# Bias Evaluation on BBQ (Averaged across Gender, Nationality, Religion)

Source: [Bias Evaluation on BBQ (Averaged across Gender, Nationality, Religion)](<https://www.sota2.com/research/sota/bias-evaluation-on-bbq-averaged-across-gender-nationality-religion>)

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

Data updated: 2026-02-26T05:34:08.275Z

Evaluation of model bias on the BBQ dataset, reporting averaged accuracy specifically across the gender, nationality, and religion domains.

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 | Bias Evaluation |
| Dataset (including any stated split) | BBQ averaged across gender, nationality, and religion domains |
| Metric | Accuracy (Ambiguous) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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

- [Accuracy (Ambiguous)](<https://www.sota2.com/research/sota/bias-evaluation-on-bbq-averaged-across-gender-nationality-religion.md?metric=Accuracy%20(Ambiguous)>): Higher is better; unit: PERCENT.

- [Accuracy (Disambiguated)](<https://www.sota2.com/research/sota/bias-evaluation-on-bbq-averaged-across-gender-nationality-religion.md?metric=Accuracy%20(Disambiguated)>): Higher is better; unit: PERCENT.

- [Average Accuracy](<https://www.sota2.com/research/sota/bias-evaluation-on-bbq-averaged-across-gender-nationality-religion.md?metric=Average%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Ambiguous) | Result date | Report paper |
| --- | --- | --- | --- |
| Self-Debiasing | 87.73 | Not recorded | 69842418a8b01e76cc8ee74f |
| Prefix Prompting | 82.81 | Not recorded | 69842418a8b01e76cc8ee74f |
| Prefix Prompting | 75.13 | Not recorded | 69842418a8b01e76cc8ee74f |
| BBA | 74.46 | Not recorded | 69842418a8b01e76cc8ee74f |
| FBA | 68.78 | Not recorded | 69842418a8b01e76cc8ee74f |
| BBA | 68.08 | Not recorded | 69842418a8b01e76cc8ee74f |
| FBA | 66.45 | Not recorded | 69842418a8b01e76cc8ee74f |
| Self-Debiasing | 66.06 | Not recorded | 69842418a8b01e76cc8ee74f |
| DDP | 60.38 | Not recorded | 69842418a8b01e76cc8ee74f |
| IG^2 | 59 | Not recorded | 69842418a8b01e76cc8ee74f |
| Base | 58.95 | Not recorded | 69842418a8b01e76cc8ee74f |
| Auto-Debias | 58.53 | Not recorded | 69842418a8b01e76cc8ee74f |
| DDP | 56.87 | Not recorded | 69842418a8b01e76cc8ee74f |
| Auto-Debias | 54.61 | Not recorded | 69842418a8b01e76cc8ee74f |
| Base | 54.5 | Not recorded | 69842418a8b01e76cc8ee74f |
| IG^2 | 52.61 | Not recorded | 69842418a8b01e76cc8ee74f |

### Self-Debiasing

Accuracy (Ambiguous): 87.73. Result ID: 69964d4f5698665ec491d11c.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa721d1e209131cea5497f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### Prefix Prompting

Accuracy (Ambiguous): 82.81. Result ID: 69964d4f5698665ec491d123.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73311e209131cea5b0a7.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### Prefix Prompting

Accuracy (Ambiguous): 75.13. Result ID: 69964d4f5698665ec491d11b.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73311e209131cea5b0a7.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### BBA

Accuracy (Ambiguous): 74.46. Result ID: 69964d4f5698665ec491d120.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### FBA

Accuracy (Ambiguous): 68.78. Result ID: 69964d4f5698665ec491d127.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### BBA

Accuracy (Ambiguous): 68.08. Result ID: 69964d4f5698665ec491d128.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### FBA

Accuracy (Ambiguous): 66.45. Result ID: 69964d4f5698665ec491d11f.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### Self-Debiasing

Accuracy (Ambiguous): 66.06. Result ID: 69964d4f5698665ec491d124.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa721d1e209131cea5497f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### DDP

Accuracy (Ambiguous): 60.38. Result ID: 69964d4f5698665ec491d125.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773d1e209131cea71e88.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### IG^2

Accuracy (Ambiguous): 59. Result ID: 69964d4f5698665ec491d11e.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73ad1e209131cea5de3d.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### Base

Accuracy (Ambiguous): 58.95. Result ID: 69964d4f5698665ec491d121.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### Auto-Debias

Accuracy (Ambiguous): 58.53. Result ID: 69964d4f5698665ec491d122.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

### DDP

Accuracy (Ambiguous): 56.87. Result ID: 69964d4f5698665ec491d11d.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773d1e209131cea71e88.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### Auto-Debias

Accuracy (Ambiguous): 54.61. Result ID: 69964d4f5698665ec491d11a.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### Base

Accuracy (Ambiguous): 54.5. Result ID: 69964d4f5698665ec491d119.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1"}. This record does not assert independent reproduction.

### IG^2

Accuracy (Ambiguous): 52.61. Result ID: 69964d4f5698665ec491d126.

[Reporting paper](<https://www.sota2.com/research/paper/69842418a8b01e76cc8ee74f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73ad1e209131cea5de3d.md>)

Recorded evaluation configuration: {"Model":"Llama-3.2"}. This record does not assert independent reproduction.

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