Spoken Language Understanding on SNIPS Far Field
74.64EM Accuracy (Hyp)AT-AT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AT-ATpretraining_tasks=SLU + ASR, finetuned=true2020.12 | 74.64 | 53.25 | |
| Googledescription=Google's DialogFlow cloud service2020.12 | 73.43 | — | |
| SNIPSdescription=Pipe-lined approach with acoustic model, language model, and slot tagging model2020.12 | 71.67 | — | |
| E2E Model w. pret. AT-AT encoderarchitecture=End-to-end model with pretrained AT-AT audio encoder, ablation_study=true2020.12 | 67.83 | 38.92 |