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.
@article{arxiv.2601.15296,
title = {Entropy-Tree: Tree-Based Decoding with Entropy-Guided Exploration},
author = {Longxuan Wei and Yubo Zhang and Zijiao Zhang and Zhihu Wang and Shiwan Zhao and Tianyu Huang and Huiting Zhao and Chenfei Liu and Shenao Zhang and Junchi Yan},
journal= {arXiv preprint arXiv:2601.15296},
year = {2026}
}