Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Reading Comprehension on SQuAD (Usage/Attack Metrics)

75.91Attack Accuracy

No-shield

1.560420.862740.16559.4673Oct 16, 2024
Updated 1mo ago

Evaluation Results

MethodLinks
2024.10
75.91--
2024.10
75.2--
2024.10
73.81--
2024.10
69.12--
2024.10
68.21--
2024.10
68.13--
2024.10
67.92--
2024.10
66.28--
2024.10
63.96--
2024.10
63.94--
2024.10
63.91--
2024.10
62.61--
2024.10
61.02--
2024.10
60.82--
2024.10
57.87--
2024.10
32.33--
2024.10
30.54--
2024.10
29.89--
2024.10
28.42--
2024.10
26.34--
2024.10
11.94--
2024.10
10.48--
2024.10
10.01073.02
2024.10
9.71--
2024.10
9.64--
2024.10
9.56--
2024.10
9.42--
2024.10
9.15--
2024.10
8.98062.11
2024.10
8.81--
2024.10
8.61--
2024.10
7.91--
2024.10
7.81043.21
2024.10
7.51--
2024.10
7.35016.5
2024.10
6.81--
2024.10
6.71--
2024.10
5.93--
2024.10
5.66--
2024.10
4.42--