AI Line
AI From Zero: Build a Neural Network with NumPy
Rated 4.7 out of 5 from 3 reviewsBeginner 18 lessons 503 students
Fare
€21All lines running
Coding preview on every course · Lifetime access · 14-day refund
Build and train your own transformer language model in PyTorch, from tokenizer to text generation.
The route
5 modules · 21 lessons · 4h 3m of video
Tokenizers
4 lessons
Attention from scratch
5 lessons
The transformer block
4 lessons
Training at small scale
5 lessons
Sampling and evaluation
3 lessons
Destination
Before you board
About this course
The best way to understand language models is to build one. You write every part yourself in PyTorch: a tokenizer, embeddings, self-attention, transformer blocks, the training loop and sampling.
Then you train a small model on a public-domain text collection on a single GPU or a free cloud notebook, measure it, and generate text with it. Nothing is hidden behind a library call.
Still unsure about something? Write to our contact page and a human answers, usually the same working day.
Nobody has reviewed this course yet, so there is no score to show. Buy it, work through it, and your review could be the one that helps the next developer decide.
One-time payment · lifetime access