📘 Overview

The RandomDomainKnowledgeRowGenerator operator is used to batch generate domain knowledge-related text by combining a system prompt (system_prompt) and a user-defined prompt template (user_prompt). It supports generating a specified number of outputs according to the generation_num parameter when there is no input data. Note: The domains referenced for the randomly generated data are manually predefined.

__init__ function

def __init__(self, llm_serving: LLMServingABC, system_prompt: str = "You are a helpful agent.", user_prompt: str = "{}")

Parameter	Type	Default	Description
llm_serving	LLMServingABC	Required	The Large Language Model serving instance, which must implement the LLMServingABC interface.
system_prompt	str	"You are a helpful agent."	The system prompt that defines the model's behavior.
user_prompt	str	"{}"	The user prompt template. It can flexibly insert input content or an entire row dictionary via the {} placeholder.

Prompt Template Descriptions

Prompt Template Name	Main Purpose	Applicable Scenarios	Feature Description

run function

def run(self, storage: DataFlowStorage, output_key: str = "generated_content", generation_num: int = 1)

Parameter	Type	Default	Description
storage	DataFlowStorage	Required	The DataFlow storage instance for reading and writing data.
output_key	str	"generated_content"	The name of the output column for the generated content.
generation_num	int	1	The number of content items to generate when there is no input data.

🧠 Example Usage

🧾 Default Output Format

Field	Type	Description
generated_content	str	The generated text content.