Predicate Detection on DROID phasic episodes (test)
96.8F1 ScoreETL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ETLTask=Obj. → paper, Number of phases (Ph.)=22026.05 | 96.8 | 97.7 | — | |
| ETLTask=White → blue cloth, Number of phases (Ph.)=32026.05 | 88.4 | 95.9 | — | |
| ETLTask=Bottles → stove, Number of phases (Ph.)=32026.05 | 84.4 | 96.6 | — | |
| ETLTask=Tap → spoon, Number of phases (Ph.)=22026.05 | 83.9 | 93.7 | — | |
| ETLTask=Mean2026.05 | 81.3 | 94 | 0.8 | |
| Qwen2-VL-2BTask=4-way chain, Number of phases (Ph.)=42026.05 | 58.5 | 81.1 | — | |
| ETLTask=4-way chain, Number of phases (Ph.)=42026.05 | 53.1 | 86.2 | — | |
| Qwen2-VL-2BTask=Obj. → paper, Number of phases (Ph.)=22026.05 | 49.3 | 38.3 | — | |
| Qwen2-VL-2BTask=Mean2026.05 | 39 | 56.7 | 0.4 | |
| Qwen2-VL-2BTask=White → blue cloth, Number of phases (Ph.)=32026.05 | 33.5 | 25.8 | — | |
| Qwen2-VL-2BTask=Tap → spoon, Number of phases (Ph.)=22026.05 | 29.2 | 70 | — | |
| Qwen2-VL-2BTask=Bottles → stove, Number of phases (Ph.)=32026.05 | 24.4 | 68.5 | — |