|
|
337
|
+
};
|
|
329
338
|
interface EvalDefinition {
|
|
330
339
|
readonly api?: readonly ApiDefinition[];
|
|
331
340
|
readonly captureSource?: boolean;
|
|
@@ -340,4 +349,4 @@ interface EvalDefinition {
|
|
|
340
349
|
readonly trials: number;
|
|
341
350
|
}
|
|
342
351
|
//#endregion
|
|
343
|
-
export { EvalTaskDefinition as a, Prepare as c, ProfileRef as d,
|
|
352
|
+
export { EvalTaskDefinition as a, Prepare as c, ProfileRef as d, SingleCaseDefinition as f, PublicStartEvalRequest as g, ValidatorResult as h, EvalDefinition as i, PrepareContext as l, ValidatorContext as m, CommandResult as n, ExecOptions as o, Validator as p, EvalCaseDefinition as r, HarnessRef as s, CaseCache as t, PrepareValue as u };
|
package/dist/validators.d.cts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { a as EvalJudge, i as EvalUser, n as EvalSimulatedUser, o as EvalJudgment, r as EvalTurn, t as EvalScriptedUser } from "./eval-turns-
|
|
1
|
+
import { a as EvalJudge, i as EvalUser, n as EvalSimulatedUser, o as EvalJudgment, r as EvalTurn, t as EvalScriptedUser } from "./eval-turns-Uwe0G7PS.cjs";
|
|
2
2
|
import { Schema } from "effect";
|
|
3
3
|
//#region ../schema/src/domain/eval-classifiers.d.ts
|
|
4
4
|
declare const EvalClassifier: Schema.filter<Schema.Struct<{
|
package/dist/validators.d.mts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { a as EvalJudge, i as EvalUser, n as EvalSimulatedUser, o as EvalJudgment, r as EvalTurn, t as EvalScriptedUser } from "./eval-turns-
|
|
1
|
+
import { a as EvalJudge, i as EvalUser, n as EvalSimulatedUser, o as EvalJudgment, r as EvalTurn, t as EvalScriptedUser } from "./eval-turns-Uwe0G7PS.mjs";
|
|
2
2
|
import { Schema } from "effect";
|
|
3
3
|
//#region ../schema/src/domain/eval-classifiers.d.ts
|
|
4
4
|
declare const EvalClassifier: Schema.filter<Schema.Struct<{
|