Binary decision task on EVS Europe (test)
62.49AccuracyOG-MAR
Evaluation results
| Method | Links | |
|---|---|---|
| OG-MARBackbone=Gemini 2.5 Flash Lite2026.01 | 62.49 | |
| OG-MARBackbone=GPT-4o mini2026.01 | 62.06 | |
| ValuesRAGBackbone=GPT-4o mini2026.01 | 61.27 | |
| OG-MARBackbone=EXAONE 3.52026.01 | 60.8 | |
| ValuesRAGBackbone=Gemini 2.5 Flash Lite2026.01 | 60.75 | |
| DebateBackbone=GPT-4o mini2026.01 | 59.85 | |
| DebateBackbone=Gemini 2.5 Flash Lite2026.01 | 59.77 | |
| OG-MARBackbone=QWEN 2.52026.01 | 58.98 | |
| RoleBackbone=GPT-4o mini2026.01 | 58.92 | |
| RoleBackbone=Gemini 2.5 Flash Lite2026.01 | 57.86 | |
| DebateBackbone=EXAONE 3.52026.01 | 57.13 | |
| Zero-shotBackbone=Gemini 2.5 Flash Lite2026.01 | 56.81 | |
| Zero-shotBackbone=GPT-4o mini2026.01 | 56.06 | |
| Self-consistencyBackbone=GPT-4o mini2026.01 | 55.58 | |
| ValuesRAGBackbone=QWEN 2.52026.01 | 55.38 | |
| DebateBackbone=QWEN 2.52026.01 | 55.11 | |
| Self-consistencyBackbone=EXAONE 3.52026.01 | 54.9 | |
| Self-consistencyBackbone=Gemini 2.5 Flash Lite2026.01 | 54.89 | |
| RoleBackbone=QWEN 2.52026.01 | 53.57 | |
| RoleBackbone=EXAONE 3.52026.01 | 53.19 | |
| Zero-shotBackbone=QWEN 2.52026.01 | 51.99 | |
| ValuesRAGBackbone=EXAONE 3.52026.01 | 51.72 | |
| Zero-shotBackbone=EXAONE 3.52026.01 | 51.43 | |
| Self-consistencyBackbone=QWEN 2.52026.01 | 50.96 |