Knowledge Distillation of Black-Box Large Language Models

Researchers introduced Proxy-KD, a method for distilling knowledge from black-box large language models (LLMs) like GPT-4 into smaller models without accessing internal states. Proxy-KD uses a proxy model to facilitate knowledge transfer, outperforming traditional white-box KD techniques. This approach offers a new way to leverage proprietary LLMs for improving smaller models.

Computer Science Computation and Language Submitted on 13 Jan 2024 v1 https://arxiv.org/abs/2401.07013v1 , last revised 9 Nov 2024 this version, v2 Title:Knowledge Distillation of Black-Box Large Language Models View PDF /pdf/2401.07013 HTML experimental https://arxiv.org/html/2401.07013v2 Abstract:Given the exceptional performance of proprietary large language models LLMs like GPT-4, recent research has increasingly focused on boosting the capabilities of smaller models through knowledge distillation KD from these powerful yet black-box teachers. While leveraging the high-quality outputs of these teachers is advantageous, the inaccessibility of their internal states often limits effective knowledge transfer. To overcome this limitation, we introduce Proxy-KD, a novel method that uses a proxy model to facilitate the efficient transfer of knowledge from black-box LLMs to smaller models. Our experiments show that Proxy-KD not only enhances the performance of KD from black-box teacher models but also surpasses traditional white-box KD techniques.~This approach presents a compelling new avenue for distilling knowledge from advanced LLMs. Submission history From: Hongzhan Chen view email /show-email/794667b0/2401.07013 Sat, 13 Jan 2024 08:43:32 UTC 359 KB v1 /abs/2401.07013v1 v2 Sat, 9 Nov 2024 01:35:32 UTC 8,288 KB References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .