Struct EvalReport 

pub struct EvalReport {
    pub outcomes: Vec<CaseOutcome>,
    pub temperature: f32,
    pub backend: String,
    pub system_prompt: String,
}

The aggregate report over a set of cases, plus latency/token statistics.

Fields

outcomes: Vec<CaseOutcome>

One outcome per case, in run order.

temperature: f32

The temperature the eval ran at (noted in the summary; default 0 for determinism).

backend: String

A short description of the backend the eval ran against (from ChatBackend::describe).

system_prompt: String

The exact system prompt every case in this run used (the loop’s compact-vs-full selection). Stored ONCE here rather than on each CaseOutcome (it is identical across cases), and shown at the top of the run’s expander in the HTML report. #[serde(default)] keeps old result JSON loadable.

Implementations

impl EvalReport

pub fn new( outcomes: Vec<CaseOutcome>, temperature: f32, backend: String, system_prompt: String, ) -> Self

Build a report from the per-case outcomes, recording the run temperature + backend label + the shared system prompt for the summary and the HTML report’s per-run expander.

pub fn total(&self) -> usize

Number of cases run.

pub fn passed(&self) -> usize

Number of cases that fully passed.

pub fn accuracy(&self) -> f64

Accuracy = fraction of cases that fully passed, in 0.0..=1.0. 0.0 for an empty report.

pub fn mean_latency(&self) -> Duration

Mean per-case latency. Duration::ZERO for an empty report.

pub fn latency_percentile(&self, p: f64) -> Duration

The p-quantile of per-case latency (p in 0.0..=1.0), via the nearest-rank method on the sorted latencies. Duration::ZERO for an empty report.

pub fn p50_latency(&self) -> Duration

Median (p50) per-case latency.

pub fn p95_latency(&self) -> Duration

p95 per-case latency.

pub fn total_tokens(&self) -> Option<u32>

Total completion tokens across cases that reported a token count. None when NO case reported any (so the summary can say “tokens unavailable” rather than print a misleading 0).

pub fn mean_tokens(&self) -> Option<f64>

Mean completion tokens over the cases that reported a count; None when none reported.

Trait Implementations

impl Clone for EvalReport

fn clone(&self) -> EvalReport

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EvalReport

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for EvalReport

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for EvalReport

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

Formats the value using the given formatter.

impl Serialize for EvalReport

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.