Enum EvalScorer 

pub enum EvalScorer {
    LlmJudge {
        model: String,
        rubric: String,
        min_score: u8,
    },
    Assertion {
        checks: Vec<String>,
    },
    Latency {
        threshold_ms: u64,
    },
    Cost {
        threshold_usd: f64,
    },
    Custom {
        module: String,
        kwargs: Value,
    },
}

A scorer within an Eval node.

Variants

LlmJudge

LLM-as-judge: sends output to a model with a rubric, expects a score 1-5.

Fields

model: String

rubric: String

min_score: u8

Minimum acceptable score (1-5). Scores below this fail.

Assertion

Deterministic Python expressions evaluated against the output.

Fields

checks: Vec<String>

Each check is a Python expression that must evaluate to truthy.

Latency

Ensures node execution completed within a latency threshold.

Fields

threshold_ms: u64

Maximum allowed duration in milliseconds.

Cost

Ensures the execution cost is within budget.

Fields

threshold_usd: f64

Maximum allowed cost in USD.

Custom

Custom Python scorer loaded via entry point or module path.

Fields

module: String

Python dotted path: “my_package.scorers:MyScorer”

kwargs: Value

Optional keyword arguments passed to the scorer.

Trait Implementations

impl Clone for EvalScorer

fn clone(&self) -> EvalScorer

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EvalScorer

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for EvalScorer

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for EvalScorer

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

where __S: Serializer,

Serialize this value into the given Serde serializer.