Showing 2026Show all
3 papers · 1 filter
cs.CL2026
The Thinking Leakage in Reinforcement Learning for LLMs
Wenxuan Zhu, Shuming Liu, Bang An +4
Hybrid reasoning models such as the Qwen3 series switch between a THINK mode and a NO-THINK mode through a few chat-template tokens. A popular recipe trains such models with reinfo…
cs.LG2026
HyperSafe: Inference-Time Safety Recovery for Fine-Tuned Language Models
Aznaur Aliev, Carlos Hinojosa, Abdelrahman Eldesokey +3
Safety alignment in large language models can be fragile under fine-tuning, as even benign task adaptation may increase harmful compliance. Existing defenses mainly follow two dire…
cs.CR2026
Defending Against Harmful Supervision Hidden in Benign Samples
Bang An, Yibo Yang, Dandan Guo +3
Existing defenses are effective when harmful content is explicitly mixed into downstream fine-tuning data, but crafted samples can instead hide harmful supervision inside benign ta…