Prefill vs. Decode in LLM Inference

2 points | by eatonphil 5 hours ago