Skip to content

SEQ2SEQ model with Attention mechanism for QA also for NMT +(Generating text using LSTM network)

License

Notifications You must be signed in to change notification settings

mahdiabdollahpour/Neural-Dialogue-System

Repository files navigation

"# RNN-Language-Model + SEQ2SEQ model for question answering"

Two main approaches

1- character level

2- word level

using LSTM layers to model recurrent neural network to predict next word or char given the previous ones

How To Use This Code

Creating QA Model

(will be completed soon)

Using Pretrained Model on Daily Dialog Dataset

(will be completed soon)

Your Dataset Format

(will be completed soon)

TO DO

  • Using state_is_tuple in char level

  • Printing number of trainable parameter

  • Training the model

  • Using Estimator

  • Converting string data to indexes takes long not to do it every time

  • Code is dirty I've got to clean it

  • Saving is not handled in seq2seq

  • Plotting loss

  • Adding Dropout

  • Last batches are ignored , not getting to batch_size

  • Handle early stopping

About

SEQ2SEQ model with Attention mechanism for QA also for NMT +(Generating text using LSTM network)

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages