Action Generation on Do-Undo (evaluation)
89DINO-RGemini
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Geminitype=Understanding & Generation2025.12 | 89 | 81.8 | 31.5 | 31 | 0.049 | 83.3 | |
| Bagel-think (7B)type=Understanding & Generation, seeds=3 (mean and std. deviation)2025.12 | 83.7 | 81.3 | 29.8 | 30.1 | 0.128 | 85 | |
| Do-Undotype=Understanding & Generation, seeds=3 (mean and std. deviation)2025.12 | 81.9 | 80.7 | 28.2 | 28.1 | 0.134 | 89.2 | |
| Qwen (20B)type=Understanding & Generation2025.12 | 81.5 | 77 | 31.2 | 31.4 | 0.114 | 76.2 | |
| Do-Undo(c)type=Understanding & Generation, seeds=3 (mean and std. deviation)2025.12 | 81.2 | 80.7 | 27.9 | 28.2 | 0.133 | 86.8 | |
| Flux Kontext (7B)type=Generation2025.12 | 77.6 | 75 | 30.3 | 29.9 | 0.129 | 69.7 |