Lion: Adversarial Distillation of Proprietary Large Language Models

Yuxin Jiang, Chunkit Chan, Mingyang Chen, Wei Wang


Abstract
The practice of transferring knowledge from a sophisticated, proprietary large language model (LLM) to a compact, open-source LLM has garnered considerable attention. Previous works have focused on a unidirectional knowledge distillation way by aligning the responses of the student model with those of the teacher model to a set of instructions. Nevertheless, they overlooked the possibility of incorporating any “feedback”–identifying challenging instructions where the student model’s performance falls short–to boost the student model’s proficiency iteratively. To this end, we propose a novel adversarial distillation framework for a more efficient knowledge transfer. Leveraging the versatile role adaptability of LLMs, we prompt the teacher model to identify “hard” instructions and generate new “hard” instructions for the student model, creating a three-stage adversarial loop of imitation, discrimination, and generation. By applying this adversarial framework, we successfully transfer knowledge from ChatGPT to a student model (named Lion), using a mere 70k training data. Our results show that Lion-13B not only achieves comparable open-ended generation capabilities to ChatGPT but surpasses conventional state-of-the-art (SOTA) instruction-tuned models like Vicuna-13B by 55.4% in challenging zero-shot reasoning benchmarks such as BIG-Bench Hard (BBH) and 16.7% on AGIEval.
Anthology ID:
2023.emnlp-main.189
Volume:
Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
Month:
December
Year:
2023
Address:
Singapore
Editors:
Houda Bouamor, Juan Pino, Kalika Bali
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3134–3154
Language:
URL:
https://aclanthology.org/2023.emnlp-main.189
DOI:
10.18653/v1/2023.emnlp-main.189
Bibkey:
Cite (ACL):
Yuxin Jiang, Chunkit Chan, Mingyang Chen, and Wei Wang. 2023. Lion: Adversarial Distillation of Proprietary Large Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 3134–3154, Singapore. Association for Computational Linguistics.
Cite (Informal):
Lion: Adversarial Distillation of Proprietary Large Language Models (Jiang et al., EMNLP 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.emnlp-main.189.pdf
Video:
 https://aclanthology.org/2023.emnlp-main.189.mp4