Dependency Parsing On Penn Treebank

评估指标

LAS

POS

UAS

评测结果

各个模型在此基准测试上的表现结果

				Paper Title	Repository
Label Attention Layer + HPSG + XLNet	96.26	97.3	97.42	Rethinking Self-Attention: Towards Interpretability in Neural Parsing
DMPar + XLNet	95.92	-	97.30	Enhancing Structure-aware Encoder with Extremely Limited Data for Graph-based Dependency Parsing	-
ACE	95.8	-	97.2	Automated Concatenation of Embeddings for Structured Prediction
Deep Biaffine + RoBERTa	95.75	-	97.29	Deep Biaffine Attention for Neural Dependency Parsing
HPSG Parser (Joint) + XLNet	95.72	97.3	97.20	Head-Driven Phrase Structure Grammar Parsing on Penn Treebank
MFVI	95.34	-	96.91	Second-Order Neural Dependency Parsing with Message Passing and End-to-End Training
CVT + Multi-Task	95.02	-	96.61	Semi-Supervised Sequence Modeling with Cross-View Training
RNG Transformer	95.01	-	96.66	Recursive Non-Autoregressive Graph-to-Graph Transformer for Dependency Parsing with Iterative Refinement
SpanRel	94.70	-	96.44	Generalizing Natural Language Analysis through Span-relation Representations
CRFPar	94.49	-	96.14	Efficient Second-Order TreeCRF for Neural Dependency Parsing
Left-to-Right Pointer Network	94.43	-	96.04	Left-to-Right Dependency Parsing with Pointer Networks
Graph-based parser with GNNs	94.31	97.3	95.97	Graph-based Dependency Parsing with Graph Neural Networks	-
Deep Biaffine	94.22	-	95.87	Deep Biaffine Attention for Neural Dependency Parsing
Stack-Pointer Network	94.19	97.3	95.87	Stack-Pointer Networks for Dependency Parsing
jPTDP	93.87	97.97	95.51	An improved neural network model for joint POS tagging and dependency parsing
Experiment-bert	93.2	-	95.42	-	-
Andor et al.	92.79	97.44	94.61	Globally Normalized Transition-Based Neural Networks
Distilled neural FOG	92.06	97.44	94.26	Distilling an Ensemble of Greedy Dependency Parsers into One MST Parser
Weiss et al.	92.06	97.3	94.01	Structured Training for Neural Network Transition-Based Parsing	-
BIST transition-based parser	91.9	97.44	93.99	Simple and Accurate Dependency Parsing Using Bidirectional LSTM Feature Representations

0 of 22 row(s) selected.