Data Science & ML

Training vs Inference

/ˈtreɪnɪŋ vɜːsəs ˈɪnfərəns/

Definition

Training updates model weights using data; inference applies a fixed model to new inputs to make predictions.

Example in context

"Training runs on 8 GPUs overnight — inference needs only one GPU and must respond in under 100ms."

Related terms

Practice this term

Master Training vs Inference in context by working through exercises in the Data Science & ML module. You'll see the term used in real engineering scenarios with multiple-choice, fill-in-the-blank, and matching drills.