List your company

Language Modeling on OpenWebText (OWT) (PPL)

11.2PPLOriginal Model
11.1311.602512.07512.5475Jun 26, 2026

Evaluation results

MethodLinks
Original ModelBackbone=Qwen-3-4B, Avg. Sparsity (L/H)=0% / 0%, Mem (GB)=5.80, Time (s)=41002026.0611.2
Learning to AllocateBackbone=Qwen-3-4B, Avg. Sparsity (L/H)=38% / 28%, Mem (GB)=5.30, Time (s)=28502026.0611.32
Oracle Static PruningBackbone=Qwen-3-4B, Avg. Sparsity (L/H)=33% / 22%, Mem (GB)=4.95, Time (s)=26802026.0611.35
FlexiDepthBackbone=Qwen-3-4B, Avg. Sparsity (L/H)=34% / –, Mem (GB)=5.75, Time (s)=29102026.0611.38
AdaSkipBackbone=Qwen-3-4B, Avg. Sparsity (L/H)=36% / –, Mem (GB)=5.65, Time (s)=28002026.0611.42
Static PruningBackbone=Qwen-3-4B, Avg. Sparsity (L/H)=33% / 0%, Mem (GB)=5.10, Time (s)=27502026.0612.95