llama-stack-client 0.3.4-rc3 → 0.4.0-rc1

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

@@ -0,0 +1,369 @@
+// 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 JobAPI from './job';
+import {
+  Job,
+  JobArtifactsParams,
+  JobArtifactsResponse,
+  JobCancelParams,
+  JobListResponse,
+  JobStatusParams,
+  JobStatusResponse,
+  ListPostTrainingJobsResponse,
+} from './job';
+
+export class PostTraining extends APIResource {
+  job: JobAPI.Job = new JobAPI.Job(this._client);
+
+  /**
+   * Run preference optimization of a model.
+   */
+  preferenceOptimize(
+    body: PostTrainingPreferenceOptimizeParams,
+    options?: Core.RequestOptions,
+  ): Core.APIPromise<PostTrainingJob> {
+    return this._client.post('/v1alpha/post-training/preference-optimize', { body, ...options });
+  }
+
+  /**
+   * Run supervised fine-tuning of a model.
+   */
+  supervisedFineTune(
+    body: PostTrainingSupervisedFineTuneParams,
+    options?: Core.RequestOptions,
+  ): Core.APIPromise<PostTrainingJob> {
+    return this._client.post('/v1alpha/post-training/supervised-fine-tune', { body, ...options });
+  }
+}
+
+/**
+ * Configuration for Low-Rank Adaptation (LoRA) fine-tuning.
+ */
+export type AlgorithmConfig = AlgorithmConfig.LoraFinetuningConfig | AlgorithmConfig.QatFinetuningConfig;
+
+export namespace AlgorithmConfig {
+  /**
+   * Configuration for Low-Rank Adaptation (LoRA) fine-tuning.
+   */
+  export interface LoraFinetuningConfig {
+    alpha: number;
+
+    apply_lora_to_mlp: boolean;
+
+    apply_lora_to_output: boolean;
+
+    lora_attn_modules: Array<string>;
+
+    rank: number;
+
+    quantize_base?: boolean | null;
+
+    type?: 'LoRA';
+
+    use_dora?: boolean | null;
+  }
+
+  /**
+   * Configuration for Quantization-Aware Training (QAT) fine-tuning.
+   */
+  export interface QatFinetuningConfig {
+    group_size: number;
+
+    quantizer_name: string;
+
+    type?: 'QAT';
+  }
+}
+
+export interface PostTrainingJob {
+  job_uuid: string;
+}
+
+export interface PostTrainingPreferenceOptimizeParams {
+  /**
+   * Configuration for Direct Preference Optimization (DPO) alignment.
+   */
+  algorithm_config: PostTrainingPreferenceOptimizeParams.AlgorithmConfig;
+
+  finetuned_model: string;
+
+  hyperparam_search_config: { [key: string]: unknown };
+
+  job_uuid: string;
+
+  logger_config: { [key: string]: unknown };
+
+  /**
+   * Comprehensive configuration for the training process.
+   */
+  training_config: PostTrainingPreferenceOptimizeParams.TrainingConfig;
+}
+
+export namespace PostTrainingPreferenceOptimizeParams {
+  /**
+   * Configuration for Direct Preference Optimization (DPO) alignment.
+   */
+  export interface AlgorithmConfig {
+    beta: number;
+
+    loss_type?: 'sigmoid' | 'hinge' | 'ipo' | 'kto_pair';
+  }
+
+  /**
+   * Comprehensive configuration for the training process.
+   */
+  export interface TrainingConfig {
+    n_epochs: number;
+
+    /**
+     * Configuration for training data and data loading.
+     */
+    data_config?: TrainingConfig.DataConfig | null;
+
+    dtype?: string | null;
+
+    /**
+     * Configuration for memory and compute efficiency optimizations.
+     */
+    efficiency_config?: TrainingConfig.EfficiencyConfig | null;
+
+    gradient_accumulation_steps?: number;
+
+    max_steps_per_epoch?: number;
+
+    max_validation_steps?: number | null;
+
+    /**
+     * Configuration parameters for the optimization algorithm.
+     */
+    optimizer_config?: TrainingConfig.OptimizerConfig | null;
+  }
+
+  export namespace TrainingConfig {
+    /**
+     * Configuration for training data and data loading.
+     */
+    export interface DataConfig {
+      batch_size: number;
+
+      /**
+       * Format of the training dataset.
+       */
+      data_format: 'instruct' | 'dialog';
+
+      dataset_id: string;
+
+      shuffle: boolean;
+
+      packed?: boolean | null;
+
+      train_on_input?: boolean | null;
+
+      validation_dataset_id?: string | null;
+    }
+
+    /**
+     * Configuration for memory and compute efficiency optimizations.
+     */
+    export interface EfficiencyConfig {
+      enable_activation_checkpointing?: boolean | null;
+
+      enable_activation_offloading?: boolean | null;
+
+      fsdp_cpu_offload?: boolean | null;
+
+      memory_efficient_fsdp_wrap?: boolean | null;
+    }
+
+    /**
+     * Configuration parameters for the optimization algorithm.
+     */
+    export interface OptimizerConfig {
+      lr: number;
+
+      num_warmup_steps: number;
+
+      /**
+       * Available optimizer algorithms for training.
+       */
+      optimizer_type: 'adam' | 'adamw' | 'sgd';
+
+      weight_decay: number;
+    }
+  }
+}
+
+export interface PostTrainingSupervisedFineTuneParams {
+  hyperparam_search_config: { [key: string]: unknown };
+
+  job_uuid: string;
+
+  logger_config: { [key: string]: unknown };
+
+  /**
+   * Comprehensive configuration for the training process.
+   */
+  training_config: PostTrainingSupervisedFineTuneParams.TrainingConfig;
+
+  /**
+   * Configuration for Low-Rank Adaptation (LoRA) fine-tuning.
+   */
+  algorithm_config?:
+    | PostTrainingSupervisedFineTuneParams.LoraFinetuningConfig
+    | PostTrainingSupervisedFineTuneParams.QatFinetuningConfig
+    | null;
+
+  checkpoint_dir?: string | null;
+
+  /**
+   * Model descriptor for training if not in provider config`
+   */
+  model?: string | null;
+}
+
+export namespace PostTrainingSupervisedFineTuneParams {
+  /**
+   * Comprehensive configuration for the training process.
+   */
+  export interface TrainingConfig {
+    n_epochs: number;
+
+    /**
+     * Configuration for training data and data loading.
+     */
+    data_config?: TrainingConfig.DataConfig | null;
+
+    dtype?: string | null;
+
+    /**
+     * Configuration for memory and compute efficiency optimizations.
+     */
+    efficiency_config?: TrainingConfig.EfficiencyConfig | null;
+
+    gradient_accumulation_steps?: number;
+
+    max_steps_per_epoch?: number;
+
+    max_validation_steps?: number | null;
+
+    /**
+     * Configuration parameters for the optimization algorithm.
+     */
+    optimizer_config?: TrainingConfig.OptimizerConfig | null;
+  }
+
+  export namespace TrainingConfig {
+    /**
+     * Configuration for training data and data loading.
+     */
+    export interface DataConfig {
+      batch_size: number;
+
+      /**
+       * Format of the training dataset.
+       */
+      data_format: 'instruct' | 'dialog';
+
+      dataset_id: string;
+
+      shuffle: boolean;
+
+      packed?: boolean | null;
+
+      train_on_input?: boolean | null;
+
+      validation_dataset_id?: string | null;
+    }
+
+    /**
+     * Configuration for memory and compute efficiency optimizations.
+     */
+    export interface EfficiencyConfig {
+      enable_activation_checkpointing?: boolean | null;
+
+      enable_activation_offloading?: boolean | null;
+
+      fsdp_cpu_offload?: boolean | null;
+
+      memory_efficient_fsdp_wrap?: boolean | null;
+    }
+
+    /**
+     * Configuration parameters for the optimization algorithm.
+     */
+    export interface OptimizerConfig {
+      lr: number;
+
+      num_warmup_steps: number;
+
+      /**
+       * Available optimizer algorithms for training.
+       */
+      optimizer_type: 'adam' | 'adamw' | 'sgd';
+
+      weight_decay: number;
+    }
+  }
+
+  /**
+   * Configuration for Low-Rank Adaptation (LoRA) fine-tuning.
+   */
+  export interface LoraFinetuningConfig {
+    alpha: number;
+
+    apply_lora_to_mlp: boolean;
+
+    apply_lora_to_output: boolean;
+
+    lora_attn_modules: Array<string>;
+
+    rank: number;
+
+    quantize_base?: boolean | null;
+
+    type?: 'LoRA';
+
+    use_dora?: boolean | null;
+  }
+
+  /**
+   * Configuration for Quantization-Aware Training (QAT) fine-tuning.
+   */
+  export interface QatFinetuningConfig {
+    group_size: number;
+
+    quantizer_name: string;
+
+    type?: 'QAT';
+  }
+}
+
+PostTraining.Job = Job;
+
+export declare namespace PostTraining {
+  export {
+    type AlgorithmConfig as AlgorithmConfig,
+    type PostTrainingJob as PostTrainingJob,
+    type PostTrainingPreferenceOptimizeParams as PostTrainingPreferenceOptimizeParams,
+    type PostTrainingSupervisedFineTuneParams as PostTrainingSupervisedFineTuneParams,
+  };
+
+  export {
+    Job as Job,
+    type ListPostTrainingJobsResponse as ListPostTrainingJobsResponse,
+    type JobListResponse as JobListResponse,
+    type JobArtifactsResponse as JobArtifactsResponse,
+    type JobStatusResponse as JobStatusResponse,
+    type JobArtifactsParams as JobArtifactsParams,
+    type JobCancelParams as JobCancelParams,
+    type JobStatusParams as JobStatusParams,
+  };
+}

package/src/resources/{post-training.ts → alpha/post-training.ts}

@@ -3,7 +3,7 @@
 //
 // 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 * from './post-training/index';

package/src/resources/alpha.ts

@@ -0,0 +1,9 @@
+// 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 * from './alpha/index';