Schematic Auditing on Diagnostic dataset for schematic auditing
100Terminal Misalignment (MA)ChatGPT-5
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ChatGPT-5V2G framework augmentation=false, Zero-shot prompting=true2026.02 | 100 | 0 | 0 | 100 | 100 | 0 | 0 | 0 | 100 | 100 | 50 | |
| Gemini 2.5 Pro +V2GV2G framework augmentation=true, Zero-shot prompting=true2026.02 | 100 | 100 | 0 | 100 | 100 | 100 | 0 | 100 | 0 | 100 | 70 | |
| Claude Sonnet-4 +V2GV2G framework augmentation=true, Zero-shot prompting=true2026.02 | 100 | 100 | 100 | 0 | 100 | 0 | 0 | 0 | 100 | 100 | 60 | |
| ChatGPT-4oV2G framework augmentation=false, Zero-shot prompting=true2026.02 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 100 | 0 | 0 | 10 | |
| ChatGPT-4o +V2GV2G framework augmentation=true, Zero-shot prompting=true2026.02 | 0 | 100 | 100 | 0 | 100 | 0 | 0 | 100 | 0 | 100 | 50 | |
| ChatGPT-5 +V2GV2G framework augmentation=true, Zero-shot prompting=true2026.02 | 0 | 100 | 100 | 100 | 100 | 0 | 0 | 100 | 0 | 100 | 60 | |
| GLM-4.5VV2G framework augmentation=false, Zero-shot prompting=true2026.02 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
| GLM-4.5V +V2GV2G framework augmentation=true, Zero-shot prompting=true2026.02 | 0 | 100 | 100 | 0 | 0 | 100 | 0 | 0 | 0 | 0 | 30 | |
| Gemini 2.5 ProV2G framework augmentation=false, Zero-shot prompting=true2026.02 | 0 | 0 | 0 | 0 | 100 | 0 | 0 | 0 | 0 | 0 | 10 | |
| Qwen2.5-VL-72BV2G framework augmentation=false, Zero-shot prompting=true2026.02 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
| Qwen2.5-VL-72B +V2GV2G framework augmentation=true, Zero-shot prompting=true2026.02 | 0 | 100 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 10 | |
| Claude Sonnet-4V2G framework augmentation=false, Zero-shot prompting=true2026.02 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |