Search papers, labs, and topics across Lattice.
8
0
8
19
ELSA3D outperforms existing unified 3D models by halving computational costs while enhancing cross-modal reasoning precision.
RECONTEXT boosts long-context reasoning in LLMs by effectively reusing evidence from the input, leading to superior performance without the need for retraining.
Forget brute-force VLMs: parameter-efficient fine-tuning with high-quality rationales unlocks surprisingly accurate and interpretable time-series anomaly detection.
Forget GAN inversions – now you can steer diffusion models with a dynamically weighted soup of differentiable rewards, including a VQA-based reward for language-vision reasoning, and get SOTA image edits.
Training video generation models to explicitly infer latent physical properties yields more physically plausible videos than simply scaling data and model size.
Hallucinations in 3D embodied agents can be significantly reduced at inference time by contrasting predictions under original and geometrically/semantically perturbed 3D scene graphs.
Text-to-3D generation gets a semantic upgrade: DreamPartGen creates 3D objects with parts that not only look right but also understand their relationships and align with textual descriptions.
Achieve SOTA multimodal performance across eight benchmarks and strong zero-shot generalization without task-specific training by decoupling understanding and generation via unified discrete flow matching.
