Text Summarization On Gigaword

评估指标

ROUGE-1

ROUGE-2

ROUGE-L

评测结果

各个模型在此基准测试上的表现结果

				Paper Title	Repository
OpenAI/o3-mini	60.12	54.22	57.21	-	-
Riple/Saanvi-v0.1	52.21	45.58	60.29	-	-
Pegasus+DotProd	40.6	21.0	37.0	Beyond Reptile: Meta-Learned Dot-Product Maximization between Gradients for Improved Single-Task Regularization	-
BART-RXF	40.45	20.69	36.56	Better Fine-Tuning by Reducing Representational Collapse
MUPPET BART Large	40.4	20.54	36.21	Muppet: Massive Multi-task Representations with Pre-Finetuning
OFA	39.81	20.66	37.11	OFA: Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework
Transformer+Rep(Uni)	39.81	20.40	36.93	Rethinking Perturbations in Encoder-Decoders for Fast Training
Transformer+Wdrop	39.66	20.45	36.59	Rethinking Perturbations in Encoder-Decoders for Fast Training
ProphetNet	39.51	20.42	36.69	ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training
ERNIE-GENLARGE (large-scale text corpora)	39.46	20.34	36.74	ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation
PALM	39.45	20.37	36.75	PALM: Pre-training an Autoencoding&Autoregressive Language Model for Context-conditioned Generation
Best Summary Length	39.27	20.40	37.75	A New Approach to Overgenerating and Scoring Abstractive Summaries
ERNIE-GENLARGE	39.25	20.25	36.53	ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation
ControlCopying + BPNorm	39.19	20.38	36.69	Controlling the Amount of Verbatim Copying in Abstractive Summarization
PEGASUS	39.12	19.86	36.24	PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization
BiSET	39.11	19.78	36.87	BiSET: Bi-directional Selective Encoding with Template for Abstractive Summarization
ControlCopying + SBWR	39.08	20.47	36.69	Controlling the Amount of Verbatim Copying in Abstractive Summarization
UniLM	38.90	20.05	36.00	Unified Language Model Pre-training for Natural Language Understanding and Generation
ERNIE-GENBASE	38.83	20.04	36.20	ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation
MASS	38.73	19.71	35.96	MASS: Masked Sequence to Sequence Pre-training for Language Generation

0 of 41 row(s) selected.