Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
alpha_magic 931d8d0370 | 1 year ago | |
---|---|---|
RVT | 1 year ago | |
easyrobust | 1 year ago | |
easyrobust.egg-info | 1 year ago | |
test_scripts | 1 year ago | |
train_configs/imagenet | 1 year ago | |
LICENSE | 1 year ago | |
README.md | 1 year ago | |
inference.py | 1 year ago | |
main.py | 1 year ago | |
main_local.py | 1 year ago | |
model.py | 1 year ago | |
requirements.txt | 1 year ago | |
setup.py | 1 year ago | |
train.sh | 1 year ago |
EasyRobust is a tool for training your robust models. Now it support adversarial / non-adversarial training of CNN / ViT models on ImageNet.
git clone https://github.com/thu-ml/ares/tree/main/easyrobust
cd easyrobust
pip install -e .
sh train.sh train_configs/imagenet/resnet50_baseline.yaml
see test_scripts
train_configs/imagenet/deit_small_baseline.yaml
: baseline training on deit_smalltrain_configs/imagenet/resnet50_baseline.yaml
: baseline training on resnet50train_configs/imagenet/advtrain/resnet50_advtrain.yaml
: adversarial training on resnet50train_configs/imagenet/advtrain/deit_small_advtrain.yaml
: adversarial training on deit_smallMore training templates will be supported in future.
All data augmentation in timm.
StyleAugmentation Style Augmentation: Data Augmentation via Style Randomization
CartoonAugmentation CartoonGAN: Generative Adversarial Networks for Photo Cartoonization
All implemented models in timm.
WaveCNet: WaveCNet: Wavelet Integrated CNNs to Suppress Aliasing Effect for Noise-Robust Image Classification
DrViT: Discrete Representations Strengthen Vision Transformer Robustness
BlurPool: Making Convolutional Networks Shift-Invariant Again
GaussianPool: Gaussian-Based Pooling for Convolutional Neural Networks
SelfNorm: CrossNorm and SelfNorm for Generalization under Distribution Shifts
pAdaIN: Permuted AdaIN: Reducing the Bias Towards Global Statistics in Image Classification
CrossNorm: CrossNorm and SelfNorm for Generalization under Distribution Shifts
Adversarial Training: Towards Deep Learning Models Resistant to Adversarial Attacks
GNT: A simple way to make neural networks robust against diverse image corruptions
Shape-Texture Debiased: Shape-Texture Debiased Neural Network Training
ProbCompactLoss: Improving Adversarial Robustness via Probabilistically Compact Loss with Logit Constraints
No Description
Python Text other
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》