Skip to content

Latest commit

 

History

History
34 lines (22 loc) · 903 Bytes

README.md

File metadata and controls

34 lines (22 loc) · 903 Bytes

A chatbot created using with the help of tensorflow, incorporating an Encoder - Decoder model ( RNN ) with bahdanau attention and GRU's

Files Description :

  1. Neural machine traslation model - NMT.ipynb

  2. Dataset - cornell_movie_dialogs_corpus

  3. Trained model Checkpoints - training_checkpoints , training_checkpoints1

  4. PreProcessing of Data - bagofwords.py , combining.py , conversations.py , conversations.pickle , dummy.py ( To View processed Data )

  5. Feature Creation - Features.py ( Creating features and labels and storing in pickle format )

The machine translation jupyter notebook can be trained for higher epochs to gain better accuracy , or pretrained (100 epochs) can be used to see the results. The NMT.ipynb contains the model and a translator function in the end , which takes in a sentance as an input and returns what the model thinks the reply should be