# Mathematical Reasoning on MATH 500 (Pass@1 & Pass@8 Accuracy and Length)

Source: [Mathematical Reasoning on MATH 500 (Pass@1 & Pass@8 Accuracy and Length)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-500-pass-at-1-and-pass-at-8-accuracy-and-length>)

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

Data updated: 2026-03-04T08:44:11.183Z

Evaluation of model performance on the MATH 500 dataset for mathematical reasoning, reporting correctness (accuracy) and average output length for both Pass@1 and Pass@8 sampling strategies.

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 500 |
| Metric | Accuracy Pass@1 |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Accuracy Pass@1](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-500-pass-at-1-and-pass-at-8-accuracy-and-length.md?metric=Accuracy%20Pass%401>): Higher is better; unit: PERCENT.

- [Accuracy Pass@8](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-500-pass-at-1-and-pass-at-8-accuracy-and-length.md?metric=Accuracy%20Pass%408>): Higher is better; unit: PERCENT.

- [Average Length Pass@1](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-500-pass-at-1-and-pass-at-8-accuracy-and-length.md?metric=Average%20Length%20Pass%401>): Lower is better; unit: SCALAR.

- [Average Length Pass@8](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math-500-pass-at-1-and-pass-at-8-accuracy-and-length.md?metric=Average%20Length%20Pass%408>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Accuracy Pass@1 | Result date | Report paper |
| --- | --- | --- | --- |
| DAPO w LATR | 62.6 | Not recorded | 69019f783ef7ccd74e5e39f0 |
| GRPO w LATR | 61.9 | Not recorded | 69019f783ef7ccd74e5e39f0 |
| DAPO w Stoch. | 60.4 | Not recorded | 69019f783ef7ccd74e5e39f0 |
| GRPO w Stoch. | 58.4 | Not recorded | 69019f783ef7ccd74e5e39f0 |
| Qwen2.5-3B | 24.7 | Not recorded | 69019f783ef7ccd74e5e39f0 |

### DAPO w LATR

Accuracy Pass@1: 62.6. Result ID: 69a7ef776dbb7075b94cd0d8.

[Reporting paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-3B","RL Algorithm":"DAPO","Rollout Strategy":"LATR"}. This record does not assert independent reproduction.

### GRPO w LATR

Accuracy Pass@1: 61.9. Result ID: 69a7ef776dbb7075b94cd0d7.

[Reporting paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-3B","RL Algorithm":"GRPO","Rollout Strategy":"LATR"}. This record does not assert independent reproduction.

### DAPO w Stoch.

Accuracy Pass@1: 60.4. Result ID: 69a7ef776dbb7075b94cd0d6.

[Reporting paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a81e209131cea79ac6.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-3B","RL Algorithm":"DAPO","Rollout Strategy":"Stochastic Sampling"}. This record does not assert independent reproduction.

### GRPO w Stoch.

Accuracy Pass@1: 58.4. Result ID: 69a7ef776dbb7075b94cd0d5.

[Reporting paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

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

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

Recorded evaluation configuration: {"Model":"Qwen2.5-3B","RL Algorithm":"GRPO","Rollout Strategy":"Stochastic Sampling"}. This record does not assert independent reproduction.

### Qwen2.5-3B

Accuracy Pass@1: 24.7. Result ID: 69a7ef776dbb7075b94cd0d4.

[Reporting paper](<https://www.sota2.com/research/paper/69019f783ef7ccd74e5e39f0.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-3B","Configuration":"Base Model"}. This record does not assert independent reproduction.

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