Controllable Mathematical Reasoning via Self-Optimizing Thought Vectors
Fuente:
arXiv
Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908611742007296 |
|---|---|
| author | LI, Xuying |
| author_facet | LI, Xuying |
| contents | We present a novel approach for controllable mathematical reasoning that leverages self-optimizing thought vectors with entropy minimization. Our method introduces learnable thought vectors that dynamically modulate the internal reasoning process of large language models. Using Gemma-2-9B on GSM8K, we achieve 90.1% accuracy with a controllability score of 0.42, demonstrating that entropy-based rewards effectively guide focused reasoning patterns without requiring external reward annotations. Our analysis reveals distinct thought vector clusters and consistent low-entropy distributions across control conditions, validating our framework for controllable AI reasoning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_22132 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Controllable Mathematical Reasoning via Self-Optimizing Thought Vectors LI, Xuying Artificial Intelligence We present a novel approach for controllable mathematical reasoning that leverages self-optimizing thought vectors with entropy minimization. Our method introduces learnable thought vectors that dynamically modulate the internal reasoning process of large language models. Using Gemma-2-9B on GSM8K, we achieve 90.1% accuracy with a controllability score of 0.42, demonstrating that entropy-based rewards effectively guide focused reasoning patterns without requiring external reward annotations. Our analysis reveals distinct thought vector clusters and consistent low-entropy distributions across control conditions, validating our framework for controllable AI reasoning. |
| title | Controllable Mathematical Reasoning via Self-Optimizing Thought Vectors |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2510.22132 |