ray.serve.llm.LLMServer.reset_prefix_cache

async LLMServer.reset_prefix_cache() → None

Reset the KV prefix cache on the engine.

Clears cached key-value pairs from previous requests.