Instruction Following Evaluation on BPO Eval (test)
58.5A Win RateBPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BPOBase LLM=gpt-3.5-turbo, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 58.5 | 41.5 | |
| BPOBase LLM=text-bison, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 53 | 47 | |
| BPOBase LLM=claude-instant-1.2, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 52.5 | 47.5 | |
| BPOBase LLM=claude-2, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 52 | 48 | |
| BPOBase LLM=gpt-4, Method A=BPO, Method B=Original, Evaluator=Claude-v1.32023.11 | 51.5 | 48.5 |