LLM Alignment on PKU-SafeRLHF
49BWR (Median)AdaBoN
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AdaBoNLM=Qwen, RM=Armo, K=5, B=120, d=0.75B2025.05 | 49 | 46 | 51 | |
| AdaBoNLM=Gemma, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 53 | 50 | 58 | |
| AdaBoNLM=Qwen, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 54 | 53 | 57 | |
| AdaBoNLM=Qwen, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 56 | 52 | 60 | |
| AdaBoNLM=Llama, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 56 | 52 | 59 | |
| AdaBoNLM=Mistral, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 57 | 55 | 60 | |
| AdaBoNLM=Mistral, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 57 | 53 | 60 | |
| AdaBoNLM=Mistral, RM=Armo, K=5, B=120, d=0.75B2025.05 | 57 | 55 | 61 | |
| AdaBoNLM=Gemma, RM=Armo, K=5, B=120, d=0.75B2025.05 | 57 | 53 | 60 | |
| AdaBoNLM=Gemma, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 58 | 55 | 60 | |
| AdaBoNLM=Llama, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 59 | 54 | 62 | |
| AdaBoNLM=Llama, RM=Armo, K=5, B=120, d=0.75B2025.05 | 61 | 58 | 62 |