Better Alignment with Instruction Back-and-Forth Translation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914911721881600 |
|---|---|
| author | Nguyen, Thao Li, Jeffrey Oh, Sewoong Schmidt, Ludwig Weston, Jason Zettlemoyer, Luke Li, Xian |
| author_facet | Nguyen, Thao Li, Jeffrey Oh, Sewoong Schmidt, Ludwig Weston, Jason Zettlemoyer, Luke Li, Xian |
| contents | We propose a new method, instruction back-and-forth translation, to construct high-quality synthetic data grounded in world knowledge for aligning large language models (LLMs). Given documents from a web corpus, we generate and curate synthetic instructions using the backtranslation approach proposed by Li et al.(2023a), and rewrite the responses to improve their quality further based on the initial documents. Fine-tuning with the resulting (backtranslated instruction, rewritten response) pairs yields higher win rates on AlpacaEval than using other common instruction datasets such as Humpback, ShareGPT, Open Orca, Alpaca-GPT4 and Self-instruct. We also demonstrate that rewriting the responses with an LLM outperforms direct distillation, and the two generated text distributions exhibit significant distinction in embedding space. Further analysis shows that our backtranslated instructions are of higher quality than other sources of synthetic instructions, while our responses are more diverse and complex than those obtained from distillation. Overall we find that instruction back-and-forth translation combines the best of both worlds -- making use of the information diversity and quantity found on the web, while ensuring the quality of the responses which is necessary for effective alignment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_04614 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Better Alignment with Instruction Back-and-Forth Translation Nguyen, Thao Li, Jeffrey Oh, Sewoong Schmidt, Ludwig Weston, Jason Zettlemoyer, Luke Li, Xian Computation and Language Artificial Intelligence Machine Learning We propose a new method, instruction back-and-forth translation, to construct high-quality synthetic data grounded in world knowledge for aligning large language models (LLMs). Given documents from a web corpus, we generate and curate synthetic instructions using the backtranslation approach proposed by Li et al.(2023a), and rewrite the responses to improve their quality further based on the initial documents. Fine-tuning with the resulting (backtranslated instruction, rewritten response) pairs yields higher win rates on AlpacaEval than using other common instruction datasets such as Humpback, ShareGPT, Open Orca, Alpaca-GPT4 and Self-instruct. We also demonstrate that rewriting the responses with an LLM outperforms direct distillation, and the two generated text distributions exhibit significant distinction in embedding space. Further analysis shows that our backtranslated instructions are of higher quality than other sources of synthetic instructions, while our responses are more diverse and complex than those obtained from distillation. Overall we find that instruction back-and-forth translation combines the best of both worlds -- making use of the information diversity and quantity found on the web, while ensuring the quality of the responses which is necessary for effective alignment. |
| title | Better Alignment with Instruction Back-and-Forth Translation |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2408.04614 |