ResearchBenchmarksLLM Pruning Evaluation on Code Dimension HumanEval MBPPLLM Pruning Evaluation on Code Dimension HumanEval MBPPFollow72.3S_Code Relative RetentionSparseGPT12.416827.963443.5159.0566Jun 2, 2026Evaluation resultsMethodMethodLinksS_Code Relative RetentionSparseGPTCalibration=mix_4_3_3_...Calibration=mix_4_3_3_code, Model Scale=LLaMA-3.1-70B, Sparsity=60% unstructured2026.0672.3WandaCalibration=mix_4_3_3_...Calibration=mix_4_3_3_code, Model Scale=LLaMA-3.1-8B, Sparsity=60% unstructured2026.0614.72