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

Biomedical Question Answering on PubMedQA (Unauthorized/Authorized/Attack Accuracy)

77Attack Accuracy

No-shield

0.5620.40540.2560.095Oct 16, 2024
Updated 1mo ago

Evaluation Results

MethodLinks
2024.10
77--
2024.10
77--
2024.10
72.5--
2024.10
72.5--
2024.10
71--
2024.10
70--
2024.10
69.5--
2024.10
69--
2024.10
68--
2024.10
65.5--
2024.10
63--
2024.10
60.5--
2024.10
60--
2024.10
58--
2024.10
56.5--
2024.10
55.5--
2024.10
55.5--
2024.10
51.5--
2024.10
49--
2024.10
47--
2024.10
12.5046
2024.10
12.5--
2024.10
12010.5
2024.10
12--
2024.10
12--
2024.10
12--
2024.10
11029
2024.10
10.5--
2024.10
10.5--
2024.10
10--
2024.10
10--
2024.10
9.5--
2024.10
9.5--
2024.10
7--
2024.10
6.5--
2024.10
6015.5
2024.10
5.5--
2024.10
5--
2024.10
4.5--
2024.10
3.5--