GitHub · #1211

linkedin/Liger-Kernel

Efficient Triton Kernels for LLM Training. On Linux: git clone, python3 -m venv .venv, then pip install -e . 6,664 GitHub stars (#1211 in this directory).

Repository
https://github.com/linkedin/Liger-Kernel
Homepage
https://linkedin.github.io/Liger-Kernel/
Owner
linkedin (Organization)
Stars
6,664
Forks
626
Open issues
195
Language
Python
License
BSD-2-Clause
Created
2024-08-06
Last push
2026-10-17
Topics
finetuning, gemma2, hacktoberfest, llama, llama3, llm-training, llms, mistral, phi3, triton, triton-kernels
Snapshot
2026-10-18 · weekly refresh

Try it on Shadowfetch Linux.

Shadowfetch Linux is a Debian testing desktop with KDE Plasma, local AI tools, and ShadowCode. Clone this repository there, follow the steps for its primary language, and ask ShadowCode to finish the build.

Clone and set up a Python environment.

git clone https://github.com/linkedin/Liger-Kernel.git
cd Liger-Kernel
python3 -m venv .venv
source .venv/bin/activate
pip install -e .

If the project has no installable package, use the README’s requirements file or conda notes instead of pip install -e .

ShadowCode prompt.

Paste this into ShadowCode after you open the project.

Clone https://github.com/linkedin/Liger-Kernel into ~/Workspaces/Liger-Kernel.
Read the README.
On this Shadowfetch Linux desktop, create a Python virtualenv, install from the README, then run or import the project.
Ask before installing packages or using the network.

Download Shadowfetch Linux — ISO, checksum and signature.

Back to the GitHub directory · Methodology