Agent 주요기능 | 다른 LLM 호출해보기

Ruah·2025년 1월 30일

Agent

목록 보기

6/11

OpenAI 외의 도구 호출 Agent 만들기(Tool Calling Agent with others)

OpenAI 외에도 Langchain은 Anthropic, Google Gemini, Mistral, Together.ai, Ollama과 같은 여러 공급자 구현을 지원한다.

여기서는 다양한 LLM으로 도구 호출 에이전트를 생성하고 실행하는 방법을 살펴보자.

Source : https://python.langchain.com/v0.1/docs/modules/agents/agent_types/tool_calling/

from dotenv import load_dotenv

load_dotenv()

from langchain_teddynote import logging
logging.langsmith("agent-other-llm")

LangSmith 추적을 시작합니다.
[프로젝트명]
agent-other-llm

from langchain.tools import tool 
from typing import List, Dict
from langchain_teddynote.tools import GoogleNews

# Define tool
@tool
def search_google_news(query: str)-> List[Dict[str, str]] :
    """Search google news by input keyword"""
    news_tool = GoogleNews()
    return news_tool.search_by_keyword(query, k=5)


print(f"도구 이름 : {search_google_news.name}")
print(f"도구 설명 : {search_google_news.description}")

도구 이름 : search_google_news
도구 설명 : Search google news by input keyword

# Define tool
tools = [search_google_news]

Agent 용 프롬프트 생성

chat_history : 이전 대화 내용을 저장하는 변수 (멀티턴을 지원하지 않는다면, 생략 가능합니다.)
agent_scratchpad : 에이전트가 임시로 저장하는 변수
input : 사용자의 입력

from langchain_core.prompts import ChatPromptTemplate
from langchain.agents import create_tool_calling_agent

# create prompt
# 프롬프트는 에이전트에게 모델이 수행할 작업을 설명하는 텍스트를 제공한다 (도구의 이름과 역할을 입력)
prompt = ChatPromptTemplate.from_messages([
    (
        "system",
        "You are a helpful assistant that can search google news by input keyword"
     ),
     ("placeholder", "{chat_history}"),
     ("human", "{input}"),
     ("placeholder", "{agent_scratchpad}"),
])

Tool Calling 을 지원하는 LLM List

실습 진행을 위해 아래 내용을 설정한다.

Anthropic
- Anthropic API 키 발급 관련
- .env 파일 내 ANTHROPIC_API_KEY 에 발급받은 키 설정
Gemini
- Gemini API 키 발급 관련
- .env 파일 내 GOOGLE_API_KEY 에 발급받은 키 설정
Together AI
- Together AI API 키 발급 관련
- .env 파일 내 TOGETHER_API_KEY 에 발급받은 키 설정
Ollama
- Ollama Tool Calling 지원 모델 리스트
- 이번 실습에 사용할 llama3.1 모델
- 터미널 창에 ollama pull llama3.1 명령어를 입력하여 모델 다운로드
- 이전에 Ollama 를 사용하지 않았다면, Ollama 를 참고한다.

langchain-ollama 설치를 한 뒤 진행한다.


# ! pip install -qU langchain-ollama==0.1.3
# ! pip install -U notebook

# ! pip install -q langchain_anthropic langchain_google_vertexai langchain_ollama

from langchain_anthropic import ChatAnthropic
from langchain_google_genai import ChatGoogleGenerativeAI
# from langchain_google_vertexai import ChatVertexAI
from langchain_openai import ChatOpenAI
from langchain_ollama import ChatOllama
import os

# GPT-4o-mini
gpt = ChatOpenAI(model="gpt-4o-mini")

# Claude 3.5 Sonnet
claude = ChatAnthropic(model="claude-3-5-sonnet-20240620", temperature=0)

# Gemini 1.5 Pro
gemini = ChatGoogleGenerativeAI(model="gemini-1.5-pro", temperature=0)
# gemini = ChatVertexAI(model="gemini-1.5-pro", temperature=0)

# TogetherAI : Llama-3.1-70B-Instruct-turbo
llama = ChatOpenAI(
    base_url="https://api.together.xyz/v1",
    api_key=os.environ["TOGETHER_API_KEY"],
    model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo"
)

# # Llama 3.1
ollama = ChatOllama(model="llama3.1", temperature=0)

# # Qwen2.5 7B (한글성능 나쁘지 않음.)
qwen = ChatOllama(model="qwen2.5:latest")

각 LLM 기반으로 Agent를 생성한다.

from langchain.agents import create_tool_calling_agent

gpt_agent = create_tool_calling_agent(gpt, tools, prompt)
claude_agent = create_tool_calling_agent(claude, tools, prompt)
gemini_agent = create_tool_calling_agent(gemini, tools, prompt)
llama_agent = create_tool_calling_agent(llama, tools, prompt)
ollama_agent = create_tool_calling_agent(ollama, tools, prompt)
qwen_agent = create_tool_calling_agent(qwen, tools, prompt)

gemini는 SDK를 설치해야한다고 하니 그냥 넘어가자.

AgentExecutor 생성 후ㅜ 실행 및 결과 확인

from langchain.agents import AgentExecutor

# gpt_agent 실행
agent_executor = AgentExecutor(
    agent=gpt_agent,
    tools=tools,
    verbose=True,
    handle_parsing_errors=True
)

result = agent_executor.invoke ({
    "input": "미국 AI 투자와 관련된 뉴스를 검색해주세요."
})

print("Agent 실행 결과 : \n", result["output"])

[1m> Entering new AgentExecutor chain...[0m
[32;1m[1;3m
Invoking: `search_google_news` with `{'query': '미국 AI 투자'}`


[0m[36;1m[1;3m[{'url': 'https://news.google.com/rss/articles/CBMiXkFVX3lxTFA3WU9zaU16TlNPWEVVMU1DaHVGQVhTeXhUNzZjSG5RVnRNbEo0Q2duV2JZdlhLOUdCUi1FVEMxVzZRSHVsX1ZlSTdMWUs2dmtORmFrMnZoYzYtVHBzeFHSAXZBVV95cUxPY24wZ05Td0JqSmRsSUZUOG1NZUtmTEdxTTI2aTV3d1RfbHdoYkFDQ1N2MnUtWWJjZmduTUI5RHBGRUs3NFJfUzd0TEgwSHlLSThneWdqSHNNV0tzcHZaazlFUTB3Rm12SG9UVGNaUGlqRm4zVHZR?oc=5', 'content': "트럼프, 'AI 패권' 야심 드러내...중국 노골적 견제 - YTN"}, {'url': 'https://news.google.com/rss/articles/CBMiUEFVX3lxTE05TlIzaS0wSlN3V3NFZ0xtTnh1aVJYOWdSSVowQlE2YkZaaklTQk5IOF9FTE0xSDZYV25RYkV4dFl5RWVYQmFjbHJKMnV6RVQt0gFTQVVfeXFMTnFTNks0eXlSX0NESm1JX1hoOG1MbnlRcnM0ZjEtTDNOeTdxR0V6RVgxdmZEWXpOWGRNMzhSS1ZzblBzcEYwQl8yYlhQSjJ6bW1NMmc?oc=5', 'content': '트럼프, AI 분야 ‘5천억 달러’ 투자 계획 발표 - VOA Korea'}, {'url': 'https://news.google.com/rss/articles/CBMihAFBVV95cUxQN3FmTE9KZ2llMzVFa0dSQUdySlUyLU9DNTliaUgtN2hNSEd2SkRqYVJkUUoxOVNRbThOZ0dOY2FSZWpaTVZTaE9wTjN0UXZMLS1UcGtJRkxFRlhNVU1aay1PRnJUYXlWRHBvMXdGWHBDY2Zka05YUWRVZU9kYlJqMnNQQ1fSAZgBQVVfeXFMTUY1bzd4TlFIeUNMOWxLY2NGZ3NLcjJnZklxbmN4alUzV19HRW1xQTdYbUdvYlJ6dzc0RndwY0xRYWJhalZOWE5Bb3AxMHhDZHIybWtJRVgtQzhGZzV2OUhEODhRQWlPN25VcDM3NWducG5sV3NmQjdLZXd5UjBYNTUwSi1vQjE0MmJiaXZIVXdnM0lEZExsLUc?oc=5', 'content': '트럼프 “美 AI인프라에 5000억 달러 투자” - 조선일보'}, {'url': 'https://news.google.com/rss/articles/CBMiWkFVX3lxTE1mYm1jSXlwV3hpZGhfRjg4eTF4Q3RwQXFSa0RpNkh2TFk2cUIyUEhEYTl1VWFOeWdyMHllZjMxODEycm1DTmt3ajVYWDhaLVozcUFQYWUxdHREQdIBVEFVX3lxTE5uaWY3LVpGT1MydXJhWjFLTERRcGpLZTJjbW9FNU9PWUtzV0c1bS1NRThsM1FUWHlSNHNkUmtILWU5VFFsSHhkd0w3WWdtQ2RIU21CbA?oc=5', 'content': '"中 \'딥시크\' 보니…" 트럼프 발언에 엔비디아 개미 \'충격\' - 한국경제'}, {'url': 'https://news.google.com/rss/articles/CBMiVkFVX3lxTE56ZXB0bWRUOHRsU2ktUW1fOFgxOWhYTjlGWjczbjVHRG9vOEF2VFY5dUszTTBjVkxfZXJoRGNoTlpVRWZoV00xTFlsRlE0bzFrbmJDVjFn?oc=5', 'content': '트럼프 "AI에 718조 투자"…미∙중 패권 3차대전, 韓엔 호재 - 중앙일보'}][0m[32;1m[1;3m다음은 미국 AI 투자와 관련된 최근 뉴스 기사들입니다:

1. [트럼프, 'AI 패권' 야심 드러내...중국 노골적 견제 - YTN](https://news.google.com/rss/articles/CBMiXkFVX3lxTFA3WU9zaU16TlNPWEVVMU1DaHVGQVhTeXhUNzZjSG5RVnRNbEo0Q2duV2JZdlhLOUdCUi1FVEMxVzZRSHVsX1ZlSTdMWUs2dmtORmFrMnZoYzYtVHBzeFHSAXZBVV95cUxPY24wZ05Td0JqSmRsSUZUOG1NZUtmTEdxTTI2aTV3d1RfbHdoYkFDQ1N2MnUtWWJjZmduTUI5RHBGRUs3NFJfUzd0TEgwSHlLSThneWdqSHNNV0tzcHZaazlFUTB3Rm12SG9UVGNaUGlqRm4zVHZR?oc=5)

2. [트럼프, AI 분야 ‘5천억 달러’ 투자 계획 발표 - VOA Korea](https://news.google.com/rss/articles/CBMiUEFVX3lxTE05TlIzaS0wSlN3V3NFZ0xtTnh1aVJYOWdSSVowQlE2YkZaaklTQk5IOF9FTE0xSDZYV25RYkV4dFl5RWVYQmFjbHJKMnV6RVQt0gFTQVVfeXFMTnFTNks0eXlSX0NESm1JX1hoOG1MbnlRcnM0ZjEtTDNOeTdxR0V6RVgxdmZEWXpOWGRNMzhSS1ZzblBzcEYwQl8yYlhQSjJ6bW1NMmc?oc=5)

3. [트럼프 “美 AI인프라에 5000억 달러 투자” - 조선일보](https://news.google.com/rss/articles/CBMihAFBVV95cUxQN3FmTE9KZ2llMzVFa0dSQUdySlUyLU9DNTliaUgtN2hNSEd2SkRqYVJkUUoxOVNRbThOZ0dOY2FSZWpaTVZTaE9wTjN0UXZMLS1UcGtJRkxFRlhNVU1aay1PRnJUYXlWRHBvMXdGWHBDY2Zka05YUWRVZU9kYlJqMnNQQ1fSAZgBQVVfeXFMTUY1bzd4TlFIeUNMOWxLY2NGZ3NLcjJnZklxbmN4alUzV19HRW1xQTdYbUdvYlJ6dzc0RndwY0xRYWJhalZOWE5Bb3AxMHhDZHIybWtJRVgtQzhGZzV2OUhEODhRQWlPN25VcDM3NWducG5sV3NmQjdLZXd5UjBYNTUwSi1vQjE0MmJiaXZIVXdnM0lEZExsLUc?oc=5)

4. [\"中 '딥시크' 보니…\" 트럼프 발언에 엔비디아 개미 '충격' - 한국경제](https://news.google.com/rss/articles/CBMiWkFVX3lxTE1mYm1jSXlwV3hpZGhfRjg4eTF4Q3RwQXFSa0RpNkh2TFk2cUIyUEhEYTl1VWFOeWdyMHllZjMxODEycm1DTmt3ajVYWDhaLVozcUFQYWUxdHREQdIBVEFVX3lxTE5uaWY3LVpGT1MydXJhWjFLTERRcGpLZTJjbW9FNU9PWUtzV0c1bS1NRThsM1FUWHlSNHNkUmtILWU5VFFsSHhkd0w3WWdtQ2RIU21CbA?oc=5)

5. [트럼프 \"AI에 718조 투자\"…미∙중 패권 3차대전, 韓엔 호재 - 중앙일보](https://news.google.com/rss/articles/CBMiVkFVX3lxTE56ZXB0bWRUOHRsU2ktUW1fOFgxOWhYTjlGWjczbjVHRG9vOEF2VFY5dUszTTBjVkxfZXJoRGNoTlpVRWZoV00xTFlsRlE0bzFrbmJDVjFn?oc=5)

이 기사들은 트럼프 전 대통령의 AI 분야에 대한 투자 계획과 관련된 내용을 다루고 있습니다.[0m

[1m> Finished chain.[0m
Agent 실행 결과 : 
 다음은 미국 AI 투자와 관련된 최근 뉴스 기사들입니다:

1. [트럼프, 'AI 패권' 야심 드러내...중국 노골적 견제 - YTN](https://news.google.com/rss/articles/CBMiXkFVX3lxTFA3WU9zaU16TlNPWEVVMU1DaHVGQVhTeXhUNzZjSG5RVnRNbEo0Q2duV2JZdlhLOUdCUi1FVEMxVzZRSHVsX1ZlSTdMWUs2dmtORmFrMnZoYzYtVHBzeFHSAXZBVV95cUxPY24wZ05Td0JqSmRsSUZUOG1NZUtmTEdxTTI2aTV3d1RfbHdoYkFDQ1N2MnUtWWJjZmduTUI5RHBGRUs3NFJfUzd0TEgwSHlLSThneWdqSHNNV0tzcHZaazlFUTB3Rm12SG9UVGNaUGlqRm4zVHZR?oc=5)

2. [트럼프, AI 분야 ‘5천억 달러’ 투자 계획 발표 - VOA Korea](https://news.google.com/rss/articles/CBMiUEFVX3lxTE05TlIzaS0wSlN3V3NFZ0xtTnh1aVJYOWdSSVowQlE2YkZaaklTQk5IOF9FTE0xSDZYV25RYkV4dFl5RWVYQmFjbHJKMnV6RVQt0gFTQVVfeXFMTnFTNks0eXlSX0NESm1JX1hoOG1MbnlRcnM0ZjEtTDNOeTdxR0V6RVgxdmZEWXpOWGRNMzhSS1ZzblBzcEYwQl8yYlhQSjJ6bW1NMmc?oc=5)

3. [트럼프 “美 AI인프라에 5000억 달러 투자” - 조선일보](https://news.google.com/rss/articles/CBMihAFBVV95cUxQN3FmTE9KZ2llMzVFa0dSQUdySlUyLU9DNTliaUgtN2hNSEd2SkRqYVJkUUoxOVNRbThOZ0dOY2FSZWpaTVZTaE9wTjN0UXZMLS1UcGtJRkxFRlhNVU1aay1PRnJUYXlWRHBvMXdGWHBDY2Zka05YUWRVZU9kYlJqMnNQQ1fSAZgBQVVfeXFMTUY1bzd4TlFIeUNMOWxLY2NGZ3NLcjJnZklxbmN4alUzV19HRW1xQTdYbUdvYlJ6dzc0RndwY0xRYWJhalZOWE5Bb3AxMHhDZHIybWtJRVgtQzhGZzV2OUhEODhRQWlPN25VcDM3NWducG5sV3NmQjdLZXd5UjBYNTUwSi1vQjE0MmJiaXZIVXdnM0lEZExsLUc?oc=5)

4. [\"中 '딥시크' 보니…\" 트럼프 발언에 엔비디아 개미 '충격' - 한국경제](https://news.google.com/rss/articles/CBMiWkFVX3lxTE1mYm1jSXlwV3hpZGhfRjg4eTF4Q3RwQXFSa0RpNkh2TFk2cUIyUEhEYTl1VWFOeWdyMHllZjMxODEycm1DTmt3ajVYWDhaLVozcUFQYWUxdHREQdIBVEFVX3lxTE5uaWY3LVpGT1MydXJhWjFLTERRcGpLZTJjbW9FNU9PWUtzV0c1bS1NRThsM1FUWHlSNHNkUmtILWU5VFFsSHhkd0w3WWdtQ2RIU21CbA?oc=5)

5. [트럼프 \"AI에 718조 투자\"…미∙중 패권 3차대전, 韓엔 호재 - 중앙일보](https://news.google.com/rss/articles/CBMiVkFVX3lxTE56ZXB0bWRUOHRsU2ktUW1fOFgxOWhYTjlGWjczbjVHRG9vOEF2VFY5dUszTTBjVkxfZXJoRGNoTlpVRWZoV00xTFlsRlE0bzFrbmJDVjFn?oc=5)

이 기사들은 트럼프 전 대통령의 AI 분야에 대한 투자 계획과 관련된 내용을 다루고 있습니다.

이제 다양한 llm을 사용하여 agent를 실행해보자.

다음은 입력받은 llm을 사용하여 Agent를 생성하고 실행하여 결과를 출력하는 함수이다.

def execute_agent(llm, tools, input_text, label):
    agent = create_tool_calling_agent(llm, tools, prompt)
    executor = AgentExecutor(agent=agent, tools=tools, verbose=False)
    result = executor.invoke({"input": input_text})
    print(f"[ '{label}' 실행 결과 ]")

    if isinstance(result["output"], list) and len(result["output"]) > 0:
        for item in result ["output"]:
            if "text" in item:
                print(item["text"])
    elif isinstance(result["output"], str):
        print(result["output"])
    else:
        print(result["output"])

각 LLM 별로 에이전트를 생성, 실행하여 결과를 출력.

query = (
    "AI 투자와 관련된 뉴스를 검색하고, 결과를 Instagram 게시글 형식으로 작성해 주세요."
)


# gpt
execute_agent(gpt, tools, query, "gpt")

[ 'gpt' 실행 결과 ]
🌟 **AI 투자 동향** 🌟

1️⃣ **MS와 메타**: 메타가 AI에 대한 지속적인 투자를 통해 주가 상승세를 이어가고 있다는 소식입니다! 🤑 [자세히 보기](https://news.google.com/rss/articles/CBMiWkFVX3lxTE8yeEdfOWxJSmtHSUJYS0ZXVDJaU1RWUTNzU01EVkJQaFpOREZNRE95REpjWU9YR3ZJTlAyODJVdzU2djNZSW5kUjRpVkdid3lfYU9USlNlLTRuZ9IBVEFVX3lxTE1xVzJZZ1lRQ3R6UnlwYkkwS1dsSVBWV2o5TmZoTjBDMGtRanZzMENMMlE2bE9PbHpGMjVsQ0ZOYlk1Wk96bWJGMzlkQ3pqVVZvTndmcw?oc=5)

2️⃣ **트럼프 발 AI 투자**: AI 투자가 한국 기업에 새로운 기회를 제공할 수 있다는 전망이 나오고 있습니다. 전력 솔루션 시장이 주목받고 있네요! 🔍 [자세히 보기](https://news.google.com/rss/articles/CBMiW0FVX3lxTE4zbUlUMW5PZ2xBWVBXQjRrdWM1ZlRlN3BoWE1VRkd3cHRlcTVKSEEtLUZrMWlKc2NkRFY5dmJ3YmlmUTctd2Q4SzNxQnJBZnJMYzJxOVo1WjB1em_SAV5BVV95cUxQWjVzN2t0ejY2akhjZGVGbnpRRWw3ZWUwckFLc3U3NXkwaFcxaTVCRWJHY0M0STNhMU1heURwWmlXa285YnJqa2h5NFdkamVEejhwQnpVNE9uVS1aWk9B?oc=5)

3️⃣ **ETF와 AI 투자**: 금리 하락기에는 ETF와 AI 투자의 조합으로 전략적 자산 배분 가이드가 강조되고 있습니다! 📊 [자세히 보기](https://news.google.com/rss/articles/CBMiVkFVX3lxTFA4WnRkSE1kZnUteWhMdmJWSWFEWUs3Y0tTTVFjc0M2YmNNOEM5b0RHa3pDN0dneXp6ZG9tcGNiQjMtYXF0X2l6a3VoMWlrTHdSVDgyLXlR?oc=5)

4️⃣ **애플 주가 급등**: AI 투자에 상대적으로 적게 참여했다는 이유로 애플의 주가가 3% 급등하며 시가총액 1위를 탈환했습니다! 🍏 [자세히 보기](https://news.google.com/rss/articles/CBMiWEFVX3lxTE5IYk80R1pzV084dlpXODE3ZWRLTGFhZ0NaWmJoTGRIVE00OWwzcVBTTElUanJjRmtDMmVZV20xUkVoaEVFRnVxYTV1VFJ4T1B2dkxMM1cwUDQ?oc=5)

5️⃣ **어닝 서프라이즈 메타**: 메타가 어닝 서프라이즈로 인해 주가가 4% 상승하며 AI 인프라에 대한 지속적인 투자 의사를 밝혔습니다! 💹 [자세히 보기](https://news.google.com/rss/articles/CBMiUkFVX3lxTE5kVkZWUWxKZXRuUzQxRkFOc2pVNzdDMkxFUzFGeE51QUZrb2tMMVo5cUptTExXS01jZEFlTXBQUWlTRXBmYUhfUjNXc24xU1daQkE?oc=5)

📈 **AI 투자 열풍이 계속되고 있습니다! 어떤 기업이 다음 주식의 주인공이 될지 기대됩니다!**

# claude
# execute_agent(claude, tools, query, "claude")

claude는 credit balance 가 부족하여 실행이 안되니 그냥 넘어가자.

# google gemini
execute_agent(gemini, tools, query, "gemini")

---------------------------------------------------------------------------

InvalidArgument                           Traceback (most recent call last)

File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_google_genai/chat_models.py:178, in _chat_with_retry.<locals>._chat_with_retry(**kwargs)
    177 try:
--> 178     return generation_method(**kwargs)
    179 # Do not retry for these errors.


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/ai/generativelanguage_v1beta/services/generative_service/client.py:1143, in GenerativeServiceClient.stream_generate_content(self, request, model, contents, retry, timeout, metadata)
   1142 # Send the request.
-> 1143 response = rpc(
   1144     request,
   1145     retry=retry,
   1146     timeout=timeout,
   1147     metadata=metadata,
   1148 )
   1150 # Done; return the response.


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/gapic_v1/method.py:131, in _GapicCallable.__call__(self, timeout, retry, compression, *args, **kwargs)
    129     kwargs["compression"] = compression
--> 131 return wrapped_func(*args, **kwargs)


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/retry/retry_unary.py:293, in Retry.__call__.<locals>.retry_wrapped_func(*args, **kwargs)
    290 sleep_generator = exponential_sleep_generator(
    291     self._initial, self._maximum, multiplier=self._multiplier
    292 )
--> 293 return retry_target(
    294     target,
    295     self._predicate,
    296     sleep_generator,
    297     timeout=self._timeout,
    298     on_error=on_error,
    299 )


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/retry/retry_unary.py:153, in retry_target(target, predicate, sleep_generator, timeout, on_error, exception_factory, **kwargs)
    151 except Exception as exc:
    152     # defer to shared logic for handling errors
--> 153     _retry_error_helper(
    154         exc,
    155         deadline,
    156         sleep,
    157         error_list,
    158         predicate,
    159         on_error,
    160         exception_factory,
    161         timeout,
    162     )
    163     # if exception not raised, sleep before next attempt


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/retry/retry_base.py:212, in _retry_error_helper(exc, deadline, next_sleep, error_list, predicate_fn, on_error_fn, exc_factory_fn, original_timeout)
    207     final_exc, source_exc = exc_factory_fn(
    208         error_list,
    209         RetryFailureReason.NON_RETRYABLE_ERROR,
    210         original_timeout,
    211     )
--> 212     raise final_exc from source_exc
    213 if on_error_fn is not None:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/retry/retry_unary.py:144, in retry_target(target, predicate, sleep_generator, timeout, on_error, exception_factory, **kwargs)
    143 try:
--> 144     result = target()
    145     if inspect.isawaitable(result):


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/timeout.py:130, in TimeToDeadlineTimeout.__call__.<locals>.func_with_timeout(*args, **kwargs)
    128     kwargs["timeout"] = remaining_timeout
--> 130 return func(*args, **kwargs)


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/google/api_core/grpc_helpers.py:174, in _wrap_stream_errors.<locals>.error_remapped_callable(*args, **kwargs)
    173 except grpc.RpcError as exc:
--> 174     raise exceptions.from_grpc_error(exc) from exc


InvalidArgument: 400 * GenerateContentRequest.contents[2].parts[0].function_response.name: Name cannot be empty.



The above exception was the direct cause of the following exception:


ChatGoogleGenerativeAIError               Traceback (most recent call last)

Cell In[37], line 2
      1 # google gemini
----> 2 execute_agent(gemini, tools, query, "gemini")


Cell In[33], line 4, in execute_agent(llm, tools, input_text, label)
      2 agent = create_tool_calling_agent(llm, tools, prompt)
      3 executor = AgentExecutor(agent=agent, tools=tools, verbose=False)
----> 4 result = executor.invoke({"input": input_text})
      5 print(f"[ '{label}' 실행 결과 ]")
      7 if isinstance(result["output"], list) and len(result["output"]) > 0:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain/chains/base.py:170, in Chain.invoke(self, input, config, **kwargs)
    168 except BaseException as e:
    169     run_manager.on_chain_error(e)
--> 170     raise e
    171 run_manager.on_chain_end(outputs)
    173 if include_run_info:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain/chains/base.py:160, in Chain.invoke(self, input, config, **kwargs)
    157 try:
    158     self._validate_inputs(inputs)
    159     outputs = (
--> 160         self._call(inputs, run_manager=run_manager)
    161         if new_arg_supported
    162         else self._call(inputs)
    163     )
    165     final_outputs: Dict[str, Any] = self.prep_outputs(
    166         inputs, outputs, return_only_outputs
    167     )
    168 except BaseException as e:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain/agents/agent.py:1624, in AgentExecutor._call(self, inputs, run_manager)
   1622 # We now enter the agent loop (until it returns something).
   1623 while self._should_continue(iterations, time_elapsed):
-> 1624     next_step_output = self._take_next_step(
   1625         name_to_tool_map,
   1626         color_mapping,
   1627         inputs,
   1628         intermediate_steps,
   1629         run_manager=run_manager,
   1630     )
   1631     if isinstance(next_step_output, AgentFinish):
   1632         return self._return(
   1633             next_step_output, intermediate_steps, run_manager=run_manager
   1634         )


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain/agents/agent.py:1332, in AgentExecutor._take_next_step(self, name_to_tool_map, color_mapping, inputs, intermediate_steps, run_manager)
   1321 def _take_next_step(
   1322     self,
   1323     name_to_tool_map: Dict[str, BaseTool],
   (...)
   1327     run_manager: Optional[CallbackManagerForChainRun] = None,
   1328 ) -> Union[AgentFinish, List[Tuple[AgentAction, str]]]:
   1329     return self._consume_next_step(
   1330         [
   1331             a
-> 1332             for a in self._iter_next_step(
   1333                 name_to_tool_map,
   1334                 color_mapping,
   1335                 inputs,
   1336                 intermediate_steps,
   1337                 run_manager,
   1338             )
   1339         ]
   1340     )


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain/agents/agent.py:1358, in AgentExecutor._iter_next_step(self, name_to_tool_map, color_mapping, inputs, intermediate_steps, run_manager)
   1355     intermediate_steps = self._prepare_intermediate_steps(intermediate_steps)
   1357     # Call the LLM to see what to do.
-> 1358     output = self._action_agent.plan(
   1359         intermediate_steps,
   1360         callbacks=run_manager.get_child() if run_manager else None,
   1361         **inputs,
   1362     )
   1363 except OutputParserException as e:
   1364     if isinstance(self.handle_parsing_errors, bool):


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain/agents/agent.py:581, in RunnableMultiActionAgent.plan(self, intermediate_steps, callbacks, **kwargs)
    573 final_output: Any = None
    574 if self.stream_runnable:
    575     # Use streaming to make sure that the underlying LLM is invoked in a
    576     # streaming
   (...)
    579     # Because the response from the plan is not a generator, we need to
    580     # accumulate the output into final output and return that.
--> 581     for chunk in self.runnable.stream(inputs, config={"callbacks": callbacks}):
    582         if final_output is None:
    583             final_output = chunk


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:3401, in RunnableSequence.stream(self, input, config, **kwargs)
   3395 def stream(
   3396     self,
   3397     input: Input,
   3398     config: Optional[RunnableConfig] = None,
   3399     **kwargs: Optional[Any],
   3400 ) -> Iterator[Output]:
-> 3401     yield from self.transform(iter([input]), config, **kwargs)


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:3388, in RunnableSequence.transform(self, input, config, **kwargs)
   3382 def transform(
   3383     self,
   3384     input: Iterator[Input],
   3385     config: Optional[RunnableConfig] = None,
   3386     **kwargs: Optional[Any],
   3387 ) -> Iterator[Output]:
-> 3388     yield from self._transform_stream_with_config(
   3389         input,
   3390         self._transform,
   3391         patch_config(config, run_name=(config or {}).get("run_name") or self.name),
   3392         **kwargs,
   3393     )


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:2188, in Runnable._transform_stream_with_config(self, input, transformer, config, run_type, **kwargs)
   2186 try:
   2187     while True:
-> 2188         chunk: Output = context.run(next, iterator)  # type: ignore
   2189         yield chunk
   2190         if final_output_supported:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:3351, in RunnableSequence._transform(self, input, run_manager, config, **kwargs)
   3348     else:
   3349         final_pipeline = step.transform(final_pipeline, config)
-> 3351 yield from final_pipeline


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:1406, in Runnable.transform(self, input, config, **kwargs)
   1403 final: Input
   1404 got_first_val = False
-> 1406 for ichunk in input:
   1407     # The default implementation of transform is to buffer input and
   1408     # then call stream.
   1409     # It'll attempt to gather all input into a single chunk using
   1410     # the `+` operator.
   1411     # If the input is not addable, then we'll assume that we can
   1412     # only operate on the last chunk,
   1413     # and we'll iterate until we get to the last chunk.
   1414     if not got_first_val:
   1415         final = ichunk


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:5559, in RunnableBindingBase.transform(self, input, config, **kwargs)
   5553 def transform(
   5554     self,
   5555     input: Iterator[Input],
   5556     config: Optional[RunnableConfig] = None,
   5557     **kwargs: Any,
   5558 ) -> Iterator[Output]:
-> 5559     yield from self.bound.transform(
   5560         input,
   5561         self._merge_configs(config),
   5562         **{**self.kwargs, **kwargs},
   5563     )


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/runnables/base.py:1424, in Runnable.transform(self, input, config, **kwargs)
   1421             final = ichunk
   1423 if got_first_val:
-> 1424     yield from self.stream(final, config, **kwargs)


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_core/language_models/chat_models.py:398, in BaseChatModel.stream(self, input, config, stop, **kwargs)
    395     self.rate_limiter.acquire(blocking=True)
    397 try:
--> 398     for chunk in self._stream(messages, stop=stop, **kwargs):
    399         if chunk.message.id is None:
    400             chunk.message.id = f"run-{run_manager.run_id}"


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_google_genai/chat_models.py:1034, in ChatGoogleGenerativeAI._stream(self, messages, stop, run_manager, tools, functions, safety_settings, tool_config, generation_config, cached_content, tool_choice, **kwargs)
   1008 def _stream(
   1009     self,
   1010     messages: List[BaseMessage],
   (...)
   1021     **kwargs: Any,
   1022 ) -> Iterator[ChatGenerationChunk]:
   1023     request = self._prepare_request(
   1024         messages,
   1025         stop=stop,
   (...)
   1032         tool_choice=tool_choice,
   1033     )
-> 1034     response: GenerateContentResponse = _chat_with_retry(
   1035         request=request,
   1036         generation_method=self.client.stream_generate_content,
   1037         **kwargs,
   1038         metadata=self.default_metadata,
   1039     )
   1041     prev_usage_metadata: UsageMetadata | None = None
   1042     for chunk in response:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_google_genai/chat_models.py:196, in _chat_with_retry(generation_method, **kwargs)
    193     except Exception as e:
    194         raise e
--> 196 return _chat_with_retry(**kwargs)


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/tenacity/__init__.py:336, in BaseRetrying.wraps.<locals>.wrapped_f(*args, **kw)
    334 copy = self.copy()
    335 wrapped_f.statistics = copy.statistics  # type: ignore[attr-defined]
--> 336 return copy(f, *args, **kw)


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/tenacity/__init__.py:475, in Retrying.__call__(self, fn, *args, **kwargs)
    473 retry_state = RetryCallState(retry_object=self, fn=fn, args=args, kwargs=kwargs)
    474 while True:
--> 475     do = self.iter(retry_state=retry_state)
    476     if isinstance(do, DoAttempt):
    477         try:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/tenacity/__init__.py:376, in BaseRetrying.iter(self, retry_state)
    374 result = None
    375 for action in self.iter_state.actions:
--> 376     result = action(retry_state)
    377 return result


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/tenacity/__init__.py:398, in BaseRetrying._post_retry_check_actions.<locals>.<lambda>(rs)
    396 def _post_retry_check_actions(self, retry_state: "RetryCallState") -> None:
    397     if not (self.iter_state.is_explicit_retry or self.iter_state.retry_run_result):
--> 398         self._add_action_func(lambda rs: rs.outcome.result())
    399         return
    401     if self.after is not None:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/concurrent/futures/_base.py:449, in Future.result(self, timeout)
    447     raise CancelledError()
    448 elif self._state == FINISHED:
--> 449     return self.__get_result()
    451 self._condition.wait(timeout)
    453 if self._state in [CANCELLED, CANCELLED_AND_NOTIFIED]:


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/concurrent/futures/_base.py:401, in Future.__get_result(self)
    399 if self._exception:
    400     try:
--> 401         raise self._exception
    402     finally:
    403         # Break a reference cycle with the exception in self._exception
    404         self = None


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/tenacity/__init__.py:478, in Retrying.__call__(self, fn, *args, **kwargs)
    476 if isinstance(do, DoAttempt):
    477     try:
--> 478         result = fn(*args, **kwargs)
    479     except BaseException:  # noqa: B902
    480         retry_state.set_exception(sys.exc_info())  # type: ignore[arg-type]


File /opt/homebrew/Caskroom/miniforge/base/envs/test_langchain/lib/python3.12/site-packages/langchain_google_genai/chat_models.py:190, in _chat_with_retry.<locals>._chat_with_retry(**kwargs)
    187         raise ValueError(error_msg)
    189 except google.api_core.exceptions.InvalidArgument as e:
--> 190     raise ChatGoogleGenerativeAIError(
    191         f"Invalid argument provided to Gemini: {e}"
    192     ) from e
    193 except Exception as e:
    194     raise e


ChatGoogleGenerativeAIError: Invalid argument provided to Gemini: 400 * GenerateContentRequest.contents[2].parts[0].function_response.name: Name cannot be empty.


# llama3.1 70B (Together.ai)
execute_agent(
    llama,
    tools,
    "Search AI related news and write it in Instagram post format",
    "llama3.1 70B",
)

[ 'llama3.1 70B' 실행 결과 ]
"Hey everyone! 

Want to stay up-to-date on the latest AI news? 

Here are some recent articles from Google News:

1. "딥시크가 '챗GPT' 베낀 증거 있어"..."오픈AI, 합성데이터 활용 방지 예고" - AI타임스
2. "오픈AI 등 ″딥시크 조사…데이터 무단 수집 여부″" - 연합뉴스TV
3. "AI로 제조업 안전사고 예방한다…산업부, R&D 사업 공모" - 연합뉴스
4. "산업부, 제조안전고도화기술개발사업 공고...AI로 제조업 안전사고 예방한다" - AI타임스
5. "AI로 제조업 산재 위험 예방…산업부, 배터리·석유화학 분야 과제 공모" - 뉴시스

Check out these articles to learn more about the latest developments in AI!

#AI #GoogleNews #LatestNews"

# llama3.1 8B (ollama)
execute_agent(ollama, tools, query, "llama3.1(Ollama)")

[ 'llama3.1(Ollama)' 실행 결과 ]
**AI 투자 뉴스**

*   **태국 BOI, 5억 달러 규모의 투자 승인** 태국 BOI는 TikTok의 데이터 호스팅 프로젝트를 포함한 총 5억 달러 규모의 투자를 승인했다. 또한 바이오 경제에 대한 혁신적인 인센티브를 발표했다.
*   **42조 원 규모의 탄약, 'AI' 전부** IT 조선은 42조 원 규모의 탄약이 모두 AI 관련 사업이라고 보도했다.
*   **트럼프 대통령, 미국 내에서의 AI 관련 사업에 77조 엔 설비 투자 발표** nhk.or.jp는 트럼프 대통령이 미국 내에서의 AI 관련 사업에 77조 엔 규모의 설비 투자를 발표했다고 보도했다.
*   **투스 시큐리티, 2025년 미국 투자 전망 발표** KIPOST는 투스 시큐리티가 2025년 미국 투자 전망을 발표했다고 보도했다.
*   **외식업 댓글 관리 AI' 르몽, 블루포인트에서 시드투자 유치** 스타트업엔(StartupN)은 외식업 댓글 관리 AI인 르몽이 블루포인트에서 시드 투자를 유치했다고 보도했다.


# qwen2.5 7B (ollama)
query = "AI 투자와 관련된 뉴스를 검색하고, 결과를 Instagram 게시글 형식으로 작성해 주세요. 한글로 답변하세요!"

execute_agent(qwen, tools, query, "qwen2.5(Ollama)")

[ 'qwen2.5(Ollama)' 실행 결과 ]
Instagram 게시글:

🚀 AI 투자 뉴스 🚀

1. MS, 메타 "AI 투자 지속"에 추가 상승 달렸다 [김현석의 월스트리트나우] - 한국경제

2. 트럼프발 AI 투자, 한국기업에 기회?… "전력 설루션 시장 주목" - 연합뉴스

3. ETF·AI 투자의 새 물결… 금리 하락기의 전략적 자산 배분 가이드 [AI PRISM*주간 금융상품 투자자 뉴스] - 서울경제신문

4. AI 투자 적게 했다는 이유로 애플은 3% 급등, 시총 1위 탈환 - 뉴스1

5. 어닝서프라이즈 메타 4% 급등.. “AI인프라 투자 계속” - 매일경제

# 참고 링크: 
1. [MS, 메타 "AI 투자 지속"에 추가 상승 달렸다](https://news.google.com/rss/articles/CBMiWkFVX3lxTE8yeEdfOWxJSmtHSUJYS0ZXVDJaU1RWUTNzU01EVkJQaFpOREZNRE95REpjWU9YR3ZJTlAyODJVdzU2djNZSW5kUjRpVkdid3lfYU9USlNlLTRuZ9IBVEFVX3lxTE1xVzJZZ1lRQ3R6UnlwYkkwS1dsSVBWV2o5TmZoTjBDMGtRanZzMENMMlE2bE9PbHpGMjVsQ0ZOYlk1Wk96bWJGMzlkQ3pqVVZvTndmcw?oc=5)
2. [트럼프발 AI 투자, 한국기업에 기회?… "전력 설루션 시장 주목"](https://news.google.com/rss/articles/CBMiW0FVX3lxTE4zbUlUMW5PZ2xBWVBXQjRrdWM1ZlRlN3BoWE1VRkd3cHRlcTVKSEEtLUZrMWlKc2NkRFY5dmJ3YmlmUTctd2Q4SzNxQnJBZnJMYzJxOVo1WjB1em_SAV5BVV95cUxQWjVzN2t0ejY2akhjZGVGbnpRRWw3ZWUwckFLc3U3NXkwaFcxaTVCRWJHY0M0STNhMU1heURwWmlXa285YnJqa2h5NFdkamVEejhwQnpVNE9uVS1aWk9B?oc=5)
3. [ETF·AI 투자의 새 물결… 금리 하락기의 전략적 자산 배분 가이드](https://news.google.com/rss/articles/CBMiVkFVX3lxTFA4WnRkSE1kZnUteWhMdmJWSWFEWUs3Y0tTTVFjc0M2YmNNOEM5b0RHa3pDN0dneXp6ZG9tcGNiQjMtYXF0X2l6a3VoMWlrTHdSVDgyLXlR?oc=5)
4. [AI 투자 적게 했다는 이유로 애플은 3% 급등, 시총 1위 탈환](https://news.google.com/rss/articles/CBMiWEFVX3lxTE5IYk80R1pzV084dlpXODE3RkFOc2pVNzdDMkxFUzFGeE51QUZrb2tMMVo5cUptTExXS01jZEFlTXBQUWlTRXBmYUhfUjNXc24xU1daQkE?oc=5)
5. [어닝서프라이즈 메타 4% 급등.. “AI인프라 투자 계속”](https://news.google.com/rss/articles/CBMiUkFVX3lxTE5kVkZWUWxKZXRuUzQxRkFOc2pVNzdDMkxFUzFGeE51QUZrb2tMMVo5cUptTExXS01jZEFlTXBQUWlTRXBmYUhfUjNXc24xU1daQkE?oc=5) 

# #AI #투자 #주식 #증권 #경제 #시황 #마켓뉴스

# (이 게시글은 참고용으로 작성되었으며, 실제 투자는 주의가 필요합니다. 전문적인 자문을 받으시기 바랍니다.)

Ruah

집요한 주니어 개발자의 호되게 당했던 기록

이전 포스트

Agent 주요기능 | Iteration & Human-in-the-loop (중간 human 개입)

다음 포스트