A compact set of Jupyter notebooks that walk from tokenization → embeddings → attention → residual+norm → pooler → MLM with minimal, runnable examples.
Reference Environment
- Python 3.10.18
- PyTorch 2.8.0
- ...
pip install -r requirements.txtClone the repository and run the notebooks step by step:
jupyter notebookPortions of this project are adapted from
bilibili_vlogs.
© Original copyright to the respective author(s). All rights reserved unless otherwise stated.
