HuatuoGPT-II, One-stage Training for Medical Adaption of LLMs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Chen, Junying, Wang, Xidong, Ji, Ke, Gao, Anningzhe, Jiang, Feng, Chen, Shunian, Zhang, Hongbo, Song, Dingjie, Xie, Wenya, Kong, Chuyi, Li, Jianquan, Wan, Xiang, Li, Haizhou, Wang, Benyou
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866929499818426368
author Chen, Junying
Wang, Xidong
Ji, Ke
Gao, Anningzhe
Jiang, Feng
Chen, Shunian
Zhang, Hongbo
Song, Dingjie
Xie, Wenya
Kong, Chuyi
Li, Jianquan
Wan, Xiang
Li, Haizhou
Wang, Benyou
author_facet Chen, Junying
Wang, Xidong
Ji, Ke
Gao, Anningzhe
Jiang, Feng
Chen, Shunian
Zhang, Hongbo
Song, Dingjie
Xie, Wenya
Kong, Chuyi
Li, Jianquan
Wan, Xiang
Li, Haizhou
Wang, Benyou
contents Adapting a language model into a specific domain, a.k.a `domain adaption', is a common practice when specialized knowledge, e.g. medicine, is not encapsulated in a general language model like Llama2. The challenge lies in the heterogeneity of data across the two training stages, as it varies in languages, genres, or formats. To tackle this and simplify the learning protocol, we propose to transform heterogeneous data, from the both pre-training and supervised stages, into a unified, simple input-output pair format. We validate the new protocol in the domains where proprietary LLMs like ChatGPT perform relatively poorly, such as Traditional Chinese Medicine. The developed model, HuatuoGPT-II, has shown state-of-the-art performance in Chinese medicine domain on a number of benchmarks, e.g. medical licensing exams. It even outperforms proprietary models like ChatGPT and GPT-4 in some aspects, especially in Traditional Chinese Medicine. Expert manual evaluations further validate HuatuoGPT-II's advantages over existing LLMs. Notably, HuatuoGPT-II was benchmarked in a fresh Chinese National Medical Licensing Examination where it achieved the best performance, showcasing not only its effectiveness but also its generalization capabilities.
format Preprint
id arxiv_https___arxiv_org_abs_2311_09774
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle HuatuoGPT-II, One-stage Training for Medical Adaption of LLMs
Chen, Junying
Wang, Xidong
Ji, Ke
Gao, Anningzhe
Jiang, Feng
Chen, Shunian
Zhang, Hongbo
Song, Dingjie
Xie, Wenya
Kong, Chuyi
Li, Jianquan
Wan, Xiang
Li, Haizhou
Wang, Benyou
Computation and Language
Artificial Intelligence
Machine Learning
Adapting a language model into a specific domain, a.k.a `domain adaption', is a common practice when specialized knowledge, e.g. medicine, is not encapsulated in a general language model like Llama2. The challenge lies in the heterogeneity of data across the two training stages, as it varies in languages, genres, or formats. To tackle this and simplify the learning protocol, we propose to transform heterogeneous data, from the both pre-training and supervised stages, into a unified, simple input-output pair format. We validate the new protocol in the domains where proprietary LLMs like ChatGPT perform relatively poorly, such as Traditional Chinese Medicine. The developed model, HuatuoGPT-II, has shown state-of-the-art performance in Chinese medicine domain on a number of benchmarks, e.g. medical licensing exams. It even outperforms proprietary models like ChatGPT and GPT-4 in some aspects, especially in Traditional Chinese Medicine. Expert manual evaluations further validate HuatuoGPT-II's advantages over existing LLMs. Notably, HuatuoGPT-II was benchmarked in a fresh Chinese National Medical Licensing Examination where it achieved the best performance, showcasing not only its effectiveness but also its generalization capabilities.
title HuatuoGPT-II, One-stage Training for Medical Adaption of LLMs
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2311.09774