Search papers, labs, and topics across Lattice.
4
0
11
Achieving state-of-the-art performance in 3D motion forecasting, MolmoMotion reveals that language instructions can significantly enhance trajectory predictions across various object categories and motion types.
Autonomous research agents are surprisingly unreliable, with existing systems hallucinating references 21% of the time and failing to align methods with code as often as 80%, but a new "Chain-of-Evidence" approach can fix this.
Ditch slow, external segmentation pipelines: TrajTok learns trajectory tokens end-to-end, boosting video understanding while staying lean and adaptable.
Sticking to a single HTML-to-text extractor in your LLM pretraining pipeline could be leaving 71% of the data on the table.
