Fetching the paper…

CodePurify: Defend Backdoor Attacks on Neural Code Models via Entropy-based Purification · Around