fairseq
fairseq copied to clipboard
got abnormal Rouge-L score when using file2rouge
test BART for summarization task on multi-news dataset
get abnormal Rouge-L score using file1rouge
1 ROUGE-1 Average_F: 0.48576 (95%-conf.int. 0.48359 - 0.48795) 1 ROUGE-2 Average_F: 0.18428 (95%-conf.int. 0.18168 - 0.18704) 1 ROUGE-L Average_F: 0.44147 (95%-conf.int. 0.43921 - 0.44366)
Is there anyone help me???