Steerable Visual Representations
Paper • 2604.02327 • Published • 55
Unchanged copies of the commercially usable SteerViT checkpoints by Jona Ruthardt, Manu Gaur, Deva Ramanan, Makarand Tapaswi and Yuki M. Asano. Cubert GmbH mirrors them with the authors' permission, so that our software can pin and download them reproducibly.
1846972410faf266b9153253fd909b73b1f30346LICENSE), as stated by the authors for the checkpoints labelled "commercial".Only the upstream checkpoints labelled "commercial" are mirrored here. The upstream research-only checkpoints
(steervit_dinov2_base.pth, steervit_mae_base.pth, steervit_siglip_base.pth, steervit_siglip2_base.pth) are not
part of this repository.
| File | Backbone | Size (bytes) | sha256 |
|---|---|---|---|
steervit_dinov2_commercial_small.pth |
DINOv2-Small | 48 284 373 | 6c23756039ebddb01c262710cb2f9ea774e905a6e134946b3176dbcc96c969ee |
steervit_dinov2_commercial_base.pth |
DINOv2-Base | 92 345 957 | f863021d5aaedbfd4b11d5be639936f56cfe3111ab90db6b8f9c60e40ee64833 |
steervit_dinov2_commercial_large.pth |
DINOv2-Large | 235 029 421 | 6c7702a5b194db96c4211f3445eacaf3693a86638e45250603fbcd3bbbb69bb6 |
Usage is the same as upstream: download a file from this repository and load it with
SteerViT.from_pretrained("steervit_dinov2_commercial_base.pth"); see the upstream model card for installation and the
quick start.
@inproceedings{ruthardt2026steervit,
title = {Steerable Visual Representations},
author = {Ruthardt, Jona and Gaur, Manu and Ramanan, Deva and Tapaswi, Makarand and Asano, Yuki M.},
booktitle = {European Conference on Computer Vision (ECCV)},
year = {2026}
}