Struct EvaluationCriteria 

pub struct EvaluationCriteria {
    pub tool_trajectory_score: Option<f64>,
    pub tool_trajectory_config: Option<ToolTrajectoryConfig>,
    pub response_similarity: Option<f64>,
    pub response_match_config: Option<ResponseMatchConfig>,
    pub semantic_match_score: Option<f64>,
    pub semantic_match_config: Option<SemanticMatchConfig>,
    pub rubric_quality_score: Option<f64>,
    pub rubric_config: Option<RubricConfig>,
    pub safety_score: Option<f64>,
    pub hallucination_score: Option<f64>,
    pub custom: Vec<CustomCriterion>,
}

Available on crate feature eval only.

Collection of evaluation criteria

Fields

tool_trajectory_score: Option<f64>

Tool trajectory matching score threshold (0.0 - 1.0) Checks if the agent called the expected tools in the expected order

tool_trajectory_config: Option<ToolTrajectoryConfig>

Tool trajectory configuration

response_similarity: Option<f64>

Response text similarity threshold (0.0 - 1.0) Uses text similarity metrics to compare expected vs actual response

response_match_config: Option<ResponseMatchConfig>

Response matching configuration

semantic_match_score: Option<f64>

LLM-judged semantic match threshold (0.0 - 1.0) Uses an LLM to judge if responses are semantically equivalent

semantic_match_config: Option<SemanticMatchConfig>

Semantic match configuration

rubric_quality_score: Option<f64>

Rubric-based quality score threshold (0.0 - 1.0) Evaluates response quality against defined rubrics

rubric_config: Option<RubricConfig>

Rubric configuration

safety_score: Option<f64>

Safety score threshold (0.0 - 1.0) Checks for unsafe or harmful content

hallucination_score: Option<f64>

Hallucination detection threshold (0.0 - 1.0) Detects factual inaccuracies or made-up information

custom: Vec<CustomCriterion>

Custom criteria for extensibility

Implementations

impl EvaluationCriteria

pub fn exact_tools() -> EvaluationCriteria

Create criteria requiring exact tool trajectory match

pub fn semantic_match(threshold: f64) -> EvaluationCriteria

Create criteria for semantic response matching

pub fn response_similarity(threshold: f64) -> EvaluationCriteria

Create criteria with response similarity

pub fn with_tool_trajectory(self, threshold: f64) -> EvaluationCriteria

Add tool trajectory requirement

pub fn with_response_similarity(self, threshold: f64) -> EvaluationCriteria

Add response similarity requirement

pub fn with_semantic_match(self, threshold: f64) -> EvaluationCriteria

Add semantic match requirement

pub fn with_rubrics( self, threshold: f64, rubrics: Vec<Rubric>, ) -> EvaluationCriteria

Add rubric-based evaluation

pub fn has_criteria(&self) -> bool

Check if any criteria are defined

Trait Implementations

impl Clone for EvaluationCriteria

fn clone(&self) -> EvaluationCriteria

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EvaluationCriteria

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

Formats the value using the given formatter.

impl Default for EvaluationCriteria

fn default() -> EvaluationCriteria

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for EvaluationCriteria

fn deserialize<__D>( __deserializer: __D, ) -> Result<EvaluationCriteria, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for EvaluationCriteria

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

where __S: Serializer,

Serialize this value into the given Serde serializer.