pub fn output_room(
window: Option<u64>,
most: Option<u64>,
estimate: u64,
) -> Option<u64>Expand description
The most output a request of estimate tokens may ask of a model with
a window of tokens and an output limit of most: most, or what the
window leaves after the request and OUTPUT_MARGIN when that is
less. None when neither is known; Some(0) when the window leaves no
room, which calls for compacting first, not for a request with no room
to answer.