runbios-sdk 0.2.19-dev.277 → 0.2.19-dev.278

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/index.d.ts CHANGED
@@ -36,7 +36,7 @@ import { Loop } from './resources/loop.js';
36
36
  * SDK version. Sent as part of the User-Agent header.
37
37
  * Must match package.json "version" -- enforced by a contract test.
38
38
  */
39
- export declare const VERSION = "0.2.19-dev.277";
39
+ export declare const VERSION = "0.2.19-dev.278";
40
40
  export declare class RunBiOS {
41
41
  /** Search models, fetch configs, check adapter compatibility. */
42
42
  readonly models: Models;
@@ -75,7 +75,7 @@ export { Training } from './resources/training.js';
75
75
  export { Wallet } from './resources/wallet.js';
76
76
  export { GPU, type GPURecommendation } from './resources/gpu.js';
77
77
  export { Inference, validateChatRequest, parseSSE, buildInferenceRequest, contextSizingBasis, type ContextSizingBasis, CONTEXT_DEFAULT_CEILING, CONTEXT_EDITABLE_FLOOR, type ChatMessage, type FunctionTool, type ChatCompletionParams, type ChatCompletionResponse, type ChatCompletionChunk, type InferenceVerbParams, type CompletionParams, type EmbeddingParams, type RerankParams, type AnthropicMessageParams, type ServerlessUsageWindow, type ServerlessTimeseriesMetric, type ServerlessWorkspaceLimits, type ServerlessUsageEnvelope, type ServerlessUsageOverviewResponse, type ServerlessUsageRowsResponse, type ServerlessUsageTimeseriesResponse, type ServerlessUsageDailyResponse, type ServerlessSavingsTotals, type ServerlessSavingsPeriod, type ServerlessUsageSavingsResponse, } from './resources/inference.js';
78
- export type { BiOSConfig, PaginatedResponse, ApiErrorBody, AvailableGpuAlternative, CapacityMinimumRequirement, InferenceBookingAccepted, Model, ModelSearchParams, ModelSearchResponse, ModelDetailResponse, InferenceModel, InferenceModelListResponse, ModelConfig, Dataset, DatasetListParams, DatasetListResponse, DatasetUploadParams, DatasetPreview, DatasetPreviewParams, DatasetImportHFParams, DatasetRegisterHFParams, DatasetHubSearchParams, DatasetHubPreviewParams, DatasetValidation, DatasetFormatVariant, DatasetFormatSpec, DatasetFormatSpecs, DatasetStorageUsage, TrainingMethod, RLHFAlgorithm, AdapterType, TrainingJobStatus, TrainingStatusFilter, TrainingCreateParams, TrainingListParams, TrainingListResponse, TrainingJob, TrainingMetrics, TrainingResourceMetrics, TrainingDeviceMetrics, MetricPoint, MetricGraphConfig, TrainingCheckpoint, TrainingLogs, TrainingLogEntry, TrainingStopResponse, TrainingResumeResponse, CanonicalTrainingRequest, TrainingPreflightDataset, TrainingPreflightWarning, TrainingPreflightResponse, TrainingCapabilityChoice, TrainingConfigFieldCapability, TrainingCapabilities, GPUChoice, WalletBalance, Transaction, TransactionListResponse, TransactionListParams, GPUInfo, GPUPricingResponse, GPUOptionsParams, TrainingRecommendParams, TrainingAdvisorVerdict, TrainingAdvisorRecommendation, TrainingAdvisorUserShape, TrainingAdvisorJustification, TrainingAdvisorBasis, GPUOption, GPUOptionSuggestion, GPUOptionsResponse, InferenceStatus, InferenceCreateParams, InferenceUpdateParams, InferenceDeployment, InferenceDeploymentSummary, InferenceCreateResponse, InferenceListResponse, InferenceUpdateResponse, InferenceLifecycleResponse, InferenceDeleteResponse, InferenceGPUOptionMarket, InferenceGPUOption, InferenceGPUAlternative, InferenceGPUOptionsParams, InferenceGPUOptionsResponse, InferenceCanonicalRequest, InferencePreflightWarning, InferencePreflightResponse, AdapterCompatibility, AdapterCompatibilityResponse, AdapterCompatibilityParams, SupportedArchitecture, ArchitectureScope, SupportedArchitecturesResponse, ApiKey, ApiKeyScope, ApiKeyIntrospection, Organization, OrgMember, OrgInvite, Workspace, Integration, IntegrationCreateParams, StorageUsage, StorageObject, PromptTokensDetails, ChatCompletionUsage, ServingKind, TrainingRulePausedReason, TrainingTrigger, TrainingRecipe, TrainingRunState, TrainingVerdict, TrainingDecision, NotifyState, EvaluationStatus, EvaluationItemStatus, EvaluationWinner, EvaluationItemWinnerFilter, EvaluationWarningCode, TrainingToolCall, TrainingMessage, TrainingRuleServing, TrainingRunPromoteRequest, TrainingRunRejectRequest, TrainingRunRollbackRequest, TrainingRunCancelRequest, TrainingRunListParams, TrainingRun, TrainingRunSummary, TrainingRunEvent, TrainingRunLinks, TrainingRunActions, EvaluationRef, EvaluationDecoding, EvaluationJudgeDimension, EvaluationDimensionScore, EvaluationWarning, EvaluationMargin, Evaluation, EvaluationItemSide, EvaluationItem, EvaluationItemListParams, BenchmarkStatus, BenchmarkSourceKind, BenchmarkRunStatus, BenchmarkScoring, BenchmarkDimensionScore, Benchmark, BenchmarkItem, BenchmarkRun, BenchmarkHistoryPoint, BenchmarkSource, BenchmarkCreateParams, BenchmarkListParams, BenchmarkItemListParams, BenchmarkHistoryParams, BenchmarkRetireParams, InferenceAlias, InferenceAliasRequest, TrainingRuleDeleteResponse, TrainingRunListResponse, TrainingRunResponse, TrainingRunActionResponse, EvaluationItemsResponse, InferenceAliasListResponse, InferenceAliasResponse, InferenceAliasDeleteResponse, BenchmarkListResponse, BenchmarkItemsResponse, BenchmarkHistoryResponse, TrainingRuleRunRefusalCode, TrainingRuleSaveRefusalCode, TrainingRuleMonthlyLimitRefusal, PipelineStatus, PipelineActiveRun, PipelineVersion, PipelineAttempt, PipelineAttemptTrigger, PipelineAttemptOutcome, Pipeline, PipelineListParams, PipelineListResponse, PipelineResponse, PipelineTrainWhen, PipelineSchedule, PipelineWriteRequest, PipelineMutationResponse, PipelineImportParams, PipelineFloors, LoopModel, LoopModelsResponse, LoopTracePipeline, LoopTracePipelinesResponse, LoopMetricsHistoryParams, LoopMetricPoint, LoopSignalDay, LoopMetricsHistory, PromotionPolicyState, RuleBenchmark, PipelineBenchmarksResponse, PipelineBenchmarkAttachParams, PipelineBenchmarkMoveParams, CompareBenchmark, VersionBenchmarkScore, PromotionMeasurement, PromotionOutcome, VersionScores, VersionScoreDifference, VersionCompareResponse, VersionCompareParams, } from './types.js';
78
+ export type { BiOSConfig, PaginatedResponse, ApiErrorBody, AvailableGpuAlternative, CapacityMinimumRequirement, InferenceBookingAccepted, Model, ModelSearchParams, ModelSearchResponse, ModelDetailResponse, InferenceModel, InferenceModelListResponse, ModelConfig, Dataset, DatasetListParams, DatasetListResponse, DatasetUploadParams, DatasetPreview, DatasetPreviewParams, DatasetImportHFParams, DatasetRegisterHFParams, DatasetHubSearchParams, DatasetHubPreviewParams, DatasetValidation, DatasetFormatVariant, DatasetFormatSpec, DatasetFormatSpecs, DatasetStorageUsage, TrainingMethod, RLHFAlgorithm, AdapterType, TrainingJobStatus, TrainingStatusFilter, TrainingCreateParams, TrainingListParams, TrainingListResponse, TrainingJob, TrainingMetrics, TrainingResourceMetrics, TrainingDeviceMetrics, MetricPoint, MetricGraphConfig, TrainingCheckpoint, TrainingLogs, TrainingLogEntry, TrainingStopResponse, TrainingResumeResponse, CanonicalTrainingRequest, TrainingPreflightDataset, TrainingPreflightWarning, TrainingPreflightResponse, TrainingCapabilityChoice, TrainingConfigFieldCapability, TrainingCapabilities, GPUChoice, WalletBalance, Transaction, TransactionListResponse, TransactionListParams, GPUInfo, GPUPricingResponse, GPUOptionsParams, TrainingRecommendParams, TrainingAdvisorVerdict, TrainingAdvisorRecommendation, TrainingAdvisorUserShape, TrainingAdvisorJustification, TrainingAdvisorBasis, GPUOption, GPUOptionSuggestion, GPUOptionsResponse, InferenceStatus, InferenceCreateParams, InferenceUpdateParams, InferenceDeployment, InferenceDeploymentSummary, InferenceCreateResponse, InferenceListResponse, InferenceUpdateResponse, InferenceLifecycleResponse, InferenceDeleteResponse, InferenceGPUOptionMarket, InferenceGPUOption, InferenceGPUAlternative, InferenceGPUOptionsParams, InferenceGPUOptionsResponse, InferenceCanonicalRequest, InferencePreflightWarning, InferencePreflightResponse, AdapterCompatibility, AdapterCompatibilityResponse, AdapterCompatibilityParams, SupportedArchitecture, ArchitectureScope, SupportedArchitecturesResponse, ApiKey, ApiKeyScope, ApiKeyIntrospection, Organization, OrgMember, OrgInvite, Workspace, Integration, IntegrationCreateParams, StorageUsage, StorageObject, PromptTokensDetails, ChatCompletionUsage, ServingKind, TrainingRulePausedReason, TrainingTrigger, TrainingRecipe, TrainingRunState, TrainingVerdict, TrainingDecision, EvaluationStatus, EvaluationItemStatus, EvaluationWinner, EvaluationItemWinnerFilter, EvaluationWarningCode, TrainingToolCall, TrainingMessage, TrainingRuleServing, TrainingRunPromoteRequest, TrainingRunRejectRequest, TrainingRunRollbackRequest, TrainingRunCancelRequest, TrainingRunListParams, TrainingRun, TrainingRunSummary, TrainingRunEvent, TrainingRunLinks, TrainingRunActions, EvaluationRef, EvaluationDecoding, EvaluationJudgeDimension, EvaluationDimensionScore, EvaluationWarning, EvaluationMargin, Evaluation, EvaluationItemSide, EvaluationItem, EvaluationItemListParams, BenchmarkStatus, BenchmarkSourceKind, BenchmarkRunStatus, BenchmarkScoring, BenchmarkDimensionScore, Benchmark, BenchmarkItem, BenchmarkRun, BenchmarkHistoryPoint, BenchmarkSource, BenchmarkCreateParams, BenchmarkListParams, BenchmarkItemListParams, BenchmarkHistoryParams, BenchmarkRetireParams, InferenceAlias, InferenceAliasRequest, TrainingRuleDeleteResponse, TrainingRunListResponse, TrainingRunResponse, TrainingRunActionResponse, EvaluationItemsResponse, InferenceAliasListResponse, InferenceAliasResponse, InferenceAliasDeleteResponse, BenchmarkListResponse, BenchmarkItemsResponse, BenchmarkHistoryResponse, TrainingRuleRunRefusalCode, TrainingRuleSaveRefusalCode, TrainingRuleMonthlyLimitRefusal, PipelineStatus, PipelineActiveRun, PipelineVersion, PipelineAttempt, PipelineAttemptTrigger, PipelineAttemptOutcome, Pipeline, PipelineListParams, PipelineListResponse, PipelineResponse, PipelineTrainWhen, PipelineSchedule, PipelineWriteRequest, PipelineMutationResponse, PipelineImportParams, PipelineFloors, LoopModel, LoopModelsResponse, LoopTracePipeline, LoopTracePipelinesResponse, LoopMetricsHistoryParams, LoopMetricPoint, LoopSignalDay, LoopMetricsHistory, PromotionPolicyState, RuleBenchmark, PipelineBenchmarksResponse, PipelineBenchmarkAttachParams, PipelineBenchmarkMoveParams, CompareBenchmark, VersionBenchmarkScore, PromotionMeasurement, PromotionOutcome, VersionScores, VersionScoreDifference, VersionCompareResponse, VersionCompareParams, } from './types.js';
79
79
  /** The closed set of run states a training run never leaves. */
80
80
  export { TERMINAL_RUN_STATES } from './types.js';
81
81
  /** The most versions a pipeline may be set to make. */
package/dist/index.js CHANGED
@@ -36,7 +36,7 @@ import { Loop } from './resources/loop.js';
36
36
  * SDK version. Sent as part of the User-Agent header.
37
37
  * Must match package.json "version" -- enforced by a contract test.
38
38
  */
39
- export const VERSION = '0.2.19-dev.277';
39
+ export const VERSION = '0.2.19-dev.278';
40
40
  export class RunBiOS {
41
41
  /** Search models, fetch configs, check adapter compatibility. */
42
42
  models;
package/dist/types.d.ts CHANGED
@@ -2611,6 +2611,11 @@ export interface LoopTrace {
2611
2611
  verdict?: LoopVerdict;
2612
2612
  /** Who gave that verdict. */
2613
2613
  verdict_source?: LoopSource;
2614
+ /**
2615
+ * The score a `scored` verdict carries: above zero it trains the answer as
2616
+ * it stands (good feedback), zero or below it trains nothing (bad).
2617
+ */
2618
+ verdict_score?: number;
2614
2619
  /** Its tags. On the single read only. */
2615
2620
  labels?: LoopLabel[];
2616
2621
  /**
@@ -2866,7 +2871,6 @@ export type TrainingRunState = 'built' | 'registering' | 'dataset_validating' |
2866
2871
  export declare const TERMINAL_RUN_STATES: readonly TrainingRunState[];
2867
2872
  export type TrainingVerdict = 'better' | 'not_better' | 'inconclusive' | 'not_evaluated';
2868
2873
  export type TrainingDecision = 'auto_promoted' | 'promoted' | 'rejected' | 'auto_rejected' | 'rolled_back';
2869
- export type NotifyState = 'none' | 'pending' | 'sending' | 'sent' | 'suppressed' | 'failed';
2870
2874
  export type EvaluationStatus = 'open' | 'done' | 'failed' | 'budget_stopped';
2871
2875
  export type EvaluationItemStatus = 'pending' | 'working' | 'done' | 'failed';
2872
2876
  export type EvaluationWinner = 'candidate' | 'incumbent' | 'tie';
@@ -3008,10 +3012,6 @@ export interface TrainingRun {
3008
3012
  decided_at: string | null;
3009
3013
  auto_promote_at: string | null;
3010
3014
  review_deadline_at: string | null;
3011
- notify_state: NotifyState;
3012
- notify_event: string | null;
3013
- notify_error: string | null;
3014
- notify_attempts: number;
3015
3015
  created_at: string;
3016
3016
  updated_at: string;
3017
3017
  finished_at: string | null;
@@ -3057,7 +3057,6 @@ export interface TrainingRunSummary {
3057
3057
  evaluation_id: string | null;
3058
3058
  review_deadline_at: string | null;
3059
3059
  auto_promote_at: string | null;
3060
- notify_state: NotifyState;
3061
3060
  created_at: string;
3062
3061
  finished_at: string | null;
3063
3062
  }
@@ -3515,8 +3514,8 @@ export type TrainingRuleSaveRefusalCode = 'PIPELINE_NAME_TAKEN' | 'OWN_TAG_TAKEN
3515
3514
  *
3516
3515
  * The test is the worst case, not the average: a run may start only if the
3517
3516
  * most the rule's runs this UTC calendar month can have cost
3518
- * (`month_spent_cents`), plus the most the run about to start may spend (its
3519
- * three ceilings added up), fits under the limit.
3517
+ * (`month_spent_cents`), plus the most the run about to start may spend
3518
+ * (`run_max_cents`), fits under the limit.
3520
3519
  */
3521
3520
  export interface TrainingRuleMonthlyLimitRefusal {
3522
3521
  error: {
@@ -3527,7 +3526,11 @@ export interface TrainingRuleMonthlyLimitRefusal {
3527
3526
  month_spent_cents: number;
3528
3527
  /** The rule's monthly limit. */
3529
3528
  monthly_ceiling_cents: number;
3530
- /** The most one run of this rule may spend: training + candidate + evaluation ceilings. */
3529
+ /**
3530
+ * The most one attempt of this rule may spend: its training, its comparison
3531
+ * machine and its judging, and -- on a pipeline, while no version is live --
3532
+ * the base model's own comparison machine beside the new model's.
3533
+ */
3531
3534
  run_max_cents: number;
3532
3535
  /**
3533
3536
  * The first instant of the next UTC month, when the counted spend starts
@@ -3820,10 +3823,11 @@ export interface Pipeline {
3820
3823
  * this UTC calendar month -- the figure the limit is enforced against. An
3821
3824
  * upper bound, not an exact spend. A run counts toward the month it was
3822
3825
  * created in, and a run still in progress also counts toward the current
3823
- * month, at its three ceilings or what it has been billed when that is
3824
- * more. A finished run counts what it was billed, and its comparison
3825
- * machine is billed at the most it could have cost (the hourly cap for the
3826
- * time it was up, never more than the candidate ceiling). A run created in
3826
+ * month, at the most it may cost (as `run_max_cents` counts it, the base
3827
+ * model's comparison machine included) or what it has been billed when that
3828
+ * is more. A finished run counts what it was billed, and its comparison
3829
+ * machines are billed at the most they could have cost (each machine's
3830
+ * hourly cap for the time it was up, never more than its own amount). A run created in
3827
3831
  * an earlier month that finishes in this one counts here only while it is
3828
3832
  * still running.
3829
3833
  */
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "runbios-sdk",
3
- "version": "0.2.19-dev.277",
3
+ "version": "0.2.19-dev.278",
4
4
  "description": "Official TypeScript SDK for the Run BiOS training and deployment platform API",
5
5
  "type": "module",
6
6
  "main": "./dist/index.js",