LLMs From Scratch - Sebastian Raschka
Code an LLM from zero: tokens → attention → training
Build a GPT-style large language model step by step to truly understand what happens inside an LLM: tokenization, embeddings, attention mechanisms, the transformer architecture, pretraining loops, and fine-tuning. Written by Sebastian Raschka; pairs with his book of the same name. For engineers who want depth, not just API calls.
Learn from ReposFreeadvanced~40h
Submitted by @yahianaimRoadmap (3 steps)
1
Foundations
Ch. 1-3#PyTorch refresher#Tokenization & embeddings#Attention mechanism
2
Architecture
Ch. 4-5#GPT transformer from scratch#Pretraining loop#Evaluation
3
Adaptation
Ch. 6-7#Fine-tuning classification#Instruction fine-tuning#Loading pretrained weights