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.
- License
- Apache-2.0
- SKILL.md tokens
- 1,077
- Bundled files
- 0
- Bundled tokens (estimated)
- 0
Cost per 1,000 loads
Estimated input cost per 1,000 loads of SKILL.md only (characters ÷ 4). Bundled files, output tokens and execution are excluded.
| Model | Input / 1M tokens | 1,000 loads |
|---|---|---|
| Claude Opus 5 | $5.00 | $5.38 |
| Claude Sonnet 5 | $2.00 | $2.15 |
| Claude Haiku 4.5 | $1.00 | $1.08 |
Safety flags
Detected automatically from files; not a security audit.
- Runs scripts
- No
- Network access
- Yes
- Shell commands
- No
Works with
Open SKILL.md format
- Claude Code
- Codex
- Cursor
- GitHub Copilot
- Gemini CLI
- OpenCode
- Goose
Install
npx skills add huggingface/skills --skill trl-trainingLast verified:
trl-training FAQ
What does the trl-training skill do?
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. Source: https://github.com/huggingface/skills/blob/main/skills/trl-training/SKILL.md
What does the trl-training skill cost to load?
SKILL.md is about 1,077 tokens. Estimated input cost per 1,000 loads: Claude Opus 5 $5.38; Claude Sonnet 5 $2.15; Claude Haiku 4.5 $1.08. Estimated input cost per 1,000 loads of SKILL.md only (characters ÷ 4). Bundled files, output tokens and execution are excluded. Prices: https://docs.anthropic.com/en/docs/about-claude/pricing
How do I install the trl-training skill?
Skills CLI: npx skills add huggingface/skills --skill trl-training. Claude Code: /plugin marketplace add huggingface/skills, then install the plugin from that marketplace. Download the trl-training skill folder as a ZIP and upload it in Claude.ai skills settings.
Is the trl-training skill safe?
Automatically detected flags: network access. Flags are detected automatically from the skill's files and are not a security audit; review the SKILL.md before installing.