DynamicHate_distilgpt2

distilgpt2 fine tuned on the "Dynamically Generated Hate Speech Dataset"

dataset: https://github.com/bvidgen/Dynamically-Generated-Hate-Speech-Dataset

Trained for 75 epochs on a single Nvidia Quadro M4000 gpu

data set was from csv and did not include line breaks, proof of concept, would need a larger database and proper preprocessing for clean results

this model could be useful to automatically generate large databases of hate speech for text classification/sentiment analysis for moderation on online platforms