This repository is an update from YU1ut/MixMatch-pytorch to work with the latest version of PyTorch and torchvision, and follow modern best practices for PyTorch code.

This README is copied from YU1ut/MixMatch-pytorch

This is an unofficial PyTorch implementation of MixMatch: A Holistic Approach to Semi-Supervised Learning. The official Tensorflow implementation is here.

Now only experiments on CIFAR-10 are available.

This repository carefully implemented important details of the official implementation to reproduce the results.

• Python 3.6+
• PyTorch 1.0
• torchvision 0.2.2 (older versions are not compatible with this code)
• tensorboardX
• progress
• matplotlib
• numpy

Train the model by 250 labeled data of CIFAR-10 dataset:

python train.py --gpu <gpu_id> --n-labeled 250 --out cifar10@250

Train the model by 4000 labeled data of CIFAR-10 dataset:

python train.py --gpu <gpu_id> --n-labeled 4000 --out cifar10@4000

tensorboard.sh --port 6006 --logdir cifar10@250

(Results of this code were evaluated on 1 run. Results of 5 runs with different seeds will be updated later. )

@article{berthelot2019mixmatch,
  title={MixMatch: A Holistic Approach to Semi-Supervised Learning},
  author={Berthelot, David and Carlini, Nicholas and Goodfellow, Ian and Papernot, Nicolas and Oliver, Avital and Raffel, Colin},
  journal={arXiv preprint arXiv:1905.02249},
  year={2019}
}