General Evaluation on PIQA, HellaSwag, WinoGrande, ARC-e, and ARC-c
69Average ScoreBaseline
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BaselineModels=LLaMA-3.1-8b, CR=-, Quantization=false2026.02 | 69 | 0 | |
| ROCKETModels=LLaMA-3.1-8b, CR=0.4, Quantization=true2026.02 | 65 | 5.79 | |
| Dobi-SVDModels=LLaMA-3.1-8b, CR=0.4, Quantization=true2026.02 | 63 | 8.7 | |
| ROCKETModels=LLaMA-3.1-8b, CR=0.6, Quantization=true2026.02 | 60 | 13 | |
| ROCKETModels=LLaMA-3.1-8b, CR=0.4, Quantization=false2026.02 | 56 | 18.84 | |
| Dobi-SVDModels=LLaMA-3.1-8b, CR=0.6, Quantization=true2026.02 | 52 | 24.6 | |
| LLM-PrunerModels=LLaMA-3.1-8b, CR=0.4, Quantization=false2026.02 | 46 | 33.3 | |
| SliceGPTModels=LLaMA-3.1-8b, CR=0.4, Quantization=false2026.02 | 46 | 33.3 | |
| BonsaiModels=LLaMA-3.1-8b, CR=0.4, Quantization=false2026.02 | 41 | 40.6 | |
| Wanda-spModels=LLaMA-3.1-8b, CR=0.4, Quantization=false2026.02 | 39 | 43.5 |