Search papers, labs, and topics across Lattice.
V, and R
6
0
9
PhysEditWorld reveals that explicit control over physical parameters can transform how game world models interact with their environments, leading to more realistic and manipulable simulations.
Block-GTQ slashes quantization error by up to 80%, transforming how we optimize key-cache performance in long-context models.
UnityShots achieves unprecedented cross-shot coherence in multi-shot video generation, outperforming open-source benchmarks and rivaling top closed-source systems.
Image generation models exhibit a surprising "Order-to-Space Bias," meaning the order you mention objects in a prompt can drastically alter their placement in the generated image, even overriding other visual cues.
Training a single point cloud encoder across diverse 3D domains not only improves perception but also unlocks emergent behaviors and enhances robotic manipulation and spatial reasoning.
AI-generated images betray themselves not by their appearance, but by their *behavior*: they are far more sensitive to small perturbations than real images, revealing a fundamental weakness exploitable for universal detection.