Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.15296 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909997534806016 |
|---|---|
| author | Wei, Longxuan Zhang, Yubo Zhang, Zijiao Wang, Zhihu Zhao, Shiwan Huang, Tianyu Zhao, Huiting Liu, Chenfei Zhang, Shenao Yan, Junchi |
| author_facet | Wei, Longxuan Zhang, Yubo Zhang, Zijiao Wang, Zhihu Zhao, Shiwan Huang, Tianyu Zhao, Huiting Liu, Chenfei Zhang, Shenao Yan, Junchi |
| contents | Large language models achieve strong reasoning performance, yet existing decoding strategies either explore blindly (random sampling) or redundantly (independent multi-sampling). We propose Entropy-Tree, a tree-based decoding method that exploits entropy as a signal for branching decisions--expanding the search tree only at positions where the model exhibits genuine uncertainty. Entropy-Tree shows superior accuracy and calibration in reasoning tasks: it achieves better pass@k than Multi-chain across multiple models and datasets, and its predictive entropy demonstrates better AUROC compared to several traditional metrics. Entropy-Tree unifies efficient structured exploration and reliable uncertainty estimation within a single decoding procedure. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_15296 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Entropy-Tree: Tree-Based Decoding with Entropy-Guided Exploration Wei, Longxuan Zhang, Yubo Zhang, Zijiao Wang, Zhihu Zhao, Shiwan Huang, Tianyu Zhao, Huiting Liu, Chenfei Zhang, Shenao Yan, Junchi Computation and Language Artificial Intelligence Large language models achieve strong reasoning performance, yet existing decoding strategies either explore blindly (random sampling) or redundantly (independent multi-sampling). We propose Entropy-Tree, a tree-based decoding method that exploits entropy as a signal for branching decisions--expanding the search tree only at positions where the model exhibits genuine uncertainty. Entropy-Tree shows superior accuracy and calibration in reasoning tasks: it achieves better pass@k than Multi-chain across multiple models and datasets, and its predictive entropy demonstrates better AUROC compared to several traditional metrics. Entropy-Tree unifies efficient structured exploration and reliable uncertainty estimation within a single decoding procedure. |
| title | Entropy-Tree: Tree-Based Decoding with Entropy-Guided Exploration |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2601.15296 |