A chatbot for emotional dialog generation built in TensorFlow 1.9 and modeled after the approaches described in Asghar et al. and Koshla et al..
Check out examples to see how to train a model and then perform inference with it after training. The project's two "main" scripts are train.py and infer.py.
To generate some more expansive documentation, simply run gen_doc.sh.
- Python 3
- tensorflow 1.9 or higher
- gensim (for training Word2Vec embeddings)
- PyYAML (for configuring a model)
- numpy
- sklearn (for principal component analysis)
- pandas (for spreadsheet processing)
- spacy (for preprocessing package)
- en_core_web_sm language model
- matplotlib (for analysis package)
- sphinx 1.7.6 or higher (for generating documentation)