List your company

Mathematical Reasoning on Capability Evaluation Suite Math Pool

69.12S_Math Scoremix_balanced
63.97265.308566.64567.9815Jun 2, 2026

Evaluation results

MethodLinks
mix_balancedSparsity=40%, Pruning Method=Wanda, Base Model=LLaMA-3.1-8B2026.0669.12
SGS-styleSparsity=40%, Pruning Method=Wanda, Base Model=LLaMA-3.1-8B2026.0667.5
Self-Gen (from C4)Sparsity=40%, Pruning Method=Wanda, Base Model=LLaMA-3.1-8B2026.0666.2
C4 (random)Sparsity=40%, Pruning Method=Wanda, Base Model=LLaMA-3.1-8B2026.0664.17