Implementation:Googleapis Python genai CreateTuningJobConfig Setup

Knowledge Sources	googleapis/python-genai Google Gen AI Python SDK
Domains	Fine_Tuning, Hyperparameter_Tuning
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete tool for specifying fine-tuning job hyperparameters and configuration provided by the google-genai types module.

Description

CreateTuningJobConfig is a Pydantic model that holds all optional configuration for a fine-tuning job. It specifies the tuning method (SUPERVISED_FINE_TUNING, DISTILLATION, PREFERENCE_TUNING), hyperparameters (epoch_count, learning_rate_multiplier), LoRA adapter size, tuned model display name, optional validation dataset, and description. It is passed as the config parameter to tunings.tune.

Usage

Create a CreateTuningJobConfig with the desired hyperparameters and pass it to tunings.tune. All fields are optional; the service applies sensible defaults. Set tuned_model_display_name for easy identification. Use method to switch between tuning approaches.

Code Reference

Source Location

• Repository: googleapis/python-genai
• File: google/genai/types.py
• Lines: L12173

Signature

class CreateTuningJobConfig(_common.BaseModel):
    """Fine-tuning job creation configuration."""
    http_options: Optional[HttpOptions] = None
    method: Optional[TuningMethod] = None
    validation_dataset: Optional[TuningValidationDataset] = None
    tuned_model_display_name: Optional[str] = None
    description: Optional[str] = None
    epoch_count: Optional[int] = None
    learning_rate_multiplier: Optional[float] = None
    adapter_size: Optional[AdapterSize] = None

Import

from google.genai import types

I/O Contract

Inputs

Name	Type	Required	Description
method	Optional[TuningMethod]	No	SUPERVISED_FINE_TUNING, DISTILLATION, or PREFERENCE_TUNING
epoch_count	Optional[int]	No	Number of training epochs
learning_rate_multiplier	Optional[float]	No	Learning rate multiplier
adapter_size	Optional[AdapterSize]	No	LoRA adapter size (rank)
tuned_model_display_name	Optional[str]	No	Display name for the tuned model
validation_dataset	Optional[TuningValidationDataset]	No	Validation dataset for evaluation
description	Optional[str]	No	Description of the tuning job

Outputs

Name	Type	Description
CreateTuningJobConfig	CreateTuningJobConfig	Configuration object to pass to tunings.tune

Usage Examples

Basic Configuration

from google.genai import types

config = types.CreateTuningJobConfig(
    epoch_count=5,
    learning_rate_multiplier=1.0,
    tuned_model_display_name="my-sentiment-classifier",
)

With LoRA Adapter Size

config = types.CreateTuningJobConfig(
    epoch_count=3,
    adapter_size="ADAPTER_SIZE_FOUR",
    tuned_model_display_name="my-lora-model",
    description="Sentiment classification fine-tune with LoRA rank 4",
)

Related Pages

Implements Principle

• Principle:Googleapis_Python_genai_Tuning_Job_Configuration