Metamaterial Generation on MetaModulus (test)
91.31VS (%)METASYMBO
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| METASYMBOBackbone LLM=GPT-4o-mini, Integration Strategy=Union2026.04 | 91.31 | 98.35 | 98.7 | 7.43 | 55.31 | 40 | |
| METASYMBOBackbone LLM=Gemini2.0, Integration Strategy=Union2026.04 | 89.65 | 95.97 | 99.2 | 10.07 | 49.66 | 56 | |
| Deepseek-ReasoningCategory=LLMs2026.04 | 85.5 | 65.3 | 86.9 | 67.7 | 49.93 | 76 | |
| METASYMBOStrategy=Ablation with Pure LLM Agents2026.04 | 78 | 29.35 | 88.9 | 51.2 | 48.32 | 61 | |
| METASYMBOBackbone LLM=GPT-4o-mini, Integration Strategy=Mix2026.04 | 76.84 | 94.17 | 98.2 | 0.83 | 52.34 | 0 | |
| METASYMBOBackbone LLM=Gemini2.0, Integration Strategy=Mix2026.04 | 64.53 | 91.74 | 93.3 | 0 | 54.64 | 0 | |
| CDVAECategory=Generative Models2026.04 | 57.03 | 0.4 | 55.85 | — | — | — | |
| GPT-4o-miniCategory=LLMs2026.04 | 47.29 | 0 | 73.6 | 39.66 | 41.55 | 59 | |
| Deepseek-chatCategory=LLMs2026.04 | 46.9 | 16.53 | 73.1 | 85.95 | 41.89 | 65 | |
| Gemini-2.0-flash-liteCategory=LLMs2026.04 | 44.31 | 41.86 | 27.5 | 92.56 | 47.55 | 67 | |
| Qwen3-235bCategory=LLMs2026.04 | 41.22 | 4.95 | 97.1 | 83.47 | 38.2 | 73 | |
| SyMatCategory=Generative Models2026.04 | 41.1 | 0 | 79.34 | — | — | — | |
| DiffCSPCategory=Generative Models2026.04 | 34.46 | 6.5 | 95.8 | — | — | — | |
| Llama-4-maverickCategory=LLMs2026.04 | 32.06 | 0.82 | 65.1 | 93.39 | 44.63 | 80 | |
| Cond-CDVAECategory=Generative Models2026.04 | 19.37 | 2 | 68.6 | — | — | — |