Multi-Intent Spoken Language Understanding on MixATIS
57.5Overall AccuracyCo-guiding Net
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Co-guiding NetBackbone=RoBERTa2022.10 | 57.5 | — | — | |
| GL-GINBackbone=RoBERTa2022.10 | 53.6 | — | — | |
| Co-guiding Net2022.10 | 51.3 | — | — | |
| AGIFBackbone=RoBERTa2022.10 | 50 | — | — | |
| SSRANScope Recognizer=true, Result Attention Network=true, Auxiliary Tasks=true2022.11 | 48.9 | 77.9 | 89.4 | |
| SSRAN w/o Auxiliary TasksAuxiliary Tasks=false2022.11 | 47.2 | 77.1 | 89.1 | |
| SSRAN w/o SRScope Recognizer=false2022.11 | 46.3 | 76.5 | 88.7 | |
| GL-GIN2022.11 | 43.5 | 76.3 | 88.3 | |
| LR-Transformer2022.11 | 43.3 | 76.1 | 88 | |
| GL-GIN2022.10 | 43 | — | — | |
| SSRAN w/o RANResult Attention Network=false2022.11 | 43 | 76.9 | 87.6 | |
| Basic modelScope Recognizer=false, Result Attention Network=false, Auxiliary Tasks=false2022.11 | 41.2 | 75.1 | 85.2 | |
| AGIF2022.10 | 40.8 | — | — | |
| AG-IF2022.11 | 40.8 | 74.4 | 86.7 | |
| Stack-Propagation2022.11 | 40.1 | 72.1 | 87.8 | |
| Joint Multiple ID-SF2022.11 | 36.1 | 73.4 | 84.6 |