Dual-Dimensional Consistency: Balancing Budget and Quality in Adaptive Inference-Time Scaling
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916013386235904 |
|---|---|
| author | Xu, Rongman Li, Yifei Zhao, Tianzhe Wu, Yanrui Li, Bo Yan, Hang |
| author_facet | Xu, Rongman Li, Yifei Zhao, Tianzhe Wu, Yanrui Li, Bo Yan, Hang |
| contents | Large Language Models (LLMs) have demonstrated remarkable abilities in reasoning. However, maximizing their potential through inference-time scaling faces challenges in trade-off between sampling budget and reasoning quality. Current strategies remain inefficient as they typically treat sampling width and depth as orthogonal objectives, where width consensus methods risk reinforcing hallucinations, while depth pruning mechanisms prematurely truncate complex yet valid reasoning chains. Therefore, we propose Dual-Dimensional Consistency (DDC), a unified framework that bridges path quality with adaptive termination. By coupling Confidence-Weighted Bayesian protocol with a Trend-Aware Stratified Pruning, our method ensures that computational resources are concentrated on high quality reasoning paths, filtering hallucinations while accelerating consensus. Evaluations across five benchmarks demonstrate that this approach reduces token consumption by over 10 times while maintaining or exceeding the accuracy of strong baselines across various LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_15100 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Dual-Dimensional Consistency: Balancing Budget and Quality in Adaptive Inference-Time Scaling Xu, Rongman Li, Yifei Zhao, Tianzhe Wu, Yanrui Li, Bo Yan, Hang Artificial Intelligence Large Language Models (LLMs) have demonstrated remarkable abilities in reasoning. However, maximizing their potential through inference-time scaling faces challenges in trade-off between sampling budget and reasoning quality. Current strategies remain inefficient as they typically treat sampling width and depth as orthogonal objectives, where width consensus methods risk reinforcing hallucinations, while depth pruning mechanisms prematurely truncate complex yet valid reasoning chains. Therefore, we propose Dual-Dimensional Consistency (DDC), a unified framework that bridges path quality with adaptive termination. By coupling Confidence-Weighted Bayesian protocol with a Trend-Aware Stratified Pruning, our method ensures that computational resources are concentrated on high quality reasoning paths, filtering hallucinations while accelerating consensus. Evaluations across five benchmarks demonstrate that this approach reduces token consumption by over 10 times while maintaining or exceeding the accuracy of strong baselines across various LLMs. |
| title | Dual-Dimensional Consistency: Balancing Budget and Quality in Adaptive Inference-Time Scaling |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2605.15100 |