paper-with-me

홈 › Papers

VocabTailor: Dynamic Vocabulary Selection for Downstream Tasks in Small Language Models

2025-08-21 · Hanling Zhang, Yayu Zhou, Tongcheng Fang, Zhihang Yuan, Guohao Dai, Wanli Ouyang, Yu Wang arxiv

Small Language Models (SLMs) provide computational advantages in resource-constrained environments, yet memory limitations remain a critical bottleneck for edge device deployment. A substantial portion of SLMs' memory footprint stems from vocabulary-related components, particularly embeddings and language modeling (LM) heads, due to large vocabulary sizes. Existing static vocabulary pruning, while reducing memory usage, suffers from rigid, one-size-fits-all designs that cause information loss during the prefill stage and lack flexibility. In this work, we identify two key principles underlying the vocabulary reduction challenge: the lexical locality principle, the observation that only a small subset of tokens is required during any single inference, and the asymmetry in computational characteristics between vocabulary-related components of SLM. Based on these insights, we introduce VocabTailor, a novel decoupled dynamic vocabulary selection framework that addresses memory constraints through offloading embedding and implements a hybrid static-dynamic vocabulary selection strategy for LM Head, enabling on-demand loading of vocabulary components. Comprehensive experiments across diverse downstream tasks demonstrate that VocabTailor achieves a reduction of up to 99% in the memory usage of vocabulary-related components with minimal or no degradation in task performance, substantially outperforming existing static vocabulary pruning. Our code is available at https://github.com/AwakenedInsects/VocabTailor.

📄 PDF Abstract BibTeX arXiv:2508.15229

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Training-Free Action Recognition and Goal Inference with Dynamic Frame Selection

2024-01-23 · Ee Yeo Keat, Zhang Hao, Alexander Matyasko, Basura Fernando

We introduce VidTFS, a Training-free, open-vocabulary video goal and action inference framework that combines the frozen vision foundational model (VFM) and large language model (LLM) with a novel dynamic Frame Selection…

Action RecognitionLanguage ModelingLanguage ModellingLarge Language Model+1

Generation with Dynamic Vocabulary

2024-10-11 · Yanting Liu, Tao Ji, Changzhi Sun, Yuanbin Wu 외

We introduce a new dynamic vocabulary for language models. It can involve arbitrary text spans during generation. These text spans act as basic generation bricks, akin to tokens in the traditional static vocabularies. We…

Language ModelingLanguage ModellingQuestion Answering

VSEC-LDA: Boosting Topic Modeling with Embedded Vocabulary Selection

2020-01-15 · Yuzhen Ding, Baoxin Li

Topic modeling has found wide application in many problems where latent structures of the data are crucial for typical inference tasks. When applying a topic model, a relatively standard pre-processing step is to first b…

Topic Models

In Search of Lost DNA Sequence Pretraining

2026-04-17 · Zhijiang Tang, Jiaxin Qi, Yan Cui, Jinli Ou 외 arxiv

DNA sequence encoding is fundamental to gene function prediction, protein synthesis, and diverse downstream biological tasks. Despite the substantial progress achieved by large-scale DNA sequence pretraining, existing st…

How Large a Vocabulary Does Text Classification Need? A Variational Approach to Vocabulary Selection

2019-02-27 · NAACL 2019 6 · Wenhu Chen, Yu Su, Yilin Shen, Zhiyu Chen 외

With the rapid development in deep learning, deep neural networks have been widely adopted in many real-life natural language applications. Under deep neural networks, a pre-defined vocabulary is required to vectorize te…

General Classificationtext-classificationText Classification