book

Hugging Face in Action

Name: Hugging Face in Action
Author: Wei-Meng Lee
ISBN: 9781633436718

by Wei-Meng Lee

October 2025

Intermediate to advanced

368 pages

9h 37m

English

Manning Publications

Read now

Unlock full access

Hugging Face in Action
copyright
contents
preface
acknowledgments
about this book
about the author
about the cover illustration
1 Introducing Hugging Face
1.1 Hugging Face Transformers library1.2 Hugging Face models1.3 Hugging Face Gradio Python library1.4 Understanding the Hugging Face mental model1.4.1 Step 1: User need1.4.2 Step 2: Model Hub discovery1.4.3 Step 3: Model card1.4.4 Step 4: Two execution paths1.4.5 Step 5: Results delivered
2 Getting started
2.1 Downloading Anaconda2.1.1 Creating virtual environments2.1.2 Starting Jupyter Notebook2.2 Installing the Transformers library2.2.1 Support for GPU2.2.2 Using GPU in the pipeline object2.3 Installing the Hugging Face Hub package2.3.1 Downloading files2.3.2 Using the Hugging Face CLI

3 Using Hugging Face transformers and pipelines for NLP tasks
3.1 Introduction to the transformer architecture3.1.1 Tokenization3.1.2 Token embeddings3.1.3 Positional encoding3.1.4 Transformer block3.1.5 Softmax3.2 Working with the Transformers library3.2.1 What are pretrained transformers models?3.2.2 What are transformers pipelines?3.2.3 Using a model directly3.2.4 Using a transformers pipeline3.3 Using transformers for NLP tasks3.3.1 Text classification3.3.2 Text generation3.3.3 Text summarization3.3.4 Text translation3.3.5 Zero-shot classification3.3.6 Question-answering tasks
4 Using Hugging Face for computer vision tasks
4.1 Hugging Face computer vision models4.2 Object detection4.2.1 Using the model directly4.2.2 Using the transformers pipeline4.2.3 Binding to a webcam4.3 Image classification4.4 Image segmentation4.4.1 Using the model programmatically4.4.2 Binding to Gradio4.5 Video classification4.5.1 Installing the prerequisites4.5.2 Downloading the videos for testing4.5.3 Using the transformers pipeline object
5 Exploring, tokenizing, and visualizing Hugging Face datasets
5.1 What are Hugging Face datasets?5.1.1 Getting the list of datasets available5.1.2 Validating the availability of a dataset5.1.3 Downloading a dataset5.1.4 Shuffling a dataset5.1.5 Streaming a dataset5.1.6 Getting the Parquet files of a dataset5.2 Tokenization in NLP5.2.1 Types of tokenization methods5.2.2 Tokenizing datasets5.3 Visualizing datasets5.3.1 Using the twitter-financial-news-topic dataset5.3.2 Using the CIFAR-10 dataset
6 Fine-tuning pretrained models and working with multimodal models
6.1 Fine-tuning pretrained models6.1.1 Loading the yelp_polarity dataset6.1.2 Filtering the yelp_polarity dataset6.1.3 Tokenizing the reduced dataset6.1.4 Setting up a pretrained model for sequence classification6.1.5 Configuring and initializing a trainer for fine-tuning a pretrained model6.1.6 Using the fine-tuned model6.1.7 Fine-tuning models for multiclass text classification6.2 Working with multimodal models6.2.1 Single-modal models6.2.2 Multimodal models
7 Creating LLM-based applications using LangChain and LlamaIndex
7.1 Introducing LLMs7.2 Introducing LangChain7.2.1 Installing LangChain7.2.2 Creating a prompt template7.2.3 Specifying an LLM7.2.4 Creating an LLM chain7.2.5 Running the chain7.2.6 Maintaining a conversation7.2.7 Using the RunnableWithMessageHistory class7.2.8 Using other LLMs7.3 Connecting LLMs to your private data using LlamaIndex7.3.1 Installing the packages7.3.2 Preparing the documents7.3.3 Loading the documents7.3.4 Using an embedding model7.3.5 Indexing the document7.3.6 Loading the embeddings7.3.7 Using an LLM for querying7.3.8 Asking questions7.3.9 Using LlamaIndex with OpenAI7.3.10 Creating a web frontend for the app7.3.11 Holding a conversation7.3.12 Creating a chatbot UI
8 Building LangChain applications visually using Langflow
8.1 What is Langflow?8.1.1 Installing Langflow using the pip command8.1.2 Installing Langflow using Docker8.1.3 Running Langflow in the cloud8.2 Creating a new Langflow project8.2.1 Adding a Prompt component8.2.2 Adding a Models component8.2.3 Adding a Chains component8.2.4 Adding Chat Input and Chat Output components8.2.5 Testing the project8.2.6 Maintaining a conversation using the Chat Memory component8.3 Asking questions on your own data8.3.1 Loading PDF documents using the File component8.3.2 Splitting long text into smaller chunks using the Parse Data component8.3.3 Getting questions using the Prompt component8.3.4 Using the HuggingFace component8.3.5 Connecting to the Chat Output component8.3.6 Testing the project8.3.7 Using an LLM with the OpenAI component8.4 Using your project programmatically8.4.1 cURL8.4.2 Python code
9 Programming agents
9.1 What are agents?9.2 Developing agents using smolagents9.2.1 Using built-in tools: DuckDuckGoSearchTool9.2.2 Using built-in tools: PythonInterpreterTool9.2.3 Writing your own custom tools9.3 Developing agents with LangChain9.3.1 Using the built-in Tool class9.3.2 Using custom tools9.4 Developing agents using LangGraph9.4.1 What is LangGraph?9.4.2 LangGraph agent basics9.4.3 Using LangGraph with tools9.4.4 Using LangGraph with a custom tool9.4.5 Using LangGraph with memory
10 Building a web-based UI using Gradio
10.1 Basics of Gradio10.1.1 Using Gradio’s Interface class10.1.2 Configuring flagging options10.1.3 Configuring authentication10.1.4 Customizing the server and port10.1.5 Sharing your Gradio application10.1.6 Deploying your Gradio application to Hugging Face Spaces10.2 Working with widgets10.2.1 Working with Textbox10.2.2 Working with Audio10.2.3 Working with Images10.2.4 Working with selection widgets10.2.5 Layout using the TabbedInterface class10.3 Creating a chatbot UI10.3.1 Creating the basic chatbot UI10.3.2 Wiring the Textbox’s submit event10.3.3 Clearing the chatbot
11 Building locally running LLM-based applications using GPT4All
11.1 Introducing GPT4All11.2 Installing GPT4All11.2.1 Installing the GPT4All application11.2.2 Installing the gpt4all Python library11.2.3 Listing all supported models11.2.4 Loading a specific model11.2.5 Asking a question11.2.6 Binding with Gradio
12 Using LLMs to query your local data
12.1 Using GPT4All to query with your own data12.1.1 Installing the required packages12.1.2 Importing the various modules from the LangChain package12.1.3 Loading the PDF documents12.1.4 Splitting the text into chunks12.1.5 Embedding12.1.6 Loading the embeddings12.1.7 Downloading the model12.1.8 Asking questions12.1.9 Loading multiple documents12.1.10 Loading CSV files12.1.11 Loading JSON files12.2 Using LLMs to write code to analyze your data12.2.1 Preparing the JSON file12.2.2 Loading the JSON file12.2.3 Asking the question using the Mistral 7B model12.2.4 Asking questions using OpenAI
13 Bridging LLMs to the real world with the Model Context Protocol
13.1 What is MCP?13.1.1 The problems MCP solves13.1.2 Understanding MCP13.1.3 MCP server deployment13.1.4 Components in an MCP server13.2 Building an MCP server13.2.1 Installing uv13.2.2 Initializing the project13.2.3 Installing the packages13.2.4 Creating the MCP server13.2.5 Inspecting the MCP server13.2.6 Implementing Resources13.2.7 Implementing Tools13.2.8 Implementing a prompt13.2.9 Testing the components13.3 Testing the MCP server using Claude Desktop13.3.1 Configuring Claude Desktop to use the MCP server13.3.2 Getting the weather13.3.3 Getting the content of a text file13.3.4 Getting the content of a PDF file13.3.5 Improving the MCP server13.4 Trying third-party MCP servers13.4.1 Get My Location13.4.2 mcp-datetime

Content preview from Hugging Face in Action

13 Bridging LLMs to the real world with the Model Context Protocol

This chapter covers

Introducing Model Context Protocol (MCP)
Developing your own MCP server
Using an MCP server with Claude Desktop
Using third-party MCP servers

As large language models (LLMs) become more advanced, developers face a key challenge: making it easier for these models to work with external data that wasn’t part of their original training. Right now, connecting LLMs to different types of data (such as files, websites, or live social media feeds) often requires a custom solution for each source, adding work and complexity.

To solve this problem, a new framework called the Model Context Protocol (MCP) was introduced. MCP provides a standard way for LLMs to ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781633436718Publisher Support Publisher Website

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Hugging Face in Action

by Wei-Meng Lee