Universal Dependency Parsing on SpokeBench Complex v1 (test)
69.2ID AccuracyBiLingua Parser
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| BiLingua ParserType=LLM-based pipeline2026.02 | 69.2 | 77.3 | 60.9 | 65.7 | 69.8 | |
| DECAPBackbone=GPT-4.12026.02 | 55.8 | 70.9 | 53.4 | 55.7 | 63.8 | |
| Bilingual Traditional ParserBackbone=Stanza2026.02 | 13.3 | 13.3 | 13 | 13 | 28 |