Eval.py uses the output of the Template library and the HuBert-NER models' output for CHECKLIST evaluation. The predictions are generated by the version of HuBERT-NER on the 20th November 2020.