# Question Answering on MMLU STEM (Acc, Len, Refl/Ans)

Source: [Question Answering on MMLU STEM (Acc, Len, Refl/Ans)](<https://www.sota2.com/research/sota/question-answering-on-mmlu-stem-acc-len-refl-ans>)

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

Data updated: 2026-05-01T07:33:03.874Z

Evaluation of model performance on the Question Answering task across the STEM subjects subset of the MMLU benchmark, reporting standard accuracy alongside length and reflection/answer counts.

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 | Question Answering |
| Dataset (including any stated split) | MMLU STEM |
| Metric | Accuracy |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/question-answering-on-mmlu-stem-acc-len-refl-ans.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Answer Length](<https://www.sota2.com/research/sota/question-answering-on-mmlu-stem-acc-len-refl-ans.md?metric=Answer%20Length>): Higher is better; unit: SCALAR.

- [Reflection/Answer Ratio](<https://www.sota2.com/research/sota/question-answering-on-mmlu-stem-acc-len-refl-ans.md?metric=Reflection%2FAnswer%20Ratio>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| PLE | 92.22 | Not recorded | 69f4336355a3c32af89fcf86 |
| PLE | 92.22 | Not recorded | 69f4336355a3c32af89fcf86 |
| Hybrid | 90.26 | Not recorded | 69f4336355a3c32af89fcf86 |
| PLE | 88.89 | Not recorded | 69f4336355a3c32af89fcf86 |
| SFT-only | 85.82 | Not recorded | 69f4336355a3c32af89fcf86 |
| Hybrid | 84.49 | Not recorded | 69f4336355a3c32af89fcf86 |
| PLE | 83.7 | Not recorded | 69f4336355a3c32af89fcf86 |
| Instruct | 62 | Not recorded | 69f4336355a3c32af89fcf86 |
| SFT-only | 59.44 | Not recorded | 69f4336355a3c32af89fcf86 |

### PLE

Accuracy: 92.22. Result ID: 69f4559747f34ce924d9ddf6.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B","Mode":"No-think"}. This record does not assert independent reproduction.

### PLE

Accuracy: 92.22. Result ID: 69f4559747f34ce924d9ddf9.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Mode":"Think"}. This record does not assert independent reproduction.

### Hybrid

Accuracy: 90.26. Result ID: 69f4559747f34ce924d9ddf7.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Mode":"Think"}. This record does not assert independent reproduction.

### PLE

Accuracy: 88.89. Result ID: 69f4559747f34ce924d9ddf5.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B","Mode":"Think"}. This record does not assert independent reproduction.

### SFT-only

Accuracy: 85.82. Result ID: 69f4559747f34ce924d9ddf3.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c1c1e209131cea8c661.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B","Mode":"Think"}. This record does not assert independent reproduction.

### Hybrid

Accuracy: 84.49. Result ID: 69f4559747f34ce924d9ddf8.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Mode":"No-think"}. This record does not assert independent reproduction.

### PLE

Accuracy: 83.7. Result ID: 69f4559747f34ce924d9ddfa.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Mode":"No-think"}. This record does not assert independent reproduction.

### Instruct

Accuracy: 62. Result ID: 69f4559747f34ce924d9ddf2.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B","Mode":"No-think"}. This record does not assert independent reproduction.

### SFT-only

Accuracy: 59.44. Result ID: 69f4559747f34ce924d9ddf4.

[Reporting paper](<https://www.sota2.com/research/paper/69f4336355a3c32af89fcf86.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c1c1e209131cea8c661.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B","Mode":"No-think"}. This record does not assert independent reproduction.

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