# LLM-as-a-judge on HH-RLHF

Source: [LLM-as-a-judge on HH-RLHF](<https://www.sota2.com/research/sota/llm-as-a-judge-on-hh-rlhf>)

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

Data updated: 2026-05-18T07:07:51.681Z

Evaluation of LLM performance using the LLM-as-a-Judge framework on the HH-RLHF dataset, measured by Coverage and Success percentages.

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 | LLM-as-a-judge |
| Dataset (including any stated split) | HH-RLHF |
| Metric | Coverage |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| 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

- [Coverage](<https://www.sota2.com/research/sota/llm-as-a-judge-on-hh-rlhf.md?metric=Coverage>): Higher is better; unit: PERCENT.

- [Success](<https://www.sota2.com/research/sota/llm-as-a-judge-on-hh-rlhf.md?metric=Success>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Coverage | Result date | Report paper |
| --- | --- | --- | --- |
| Heuristic Selection | 81.3 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| Heuristic Selection | 72.4 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Predictive | 72.3 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Predictive | 58.6 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Ours | 47.2 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Random | 45.4 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Simulated | 42.5 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Vanilla | 41.3 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Ours | 36.9 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Random | 34.2 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Vanilla | 32.7 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |
| CSE + Simulated | 31 | Not recorded | 6a0a9d11dbc836b87d82b3e2 |

### Heuristic Selection

Coverage: 81.3. Result ID: 6a0abaae47f34ce924f05dbf.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"M → L → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### Heuristic Selection

Coverage: 72.4. Result ID: 6a0abaae47f34ce924f05db9.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"L → Q → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Predictive

Coverage: 72.3. Result ID: 6a0abaae47f34ce924f05dc0.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"M → L → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Predictive

Coverage: 58.6. Result ID: 6a0abaae47f34ce924f05dba.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"L → Q → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Ours

Coverage: 47.2. Result ID: 6a0abaae47f34ce924f05dc4.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"M → L → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Random

Coverage: 45.4. Result ID: 6a0abaae47f34ce924f05dc2.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74421e209131cea60928.md>)

Recorded evaluation configuration: {"Cascade":"M → L → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Simulated

Coverage: 42.5. Result ID: 6a0abaae47f34ce924f05dc1.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74421e209131cea60928.md>)

Recorded evaluation configuration: {"Cascade":"M → L → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Vanilla

Coverage: 41.3. Result ID: 6a0abaae47f34ce924f05dc3.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"M → L → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Ours

Coverage: 36.9. Result ID: 6a0abaae47f34ce924f05dbe.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"L → Q → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Random

Coverage: 34.2. Result ID: 6a0abaae47f34ce924f05dbc.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74421e209131cea60928.md>)

Recorded evaluation configuration: {"Cascade":"L → Q → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Vanilla

Coverage: 32.7. Result ID: 6a0abaae47f34ce924f05dbd.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

Recorded evaluation configuration: {"Cascade":"L → Q → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

### CSE + Simulated

Coverage: 31. Result ID: 6a0abaae47f34ce924f05dbb.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d11dbc836b87d82b3e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74421e209131cea60928.md>)

Recorded evaluation configuration: {"Cascade":"L → Q → O","Target agreement level":"0.85"}. This record does not assert independent reproduction.

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