You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is a simple chatbot Project in English. Using Java swing to develop GUI and python pytorch
to do the trainning and neural model.
This Project only achieve seq2seq model and do not accomplish parallel training (Thus slow in training)
This Project is modified base on original Project : https://github.com/zhaoyingjun/chatbot (A Great Thanks!)
Explain to Each Folder and File:
ChatApp/: All Java GUI implemetation
config/: all resource Path and some static parameter
icons/: image resources
train_data/: all train data and trained model
data.py: formatting all the dialog inputs and creating dataset
execute.py: do training
runner.py: interface between Java GUI and pytorch model
seq2seqModel.py: model class
There is already a 30-epoch training model base on the given dialog and wordlist
So if just want to run without any more training, then type "python runner.py" in terminal
If want to redo training by yourself, delete train_data/data_model.pt, and run "python execute.py"
If want to redo training with your own data, delete train_data/data_model.pt, train_data/data_running.pt.
And Strongly recommend go over the code data.py and execute.py to format your new dataset.