Enum TokenScaling 

pub enum TokenScaling {
    Double,
    Linear {
        step: u32,
    },
    Fixed(Vec<u32>),
}

How successive recovery attempts grow max_output_tokens.

Variants

Double

Double per attempt: 4096 → 8192 → 16384 → … Worst case 2^max_attempts × the starting cap. Default for callers that prefer a small ladder with big steps.

Linear

Add a fixed step per attempt: 4096 → 4096+step → 4096+2·step. Predictable cost ladder; better when the model usually only needs a little more room.

Fields

step: u32

Fixed(Vec<u32>)

Explicit progression: caps[0] for the first retry, caps[1] for the second, etc. Lets callers express “try 8k then 16k then give up” without computing scales.

Trait Implementations

impl Clone for TokenScaling

fn clone(&self) -> TokenScaling

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TokenScaling

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.