Personalization on PRISM
81.62Personalization Win RateMIPO
Evaluation Results
| Method | Links | |
|---|---|---|
| MIPOBase Model=Llama-3.2-3B-Instruct2026.03 | 81.62 | |
| INFONCEBase Model=Llama-3.2-1B-Instruct2026.03 | 80.93 | |
| MIPOBase Model=Llama-3.2-1B-Instruct2026.03 | 80.93 | |
| RLAIF (MI)Base Model=Llama-3.2-3B-Instruct2026.03 | 80.58 | |
| INFONCEBase Model=Llama-3.2-3B-Instruct2026.03 | 80.58 | |
| RLAIFBase Model=Llama-3.2-3B-Instruct2026.03 | 80.07 | |
| SFTBase Model=Llama-3.2-3B-Instruct2026.03 | 78.52 | |
| SFT (Revision)Base Model=Llama-3.2-3B-Instruct2026.03 | 78.35 | |
| Personalized PromptingBase Model=Llama-3.2-3B-Instruct2026.03 | 76.8 | |
| SFT (Revision)Base Model=Llama-3.2-1B-Instruct2026.03 | 73.37 | |
| PPO (MI)Base Model=Llama-3.2-1B-Instruct2026.03 | 73.37 | |
| PPO (MI)Base Model=Llama-3.2-3B-Instruct2026.03 | 73.02 | |
| Personalized PromptingBase Model=Llama-3.2-1B-Instruct2026.03 | 72.17 | |
| SFTBase Model=Llama-3.2-1B-Instruct2026.03 | 69.76 | |
| RLAIF (MI)Base Model=Llama-3.2-1B-Instruct2026.03 | 69.07 | |
| RLAIF (MI)Base Model=Qwen2.5-7B-Instruct2026.03 | 64.43 | |
| RLAIFBase Model=Qwen2.5-7B-Instruct2026.03 | 63.4 | |
| MIPOBase Model=Qwen2.5-3B-Instruct2026.03 | 59.79 | |
| SFT (Revision)Base Model=Qwen2.5-7B-Instruct2026.03 | 59.28 | |
| MIPOBase Model=Qwen2.5-7B-Instruct2026.03 | 58.76 | |
| Personalized PromptingBase Model=Qwen2.5-7B-Instruct2026.03 | 58.25 | |
| INFONCEBase Model=Qwen2.5-7B-Instruct2026.03 | 58.25 | |
| INFONCEBase Model=Qwen2.5-1.5B-Instruct2026.03 | 57.87 | |
| RLAIFBase Model=Llama-3.2-1B-Instruct2026.03 | 56.87 | |
| SFTBase Model=Qwen2.5-7B-Instruct2026.03 | 56.7 | |
| MIPOBase Model=Qwen2.5-1.5B-Instruct2026.03 | 56.53 | |
| PPO (MI)Base Model=Qwen2.5-7B-Instruct2026.03 | 54.12 | |
| SFT (Revision)Base Model=Qwen2.5-3B-Instruct2026.03 | 53.09 | |
| INFONCEBase Model=Qwen2.5-3B-Instruct2026.03 | 51.89 | |
| RLAIF (MI)Base Model=Qwen2.5-3B-Instruct2026.03 | 51.72 | |
| PPO (MI)Base Model=Qwen2.5-3B-Instruct2026.03 | 50.52 | |
| Personalized PromptingBase Model=Qwen2.5-3B-Instruct2026.03 | 49.49 | |
| RLAIFBase Model=Qwen2.5-3B-Instruct2026.03 | 48.63 | |
| SFT (Revision)Base Model=Qwen2.5-1.5B-Instruct2026.03 | 46.56 | |
| SFTBase Model=Qwen2.5-3B-Instruct2026.03 | 44.67 | |
| Llama-3.2-3B-InstructBase Model=Llama-3.2-3B-Instruct2026.03 | 40.72 | |
| Personalized PromptingBase Model=Qwen2.5-1.5B-Instruct2026.03 | 39.18 | |
| Llama-3.2-1B-InstructBase Model=Llama-3.2-1B-Instruct2026.03 | 36.6 | |
| SFTBase Model=Qwen2.5-1.5B-Instruct2026.03 | 35.05 | |
| PPO (MI)Base Model=Qwen2.5-1.5B-Instruct2026.03 | 32.47 | |
| Qwen2.5-7B-InstructBase Model=Qwen2.5-7B-Instruct2026.03 | 20.62 | |
| RLAIF (MI)Base Model=Qwen2.5-1.5B-Instruct2026.03 | 16.83 | |
| RLAIFBase Model=Qwen2.5-1.5B-Instruct2026.03 | 13.92 | |
| Qwen2.5-3B-InstructBase Model=Qwen2.5-3B-Instruct2026.03 | 11.86 | |
| Qwen2.5-1.5B-InstructBase Model=Qwen2.5-1.5B-Instruct2026.03 | 8.25 |