Multi-domain Dialog on MultiWOZ
85.3Success RateDyBBT-8B/GPT-4.0
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DyBBT-8B/GPT-4.0System 1=Qwen3-8B, System 2=GPT-4.02025.09 | 85.3 | 92.2 | 87.8 | 13.9 | |
| AutoTODYear=20242025.09 | 84.4 | 91.7 | 86.7 | — | |
| DyBBT-8BBackbone=8B2025.09 | 84.1 | 91.2 | 86.9 | 14.6 | |
| ProTODYear=20252025.09 | 83.3 | 91.7 | 87 | — | |
| DyBBT-4BBackbone=4B2025.09 | 82.5 | 90.9 | 86.4 | 15.2 | |
| DyBBT-1.7BBackbone=1.7B2025.09 | 81.3 | 89.6 | 85.3 | 15.6 | |
| DyBBT-0.6BBackbone=0.6B2025.09 | 78.2 | 88.1 | 84.2 | 16.1 | |
| MACRMYear=20252025.09 | 74.3 | 78.8 | 84 | 8.03 | |
| PPOYear=20232025.09 | 71.7 | 74.1 | 86.6 | 17.8 | |
| REINFORCEYear=20232025.09 | 31.7 | 56.9 | 17.4 | 25.3 | |
| EIERLYear=20252025.09 | 18.5 | — | — | — | |
| LLM_DPYear=20242025.09 | 8 | — | — | — | |
| DQNYear=20152025.09 | 3.5 | — | — | — |