# Mathematical Reasoning on MATH (Acc@8 and Relative Change)

Source: [Mathematical Reasoning on MATH (Acc@8 and Relative Change)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-acc-at-8-and-relative-change>)

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

Data updated: 2026-07-03T07:18:15.286Z

Evaluation of model performance on the MATH dataset focusing on the Acc@8 score and the reported Relative Change metric.

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 | Mathematical Reasoning |
| Dataset (including any stated split) | MATH |
| Metric | Acc@8 |
| Unit | PERCENT |
| Direction | Higher 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

- [Acc@8](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-acc-at-8-and-relative-change.md?metric=Acc%408>): Higher is better; unit: PERCENT.

- [Relative Change](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-acc-at-8-and-relative-change.md?metric=Relative%20Change>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Acc@8 | Result date | Report paper |
| --- | --- | --- | --- |
| GRPO | 57.29 | Not recorded | 6a47420c143ebdcacacc5c88 |
| SDPO0 | 56.42 | Not recorded | 6a47420c143ebdcacacc5c88 |
| GRPO | 44.67 | Not recorded | 6a47420c143ebdcacacc5c88 |
| SDPOCoT0 | 38.13 | Not recorded | 6a47420c143ebdcacacc5c88 |
| SDPO0 | 34.38 | Not recorded | 6a47420c143ebdcacacc5c88 |
| SDPO5% | 34.38 | Not recorded | 6a47420c143ebdcacacc5c88 |
| Baseline | 32.71 | Not recorded | 6a47420c143ebdcacacc5c88 |
| SDPOCoT0 | 14.17 | Not recorded | 6a47420c143ebdcacacc5c88 |
| SDPO5% | 0 | Not recorded | 6a47420c143ebdcacacc5c88 |

### GRPO

Acc@8: 57.29. Result ID: 6a47623560b88bb33992a4ec.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Stage":"Last"}. This record does not assert independent reproduction.

### SDPO0

Acc@8: 56.42. Result ID: 6a47623560b88bb33992a4ed.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Implementation](<https://github.com/Moenupa/SDPO-CL>)

Recorded evaluation configuration: {"Stage":"Current"}. This record does not assert independent reproduction.

### GRPO

Acc@8: 44.67. Result ID: 6a47623560b88bb33992a4eb.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Stage":"Current"}. This record does not assert independent reproduction.

### SDPOCoT0

Acc@8: 38.13. Result ID: 6a47623560b88bb33992a4ef.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Implementation](<https://github.com/Moenupa/SDPO-CL>)

Recorded evaluation configuration: {"Stage":"Current"}. This record does not assert independent reproduction.

### SDPO0

Acc@8: 34.38. Result ID: 6a47623560b88bb33992a4ee.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Implementation](<https://github.com/Moenupa/SDPO-CL>)

Recorded evaluation configuration: {"Stage":"Last"}. This record does not assert independent reproduction.

### SDPO5%

Acc@8: 34.38. Result ID: 6a47623560b88bb33992a4f1.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Implementation](<https://github.com/Moenupa/SDPO-CL>)

Recorded evaluation configuration: {"Stage":"Current"}. This record does not assert independent reproduction.

### Baseline

Acc@8: 32.71. Result ID: 6a47623560b88bb33992a4ea.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

Recorded evaluation configuration: {"Stage":"Baseline"}. This record does not assert independent reproduction.

### SDPOCoT0

Acc@8: 14.17. Result ID: 6a47623560b88bb33992a4f0.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Implementation](<https://github.com/Moenupa/SDPO-CL>)

Recorded evaluation configuration: {"Stage":"Last"}. This record does not assert independent reproduction.

### SDPO5%

Acc@8: 0. Result ID: 6a47623560b88bb33992a4f2.

[Reporting paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a47420c143ebdcacacc5c88.md>)

[Implementation](<https://github.com/Moenupa/SDPO-CL>)

Recorded evaluation configuration: {"Stage":"Last"}. This record does not assert independent reproduction.

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