https://github.com/akihironitta/calconv

A simple calculator for size of neural networks
https://github.com/akihironitta/calconv

Last synced: 26 days ago
JSON representation

A simple calculator for size of neural networks

• Host: GitHub
• URL: https://github.com/akihironitta/calconv
• Owner: akihironitta
• License: gpl-3.0
• Created: 2020-02-12T08:20:28.000Z (almost 5 years ago)
• Default Branch: master
• Last Pushed: 2020-02-12T08:23:57.000Z (almost 5 years ago)
• Last Synced: 2024-12-05T17:52:07.825Z (about 1 month ago)
• Language: Python
• Size: 15.6 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.org
  • License: LICENSE

Awesome Lists containing this project

README

        
* NAME

  =calconv= - A simple calculator for size of neural networks.

* DESCRIPTION

  =calconv= calculates output size of single or multiple convolution layers.

  While =torchsummary= gets information about size of each layer by passing a random tensor through the network

  and thus need to actually allocate memory, =calconv= only uses parameters of convolution layers and works even under very limited resources.

  =calconv= is still the best choice on multiple GPUs

  when you have to deal with very high resolution (VHR) imagery like satellite imagery.