Controllable Generation on Code-UltraFeedback
90.7DiversityBase
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BaseBackbone=Phi-4-mini, Target Score=2,3,2,2,32025.10 | 90.7 | 2.5 | — | |
| PromptingBackbone=Phi-4-mini, Target Score=2,3,2,2,32025.10 | 90.6 | 2.51 | 72 | |
| PromptingBackbone=Phi-4-mini, Target Score=3,3,3,3,32025.10 | 90.3 | 1.47 | 946 | |
| BaseBackbone=Phi-4-mini, Target Score=3,3,3,3,32025.10 | 90.2 | 1.57 | — | |
| Re-ControlBackbone=LLaMA-3.2-3B, Target Score=3,3,3,3,32025.10 | 88 | 2.21 | 7.54 | |
| PromptingBackbone=LLaMA-3.2-3B, Target Score=3,3,3,3,32025.10 | 87.9 | 2.21 | 6.8 | |
| BaseBackbone=LLaMA-3.2-3B, Target Score=3,3,3,3,32025.10 | 87.6 | 2.29 | — | |
| BaseBackbone=LLaMA-3.2-3B, Target Score=2,2,2,2,22025.10 | 87.4 | 2.95 | — | |
| PromptingBackbone=Phi-4-mini, Target Score=2,2,2,2,22025.10 | 86.9 | 3.64 | 215 | |
| BaseBackbone=Phi-4-mini, Target Score=2,2,2,2,22025.10 | 86.8 | 3.65 | — | |
| PromptingBackbone=LLaMA-3.2-3B, Target Score=2,2,2,2,22025.10 | 86.5 | 2.85 | 6.06 | |
| Re-ControlBackbone=LLaMA-3.2-3B, Target Score=2,2,2,2,22025.10 | 85.6 | 2.78 | 6.57 | |
| BaseBackbone=LLaMA-3.2-3B, Target Score=2,3,3,2,32025.10 | 83.8 | 2.24 | — | |
| PromptingBackbone=LLaMA-3.2-3B, Target Score=2,3,3,2,32025.10 | 83.8 | 2.23 | 1.85 | |
| Re-ControlBackbone=LLaMA-3.2-3B, Target Score=2,3,3,2,32025.10 | 83.1 | 2.18 | 2.06 | |
| ITIBackbone=Phi-4-mini, Target Score=3,3,3,3,32025.10 | 78.9 | 1.55 | 1,649 | |
| Re-ControlBackbone=Phi-4-mini, Target Score=3,3,3,3,32025.10 | 78.6 | 1.43 | 1,725 | |
| MAT-SteerBackbone=LLaMA-3.2-3B, Target Score=3,3,3,3,32025.10 | 77.8 | 2.41 | 13.63 | |
| OursBackbone=Phi-4-mini, Target Score=3,3,3,3,32025.10 | 75.5 | 1.33 | 2,412 | |
| ITIBackbone=LLaMA-3.2-3B, Target Score=3,3,3,3,32025.10 | 74.1 | 2.62 | 12.72 | |
| MAT-SteerBackbone=Phi-4-mini, Target Score=3,3,3,3,32025.10 | 70 | 1.43 | 1,892 | |
| ITIBackbone=LLaMA-3.2-3B, Target Score=2,3,3,2,32025.10 | 67 | 2.33 | 1.82 | |
| ITIBackbone=Phi-4-mini, Target Score=2,3,2,2,32025.10 | 64.7 | 2.5 | 133 | |
| ITIBackbone=Phi-4-mini, Target Score=2,2,2,2,22025.10 | 62.3 | 3.66 | 454 | |
| Re-ControlBackbone=Phi-4-mini, Target Score=2,2,2,2,22025.10 | 61.4 | 3.53 | 692 | |
| PRE-CONTROLBackbone=LLaMA-3.2-3B, Target Score=3,3,3,3,32025.10 | 61.4 | 2.08 | 17.46 | |
| MAT-SteerBackbone=LLaMA-3.2-3B, Target Score=2,3,3,2,32025.10 | 58.7 | 2.38 | 1.64 | |
| MAT-SteerBackbone=Phi-4-mini, Target Score=2,3,2,2,32025.10 | 58.6 | 2.49 | 189 | |
| Re-ControlBackbone=Phi-4-mini, Target Score=2,3,2,2,32025.10 | 57 | 2.48 | 246 | |
| PRE-CONTROLBackbone=LLaMA-3.2-3B, Target Score=2,3,3,2,32025.10 | 51.2 | 2.17 | 2.77 | |
| MAT-SteerBackbone=LLaMA-3.2-3B, Target Score=2,2,2,2,22025.10 | 48 | 2.59 | 16.67 | |
| OursBackbone=Phi-4-mini, Target Score=2,3,2,2,32025.10 | 45.4 | 2.42 | 266 | |
| ITIBackbone=LLaMA-3.2-3B, Target Score=2,2,2,2,22025.10 | 44.1 | 2.83 | 6.79 | |
| PRE-CONTROLBackbone=LLaMA-3.2-3B, Target Score=2,2,2,2,22025.10 | 44 | 1.95 | 30.68 | |
| OursBackbone=Phi-4-mini, Target Score=2,2,2,2,22025.10 | 32.2 | 2.58 | 2,483 | |
| MAT-SteerBackbone=Phi-4-mini, Target Score=2,2,2,2,22025.10 | 31.8 | 2.89 | 838 |