List your company
ResearchFormerly WizWand

Corpus Composition Inference on LLMScan Fine-Grained 1.0 (test)

30.37Overlap AccuracyLLMSurgeon
20.687623.201325.71528.2287May 28, 2026
Updated 5mo ago

Evaluation results

MethodLinks
LLMSurgeonTarget Model=StarCoder-15.5B2026.0530.37
GradNormTarget Model=StarCoder-15.5B2026.0527.54
DC-PDDTarget Model=StarCoder-15.5B2026.0525.98
RecallTarget Model=StarCoder-15.5B2026.0525.91
LossTarget Model=StarCoder-15.5B2026.0525.76
Joint-LogitTarget Model=StarCoder-15.5B2026.0525.6
NeighborTarget Model=StarCoder-15.5B2026.0525.47
RefTarget Model=StarCoder-15.5B2026.0525.46
DUCITarget Model=StarCoder-15.5B2026.0525.44
Min-K%++Target Model=StarCoder-15.5B2026.0523.57
Min-K%Target Model=StarCoder-15.5B2026.0523.41
ZlibTarget Model=StarCoder-15.5B2026.0521.06