Buchumschlag

Thinking-Free Policy Initialization Makes Distilled Reasoning Models More Effective and Efficient Reasoners

Fuente: arXiv

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Xu, Xin, AI, Cliveb, Yang, Kai, Chen, Tianhao, Wang, Yang, Yang, Saiyong, Yang, Can
Format:	Preprint
Veröffentlicht:	2025
Schlagworte:	Machine Learning Computation and Language
Online-Zugang:	Acceder al recurso
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Ähnliche Einträge

Learning beyond Teacher: Generalized On-Policy Distillation with Reward Extrapolation
von: Yang, Wenkai, et al.
Veröffentlicht: (2026)

Reasoning-preserved Efficient Distillation of Large Language Models via Activation-aware Initialization
von: He, Junlin, et al.
Veröffentlicht: (2026)

ROSD: Reflective On-Policy Self-Distillation for Language Model Reasoning across Domains
von: Zhao, Ziqi, et al.
Veröffentlicht: (2026)

Think Beyond Size: Adaptive Prompting for More Effective Reasoning
von: R, Kamesh
Veröffentlicht: (2024)

Dynamic Thinking-Token Selection for Efficient Reasoning in Large Reasoning Models
von: Guo, Zhenyuan, et al.
Veröffentlicht: (2026)

Scaling Reasoning Efficiently via Relaxed On-Policy Distillation
von: Ko, Jongwoo, et al.
Veröffentlicht: (2026)

Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning
von: Shrivastava, Vaishnavi, et al.
Veröffentlicht: (2025)

More Thinking, More Bias: Length-Driven Position Bias in Reasoning Models
von: Wang, Xiao
Veröffentlicht: (2026)

Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models
von: Zhao, Siyan, et al.
Veröffentlicht: (2026)

ThinkRouter: Efficient Reasoning via Routing Thinking between Latent and Discrete Spaces
von: Xu, Xin, et al.
Veröffentlicht: (2026)

Reverse Thinking Makes LLMs Stronger Reasoners
von: Chen, Justin Chih-Yao, et al.
Veröffentlicht: (2024)

Democratizing Tool Learning with Environments Fully Simulated by a Free 8B Language Model
von: Tang, Chenming, et al.
Veröffentlicht: (2026)

Efficient Reasoning with Balanced Thinking
von: Li, Yulin, et al.
Veröffentlicht: (2026)

Effectively Controlling Reasoning Models through Thinking Intervention
von: Wu, Tong, et al.
Veröffentlicht: (2025)

Efficient Reasoning with Hidden Thinking
von: Shen, Xuan, et al.
Veröffentlicht: (2025)

Mid-Think: Training-Free Intermediate-Budget Reasoning via Token-Level Triggers
von: Yang, Wang, et al.
Veröffentlicht: (2026)

Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMs
von: Yang, Dayu, et al.
Veröffentlicht: (2025)

Relative Kinetic Utility for Reasoning-Aware Structural Pruning in Large Language Models
von: Qian, Tianhao
Veröffentlicht: (2026)

Think Outside the Policy: In-Context Steered Policy Optimization
von: Huang, Hsiu-Yuan, et al.
Veröffentlicht: (2025)

Internalize the Temperature: On-Policy Self-Distillation as Policy Reheater for Reinforcement Learning
von: Yang, Xuewei, et al.
Veröffentlicht: (2026)

AdapThink: Adaptive Thinking Preferences for Reasoning Language Model
von: Wan, Xu, et al.
Veröffentlicht: (2025)

Are Reasoning Models More Prone to Hallucination?
von: Yao, Zijun, et al.
Veröffentlicht: (2025)

Reasoning Models Don't Just Think Longer, They Move Differently
von: Gjølbye, Anders, et al.
Veröffentlicht: (2026)

Asynchronous Reasoning: Training-Free Interactive Thinking LLMs
von: Yakushev, George, et al.
Veröffentlicht: (2025)

Composition-RL: Compose Your Verifiable Prompts for Reinforcement Learning of Large Language Models
von: Xu, Xin, et al.
Veröffentlicht: (2026)

What makes Reasoning Models Different? Follow the Reasoning Leader for Efficient Decoding
von: Li, Ming, et al.
Veröffentlicht: (2025)

Rebellious Student: Reversing Teacher Signals for Reasoning Exploration with Self-Distilled RLVR
von: Kim, Jeonghye, et al.
Veröffentlicht: (2026)

Agentic-R1: Distilled Dual-Strategy Reasoning
von: Du, Weihua, et al.
Veröffentlicht: (2025)

An Analysis for Reasoning Bias of Language Models with Small Initialization
von: Yao, Junjie, et al.
Veröffentlicht: (2025)

To Think or Not to Think: Exploring the Unthinking Vulnerability in Large Reasoning Models
von: Zhu, Zihao, et al.
Veröffentlicht: (2025)

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

Exploring the Reversal Curse and Other Deductive Logical Reasoning in BERT and GPT-Based Large Language Models
von: Wu, Da, et al.
Veröffentlicht: (2023)

Think Before You Prune: Self-Reflective Structured Pruning for Reasoning Language Models
von: Wang, Ziyan, et al.
Veröffentlicht: (2025)

Recursive Models for Long-Horizon Reasoning
von: Yang, Chenxiao, et al.
Veröffentlicht: (2026)

AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agent
von: Luo, Haipeng, et al.
Veröffentlicht: (2025)

Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models
von: Qu, Yun, et al.
Veröffentlicht: (2026)

Longer Context, Deeper Thinking: Uncovering the Role of Long-Context Ability in Reasoning
von: Yang, Wang, et al.
Veröffentlicht: (2025)

Good Learners Think Their Thinking: Generative PRM Makes Large Reasoning Model More Efficient Math Learner
von: He, Tao, et al.
Veröffentlicht: (2025)

Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?
von: Kim, Jeonghye, et al.
Veröffentlicht: (2026)

Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation
von: Wang, Xinyi, et al.
Veröffentlicht: (2024)