A deep learning based sequence model used to predict forthcoming words in a sentence. In order to train this model, the popular drama/indie screenplay: Good Will Hunting was used. This corpus comprised of a total of 3293 unique words. Data preprocessing steps included removing whitespaces from the text file, converting the texts to lowercase and tokenizing them. The model uses 3 layers in total: 1) An Eembedding Layer 2) A bidirectional LSTM layer with 150 units 3) A Dense Layer with softmax activation.
- Generate auto-replies
- Spell check/Autocorrection
- Sentence completion
-
Pre-requisites:
- TensorFlow 2.X
- Streamlit
- Python 3.5-3.8
-
Use pip3 to install the current stable release of TensorFlow 2.X in your system
$ pip3 install --user --upgrade tensorflow- Install Streamlit
$ pip install streamlit- Executing the source file
$ streamlit run app.pyFor explanation and intuition behind the source code, refer to the contributors article at: https://towardsdatascience.com/how-our-device-thinks-e1f5ab15071e
|
Ujjwal Kumar
|
Made with ❤️ by DS Community SRM