A Law of Next-Token Prediction in Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: He, Hangfeng, Su, Weijie J.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916926261821440
author He, Hangfeng
Su, Weijie J.
author_facet He, Hangfeng
Su, Weijie J.
contents Large language models (LLMs) have been widely employed across various application domains, yet their black-box nature poses significant challenges to understanding how these models process input data internally to make predictions. In this paper, we introduce a precise and quantitative law that governs the learning of contextualized token embeddings through intermediate layers in pre-trained LLMs for next-token prediction. Our findings reveal that each layer contributes equally to enhancing prediction accuracy, from the lowest to the highest layer -- a universal phenomenon observed across a diverse array of open-source LLMs, irrespective of their architectures or pre-training data. We demonstrate that this law offers new perspectives and actionable insights to inform and guide practices in LLM development and applications, including model scaling, pre-training tasks, and interpretation.
format Preprint
id arxiv_https___arxiv_org_abs_2408_13442
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Law of Next-Token Prediction in Large Language Models
He, Hangfeng
Su, Weijie J.
Machine Learning
Artificial Intelligence
Computation and Language
Large language models (LLMs) have been widely employed across various application domains, yet their black-box nature poses significant challenges to understanding how these models process input data internally to make predictions. In this paper, we introduce a precise and quantitative law that governs the learning of contextualized token embeddings through intermediate layers in pre-trained LLMs for next-token prediction. Our findings reveal that each layer contributes equally to enhancing prediction accuracy, from the lowest to the highest layer -- a universal phenomenon observed across a diverse array of open-source LLMs, irrespective of their architectures or pre-training data. We demonstrate that this law offers new perspectives and actionable insights to inform and guide practices in LLM development and applications, including model scaling, pre-training tasks, and interpretation.
title A Law of Next-Token Prediction in Large Language Models
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2408.13442