Pruning on Qwen3-32B
1.33Pruning Time (hours)SparseGPT
Evaluation Results
| Method | Links | |
|---|---|---|
| SparseGPTHardware=single B200 GPU2026.06 | 1.33 | |
| SpenseGPTHardware=single B200 GPU2026.06 | 1.33 | |
| SpenseGPT+Hardware=single B200 GPU2026.06 | 1.87 | |
| HyperPruneHardware=single B200 GPU2026.06 | 3.8 |