---
license: cc-by-nc-4.0
library_name: timm
tags:
  - vision
  - self-supervised-learning
  - image-classification
  - feature-extraction
  - vit
datasets:
  - ILSVRC/imagenet-1k
pipeline_tag: image-feature-extraction
---

<h1 style="font-size: 2.5em; text-align: center;">VISReg: Variance-Invariance-Sketching Regularization for JEPA training</h1>

<p align="center">
    <a href="https://arxiv.org/abs/2606.02572v1"><img src="https://img.shields.io/badge/arXiv-2606.02572-b31b1b.svg" alt="arXiv"></a>
    <a href="https://haiyuwu.github.io/visreg/"><img src="https://img.shields.io/badge/Project-Page-blue" alt="Project Page"></a>
    <a href="https://github.com/HaiyuWu/visreg"><img src="https://img.shields.io/badge/GitHub-Code-black?logo=github" alt="GitHub"></a>
</p>

**Key results:**
- 💪 **Strong collapse prevention**: High gradient when embedding collapse
- ⚡ **Friendly to scale training**: Linear complexity to scaling factors
- 🧩 **Easy to train**: Similar to LeJEPA, it is a heuristic-free method
- 🏆 **Best OOD performance**: Achieve the best accuracy on 6 OOD datasets
- 📉 **Data efficiency**: Achieving a similar average accuracy to DINOv2 with 90% less data
- 🧬 **Robust to low-quality datasets**: It is robust to long-tailed and sparse datasets

<h2 style="font-size: 1.8em;">Available Checkpoints</h2>

| File | Architecture | Patch Size | Embed Dim | Params | Pre-training Data |
|------|-------------|------------|-----------|--------|-------------------|
| `visreg-vit-b-inet1k.pth` | ViT-Base | 16 | 768 | 86M | ImageNet-1K |
| `visreg-vit-l-inet1k.pth` | ViT-Large | 14 | 1024 | 304M | ImageNet-1K |

<h2 style="font-size: 1.8em;">Usage</h2>

<h3 style="font-size: 1.4em;">Load with timm</h3>

```python
import timm
import torch

# ViT-Base/16
model = timm.create_model("vit_base_patch16_224", pretrained=False, num_classes=0, dynamic_img_size=True)
state_dict = torch.load("visreg-vit-b-inet1k.pth", map_location="cpu")
model.load_state_dict(state_dict)

# ViT-Large/14
model = timm.create_model("vit_large_patch14_224", pretrained=False, num_classes=0, dynamic_img_size=True)
state_dict = torch.load("visreg-vit-l-inet1k.pth", map_location="cpu")
model.load_state_dict(state_dict)
```

<h3 style="font-size: 1.4em;">Download with huggingface_hub</h3>

```python
from huggingface_hub import hf_hub_download

# ViT-Base/16
path = hf_hub_download(repo_id="BooBooWu/visreg", filename="visreg-vit-b-inet1k.pth")

# ViT-Large/14
path = hf_hub_download(repo_id="BooBooWu/visreg", filename="visreg-vit-l-inet1k.pth")
```

<h3 style="font-size: 1.4em;">Feature extraction</h3>

```python
from PIL import Image
from torchvision import transforms

transform = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])

img = transform(Image.open("image.jpg")).unsqueeze(0)

with torch.no_grad():
    features = model(img)  # [1, embed_dim]
```

<h2 style="font-size: 1.8em;">Evaluation</h2>

Full evaluation suite (linear probe, segmentation, fine-tuning) is available in the [GitHub repo](https://github.com/HaiyuWu/visreg):

```bash
# Linear probe on 10+ datasets
python downstream/linear_prob/run_evaluation.py \
    --checkpoint visreg-vit-b-inet1k.pth \
    --model vit_b \
    --datasets all
```

<h2 style="font-size: 1.8em;">Citation</h2>

```bibtex
@inproceedings{wu2026visreg,
  title     = {VISReg: Variance-Invariance-Sketching Regularization for JEPA training},
  author    = {Wu, Haiyu and Balestriero, Randall and Levine, Morgan},
  booktitle = {arXiv},
  year      = {2026}
}
```

<h2 style="font-size: 1.8em;">License</h2>

This project (code and pretrained weights) is released under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/) for non-commercial use only.