LLM4AD: A Platform for Algorithm Design with Large Language Model
About
We introduce LLM4AD, a unified Python platform for algorithm design (AD) with large language models (LLMs). LLM4AD is a generic framework with modularized blocks for search methods, algorithm design tasks, and LLM interface. The platform integrates numerous key methods and supports a wide range of algorithm design tasks across various domains including optimization, machine learning, and scientific discovery. We have also designed a unified evaluation sandbox to ensure a secure and robust assessment of algorithms. Additionally, we have compiled a comprehensive suite of support resources, including tutorials, examples, a user manual, online resources, and a dedicated graphical user interface (GUI) to enhance the usage of LLM4AD. We believe this platform will serve as a valuable tool for fostering future development in the merging research direction of LLM-assisted algorithm design.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Online Bin Packing | Online BPP (test) | Gap (%)0.445 | 120 | |
| Traveling Salesperson Problem | TSP N=200 (Generalization (128 instances)) | Optimality Gap14.403 | 35 | |
| Traveling Salesperson Problem | TSP N=1000 Generalization (128 instances) | Optimality Gap16.076 | 30 | |
| Traveling Salesperson Problem | TSP N=500 Generalization (128 instances) | Optimality Gap15.232 | 30 | |
| Traveling Salesperson Problem | TSP n=100 (train) | Objective Value8.786 | 26 | |
| Traveling Salesperson Problem | TSP-20 (train) | Objective Value4.197 | 17 | |
| Traveling Salesperson Problem | TSP-50 (train) | Objective Value6.399 | 17 | |
| Traveling Salesperson Problem | TSP Overall | Average Gap13.343 | 16 | |
| Capacitated Vehicle Routing Problem | CVRP (train) | CVRP Cost (20)4.826 | 4 | |
| Offline Bin Packing Problem | Offline BPP Generalization | Bins Used (N=500, C=300)102.4 | 4 |