Reproduce “NextEval: An Agent-Based Solver and Benchmark for Derivative-Free Optimization” 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
Shuo Chen, Li Cui, Cunxin Huang, Ya-Feng Liu, Minghang Xie, Meng Xu, Zaikun Zhang
NextEval uses a persistent language-model agent to select and configure numerical methods for derivative-free optimization; the methods, not the agent, calculate candidate points. The agent allocates evaluation budgets and can switch tools. On 92 public problems of dimension at most five, the complete system scored 1.000 versus 0.820 for NEWUOA (PDFO) on objective-evaluation profiles, not elapsed time or broader generalization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.