# Reasoning Quality Evaluation on HelpSteer1 sampled (train)

Source: [Reasoning Quality Evaluation on HelpSteer1 sampled (train)](<https://www.sota2.com/research/sota/reasoning-quality-evaluation-on-helpsteer1-sampled-train>)

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

Data updated: 2026-04-01T06:51:59.819Z

Evaluation of model reasoning quality on the sampled training set of HelpSteer1, focusing on Usefulness, Accuracy, and Consistency.

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 | Reasoning Quality Evaluation |
| Dataset (including any stated split) | HelpSteer1 sampled (train) |
| Metric | Usefulness Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [Usefulness Score](<https://www.sota2.com/research/sota/reasoning-quality-evaluation-on-helpsteer1-sampled-train.md?metric=Usefulness%20Score>): Higher is better; unit: SCALAR.

- [Accuracy Score](<https://www.sota2.com/research/sota/reasoning-quality-evaluation-on-helpsteer1-sampled-train.md?metric=Accuracy%20Score>): Higher is better; unit: SCALAR.

- [Consistency Score](<https://www.sota2.com/research/sota/reasoning-quality-evaluation-on-helpsteer1-sampled-train.md?metric=Consistency%20Score>): Higher is better; unit: SCALAR.

- [Mean Score](<https://www.sota2.com/research/sota/reasoning-quality-evaluation-on-helpsteer1-sampled-train.md?metric=Mean%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Usefulness Score | Result date | Report paper |
| --- | --- | --- | --- |
| MA-SAPO | 3.89 | Not recorded | 68fa7c551e209131cea8d9ae |
| Single-Agent | 3.64 | Not recorded | 68fa7c551e209131cea8d9ae |

### MA-SAPO

Usefulness Score: 3.89. Result ID: 69ccbf6747f34ce924b5da8b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c551e209131cea8d9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c551e209131cea8d9ae.md>)

Recorded evaluation configuration: {"Configuration":"Multi-agent reasoning pipeline"}. This record does not assert independent reproduction.

### Single-Agent

Usefulness Score: 3.64. Result ID: 69ccbf6747f34ce924b5da8a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c551e209131cea8d9ae.md>)

Recorded evaluation configuration: {"Configuration":"Single-agent baseline"}. This record does not assert independent reproduction.

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