Self-Alignment with Instruction Backtranslation
Xian Li, Ping Yu, Chunting Zhou, Timo Schick, Omer Levy, Luke Zettlemoyer, Jason E Weston, Mike Lewis
Abstract
We present a scalable method to build a high quality instruction following language model by automatically labelling human-written text with corresponding instructions. Our approach, named instruction backtranslation, starts with a language model finetuned on a small amount of seed data, and a given web corpus. The seed model is used to construct training examples by generating instruction prompts for web documents (self-augmentation), and then selecting high quality examples from among these candidates (self-curation). This data is then used to finetune a stronger model. Finetuning LLaMa on two iterations of our approach yields a model that outperforms all other LLaMa-based models on the Alpaca leaderboard not relying on distillation data, demonstrating highly effective self-alignment.
BibTeX
@inproceedings{
li2024selfalignment,
title={Self-Alignment with Instruction Backtranslation},
author={Xian Li and Ping Yu and Chunting Zhou and Timo Schick and Omer Levy and Luke Zettlemoyer and Jason E Weston and Mike Lewis},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=1oijHJBRsT}
}