A Gravitational Interpretation of Fine-Tuning Reversion
Paper • 2606.28525 • Published • 5
Nils Lukas
nilslukas
·
AI & ML interests
None yet
Recent Activity
authored a paper 10 days ago
A Gravitational Interpretation of Fine-Tuning Reversion authored a paper 10 days ago
When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models authored a paper 10 days ago
Watermarking Should Be Treated as a Monitoring PrimitiveOrganizations
authored 13
papers 10 days ago
When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models
Paper • 2606.10740 • Published • 2
Watermarking Should Be Treated as a Monitoring Primitive
Paper • 2605.13095 • Published
Robust Safety Monitoring of Language Models via Activation Watermarking
Paper • 2603.23171 • Published
Mitigating Watermark Forgery in Generative Models via Randomized Key Selection
Paper • 2507.07871 • Published • 1
SD-E$^2$: Semantic Exploration for Reasoning Under Token Budgets
Paper • 2601.17982 • Published • 1
Forest Before Trees: Latent Superposition for Efficient Visual Reasoning
Paper • 2601.06803 • Published • 10
Robust and Calibrated Detection of Authentic Multimedia Content
Paper • 2512.15182 • Published • 17
First-Place Solution to NeurIPS 2024 Invisible Watermark Removal Challenge
Paper • 2508.21072 • Published
SPIRIT: Patching Speech Language Models against Jailbreak Attacks
Paper • 2505.13541 • Published • 2
Mask Image Watermarking
Paper • 2504.12739 • Published • 1
Optimizing Adaptive Attacks against Content Watermarks for Language Models
Paper • 2410.02440 • Published
Leveraging Optimization for Adaptive Attacks on Image Watermarks
Paper • 2309.16952 • Published