Refer to mlm_environment.yml
. This method mainly relies on huggingface's transformers library.
This uses a depreciated library bert-embeddings
. Thus, it would be difficult to set up using conda. A reproducible environment would be to use python=3.7 and then run pip install bert-embeddings
.
This jupyter notebook has been used to easily explore/test our method through google colab. It will output the top_10_100000_corpus.json file.
This jupyter notebook has been used to analyze the euphemisms generated by the baselines
This jupyter notebook has been used to analyze the euphemisms generated by the ConSim. It also has been used to scrape videos for good euphemistic hashtags for each of the examined euphemism generation method.