← All skills

simpo-training

65 / 100 · Review before installing

Check before installing.

By artubss/SKILLS-CLAUDE-CODE (GitHub) · other · 11 stars · scanned at commit ac67ed8

Simple Preference Optimization para alinhamento de LLMs. Alternativa sem modelo de referência ao DPO com melhor desempenho (+6.4 pontos no AlpacaEval 2.0). Sem modelo de referência necessário, mais eficiente que DPO. Use para alinhamento de preferências quando quer treinamento mais simples e rápido que DPO/PPO.

Works with any agent that supports SKILL.md skills.

What this skill can do

Safer alternatives

Low-risk skills in the same category (other) that look related. Check that one does what you need.

• 100 Low risksimpo-losscxcscmu/SkillLearnBench · 93 stars
• 100 Low riskrun2_simpo_losscxcscmu/SkillLearnBench · 93 stars
• 80 Low riskml-trainingain3sh/.agents · 14 stars

Install

Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.

if [ -e ~/.agents/skills/'simpo-training' ] || [ -L ~/.agents/skills/'simpo-training' ]; then echo 'A skill folder named simpo-training already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/artubss/SKILLS-CLAUDE-CODE' "$d" && git -C "$d" checkout --quiet 'ac67ed85889b15fa9e784e2a93a83f8916870698' -- 'skills/skills/ai-research/post-training-simpo' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/skills/ai-research/post-training-simpo' ~/.agents/skills/'simpo-training' && rm -rf "$d"; fi

Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'simpo-training' ~/.claude/skills/'simpo-training' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.

Badge for your README

The badge for artubss/SKILLS-CLAUDE-CODE shows the score of its lowest-scoring skill.

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/artubss/SKILLS-CLAUDE-CODE.svg)](https://zalavijaysinh.github.io/skilltrust/repos/artubss/SKILLS-CLAUDE-CODE/)

Score history