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.

Aligning a model without a reference model with SimPO v1.0.0

Use this resource to align an LLM on preferences without a reference model, more simply than DPO

New

  • 567 downloads
  • 106 likes
  • 126 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install simpo-preference-alignment-2

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 guides you through installing alignment-handbook and Flash Attention 2, configuring SimPO training from a base model and preference data, then tuning hyperparameters. You get baseline settings and their loss functions.

Mostly picked by

Updated November 2, 2026

Card for the “Aligning a model without a reference model with SimPO” skill

Details

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

Similar resources

View all

What you get

  • SKILL.md: installation, baseline training and hyperparameters
  • references/datasets.md, hyperparameters.md and loss-functions.md: data, settings and loss functions

Good fit for

  • You want to align a model more simply than with DPO
  • You lack the memory for a second reference model

Not for

    Tags

    • alignment
    • preferences
    • simpo
    • llm

    About

    You say

    “Train a Mistral 7B with SimPO on my preferred response pairs.”

    Card for the “Aligning a model without a reference model with SimPO” skill

    Skill published by Orchestra Research 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 align an LLM on preferences without a reference model, more simply than DPO
    3. Review and adjust the result

    Content preview

    Files

    • LICENSE1.1 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • references/datasets.md11.1 KB
    • references/hyperparameters.md8.7 KB
    • references/loss-functions.md7.2 KB
    • SKILL.md6 KB

    Origin and licence

    Listed unmodified

    Author
    Orchestra Research
    Source
    github.com/orchestra-research/ai-research-skills/tree/773a52944ba4747a18bd4ae9ade53fff041adcbc/06-post-training/simpo
    Licence
    MIT

    Notice

    « simpo-training » (Orchestra Research, 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.