Sycophancy Evaluation on Sycophancy Evaluation
13.3BRRBaseline
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| BaselineModel=Nemotron-30B2026.05 | 13.3 | 18.8 | 81.2 | 78.4 | |
| BaselineModel=Qwen3-8B2026.05 | 10.3 | 16.5 | 83.5 | 84.3 | |
| SFTModel=Qwen3-8B2026.05 | 6.6 | 12.9 | 87.1 | 84.6 | |
| SFTModel=Nemotron-30B2026.05 | 6.6 | 12 | 88 | 86 | |
| BaselineModel=gpt-oss-20b2026.05 | 5.6 | 10.9 | 89.1 | 86.2 | |
| OPCTModel=Nemotron-30B2026.05 | 3.5 | 8.9 | 91.1 | 86.7 | |
| SFTModel=gpt-oss-20b2026.05 | 3.3 | 8.7 | 91.3 | 87.6 | |
| OPCTModel=Qwen3-8B2026.05 | 2.5 | 8.6 | 91.4 | 89.5 | |
| OPCTModel=gpt-oss-20b2026.05 | 1.2 | 6.7 | 93.3 | 90.3 |