Skip to main content

run_eval_with_meta

Function run_eval_with_meta 

Source
pub fn run_eval_with_meta<H, S>(
    harness: &H,
    scorer: &S,
    cases: &[EvalCase<H::Setup, S::Expect>],
    meta: RunMeta,
) -> EvalReport
where H: Harness, S: Scorer<World = H::World>,
Expand description

As run_eval, but with explicit run RunMeta (backend label, temperature, shared system prompt) recorded on the resulting EvalReport.

This is the single convergence point that owns the progress logging: a one-time startup banner, then a [i/total] line BEFORE and AFTER each case (the BEFORE line — the anti-hang signal — prints the instant the case starts). ALL progress goes to stderr; stdout is left clean for any report payload a host wants to print.