HyperAI超神経

プラットフォーム

Data To Text Generation On Webnlg

評価指標

BLEU

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

		Paper Title
Control Prefixes (A1, T5-large)	67.32	Control Prefixes for Parameter-Efficient Text Generation
Control Prefixes (A1, A2, T5-large)	67.15	Control Prefixes for Parameter-Efficient Text Generation
JointGT Baseline	67.08	FactSpotter: Evaluating the Factual Faithfulness of Graph-to-Text Generation
T5-B Baseline	67.04	FactSpotter: Evaluating the Factual Faithfulness of Graph-to-Text Generation
T5-large + Wiki + Position	66.07	Stage-wise Fine-tuning for Graph-to-Text Generation
HTML (fine-tuning)	65.4	HTLM: Hyper-Text Pre-Training and Prompting of Language Models
T5-small	65.05	Investigating Pretrained Language Models for Graph-to-Text Generation
TrICy (trK = trk* = 0.24)	64.73	TrICy: Trigger-guided Data-to-text Generation with Intent aware Attention-Copy
T5-Base	64.7	Text-to-Text Pre-Training for Data-to-Text Tasks
TrICy (trK = 0)	64.08	TrICy: Trigger-guided Data-to-text Generation with Intent aware Attention-Copy
CGE-LW (Levi Graph)	63.69	Modeling Global and Local Node Contexts for Text Generation from Knowledge Graphs
Multiview-G2S	62.89	Structural Information Preserving for Graph-to-Text Generation
Graformer	61.15	Modeling Graph Structure via Relative Position for Text Generation from Knowledge Graphs
GTR-LSTM (entity masking)	58.6	GTR-LSTM: A Triple Encoder for Sentence Generation from RDF Data
E2E GRU	57.20	Neural data-to-text generation: A comparison between pipeline and end-to-end architectures
GCN EC	55.9	Deep Graph Convolutional Encoders for Structured Data to Text Generation
BestPlan	47.4	Step-by-Step: Separating Planning from Realization in Neural Data-to-Text Generation
BART (TextBox 2.0)	-	TextBox 2.0: A Text Generation Library with Pre-trained Language Models

0 of 18 row(s) selected.

Data To Text Generation On Webnlg | SOTA | HyperAI超神経