MNIST Challenge 2017 by TF Kr

MNIST classification using residual network

Result

Requirements

• Python 2.7
• TensorFlow 1.1.0

Environments

• NVIDIA GTX1080 (gpu mem > 7GB)

Usage

Input data

└── data
    ├── t10k-images-idx3-ubyte.gz
    ├── t10k-labels-idx1-ubyte.gz
    └── ... 

First, create checkpoint dir and download trained parameter files

└── model
    └── mnist_weight
        ├── checkpoint 
        ├── resnet-model-xx.index
        └── ...

You can download CHECKPOINT files (you don't need download. ckpt files were already included.)

To test a model

$ python testing_mnist.py  

To train a model

$ python training_mnist.py

Weight visualization

REF

code : https://github.com/tensorflow/models/tree/master/resnet

Author

tkwoo

---

| |__ _ _ |_ | | |/ / \ \ / /_ ___
| '_ | | | | | | | ' / \ \ /\ / / _ \ / _ \ | |) | || | | |_ | . \ _ \ V V / () | () | |.__/ _, | |() ||_() _/_/ _/ _/ |___/