Self-Improvement for Fast, High-Quality Plan Generation

TL;DR

This paper presents a method using self-improving generative models to produce high-quality plans efficiently, outperforming traditional symbolic planners in plan length and scalability across multiple domains.

Contribution

It introduces a self-improvement framework for transformer-based models that iteratively enhances plan quality through combined model and graph search techniques.

Findings

30% reduction in plan length compared to symbolic planners

Over 80% of generated plans are optimal

Model latency scales sub-exponentially

Abstract

Generative models trained on synthetic plan data are a promising approach to generalized planning. Recent work has focused on finding any valid plan, rather than a high-quality solution. We address the challenge of producing high-quality plans, a computationally hard problem, in sub-exponential time. First, we demonstrate that, given optimal data, a decoder-only transformer can generate high-quality plans for unseen problem instances. Second, we show how to self-improve an initial model trained on sub-optimal data. Each round of self-improvement combines multiple model calls with graph search to generate improved plans, used for model fine-tuning. An experimental study on four domains: Blocksworld, Logistics, Labyrinth, and Sokoban, shows on average a 30% reduction in plan length over the source symbolic planner, with over 80% of plans being optimal, where the optimum is known. Plan quality is further improved by inference-time search. The model's latency scales sub-exponentially in contrast to the satisficing and optimal symbolic planners to which we compare. Together, these results suggest that self-improvement with generative models offers a scalable approach for high-quality plan generation.