Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/prlz77/ResNeXt.pytorch
Reproduces ResNet-V3 with pytorch
https://github.com/prlz77/ResNeXt.pytorch
cifar pytorch resnet resnext
Last synced: 14 days ago
JSON representation
Reproduces ResNet-V3 with pytorch
- Host: GitHub
- URL: https://github.com/prlz77/ResNeXt.pytorch
- Owner: prlz77
- License: mit
- Created: 2017-03-30T07:35:24.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2020-08-01T23:01:12.000Z (almost 4 years ago)
- Last Synced: 2024-02-04T21:58:54.542Z (4 months ago)
- Topics: cifar, pytorch, resnet, resnext
- Language: Python
- Homepage:
- Size: 154 KB
- Stars: 500
- Watchers: 15
- Forks: 124
- Open Issues: 7
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Lists
- awesome-image-classification - unofficial-pytorch : https://github.com/prlz77/ResNeXt.pytorch
- Awesome-pytorch-list - ResNeXt.pytorch - V3 (Aggregated Residual Transformations for Deep Neural Networks) with pytorch. (Paper implementations / Other libraries:)
- Awesome-pytorch-list-CNVersion - ResNeXt.pytorch - V3 (深度神经网络的聚集残差变换)。 (Paper implementations｜论文实现 / Other libraries｜其他库:)
- awesome-image-classification - unofficial-pytorch : https://github.com/prlz77/ResNeXt.pytorch
README
# ResNeXt.pytorch
Reproduces ResNet-V3 (Aggregated Residual Transformations for Deep Neural Networks) with pytorch.
- [x] Tried on pytorch 1.6
- [x] Trains on Cifar10 and Cifar100
- [x] Upload Cifar Training Curves
- [x] Upload Cifar Trained Models
- [x] Pytorch 0.4.0
- [ ] Train Imagenet
## Download
```bash
git clone https://github.com/prlz77/resnext.pytorch
cd resnext.pytorch
# git checkout R4.0 or R3.0 for backwards compatibility (not recommended).
```
## Usage
To train on Cifar-10 using 2 gpu:
```bash
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.
After train phase, you can check saved model.
**Thanks to [@AppleHolic](https://github.com/AppleHolic) we have now a test script:**
To test on Cifar-10 using 2 gpu:
```bash
python test.py ~/DATASETS/cifar.python cifar10 --ngpu 2 --load ./snapshots/model.pytorch --test_bs 128
```
## Configurations
From [the original paper](https://arxiv.org/pdf/1611.05431.pdf):
| cardinality | base_width | parameters | Error cifar10 | error cifar100 | default |
|:-----------:|:------------:|:----------:|:----------------:|:-----------------:|:-------:|
| 8 | 64 | 34.4M | 3.65 | 17.77 | x |
| 16 | 64 | 68.1M | 3.58 | 17.31 | |
**Update:** ``widen_factor`` has been disentangled from ``base_width`` because it was confusing. Now widen factor is set to consant 4, and ``base_width`` is the same as in the original paper.
## Trained models and curves
[Link](https://mega.nz/#F!wbJXDS6b!YN3hCDi1tT3SdNFrLPm7mA) to trained models corresponding to the following curves:
**Update:** several commits have been pushed after training the models in Mega, so it is recommended to revert to ``e10c37d8cf7a958048bc0f58cd86c3e8ac4e707d``
## Other frameworks
* [torch (@facebookresearch)](https://github.com/facebookresearch/ResNeXt). (Original) Cifar and Imagenet
* [caffe (@terrychenism)](https://github.com/terrychenism/ResNeXt). Imagenet
* [MXNet (@dmlc)](https://github.com/dmlc/mxnet/tree/master/example/image-classification#imagenet-1k). Imagenet
## Cite
```
@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}
}
```