Task-Focused Dialogue on Mgshop (TSE, Reward, BLEU)
94.75TSEGOPO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GOPOBackbone=Qwen3-14B2026.01 | 94.75 | 7.63 | 27.9 | |
| DeepSeekModel Version=R12026.01 | 93.81 | 7.46 | 14.3 | |
| GLMModel Version=4.72026.01 | 93.65 | 7.25 | 15 | |
| GPTModel Version=5.22026.01 | 93.38 | 7.54 | 9.7 | |
| GeminiModel Version=2.52026.01 | 92.87 | 7.35 | 13.3 | |
| GOPOBackbone=Qwen-7B-Chat2026.01 | 92.43 | 7.38 | 21.1 | |
| QwenModel Version=235B2026.01 | 92.27 | 7.24 | 18.2 | |
| PPO2026.01 | 85.84 | 7.09 | 19 | |
| Memento2026.01 | 83.81 | 7.13 | 18.8 | |
| SFT2026.01 | 83.63 | 6.25 | 18.7 | |
| Untrained2026.01 | 74.54 | 5.97 | 9.1 |