Conversation Evaluation on Proprietary Chinese 500 Multi-turn Dialogues (test)
79.4Win Rate vs GPT4C-SFT-Empathy
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| C-SFT-EmpathyParameters=70B2024.09 | 79.4 | 97.5 | 77.3 | |
| Llama3-70B-InstructParameters=70B2024.09 | 10.2 | — | 9 |
Loading the SOTA2 catalog…