Text-to-model translation on Text2Zinc n=1318 (test)
1,271Execution Accuracy (Eacc)CoT + Grammar
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CoT + GrammarBackbone=GPT-5.2, Strategy=Chain-of-Thought + Grammar2026.04 | 1,271 | 956 | |
| CoT + Code & GrammarBackbone=GPT-5.2, Strategy=Chain-of-Thought + Code & Grammar2026.04 | 1,271 | 949 | |
| Agentic + CodeBackbone=GPT-5.2, Strategy=Agentic + Code2026.04 | 1,253 | 956 | |
| CoT + CodeBackbone=GPT-5.2, Strategy=Chain-of-Thought + Code2026.04 | 1,246 | 927 | |
| CoTBackbone=GPT-5.2, Strategy=Chain-of-Thought2026.04 | 1,163 | 913 | |
| GalaBackbone=GPT-5.22026.04 | 1,155 | 880 | |
| AgenticBackbone=GPT-5.2, Strategy=Agentic2026.04 | 1,141 | 904 | |
| Zero ShotBackbone=GPT-5.2, Strategy=Zero Shot2026.04 | 1,094 | 848 |