Benchmark your MAX model against standard tests.
What it does
Measures benchmark accuracy for MAX-served text models and produces reproducible per-task scores for comparison with reference results.
When to use it
Use when benchmarking a served model, comparing reference scores, checking a newly imported MAX model's answers, or repeating dataset evaluations.
How to use it
Give it a served model ID; it evaluates benchmark accuracy and saves per-task and combined summary files.
What you provide
- the served model ID
Uses
Latest review
What you need · 4
Requires Pixi to install and run the evaluator environment.
Requires a running MAX endpoint; multiple-choice tasks additionally require the server to have been started with --enable-echo.
Access · 4
This skill
MAX
Write
Sends benchmark prompts to MAX
Hugging Face
Read
Downloads benchmark datasets