Common Request Parameters

Parameters like top_p and stop give finer control over sampling and where generation ends.

What does a 'stop' parameter typically do in an LLM API request?