← Search

Paul M. Riechers

2 accepted papers

2025

Constrained Belief Updates Explain Geometric Structures in Transformer Representations

ICML 2025poster

What computational structures emerge in transformers trained on next-token prediction? In this work, we provide evidence that transformers implement constrained Bayesian belief updating---a parallelized version of partial Bayesian inference shaped by architectural constraints. We integrate the model…

Cited by 19Source ↗PDF ↗Scholar ↗
2024

Transformers Represent Belief State Geometry in their Residual Stream

NeurIPS 2024poster

What computational structure are we building into large language models when we train them on next-token prediction? Here, we present evidence that this structure is given by the meta-dynamics of belief updating over hidden states of the data- generating process. Leveraging the theory of optimal pre…

Cited by 91Source ↗PDF ↗Scholar ↗