arrow_backBack to research feed
otherPublished: June 15, 2022

Emergent abilities of large language models

By Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama

Research TL;DR

"Surveys emergent behaviors in LLMs. Identifies capabilities like arithmetic and translation that appear abruptly as training compute scales."

Abstract

We discuss emergent abilities of large language models, which are abilities that are not present in smaller models but are present in larger models. We analyze how emergence occurs across various tasks like multi-step math and reasoning.

Read full paper on arXiv →

Related Research

No other papers in this category.