Qinghua Mao

maopopovich's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

upvoted a paper 2 days ago

Imprint Reader: From Weight-Update Readout to Behavioral Intervention

Paper • 2609.35261 • Published 4 days ago • 16

upvoted 2 papers 4 months ago

COLLEAGUE.SKILL: Automated AI Skill Generation via Expert Knowledge Distillation

Paper • 2605.31264 • Published May 29 • 130

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

Paper • 2605.29801 • Published May 28 • 145

upvoted 2 papers 6 months ago

Rethinking Generalization in Reasoning SFT: A Conditional Analysis on Optimization, Data, and Model Capability

Paper • 2604.06628 • Published Apr 8 • 130

ATBench: A Diverse and Realistic Trajectory Benchmark for Long-Horizon Agent Safety

Paper • 2604.02022 • Published Apr 2 • 16

upvoted a paper 7 months ago

Code2Math: Can Your Code Agent Effectively Evolve Math Problems Through Exploration?

Paper • 2603.03202 • Published Mar 3 • 16

upvoted an article over 1 year ago

view article

Article

What is test-time compute and how to scale it?

• 

Kseniase

•

Feb 6, 2025

• 126