Activation Approximations Can Incur Safety Vulnerabilities Even in Aligned LLMs: Comprehensive Analysis and Defense
Paper
•
2502.00840
•
Published
•
1
•
3
Lipeng (Tony) He
ttttonyhe
·
AI & ML interests
Trustworthy Machine Learning
Recent Activity
commented on
a paper
about 1 month ago
Activation Approximations Can Incur Safety Vulnerabilities Even in
Aligned LLMs: Comprehensive Analysis and Defense
liked
a dataset
about 2 months ago
PKU-Alignment/BeaverTails
liked
a model
3 months ago
cais/HarmBench-Llama-2-13b-cls
Organizations
None yet
ttttonyhe's activity
commented
a
paper
about 1 month ago