Yeah considering in terms of a binary classification the accuracy is not super impressive imo. It could just be learning statistical patterns for code that is m

Yeah considering in terms of a binary classification the accuracy is not super impressive imo. It could just be learning statistical patterns for code that is more likely to have issues rather than identifying actual bugs. They did mention is does other things such as suggesting improvements, so maybe it does provide more tangible value in those ways. Hard to say without more information!