Xsum
Benchmarks
TaskDatasetSOTA resultTrendResults
XSum (test)
60.61ROUGE-2
276
Xsum
27.1ROUGE-2
108
XSum
0.806AUROC
98
XSum
0.163Hypervolume (HV)
72
XSum
0.617PRR
66
XSum
85.9ROC-AUC
66
XSum Generated by Claude3 (test)
100AUROC
60
XSum Generated by GPT-4 (test)
1AUROC
60
XSum Generated by ChatGPT (test)
1AUROC
60
XSum Fast-DetectGPT benchmark
100AUROC
54
XSum
9.16ROUGE-2
46
XSUM (test)
40.4ROUGE-L
44
XSum (test)
0.945AUC
43
XSum
27.45ROUGE-L
42
Xsum
0Empirical Privacy Loss (epsilon)
40
XSum
100AUROC
40
XSum on Gemma-7B
0.147Average Soft Constrained Reward
36
XSum
24.89Accuracy
35
XSum
21.3ROUGE-2
28
XSum
14.71Perplexity
26
XSum
40.67ROUGE-1
25
XSum
100TPR
22
XSum-Faithful (XSF)
47Spearman Correlation
22
XSumFaith (test)
42.5Pearson Correlation Coefficient
22
XSum
46.76Score
20