Understanding In-Context Learning from Repetitions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yan, Jianhao, Xu, Jin, Song, Chiyu, Wu, Chenming, Li, Yafu, Zhang, Yue
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909114151469056
author Yan, Jianhao
Xu, Jin
Song, Chiyu
Wu, Chenming
Li, Yafu
Zhang, Yue
author_facet Yan, Jianhao
Xu, Jin
Song, Chiyu
Wu, Chenming
Li, Yafu
Zhang, Yue
contents This paper explores the elusive mechanism underpinning in-context learning in Large Language Models (LLMs). Our work provides a novel perspective by examining in-context learning via the lens of surface repetitions. We quantitatively investigate the role of surface features in text generation, and empirically establish the existence of \emph{token co-occurrence reinforcement}, a principle that strengthens the relationship between two tokens based on their contextual co-occurrences. By investigating the dual impacts of these features, our research illuminates the internal workings of in-context learning and expounds on the reasons for its failures. This paper provides an essential contribution to the understanding of in-context learning and its potential limitations, providing a fresh perspective on this exciting capability.
format Preprint
id arxiv_https___arxiv_org_abs_2310_00297
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Understanding In-Context Learning from Repetitions
Yan, Jianhao
Xu, Jin
Song, Chiyu
Wu, Chenming
Li, Yafu
Zhang, Yue
Computation and Language
This paper explores the elusive mechanism underpinning in-context learning in Large Language Models (LLMs). Our work provides a novel perspective by examining in-context learning via the lens of surface repetitions. We quantitatively investigate the role of surface features in text generation, and empirically establish the existence of \emph{token co-occurrence reinforcement}, a principle that strengthens the relationship between two tokens based on their contextual co-occurrences. By investigating the dual impacts of these features, our research illuminates the internal workings of in-context learning and expounds on the reasons for its failures. This paper provides an essential contribution to the understanding of in-context learning and its potential limitations, providing a fresh perspective on this exciting capability.
title Understanding In-Context Learning from Repetitions
topic Computation and Language
url https://arxiv.org/abs/2310.00297