skills.rest
Clay-HHK
Clay-HHKCOMMUNITY
Explore all skills by this author →
GitHub

llama-factory

Fine-tune LLMs via WebUI/CLI with LoRA/QLoRA workflows and adapter merging.

2|Updated Apr 12, 2026
Search Tags:#llm#quantization#lora#webui#finetuning#qlora#llama-factory
One-click install
npx skills add https://github.com/Clay-HHK/claude-config --skill llama-factory-clay-hhk
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: llama-factory
Source: https://github.com/Clay-HHK/claude-config/tree/main/skills/AI-research-SKILLs/03-fine-tuning/llama-factory
Command: npx skills add https://github.com/Clay-HHK/claude-config --skill llama-factory-clay-hhk

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

LLaMA-Factory streamlines the complex and error-prone process of fine-tuning large language models by providing a unified WebUI and CLI that support LoRA/QLoRA, quantization, adapter merging, multimodal inputs, and distributed or NPU training, reducing setup friction and operational overhead.

Core Features & Use Cases

  • No-code WebUI & CLI: Launch training, evaluation, export, and inference workflows without deep engineering effort.
  • Flexible Fine-tuning Methods: Supports LoRA, QLoRA in multiple bit widths (2/3/4/5/6/8-bit), full-parameter, freeze, and advanced adapters like LoRA+, PiSSA, DoRA.
  • Production Deployment Workflows: Merge adapters, export quantized models for vLLM or transformers inference, and use distributed, DeepSpeed or FSDP training across GPUs and NPUs.
  • Use Case: Quickly run a QLoRA experiment on a target model, monitor metrics with LlamaBoard or Wandb, merge the adapter into a deployable model, and export a quantized artifact for inference.

Quick Start

Use the LLaMA-Factory WebUI to run a QLoRA fine-tuning job for your chosen model and dataset.

Frequently Asked Questions about llama-factory

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

FAQPage Schema
How do I fine-tune a large language model without writing code?▼

You can perform no-code LLM fine-tuning using the LLaMA-Factory WebUI to launch training, evaluation, and export workflows without deep engineering effort.

What QLoRA bit widths are supported for LLM quantization?▼

LLM quantization supports multiple QLoRA bit widths including 2, 3, 4, 5, 6, and 8-bit configurations, alongside full-parameter, freeze, and advanced adapters like LoRA+, PiSSA, and DoRA.

Can I use distributed GPU and NPU training for LLM fine-tuning?▼

Distributed GPU and NPU training are supported for LLM fine-tuning, allowing you to utilize DeepSpeed or FSDP across multiple devices to scale your training workloads.

How do I merge LoRA adapters and export a quantized model for inference?▼

You can merge LoRA adapters into a deployable base model and export a quantized artifact for inference using either the vLLM or transformers backends via the CLI or WebUI.

Does LLaMA-Factory integrate with HuggingFace and ModelScope for LLM training?▼

HuggingFace and ModelScope integration is supported for LLM training, allowing you to seamlessly load target models and datasets for your QLoRA or full-parameter experiments.

What is the best way to monitor LLM fine-tuning metrics during a run?▼

Monitor LLM fine-tuning metrics using LlamaBoard or Wandb, which track evaluation stats and training progress while you run QLoRA experiments through the WebUI or CLI.

Related Skills

Explore More Software Engineering Skills→
ombhanushaliiiombhanushaliii
1

Model Fine-Tuning

Upload datasets and configure training on distributed GPU resources.

Community
Intermediate
DaddyElonMusk69DaddyElonMusk69

serving-llms-vllm

Serve large language models with high throughput using vLLM.

Community
Advanced
Estrella-231Estrella-231

pytorch-lightning

Organize PyTorch training into LightningModule, LightningDataModule, and Trainer abstractions.

Community
Intermediate
eliottbusinesseliottbusiness

axolotl

Fine-tune large language models with Axolotl using YAML configurations.

Community
Advanced