# OOD Bias Evaluation on Winobias

Source: [OOD Bias Evaluation on Winobias](<https://www.sota2.com/research/sota/ood-bias-evaluation-on-winobias>)

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

Data updated: 2026-02-26T05:31:07.319Z

Evaluation of model performance regarding Out-of-Domain Bias using the Winobias dataset, typically measured by Accuracy (Acc).

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 | Out-of-Domain (OOD) Bias Evaluation |
| Dataset (including any stated split) | Winobias |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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](<https://www.sota2.com/research/sota/ood-bias-evaluation-on-winobias.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| DPO | 0.507 | Not recorded | 69535c9853fdd5c36173db27 |
| FR | 0.506 | Not recorded | 69535c9853fdd5c36173db27 |
| IPO | 0.501 | Not recorded | 69535c9853fdd5c36173db27 |
| C2PO | 0.501 | Not recorded | 69535c9853fdd5c36173db27 |
| CPO | 0.5 | Not recorded | 69535c9853fdd5c36173db27 |
| BCO | 0.5 | Not recorded | 69535c9853fdd5c36173db27 |
| GRPO | 0.5 | Not recorded | 69535c9853fdd5c36173db27 |
| CPO | 0.499 | Not recorded | 69535c9853fdd5c36173db27 |
| FR | 0.499 | Not recorded | 69535c9853fdd5c36173db27 |
| C2PO | 0.496 | Not recorded | 69535c9853fdd5c36173db27 |
| GRPO | 0.494 | Not recorded | 69535c9853fdd5c36173db27 |
| DPO | 0.492 | Not recorded | 69535c9853fdd5c36173db27 |
| IPO | 0.491 | Not recorded | 69535c9853fdd5c36173db27 |
| BCO | 0.484 | Not recorded | 69535c9853fdd5c36173db27 |

### DPO

Accuracy: 0.507. Result ID: 6996ac3e5698665ec4f65185.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

### FR

Accuracy: 0.506. Result ID: 6996ac3e5698665ec4f6518a.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78ef1e209131cea7b4cf.md>)

[Implementation](<https://github.com/shoyua/Towards-Reward-Fairness>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

### IPO

Accuracy: 0.501. Result ID: 6996ac3e5698665ec4f6518f.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa777e1e209131cea738f3.md>)

[Implementation](<https://github.com/shivank21/Implicit_Preference_Optimization>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### C2PO

Accuracy: 0.501. Result ID: 6996ac3e5698665ec4f65192.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### CPO

Accuracy: 0.5. Result ID: 6996ac3e5698665ec4f6518d.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c31e.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### BCO

Accuracy: 0.5. Result ID: 6996ac3e5698665ec4f6518e.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa792f1e209131cea7c79e.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### GRPO

Accuracy: 0.5. Result ID: 6996ac3e5698665ec4f65190.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73601e209131cea5c09a.md>)

[Implementation](<https://github.com/rsshyam/Group-robust-preference-optimization>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### CPO

Accuracy: 0.499. Result ID: 6996ac3e5698665ec4f65186.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c31e.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

### FR

Accuracy: 0.499. Result ID: 6996ac3e5698665ec4f65191.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78ef1e209131cea7b4cf.md>)

[Implementation](<https://github.com/shoyua/Towards-Reward-Fairness>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### C2PO

Accuracy: 0.496. Result ID: 6996ac3e5698665ec4f6518b.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

### GRPO

Accuracy: 0.494. Result ID: 6996ac3e5698665ec4f65189.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73601e209131cea5c09a.md>)

[Implementation](<https://github.com/rsshyam/Group-robust-preference-optimization>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

### DPO

Accuracy: 0.492. Result ID: 6996ac3e5698665ec4f6518c.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Qwen-14B"}. This record does not assert independent reproduction.

### IPO

Accuracy: 0.491. Result ID: 6996ac3e5698665ec4f65188.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa777e1e209131cea738f3.md>)

[Implementation](<https://github.com/shivank21/Implicit_Preference_Optimization>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

### BCO

Accuracy: 0.484. Result ID: 6996ac3e5698665ec4f65187.

[Reporting paper](<https://www.sota2.com/research/paper/69535c9853fdd5c36173db27.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa792f1e209131cea7c79e.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA-2-13B-Chat"}. This record does not assert independent reproduction.

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