# Correctness Assessment on ProcessBench (test)

Source: [Correctness Assessment on ProcessBench (test)](<https://www.sota2.com/research/sota/correctness-assessment-on-processbench-test>)

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

Data updated: 2026-04-02T09:23:24.060Z

Evaluation of model performance on the test split of the ProcessBench dataset, specifically assessing correctness using distortion metrics derived from QwenPRM and MathShepherd-PRM.

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 | Correctness Assessment |
| Dataset (including any stated split) | ProcessBench (test) |
| Metric | Worst-case Size Distortion (QwenPRM) |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 9 of 9 |
| 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

- [Worst-case Size Distortion (QwenPRM)](<https://www.sota2.com/research/sota/correctness-assessment-on-processbench-test.md?metric=Worst-case%20Size%20Distortion%20(QwenPRM)>): Lower is better; unit: SCALAR.

- [Worst-case Size Distortion (MathShepherd-PRM)](<https://www.sota2.com/research/sota/correctness-assessment-on-processbench-test.md?metric=Worst-case%20Size%20Distortion%20(MathShepherd-PRM)>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Worst-case Size Distortion (QwenPRM) | Result date | Report paper |
| --- | --- | --- | --- |
| naive (1) | 0.24 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| e-score (2) | 0.79 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| e-score (combined) | 0.94 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| e-score (1) | 1 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| e-score (3) | 1.01 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| naive (3) | 1.39 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| naive (2) | 1.89 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| p-score | 7.21 | Not recorded | 6902f1013ef7ccd74e5e4436 |
| p-score (randomized) | 15.8 | Not recorded | 6902f1013ef7ccd74e5e4436 |

### naive (1)

Worst-case Size Distortion (QwenPRM): 0.24. Result ID: 69ce358847f34ce924b7a7c5.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"naive (1)"}. This record does not assert independent reproduction.

### e-score (2)

Worst-case Size Distortion (QwenPRM): 0.79. Result ID: 69ce358847f34ce924b7a7cb.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"e-score (2)"}. This record does not assert independent reproduction.

### e-score (combined)

Worst-case Size Distortion (QwenPRM): 0.94. Result ID: 69ce358847f34ce924b7a7cd.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"e-score (combined)"}. This record does not assert independent reproduction.

### e-score (1)

Worst-case Size Distortion (QwenPRM): 1. Result ID: 69ce358847f34ce924b7a7ca.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"e-score (1)"}. This record does not assert independent reproduction.

### e-score (3)

Worst-case Size Distortion (QwenPRM): 1.01. Result ID: 69ce358847f34ce924b7a7cc.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"e-score (3)"}. This record does not assert independent reproduction.

### naive (3)

Worst-case Size Distortion (QwenPRM): 1.39. Result ID: 69ce358847f34ce924b7a7c7.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"naive (3)"}. This record does not assert independent reproduction.

### naive (2)

Worst-case Size Distortion (QwenPRM): 1.89. Result ID: 69ce358847f34ce924b7a7c6.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"naive (2)"}. This record does not assert independent reproduction.

### p-score

Worst-case Size Distortion (QwenPRM): 7.21. Result ID: 69ce358847f34ce924b7a7c8.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"p-score"}. This record does not assert independent reproduction.

### p-score (randomized)

Worst-case Size Distortion (QwenPRM): 15.8. Result ID: 69ce358847f34ce924b7a7c9.

[Reporting paper](<https://www.sota2.com/research/paper/6902f1013ef7ccd74e5e4436.md>)

Recorded evaluation configuration: {"Score":"p-score (randomized)"}. This record does not assert independent reproduction.

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