Instruction-following and safety evaluation on Norwegian pack
14.2Overall Delta ScoreBorealis
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| BorealisSize=4B, Judge=Qwen3.5-122B-A10B (Q4_K_S)2026.05 | 14.2 | -23.9 | 24 | -30 | 4.1 | -7.5 | 15.2 | -27 | 10.6 | -28.8 | |
| BorealisSize=12B, Judge=Qwen3.5-122B-A10B (Q4_K_S)2026.05 | 7 | -4.7 | 9.5 | -5 | -0.9 | 2.5 | 10.2 | -7 | 7.8 | -8.8 | |
| BorealisSize=27B, Judge=Qwen3.5-122B-A10B (Q4_K_S)2026.05 | 6 | -7.5 | 8.2 | -9 | 1.6 | 3.8 | 8.2 | -12 | 5 | -11.2 | |
| BorealisSize=1B, Judge=Qwen3.5-122B-A10B (Q4_K_S)2026.05 | 0.8 | -2.5 | 2.5 | -7 | -1.2 | 3.8 | -0.2 | 1 | 1.9 | -7.5 |