Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Kim, Juno, Wu, Denny, Lee, Jason, Suzuki, Taiji
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Transformers Provably Solve Parity Efficiently with Chain of Thought
by: Kim, Juno, et al.
Published: (2024)

Transformers Learn Nonlinear Features In Context: Nonconvex Mean-field Dynamics on the Attention Landscape
by: Kim, Juno, et al.
Published: (2024)

Transformers Provably Learn Chain-of-Thought Reasoning with Length Generalization
by: Huang, Yu, et al.
Published: (2025)

Drop the Act: Probe-Filtered RL for Faithful Chain-of-Thought Reasoning
by: Parekh, Swapnil
Published: (2026)

A Hierarchical Language Model with Predictable Scaling Laws and Provable Benefits of Reasoning
by: Gaitonde, Jason, et al.
Published: (2026)

Learning from Partial Chain-of-Thought via Truncated-Reasoning Self-Distillation
by: Silvestri, Gianluigi, et al.
Published: (2026)

Dynamic Chain-of-Thought: Towards Adaptive Deep Reasoning
by: Wang, Libo
Published: (2025)

Provably Learning Diffusion Models under the Manifold Hypothesis: Collapse and Refine
by: Huang, Wei, et al.
Published: (2026)

Provable In-Context Vector Arithmetic via Retrieving Task Concepts
by: Bu, Dake, et al.
Published: (2025)

Value-Guided Search for Efficient Chain-of-Thought Reasoning
by: Wang, Kaiwen, et al.
Published: (2025)

Transformers Learn to Implement Multi-step Gradient Descent with Chain of Thought
by: Huang, Jianhao, et al.
Published: (2025)

Optimizing Chain-of-Thought Reasoners via Gradient Variance Minimization in Rejection Sampling and RL
by: Yao, Jiarui, et al.
Published: (2025)

Outcome-Based RL Provably Leads Transformers to Reason, but Only With the Right Data
by: Ran-Milo, Yuval, et al.
Published: (2026)

Unveiling Confirmation Bias in Chain-of-Thought Reasoning
by: Wan, Yue, et al.
Published: (2025)

CoTox: Chain-of-Thought-Based Molecular Toxicity Reasoning and Prediction
by: Park, Jueon, et al.
Published: (2025)

Diffusion of Thoughts: Chain-of-Thought Reasoning in Diffusion Language Models
by: Ye, Jiacheng, et al.
Published: (2024)

Fractured Chain-of-Thought Reasoning
by: Liao, Baohao, et al.
Published: (2025)

Reinforcing Chain-of-Thought Reasoning with Self-Evolving Rubrics
by: Sheng, Leheng, et al.
Published: (2026)

DAG-Math: Graph-of-Thought Guided Mathematical Reasoning in LLMs
by: Zhang, Yuanhe, et al.
Published: (2025)

The Ends Justify the Thoughts: RL-Induced Motivated Reasoning in LLM CoTs
by: Howe, Nikolaus, et al.
Published: (2025)

Transformers are Minimax Optimal Nonparametric In-Context Learners
by: Kim, Juno, et al.
Published: (2024)

Provable Benefits of In-Tool Learning for Large Language Models
by: Houliston, Sam, et al.
Published: (2025)

Provable Long-Range Benefits of Next-Token Prediction
by: Cao, Xinyuan, et al.
Published: (2025)

Provable Benefit of Cutout and CutMix for Feature Learning
by: Oh, Junsoo, et al.
Published: (2024)

Accelerating RL for LLM Reasoning with Optimal Advantage Regression
by: Brantley, Kianté, et al.
Published: (2025)

The Kinetics of Reasoning: How Chain-of-Thought Shapes Learning in Transformers?
by: Pengmei, Zihan, et al.
Published: (2025)

LongCoT: Benchmarking Long-Horizon Chain-of-Thought Reasoning
by: Motwani, Sumeet Ramesh, et al.
Published: (2026)

Neural network learns low-dimensional polynomials with SGD near the information-theoretic limit
by: Lee, Jason D., et al.
Published: (2024)

Curriculum Learning for Efficient Chain-of-Thought Distillation via Structure-Aware Masking and GRPO
by: Yu, Bowen, et al.
Published: (2026)

Provable Benefit of Curriculum in Transformer Tree-Reasoning Post-Training
by: Bu, Dake, et al.
Published: (2025)

Symbolic Analysis of Grover Search Algorithm via Chain-of-Thought Reasoning and Quantum-Native Tokenization
by: Chen, Min, et al.
Published: (2025)

Token-Efficient RL for LLM Reasoning
by: Lee, Alan, et al.
Published: (2025)

State Space Models are Provably Comparable to Transformers in Dynamic Token Selection
by: Nishikawa, Naoki, et al.
Published: (2024)

Re-FORC: Adaptive Reward Prediction for Efficient Chain-of-Thought Reasoning
by: Zabounidis, Renos, et al.
Published: (2025)

Temporalizing Confidence: Evaluation of Chain-of-Thought Reasoning with Signal Temporal Logic
by: Mao, Zhenjiang, et al.
Published: (2025)

Improving Chain-of-Thought for Logical Reasoning via Attention-Aware Intervention
by: Phuong, Nguyen Minh, et al.
Published: (2026)

Chain-of-Thought Reasoning In The Wild Is Not Always Faithful
by: Arcuschin, Iván, et al.
Published: (2025)

Compositional Reasoning with Transformers, RNNs, and Chain of Thought
by: Yehudai, Gilad, et al.
Published: (2025)

Scalable Chain of Thoughts via Elastic Reasoning
by: Xu, Yuhui, et al.
Published: (2025)

Long Chain-of-Thought Reasoning Across Languages
by: Barua, Josh, et al.
Published: (2025)