List your company

Personalization on Community Alignment (CA)

93.67Personalization Win-RateMIPO
22.773241.179159.58577.9909Mar 10, 2026

Evaluation results

MethodLinks
MIPOBase Model=Llama-3.2-1B-Instruct2026.0393.67
MIPOBase Model=Llama-3.2-3B-Instruct2026.0390.33
RLAIFBase Model=Llama-3.2-3B-Instruct2026.0390.17
INFONCEBase Model=Llama-3.2-1B-Instruct2026.0389.5
RLAIF (MI)Base Model=Qwen2.5-7B-Instruct2026.0386
SFTBase Model=Llama-3.2-3B-Instruct2026.0384.17
SFTBase Model=Llama-3.2-1B-Instruct2026.0383.83
INFONCEBase Model=Llama-3.2-3B-Instruct2026.0383.67
INFONCEBase Model=Qwen2.5-7B-Instruct2026.0382.5
SFT (Revision)Base Model=Llama-3.2-3B-Instruct2026.0381.67
MIPOBase Model=Qwen2.5-7B-Instruct2026.0381.5
RLAIF (MI)Base Model=Llama-3.2-1B-Instruct2026.0381
RLAIFBase Model=Qwen2.5-7B-Instruct2026.0380.5
MIPOBase Model=Qwen2.5-3B-Instruct2026.0380.33
PPO (MI)Base Model=Llama-3.2-1B-Instruct2026.0380.17
SFT (Revision)Base Model=Llama-3.2-1B-Instruct2026.0380
RLAIFBase Model=Llama-3.2-1B-Instruct2026.0379.83
PPO (MI)Base Model=Qwen2.5-7B-Instruct2026.0379
MIPOBase Model=Qwen2.5-1.5B-Instruct2026.0378.83
Personalized PromptingBase Model=Qwen2.5-7B-Instruct2026.0378.5
Personalized PromptingBase Model=Llama-3.2-1B-Instruct2026.0378
Personalized PromptingBase Model=Llama-3.2-3B-Instruct2026.0378
RLAIFBase Model=Qwen2.5-3B-Instruct2026.0377.33
SFTBase Model=Qwen2.5-7B-Instruct2026.0377
SFT (Revision)Base Model=Qwen2.5-7B-Instruct2026.0377
Personalized PromptingBase Model=Qwen2.5-3B-Instruct2026.0376
SFT (Revision)Base Model=Qwen2.5-3B-Instruct2026.0374.83
RLAIF (MI)Base Model=Llama-3.2-3B-Instruct2026.0374.5
RLAIF (MI)Base Model=Qwen2.5-3B-Instruct2026.0374.5
SFTBase Model=Qwen2.5-3B-Instruct2026.0374.33
PPO (MI)Base Model=Llama-3.2-3B-Instruct2026.0372.67
PPO (MI)Base Model=Qwen2.5-3B-Instruct2026.0372.67
INFONCEBase Model=Qwen2.5-3B-Instruct2026.0370
SFT (Revision)Base Model=Qwen2.5-1.5B-Instruct2026.0367
Personalized PromptingBase Model=Qwen2.5-1.5B-Instruct2026.0363.5
SFTBase Model=Qwen2.5-1.5B-Instruct2026.0360.5
RLAIF (MI)Base Model=Qwen2.5-1.5B-Instruct2026.0359.83
INFONCEBase Model=Qwen2.5-1.5B-Instruct2026.0358.83
RLAIFBase Model=Qwen2.5-1.5B-Instruct2026.0350.33
PPO (MI)Base Model=Qwen2.5-1.5B-Instruct2026.0339.67
Qwen2.5-7B-InstructBase Model=Qwen2.5-7B-Instruct2026.0339
Llama-3.2-3B-InstructBase Model=Llama-3.2-3B-Instruct2026.0337
Qwen2.5-3B-InstructBase Model=Qwen2.5-3B-Instruct2026.0334
Llama-3.2-1B-InstructBase Model=Llama-3.2-1B-Instruct2026.0331
Qwen2.5-1.5B-InstructBase Model=Qwen2.5-1.5B-Instruct2026.0325.5