Search papers, labs, and topics across Lattice.
2
0
5
On-device fine-tuning of Transformers is now feasible on ultra-low-power, memory-constrained edge devices thanks to TrainDeeploy, which achieves up to 11 trained images per second on a RISC-V SoC.
NeRFs can now fly: Tiny-DroNeRF slashes the memory footprint of neural radiance fields by 96% to enable 3D scene reconstruction on nano-drones with ultra-low-power MCUs, and then uses federated learning to boost accuracy.
