Instruction Following Evaluation on Vicuna Eval
63.8Win Rate (A)BPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BPOBase LLM=gpt-3.5-turbo, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 63.8 | 36.2 | |
| BPOBase LLM=claude-2, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 60 | 40 | |
| BPOBase LLM=text-bison, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 58.8 | 41.2 | |
| BPOBase LLM=claude-instant-1.2, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 56.3 | 43.7 | |
| BPOBase LLM=gpt-4, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 53.8 | 46.2 |