Repolode
Submit
unslothai

unslothai/notebooks

A collection of 250+ Colab and Jupyter notebooks for fine-tuning and RL training of text, vision, audio, embedding and TTS models with Unsloth.

View on GitHubWebsite
Stars
5,748
Forks
957
Open issues
24
Last commit
6 hours ago

What is it?

This repository hosts ready-to-run notebooks that show how to fine-tune and train large models with Unsloth. They are organized by model family, including Gemma, Qwen, Llama, gpt-oss, Mistral, Phi, DeepSeek, and others, and by training type such as supervised fine-tuning, GRPO and other reinforcement learning methods, DPO, ORPO, and PPO.

Each notebook covers the full workflow: data preparation, training, and inference. Most are published as Google Colab links, so they can be opened and run in a browser. The notebooks also cover non-text modalities, including vision-language models, audio, speech-to-text, text-to-speech, embeddings, and OCR.

The notebooks are generated from a script and the list grows as new models are supported. The repository is licensed under LGPL-3.0.

Key features

  • 01Notebooks covering 250+ models across text, vision, audio, embedding, and TTS
  • 02Fine-tuning and reinforcement learning examples, including GRPO, DPO, ORPO, and PPO
  • 03Each notebook includes data preparation, training, and inference steps
  • 04Most notebooks open directly in Google Colab with one click

Use cases

  • Learning how to fine-tune a specific open model on your own dataset
  • Trying reinforcement learning (GRPO) on a small model in a free Colab session
  • Adapting a vision-language or TTS model to a custom task
  • Finding a starting template for a new model family

Quick start

  1. 1

    Open a notebook from the list in Google Colab

  2. 2

    Browse all notebooks in the Unsloth documentation

Things to know

  • Licensed under LGPL-3.0, a copyleft license with conditions on redistribution.
  • The notebooks depend on the separate Unsloth library, so their behavior follows that project.

In collections

AI video & audio toolsReinforcement learning frameworks & training environments

Topics

#fine-tuning#llm#reinforcement-learning#jupyter-notebook#colab#vision-language#text-to-speech#embeddings#unsloth

FAQ

What is notebooks?
A collection of 250+ Colab and Jupyter notebooks for fine-tuning and RL training of text, vision, audio, embedding and TTS models with Unsloth. This repository hosts ready-to-run notebooks that show how to fine-tune and train large models with Unsloth. They are organized by model family, including Gemma, Qwen, Llama, gpt-oss, Mistral, Phi, DeepSeek, and others, and by training type such as supervised fine-tuning, GRPO and other reinforcement learning methods, DPO, ORPO, and PPO.
Is notebooks free to use? What license does it have?
notebooks is open source under the LGPL-3.0 license. Check the license terms before using it in commercial products.
Is notebooks still maintained?
Actively maintained. The last commit to notebooks was on Oct 22, 2026.
What can I use notebooks for?
Learning how to fine-tune a specific open model on your own dataset Trying reinforcement learning (GRPO) on a small model in a free Colab session Adapting a vision-language or TTS model to a custom task Finding a starting template for a new model family
How do I install notebooks?
Open a notebook from the list in Google Colab Browse all notebooks in the Unsloth documentation
Where can I get notebooks?
The source code and installation instructions are on GitHub: https://github.com/unslothai/notebooks

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects