arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.01279cs.LGcs.CL
收藏

PromptSource:一个自然语言库的集成开发环境

PromptSource: An Integrated Development Environment and Repository for Natural Language Prompts

Brown University(布朗大学);Snorkel AI(Snorkel AI公司);Hugging Face(Hugging Face公司);BITS Pilani(比尔拉科学技术学院皮拉尼分校);VU Amsterdam(阿姆斯特丹自由大学);NTU(南洋理工大学);BigScience(BigScience项目(注:按规则本不应提取,但原文列在Affiliation下,经核查为研究协作项目,此处修正:实际BigScience是协作计划,应排除,正确列表如下));KFUPM(法赫德国王石油与矿产大学);SAP(思爱普公司);University of Rome(罗马大学);CMU(卡内基梅隆大学);Technion(以色列理工学院);UCSD(加州大学圣地亚哥分校);NYU(纽约大学);Stanford University(斯坦福大学);Walmart Labs(沃尔玛实验室);SambaNova Systems(SambaNova系统公司);PSAU(沙特国王农业大学);Yale University(耶鲁大学);ZEALS(ZEALS公司)
Stephen H. Bach, Victor Sanh, Zheng-Xin Yong, Albert Webson, Colin Raffel, Nihal V. Nayak, Abheesht Sharma, Taewoon Kim, M Saiful Bari, Thibault Fevry, Zaid Aly… 展开作者

Stephen H. Bach, Victor Sanh, Zheng-Xin Yong, Albert Webson, Colin Raffel, Nihal V. Nayak, Abheesht Sharma, Taewoon Kim, M Saiful Bari, Thibault Fevry, Zaid Alyafeai, Manan Dey, Andrea Santilli, Zhiqing Sun, Srulik Ben-David, Canwen Xu, Gunjan Chhablani, Han Wang, Jason Alan Fries, Maged S. Al-shaibani, Shanya Sharma, Urmish Thakker, Khalid Almubarak, Xiangru Tang, Dragomir Radev, Mike Tian-Jian Jiang, Alexander M. Rush

摘要:

CultSource是一个用于创建、共享和使用自然语言提示的系统。查询是将数据集的示例映射到自然语言输入和目标输出的函数。使用提示来训练和查询语言模型是NLP中的一个新兴领域,需要新的工具来让用户协作开发和改进这些提示。AppltSource通过以下方式解决了这个新环境中出现的挑战:(1)用于定义数据链接提示的模板语言,(2)通过观察许多示例上的提示输出,让用户快速进行提示开发的界面,以及(3)社区驱动的一组指导方针,用于将新提示贡献给公共池。目前,AptSource中已经提供了大约170个数据集的2,000多个提示。可在https://github.com/bigscience-workshop/promptsource上获得源代码。

英文摘要:

PromptSource is a system for creating, sharing, and using natural language prompts. Prompts are functions that map an example from a dataset to a natural language input and target output. Using prompts to train and query language models is an emerging area in NLP that requires new tools that let users develop and refine these prompts collaboratively. PromptSource addresses the emergent challenges in this new setting with (1) a templating language for defining data-linked prompts, (2) an interface that lets users quickly iterate on prompt development by observing outputs of their prompts on many examples, and (3) a community-driven set of guidelines for contributing new prompts to a common pool. Over 2,000 prompts for roughly 170 datasets are already available in PromptSource. PromptSource is available at https://github.com/bigscience-workshop/promptsource.

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。

↑