AMR
Benchmarks
TaskDatasetSOTA resultTrendResults
LDC2017T10 AMR 2.0 (test)
86.26Smatch
168
AMR 3.0 (test)
86.5SMATCH
45
AMR 1.0 (test)
81.7Smatch
45
AMR 1.0 LDC2014T12 (test)
75.4SMATCH F1
23
LDC2017T10 AMR17 (test)
74.79chrF++
17
AMR Italian (IT) human-translated 2.0 (test)
76.7Smatch Score
15
AMR Spanish (ES) human-translated 2.0 (test)
77.1Smatch Score
15
AMR German (DE) human-translated 2.0 (test)
0.746Smatch
15
AMR 3.0 LDC2020T02 (test)
84.2Smatch Labeled
14
AMR Chinese (ZH) human-translated 2.0 (test)
64Smatch
13
AMR 2.0 (test)
49.8BLEU
10
AMR 3.0 (test)
49.2BLEU
5
AMR (Human Evaluation)
6.7Fluency
5