facebookresearch/swav

Health v2 · maintenance only

Flags: no_releases archived no_license

Adoption not part of the score

2096 stars · 286 forks observed · 2026-08-28

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

PyTorch implementation of SwAV, a self-supervised method for pre-training convolutional networks without labels by swapping cluster assignments between augmented image views. Includes pretrained ResNet-50 models and training scripts for ImageNet-scale experiments.

Use cases

• pretrain a vision backbone without labels
• download SwAV pretrained ResNet-50 weights
• reproduce self-supervised learning research results
• replace supervised ImageNet backbone with self-supervised features
• train self-supervised models with large or small batches

When to choose

• you need label-free visual representation pretraining
• you want efficient self-supervised training without large memory banks or momentum encoders
• you want a strong pretrained ResNet-50 backbone for transfer learning

When to avoid

• you need transformer-based self-supervised methods like DINO or MAE
• you need a production-ready maintained library rather than research code
• you work outside PyTorch

Facets

library · maturity maintenance

machine-learning deep-learning computer-vision machine-learning deep-learning python self-supervised-learning contrastive-learning pytorch pretrained-models representation-learning research-code gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/swav")