Skip to main content

PseudoLabelJobConfig

Configuration for LLM Judge labeling job.

    name string

    Optional name for the job.

    description string

    Optional description for the job.

    datasetId stringrequired

    ID of the main dataset to be pseudo-labeled

    fewShotDatasetId string

    ID of the dataset containing few-shot examples. Optional.

    activeLabeledDatasetId string

    ID of the actively labeled dataset. Optional. If null, this job is for active learning.

    selectedColumns string[]required

    chatCompletionConfig

    object

    required

    For Chat LLM based labeling, the configuration to use with the requests (messages omitted)

    vendor stringrequired
    model stringrequired

    The ID of the model to use for the completion.

    messages

    object[]

    required

  • Array [

    • role stringrequired

    Role can be 'system', 'user', or 'assistant'.

    content stringrequired

    The content of the message.

  • ]

    • maxTokens integerrequired

    The maximum number of tokens to generate.

    temperature numberrequired

    The temperature to use for the completion.

    topP numberrequired

    The top_p value to use for the completion.

    promptTemplate

    object

    required

    id stringrequired
    template stringrequired

    The template string that defines the prompt

    baseEvaluationMetric stringrequired

    Reserved field. Do not use at the moment.

    skipActiveLabeling booleanrequired

    If true, skip active labeling, which involves an intermediate Dataset created for human labeling.