Commonsense Reasoning on HellaSwag Spanish
54.4Normalized AccuracyML (15%)
Evaluation Results
| Method | Links | |
|---|---|---|
| ML (15%)Filtering strategy=ML (15%)2026.04 | 54.4 | |
| ML (Q3)Filtering strategy=ML (Q3)2026.04 | 53.72 | |
| MLFiltering strategy=ML2026.04 | 53.1 | |
| HQFiltering strategy=HQ2026.04 | 52.63 | |
| HQ seedFiltering strategy=HQ seed2026.04 | 52.19 | |
| MKC-eFiltering strategy=MKC-e2026.04 | 50.75 | |
| No filteringFiltering strategy=No filtering2026.04 | 50.06 |