Task-oriented dialogue on MultiWOZ 2.4
99InformQwen-3-32B
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Qwen-3-32BSystem Type=Multi-Agent, Optimization Strategy=Mean of L1 Norm2026.06 | 99 | 94 | 54.4 | 98.1 | 85.5 | 91.4 | |
| Qwen-3-32BSystem Type=Multi-Agent, Optimization Strategy=Max of L1 Norm2026.06 | 99 | 95 | 53 | 98.3 | 85.2 | 91.3 | |
| Qwen-3-32BSystem Type=Multi-Agent, Optimization Strategy=Mean of Product with Input2026.06 | 98 | 83 | 51.9 | 96.9 | 84.6 | 90.3 | |
| Qwen-3-32BSystem Type=Multi-Agent, Optimization Strategy=Before Optimization2026.06 | 95 | 80 | 28.9 | 92.5 | 69.4 | 79.3 | |
| Qwen-3-32BSystem Type=Multi-Agent, Optimization Strategy=Max of Product with Input2026.06 | 95 | 86 | 51 | 97.4 | 79.3 | 87.4 | |
| Qwen-3-32BSystem Type=Single-Agent, Optimization Strategy=None2026.06 | 88 | 40 | 44.4 | 97.7 | 80.9 | 88.5 | |
| Llama-3.3-70B-ItSystem Type=Multi-Agent, Optimization Strategy=Before Optimization2026.06 | 87 | 57 | 24.3 | 97.4 | 54.1 | 69.5 | |
| Llama-3.3-70B-ItSystem Type=Multi-Agent, Optimization Strategy=Max of L1 Norm2026.06 | 87 | 54 | 36.7 | 94.5 | 78.9 | 86 | |
| Llama-3.3-70B-ItSystem Type=Multi-Agent, Optimization Strategy=Max of Product with Input2026.06 | 85 | 62 | 38.9 | 94.7 | 81.7 | 87.7 | |
| Llama-3.3-70B-ItSystem Type=Single-Agent, Optimization Strategy=None2026.06 | 84 | 71 | 40.3 | 96.9 | 81.5 | 88.5 | |
| Llama-3.3-70B-ItSystem Type=Multi-Agent, Optimization Strategy=Mean of Product with Input2026.06 | 82 | 59 | 39 | 96.7 | 79.6 | 87.4 | |
| Llama-3.3-70B-ItSystem Type=Multi-Agent, Optimization Strategy=Mean of L1 Norm2026.06 | 42 | 7 | 28.6 | 90.7 | 78 | 83.9 |