Search papers, labs, and topics across Lattice.
4
6
8
4
Ditching pixel-space translation unlocks a unified model (LatentUM) that reasons across modalities with SOTA results, opening doors to more efficient and aligned visual AI.
Skip the costly generative evals: a simple probe trained on internal LLM representations can accurately predict downstream task performance during training, slashing evaluation time from an hour to just three minutes.
Generate safety-critical driving scenarios with full trajectory control, even *beyond* your training data, using RL to fine-tune a video diffusion model.
Forget generating 4D from text or a single image – this work lets you create compelling 3D animations by simply specifying the start and end poses in two images.
