Quantization Optimization on LLaMA
0.1Optimization Time (hours)BWLA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BWLAModel=LLaMA2-7B2026.05 | 0.1 | 3 | |
| BWLAModel=LLaMA3-8B2026.05 | 0.12 | 3.3 | |
| BWLAModel=LLaMA2-13B2026.05 | 0.25 | 3.2 | |
| OSTQuantModel=LLaMA2-7B2026.05 | 0.3 | — | |
| OSTQuantModel=LLaMA3-8B2026.05 | 0.4 | — | |
| OSTQuantModel=LLaMA2-13B2026.05 | 0.8 | — | |
| BWLAModel=LLaMA2-70B2026.05 | 1.4 | 3.9 | |
| OmniquantModel=LLaMA2-7B2026.05 | 1.6 | — | |
| OmniquantModel=LLaMA3-8B2026.05 | 1.8 | — | |
| OmniquantModel=LLaMA2-13B2026.05 | 3.3 | — | |
| OSTQuantModel=LLaMA2-70B2026.05 | 5.5 | — | |
| OmniquantModel=LLaMA2-70B2026.05 | 9.5 | — |