Gradient-based selection that drops high-gradient samples during continual fine-tuning preserves safety alignment in LLMs better than standard fine-tuning while keeping task performance competitive.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Continual Safety Alignment via Gradient-Based Sample Selection
Gradient-based selection that drops high-gradient samples during continual fine-tuning preserves safety alignment in LLMs better than standard fine-tuning while keeping task performance competitive.