HomebrewNLP

HomebrewNLP

Overview

A case study of efficient training of large language models using commodity hardware.

Example Command

python3 main.py train --config_path configs/small.yaml

---

| Discord | WandB

Datasets

• Book Dataset
• 200MB Slice of ThePile

Citing

BibTeX

@misc{nestler2021homebrewnlp,
  title = {{HomebrewNLP}},
  author = {Nestler, Lucas and Gill, David},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub repository},
  doi = {10.5281/zenodo.5553247},
  howpublished = {url{https://github.com/HomebrewNLP/HomebrewNLP}}
}

Latest DOI