Model Utility on Alpaca-Llama 3 Utility Benchmarks
47.36ARC-c ScoreDualOptim
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DualOptimquantization=full2026.05 | 47.36 | 50.13 | 33.58 | 14.02 | 12.99 | 31.62 | |
| DualOptim+quantization=full2026.05 | 47.27 | 51.89 | 32.25 | 15.39 | 14.23 | 32.81 | |
| DualOptimquantization=8-bit2026.05 | 47.1 | 50.78 | 33.58 | 13.82 | 12.96 | 31.65 | |
| Joint2026.05 | 47.04 | 51.63 | 33.74 | 12.18 | 14.1 | 31.74 | |
| DualOptim+quantization=8-bit2026.05 | 46.93 | 51.66 | 34.47 | 13.71 | 14.73 | 32.3 | |
| Alternate2026.05 | 46.81 | 50.83 | 34.6 | 13.83 | 12.94 | 31.8 | |
| Initialstatus=baseline2026.05 | 45.56 | 52.53 | 29.74 | 12.11 | 13.12 | 30.61 |