vturrisi/solo-learn
A PyTorch Lightning library providing state-of-the-art self-supervised learning methods for visual representation learning.
Not currently ranked — collecting fresh signals.
star history
solo-learn is a self-contained library implementing various self-supervised learning algorithms (BYOL, SimCLR, MoCo, DINO, MAE, Barlow Twins, VICReg, SimSiam, SwAV) for unsupervised visual representation learning. It uses PyTorch Lightning for training, supports NVIDIA DALI for data loading, and allows using the trained backbones outside the library. The project provides SOTA implementations with training tricks and supports finetuning, mixing augmentations, and checkpointing.
Frequently asked
- What is vturrisi/solo-learn?
- A PyTorch Lightning library providing state-of-the-art self-supervised learning methods for visual representation learning.
- Is solo-learn open source?
- Yes — vturrisi/solo-learn is open source, released under the MIT license.
- What language is solo-learn written in?
- vturrisi/solo-learn is primarily written in Python.
- How popular is solo-learn?
- vturrisi/solo-learn has 1.6k stars on GitHub.
- Where can I find solo-learn?
- vturrisi/solo-learn is on GitHub at https://github.com/vturrisi/solo-learn.