app.py

from time import time
from typing import Iterable

# import gradio as gr
import streamlit as st
from langchain.chains import RetrievalQA
from langchain.embeddings import OpenAIEmbeddings
from langchain.embeddings import HuggingFaceEmbeddings

# from langchain.prompts import PromptTemplate
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
from langchain.llms import HuggingFacePipeline

# from langchain.llms import OpenAI
from langchain.chat_models import ChatOpenAI
from langchain.vectorstores import Qdrant
from openai.error import InvalidRequestError
from qdrant_client import QdrantClient
from config import DB_CONFIG, DB_E5_CONFIG


@st.cache_resource
def load_e5_embeddings():
    model_name = "intfloat/multilingual-e5-large"
    model_kwargs = {"device": "cuda:0" if torch.cuda.is_available() else "cpu"}
    encode_kwargs = {"normalize_embeddings": False}
    embeddings = HuggingFaceEmbeddings(
        model_name=model_name,
        model_kwargs=model_kwargs,
        encode_kwargs=encode_kwargs,
    )
    return embeddings


@st.cache_resource
def load_rinna_model():
    if torch.cuda.is_available():
        model_name = "rinna/bilingual-gpt-neox-4b-instruction-ppo"
        tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)
        model = AutoModelForCausalLM.from_pretrained(
            model_name,
            load_in_8bit=True,
            torch_dtype=torch.float16,
            device_map="auto",
        )
        return tokenizer, model
    else:
        return None, None


E5_EMBEDDINGS = load_e5_embeddings()
RINNA_TOKENIZER, RINNA_MODEL = load_rinna_model()


def _get_config_and_embeddings(collection_name: str | None) -> tuple:
    if collection_name is None or collection_name == "E5":
        db_config = DB_E5_CONFIG
        embeddings = E5_EMBEDDINGS
    elif collection_name == "OpenAI":
        db_config = DB_CONFIG
        embeddings = OpenAIEmbeddings()
    else:
        raise ValueError("Unknow collection name")
    return db_config, embeddings


@st.cache_resource
def _get_rinna_llm(temperature: float) -> HuggingFacePipeline | None:
    if RINNA_MODEL is not None:
        pipe = pipeline(
            "text-generation",
            model=RINNA_MODEL,
            tokenizer=RINNA_TOKENIZER,
            max_new_tokens=1024,
            temperature=temperature,
        )
        llm = HuggingFacePipeline(pipeline=pipe)
    else:
        llm = None
    return llm


def _get_llm_model(
    model_name: str | None,
    temperature: float,
):
    if model_name is None:
        model = "gpt-3.5-turbo"
    elif model_name == "rinna":
        model = "rinna"
    elif model_name == "GPT-3.5":
        model = "gpt-3.5-turbo"
    elif model_name == "GPT-4":
        model = "gpt-4"
    else:
        raise ValueError("Unknow model name")
    if model.startswith("gpt"):
        llm = ChatOpenAI(model=model, temperature=temperature)
    elif model == "rinna":
        llm = _get_rinna_llm(temperature)
    return llm


def get_retrieval_qa(
    collection_name: str | None,
    model_name: str | None,
    temperature: float,
    option: str | None,
):
    db_config, embeddings = _get_config_and_embeddings(collection_name)
    db_url, db_api_key, db_collection_name = db_config
    client = QdrantClient(url=db_url, api_key=db_api_key)
    db = Qdrant(
        client=client, collection_name=db_collection_name, embeddings=embeddings
    )

    if option is None or option == "All":
        retriever = db.as_retriever()
    else:
        retriever = db.as_retriever(
            search_kwargs={
                "filter": {"category": option},
            }
        )

    llm = _get_llm_model(model_name, temperature)

    # chain_type_kwargs = {"prompt": PROMPT}
    result = RetrievalQA.from_chain_type(
        llm=llm,
        chain_type="stuff",
        retriever=retriever,
        return_source_documents=True,
        # chain_type_kwargs=chain_type_kwargs,
    )
    return result


def get_related_url(metadata) -> Iterable[str]:
    urls = set()
    for m in metadata:
        # p = m['source']
        url = m["url"]
        if url in urls:
            continue
        urls.add(url)
        category = m["category"]
        # print(m)
        yield f'<p>URL: <a href="{url}">{url}</a> (category: {category})</p>'


def run_qa(query: str, qa: RetrievalQA) -> tuple[str, str]:
    now = time()
    try:
        result = qa(query)
    except InvalidRequestError as e:
        return "回答が見つかりませんでした。別な質問をしてみてください", str(e)
    else:
        metadata = [s.metadata for s in result["source_documents"]]
        sec_html = f"<p>実行時間: {(time() - now):.2f}秒</p>"
        html = "<div>" + sec_html + "\n".join(get_related_url(metadata)) + "</div>"

    return result["result"], html


def main(
    query: str,
    collection_name: str | None,
    model_name: str | None,
    option: str | None,
    temperature: float,
    e5_option: list[str],
) -> Iterable[tuple[str, tuple[str, str]]]:
    qa = get_retrieval_qa(collection_name, model_name, temperature, option)
    if collection_name == "E5":
        for option in e5_option:
            if option == "No":
                yield "E5 No", run_qa(query, qa)
            elif option == "Query":
                yield "E5 Query", run_qa("query: " + query, qa)
            elif option == "Passage":
                yield "E5 Passage", run_qa("passage: " + query, qa)
            else:
                raise ValueError("Unknow option")
    else:
        yield "OpenAI", run_qa(query, qa)


AVAILABLE_LLMS = ["GPT-3.5", "GPT-4"]

if RINNA_MODEL is not None:
    AVAILABLE_LLMS.append("rinna")

with st.form("my_form"):
    query = st.text_input(label="query")
    collection_name = st.radio(options=["E5", "OpenAI"], label="Embedding")

    # if collection_name == "E5":  # TODO : 選択肢で選べるようにする
    e5_option = st.multiselect("E5 option", ["No", "Query", "Passage"], default="No")

    model_name = st.radio(
        options=AVAILABLE_LLMS,
        label="Model",
        help="GPU環境だとrinnaが選択可能",
    )
    option = st.radio(
        options=["All", "ja-book", "ja-nvda-user-guide", "en-nvda-user-guide"],
        label="絞り込み",
        help="ドキュメント制限する?",
    )
    temperature = st.slider(label="temperature", min_value=0, max_value=2)

    submitted = st.form_submit_button("Submit")
    if submitted:
        with st.spinner("Searching..."):
            results = main(
                query, collection_name, model_name, option, temperature, e5_option
            )
            for type_, (answer, html) in results:
                with st.container():
                    st.header(type_)
                    st.write(answer)
                    st.markdown(html, unsafe_allow_html=True)
                    st.divider()