Temporal Hallucination Detection on Charades-STA (test)
0.668AUROCMTLA
Evaluation Results
| Method | Links | |
|---|---|---|
| MTLABackbone=Qwen3-VL-8B2026.07 | 0.668 | |
| InternalConfBackbone=Qwen3-VL-8B2026.07 | 0.528 | |
| last_logpBackbone=Qwen3-VL-8B2026.07 | 0.527 | |
| SVARBackbone=Qwen3-VL-8B2026.07 | 0.511 |