At the bottom of the main.py code, there is a code for setting the model_path and dataset_path as follows.
If you run the main.py code, model and dataset file are saved in the data folder as a pickle file. Description of the files are as follows.
- model: It contains the network variables. Therefore, you must write this, if you want to continue training or test the algorithm.
- dataset: It contains replay memory data. If you don't add it and only load the model, It continues training with empty replay memory. Therefore, it is not an essential thing when loading the model, but I recommend to load it.