Xsum
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
XSum 1,000 samples (test)
23.11ROUGE-L
12
XSUM
22.4AUPC
12
XSum
18.87ROUGE-L
10
XSum
27.27ROUGE-1
10
XSum
56.4Balanced Accuracy
10
XSum
8.38ROUGE-2
10
XSum
25.76F1 Score
10
XSUM balanced (test)
76Mage Macro F1
10
XSum
18.87ROUGE-L F1
9
GEM XSum en
23.2ROUGE-2
9
XSum 50 document sample (sampled)
40.9RL Score
9
XSUM FRANK
0.183Partial Pearson's ρ
9
XSum
0.37PRR
8
XSum T=1 (test)
18.99ROUGE-2
8
XSum
87.5Pairwise Preference Rate
8
XSUM
80Win Rate
8
XSum Polish 4o-mini
84.86Detection Performance
8
XSum Revise 4o-mini
87.59Detection Performance
8
XSum Paraphrase 4o
0.826Detection Performance
8
XSum Paraphrase 4o-mini
86.21Detection Performance
8
XSum Polish 4o-mini short text
0.746AUROC
8
XSum Revise 4o-mini short text
0.805AUROC
8
XSum Paraphrase 4o short text
0.823AUROC
8
XSum Paraphrase 4o-mini short text
80.99AUROC
8
XSum
0.118Paraphrase (4o-mini)
8