Search papers, labs, and topics across Lattice.
Hefei University of Technology, Hefei Comprehensive National Science Center
3
0
6
Achieving 74.419% accuracy, this ensemble approach shows that self-supervised learning can dramatically enhance micro-gesture recognition performance.
Current LLM judges show a troubling reliability gap in long-form evaluations, raising questions about their effectiveness in real-world applications.
Coding agents may be less prone to tangled refactorings than humans, but those refactorings tank compilability.
