Self-doubt detection on MuSR 90-trace
83.66AUROC (Self-doubt)SELFDOUBT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SELFDOUBTModel=gpt-oss-120b2026.04 | 83.66 | 0.004 | |
| SELFDOUBTModel=Claude Sonnet 4.62026.04 | 83.58 | 0 | |
| SELFDOUBTModel=Grok 4.1 Fast2026.04 | 80.89 | 0.001 | |
| SELFDOUBTModel=gpt-oss-20b2026.04 | 78.67 | 0.004 | |
| SELFDOUBTModel=Qwen3-14B2026.04 | 78.53 | 0.002 | |
| SELFDOUBTModel=Qwen32026.04 | 76.54 | 0.009 | |
| SELFDOUBTModel=Gemini 2.5 Flash2026.04 | 64.75 | 0.04 |