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

Common-sense Reasoning on lm-eval-harness 11 tasks

63.430-shot Accuracy

MoE Transformer w/ dynamic conv. (head-wise)

45.07449.839554.60559.3705Jun 2, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
63.43
2026.06
63.42
2026.06
62.97
2026.06
62.46
2026.06
60.7
2026.06
59.7
2026.06
58.94
2026.06
58.46
2026.06
58.35
2026.06
58.23
2026.06
49.22
2026.06
48.9
2026.06
48.81
2026.06
47.61
2026.06
47.34
2026.06
47.27
2026.06
47.26
2026.06
47.24
2026.06
46.93
2026.06
46.86
2026.06
46.76
2026.06
46.64
2026.06
46.41
2026.06
45.78