# Bayesian Assessment of Sycophancy on BASIL Third-p. belief setting 1.0 (test) Source: [Bayesian Assessment of Sycophancy on BASIL Third-p. belief setting 1.0 (test)]() Collection index: [llms.txt]() Data updated: 2026-04-21T07:04:33.684Z Evaluation of model performance on the Bayesian Assessment of Sycophancy task using the test split of the BASIL Third-p. belief setting 1.0 dataset, measured by Bayesian error (RMSE). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Bayesian Assessment of Sycophancy | | Dataset (including any stated split) | BASIL Third-p. belief setting 1.0 (test) | | Metric | Bayesian Error (RMSE) | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 18 of 18 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Bayesian Error (RMSE)](): Lower is better; unit: SCALAR. ## Recorded results | Method | Bayesian Error (RMSE) | Result date | Report paper | | --- | --- | --- | --- | | gpt-4o-mini | 0.16 | Not recorded | 68fa7c421e209131cea8d346 | | gpt-4o-mini | 0.189 | Not recorded | 68fa7c421e209131cea8d346 | | claude-haiku-4-5 | 0.23 | Not recorded | 68fa7c421e209131cea8d346 | | phi-4:14b | 0.258 | Not recorded | 68fa7c421e209131cea8d346 | | claude-haiku-4-5 | 0.259 | Not recorded | 68fa7c421e209131cea8d346 | | phi-4:14b | 0.259 | Not recorded | 68fa7c421e209131cea8d346 | | gpt-4o-mini | 0.271 | Not recorded | 68fa7c421e209131cea8d346 | | llama-3.2:3b | 0.292 | Not recorded | 68fa7c421e209131cea8d346 | | llama-3.2:1b | 0.302 | Not recorded | 68fa7c421e209131cea8d346 | | llama-3.2:1b | 0.309 | Not recorded | 68fa7c421e209131cea8d346 | | llama-3.2:3b | 0.312 | Not recorded | 68fa7c421e209131cea8d346 | | llama-3.2:3b | 0.327 | Not recorded | 68fa7c421e209131cea8d346 | | llama-3.2:1b | 0.358 | Not recorded | 68fa7c421e209131cea8d346 | | mistral:7b | 0.386 | Not recorded | 68fa7c421e209131cea8d346 | | phi-4:14b | 0.443 | Not recorded | 68fa7c421e209131cea8d346 | | mistral:7b | 0.449 | Not recorded | 68fa7c421e209131cea8d346 | | claude-haiku-4-5 | 0.467 | Not recorded | 68fa7c421e209131cea8d346 | | mistral:7b | 0.498 | Not recorded | 68fa7c421e209131cea8d346 | ### gpt-4o-mini Bayesian Error (RMSE): 0.16. Result ID: 69e720a447f34ce924cccb41. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"5"}. This record does not assert independent reproduction. ### gpt-4o-mini Bayesian Error (RMSE): 0.189. Result ID: 69e720a447f34ce924cccb35. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"1"}. This record does not assert independent reproduction. ### claude-haiku-4-5 Bayesian Error (RMSE): 0.23. Result ID: 69e720a447f34ce924cccb42. [Reporting paper]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"5"}. This record does not assert independent reproduction. ### phi-4:14b Bayesian Error (RMSE): 0.258. Result ID: 69e720a447f34ce924cccb34. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"1"}. This record does not assert independent reproduction. ### claude-haiku-4-5 Bayesian Error (RMSE): 0.259. Result ID: 69e720a447f34ce924cccb36. [Reporting paper]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"1"}. This record does not assert independent reproduction. ### phi-4:14b Bayesian Error (RMSE): 0.259. Result ID: 69e720a447f34ce924cccb40. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"5"}. This record does not assert independent reproduction. ### gpt-4o-mini Bayesian Error (RMSE): 0.271. Result ID: 69e720a447f34ce924cccb3b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Hybrid"}. This record does not assert independent reproduction. ### llama-3.2:3b Bayesian Error (RMSE): 0.292. Result ID: 69e720a447f34ce924cccb38. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Hybrid"}. This record does not assert independent reproduction. ### llama-3.2:1b Bayesian Error (RMSE): 0.302. Result ID: 69e720a447f34ce924cccb37. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Hybrid"}. This record does not assert independent reproduction. ### llama-3.2:1b Bayesian Error (RMSE): 0.309. Result ID: 69e720a447f34ce924cccb3d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"5"}. This record does not assert independent reproduction. ### llama-3.2:3b Bayesian Error (RMSE): 0.312. Result ID: 69e720a447f34ce924cccb3e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"5"}. This record does not assert independent reproduction. ### llama-3.2:3b Bayesian Error (RMSE): 0.327. Result ID: 69e720a447f34ce924cccb32. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"1"}. This record does not assert independent reproduction. ### llama-3.2:1b Bayesian Error (RMSE): 0.358. Result ID: 69e720a447f34ce924cccb31. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"1"}. This record does not assert independent reproduction. ### mistral:7b Bayesian Error (RMSE): 0.386. Result ID: 69e720a447f34ce924cccb3f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"5"}. This record does not assert independent reproduction. ### phi-4:14b Bayesian Error (RMSE): 0.443. Result ID: 69e720a447f34ce924cccb3a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Hybrid"}. This record does not assert independent reproduction. ### mistral:7b Bayesian Error (RMSE): 0.449. Result ID: 69e720a447f34ce924cccb33. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Direct Probing","Number of Samples":"1"}. This record does not assert independent reproduction. ### claude-haiku-4-5 Bayesian Error (RMSE): 0.467. Result ID: 69e720a447f34ce924cccb3c. [Reporting paper]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Hybrid"}. This record does not assert independent reproduction. ### mistral:7b Bayesian Error (RMSE): 0.498. Result ID: 69e720a447f34ce924cccb39. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Confidence Elicitation Method":"Hybrid"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.