# Zero-shot Task Completion on HellaSwag 0-shot (N=500)

Source: [Zero-shot Task Completion on HellaSwag 0-shot (N=500)](<https://www.sota2.com/research/sota/zero-shot-task-completion-on-hellaswag-0-shot-n-500>)

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

Data updated: 2026-03-23T10:44:12.274Z

Evaluation of model performance on a specific 500-example subset of the HellaSwag dataset under zero-shot task completion settings, 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 | Zero-shot Task Completion |
| Dataset (including any stated split) | HellaSwag 0-shot (N=500) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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/zero-shot-task-completion-on-hellaswag-0-shot-n-500.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Full cache | 47.5 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| KV-Direct | 47.5 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Window-64 | 47.5 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Full cache | 44 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| KV-Direct | 44 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Window-64 | 44 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Full cache | 42 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| KV-Direct | 42 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Window-64 | 42 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Full cache | 41 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| KV-Direct | 41 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Window-64 | 41 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Full cache | 39.4 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| KV-Direct | 39.4 | Not recorded | 69c0c8d90636e5bf4afc3967 |
| Window-64 | 39.4 | Not recorded | 69c0c8d90636e5bf4afc3967 |

### Full cache

Accuracy: 47.5. Result ID: 69c1192e47f34ce924a94836.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49db1.md>)

[Implementation](<https://github.com/tensorflow/tensor2tensor>)

Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B"}. This record does not assert independent reproduction.

### KV-Direct

Accuracy: 47.5. Result ID: 69c1192e47f34ce924a94837.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B"}. This record does not assert independent reproduction.

### Window-64

Accuracy: 47.5. Result ID: 69c1192e47f34ce924a94838.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B"}. This record does not assert independent reproduction.

### Full cache

Accuracy: 44. Result ID: 69c1192e47f34ce924a94830.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49db1.md>)

[Implementation](<https://github.com/tensorflow/tensor2tensor>)

Recorded evaluation configuration: {"Model":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### KV-Direct

Accuracy: 44. Result ID: 69c1192e47f34ce924a94831.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### Window-64

Accuracy: 44. Result ID: 69c1192e47f34ce924a94832.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### Full cache

Accuracy: 42. Result ID: 69c1192e47f34ce924a94833.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49db1.md>)

[Implementation](<https://github.com/tensorflow/tensor2tensor>)

Recorded evaluation configuration: {"Model":"DS-R1-1.5B"}. This record does not assert independent reproduction.

### KV-Direct

Accuracy: 42. Result ID: 69c1192e47f34ce924a94834.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"DS-R1-1.5B"}. This record does not assert independent reproduction.

### Window-64

Accuracy: 42. Result ID: 69c1192e47f34ce924a94835.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"DS-R1-1.5B"}. This record does not assert independent reproduction.

### Full cache

Accuracy: 41. Result ID: 69c1192e47f34ce924a9482d.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49db1.md>)

[Implementation](<https://github.com/tensorflow/tensor2tensor>)

Recorded evaluation configuration: {"Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### KV-Direct

Accuracy: 41. Result ID: 69c1192e47f34ce924a9482e.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### Window-64

Accuracy: 41. Result ID: 69c1192e47f34ce924a9482f.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### Full cache

Accuracy: 39.4. Result ID: 69c1192e47f34ce924a9482a.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49db1.md>)

[Implementation](<https://github.com/tensorflow/tensor2tensor>)

Recorded evaluation configuration: {"Model":"SmolLM2-135M"}. This record does not assert independent reproduction.

### KV-Direct

Accuracy: 39.4. Result ID: 69c1192e47f34ce924a9482b.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"SmolLM2-135M"}. This record does not assert independent reproduction.

### Window-64

Accuracy: 39.4. Result ID: 69c1192e47f34ce924a9482c.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8d90636e5bf4afc3967.md>)

Recorded evaluation configuration: {"Model":"SmolLM2-135M"}. This record does not assert independent reproduction.

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