arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2211.09085cs.CLstat.ML
收藏

Galactica:一个面向科学的大语言模型

Galactica: A Large Language Model for Science

Meta AI

Ross Taylor, Marcin Kardas, Guillem Cucurull, Thomas Scialom, Anthony Hartshorn, Elvis Saravia, Andrew Poulton, Viktor Kerkez, Robert Stojnic

摘要:

信息过载是科学进步的主要障碍。科学文献和数据的爆炸式增长使得从大量信息中发现有用的见解变得越来越困难。今天,科学知识可以通过搜索引擎获取,但它们无法单独组织科学知识。本文介绍了Galactica:一个能够存储、组合和推理科学知识的大语言模型。我们训练的论文,参考资料,知识库和许多其他来源的大型科学语料库。我们在一系列科学任务上优于现有模型。在LaTeX方程等技术知识探测方面,Galactica的表现比最新的GPT-3高出68.2%,而后者仅为49.0%。卡拉狄加在推理方面也表现出色,在数学MMLU上以41.3%对35.7%的成绩超过了Chinchilla,在数学上以20.4%对8.8%的成绩超过了PaLM 540 B。它还为下游任务设定了新的最先进水平,例如PubMedQA和MedMCQA dev分别为77.6%和52.9%。尽管没有在通用语料库上进行训练,卡拉狄加在BIG-bench上的表现优于布卢姆和OPT-175 B。我们相信这些结果表明了语言模型作为科学新接口的潜力。我们将该模型开源,以造福科学界。

英文摘要:

Information overload is a major obstacle to scientific progress. The explosive growth in scientific literature and data has made it ever harder to discover useful insights in a large mass of information. Today scientific knowledge is accessed through search engines, but they are unable to organize scientific knowledge alone. In this paper we introduce Galactica: a large language model that can store, combine and reason about scientific knowledge. We train on a large scientific corpus of papers, reference material, knowledge bases and many other sources. We outperform existing models on a range of scientific tasks. On technical knowledge probes such as LaTeX equations, Galactica outperforms the latest GPT-3 by 68.2% versus 49.0%. Galactica also performs well on reasoning, outperforming Chinchilla on mathematical MMLU by 41.3% to 35.7%, and PaLM 540B on MATH with a score of 20.4% versus 8.8%. It also sets a new state-of-the-art on downstream tasks such as PubMedQA and MedMCQA dev of 77.6% and 52.9%. And despite not being trained on a general corpus, Galactica outperforms BLOOM and OPT-175B on BIG-bench. We believe these results demonstrate the potential for language models as a new interface for science. We open source the model for the benefit of the scientific community.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑