paper-with-me

홈 › Papers

When Every Token Counts: Optimal Segmentation for Low-Resource Language Models

2024-12-09 · Bharath Raj, Garvit Suri, Vikrant Dewangan, Raghav Sonavane

Traditional greedy tokenization methods have been a critical step in Natural Language Processing (NLP), influencing how text is converted into tokens and directly impacting model performance. While subword tokenizers like Byte-Pair Encoding (BPE) are widely used, questions remain about their optimality across model scales and languages. In this work, we demonstrate through extensive experiments that an optimal BPE configuration significantly reduces token count compared to greedy segmentation, yielding improvements in token-saving percentages and performance benefits, particularly for smaller models. We evaluate tokenization performance across various intrinsic and extrinsic tasks, including generation and classification. Our findings suggest that compression-optimized tokenization strategies could provide substantial advantages for multilingual and low-resource language applications, highlighting a promising direction for further research and inclusive NLP.

📄 PDF Abstract BibTeX arXiv:2412.06926

Code (1)

rath772k/optimal_segmentation 공식 구현

Methods 이 논문이 사용한 방법론

BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…

Similar Papers 제목 키워드 기반

Token Inflation: How Dishonest Providers Can Overcharge for Large Language Model Usage

2026-05-28 · Shahinul Hoque, Jinghuai Zhang, Jinyuan Sun, Fnu Suya arxiv

Per-token billing is now the standard pricing model for commercial large language models (LLMs), so the honesty of reported token counts directly affects what users pay. We show that this kind of billing is hard to audit…

Writing-System-Level Tokenizer Adaptation for Byte-Level BPE

2026-08-01 · Bohdan Didenko arxiv

Pretrained byte-level BPE tokenizers can segment underrepresented languages inefficiently. Replacing a tokenizer changes the meaning of nearly every token ID, while vocabulary expansion enlarges the model's embedding and…

Dynamic Token Pruning in Plain Vision Transformers for Semantic Segmentation

2023-08-02 · ICCV 2023 1 · Quan Tang, BoWen Zhang, Jiajun Liu, Fagui Liu 외

Vision transformers have achieved leading performance on various visual tasks yet still suffer from high computational complexity. The situation deteriorates in dense prediction tasks like semantic segmentation, as high-…

image-classificationImage ClassificationSegmentationSemantic Segmentation

VLTP: Vision-Language Guided Token Pruning for Task-Oriented Segmentation

2024-09-13 · Hanning Chen, Yang Ni, Wenjun Huang, Yezi Liu 외

Vision Transformers (ViTs) have emerged as the backbone of many segmentation models, consistently achieving state-of-the-art (SOTA) performance. However, their success comes at a significant computational cost. Image tok…

DecoderLanguage ModellingLarge Language ModelSegmentation

T-REN: Learning Text-Aligned Region Tokens Improves Dense Vision-Language Alignment and Scalability

2026-04-20 · Savya Khosla, Sethuraman T, Aryan Chadha, Alex Schwing 외 arxiv

Despite recent progress, vision-language encoders struggle with two core limitations: (1) weak alignment between language and dense vision features, which hurts tasks like open-vocabulary semantic segmentation; and (2) h…

Semantic SegmentationObject LocalizationImage RetrievalScene Parsing