Hugging Face · Agent Skill

trl-training

Post-train LLMs with TRL (Transformers Reinforcement Learning) — SFT, DPO, GRPO, KTO, and reward-model training. Use when writing or debugging training code with the TRL Python API or the trl CLI.

Published by Hugging FaceApache-2.077 lines in SKILL.md

Source: huggingface/skills/skills/trl-training
Repository owner: huggingface

Install

Read the SKILL.md and any scripts before installing: a skill runs with your agent's permissions. This copies just this skill into Claude Code's user skills folder; other agents read skills from their own folder.

Claude Code (user-level)

git clone --depth 1 --filter=blob:none --sparse https://github.com/huggingface/skills.git /tmp/skills
cd /tmp/skills && git sparse-checkout set "skills/trl-training"
mkdir -p ~/.claude/skills && cp -r "skills/trl-training" ~/.claude/skills/trl-training

Skills folder docs:

Claude Code Codex Gemini CLI OpenCode Cursor GitHub Copilot goose OpenHands Amp Factory Droid Junie Kiro Roo Code Mistral Vibe pi Hermes Agent OpenClaw Letta

Works with

Agent Skills is an open format, so this skill loads in any harness that supports it, including Claude Code, Codex, Gemini CLI, OpenCode, Cursor, GitHub Copilot, goose, OpenHands. Some skills are written for one product and say so in their description.

More skills from Hugging Face

hf-cli hf-cloud-aws-context-discovery hf-cloud-python-env-setup hf-cloud-sagemaker-deployment-planner hf-cloud-sagemaker-iam-preflight hf-cloud-sagemaker-production-defaults hf-cloud-serving-image-selection hf-mem huggingface-best huggingface-community-evals huggingface-datasets huggingface-gradio

Reviewed Oct 5, 2026. Name and description are the skill's own frontmatter.