# Explanation Faithfulness Evaluation on BiasQA (test)

Source: [Explanation Faithfulness Evaluation on BiasQA (test)](<https://www.sota2.com/research/sota/explanation-faithfulness-evaluation-on-biasqa-test>)

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

Data updated: 2026-07-08T07:41:24.761Z

Evaluation of explanation faithfulness on the test split of the BiasQA dataset, measured using metrics like Greedy and Maj@16.

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 | Explanation Faithfulness Evaluation |
| Dataset (including any stated split) | BiasQA (test) |
| Metric | Greedy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [Greedy](<https://www.sota2.com/research/sota/explanation-faithfulness-evaluation-on-biasqa-test.md?metric=Greedy>): Higher is better; unit: PERCENT.

- [Maj@16](<https://www.sota2.com/research/sota/explanation-faithfulness-evaluation-on-biasqa-test.md?metric=Maj%4016>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Greedy | Result date | Report paper |
| --- | --- | --- | --- |
| DPO + RM | 20.7 | Not recorded | 68fa7a0d1e209131cea80d2f |
| DPO + RMC | 18.2 | Not recorded | 68fa7a0d1e209131cea80d2f |
| DPO + RMD | 16.5 | Not recorded | 68fa7a0d1e209131cea80d2f |
| DPO + RMD | 15.3 | Not recorded | 68fa7a0d1e209131cea80d2f |
| Base | 14.9 | Not recorded | 68fa7a0d1e209131cea80d2f |
| DPO + RM | 14.1 | Not recorded | 68fa7a0d1e209131cea80d2f |
| DPO + RMC | 13.8 | Not recorded | 68fa7a0d1e209131cea80d2f |

### DPO + RM

Greedy: 20.7. Result ID: 6a4dff1460b88bb339d75b9e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

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

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"DPO","Reward Model":"SK-GEMMA-27B"}. This record does not assert independent reproduction.

### DPO + RMC

Greedy: 18.2. Result ID: 6a4dff1460b88bb339d75ba0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"DPO","Reward Model":"SK-GEMMA-27B"}. This record does not assert independent reproduction.

### DPO + RMD

Greedy: 16.5. Result ID: 6a4dff1460b88bb339d75b9f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"DPO","Reward Model":"SK-GEMMA-27B"}. This record does not assert independent reproduction.

### DPO + RMD

Greedy: 15.3. Result ID: 6a4dff1460b88bb339d75b9c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"DPO","Reward Model":"SK-LLAMA-8B"}. This record does not assert independent reproduction.

### Base

Greedy: 14.9. Result ID: 6a4dff1460b88bb339d75b9a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"Base","Reward Model":"None"}. This record does not assert independent reproduction.

### DPO + RM

Greedy: 14.1. Result ID: 6a4dff1460b88bb339d75b9b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

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

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"DPO","Reward Model":"SK-LLAMA-8B"}. This record does not assert independent reproduction.

### DPO + RMC

Greedy: 13.8. Result ID: 6a4dff1460b88bb339d75b9d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d2f.md>)

Recorded evaluation configuration: {"Base Model":"LLAMA-3.2-3B-IT","Training Method":"DPO","Reward Model":"SK-LLAMA-8B"}. This record does not assert independent reproduction.

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