# Commonsense Reasoning on Winogrande first 1000 examples (standard)

Source: [Commonsense Reasoning on Winogrande first 1000 examples (standard)](<https://www.sota2.com/research/sota/commonsense-reasoning-on-winogrande-first-1000-examples-standard>)

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

Data updated: 2026-04-21T12:46:24.325Z

Evaluation of model performance on the first 1000 standard examples of the Winogrande dataset for Commonsense Reasoning, measured by 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 | Commonsense Reasoning |
| Dataset (including any stated split) | Winogrande first 1000 examples (standard) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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/commonsense-reasoning-on-winogrande-first-1000-examples-standard.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| k=64 + sink | 48.9 | Not recorded | 69e704df7d0172140ec7f0a5 |
| k=0 (baseline) | 48.8 | Not recorded | 69e704df7d0172140ec7f0a5 |
| k=64 (DR only) | 48.8 | Not recorded | 69e704df7d0172140ec7f0a5 |

### k=64 + sink

Accuracy: 48.9. Result ID: 69e7706247f34ce924cf1f65.

[Reporting paper](<https://www.sota2.com/research/paper/69e704df7d0172140ec7f0a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704df7d0172140ec7f0a5.md>)

Recorded evaluation configuration: {"Model size":"300M","Tokens":"5B","Precision":"FP16","Decoding":"greedy","Mode":"zero-shot","Examples":"1000","Depth Registers":"k=64","Attention-sink loss":"included"}. This record does not assert independent reproduction.

### k=0 (baseline)

Accuracy: 48.8. Result ID: 69e7706247f34ce924cf1f63.

[Reporting paper](<https://www.sota2.com/research/paper/69e704df7d0172140ec7f0a5.md>)

Recorded evaluation configuration: {"Model size":"300M","Tokens":"5B","Precision":"FP16","Decoding":"greedy","Mode":"zero-shot","Examples":"1000"}. This record does not assert independent reproduction.

### k=64 (DR only)

Accuracy: 48.8. Result ID: 69e7706247f34ce924cf1f64.

[Reporting paper](<https://www.sota2.com/research/paper/69e704df7d0172140ec7f0a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704df7d0172140ec7f0a5.md>)

Recorded evaluation configuration: {"Model size":"300M","Tokens":"5B","Precision":"FP16","Decoding":"greedy","Mode":"zero-shot","Examples":"1000","Depth Registers":"k=64"}. This record does not assert independent reproduction.

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