Advanced RAG Techniques: Choosing the Right Approach/

...

/

Query Decomposition for Better Precision

Query Decomposition for Better Precision

Learn about the decomposition technique, how it works, and its step-by-step implementation.

We'll cover the following...

In RAG tasks, we often encounter complex questions that require in-depth analysis and the gathering of information from various sources. This is where decomposition comes into play. Decomposition is a powerful technique that breaks down a large, intricate problem into smaller, more manageable sub-problems. By addressing these sub-problems independently, we can simplify the overall task and ultimately create a more comprehensive and accurate response.

What is decomposition?

In the context of RAG, decomposition involves dividing a primary question into a series of smaller, more focused sub-questions. Each sub-question can be answered independently, and the answers are then combined to form a comprehensive response to the original question. This approach offers several advantages:

  • Enhanced efficiency: By tackling smaller sub-problems, the retrieval and generation processes become more efficient as the system focuses on specific aspects of the main question.

  • Improved accuracy: Decomposing the question allows for a deeper exploration of each sub-question, potentially leading to more accurate and relevant answers.

  • Structured response: Decomposition facilitates the organization of the final answer by presenting the sub-questions and their corresponding answers in a clear, structured format.

Press + to interact
High-level overview of the decomposition technique
High-level overview of the decomposition technique

Step-by-step implementation

The following are the steps to implement the decomposition technique:

Press + to interact
Steps for implementing decomposition
Steps for implementing decomposition

1. Import necessary modules

We’ll import the required modules from the installed libraries to implement multi-query:

Press + to interact
import os
import bs4
from langchain_community.document_loaders import WebBaseLoader
from langchain.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_openai import ChatOpenAI
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_openai import OpenAIEmbeddings
from langchain_community.vectorstores import Chroma
from langchain import hub
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough, RunnableLambda

These libraries and modules are essential for the subsequent steps in the process.

2. Set up the LangSmith and OpenAI API keys

The following code snippet sets up your LangChain API key and OpenAI API key from environment variables. We’ll need valid API keys to interact with the LangChain and OpenAI language models:

Press + to interact
os.environ['LANGCHAIN_TRACING_V2'] = 'true'
os.environ['LANGCHAIN_ENDPOINT'] = 'https://api.smith.langchain.com'
os.environ['LANGCHAIN_API_KEY'] = '' # Add your LangSmith LangChain API key
os.environ['LANGCHAIN_PROJECT']='Decomposition'
OPENAI_API_KEY = os.environ["OPENAI_API_KEY"] = ""  # Add your OpenAI API key
if OPENAI_API_KEY == "":
    raise ValueError("Please set the OPENAI_API_KEY environment variable")

Code explanation

  • Lines 1–4: Sets up the LangChain environment variables:

    • LANGCHAIN_TRACING_V2: Enables tracing for LangChain operations.

    • LANGCHAIN_ENDPOINT: Specifies the endpoint for the LangChain API. ...