Mitigating Strategy-Selection Bias in Reasoning for More Effective Test-Time Scaling
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908554461446144 |
|---|---|
| author | Wu, Zongqian Xu, Baoduo Li, Tianyu Sun, Zhu Zhu, Xiaofeng Feng, Lei |
| author_facet | Wu, Zongqian Xu, Baoduo Li, Tianyu Sun, Zhu Zhu, Xiaofeng Feng, Lei |
| contents | Test-time scaling (TTS) has been shown to improve the performance of large language models (LLMs) by sampling and aggregating diverse reasoning paths. However, existing research has overlooked a critical issue: selection bias of reasoning strategies during scaling. Specifically, when generating reasoning processes, LLMs tend to follow certain strategies (e.g., algebraic solutions for math problems) while neglecting other valid alternatives (e.g., geometric solutions), resulting in insufficient exploration of the solution space. To further understand the impact of this bias, we present a theoretical analysis that reveals when it undermines the effectiveness of test-time scaling. Motivated by this theoretical insight, we introduce TTS-Uniform, a framework designed to mitigate the selection bias of reasoning strategies. It (i) identifies potential strategies, (ii) uniformly allocates the sampling budget across them, and (iii) filters out unstable strategies prior to aggregation. Experimental results show that TTS-Uniform significantly enhances scaling effectiveness across multiple mainstream LLMs and benchmark datasets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_17905 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Mitigating Strategy-Selection Bias in Reasoning for More Effective Test-Time Scaling Wu, Zongqian Xu, Baoduo Li, Tianyu Sun, Zhu Zhu, Xiaofeng Feng, Lei Artificial Intelligence Test-time scaling (TTS) has been shown to improve the performance of large language models (LLMs) by sampling and aggregating diverse reasoning paths. However, existing research has overlooked a critical issue: selection bias of reasoning strategies during scaling. Specifically, when generating reasoning processes, LLMs tend to follow certain strategies (e.g., algebraic solutions for math problems) while neglecting other valid alternatives (e.g., geometric solutions), resulting in insufficient exploration of the solution space. To further understand the impact of this bias, we present a theoretical analysis that reveals when it undermines the effectiveness of test-time scaling. Motivated by this theoretical insight, we introduce TTS-Uniform, a framework designed to mitigate the selection bias of reasoning strategies. It (i) identifies potential strategies, (ii) uniformly allocates the sampling budget across them, and (iii) filters out unstable strategies prior to aggregation. Experimental results show that TTS-Uniform significantly enhances scaling effectiveness across multiple mainstream LLMs and benchmark datasets. |
| title | Mitigating Strategy-Selection Bias in Reasoning for More Effective Test-Time Scaling |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2509.17905 |