Multitask Prompted Training Enables Zero-Shot Task Generalization

Victor Sanh Hugging Face and Albert Webson∗ Brown University and Colin Raffel∗ Hugging Face and Stephen H. Bach∗ Brown、Snorkel AI and Lintang Sutawika BigScience and Zaid Alyafeai KFUPM and Antoine Chaffin IRISA、IMATAG and Arnaud Stiegler Hyperscience and Teven Le Scao Hugging Face and Arun Raja I2R, Singapore and Manan Dey SAP and M Saiful Bari NTU, Singapore and Canwen Xu UCSD、Hugging Face and Urmish Thakker SambaNova Systems and Shanya Sharma Walmart Labs and Eliza Szczechla BigScience and Taewoon Kim VU Amsterdam and Gunjan Chhablani BigScience and Nihal V. Nayak Brown University and Debajyoti Datta University of Virginia and Jonathan Chang ASUS and Mike Tian-Jian Jiang ZEALS, Japan and Han Wang NYU and Matteo Manica IBM Research and Sheng Shen UC Berkeley and Zheng-Xin Yong Brown University and Harshit Pandey BigScience and Michael McKenna Parity and Rachel Bawden Inria, France and Thomas Wang Inria, France and Trishala Neeraj BigScience and Jos Rozen Naver Labs Europe and Abheesht Sharma BITS Pilani, India and Andrea Santilli University of Rome and Thibault Fevry BigScience and Jason Alan Fries Stanford、Snorkel AI and Ryan Teehan Charles River Analytics and Tali Bers Brown University and Stella Biderman Booz Allen、EleutherAI and Leo Gao EleutherAI and Thomas Wolf Hugging Face and Alexander M. Rush Hugging Face Thanks: Equal contribution. Full list of individual contributions detailed in Appendix A. Corresponding authors: and

原文 arXiv:2110.08207;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2110.08207v3