Conflict Discrimination on E-VQA (test)
93.4MCCRPA-SFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RPA-SFTbackbone=Qwen3-VL-8B2026.02 | 93.4 | 95.5 | |
| Few-shot CoTbackbone=Qwen3-VL-8B2026.02 | 86.9 | 95.3 | |
| Zero-shotbackbone=Qwen3-VL-8B2026.02 | 85.4 | 94.5 | |
| SFTbackbone=Qwen3-VL-8B2026.02 | 82.6 | 91.4 | |
| RPA-SFTbackbone=InternVL3.5-8B2026.02 | 78.6 | 90.5 | |
| SFTbackbone=InternVL3.5-8B2026.02 | 77.7 | 89.1 | |
| Few-shot CoTbackbone=InternVL3.5-8B2026.02 | 72.5 | 89.5 | |
| Zero-shotbackbone=InternVL3.5-8B2026.02 | 52.2 | 82.6 | |
| SFTbackbone=LLaVA-1.5-7B2026.02 | 38.4 | 69.8 | |
| RPA-SFTbackbone=LLaVA-1.5-7B2026.02 | 37.7 | 70 | |
| Zero-shotbackbone=LLaVA-1.5-7B2026.02 | 5.5 | 50.9 | |
| Few-shot CoTbackbone=LLaVA-1.5-7B2026.02 | 5.1 | 49.6 |