Search papers, labs, and topics across Lattice.
3
0
7
Robot manipulation models trained on mostly VR data can perform as well as those trained on real-world data, but at 1/20th the cost.
By decoupling camera tracking and geometric mapping with a hybrid memory design, Mem3R achieves state-of-the-art streaming 3D reconstruction while shrinking model size by 19%.
Recurrent models can now achieve Transformer-competitive performance on recall-intensive tasks, thanks to a simple memory caching mechanism that grows memory capacity with sequence length.
