Cover Image

Anti-Self-Distillation for Reasoning RL via Pointwise Mutual Information

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Shen, Guobin, Cheng, Xiang, Zhao, Chenxiao, Huang, Lei, Li, Jindong, Zhao, Dongcheng, Yu, Xing
Format:	Preprint
Published:	2026
Subjects:	Machine Learning Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

From Generic Correlation to Input-Specific Credit in On-Policy Self Distillation
by: Shen, Guobin, et al.
Published: (2026)

VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training
by: Shen, Guobin, et al.
Published: (2026)

Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning
by: Huang, Lei, et al.
Published: (2026)

Pointwise Mutual Information as a Performance Gauge for Retrieval-Augmented Generation
by: Liu, Tianyu, et al.
Published: (2024)

Safety Instincts: LLMs Learn to Trust Their Internal Compass for Self-Defense
by: Shen, Guobin, et al.
Published: (2025)

Towards More Accurate US Presidential Election via Multi-step Reasoning with Large Language Models
by: Yu, Chenxiao, et al.
Published: (2024)

Mitigating Hallucinations in Large Language Models via Causal Reasoning
by: Li, Yuangang, et al.
Published: (2025)

Diverse Policies Recovering via Pointwise Mutual Information Weighted Imitation Learning
by: Yang, Hanlin, et al.
Published: (2024)

StressPrompt: Does Stress Impact Large Language Models and Human Performance Similarly?
by: Shen, Guobin, et al.
Published: (2024)

Light Alignment Improves LLM Safety via Model Self-Reflection with a Single Neuron
by: Shen, Sicheng, et al.
Published: (2026)

Learning to Maximize Mutual Information for Chain-of-Thought Distillation
by: Chen, Xin, et al.
Published: (2024)

MVPBench: A Benchmark and Fine-Tuning Framework for Aligning Large Language Models with Diverse Human Values
by: Liang, Yao, et al.
Published: (2025)

Decoupling KL and Trajectories: A Unified Perspective for SFT, DAgger, Offline RL, and OPD in LLM Distillation
by: Zhao, Anhao, et al.
Published: (2026)

Multi-Level Safety Continual Projection for Fine-Tuned Large Language Models without Retraining
by: Han, Bing, et al.
Published: (2025)

Multilingual Safety Alignment via Self-Distillation
by: Qin, Ruiyang, et al.
Published: (2026)

PromptBench: A Unified Library for Evaluation of Large Language Models
by: Zhu, Kaijie, et al.
Published: (2023)

Dynamic Evaluation of Large Language Models by Meta Probing Agents
by: Zhu, Kaijie, et al.
Published: (2024)

GEAR: Granularity-Adaptive Advantage Reweighting for LLM Agents via Self-Distillation
by: Li, Sijia, et al.
Published: (2026)

Self-Distillation for Multi-Token Prediction
by: Zhao, Guoliang, et al.
Published: (2026)

MITS: Enhanced Tree Search Reasoning for LLMs via Pointwise Mutual Information
by: Li, Jiaxi, et al.
Published: (2025)

Scaling Up RL: Unlocking Diverse Reasoning in LLMs via Prolonged Training
by: Liu, Mingjie, et al.
Published: (2025)

Skill-Conditioned Gated Self-Distillation for LLM Reasoning
by: Huang, Jiazhen, et al.
Published: (2026)

SAGE-32B: Agentic Reasoning via Iterative Distillation
by: Jha, Basab, et al.
Published: (2026)

Efficient Mathematical Reasoning Models via Dynamic Pruning and Knowledge Distillation
by: Yu, Fengming, et al.
Published: (2025)

DyVal: Dynamic Evaluation of Large Language Models for Reasoning Tasks
by: Zhu, Kaijie, et al.
Published: (2023)

RL Tango: Reinforcing Generator and Verifier Together for Language Reasoning
by: Zha, Kaiwen, et al.
Published: (2025)

StealthRank: LLM Ranking Manipulation via Stealthy Prompt Optimization
by: Tang, Yiming, et al.
Published: (2025)

UniSD: Towards a Unified Self-Distillation Framework for Large Language Models
by: Jin, Yiqiao, et al.
Published: (2026)

CultureLLM: Incorporating Cultural Differences into Large Language Models
by: Li, Cheng, et al.
Published: (2024)

Efficient Process Reward Modeling via Contrastive Mutual Information
by: Lee, Nakyung, et al.
Published: (2026)

Merge-of-Thought Distillation
by: Shen, Zhanming, et al.
Published: (2025)

Multimodal Generative Engine Optimization: Rank Manipulation for Vision-Language Model Rankers
by: Du, Yixuan, et al.
Published: (2026)

UnStar: Unlearning with Self-Taught Anti-Sample Reasoning for LLMs
by: Sinha, Yash, et al.
Published: (2024)

SDMPrune: Self-Distillation MLP Pruning for Efficient Large Language Models
by: Zhu, Hourun, et al.
Published: (2025)

TTSR: Test-Time Self-Reflection for Continual Reasoning Improvement
by: He, Haoyang, et al.
Published: (2026)

CoVerRL: Breaking the Consensus Trap in Label-Free Reasoning via Generator-Verifier Co-Evolution
by: Pan, Teng, et al.
Published: (2026)

Structural Rationale Distillation via Reasoning Space Compression
by: Yang, Jialin, et al.
Published: (2026)

Self-ReSET: Learning to Self-Recover from Unsafe Reasoning Trajectories
by: Zhang, Dongcheng, et al.
Published: (2026)

NeSTR: A Neuro-Symbolic Abductive Framework for Temporal Reasoning in Large Language Models
by: Liang, Feng, et al.
Published: (2025)

Steering Large Reasoning Models towards Concise Reasoning via Flow Matching
by: Li, Yawei, et al.
Published: (2026)