Run inference by task pipeline

Run inference using a recommended model for a specific task pipeline. Hugging Face selects the best available model for the given task.