Open-source embedding models in LangChain - Mini Project: Build & Apply

Using Open-source Embedding Models with Langchain

📖 Scenario: You want to build a simple text search tool using open-source embedding models with Langchain. This tool will convert text into vectors that computers can understand and compare.

🎯 Goal: Build a small Langchain program that loads an open-source embedding model, creates embeddings for a list of texts, and stores them in a vector store.

📋 What You'll Learn

Create a list of texts to embed

Set up the embedding model configuration

Generate embeddings for the texts using the model

Store the embeddings in a vector store

💡 Why This Matters

🌍 Real World

Embedding models help computers understand text by turning words into numbers. This is useful for search engines, chatbots, and recommendation systems.

💼 Career

Knowing how to use open-source embedding models with Langchain is valuable for roles in AI development, data science, and software engineering focused on natural language processing.

1

Create a list of texts to embed

Create a Python list called documents with these exact strings: 'Hello world', 'Langchain is great', and 'Open-source embeddings'.

LangChain

# Create a list called documents with the specified texts
# Your code here

Need a hint?

Use square brackets [] to create a list and separate strings with commas.

2

Set up the embedding model configuration

Import HuggingFaceEmbeddings from langchain.embeddings and create a variable called embedding_model that initializes HuggingFaceEmbeddings with the model name 'sentence-transformers/all-MiniLM-L6-v2'.

LangChain

documents = ['Hello world', 'Langchain is great', 'Open-source embeddings']
# Import HuggingFaceEmbeddings and create embedding_model with the specified model name
# Your code here

Need a hint?

Use the exact model name string and import statement as shown.

3

Generate embeddings for the texts using the model

Create a variable called embeddings and assign it the result of calling embedding_model.embed_documents(documents) to generate embeddings for the documents list.

LangChain

from langchain.embeddings import HuggingFaceEmbeddings

documents = ['Hello world', 'Langchain is great', 'Open-source embeddings']

embedding_model = HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2')
# Generate embeddings for documents and store in embeddings
# Your code here

Need a hint?

Call embed_documents on embedding_model with documents as argument.

4

Store the embeddings in a vector store

Import FAISS from langchain.vectorstores and create a variable called vector_store by calling FAISS.from_texts(documents, embedding_model) to store the documents and their embeddings.

LangChain

from langchain.embeddings import HuggingFaceEmbeddings

documents = ['Hello world', 'Langchain is great', 'Open-source embeddings']

embedding_model = HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2')

embeddings = embedding_model.embed_documents(documents)
# Import FAISS and create vector_store = FAISS.from_texts(documents, embedding_model)
# Your code here

Need a hint?

Use FAISS.from_texts(documents, embedding_model) to create the vector store.