A demonstration of the capabilities of keras embedding layer with Bidirectional GRU. Spacy 3 is used for tokenization, and training is done on google colab. This model is is trained on the first book of harry potter.
Info : Please note that this model has only been trained for 10 minutes. This model is just a just a demonstration, so its in early stage of training. Model trained for longer duration of time can provide more accurate results.
Requirements : Python 3 + Tensorflow + Numpy
Dataset used : https://www.kaggle.com/datasets/utsavk02/4-light-novel-for-text-generation
Intructions : Execute runmodel.py after making sure that the model and the tokenizer files are in the same directory
Other info : 'Showcase.jpg' contains a demonstration of the model's capabilities.