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

Physical Reasoning on PIQA (Accuracy, Accuracy Drop)

82.21Accuracy

BF16

47.349256.399665.4574.5004Feb 11, 2026
Updated 3d ago

Evaluation Results

MethodLinks
2026.02
82.21-
2026.02
81.91-0.3
2026.02
81.88-
2026.02
81.72-0.16
2026.02
81.67-0.21
2026.02
81.66-0.22
2026.02
81.29-0.59
2026.02
81.01-1.2
2026.02
80.85-
2026.02
80.85-1.36
2026.02
79.76-1.09
2026.02
79.6-1.25
2026.02
79.11-
2026.02
79.09-1.76
2026.02
78.56-2.29
2026.02
78.4-0.71
2026.02
77.91-1.2
2026.02
77.86-1.25
2026.02
77.37-1.74
2026.02
48.69-33.52