Start here
For readers refreshing machine-learning foundations, studying deep learning, and implementing a small language model. Choose an entry point that fits your background; order, levels, prerequisites, and learning goals are editorial guidance, not a curriculum endorsed by the recommenders.
Editorial starting points from From The Greats. Level, prerequisites, learning goals and reading order are our guidance, not statements attributed to the recommenders. Inspect their actual recommendation evidence below.
Editorial guidance
Refresh the mathematics that supports machine learning before choosing a deeper technical text.
Level: Prerequisite refresher
Prerequisites: Editorial suggestion: some prior algebra and calculus; supplement unfamiliar topics with exercises and additional explanations.
Learning goal: Connect linear algebra, calculus, probability, and optimization to machine-learning methods.
Caveats: Abbeel’s recommendation is an undated author-hosted endorsement of the 2020 work. It does not establish endorsement of every later online PDF revision. This refresher may require supplementary resources if the mathematics is new to you.
Recommended by Pieter Abbeel — recommendation evidence.
Open the bookEditorial guidance
Start with the beginner resource Karpathy recommended in 2015.
Level: Foundations
Prerequisites: Editorial suggestion: basic programming and comfort with algebra; refresh calculus and vectors as needed.
Learning goal: Build an initial understanding of neural networks before moving to denser mathematical treatments.
Caveats: Karpathy’s recommendation dates to 2015. The original companion code targets Python 2.6/2.7 and old Theano versions; it is not a current software setup. The author’s repository was inspected; the linked book website was unavailable during this review.
Recommended by Andrej Karpathy — recommendation evidence.
Open the bookEditorial guidance
Learn through mathematical explanations and hands-on deep-learning notebooks.
Level: Practical deep learning
Prerequisites: Editorial suggestion: Python, basic linear algebra, derivatives, and probability; the book includes introductory material for review.
Learning goal: Connect deep-learning concepts to code and experiments, including attention and transformers.
Caveats: Huang’s recommendation is an undated author-hosted endorsement. The online book and its framework implementations change over time; no specific later revision is asserted as his endorsed edition. Notebook dependencies and exercises have not been tested as part of this catalog review.
Recommended by Jensen Huang — recommendation evidence.
Open the bookEditorial guidance
Move to an advanced mathematical textbook when you are ready for more depth.
Level: Advanced textbook
Prerequisites: Editorial suggestion: working knowledge of linear algebra, probability, calculus, and programming; Part I offers a mathematics and machine-learning refresher.
Learning goal: Study the foundations, training methods, and research topics of deep learning in a sustained technical treatment.
Caveats: Karpathy recommended the work in progress in October 2015, before the final 2016 edition. It is not a guide to current LLM practice; no endorsement of later revisions is implied.
Recommended by Andrej Karpathy — recommendation evidence, Sebastian Raschka — recommendation evidence.
Open the bookEditorial guidance
Implement a small language model after building comfort with Python and machine-learning basics.
Level: LLM implementation
Prerequisites: Editorial suggestion: Python and basic machine-learning concepts; be ready to work through PyTorch code and tensor operations.
Learning goal: Understand a GPT-style model by implementing its components and working through pretraining and fine-tuning.
Caveats: Huyen’s recommendation is a publisher-hosted endorsement of the 2024 book. Its companion code uses Python and PyTorch and has not been tested in this catalog review. The exercises build an educational model; they are not a complete path to training a frontier model.
Recommended by Chip Huyen — recommendation evidence.
Open the book
From The Greats
