Search papers, labs, and topics across Lattice.
Ant Group
7
0
10
Delegation intelligence in LLMs can significantly enhance their ability to tackle complex, long-horizon tasks, as demonstrated by SearchSwarm's superior performance on challenging benchmarks.
Code agents are only eliminating 50% of code smells, revealing critical gaps in their understanding of maintainability and cross-file dependencies.
Current AI agents falter in autonomous development, revealing critical gaps in robustness and alignment as they struggle against human-engineered solutions.
Forget benchmarks - SuperValid's capability-aligned validation loss robustly predicts downstream LLM performance across architectures, scales, and training distributions.
Ditch SwiGLU's quadratic instability: PowLU offers a rational power function that stabilizes LLM pre-training without sacrificing performance.
Low-light 3D reconstruction gets a boost from NAKA-GS, a bionics-inspired approach that significantly outperforms baselines by intelligently correcting color and pruning noisy points.
Reconstructing 3D scenes from images obscured by smoke and extreme darkness is now significantly more achievable, thanks to insights gleaned from the NTIRE 2026 challenge.
