A Systematic Study of Compositional Syntactic Transformer Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Yida, Xve, Hao, Hu, Xiang, Tu, Kewei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909665223245824
author Zhao, Yida
Xve, Hao
Hu, Xiang
Tu, Kewei
author_facet Zhao, Yida
Xve, Hao
Hu, Xiang
Tu, Kewei
contents Syntactic language models (SLMs) enhance Transformers by incorporating syntactic biases through the modeling of linearized syntactic parse trees alongside surface sentences. This paper focuses on compositional SLMs that are based on constituency parse trees and contain explicit bottom-up composition of constituent representations. We identify key aspects of design choices in existing compositional SLMs and propose a unified framework encompassing both existing models and novel variants. We conduct a comprehensive empirical evaluation of all the variants in our framework across language modeling, syntactic generalization, summarization, dialogue, and inference efficiency. Based on the experimental results, we make multiple recommendations on the design of compositional SLMs. Our code is released at https://github.com/zhaoyd1/compositional_SLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2506_22978
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Systematic Study of Compositional Syntactic Transformer Language Models
Zhao, Yida
Xve, Hao
Hu, Xiang
Tu, Kewei
Computation and Language
Artificial Intelligence
Syntactic language models (SLMs) enhance Transformers by incorporating syntactic biases through the modeling of linearized syntactic parse trees alongside surface sentences. This paper focuses on compositional SLMs that are based on constituency parse trees and contain explicit bottom-up composition of constituent representations. We identify key aspects of design choices in existing compositional SLMs and propose a unified framework encompassing both existing models and novel variants. We conduct a comprehensive empirical evaluation of all the variants in our framework across language modeling, syntactic generalization, summarization, dialogue, and inference efficiency. Based on the experimental results, we make multiple recommendations on the design of compositional SLMs. Our code is released at https://github.com/zhaoyd1/compositional_SLMs.
title A Systematic Study of Compositional Syntactic Transformer Language Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2506.22978