# Sequential statistical inference for Large Language Models: Representation, validity, and monitoring

arXiv: https://arxiv.org/abs/2606.07624 · HTML: https://mlsanity.com/paper/2606.07624/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 13 references that are in the map.
- First posted on arXiv 30 May 2026.

## Builds on (most cited first, 7 of 7)

- [Time-uniform, nonparametric, nonasymptotic confidence sequences](https://mlsanity.com/paper/1810.08240/) (2018)
- [Conformal Inference for Online Prediction with Arbitrary Distribution Shifts](https://mlsanity.com/paper/2208.08401/) (2022)
- [An Overview of Large Language Models for Statisticians](https://mlsanity.com/paper/2502.17814/) (2025)
- [Generalized generalized linear models: Convex estimation and online bounds](https://mlsanity.com/paper/2304.13793/) (2023)
- [Leave a Window Out: Modifying the Jackknife for Predictive Inference in Time Series](https://mlsanity.com/paper/2605.30292/) (2026)
- [Nonlinear time-series embedding by monotone variational inequality](https://mlsanity.com/paper/2406.06894/) (2024)
- [Control Charts for Multi-agent Systems](https://mlsanity.com/paper/2605.11135/) (2026)

## Built on by (most cited first, 2 of 2)

- [Quickest Detection of Hallucination Onset: Delay Bounds and Learned CUSUM Statistics](https://mlsanity.com/paper/2606.12476/) (2026)
- [Adaptive Triggering for Bias Correction in LLM Reasoning](https://mlsanity.com/paper/2608.25379/) (2026)

## Abstract (arXiv)

This discussion argues that sequential statistical inference can naturally contribute to LLM trustworthiness. In deployment, LLM systems are queried repeatedly, conditioned on evolving contexts, and incorporate user or tool feedback, and may exhibit behavioral shifts after model updates or distribution changes. The discussion is organized around three tasks: representation, modeling LLM interactions as dependent stochastic processes rather than isolated prompt--response pairs; validity, developing uncertainty guarantees that remain meaningful under dependence, repeated use, and adaptation; and monitoring, using sequential alarms and change-point detection to identify shifts in calibration, hallucination rates, refusal behavior, fairness, or other task-relevant properties. This perspective complements recent surveys by viewing trustworthy LLM deployment as a problem of statistical process control.