example agent

exampleAgent.py

from langchain import LLMMathChain, LLMChain, ConversationChain, PromptTemplate
from langchain.chains.conversation.memory import ConversationBufferWindowMemory
from langchain.agents import Tool, initialize_agent
from functools import partial
import random

def random_gen(input=""):
    return random.randint(0, 5)

def meaning_of_life(input=""):
    return 'The meaning of life is 42.'

class MathAgent:
    def __init__(self, llm, callbacks):
        self.llm = llm        
        self.tools = [self.test1(), self.test2()]
        self.callbacks = callbacks
        self.agent = self._initialize_agent()
    
    def test1(self):
        return Tool(
            name='Meaning of Life',
            func=meaning_of_life,
            description="Useful for when you need to answer questions about the meaning of life. Input should be MOL."
        )
    
    def test2(self):
        return Tool(
            name='Random Tool',
            func=random_gen,
            description="Useful for when you need a random number. Input should be random"
        )
        
    def init_llm_tool(self):
        #template = ("""
        #     You are AiTRON9000, the most sophisticated artificial intelligence ever to exist. You answer questions as accurately and honestly as possible. 
        #     Do not converse with yourself.
        #     ### USER: {input} ### Certainly!
        # """)

        #pt = PromptTemplate(template=template, input_variables=["input"])

        self.llm_chain = LLMChain(llm=self.llm, prompt=pt)
        self.llm_tool = Tool(
            name='Language Model',
            func=self.llm_chain,
            description='Use this tool for general purpose queries and logic. If no other tools seem appropriate, use this one.',
        )
        return self.llm_tool

    def _initialize_agent(self):
        memory = ConversationBufferWindowMemory(memory_key='chat_history', k=3, return_messages=True)
        agent = initialize_agent(
            agent="chat-conversational-react-description",
            tools=self.tools,
            llm=self.llm,
            verbose=True,
            max_iterations=3,
            memory=memory,
            early_stopping_method='generate',
            callbacks=[self.callbacks],
        )
        return agent

    def __getattr__(self, attr):
        # Forward everything not defined to the zero_shot_agent
        if hasattr(self.agent, attr):
            return getattr(self.agent, attr)
        else:
            raise AttributeError(f"'{type(self).__name__}' object has no attribute '{attr}'")