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

Commonsense Reasoning on Winogrande (Accuracy, Delta Accuracy)

88.8Accuracy

LoRA

47.61658.3086979.692Feb 28, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
88.82.4
2026.02
87.20.8
2026.02
86.40
2026.02
84.61.6
2026.02
830
2026.02
82.1-4.5
2026.02
80.6-2.4
2026.02
64.9-18.1
2026.02
60.11.9
2026.02
58.20
2026.02
57-1.2
2026.02
51.60.2
2026.02
51.40
2026.02
50.8-0.6
2026.02
49.3-8.9
2026.02
49.2-2.2