playrote.play.v1

aamanm006/eval-contamination-detectorv1.0.1Public

Eval contamination detector

Detect train/eval benchmark contamination and verbatim/near-duplicate leakage with 4-stage DAG verification and balanced candidate arithmetic.

  • Published 4 Sep 2026
  • Download 1
  • Needs rote 0.62.0+
  • Digest sha256:adee86f1d4a3…
  • Reference amanm006/eval-contamination-detector@1.0.1

Before you run it1 local tool

Everything here is checked on your machine when rote inspects the play. This page never learns what you have installed or which credentials you hold.

  • ○
    python3local tool

    Requires >=3.9. Run rote play inspect to check your host; unverified versions remain unknown.

Inputs2, all optional

rote asks for these before running and shows the defaults it will use.

  • Train settrain_set

    Absolute path to training dataset JSONL/text (defaults to bundled benchmark fixture)

    optionalstringdefault defaultcustom value allowed
  • Eval seteval_set

    Absolute path to evaluation/benchmark dataset JSONL/text (defaults to bundled benchmark fixture)

    optionalstringdefault defaultcustom value allowed

What it changes

No write operations are declared in this version's manifest.

Credentials remain in the recipient's local rote vault; this page and the publisher never receive them.

Runtime details
  • rote CLIversion 0.62.0 or newer
  • Servicesnone declared
  • Digestsha256:adee86f1d4a3cc834e8d4a056e01399d7bea9ec40457e64b67badf76cf1ce866
  • Archive size9246 bytes
  • Published2026-09-04T10:16:30.745189+00:00
  • Published withrote 0.79.0

Credential values and current readiness remain local. Opening this page never prompts for a credential, opens an authorization browser, or calls a service this play uses.

A play is a published, versioned automation package that runs on your own machine through rote, a free command-line tool. Opening this page installs nothing and runs nothing.