Beyond Greedy Decoding: Model-Specific Strategy Selection via Multi-faceted Uncertainty Decomposition

Kwangje Baeg, Yubin Lim
Conference on Parsimony and Learning, PMLR 328:735-755, 2026.

Abstract

Large Language Models (LLMs) rely on static decoding strategies despite significant differences in the difficulty of generation. Recent uncertainty-based approaches aggregate diverse signals, overlooking model heterogeneity—particularly pronounced in morphologically rich languages (e.g., Korean) where tokenization variations lead to unique uncertainty traits. We focus on Korean instruction-tuned LLMs and decompose uncertainty into three largely independent components—Semantic Entropy, Graph Laplacian, and Trajectory Consistency. Unsupervised clustering reveals model-specific behavioral profiles with marked heterogeneity, challenging aggregation-based approaches and supporting uncertainty-guided strategy selection. High generation quality does not correlate with low output diversity, and universal decoding strategies fail for heterogeneous models. Cross-dataset validation shows that uncertainty patterns capture transferable model characteristics, enabling practitioners to systematically select strategies based on generation context.

Cite this Paper

BibTeX
@InProceedings{pmlr-v328-baeg26a, title = {Beyond Greedy Decoding: Model-Specific Strategy Selection via Multi-faceted Uncertainty Decomposition}, author = {Baeg, Kwangje and Lim, Yubin}, booktitle = {Conference on Parsimony and Learning}, pages = {735--755}, year = {2026}, editor = {Burkholz, Rebekka and Liu, Shiwei and Ravishankar, Saiprasad and Redman, William and Huang, Wei and Su, Weijie and Zhu, Zhihui}, volume = {328}, series = {Proceedings of Machine Learning Research}, month = {23--26 Mar}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v328/main/assets/baeg26a/baeg26a.pdf}, url = {https://proceedings.mlr.press/v328/baeg26a.html}, abstract = {Large Language Models (LLMs) rely on static decoding strategies despite significant differences in the difficulty of generation. Recent uncertainty-based approaches aggregate diverse signals, overlooking model heterogeneity—particularly pronounced in morphologically rich languages (e.g., Korean) where tokenization variations lead to unique uncertainty traits. We focus on Korean instruction-tuned LLMs and decompose uncertainty into three largely independent components—Semantic Entropy, Graph Laplacian, and Trajectory Consistency. Unsupervised clustering reveals model-specific behavioral profiles with marked heterogeneity, challenging aggregation-based approaches and supporting uncertainty-guided strategy selection. High generation quality does not correlate with low output diversity, and universal decoding strategies fail for heterogeneous models. Cross-dataset validation shows that uncertainty patterns capture transferable model characteristics, enabling practitioners to systematically select strategies based on generation context.} }
Endnote
%0 Conference Paper %T Beyond Greedy Decoding: Model-Specific Strategy Selection via Multi-faceted Uncertainty Decomposition %A Kwangje Baeg %A Yubin Lim %B Conference on Parsimony and Learning %C Proceedings of Machine Learning Research %D 2026 %E Rebekka Burkholz %E Shiwei Liu %E Saiprasad Ravishankar %E William Redman %E Wei Huang %E Weijie Su %E Zhihui Zhu %F pmlr-v328-baeg26a %I PMLR %P 735--755 %U https://proceedings.mlr.press/v328/baeg26a.html %V 328 %X Large Language Models (LLMs) rely on static decoding strategies despite significant differences in the difficulty of generation. Recent uncertainty-based approaches aggregate diverse signals, overlooking model heterogeneity—particularly pronounced in morphologically rich languages (e.g., Korean) where tokenization variations lead to unique uncertainty traits. We focus on Korean instruction-tuned LLMs and decompose uncertainty into three largely independent components—Semantic Entropy, Graph Laplacian, and Trajectory Consistency. Unsupervised clustering reveals model-specific behavioral profiles with marked heterogeneity, challenging aggregation-based approaches and supporting uncertainty-guided strategy selection. High generation quality does not correlate with low output diversity, and universal decoding strategies fail for heterogeneous models. Cross-dataset validation shows that uncertainty patterns capture transferable model characteristics, enabling practitioners to systematically select strategies based on generation context.
APA
Baeg, K. & Lim, Y.. (2026). Beyond Greedy Decoding: Model-Specific Strategy Selection via Multi-faceted Uncertainty Decomposition. Conference on Parsimony and Learning, in Proceedings of Machine Learning Research 328:735-755 Available from https://proceedings.mlr.press/v328/baeg26a.html.

Related Material