New KV cache compaction technique cuts LLM memory 50x without accuracy loss

MIT researchers developed Attention Matching, a KV cache compaction technique that compresses LLM memory by 50x in seconds — ...
IEEE

Intelligent Caching Based on Popular Content in Vehicular Networks: A Deep Transfer Learning Approach

Abstract: Information-centric networking (ICN) allows data to be cached at each node in the network. It is vital in vehicular networks (VNs) to improve caching performance and reduce content delay in ...