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

Robustness Evaluation on HellaSwag SAGE-generated

74.17Overall Accuracy (OA)

GPT-4o

52.8558.38563.9269.455May 12, 2026
Updated 21d ago

Evaluation Results

MethodLinks
2026.05
74.1765.6811.4450.67
2026.05
71.7937.634.229.26
2026.05
71.5435.6835.8627.93
2026.05
66.2136.7329.4826.86
2026.05
66.1734.4131.7625.33
2026.05
64.7134.1130.624.79
2026.05
64.533.7330.7724.33
2026.05
60.7933.3927.423.15
2026.05
60.534.7325.7723.83
2026.05
59.5432.8426.722.57
2026.05
54.1232.521.6220.57
2026.05
53.6731.8321.8320.29