# General Language Understanding on AI2 OLMES (test)

Source: [General Language Understanding on AI2 OLMES (test)](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test>)

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

Data updated: 2026-05-12T11:21:38.249Z

Evaluation of general language understanding capabilities on the test split of the AI2 OLMES benchmark, reporting scores for ARC-E, ARC-C, BoolQ, CSQA, HellaSwag, OBQA, PIQA, SIQA, and WinoG.

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 | General Language Understanding |
| Dataset (including any stated split) | AI2 OLMES (test) |
| Metric | ARC-E Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [ARC-E Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=ARC-E%20Accuracy>): Higher is better; unit: PERCENT.

- [ARC-C Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=ARC-C%20Accuracy>): Higher is better; unit: PERCENT.

- [BoolQ Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=BoolQ%20Accuracy>): Higher is better; unit: PERCENT.

- [CSQA Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=CSQA%20Accuracy>): Higher is better; unit: PERCENT.

- [HellaSwag Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=HellaSwag%20Accuracy>): Higher is better; unit: PERCENT.

- [OBQA Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=OBQA%20Accuracy>): Higher is better; unit: PERCENT.

- [PIQA Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=PIQA%20Accuracy>): Higher is better; unit: PERCENT.

- [SIQA Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=SIQA%20Accuracy>): Higher is better; unit: PERCENT.

- [WinoG Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=WinoG%20Accuracy>): Higher is better; unit: PERCENT.

- [Core 9 Average Accuracy](<https://www.sota2.com/research/sota/general-language-understanding-on-ai2-olmes-test.md?metric=Core%209%20Average%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | ARC-E Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Looped-MoE | 40.4 | Not recorded | 6a02b45edbc836b87d824a0c |
| Base | 39.3 | Not recorded | 6a02b45edbc836b87d824a0c |
| Looped | 39.1 | Not recorded | 6a02b45edbc836b87d824a0c |
| MoE | 38.9 | Not recorded | 6a02b45edbc836b87d824a0c |

### Looped-MoE

ARC-E Accuracy: 40.4. Result ID: 6a030d3347f34ce924e6d051.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b45edbc836b87d824a0c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b45edbc836b87d824a0c.md>)

Recorded evaluation configuration: {"Params":"216M","Training FLOPs":"10^18"}. This record does not assert independent reproduction.

### Base

ARC-E Accuracy: 39.3. Result ID: 6a030d3347f34ce924e6d050.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b45edbc836b87d824a0c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66eb1e209131cea132ee.md>)

Recorded evaluation configuration: {"Params":"246M","Training FLOPs":"10^18"}. This record does not assert independent reproduction.

### Looped

ARC-E Accuracy: 39.1. Result ID: 6a030d3347f34ce924e6d04f.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b45edbc836b87d824a0c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69df1bc77d0172140ec7bbd3.md>)

[Implementation](<https://github.com/sandyresearch/parcae>)

Recorded evaluation configuration: {"Params":"168M","Training FLOPs":"10^18"}. This record does not assert independent reproduction.

### MoE

ARC-E Accuracy: 38.9. Result ID: 6a030d3347f34ce924e6d052.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b45edbc836b87d824a0c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63171e209131ce9fc941.md>)

Recorded evaluation configuration: {"Params":"366M","Training FLOPs":"10^18"}. This record does not assert independent reproduction.

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