Overview

Llama-factory provides a WebUI-based, no-code approach to fine-tuning LLMs. It supports 100+ models, 2/3/4/5/6/8-bit QLoRA, and multimodal training, with guidance sourced from official documentation to streamline implementation and debugging.

How This Skill Works

Users configure fine-tuning pipelines through a WebUI, select from 100+ supported models, and apply QLoRA bit-widths (2/3/4/5/6/8). The skill derives actionable steps from references in references/ (getting_started, advanced, api/guides) and relies on dependencies like llmtuner, torch, transformers, datasets, peft, accelerate, and gradio to execute training and evaluation.

When to Use It

• Working with llama-factory
• Asking about llama-factory features or APIs
• Implementing llama-factory solutions
• Debugging llama-factory code
• Learning llama-factory best practices

Quick Start

1. Step 1: Review the getting_started.md in references and install dependencies (llmtuner, torch, transformers, datasets, peft, accelerate, gradio)
2. Step 2: Launch the WebUI, select a model from the 100+ catalog, and configure QLoRA bit-widths (2/3/4/5/6/8) with multimodal options
3. Step 3: Run a small fine-tuning task, monitor logs, and use the reference guides to refine hyperparameters

Best Practices

• Start with the getting_started.md and tutorials in the references before building workflows
• Use the appropriate reference files (api, guides, etc.) for detailed feature information
• Read the reference files in references/ for comprehensive explanations and code examples
• Use the view command to read specific reference files when detailed information is needed
• Check code examples, which include language annotations for clearer understanding

Example Use Cases

• Set up a no-code fine-tuning workflow for a new model via the WebUI and validate results
• Experiment with 2/3/4/5/6/8-bit QLoRA configurations to optimize memory usage
• Fine-tune multiple HuggingFace models (Llama, Qwen, Gemma) from the 100+ model catalog
• Diagnose and fix issues in llama-factory code using official docs and reference guides
• Migrate an existing fine-tuning pipeline to llama-factory using getting_started and advanced guides

Frequently Asked Questions