Train TRL-based language models on Hugging Face Jobs with cloud GPUs and GGUF conversion.
npx skills add https://github.com/Nymbo/Skills --skill model-trainer
Please help me install this Agent Skill. Skill: model-trainer Source: https://github.com/Nymbo/Skills/tree/main/model-trainer Command: npx skills add https://github.com/Nymbo/Skills --skill model-trainer
This skill automates cloud-based TRL training workflows on Hugging Face Jobs, enabling SFT, DPO, GRPO and reward modeling training without local GPU infrastructure, plus GGUF conversion for local deployment.
Submit a complete training job via hf_jobs() using the template script (e.g., scripts/train_sft_example.py) for your model, dataset, and preferred hardware. The job will run on cloud GPUs, monitor via Trackio, and push results to the Hub.
High-intent search queries and answers about installing and using this skill.
Train language models using TRL on Hugging Face Jobs, which handles cloud GPU provisioning, monitoring via Trackio, and automatic model persistence to the Hub. Submit training scripts for SFT, DPO, GRPO, or reward-model workflows directly through hf_jobs().
Yes, GGUF conversion is included in the end-to-end workflow, enabling local inference with Ollama or llama.cpp after cloud training completes. The skill handles conversion automatically as part of model persistence.
The skill supports SFT (Supervised Fine-Tuning), DPO (Direct Preference Optimization), GRPO (Group Relative Policy Optimization), and reward-model training, all executable on cloud GPUs with hardware selection and cost estimation included.
Dataset validation is built into the workflow to ensure format compatibility before cloud execution. The skill verifies dataset structure as part of the end-to-end pipeline, reducing failed job submissions.
Trackio integration provides real-time monitoring of training metrics and resource usage, while built-in cost estimation helps forecast GPU expenses. Hub authentication ensures secure result persistence and monitoring continuity.
Template scripts (e.g., scripts/train_sft_example.py) are provided for common workflows; customize model, dataset, and hardware selection without writing scripts from scratch. PEP 723 compliant dependencies headers simplify execution.
DMT123Train and fine-tune language models with TRL on Hugging Face Jobs infrastructure.
evalstateSubmit TRL training jobs on Hugging Face Jobs and convert models to GGUF.
huggingfaceTrain and fine-tune language models on Hugging Face Jobs using TRL methods.
luokai25Automate LLM training on Hugging Face Jobs with TRL methods.
