Memory-Efficient Implementation of DenseNets

Code

• github.com/gpleiss/efficient_densenet_pytorch
  OfficialIn paperpytorch★ 0
• github.com/npielawski/pytorch_tiramisu
  pytorch★ 0
• github.com/zhourunlong/densenet-ag_news-nlp
  pytorch★ 0
• github.com/facebookresearch/ResNeXt
  torch★ 0
• github.com/code-implementation1/Code2/tree/main/3D_DenseNet
  mindspore★ 0
• github.com/joeyearsley/efficient_densenet_tensorflow
  tf★ 0

Abstract

The DenseNet architecture is highly computationally efficient as a result of feature reuse. However, a naive DenseNet implementation can require a significant amount of GPU memory: If not properly managed, pre-activation batch normalization and contiguous convolution operations can produce feature maps that grow quadratically with network depth. In this technical report, we introduce strategies to reduce the memory consumption of DenseNets during training. By strategically using shared memory allocations, we reduce the memory cost for storing feature maps from quadratic to linear. Without the GPU memory bottleneck, it is now possible to train extremely deep DenseNets. Networks with 14M parameters can be trained on a single GPU, up from 4M. A 264-layer DenseNet (73M parameters), which previously would have been infeasible to train, can now be trained on a single workstation with 8 NVIDIA Tesla M40 GPUs. On the ImageNet ILSVRC classification dataset, this large DenseNet obtains a state-of-the-art single-crop top-1 error of 20.26%.

Tasks

Reproductions