HyperAI超神经

首页资讯论文教程数据集百科 SOTA LLM 模型天梯 GPU 天梯顶会

中文

HyperAI超神经

Question Generation On Squad11

评估指标

BLEU-4

METEOR

ROUGE-L

评测结果

各个模型在此基准测试上的表现结果

模型名称	BLEU-4	METEOR	ROUGE-L	Paper Title	Repository
ProphetNet + syn. mask + localness	24.37	26.26	52.77	Enhancing Pre-trained Models with Text Structure Knowledge for Question Generation	-
UniLMv2	24.43	-	-	UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training
ERNIE-GENLARGE (beam size=5)	25.41	-	-	ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation
ProphetNet	23.91	26.6	52.3	ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training
UniLM + ASGen	23.7	25.9	52.3	Learning to Generate Questions by Recovering Answer-containing Sentences	-
MPQG	13.91	-	-	Leveraging Context Information for Natural Question Generation	-
ProphetNet + ASGen	24.44	26.73	52.8	Learning to Generate Questions by Recovering Answer-containing Sentences	-
RNN +attn +copy	13.5	-	-	Evaluating Rewards for Question Generation Models
UniLM	22.78	25.1	51.1	Unified Language Model Pre-training for Natural Language Understanding and Generation
NQG++	13.27	-	-	Neural Question Generation from Text: A Preliminary Study
BART (TextBox 2.0)	25.08	26.73	52.55	TextBox 2.0: A Text Generation Library with Pre-trained Language Models
BERTSQG	22.17	-	-	A Recurrent BERT-based Model for Question Generation	-
Selector & NQG++	15.874	-	-	Mixture Content Selection for Diverse Sequence Generation

0 of 13 row(s) selected.