AMR Parsing on AMR 3.0 LDC2020T02 (test)
84.2Smatch LabeledAMRBART (large)
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| AMRBART (large)Pre-trained=true, Model size=large2022.03 | 84.2 | 87.1 | 84.6 | 90.2 | 80.3 | 72.4 | 72.1 | 88.5 | 78.9 | |
| Bevilacqua+ (2021, large)Pre-trained=true, Model size=large2022.03 | 83 | 85.4 | 83.5 | 89.8 | 78.9 | 70.4 | 73 | 87.2 | 82.7 | |
| Bevilacqua+ (2021, large)Pre-trained=true, Model size=large, Silver data fine-tuning=true2022.03 | 83 | 85.4 | 83.5 | 89.5 | 79.1 | 71.3 | 71.7 | 87.1 | 81.2 | |
| AMRBART (base)Pre-trained=true, Model size=base2022.03 | 82.5 | 85.7 | 82.9 | 89.4 | 78.2 | 69.9 | 70.3 | 86.8 | 76.1 | |
| ND + AD + LVreference=§3.4, structure=Levi graph2021.07 | 77 | 79.8 | 77.5 | 86.1 | 73.6 | 62.6 | 71.3 | 73.3 | 79.5 | |
| ND + AD + LVNode Decoder=Arc-aware, Arc Decoder=Levi graph2021.07 | 77 | 79.8 | 77.5 | 86.1 | 73.6 | 62.6 | 71.3 | 73.3 | 79.5 | |
| CL20implementation=original Cai and Lam (2020), runs=32021.07 | 76.8 | 79.9 | 77.3 | 86.3 | 73.2 | 63.4 | 72.3 | 73 | 79.5 | |
| ND + AD + BDreference=§3.32021.07 | 76.8 | 80.1 | 77.3 | 86.5 | 73.1 | 63.6 | 73.2 | 73 | 79.6 | |
| CL20Description=Original implementation of Cai and Lam (2020)2021.07 | 76.8 | 79.9 | 77.3 | 86.3 | 73.2 | 63.4 | 72.3 | 73 | 79.5 | |
| ND + AD + BDNode Decoder=Arc-aware, Arc Decoder=Basic2021.07 | 76.8 | 80.1 | 77.3 | 86.5 | 73.1 | 63.6 | 73.2 | 73 | 79.6 | |
| Lyu et al.2021.07 | 75.8 | — | — | 88 | 72.6 | — | — | — | — | |
| ND + BD + BDreference=§3.22021.07 | 75.8 | 79 | 76.2 | 84.6 | 72.1 | 61.7 | 72.6 | 71.6 | 78.7 | |
| Lyu et al. (2020)2021.07 | 75.8 | — | 88 | — | 72.6 | — | — | — | — | |
| ND + BD + BDNode Decoder=Basic, Arc Decoder=Basic2021.07 | 75.8 | 79 | 76.2 | 84.6 | 72.1 | 61.7 | 72.6 | 71.6 | 78.7 |