# Language Understanding on HellaSwag, PIQA, MMLU Suite

Source: [Language Understanding on HellaSwag, PIQA, MMLU Suite](<https://www.sota2.com/research/sota/language-understanding-on-hellaswag-piqa-mmlu-suite>)

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

Data updated: 2026-02-26T05:46:26.381Z

Aggregate evaluation of language understanding capabilities across the HellaSwag, PIQA, and MMLU benchmarks.

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 | Language Understanding |
| Dataset (including any stated split) | Language Understanding Suite HellaSwag, PIQA, MMLU |
| Metric | HellaSwag Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [HellaSwag Accuracy](<https://www.sota2.com/research/sota/language-understanding-on-hellaswag-piqa-mmlu-suite.md?metric=HellaSwag%20Accuracy>): Higher is better; unit: PERCENT.

- [PIQA Accuracy](<https://www.sota2.com/research/sota/language-understanding-on-hellaswag-piqa-mmlu-suite.md?metric=PIQA%20Accuracy>): Higher is better; unit: PERCENT.

- [MMLU Accuracy](<https://www.sota2.com/research/sota/language-understanding-on-hellaswag-piqa-mmlu-suite.md?metric=MMLU%20Accuracy>): Higher is better; unit: PERCENT.

- [Average Score](<https://www.sota2.com/research/sota/language-understanding-on-hellaswag-piqa-mmlu-suite.md?metric=Average%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | HellaSwag Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| OpenLLaMA (V1) | 49 | Not recorded | 68fa71a71e209131cea52555 |
| INCITE (V1) | 48 | Not recorded | 68fa71a71e209131cea52555 |
| MobileLLaMA | 48 | Not recorded | 68fa71a71e209131cea52555 |
| MobileLLaMA | 43 | Not recorded | 68fa71a71e209131cea52555 |
| OPT | 41 | Not recorded | 68fa71a71e209131cea52555 |
| TinyLLaMA (2T) | 40 | Not recorded | 68fa71a71e209131cea52555 |
| Pythia | 40 | Not recorded | 68fa71a71e209131cea52555 |
| Galactica | 34 | Not recorded | 68fa71a71e209131cea52555 |

### OpenLLaMA (V1)

HellaSwag Accuracy: 49. Result ID: 6991f3df3a6c3148490ba920.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

Recorded evaluation configuration: {"Parameters":"3B"}. This record does not assert independent reproduction.

### INCITE (V1)

HellaSwag Accuracy: 48. Result ID: 6991f3df3a6c3148490ba91f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

Recorded evaluation configuration: {"Parameters":"3B"}. This record does not assert independent reproduction.

### MobileLLaMA

HellaSwag Accuracy: 48. Result ID: 6991f3df3a6c3148490ba921.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Implementation](<https://github.com/Meituan-AutoML/MobileVLM>)

Recorded evaluation configuration: {"Parameters":"2.7B"}. This record does not assert independent reproduction.

### MobileLLaMA

HellaSwag Accuracy: 43. Result ID: 6991f3df3a6c3148490ba926.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Implementation](<https://github.com/Meituan-AutoML/MobileVLM>)

Recorded evaluation configuration: {"Parameters":"1.4B"}. This record does not assert independent reproduction.

### OPT

HellaSwag Accuracy: 41. Result ID: 6991f3df3a6c3148490ba924.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cda1e209131cea36474.md>)

[Implementation](<https://github.com/facebookresearch/metaseq>)

Recorded evaluation configuration: {"Parameters":"1.3B"}. This record does not assert independent reproduction.

### TinyLLaMA (2T)

HellaSwag Accuracy: 40. Result ID: 6991f3df3a6c3148490ba922.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

Recorded evaluation configuration: {"Parameters":"1.1B"}. This record does not assert independent reproduction.

### Pythia

HellaSwag Accuracy: 40. Result ID: 6991f3df3a6c3148490ba925.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f971e209131cea467b6.md>)

[Implementation](<https://github.com/EleutherAI/pythia>)

Recorded evaluation configuration: {"Parameters":"1.4B"}. This record does not assert independent reproduction.

### Galactica

HellaSwag Accuracy: 34. Result ID: 6991f3df3a6c3148490ba923.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71a71e209131cea52555.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dff1e209131cea3d479.md>)

[Implementation](<https://www.sota2.com/galactica.org>)

Recorded evaluation configuration: {"Parameters":"1.3B"}. This record does not assert independent reproduction.

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