List your company

Harmlessness on Harmlessness

96Average Win RateCurriculum-RLAIF
78.3282.9187.592.09May 26, 2025

Evaluation results

MethodLinks
Curriculum-RLAIFBase Model=Qwen2.5-32B, Category=Curriculum2025.0596
Implicit Eval. (DPO)Base Model=Qwen2.5-32B, Category=Curriculum2025.0594
Curriculum-RLAIFBase Model=LLaMA-3-8B, Category=Curriculum2025.0593
Internal Eval.Base Model=Qwen2.5-32B, Category=Curriculum2025.0593
Curriculum-RLAIFBase Model=Gemma-1-2B, Category=Curriculum2025.0592
Conventional RLAIFBase Model=Qwen2.5-32B, Category=Non-Curriculum2025.0591
Internal Eval.Base Model=Gemma-1-2B, Category=Curriculum2025.0590
Implicit Eval. (DPO)Base Model=LLaMA-3-8B, Category=Curriculum2025.0590
External Eval.Base Model=Qwen2.5-32B, Category=Curriculum2025.0590
Internal Eval.Base Model=LLaMA-3-8B, Category=Curriculum2025.0589
RLCDBase Model=Qwen2.5-32B, Category=Non-Curriculum2025.0589
External Eval.Base Model=Gemma-1-2B, Category=Curriculum2025.0588
Conventional RLAIFBase Model=LLaMA-3-8B, Category=Non-Curriculum2025.0588
CAIBase Model=Qwen2.5-32B, Category=Non-Curriculum2025.0588
Implicit Eval. (DPO)Base Model=Gemma-1-2B, Category=Curriculum2025.0586
RLCDBase Model=LLaMA-3-8B, Category=Non-Curriculum2025.0585
External Eval.Base Model=LLaMA-3-8B, Category=Curriculum2025.0585
RLCDBase Model=Gemma-1-2B, Category=Non-Curriculum2025.0583
Conventional RLAIFBase Model=Gemma-1-2B, Category=Non-Curriculum2025.0583
CAIBase Model=LLaMA-3-8B, Category=Non-Curriculum2025.0583
CAIBase Model=Gemma-1-2B, Category=Non-Curriculum2025.0579