Cover Image

Text or Pixels? It Takes Half: On the Token Efficiency of Visual Text Inputs in Multimodal LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Li, Yanhong, Lan, Zixuan, Zhou, Jiawei
Format:	Preprint
Published:	2025
Subjects:	Computation and Language Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Beyond Text: Optimizing RAG with Multimodal Inputs for Industrial Applications
by: Riedler, Monica, et al.
Published: (2024)

CSRP: Chain-of-Thought Reasoning for Chinese Text Correction via Reinforcement Learning with Efficiency-Aware Rewards
by: Tian, Wei, et al.
Published: (2026)

Chunk-Distilled Language Modeling
by: Li, Yanhong, et al.
Published: (2024)

Information Capacity: Evaluating the Efficiency of Large Language Models via Text Compression
by: Yuan, Cheng, et al.
Published: (2025)

Unsupervised Text Style Transfer via LLMs and Attention Masking with Multi-way Interactions
by: Pan, Lei, et al.
Published: (2024)

Cross-Attention is Half Explanation in Speech-to-Text Models
by: Papi, Sara, et al.
Published: (2025)

Multimodal LLMs as Customized Reward Models for Text-to-Image Generation
by: Zhou, Shijie, et al.
Published: (2025)

Text as Images: Can Multimodal Large Language Models Follow Printed Instructions in Pixels?
by: Li, Xiujun, et al.
Published: (2023)

On the Predictive Power of Representation Dispersion in Language Models
by: Li, Yanhong, et al.
Published: (2025)

ToBlend: Token-Level Blending With an Ensemble of LLMs to Attack AI-Generated Text Detection
by: Huang, Fan, et al.
Published: (2024)

Text Generation Beyond Discrete Token Sampling
by: Zhuang, Yufan, et al.
Published: (2025)

Discrete Tokenization for Multimodal LLMs: A Comprehensive Survey
by: Li, Jindong, et al.
Published: (2025)

Empowering Character-level Text Infilling by Eliminating Sub-Tokens
by: Ren, Houxing, et al.
Published: (2024)

LLMs Enable Bag-of-Texts Representations for Short-Text Clustering
by: Lin, I-Fan, et al.
Published: (2025)

Wings: Learning Multimodal LLMs without Text-only Forgetting
by: Zhang, Yi-Kai, et al.
Published: (2024)

ConTextual: Improving Clinical Text Summarization in LLMs with Context-preserving Token Filtering and Knowledge Graphs
by: Piya, Fahmida Liza, et al.
Published: (2025)

Reinforcement Learning with Token-level Feedback for Controllable Text Generation
by: Li, Wendi, et al.
Published: (2024)

PixelPrune: Pixel-Level Adaptive Visual Token Reduction via Predictive Coding
by: Wang, Nan, et al.
Published: (2026)

TextQuests: How Good are LLMs at Text-Based Video Games?
by: Phan, Long, et al.
Published: (2025)

Towards Visual Text Grounding of Multimodal Large Language Model
by: Li, Ming, et al.
Published: (2025)

Token Masking Improves Transformer-Based Text Classification
by: Xu, Xianglong, et al.
Published: (2025)

Explainability-Based Token Replacement on LLM-Generated Text
by: Mohammadi, Hadi, et al.
Published: (2025)

Training Text-to-Molecule Models with Context-Aware Tokenization
by: Kim, Seojin, et al.
Published: (2025)

From Pixels to Tokens: Byte-Pair Encoding on Quantized Visual Modalities
by: Zhang, Wanpeng, et al.
Published: (2024)

Visual Text Matters: Improving Text-KVQA with Visual Text Entity Knowledge-aware Large Multimodal Assistant
by: Penamakuri, Abhirama Subramanyam, et al.
Published: (2024)

Chem-FINESE: Validating Fine-Grained Few-shot Entity Extraction through Text Reconstruction
by: Wang, Qingyun, et al.
Published: (2024)

Alternatives To Next Token Prediction In Text Generation -- A Survey
by: Wyatt, Charlie, et al.
Published: (2025)

Sticking to the Mean: Detecting Sticky Tokens in Text Embedding Models
by: Chen, Kexin, et al.
Published: (2025)

Exploring the Capabilities of Large Multimodal Models on Dense Text
by: Zhang, Shuo, et al.
Published: (2024)

JoyAI-LLM Flash: Advancing Mid-Scale LLMs with Token Efficiency
by: Cai, Aichen, et al.
Published: (2026)

Thunder-Tok: Minimizing Tokens per Word in Tokenizing Korean Texts for Generative Language Models
by: Cho, Gyeongje, et al.
Published: (2025)

Does Visual Rendering Bypass Tokenization? Investigating Script-Tokenizer Misalignment in Pixel-Based Language Models
by: Susanto, Lucky, et al.
Published: (2026)

Selective Attention Federated Learning: Improving Privacy and Efficiency for Clinical Text Classification
by: Li, Yue, et al.
Published: (2025)

Conditioning LLMs to Generate Code-Switched Text
by: Heredia, Maite, et al.
Published: (2025)

What Happened in LLMs Layers when Trained for Fast vs. Slow Thinking: A Gradient Perspective
by: Li, Ming, et al.
Published: (2024)

Grasp Any Region: Towards Precise, Contextual Pixel Understanding for Multimodal LLMs
by: Wang, Haochen, et al.
Published: (2025)

Distilling Text Style Transfer With Self-Explanation From LLMs
by: Zhang, Chiyu, et al.
Published: (2024)

Exploiting Text Semantics for Few and Zero Shot Node Classification on Text-attributed Graph
by: Wang, Yuxiang, et al.
Published: (2025)

Adaptive Token Boundaries: Integrating Human Chunking Mechanisms into Multimodal LLMs
by: Yu, Dongxing
Published: (2025)

TextBind: Multi-turn Interleaved Multimodal Instruction-following in the Wild
by: Li, Huayang, et al.
Published: (2023)