Semantic Parsing on CFQ MCD1, MCD2, MCD3
94.3MCD1 AccuracyDynamic Least-to-Most
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Dynamic Least-to-MostSupervision Paradigm=Prompting2022.09 | 94.3 | 95.3 | 95.5 | 95 | |
| LeARSupervision Paradigm=Fully Supervised2022.09 | 91.7 | 89.2 | 91.7 | 90.9 | |
| T5-large + IRSupervision Paradigm=Fully Supervised, Backbone=T5-large, Intermediate Representation (IR)=true2022.09 | 88.6 | 79.2 | 72.7 | 80.2 | |
| T5-3B + IRSupervision Paradigm=Fully Supervised, Backbone=T5-3B, Intermediate Representation (IR)=true2022.09 | 88.4 | 85.3 | 77.9 | 83.9 | |
| T5-base + IRSupervision Paradigm=Fully Supervised, Backbone=T5-base, Intermediate Representation (IR)=true2022.09 | 85.8 | 64 | 53.6 | 67.8 | |
| HPDSupervision Paradigm=Fully Supervised2022.09 | 79.6 | 59.6 | 67.8 | 69 | |
| T5-largeSupervision Paradigm=Fully Supervised, Backbone=T5-large, Intermediate Representation (IR)=false2022.09 | 65.1 | 32.3 | 25.4 | 40.9 | |
| T5-3BSupervision Paradigm=Fully Supervised, Backbone=T5-3B, Intermediate Representation (IR)=false2022.09 | 65 | 41 | 42.6 | 49.5 | |
| T5-baseSupervision Paradigm=Fully Supervised, Backbone=T5-base, Intermediate Representation (IR)=false2022.09 | 58.5 | 27 | 18.4 | 34.6 |