Learn AIML — tinygrad prep

A start-to-finish reading path for understanding and contributing to tinygrad.

This page uses only legal links: official free versions, open resources, publisher pages, and library/borrowing routes.

Recommended sequence

Understanding Deep Learning
From Python to NumPy
Grokking Deep Learning
The Matrix Calculus You Need for Deep Learning
micrograd
Dive into Deep Learning
tinygrad
Computer Systems: A Programmer's Perspective
Programming Massively Parallel Processors

Start-to-end path

1. Understanding Deep Learning — Simon J. D. Prince

read first

Purpose: theory, backprop, neural nets, optimization.
Official free site: https://udlbook.github.io/udlbook/

2. From Python to NumPy — Nicolas P. Rougier

read early

Purpose: arrays, vectorization, shapes, tensor thinking.
Official online version: https://www.labri.fr/perso/nrougier/from-python-to-numpy/
Repository: https://github.com/rougier/from-python-to-numpy

3. Grokking Deep Learning — Andrew Trask

implementation bridge

Purpose: code-first intuition and implementation mindset.
Publisher: https://www.manning.com/books/grokking-deep-learning
Library search: https://search.worldcat.org/

4. The Matrix Calculus You Need for Deep Learning

reference / as needed

Purpose: matrix derivatives and gradient shape intuition.
arXiv: https://arxiv.org/abs/1802.01528

5. micrograd — Andrej Karpathy

must do

Purpose: autodiff, computational graphs, minimal backprop code.
Repository: https://github.com/karpathy/micrograd

6. Dive into Deep Learning

selective reference

Purpose: practical modern DL context and implementation reference.
Official site: https://d2l.ai/

7. tinygrad

start here after 1-5

Purpose: read source, run examples, trace ops, start contributing.
Repository: https://github.com/tinygrad/tinygrad

8. Computer Systems: A Programmer's Perspective

later

Purpose: systems/performance intuition.
Official site: https://csapp.cs.cmu.edu/
Library search: https://search.worldcat.org/

9. Programming Massively Parallel Processors

later / backend work

Purpose: GPU execution model, memory hierarchy, kernels.
Publisher: https://www.elsevier.com/books/programming-massively-parallel-processors/kirk/978-0-323-91231-0
Library search: https://search.worldcat.org/

Shortest practical path

Understanding Deep Learning
From Python to NumPy
Grokking Deep Learning
micrograd
tinygrad

Generated for Parth's tinygrad learning path.