Theory of Mind Inference on SAVVY
50.66AccuracyObserve-to-Believe Pipeline (Exp 4)
Evaluation Results
| Method | Links | |
|---|---|---|
| Observe-to-Believe Pipeline (Exp 4)Context=Full, Audio=✗2026.05 | 50.66 | |
| Observe-to-Believe Pipeline (Exp 5)Context=Full, Audio=✓2026.05 | 50 | |
| Observe-to-Believe Pipeline (Exp 3)Context=Partial, Audio=✗2026.05 | 49.78 | |
| Observe-to-Believe Pipeline (Exp 2)Context=Partial, Audio=✓2026.05 | 48.46 | |
| SAVVY Egocentric (Baseline 1)Context=None, Audio=✓2026.05 | 34.36 | |
| SAVVY Allocentric (Baseline 2)Context=None, Audio=✓2026.05 | 24.42 |