Implementation of CIFAR-10 classifier with PyTorch

• Python 2.7
• PyTorch
• LMDB
• Visdom

1. Convert to LMDB format

   $ python convert_to_lmdb.py --data_dir ../data
   

2. (Optional) Test for reading LMDBs

   Open read_lmdb_sample.ipynb in Jupyter

3. Train

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

4. Retrain if you need

   $ python train.py --data_dir ./data --logdir ./logs_retrain --restore_checkpoint ./logs/model-100.tar
   

5. Evaluate

   $ python eval.py --data_dir ../data ./logs/model-100.tar
   

6. Visualize

   $ python -m visdom.server
   $ python visualize.py --logdir ./logs
   

7. (Optional) Try to make an inference

   Open inference_sample.ipynb in Jupyter

8. Clean

   $ rm -rf ./logs
   or
   $ rm -rf ./logs_retrain