Vision-Language Reasoning on MME (test)
78.98Simple AccuracyNo intervention baseline
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| No intervention baselineIntervention=None2026.01 | 78.98 | 30.77 | 42.29 | |
| Q4 system ablIntervention=Q4 system ablation2026.01 | 78.9 | 30 | 41.2 | |
| PAIIntervention=PAI2026.01 | 78.73 | 30.77 | 42.62 | |
| AD-HHIntervention=AD-HH2026.01 | 78.73 | 29.23 | 42.88 | |
| Image×2.0Intervention=Image×2.02026.01 | 78.35 | 30.77 | 40.14 | |
| Q4 text redistr (prop)Intervention=Q4 text redistribution, strategy=proportional2026.01 | 78.26 | 29.23 | 41.07 | |
| Q4 system redistr (prop)Intervention=Q4 system redistribution, strategy=proportional2026.01 | 72.83 | 21.54 | 28.98 |