Books

Master language models through mathematics, illustrations, and code―and build your own from scratch!

A clear, illustrated guide to large language models, covering key concepts and practical applications. Ideal for projects, interviews, or personal learning.

Build GPT-2, Llama 3, and DeepSeek from scratch in PyTorch. Every chapter has runnable end-to-end code and loads real pretrained weights. Goes well past where most LLM tutorials stop.

Bridge AI and science with this hands-on guide. Whether you're a researcher learning ML or an engineer entering scientific applications, build real systems across chemistry, biology, physics & climate. Master Transformers, Diffusion Models & GNNs for scientific discovery. 500+ pages, 50+ Colab notebooks. Design molecules, predict proteins, accelerate climate models—all hands-on, zero setup required.

It's never been easier to build an AI agent — and never been harder to make one that actually works. This book takes you from language model foundations to production-ready multi-agent systems with the depth to predict failure before it happens, engineer graceful degradation over catastrophic failure, and take absolute architectural ownership. Get the paperback from amazon.

I wanted to understand how ChatGPT and other large language models (LLMs) really work, so I read a lot of books, watched YouTube videos, asked hundreds of questions, and wrote it all down. This book is the result. If you want to understand how large language models like ChatGPT actually work, from tokens and vectors to transformers and training, this book will explain it in a clear, approachable way.

A practical guide to fine-tuning Large Language Models (LLMs), offering both a high-level overview and detailed instructions on how to train these models for specific tasks.Get the paperback version here. Get the Kindle version here.

Revised for PyTorch 2.x! In 2019, I published a PyTorch tutorial on Towards Data Science and I was amazed by the reaction from the readers! Their feedback motivated me to write this book to help beginners start their journey into Deep Learning and PyTorch. I hope you enjoy reading this book as much as I enjoy writing it.

AIエージェントの構築が、これほど容易だった時代はない。そして、実際に機能するものを作ることが、これほど難しい時代もない。本書は言語モデルの基礎から本番対応マルチエージェントシステムまで、失敗が起こる前に予測し、壊滅的な障害ではなく優雅な劣化を設計し、完全なアーキテクチャの所有権を確立するための深さをもって、あなたを導く。ペーパーバック版はamazonにて好評発売中。

Artificial Intelligence is powered by neural networks.But how do neural networks actually learn?How do systems recognize faces, understand speech, generate text, and create images?Neural Networks and Architectures: A Comprehensive Guide for Students provides a complete learning pathway from fundamental concepts to state-of-the-art deep learning architectures.Inside this book, you will discover:✓ Biological inspiration behind neural networks✓ Mathematical foundations of deep learning✓ Perceptrons and Multi-Layer Perceptrons (MLPs)✓ Forward Propagation and Backpropagation✓ Activation Functions and Regularization Techniques✓ Convolutional Neural Networks (CNNs)✓ Recurrent Neural Networks (RNNs), LSTM, and GRU✓ Autoencoders and Generative Adversarial Networks (GANs)✓ Transformer Architecture, BERT, and GPT✓ Optimization Algorithms and Model Training✓ Practical Projects using TensorFlow, PyTorch, and Keras✓ Ethical Considerations and Future Research DirectionsDesigned for students, educators, researchers, and AI enthusiasts, this book combines theory, mathematics, implementation, and applications into one comprehensive learning resource.Build a strong foundation in neural networks and prepare yourself for the future of Artificial Intelligence.

How does a machine recognize a face?How can AI distinguish speech from noise?Why do modern computer vision systems still rely on mathematical techniques developed decades ago?The answer lies in Fourier and Wavelet Analysis.In Fourier and Wavelet Analysis in Artificial Intelligence, Anshuman Mishra reveals how frequency-domain representations, multi-resolution analysis, and signal-processing techniques continue to shape the future of Machine Learning, Deep Learning, Computer Vision, Speech Recognition, Biomedical AI, and Edge Intelligence.From Fourier Transforms and Fast Fourier Algorithms to Wavelet Scattering Networks and Hybrid CNN Architectures, this book demonstrates how mathematical signal analysis becomes intelligent feature extraction.Discover the mathematics behind perception, representation, and intelligent decision-making.

Can a neural network be viewed as a differential equation?Why does gradient descent behave like a dynamical system?How do biological neurons inspire modern AI architectures?What mathematical principles govern stability, learning, adaptation, and intelligence?In Differential Equations in AI and Neural Dynamics, Anshuman Mishra explores the mathematical framework that underlies modern Artificial Intelligence.From Ordinary Differential Equations and Neural Population Models to Neural ODEs, Stochastic Learning, Reinforcement Learning, and Brain-Inspired Computation, this book reveals how continuous-time mathematics drives intelligent behavior.Discover how equations of change become equations of intelligence.

Prediction is only the beginning.Modern AI systems must explain their predictions, quantify uncertainty, scale to massive datasets, and operate reliably in real-world environments.How do Support Vector Regression models capture nonlinear patterns?How do Bayesian methods estimate uncertainty?How do Gaussian Processes make predictions with confidence intervals?How can complex AI models remain interpretable and trustworthy?In this advanced second volume, Anshuman Mishra explores the cutting-edge regression techniques that power intelligent prediction systems across machine learning, data science, healthcare, finance, robotics, and scientific research.Discover how modern Artificial Intelligence transforms data into reliable, explainable, and scalable predictions.

Artificial Intelligence does not learn in certainty.It learns through uncertainty, exploration, randomness, and adaptation.How does AlphaGo evaluate millions of possible moves?How do reinforcement learning agents discover optimal strategies?How do diffusion models generate realistic images?How do autonomous robots navigate uncertain environments?The answer lies in stochastic processes.In this advanced second volume, Anshuman Mishra explores the mathematical foundations behind reinforcement learning, probabilistic deep learning, robotics, generative AI, and emerging stochastic algorithms that are shaping the future of intelligent systems.Discover how randomness becomes learning—and how uncertainty becomes intelligence.

Artificial Intelligence is often described as learning from data.But beneath every learning algorithm lies something even more fundamental:Probability, randomness, and uncertainty.From Hidden Markov Models and stochastic gradient descent to Monte Carlo methods and reinforcement learning, modern AI systems depend on stochastic processes to make predictions, learn from experience, and adapt to changing environments.In Stochastic Processes in Artificial Intelligence, Anshuman Mishra provides a structured and accessible journey through the mathematical foundations that power intelligent systems.Discover how uncertainty becomes intelligence—and why stochastic thinking is essential for the future of Artificial Intelligence.