Search papers, labs, and topics across Lattice.
2
0
5
Humanoid robots can now sprint at 6 m/s with zero-shot sim-to-real transfer, thanks to a surprisingly data-efficient approach using frequency-domain priors learned from just five human motion sequences.
By recombining subgraphs from sparse models without retraining, "model stitching" creates a diverse set of model variants that significantly improves the efficiency of multi-DNN inference on edge SoCs.
