Archives/langchain
2
0
Fork 1
mirror of https://github.com/hwchase17/langchain synced 2024-11-06 03:20:49 +00:00
Code Issues Packages Projects Releases Wiki Activity

Harrison/indexing pipeline (#1317)

Browse Source
This commit is contained in:
Harrison Chase 2023-02-27 00:31:36 -08:00 committed by GitHub
parent a0bf856c70
commit 0824d65a5c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 211 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
docs/modules/indexes/chain_examples/question_answering.ipynb
View File

@ -30,29 +30,24 @@
"from langchain.text_splitter import CharacterTextSplitter\n",
"from langchain.vectorstores import Chroma\n",
"from langchain.docstore.document import Document\n",
"from langchain.prompts import PromptTemplate"
"from langchain.prompts import PromptTemplate\n",
"from langchain.indexes.vectorstore import VectorstoreIndexCreator"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "291f0117",
"id": "ef9305cc",
"metadata": {},
"outputs": [],
"source": [
"from langchain.document_loaders import TextLoader\n",
"loader = TextLoader('../../state_of_the_union.txt')\n",
"documents = loader.load()\n",
"text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",
"texts = text_splitter.split_documents(documents)\n",
"\n",
"embeddings = OpenAIEmbeddings()"
"index_creator = VectorstoreIndexCreator()"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "fd9666a9",
"execution_count": 3,
"id": "291f0117",
"metadata": {},
"outputs": [
{
@ -65,12 +60,14 @@
}
],
"source": [
"docsearch = Chroma.from_documents(texts, embeddings)"
"from langchain.document_loaders import TextLoader\n",
"loader = TextLoader('../../state_of_the_union.txt')\n",
"docsearch = index_creator.from_loaders([loader])"
]
},
{
"cell_type": "code",
"execution_count": 5,
"execution_count": 4,
"id": "d1eaf6e6",
"metadata": {},
"outputs": [],

184
docs/modules/indexes/getting_started.ipynb
View File

@ -2,45 +2,152 @@
"cells": [
{
"cell_type": "markdown",
"id": "07c1e3b9",
"id": "2244801b",
"metadata": {},
"source": [
"# Getting Started\n",
"\n",
"This example showcases question answering over a vector database.\n",
"We have chosen this as the example for getting started because it nicely combines a lot of different elements (Text splitters, embeddings, vectorstores) and then also shows how to use them in a chain."
"This example showcases question answering over documents.\n",
"We have chosen this as the example for getting started because it nicely combines a lot of different elements (Text splitters, embeddings, vectorstores) and then also shows how to use them in a chain.\n",
"\n",
"Question answering over documents consists of three steps:\n",
"\n",
"1. Create an index\n",
"2. Create a question answering chain\n",
"3. Ask questions!\n",
"\n",
"Each of the steps has multiple sub steps and potential configurations. In this notebook we will primarily focus on (1). We will start by showing the one-liner for doing so, but then break down what is actually going on.\n",
"\n",
"First, let's import some common classes we'll use no matter what."
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "82525493",
"id": "8d369452",
"metadata": {},
"outputs": [],
"source": [
"from langchain.embeddings.openai import OpenAIEmbeddings\n",
"from langchain.vectorstores import Chroma\n",
"from langchain.text_splitter import CharacterTextSplitter\n",
"from langchain import OpenAI, VectorDBQA"
"from langchain.chains import VectorDBQA\n",
"from langchain.llms import OpenAI"
]
},
{
"cell_type": "markdown",
"id": "0b7adc54",
"id": "07c1e3b9",
"metadata": {},
"source": [
"Here we load in the documents we want to use to create our index."
"Next in the generic setup, let's specify the document loader we want to use."
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "33958a86",
"metadata": {},
"outputs": [],
"source": [
"from langchain.document_loaders import TextLoader\n",
"loader = TextLoader('../state_of_the_union.txt')"
]
},
{
"cell_type": "markdown",
"id": "489c74bb",
"metadata": {},
"source": [
"## One Line Index Creation\n",
"\n",
"To get started as quickly as possible, we can use the `VectorstoreIndexCreator`."
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "611e0c19",
"id": "403fc231",
"metadata": {},
"outputs": [],
"source": [
"from langchain.indexes import VectorstoreIndexCreator"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "57a8a199",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Running Chroma using direct local API.\n",
"Using DuckDB in-memory for database. Data will be transient.\n"
]
}
],
"source": [
"index = VectorstoreIndexCreator().from_loaders([loader])"
]
},
{
"cell_type": "markdown",
"id": "f3493fa4",
"metadata": {},
"source": [
"Now that the index is created, we can use it in a VectorDBQAChain to ask questions of the data!"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "23d0d234",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"\" The president said that Ketanji Brown Jackson is one of the nation's top legal minds, a consensus builder, and has gained a broad range of support. He also said that she is a former top litigator in private practice, a former federal public defender, and from a family of public school educators and police officers.\""
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"qa = VectorDBQA.from_chain_type(llm=OpenAI(), chain_type=\"stuff\", vectorstore=index)\n",
"query = \"What did the president say about Ketanji Brown Jackson\"\n",
"qa.run(query)"
]
},
{
"cell_type": "markdown",
"id": "2cb6d2eb",
"metadata": {},
"source": [
"## Walkthrough\n",
"\n",
"Okay, so what's actually going on? How is this index getting created?\n",
"\n",
"A lot of the magic is being hid in this `VectorstoreIndexCreator`. What is this doing?\n",
"\n",
"There are three main steps going on after the documents are loaded:\n",
"\n",
"1. Splitting documents into chunks\n",
"2. Creating embeddings for each document\n",
"3. Storing documents and embeddings in a vectorstore\n",
"\n",
"Let's walk through this in code"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "54270abc",
"metadata": {},
"outputs": [],
"source": [
"from langchain.document_loaders import TextLoader\n",
"loader = TextLoader('../state_of_the_union.txt')\n",
"documents = loader.load()"
]
},
@ -54,11 +161,12 @@
},
{
"cell_type": "code",
"execution_count": 4,
"execution_count": 8,
"id": "afecb8cf",
"metadata": {},
"outputs": [],
"source": [
"from langchain.text_splitter import CharacterTextSplitter\n",
"text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",
"texts = text_splitter.split_documents(documents)"
]
@ -73,11 +181,12 @@
},
{
"cell_type": "code",
"execution_count": 5,
"execution_count": 10,
"id": "9eaaa735",
"metadata": {},
"outputs": [],
"source": [
"from langchain.embeddings import OpenAIEmbeddings\n",
"embeddings = OpenAIEmbeddings()"
]
},
@ -91,7 +200,7 @@
},
{
"cell_type": "code",
"execution_count": 6,
"execution_count": 11,
"id": "5c7049db",
"metadata": {},
"outputs": [
@ -105,6 +214,7 @@
}
],
"source": [
"from langchain.vectorstores import Chroma\n",
"db = Chroma.from_documents(texts, embeddings)"
]
},
@ -113,12 +223,12 @@
"id": "30c4e5c6",
"metadata": {},
"source": [
"Finally, we create a chain and use it to answer questions!"
"Then, as before, we create a chain and use it to answer questions!"
]
},
{
"cell_type": "code",
"execution_count": 9,
"execution_count": 12,
"id": "3018f865",
"metadata": {},
"outputs": [],
@ -128,17 +238,17 @@
},
{
"cell_type": "code",
"execution_count": 10,
"execution_count": 13,
"id": "032a47f8",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"\" The President said that Ketanji Brown Jackson is one of the nation's top legal minds, a former top litigator in private practice, a former federal public defender, and from a family of public school educators and police officers. He said that she is a consensus builder and has received a broad range of support from the Fraternal Order of Police to former judges appointed by Democrats and Republicans.\""
"\" The President said that Ketanji Brown Jackson is one of the nation's top legal minds and a consensus builder, with a broad range of support from the Fraternal Order of Police to former judges appointed by Democrats and Republicans. She is a former top litigator in private practice, a former federal public defender, and from a family of public school educators and police officers.\""
]
},
"execution_count": 10,
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
@ -148,10 +258,40 @@
"qa.run(query)"
]
},
{
"cell_type": "markdown",
"id": "9464690e",
"metadata": {},
"source": [
"`VectorstoreIndexCreator` is just a wrapper around all this logic. It is configurable in the text splitter it uses, the embeddings it uses, and the vectorstore it uses. For example, you can configure it as below:"
]
},
{
"cell_type": "code",
"execution_count": 14,
"id": "4001bbc6",
"metadata": {},
"outputs": [],
"source": [
"index_creator = VectorstoreIndexCreator(\n",
" vectorstore_cls=Chroma, \n",
" embedding=OpenAIEmbeddings(),\n",
" text_splitter=CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",
")"
]
},
{
"cell_type": "markdown",
"id": "78d8d143",
"metadata": {},
"source": [
"Hopefully this highlights what is going on under the hood of `VectorstoreIndexCreator`. While we think it's important to have a simple way to create indexes, we also think it's important to understand what's going on under the hood."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8b403637",
"id": "dd7257bf",
"metadata": {},
"outputs": [],
"source": []

3
langchain/indexes/__init__.py
View File

@ -1,4 +1,5 @@
"""All index utils."""
from langchain.indexes.graph import GraphIndexCreator
from langchain.indexes.vectorstore import VectorstoreIndexCreator
__all__ = ["GraphIndexCreator"]
__all__ = ["GraphIndexCreator", "VectorstoreIndexCreator"]

37
langchain/indexes/vectorstore.py Normal file
View File

@ -0,0 +1,37 @@
from typing import List, Type
from pydantic import BaseModel, Extra, Field
from langchain.document_loaders.base import BaseLoader
from langchain.embeddings.base import Embeddings
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter, TextSplitter
from langchain.vectorstores.base import VectorStore
from langchain.vectorstores.chroma import Chroma
def _get_default_text_splitter() -> TextSplitter:
return RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
class VectorstoreIndexCreator(BaseModel):
"""Logic for creating indexes."""
vectorstore_cls: Type[VectorStore] = Chroma
embedding: Embeddings = Field(default_factory=OpenAIEmbeddings)
text_splitter: TextSplitter = Field(default_factory=_get_default_text_splitter)
class Config:
"""Configuration for this pydantic object."""
extra = Extra.forbid
arbitrary_types_allowed = True
def from_loaders(self, loaders: List[BaseLoader]) -> VectorStore:
"""Create a vectorstore index from loaders."""
docs = []
for loader in loaders:
docs.extend(loader.load())
sub_docs = self.text_splitter.split_documents(docs)
vectorstore = self.vectorstore_cls.from_documents(sub_docs, self.embedding)
return vectorstore