How does Architecture Influence the Base Capabilities of Pre-trained Language Models? A Case Study Based on FFN-Wider and MoE Transformers

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lu, Xin, Zhao, Yanyan, Qin, Bing, Huo, Liangyu, Yang, Qing, Xu, Dongliang
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912096593117184
author Lu, Xin
Zhao, Yanyan
Qin, Bing
Huo, Liangyu
Yang, Qing
Xu, Dongliang
author_facet Lu, Xin
Zhao, Yanyan
Qin, Bing
Huo, Liangyu
Yang, Qing
Xu, Dongliang
contents Pre-trained language models have been proven to possess strong base capabilities, which not only excel in in-distribution language modeling but also show powerful abilities in out-of-distribution language modeling, transfer learning and few-shot learning. Unlike existing work focusing on the influence of scale on base capabilities, our work examines the influence of architecture on those. Specifically, our concern is: How does architecture influence the base capabilities of pre-trained language models? In this work, we attempt to explain and reverse the decline in base capabilities caused by the architecture of FFN-Wider Transformers, seeking to provide some insights. Through analysis, we found the contribution ratio of Multi-Head Attention (a combination function) to pre-trained language modeling is a key factor affecting base capabilities. FFN-Wider Transformers reduce the contribution ratio of this combination function, leading to a decline in base capabilities. We confirmed this by experiments and proposed Combination Enhanced Architecture (CEA) to address the decline in base capabilities of such models. Significantly, we extended our explanation and CEA to Mixture of Experts (MoE) Transformers. We successfully achieved significant improvements in base capabilities on a 14B parameter MoE model, demonstrating the practical application value of our work. This also indicates that our analysis has a certain guiding significance for architecture analysis, architecture improvement and architecture design.
format Preprint
id arxiv_https___arxiv_org_abs_2403_02436
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle How does Architecture Influence the Base Capabilities of Pre-trained Language Models? A Case Study Based on FFN-Wider and MoE Transformers
Lu, Xin
Zhao, Yanyan
Qin, Bing
Huo, Liangyu
Yang, Qing
Xu, Dongliang
Computation and Language
Pre-trained language models have been proven to possess strong base capabilities, which not only excel in in-distribution language modeling but also show powerful abilities in out-of-distribution language modeling, transfer learning and few-shot learning. Unlike existing work focusing on the influence of scale on base capabilities, our work examines the influence of architecture on those. Specifically, our concern is: How does architecture influence the base capabilities of pre-trained language models? In this work, we attempt to explain and reverse the decline in base capabilities caused by the architecture of FFN-Wider Transformers, seeking to provide some insights. Through analysis, we found the contribution ratio of Multi-Head Attention (a combination function) to pre-trained language modeling is a key factor affecting base capabilities. FFN-Wider Transformers reduce the contribution ratio of this combination function, leading to a decline in base capabilities. We confirmed this by experiments and proposed Combination Enhanced Architecture (CEA) to address the decline in base capabilities of such models. Significantly, we extended our explanation and CEA to Mixture of Experts (MoE) Transformers. We successfully achieved significant improvements in base capabilities on a 14B parameter MoE model, demonstrating the practical application value of our work. This also indicates that our analysis has a certain guiding significance for architecture analysis, architecture improvement and architecture design.
title How does Architecture Influence the Base Capabilities of Pre-trained Language Models? A Case Study Based on FFN-Wider and MoE Transformers
topic Computation and Language
url https://arxiv.org/abs/2403.02436