📘 Overview

ConsistentChatGenerator is an operator designed to synthesize multi-turn dialogue data from scratch. It operates in a two-stage process, generating conversations based on a predefined set of topics and human intents, making it ideal for creating consistent and context-aware conversational datasets.

__init__ function

def __init__(self, 
             llm_serving: LLMServingABC = None, 
             num_dialogs_per_intent = 20, 
             num_turns_per_dialog = 6, 
             temperature = 0.9):

init Parameters

Parameter	Type	Default Value	Description
llm_serving	LLMServingABC	None	The Large Language Model serving instance used for generation.
num_dialogs_per_intent	int	20	The number of dialogs to generate for each predefined intent.
num_turns_per_dialog	int	6	The number of turns (user and assistant messages) in each dialog.
temperature	float	0.9	The sampling temperature to control the randomness of the generation.

Prompt Template Descriptions

Prompt Template Name	Main Purpose	Applicable Scenarios	Feature Description

run function

def run(self, storage: DataFlowStorage):

Parameters

Name	Type	Default Value	Description
storage	DataFlowStorage	Required	The data flow storage instance where the generated DataFrame will be written.

🧠 Example Usage

from dataflow.operators.conversations import ConsistentChatGenerator
from dataflow.utils.storage import FileStorage
from dataflow.serving import APILLMServing_request
from dataflow.core import LLMServingABC

class ConsistentChatGeneratorExample:
    def __init__(self, llm_serving: LLMServingABC = None):
        self.storage = FileStorage(
            first_entry_file_name="",
            cache_path="./cache_local",
            file_name_prefix="dataflow_cache_step",
            cache_type="jsonl",
        )

        self.llm_serving = APILLMServing_request(
            api_url="",
            model_name="gpt-4o",
            max_workers=30
        )

        self.generator = ConsistentChatGenerator(
            llm_serving=self.llm_serving,
            num_dialogs_per_intent=20,
            num_turns_per_dialog=6,
            temperature=0.9
        )

    def forward(self):
        self.generator.run(
            storage=self.storage.step()
        )

if __name__ == "__main__":
    pl = ConsistentChatGeneratorExample()
    pl.forward()

🧾 Default Output Format

Field	Type	Description
category	str	The category or intent of the generated dialogue.
conversation	list	A list of dictionaries representing the multi-turn dialogue.

Example Input:

This operator generates data from scratch and does not require an input file.

Example Output:

{
  "category": "Travel Planning",
  "conversation": [
    {
      "role": "user",
      "value": "I want to plan a 5-day trip to Paris. Can you help me with an itinerary?"
    },
    {
      "role": "assistant",
      "value": "Of course! For a 5-day trip, I'd suggest dedicating the first day to iconic landmarks like the Eiffel Tower and the Louvre Museum. Have you been to Paris before?"
    },
    {
      "role": "user",
      "value": "No, this is my first time. I'd also like to see some less touristy spots if possible."
    },
    {
      "role": "assistant",
      "value": "Great! For a more local experience, you could explore the Le Marais district for its unique shops and historic architecture. On day two, perhaps a visit to Montmartre and the Sacré-Cœur Basilica?"
    }
  ]
}