Skip to content

aaamourao/stochastic_depth_keras

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Deep Networks with Stochastic Depth

Rewriting code for "Deep Networks with Stochastic Depth" in Keras.

Original code is at yueatsprograms/Stochastic_Depth.

Fork by @aaamourao : adriano-dev branch

mourao.aaa@gmail.com

Encrypted mail: adrianomourao@protonmail.com

Usage

System requirements

You need to install the following libraries:

  • Developer version of libblas
  • Developer version of liblapack
  • Developer version of libatlas
  • GNU Fortran
Fedora install instructions
sudo dnf install blas blas-devel lapack lapack-devel atlas atlas-devel gcc-fortran
Ubuntu install instructions
sudo apt-get install libblas-dev liblapack-dev libatlas-base-dev gfortran

[optional] Create a virtual environment

Run the following commands and replace <virtual-dir-path> for your target directory on your workspace

virtualenv --distribute <virtual-dir-path>
source <virtual-dir-path>/bin/activate

Install python requirements

There are some python libraries required to run this project

pip install -r requirements.txt

Hint : if pip doesn't find Keras library install it manually:

pip install -r requirements.txt```

Install python libraries from repo

Just run python train.py

Results

  • Initial learning rate == 0.1
  • Number of layers == 50
  • (other configs are same as train.py)

results

About

Rewriting code for "Deep Networks with Stochastic Depth" http://arxiv.org/abs/1603.09382 in keras

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%