llamafactory
LlamaFactory streamlines LLM fine-tuning through YAML-driven configs, supporting LoRA, QLoRA, and full-parameter training methods like SFT and DPO. Register datasets, train your model, merge adapters, and export for standalone deployment or API serving. The skill guides you through GPU requirements, base model selection, and inference setup.
LlamaFactory fine-tunes open-weight LLMs via YAML configs, supporting LoRA, QLoRA, and full-parameter methods.
AI-generated summary based on this skill's SKILL.md
Decision gist · record as of 2026-07-27
LlamaFactory fine-tunes open-weight LLMs via YAML configs, supporting LoRA, QLoRA, and full-parameter methods. LlamaFactory streamlines LLM fine-tuning through YAML-driven configs, supporting LoRA, QLoRA, and full-parameter training methods like SFT and DPO. Register datasets, train your model, merge adapters, and export for standalone deployment or API serving. The skill guides you through GPU requirements, base model selection, and inference setup.
Use it when
- LlamaFactory's LoRA adapter setup involves configuring rank, alpha, and target modules in your YAML training config.
- LlamaFactory provides merge and export utilities to combine trained LoRA adapters with base model weights into a standalone model.
Verify before relying
Read SKILL.md below before installing (2 files). Open directory: indexed for reading, not audited.
Install
Prism-Shadow/penguin-harness/llamafactory · repository language: TypeScript
Open directory. Skills are indexed for reading, not audited. Review a skill's body before installing it.
Frequently asked questions
AI-generated answers based on this skill's SKILL.md and metadata
How do I fine-tune a model with LlamaFactory?
LlamaFactory streamlines LLM fine-tuning through YAML-driven configs. Define your base model, dataset, training method (SFT, DPO, or LoRA), and hardware specs in a YAML file, then run the CLI. LlamaFactory handles dataset registration, parameter initialization, and training loops. You can choose full-parameter tuning or memory-efficient LoRA/QLoRA adapters depending on your GPU resources.
What is the LlamaFactory LoRA adapter setup process?
LlamaFactory's LoRA adapter setup involves configuring rank, alpha, and target modules in your YAML training config. Register your dataset in the framework's format (Alpaca, ShareGPT, or custom), specify the base model, set adapter output directory, and launch training. LlamaFactory applies low-rank decomposition to model weights, keeping memory footprint minimal while maintaining adaptation quality.
How do I merge LoRA adapters and export the model?
LlamaFactory provides merge and export utilities to combine trained LoRA adapters with base model weights into a standalone model. Use the export command specifying adapter paths and output format. The merged model can then be deployed via vLLM, Ollama, or other inference servers without requiring adapter files, simplifying production serving.
Can I use LlamaFactory's web UI for no-code training?
Yes, LlamaFactory includes a web UI for no-code model training and parameter tuning. The interface guides you through dataset upload, base model selection, training method choice, and hyperparameter adjustment without writing YAML manually. This makes LlamaFactory accessible to users without CLI experience while maintaining full training flexibility.
What GPU memory does LlamaFactory require for fine-tuning?
LlamaFactory's GPU requirements depend on your training method. Full-parameter training demands substantial VRAM; LoRA and QLoRA drastically reduce memory needs, enabling fine-tuning on consumer GPUs. The framework provides memory optimization guides and supports multi-GPU setups. Your actual requirements vary by model size, batch size, and sequence length configured in YAML.
How do I deploy a fine-tuned model after training?
After training, LlamaFactory supports deployment via inference APIs or interactive chat interfaces. Export your merged model, then serve it using vLLM, Ollama, or LlamaFactory's built-in inference server. Configure API endpoints, batch settings, and quantization as needed. The exported model runs standalone without training infrastructure, ready for production workloads.
SKILL.md
Rendered from the published skill. Quoted content, verbatim.
LlamaFactory Fine-Tuning
LlamaFactory fine-tunes open-weight LLMs (LoRA/QLoRA and full-parameter; SFT, DPO and more) through the llamafactory-cli command driven by YAML configs.
Before you start
If the user's message only invokes this skill (e.g. "use llamafactory skill") without a concrete request, ask the user what they want to fine-tune. Do not run any command until the goal is clear.
Confirm before training:
- GPU memory (
nvidia-smi) — it bounds the model size and method; LoRA needs far less than full fine-tuning. - The base model: a Hugging Face id or a local path.
- The dataset: where it lives and which format it is in.
- The goal: SFT with LoRA is the usual starting point.
Install
```bash git clone --depth 1 https://github.com/hiyouga/LlamaFactory.git cd LlamaFactory pip install -e . pip install -r requirements/metrics.txt # optional: evaluation
(truncated - see the full file via the links below)
File tree — 2 files
packages/skills/skills/llamafactory/SKILL.md
packages/skills/skills/llamafactory/icon.svg
Let your AI agent find skills like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 56,283 agent skills by what they can do, searchable in plain language.
wish › “Fine-tune an open-weight LLM using LoRA or full parameters with YAML configuration”
Give your agent the search over MCP, or paste the wish link into any chat. No install? Search from any chat →
Related skills
Fine-tuning with TRL provides post-training methods to align language models with human preferences through multiple approaches. Train models on instruction data with SFT, optimize for preference alignment via DPO without a separate reward model, or run full RLHF pipelines combining supervised fine-tuning, reward modeling, and PPO optimization. GRPO offers memory-efficient online reinforcement learning for resource-constrained setups.
This skill teaches you to apply reinforcement learning techniques for aligning language models with human preferences. It covers supervised fine-tuning for instruction following, direct preference optimization for preference alignment, PPO and GRPO for reward-based training, and reward model development—all integrated with HuggingFace Transformers.
This skill teaches post-training techniques for aligning language models to human preferences. It covers supervised fine-tuning, direct preference optimization (DPO), and online reinforcement learning methods like RLOO and GRPO, with complete workflows and practical examples.
Unsloth Training accelerates LLM fine-tuning through GRPO (reinforcement learning with reward functions) and SFT (supervised learning with input-output pairs). It cuts VRAM usage by 60% with FP8 training, speeds up mixed-length datasets via automatic packing, and supports vision model tuning, mobile deployment via ExecuTorch, and export to GGUF, Ollama, and vLLM.
Unsloth accelerates LLM fine-tuning on consumer and datacenter GPUs through optimized LoRA and QLoRA training, cutting both speed and memory requirements dramatically. It handles supervised fine-tuning, reinforcement learning with GRPO, vision model adaptation, and TTS training across 300+ model architectures, with direct export to GGUF for deployment on Ollama and llama.cpp.
Ollama lets you deploy and serve open-weight language models on your machine with automatic GPU detection. It exposes an OpenAI-compatible API endpoint, integrates with PenguinHarness for model registration, and supports flexible context window configuration for agent workloads.
More skills vllm (Apache-2.0)