Jina Reranker

本筆記本展示如何使用 Jina Reranker 進行文件壓縮和檢索。

%pip install -qU langchain langchain-openai langchain-community langchain-text-splitters langchainhub

%pip install --upgrade --quiet  faiss

# OR  (depending on Python version)

%pip install --upgrade --quiet  faiss_cpu

# Helper function for printing docs


def pretty_print_docs(docs):
    print(
        f"\n{'-' * 100}\n".join(
            [f"Document {i+1}:\n\n" + d.page_content for i, d in enumerate(docs)]
        )
    )

設定基礎向量儲存檢索器

讓我們先初始化一個簡單的向量儲存檢索器，並儲存 2023 年國情咨文演講（分塊）。我們可以設定檢索器來檢索大量 (20) 文件。

設定 Jina 和 OpenAI API 金鑰

import getpass
import os

os.environ["OPENAI_API_KEY"] = getpass.getpass()
os.environ["JINA_API_KEY"] = getpass.getpass()

from langchain_community.document_loaders import TextLoader
from langchain_community.embeddings import JinaEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_text_splitters import RecursiveCharacterTextSplitter

documents = TextLoader(
    "../../how_to/state_of_the_union.txt",
).load()
text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
texts = text_splitter.split_documents(documents)

embedding = JinaEmbeddings(model_name="jina-embeddings-v2-base-en")
retriever = FAISS.from_documents(texts, embedding).as_retriever(search_kwargs={"k": 20})

query = "What did the president say about Ketanji Brown Jackson"
docs = retriever.get_relevant_documents(query)
pretty_print_docs(docs)

API 參考：TextLoader | JinaEmbeddings | FAISS | RecursiveCharacterTextSplitter

使用 JinaRerank 進行重新排序

現在讓我們使用 ContextualCompressionRetriever 包裝我們的基礎檢索器，並使用 Jina Reranker 作為壓縮器。

from langchain.retrievers import ContextualCompressionRetriever
from langchain_community.document_compressors import JinaRerank

compressor = JinaRerank()
compression_retriever = ContextualCompressionRetriever(
    base_compressor=compressor, base_retriever=retriever
)

compressed_docs = compression_retriever.get_relevant_documents(
    "What did the president say about Ketanji Jackson Brown"
)

API 參考：ContextualCompressionRetriever | JinaRerank

pretty_print_docs(compressed_docs)

使用 Jina Reranker 進行 QA 重新排序

from langchain import hub
from langchain.chains import create_retrieval_chain
from langchain.chains.combine_documents import create_stuff_documents_chain

retrieval_qa_chat_prompt = hub.pull("langchain-ai/retrieval-qa-chat")
retrieval_qa_chat_prompt.pretty_print()

API 參考：hub | create_retrieval_chain | create_stuff_documents_chain

================================ System Message ================================

Answer any use questions based solely on the context below:

<context>
{context}
</context>

============================= Messages Placeholder =============================

{chat_history}

================================ Human Message =================================

{input}

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0)
combine_docs_chain = create_stuff_documents_chain(llm, retrieval_qa_chat_prompt)
chain = create_retrieval_chain(compression_retriever, combine_docs_chain)

API 參考：ChatOpenAI

chain.invoke({"input": query})