Copertina

CyclicReflex: Improving Reasoning Models via Cyclical Reflection Token Scheduling

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Fan, Chongyu, Zhang, Yihua, Jia, Jinghan, Hero, Alfred, Liu, Sijia
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond
di: Fan, Chongyu, et al.
Pubblicazione: (2025)

Challenging Forgets: Unveiling the Worst-Case Forget Sets in Machine Unlearning
di: Fan, Chongyu, et al.
Pubblicazione: (2024)

Beyond SFT: Reinforcement Learning for Safer Large Reasoning Models with Better Reasoning Ability
di: Jia, Jinghan, et al.
Pubblicazione: (2025)

Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning
di: Fan, Chongyu, et al.
Pubblicazione: (2024)

SEUF: Is Unlearning One Expert Enough for Mixture-of-Experts LLMs?
di: Zhuang, Haomin, et al.
Pubblicazione: (2024)

SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning
di: Jia, Jinghan, et al.
Pubblicazione: (2024)

Reasoning Model Unlearning: Forgetting Traces, Not Just Answers, While Preserving Reasoning Skills
di: Wang, Changsheng, et al.
Pubblicazione: (2025)

One Token Embedding Is Enough to Deadlock Your Large Reasoning Model
di: Zhang, Mohan, et al.
Pubblicazione: (2025)

Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models
di: Zhang, Yimeng, et al.
Pubblicazione: (2024)

Downgrade to Upgrade: Optimizer Simplification Enhances Robustness in LLM Unlearning
di: Lang, Yicheng, et al.
Pubblicazione: (2025)

LLM Unlearning Under the Microscope: A Full-Stack View on Methods and Metrics
di: Fan, Chongyu, et al.
Pubblicazione: (2025)

Zero Token-Driven Deep Thinking in LLMs: Unlocking the Full Potential of Existing Parameters via Cyclic Refinement
di: Li, Guanghao, et al.
Pubblicazione: (2025)

Improved Neural Protoform Reconstruction via Reflex Prediction
di: Lu, Liang, et al.
Pubblicazione: (2024)

A Multiscale Geometric Method for Capturing Relational Topic Alignment
di: Hougen, Conrad D., et al.
Pubblicazione: (2025)

UnlearnCanvas: Stylized Image Dataset for Enhanced Machine Unlearning Evaluation in Diffusion Models
di: Zhang, Yihua, et al.
Pubblicazione: (2024)

Vision-Language Models Can Self-Improve Reasoning via Reflection
di: Cheng, Kanzhi, et al.
Pubblicazione: (2024)

Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning
di: Su, DiJia, et al.
Pubblicazione: (2025)

Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning
di: Jia, Jinghan, et al.
Pubblicazione: (2026)

Improve Decoding Factuality by Token-wise Cross Layer Entropy of Large Language Models
di: Wu, Jialiang, et al.
Pubblicazione: (2025)

Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearning
di: Shang, Bingqi, et al.
Pubblicazione: (2025)

EPiC: Towards Lossless Speedup for Reasoning Training through Edge-Preserving CoT Condensation
di: Jia, Jinghan, et al.
Pubblicazione: (2025)

Bring Reason to Vision: Understanding Perception and Reasoning through Model Merging
di: Chen, Shiqi, et al.
Pubblicazione: (2025)

Cyclic Proofs in Hoare Logic and its Reverse
di: Brotherston, James, et al.
Pubblicazione: (2025)

Disentangling Reasoning Tokens and Boilerplate Tokens For Language Model Fine-tuning
di: Ye, Ziang, et al.
Pubblicazione: (2024)

CyclicJudge: Mitigating Judge Bias Efficiently in LLM-based Evaluation
di: Zhu, Ziyi, et al.
Pubblicazione: (2026)

Complete the Cycle: Reachability Types with Expressive Cyclic References (Extended Version)
di: Deng, Haotian, et al.
Pubblicazione: (2025)

Instruct-of-Reflection: Enhancing Large Language Models Iterative Reflection Capabilities via Dynamic-Meta Instruction
di: Liu, Liping, et al.
Pubblicazione: (2025)

FINEREASON: Evaluating and Improving LLMs' Deliberate Reasoning through Reflective Puzzle Solving
di: Chen, Guizhen, et al.
Pubblicazione: (2025)

Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoning
di: Zhang, Zhihan, et al.
Pubblicazione: (2024)

Transitivity Meets Cyclicity: Explicit Preference Decomposition for Dynamic Large Language Model Alignment
di: Huang, Yucong, et al.
Pubblicazione: (2026)

Think Clearly: Improving Reasoning via Redundant Token Pruning
di: Choi, Daewon, et al.
Pubblicazione: (2025)

iCLP: Large Language Model Reasoning with Implicit Cognition Latent Planning
di: Chen, Sijia, et al.
Pubblicazione: (2025)

Blind Spot Navigation in Large Language Model Reasoning with Thought Space Explorer
di: Zhang, Jinghan, et al.
Pubblicazione: (2024)

Enhancing Large Language Model Reasoning via Selective Critical Token Fine-Tuning
di: Ruan, Zhiwen, et al.
Pubblicazione: (2025)

One Token Is Enough: Improving Diffusion Language Models with a Sink Token
di: Zhang, Zihou, et al.
Pubblicazione: (2026)

Efficient Reasoning Through Suppression of Self-Affirmation Reflections in Large Reasoning Models
di: Liu, Kaiyuan, et al.
Pubblicazione: (2025)

Arithmetic in the Wild: Llama uses Base-10 Addition to Reason About Cyclic Concepts
di: Feucht, Sheridan, et al.
Pubblicazione: (2026)

RATT: A Thought Structure for Coherent and Correct LLM Reasoning
di: Zhang, Jinghan, et al.
Pubblicazione: (2024)

SEED: Accelerating Reasoning Tree Construction via Scheduled Speculative Decoding
di: Wang, Zhenglin, et al.
Pubblicazione: (2024)

Order Matters in Hallucination: Reasoning Order as Benchmark and Reflexive Prompting for Large-Language-Models
di: Xie, Zikai
Pubblicazione: (2024)