I am using BERT model for sentence similarity task. However my dataset with sentence is very specific and I want to fine tune my model on it first. My dataset is unlabelled. And BERT model that I want to use was trained with natural language inference method (where sentences are labeled as neutral, entailment or contradiction. I found articles about fine tuning BERT model with MLM method, but I don't think that I can apply to my model since it was trained in different way. Is it correct or MLM is applicable for fine tuning model trained with NLI method?