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

Knowledge-Based QA on Global-MMLU Lite (5 Unseen Languages)

71.9MMLU Score

CAGrad

41.42849.33957.2565.161Jun 4, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
71.9
2026.06
71
2026.06
70.3
2026.06
70
2026.06
69.1
2026.06
66
2026.06
65.5
2026.06
64
2026.06
63.6
2026.06
62.3
2026.06
52.6
2026.06
52.1
2026.06
51.7
2026.06
50.6
2026.06
50.5
2026.06
50.5
2026.06
49
2026.06
48.9
2026.06
48.9
2026.06
42.6