LLMs-from-scratch-CN
A Chinese translation of the open-source "LLMs-from-scratch" book and code, teaching how to build, pretrain, and fine-tune a GPT-style language model from scratch.
LLMs-from-scratch-CN is a Chinese translation of an open-source book and code repository that teaches how large language models work by building one from the ground up. It follows the original chapter by chapter, adding more detailed Chinese-language code comments and supplementary material aimed at Chinese-speaking learners, without altering the original's technical content.
The material covers building a GPT-like model, pretraining it, and fine-tuning it, including loading larger pretrained weights, and is provided as Markdown notes alongside Jupyter notebooks with runnable code for each chapter. It links back to the original English repository and book for readers who prefer the source material.
It is distributed as Jupyter notebooks and Markdown documentation.