Evaluating LLMs' Reasoning Over Ordered Procedural Steps
This work addresses a critical capability for LLMs in domains like cooking where step order matters, but it is incremental as it focuses on evaluation rather than proposing new methods.
The paper tackled the problem of evaluating large language models' ability to reason over ordered procedural steps by reconstructing sequences from shuffled steps, using a dataset of food recipes, and found that performance declines with longer sequences and more severe shuffling.
Reasoning over procedural sequences, where the order of steps directly impacts outcomes, is a critical capability for large language models (LLMs). In this work, we study the task of reconstructing globally ordered sequences from shuffled procedural steps, using a curated dataset of food recipes, a domain where correct sequencing is essential for task success. We evaluate several LLMs under zero-shot and few-shot settings and present a comprehensive evaluation framework that adapts established metrics from ranking and sequence alignment. These include Kendall's Tau, Normalized Longest Common Subsequence (NLCS), and Normalized Edit Distance (NED), which capture complementary aspects of ordering quality. Our analysis shows that model performance declines with increasing sequence length, reflecting the added complexity of longer procedures. We also find that greater step displacement in the input, corresponding to more severe shuffling, leads to further degradation. These findings highlight the limitations of current LLMs in procedural reasoning, especially with longer and more disordered inputs.