benbergner/cropr
A token pruning method that accelerates ViTs for various tasks while maintaining high performance.
This tool helps machine learning engineers and researchers accelerate the performance of Vision Transformers (ViTs) for computer vision tasks. It takes an existing ViT model and training data, and outputs a faster ViT that processes images more quickly while maintaining high accuracy. This is ideal for those deploying ViTs in applications where processing speed is critical.
No commits in the last 6 months.
Use this if you need to speed up your Vision Transformer models for tasks like image classification, object detection, or semantic segmentation without significant loss of accuracy.
Not ideal if your primary concern is developing new ViT architectures from scratch rather than optimizing existing ones for performance.
Stars
27
Forks
2
Language
Python
License
MIT
Category
Last pushed
Jul 21, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/benbergner/cropr"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Jittor/jittor
Jittor is a high-performance deep learning framework based on JIT compiling and meta-operators.
berniwal/swin-transformer-pytorch
Implementation of the Swin Transformer in PyTorch.
zhanghang1989/ResNeSt
ResNeSt: Split-Attention Networks
NVlabs/FasterViT
[ICLR 2024] Official PyTorch implementation of FasterViT: Fast Vision Transformers with...
ViTAE-Transformer/ViTPose
The official repo for [NeurIPS'22] "ViTPose: Simple Vision Transformer Baselines for Human Pose...