Visual Question Answering on MathVision mini
22.7AccuracyVanilla RoPE
Evaluation Results
| Method | Links | |
|---|---|---|
| Vanilla RoPEPositional Encoding Method=Vanilla RoPE, DIPE Enhancement=+DIPE2026.03 | 22.7 | |
| Vanilla RoPEPositional Encoding Method=Vanilla RoPE, DIPE Enhancement=Base2026.03 | 19.74 | |
| MRoPEPositional Encoding Method=MRoPE, DIPE Enhancement=+DIPE2026.03 | 19.08 | |
| MRoPE-IPositional Encoding Method=MRoPE-I, DIPE Enhancement=Base2026.03 | 19.08 | |
| MRoPE-IPositional Encoding Method=MRoPE-I, DIPE Enhancement=+DIPE2026.03 | 18.75 | |
| MRoPEPositional Encoding Method=MRoPE, DIPE Enhancement=Base2026.03 | 18.09 |