Llama Factory
Setup
How to Install Llama Factory on Linux, Windows and Docker
Step-by-step install of Llama Factory from source, with Docker, uv and the Windows extras for PyTorch and bitsandbytes.
Read the article →Guide
Llama Factory Guide
Nine articles that follow the order of a real project: install, size the hardware, prepare data, choose a method, train, align, run, export and fix what breaks.
Hardware
GPU and VRAM Requirements for Llama Factory Fine-Tuning
How much GPU memory each Llama Factory training method needs, from full fine-tuning to 2-bit QLoRA, with sizing examples.
Read the article →
Data
Preparing Your Dataset for Llama Factory (Alpaca and ShareGPT)
Learn the two dataset formats Llama Factory reads, how to register a file in dataset_info.json, and how to check it before training.
Read the article →
Training
LoRA vs QLoRA in Llama Factory: Which Should You Use?
A plain comparison of LoRA and QLoRA inside Llama Factory, covering memory, speed, quality and the config keys that control each.
Read the article →
Web UI
Using LLaMA Board: The Llama Factory Web UI
A tour of the Gradio web interface in Llama Factory: choosing a model, setting training options, previewing data, chatting and exporting.
Read the article →
Alignment
DPO, KTO and ORPO Preference Tuning in Llama Factory
What preference tuning does, how DPO, KTO and ORPO differ, and the Llama Factory configs that run each one after supervised fine-tuning.
Read the article →
Inference
Running Inference and Chat With a Llama Factory Model
How to chat with a fine-tuned model in the terminal, the web UI and through an OpenAI-style API, including vLLM and SGLang backends.
Read the article →
Export
Merging LoRA Adapters and Exporting Models in Llama Factory
Merge a trained adapter into the base weights, quantize the result and push it to the Hugging Face Hub using the Llama Factory export command.
Read the article →
Troubleshooting
Fixing Common Llama Factory Errors
Solutions for the errors people hit most in Llama Factory: CUDA out of memory, version mismatches, template problems and Windows issues.
Read the article →