Safety Alignment as Continual Learning: Mitigating the Alignment Tax via Orthogonal Gradient Projection
Paper • 2602.07892 • Published
Guanglong Sun
long2333
·
AI & ML interests
None yet
Recent Activity
submitted a paper about 2 hours ago
Safety Alignment as Continual Learning: Mitigating the Alignment Tax via Orthogonal Gradient Projection authored a paper about 13 hours ago
Safety Alignment as Continual Learning: Mitigating the Alignment Tax via Orthogonal Gradient Projection authored a paper about 13 hours ago
MePo: Meta Post-Refinement for Rehearsal-Free General Continual LearningOrganizations
None yet
submitted a
paper to Daily Papers about 2 hours ago
authored 2
papers about 13 hours ago
updated 2
models 2 days ago
published a
model 8 days ago
published a
model 3 months ago