Rare Event Analysis of Large Language Models

Jake Mcallister Dorman, Edward Gillman, Dominic C Rose, Jamie F. Mair, Juan P. Garrahan
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:26179-26201, 2026.

Abstract

Being probabilistic models, during inference large language models (LLMs) display rare events: behaviour that is far from typical but highly significant. By definition all rare events are hard to see, but the enormous scale of LLM usage means that events completely unobserved during development are likely to become prominent in deployment. Here we present an end-to-end framework for the systematic analysis of rare events in LLMs. We provide a practical implementation spanning theory, efficient generation strategies, probability estimation and error analysis, which we illustrate with concrete examples. We outline extensions and applications to other models and contexts, highlighting the generality of the concepts and techniques presented here.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-dorman26a, title = {Rare Event Analysis of Large Language Models}, author = {Dorman, Jake Mcallister and Gillman, Edward and Rose, Dominic C and Mair, Jamie F. and Garrahan, Juan P.}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {26179--26201}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/dorman26a/dorman26a.pdf}, url = {https://proceedings.mlr.press/v306/dorman26a.html}, abstract = {Being probabilistic models, during inference large language models (LLMs) display rare events: behaviour that is far from typical but highly significant. By definition all rare events are hard to see, but the enormous scale of LLM usage means that events completely unobserved during development are likely to become prominent in deployment. Here we present an end-to-end framework for the systematic analysis of rare events in LLMs. We provide a practical implementation spanning theory, efficient generation strategies, probability estimation and error analysis, which we illustrate with concrete examples. We outline extensions and applications to other models and contexts, highlighting the generality of the concepts and techniques presented here.} }
Endnote
%0 Conference Paper %T Rare Event Analysis of Large Language Models %A Jake Mcallister Dorman %A Edward Gillman %A Dominic C Rose %A Jamie F. Mair %A Juan P. Garrahan %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-dorman26a %I PMLR %P 26179--26201 %U https://proceedings.mlr.press/v306/dorman26a.html %V 306 %X Being probabilistic models, during inference large language models (LLMs) display rare events: behaviour that is far from typical but highly significant. By definition all rare events are hard to see, but the enormous scale of LLM usage means that events completely unobserved during development are likely to become prominent in deployment. Here we present an end-to-end framework for the systematic analysis of rare events in LLMs. We provide a practical implementation spanning theory, efficient generation strategies, probability estimation and error analysis, which we illustrate with concrete examples. We outline extensions and applications to other models and contexts, highlighting the generality of the concepts and techniques presented here.
APA
Dorman, J.M., Gillman, E., Rose, D.C., Mair, J.F. & Garrahan, J.P.. (2026). Rare Event Analysis of Large Language Models. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:26179-26201 Available from https://proceedings.mlr.press/v306/dorman26a.html.

Related Material