Binary decision task on AFRO Africa (test)
57.01AccuracyOG-MAR
Evaluation results
| Method | Links | |
|---|---|---|
| OG-MARBackbone=Gemini 2.5 Flash Lite2026.01 | 57.01 | |
| ValuesRAGBackbone=GPT-4o mini2026.01 | 56.54 | |
| RoleBackbone=GPT-4o mini2026.01 | 55.05 | |
| ValuesRAGBackbone=Gemini 2.5 Flash Lite2026.01 | 54.72 | |
| OG-MARBackbone=GPT-4o mini2026.01 | 53.89 | |
| DebateBackbone=GPT-4o mini2026.01 | 53.43 | |
| Zero-shotBackbone=GPT-4o mini2026.01 | 53.24 | |
| RoleBackbone=Gemini 2.5 Flash Lite2026.01 | 53.13 | |
| Zero-shotBackbone=Gemini 2.5 Flash Lite2026.01 | 52.82 | |
| Self-consistencyBackbone=GPT-4o mini2026.01 | 52.24 | |
| Self-consistencyBackbone=Gemini 2.5 Flash Lite2026.01 | 51.82 | |
| OG-MARBackbone=QWEN 2.52026.01 | 51.8 | |
| DebateBackbone=Gemini 2.5 Flash Lite2026.01 | 50.46 | |
| OG-MARBackbone=EXAONE 3.52026.01 | 50.45 | |
| RoleBackbone=QWEN 2.52026.01 | 50.14 | |
| DebateBackbone=EXAONE 3.52026.01 | 49.95 | |
| DebateBackbone=QWEN 2.52026.01 | 48.75 | |
| Zero-shotBackbone=QWEN 2.52026.01 | 48.14 | |
| ValuesRAGBackbone=EXAONE 3.52026.01 | 47.94 | |
| ValuesRAGBackbone=QWEN 2.52026.01 | 47.24 | |
| Self-consistencyBackbone=EXAONE 3.52026.01 | 40.86 | |
| Self-consistencyBackbone=QWEN 2.52026.01 | 40.8 | |
| RoleBackbone=EXAONE 3.52026.01 | 40.77 | |
| Zero-shotBackbone=EXAONE 3.52026.01 | 40.54 |