Imagen de Portada

Not All Tokens Matter: Towards Efficient LLM Reasoning via Token Significance in Reinforcement Learning

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Liu, Hanbing, Cao, Lang, Ren, Yuanyi, Zhou, Mengyu, Dong, Haoyu, Ma, Xiaojun, Han, Shi, Zhang, Dongmei
Formato:	Preprint
Publicado:	2025
Materias:	Machine Learning Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Formula-R1: Incentivizing LLM Reasoning over Complex Tables with Numerical Computation via Formula-Driven Reinforcement Learning
por: Cao, Lang, et al.
Publicado: (2025)

TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers' Guidance
por: Xu, Jingxian, et al.
Publicado: (2025)

SuperRL: Reinforcement Learning with Supervision to Boost Language Model Reasoning
por: Liu, Yihao, et al.
Publicado: (2025)

Efficient Reinforcement Learning with Semantic and Token Entropy for LLM Reasoning
por: Cao, Hongye, et al.
Publicado: (2025)

TablePilot: Recommending Human-Preferred Tabular Data Analysis with Large Language Models
por: Yi, Deyin, et al.
Publicado: (2025)

Dynamic Token Compression for Efficient Video Understanding through Reinforcement Learning
por: Wang, Shida, et al.
Publicado: (2026)

SheetDesigner: MLLM-Powered Spreadsheet Layout Generation with Rule-Based and Vision-Based Reflection
por: Chen, Qin, et al.
Publicado: (2025)

Critical Tokens Matter: Token-Level Contrastive Estimation Enhances LLM's Reasoning Capability
por: Lin, Zicheng, et al.
Publicado: (2024)

SheetBrain: A Neuro-Symbolic Agent for Accurate Reasoning over Complex and Large Spreadsheets
por: Wang, Ziwei, et al.
Publicado: (2025)

SelfBudgeter: Adaptive Token Allocation for Efficient LLM Reasoning
por: Li, Zheng, et al.
Publicado: (2025)

Table-LLM-Specialist: Language Model Specialists for Tables using Iterative Generator-Validator Fine-tuning
por: Xing, Junjie, et al.
Publicado: (2024)

Token-Budget-Aware LLM Reasoning
por: Han, Tingxu, et al.
Publicado: (2024)

Pruning the Unsurprising: Efficient LLM Reasoning via First-Token Surprisal
por: Zeng, Wenhao, et al.
Publicado: (2025)

RAG or Learning? Understanding the Limits of LLM Adaptation under Continuous Knowledge Drift in the Real World
por: Liu, Hanbing, et al.
Publicado: (2026)

Token-Efficient RL for LLM Reasoning
por: Lee, Alan, et al.
Publicado: (2025)

Large Language Models for Predictive Analysis: How Far Are They?
por: Chen, Qin, et al.
Publicado: (2025)

A Single Revision Step Improves Token-Efficient LLM Reasoning
por: Zhang, Yingchuan, et al.
Publicado: (2026)

Contribution-aware Token Compression for Efficient Video Understanding via Reinforcement Learning
por: Ma, Yinchao, et al.
Publicado: (2026)

Token-Efficient Prompt Injection Attack: Provoking Cessation in LLM Reasoning via Adaptive Token Compression
por: Cui, Yu, et al.
Publicado: (2025)

Not All Tokens Matter: Data-Centric Optimization for Efficient Code Summarization
por: Afrin, Saima, et al.
Publicado: (2026)

Table Meets LLM: Can Large Language Models Understand Structured Table Data? A Benchmark and Empirical Study
por: Sui, Yuan, et al.
Publicado: (2023)

TableMaster: A Recipe to Advance Table Understanding with Language Models
por: Cao, Lang, et al.
Publicado: (2025)

Jupiter: Enhancing LLM Data Analysis Capabilities via Notebook and Inference-Time Value-Guided Search
por: Li, Shuocheng, et al.
Publicado: (2025)

Mettle: Meta-Token Learning for Memory-Efficient Audio-Visual Adaptation
por: Zhou, Jinxing, et al.
Publicado: (2025)

Action Tokenizer Matters in In-Context Imitation Learning
por: Vuong, An Dinh, et al.
Publicado: (2025)

MMTU: A Massive Multi-Task Table Understanding and Reasoning Benchmark
por: Xing, Junjie, et al.
Publicado: (2025)

Video-KTR: Reinforcing Video Reasoning via Key Token Attribution
por: Wang, Ziyue, et al.
Publicado: (2026)

TAP4LLM: Table Provider on Sampling, Augmenting, and Packing Semi-structured Data for Large Language Model Reasoning
por: Sui, Yuan, et al.
Publicado: (2023)

HeartLLM: Discretized ECG Tokenization for LLM-Based Diagnostic Reasoning
por: Yang, Jinning, et al.
Publicado: (2025)

One Tokenizer To Rule Them All: Emergent Language Plasticity via Multilingual Tokenizers
por: Abagyan, Diana, et al.
Publicado: (2025)

Reasoning with Reinforced Functional Token Tuning
por: Zhang, Kongcheng, et al.
Publicado: (2025)

DUET: Optimize Token-Budget Allocation for Reinforcement Learning with Verifiable Rewards
por: Hu, Haoyu, et al.
Publicado: (2026)

In-Token Rationality Optimization: Towards Accurate and Concise LLM Reasoning via Self-Feedback
por: Zhu, Mingye, et al.
Publicado: (2025)

OrthoRank: Token Selection via Sink Token Orthogonality for Efficient LLM inference
por: Shin, Seungjun, et al.
Publicado: (2025)

Efficient Cold-Start Recommendation via BPE Token-Level Embedding Initialization with LLM
por: Zhao, Yushang, et al.
Publicado: (2025)

Reasoning Physical Video Generation with Diffusion Timestep Tokens via Reinforcement Learning
por: Lin, Wang, et al.
Publicado: (2025)

Is One Token All It Takes? Graph Pooling Tokens for LLM-based GraphQA
por: Grover, Ankit, et al.
Publicado: (2026)

Extending Token Computation for LLM Reasoning
por: Liao, Bingli, et al.
Publicado: (2024)

Beyond the Next Token: Towards Prompt-Robust Zero-Shot Classification via Efficient Multi-Token Prediction
por: Qian, Junlang, et al.
Publicado: (2025)

Adaptive Token-Weighted Differential Privacy for LLMs: Not All Tokens Require Equal Protection
por: Yu, Manjiang, et al.
Publicado: (2025)