General knowledge evaluation on MMLU Pro
68.41Utility ScoreICL
Evaluation Results
| Method | Links | |
|---|---|---|
| ICLBackbone=Qwen3-4B2026.06 | 68.41 | |
| PolicyAlignBackbone=Qwen3-4B2026.06 | 68.29 | |
| BaseBackbone=Qwen3-4B2026.06 | 66.29 | |
| SFTBackbone=Qwen3-4B2026.06 | 65.88 |
| Method | Links | |
|---|---|---|
| ICLBackbone=Qwen3-4B2026.06 | 68.41 | |
| PolicyAlignBackbone=Qwen3-4B2026.06 | 68.29 | |
| BaseBackbone=Qwen3-4B2026.06 | 66.29 | |
| SFTBackbone=Qwen3-4B2026.06 | 65.88 |