Dialogue on Dialogue
77.79PandaLMHBAT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HBAT#Param=13B, PPO=false, DPO=true2024.06 | 77.79 | -4.78 | 67.45 | |
| HBAT-Freeze#Param=13B, PPO=false, DPO=true2024.06 | 76.87 | -5.01 | 65.62 | |
| DPO#Param=13B, PPO=false, DPO=true2024.06 | 75 | -5.07 | 64.31 | |
| HBAT#Param=7B, PPO=false, DPO=true2024.06 | 74.78 | -5.22 | 58.1 | |
| HBAT#Param=13B, PPO=true, DPO=false2024.06 | 72.83 | -5.11 | 62.32 | |
| RLHF#Param=13B, PPO=true, DPO=false2024.06 | 72.21 | -5.65 | 61.16 | |
| HBAT-Freeze#Param=13B, PPO=true, DPO=false2024.06 | 71.31 | -5.49 | 56.37 | |
| HBAT#Param=7B, PPO=true, DPO=false2024.06 | 70.88 | -5.37 | 57.12 | |
| DPO#Param=7B, PPO=false, DPO=true2024.06 | 70.74 | -5.72 | 54.23 | |
| HBAT-Freeze#Param=7B, PPO=true, DPO=false2024.06 | 69.91 | -5.91 | 56.19 | |
| RLHF#Param=7B, PPO=true, DPO=false2024.06 | 69.79 | -5.81 | 55.63 | |
| HBAT-Freeze#Param=7B, PPO=false, DPO=true2024.06 | 68.78 | -5.41 | 56.95 | |
| RLHF+pt#Param=13B, PPO=true, DPO=false2024.06 | 63.67 | -6.97 | 54.45 | |
| SFT#Param=13B, PPO=false, DPO=false2024.06 | 62.16 | -6.32 | 46.11 | |
| SFT#Param=7B, PPO=false, DPO=false2024.06 | 54.76 | -6.79 | 43.49 | |
| SFT+ppo#Param=13B, PPO=true, DPO=false2024.06 | 54 | -7.93 | 43.12 | |
| RLHF+pt#Param=7B, PPO=true, DPO=false2024.06 | 53.52 | -7.09 | 54.18 | |
| SFT+ppo#Param=7B, PPO=true, DPO=false2024.06 | 45.32 | -8.6 | 42.25 |