Using Recurrent Neural Network architecture for predicting new script for Famous American Sitcom Friends.
I have taken dataset from a GitHub of Uragiri
I used popular sitcom Friends Dataset for generating new scripts using RNN architecture
Info on Friends sitcom(1994-2004) can be found here
I have used Kaggle notebook for demonstrating this project and due to availinility of GPU/TPU on it. You can also use Colab for this
I have taken insights from tensorflow's original notebook for understanding and analyzing the script data
Easy to understand approach was made and you can see it either here on the repo or in the above mentioned Kaggle Notebook
RNN model architecture used
Loss Plot
Model layout
Predictions
Although after 30 iterations still there are some grammatical errors maybe more number of iterations can be considered as an alternative. Hope this helps you out!!
Do upvote my notebook on Kaggle and comment if any changes can be made, Happy learning!!