Search papers, labs, and topics across Lattice.
6
0
7
0
This work introduces ToolCompass, a post-training framework that guides tool trialing by organizing tool-call representations according to shared functions and jointly reduces intra-function variation across domains and increases inter-function separation.
Humanoid robots can now traverse complex 3D structures with remarkable agility and precision, completing intricate maneuvers in real-world trials.
A new robustness metric based on Fisher Information reveals critical insights into model sensitivity and adversarial vulnerability across popular architectures.
Untangling the chaotic web of microservice failures just got easier: a new model uses temporal graph neural networks to pinpoint faults by jointly learning how services evolve and interact.
Humanoid robots can now traverse complex terrains with whole-body coordination thanks to a novel framework that blends diffusion-based motion generation with robust RL-based tracking.
FLASH enables robots to master complex deformable manipulation tasks in minutes using only synthetic data, eliminating the need for labor-intensive real-world training.
