List your company
ResearchFormerly WizWand

Dependency Parsing on UD CHILDES (test)

0.9491UASSuPar
0.3780360.5262930.674550.822807Feb 15, 2026Mar 2, 2026Mar 18, 2026Apr 2, 2026Apr 18, 2026May 3, 2026May 19, 2026
Updated 5mo ago

Evaluation results

MethodLinks
SuParBackbone Model=RoBERTa-large, Training Data=gold2026.050.9491———92.51
StanzaBackbone Model=RoBERTa-large, Training Data=gold2026.050.9421———91.39
StanzaBackbone Model=RoBERTa-base, Training Data=gold2026.050.9416———91.4
SuParBackbone Model=RoBERTa-large, Training Data=10k silver + gold2026.050.9383———91.2
StanzaBackbone Model=BiLSTM, Training Data=gold2026.050.9344———90.34
SuParBackbone Model=RoBERTa-large, Training Data=10k silver → gold2026.050.9334———90.41
DiaParserBackbone Model=RoBERTa-large, Training Data=gold2026.050.9303———89.71
DiaParserBackbone Model=RoBERTa-large, Training Data=10k silver + gold2026.050.9133———87.77
StanzaBackbone Model=off-the-shelf, Training Data=all EN-UD treebanks2026.050.8918———85.19
spaCyBackbone Model=en_core_web_trf, Training Data=OntoNotes 52026.050.6811———61.45
spaCy2026.020.6838,783———
Current ParserNumber of rules=1452026.020.634,172———
Current ParserNumber of rules=632026.020.422,963———
DiaParserBackbone Model=RoBERTa-large, Training Data=gold data2026.05——71.2483.65—
DiaParserBackbone Model=RoBERTa-large, Training Data=10k silver data + gold data2026.05——67.7980.59—
MachampBackbone Model=RoBERTa-large, Training Data=gold2026.05————79
MachampBackbone Model=RoBERTa-base, Training Data=gold2026.05————78.12
spaCyBackbone Model=en_core_web_trf, Training Data=OntoNotes 52026.05——31.4450.18—
StanzaBackbone Model=BiLSTM, Training Data=gold data2026.05——73.4284.89—
StanzaBackbone Model=RoBERTa-large, Training Data=gold data2026.05——75.4986.39—
StanzaBackbone Model=RoBERTa-base, Training Data=gold data2026.05——75.6586.35—
StanzaBackbone Model=Off-the-shelf, Training Data=all EN-UD treebanks2026.05——64.0776.9—
SuParBackbone Model=RoBERTa-large, Training Data=gold data2026.05——78.2287.61—
SuParBackbone Model=RoBERTa-large, Training Data=10k silver data + gold data2026.05——75.7286.01—
SuParBackbone Model=RoBERTa-large, Training Data=10k silver data → gold data2026.05——77.4587.4—