# Process-level Reward Modeling on PROCESSBENCH Olymp.Bench

Source: [Process-level Reward Modeling on PROCESSBENCH Olymp.Bench](<https://www.sota2.com/research/sota/process-level-reward-modeling-on-processbench-olymp-bench>)

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

Data updated: 2026-03-04T15:12:38.726Z

Evaluation of model performance on the Olymp.Bench subset of the PROCESSBENCH dataset, specifically for Process-level Reward Modeling, measured using Error, Correctness, and F1 scores.

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 | Process-level Reward Modeling |
| Dataset (including any stated split) | PROCESSBENCH Olymp.Bench |
| Metric | Error |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 6 of 6 |
| 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

- [Error](<https://www.sota2.com/research/sota/process-level-reward-modeling-on-processbench-olymp-bench.md?metric=Error>): Lower is better; unit: PERCENT.

- [Correctness](<https://www.sota2.com/research/sota/process-level-reward-modeling-on-processbench-olymp-bench.md?metric=Correctness>): Higher is better; unit: PERCENT.

- [F1 Score](<https://www.sota2.com/research/sota/process-level-reward-modeling-on-processbench-olymp-bench.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Error | Result date | Report paper |
| --- | --- | --- | --- |
| SPARE-Llama3-8B | 3.3 | Not recorded | 68fa7ad71e209131cea8574d |
| RLHFlow-Deepseek-8B | 10.1 | Not recorded | 68fa7ad71e209131cea8574d |
| SPARE-Qwen2.5-3B | 11.1 | Not recorded | 68fa7ad71e209131cea8574d |
| Math-Shepherd-7B | 15 | Not recorded | 68fa7ad71e209131cea8574d |
| Skywork-7B | 17.9 | Not recorded | 68fa7ad71e209131cea8574d |
| Qwen-2.5-Math-7B-PRM800K (Human) | 35.7 | Not recorded | 68fa7ad71e209131cea8574d |

### SPARE-Llama3-8B

Error: 3.3. Result ID: 69a84bdf6dbb7075b94f564f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

Recorded evaluation configuration: {"# Train":"40.5K"}. This record does not assert independent reproduction.

### RLHFlow-Deepseek-8B

Error: 10.1. Result ID: 69a84bdf6dbb7075b94f564d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dc1e209131cea7adb5.md>)

Recorded evaluation configuration: {"# Train":"250K"}. This record does not assert independent reproduction.

### SPARE-Qwen2.5-3B

Error: 11.1. Result ID: 69a84bdf6dbb7075b94f5650.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

Recorded evaluation configuration: {"# Train":"40.5K"}. This record does not assert independent reproduction.

### Math-Shepherd-7B

Error: 15. Result ID: 69a84bdf6dbb7075b94f564c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72381e209131cea550cf.md>)

Recorded evaluation configuration: {"# Train":"440K"}. This record does not assert independent reproduction.

### Skywork-7B

Error: 17.9. Result ID: 69a84bdf6dbb7075b94f564e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dc1e209131cea7adb5.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen-2.5-Math-7B-PRM800K (Human)

Error: 35.7. Result ID: 69a84bdf6dbb7075b94f5651.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ad71e209131cea8574d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"# Train":"250K"}. This record does not assert independent reproduction.

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