A Comprehensive Survey of Abstractive Text Summarization Based on Deep Learning

Table 3

The statistics of the standard datasets.

Dataset	Lang.	#Train	#Valid.	#Test.	Ave. source length	Ave. target length
Gigaword	Eng.	3,800,000	189,000	1951	31.4	8.3
CNN/Daily Mail	Eng.	287,226	13,368	11,490	780	56
NYT	Eng.	589,284	32,736	32,739	549	40
Newsroom	Eng.	995,041	105,760	105,760	658.6	26.7
LCSTS	Chi.	2,400,591	10,666	1,106	103.7	17.8