skills.rest
NagyVikt
NagyViktCOMMUNITY
Explore all skills by this author →
Website𝕏@NagyViktordpGitHub

huggingface-llm-trainer

Submit SFT, DPO, and GRPO training jobs to Hugging Face infrastructure.

2|Updated Mar 29, 2026
Search Tags:#llm#fine-tuning#gpu#training#huggingface#gguf#trl
One-click install
npx skills add https://github.com/NagyVikt/codex-plugins --skill huggingface-llm-trainer-nagyvikt
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: huggingface-llm-trainer
Source: https://github.com/NagyVikt/codex-plugins/tree/main/plugins/hugging-face/skills/llm-trainer
Command: npx skills add https://github.com/NagyVikt/codex-plugins --skill huggingface-llm-trainer-nagyvikt

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires transformers, peft, torch, accelerate, huggingface_hub, sentencepiece, protobuf, numpy, gguf, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill removes the complexity and hardware requirements of training language models by providing a streamlined, managed workflow for fine-tuning models on cloud GPUs.

Core Features & Use Cases

  • Managed Training: Execute SFT, DPO, and GRPO training jobs on Hugging Face infrastructure without local GPU setup.
  • GGUF Conversion: Automatically convert trained models into GGUF format for local deployment in Ollama or LM Studio.
  • Validation & Monitoring: Includes automated dataset format validation and real-time training metrics via Trackio.

Quick Start

Submit a training job to fine-tune the Qwen 0.5B model on the Capybara dataset using the huggingface-llm-trainer skill.

Frequently Asked Questions about huggingface-llm-trainer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fine-tune an LLM on cloud GPUs without setting up local hardware?▼

You can fine-tune an LLM on cloud GPUs by submitting managed training jobs to Hugging Face infrastructure. This removes local hardware requirements by handling SFT, DPO, and GRPO workflows remotely using TRL methods.

How do I convert a fine-tuned Hugging Face model to GGUF format for local inference?▼

Converting a fine-tuned model to GGUF format is handled automatically after training. The workflow outputs GGUF files directly, enabling local deployment in environments like Ollama or LM Studio.

Do I need Hugging Face Hub authentication to run managed training jobs?▼

Yes, Hugging Face Hub authentication is required to run managed training jobs. Validated dataset formats and proper authentication ensure successful cloud-based persistence of your trained models.

What is the difference between SFT, DPO, and GRPO training workflows?▼

SFT, DPO, and GRPO are distinct TRL training workflows supported for managed LLM fine-tuning. They represent different reinforcement learning and supervised techniques for adapting language models to specific datasets.

How can I monitor LLM training metrics when running jobs on Hugging Face infrastructure?▼

You can monitor LLM training metrics using the integrated Trackio tool. It provides real-time tracking of your managed cloud-based training jobs alongside automated dataset format validation.

Can I use peft and transformers to fine-tune models without a local GPU?▼

Yes, you can use peft and transformers without a local GPU by submitting jobs to Hugging Face infrastructure. The managed service executes the training remotely, eliminating local hardware dependencies.

Related Skills

Explore More Software Engineering Skills→
hongmaple0820hongmaple0820
9

hugging-face-model-trainer

Train language models on Hugging Face Jobs using TRL workflows.

Community
Advanced
afel6afel6

gguf-quantization

Convert and quantize transformer models into GGUF format for CPU inference.

Community
Advanced
Aniket-a14Aniket-a14
1

hugging-face-model-trainer

Train language models with TRL on Hugging Face Jobs.

Community
Advanced
zhouboyu-xrealzhouboyu-xreal

axolotl

Fine-tune large language models with Axolotl using LoRA, DPO, GRPO, and multimodal training.

Community
Advanced