Artificial intelligence training in low memory
Soup is a Python library that allows fine-tuning of large language models via a single YAML file. It can train models with 8 billion parameters using the layer streaming method on laptop graphics processors with 4 GB memory.
Updates
- October 15, 2026: Stars 8,615 → 9,069, latest release v0.76.1 (October 14, 2026).
- October 12, 2026: Stars 7,988 → 8,615, latest release v0.76.0 (October 12, 2026).
- October 2, 2026: Stars 7,309 → 7,988, latest release v0.75.2 (October 1, 2026).
- September 27, 2026: Stars 6,267 → 7,309, latest release v0.75.1 (September 21, 2026).
What you get
- You can train models with 8 billion parameters on laptops with 4 GB graphics memory.
- With the layer flow method, you do not have to deal with complex installations by overcoming hardware restrictions.
- You can manage the entire training process through a single configuration file.
Installation
pip install "soup-cli[train]"pip install "soup-cli[all]"Running it
soup init --template chat
soup trainIf you don't write code
I want to train an artificial intelligence model with 8 billion parameters using the Soup library on my computer with 4 GB graphics memory. Help me create a YAML configuration file that enables layer streaming and uses 4-bit quantization. Prepare the content of the soup.yaml file required to start the training process, and then explain step by step how to start the training using this file.
Repository: https://github.com/MakazhanAlpamys/Soup. Follow the official instructions in the README of this repository for setup and execution.
Before starting, check which OS we are on and if the necessary tools (e.g., Python, Node.js, Git, Docker) are installed. Install any missing ones step-by-step according to this system.
Create a new, empty folder to work in. Do not touch any files outside this folder and do not delete anything. Tell me what you are going to do in one sentence before running each command.
Tell me what I should see on the screen at the end of each step. If an error occurs, I will paste the error message to you exactly as it is, and we will fix it together.
Related dictionary terms
Links
TreScout did not build this tool · we found it in GitHub trends and wrote it up. This page describes the repository as of 2026-08-16: The star count and our text belong to that day, the repository may have changed since. Check the repository link for the current state. The introductory text was prepared with AI from the repository's own documentation, and this page was machine-translated from the Turkish original · the Turkish version prevails.