Skip to content

cayu.evals.suite_authoring

Public Cayu API. Definitions are grouped by their public source module, and definitions from private implementation modules appear on their package’s page; adding a symbol does not move existing links. The import shown at each entry is a supported public namespace. Read the Python API overview and install the required extra.

from cayu.evals import add_eval_case

Return a new suite revision containing one new stable case ID.

add_eval_case(document: EvalSuiteDocumentV1, case: EvalCaseDraftV1) -> EvalSuiteDocumentV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import compile_eval_suite_authoring_draft

Compile the exact draft wire version selected by the caller.

compile_eval_suite_authoring_draft(draft: EvalSuiteDraft) -> EvalSuiteDocument

Release source · Runtime guarantees · Capability guides

from cayu.evals import compile_eval_suite_draft

Canonicalize bounded editor material into one immutable suite revision.

compile_eval_suite_draft(draft: EvalSuiteDraftV1) -> EvalSuiteDocumentV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import compile_eval_suite_draft_v2

Canonicalize V2 editor material without losing structured judge fields.

compile_eval_suite_draft_v2(draft: EvalSuiteDraftV2) -> EvalSuiteDocumentV2

Release source · Runtime guarantees · Capability guides

from cayu.evals import compile_eval_suite_draft_v3

Canonicalize V3 material with its explicit suite trial policy.

compile_eval_suite_draft_v3(draft: EvalSuiteDraftV3) -> EvalSuiteDocumentV3

Release source · Runtime guarantees · Capability guides

from cayu.evals import duplicate_eval_case

Copy one case under an explicit new identity without aliasing history.

duplicate_eval_case(document: EvalSuiteDocumentV1, case_id: str, *, new_case_id: str, new_name: str | None=None) -> EvalSuiteDocumentV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EVAL_SUITE_AUTHORING_MAX_BYTES
EVAL_SUITE_AUTHORING_MAX_BYTES = EVAL_CORPUS_MAX_BYTES

Release source · Runtime guarantees · Capability guides

from cayu.evals import EVAL_SUITE_AUTHORING_SCHEMA_VERSION
EVAL_SUITE_AUTHORING_SCHEMA_VERSION = 1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EVAL_SUITE_AUTHORING_V2_SCHEMA_VERSION
EVAL_SUITE_AUTHORING_V2_SCHEMA_VERSION = 2

Release source · Runtime guarantees · Capability guides

from cayu.evals import EVAL_SUITE_AUTHORING_V3_SCHEMA_VERSION
EVAL_SUITE_AUTHORING_V3_SCHEMA_VERSION = 3

Release source · Runtime guarantees · Capability guides

from cayu.evals import eval_suite_document_from_json

Load one bounded authored-suite document from strict JSON text.

eval_suite_document_from_json(source: str) -> EvalSuiteDocument

Release source · Runtime guarantees · Capability guides

from cayu.evals import eval_suite_document_to_json

Return deterministic, human-readable authored-suite JSON.

eval_suite_document_to_json(document: EvalSuiteDocument) -> str

Release source · Runtime guarantees · Capability guides

from cayu.evals import eval_suite_selection

Freeze either every case or an explicit non-empty subset by revision.

eval_suite_selection(document: EvalSuiteDocument, case_ids: Sequence[str] | None=None) -> EvalSuiteSelectionV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EVAL_SUITE_SELECTION_SCHEMA_VERSION
EVAL_SUITE_SELECTION_SCHEMA_VERSION = 1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalCaseDefinitionV1

One immutable authored case revision with truthful optional provenance.

Bases: _EvalCaseAuthoringMaterial.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
name StrictStr required
description StrictStr | None None
source EvaluationSourceIdentityV1 | None None
stimulus EvalCaseStimulusV1 required
assertions tuple[EvalSuiteAuthoringAssertionSpecV1, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)
revision StrictStr required
EvalCaseDefinitionV1.validate_id(value: str, info) -> str
EvalCaseDefinitionV1.validate_name(value: str, info) -> str
EvalCaseDefinitionV1.validate_description(value: str | None, info) -> str | None
EvalCaseDefinitionV1.validate_stimulus(value: EvalCaseStimulusV1) -> EvalCaseStimulusV1
EvalCaseDefinitionV1.validate_source(value: EvaluationSourceIdentityV1 | None) -> EvaluationSourceIdentityV1 | None
EvalCaseDefinitionV1.validate_assertions_are_ordered(value: object, info) -> object
EvalCaseDefinitionV1.validate_assertions(value: tuple[EvalSuiteAuthoringAssertionSpecV1, ...]) -> tuple[EvalSuiteAuthoringAssertionSpecV1, ...]
EvalCaseDefinitionV1.validate_assertion_ids() -> _EvalCaseAuthoringMaterial
EvalCaseDefinitionV1.validate_revision_shape(value: str, info) -> str
EvalCaseDefinitionV1.validate_revision() -> EvalCaseDefinitionV1
EvalCaseDefinitionV1.create(draft: EvalCaseDraftV1) -> EvalCaseDefinitionV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalCaseDefinitionV2

One immutable V2 case revision with structured-judge fidelity.

Bases: _EvalCaseAuthoringMaterialV2.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
name StrictStr required
description StrictStr | None None
source EvaluationSourceIdentityV1 | None None
stimulus EvalCaseStimulusV1 required
assertions tuple[EvalSuiteAuthoringAssertionSpecV2, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)
revision StrictStr required
EvalCaseDefinitionV2.validate_id(value: str, info) -> str
EvalCaseDefinitionV2.validate_name(value: str, info) -> str
EvalCaseDefinitionV2.validate_description(value: str | None, info) -> str | None
EvalCaseDefinitionV2.validate_stimulus(value: EvalCaseStimulusV1) -> EvalCaseStimulusV1
EvalCaseDefinitionV2.validate_source(value: EvaluationSourceIdentityV1 | None) -> EvaluationSourceIdentityV1 | None
EvalCaseDefinitionV2.validate_assertions_are_ordered(value: object, info) -> object
EvalCaseDefinitionV2.validate_assertions(value: tuple[EvalSuiteAuthoringAssertionSpecV2, ...]) -> tuple[EvalSuiteAuthoringAssertionSpecV2, ...]
EvalCaseDefinitionV2.validate_assertion_ids() -> _EvalCaseAuthoringMaterialV2
EvalCaseDefinitionV2.validate_revision_shape(value: str, info) -> str
EvalCaseDefinitionV2.validate_revision() -> EvalCaseDefinitionV2
EvalCaseDefinitionV2.create(draft: EvalCaseDraftV2) -> EvalCaseDefinitionV2

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalCaseDraftV1

Bounded, revision-free case material accepted by authoring previews.

Bases: _EvalCaseAuthoringMaterial.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
name StrictStr required
description StrictStr | None None
source EvaluationSourceIdentityV1 | None None
stimulus EvalCaseStimulusV1 required
assertions tuple[EvalSuiteAuthoringAssertionSpecV1, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)
EvalCaseDraftV1.validate_id(value: str, info) -> str
EvalCaseDraftV1.validate_name(value: str, info) -> str
EvalCaseDraftV1.validate_description(value: str | None, info) -> str | None
EvalCaseDraftV1.validate_stimulus(value: EvalCaseStimulusV1) -> EvalCaseStimulusV1
EvalCaseDraftV1.validate_source(value: EvaluationSourceIdentityV1 | None) -> EvaluationSourceIdentityV1 | None
EvalCaseDraftV1.validate_assertions_are_ordered(value: object, info) -> object
EvalCaseDraftV1.validate_assertions(value: tuple[EvalSuiteAuthoringAssertionSpecV1, ...]) -> tuple[EvalSuiteAuthoringAssertionSpecV1, ...]
EvalCaseDraftV1.validate_assertion_ids() -> _EvalCaseAuthoringMaterial
EvalCaseDraftV1.from_case(case: EvalCaseDefinitionV1) -> EvalCaseDraftV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalCaseDraftV2

Revision-free V2 case material accepted by authoring previews.

Bases: _EvalCaseAuthoringDraftMaterialV2.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
name StrictStr required
description StrictStr | None None
source EvaluationSourceIdentityV1 | None None
stimulus EvalCaseStimulusV1 required
assertions tuple[EvalSuiteAuthoringAssertionDraftSpecV2, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)
EvalCaseDraftV2.validate_id(value: str, info) -> str
EvalCaseDraftV2.validate_name(value: str, info) -> str
EvalCaseDraftV2.validate_description(value: str | None, info) -> str | None
EvalCaseDraftV2.validate_stimulus(value: EvalCaseStimulusV1) -> EvalCaseStimulusV1
EvalCaseDraftV2.validate_source(value: EvaluationSourceIdentityV1 | None) -> EvaluationSourceIdentityV1 | None
EvalCaseDraftV2.validate_assertions_are_ordered(value: object, info) -> object
EvalCaseDraftV2.validate_assertions(value: tuple[EvalSuiteAuthoringAssertionDraftSpecV2, ...]) -> tuple[EvalSuiteAuthoringAssertionDraftSpecV2, ...]
EvalCaseDraftV2.validate_assertion_ids() -> _EvalCaseAuthoringDraftMaterialV2
EvalCaseDraftV2.from_case(case: EvalCaseDefinitionV1 | EvalCaseDefinitionV2) -> EvalCaseDraftV2

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalCaseStimulusV1
EvalCaseStimulusV1: TypeAlias = Annotated[EvalSimpleInputStimulusV1 | EvalScenarioStimulusV1, Field(discriminator='kind')]

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalScenarioStimulusV1

A content-addressed reference to one separately persisted scenario.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
kind Literal['scenario'] 'scenario'
scenario_id StrictStr required
scenario_revision StrictStr required
EvalScenarioStimulusV1.validate_scenario_id(value: str, info) -> str
EvalScenarioStimulusV1.validate_scenario_revision(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSelectedCaseV1

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
revision StrictStr required
EvalSelectedCaseV1.validate_id(value: str, info) -> str
EvalSelectedCaseV1.validate_revision(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSimpleInputStimulusV1

One ordinary fresh invocation authored without captured provenance.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
kind Literal['simple_input'] 'simple_input'
input RunInputSpec required
EvalSimpleInputStimulusV1.validate_input(value: RunInputSpec) -> RunInputSpec

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteDocumentV1

One immutable, authority-free authored suite and its exact case revisions.

Bases: _SchemaV1PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[1] EVAL_SUITE_AUTHORING_SCHEMA_VERSION
revision StrictStr required
target_key StrictStr required
suite EvalSuiteSpec required
cases tuple[EvalCaseDefinitionV1, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteDocumentV1.validate_schema_version_type(value: object) -> object
EvalSuiteDocumentV1.validate_revision_shape(value: str, info) -> str
EvalSuiteDocumentV1.validate_target_key(value: str, info) -> str
EvalSuiteDocumentV1.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteDocumentV1.validate_suite(value: EvalSuiteSpec) -> EvalSuiteSpec
EvalSuiteDocumentV1.validate_cases(value: tuple[EvalCaseDefinitionV1, ...]) -> tuple[EvalCaseDefinitionV1, ...]
EvalSuiteDocumentV1.validate_contract() -> EvalSuiteDocumentV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteDocumentV2

Immutable V2 suite preserving structured rubric and judge contracts.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[2] EVAL_SUITE_AUTHORING_V2_SCHEMA_VERSION
revision StrictStr required
target_key StrictStr required
suite EvalSuiteSpec required
cases tuple[EvalCaseDefinitionV2, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteDocumentV2.validate_revision_shape(value: str, info) -> str
EvalSuiteDocumentV2.validate_target_key(value: str, info) -> str
EvalSuiteDocumentV2.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteDocumentV2.validate_suite(value: EvalSuiteSpec) -> EvalSuiteSpec
EvalSuiteDocumentV2.validate_cases(value: tuple[EvalCaseDefinitionV2, ...]) -> tuple[EvalCaseDefinitionV2, ...]
EvalSuiteDocumentV2.validate_contract() -> EvalSuiteDocumentV2

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteDocumentV3

Immutable V3 suite with an explicit revisioned trial decision policy.

Bases: EvalSuiteDocumentV2.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[3] EVAL_SUITE_AUTHORING_V3_SCHEMA_VERSION
revision StrictStr required
target_key StrictStr required
suite EvalSuiteSpec required
cases tuple[EvalCaseDefinitionV2, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteDocumentV3.validate_revision_shape(value: str, info) -> str
EvalSuiteDocumentV3.validate_target_key(value: str, info) -> str
EvalSuiteDocumentV3.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteDocumentV3.validate_suite(value: EvalSuiteSpec) -> EvalSuiteSpec
EvalSuiteDocumentV3.validate_cases(value: tuple[EvalCaseDefinitionV2, ...]) -> tuple[EvalCaseDefinitionV2, ...]
EvalSuiteDocumentV3.validate_contract() -> EvalSuiteDocumentV2
EvalSuiteDocumentV3.validate_schema_version_type(value: object) -> object
EvalSuiteDocumentV3.validate_v3_contract() -> EvalSuiteDocumentV3

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteDraftV1

Revision-free suite material edited by an SDK or Control Plane client.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
target_key StrictStr required
name StrictStr required
description StrictStr | None None
trial_request TrialRequestSpec Field(default_factory=TrialRequestSpec)
cases tuple[EvalCaseDraftV1, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteDraftV1.validate_ids(value: str, info) -> str
EvalSuiteDraftV1.validate_name(value: str, info) -> str
EvalSuiteDraftV1.validate_description(value: str | None, info) -> str | None
EvalSuiteDraftV1.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteDraftV1.validate_trial_request(value: TrialRequestSpec) -> TrialRequestSpec
EvalSuiteDraftV1.validate_cases(value: tuple[EvalCaseDraftV1, ...]) -> tuple[EvalCaseDraftV1, ...]
EvalSuiteDraftV1.validate_case_ids() -> EvalSuiteDraftV1
EvalSuiteDraftV1.from_document(document: EvalSuiteDocumentV1) -> EvalSuiteDraftV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteDraftV2

Explicit V2 editor material with structured-judge authoring support.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[2] EVAL_SUITE_AUTHORING_V2_SCHEMA_VERSION
id StrictStr required
target_key StrictStr required
name StrictStr required
description StrictStr | None None
trial_request TrialRequestSpec Field(default_factory=TrialRequestSpec)
cases tuple[EvalCaseDraftV2, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteDraftV2.validate_ids(value: str, info) -> str
EvalSuiteDraftV2.validate_name(value: str, info) -> str
EvalSuiteDraftV2.validate_description(value: str | None, info) -> str | None
EvalSuiteDraftV2.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteDraftV2.validate_trial_request(value: TrialRequestSpec) -> TrialRequestSpec
EvalSuiteDraftV2.validate_cases(value: tuple[EvalCaseDraftV2, ...]) -> tuple[EvalCaseDraftV2, ...]
EvalSuiteDraftV2.validate_case_ids() -> EvalSuiteDraftV2
EvalSuiteDraftV2.from_document(document: EvalSuiteDocumentV1 | EvalSuiteDocumentV2) -> EvalSuiteDraftV2

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteDraftV3

V3 editor material with revision-free suite trial settings.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[3] EVAL_SUITE_AUTHORING_V3_SCHEMA_VERSION
id StrictStr required
target_key StrictStr required
name StrictStr required
description StrictStr | None None
trial_request EvalSuiteTrialRequestDraftV3 Field(default_factory=EvalSuiteTrialRequestDraftV3)
cases tuple[EvalCaseDraftV2, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteDraftV3.validate_schema_version_type(value: object) -> object
EvalSuiteDraftV3.validate_ids(value: str, info) -> str
EvalSuiteDraftV3.validate_name(value: str, info) -> str
EvalSuiteDraftV3.validate_description(value: str | None, info) -> str | None
EvalSuiteDraftV3.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteDraftV3.validate_trial_request(value: EvalSuiteTrialRequestDraftV3) -> EvalSuiteTrialRequestDraftV3
EvalSuiteDraftV3.validate_cases(value: tuple[EvalCaseDraftV2, ...]) -> tuple[EvalCaseDraftV2, ...]
EvalSuiteDraftV3.validate_case_ids() -> EvalSuiteDraftV3
EvalSuiteDraftV3.from_document(document: EvalSuiteDocumentV1 | EvalSuiteDocumentV2 | EvalSuiteDocumentV3) -> EvalSuiteDraftV3

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteSelectionV1

Content identity for a full-suite or explicit-subset launch request.

Bases: _SchemaV1PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[1] EVAL_SUITE_SELECTION_SCHEMA_VERSION
revision StrictStr required
suite_document_revision StrictStr required
suite_id StrictStr required
suite_revision StrictStr required
mode Literal['full_suite', 'subset'] required
cases tuple[EvalSelectedCaseV1, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_CASES)
EvalSuiteSelectionV1.validate_schema_version_type(value: object) -> object
EvalSuiteSelectionV1.validate_revisions(value: str, info) -> str
EvalSuiteSelectionV1.validate_suite_id(value: str, info) -> str
EvalSuiteSelectionV1.validate_cases_are_ordered(value: object, info) -> object
EvalSuiteSelectionV1.validate_cases(value: tuple[EvalSelectedCaseV1, ...]) -> tuple[EvalSelectedCaseV1, ...]
EvalSuiteSelectionV1.validate_contract() -> EvalSuiteSelectionV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import EvalSuiteTrialRequestDraftV3

Revision-free trial settings edited by SDK and Control Plane clients.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
trials StrictInt Field(default=1, ge=1, le=EVAL_CORPUS_MAX_TRIALS)
timeout_seconds StrictInt Field(default=300, ge=1, le=EVAL_CORPUS_MAX_TIMEOUT_SECONDS)
minimum_passed_trials StrictInt Field(default=1, ge=1, le=EVAL_CORPUS_MAX_TRIALS)
max_concurrency StrictInt Field(default=1, ge=1, le=EVAL_SUITE_MAX_CONCURRENCY)
EvalSuiteTrialRequestDraftV3.validate_policy_bounds() -> EvalSuiteTrialRequestDraftV3
EvalSuiteTrialRequestDraftV3.from_request(request: TrialRequestSpec) -> EvalSuiteTrialRequestDraftV3

Release source · Runtime guarantees · Capability guides

from cayu.evals import PublicJudgeReferenceDraftV1

Revision-free public reference truth accepted from an editor.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[1] 1
kind Literal['public_reference'] 'public_reference'
id StrictStr required
expected_answer StrictStr | None None
expected_facts tuple[StrictStr, ...] Field(default=(), max_length=EVAL_CORPUS_MAX_JUDGE_REFERENCE_FACTS)
PublicJudgeReferenceDraftV1.validate_id(value: str, info) -> str

validate_expected_facts_are_ordered (method)

Section titled “validate_expected_facts_are_ordered (method)”
PublicJudgeReferenceDraftV1.validate_expected_facts_are_ordered(value: object, info) -> object
PublicJudgeReferenceDraftV1.validate_contract() -> PublicJudgeReferenceDraftV1
PublicJudgeReferenceDraftV1.from_reference(reference: PublicJudgeReferenceV1) -> PublicJudgeReferenceDraftV1
PublicJudgeReferenceDraftV1.compile() -> PublicJudgeReferenceV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import revise_eval_case

Replace one case only when the caller reviewed its current revision.

revise_eval_case(document: EvalSuiteDocumentV1, replacement: EvalCaseDraftV1, *, expected_case_revision: str) -> EvalSuiteDocumentV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import StructuredModelJudgeAssertionDraftV1

Revision-free rubric assertion material compiled only by the server.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
id StrictStr required
kind Literal['structured_model_judge'] 'structured_model_judge'
description StrictStr | None None
judge_profile_key StrictStr required
judge_profile_revision StrictStr required
rubric StructuredRubricDraftV1 required
reference JudgeReferenceDraftV1 | None None
threshold StrictStr '0.5'
evidence EvalJudgeEvidenceSelectionV1 Field(default_factory=EvalJudgeEvidenceSelectionV1)
StructuredModelJudgeAssertionDraftV1.validate_ids(value: str, info) -> str
StructuredModelJudgeAssertionDraftV1.validate_description(value: str | None, info) -> str | None
StructuredModelJudgeAssertionDraftV1.validate_judge_profile_revision(value: str, info) -> str
StructuredModelJudgeAssertionDraftV1.validate_threshold(value: str, info) -> str
StructuredModelJudgeAssertionDraftV1.from_assertion(assertion: StructuredModelJudgeAssertionSpec) -> StructuredModelJudgeAssertionDraftV1
StructuredModelJudgeAssertionDraftV1.compile() -> StructuredModelJudgeAssertionSpec

Release source · Runtime guarantees · Capability guides

from cayu.evals import StructuredRubricDraftV1

Revision-free structured rubric material accepted from an editor.

Bases: _PortableModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal[1] 1
id StrictStr required
criteria tuple[StructuredRubricCriterionV1, ...] Field(min_length=1, max_length=EVAL_CORPUS_MAX_JUDGE_CRITERIA)
StructuredRubricDraftV1.validate_id(value: str, info) -> str
StructuredRubricDraftV1.validate_criteria_are_ordered(value: object, info) -> object
StructuredRubricDraftV1.validate_criteria(value: tuple[StructuredRubricCriterionV1, ...]) -> tuple[StructuredRubricCriterionV1, ...]
StructuredRubricDraftV1.validate_contract() -> StructuredRubricDraftV1
StructuredRubricDraftV1.from_rubric(rubric: StructuredRubricV1) -> StructuredRubricDraftV1
StructuredRubricDraftV1.compile() -> StructuredRubricV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import validate_eval_suite_selection

Match one selection to the exact immutable suite and case revisions.

validate_eval_suite_selection(selection: EvalSuiteSelectionV1, document: EvalSuiteDocument) -> EvalSuiteSelectionV1

Release source · Runtime guarantees · Capability guides

from cayu.evals import validate_expected_eval_suite_revision
validate_expected_eval_suite_revision(document: EvalSuiteDocumentV1, expected_revision: str) -> EvalSuiteDocumentV1

Release source · Runtime guarantees · Capability guides