This is the Python 3 companion codebase for the NAACL-HLT 2019 paper Evaluating Style Transfer for Text.
📁 code - metrics to evaluate outputs of examined style transfer models
📁 data - training and test datasets used in experiments
📁 evaluations - human and automated evaluations of style transfer model outputs
📁 models - miscellaneous models used for running experiments (not the style transfer models themselves)
📁 style_lexicon - words used to compile style lexicon used in evaluation of content preservation
📁 transfer_model_outputs - outputs of examined style transfer models
Dependencies: gensim, keras, matplotlib, pyemd, scipy, sklearn