PACE: Parameter Change for Unsupervised Environment Design

TL;DR

PACE introduces a new environment evaluation method based on policy parameter change, improving reinforcement learning generalization by efficiently measuring actual learning progress without high variance or computational costs.

Contribution

It proposes a novel, low-variance, computation-efficient environment evaluation method grounded in realized learning progress, outperforming existing UED approaches.

Findings

PACE outperforms baseline UED methods on MiniGrid and Craftax.

Achieves an IQM of 96.4% on MiniGrid.

Reduces the Optimality Gap to 17.2% on MiniGrid.

Abstract

Unsupervised Environment Design (UED) offers a promising paradigm for improving reinforcement learning generalization by adaptively shaping training environments, but it requires reliable environment evaluation to remain effective. However, existing UED methods evaluate environments using indirect proxy signals such as regret, value-based errors, or Monte Carlo, which suffer from bias, high variance, or substantial computational overhead and fail to reflect agent realized learning progress. To address these limitations, we propose Parameter Change Environment Design (PACE), which evaluates an environment through the policy parameter change induced by training on that environment, directly grounding environment selection in realized learning progress. Specifically, PACE assigns environment value using a first-order approximation of the policy optimization objective, where the improvement induced by an environment is proportional to the squared L2 norm of the corresponding parameter update, enabling low-variance and computation-efficient evaluation without additional rollouts. Experiments on MiniGrid and Craftax show that PACE consistently outperforms established UED baselines, achieving higher IQM and smaller Optimality Gap on OOD evaluations, including an IQM of 96.4% and an Optimality Gap of 17.2% on MiniGrid.