Tools
Where you write and run code.
The libraries models are built with.
The dominant deep learning framework; the one this course uses.
NumPy with autodiff and compilation; popular for research at scale.
Classical ML done right: models, pipelines, metrics.
Load, run and fine-tune almost any open model.
SFT, DPO and GRPO trainers for post-training.
Fast, memory-efficient LoRA fine-tuning on one GPU.
Interactive visualizations that build intuition.
Watch GPT-2 compute the next token, step by step.
A 3-D walk through every operation inside a small GPT.
Train a tiny neural net in the browser and see it learn.
Open any model file and inspect its layers and shapes.
Plot losses, histograms and embeddings during training.
Track runs, share models and data.
Find, read and connect research.
From a laptop to a GPU server.
Competitions, exercises and courses.
Real datasets, leaderboards and winning write-ups.
LeetCode-style problems for ML and linear algebra.
Karpathy builds GPT from scratch, on video.
Practical deep learning for coders.
Short courses and specializations on GenAI and ML.