Transformer with Expanded Decoder

Functions of each folder

1. data. You can follow the instruction of the Readme in the data folder and get the Filtered Reddit and Wizard data sets. Put them in the data folder.
2. configuration. Set the file path in the config.ini to fit with your system. You can just set the Reddit and Wizard path and do nothing to the rest of setting.
3. run_script. The models tunable parameters are able to be set in the args_parser.py. You can also keep it to default.
4. commonly_used_code. There are some commonly used code in this folder.
5. models. The TED model is in this folder.
6. example. The training process entrance is in this folder.

For more models and usage, please refers to https://github.com/tonywenuon/keras_dialogue_generation_toolkit.