Q&A over Documents

LangChain Basics 4

Jan Kirenz

LangChain: Q&A over Documents

Let’s build a tool that would allow you to query a product catalog for items of interest.

Setup

Python

from langchain.embeddings import OpenAIEmbeddings
from langchain.indexes import VectorstoreIndexCreator
from IPython.display import display, Markdown
from langchain.vectorstores import DocArrayInMemorySearch  # vector store
from langchain.document_loaders import CSVLoader
from langchain.chat_models import ChatOpenAI
from langchain.chains import RetrievalQA
import datetime
import os

from dotenv import load_dotenv, find_dotenv
_ = load_dotenv(find_dotenv())  # read local .env file
# pip install --upgrade langchain

Simple example

Load Data

file = '../data/OutdoorClothingCatalog_1000.csv'
loader = CSVLoader(file_path=file)
  • Create vector store
index = VectorstoreIndexCreator(
    vectorstore_cls=DocArrayInMemorySearch
).from_loaders([loader])

Query and response

query = "Please list all your shirts with sun protection \
in a table in markdown and summarize each one."
response = index.query(query)
display(Markdown(response))

Query and response

Step By Step

Basics

  • Language models ca only inspect a few thousands word at a time

  • This is why we need embeddings and vector stores

Loader

  • We use our loader from before (loader = CSVLoader(file_path=file)
docs = loader.load()
docs[0]
  • Document(page_content=“: 0: Women’s Campside Oxfords: This ultracomfortable lace-to-toe Oxford boasts a super-soft canvas, thick cushioning, and quality construction for a broken-in feel from the first time you put them on. & Fit: Order regular shoe size. For half sizes not offered, order up to next whole size. : Approx. weight: 1 lb.1 oz. per pair. : Soft canvas material for a broken-in feel and look. Comfortable EVA innersole with Cleansport NXT® antimicrobial odor control. Vintage hunt, fish and camping motif on innersole. Moderate arch contour of innersole. EVA foam midsole for cushioning and support. Chain-tread-inspired molded rubber outsole with modified chain-tread pattern. Imported. ? Please contact us for any inquiries.”, metadata={‘source’: ‘../data/OutdoorClothingCatalog_1000.csv’, ‘row’: 0})

Embeddings

  • Our documents are so small that we dont need to chunck them first
embeddings = OpenAIEmbeddings()
embed = embeddings.embed_query("Hi my name is Jan")
print(len(embed))
  • 1536
print(embed[:5])
  • [-0.015501204878091812, -0.0016401495086029172, -0.01953849568963051, -0.016909271478652954, -0.021893581375479698]

Create vector store

db = DocArrayInMemorySearch.from_documents(
    docs,
    embeddings
)

Query

query = "Please suggest a shirt with sunblocking"
docs = db.similarity_search(query)
len(docs)
  • 4
docs[0]
  • Document(page_content=‘: 255: Sun Shield Shirt by: “Block the sun, not the fun – our high-performance sun shirt is guaranteed to protect from harmful UV rays. & Fit: Slightly Fitted: Softly shapes the body. Falls at hip.& Care: 78% nylon, 22% Lycra Xtra Life fiber. UPF 50+ rated – the highest rated sun protection possible. Handwash, line dry.Features: Wicks moisture for quick-drying comfort. Fits comfortably over your favorite swimsuit. Abrasion resistant for season after season of wear. Imported.Protection That Won't Wear Offhigh-performance fabric provides SPF 50+ sun protection, blocking 98% of the sun's harmful rays. This fabric is recommended by The Skin Cancer Foundation as an effective UV protectant.’, metadata={‘source’: ‘../data/OutdoorClothingCatalog_1000.csv’, ‘row’: 255})

Manual Example

llm_model = "gpt-3.5-turbo"
llm = ChatOpenAI(temperature=0.0, model=llm_model)
  • If we would do it manually, we would combine the text with a join
qdocs = "".join([docs[i].page_content for i in range(len(docs))])
  • And then we would provide qdocs to the llm:
# This may take a while
response = llm.call_as_llm(f"{qdocs} Question: Please list all your \
shirts with sun protection in a table in markdown and summarize each one.")
display(Markdown(response))

XYZ Error

Retrieval Chain

Chain

retriever = db.as_retriever()
qa_stuff = RetrievalQA.from_chain_type(
    llm=llm,
    chain_type="stuff",  # one propmt and one result
    retriever=retriever,
    verbose=True
)

Query

query = "Please list all your shirts with sun protection in a table \
in markdown and summarize each one."
response = qa_stuff.run(query)
display(Markdown(response))
response = index.query(query, llm=llm)

Acknowledgments

This tutorial is mainly based on the excellent course “LangChain for LLM Application Development” provided by Harrison Chase and Andrew Ng

What’s next?

Congratulations! You have completed this tutorial 👍

Next, you may want to go back to the lab’s website

Jan Kirenz