Reasoning is Periodicity? Improving Large Language Models Through Effective Periodicity Modeling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dong, Yihong, Li, Ge, Jiang, Xue, Tao, Yongding, Zhang, Kechi, Zhu, Hao, Liu, Huanyu, Ding, Jiazheng, Li, Jia, Deng, Jinliang, Mei, Hong
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908613302288384
author Dong, Yihong
Li, Ge
Jiang, Xue
Tao, Yongding
Zhang, Kechi
Zhu, Hao
Liu, Huanyu
Ding, Jiazheng
Li, Jia
Deng, Jinliang
Mei, Hong
author_facet Dong, Yihong
Li, Ge
Jiang, Xue
Tao, Yongding
Zhang, Kechi
Zhu, Hao
Liu, Huanyu
Ding, Jiazheng
Li, Jia
Deng, Jinliang
Mei, Hong
contents Periodicity, as one of the most important basic characteristics, lays the foundation for facilitating structured knowledge acquisition and systematic cognitive processes within human learning paradigms. However, the potential flaws of periodicity modeling in Transformer affect the learning efficiency and establishment of underlying principles from data for large language models (LLMs) built upon it. In this paper, we demonstrate that integrating effective periodicity modeling can improve the learning efficiency and performance of LLMs. We introduce FANformer, which adapts Fourier Analysis Network (FAN) into attention mechanism to achieve efficient periodicity modeling, by modifying the feature projection process of attention mechanism. Extensive experimental results on language modeling show that FANformer consistently outperforms Transformer when scaling up model size and training tokens, underscoring its superior learning efficiency. Our pretrained FANformer-1B exhibits marked improvements on downstream tasks compared to open-source LLMs with similar model parameters or training tokens. Moreover, we reveal that FANformer exhibits superior ability to learn and apply rules for reasoning compared to Transformer. The results position FANformer as an effective and promising architecture for advancing LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2502_21309
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Reasoning is Periodicity? Improving Large Language Models Through Effective Periodicity Modeling
Dong, Yihong
Li, Ge
Jiang, Xue
Tao, Yongding
Zhang, Kechi
Zhu, Hao
Liu, Huanyu
Ding, Jiazheng
Li, Jia
Deng, Jinliang
Mei, Hong
Computation and Language
Artificial Intelligence
Machine Learning
Periodicity, as one of the most important basic characteristics, lays the foundation for facilitating structured knowledge acquisition and systematic cognitive processes within human learning paradigms. However, the potential flaws of periodicity modeling in Transformer affect the learning efficiency and establishment of underlying principles from data for large language models (LLMs) built upon it. In this paper, we demonstrate that integrating effective periodicity modeling can improve the learning efficiency and performance of LLMs. We introduce FANformer, which adapts Fourier Analysis Network (FAN) into attention mechanism to achieve efficient periodicity modeling, by modifying the feature projection process of attention mechanism. Extensive experimental results on language modeling show that FANformer consistently outperforms Transformer when scaling up model size and training tokens, underscoring its superior learning efficiency. Our pretrained FANformer-1B exhibits marked improvements on downstream tasks compared to open-source LLMs with similar model parameters or training tokens. Moreover, we reveal that FANformer exhibits superior ability to learn and apply rules for reasoning compared to Transformer. The results position FANformer as an effective and promising architecture for advancing LLMs.
title Reasoning is Periodicity? Improving Large Language Models Through Effective Periodicity Modeling
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2502.21309