Share your thoughts, 1 month free Claude Pro on usSee more

Multitask Language Understanding on GMMLU c

30.75Acc (Normalized)

HQ seed (LLM Training)

Updated 3mo ago

Evaluation Results

Method	Links
HQ seed (LLM Training) 2026.04		30.75
Nordic 2026.04		30.75
ML (Q3) 2026.04		30.75
Q3 negatives 2026.04		30.25
HQ 2026.04		29.25
HQ seed 2026.04		29
ML 2026.04		29
HQ all 2026.04		28.75
ML (15%) 2026.04		28.75
HQ Romance (No Fra) 2026.04		27.5
ML (10%) 2026.04		27.25
ML 2026.04		27
MKC-e 2026.04		26.5
No Filtering 2026.04		26.25
HQ 2026.04		25.75
ML (Q3) 2026.04		25.75
MKC-e 2026.04		25.5
HQ (20%) 2026.04		25.5
ML (20%) 2026.04		25.25
No Filtering 2026.04		24.75
HQ (10%) 2026.04		24.75
HQ seed 2026.04		24.5