Follow the Gradient: Tracing Safety Degradation Back to Training Data Paper • 2610.37481 • Published 8 days ago • 29
ÜberWeb: Insights from Multilingual Curation for a 20-Trillion-Token Dataset Paper • 2602.15210 • Published Feb 25 • 1
The Finetuner's Fallacy: When to Pretrain with Your Finetuning Data Paper • 2603.16177 • Published Mar 21
20/20 Vision Language Models: A Prescription for Better VLMs through Data Curation Alone Paper • 2605.11405 • Published May 13
SAGA: A Security Architecture for Governing AI Agentic Systems Paper • 2504.21034 • Published Aug 29, 2025
Follow the Gradient: Tracing Safety Degradation Back to Training Data Paper • 2610.37481 • Published 8 days ago • 29