인프런 커뮤니티 질문&답변

else:
  print("TOOL NAME : ", output.tool)
  print("TOOL INPUT : ", output.tool_input)

Traceback (most recent call last):
  File "/Users/chaejinjeong/Desktop/myGit/Study/langchain/inflearn/5.Agent/basic.py", line 75, in <module>
    print("TOOL NAME : ", output.tool)

output : [OpenAIToolAgentAction(tool='get_word_length', tool_input={'word': 'eudca'}, log="\nInvoking: `get_word_length` with `{'word': 'eudca'}`\n\n\n", message_log=[AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_DlTa5qRanu9HhiWGneMz2V3C', 'function': {'arguments': '{"word":"eudca"}', 'name': 'get_word_length'}, 'type': 'function'}]})], tool_call_id='call_DlTa5qRanu9HhiWGneMz2V3C')], <class 'list'>

• output.tool 이라는 속성이 없어서 print 해보니 AgentAction List 형태로 출력이 되었습니다.

intermediate_steps = []
final_result = dict()
while True:
    query.update({"intermediate_steps": intermediate_steps})
    output: Union[AgentFinish, AgentAction] = agent.invoke(query)

    if isinstance(output, AgentFinish):
        final_result = output.return_values
        final_result = final_result.get("output")
        break
    else:
        if isinstance(output, list):
            output = output[0]
        print("TOOL NAME : ", output.tool)
        print("TOOL INPUT : ", output.tool_input)
        _tool = tools.get(output.tool)
        observation = _tool.run(output.tool_input)
        intermediate_steps.append((output, observation))


TOOL NAME :  get_word_length
TOOL INPUT :  {'word': 'eudca'}

• 위처럼 인덱스 0으로 접근해서 처리해야할 것 같아 글 남깁니다.