Huggingface Llm Trainer

Train language models on Hugging Face's cloud GPUs without setting up local infrastructure. This handles the full pipeline: SFT, DPO, GRPO, and reward modeling through TRL, plus GGUF conversion for local deployment in Ollama or llama.cpp. The workflow uses UV scripts with PEP 723 inline dependencies, which keeps things clean and self-contained. It includes Trackio monitoring by default and handles the critical parts people usually forget, like making sure your HF_TOKEN gets passed through so trained models actually push to the Hub instead of vanishing when the ephemeral training environment shuts down. Also covers Unsloth integration if you need 60% less VRAM or are training vision-language models where memory gets tight.

skills.sh installs: 2.4K
GitHub stars: 11.3K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add huggingface/skills --skill huggingface-llm-trainer --agent claude-code

Installs into .claude/skills of the current project.

Files

TRL Training on Hugging Face Jobs

Overview

Train language models using TRL (Transformer Reinforcement Learning) on fully managed Hugging Face infrastructure. No local GPU setup required—models train on cloud GPUs and results are automatically saved to the Hugging Face Hub.

TRL provides multiple training methods:

SFT (Supervised Fine-Tuning) - Standard instruction tuning
DPO (Direct Preference Optimization) - Alignment from preference data
GRPO (Group Relative Policy Optimization) - Online RL training
Reward Modeling - Train reward models for RLHF

For detailed TRL method documentation:

hf_doc_search("your query", product="trl")
hf_doc_fetch("https://huggingface.co/docs/trl/sft_trainer")  # SFT
hf_doc_fetch("https://huggingface.co/docs/trl/dpo_trainer")  # DPO
# etc.

See also: references/training_methods.md for method overviews and selection guidance