nemo_microservices.types.customization.customization_config_job_value#

Module Contents#

Classes#

API#

class nemo_microservices.types.customization.customization_config_job_value.CustomizationConfigJobValue(/, **data: typing.Any)#

Bases: nemo_microservices._models.BaseModel

base_model: str#

None

The base model that will be customized.

dataset_schema: Optional[Dict[str, object]]#

None

Description of the expected format of the dataset

max_seq_length: int#

None

precision: nemo_microservices.types.shared.model_precision.ModelPrecision#

None

Type of model precision.

Values

  • "int8" - 8-bit integer precision

  • "bf16" - Brain floating point precision

  • "fp16" - 16-bit floating point precision

  • "fp32" - 32-bit floating point precision

  • "fp8-mixed" - Mixed 8-bit floating point precision available on Hopper and later architectures.

  • "bf16-mixed" - Mixed Brain floating point precision

prompt_template: Optional[str]#

None

Prompt template used to extract keys from the dataset.

E.g. prompt_template=’{input} {output}’, and sample looks like ‘{“input”: “Q: 2x2 A:”, “output”: “4”}’ then the model sees ‘Q: 2x2 A: 4’

training_option: nemo_microservices.types.customization_training_option.CustomizationTrainingOption#

None

Resource configuration for model training.

Specifies the hardware and parallelization settings for training.