DataGenerationConfig

Generate synthetic data using a model for finetuning an LLM.

KEY TYPE Description
generationInstructions str Instructions for the data generation model.
tokenBudget int Token budget for generation.
descriptionCol str Name of the description column.
frequencyPenalty float Penalty for frequency of token appearance.
idCol str Name of the identifier column.
fewshotExamples int Number of fewshot examples used to prompt the model.
oversample bool Whether to oversample the data.
model str Model to use for data generation.
documentationCharLimit int Character limit for documentation.
seed Optional[int] Seed for random number generation.
promptCol str Name of the input prompt column.
verifyResponse bool Whether to verify the response.
completionCol str Name of the output completion column.
temperature float Sampling temperature for the model.
concurrency int Number of concurrent processes.
examplesPerTarget int Number of examples per target.
subsetSize Optional[int] Size of the subset to use for generation.