Attention as In-Context Empirical Bayes: A Two-Stage View via Particle Dynamics

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Smart, Matthew, Ganguly, Soumya, Metya, Nilava, Morozov, Alexandre V., Sengupta, Anirvan M.
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917542890569728
author Smart, Matthew
Ganguly, Soumya
Metya, Nilava
Morozov, Alexandre V.
Sengupta, Anirvan M.
author_facet Smart, Matthew
Ganguly, Soumya
Metya, Nilava
Morozov, Alexandre V.
Sengupta, Anirvan M.
contents We study minimal attention-only transformers under all-token corruption and show they admit a two-stage empirical Bayes interpretation. A single attention step computes a kernel-weighted posterior mean with respect to the empirical distribution defined by the context. Depth refines this distribution through particle dynamics (Stage 1), while a long-range skip-connection carries the noisy input as a query for posterior inference (Stage 2), revealing distinct statistical roles for depth and attention residuals. The framework isolates a minimal setting in which the context itself induces a depth-dependent energy landscape governing in-context inference. We show that effective denoising can emerge without an explicit noise schedule: a fixed kernel bandwidth and finite integration horizon suffice, yielding a principled depth-noise relationship. We further establish a posterior-mean recovery guarantee for a class of well-behaved priors, where the empirical estimator converges to the Bayes-optimal predictor under asymptotic conditions. Connecting these dynamics to reverse-diffusion limits, our results provide a statistical interpretation of attention as in-context inference via sample-based posterior estimation, without explicit density modeling.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29351
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Attention as In-Context Empirical Bayes: A Two-Stage View via Particle Dynamics
Smart, Matthew
Ganguly, Soumya
Metya, Nilava
Morozov, Alexandre V.
Sengupta, Anirvan M.
Machine Learning
Dynamical Systems
We study minimal attention-only transformers under all-token corruption and show they admit a two-stage empirical Bayes interpretation. A single attention step computes a kernel-weighted posterior mean with respect to the empirical distribution defined by the context. Depth refines this distribution through particle dynamics (Stage 1), while a long-range skip-connection carries the noisy input as a query for posterior inference (Stage 2), revealing distinct statistical roles for depth and attention residuals. The framework isolates a minimal setting in which the context itself induces a depth-dependent energy landscape governing in-context inference. We show that effective denoising can emerge without an explicit noise schedule: a fixed kernel bandwidth and finite integration horizon suffice, yielding a principled depth-noise relationship. We further establish a posterior-mean recovery guarantee for a class of well-behaved priors, where the empirical estimator converges to the Bayes-optimal predictor under asymptotic conditions. Connecting these dynamics to reverse-diffusion limits, our results provide a statistical interpretation of attention as in-context inference via sample-based posterior estimation, without explicit density modeling.
title Attention as In-Context Empirical Bayes: A Two-Stage View via Particle Dynamics
topic Machine Learning
Dynamical Systems
url https://arxiv.org/abs/2605.29351