Addressing the Challenges of Planning Language Generation

ACL ARR 2025 May Submission4648 Authors

Abstract: Using LLMs to generate formal planning languages such as PDDL that invoke symbolic solvers to deterministically derive plans has been shown to outperform generating plans. While this success has been limited to closed-sourced models or particular LLM pipelines, we design and evaluate 8 different PDDL generation pipelines with open-source models under 50 billion parameters, previously shown to be incapable of this task. We find that intuitive approaches, such as using a high-resource language wrapper or constrained decoding with grammar, decrease performance. However, inference-time scaling approaches, such as revision with feedback from the solver and plan validator, more than double the performance.

Paper Type: Short

Research Area: NLP Applications

Research Area Keywords: planning, large language models, pddl, formal reasoning, inference-time scaling

Contribution Types: NLP engineering experiment

Languages Studied: PDDL

Submission Number: 4648