Bayesian Assessment of Sycophancy on BASIL Third-p. belief setting 1.0 (test)
0.16Bayesian Error (RMSE)gpt-4o-mini
Evaluation Results
| Method | Links | |
|---|---|---|
| gpt-4o-miniConfidence Elicitation Method=Direct Probing, Number of Samples=52025.08 | 0.16 | |
| gpt-4o-miniConfidence Elicitation Method=Direct Probing, Number of Samples=12025.08 | 0.189 | |
| claude-haiku-4-5Confidence Elicitation Method=Direct Probing, Number of Samples=52025.08 | 0.23 | |
| phi-4:14bConfidence Elicitation Method=Direct Probing, Number of Samples=12025.08 | 0.258 | |
| claude-haiku-4-5Confidence Elicitation Method=Direct Probing, Number of Samples=12025.08 | 0.259 | |
| phi-4:14bConfidence Elicitation Method=Direct Probing, Number of Samples=52025.08 | 0.259 | |
| gpt-4o-miniConfidence Elicitation Method=Hybrid2025.08 | 0.271 | |
| llama-3.2:3bConfidence Elicitation Method=Hybrid2025.08 | 0.292 | |
| llama-3.2:1bConfidence Elicitation Method=Hybrid2025.08 | 0.302 | |
| llama-3.2:1bConfidence Elicitation Method=Direct Probing, Number of Samples=52025.08 | 0.309 | |
| llama-3.2:3bConfidence Elicitation Method=Direct Probing, Number of Samples=52025.08 | 0.312 | |
| llama-3.2:3bConfidence Elicitation Method=Direct Probing, Number of Samples=12025.08 | 0.327 | |
| llama-3.2:1bConfidence Elicitation Method=Direct Probing, Number of Samples=12025.08 | 0.358 | |
| mistral:7bConfidence Elicitation Method=Direct Probing, Number of Samples=52025.08 | 0.386 | |
| phi-4:14bConfidence Elicitation Method=Hybrid2025.08 | 0.443 | |
| mistral:7bConfidence Elicitation Method=Direct Probing, Number of Samples=12025.08 | 0.449 | |
| claude-haiku-4-5Confidence Elicitation Method=Hybrid2025.08 | 0.467 | |
| mistral:7bConfidence Elicitation Method=Hybrid2025.08 | 0.498 |