Mistake detection on BridgeData V2 (test)
59.5APQwen 2.5 (ft)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen 2.5 (ft)Evaluation protocol=fine-tuned, Inf.Time (m)=100.32026.03 | 59.5 | 44.87 | 51.16 | |
| Qwen 2.5Evaluation protocol=zero-shot, Inf.Time (m)=119.72026.03 | 58.21 | 29.8 | 39.42 | |
| TIMIDEvaluation protocol=trained, Inf.Time (m)=0.022026.03 | 49.72 | 33.77 | 40.22 | |
| PEL4VADEvaluation protocol=trained, Inf.Time (m)=0.022026.03 | 10.18 | 19.29 | 13.33 | |
| Auto-EncoderEvaluation protocol=semi-supervised, Inf.Time (m)=1.162026.03 | 9.33 | 8.09 | 8.67 |