Deception Detection on InstrumentalEval (InstEval)
0.867AUROCSTATEWITNESS
Evaluation results
| Method | Links | |
|---|---|---|
| STATEWITNESSTarget Model=GPT-OSS-20B2026.06 | 0.867 | |
| STATEWITNESSTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.823 | |
| LLM-as-a-JudgeTarget Model=GPT-OSS-20B2026.06 | 0.741 | |
| LLM-as-a-JudgeTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.681 | |
| CMMS probeTarget Model=GPT-OSS-20B2026.06 | 0.665 | |
| MMS probeTarget Model=GPT-OSS-20B2026.06 | 0.663 | |
| LLM-as-a-Judge w/ CoTTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.636 | |
| CMMS probeTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.634 | |
| LLM-as-a-Judge w/ CoTTarget Model=GPT-OSS-20B2026.06 | 0.622 | |
| Self-EvaluationTarget Model=GPT-OSS-20B2026.06 | 0.607 | |
| LR probeTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.588 | |
| Self-EvaluationTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.558 | |
| LAT probeTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.506 | |
| MMS probeTarget Model=GEMMA4-26B-A4B-IT2026.06 | 0.5 | |
| LR probeTarget Model=GPT-OSS-20B2026.06 | 0.33 | |
| LAT probeTarget Model=GPT-OSS-20B2026.06 | 0.19 |