AI Blocks › Build and ship software
LLM fine-tuning guide
Say your model and GPU. Get a fine-tuning plan and code with Unsloth, sized to your memory, ready to export.
LLM fine-tuning guide is a free AI skill for the job "Build and ship software". Give the instructions to an AI assistant that supports the required files and tools, then review its output.
Free · No account · CC BY 4.0 licence · Made by sickn33 · Level: expert · Thinking: high
Made by sickn33. See the original on GitHub; we copied it as it is, with its licence.
What it does
A fine-tuning plan and working code for your model and your graphics card, sized to the memory you have so it does not run out halfway through. Tell it the model, the GPU and the goal. It works through six steps with Unsloth. First, it estimates memory before any code. Then it loads the model and fixes the chat template, a common cause of models that ignore stop tokens. It attaches LoRA adapters, runs the training, and exports to the format your runtime needs, such as GGUF. It covers QLoRA on a 16 GB GPU and reward-based methods like GRPO and DPO. You need Python and a GPU.
You give it
- Base model
- GPU and memory
- Training data
- The goal of the tuned model
- Export target, like GGUF (optional)
- Codebase (optional)
You get
- A memory estimate
- Training code
- A fixed chat template
- A tuning run setup
- Export steps
Use it when
- You want a model that follows your format and have one consumer GPU.
- Training crashes with out of memory part way through.
- Your tuned model repeats the prompt or never stops.
- You need a file you can run locally after training.
How to use it
- Add it to your AI once: download it and add the folder to Claude, ChatGPT or Codex as a skill.
- Ask for the job in your own words. Your AI opens the skill when it fits.
- It works best when your AI can run code.