List your company

Mathematical Reasoning on NuminaMath (Accuracy, Delta Metrics)

62.45Math AccuracyCKA
3.762818.998934.23549.4711Mar 17, 2026

Evaluation results

MethodLinks
CKAPruning Budget=10%, Fine-tuning Protocol=without fine-tuning2026.0362.45-11.04-15
RandomPruning Budget=10%, Fine-tuning Protocol=without fine-tuning2026.0356.83-16.66-22.7
Ours-NonMathPruning Budget=10%, Fine-tuning Protocol=without fine-tuning2026.0355.5-17.99-24.5
Ours-MixedPruning Budget=10%, Fine-tuning Protocol=without fine-tuning2026.0354.42-19.07-25.9
Ours-MathPruning Budget=10%, Fine-tuning Protocol=without fine-tuning2026.0352.41-21.08-28.7
InterlacePruning Budget=10%, Fine-tuning Protocol=without fine-tuning2026.0347.99-25.5-34.7
Ours-NonMathPruning Budget=25%, Fine-tuning Protocol=without fine-tuning2026.0328.92-44.57-60.6
Ours-MixedPruning Budget=25%, Fine-tuning Protocol=without fine-tuning2026.0328.11-45.38-61.7
Ours-MathPruning Budget=25%, Fine-tuning Protocol=without fine-tuning2026.0325.7-47.79-65
InterlacePruning Budget=25%, Fine-tuning Protocol=without fine-tuning2026.0322.69-50.8-69.1
RandomPruning Budget=25%, Fine-tuning Protocol=without fine-tuning2026.0322.29-51.2-69.7
CKAPruning Budget=25%, Fine-tuning Protocol=without fine-tuning2026.0321.89-51.6-70.2
InterlacePruning Budget=40%, Fine-tuning Protocol=without fine-tuning2026.0319.48-54.01-73.5
Ours-NonMathPruning Budget=40%, Fine-tuning Protocol=without fine-tuning2026.0313.65-59.84-81.4
Ours-MathPruning Budget=40%, Fine-tuning Protocol=without fine-tuning2026.0310.44-63.05-85.8
Ours-MixedPruning Budget=40%, Fine-tuning Protocol=without fine-tuning2026.0310.04-63.45-86.3
RandomPruning Budget=40%, Fine-tuning Protocol=without fine-tuning2026.038.23-65.26-88.8
CKAPruning Budget=40%, Fine-tuning Protocol=without fine-tuning2026.036.02-67.47-91.8