Common Sense Reasoning on HellaSwag topic-filtered
68.2AccuracyQwen3-4B
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen3-4B2026.06 | 68.2 | |
| Talkie-1930variant=base2026.06 | 51.2 | |
| GPT-2 XL2026.06 | 49.9 | |
| TYPEWRITERLMvariant=base2026.06 | 36.5 | |
| GPT-1900variant=base2026.06 | 35.1 |