Research map / Large language models / Agents, code and math
Rethinking Code Similarity for Automated Algorithm Design with LLMs
Rui Zhang, Zhichao Lu
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 BehaveSim niche-based diversity management to FunSearch | FunSearch · TSP top-1 | relative gap to optimum (%): 25.22 -> 17.37 | Improved |
| add BehaveSim niche-based diversity management to FunSearch | FunSearch · ASP top-1 | relative gap to optimum (%): 5.84 -> 5.24 | Improved |
| add BehaveSim niche-based diversity management to FunSearch | FunSearch · CPP top-1 | sum of radii: 2.6259 -> 2.6293 | Improved |
| add BehaveSim as helper objective to EoH | EoH · TSP top-1 | relative gap to optimum (%): 18.86 -> 16.55 | Improved |
| add BehaveSim as helper objective to EoH | EoH · CPP top-1 | sum of radii: 2.5724 -> 2.6263 | Improved |
| add BehaveSim as helper objective to EoH | EoH · ASP top-1 | relative gap to optimum (%): 5.99 -> 5.33 | Improved |
| raise inter-island selection probability p_s1 from 0 to 0.5 (ablation) | FunSearch+BehaveSim · ASP | performance | Improved |
| remove clustering within islands during database construction (ablation) | FunSearch+BehaveSim · ASP | performance | Improved |
| truncate 10-30% of the trajectory (ablation) | BehaveSim · code-similarity benchmark | average similarity: Type-4 0.46 -> 0.47-0.52 | No clear effect |
| truncate 40-50% or sample every 3-5 steps (ablation) | BehaveSim · code-similarity benchmark Type-4 | average similarity: 0.46 -> 0.64 (k=0.5); 0.46 -> 0.74 (n=5) | Hurt |
Closest papers by meaning
- How efficient is LLM-generated code? A rigorous & high-standard benchmark (ICLR 2025 · similarity 0.86)
- CodeMMLU: A Multi-Task Benchmark for Assessing Code Understanding & Reasoning Capabilities of CodeLLMs (ICLR 2025 · similarity 0.86)
- LiveCodeBench Pro: How Do Olympiad Medalists Judge LLMs in Competitive Programming? (NeurIPS 2025 · similarity 0.86)
- Automated Creation of Reusable and Diverse Toolsets for Enhancing LLM Reasoning (AAAI 2025 · similarity 0.85)
- rStar-Coder: Scaling Competitive Code Reasoning with a Large-Scale Verified Dataset (NeurIPS 2025 · similarity 0.85)
- From I/O to Code with Discovery Agent (NeurIPS 2026 · similarity 0.85)
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.
