Skip to content

Cuongvn08/tensorflow_kaggle_mnist

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

A tensorflow based standard NN for Kaggle mnist competition

This project presents how to use Tensorflow to implement a standard NN for Kaggle mnist competition.

  • config.py: consists of all settings, hyper-parameters
  • data.py: to load data from csv and separate into data and label
    • train.csv, test.csv: were downloaded from kaggle homepage, mnist project
  • model.py: implement Model class to design expected model, get logit and weights or biases
  • train.py: run this file to train and evaluate the model after setting, automatically save logs and checkpoint into result folder
  • run.py: set the best step with the lowest cost and run this file to get prediction of test data and submit.

The best score of this model is 0.995 (Top 8%). Hyper-parameters tunning and more complex model would get better result.

Development enviroment:

  • Window 10
  • Anaconda 3
  • Python 3.5
  • Tensorflow 1.1
  • Pycharm community

About

[Done] [Top 7%] A tensorflow based standard NN for Kaggle mnist competition (score 0.995)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages