Multi-task and Multi-image Reasoning on MMT-Bench
58.5SI ScoreGRACE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GRACEData %=20%2026.05 | 58.5 | 56.3 | |
| GRACEData %=10%2026.05 | 57.7 | 55.1 | |
| GRACEData %=15%2026.05 | 57.6 | 54.9 | |
| CADCData %=20%2026.05 | 57.5 | 55.2 | |
| RandomData %=20%2026.05 | 57.4 | 54.8 | |
| StepmaxData %=20%2026.05 | 56.7 | 54.5 | |
| GRACEData %=5%2026.05 | 56.5 | 55.1 | |
| LongestData %=20%2026.05 | 56.4 | 54.6 | |
| LESSData %=20%2026.05 | 56 | 53.7 | |
| ICONSData %=20%2026.05 | 55.5 | 53.8 | |
| FullData %=100%2026.05 | 55 | 53.4 |