Reproduces ResNet-V3 (Aggregated Residual Transformations for Deep Neural Networks) with pytorch.
- Trains on Cifar10 and Cifar100
- Upload Cifar Training Curves
- Upload Cifar Trained Models
- Train Imagenet
To train on Cifar-10 using 2 gpu:
python train.py ~/DATASETS/cifar.python cifar10 -s ./snapshots --log ./logs --ngpu 2 --learning_rate 0.05 -b 128
It should reach ~3.65% on Cifar-10, and ~17.77% on Cifar-100.
From the original paper:
cardinality | widen_factor | parameters | accuracy cifar10 | accuracy cifar100 | default |
---|---|---|---|---|---|
8 | 4 | 34.4M | 3.65 | 17.77 | x |
16 | 64 | 68.1M | 3.58 | 17.31 |
Link to trained models corresponding to the following curves:
- torch (@facebookresearch). (Original) Cifar and Imagenet
- caffe (@terrychenism). Imagenet
- MXNet (@dmlc). Imagenet
@article{xie2016aggregated,
title={Aggregated residual transformations for deep neural networks},
author={Xie, Saining and Girshick, Ross and Doll{\'a}r, Piotr and Tu, Zhuowen and He, Kaiming},
journal={arXiv preprint arXiv:1611.05431},
year={2016}
}