llama-stack-client 0.3.4 → 0.4.0-rc1

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

@@ -1,516 +0,0 @@
-// 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,
-} 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('/v1/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('/v1/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 {
-    /**
-     * LoRA scaling parameter that controls adaptation strength
-     */
-    alpha: number;
-
-    /**
-     * Whether to apply LoRA to MLP layers
-     */
-    apply_lora_to_mlp: boolean;
-
-    /**
-     * Whether to apply LoRA to output projection layers
-     */
-    apply_lora_to_output: boolean;
-
-    /**
-     * List of attention module names to apply LoRA to
-     */
-    lora_attn_modules: Array<string>;
-
-    /**
-     * Rank of the LoRA adaptation (lower rank = fewer parameters)
-     */
-    rank: number;
-
-    /**
-     * Algorithm type identifier, always "LoRA"
-     */
-    type: 'LoRA';
-
-    /**
-     * (Optional) Whether to quantize the base model weights
-     */
-    quantize_base?: boolean;
-
-    /**
-     * (Optional) Whether to use DoRA (Weight-Decomposed Low-Rank Adaptation)
-     */
-    use_dora?: boolean;
-  }
-
-  /**
-   * Configuration for Quantization-Aware Training (QAT) fine-tuning.
-   */
-  export interface QatFinetuningConfig {
-    /**
-     * Size of groups for grouped quantization
-     */
-    group_size: number;
-
-    /**
-     * Name of the quantization algorithm to use
-     */
-    quantizer_name: string;
-
-    /**
-     * Algorithm type identifier, always "QAT"
-     */
-    type: 'QAT';
-  }
-}
-
-export interface ListPostTrainingJobsResponse {
-  data: Array<ListPostTrainingJobsResponse.Data>;
-}
-
-export namespace ListPostTrainingJobsResponse {
-  export interface Data {
-    job_uuid: string;
-  }
-}
-
-export interface PostTrainingJob {
-  job_uuid: string;
-}
-
-export interface PostTrainingPreferenceOptimizeParams {
-  /**
-   * The algorithm configuration.
-   */
-  algorithm_config: PostTrainingPreferenceOptimizeParams.AlgorithmConfig;
-
-  /**
-   * The model to fine-tune.
-   */
-  finetuned_model: string;
-
-  /**
-   * The hyperparam search configuration.
-   */
-  hyperparam_search_config: { [key: string]: boolean | number | string | Array<unknown> | unknown | null };
-
-  /**
-   * The UUID of the job to create.
-   */
-  job_uuid: string;
-
-  /**
-   * The logger configuration.
-   */
-  logger_config: { [key: string]: boolean | number | string | Array<unknown> | unknown | null };
-
-  /**
-   * The training configuration.
-   */
-  training_config: PostTrainingPreferenceOptimizeParams.TrainingConfig;
-}
-
-export namespace PostTrainingPreferenceOptimizeParams {
-  /**
-   * The algorithm configuration.
-   */
-  export interface AlgorithmConfig {
-    /**
-     * Temperature parameter for the DPO loss
-     */
-    beta: number;
-
-    /**
-     * The type of loss function to use for DPO
-     */
-    loss_type: 'sigmoid' | 'hinge' | 'ipo' | 'kto_pair';
-  }
-
-  /**
-   * The training configuration.
-   */
-  export interface TrainingConfig {
-    /**
-     * Number of steps to accumulate gradients before updating
-     */
-    gradient_accumulation_steps: number;
-
-    /**
-     * Maximum number of steps to run per epoch
-     */
-    max_steps_per_epoch: number;
-
-    /**
-     * Number of training epochs to run
-     */
-    n_epochs: number;
-
-    /**
-     * (Optional) Configuration for data loading and formatting
-     */
-    data_config?: TrainingConfig.DataConfig;
-
-    /**
-     * (Optional) Data type for model parameters (bf16, fp16, fp32)
-     */
-    dtype?: string;
-
-    /**
-     * (Optional) Configuration for memory and compute optimizations
-     */
-    efficiency_config?: TrainingConfig.EfficiencyConfig;
-
-    /**
-     * (Optional) Maximum number of validation steps per epoch
-     */
-    max_validation_steps?: number;
-
-    /**
-     * (Optional) Configuration for the optimization algorithm
-     */
-    optimizer_config?: TrainingConfig.OptimizerConfig;
-  }
-
-  export namespace TrainingConfig {
-    /**
-     * (Optional) Configuration for data loading and formatting
-     */
-    export interface DataConfig {
-      /**
-       * Number of samples per training batch
-       */
-      batch_size: number;
-
-      /**
-       * Format of the dataset (instruct or dialog)
-       */
-      data_format: 'instruct' | 'dialog';
-
-      /**
-       * Unique identifier for the training dataset
-       */
-      dataset_id: string;
-
-      /**
-       * Whether to shuffle the dataset during training
-       */
-      shuffle: boolean;
-
-      /**
-       * (Optional) Whether to pack multiple samples into a single sequence for
-       * efficiency
-       */
-      packed?: boolean;
-
-      /**
-       * (Optional) Whether to compute loss on input tokens as well as output tokens
-       */
-      train_on_input?: boolean;
-
-      /**
-       * (Optional) Unique identifier for the validation dataset
-       */
-      validation_dataset_id?: string;
-    }
-
-    /**
-     * (Optional) Configuration for memory and compute optimizations
-     */
-    export interface EfficiencyConfig {
-      /**
-       * (Optional) Whether to use activation checkpointing to reduce memory usage
-       */
-      enable_activation_checkpointing?: boolean;
-
-      /**
-       * (Optional) Whether to offload activations to CPU to save GPU memory
-       */
-      enable_activation_offloading?: boolean;
-
-      /**
-       * (Optional) Whether to offload FSDP parameters to CPU
-       */
-      fsdp_cpu_offload?: boolean;
-
-      /**
-       * (Optional) Whether to use memory-efficient FSDP wrapping
-       */
-      memory_efficient_fsdp_wrap?: boolean;
-    }
-
-    /**
-     * (Optional) Configuration for the optimization algorithm
-     */
-    export interface OptimizerConfig {
-      /**
-       * Learning rate for the optimizer
-       */
-      lr: number;
-
-      /**
-       * Number of steps for learning rate warmup
-       */
-      num_warmup_steps: number;
-
-      /**
-       * Type of optimizer to use (adam, adamw, or sgd)
-       */
-      optimizer_type: 'adam' | 'adamw' | 'sgd';
-
-      /**
-       * Weight decay coefficient for regularization
-       */
-      weight_decay: number;
-    }
-  }
-}
-
-export interface PostTrainingSupervisedFineTuneParams {
-  /**
-   * The hyperparam search configuration.
-   */
-  hyperparam_search_config: { [key: string]: boolean | number | string | Array<unknown> | unknown | null };
-
-  /**
-   * The UUID of the job to create.
-   */
-  job_uuid: string;
-
-  /**
-   * The logger configuration.
-   */
-  logger_config: { [key: string]: boolean | number | string | Array<unknown> | unknown | null };
-
-  /**
-   * The training configuration.
-   */
-  training_config: PostTrainingSupervisedFineTuneParams.TrainingConfig;
-
-  /**
-   * The algorithm configuration.
-   */
-  algorithm_config?: AlgorithmConfig;
-
-  /**
-   * The directory to save checkpoint(s) to.
-   */
-  checkpoint_dir?: string;
-
-  /**
-   * The model to fine-tune.
-   */
-  model?: string;
-}
-
-export namespace PostTrainingSupervisedFineTuneParams {
-  /**
-   * The training configuration.
-   */
-  export interface TrainingConfig {
-    /**
-     * Number of steps to accumulate gradients before updating
-     */
-    gradient_accumulation_steps: number;
-
-    /**
-     * Maximum number of steps to run per epoch
-     */
-    max_steps_per_epoch: number;
-
-    /**
-     * Number of training epochs to run
-     */
-    n_epochs: number;
-
-    /**
-     * (Optional) Configuration for data loading and formatting
-     */
-    data_config?: TrainingConfig.DataConfig;
-
-    /**
-     * (Optional) Data type for model parameters (bf16, fp16, fp32)
-     */
-    dtype?: string;
-
-    /**
-     * (Optional) Configuration for memory and compute optimizations
-     */
-    efficiency_config?: TrainingConfig.EfficiencyConfig;
-
-    /**
-     * (Optional) Maximum number of validation steps per epoch
-     */
-    max_validation_steps?: number;
-
-    /**
-     * (Optional) Configuration for the optimization algorithm
-     */
-    optimizer_config?: TrainingConfig.OptimizerConfig;
-  }
-
-  export namespace TrainingConfig {
-    /**
-     * (Optional) Configuration for data loading and formatting
-     */
-    export interface DataConfig {
-      /**
-       * Number of samples per training batch
-       */
-      batch_size: number;
-
-      /**
-       * Format of the dataset (instruct or dialog)
-       */
-      data_format: 'instruct' | 'dialog';
-
-      /**
-       * Unique identifier for the training dataset
-       */
-      dataset_id: string;
-
-      /**
-       * Whether to shuffle the dataset during training
-       */
-      shuffle: boolean;
-
-      /**
-       * (Optional) Whether to pack multiple samples into a single sequence for
-       * efficiency
-       */
-      packed?: boolean;
-
-      /**
-       * (Optional) Whether to compute loss on input tokens as well as output tokens
-       */
-      train_on_input?: boolean;
-
-      /**
-       * (Optional) Unique identifier for the validation dataset
-       */
-      validation_dataset_id?: string;
-    }
-
-    /**
-     * (Optional) Configuration for memory and compute optimizations
-     */
-    export interface EfficiencyConfig {
-      /**
-       * (Optional) Whether to use activation checkpointing to reduce memory usage
-       */
-      enable_activation_checkpointing?: boolean;
-
-      /**
-       * (Optional) Whether to offload activations to CPU to save GPU memory
-       */
-      enable_activation_offloading?: boolean;
-
-      /**
-       * (Optional) Whether to offload FSDP parameters to CPU
-       */
-      fsdp_cpu_offload?: boolean;
-
-      /**
-       * (Optional) Whether to use memory-efficient FSDP wrapping
-       */
-      memory_efficient_fsdp_wrap?: boolean;
-    }
-
-    /**
-     * (Optional) Configuration for the optimization algorithm
-     */
-    export interface OptimizerConfig {
-      /**
-       * Learning rate for the optimizer
-       */
-      lr: number;
-
-      /**
-       * Number of steps for learning rate warmup
-       */
-      num_warmup_steps: number;
-
-      /**
-       * Type of optimizer to use (adam, adamw, or sgd)
-       */
-      optimizer_type: 'adam' | 'adamw' | 'sgd';
-
-      /**
-       * Weight decay coefficient for regularization
-       */
-      weight_decay: number;
-    }
-  }
-}
-
-PostTraining.Job = Job;
-
-export declare namespace PostTraining {
-  export {
-    type AlgorithmConfig as AlgorithmConfig,
-    type ListPostTrainingJobsResponse as ListPostTrainingJobsResponse,
-    type PostTrainingJob as PostTrainingJob,
-    type PostTrainingPreferenceOptimizeParams as PostTrainingPreferenceOptimizeParams,
-    type PostTrainingSupervisedFineTuneParams as PostTrainingSupervisedFineTuneParams,
-  };
-
-  export {
-    Job as Job,
-    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/synthetic-data-generation.ts

@@ -1,65 +0,0 @@
-// 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 Shared from './shared';
-
-export class SyntheticDataGeneration extends APIResource {
-  /**
-   * Generate synthetic data based on input dialogs and apply filtering.
-   */
-  generate(
-    body: SyntheticDataGenerationGenerateParams,
-    options?: Core.RequestOptions,
-  ): Core.APIPromise<SyntheticDataGenerationResponse> {
-    return this._client.post('/v1/synthetic-data-generation/generate', { body, ...options });
-  }
-}
-
-/**
- * Response from the synthetic data generation. Batch of (prompt, response, score)
- * tuples that pass the threshold.
- */
-export interface SyntheticDataGenerationResponse {
-  /**
-   * List of generated synthetic data samples that passed the filtering criteria
-   */
-  synthetic_data: Array<{ [key: string]: boolean | number | string | Array<unknown> | unknown | null }>;
-
-  /**
-   * (Optional) Statistical information about the generation process and filtering
-   * results
-   */
-  statistics?: { [key: string]: boolean | number | string | Array<unknown> | unknown | null };
-}
-
-export interface SyntheticDataGenerationGenerateParams {
-  /**
-   * List of conversation messages to use as input for synthetic data generation
-   */
-  dialogs: Array<Shared.Message>;
-
-  /**
-   * Type of filtering to apply to generated synthetic data samples
-   */
-  filtering_function: 'none' | 'random' | 'top_k' | 'top_p' | 'top_k_top_p' | 'sigmoid';
-
-  /**
-   * (Optional) The identifier of the model to use. The model must be registered with
-   * Llama Stack and available via the /models endpoint
-   */
-  model?: string;
-}
-
-export declare namespace SyntheticDataGeneration {
-  export {
-    type SyntheticDataGenerationResponse as SyntheticDataGenerationResponse,
-    type SyntheticDataGenerationGenerateParams as SyntheticDataGenerationGenerateParams,
-  };
-}