Research map / Large language models / Agents, code and math
OptiHive: Ensemble Selection for LLM-Based Optimization via Statistical Modeling
Maxime Bouscary, Saurabh Amin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply OptiHive selection over 50 candidate solvers instead of a single generated solver | gpt-4.1-mini / o3 solvers · MDVRP+OBS | optimality rate: 5% -> 92.1% | Improved |
| apply OptiHive selection | gpt-4.1-mini / o3 solvers · DCMDVRP | optimality rate: 40% -> 97.03% | Improved |
| apply OptiHive selection | gpt-4.1-mini solvers · Time-dependent WSCP | optimality rate: 3% -> 64.1% | Improved |
| use weaker gpt-4.1-nano to generate solvers (ablation) | OptiHive · K-robust WSCP | optimality rate: 100% -> 83.1% | Hurt |
| use weaker gpt-4.1-nano to generate instances (ablation) | OptiHive · Time-dependent WSCP | optimality rate: 64.1% -> 23.5% | Hurt |
| use weaker gpt-4.1-nano to generate tests (ablation) | OptiHive · Time-dependent WSCP | optimality rate: 64.1% -> 19.7% | Hurt |
| use weaker gpt-4.1-nano to generate instances or tests (ablation) | OptiHive · K-robust and probabilistic WSCP | optimality rate: 100% -> 100% (probabilistic, nano tests) | No clear effect |
| increase number of generated tests beyond five (ablation) | OptiHive · DCMDVRP / MDVRP+OBS | optimality rate | No clear effect |
| increase number of candidate solvers on hard problem (ablation) | OptiHive · MDVRP+OBS | optimality rate | Mixed |
Closest papers by meaning
- LLMOPT: Learning to Define and Solve General Optimization Problems from Scratch (ICLR 2025 · similarity 0.88)
- OptiBench Meets ReSocratic: Measure and Improve LLMs for Optimization Modeling (ICLR 2025 · similarity 0.88)
- Autoformulation of Mathematical Optimization Models Using LLMs (ICML 2025 · similarity 0.88)
- MM-OptBench: A Solver-Grounded Benchmark for Multimodal Optimization Modeling (NeurIPS 2026 · similarity 0.88)
- OptMATH: A Scalable Bidirectional Data Synthesis Framework for Optimization Modeling (ICML 2025 · similarity 0.87)
- Large Language Models as End-to-end Combinatorial Optimization Solvers (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
