# Offline Constrained RLHF on PKU-SafeRLHF 7k (test)

Source: [Offline Constrained RLHF on PKU-SafeRLHF 7k (test)](<https://www.sota2.com/research/sota/offline-constrained-rlhf-on-pku-saferlhf-7k-test>)

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

Data updated: 2026-04-02T08:43:04.717Z

Evaluation of model performance on the test split of the PKU-SafeRLHF 7k dataset, focusing on Offline Constrained RLHF, measured by safety reward (r1), helpfulness reward (r2), and violation rate.

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 | Offline Constrained RLHF |
| Dataset (including any stated split) | PKU-SafeRLHF 7k (test) |
| Metric | Safety Reward (r1) |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 1 of 1 |
| 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

- [Safety Reward (r1)](<https://www.sota2.com/research/sota/offline-constrained-rlhf-on-pku-saferlhf-7k-test.md?metric=Safety%20Reward%20(r1)>): Higher is better; unit: SCALAR.

- [Helpfulness Reward (r2)](<https://www.sota2.com/research/sota/offline-constrained-rlhf-on-pku-saferlhf-7k-test.md?metric=Helpfulness%20Reward%20(r2)>): Higher is better; unit: SCALAR.

- [Violation Rate](<https://www.sota2.com/research/sota/offline-constrained-rlhf-on-pku-saferlhf-7k-test.md?metric=Violation%20Rate>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Safety Reward (r1) | Result date | Report paper |
| --- | --- | --- | --- |
| πλ | -1.6044 | Not recorded | 69cdf7f20636e5bf4afcc99b |

### πλ

Safety Reward (r1): -1.6044. Result ID: 69ce2bef47f34ce924b75d87.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf7f20636e5bf4afcc99b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cdf7f20636e5bf4afcc99b.md>)

Recorded evaluation configuration: {"Policy Type":"Dual-only learned policy","Jmin":"−0.6473","η":"0.3","T":"1000"}. This record does not assert independent reproduction.

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