Skip to content

🐝 The First Self-Improving agents with RL / Prompting Optimization

2.3 KB
Raw Blame
#!/usr/bin/env python
# -*- coding: utf-8 -*-

from copy import deepcopy
from collections import defaultdict
from typing import List, Any, Optional

from swarm.llm.format import Message
from swarm.graph import Node
from swarm.memory.memory import GlobalMemory
from swarm.utils.log import logger, swarmlog
from swarm.utils.globals import Cost
from swarm.environment.prompt.prompt_set_registry import PromptSetRegistry
from swarm.llm import LLMRegistry


class AdversarialAnswer(Node):
    def __init__(self, 
                 domain: str,
                 model_name: Optional[str] = None,
                 operation_description: str = "Directly output an answer.",
                 id=None):
        super().__init__(operation_description, id, True)
        self.domain = domain
        self.llm = LLMRegistry.get(model_name)
        self.prompt_set = PromptSetRegistry.get(domain)
        self.role = self.prompt_set.get_role()
        self.constraint = self.prompt_set.get_constraint()

    @property
    def node_name(self):
        return self.__class__.__name__

    def meta_prompt(self, input, meta_init=False):

        task = input["task"]
        self.prompt_set = PromptSetRegistry.get(self.domain)
        role = self.prompt_set.get_role()
        constraint = self.prompt_set.get_constraint()
        prompt = self.prompt_set.get_adversarial_answer_prompt(question=task)    

        if meta_init:
            pass #TODO

        return role, constraint, prompt


    async def _execute(self, inputs: List[Any] = [], **kwargs):
        
        node_inputs = self.process_input(inputs)
        inputs = []
        for input in node_inputs:
            role, constraint, prompt= self.meta_prompt(input, meta_init=False)
            message = [Message(role="system", content=f"You are a {role}. {constraint}"),
                    Message(role="user", content=prompt)]
            response = await self.llm.agen(message)

            _memory = {
                "operation": self.node_name,
                #"task_id": input["task_id"], 
                "task": input["task"],
                "files": input.get("files", []),
                "input": input["task"],
                "subtask": prompt,
                "output": response,
                "format": "natural language"
            }

            # self.log()
            inputs.append(_memory)
        return inputs