List your company

Active Testing on MMLU-Redux

144Labeling BudgetActive Testing via Approximate Neyman Allocation
141.76156.88172187.12May 11, 2026

Evaluation results

MethodLinks
Active Testing via Approximate Neyman AllocationSurrogate Model=Qwen-3.5 4B, Target Model=Qwen-3.5 9B2026.0514428
Uniform SamplingSurrogate Model=Qwen-3.5 4B, Target Model=Qwen-3.5 9B2026.05200—