Back: Catalog

how-to-train-your-gpt

This project is a 12-part interactive textbook about building, training and running a modern language model from scratch. It offers examples of code that are interpreted each line containing the basic components of the Transformer architecture. Tokenizer explains issues such as burials, attention mechanisms, and training cycles by clearing them of technical terms.

HO

raiyanyahya · Local AI / Self-Hosted LLM

STARS4.2K
FORKS505
COMMITS0
CONTRIBUTORS0
VERSION—
UPDATED4 days ago
Jupyter NotebookLLMPythonTransformerMIT

Common commercial products in this category: ChatGPT, Claude, Gemini

What can it do?
  • You build and run your own language model from scratch.
  • Tokenizer, you learn by coding the layers of burial and the mechanisms of attention.
  • You study technical issues such as RoPE and RMSNorm in detail.
Heads up
  • • No published release
GitHub

My note

Alternative pages

More in: Local AI / Self-Hosted LLM