Search papers, labs, and topics across Lattice.
Tianjin University
1
0
SharQ recovers up to 63% of the accuracy gap in low-bit quantization while achieving over 2x latency reduction in LLM inference.
