ToolCallEvaluator
An evaluator that verifies agentic tool calls against an expected trajectory.
Declaration
struct ToolCallEvaluator<Input> where Input : ModelSampleProtocol, Input.Expectation == TrajectoryExpectationMentioned in
Overview
This evaluator produces both a strict and partial result from a single evaluation pass. It also supports ordered sequences, unordered expectations, disallowed tool checks, and group steps.
Use the toolsAllPass and toolsPercentagePass conveniences for the standard metrics:
let evaluator = ToolCallEvaluator<ModelSample<String>>(
allPass: .toolsAllPass, percentagePass: .toolsPercentagePass
)