资源分类
》 资源 》算法》 pytorch-revnet

pytorch-revnet

2020-02-17 | |  42 |   0 |   0

revnet

PyTorch implementation of the reversible residual network.

Requirements

The main requirement ist obviously PyTorch. CUDA is strongly recommended.

The training script requires tqdm for the progress bar.

The unittests require the TestCase implemented by the PyTorch project. The module can be downloaded here.

Note

The revnet models in this project tend to have exploding gradients. To counteract this, I used gradient norm clipping. For the experiments below you would call the following command:

python train_cifar.py --model revnet38 --clip 0.25

Results

CIFAR-10

ModelAccuracyMemory UsageParams
resnet3292.02%1271 MB0.47 M
revnet3891.98%660 MB0.47 M


上一篇： revnet-public

下一篇：hamiltonian-revnet

用户评价
全部评价

热门资源

  • Keras-ResNeXt

    Keras ResNeXt Implementation of ResNeXt models...

  • seetafaceJNI

    项目介绍 基于中科院seetaface2进行封装的JAVA...

  • spark-corenlp

    This package wraps Stanford CoreNLP annotators ...

  • capsnet-with-caps...

    CapsNet with capsule-wise convolution Project ...

  • inferno-boilerplate

    This is a very basic boilerplate example for pe...