From Generalist to Specialist Representation

TL;DR

This paper establishes nonparametric identifiability guarantees for learning task-specific representations from generalist models, enabling a transition to specialist models without structural assumptions.

Contribution

It proves that task structure and relevant representations are identifiable in a fully unsupervised, nonparametric setting, even with complex, interleaved task sequences.

Findings

Task structure is identifiable across time steps without supervision.

Task-relevant representations can be disentangled within each step using sparsity.

First nonparametric guarantees for hierarchical task representation identifiability.

Abstract

Given a generalist model, learning a task-relevant specialist representation is fundamental for downstream applications. Identifiability, the asymptotic guarantee of recovering the ground-truth representation, is critical because it sets the ultimate limit of any model, even with infinite data and computation. We study this problem in a completely nonparametric setting, without relying on interventions, parametric forms, or structural constraints. We first prove that the structure between time steps and tasks is identifiable in a fully unsupervised manner, even when sequences lack strict temporal dependence and may exhibit disconnections, and task assignments can follow arbitrarily complex and interleaving structures. We then prove that, within each time step, the task-relevant latent representation can be disentangled from the irrelevant part under a simple sparsity regularization, without any additional information or parametric constraints. Together, these results establish a hierarchical foundation: task structure is identifiable across time steps, and task-relevant latent representations are identifiable within each step. To our knowledge, each result provides a first general nonparametric identifiability guarantee, and together they mark a step toward provably moving from generalist to specialist models.