Joint Slot Filling and Intent Detection on SNIPS-NLU (test)
0.981Intent AccuracyAGIF
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| AGIF2020.04 | 0.981 | 0.948 | 0.873 | 98.3 | |
| Stack-Propagation2020.04 | 0.98 | 0.942 | 0.869 | — | |
| CAPSULE-NLUtraining=two-stage, ablation=without Joint Training2018.12 | 0.977 | 0.902 | 0.804 | — | |
| IntentCapsNet2018.12 | 0.974 | — | — | — | |
| CAPSULE-NLUtraining=joint2018.12 | 0.973 | 0.918 | 0.809 | — | |
| Joint Multiple ID-SF2020.04 | 0.972 | 0.88 | — | — | |
| Slot-Gated Full Atten.2018.12 | 0.97 | 0.888 | 0.755 | — | |
| SF-ID2020.04 | 0.97 | 0.905 | 0.784 | — | |
| Joint Seq.2018.12 | 0.969 | 0.873 | 0.732 | — | |
| Attention BiRNN2018.12 | 0.967 | 0.878 | 0.741 | — | |
| DR-AGG2018.12 | 0.966 | — | — | — | |
| CAPSULE-NLUablation=without Intent Detection2018.12 | — | 0.902 | — | — |