Every premium and members-only resource for $9.99/month, as a no-commitment subscription. See the plans

PalateKit

Sign in
Menu

Request a resource

Describe the resource you are missing in one sentence. We will email you when it is published.

0 / 280

A page, a video or an image online.

Unlock PalateKit

Unlock everything

Plan

Paying means accepting our terms of sale.

Secure payment by Stripe. Receipts and invoices sent by email.

Welcome back

Sign in to install, download or save this resource. A free account is enough for free resources.

Calibrating an LLM judge on human labels v1.0.0

Use this resource to check that an LLM judge agrees with human evaluations before relying on it

New

  • 435 downloads
  • 106 likes
  • 146 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install llm-judge-calibration

With no option, the skill goes to ~/.agents/skills/; add --agent claude-code (or codex, cursor…) to target an agent.

First time: connect the CLI with

npx --yes palatekit@latest login

The agent splits your human-labelled traces into training, dev and test sets, then measures the judge's true positive and true negative rates. It helps you iterate on the judge prompt and inspect disagreements. You get a validated judge and a bias-correction formula for production data.

Mostly picked by

Updated November 2, 2026

Card for the “Calibrating an LLM judge on human labels” skill

Details

Works in
Install guides
Generates with
Your agent's model
Licence
MIT

Similar resources

View all

What you get

  • SKILL.md: data splits (train, dev, test) and TPR and TNR measurement
  • SKILL.md: disagreement review, iteration and bias correction

Good fit for

  • You wrote a judge prompt and want to check that it follows human labels
  • You have about 100 binary Pass/Fail labelled traces

Not for

    Tags

    • llm judge
    • calibration
    • tpr tnr
    • evaluation

    About

    You say

    “Calibrate my judge on the 100 labelled traces. I need a TPR and a TNR above 90%.”

    Card for the “Calibrating an LLM judge on human labels” skill

    Skill published by hamelsmu under the MIT license, delivered in its original version.

    Show more

    How to use it

    1. Install the skill in your agent's skills folder
    2. Ask your agent to check that an LLM judge agrees with human evaluations before relying on it
    3. Review and adjust the result

    Content preview

    Files

    • LICENSE1.1 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • SKILL.md8.9 KB

    Origin and licence

    Listed unmodified

    Author
    hamelsmu
    Source
    github.com/hamelsmu/evals-skills/tree/22418da2bfb159f28a1b0dcf64e969e14ae56c99/skills/validate-evaluator
    Licence
    MIT

    Notice

    « validate-evaluator » (hamelsmu, MIT) livré dans sa version d'origine : fiche en français et en anglais par PalateKit.

    User reviews

    –

    No ratings yet

    Loading reviews…

    Delete your review?

    Your rating and comment will be removed from the page.