Lakonik/LakonLab
Official PyTorch implementations of AsymFlow, pi-Flow, and GMFlow—flow matching models for text-to-image generation published at arXiv, ICLR 2026, and ICML 2025.
Not currently ranked — collecting fresh signals.
star history
This repository contains the official implementations of three research papers on generative models using flow matching. AsymFlow and pi-Flow focus on accelerating text-to-image diffusion models through asymmetric flow designs and imitation distillation respectively, producing high-quality images in fewer steps. GMFlow extends flow matching with Gaussian mixture formulations. The implementations are built in PyTorch with support for ComfyUI integration and HuggingFace demos.
Frequently asked
- What is Lakonik/LakonLab?
- Official PyTorch implementations of AsymFlow, pi-Flow, and GMFlow—flow matching models for text-to-image generation published at arXiv, ICLR 2026, and ICML 2025.
- Is LakonLab open source?
- Yes — Lakonik/LakonLab is open source, released under the Apache-2.0 license.
- What language is LakonLab written in?
- Lakonik/LakonLab is primarily written in Python.
- How popular is LakonLab?
- Lakonik/LakonLab has 441 stars on GitHub.
- Where can I find LakonLab?
- Lakonik/LakonLab is on GitHub at https://github.com/Lakonik/LakonLab.