Search papers, labs, and topics across Lattice.
Shanghai Jiao Tong University
3
0
6
StreamKL slashes memory usage from quadratic to constant, enabling efficient long-context attention distillation on a single GPU.
Forget hand-tuning: CODO automatically compiles efficient FPGA dataflow accelerators, delivering up to 33.8x speedups on DNN models compared to existing frameworks.
Helical peptides can be modeled as piecewise integrable systems, allowing for accurate prediction and design of local protein geometry by focusing on torsion uniformity within quantitatively defined integrability zones.
