When you initialize Editor, use the argument model_name (default is roberta-base). The backend will initialize a tokenizer and a model using AutoTokenizer.from_pretrained(model_name) and AutoModelForMaskedLM.from_pretrained(model_name), so either a name or a path should work.
When you initialize
Editor
, use the argumentmodel_name
(default isroberta-base
). The backend will initialize a tokenizer and a model usingAutoTokenizer.from_pretrained(model_name)
andAutoModelForMaskedLM.from_pretrained(model_name)
, so either a name or a path should work.