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.

Quantize an LLM after training with Quark v1.0.0

Use this resource to post-training quantize a PyTorch or Hugging Face LLM with AMD Quark

New

  • 466 downloads
  • 130 likes
  • 159 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install llm-post-training-quantization-amd

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 identifies the model on the Hub or locally, picks a quantization plan (FP8, INT4 or another scheme), produces a reproducible manifest, asks for approval before running, then verifies the quantized model. It does not accept .onnx inputs and does not train or fine-tune the model.

Mostly picked by

Updated November 2, 2026

Card for the “Quantize an LLM after training with Quark” skill

Details

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

Similar resources

View all

What you get

  • SKILL.md: model intake, plan, manifest and execution
  • references/quant-plan.md and troubleshooting.md: quantization plan and troubleshooting

Good fit for

  • You want to shrink a Llama, Qwen or Mistral-type LLM to FP8 or INT4
  • You need a reproducible quantized model

Not for

    Tags

    • quantization
    • llm
    • fp8

    About

    You say

    “Quantize this Qwen model to FP8 with Quark and give me the manifest.”

    Card for the “Quantize an LLM after training with Quark” skill

    Skill published by amd 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 post-training quantize a PyTorch or Hugging Face LLM with AMD Quark
    3. Review and adjust the result

    Content preview

    Files

    • .federated.json0.3 KB
    • evals/evals.json2.6 KB
    • LICENSE1.1 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • references/contracts/model_analysis.schema.json1.9 KB
    • references/contracts/quant_plan.schema.json1.5 KB
    • references/contracts/run_manifest.schema.json1.7 KB
    • references/environment.md3.8 KB
    • references/example-fp8-qwen3-8b.md3.3 KB
    • references/model-intake.md4.5 KB
    • references/quant-plan.md4.8 KB
    • references/troubleshooting.md5.1 KB
    • skill-card.md0.2 KB
    • SKILL.md9.4 KB

    Origin and licence

    Listed unmodified

    Author
    amd
    Source
    github.com/amd/skills/tree/0b64b9c4decd14af2e9789f1ef4c1b66909e68b1/skills/quark-torch-llm-ptq
    Licence
    MIT

    Notice

    « quark-torch-llm-ptq » (amd, 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.