📘 Overview

ReasoningAnswerGenerator is a reasoning-based answer generation operator designed to automatically construct prompts and invoke large language models (LLMs) to produce reasoning-style answers based on input questions. This operator can be used in conjunction with various Prompt templates (Mathematical, General, DIY) and LLM service modules to automate the question–answer generation workflow.

__init__ Function

@prompt_restrict(
    MathAnswerGeneratorPrompt,
    GeneralAnswerGeneratorPrompt,
    DiyAnswerGeneratorPrompt
)
@OPERATOR_REGISTRY.register()
class ReasoningAnswerGenerator(OperatorABC):
    '''
    Answer Generator is a class that generates answers for given questions.
    '''
    def __init__(self,
                llm_serving: LLMServingABC,
                prompt_template = MathAnswerGeneratorPrompt | GeneralAnswerGeneratorPrompt | DiyAnswerGeneratorPrompt | DIYPromptABC
                ):

__init__ Parameters

Parameter Name	Type	Default Value	Description
llm_serving	LLMServingABC	Required	Instance of the LLM service used to perform reasoning and answer generation.
prompt_template	PromptABC	Supports MathAnswerGeneratorPrompt, GeneralAnswerGeneratorPrompt, DiyAnswerGeneratorPrompt, or user-defined DIYPromptABC	Prompt template object used to construct the input prompt. Supports mathematical, general, and custom templates.

Prompt Template Descriptions

Prompt Template Name	Primary Purpose	Applicable Scenario	Key Features
MathAnswerGeneratorPrompt	Generates reasoning and answers for mathematical problems	Math problem solving, formula derivation, theorem proofs, and calculation tasks	Built-in math-specific template emphasizing logical reasoning and formula representation, supporting step-by-step reasoning structures.
GeneralAnswerGeneratorPrompt	Generates general reasoning answers	Common reasoning, reading comprehension, general knowledge, and analytical questions	Highly versatile template supporting multi-domain logical reasoning and fluent natural language generation.
DiyAnswerGeneratorPrompt	Custom reasoning template defined by the user	When the user wants full control over prompt structure or output style	Allows user-defined prompts and generation rules for maximum flexibility and controllability; can be integrated with system parameters.

run Function

def run(storage, input_key="instruction", output_key="generated_cot")

Executes the main operator logic — reads input data from storage, generates reasoning-based answers using the LLM with the specified prompt, and writes the results back to storage.

Parameters

Name	Type	Default Value	Description
storage	DataFlowStorage	Required	DataFlow storage instance responsible for reading and writing data.
input_key	str	"instruction"	Name of the input column corresponding to the question field.
output_key	str	"generated_cot"	Name of the output column corresponding to the generated answer field.

🧠 Example Usage

from dataflow.operators.reasoning import ReasoningAnswerGenerator
from dataflow.utils.storage import FileStorage
from dataflow.serving import APILLMServing_request
from dataflow.core import LLMServingABC
from dataflow.prompts.reasoning.general import GeneralAnswerGeneratorPrompt

class ReasoningAnswerGeneratorTest():
    def __init__(self, llm_serving: LLMServingABC = None):
        
        self.storage = FileStorage(
            first_entry_file_name="test_step3.jsonl",
            cache_path="./cache_local",
            file_name_prefix="dataflow_cache_step",
            cache_type="jsonl",
        )

        # Use API server as LLM serving
        self.llm_serving = APILLMServing_request(
                    api_url="",
                    model_name="gpt-4o",
                    max_workers=30
        )
        
        self.answer_generator = ReasoningAnswerGenerator(
            llm_serving=self.llm_serving,
            prompt_template=GeneralAnswerGeneratorPrompt()
        )
        
    def forward(self):
        self.answer_generator.run(
            storage = self.storage.step(),
            input_key = "instruction", 
            output_key = "generated_cot"
        )

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

🧾 Default Output Format

Field	Type	Description
instruction	str	The input question text.
generated_cot	str	The reasoning-based answer generated by the model.

Example Input:

{
"instruction":"A triangle has sides of lengths 7, 24, and 25. Determine if it is a right triangle."
}

Example Output:

{
"instruction":"A triangle has sides of lengths 7, 24, and 25. Determine if it is a right triangle.",
"generated_cot":"Solution:\n1. Identify key components and premises of the task\n→ Sides of the triangle are 7, 24, and 25.\n\n2. Apply relevant principles, theorems, or methods with step-by-step derivation or argument\n→ Use the Pythagorean theorem for a right triangle: a^2 + b^2 = c^2.\n→ Assume 25 is the hypotenuse (largest side), then check: 7^2 + 24^2 = 25^2.\n\n3. Perform any necessary calculations or logical checks with intermediate verification\n→ Calculate 7^2: 7^2 = 49.\n→ Calculate 24^2: 24^2 = 576.\n→ Calculate 25^2: 25^2 = 625.\n→ Verify: 49 + 576 = 625.\n\n4. Present the final answer or conclusion in a clear, unambiguous notation\n→ Since 7^2 + 24^2 = 25^2 holds true, the triangle is a right triangle.\n→ The triangle is a right triangle: \\\\boxed{\\text{Yes}}."
}