Search papers, labs, and topics across Lattice.
1
0
3
You can slash LLM inference energy by 35% on edge devices just by intelligently managing eDRAM refresh rates based on activation data type and lifespan.
