Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Zero-shot Commonsense Reasoning on ARC-e, ARC-c, HellaSwag, PIQA, WinoGrande (test)

74.5ARC-e Accuracy

BF16

34.56444.93255.365.668Jan 28, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.01
74.542.255.276.869.163.6
2026.01
74.139.349.175.762.260.1
2026.01
70.234.646.473.962.757.6
2026.01
69.235.146.273.458.656.5
2026.01
68.734.842.372.755.854.9
2026.01
67.232.943.973.558.255.1
2026.01
66.53541.872.657.754.7
2026.01
65.431.347.774.260.355.8
2026.01
64.530.539.17154.251.9
2026.01
61.926.235.469.252.749.1
2026.01
60.325.93668.354.148.9
2026.01
49.823.130.364.552.944.1
2026.01
43.12029.459.952.240.9
2026.01
42.417.425.656.351.338.6
2026.01
42.11827.360.45139.8
2026.01
40.72228.460.149.240.1
2026.01
37.617.725.857.450.637.8
2026.01
36.116.12657.249.637