■ RunnablePassthrough 클래스의 assign 메소드를 사용해 대화 기록을 관리하는 방법을 보여준다.
• 챗봇을 구축할 때 중요한 점은 대화 기록을 관리하는 방법이다.
• 메시지 목록이 무제한으로 늘어나면 LLM의 컨텍스트 창은 오버플로우 할 가능성이 있다.
• 전달하는 메시지의 크기를 제한하는 단계를 추가하는 것이 중요하다.
▶ main.py
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
import os from langchain_core.messages import HumanMessage, AIMessage from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder from langchain_core.runnables import RunnablePassthrough from langchain_openai import ChatOpenAI os.environ["OPENAI_API_KEY"] = "<OPENAI_API_KEY>" def filterMessageList(messageList, k = 10): return messageList[-k:] runnableAssign = RunnablePassthrough.assign(messages = lambda x : filterMessageList(x["messages"])) chatPromptTemplate = ChatPromptTemplate.from_messages( [ ("system", "You are a helpful assistant. Answer all questions to the best of your ability in {language}."), MessagesPlaceholder(variable_name = "messages"), ] ) chatOpenAI = ChatOpenAI(model = "gpt-3.5-turbo") runnableSequence = runnableAssign | chatPromptTemplate | chatOpenAI chatMessageList = [ HumanMessage(content = "hi! I'm david" ), AIMessage (content = "hi!" ), HumanMessage(content = "I like vanilla ice cream"), AIMessage (content = "nice" ), HumanMessage(content = "whats 2 + 2" ), AIMessage (content = "4" ), HumanMessage(content = "thanks" ), AIMessage (content = "no problem!" ), HumanMessage(content = "having fun?" ), AIMessage (content = "yes!" ) ] response = runnableSequence.invoke( { "messages" : chatMessageList + [HumanMessage(content = "what's my name?")], "language" : "English", } ) print(response.content) """ Sorry, I don't have access to your personal information. """ |
▶ requirements.txt
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
aiohttp==3.9.5 aiosignal==1.3.1 annotated-types==0.7.0 anyio==4.4.0 async-timeout==4.0.3 attrs==23.2.0 certifi==2024.6.2 charset-normalizer==3.3.2 distro==1.9.0 exceptiongroup==1.2.1 frozenlist==1.4.1 greenlet==3.0.3 h11==0.14.0 httpcore==1.0.5 httpx==0.27.0 idna==3.7 jsonpatch==1.33 jsonpointer==2.4 langchain==0.2.3 langchain-core==0.2.5 langchain-openai==0.1.8 langchain-text-splitters==0.2.1 langsmith==0.1.75 multidict==6.0.5 numpy==1.26.4 openai==1.33.0 orjson==3.10.3 packaging==23.2 pydantic==2.7.3 pydantic_core==2.18.4 PyYAML==6.0.1 regex==2024.5.15 requests==2.32.3 sniffio==1.3.1 SQLAlchemy==2.0.30 tenacity==8.3.0 tiktoken==0.7.0 tqdm==4.66.4 typing_extensions==4.12.2 urllib3==2.2.1 yarl==1.9.4 |
※ pip install langchain langchain-openai 명령을 실행했다.