Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
2
0
4
0
A design space spanning High-Bandwidth Flash and DRAM configurations, bandwidth exposure, and near-data compute is explored, finding that state bandwidth and HBF transport form two largely orthogonal knees and address two provisioning questions.
CAEE slashes inference latency in MoE models by intelligently pruning low-value experts, achieving efficiency gains without sacrificing accuracy.