New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Simpo

by Luciole-Studio in Luciole-Studio/Misaka-Agent

Reference-free preference alignment, simpler than DPO. An agent skill from Luciole-Studio/Misaka-Agent.

MITAuto-check passedAI & LLM Engineering

Install Simpo

skills CLI
$ npx skills add Luciole-Studio/Misaka-Agent --skill simpo -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install Luciole-Studio/Misaka-Agent simpo --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/Luciole-Studio/Misaka-Agent.git skills-src && mkdir -p .claude/skills && cp -r skills-src/misaka/core/skills/assets/optional/mlops/simpo .claude/skills/simpo && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
simpo
GitHub stars
139
Used in
1 other repo
Token cost
~1.4k tokens
SKILL.md length
275 words
Files
4 (incl. references)
Skills in repo
76
Repo updated
First seen
Licence
MIT

At a glance

Reference-free preference alignment, simpler than DPO. An agent skill from Luciole-Studio/Misaka-Agent.

  • Tasks that involve MLOps
  • SKILL.md covers Quick start, Common workflows, When to use vs alternatives and Common issues, plus 3 more sections
  • Calls conda, python and git; reaches github.com and pytorch.org
  • Tasks that involve Fine-tuning

What it does

Simpo is an agent skill from Luciole-Studio/Misaka-Agent. Reference-free preference alignment, simpler than DPO.

Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 4 other files, including reference files (for example `references/datasets.md`, `references/hyperparameters.md` and `references/loss-functions.md`).

It sits in AI & LLM Engineering, covering MLOps and Fine-tuning. It works with Mistral AI. The repository describes itself as: A multi-agent research system for the humanities and social sciences. The licence is MIT.

When your agent uses it

  • Tasks that involve MLOps
  • Tasks that involve Fine-tuning

Example prompts

  • “/simpo”

Requirements

  • Python 3

What it can do on your machine

Read from SKILL.md and the folder at commit b94464a. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • conda
    • python
    • git

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Hosts in commands or code, which the agent is likely to contact:

    • github.com
    • pytorch.org

    Also links to:

    • arxiv.org
    • huggingface.co

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Simpo loads about 1.4k tokens when it runs, and up to ~8.1k if it reads all its reference files. Until then it costs about 15 tokens; SKILL.md has 275 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~15
When it runs · the whole SKILL.md, loaded when a task matches
~1.4k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~8.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from Luciole-Studio/Misaka-Agent at commit b94464a, republished under its MIT licence (© Luciole-Studio). 275 words, ~1,437 tokens.

Download SKILL.mdSave it as .claude/skills/simpo/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.
name
simpo
description
Reference-free preference alignment, simpler than DPO.
version
1.0.0
author
Orchestra Research
license
MIT
dependencies
torch, transformers, datasets, trl, accelerate
platforms
linux, macos, windows

SimPO - Simple Preference Optimization

Quick start

SimPO is a reference-free preference optimization method that outperforms DPO without needing a reference model.

Installation:

bash
# Create environment
conda create -n simpo python=3.10 && conda activate simpo

# Install PyTorch 2.2.2
# Visit: https://pytorch.org/get-started/locally/

# Install alignment-handbook
git clone https://github.com/huggingface/alignment-handbook.git
cd alignment-handbook
python -m pip install .

# Install Flash Attention 2
python -m pip install flash-attn --no-build-isolation

Training (Mistral 7B):

bash
ACCELERATE_LOG_LEVEL=info accelerate launch \
  --config_file accelerate_configs/deepspeed_zero3.yaml \
  scripts/run_simpo.py \
  training_configs/mistral-7b-base-simpo.yaml

Common workflows

Workflow 1: Train from base model (Mistral 7B)

Config (mistral-7b-base-simpo.yaml):

yaml
# Model
model_name_or_path: mistralai/Mistral-7B-v0.1
torch_dtype: bfloat16

# Dataset
dataset_mixer:
  HuggingFaceH4/ultrafeedback_binarized: 1.0
dataset_splits:
  - train_prefs
  - test_prefs

# SimPO hyperparameters
beta: 2.0                  # Reward scaling (2.0-10.0)
gamma_beta_ratio: 0.5       # Target margin (0-1)
loss_type: sigmoid          # sigmoid or hinge
sft_weight: 0.0             # Optional SFT regularization

# Training
learning_rate: 5e-7         # Critical: 3e-7 to 1e-6
num_train_epochs: 1
per_device_train_batch_size: 1
gradient_accumulation_steps: 8

# Output
output_dir: ./outputs/mistral-7b-simpo

Launch training:

bash
accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml \
  scripts/run_simpo.py training_configs/mistral-7b-base-simpo.yaml
Workflow 2: Fine-tune instruct model (Llama 3 8B)

Config (llama3-8b-instruct-simpo.yaml):

yaml
model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct

dataset_mixer:
  argilla/ultrafeedback-binarized-preferences-cleaned: 1.0

beta: 2.5
gamma_beta_ratio: 0.5
learning_rate: 5e-7
sft_weight: 0.1             # Add SFT loss to preserve capabilities

num_train_epochs: 1
per_device_train_batch_size: 2
gradient_accumulation_steps: 4
output_dir: ./outputs/llama3-8b-simpo

Launch:

bash
accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml \
  scripts/run_simpo.py training_configs/llama3-8b-instruct-simpo.yaml
Workflow 3: Reasoning-intensive tasks (lower LR)

For math/code tasks:

yaml
model_name_or_path: deepseek-ai/deepseek-math-7b-base

dataset_mixer:
  argilla/distilabel-math-preference-dpo: 1.0

beta: 5.0                   # Higher for stronger signal
gamma_beta_ratio: 0.7       # Larger margin
learning_rate: 3e-7         # Lower LR for reasoning
sft_weight: 0.0

num_train_epochs: 1
per_device_train_batch_size: 1
gradient_accumulation_steps: 16

When to use vs alternatives

Use SimPO when:

  • Want simpler training than DPO (no reference model)
  • Have preference data (chosen/rejected pairs)
  • Need better performance than DPO
  • Limited compute resources
  • Single-node training sufficient

Algorithm selection:

  • SimPO: Simplest, best performance, no reference model
  • DPO: Need reference model baseline, more conservative
  • PPO: Maximum control, need reward model, complex setup
  • GRPO: Memory-efficient RL, no critic

Use alternatives instead:

  • OpenRLHF: Multi-node distributed training, PPO/GRPO
  • TRL: Need multiple methods in one framework
  • DPO: Established baseline comparison

Common issues

Issue: Loss divergence

Reduce learning rate:

yaml
learning_rate: 3e-7  # Reduce from 5e-7

Reduce beta:

yaml
beta: 1.0  # Reduce from 2.0

Issue: Model forgets capabilities

Add SFT regularization:

yaml
sft_weight: 0.1  # Add SFT loss component

Issue: Poor preference separation

Increase beta and margin:

yaml
beta: 5.0            # Increase from 2.0
gamma_beta_ratio: 0.8  # Increase from 0.5

Issue: OOM during training

Reduce batch size:

yaml
per_device_train_batch_size: 1
gradient_accumulation_steps: 16  # Maintain effective batch

Enable gradient checkpointing:

yaml
gradient_checkpointing: true

Advanced topics

Loss functions: See references/loss-functions.md for sigmoid vs hinge loss, mathematical formulations, and when to use each.

Hyperparameter tuning: See references/hyperparameters.md for beta, gamma, learning rate selection guide, and model-size-specific recommendations.

Dataset preparation: See references/datasets.md for preference data formats, quality filtering, and custom dataset creation.

Hardware requirements

  • GPU: NVIDIA A100/H100 recommended
  • VRAM:
    • 7B model: 1× A100 40GB (DeepSpeed ZeRO-3)
    • 8B model: 2× A100 40GB
    • 70B model: 8× A100 80GB
  • Single-node: DeepSpeed ZeRO-3 sufficient
  • Mixed precision: BF16 recommended

Memory optimization:

  • DeepSpeed ZeRO-3 (default config)
  • Gradient checkpointing
  • Flash Attention 2

Resources

© Luciole-Studio, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (references) in misaka/core/skills/assets/optional/mlops/simpo of Luciole-Studio/Misaka-Agent.

  • SKILL.md
  • references/datasets.md
  • references/hyperparameters.md
  • references/loss-functions.md

Open the folder on GitHubat commit b94464a

Used in 1 other repository

We found 1 copy of this SKILL.md (exact, near-identical or edited) in other folders, from 1 other GitHub owner. This page covers the copy in Luciole-Studio/Misaka-Agent, which our catalogue first saw on October 7, 2026.

Compare with similar skills

Simpo next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Simpo compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Simpo this skillLuciole-Studio/Misaka-Agent1391 repos~1.4kAutomated safety check: PassMIT
AWS AI MLaws/agent-toolkit-for-aws2.8k—~1.7kAutomated safety check: PassApache-2.0
Vertex AI Geminimajiayu000/claude-skill-registry6661 repos~2.1kAutomated safety check: PassApache-2.0
Open Weightsericrisco/rsc-harness167—~4.1kAutomated safety check: PassMIT
ML Engineermajiayu000/claude-skill-registry6661 repos~2.8kAutomated safety check: PassMIT
Safactory WorkflowsAI45Lab/SAfactory236—~1.8kAutomated safety check: PassNone

Similar skills

  • AWS AI ML

    aws/agent-toolkit-for-aws

    Official

    Selects, deploys, and customizes AI models on Amazon SageMaker.

    2.8k GitHub stars~1.7k tokensUpdated today
    AI & LLM EngineeringAuto-check passed

  • Vertex AI Gemini

    majiayu000/claude-skill-registry

    Google Cloud Vertex AI for enterprise Gemini deployments — production scaling, fine-tuning, and MLOps.

    666 GitHub starsUsed in 1 repo~2.1k tokens
    AI & LLM EngineeringAuto-check passed

  • Open Weights

    ericrisco/rsc-harness

    A skill your agent uses when choosing an open-weight LLM and clearing it for use — which family and size fit the task, the hardware and the budget, and above all whether the license permits shipping.

    167 GitHub stars~4.1k tokensUpdated today
    AI & LLM EngineeringAuto-check passed

  • ML Engineer

    majiayu000/claude-skill-registry

    Expert in building scalable ML systems, from data pipelines and model training to production deployment and monitoring.

    666 GitHub starsUsed in 1 repo~2.8k tokens
    DevOps & CloudAuto-check passed

  • Safactory Workflows

    AI45Lab/SAfactory

    Integrate a benchmark or custom environment into SAfactory using fixed adapter templates and local contract tests, optionally run Docker/RJob evaluation, or prepare GRPO/RL training.

    236 GitHub stars~1.8k tokensUpdated 14 days ago
    AI & LLM EngineeringAuto-check passed

  • SimPO Preference Training

    Orchestra-Research/AI-Research-SKILLs

    Walks through aligning language models with SimPO, a reference-free preference optimization method, using accelerate configs for Mistral 7B, Llama 3 8B and math-focused models.

    13k GitHub starsUsed in 5 repos~1.5k tokens
    AI & LLM EngineeringAuto-check passed

More from Luciole-Studio/Misaka-Agent

All 76 skills in this repo

  • Kanban Video Orchestrator

    Luciole-Studio/Misaka-Agent

    Plan and run multi-agent video production pipelines. An agent skill from Luciole-Studio/Misaka-Agent.

    139 GitHub starsUsed in 2 repos~2.4k tokens
    Auto-check: notes

  • Ast Grep

    Luciole-Studio/Misaka-Agent

    AST-aware structural code search and rewrite via ast-grep. An agent skill from Luciole-Studio/Misaka-Agent.

    139 GitHub starsUsed in 1 repo~3.2k tokens
    Auto-check passed

  • Drug Discovery

    Luciole-Studio/Misaka-Agent

    Drug discovery: ChEMBL search, drug-likeness, interactions. An agent skill from Luciole-Studio/Misaka-Agent.

    139 GitHub starsUsed in 1 repo~2.2k tokens
    Auto-check passed

  • Fitness Nutrition

    Luciole-Studio/Misaka-Agent

    Workout planning, macros, and body metrics via wger/USDA. An agent skill from Luciole-Studio/Misaka-Agent.

    139 GitHub starsUsed in 1 repo~2.4k tokens
    Auto-check passed

  • Hyperframes

    Luciole-Studio/Misaka-Agent

    Render MP4/WebM videos from HTML compositions. An agent skill from Luciole-Studio/Misaka-Agent.

    139 GitHub starsUsed in 1 repo~3.9k tokens
    Auto-check passed

  • Osint Investigation

    Luciole-Studio/Misaka-Agent

    Follow the money via public records and sanctions data. An agent skill from Luciole-Studio/Misaka-Agent.

    139 GitHub starsUsed in 1 repo~2.9k tokens
    Auto-check passed

Works with

Categories

Questions about Simpo

What does Simpo do?

Reference-free preference alignment, simpler than DPO. An agent skill from Luciole-Studio/Misaka-Agent. Simpo is an agent skill from Luciole-Studio/Misaka-Agent. Reference-free preference alignment, simpler than DPO.

When should I use Simpo?

Simpo fits situations like: tasks that involve MLOps; tasks that involve Fine-tuning.

How do I install Simpo in Claude Code?

Run `npx skills add Luciole-Studio/Misaka-Agent --skill simpo -a claude-code`. Or copy the skill folder (misaka/core/skills/assets/optional/mlops/simpo in Luciole-Studio/Misaka-Agent) into .claude/skills/simpo in your project. Claude Code loads it when a task matches its description.

How do I install Simpo in Codex?

Run `npx skills add Luciole-Studio/Misaka-Agent --skill simpo -a codex`. Or copy the skill folder (misaka/core/skills/assets/optional/mlops/simpo in Luciole-Studio/Misaka-Agent) into .agents/skills/simpo in your project. Codex loads it when a task matches its description.

Can I use Simpo in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add Luciole-Studio/Misaka-Agent --skill simpo -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/simpo, .gemini/skills/simpo, .github/skills/simpo and .opencode/skills/simpo in your project.

What does Simpo need to run?

Going by SKILL.md and its folder, Simpo needs the command-line tools its instructions call (conda, python and git). Our summary lists: Python 3.

Does Simpo access the network?

SKILL.md names 4 domains. In commands or code: github.com and pytorch.org; the agent is likely to contact these when it follows the instructions. As links in the text: arxiv.org and huggingface.co. This is read from the text; nothing was executed.

Is Simpo safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Simpo use?

Simpo is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Simpo use?

About 1.4k tokens (SKILL.md is roughly 5.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 6.6k tokens, read only when the agent opens those files.

What are the alternatives to Simpo?

Skills that share tags, products or a category with Simpo: AWS AI ML (aws/agent-toolkit-for-aws, 2.8k stars), Vertex AI Gemini (majiayu000/claude-skill-registry, 666 stars), Open Weights (ericrisco/rsc-harness, 167 stars) and ML Engineer (majiayu000/claude-skill-registry, 666 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Simpo?

Luciole-Studio (a GitHub organization) maintains it in Luciole-Studio/Misaka-Agent, which has 139 GitHub stars. The repository holds 76 skills in this directory. The repository was last updated on October 8, 2026.

Source: Luciole-Studio/Misaka-Agent on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.