Self-Improvement for Fast, High-Quality Plan Generation

2026-05-05

AI Summary

This work addresses the challenge of efficiently generating high-quality (near-optimal) plans in generalized planning by proposing a decoder-only Transformer-based self-improvement framework. The approach begins with pretraining on synthetic data and iteratively refines plan quality through alternating rounds of model-generated plans and graph-based search, using the improved plans for subsequent fine-tuning. It achieves, for the first time, high-quality planning within sub-exponential time, substantially outperforming conventional symbolic planners: across four domains, it reduces plan length by 30% on average, with over 80% of solutions matching known optima, while exhibiting inference latency that grows slower than 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.

Problem

• high-quality plan generation
• self-improvement
• generative models
• planning
• computational hardness

Innovation

• self-improvement
• generative planning
• decoder-only transformer
• sub-exponential latency
• plan quality optimization