Cover Image

Thinking as Compression: Your Reasoning Model is Secretly a Context Compressor

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Ma, Guoxin, Liu, Yibing, Li, Chengzhengxu, Liang, Yu, Wang, Yan, Zhang, Yueyang, Chen, Kecheng, Zhang, Zhaohan, Sun, Zhiyuan, Shi, Daiting
Format:	Preprint
Published:	2026
Subjects:	Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

When Less is More: The LLM Scaling Paradox in Context Compression
by: Guo, Ruishan, et al.
Published: (2026)

Upfront Chain-of-Thought: A Cooperative Framework for Chain-of-Thought Compression
by: Li, Chengzhengxu, et al.
Published: (2025)

Advancing General-Purpose Reasoning Models with Modular Gradient Surgery
by: Cai, Min, et al.
Published: (2026)

MGT-Prism: Enhancing Domain Generalization for Machine-Generated Text Detection via Spectral Alignment
by: Liu, Shengchao, et al.
Published: (2025)

ConsistRM: Improving Generative Reward Models via Consistency-Aware Self-Training
by: Liang, Yu, et al.
Published: (2026)

ReflectRM: Boosting Generative Reward Models via Self-Reflection within a Unified Judgment Framework
by: Qin, Kai, et al.
Published: (2026)

DEER: Disentangled Mixture of Experts with Instance-Adaptive Routing for Generalizable Machine-Generated Text Detection
by: Ma, Guoxin, et al.
Published: (2025)

Concentrate Attention: Towards Domain-Generalizable Prompt Optimization for Language Models
by: Li, Chengzhengxu, et al.
Published: (2024)

Iron Sharpens Iron: Defending Against Attacks in Machine-Generated Text Detection with Adversarial Training
by: Li, Yuanfan, et al.
Published: (2025)

StablePT: Towards Stable Prompting for Few-shot Learning via Input Separation
by: Liu, Xiaoming, et al.
Published: (2024)

All Your Knowledge Belongs to Us: Stealing Knowledge Graphs via Reasoning APIs
by: Xi, Zhaohan
Published: (2025)

Squeeze3D: Your 3D Generation Model is Secretly an Extreme Neural Compressor
by: Dagli, Rishit, et al.
Published: (2025)

Confidence Should Be Calibrated More Than One Turn Deep
by: Zhang, Zhaohan, et al.
Published: (2026)

Large Language Models for Lossless Image Compression: Next-Pixel Prediction in Language Space is All You Need
by: Chen, Kecheng, et al.
Published: (2024)

Your Dense Retriever is Secretly an Expeditious Reasoner
by: Zhang, Yichi, et al.
Published: (2025)

TRE: Encouraging Exploration in the Trust Region
by: Huang, Chao, et al.
Published: (2026)

Does DetectGPT Fully Utilize Perturbation? Bridging Selective Perturbation to Fine-tuned Contrastive Learning Detector would be Better
by: Liu, Shengchao, et al.
Published: (2024)

Perception Compressor: A Training-Free Prompt Compression Framework in Long Context Scenarios
by: Tang, Jiwei, et al.
Published: (2024)

Flora: Low-Rank Adapters Are Secretly Gradient Compressors
by: Hao, Yongchang, et al.
Published: (2024)

Does Your Reasoning Model Implicitly Know When to Stop Thinking?
by: Huang, Zixuan, et al.
Published: (2026)

LightTransfer: Your Long-Context LLM is Secretly a Hybrid Model with Effortless Adaptation
by: Zhang, Xuan, et al.
Published: (2024)

Latent Thinking Optimization: Your Latent Reasoning Language Model Secretly Encodes Reward Signals in Its Latent Thoughts
by: Du, Hanwen, et al.
Published: (2025)

MGTEVAL: An Interactive Platform for Systemtic Evaluation of Machine-Generated Text Detectors
by: Li, Yuanfan, et al.
Published: (2026)

MC#: Mixture Compressor for Mixture-of-Experts Large Models
by: Huang, Wei, et al.
Published: (2025)

It Takes Two: Your GRPO Is Secretly DPO
by: Wu, Yihong, et al.
Published: (2025)

InftyThink: Breaking the Length Limits of Long-Context Reasoning in Large Language Models
by: Yan, Yuchen, et al.
Published: (2025)

GenCRF: Generative Clustering and Reformulation Framework for Enhanced Intent-Driven Information Retrieval
by: Seo, Wonduk, et al.
Published: (2024)

Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator
by: Zheng, Kaiwen, et al.
Published: (2025)

Large Language Models as an Indirect Reasoner: Contrapositive and Contradiction for Automated Reasoning
by: Zhang, Yanfang, et al.
Published: (2024)

Compressor-VLA: Instruction-Guided Visual Token Compression for Efficient Robotic Manipulation
by: Gao, Juntao, et al.
Published: (2025)

On the Context-Hiding Property of Shamir-Based Homomorphic Secret Sharing
by: Feng, Shuai, et al.
Published: (2025)

500xCompressor: Generalized Prompt Compression for Large Language Models
by: Li, Zongqian, et al.
Published: (2024)

YODA: Yet Another One-step Diffusion-based Video Compressor
by: Li, Xingchen, et al.
Published: (2026)

Your Language Model Secretly Contains Personality Subnetworks
by: Ye, Ruimeng, et al.
Published: (2026)

Beyond Pairs: Your Language Model is Secretly Optimizing a Preference Graph
by: Liu, Ning, et al.
Published: (2026)

Is Your LLM Secretly a World Model of the Internet? Model-Based Planning for Web Agents
by: Gu, Yu, et al.
Published: (2024)

Think-RM: Enabling Long-Horizon Reasoning in Generative Reward Models
by: Hong, Ilgee, et al.
Published: (2025)

TourRank: Utilizing Large Language Models for Documents Ranking with a Tournament-Inspired Strategy
by: Chen, Yiqun, et al.
Published: (2024)

Think How to Think: Mitigating Overthinking with Autonomous Difficulty Cognition in Large Reasoning Models
by: Liu, Yongjiang, et al.
Published: (2025)

DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models
by: Shen, Yi, et al.
Published: (2025)