PlanBench: An Extensible Benchmark for Evaluating Large Language Models on Planning and Reasoning about Change

By Karthik Valmeekam, Alberto Olmo, Sarath Sreedharan, Subbarao Kambhampati. Neural Information Processing Systems, 2022

438 citations recorded by the source corpus.

Generating plans of action, and reasoning about change have long been considered a core competence of intelligent agents. It is thus no surprise that evaluating the planning and reasoning capabilities of large language models (LLMs) has become a hot topic of research. Most claims about LLM planning capabilities are however based on common sense tasks-where it becomes hard to tell whether LLMs are planning or merely retrieving from their vast world knowledge. There is a strong need for systematic and extensible planning benchmarks with sufficient diversity to evaluate whether LLMs have innate planning capabilities. Motivated by this, we propose PlanBench, an extensible benchmark suite based on the kinds of domains used in the automated planning community, especially in the International Planning Competition, to test the capabilities of LLMs in planning or reasoning about actions and change. PlanBench provides sufficient diversity in both the task domains and the specific planning capabilities. Our studies also show that on many critical capabilities-including plan generation-LLM performance falls quite short, even with the SOTA models. PlanBench can thus function as a useful marker of progress of LLMs in planning and reasoning.

Subfields: Artificial Intelligence, Management Information Systems

View Paper: https://www.semanticscholar.org/paper/e850d4c0dad3aee3b8b40be5e5d5e5c31354d8cc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

To CoT or not to CoT? Chain-of-thought helps mainly on math and symbolic reasoning (2024, neutral): “…I.1 L ONG H ORIZON P LANNING One set of tasks where symbolic reasoning helps substantially that our experiments haven’t covered as thoroughly (with the exception of BiGGen-Bench) is long-horizon planning (Valmeekam et al., 2023; Xie et al., 2024; Gundawar et al., 2024; Valmeekam et al., 2024).”
Faith and Fate: Limits of Transformers on Compositionality (2023, background): “[58] Karthik Valmeekam, Alberto Olmo, Sarath Sreedharan, and Subbarao Kambhampati.”
A Survey on Vision--Language--Action Models for Embodied AI (2024, positive): “PlanBench [219], [220] comprehensively assesses various aspects of task planning ability, such as cost optimality, plan verification, and replanning.”
TravelPlanner: A Benchmark for Real-World Planning with Language Agents (2024, background): “Researchers have therefore investigated their potential in an array of planning tasks ranging from classic planning settings like Blocksworld (Valmeekam et al., 2023) to embodied agents (Huang et al., 2022; Ichter et al., 2022; Song et al., 2023; Wang et al., 2023) and web agents (Deng et al., 2023;”
Do Embodied Agents Dream of Pixelated Sheep?: Embodied Decision Making using Language Guided World Modelling (2023, background): “However, by default, the generated plans are not grounded in environment dynamics and constraining output can harm model performance, both of which lead to subpar performance of out-of-the-box LLMs on decision-making tasks (Valmeekam et al., 2022).”