license: llama3.2
This is the KVLink5 model of the paper "KVLink: Accelerating LLMs via Efficient KV Cache Reuse."