Implementation of CIFAR-10 classifier with TensorFlow

• Python 2.7
• Tensorflow

1. Convert to TFRecords format

   $ python convert_to_tfrecords.py --data_dir ./data
   

2. (Optional) Test for reading TFRecords files

   Open `read_tfrecords_sample.ipynb` in Jupyter
   

3. Train

   $ python train.py --data_dir ./data --train_logdir ./logs/train
   

4. Retrain if you need

   $ python train.py --data_dir ./data --train_logdir ./logs/train2 --restore_checkpoint ./logs/train/latest.ckpt
   

5. Evaluate

   $ python eval.py --data_dir ./data --checkpoint_dir ./logs/train --eval_logdir ./logs/eval
   

6. Visualize

   $ tensorboard --logdir ./logs
   

7. (Optional) Try to make an inference

   Open `inference_sample.ipynb` in Jupyter
   

8. Clean

   $ rm -rf ./logs
   or
   $ rm -rf ./logs/train2
   or
   $ rm -rf ./logs/eval