Download your dataset in baidu yun

run the command python chinese_rec.py --mode=train --max_steps=200000 --eval_steps=1000 --save_steps=10000

run the command python chinese_rec.py --mode=validation

run the command python chinese_rec.py--mode=inference

The detailed you can get in TensorFlow与中文手写汉字识别

TODOs：

• delete the placeholders in chinese_rec.py