List your company

Multitask Language Understanding on GMMLU c

30.75Acc (Normalized)HQ seed (LLM Training)
24.2525.937527.62529.3125Apr 22, 2026

Evaluation results

MethodLinks
HQ seed (LLM Training)Filtering Strategy=HQ seed (LLM Training)2026.0430.75
NordicFiltering Strategy=Nordic2026.0430.75
ML (Q3)Filtering Strategy=ML (Q3)2026.0430.75
Q3 negativesFiltering Strategy=Q3 negatives2026.0430.25
HQFiltering Strategy=HQ2026.0429.25
HQ seedFiltering Strategy=HQ seed2026.0429
MLFiltering Strategy=ML2026.0429
HQ allFiltering Strategy=HQ all2026.0428.75
ML (15%)Filtering Strategy=ML (15%)2026.0428.75
HQ Romance (No Fra)Filtering Strategy=HQ Romance (No Fra)2026.0427.5
ML (10%)Retention=10%2026.0427.25
ML2026.0427
MKC-eFiltering Strategy=MKC-e2026.0426.5
No FilteringFiltering Strategy=No Filtering2026.0426.25
HQ2026.0425.75
ML (Q3)Sampling Strategy=Q32026.0425.75
MKC-e2026.0425.5
HQ (20%)Retention=20%2026.0425.5
ML (20%)Retention=20%2026.0425.25
No Filtering2026.0424.75
HQ (10%)Retention=10%2026.0424.75
HQ seedVariant=Seed2026.0424.5