# Sycophancy Bias Detection on Anthropic-HH

Source: [Sycophancy Bias Detection on Anthropic-HH](<https://www.sota2.com/research/sota/sycophancy-bias-detection-on-anthropic-hh>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:24:18.705Z

Evaluation of model performance specifically targeting the detection of sycophancy bias within responses generated on the Anthropic-HH dataset, typically assessed using metrics like AUC, AP, and TNR80.

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 | Sycophancy Bias Detection |
| Dataset (including any stated split) | Anthropic-HH |
| Metric | AUC |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [AUC](<https://www.sota2.com/research/sota/sycophancy-bias-detection-on-anthropic-hh.md?metric=AUC>): Higher is better; unit: SCALAR.

- [AP](<https://www.sota2.com/research/sota/sycophancy-bias-detection-on-anthropic-hh.md?metric=AP>): Higher is better; unit: SCALAR.

- [TNR80](<https://www.sota2.com/research/sota/sycophancy-bias-detection-on-anthropic-hh.md?metric=TNR80>): Higher is better; unit: SCALAR.

## Recorded results

| Method | AUC | Result date | Report paper |
| --- | --- | --- | --- |
| Influence | 0.711 | Not recorded | 68fa7af61e209131cea85fc4 |
| Confidence | 0.585 | Not recorded | 68fa7af61e209131cea85fc4 |
| Mahalanobis | 0.56 | Not recorded | 68fa7af61e209131cea85fc4 |
| Entropy | 0.533 | Not recorded | 68fa7af61e209131cea85fc4 |
| KNN | 0.533 | Not recorded | 68fa7af61e209131cea85fc4 |

### Influence

AUC: 0.711. Result ID: 698b80377a967eddb76f1b6e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fc4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68f81e209131cea1f643.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Confidence

AUC: 0.585. Result ID: 698b80377a967eddb76f1b6f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fc4.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Mahalanobis

AUC: 0.56. Result ID: 698b80377a967eddb76f1b71.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fc4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64f91e209131cea077ac.md>)

[Implementation](<https://github.com/pokaxpoka/deep_Mahalanobis_detector>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Entropy

AUC: 0.533. Result ID: 698b80377a967eddb76f1b70.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fc4.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### KNN

AUC: 0.533. Result ID: 698b80377a967eddb76f1b72.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fc4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Implementation](<https://github.com/deeplearning-wisc/knn-ood>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.