google-genai-kotlin/com.google.genai.kotlin.types/DistillationSpec

DistillationSpec

@Serializable
data class DistillationSpec(val promptDatasetUri: String? = null, val baseTeacherModel: String? = null, val hyperParameters: DistillationHyperParameters? = null, val pipelineRootDirectory: String? = null, val studentModel: String? = null, val trainingDatasetUri: String? = null, val tunedTeacherModelSource: String? = null, val validationDatasetUri: String? = null, val tuningMode: TuningMode? = null)

Distillation tuning spec for tuning.

Constructors

Link copied to clipboard
constructor(promptDatasetUri: String? = null, baseTeacherModel: String? = null, hyperParameters: DistillationHyperParameters? = null, pipelineRootDirectory: String? = null, studentModel: String? = null, trainingDatasetUri: String? = null, tunedTeacherModelSource: String? = null, validationDatasetUri: String? = null, tuningMode: TuningMode? = null)

Properties

Link copied to clipboard
val baseTeacherModel: String? = null

The base teacher model that is being distilled. See Supported models.

Link copied to clipboard
val hyperParameters: DistillationHyperParameters? = null

Optional. Hyperparameters for Distillation.

Link copied to clipboard
val pipelineRootDirectory: String? = null

Deprecated. A path in a Cloud Storage bucket, which will be treated as the root output directory of the distillation pipeline. It is used by the system to generate the paths of output artifacts.

Link copied to clipboard
val promptDatasetUri: String? = null

Optional. Cloud Storage path to file containing prompt dataset for distillation. The dataset must be formatted as a JSONL file.

Link copied to clipboard
val studentModel: String? = null

The student model that is being tuned, e.g., "google/gemma-2b-1.1-it". Deprecated. Use base_model instead.

Link copied to clipboard
val trainingDatasetUri: String? = null

Deprecated. Cloud Storage path to file containing training dataset for tuning. The dataset must be formatted as a JSONL file.

Link copied to clipboard
val tunedTeacherModelSource: String? = null

The resource name of the Tuned teacher model. Format: projects/{project}/locations/{location}/models/{model}.

Link copied to clipboard
val tuningMode: TuningMode? = null

Optional. Specifies the tuning mode for distillation (sft part). This feature is only available for open source models.

Link copied to clipboard
val validationDatasetUri: String? = null

Optional. Cloud Storage path to file containing validation dataset for tuning. The dataset must be formatted as a JSONL file.