easy5 pts · first answer countsmodel-servingllm-inference

Two phases of LLM generation

LLM generation has a prefill phase and a decode phase. What happens in prefill?

A.Output tokens are produced one at a timeB.Weights are loaded from disk to the GPUC.The response is checked against a safety filterD.All prompt tokens are processed at once to build the KV cache

Discussion 0

Join free to post in the discussion.

Two phases of LLM generation

Discussion 0

Related questions