** please comment the wandb (wandb is used to track the training through web)
1/ setup the config you want in the config.json
2/ start training tf1 by train_tf1.py
3/ edit the quantization.py to correct input and output feature extractor => tflite