paper-with-me

홈 › Papers

Entropy-Gradient Inversion: Moving Toward Internal Mechanism of Large Reasoning Models

2026-05-18 · Junyao Yang, Chen Qian, Kun Wang, Linfeng Zhang, Quanshi Zhang, Yong Liu, Dongrui Liu arxiv

The advancement of Large Reasoning Models (LRMs) has catalyzed a paradigm shift from reactive `fast thinking'' text generation to systematic, step-by-step `slow thinking'' reasoning, unlocking state-of-the-art performance in complex mathematical and logical tasks. However, the field faces \textit{the fundamental gap between token-level behavioral analysis and internal reasoning mechanisms, and the instability of reinforcement learning (RL) for reasoning optimization relying on costly external verifiers}. We identify and formally define \textbf{Entropy-Gradient Inversion}, a robust negative correlation between token entropy and logit gradients that acts as a definitive geometric fingerprint for LRM reasoning capability. Building on this, we propose \textbf{Correlation-Regularized Group Policy Optimization (CorR-PO)}, which embeds this inversion signature into RL reward regularization. Extensive experiments on various reasoning benchmarks across multiple model scales show CorR-PO consistently outperforms state-of-the-art baselines, confirming that stronger inversion directly correlates with superior reasoning performance.

📄 PDF Abstract BibTeX arXiv:2605.17770

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningText Generation

Similar Papers 제목 키워드 기반

Each Prompt Matters: Scaling Reinforcement Learning Without Wasting Rollouts on Hundred-Billion-Scale MoE

2025-12-08 · Anxiang Zeng, Haibo Zhang, Hailing Zhang, Kaixiang Mo 외 arxiv

We present CompassMax-V3-Thinking, a hundred-billion-scale MoE reasoning model trained with a new RL framework built on one principle: each prompt must matter. Scaling RL to this size exposes critical inefficiencies-zero…

Reinforcement Learning

Temperature-driven inversion and nonlinear dynamics in ChatGPT-like AIs

2026-08-02 · Neil F. Johnson, Frank Yingjie Huo, Bella Xinrui Li arxiv

Increasing the temperature of an ordinary many-state system increases access to a wider range of states and hence increases its entropy. We find the opposite in ChatGPT-like AIs, even though raising the decoder temperatu…

Mitigating Gradient Inversion Risks in Language Models via Token Obfuscation

2026-02-11 · Xinguo Feng, Zhongkui Ma, Zihan Wang, Alsharif Abuadbba 외 arxiv

Training and fine-tuning large-scale language models largely benefit from collaborative learning, but the approach has been proven vulnerable to gradient inversion attacks (GIAs), which allow adversaries to reconstruct p…

Theoretical Insights in Model Inversion Robustness and Conditional Entropy Maximization for Collaborative Inference Systems

2025-03-01 · CVPR 2025 1 · Song Xia, Yi Yu, Wenhan Yang, Meiwen Ding 외

By locally encoding raw data into intermediate features, collaborative inference enables end users to leverage powerful deep learning models without exposure of sensitive raw data to cloud servers. However, recent studie…

Collaborative InferenceRepresentation Learning

Neurons as an Information-theoretic Engine

2017-12-27

We show that dynamical gain modulation of neurons' stimulus response is described as an information-theoretic cycle that generates entropy associated with the stimulus-related activity from entropy produced by the modula…