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

Juno Kim, Denny Wu, Jason D. Lee, Taiji Suzuki. Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation. In Forty-second International Conference on Machine Learning, ICML 2025, Vancouver, BC, Canada, July 13-19, 2025. OpenReview.net, 2025. [doi]

@inproceedings{KimWLS25,
  title = {Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation},
  author = {Juno Kim and Denny Wu and Jason D. Lee and Taiji Suzuki},
  year = {2025},
  url = {https://openreview.net/forum?id=2HJcVtuovs},
  researchr = {https://researchr.org/publication/KimWLS25},
  cites = {0},
  citedby = {0},
  booktitle = {Forty-second International Conference on Machine Learning, ICML 2025, Vancouver, BC, Canada, July 13-19, 2025},
  publisher = {OpenReview.net},
}