Unsloth

Unsloth is a free, open-source tool for fine-tuning AI models (training them a little more with your own examples so they adapt to a task) using less memory and less time than the usual methods. Thanks to that, you can fine-tune a small or medium-sized model with a single consumer graphics card, without renting big servers.

What usually confuses people is mixing up fine-tuning with using. Ollama or LM Studio are for running a model as it comes; Unsloth is mainly for modifying it, although its recent versions also include an interface for running models locally. And fine-tuning isn't "teaching it your documents": for a model to look up your information there are simpler techniques. Fine-tuning makes sense when you want it to always answer with a specific style, format or judgement. Unsloth also publishes compressed versions of many well-known models, and that's how a lot of people discover it.

Unsloth is one of the four programs in my local AI setup, along with Ollama, LM Studio and llama.cpp, and it's the one I use for fine-tuning models. What each one brings and the real problems each has given me I explain in "Ollama, LM Studio, llama.cpp and Unsloth: which one I use for what". If you're just starting, leave fine-tuning for later: first check whether a good model with good instructions already solves the task for you.