DynaGen

Created an optimization methodology design of LLM KV Cache management policy and enhanced LLM inference throughput by at least 30%, ensuring smooth user experience of LLM services on low-end PCs.