llama-stack-client 0.4.3 → 0.4.5

package/dist/src/resources/alpha/inference.ts

@@ -0,0 +1,183 @@
+// Copyright (c) Meta Platforms, Inc. and affiliates.
+// All rights reserved.
+//
+// This source code is licensed under the terms described in the LICENSE file in
+// the root directory of this source tree.
+//
+// File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+import { APIResource } from '../../resource';
+import * as Core from '../../core';
+
+export class Inference extends APIResource {
+  /**
+   * Rerank a list of documents based on their relevance to a query.
+   */
+  rerank(
+    body: InferenceRerankParams,
+    options?: Core.RequestOptions,
+  ): Core.APIPromise<InferenceRerankResponse> {
+    return (
+      this._client.post('/v1alpha/inference/rerank', { body, ...options }) as Core.APIPromise<{
+        data: InferenceRerankResponse;
+      }>
+    )._thenUnwrap((obj) => obj.data);
+  }
+}
+
+/**
+ * List of rerank result objects, sorted by relevance score (descending).
+ */
+export type InferenceRerankResponse = Array<InferenceRerankResponse.InferenceRerankResponseItem>;
+
+export namespace InferenceRerankResponse {
+  /**
+   * A single rerank result from a reranking response.
+   */
+  export interface InferenceRerankResponseItem {
+    /**
+     * The original index of the document in the input list.
+     */
+    index: number;
+
+    /**
+     * The relevance score from the model output. Higher scores indicate greater
+     * relevance.
+     */
+    relevance_score: number;
+  }
+}
+
+export interface InferenceRerankParams {
+  /**
+   * List of items to rerank. Each item can be a string, text content part, or image
+   * content part.
+   */
+  items: Array<
+    | string
+    | InferenceRerankParams.OpenAIChatCompletionContentPartTextParam
+    | InferenceRerankParams.OpenAIChatCompletionContentPartImageParam
+  >;
+
+  /**
+   * The identifier of the reranking model to use.
+   */
+  model: string;
+
+  /**
+   * The search query to rank items against. Can be a string, text content part, or
+   * image content part.
+   */
+  query:
+    | string
+    | InferenceRerankParams.OpenAIChatCompletionContentPartTextParam
+    | InferenceRerankParams.OpenAIChatCompletionContentPartImageParam;
+
+  /**
+   * Maximum number of results to return. Default: returns all.
+   */
+  max_num_results?: number | null;
+}
+
+export namespace InferenceRerankParams {
+  /**
+   * Text content part for OpenAI-compatible chat completion messages.
+   */
+  export interface OpenAIChatCompletionContentPartTextParam {
+    /**
+     * The text content of the message.
+     */
+    text: string;
+
+    /**
+     * Must be 'text' to identify this as text content.
+     */
+    type?: 'text';
+  }
+
+  /**
+   * Image content part for OpenAI-compatible chat completion messages.
+   */
+  export interface OpenAIChatCompletionContentPartImageParam {
+    /**
+     * Image URL specification and processing details.
+     */
+    image_url: OpenAIChatCompletionContentPartImageParam.ImageURL;
+
+    /**
+     * Must be 'image_url' to identify this as image content.
+     */
+    type?: 'image_url';
+  }
+
+  export namespace OpenAIChatCompletionContentPartImageParam {
+    /**
+     * Image URL specification and processing details.
+     */
+    export interface ImageURL {
+      /**
+       * URL of the image to include in the message.
+       */
+      url: string;
+
+      /**
+       * Level of detail for image processing. Can be 'low', 'high', or 'auto'.
+       */
+      detail?: 'low' | 'high' | 'auto' | null;
+    }
+  }
+
+  /**
+   * Text content part for OpenAI-compatible chat completion messages.
+   */
+  export interface OpenAIChatCompletionContentPartTextParam {
+    /**
+     * The text content of the message.
+     */
+    text: string;
+
+    /**
+     * Must be 'text' to identify this as text content.
+     */
+    type?: 'text';
+  }
+
+  /**
+   * Image content part for OpenAI-compatible chat completion messages.
+   */
+  export interface OpenAIChatCompletionContentPartImageParam {
+    /**
+     * Image URL specification and processing details.
+     */
+    image_url: OpenAIChatCompletionContentPartImageParam.ImageURL;
+
+    /**
+     * Must be 'image_url' to identify this as image content.
+     */
+    type?: 'image_url';
+  }
+
+  export namespace OpenAIChatCompletionContentPartImageParam {
+    /**
+     * Image URL specification and processing details.
+     */
+    export interface ImageURL {
+      /**
+       * URL of the image to include in the message.
+       */
+      url: string;
+
+      /**
+       * Level of detail for image processing. Can be 'low', 'high', or 'auto'.
+       */
+      detail?: 'low' | 'high' | 'auto' | null;
+    }
+  }
+}
+
+export declare namespace Inference {
+  export {
+    type InferenceRerankResponse as InferenceRerankResponse,
+    type InferenceRerankParams as InferenceRerankParams,
+  };
+}

package/dist/src/resources/alpha/post-training/index.ts

@@ -0,0 +1,17 @@
+// Copyright (c) Meta Platforms, Inc. and affiliates.
+// All rights reserved.
+//
+// This source code is licensed under the terms described in the LICENSE file in
+// the root directory of this source tree.
+//
+// File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+export { Job, type JobListResponse, type JobArtifactsResponse, type JobStatusResponse } from './job';
+export {
+  PostTraining,
+  type AlgorithmConfig,
+  type ListPostTrainingJobsResponse,
+  type PostTrainingJob,
+  type PostTrainingPreferenceOptimizeParams,
+  type PostTrainingSupervisedFineTuneParams,
+} from './post-training';

package/dist/src/resources/alpha/post-training/job.ts

@@ -0,0 +1,161 @@
+// Copyright (c) Meta Platforms, Inc. and affiliates.
+// All rights reserved.
+//
+// This source code is licensed under the terms described in the LICENSE file in
+// the root directory of this source tree.
+//
+// File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+import { APIResource } from '../../../resource';
+import * as Core from '../../../core';
+import * as PostTrainingAPI from './post-training';
+
+export class Job extends APIResource {
+  /**
+   * Get all training jobs.
+   */
+  list(options?: Core.RequestOptions): Core.APIPromise<JobListResponse> {
+    return (
+      this._client.get('/v1alpha/post-training/jobs', options) as Core.APIPromise<{ data: JobListResponse }>
+    )._thenUnwrap((obj) => obj.data);
+  }
+
+  /**
+   * Get the artifacts of a training job.
+   */
+  artifacts(options?: Core.RequestOptions): Core.APIPromise<JobArtifactsResponse> {
+    return this._client.get('/v1alpha/post-training/job/artifacts', options);
+  }
+
+  /**
+   * Cancel a training job.
+   */
+  cancel(options?: Core.RequestOptions): Core.APIPromise<void> {
+    return this._client.post('/v1alpha/post-training/job/cancel', {
+      ...options,
+      headers: { Accept: '*/*', ...options?.headers },
+    });
+  }
+
+  /**
+   * Get the status of a training job.
+   */
+  status(options?: Core.RequestOptions): Core.APIPromise<JobStatusResponse> {
+    return this._client.get('/v1alpha/post-training/job/status', options);
+  }
+}
+
+export type JobListResponse = Array<PostTrainingAPI.PostTrainingJob>;
+
+/**
+ * Artifacts of a finetuning job.
+ */
+export interface JobArtifactsResponse {
+  job_uuid: string;
+
+  checkpoints?: Array<JobArtifactsResponse.Checkpoint>;
+}
+
+export namespace JobArtifactsResponse {
+  /**
+   * Checkpoint created during training runs.
+   */
+  export interface Checkpoint {
+    created_at: string;
+
+    epoch: number;
+
+    identifier: string;
+
+    path: string;
+
+    post_training_job_id: string;
+
+    /**
+     * Training metrics captured during post-training jobs.
+     */
+    training_metrics?: Checkpoint.TrainingMetrics | null;
+  }
+
+  export namespace Checkpoint {
+    /**
+     * Training metrics captured during post-training jobs.
+     */
+    export interface TrainingMetrics {
+      epoch: number;
+
+      perplexity: number;
+
+      train_loss: number;
+
+      validation_loss: number;
+    }
+  }
+}
+
+/**
+ * Status of a finetuning job.
+ */
+export interface JobStatusResponse {
+  job_uuid: string;
+
+  /**
+   * Status of a job execution.
+   */
+  status: 'completed' | 'in_progress' | 'failed' | 'scheduled' | 'cancelled';
+
+  checkpoints?: Array<JobStatusResponse.Checkpoint>;
+
+  completed_at?: string | null;
+
+  resources_allocated?: { [key: string]: unknown } | null;
+
+  scheduled_at?: string | null;
+
+  started_at?: string | null;
+}
+
+export namespace JobStatusResponse {
+  /**
+   * Checkpoint created during training runs.
+   */
+  export interface Checkpoint {
+    created_at: string;
+
+    epoch: number;
+
+    identifier: string;
+
+    path: string;
+
+    post_training_job_id: string;
+
+    /**
+     * Training metrics captured during post-training jobs.
+     */
+    training_metrics?: Checkpoint.TrainingMetrics | null;
+  }
+
+  export namespace Checkpoint {
+    /**
+     * Training metrics captured during post-training jobs.
+     */
+    export interface TrainingMetrics {
+      epoch: number;
+
+      perplexity: number;
+
+      train_loss: number;
+
+      validation_loss: number;
+    }
+  }
+}
+
+export declare namespace Job {
+  export {
+    type JobListResponse as JobListResponse,
+    type JobArtifactsResponse as JobArtifactsResponse,
+    type JobStatusResponse as JobStatusResponse,
+  };
+}