Preference Evaluation on HH-Helpful
52Win CountDLMA-13B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DLMA-13BComparison Baseline=Llama2-13B2024.02 | 52 | 14 | 34 | |
| DLMA-7BComparison Baseline=RLAIF-7B2024.02 | 48 | 14 | 38 | |
| DLMA-13BComparison Baseline=RLAIF-13B2024.02 | 47 | 18 | 35 | |
| DLMA-7BComparison Baseline=Llama2-7B2024.02 | 46 | 15 | 39 | |
| DLMA-13BComparison Baseline=CD-13B2024.02 | 46 | 21 | 33 | |
| DLMA-7BComparison Baseline=CD-7B2024.02 | 43 | 18 | 39 | |
| DLMA-13BComparison Baseline=RLCD-13B2024.02 | 41 | 20 | 39 | |
| DLMA-7BComparison Baseline=RLCD-7B2024.02 | 39 | 21 | 40 |