Compositional Visual Reasoning on CVR (Sample Size Sweep)
47.4Accuracy (Joint)
PR-A2CL
Updated 1mo ago
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
2026.03 | 47.4 | 47.2 | 68.9 | 67.5 | |
2026.03 | 46.5 | 45.3 | 67.3 | 65.6 | |
2026.03 | 44.8 | 40.9 | 63.8 | 60.3 | |
2026.03 | 44.3 | 40.5 | 59.6 | 54.5 | |
2026.03 | 39.3 | 46.7 | 46.3 | 56.4 | |
2026.03 | 32 | 33.9 | 46.2 | 46.6 | |
2026.03 | 31.1 | 33.9 | 52.7 | 48.9 | |
2026.03 | 28.7 | 34 | 47.1 | 48.5 | |
2026.03 | 27.9 | 31.7 | 32.9 | 37 | |
2026.03 | 27.5 | 28 | 38.4 | 34.9 | |
2026.03 | 27.3 | 28.6 | 28.7 | 31.7 | |
2026.03 | 26.9 | 28.9 | 37.5 | 38.5 | |
2026.03 | 26.8 | 30 | 32 | 34.1 | |
2026.03 | 26.4 | 31.4 | 40.4 | 39.5 | |
2026.03 | 26.2 | 27.1 | 27.4 | 28.4 | |
2026.03 | 25.8 | 26.9 | 33.9 | 30.3 |