VisionHOPE: Visual Backbones as Self-Modifying Learning Systems
Paper • 2609.33325 • Published • 2
VisionHOPE
Official pretrained weights for VisionHOPE: Visual Backbones as Self-Modifying Learning Systems.
This repository contains the hierarchical VisionHOPE-T/S/B checkpoints for ImageNet-1K classification, COCO object detection and instance segmentation, and ADE20K semantic segmentation.
Pretrained weights
| Task | VisionHOPE-T | VisionHOPE-S | VisionHOPE-B |
|---|---|---|---|
| ImageNet-1K classification | Download | Download | Download |
| COCO · Mask R-CNN 1× | Download | Download | Download |
| COCO · Mask R-CNN 3× | Download | Download | — |
| ADE20K · UPerNet | Download | Download | Download |
Download
Download an individual checkpoint with the Hugging Face CLI:
hf download PSRben/VisionHOPE visionhope_small.pth --local-dir weights
To download all checkpoints:
hf download PSRben/VisionHOPE --include "visionhope_*.pth" --local-dir weights
See the code repository for installation, model definitions, and training and evaluation commands.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support