Segment-level mistake detection on CaptainCook4D (Average)
72.3EDAAMNAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AMNAREvaluation Protocol=Fully supervised2026.06 | 72.3 | — | |
| AEMEvaluation Protocol=Fully supervised2026.06 | 71.9 | — | |
| EgoPEDEvaluation Protocol=Fully supervised2026.06 | 69.8 | — | |
| ZeProM (Q-M)Evaluation Protocol=Zero-shot, Backbone=Qwen3.5-35B-A3B2026.06 | 69.6 | 14 | |
| ZeProM (Q-L)Evaluation Protocol=Zero-shot, Backbone=Qwen3.5-397B-A17B-GPTQ-Int42026.06 | 69 | 17.9 |