arxiv
PublishedJune 10, 2026 at 4:00 AM
▲bullish
Two to Tango: Coupled Task-Reference Selection for Safe LLM Fine-tuning
Publisher summary· verbatim
arXiv:2606.09866v1 Announce Type: cross Abstract: Fine-tuning safety aligned large language models (LLMs) on downstream data improves adaptation but may erode learned safety behavior. Existing methods use fixed safety examples, global constraints, or one-sided task filtering. Our diagnostics show ta
Stay posted· Newsletter
A 5-min weekly brief — top movers, price watch, story of the week.
Discussion
No replies yet. Be first.
Related coverage
More from ARXIV
arxivNovel Claim or D\'ej\`a Vu? Rethinking "Contamination-Free'' Dynamic Evaluation for Multimodal Automated Fact-Checking1harxivMission-Level Runtime Assurance for LLM-Assisted ISR Swarms over a Verification-Aware Fabric1harxivNeonatal Hypoxic-ischaemic Encephalopathy Classification from the EEG and HRV Signals Using a Conformer based Masked Autoencoder1harxivGTIN: A Unified Framework for Joint Event and Time Prediction in Temporal Graphs1hThe Bubble Brief
WEEKLYRead safety insights every Tuesday — top movers, new releases, story of the week.
Originally published on arxiv ↗