Interface DistillationHyperParameters

Distillation hyperparameters for tuning.

interface DistillationHyperParameters {
    adapterSize?: AdapterSize;
    batchSize?: number;
    epochCount?: string;
    generationConfig?: GenerationConfig;
    learningRate?: number;
    learningRateMultiplier?: number;
}

Properties

adapterSize? batchSize? epochCount? generationConfig? learningRate? learningRateMultiplier?

Properties

OptionaladapterSize

adapterSize?: AdapterSize

The size of the adapter. Can be 'small', 'medium', or 'large'.

OptionalbatchSize

batchSize?: number

Batch size for tuning. This feature is only available for open source models.

OptionalepochCount

epochCount?: string

Number of complete passes the model makes over the entire training dataset during training.

OptionalgenerationConfig

generationConfig?: GenerationConfig

Generation config for Distillation teacher model sampling. Only the following fields are supported for distillation teacher samplings:

  • temperature
  • top_p
  • top_k
  • candidate_count
  • thinking_config

OptionallearningRate

learningRate?: number

The learning rate for distillation tuning.

OptionallearningRateMultiplier

learningRateMultiplier?: number

Multiplier for adjusting the default learning rate.

Settings

On This Page

Properties