vturrisi/solo-learn

Health v2 · maintenance only

Adoption not part of the score

1573 stars · 206 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

solo-learn is a Python library of state-of-the-art self-supervised methods for unsupervised visual representation learning, built on PyTorch Lightning. It implements methods like SimCLR, MoCo, BYOL, SwAV, DINO, MAE, and VicReg in a comparable, configurable environment with training tricks and utilities.

Use cases

• train self-supervised models on images without labels
• pretrain a vision backbone with contrastive learning
• compare SSL methods like SimCLR, BYOL, and DINO under the same setup
• learn visual representations for downstream classification
• pretrain MAE or transformer models on ImageNet
• run linear evaluation and finetuning of pretrained backbones

When to choose

• you need many self-supervised vision methods in one consistent framework
• you want PyTorch Lightning-based training with Hydra/OmegaConf configs
• you need ImageNet-scale SSL training with GPU/DALI support

When to avoid

• you need supervised image classification out of the box
• you work outside computer vision (e.g., NLP or audio SSL)
• you want a maintained library with frequent updates

Facets

library · maturity maintenance

machine-learning deep-learning nlp machine-learning deep-learning computer-vision image-processing python self-supervised-learning contrastive-learning pytorch-lightning simclr byol moco swav dino mae vicreg barlow-twins representation-learning nvidia-dali gpu linux docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="vturrisi/solo-learn")