See the relevant 🤗 Model Card for this large multilabel text classificaiton model that I fine-tuned atop the RoBERTa-base checkpoint. This model utilizes a set of classes constructed as a work around to a current bug in the 🤗 Evaluate library. The temporary code can be found in the custom_eval_classes folder, as well as under 🤗 Spaces where it is briefly demonstrated. The distributed training notebook is found in the RoBERTa-base-DReiFT.ipynb file.