Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Automated Probing on HellaSwag
Loading...
35
Error Rate
PAIR
33.4
44.2
55
65.8
Feb 13, 2026
Error Rate
Attack Success Rate
Updated 1mo ago
Evaluation Results
Method
Method
Links
Error Rate
Attack Success Rate
PAIR
Generator Model=GPT-5....
2026.02
35
99.06
AutoDetect
Generator Model=GPT-5....
2026.02
45
-
PROBELLM
Generator Model=GPT-5....
2026.02
75
-
Feedback
Search any
task
Search any
task
