Task Vector Geometry Underlies Dual Modes of Task Inference in Transformers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yan, Hao, Yang, Haolin, Zhong, Yiqiao
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915981084852224
author Yan, Hao
Yang, Haolin
Zhong, Yiqiao
author_facet Yan, Hao
Yang, Haolin
Zhong, Yiqiao
contents Transformers are effective at inferring the latent task from context via two inference modes: recognizing a task seen during training, and adapting to a novel one. Recent interpretability studies have identified from middle-layer representations task-specific directions, or task vectors, that steer model behavior. However, a lack of rigorous foundations hinders connecting internal representations to external model behavior: existing work fails to explain how task-vector geometry is shaped by the training distribution, and what geometry enables out-of-distribution (OOD) generalization. In this paper, we study these questions in a controlled synthetic setting by training small transformers from scratch on latent-task sequence distributions, which allows a principled mathematical characterization. We show that two inference modes can coexist within a single model. In-distribution behavior is governed by Bayesian task retrieval, implemented internally through convex combinations of learned task vectors. OOD behavior, by contrast, arises through extrapolative task learning, whose representations occupy a subspace nearly orthogonal to the task-vector subspace. Taken together, our results suggest that task-vector geometry, training distributions, and generalization behaviors are closely related.
format Preprint
id arxiv_https___arxiv_org_abs_2605_03780
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Task Vector Geometry Underlies Dual Modes of Task Inference in Transformers
Yan, Hao
Yang, Haolin
Zhong, Yiqiao
Machine Learning
Computation and Language
Transformers are effective at inferring the latent task from context via two inference modes: recognizing a task seen during training, and adapting to a novel one. Recent interpretability studies have identified from middle-layer representations task-specific directions, or task vectors, that steer model behavior. However, a lack of rigorous foundations hinders connecting internal representations to external model behavior: existing work fails to explain how task-vector geometry is shaped by the training distribution, and what geometry enables out-of-distribution (OOD) generalization. In this paper, we study these questions in a controlled synthetic setting by training small transformers from scratch on latent-task sequence distributions, which allows a principled mathematical characterization. We show that two inference modes can coexist within a single model. In-distribution behavior is governed by Bayesian task retrieval, implemented internally through convex combinations of learned task vectors. OOD behavior, by contrast, arises through extrapolative task learning, whose representations occupy a subspace nearly orthogonal to the task-vector subspace. Taken together, our results suggest that task-vector geometry, training distributions, and generalization behaviors are closely related.
title Task Vector Geometry Underlies Dual Modes of Task Inference in Transformers
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2605.03780