examples/aflow/pubmedxqa/optimized/round_19/graph.py

import evoagentx.workflow.operators as operator
import examples.aflow.pubmedxqa.optimized.round_19.prompt as prompt_custom
from evoagentx.models.model_configs import LLMConfig
from evoagentx.benchmark.benchmark import Benchmark
from evoagentx.models.model_utils import create_llm_instance

class Workflow:
    
    def __init__(
        self,
        name: str,
        llm_config: LLMConfig,
        benchmark: Benchmark
    ):
        self.name = name
        self.llm = create_llm_instance(llm_config)
        self.benchmark = benchmark
        self.custom = operator.Custom(self.llm)
        self.answer_generate = operator.AnswerGenerate(self.llm)
        self.qas_ensemble = operator.QAScEnsemble(self.llm)  # Initialize QAScEnsemble operator
    
    async def __call__(self, problem: str):
        """
        Implementation of the workflow
        """
        # Generate multiple answers for the problem
        solutions = [await self.answer_generate(input=problem) for _ in range(10)]  # Generate 10 answers
        # Aggregate the answers using QAScEnsemble
        ensemble_response = await self.qas_ensemble(solutions=[sol['answer'] for sol in solutions])
        
        # Self-consistency check
        if ensemble_response['response'] in [sol['answer'] for sol in solutions]:
            # Use custom method to refine the final answer with additional context
            refined_response = await self.custom(input=problem + " Context: " + ensemble_response['response'], instruction="Refine the answer based on the provided context.")
        else:
            # If no consistency, return ensemble response and add a note for further review
            refined_response = {'response': ensemble_response['response'] + " Note: Further review may be needed."}  
        
        return refined_response['response']  # Return the refined answer