# Science Question Answering on MMLU-Pro OOD

Source: [Science Question Answering on MMLU-Pro OOD](<https://www.sota2.com/research/sota/science-question-answering-on-mmlu-pro-ood>)

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

Data updated: 2026-04-16T06:55:23.930Z

Evaluation of model performance on the Out-of-Distribution (OOD) subset of the MMLU-Pro dataset specifically for Science Question Answering tasks, reporting standard accuracy.

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 | Science Question Answering |
| Dataset (including any stated split) | MMLU-Pro OOD |
| Metric | Accuracy (MMLU-Pro OOD Science) |
| 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 (MMLU-Pro OOD Science)](<https://www.sota2.com/research/sota/science-question-answering-on-mmlu-pro-ood.md?metric=Accuracy%20(MMLU-Pro%20OOD%20Science)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (MMLU-Pro OOD Science) | Result date | Report paper |
| --- | --- | --- | --- |
| RL-PLUS | 54.7 | Not recorded | 68fa7c411e209131cea8d325 |
| GRPO | 47.5 | Not recorded | 68fa7c411e209131cea8d325 |
| SFT | 42.7 | Not recorded | 68fa7c411e209131cea8d325 |
| SFT+GRPO | 37.7 | Not recorded | 68fa7c411e209131cea8d325 |
| Base Model | 30.2 | Not recorded | 68fa7c411e209131cea8d325 |

### RL-PLUS

Accuracy (MMLU-Pro OOD Science): 54.7. Result ID: 69e087c347f34ce924c9411c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c411e209131cea8d325.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c411e209131cea8d325.md>)

Recorded evaluation configuration: {"Base model":"Qwen2.5-Math-7B"}. This record does not assert independent reproduction.

### GRPO

Accuracy (MMLU-Pro OOD Science): 47.5. Result ID: 69e087c347f34ce924c9411a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c411e209131cea8d325.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c261e209131cea8cb63.md>)

Recorded evaluation configuration: {"Base model":"Qwen2.5-Math-7B"}. This record does not assert independent reproduction.

### SFT

Accuracy (MMLU-Pro OOD Science): 42.7. Result ID: 69e087c347f34ce924c94119.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c411e209131cea8d325.md>)

Recorded evaluation configuration: {"Base model":"Qwen2.5-Math-7B"}. This record does not assert independent reproduction.

### SFT+GRPO

Accuracy (MMLU-Pro OOD Science): 37.7. Result ID: 69e087c347f34ce924c9411b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c411e209131cea8d325.md>)

Recorded evaluation configuration: {"Base model":"Qwen2.5-Math-7B"}. This record does not assert independent reproduction.

### Base Model

Accuracy (MMLU-Pro OOD Science): 30.2. Result ID: 69e087c347f34ce924c94118.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c411e209131cea8d325.md>)

Recorded evaluation configuration: {"Base model":"Qwen2.5-Math-7B"}. This record does not assert independent reproduction.

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