DEVOCEAN

from langchain_community.chat_models import ChatOllama
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate

# LangChain이 지원하는 다른 채팅 모델을 사용합니다. 여기서는 Ollama를 사용합니다.
llm = ChatOllama(model="ggml-q5-km:latest")

from fastapi import FastAPI
from fastapi.responses import RedirectResponse
from fastapi.middleware.cors import CORSMiddleware
from typing import List, Union
from langserve.pydantic_v1 import BaseModel, Field
from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
from langserve import add_routes
from llm import llm as model


app = FastAPI()

# Set all CORS enabled originsapp.add_middleware(
    CORSMiddleware,    allow_origins=["*"],    allow_credentials=True,    allow_methods=["*"],    allow_headers=["*"],    expose_headers=["*"],)


@app.get("/")
async def redirect_root_to_docs():
    return RedirectResponse("/llm/playground")


class InputChat(BaseModel):
    """Input for the chat endpoint."""    messages: List[Union[HumanMessage, AIMessage, SystemMessage]] = Field(
        ...,        description="The chat messages representing the current conversation.",    )


add_routes(app, model, path="/llm")


if __name__ == "__main__":
    import uvicorn

    uvicorn.run(app, host="0.0.0.0", port=8000)

import os
import streamlit as st
from langchain.embeddings import CacheBackedEmbeddings
from langchain.storage import LocalFileStore
from langchain_openai import OpenAIEmbeddings
from langchain_community.embeddings.huggingface import HuggingFaceEmbeddings
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain_core.messages import ChatMessage
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.document_loaders.unstructured import UnstructuredFileLoader
from langchain_community.vectorstores.faiss import FAISS
from langserve import RemoteRunnable
from langchain_openai import ChatOpenAI
from langchain_core.callbacks.streaming_stdout import StreamingStdOutCallbackHandler


# Embedding 설정# USE_BGE_EMBEDDING = True 로 설정시 HuggingFace BAAI/bge-m3 임베딩 사용 (2.7GB 다운로드 시간 걸릴 수 있습니다)# USE_BGE_EMBEDDING = False 로 설정시 OpenAIEmbeddings 사용 (OPENAI_API_KEY 입력 필요. 과금)USE_BGE_EMBEDDING = True# LangServe 모델 설정(EndPoint)# 1) REMOTE 접속: 본인의 REMOTE LANGSERVE 주소 입력# (예시)# LANGSERVE_ENDPOINT = "https://poodle-deep-marmot.ngrok-free.app/llm/"LANGSERVE_ENDPOINT = "https://NGROK에서_할당받은_URL/llm/"# 2) LocalHost 접속: 끝에 붙는 N4XyA 는 각자 다르니# http://localhost:8000/llm/playground 에서 python SDK 에서 확인!# LANGSERVE_ENDPOINT = "http://localhost:8000/llm/c/N4XyA"LANGSERVE_ENDPOINT = "http://localhost:8000/llm/c/N4XyA"# 필수 디렉토리 생성 @Mineruif not os.path.exists(".cache"):
    os.mkdir(".cache")
if not os.path.exists(".cache/embeddings"):
    os.mkdir(".cache/embeddings")
if not os.path.exists(".cache/files"):
    os.mkdir(".cache/files")

# 프롬프트를 자유롭게 수정해 보세요!RAG_PROMPT_TEMPLATE = """당신은 질문에 친절하게 답변하는 AI 입니다. 검색된 다음 문맥을 사용하여 질문에 답하세요. 답을 모른다면 모른다고 답변하세요.Question: {question} Context: {context} Answer:"""st.set_page_config(page_title="OpenLLM local 테스트", page_icon="💬")
st.title("OpenLLM Local 테스트")


if "messages" not in st.session_state:
    st.session_state["messages"] = [
        ChatMessage(role="assistant", content="무엇을 도와드릴까요?")
    ]


def print_history():
    for msg in st.session_state.messages:
        st.chat_message(msg.role).write(msg.content)


def add_history(role, content):
    st.session_state.messages.append(ChatMessage(role=role, content=content))


def format_docs(docs):
    # 검색한 문서 결과를 하나의 문단으로 합쳐줍니다.    return "\n\n".join(doc.page_content for doc in docs)


@st.cache_resource(show_spinner="Embedding file...")
def embed_file(file):
    file_content = file.read()
    file_path = f"./.cache/files/{file.name}"    with open(file_path, "wb") as f:
        f.write(file_content)

    cache_dir = LocalFileStore(f"./.cache/embeddings/{file.name}")

    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=500,        chunk_overlap=50,        separators=["\n\n", "\n", "(?<=\. )", " ", ""],        length_function=len,    )
    loader = UnstructuredFileLoader(file_path)
    docs = loader.load_and_split(text_splitter=text_splitter)

    if USE_BGE_EMBEDDING:
        # BGE Embedding: @Mineru        model_name = "BAAI/bge-m3"        # GPU Device 설정:        # - NVidia GPU: "cuda"        # - Mac M1, M2, M3: "mps"        # - CPU: "cpu"        model_kwargs = {
            # "device": "cuda"            "device": "mps"            # "device": "cpu"        }
        encode_kwargs = {"normalize_embeddings": True}
        embeddings = HuggingFaceEmbeddings(
            model_name=model_name,            model_kwargs=model_kwargs,            encode_kwargs=encode_kwargs,        )
    
    cached_embeddings = CacheBackedEmbeddings.from_bytes_store(embeddings, cache_dir)
    vectorstore = FAISS.from_documents(docs, embedding=cached_embeddings)
    retriever = vectorstore.as_retriever()
    return retriever


def format_docs(docs):
    # 검색한 문서 결과를 하나의 문단으로 합쳐줍니다.    return "\n\n".join(doc.page_content for doc in docs)


with st.sidebar:
    file = st.file_uploader(
        "파일 업로드",        type=["pdf", "txt", "docx"],    )

if file:
    retriever = embed_file(file)

print_history()


if user_input := st.chat_input():
    add_history("user", user_input)
    st.chat_message("user").write(user_input)
    with st.chat_message("assistant"):
        # ngrok remote 주소 설정        ollama = RemoteRunnable(LANGSERVE_ENDPOINT)
        chat_container = st.empty()
        if file is not None:
            prompt = ChatPromptTemplate.from_template(RAG_PROMPT_TEMPLATE)

            # 체인을 생성합니다.            rag_chain = (
                {
                    "context": retriever | format_docs,                    "question": RunnablePassthrough(),                }
                | prompt
                | ollama
                | StrOutputParser()
            )
            # 문서에 대한 질의를 입력하고, 답변을 출력합니다.            answer = rag_chain.stream(user_input)  # 문서에 대한 질의            chunks = []
            for chunk in answer:
                chunks.append(chunk)
                chat_container.markdown("".join(chunks))
            add_history("ai", "".join(chunks))
        else:
            prompt = ChatPromptTemplate.from_template(
                "다음의 질문에 간결하게 답변해 주세요:\n{input}"            )

            # 체인을 생성합니다.            chain = prompt | ollama | StrOutputParser()

            answer = chain.stream(user_input)  # 문서에 대한 질의            chunks = []
            for chunk in answer:
                chunks.append(chunk)
                chat_container.markdown("".join(chunks))
            add_history("ai", "".join(chunks))

$ python langserv_server.py
INFO:     Started server process [28441]
INFO:     Waiting for application startup.

 __          ___      .__   __.   _______      _______. _______ .______     ____    ____  _______
|  |        /   \     |  \ |  |  /  _____|    /       ||   ____||   _  \    \   \  /   / |   ____|
|  |       /  ^  \    |   \|  | |  |  __     |   (----`|  |__   |  |_)  |    \   \/   /  |  |__
|  |      /  /_\  \   |  . `  | |  | |_ |     \   \    |   __|  |      /      \      /   |   __|
|  `----./  _____  \  |  |\   | |  |__| | .----)   |   |  |____ |  |\  \----.  \    /    |  |____
|_______/__/     \__\ |__| \__|  \______| |_______/    |_______|| _| `._____|   \__/     |_______|

LANGSERVE: Playground for chain "/llm/" is live at:
LANGSERVE:  │
LANGSERVE:  └──> /llm/playground/
LANGSERVE:
LANGSERVE: See all available routes at /docs/

LANGSERVE: ⚠️ Using pydantic 2.7.0. OpenAPI docs for invoke, batch, stream, stream_log endpoints will not be generated. API endpoints and playground should work as expected. If you need to see the docs, you can downgrade to pydantic 1. For example, `pip install pydantic==1.10.13`. See https://github.com/tiangolo/fastapi/issues/10360 for details.

INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

$ streamlit run stream_main.py --server.port 8080

  You can now view your Streamlit app in your browser.

  Local URL: http://localhost:8080
  Network URL: http://192.168.0.69:8080

제1장 KBO 정규시즌
제1조 경기방식
1. 단일 리그제로 각 구단은 144경기(구단간 16차전)씩 총 720경기를 거행한다.
2. 연장전은 12회(KBO 포스트시즌 15회)까지로 하고 승패를 가리지 못할 경우에는 무승부로 한다.
제2조 승률계산법
승률은 승수/(승수+패수)로 한다.