Reproduce “Beyond the Hype: Benchmarking LLM-Evolved Heuristics for Bin Packing” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Kevin Sim, Quentin Renau, Emma Hart
The authors test whether LLM-generated online bin-packing heuristics generalize across diverse item streams. They compare five such rules with five hand-designed heuristics on 6,064 instances across 12 datasets. Average Excess Bins measures bins above a lower bound based on item sizes and capacity. Best-Fit ranks first by summed dataset-level AEB (5.85), and FS1 ranks second; FSW’s score is 113.21, showing that performance varies among LLM heuristics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.