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

Multi-task Language Evaluation on 16-task evaluation suite 1B/25B checkpoints (test)

42.22Average Score

HERMES + L12 quality top-30%

37.810438.955240.141.2448Jul 2, 2026
Updated 23d ago

Evaluation Results

MethodLinks
2026.07
42.221.628
2026.07
41.551.048
2026.07
41.531.059
2026.07
41.340.949
2026.07
40.820.508
2026.07
40.790.469
2026.07
40.610.317
2026.07
40.450.114
2026.07
40.290.128
2026.07
40.280.048
2026.07
39.940.312
2026.07
39.920.362
2026.07
39.880.294
2026.07
39.860.302
2026.07
39.690.458
2026.07
39.340.747
2026.07
38.891.147
2026.07
38.871.163
2026.07
38.861.181
2026.07
37.981.885