Search papers, labs, and topics across Lattice.
University of Illinois Urbana-Champaign 鈭桭irst Author
8
0
15
Achieving up to 10x faster rendering speeds, CaT-GS redefines efficiency in 3D Gaussian Splatting for large-scale scenes.
Achieving $O(W)$ storage efficiency and high cache hit rates in a large-scale LLM serving system could redefine performance benchmarks for hybrid architectures in production.
Discrepancies between LLM annotations and keyword lexicons can lead to fundamentally different interpretations of public discourse, especially for well-sampled speakers.
Adversaries can exploit overthinking in LVLMs to induce inference delays of up to 6.96x, jeopardizing robotic safety and performance.
Informative layers in speech models cluster in depth zones, and selecting them can boost audio deepfake detection performance by 28% over previous methods.
UniFormer achieves up to a 1.113% increase in Watch Time, showcasing a breakthrough in user engagement for industrial recommendation systems.
Keyword lexicons can mislead researchers into thinking they understand a speaker's psychology when they may simply be capturing a flawed measurement artifact.
Hierarchical latent actions in HiMem-WAM boost long-horizon manipulation robustness, outperforming existing models in real-world scenarios.
