Search papers, labs, and topics across Lattice.
1
0
2
HBQ achieves state-of-the-art efficiency in LLM inference, delivering up to 2.3x higher area/energy efficiency while maintaining accuracy levels that challenge conventional methods.
