Detection Accuracy on LongForm QA
99.88AccuracyMC2MARK
Evaluation Results
| Method | Links | |
|---|---|---|
| MC2MARKmessage_length=16, base_model=Phi-3.5-mini-instruct2026.02 | 99.88 | |
| MC2MARKmessage_length=64, base_model=Phi-3.5-mini-instruct2026.02 | 99.33 | |
| MC2MARKmessage_length=32, base_model=Phi-3.5-mini-instruct2026.02 | 98.91 | |
| MC2MARKmessage_length=128, base_model=Phi-3.5-mini-instruct2026.02 | 98.36 | |
| BiMarkmessage_length=16, base_model=Phi-3.5-mini-instruct2026.02 | 97.21 | |
| MPAC(1.5)message_length=16, base_model=Phi-3.5-mini-instruct2026.02 | 94.16 | |
| MC2MARKmessage_length=256, base_model=Phi-3.5-mini-instruct2026.02 | 93.2 | |
| BiMarkmessage_length=32, base_model=Phi-3.5-mini-instruct2026.02 | 92.12 | |
| MC2MARKmessage_length=512, base_model=Phi-3.5-mini-instruct2026.02 | 87.37 | |
| MPAC(1.0)message_length=16, base_model=Phi-3.5-mini-instruct2026.02 | 83.69 | |
| BiMarkmessage_length=64, base_model=Phi-3.5-mini-instruct2026.02 | 82.1 | |
| MPAC(1.5)message_length=32, base_model=Phi-3.5-mini-instruct2026.02 | 81.11 | |
| MPAC(1.5)message_length=64, base_model=Phi-3.5-mini-instruct2026.02 | 72.66 | |
| BiMarkmessage_length=128, base_model=Phi-3.5-mini-instruct2026.02 | 69.62 | |
| MPAC(1.0)message_length=32, base_model=Phi-3.5-mini-instruct2026.02 | 68.44 | |
| MPAC(1.5)message_length=128, base_model=Phi-3.5-mini-instruct2026.02 | 67.35 | |
| MPAC(1.0)message_length=64, base_model=Phi-3.5-mini-instruct2026.02 | 67 | |
| MPAC(1.5)message_length=256, base_model=Phi-3.5-mini-instruct2026.02 | 61.89 | |
| MPAC(1.0)message_length=128, base_model=Phi-3.5-mini-instruct2026.02 | 59.57 | |
| MPAC(1.0)message_length=256, base_model=Phi-3.5-mini-instruct2026.02 | 57.66 | |
| MPAC(1.5)message_length=512, base_model=Phi-3.5-mini-instruct2026.02 | 56.98 | |
| MPAC(1.0)message_length=512, base_model=Phi-3.5-mini-instruct2026.02 | 55.42 | |
| BiMarkmessage_length=256, base_model=Phi-3.5-mini-instruct2026.02 | 52.87 | |
| BiMarkmessage_length=512, base_model=Phi-3.5-mini-instruct2026.02 | 34.45 |