Struct Prediction 

pub struct Prediction {
    pub model: String,
    pub answers: Map<String, Value>,
    pub typed: Vec<(String, Answer)>,
    pub input_tokens: usize,
    pub output_tokens: usize,
    pub sequence_lengths: Vec<usize>,
    pub truncated: bool,
    pub routing: Option<Value>,
}

The full result of one predict call.

Fields

model: String

answers: Map<String, Value>

Answers in question order.

typed: Vec<(String, Answer)>

Typed answers, same order and keys as answers.

input_tokens: usize

Total across every sequence in the call, as upstream reports it.

output_tokens: usize

sequence_lengths: Vec<usize>

Token count of each question’s sequence, in question order.

input_tokens is a batch total, so it cannot distinguish several short sequences from one that was cut at max_len; these can.

truncated: bool

True when any sequence lost part of its state to max_len.

routing: Option<Value>

Present only on results produced through a crate::Router.

Implementations

impl Prediction

pub fn to_value(&self) -> Value

Render in upstream’s exact shape.

pub fn to_value_detailed(&self) -> Value

Upstream’s shape plus the per-sequence usage detail.

Kept separate from Self::to_value so the default rendering stays byte-compatible with upstream, which emits only the two token counts.

pub fn answer(&self, question_id: &str) -> Option<&Answer>

Look up one answer by question id.

Trait Implementations

impl Clone for Prediction

fn clone(&self) -> Prediction

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Prediction

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

Formats the value using the given formatter.