从头构建 RAG（仅限开源！）¶

在本教程中，我们将从头开始，向您展示如何构建一个数据摄取管道到向量数据库中，然后从该向量数据库构建一个检索管道。

值得注意的是，我们使用了完全开源的技术栈：

• Sentence Transformers 作为嵌入模型
• Postgres 作为向量存储（我们也支持许多其他向量存储！）
• Llama 2 作为 LLM（通过 llama.cpp）

设置¶

我们设置开源组件。

1. Sentence Transformers
2. Llama 2
3. 我们初始化 postgres 并使用我们的包装器/抽象对其进行封装。

Sentence Transformers¶

In [ ]

%pip install llama-index-readers-file pymupdf
%pip install llama-index-vector-stores-postgres
%pip install llama-index-embeddings-huggingface
%pip install llama-index-llms-llama-cpp

%pip install llama-index-readers-file pymupdf %pip install llama-index-vector-stores-postgres %pip install llama-index-embeddings-huggingface %pip install llama-index-llms-llama-cpp

In [ ]

# sentence transformers
from llama_index.embeddings.huggingface import HuggingFaceEmbedding

embed_model = HuggingFaceEmbedding(model_name="BAAI/bge-small-en")

# sentence transformers from llama_index.embeddings.huggingface import HuggingFaceEmbedding embed_model = HuggingFaceEmbedding(model_name="BAAI/bge-small-en")

Llama CPP¶

在本 Notebook 中，我们使用 llama-2-chat-13b-ggml 模型以及正确的提示格式。

请查看我们的 Llama CPP 指南以获取完整的设置说明/详细信息。

In [ ]

!pip install llama-cpp-python

!pip install llama-cpp-python

Requirement already satisfied: llama-cpp-python in /Users/jerryliu/Programming/gpt_index/.venv/lib/python3.10/site-packages (0.2.7)
Requirement already satisfied: numpy>=1.20.0 in /Users/jerryliu/Programming/gpt_index/.venv/lib/python3.10/site-packages (from llama-cpp-python) (1.23.5)
Requirement already satisfied: typing-extensions>=4.5.0 in /Users/jerryliu/Programming/gpt_index/.venv/lib/python3.10/site-packages (from llama-cpp-python) (4.7.1)
Requirement already satisfied: diskcache>=5.6.1 in /Users/jerryliu/Programming/gpt_index/.venv/lib/python3.10/site-packages (from llama-cpp-python) (5.6.3)

[notice] A new release of pip available: 22.3.1 -> 23.2.1
[notice] To update, run: pip install --upgrade pip

In [ ]

from llama_index.llms.llama_cpp import LlamaCPP

# model_url = "https://hugging-face.cn/TheBloke/Llama-2-13B-chat-GGML/resolve/main/llama-2-13b-chat.ggmlv3.q4_0.bin"
model_url = "https://hugging-face.cn/TheBloke/Llama-2-13B-chat-GGUF/resolve/main/llama-2-13b-chat.Q4_0.gguf"

llm = LlamaCPP(
    # You can pass in the URL to a GGML model to download it automatically
    model_url=model_url,
    # optionally, you can set the path to a pre-downloaded model instead of model_url
    model_path=None,
    temperature=0.1,
    max_new_tokens=256,
    # llama2 has a context window of 4096 tokens, but we set it lower to allow for some wiggle room
    context_window=3900,
    # kwargs to pass to __call__()
    generate_kwargs={},
    # kwargs to pass to __init__()
    # set to at least 1 to use GPU
    model_kwargs={"n_gpu_layers": 1},
    verbose=True,
)

from llama_index.llms.llama_cpp import LlamaCPP # model_url = "https://hugging-face.cn/TheBloke/Llama-2-13B-chat-GGML/resolve/main/llama-2-13b-chat.ggmlv3.q4_0.bin" model_url = "https://hugging-face.cn/TheBloke/Llama-2-13B-chat-GGUF/resolve/main/llama-2-13b-chat.Q4_0.gguf" llm = LlamaCPP( # You can pass in the URL to a GGML model to download it automatically model_url=model_url, # optionally, you can set the path to a pre-downloaded model instead of model_url model_path=None, temperature=0.1, max_new_tokens=256, # llama2 has a context window of 4096 tokens, but we set it lower to allow for some wiggle room context_window=3900, # kwargs to pass to __call__() generate_kwargs={}, # kwargs to pass to __init__() # set to at least 1 to use GPU model_kwargs={"n_gpu_layers": 1}, verbose=True, )

初始化 Postgres¶

使用在 localhost 上运行的现有 postgres，创建我们将要使用的数据库。

注意：当然还有许多其他开源/自托管数据库可供您使用！例如 Chroma、Qdrant、Weaviate 等等。请参阅我们的向量存储指南。

注意：您需要在本地系统上设置 postgres。这里有一个如何在 OSX 上设置的示例：https://www.sqlshack.com/setting-up-a-postgresql-database-on-mac/。

注意：您还需要安装 pgvector (https://github.com/pgvector/pgvector)。

您可以添加一个角色，如下所示：

CREATE ROLE <user> WITH LOGIN PASSWORD '<password>';
ALTER ROLE <user> SUPERUSER;

In [ ]

!pip install psycopg2-binary pgvector asyncpg "sqlalchemy[asyncio]" greenlet

!pip install psycopg2-binary pgvector asyncpg "sqlalchemy[asyncio]" greenlet

In [ ]

import psycopg2

db_name = "vector_db"
host = "localhost"
password = "password"
port = "5432"
user = "jerry"
# conn = psycopg2.connect(connection_string)
conn = psycopg2.connect(
    dbname="postgres",
    host=host,
    password=password,
    port=port,
    user=user,
)
conn.autocommit = True

with conn.cursor() as c:
    c.execute(f"DROP DATABASE IF EXISTS {db_name}")
    c.execute(f"CREATE DATABASE {db_name}")

import psycopg2 db_name = "vector_db" host = "localhost" password = "password" port = "5432" user = "jerry" # conn = psycopg2.connect(connection_string) conn = psycopg2.connect( dbname="postgres", host=host, password=password, port=port, user=user, ) conn.autocommit = True with conn.cursor() as c: c.execute(f"DROP DATABASE IF EXISTS {db_name}") c.execute(f"CREATE DATABASE {db_name}")

In [ ]

from sqlalchemy import make_url
from llama_index.vector_stores.postgres import PGVectorStore

vector_store = PGVectorStore.from_params(
    database=db_name,
    host=host,
    password=password,
    port=port,
    user=user,
    table_name="llama2_paper",
    embed_dim=384,  # openai embedding dimension
)

from sqlalchemy import make_url from llama_index.vector_stores.postgres import PGVectorStore vector_store = PGVectorStore.from_params( database=db_name, host=host, password=password, port=port, user=user, table_name="llama2_paper", embed_dim=384, # openai embedding dimension )

从头构建摄取管道¶

我们将展示如何构建引言中提到的摄取管道。

这里我们快速介绍这些步骤（可以跳过元数据提取）。更多详细信息请参阅我们的专门摄取指南。

1. 加载数据¶

In [ ]

!mkdir data
!wget --user-agent "Mozilla" "https://arxiv.org/pdf/2307.09288.pdf" -O "data/llama2.pdf"

!mkdir data !wget --user-agent "Mozilla" "https://arxiv.org/pdf/2307.09288.pdf" -O "data/llama2.pdf"

In [ ]

from pathlib import Path
from llama_index.readers.file import PyMuPDFReader

from pathlib import Path from llama_index.readers.file import PyMuPDFReader

In [ ]

loader = PyMuPDFReader()
documents = loader.load(file_path="./data/llama2.pdf")

loader = PyMuPDFReader() documents = loader.load(file_path="./data/llama2.pdf")

2. 使用文本分割器分割文档¶

In [ ]

from llama_index.core.node_parser import SentenceSplitter

from llama_index.core.node_parser import SentenceSplitter

In [ ]

text_parser = SentenceSplitter(
    chunk_size=1024,
    # separator=" ",
)

text_parser = SentenceSplitter( chunk_size=1024, # separator=" ", )

In [ ]

text_chunks = []
# maintain relationship with source doc index, to help inject doc metadata in (3)
doc_idxs = []
for doc_idx, doc in enumerate(documents):
    cur_text_chunks = text_parser.split_text(doc.text)
    text_chunks.extend(cur_text_chunks)
    doc_idxs.extend([doc_idx] * len(cur_text_chunks))

text_chunks = [] # maintain relationship with source doc index, to help inject doc metadata in (3) doc_idxs = [] for doc_idx, doc in enumerate(documents): cur_text_chunks = text_parser.split_text(doc.text) text_chunks.extend(cur_text_chunks) doc_idxs.extend([doc_idx] * len(cur_text_chunks))

3. 从文本块手动构建节点¶

In [ ]

from llama_index.core.schema import TextNode

nodes = []
for idx, text_chunk in enumerate(text_chunks):
    node = TextNode(
        text=text_chunk,
    )
    src_doc = documents[doc_idxs[idx]]
    node.metadata = src_doc.metadata
    nodes.append(node)

from llama_index.core.schema import TextNode nodes = [] for idx, text_chunk in enumerate(text_chunks): node = TextNode( text=text_chunk, ) src_doc = documents[doc_idxs[idx]] node.metadata = src_doc.metadata nodes.append(node)

4. 为每个节点生成嵌入¶

这里我们使用 sentence_transformers 模型为每个节点生成嵌入。

In [ ]

for node in nodes:
    node_embedding = embed_model.get_text_embedding(
        node.get_content(metadata_mode="all")
    )
    node.embedding = node_embedding

for node in nodes: node_embedding = embed_model.get_text_embedding( node.get_content(metadata_mode="all") ) node.embedding = node_embedding

5. 将节点加载到向量存储中¶

现在我们将这些节点插入到我们的 PostgresVectorStore 中。

In [ ]

vector_store.add(nodes)

vector_store.add(nodes)

从头构建检索管道¶

我们将展示如何构建一个检索管道。与摄取类似，我们快速介绍这些步骤。请查阅我们的检索指南了解更多详细信息！

In [ ]

query_str = "Can you tell me about the key concepts for safety finetuning"

query_str = "Can you tell me about the key concepts for safety finetuning"

1. 生成查询嵌入¶

In [ ]

query_embedding = embed_model.get_query_embedding(query_str)

query_embedding = embed_model.get_query_embedding(query_str)

2. 查询向量数据库¶

In [ ]

# construct vector store query
from llama_index.core.vector_stores import VectorStoreQuery

query_mode = "default"
# query_mode = "sparse"
# query_mode = "hybrid"

vector_store_query = VectorStoreQuery(
    query_embedding=query_embedding, similarity_top_k=2, mode=query_mode
)

# construct vector store query from llama_index.core.vector_stores import VectorStoreQuery query_mode = "default" # query_mode = "sparse" # query_mode = "hybrid" vector_store_query = VectorStoreQuery( query_embedding=query_embedding, similarity_top_k=2, mode=query_mode )

In [ ]

# returns a VectorStoreQueryResult
query_result = vector_store.query(vector_store_query)
print(query_result.nodes[0].get_content())

# returns a VectorStoreQueryResult query_result = vector_store.query(vector_store_query) print(query_result.nodes[0].get_content())

3. 将结果解析为一组节点¶

In [ ]

from llama_index.core.schema import NodeWithScore
from typing import Optional

nodes_with_scores = []
for index, node in enumerate(query_result.nodes):
    score: Optional[float] = None
    if query_result.similarities is not None:
        score = query_result.similarities[index]
    nodes_with_scores.append(NodeWithScore(node=node, score=score))

from llama_index.core.schema import NodeWithScore from typing import Optional nodes_with_scores = [] for index, node in enumerate(query_result.nodes): score: Optional[float] = None if query_result.similarities is not None: score = query_result.similarities[index] nodes_with_scores.append(NodeWithScore(node=node, score=score))

4. 放入检索器中¶

In [ ]

from llama_index.core import QueryBundle
from llama_index.core.retrievers import BaseRetriever
from typing import Any, List


class VectorDBRetriever(BaseRetriever):
    """Retriever over a postgres vector store."""

    def __init__(
        self,
        vector_store: PGVectorStore,
        embed_model: Any,
        query_mode: str = "default",
        similarity_top_k: int = 2,
    ) -> None:
        """Init params."""
        self._vector_store = vector_store
        self._embed_model = embed_model
        self._query_mode = query_mode
        self._similarity_top_k = similarity_top_k
        super().__init__()

    def _retrieve(self, query_bundle: QueryBundle) -> List[NodeWithScore]:
        """Retrieve."""
        query_embedding = embed_model.get_query_embedding(
            query_bundle.query_str
        )
        vector_store_query = VectorStoreQuery(
            query_embedding=query_embedding,
            similarity_top_k=self._similarity_top_k,
            mode=self._query_mode,
        )
        query_result = vector_store.query(vector_store_query)

        nodes_with_scores = []
        for index, node in enumerate(query_result.nodes):
            score: Optional[float] = None
            if query_result.similarities is not None:
                score = query_result.similarities[index]
            nodes_with_scores.append(NodeWithScore(node=node, score=score))

        return nodes_with_scores

from llama_index.core import QueryBundle from llama_index.core.retrievers import BaseRetriever from typing import Any, List class VectorDBRetriever(BaseRetriever): """Retriever over a postgres vector store.""" def __init__( self, vector_store: PGVectorStore, embed_model: Any, query_mode: str = "default", similarity_top_k: int = 2, ) -> None: """Init params.""" self._vector_store = vector_store self._embed_model = embed_model self._query_mode = query_mode self._similarity_top_k = similarity_top_k super().__init__() def _retrieve(self, query_bundle: QueryBundle) -> List[NodeWithScore]: """Retrieve.""" query_embedding = embed_model.get_query_embedding( query_bundle.query_str ) vector_store_query = VectorStoreQuery( query_embedding=query_embedding, similarity_top_k=self._similarity_top_k, mode=self._query_mode, ) query_result = vector_store.query(vector_store_query) nodes_with_scores = [] for index, node in enumerate(query_result.nodes): score: Optional[float] = None if query_result.similarities is not None: score = query_result.similarities[index] nodes_with_scores.append(NodeWithScore(node=node, score=score)) return nodes_with_scores

In [ ]

retriever = VectorDBRetriever(
    vector_store, embed_model, query_mode="default", similarity_top_k=2
)

retriever = VectorDBRetriever( vector_store, embed_model, query_mode="default", similarity_top_k=2 )

将其接入我们的 RetrieverQueryEngine 以合成响应¶

In [ ]

from llama_index.core.query_engine import RetrieverQueryEngine

query_engine = RetrieverQueryEngine.from_args(retriever, llm=llm)

from llama_index.core.query_engine import RetrieverQueryEngine query_engine = RetrieverQueryEngine.from_args(retriever, llm=llm)

In [ ]

query_str = "How does Llama 2 perform compared to other open-source models?"

response = query_engine.query(query_str)

query_str = "How does Llama 2 perform compared to other open-source models?" response = query_engine.query(query_str)

In [ ]

print(str(response))

print(str(response))

 Based on the results shown in Table 3, Llama 2 outperforms all open-source models on most of the benchmarks, with an average improvement of around 5 points over the next best model (GPT-3.5).

In [ ]

print(response.source_nodes[0].get_content())

print(response.source_nodes[0].get_content())