In the textual domain, pre-trained language models such as Bidirectional Encoder Representations from Transformers (BERT) have proven to be helpful as they provide a good baseline for further fine-tuning. However, due to the domain-knowledge and the many technical terms in cysec, general language...