Best-of-N Alignment on HH-RLHF (test)
98Percent batches with BWR > 0.50AdaBoN
Evaluation Results
| Method | Links | |
|---|---|---|
| AdaBoNLM=Llama, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 98 | |
| AdaBoNLM=Mistral, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 94 | |
| AdaBoNLM=Mistral, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 94 | |
| AdaBoNLM=Mistral, RM=Armo, K=5, B=120, d=0.75B2025.05 | 92 | |
| AdaBoNLM=Llama, RM=Armo, K=5, B=120, d=0.75B2025.05 | 92 | |
| AdaBoNLM=Llama, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 82 | |
| AdaBoNLM=Qwen, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 80 | |
| AdaBoNLM=Qwen, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 76 | |
| AdaBoNLM=Gemma, RM=Armo, K=5, B=120, d=0.75B2025.05 | 76 | |
| AdaBoNLM=Qwen, RM=Armo, K=5, B=120, d=0.75B2025.05 | 72 | |
| AdaBoNLM=Gemma, RM=Mistral, K=5, B=120, d=0.75B2025.05 | 70 | |
| AdaBoNLM=Gemma, RM=FsfairX, K=5, B=120, d=0.75B2025.05 | 54 |