Research map / Large language models / Agents, code and math
NEMO: Execution-Aware Optimization Modeling via Autonomous Coding Agents
Yang Song, Anoushka Vyas, Zirui Wei, Sina Khoshfetrat Pakazad, Henrik Ohlsson, Graham Neubig
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 |
|---|---|---|---|
| add simulator-optimizer asymmetric validation (ablation) | NEMO · BWOR | accuracy: 71.9% -> 75.6% | Improved |
| add simulator-optimizer asymmetric validation (ablation) | NEMO · MAMO-Complex | accuracy: 50.9% -> 54.2% | Improved |
| add simulator-optimizer asymmetric validation (ablation) | NEMO · IndustryOR | accuracy: 60.0% -> 60.0% | No clear effect |
| add external memory of retrieved OptMATH examples (ablation) | NEMO Base · MAMO-Complex | accuracy: 54.2% -> 61.9% | Improved |
| add external memory of retrieved OptMATH examples (ablation) | NEMO Base · BWOR | accuracy: 75.6% -> 80.4% | Improved |
| add component-wise MBR decoding with LLM re-ranking (ablation) | NEMO +Mem · MAMO-Complex | accuracy: 61.9% -> 71.0% | Improved |
| add multiple optimizer backends and self-consistency (ablation) | NEMO +Mem+MBR · BWOR | accuracy: 80.4% -> 82.9% | Improved |
| add multiple optimizer backends (ablation) | NEMO +Mem+MBR · IndustryOR | accuracy: 63.0% -> 63.0% | No clear effect |
| use NEMO instead of compute-matched Best-of-5 ACA | Claude 4.5 Sonnet ACA · BWOR | accuracy: 79.3% -> 86.6% | Improved |
| use NEMO instead of compute-matched Best-of-5 ACA | Claude 4.5 Sonnet ACA · OptMATH-Bench | accuracy: 62.7% -> 68.1% | Improved |
| add simulator on top of full pipeline (ablation) | NEMO (Claude 4.5 Sonnet) · BWOR | accuracy: 80.5% -> 86.6% | Improved |
Closest papers by meaning
- EvoCodeBench: Evaluating Coding Agents in Multi-Turn Iterative Interactions (NeurIPS 2026 · similarity 0.87)
- GSO: Challenging Software Optimization Tasks for Evaluating SWE-Agents (NeurIPS 2025 · similarity 0.87)
- VeRO: A Harness for Agents to Optimize Agents (ICML 2026 · similarity 0.86)
- Automated Reformulation of Robust Optimization via Memory-Augmented Large Language Models (NeurIPS 2026 · similarity 0.86)
- Generalizable End-to-End Tool-Use RL with Synthetic CodeGym (ICLR 2026 · similarity 0.86)
- CATArena: Evaluating Evolutionary Capabilities of Code Agents via Iterative Tournaments (ICML 2026 · similarity 0.86)
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.
