Explainable Chain-of-Thought Reasoning: An Empirical Analysis on State-Aware Reasoning Dynamics
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915760960438272 |
|---|---|
| author | Yu, Sheldon Xiong, Yuxin Wu, Junda Li, Xintong Yu, Tong Chen, Xiang Sinha, Ritwik Shang, Jingbo McAuley, Julian |
| author_facet | Yu, Sheldon Xiong, Yuxin Wu, Junda Li, Xintong Yu, Tong Chen, Xiang Sinha, Ritwik Shang, Jingbo McAuley, Julian |
| contents | Recent advances in chain-of-thought (CoT) prompting have enabled large language models (LLMs) to perform multi-step reasoning. However, the explainability of such reasoning remains limited, with prior work primarily focusing on local token-level attribution, such that the high-level semantic roles of reasoning steps and their transitions remain underexplored. In this paper, we introduce a state-aware transition framework that abstracts CoT trajectories into structured latent dynamics. Specifically, to capture the evolving semantics of CoT reasoning, each reasoning step is represented via spectral analysis of token-level embeddings and clustered into semantically coherent latent states. To characterize the global structure of reasoning, we model their progression as a Markov chain, yielding a structured and interpretable view of the reasoning process. This abstraction supports a range of analyses, including semantic role identification, temporal pattern visualization, and consistency evaluation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_00190 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Explainable Chain-of-Thought Reasoning: An Empirical Analysis on State-Aware Reasoning Dynamics Yu, Sheldon Xiong, Yuxin Wu, Junda Li, Xintong Yu, Tong Chen, Xiang Sinha, Ritwik Shang, Jingbo McAuley, Julian Computation and Language Artificial Intelligence Recent advances in chain-of-thought (CoT) prompting have enabled large language models (LLMs) to perform multi-step reasoning. However, the explainability of such reasoning remains limited, with prior work primarily focusing on local token-level attribution, such that the high-level semantic roles of reasoning steps and their transitions remain underexplored. In this paper, we introduce a state-aware transition framework that abstracts CoT trajectories into structured latent dynamics. Specifically, to capture the evolving semantics of CoT reasoning, each reasoning step is represented via spectral analysis of token-level embeddings and clustered into semantically coherent latent states. To characterize the global structure of reasoning, we model their progression as a Markov chain, yielding a structured and interpretable view of the reasoning process. This abstraction supports a range of analyses, including semantic role identification, temporal pattern visualization, and consistency evaluation. |
| title | Explainable Chain-of-Thought Reasoning: An Empirical Analysis on State-Aware Reasoning Dynamics |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2509.00190 |