An Information-Theoretic Parameter-Free Bayesian Framework for Probing Labeled Dependency Trees from Attention Score
Hongxu Liu, Jing Ma, Xiaojie Wang, Caixia Yuan, Fangxiang Feng
Abstract
Figuring out how neural language models comprehend syntax acts as a key to revealing how they understand languages. We systematically analyzed methods of extracting syntax from models, namely _probing_, and found limitations yet widely exist in previous probing practice. We proposed a method capable of estimating mutual information (MI) and directly extracting dependency trees from attention scores in a mathematical-rigorous way, requiring no additional network training effort. Compared with previous approaches, it has a much simpler model, while being able to probe more complex dependency trees, also transparent for fine-grained explanation. We tested our method on several open-source LLMs and demonstrated its effectiveness by systematically comparing it with a great many competitive baselines. Several informative conclusions can be drawn by further analysis of the results, shedding light on our method’s explanatory potential. An anonymous and off-the-shelf version of our code is released at https://anonymous.4open.science/r/IPBP-99F1.
BibTeX
@inproceedings{
liu2026an,
title={An Information-Theoretic Parameter-Free Bayesian Framework for Probing Labeled Dependency Trees from Attention Score},
author={Hongxu Liu and Jing Ma and Xiaojie Wang and Caixia Yuan and Fangxiang Feng},
booktitle={The Fourteenth International Conference on Learning Representations},
year={2026},
url={https://openreview.net/forum?id=q7raIuTQDK}
}