metrale_bench/benchmarks/kat_equality/mod.rs
1// SPDX-License-Identifier: MIT OR Apache-2.0
2
3//! 2026-09-26: KAT equality: the same sample, in different request orders,
4//! byte for byte.
5//!
6//! A known-answer test is a claim about its input only if the output does not
7//! also depend on what ran before it. This gate issues one BFCL draw against
8//! one server in two or more orders and requires every `sample_id`'s reply to
9//! be byte-identical across them.
10//!
11//! `ssm-state-poisoning-gate` cannot answer this: it replays one fixed script
12//! in the same order every round, and it counts a reworded reply
13//! (`RoundVerdict::Jittered`) as a pass. Here a reworded reply is the finding.
14//!
15//! Owner: bench, kat_equality.
16//! Invariants: none beyond the types.
17
18pub mod compare;
19pub mod driver;
20pub mod nemotron;
21pub mod report;
22
23pub use compare::{
24 Observation, OrderRun, SampleVerdict, Score, permutation, score, verdict, verdict_for,
25};
26
27pub use driver::{DESCRIPTOR, METADATA};
28pub use nemotron::{NEMOTRON_NANO_DESCRIPTOR, NEMOTRON_SUPER_DESCRIPTOR};
29
30#[cfg(test)]
31#[path = "compare_tests.rs"]
32mod compare_tests;
33
34#[cfg(test)]
35#[path = "driver_tests.rs"]
36mod driver_tests;