List your company

Active Testing on MMLU-Pro

266Labeling BudgetActive Testing via Approximate Neyman Allocation
263.04283.02303322.98May 11, 2026

Evaluation results

MethodLinks
Active Testing via Approximate Neyman AllocationSurrogate Model=Qwen-3.5 2B, Target Model=Qwen-3.5 4B2026.0526621.8
Uniform SamplingSurrogate Model=Qwen-3.5 2B, Target Model=Qwen-3.5 4B2026.05340—